disk-io.c 110.7 KB
Newer Older
C
Chris Mason 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
/*
 * Copyright (C) 2007 Oracle.  All rights reserved.
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public
 * License v2 as published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * General Public License for more details.
 *
 * You should have received a copy of the GNU General Public
 * License along with this program; if not, write to the
 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
 * Boston, MA 021110-1307, USA.
 */

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

C
Chris Mason 已提交
76 77 78 79 80
/*
 * end_io_wq structs are used to do processing in task context when an IO is
 * complete.  This is used during reads to verify checksums, and it is used
 * by writes to insert metadata for new file extents after IO is complete.
 */
81 82 83 84 85 86
struct end_io_wq {
	struct bio *bio;
	bio_end_io_t *end_io;
	void *private;
	struct btrfs_fs_info *info;
	int error;
87
	int metadata;
88
	struct list_head list;
89
	struct btrfs_work work;
90
};
91

C
Chris Mason 已提交
92 93 94 95 96
/*
 * async submit bios are used to offload expensive checksumming
 * onto the worker threads.  They checksum file and metadata bios
 * just before they are sent down the IO stack.
 */
97 98 99 100
struct async_submit_bio {
	struct inode *inode;
	struct bio *bio;
	struct list_head list;
C
Chris Mason 已提交
101 102
	extent_submit_bio_hook_t *submit_bio_start;
	extent_submit_bio_hook_t *submit_bio_done;
103 104
	int rw;
	int mirror_num;
C
Chris Mason 已提交
105
	unsigned long bio_flags;
106 107 108 109 110
	/*
	 * bio_offset is optional, can be used if the pages in the bio
	 * can't tell us where in the file the bio should go
	 */
	u64 bio_offset;
111
	struct btrfs_work work;
112
	int error;
113 114
};

115 116 117 118 119 120 121 122 123 124
/*
 * Lockdep class keys for extent_buffer->lock's in this root.  For a given
 * eb, the lockdep key is determined by the btrfs_root it belongs to and
 * the level the eb occupies in the tree.
 *
 * Different roots are used for different purposes and may nest inside each
 * other and they require separate keysets.  As lockdep keys should be
 * static, assign keysets according to the purpose of the root as indicated
 * by btrfs_root->objectid.  This ensures that all special purpose roots
 * have separate keysets.
125
 *
126 127 128
 * Lock-nesting across peer nodes is always done with the immediate parent
 * node locked thus preventing deadlock.  As lockdep doesn't know this, use
 * subclass to avoid triggering lockdep warning in such cases.
129
 *
130 131 132
 * The key is set by the readpage_end_io_hook after the buffer has passed
 * csum validation but before the pages are unlocked.  It is also set by
 * btrfs_init_new_buffer on freshly allocated blocks.
133
 *
134 135 136
 * We also add a check to make sure the highest level of the tree is the
 * same as our lockdep setup here.  If BTRFS_MAX_LEVEL changes, this code
 * needs update as well.
137 138 139 140 141
 */
#ifdef CONFIG_DEBUG_LOCK_ALLOC
# if BTRFS_MAX_LEVEL != 8
#  error
# endif
142 143 144 145 146 147 148 149 150 151 152 153 154

static struct btrfs_lockdep_keyset {
	u64			id;		/* root objectid */
	const char		*name_stem;	/* lock name stem */
	char			names[BTRFS_MAX_LEVEL + 1][20];
	struct lock_class_key	keys[BTRFS_MAX_LEVEL + 1];
} btrfs_lockdep_keysets[] = {
	{ .id = BTRFS_ROOT_TREE_OBJECTID,	.name_stem = "root"	},
	{ .id = BTRFS_EXTENT_TREE_OBJECTID,	.name_stem = "extent"	},
	{ .id = BTRFS_CHUNK_TREE_OBJECTID,	.name_stem = "chunk"	},
	{ .id = BTRFS_DEV_TREE_OBJECTID,	.name_stem = "dev"	},
	{ .id = BTRFS_FS_TREE_OBJECTID,		.name_stem = "fs"	},
	{ .id = BTRFS_CSUM_TREE_OBJECTID,	.name_stem = "csum"	},
155
	{ .id = BTRFS_QUOTA_TREE_OBJECTID,	.name_stem = "quota"	},
156 157 158
	{ .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"	},
159
	{ .id = BTRFS_UUID_TREE_OBJECTID,	.name_stem = "uuid"	},
160
	{ .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);
303 304 305 306 307
			printk_ratelimited(KERN_INFO
				"BTRFS: %s checksum verify failed on %llu wanted %X found %X "
				"level %d\n",
				root->fs_info->sb->s_id, buf->start,
				val, found, btrfs_header_level(buf));
308 309
			if (result != (char *)&inline_result)
				kfree(result);
310 311 312
			return 1;
		}
	} else {
313
		write_extent_buffer(buf, result, 0, csum_size);
314
	}
315 316
	if (result != (char *)&inline_result)
		kfree(result);
317 318 319
	return 0;
}

C
Chris Mason 已提交
320 321 322 323 324 325
/*
 * we can't consider a given block up to date unless the transid of the
 * block matches the transid in the parent node's pointer.  This is how we
 * detect blocks that either didn't get written at all or got written
 * in the wrong place.
 */
326
static int verify_parent_transid(struct extent_io_tree *io_tree,
327 328
				 struct extent_buffer *eb, u64 parent_transid,
				 int atomic)
329
{
330
	struct extent_state *cached_state = NULL;
331 332 333 334 335
	int ret;

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

336 337 338
	if (atomic)
		return -EAGAIN;

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

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

		if (ret && btrfs_super_generation(disk_sb) < 10) {
386 387
			printk(KERN_WARNING
				"BTRFS: super block crcs don't match, older mkfs detected\n");
388 389
			ret = 0;
		}
D
David Sterba 已提交
390 391 392
	}

	if (csum_type >= ARRAY_SIZE(btrfs_csum_sizes)) {
393
		printk(KERN_ERR "BTRFS: unsupported checksum algorithm %u\n",
D
David Sterba 已提交
394 395 396 397 398 399 400
				csum_type);
		ret = 1;
	}

	return ret;
}

C
Chris Mason 已提交
401 402 403 404
/*
 * helper to read a given tree block, doing retries as required when
 * the checksums don't match and we have alternate mirrors to try.
 */
405 406
static int btree_read_extent_buffer_pages(struct btrfs_root *root,
					  struct extent_buffer *eb,
407
					  u64 start, u64 parent_transid)
408 409
{
	struct extent_io_tree *io_tree;
410
	int failed = 0;
411 412 413
	int ret;
	int num_copies = 0;
	int mirror_num = 0;
414
	int failed_mirror = 0;
415

416
	clear_bit(EXTENT_BUFFER_CORRUPT, &eb->bflags);
417 418
	io_tree = &BTRFS_I(root->fs_info->btree_inode)->io_tree;
	while (1) {
419 420
		ret = read_extent_buffer_pages(io_tree, eb, start,
					       WAIT_COMPLETE,
421
					       btree_get_extent, mirror_num);
422 423
		if (!ret) {
			if (!verify_parent_transid(io_tree, eb,
424
						   parent_transid, 0))
425 426 427 428
				break;
			else
				ret = -EIO;
		}
C
Chris Mason 已提交
429

430 431 432 433 434 435
		/*
		 * 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))
436 437
			break;

438
		num_copies = btrfs_num_copies(root->fs_info,
439
					      eb->start, eb->len);
C
Chris Mason 已提交
440
		if (num_copies == 1)
441
			break;
C
Chris Mason 已提交
442

443 444 445 446 447
		if (!failed_mirror) {
			failed = 1;
			failed_mirror = eb->read_mirror;
		}

448
		mirror_num++;
449 450 451
		if (mirror_num == failed_mirror)
			mirror_num++;

C
Chris Mason 已提交
452
		if (mirror_num > num_copies)
453
			break;
454
	}
455

456
	if (failed && !ret && failed_mirror)
457 458 459
		repair_eb_io_failure(root, eb, failed_mirror);

	return ret;
460
}
461

C
Chris Mason 已提交
462
/*
C
Chris Mason 已提交
463 464
 * 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
465
 */
C
Chris Mason 已提交
466

467
static int csum_dirty_buffer(struct btrfs_root *root, struct page *page)
468
{
M
Miao Xie 已提交
469
	u64 start = page_offset(page);
470 471
	u64 found_start;
	struct extent_buffer *eb;
472

J
Josef Bacik 已提交
473 474 475
	eb = (struct extent_buffer *)page->private;
	if (page != eb->pages[0])
		return 0;
476
	found_start = btrfs_header_bytenr(eb);
477
	if (WARN_ON(found_start != start || !PageUptodate(page)))
J
Josef Bacik 已提交
478
		return 0;
479 480 481 482
	csum_tree_block(root, eb, 0);
	return 0;
}

Y
Yan Zheng 已提交
483 484 485 486 487 488 489
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;

490
	read_extent_buffer(eb, fsid, btrfs_header_fsid(), BTRFS_FSID_SIZE);
Y
Yan Zheng 已提交
491 492 493 494 495 496 497 498 499 500
	while (fs_devices) {
		if (!memcmp(fsid, fs_devices->fsid, BTRFS_FSID_SIZE)) {
			ret = 0;
			break;
		}
		fs_devices = fs_devices->seed;
	}
	return ret;
}

501
#define CORRUPT(reason, eb, root, slot)				\
502 503
	btrfs_crit(root->fs_info, "corrupt leaf, %s: block=%llu,"	\
		   "root=%llu, slot=%d", reason,			\
504
	       btrfs_header_bytenr(eb),	root->objectid, slot)
505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566

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;
}

567 568 569
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)
570 571 572 573 574
{
	u64 found_start;
	int found_level;
	struct extent_buffer *eb;
	struct btrfs_root *root = BTRFS_I(page->mapping->host)->root;
575
	int ret = 0;
576
	int reads_done;
577 578 579

	if (!page->private)
		goto out;
C
Chris Mason 已提交
580

J
Josef Bacik 已提交
581
	eb = (struct extent_buffer *)page->private;
C
Chris Mason 已提交
582

583 584 585 586 587 588
	/* 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);
589 590
	if (!reads_done)
		goto err;
591

592
	eb->read_mirror = mirror;
593 594 595 596 597
	if (test_bit(EXTENT_BUFFER_IOERR, &eb->bflags)) {
		ret = -EIO;
		goto err;
	}

598
	found_start = btrfs_header_bytenr(eb);
599
	if (found_start != eb->start) {
600
		printk_ratelimited(KERN_INFO "BTRFS: bad tree block start "
C
Chris Mason 已提交
601
			       "%llu %llu\n",
602
			       found_start, eb->start);
603
		ret = -EIO;
604 605
		goto err;
	}
Y
Yan Zheng 已提交
606
	if (check_tree_block_fsid(root, eb)) {
607
		printk_ratelimited(KERN_INFO "BTRFS: bad fsid on block %llu\n",
608
			       eb->start);
609 610 611
		ret = -EIO;
		goto err;
	}
612
	found_level = btrfs_header_level(eb);
613
	if (found_level >= BTRFS_MAX_LEVEL) {
614
		btrfs_info(root->fs_info, "bad tree block level %d",
615 616 617 618
			   (int)btrfs_header_level(eb));
		ret = -EIO;
		goto err;
	}
619

620 621
	btrfs_set_buffer_lockdep_class(btrfs_header_owner(eb),
				       eb, found_level);
622

623
	ret = csum_tree_block(root, eb, 1);
624
	if (ret) {
625
		ret = -EIO;
626 627 628 629 630 631 632 633 634 635 636 637
		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;
	}
638

639 640
	if (!ret)
		set_extent_buffer_uptodate(eb);
641
err:
642 643
	if (reads_done &&
	    test_and_clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags))
A
Arne Jansen 已提交
644 645
		btree_readahead_hook(root, eb, eb->start, ret);

D
David Woodhouse 已提交
646 647 648 649 650 651 652
	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);
653
		clear_extent_buffer_uptodate(eb);
D
David Woodhouse 已提交
654
	}
655
	free_extent_buffer(eb);
656
out:
657
	return ret;
658 659
}

660
static int btree_io_failed_hook(struct page *page, int failed_mirror)
A
Arne Jansen 已提交
661 662 663 664
{
	struct extent_buffer *eb;
	struct btrfs_root *root = BTRFS_I(page->mapping->host)->root;

J
Josef Bacik 已提交
665
	eb = (struct extent_buffer *)page->private;
666
	set_bit(EXTENT_BUFFER_IOERR, &eb->bflags);
667
	eb->read_mirror = failed_mirror;
D
David Woodhouse 已提交
668
	atomic_dec(&eb->io_pages);
669
	if (test_and_clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags))
A
Arne Jansen 已提交
670 671 672 673
		btree_readahead_hook(root, eb, eb->start, -EIO);
	return -EIO;	/* we fixed nothing */
}

674 675 676 677 678 679 680
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;
681 682
	end_io_wq->work.func = end_workqueue_fn;
	end_io_wq->work.flags = 0;
683

684
	if (bio->bi_rw & REQ_WRITE) {
D
David Woodhouse 已提交
685
		if (end_io_wq->metadata == BTRFS_WQ_ENDIO_METADATA)
686 687
			btrfs_queue_worker(&fs_info->endio_meta_write_workers,
					   &end_io_wq->work);
D
David Woodhouse 已提交
688
		else if (end_io_wq->metadata == BTRFS_WQ_ENDIO_FREE_SPACE)
J
Josef Bacik 已提交
689 690
			btrfs_queue_worker(&fs_info->endio_freespace_worker,
					   &end_io_wq->work);
D
David Woodhouse 已提交
691 692 693
		else if (end_io_wq->metadata == BTRFS_WQ_ENDIO_RAID56)
			btrfs_queue_worker(&fs_info->endio_raid56_workers,
					   &end_io_wq->work);
694 695 696
		else
			btrfs_queue_worker(&fs_info->endio_write_workers,
					   &end_io_wq->work);
697
	} else {
D
David Woodhouse 已提交
698 699 700 701
		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)
702 703 704 705 706 707
			btrfs_queue_worker(&fs_info->endio_meta_workers,
					   &end_io_wq->work);
		else
			btrfs_queue_worker(&fs_info->endio_workers,
					   &end_io_wq->work);
	}
708 709
}

