disk-io.c 97.7 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 <asm/unaligned.h>
C
Chris Mason 已提交
34
#include "compat.h"
35 36
#include "ctree.h"
#include "disk-io.h"
37
#include "transaction.h"
38
#include "btrfs_inode.h"
39
#include "volumes.h"
40
#include "print-tree.h"
41
#include "async-thread.h"
42
#include "locking.h"
43
#include "tree-log.h"
44
#include "free-space-cache.h"
45
#include "inode-map.h"
46
#include "check-integrity.h"
47

48
static struct extent_io_ops btree_extent_io_ops;
49
static void end_workqueue_fn(struct btrfs_work *work);
50
static void free_fs_root(struct btrfs_root *root);
L
liubo 已提交
51 52 53 54 55 56 57 58 59 60 61 62 63 64
static void btrfs_check_super_valid(struct btrfs_fs_info *fs_info,
				    int read_only);
static int btrfs_destroy_ordered_operations(struct btrfs_root *root);
static int btrfs_destroy_ordered_extents(struct btrfs_root *root);
static int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans,
				      struct btrfs_root *root);
static int btrfs_destroy_pending_snapshots(struct btrfs_transaction *t);
static int btrfs_destroy_delalloc_inodes(struct btrfs_root *root);
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);
static int btrfs_cleanup_transaction(struct btrfs_root *root);
65

C
Chris Mason 已提交
66 67 68 69 70
/*
 * 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.
 */
71 72 73 74 75 76
struct end_io_wq {
	struct bio *bio;
	bio_end_io_t *end_io;
	void *private;
	struct btrfs_fs_info *info;
	int error;
77
	int metadata;
78
	struct list_head list;
79
	struct btrfs_work work;
80
};
81

C
Chris Mason 已提交
82 83 84 85 86
/*
 * 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.
 */
87 88 89 90
struct async_submit_bio {
	struct inode *inode;
	struct bio *bio;
	struct list_head list;
C
Chris Mason 已提交
91 92
	extent_submit_bio_hook_t *submit_bio_start;
	extent_submit_bio_hook_t *submit_bio_done;
93 94
	int rw;
	int mirror_num;
C
Chris Mason 已提交
95
	unsigned long bio_flags;
96 97 98 99 100
	/*
	 * 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;
101
	struct btrfs_work work;
102 103
};

104 105 106 107 108 109 110 111 112 113
/*
 * 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.
114
 *
115 116 117
 * 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.
118
 *
119 120 121
 * 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.
122
 *
123 124 125
 * 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.
126 127 128 129 130
 */
#ifdef CONFIG_DEBUG_LOCK_ALLOC
# if BTRFS_MAX_LEVEL != 8
#  error
# endif
131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148

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"	},
	{ .id = BTRFS_ORPHAN_OBJECTID,		.name_stem = "orphan"	},
	{ .id = BTRFS_TREE_LOG_OBJECTID,	.name_stem = "log"	},
	{ .id = BTRFS_TREE_RELOC_OBJECTID,	.name_stem = "treloc"	},
	{ .id = BTRFS_DATA_RELOC_TREE_OBJECTID,	.name_stem = "dreloc"	},
	{ .id = 0,				.name_stem = "tree"	},
149
};
150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180

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]);
}

181 182
#endif

C
Chris Mason 已提交
183 184 185 186
/*
 * extents on the btree inode are pretty simple, there's one extent
 * that covers the entire device
 */
187
static struct extent_map *btree_get_extent(struct inode *inode,
188
		struct page *page, size_t pg_offset, u64 start, u64 len,
189
		int create)
190
{
191 192 193 194
	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
	struct extent_map *em;
	int ret;

195
	read_lock(&em_tree->lock);
196
	em = lookup_extent_mapping(em_tree, start, len);
197 198 199
	if (em) {
		em->bdev =
			BTRFS_I(inode)->root->fs_info->fs_devices->latest_bdev;
200
		read_unlock(&em_tree->lock);
201
		goto out;
202
	}
203
	read_unlock(&em_tree->lock);
204