J
Josef Bacik 已提交
710 711 712 713 714 715
/*
 * 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
716
 * 3 - raid parity work
J
Josef Bacik 已提交
717
 */
718 719
int btrfs_bio_wq_end_io(struct btrfs_fs_info *info, struct bio *bio,
			int metadata)
720
{
721 722 723 724 725 726 727
	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;
728
	end_io_wq->info = info;
729 730
	end_io_wq->error = 0;
	end_io_wq->bio = bio;
731
	end_io_wq->metadata = metadata;
732 733 734

	bio->bi_private = end_io_wq;
	bio->bi_end_io = end_workqueue_bio;
735 736 737
	return 0;
}

738
unsigned long btrfs_async_submit_limit(struct btrfs_fs_info *info)
739
{
740 741 742 743 744
	unsigned long limit = min_t(unsigned long,
				    info->workers.max_workers,
				    info->fs_devices->open_devices);
	return 256 * limit;
}
745

C
Chris Mason 已提交
746 747 748
static void run_one_async_start(struct btrfs_work *work)
{
	struct async_submit_bio *async;
749
	int ret;
C
Chris Mason 已提交
750 751

	async = container_of(work, struct  async_submit_bio, work);
752 753 754 755 756
	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
757 758 759
}

static void run_one_async_done(struct btrfs_work *work)
760 761 762
{
	struct btrfs_fs_info *fs_info;
	struct async_submit_bio *async;
763
	int limit;
764 765 766

	async = container_of(work, struct  async_submit_bio, work);
	fs_info = BTRFS_I(async->inode)->root->fs_info;
767

768
	limit = btrfs_async_submit_limit(fs_info);
769 770
	limit = limit * 2 / 3;

771
	if (atomic_dec_return(&fs_info->nr_async_submits) < limit &&
772
	    waitqueue_active(&fs_info->async_submit_wait))
773 774
		wake_up(&fs_info->async_submit_wait);

775 776 777 778 779 780
	/* 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
781
	async->submit_bio_done(async->inode, async->rw, async->bio,
782 783
			       async->mirror_num, async->bio_flags,
			       async->bio_offset);
C
Chris Mason 已提交
784 785 786 787 788 789 790
}

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

	async = container_of(work, struct  async_submit_bio, work);
791 792 793
	kfree(async);
}

794 795
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
796
			unsigned long bio_flags,
797
			u64 bio_offset,
C
Chris Mason 已提交
798 799
			extent_submit_bio_hook_t *submit_bio_start,
			extent_submit_bio_hook_t *submit_bio_done)
800 801 802 803 804 805 806 807 808 809 810
{
	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
811 812 813 814 815 816 817
	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;

818
	async->work.flags = 0;
C
Chris Mason 已提交
819
	async->bio_flags = bio_flags;
820
	async->bio_offset = bio_offset;
821

822 823
	async->error = 0;

824
	atomic_inc(&fs_info->nr_async_submits);
825

826
	if (rw & REQ_SYNC)
827 828
		btrfs_set_work_high_prio(&async->work);

829
	btrfs_queue_worker(&fs_info->workers, &async->work);
830

C
Chris Mason 已提交
831
	while (atomic_read(&fs_info->async_submit_draining) &&
832 833 834 835 836
	      atomic_read(&fs_info->nr_async_submits)) {
		wait_event(fs_info->async_submit_wait,
			   (atomic_read(&fs_info->nr_async_submits) == 0));
	}

837 838 839
	return 0;
}

840 841 842 843 844
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;
845
	int ret = 0;
846 847

	WARN_ON(bio->bi_vcnt <= 0);
C
Chris Mason 已提交
848
	while (bio_index < bio->bi_vcnt) {
849
		root = BTRFS_I(bvec->bv_page->mapping->host)->root;
850 851 852
		ret = csum_dirty_buffer(root, bvec->bv_page);
		if (ret)
			break;
853 854 855
		bio_index++;
		bvec++;
	}
856
	return ret;
857 858
}

C
Chris Mason 已提交
859 860
static int __btree_submit_bio_start(struct inode *inode, int rw,
				    struct bio *bio, int mirror_num,
861 862
				    unsigned long bio_flags,
				    u64 bio_offset)
863
{
864 865
	/*
	 * when we're called for a write, we're already in the async
866
	 * submission context.  Just jump into btrfs_map_bio
867
	 */
868
	return btree_csum_one_bio(bio);
C
Chris Mason 已提交
869
}
870

C
Chris Mason 已提交
871
static int __btree_submit_bio_done(struct inode *inode, int rw, struct bio *bio,
872 873
				 int mirror_num, unsigned long bio_flags,
				 u64 bio_offset)
C
Chris Mason 已提交
874
{
875 876
	int ret;

877
	/*
C
Chris Mason 已提交
878 879
	 * when we're called for a write, we're already in the async
	 * submission context.  Just jump into btrfs_map_bio
880
	 */
881 882 883 884
	ret = btrfs_map_bio(BTRFS_I(inode)->root, rw, bio, mirror_num, 1);
	if (ret)
		bio_endio(bio, ret);
	return ret;
885 886
}

887 888 889 890 891 892 893 894 895 896 897
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;
}

898
static int btree_submit_bio_hook(struct inode *inode, int rw, struct bio *bio,
899 900
				 int mirror_num, unsigned long bio_flags,
				 u64 bio_offset)
901
{
902
	int async = check_async_write(inode, bio_flags);
903 904
	int ret;

905
	if (!(rw & REQ_WRITE)) {
C
Chris Mason 已提交
906 907 908 909
		/*
		 * called for a read, do the setup so that checksum validation
		 * can happen in the async kernel threads
		 */
910 911
		ret = btrfs_bio_wq_end_io(BTRFS_I(inode)->root->fs_info,
					  bio, 1);
912
		if (ret)
913 914 915
			goto out_w_error;
		ret = btrfs_map_bio(BTRFS_I(inode)->root, rw, bio,
				    mirror_num, 0);
916 917 918
	} else if (!async) {
		ret = btree_csum_one_bio(bio);
		if (ret)
919 920 921 922 923 924 925 926 927 928 929 930 931
			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);
932
	}
933

934 935 936 937 938
	if (ret) {
out_w_error:
		bio_endio(bio, ret);
	}
	return ret;
939 940
}

J
Jan Beulich 已提交
941
#ifdef CONFIG_MIGRATION
942
static int btree_migratepage(struct address_space *mapping,
943 944
			struct page *newpage, struct page *page,
			enum migrate_mode mode)
945 946 947 948 949 950 951 952 953 954 955 956 957 958
{
	/*
	 * 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;
959
	return migrate_page(mapping, newpage, page, mode);
960
}
J
Jan Beulich 已提交
961
#endif
962

963 964 965 966

static int btree_writepages(struct address_space *mapping,
			    struct writeback_control *wbc)
{
967 968 969
	struct btrfs_fs_info *fs_info;
	int ret;

970
	if (wbc->sync_mode == WB_SYNC_NONE) {
971 972 973 974

		if (wbc->for_kupdate)
			return 0;

975
		fs_info = BTRFS_I(mapping->host)->root->fs_info;
976
		/* this is a bit racy, but that's ok */
977 978 979
		ret = percpu_counter_compare(&fs_info->dirty_metadata_bytes,
					     BTRFS_DIRTY_METADATA_THRESH);
		if (ret < 0)
980 981
			return 0;
	}
982
	return btree_write_cache_pages(mapping, wbc);
983 984
}

985
static int btree_readpage(struct file *file, struct page *page)
986
{
987 988
	struct extent_io_tree *tree;
	tree = &BTRFS_I(page->mapping->host)->io_tree;
989
	return extent_read_full_page(tree, page, btree_get_extent, 0);
990
}
C
Chris Mason 已提交
991

992
static int btree_releasepage(struct page *page, gfp_t gfp_flags)
993
{
994
	if (PageWriteback(page) || PageDirty(page))
C
Chris Mason 已提交
995
		return 0;
996

997
	return try_release_extent_buffer(page);
998 999
}

1000 1001
static void btree_invalidatepage(struct page *page, unsigned int offset,
				 unsigned int length)
1002
{
1003 1004
	struct extent_io_tree *tree;
	tree = &BTRFS_I(page->mapping->host)->io_tree;
1005 1006
	extent_invalidatepage(tree, page, offset);
	btree_releasepage(page, GFP_NOFS);
1007
	if (PagePrivate(page)) {
1008 1009 1010
		btrfs_warn(BTRFS_I(page->mapping->host)->root->fs_info,
			   "page private not zero on page %llu",
			   (unsigned long long)page_offset(page));
1011 1012 1013 1014
		ClearPagePrivate(page);
		set_page_private(page, 0);
		page_cache_release(page);
	}
1015 1016
}

1017 1018
static int btree_set_page_dirty(struct page *page)
{
1019
#ifdef DEBUG
1020 1021 1022 1023 1024 1025 1026 1027
	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);
1028
#endif
1029 1030 1031
	return __set_page_dirty_nobuffers(page);
}

1032
static const struct address_space_operations btree_aops = {
1033
	.readpage	= btree_readpage,
1034
	.writepages	= btree_writepages,
1035 1036
	.releasepage	= btree_releasepage,
	.invalidatepage = btree_invalidatepage,
1037
#ifdef CONFIG_MIGRATION
1038
	.migratepage	= btree_migratepage,
1039
#endif
1040
	.set_page_dirty = btree_set_page_dirty,
1041 1042
};

1043 1044
int readahead_tree_block(struct btrfs_root *root, u64 bytenr, u32 blocksize,
			 u64 parent_transid)
C
Chris Mason 已提交
1045
{
1046 1047
	struct extent_buffer *buf = NULL;
	struct inode *btree_inode = root->fs_info->btree_inode;
1048
	int ret = 0;
C
Chris Mason 已提交
1049

1050
	buf = btrfs_find_create_tree_block(root, bytenr, blocksize);
1051
	if (!buf)
C
Chris Mason 已提交
1052
		return 0;
1053
	read_extent_buffer_pages(&BTRFS_I(btree_inode)->io_tree,
1054
				 buf, 0, WAIT_NONE, btree_get_extent, 0);
1055
	free_extent_buffer(buf);
1056
	return ret;
C
Chris Mason 已提交
1057 1058
}

1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082
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;
1083
	} else if (extent_buffer_uptodate(buf)) {
1084 1085 1086 1087 1088 1089 1090
		*eb = buf;
	} else {
		free_extent_buffer(buf);
	}
	return 0;
}

1091 1092 1093
struct extent_buffer *btrfs_find_tree_block(struct btrfs_root *root,
					    u64 bytenr, u32 blocksize)
{
1094
	return find_extent_buffer(root->fs_info, bytenr);
1095 1096 1097 1098 1099
}

struct extent_buffer *btrfs_find_create_tree_block(struct btrfs_root *root,
						 u64 bytenr, u32 blocksize)
{
1100
	return alloc_extent_buffer(root->fs_info, bytenr, blocksize);
1101 1102 1103
}


1104 1105
int btrfs_write_tree_block(struct extent_buffer *buf)
{
1106
	return filemap_fdatawrite_range(buf->pages[0]->mapping, buf->start,
1107
					buf->start + buf->len - 1);
1108 1109 1110 1111
}

int btrfs_wait_tree_block_writeback(struct extent_buffer *buf)
{
1112
	return filemap_fdatawait_range(buf->pages[0]->mapping,
1113
				       buf->start, buf->start + buf->len - 1);
1114 1115
}

1116
struct extent_buffer *read_tree_block(struct btrfs_root *root, u64 bytenr,
1117
				      u32 blocksize, u64 parent_transid)
1118 1119 1120 1121 1122 1123 1124 1125
{
	struct extent_buffer *buf = NULL;
	int ret;

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

1126
	ret = btree_read_extent_buffer_pages(root, buf, 0, parent_transid);
1127 1128 1129 1130
	if (ret) {
		free_extent_buffer(buf);
		return NULL;
	}
1131
	return buf;
1132

1133 1134
}

1135 1136
void clean_tree_block(struct btrfs_trans_handle *trans, struct btrfs_root *root,
		      struct extent_buffer *buf)
1137
{
1138 1139
	struct btrfs_fs_info *fs_info = root->fs_info;

1140
	if (btrfs_header_generation(buf) ==
1141
	    fs_info->running_transaction->transid) {
1142
		btrfs_assert_tree_locked(buf);
1143

1144
		if (test_and_clear_bit(EXTENT_BUFFER_DIRTY, &buf->bflags)) {
1145 1146 1147
			__percpu_counter_add(&fs_info->dirty_metadata_bytes,
					     -buf->len,
					     fs_info->dirty_metadata_batch);
1148 1149 1150 1151
			/* ugh, clear_extent_buffer_dirty needs to lock the page */
			btrfs_set_lock_blocking(buf);
			clear_extent_buffer_dirty(buf);
		}
1152
	}
1153 1154
}

1155 1156 1157 1158
static void __setup_root(u32 nodesize, u32 leafsize, u32 sectorsize,
			 u32 stripesize, struct btrfs_root *root,
			 struct btrfs_fs_info *fs_info,
			 u64 objectid)
1159
{
C
Chris Mason 已提交
1160
	root->node = NULL;
1161
	root->commit_root = NULL;
1162 1163 1164
	root->sectorsize = sectorsize;
	root->nodesize = nodesize;
	root->leafsize = leafsize;
1165
	root->stripesize = stripesize;
C
Chris Mason 已提交
1166
	root->ref_cows = 0;
1167
	root->track_dirty = 0;
1168
	root->in_radix = 0;
1169 1170
	root->orphan_item_inserted = 0;
	root->orphan_cleanup_state = 0;
1171

1172 1173
	root->objectid = objectid;
	root->last_trans = 0;
1174
	root->highest_objectid = 0;
1175
	root->nr_delalloc_inodes = 0;
1176
	root->nr_ordered_extents = 0;
1177
	root->name = NULL;
1178
	root->inode_tree = RB_ROOT;
1179
	INIT_RADIX_TREE(&root->delayed_nodes_tree, GFP_ATOMIC);
1180
	root->block_rsv = NULL;
1181
	root->orphan_block_rsv = NULL;
1182 1183

	INIT_LIST_HEAD(&root->dirty_list);
1184
	INIT_LIST_HEAD(&root->root_list);
1185 1186
	INIT_LIST_HEAD(&root->delalloc_inodes);
	INIT_LIST_HEAD(&root->delalloc_root);
1187 1188
	INIT_LIST_HEAD(&root->ordered_extents);
	INIT_LIST_HEAD(&root->ordered_root);
1189 1190
	INIT_LIST_HEAD(&root->logged_list[0]);
	INIT_LIST_HEAD(&root->logged_list[1]);
1191
	spin_lock_init(&root->orphan_lock);
1192
	spin_lock_init(&root->inode_lock);
1193
	spin_lock_init(&root->delalloc_lock);
1194
	spin_lock_init(&root->ordered_extent_lock);
1195
	spin_lock_init(&root->accounting_lock);
1196 1197
	spin_lock_init(&root->log_extents_lock[0]);
	spin_lock_init(&root->log_extents_lock[1]);
1198
	mutex_init(&root->objectid_mutex);
1199
	mutex_init(&root->log_mutex);
Y
Yan Zheng 已提交
1200 1201 1202 1203 1204 1205
	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
1206
	atomic_set(&root->log_batch, 0);
1207
	atomic_set(&root->orphan_inodes, 0);
1208
	atomic_set(&root->refs, 1);
Y
Yan Zheng 已提交
1209
	root->log_transid = 0;
1210
	root->last_log_commit = 0;
1211 1212 1213
	if (fs_info)
		extent_io_tree_init(&root->dirty_log_pages,
				     fs_info->btree_inode->i_mapping);
C
Chris Mason 已提交
1214

1215 1216
	memset(&root->root_key, 0, sizeof(root->root_key));
	memset(&root->root_item, 0, sizeof(root->root_item));
1217
	memset(&root->defrag_progress, 0, sizeof(root->defrag_progress));
1218
	memset(&root->root_kobj, 0, sizeof(root->root_kobj));
1219 1220 1221 1222
	if (fs_info)
		root->defrag_trans_start = fs_info->generation;
	else
		root->defrag_trans_start = 0;
1223
	init_completion(&root->kobj_unregister);
1224
	root->defrag_running = 0;
1225
	root->root_key.objectid = objectid;
1226
	root->anon_dev = 0;
1227

1228
	spin_lock_init(&root->root_item_lock);
1229 1230
}

1231
static struct btrfs_root *btrfs_alloc_root(struct btrfs_fs_info *fs_info)
A
Al Viro 已提交
1232 1233 1234 1235 1236 1237 1238
{
	struct btrfs_root *root = kzalloc(sizeof(*root), GFP_NOFS);
	if (root)
		root->fs_info = fs_info;
	return root;
}

1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254
#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
/* Should only be used by the testing infrastructure */
struct btrfs_root *btrfs_alloc_dummy_root(void)
{
	struct btrfs_root *root;

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

	return root;
}
#endif

1255 1256 1257 1258 1259 1260 1261 1262 1263
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;
1264
	uuid_le uuid;
1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280

	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);
1281
		leaf = NULL;
1282 1283 1284 1285 1286 1287 1288 1289 1290 1291
		goto fail;
	}

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

1292
	write_extent_buffer(leaf, fs_info->fsid, btrfs_header_fsid(),
1293 1294
			    BTRFS_FSID_SIZE);
	write_extent_buffer(leaf, fs_info->chunk_tree_uuid,
1295
			    btrfs_header_chunk_tree_uuid(leaf),
1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311
			    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);
1312 1313
	uuid_le_gen(&uuid);
	memcpy(root->root_item.uuid, uuid.b, BTRFS_UUID_SIZE);
1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324
	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);

1325 1326
	return root;

1327
fail:
1328 1329 1330 1331 1332
	if (leaf) {
		btrfs_tree_unlock(leaf);
		free_extent_buffer(leaf);
	}
	kfree(root);
1333

1334
	return ERR_PTR(ret);
1335 1336
}

Y
Yan Zheng 已提交
1337 1338
static struct btrfs_root *alloc_log_tree(struct btrfs_trans_handle *trans,
					 struct btrfs_fs_info *fs_info)
1339 1340 1341
{
	struct btrfs_root *root;
	struct btrfs_root *tree_root = fs_info->tree_root;
Y
Yan Zheng 已提交
1342
	struct extent_buffer *leaf;
1343

A
Al Viro 已提交
1344
	root = btrfs_alloc_root(fs_info);
1345
	if (!root)
Y
Yan Zheng 已提交
1346
		return ERR_PTR(-ENOMEM);
1347 1348 1349 1350 1351 1352 1353 1354

	__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
1355 1356 1357 1358 1359 1360
	/*
	 * 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).
	 */
1361 1362
	root->ref_cows = 0;

1363
	leaf = btrfs_alloc_free_block(trans, root, root->leafsize, 0,
A
Arne Jansen 已提交
1364
				      BTRFS_TREE_LOG_OBJECTID, NULL,
1365
				      0, 0, 0);
Y
Yan Zheng 已提交
1366 1367 1368 1369
	if (IS_ERR(leaf)) {
		kfree(root);
		return ERR_CAST(leaf);
	}
1370

1371 1372 1373 1374 1375
	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
1376
	root->node = leaf;
1377 1378

	write_extent_buffer(root->node, root->fs_info->fsid,
1379
			    btrfs_header_fsid(), BTRFS_FSID_SIZE);
1380 1381
	btrfs_mark_buffer_dirty(root->node);
	btrfs_tree_unlock(root->node);
Y
Yan Zheng 已提交
1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411
	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;
1412 1413 1414 1415 1416
	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
1417

1418
	btrfs_set_root_node(&log_root->root_item, log_root->node);
Y
Yan Zheng 已提交
1419 1420 1421 1422

	WARN_ON(root->log_root);
	root->log_root = log_root;
	root->log_transid = 0;
1423
	root->last_log_commit = 0;
1424 1425 1426
	return 0;
}

1427 1428
static struct btrfs_root *btrfs_read_tree_root(struct btrfs_root *tree_root,
					       struct btrfs_key *key)
1429 1430 1431
{
	struct btrfs_root *root;
	struct btrfs_fs_info *fs_info = tree_root->fs_info;
1432
	struct btrfs_path *path;
1433
	u64 generation;
1434
	u32 blocksize;
1435
	int ret;
1436

1437 1438
	path = btrfs_alloc_path();
	if (!path)
1439
		return ERR_PTR(-ENOMEM);
1440 1441 1442 1443 1444

	root = btrfs_alloc_root(fs_info);
	if (!root) {
		ret = -ENOMEM;
		goto alloc_fail;
1445 1446
	}

1447
	__setup_root(tree_root->nodesize, tree_root->leafsize,
1448
		     tree_root->sectorsize, tree_root->stripesize,
1449
		     root, fs_info, key->objectid);
1450

1451 1452
	ret = btrfs_find_root(tree_root, key, path,
			      &root->root_item, &root->root_key);
1453
	if (ret) {
1454 1455
		if (ret > 0)
			ret = -ENOENT;
1456
		goto find_fail;
1457
	}
1458

1459
	generation = btrfs_root_generation(&root->root_item);
1460 1461
	blocksize = btrfs_level_size(root, btrfs_root_level(&root->root_item));
	root->node = read_tree_block(root, btrfs_root_bytenr(&root->root_item),
1462
				     blocksize, generation);
1463 1464 1465 1466 1467 1468
	if (!root->node) {
		ret = -ENOMEM;
		goto find_fail;
	} else if (!btrfs_buffer_uptodate(root->node, generation, 0)) {
		ret = -EIO;
		goto read_fail;
1469
	}
1470
	root->commit_root = btrfs_root_node(root);
1471
out:
1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493
	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) {
1494
		root->ref_cows = 1;
1495 1496
		btrfs_check_and_init_root_item(&root->root_item);
	}
1497

1498 1499 1500
	return root;
}

1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527
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;
}

1528 1529
static struct btrfs_root *btrfs_lookup_fs_root(struct btrfs_fs_info *fs_info,
					       u64 root_id)
1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560
{
	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;
}

1561 1562 1563
struct btrfs_root *btrfs_get_fs_root(struct btrfs_fs_info *fs_info,
				     struct btrfs_key *location,
				     bool check_ref)
1564 1565 1566 1567
{
	struct btrfs_root *root;
	int ret;

1568 1569 1570 1571
	if (location->objectid == BTRFS_ROOT_TREE_OBJECTID)
		return fs_info->tree_root;
	if (location->objectid == BTRFS_EXTENT_TREE_OBJECTID)
		return fs_info->extent_root;
1572 1573 1574 1575
	if (location->objectid == BTRFS_CHUNK_TREE_OBJECTID)
		return fs_info->chunk_root;
	if (location->objectid == BTRFS_DEV_TREE_OBJECTID)
		return fs_info->dev_root;
1576 1577
	if (location->objectid == BTRFS_CSUM_TREE_OBJECTID)
		return fs_info->csum_root;
1578 1579 1580
	if (location->objectid == BTRFS_QUOTA_TREE_OBJECTID)
		return fs_info->quota_root ? fs_info->quota_root :
					     ERR_PTR(-ENOENT);
1581 1582 1583
	if (location->objectid == BTRFS_UUID_TREE_OBJECTID)
		return fs_info->uuid_root ? fs_info->uuid_root :
					    ERR_PTR(-ENOENT);
1584
again:
1585
	root = btrfs_lookup_fs_root(fs_info, location->objectid);
1586
	if (root) {
1587
		if (check_ref && btrfs_root_refs(&root->root_item) == 0)
1588
			return ERR_PTR(-ENOENT);
1589
		return root;
1590
	}
1591

1592
	root = btrfs_read_fs_root(fs_info->tree_root, location);
1593 1594
	if (IS_ERR(root))
		return root;
1595

1596
	if (check_ref && btrfs_root_refs(&root->root_item) == 0) {
1597
		ret = -ENOENT;
1598
		goto fail;
1599
	}
1600

1601
	ret = btrfs_init_fs_root(root);
1602 1603
	if (ret)
		goto fail;
1604

1605 1606
	ret = btrfs_find_item(fs_info->tree_root, NULL, BTRFS_ORPHAN_OBJECTID,
			location->objectid, BTRFS_ORPHAN_ITEM_KEY, NULL);
1607 1608 1609 1610 1611
	if (ret < 0)
		goto fail;
	if (ret == 0)
		root->orphan_item_inserted = 1;

1612
	ret = btrfs_insert_fs_root(fs_info, root);
1613
	if (ret) {
1614 1615 1616 1617 1618
		if (ret == -EEXIST) {
			free_fs_root(root);
			goto again;
		}
		goto fail;
1619
	}
1620
	return root;
1621 1622 1623
fail:
	free_fs_root(root);
	return ERR_PTR(ret);
1624 1625
}

C
Chris Mason 已提交
1626 1627 1628 1629 1630 1631
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
1632

1633 1634
	rcu_read_lock();
	list_for_each_entry_rcu(device, &info->fs_devices->devices, dev_list) {
1635 1636
		if (!device->bdev)
			continue;
C
Chris Mason 已提交
1637 1638 1639 1640 1641 1642
		bdi = blk_get_backing_dev_info(device->bdev);
		if (bdi && bdi_congested(bdi, bdi_bits)) {
			ret = 1;
			break;
		}
	}
1643
	rcu_read_unlock();
C
Chris Mason 已提交
1644 1645 1646
	return ret;
}

1647 1648 1649 1650
/*
 * If this fails, caller must call bdi_destroy() to get rid of the
 * bdi again.
 */
C
Chris Mason 已提交
1651 1652
static int setup_bdi(struct btrfs_fs_info *info, struct backing_dev_info *bdi)
{
1653 1654 1655
	int err;

	bdi->capabilities = BDI_CAP_MAP_COPY;
1656
	err = bdi_setup_and_register(bdi, "btrfs", BDI_CAP_MAP_COPY);
1657 1658 1659
	if (err)
		return err;

1660
	bdi->ra_pages	= default_backing_dev_info.ra_pages;
C
Chris Mason 已提交
1661 1662 1663 1664 1665
	bdi->congested_fn	= btrfs_congested_fn;
	bdi->congested_data	= info;
	return 0;
}

1666 1667 1668 1669 1670
/*
 * 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)
1671 1672
{
	struct bio *bio;
1673
	struct end_io_wq *end_io_wq;
1674 1675
	int error;

1676 1677
	end_io_wq = container_of(work, struct end_io_wq, work);
	bio = end_io_wq->bio;
1678

1679 1680 1681 1682 1683
	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);
1684 1685
}

1686 1687 1688
static int cleaner_kthread(void *arg)
{
	struct btrfs_root *root = arg;
1689
	int again;
1690 1691

	do {
1692
		again = 0;
1693

1694
		/* Make the cleaner go to sleep early. */
1695
		if (btrfs_need_cleaner_sleep(root))
1696 1697 1698 1699 1700
			goto sleep;

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

1701 1702 1703 1704
		/*
		 * Avoid the problem that we change the status of the fs
		 * during the above check and trylock.
		 */
1705
		if (btrfs_need_cleaner_sleep(root)) {
1706 1707
			mutex_unlock(&root->fs_info->cleaner_mutex);
			goto sleep;
1708
		}
1709

1710 1711 1712 1713 1714
		btrfs_run_delayed_iputs(root);
		again = btrfs_clean_one_deleted_snapshot(root);
		mutex_unlock(&root->fs_info->cleaner_mutex);

		/*
1715 1716
		 * The defragger has dealt with the R/O remount and umount,
		 * needn't do anything special here.
1717 1718 1719
		 */
		btrfs_run_defrag_inodes(root->fs_info);
sleep:
D
David Sterba 已提交
1720
		if (!try_to_freeze() && !again) {
1721
			set_current_state(TASK_INTERRUPTIBLE);
1722 1723
			if (!kthread_should_stop())
				schedule();
1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734
			__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;
1735
	u64 transid;
1736 1737
	unsigned long now;
	unsigned long delay;
1738
	bool cannot_commit;
1739 1740

	do {
1741
		cannot_commit = false;
1742
		delay = HZ * root->fs_info->commit_interval;
1743 1744
		mutex_lock(&root->fs_info->transaction_kthread_mutex);

J
Josef Bacik 已提交
1745
		spin_lock(&root->fs_info->trans_lock);
1746 1747
		cur = root->fs_info->running_transaction;
		if (!cur) {
J
Josef Bacik 已提交
1748
			spin_unlock(&root->fs_info->trans_lock);
1749 1750
			goto sleep;
		}
Y
Yan Zheng 已提交
1751

1752
		now = get_seconds();
1753
		if (cur->state < TRANS_STATE_BLOCKED &&
1754 1755
		    (now < cur->start_time ||
		     now - cur->start_time < root->fs_info->commit_interval)) {
J
Josef Bacik 已提交
1756
			spin_unlock(&root->fs_info->trans_lock);
1757 1758 1759
			delay = HZ * 5;
			goto sleep;
		}
1760
		transid = cur->transid;
J
Josef Bacik 已提交
1761
		spin_unlock(&root->fs_info->trans_lock);
1762

1763
		/* If the file system is aborted, this will always fail. */
1764
		trans = btrfs_attach_transaction(root);
1765
		if (IS_ERR(trans)) {
1766 1767
			if (PTR_ERR(trans) != -ENOENT)
				cannot_commit = true;
1768
			goto sleep;
1769
		}
1770
		if (transid == trans->transid) {
1771
			btrfs_commit_transaction(trans, root);
1772 1773 1774
		} else {
			btrfs_end_transaction(trans, root);
		}
1775 1776 1777 1778
sleep:
		wake_up_process(root->fs_info->cleaner_kthread);
		mutex_unlock(&root->fs_info->transaction_kthread_mutex);

J
Josef Bacik 已提交
1779 1780 1781
		if (unlikely(test_bit(BTRFS_FS_STATE_ERROR,
				      &root->fs_info->fs_state)))
			btrfs_cleanup_transaction(root);
1782
		if (!try_to_freeze()) {
1783
			set_current_state(TASK_INTERRUPTIBLE);
1784
			if (!kthread_should_stop() &&
1785 1786
			    (!btrfs_transaction_blocked(root->fs_info) ||
			     cannot_commit))
1787
				schedule_timeout(delay);
1788 1789 1790 1791 1792 1793
			__set_current_state(TASK_RUNNING);
		}
	} while (!kthread_should_stop());
	return 0;
}

C
Chris Mason 已提交
1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 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
/*
 * 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));

1900 1901 1902 1903 1904 1905 1906 1907
	/*
	 * 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
1908
			       btrfs_header_generation(info->fs_root->node));
1909
		btrfs_set_backup_fs_root_level(root_backup,
C
Chris Mason 已提交
1910
			       btrfs_header_level(info->fs_root->node));
1911
	}
C
Chris Mason 已提交
1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 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

	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
1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011
/* 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
2012
	btrfs_stop_workers(&fs_info->qgroup_rescan_workers);
L
Liu Bo 已提交
2013 2014
}

2015 2016 2017 2018 2019 2020 2021 2022 2023 2024
static void free_root_extent_buffers(struct btrfs_root *root)
{
	if (root) {
		free_extent_buffer(root->node);
		free_extent_buffer(root->commit_root);
		root->node = NULL;
		root->commit_root = NULL;
	}
}

C
Chris Mason 已提交
2025 2026 2027
/* helper to cleanup tree roots */
static void free_root_pointers(struct btrfs_fs_info *info, int chunk_root)
{
2028
	free_root_extent_buffers(info->tree_root);
2029

2030 2031 2032 2033 2034 2035 2036
	free_root_extent_buffers(info->dev_root);
	free_root_extent_buffers(info->extent_root);
	free_root_extent_buffers(info->csum_root);
	free_root_extent_buffers(info->quota_root);
	free_root_extent_buffers(info->uuid_root);
	if (chunk_root)
		free_root_extent_buffers(info->chunk_root);
C
Chris Mason 已提交
2037 2038
}

2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050
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) {
2051
			btrfs_drop_and_free_fs_root(fs_info, gang[0]);
2052 2053 2054
		} else {
			free_extent_buffer(gang[0]->node);
			free_extent_buffer(gang[0]->commit_root);
2055
			btrfs_put_fs_root(gang[0]);
2056 2057 2058 2059 2060 2061 2062 2063 2064 2065
		}
	}

	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++)
2066
			btrfs_drop_and_free_fs_root(fs_info, gang[i]);
2067 2068
	}
}
C
Chris Mason 已提交
2069

A
Al Viro 已提交
2070 2071 2072
int open_ctree(struct super_block *sb,
	       struct btrfs_fs_devices *fs_devices,
	       char *options)
2073
{
2074 2075 2076 2077
	u32 sectorsize;
	u32 nodesize;
	u32 leafsize;
	u32 blocksize;
2078
	u32 stripesize;
2079
	u64 generation;
2080
	u64 features;
2081
	struct btrfs_key location;
2082
	struct buffer_head *bh;
2083
	struct btrfs_super_block *disk_super;
2084
	struct btrfs_fs_info *fs_info = btrfs_sb(sb);
2085
	struct btrfs_root *tree_root;
2086 2087 2088 2089
	struct btrfs_root *extent_root;
	struct btrfs_root *csum_root;
	struct btrfs_root *chunk_root;
	struct btrfs_root *dev_root;
2090
	struct btrfs_root *quota_root;
2091
	struct btrfs_root *uuid_root;
2092
	struct btrfs_root *log_tree_root;
2093
	int ret;
2094
	int err = -EINVAL;
C
Chris Mason 已提交
2095 2096
	int num_backups_tried = 0;
	int backup_index = 0;
2097 2098
	bool create_uuid_tree;
	bool check_uuid_tree;
2099

2100
	tree_root = fs_info->tree_root = btrfs_alloc_root(fs_info);
A
Al Viro 已提交
2101
	chunk_root = fs_info->chunk_root = btrfs_alloc_root(fs_info);
2102
	if (!tree_root || !chunk_root) {
C
Chris Mason 已提交
2103 2104 2105
		err = -ENOMEM;
		goto fail;
	}
2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118

	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;
	}

2119 2120 2121 2122 2123 2124 2125 2126
	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));

2127 2128 2129 2130 2131 2132
	ret = percpu_counter_init(&fs_info->delalloc_bytes, 0);
	if (ret) {
		err = ret;
		goto fail_dirty_metadata_bytes;
	}

2133 2134 2135
	fs_info->btree_inode = new_inode(sb);
	if (!fs_info->btree_inode) {
		err = -ENOMEM;
2136
		goto fail_delalloc_bytes;
2137 2138
	}

2139
	mapping_set_gfp_mask(fs_info->btree_inode->i_mapping, GFP_NOFS);
2140

2141
	INIT_RADIX_TREE(&fs_info->fs_roots_radix, GFP_ATOMIC);
2142
	INIT_RADIX_TREE(&fs_info->buffer_radix, GFP_ATOMIC);
C
Chris Mason 已提交
2143
	INIT_LIST_HEAD(&fs_info->trans_list);
2144
	INIT_LIST_HEAD(&fs_info->dead_roots);
Y
Yan, Zheng 已提交
2145
	INIT_LIST_HEAD(&fs_info->delayed_iputs);
2146
	INIT_LIST_HEAD(&fs_info->delalloc_roots);
2147
	INIT_LIST_HEAD(&fs_info->caching_block_groups);
2148
	spin_lock_init(&fs_info->delalloc_root_lock);
J
Josef Bacik 已提交
2149
	spin_lock_init(&fs_info->trans_lock);
2150
	spin_lock_init(&fs_info->fs_roots_radix_lock);
Y
Yan, Zheng 已提交
2151
	spin_lock_init(&fs_info->delayed_iput_lock);
C
Chris Mason 已提交
2152
	spin_lock_init(&fs_info->defrag_inodes_lock);
2153
	spin_lock_init(&fs_info->free_chunk_lock);
J
Jan Schmidt 已提交
2154
	spin_lock_init(&fs_info->tree_mod_seq_lock);
2155
	spin_lock_init(&fs_info->super_lock);
2156
	spin_lock_init(&fs_info->buffer_lock);
J
Jan Schmidt 已提交
2157
	rwlock_init(&fs_info->tree_mod_log_lock);
C
Chris Mason 已提交
2158
	mutex_init(&fs_info->reloc_mutex);
2159
	seqlock_init(&fs_info->profiles_lock);
2160

2161
	init_completion(&fs_info->kobj_unregister);
2162
	INIT_LIST_HEAD(&fs_info->dirty_cowonly_roots);
2163
	INIT_LIST_HEAD(&fs_info->space_info);
J
Jan Schmidt 已提交
2164
	INIT_LIST_HEAD(&fs_info->tree_mod_seq_list);
2165
	btrfs_mapping_init(&fs_info->mapping_tree);
2166 2167 2168 2169 2170 2171 2172 2173 2174
	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);
2175
	atomic_set(&fs_info->nr_async_submits, 0);
2176
	atomic_set(&fs_info->async_delalloc_pages, 0);
2177
	atomic_set(&fs_info->async_submit_draining, 0);
2178
	atomic_set(&fs_info->nr_async_bios, 0);
C
Chris Mason 已提交
2179
	atomic_set(&fs_info->defrag_running, 0);
2180
	atomic64_set(&fs_info->tree_mod_seq, 0);
C
Chris Mason 已提交
2181
	fs_info->sb = sb;
2182
	fs_info->max_inline = 8192 * 1024;
J
Josef Bacik 已提交
2183
	fs_info->metadata_ratio = 0;
C
Chris Mason 已提交
2184
	fs_info->defrag_inodes = RB_ROOT;
2185
	fs_info->free_chunk_space = 0;
J
Jan Schmidt 已提交
2186
	fs_info->tree_mod_log = RB_ROOT;
2187
	fs_info->commit_interval = BTRFS_DEFAULT_COMMIT_INTERVAL;
2188
	fs_info->avg_delayed_ref_runtime = div64_u64(NSEC_PER_SEC, 64);
2189 2190 2191
	/* 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
2192

2193 2194
	fs_info->thread_pool_size = min_t(unsigned long,
					  num_online_cpus() + 2, 8);
2195

2196 2197
	INIT_LIST_HEAD(&fs_info->ordered_roots);
	spin_lock_init(&fs_info->ordered_root_lock);
2198 2199 2200 2201 2202 2203 2204
	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);
2205

A
Arne Jansen 已提交
2206 2207 2208 2209 2210 2211 2212
	mutex_init(&fs_info->scrub_lock);
	atomic_set(&fs_info->scrubs_running, 0);
	atomic_set(&fs_info->scrub_pause_req, 0);
	atomic_set(&fs_info->scrubs_paused, 0);
	atomic_set(&fs_info->scrub_cancel_req, 0);
	init_waitqueue_head(&fs_info->scrub_pause_wait);
	fs_info->scrub_workers_refcnt = 0;
2213 2214 2215
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
	fs_info->check_integrity_print_mask = 0;
#endif
A
Arne Jansen 已提交
2216

2217 2218
	spin_lock_init(&fs_info->balance_lock);
	mutex_init(&fs_info->balance_mutex);
2219 2220
	atomic_set(&fs_info->balance_running, 0);
	atomic_set(&fs_info->balance_pause_req, 0);
2221
	atomic_set(&fs_info->balance_cancel_req, 0);
2222
	fs_info->balance_ctl = NULL;
2223
	init_waitqueue_head(&fs_info->balance_wait_q);
A
Arne Jansen 已提交
2224

2225 2226
	sb->s_blocksize = 4096;
	sb->s_blocksize_bits = blksize_bits(4096);
J
Jens Axboe 已提交
2227
	sb->s_bdi = &fs_info->bdi;
2228

2229
	fs_info->btree_inode->i_ino = BTRFS_BTREE_INODE_OBJECTID;
M
Miklos Szeredi 已提交
2230
	set_nlink(fs_info->btree_inode, 1);
2231 2232 2233 2234 2235 2236
	/*
	 * 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;
2237
	fs_info->btree_inode->i_mapping->a_ops = &btree_aops;
C
Chris Mason 已提交
2238 2239
	fs_info->btree_inode->i_mapping->backing_dev_info = &fs_info->bdi;

2240
	RB_CLEAR_NODE(&BTRFS_I(fs_info->btree_inode)->rb_node);
2241
	extent_io_tree_init(&BTRFS_I(fs_info->btree_inode)->io_tree,
2242
			     fs_info->btree_inode->i_mapping);
2243
	BTRFS_I(fs_info->btree_inode)->io_tree.track_uptodate = 0;
2244
	extent_map_tree_init(&BTRFS_I(fs_info->btree_inode)->extent_tree);
2245 2246

	BTRFS_I(fs_info->btree_inode)->io_tree.ops = &btree_extent_io_ops;
2247

2248 2249 2250
	BTRFS_I(fs_info->btree_inode)->root = tree_root;
	memset(&BTRFS_I(fs_info->btree_inode)->location, 0,
	       sizeof(struct btrfs_key));
2251 2252
	set_bit(BTRFS_INODE_DUMMY,
		&BTRFS_I(fs_info->btree_inode)->runtime_flags);
2253
	btrfs_insert_inode_hash(fs_info->btree_inode);
2254

J
Josef Bacik 已提交
2255
	spin_lock_init(&fs_info->block_group_cache_lock);
2256
	fs_info->block_group_cache_tree = RB_ROOT;
2257
	fs_info->first_logical_byte = (u64)-1;
J
Josef Bacik 已提交
2258

2259
	extent_io_tree_init(&fs_info->freed_extents[0],
2260
			     fs_info->btree_inode->i_mapping);
2261
	extent_io_tree_init(&fs_info->freed_extents[1],
2262
			     fs_info->btree_inode->i_mapping);
2263
	fs_info->pinned_extents = &fs_info->freed_extents[0];
2264
	fs_info->do_barriers = 1;
2265

C
Chris Mason 已提交
2266

2267
	mutex_init(&fs_info->ordered_operations_mutex);
2268
	mutex_init(&fs_info->ordered_extent_flush_mutex);
2269
	mutex_init(&fs_info->tree_log_mutex);
2270
	mutex_init(&fs_info->chunk_mutex);
2271 2272
	mutex_init(&fs_info->transaction_kthread_mutex);
	mutex_init(&fs_info->cleaner_mutex);
2273
	mutex_init(&fs_info->volume_mutex);
2274
	init_rwsem(&fs_info->extent_commit_sem);
2275
	init_rwsem(&fs_info->cleanup_work_sem);
2276
	init_rwsem(&fs_info->subvol_sem);
S
Stefan Behrens 已提交
2277
	sema_init(&fs_info->uuid_tree_rescan_sem, 1);
2278 2279 2280 2281 2282
	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);
2283

2284
	spin_lock_init(&fs_info->qgroup_lock);
2285
	mutex_init(&fs_info->qgroup_ioctl_lock);
2286 2287 2288 2289 2290
	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;
2291
	fs_info->qgroup_ulist = NULL;
J
Jan Schmidt 已提交
2292
	mutex_init(&fs_info->qgroup_rescan_lock);
2293

2294 2295 2296
	btrfs_init_free_cluster(&fs_info->meta_alloc_cluster);
	btrfs_init_free_cluster(&fs_info->data_alloc_cluster);

2297
	init_waitqueue_head(&fs_info->transaction_throttle);
2298
	init_waitqueue_head(&fs_info->transaction_wait);
S
Sage Weil 已提交
2299
	init_waitqueue_head(&fs_info->transaction_blocked_wait);
2300
	init_waitqueue_head(&fs_info->async_submit_wait);
2301

D
David Woodhouse 已提交
2302 2303
	ret = btrfs_alloc_stripe_hash_table(fs_info);
	if (ret) {
2304
		err = ret;
D
David Woodhouse 已提交
2305 2306 2307
		goto fail_alloc;
	}

2308
	__setup_root(4096, 4096, 4096, 4096, tree_root,
C
Chris Mason 已提交
2309
		     fs_info, BTRFS_ROOT_TREE_OBJECTID);
2310

2311
	invalidate_bdev(fs_devices->latest_bdev);
D
David Sterba 已提交
2312 2313 2314 2315

	/*
	 * Read super block and check the signature bytes only
	 */
Y
Yan Zheng 已提交
2316
	bh = btrfs_read_dev_super(fs_devices->latest_bdev);
2317 2318
	if (!bh) {
		err = -EINVAL;
2319
		goto fail_alloc;
2320
	}
C
Chris Mason 已提交
2321

D
David Sterba 已提交
2322 2323 2324 2325 2326
	/*
	 * 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)) {
2327
		printk(KERN_ERR "BTRFS: superblock checksum mismatch\n");
D
David Sterba 已提交
2328 2329 2330 2331 2332 2333 2334 2335 2336
		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
	 */
2337 2338 2339
	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));
2340
	brelse(bh);
2341

2342
	memcpy(fs_info->fsid, fs_info->super_copy->fsid, BTRFS_FSID_SIZE);
2343

D
David Sterba 已提交
2344 2345
	ret = btrfs_check_super_valid(fs_info, sb->s_flags & MS_RDONLY);
	if (ret) {
2346
		printk(KERN_ERR "BTRFS: superblock contains fatal errors\n");
D
David Sterba 已提交
2347 2348 2349 2350
		err = -EINVAL;
		goto fail_alloc;
	}

2351
	disk_super = fs_info->super_copy;
2352
	if (!btrfs_super_root(disk_super))
2353
		goto fail_alloc;
2354

L
liubo 已提交
2355
	/* check FS state, whether FS is broken. */
2356 2357
	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
2358

C
Chris Mason 已提交
2359 2360 2361 2362 2363 2364 2365
	/*
	 * 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);

2366 2367 2368 2369 2370 2371
	/*
	 * 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
2372 2373 2374
	ret = btrfs_parse_options(tree_root, options);
	if (ret) {
		err = ret;
2375
		goto fail_alloc;
Y
Yan Zheng 已提交
2376
	}
2377

2378 2379 2380 2381 2382
	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",
2383
		       features);
2384
		err = -EINVAL;
2385
		goto fail_alloc;
2386 2387
	}

2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404
	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;
	}

2405
	features = btrfs_super_incompat_flags(disk_super);
L
Li Zefan 已提交
2406
	features |= BTRFS_FEATURE_INCOMPAT_MIXED_BACKREF;
L
Li Zefan 已提交
2407
	if (tree_root->fs_info->compress_type == BTRFS_COMPRESS_LZO)
L
Li Zefan 已提交
2408
		features |= BTRFS_FEATURE_INCOMPAT_COMPRESS_LZO;
2409

2410
	if (features & BTRFS_FEATURE_INCOMPAT_SKINNY_METADATA)
2411
		printk(KERN_ERR "BTRFS: has skinny extents\n");
2412

2413 2414 2415 2416 2417 2418
	/*
	 * 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))
2419
			printk(KERN_INFO "BTRFS: flagging fs with big metadata feature\n");
2420 2421 2422
		features |= BTRFS_FEATURE_INCOMPAT_BIG_METADATA;
	}

2423 2424 2425 2426
	nodesize = btrfs_super_nodesize(disk_super);
	leafsize = btrfs_super_leafsize(disk_super);
	sectorsize = btrfs_super_sectorsize(disk_super);
	stripesize = btrfs_super_stripesize(disk_super);
2427
	fs_info->dirty_metadata_batch = leafsize * (1 + ilog2(nr_cpu_ids));
2428
	fs_info->delalloc_batch = sectorsize * 512 * (1 + ilog2(nr_cpu_ids));
2429 2430 2431 2432 2433 2434 2435

	/*
	 * 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)) {
2436
		printk(KERN_WARNING "BTRFS: unequal leaf/node/sector sizes "
2437 2438 2439 2440 2441
				"are not allowed for mixed block groups on %s\n",
				sb->s_id);
		goto fail_alloc;
	}

2442 2443 2444 2445
	/*
	 * Needn't use the lock because there is no other task which will
	 * update the flag.
	 */
L
Li Zefan 已提交
2446
	btrfs_set_super_incompat_flags(disk_super, features);
2447

2448 2449 2450 2451 2452
	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",
2453
		       features);
2454
		err = -EINVAL;
2455
		goto fail_alloc;
2456
	}
2457 2458 2459 2460

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

2461
	btrfs_init_workers(&fs_info->workers, "worker",
2462 2463
			   fs_info->thread_pool_size,
			   &fs_info->generic_worker);
C
Chris Mason 已提交
2464

2465
	btrfs_init_workers(&fs_info->delalloc_workers, "delalloc",
2466
			   fs_info->thread_pool_size, NULL);
2467

2468
	btrfs_init_workers(&fs_info->flush_workers, "flush_delalloc",
2469
			   fs_info->thread_pool_size, NULL);
2470

2471
	btrfs_init_workers(&fs_info->submit_workers, "submit",
2472
			   min_t(u64, fs_devices->num_devices,
2473
			   fs_info->thread_pool_size), NULL);
2474

2475
	btrfs_init_workers(&fs_info->caching_workers, "cache",
2476
			   fs_info->thread_pool_size, NULL);
2477

2478 2479 2480 2481 2482
	/* 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;
2483

2484
	fs_info->workers.idle_thresh = 16;
C
Chris Mason 已提交
2485
	fs_info->workers.ordered = 1;
2486

2487 2488 2489
	fs_info->delalloc_workers.idle_thresh = 2;
	fs_info->delalloc_workers.ordered = 1;

2490 2491
	btrfs_init_workers(&fs_info->fixup_workers, "fixup", 1,
			   &fs_info->generic_worker);
2492
	btrfs_init_workers(&fs_info->endio_workers, "endio",
2493 2494
			   fs_info->thread_pool_size,
			   &fs_info->generic_worker);
2495
	btrfs_init_workers(&fs_info->endio_meta_workers, "endio-meta",
2496 2497
			   fs_info->thread_pool_size,
			   &fs_info->generic_worker);
2498
	btrfs_init_workers(&fs_info->endio_meta_write_workers,
2499 2500
			   "endio-meta-write", fs_info->thread_pool_size,
			   &fs_info->generic_worker);
D
David Woodhouse 已提交
2501 2502 2503 2504 2505 2506
	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);
2507
	btrfs_init_workers(&fs_info->endio_write_workers, "endio-write",
2508 2509
			   fs_info->thread_pool_size,
			   &fs_info->generic_worker);
J
Josef Bacik 已提交
2510 2511
	btrfs_init_workers(&fs_info->endio_freespace_worker, "freespace-write",
			   1, &fs_info->generic_worker);
2512 2513 2514
	btrfs_init_workers(&fs_info->delayed_workers, "delayed-meta",
			   fs_info->thread_pool_size,
			   &fs_info->generic_worker);
2515 2516 2517
	btrfs_init_workers(&fs_info->readahead_workers, "readahead",
			   fs_info->thread_pool_size,
			   &fs_info->generic_worker);
J
Jan Schmidt 已提交
2518 2519
	btrfs_init_workers(&fs_info->qgroup_rescan_workers, "qgroup-rescan", 1,
			   &fs_info->generic_worker);
2520 2521 2522 2523 2524 2525

	/*
	 * endios are largely parallel and should have a very
	 * low idle thresh
	 */
	fs_info->endio_workers.idle_thresh = 4;
2526
	fs_info->endio_meta_workers.idle_thresh = 4;
D
David Woodhouse 已提交
2527 2528
	fs_info->endio_raid56_workers.idle_thresh = 4;
	fs_info->rmw_workers.idle_thresh = 2;
2529

2530 2531
	fs_info->endio_write_workers.idle_thresh = 2;
	fs_info->endio_meta_write_workers.idle_thresh = 2;
2532
	fs_info->readahead_workers.idle_thresh = 2;
2533

2534 2535 2536 2537 2538 2539 2540 2541 2542 2543 2544
	/*
	 * 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
2545 2546
	ret |= btrfs_start_workers(&fs_info->rmw_workers);
	ret |= btrfs_start_workers(&fs_info->endio_raid56_workers);
2547 2548 2549 2550 2551 2552
	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);
2553
	ret |= btrfs_start_workers(&fs_info->flush_workers);
J
Jan Schmidt 已提交
2554
	ret |= btrfs_start_workers(&fs_info->qgroup_rescan_workers);
2555
	if (ret) {
2556
		err = -ENOMEM;
2557 2558
		goto fail_sb_buffer;
	}
2559

2560
	fs_info->bdi.ra_pages *= btrfs_super_num_devices(disk_super);
C
Chris Mason 已提交
2561 2562
	fs_info->bdi.ra_pages = max(fs_info->bdi.ra_pages,
				    4 * 1024 * 1024 / PAGE_CACHE_SIZE);
2563

2564 2565 2566
	tree_root->nodesize = nodesize;
	tree_root->leafsize = leafsize;
	tree_root->sectorsize = sectorsize;
2567
	tree_root->stripesize = stripesize;
2568 2569 2570

	sb->s_blocksize = sectorsize;
	sb->s_blocksize_bits = blksize_bits(sectorsize);
2571

2572
	if (btrfs_super_magic(disk_super) != BTRFS_MAGIC) {
2573
		printk(KERN_INFO "BTRFS: valid FS not found on %s\n", sb->s_id);
C
Chris Mason 已提交
2574 2575
		goto fail_sb_buffer;
	}
2576

2577
	if (sectorsize != PAGE_SIZE) {
2578
		printk(KERN_WARNING "BTRFS: Incompatible sector size(%lu) "
2579
		       "found on %s\n", (unsigned long)sectorsize, sb->s_id);
2580 2581 2582
		goto fail_sb_buffer;
	}

2583
	mutex_lock(&fs_info->chunk_mutex);
Y
Yan Zheng 已提交
2584
	ret = btrfs_read_sys_array(tree_root);
2585
	mutex_unlock(&fs_info->chunk_mutex);
2586
	if (ret) {
2587
		printk(KERN_WARNING "BTRFS: failed to read the system "
C
Chris Mason 已提交
2588
		       "array on %s\n", sb->s_id);
2589
		goto fail_sb_buffer;
2590
	}
2591 2592 2593

	blocksize = btrfs_level_size(tree_root,
				     btrfs_super_chunk_root_level(disk_super));
2594
	generation = btrfs_super_chunk_root_generation(disk_super);
2595 2596 2597 2598 2599 2600

	__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),
2601
					   blocksize, generation);
2602 2603
	if (!chunk_root->node ||
	    !test_bit(EXTENT_BUFFER_UPTODATE, &chunk_root->node->bflags)) {
2604
		printk(KERN_WARNING "BTRFS: failed to read chunk root on %s\n",
2605
		       sb->s_id);
C
Chris Mason 已提交
2606
		goto fail_tree_roots;
2607
	}
2608 2609
	btrfs_set_root_node(&chunk_root->root_item, chunk_root->node);
	chunk_root->commit_root = btrfs_root_node(chunk_root);
2610

2611
	read_extent_buffer(chunk_root->node, fs_info->chunk_tree_uuid,
2612
	   btrfs_header_chunk_tree_uuid(chunk_root->node), BTRFS_UUID_SIZE);
2613

2614
	ret = btrfs_read_chunk_tree(chunk_root);
Y
Yan Zheng 已提交
2615
	if (ret) {
2616
		printk(KERN_WARNING "BTRFS: failed to read chunk tree on %s\n",
C
Chris Mason 已提交
2617
		       sb->s_id);
C
Chris Mason 已提交
2618
		goto fail_tree_roots;
Y
Yan Zheng 已提交
2619
	}
2620

2621 2622 2623 2624 2625
	/*
	 * 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);
2626

2627
	if (!fs_devices->latest_bdev) {
2628
		printk(KERN_CRIT "BTRFS: failed to read devices on %s\n",
2629 2630 2631 2632
		       sb->s_id);
		goto fail_tree_roots;
	}

C
Chris Mason 已提交
2633
retry_root_backup:
2634 2635
	blocksize = btrfs_level_size(tree_root,
				     btrfs_super_root_level(disk_super));
2636
	generation = btrfs_super_generation(disk_super);
2637

C
Chris Mason 已提交
2638
	tree_root->node = read_tree_block(tree_root,
2639
					  btrfs_super_root(disk_super),
2640
					  blocksize, generation);
C
Chris Mason 已提交
2641 2642
	if (!tree_root->node ||
	    !test_bit(EXTENT_BUFFER_UPTODATE, &tree_root->node->bflags)) {
2643
		printk(KERN_WARNING "BTRFS: failed to read tree root on %s\n",
2644
		       sb->s_id);
C
Chris Mason 已提交
2645 2646

		goto recovery_tree_root;
2647
	}
C
Chris Mason 已提交
2648

2649 2650
	btrfs_set_root_node(&tree_root->root_item, tree_root->node);
	tree_root->commit_root = btrfs_root_node(tree_root);
2651
	btrfs_set_root_refs(&tree_root->root_item, 1);
2652

2653 2654 2655 2656 2657 2658 2659
	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
2660
		goto recovery_tree_root;
2661
	}
2662
	extent_root->track_dirty = 1;
2663
	fs_info->extent_root = extent_root;
2664

2665 2666 2667 2668
	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
2669
		goto recovery_tree_root;
2670
	}
2671
	dev_root->track_dirty = 1;
2672 2673
	fs_info->dev_root = dev_root;
	btrfs_init_devices_late(fs_info);
2674

2675 2676 2677 2678
	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
2679
		goto recovery_tree_root;
2680
	}
2681
	csum_root->track_dirty = 1;
2682
	fs_info->csum_root = csum_root;
2683

2684 2685 2686
	location.objectid = BTRFS_QUOTA_TREE_OBJECTID;
	quota_root = btrfs_read_tree_root(tree_root, &location);
	if (!IS_ERR(quota_root)) {
2687 2688 2689
		quota_root->track_dirty = 1;
		fs_info->quota_enabled = 1;
		fs_info->pending_quota_state = 1;
2690
		fs_info->quota_root = quota_root;
2691 2692
	}

2693 2694 2695 2696 2697 2698 2699
	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;
2700
		check_uuid_tree = false;
2701 2702 2703
	} else {
		uuid_root->track_dirty = 1;
		fs_info->uuid_root = uuid_root;
2704 2705 2706
		create_uuid_tree = false;
		check_uuid_tree =
		    generation != btrfs_super_uuid_tree_generation(disk_super);
2707 2708
	}

2709 2710 2711
	fs_info->generation = generation;
	fs_info->last_trans_committed = generation;

2712 2713
	ret = btrfs_recover_balance(fs_info);
	if (ret) {
2714
		printk(KERN_WARNING "BTRFS: failed to recover balance\n");
2715 2716 2717
		goto fail_block_groups;
	}

2718 2719
	ret = btrfs_init_dev_stats(fs_info);
	if (ret) {
2720
		printk(KERN_ERR "BTRFS: failed to init dev_stats: %d\n",
2721 2722 2723 2724
		       ret);
		goto fail_block_groups;
	}

2725 2726
	ret = btrfs_init_dev_replace(fs_info);
	if (ret) {
2727
		pr_err("BTRFS: failed to init dev_replace: %d\n", ret);
2728 2729 2730 2731 2732
		goto fail_block_groups;
	}

	btrfs_close_extra_devices(fs_info, fs_devices, 1);

2733 2734
	ret = btrfs_sysfs_add_one(fs_info);
	if (ret) {
2735
		pr_err("BTRFS: failed to init sysfs interface: %d\n", ret);
2736 2737 2738
		goto fail_block_groups;
	}

2739 2740
	ret = btrfs_init_space_info(fs_info);
	if (ret) {
2741
		printk(KERN_ERR "BTRFS: Failed to initial space info: %d\n", ret);
2742 2743 2744
		goto fail_block_groups;
	}

2745 2746
	ret = btrfs_read_block_groups(extent_root);
	if (ret) {
2747
		printk(KERN_ERR "BTRFS: Failed to read block groups: %d\n", ret);
2748 2749
		goto fail_block_groups;
	}
2750 2751
	fs_info->num_tolerated_disk_barrier_failures =
		btrfs_calc_num_tolerated_disk_barrier_failures(fs_info);
2752 2753 2754
	if (fs_info->fs_devices->missing_devices >
	     fs_info->num_tolerated_disk_barrier_failures &&
	    !(sb->s_flags & MS_RDONLY)) {
2755 2756
		printk(KERN_WARNING "BTRFS: "
			"too many missing devices, writeable mount is not allowed\n");
2757 2758
		goto fail_block_groups;
	}
C
Chris Mason 已提交
2759

2760 2761
	fs_info->cleaner_kthread = kthread_run(cleaner_kthread, tree_root,
					       "btrfs-cleaner");
2762
	if (IS_ERR(fs_info->cleaner_kthread))
2763
		goto fail_block_groups;
2764 2765 2766 2767

	fs_info->transaction_kthread = kthread_run(transaction_kthread,
						   tree_root,
						   "btrfs-transaction");
2768
	if (IS_ERR(fs_info->transaction_kthread))
2769
		goto fail_cleaner;
2770

C
Chris Mason 已提交
2771 2772 2773
	if (!btrfs_test_opt(tree_root, SSD) &&
	    !btrfs_test_opt(tree_root, NOSSD) &&
	    !fs_info->fs_devices->rotating) {
2774
		printk(KERN_INFO "BTRFS: detected SSD devices, enabling SSD "
C
Chris Mason 已提交
2775 2776 2777 2778
		       "mode\n");
		btrfs_set_opt(fs_info->mount_opt, SSD);
	}

2779 2780 2781 2782
	/* Set the real inode map cache flag */
	if (btrfs_test_opt(tree_root, CHANGE_INODE_CACHE))
		btrfs_set_opt(tree_root->fs_info->mount_opt, INODE_MAP_CACHE);

2783 2784 2785 2786 2787 2788 2789 2790
#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)
2791
			printk(KERN_WARNING "BTRFS: failed to initialize"
2792 2793 2794
			       " integrity check module %s\n", sb->s_id);
	}
#endif
2795 2796 2797
	ret = btrfs_read_qgroup_config(fs_info);
	if (ret)
		goto fail_trans_kthread;
2798

L
liubo 已提交
2799
	/* do not make disk changes in broken FS */
2800
	if (btrfs_super_log_root(disk_super) != 0) {
2801 2802
		u64 bytenr = btrfs_super_log_root(disk_super);

2803
		if (fs_devices->rw_devices == 0) {
2804
			printk(KERN_WARNING "BTRFS: log replay required "
C
Chris Mason 已提交
2805
			       "on RO media\n");
2806
			err = -EIO;
2807
			goto fail_qgroup;
2808
		}
2809 2810 2811
		blocksize =
		     btrfs_level_size(tree_root,
				      btrfs_super_log_root_level(disk_super));
C
Chris Mason 已提交
2812

A
Al Viro 已提交
2813
		log_tree_root = btrfs_alloc_root(fs_info);
2814 2815
		if (!log_tree_root) {
			err = -ENOMEM;
2816
			goto fail_qgroup;
2817
		}
2818 2819 2820 2821 2822

		__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,
2823 2824
						      blocksize,
						      generation + 1);
2825 2826
		if (!log_tree_root->node ||
		    !extent_buffer_uptodate(log_tree_root->node)) {
2827
			printk(KERN_ERR "BTRFS: failed to read log tree\n");
2828 2829 2830 2831
			free_extent_buffer(log_tree_root->node);
			kfree(log_tree_root);
			goto fail_trans_kthread;
		}
2832
		/* returns with log_tree_root freed on success */
2833
		ret = btrfs_recover_log_trees(log_tree_root);
2834 2835 2836 2837 2838 2839 2840
		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;
		}
2841 2842

		if (sb->s_flags & MS_RDONLY) {
2843 2844 2845
			ret = btrfs_commit_super(tree_root);
			if (ret)
				goto fail_trans_kthread;
2846
		}
2847
	}
Z
Zheng Yan 已提交
2848

2849
	ret = btrfs_find_orphan_roots(tree_root);
2850 2851
	if (ret)
		goto fail_trans_kthread;
2852

2853
	if (!(sb->s_flags & MS_RDONLY)) {
2854
		ret = btrfs_cleanup_fs_roots(fs_info);
2855 2856
		if (ret)
			goto fail_trans_kthread;
2857

2858
		ret = btrfs_recover_relocation(tree_root);
2859 2860
		if (ret < 0) {
			printk(KERN_WARNING
2861
			       "BTRFS: failed to recover relocation\n");
2862
			err = -EINVAL;
2863
			goto fail_qgroup;
2864
		}
2865
	}
Z
Zheng Yan 已提交
2866

2867 2868
	location.objectid = BTRFS_FS_TREE_OBJECTID;
	location.type = BTRFS_ROOT_ITEM_KEY;
2869
	location.offset = 0;
2870 2871

	fs_info->fs_root = btrfs_read_fs_root_no_name(fs_info, &location);
2872 2873
	if (IS_ERR(fs_info->fs_root)) {
		err = PTR_ERR(fs_info->fs_root);
2874
		goto fail_qgroup;
2875
	}
C
Chris Mason 已提交
2876

2877 2878
	if (sb->s_flags & MS_RDONLY)
		return 0;
I
Ilya Dryomov 已提交
2879

2880 2881 2882
	down_read(&fs_info->cleanup_work_sem);
	if ((ret = btrfs_orphan_cleanup(fs_info->fs_root)) ||
	    (ret = btrfs_orphan_cleanup(fs_info->tree_root))) {
2883
		up_read(&fs_info->cleanup_work_sem);
2884 2885 2886 2887
		close_ctree(tree_root);
		return ret;
	}
	up_read(&fs_info->cleanup_work_sem);
I
Ilya Dryomov 已提交
2888

2889 2890
	ret = btrfs_resume_balance_async(fs_info);
	if (ret) {
2891
		printk(KERN_WARNING "BTRFS: failed to resume balance\n");
2892 2893
		close_ctree(tree_root);
		return ret;
2894 2895
	}

2896 2897
	ret = btrfs_resume_dev_replace_async(fs_info);
	if (ret) {
2898
		pr_warn("BTRFS: failed to resume dev_replace\n");
2899 2900 2901 2902
		close_ctree(tree_root);
		return ret;
	}

2903 2904
	btrfs_qgroup_rescan_resume(fs_info);

2905
	if (create_uuid_tree) {
2906
		pr_info("BTRFS: creating UUID tree\n");
2907 2908
		ret = btrfs_create_uuid_tree(fs_info);
		if (ret) {
2909
			pr_warn("BTRFS: failed to create the UUID tree %d\n",
2910 2911 2912 2913
				ret);
			close_ctree(tree_root);
			return ret;
		}
2914 2915
	} else if (check_uuid_tree ||
		   btrfs_test_opt(tree_root, RESCAN_UUID_TREE)) {
2916
		pr_info("BTRFS: checking UUID tree\n");
2917 2918
		ret = btrfs_check_uuid_tree(fs_info);
		if (ret) {
2919
			pr_warn("BTRFS: failed to check the UUID tree %d\n",
2920 2921 2922 2923 2924 2925
				ret);
			close_ctree(tree_root);
			return ret;
		}
	} else {
		fs_info->update_uuid_tree_gen = 1;
2926 2927
	}

A
Al Viro 已提交
2928
	return 0;
C
Chris Mason 已提交
2929

2930 2931
fail_qgroup:
	btrfs_free_qgroup_config(fs_info);
2932 2933
fail_trans_kthread:
	kthread_stop(fs_info->transaction_kthread);
J
Josef Bacik 已提交
2934
	btrfs_cleanup_transaction(fs_info->tree_root);
2935
	del_fs_roots(fs_info);
2936
fail_cleaner:
2937
	kthread_stop(fs_info->cleaner_kthread);
2938 2939 2940 2941 2942 2943 2944

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

2945
fail_block_groups:
J
Josef Bacik 已提交
2946
	btrfs_put_block_group_cache(fs_info);
2947
	btrfs_free_block_groups(fs_info);
C
Chris Mason 已提交
2948 2949 2950

fail_tree_roots:
	free_root_pointers(fs_info, 1);
2951
	invalidate_inode_pages2(fs_info->btree_inode->i_mapping);
C
Chris Mason 已提交
2952

C
Chris Mason 已提交
2953
fail_sb_buffer:
L
Liu Bo 已提交
2954
	btrfs_stop_all_workers(fs_info);
2955
fail_alloc:
2956
fail_iput:
2957 2958
	btrfs_mapping_tree_free(&fs_info->mapping_tree);

2959
	iput(fs_info->btree_inode);
2960 2961
fail_delalloc_bytes:
	percpu_counter_destroy(&fs_info->delalloc_bytes);
2962 2963
fail_dirty_metadata_bytes:
	percpu_counter_destroy(&fs_info->dirty_metadata_bytes);
2964
fail_bdi:
2965
	bdi_destroy(&fs_info->bdi);
2966 2967
fail_srcu:
	cleanup_srcu_struct(&fs_info->subvol_srcu);
2968
fail:
D
David Woodhouse 已提交
2969
	btrfs_free_stripe_hash_table(fs_info);
2970
	btrfs_close_devices(fs_info->fs_devices);
A
Al Viro 已提交
2971
	return err;
C
Chris Mason 已提交
2972 2973 2974 2975 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989

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;
2990 2991
}

2992 2993 2994 2995 2996
static void btrfs_end_buffer_write_sync(struct buffer_head *bh, int uptodate)
{
	if (uptodate) {
		set_buffer_uptodate(bh);
	} else {
2997 2998 2999
		struct btrfs_device *device = (struct btrfs_device *)
			bh->b_private;

3000
		printk_ratelimited_in_rcu(KERN_WARNING "BTRFS: lost page write due to "
3001 3002
					  "I/O error on %s\n",
					  rcu_str_deref(device->name));
3003 3004 3005
		/* note, we dont' set_buffer_write_io_error because we have
		 * our own ways of dealing with the IO errors
		 */
3006
		clear_buffer_uptodate(bh);
3007
		btrfs_dev_stat_inc_and_print(device, BTRFS_DEV_STAT_WRITE_ERRS);
3008 3009 3010 3011 3012
	}
	unlock_buffer(bh);
	put_bh(bh);
}

Y
Yan Zheng 已提交
3013 3014 3015 3016 3017 3018 3019 3020 3021 3022 3023 3024 3025 3026 3027 3028
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);
3029 3030
		if (bytenr + BTRFS_SUPER_INFO_SIZE >=
					i_size_read(bdev->bd_inode))
Y
Yan Zheng 已提交
3031
			break;
3032 3033
		bh = __bread(bdev, bytenr / 4096,
					BTRFS_SUPER_INFO_SIZE);
Y
Yan Zheng 已提交
3034 3035 3036 3037 3038
		if (!bh)
			continue;

		super = (struct btrfs_super_block *)bh->b_data;
		if (btrfs_super_bytenr(super) != bytenr ||
3039
		    btrfs_super_magic(super) != BTRFS_MAGIC) {
Y
Yan Zheng 已提交
3040 3041 3042 3043 3044 3045 3046 3047 3048 3049 3050 3051 3052 3053 3054
			brelse(bh);
			continue;
		}

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

3055 3056 3057 3058 3059 3060 3061 3062 3063 3064 3065
/*
 * 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
3066 3067 3068 3069 3070 3071 3072 3073 3074 3075 3076 3077 3078 3079 3080 3081 3082 3083 3084 3085 3086 3087
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);
3088 3089 3090 3091
			if (!bh) {
				errors++;
				continue;
			}
Y
Yan Zheng 已提交
3092
			wait_on_buffer(bh);
3093 3094 3095 3096 3097 3098 3099 3100 3101
			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
3102 3103 3104 3105
		} else {
			btrfs_set_super_bytenr(sb, bytenr);

			crc = ~(u32)0;
3106
			crc = btrfs_csum_data((char *)sb +
Y
Yan Zheng 已提交
3107 3108 3109 3110 3111
					      BTRFS_CSUM_SIZE, crc,
					      BTRFS_SUPER_INFO_SIZE -
					      BTRFS_CSUM_SIZE);
			btrfs_csum_final(crc, sb->csum);

3112 3113 3114 3115
			/*
			 * one reference for us, and we leave it for the
			 * caller
			 */
Y
Yan Zheng 已提交
3116 3117
			bh = __getblk(device->bdev, bytenr / 4096,
				      BTRFS_SUPER_INFO_SIZE);
3118
			if (!bh) {
3119
				printk(KERN_ERR "BTRFS: couldn't get super "
3120 3121 3122 3123 3124
				       "buffer head for bytenr %Lu\n", bytenr);
				errors++;
				continue;
			}

Y
Yan Zheng 已提交
3125 3126
			memcpy(bh->b_data, sb, BTRFS_SUPER_INFO_SIZE);

3127
			/* one reference for submit_bh */
Y
Yan Zheng 已提交
3128
			get_bh(bh);
3129 3130

			set_buffer_uptodate(bh);
Y
Yan Zheng 已提交
3131 3132
			lock_buffer(bh);
			bh->b_end_io = btrfs_end_buffer_write_sync;
3133
			bh->b_private = device;
Y
Yan Zheng 已提交
3134 3135
		}

C
Chris Mason 已提交
3136 3137 3138 3139
		/*
		 * we fua the first super.  The others we allow
		 * to go down lazy.
		 */
3140 3141 3142 3143
		if (i == 0)
			ret = btrfsic_submit_bh(WRITE_FUA, bh);
		else
			ret = btrfsic_submit_bh(WRITE_SYNC, bh);
3144
		if (ret)
Y
Yan Zheng 已提交
3145 3146 3147 3148 3149
			errors++;
	}
	return errors < i ? 0 : -1;
}

C
Chris Mason 已提交
3150 3151 3152 3153 3154 3155 3156 3157 3158 3159 3160 3161 3162 3163 3164 3165 3166 3167 3168 3169 3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180 3181 3182 3183 3184 3185 3186 3187 3188
/*
 * endio for the write_dev_flush, this will wake anyone waiting
 * for the barrier when it is done
 */
static void btrfs_end_empty_barrier(struct bio *bio, int err)
{
	if (err) {
		if (err == -EOPNOTSUPP)
			set_bit(BIO_EOPNOTSUPP, &bio->bi_flags);
		clear_bit(BIO_UPTODATE, &bio->bi_flags);
	}
	if (bio->bi_private)
		complete(bio->bi_private);
	bio_put(bio);
}

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

	if (device->nobarriers)
		return 0;

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

		wait_for_completion(&device->flush_wait);

		if (bio_flagged(bio, BIO_EOPNOTSUPP)) {
3189
			printk_in_rcu("BTRFS: disabling barriers on dev %s\n",
3190
				      rcu_str_deref(device->name));
C
Chris Mason 已提交
3191
			device->nobarriers = 1;
3192
		} else if (!bio_flagged(bio, BIO_UPTODATE)) {
C
Chris Mason 已提交
3193
			ret = -EIO;
3194 3195
			btrfs_dev_stat_inc_and_print(device,
				BTRFS_DEV_STAT_FLUSH_ERRS);
C
Chris Mason 已提交
3196 3197 3198 3199 3200 3201 3202 3203 3204 3205 3206 3207 3208
		}

		/* drop the reference from the wait == 0 run */
		bio_put(bio);
		device->flush_bio = NULL;

		return ret;
	}

	/*
	 * one reference for us, and we leave it for the
	 * caller
	 */
3209
	device->flush_bio = NULL;
3210
	bio = btrfs_io_bio_alloc(GFP_NOFS, 0);
C
Chris Mason 已提交
3211 3212 3213 3214 3215 3216 3217 3218 3219 3220
	if (!bio)
		return -ENOMEM;

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

	bio_get(bio);
3221
	btrfsic_submit_bio(WRITE_FLUSH, bio);
C
Chris Mason 已提交
3222 3223 3224 3225 3226 3227 3228 3229 3230 3231 3232 3233

	return 0;
}

/*
 * send an empty flush down to each device in parallel,
 * then wait for them
 */
static int barrier_all_devices(struct btrfs_fs_info *info)
{
	struct list_head *head;
	struct btrfs_device *dev;
3234 3235
	int errors_send = 0;
	int errors_wait = 0;
C
Chris Mason 已提交
3236 3237 3238 3239 3240 3241
	int ret;

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

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

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

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

3272 3273 3274 3275 3276 3277 3278 3279 3280 3281 3282 3283 3284 3285 3286 3287 3288 3289 3290 3291 3292 3293 3294 3295 3296 3297 3298 3299 3300 3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326 3327 3328 3329 3330 3331
int btrfs_calc_num_tolerated_disk_barrier_failures(
	struct btrfs_fs_info *fs_info)
{
	struct btrfs_ioctl_space_info space;
	struct btrfs_space_info *sinfo;
	u64 types[] = {BTRFS_BLOCK_GROUP_DATA,
		       BTRFS_BLOCK_GROUP_SYSTEM,
		       BTRFS_BLOCK_GROUP_METADATA,
		       BTRFS_BLOCK_GROUP_DATA | BTRFS_BLOCK_GROUP_METADATA};
	int num_types = 4;
	int i;
	int c;
	int num_tolerated_disk_barrier_failures =
		(int)fs_info->fs_devices->num_devices;

	for (i = 0; i < num_types; i++) {
		struct btrfs_space_info *tmp;

		sinfo = NULL;
		rcu_read_lock();
		list_for_each_entry_rcu(tmp, &fs_info->space_info, list) {
			if (tmp->flags == types[i]) {
				sinfo = tmp;
				break;
			}
		}
		rcu_read_unlock();

		if (!sinfo)
			continue;

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

				btrfs_get_block_group_info(
					&sinfo->block_groups[c], &space);
				if (space.total_bytes == 0 ||
				    space.used_bytes == 0)
					continue;
				flags = space.flags;
				/*
				 * return
				 * 0: if dup, single or RAID0 is configured for
				 *    any of metadata, system or data, else
				 * 1: if RAID5 is configured, or if RAID1 or
				 *    RAID10 is configured and only two mirrors
				 *    are used, else
				 * 2: if RAID6 is configured, else
				 * num_mirrors - 1: if RAID1 or RAID10 is
				 *                  configured and more than
				 *                  2 mirrors are used.
				 */
				if (num_tolerated_disk_barrier_failures > 0 &&
				    ((flags & (BTRFS_BLOCK_GROUP_DUP |
					       BTRFS_BLOCK_GROUP_RAID0)) ||
				     ((flags & BTRFS_BLOCK_GROUP_PROFILE_MASK)
				      == 0)))
					num_tolerated_disk_barrier_failures = 0;
D
David Woodhouse 已提交
3332 3333 3334 3335 3336 3337
				else if (num_tolerated_disk_barrier_failures > 1) {
					if (flags & (BTRFS_BLOCK_GROUP_RAID1 |
					    BTRFS_BLOCK_GROUP_RAID5 |
					    BTRFS_BLOCK_GROUP_RAID10)) {
						num_tolerated_disk_barrier_failures = 1;
					} else if (flags &
3338
						   BTRFS_BLOCK_GROUP_RAID6) {
D
David Woodhouse 已提交
3339 3340 3341
						num_tolerated_disk_barrier_failures = 2;
					}
				}
3342 3343 3344 3345 3346 3347 3348 3349
			}
		}
		up_read(&sinfo->groups_sem);
	}

	return num_tolerated_disk_barrier_failures;
}

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

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

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

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

3372 3373 3374 3375 3376 3377 3378 3379 3380 3381
	if (do_barriers) {
		ret = barrier_all_devices(root->fs_info);
		if (ret) {
			mutex_unlock(
				&root->fs_info->fs_devices->device_list_mutex);
			btrfs_error(root->fs_info, ret,
				    "errors while submitting device barriers.");
			return ret;
		}
	}
C
Chris Mason 已提交
3382

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

Y
Yan Zheng 已提交
3391
		btrfs_set_stack_device_generation(dev_item, 0);
3392 3393 3394 3395 3396 3397 3398 3399
		btrfs_set_stack_device_type(dev_item, dev->type);
		btrfs_set_stack_device_id(dev_item, dev->devid);
		btrfs_set_stack_device_total_bytes(dev_item, dev->total_bytes);
		btrfs_set_stack_device_bytes_used(dev_item, dev->bytes_used);
		btrfs_set_stack_device_io_align(dev_item, dev->io_align);
		btrfs_set_stack_device_io_width(dev_item, dev->io_width);
		btrfs_set_stack_device_sector_size(dev_item, dev->sector_size);
		memcpy(dev_item->uuid, dev->uuid, BTRFS_UUID_SIZE);
Y
Yan Zheng 已提交
3400
		memcpy(dev_item->fsid, dev->fs_devices->fsid, BTRFS_UUID_SIZE);
Y
Yan Zheng 已提交
3401

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

Y
Yan Zheng 已提交
3521
	mutex_lock(&root->fs_info->cleaner_mutex);
Y
Yan, Zheng 已提交
3522
	btrfs_run_delayed_iputs(root);
Y
Yan Zheng 已提交
3523
	mutex_unlock(&root->fs_info->cleaner_mutex);
D
David Sterba 已提交
3524
	wake_up_process(root->fs_info->cleaner_kthread);
3525 3526 3527 3528 3529

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

3530
	trans = btrfs_join_transaction(root);
3531 3532
	if (IS_ERR(trans))
		return PTR_ERR(trans);
3533
	return btrfs_commit_transaction(trans, root);
Y
Yan Zheng 已提交
3534 3535 3536 3537 3538 3539 3540 3541 3542 3543
}

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
3544 3545 3546 3547 3548
	/* 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);

3549
	/* pause restriper - we want to resume on mount */
3550
	btrfs_pause_balance(fs_info);
3551

3552 3553
	btrfs_dev_replace_suspend_for_unmount(fs_info);

3554
	btrfs_scrub_cancel(fs_info);
C
Chris Mason 已提交
3555 3556 3557 3558 3559 3560

	/* 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 */
3561
	btrfs_cleanup_defrag_inodes(fs_info);
C
Chris Mason 已提交
3562

Y
Yan Zheng 已提交
3563
	if (!(fs_info->sb->s_flags & MS_RDONLY)) {
L
liubo 已提交
3564 3565
		ret = btrfs_commit_super(root);
		if (ret)
3566
			btrfs_err(root->fs_info, "commit super ret %d", ret);
L
liubo 已提交
3567 3568
	}

3569
	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state))
3570
		btrfs_error_commit_super(root);
3571

3572 3573
	btrfs_put_block_group_cache(fs_info);

A
Al Viro 已提交
3574 3575
	kthread_stop(fs_info->transaction_kthread);
	kthread_stop(fs_info->cleaner_kthread);
3576

3577 3578 3579
	fs_info->closing = 2;
	smp_mb();

3580 3581
	btrfs_free_qgroup_config(root->fs_info);

3582
	if (percpu_counter_sum(&fs_info->delalloc_bytes)) {
3583
		btrfs_info(root->fs_info, "at unmount delalloc count %lld",
3584
		       percpu_counter_sum(&fs_info->delalloc_bytes));
C
Chris Mason 已提交
3585
	}
3586

3587 3588
	btrfs_sysfs_remove_one(fs_info);

Y
Yan Zheng 已提交
3589
	del_fs_roots(fs_info);
3590

3591 3592
	btrfs_free_block_groups(fs_info);

3593 3594
	btrfs_stop_all_workers(fs_info);

3595
	free_root_pointers(fs_info, 1);
3596

3597
	iput(fs_info->btree_inode);
3598

3599 3600 3601 3602 3603
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
	if (btrfs_test_opt(root, CHECK_INTEGRITY))
		btrfsic_unmount(root, fs_info->fs_devices);
#endif

3604
	btrfs_close_devices(fs_info->fs_devices);
3605
	btrfs_mapping_tree_free(&fs_info->mapping_tree);
3606

3607
	percpu_counter_destroy(&fs_info->dirty_metadata_bytes);
3608
	percpu_counter_destroy(&fs_info->delalloc_bytes);
C
Chris Mason 已提交
3609
	bdi_destroy(&fs_info->bdi);
3610
	cleanup_srcu_struct(&fs_info->subvol_srcu);
3611

D
David Woodhouse 已提交
3612 3613
	btrfs_free_stripe_hash_table(fs_info);

3614 3615 3616
	btrfs_free_block_rsv(root, root->orphan_block_rsv);
	root->orphan_block_rsv = NULL;

3617 3618 3619
	return 0;
}

3620 3621
int btrfs_buffer_uptodate(struct extent_buffer *buf, u64 parent_transid,
			  int atomic)
3622
{
3623
	int ret;
3624
	struct inode *btree_inode = buf->pages[0]->mapping->host;
3625

3626
	ret = extent_buffer_uptodate(buf);
3627 3628 3629 3630
	if (!ret)
		return ret;

	ret = verify_parent_transid(&BTRFS_I(btree_inode)->io_tree, buf,
3631 3632 3633
				    parent_transid, atomic);
	if (ret == -EAGAIN)
		return ret;
3634
	return !ret;
3635 3636 3637
}

int btrfs_set_buffer_uptodate(struct extent_buffer *buf)
C
Chris Mason 已提交
3638
{
3639
	return set_extent_buffer_uptodate(buf);
3640
}
3641

3642 3643
void btrfs_mark_buffer_dirty(struct extent_buffer *buf)
{
3644
	struct btrfs_root *root;
3645
	u64 transid = btrfs_header_generation(buf);
3646
	int was_dirty;
3647

3648 3649 3650 3651 3652 3653 3654 3655 3656 3657
#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
	/*
	 * This is a fast path so only do this check if we have sanity tests
	 * enabled.  Normal people shouldn't be marking dummy buffers as dirty
	 * outside of the sanity tests.
	 */
	if (unlikely(test_bit(EXTENT_BUFFER_DUMMY, &buf->bflags)))
		return;
#endif
	root = BTRFS_I(buf->pages[0]->mapping->host)->root;
3658
	btrfs_assert_tree_locked(buf);
J
Julia Lawall 已提交
3659 3660
	if (transid != root->fs_info->generation)
		WARN(1, KERN_CRIT "btrfs transid mismatch buffer %llu, "
C
Chris Mason 已提交
3661
		       "found %llu running %llu\n",
3662
			buf->start, transid, root->fs_info->generation);
3663
	was_dirty = set_extent_buffer_dirty(buf);
3664 3665 3666 3667
	if (!was_dirty)
		__percpu_counter_add(&root->fs_info->dirty_metadata_bytes,
				     buf->len,
				     root->fs_info->dirty_metadata_batch);
3668 3669
}

3670 3671
static void __btrfs_btree_balance_dirty(struct btrfs_root *root,
					int flush_delayed)
3672 3673 3674 3675 3676
{
	/*
	 * looks as though older kernels can get into trouble with
	 * this code, they end up stuck in balance_dirty_pages forever
	 */
3677
	int ret;
3678 3679 3680 3681

	if (current->flags & PF_MEMALLOC)
		return;

3682 3683
	if (flush_delayed)
		btrfs_balance_delayed_items(root);
3684

3685 3686 3687
	ret = percpu_counter_compare(&root->fs_info->dirty_metadata_bytes,
				     BTRFS_DIRTY_METADATA_THRESH);
	if (ret > 0) {
3688 3689
		balance_dirty_pages_ratelimited(
				   root->fs_info->btree_inode->i_mapping);
3690 3691 3692 3693
	}
	return;
}

3694
void btrfs_btree_balance_dirty(struct btrfs_root *root)
C
Chris Mason 已提交
3695
{
3696 3697
	__btrfs_btree_balance_dirty(root, 1);
}
3698

3699 3700 3701
void btrfs_btree_balance_dirty_nodelay(struct btrfs_root *root)
{
	__btrfs_btree_balance_dirty(root, 0);
C
Chris Mason 已提交
3702
}
3703

3704
int btrfs_read_buffer(struct extent_buffer *buf, u64 parent_transid)
3705
{
3706
	struct btrfs_root *root = BTRFS_I(buf->pages[0]->mapping->host)->root;
3707
	return btree_read_extent_buffer_pages(root, buf, 0, parent_transid);
3708
}
3709

3710
static int btrfs_check_super_valid(struct btrfs_fs_info *fs_info,
L
liubo 已提交
3711 3712
			      int read_only)
{
D
David Sterba 已提交
3713 3714 3715
	/*
	 * Placeholder for checks
	 */
3716
	return 0;
L
liubo 已提交
3717 3718
}

3719
static void btrfs_error_commit_super(struct btrfs_root *root)
L
liubo 已提交
3720 3721 3722 3723 3724 3725 3726 3727 3728 3729 3730 3731
{
	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);
}

3732 3733
static void btrfs_destroy_ordered_operations(struct btrfs_transaction *t,
					     struct btrfs_root *root)
L
liubo 已提交
3734 3735 3736 3737 3738 3739 3740
{
	struct btrfs_inode *btrfs_inode;
	struct list_head splice;

	INIT_LIST_HEAD(&splice);

	mutex_lock(&root->fs_info->ordered_operations_mutex);
3741
	spin_lock(&root->fs_info->ordered_root_lock);
L
liubo 已提交
3742

3743
	list_splice_init(&t->ordered_operations, &splice);
L
liubo 已提交
3744 3745 3746 3747 3748
	while (!list_empty(&splice)) {
		btrfs_inode = list_entry(splice.next, struct btrfs_inode,
					 ordered_operations);

		list_del_init(&btrfs_inode->ordered_operations);
3749
		spin_unlock(&root->fs_info->ordered_root_lock);
L
liubo 已提交
3750 3751

		btrfs_invalidate_inodes(btrfs_inode->root);
3752

3753
		spin_lock(&root->fs_info->ordered_root_lock);
L
liubo 已提交
3754 3755
	}

3756
	spin_unlock(&root->fs_info->ordered_root_lock);
L
liubo 已提交
3757 3758 3759
	mutex_unlock(&root->fs_info->ordered_operations_mutex);
}

3760
static void btrfs_destroy_ordered_extents(struct btrfs_root *root)
L
liubo 已提交
3761 3762 3763
{
	struct btrfs_ordered_extent *ordered;

3764
	spin_lock(&root->ordered_extent_lock);
3765 3766 3767 3768
	/*
	 * This will just short circuit the ordered completion stuff which will
	 * make sure the ordered extent gets properly cleaned up.
	 */
3769
	list_for_each_entry(ordered, &root->ordered_extents,
3770 3771
			    root_extent_list)
		set_bit(BTRFS_ORDERED_IOERR, &ordered->flags);
3772 3773 3774 3775 3776 3777 3778 3779 3780 3781 3782 3783 3784 3785 3786
	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);
3787 3788
		list_move_tail(&root->ordered_root,
			       &fs_info->ordered_roots);
3789 3790 3791 3792 3793 3794

		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
3795 3796
}

3797 3798
static int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans,
				      struct btrfs_root *root)
L
liubo 已提交
3799 3800 3801 3802 3803 3804 3805 3806 3807
{
	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);
3808
	if (atomic_read(&delayed_refs->num_entries) == 0) {
3809
		spin_unlock(&delayed_refs->lock);
3810
		btrfs_info(root->fs_info, "delayed_refs has NO entry");
L
liubo 已提交
3811 3812 3813
		return ret;
	}

3814 3815
	while ((node = rb_first(&delayed_refs->href_root)) != NULL) {
		struct btrfs_delayed_ref_head *head;
3816
		bool pin_bytes = false;
L
liubo 已提交
3817

3818 3819 3820 3821 3822
		head = rb_entry(node, struct btrfs_delayed_ref_head,
				href_node);
		if (!mutex_trylock(&head->mutex)) {
			atomic_inc(&head->node.refs);
			spin_unlock(&delayed_refs->lock);
L
liubo 已提交
3823

3824
			mutex_lock(&head->mutex);
3825
			mutex_unlock(&head->mutex);
3826 3827 3828 3829 3830 3831 3832 3833 3834 3835 3836 3837 3838
			btrfs_put_delayed_ref(&head->node);
			spin_lock(&delayed_refs->lock);
			continue;
		}
		spin_lock(&head->lock);
		while ((node = rb_first(&head->ref_root)) != NULL) {
			ref = rb_entry(node, struct btrfs_delayed_ref_node,
				       rb_node);
			ref->in_tree = 0;
			rb_erase(&ref->rb_node, &head->ref_root);
			atomic_dec(&delayed_refs->num_entries);
			btrfs_put_delayed_ref(ref);
			cond_resched_lock(&head->lock);
3839
		}
3840 3841 3842 3843 3844 3845 3846 3847 3848 3849 3850 3851
		if (head->must_insert_reserved)
			pin_bytes = true;
		btrfs_free_delayed_extent_op(head->extent_op);
		delayed_refs->num_heads--;
		if (head->processing == 0)
			delayed_refs->num_heads_ready--;
		atomic_dec(&delayed_refs->num_entries);
		head->node.in_tree = 0;
		rb_erase(&head->href_node, &delayed_refs->href_root);
		spin_unlock(&head->lock);
		spin_unlock(&delayed_refs->lock);
		mutex_unlock(&head->mutex);
L
liubo 已提交
3852

3853 3854 3855 3856
		if (pin_bytes)
			btrfs_pin_extent(root, head->node.bytenr,
					 head->node.num_bytes, 1);
		btrfs_put_delayed_ref(&head->node);
L
liubo 已提交
3857 3858 3859 3860 3861 3862 3863 3864 3865
		cond_resched();
		spin_lock(&delayed_refs->lock);
	}

	spin_unlock(&delayed_refs->lock);

	return ret;
}

3866
static void btrfs_destroy_delalloc_inodes(struct btrfs_root *root)
L
liubo 已提交
3867 3868 3869 3870 3871 3872
{
	struct btrfs_inode *btrfs_inode;
	struct list_head splice;

	INIT_LIST_HEAD(&splice);

3873 3874
	spin_lock(&root->delalloc_lock);
	list_splice_init(&root->delalloc_inodes, &splice);
L
liubo 已提交
3875 3876

	while (!list_empty(&splice)) {
3877 3878
		btrfs_inode = list_first_entry(&splice, struct btrfs_inode,
					       delalloc_inodes);
L
liubo 已提交
3879 3880

		list_del_init(&btrfs_inode->delalloc_inodes);
3881 3882
		clear_bit(BTRFS_INODE_IN_DELALLOC_LIST,
			  &btrfs_inode->runtime_flags);
3883
		spin_unlock(&root->delalloc_lock);
L
liubo 已提交
3884 3885

		btrfs_invalidate_inodes(btrfs_inode->root);
3886

3887
		spin_lock(&root->delalloc_lock);
L
liubo 已提交
3888 3889
	}

3890 3891 3892 3893 3894 3895 3896 3897 3898 3899 3900 3901 3902 3903 3904 3905 3906 3907 3908 3909 3910 3911 3912 3913 3914 3915
	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
3916 3917 3918 3919 3920 3921 3922 3923 3924 3925 3926 3927 3928
}

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,
3929
					    mark, NULL);
L
liubo 已提交
3930 3931 3932 3933 3934
		if (ret)
			break;

		clear_extent_bits(dirty_pages, start, end, mark, GFP_NOFS);
		while (start <= end) {
3935 3936
			eb = btrfs_find_tree_block(root, start,
						   root->leafsize);
3937
			start += root->leafsize;
3938
			if (!eb)
L
liubo 已提交
3939
				continue;
3940
			wait_on_extent_buffer_writeback(eb);
L
liubo 已提交
3941

3942 3943 3944 3945
			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
3946 3947 3948 3949 3950 3951 3952 3953 3954 3955 3956 3957 3958
		}
	}

	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;
3959
	bool loop = true;
L
liubo 已提交
3960 3961

	unpin = pinned_extents;
3962
again:
L
liubo 已提交
3963 3964
	while (1) {
		ret = find_first_extent_bit(unpin, 0, &start, &end,
3965
					    EXTENT_DIRTY, NULL);
L
liubo 已提交
3966 3967 3968 3969
		if (ret)
			break;

		/* opt_discard */
3970 3971 3972 3973
		if (btrfs_test_opt(root, DISCARD))
			ret = btrfs_error_discard_extent(root, start,
							 end + 1 - start,
							 NULL);
L
liubo 已提交
3974 3975 3976 3977 3978 3979

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

3980 3981 3982 3983 3984 3985 3986 3987 3988
	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
3989 3990 3991
	return 0;
}

3992 3993 3994
void btrfs_cleanup_one_transaction(struct btrfs_transaction *cur_trans,
				   struct btrfs_root *root)
{
3995 3996
	btrfs_destroy_ordered_operations(cur_trans, root);

3997 3998
	btrfs_destroy_delayed_refs(cur_trans, root);

3999
	cur_trans->state = TRANS_STATE_COMMIT_START;
4000
	wake_up(&root->fs_info->transaction_blocked_wait);
4001

4002
	cur_trans->state = TRANS_STATE_UNBLOCKED;
4003
	wake_up(&root->fs_info->transaction_wait);
4004

4005 4006
	btrfs_destroy_delayed_inodes(root);
	btrfs_assert_delayed_root_empty(root);
4007 4008 4009

	btrfs_destroy_marked_extents(root, &cur_trans->dirty_pages,
				     EXTENT_DIRTY);
4010 4011
	btrfs_destroy_pinned_extent(root,
				    root->fs_info->pinned_extents);
4012

4013 4014 4015
	cur_trans->state =TRANS_STATE_COMPLETED;
	wake_up(&cur_trans->commit_wait);

4016 4017 4018 4019 4020 4021
	/*
	memset(cur_trans, 0, sizeof(*cur_trans));
	kmem_cache_free(btrfs_transaction_cachep, cur_trans);
	*/
}

4022
static int btrfs_cleanup_transaction(struct btrfs_root *root)
L
liubo 已提交
4023 4024 4025 4026 4027
{
	struct btrfs_transaction *t;

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

J
Josef Bacik 已提交
4028
	spin_lock(&root->fs_info->trans_lock);
4029 4030 4031 4032 4033 4034 4035 4036 4037 4038 4039 4040 4041 4042 4043 4044 4045 4046 4047 4048 4049 4050 4051 4052
	while (!list_empty(&root->fs_info->trans_list)) {
		t = list_first_entry(&root->fs_info->trans_list,
				     struct btrfs_transaction, list);
		if (t->state >= TRANS_STATE_COMMIT_START) {
			atomic_inc(&t->use_count);
			spin_unlock(&root->fs_info->trans_lock);
			btrfs_wait_for_commit(root, t->transid);
			btrfs_put_transaction(t);
			spin_lock(&root->fs_info->trans_lock);
			continue;
		}
		if (t == root->fs_info->running_transaction) {
			t->state = TRANS_STATE_COMMIT_DOING;
			spin_unlock(&root->fs_info->trans_lock);
			/*
			 * We wait for 0 num_writers since we don't hold a trans
			 * handle open currently for this transaction.
			 */
			wait_event(t->writer_wait,
				   atomic_read(&t->num_writers) == 0);
		} else {
			spin_unlock(&root->fs_info->trans_lock);
		}
		btrfs_cleanup_one_transaction(t, root);
4053

4054 4055 4056
		spin_lock(&root->fs_info->trans_lock);
		if (t == root->fs_info->running_transaction)
			root->fs_info->running_transaction = NULL;
L
liubo 已提交
4057
		list_del_init(&t->list);
4058
		spin_unlock(&root->fs_info->trans_lock);
L
liubo 已提交
4059

4060 4061 4062 4063 4064 4065 4066 4067 4068 4069
		btrfs_put_transaction(t);
		trace_btrfs_transaction_commit(root);
		spin_lock(&root->fs_info->trans_lock);
	}
	spin_unlock(&root->fs_info->trans_lock);
	btrfs_destroy_all_ordered_extents(root->fs_info);
	btrfs_destroy_delayed_inodes(root);
	btrfs_assert_delayed_root_empty(root);
	btrfs_destroy_pinned_extent(root, root->fs_info->pinned_extents);
	btrfs_destroy_all_delalloc_inodes(root->fs_info);
L
liubo 已提交
4070 4071 4072 4073 4074
	mutex_unlock(&root->fs_info->transaction_kthread_mutex);

	return 0;
}

4075
static struct extent_io_ops btree_extent_io_ops = {
4076
	.readpage_end_io_hook = btree_readpage_end_io_hook,
A
Arne Jansen 已提交
4077
	.readpage_io_failed_hook = btree_io_failed_hook,
4078
	.submit_bio_hook = btree_submit_bio_hook,
4079 4080
	/* note we're sharing with inode.c for the merge bio hook */
	.merge_bio_hook = btrfs_merge_bio_hook,
4081
};