205
	em = alloc_extent_map();
206 207 208 209 210
	if (!em) {
		em = ERR_PTR(-ENOMEM);
		goto out;
	}
	em->start = 0;
211
	em->len = (u64)-1;
C
Chris Mason 已提交
212
	em->block_len = (u64)-1;
213
	em->block_start = 0;
214
	em->bdev = BTRFS_I(inode)->root->fs_info->fs_devices->latest_bdev;
215

216
	write_lock(&em_tree->lock);
217 218
	ret = add_extent_mapping(em_tree, em);
	if (ret == -EEXIST) {
219 220 221
		u64 failed_start = em->start;
		u64 failed_len = em->len;

222
		free_extent_map(em);
223
		em = lookup_extent_mapping(em_tree, start, len);
224
		if (em) {
225
			ret = 0;
226 227 228
		} else {
			em = lookup_extent_mapping(em_tree, failed_start,
						   failed_len);
229
			ret = -EIO;
230
		}
231
	} else if (ret) {
232 233
		free_extent_map(em);
		em = NULL;
234
	}
235
	write_unlock(&em_tree->lock);
236 237 238

	if (ret)
		em = ERR_PTR(ret);
239 240
out:
	return em;
241 242
}

243 244
u32 btrfs_csum_data(struct btrfs_root *root, char *data, u32 seed, size_t len)
{
245
	return crc32c(seed, data, len);
246 247 248 249
}

void btrfs_csum_final(u32 crc, char *result)
{
250
	put_unaligned_le32(~crc, result);
251 252
}

C
Chris Mason 已提交
253 254 255 256
/*
 * compute the csum for a btree block, and either verify it or write it
 * into the csum field of the block.
 */
257 258 259
static int csum_tree_block(struct btrfs_root *root, struct extent_buffer *buf,
			   int verify)
{
260
	u16 csum_size = btrfs_super_csum_size(root->fs_info->super_copy);
261
	char *result = NULL;
262 263 264 265 266 267 268 269
	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;
270
	unsigned long inline_result;
271 272

	len = buf->len - offset;
C
Chris Mason 已提交
273
	while (len > 0) {
274
		err = map_private_extent_buffer(buf, offset, 32,
275
					&kaddr, &map_start, &map_len);
C
Chris Mason 已提交
276
		if (err)
277 278 279 280 281 282 283
			return 1;
		cur_len = min(len, map_len - (offset - map_start));
		crc = btrfs_csum_data(root, kaddr + offset - map_start,
				      crc, cur_len);
		len -= cur_len;
		offset += cur_len;
	}
284 285 286 287 288 289 290 291
	if (csum_size > sizeof(inline_result)) {
		result = kzalloc(csum_size * sizeof(char), GFP_NOFS);
		if (!result)
			return 1;
	} else {
		result = (char *)&inline_result;
	}

292 293 294
	btrfs_csum_final(crc, result);

	if (verify) {
295
		if (memcmp_extent_buffer(buf, result, 0, csum_size)) {
296 297
			u32 val;
			u32 found = 0;
298
			memcpy(&found, result, csum_size);
299

300
			read_extent_buffer(buf, &val, 0, csum_size);
301
			printk_ratelimited(KERN_INFO "btrfs: %s checksum verify "
C
Chris Mason 已提交
302 303 304 305 306
				       "failed on %llu wanted %X found %X "
				       "level %d\n",
				       root->fs_info->sb->s_id,
				       (unsigned long long)buf->start, val, found,
				       btrfs_header_level(buf));
307 308
			if (result != (char *)&inline_result)
				kfree(result);
309 310 311
			return 1;
		}
	} else {
312
		write_extent_buffer(buf, result, 0, csum_size);
313
	}
314 315
	if (result != (char *)&inline_result)
		kfree(result);
316 317 318
	return 0;
}

C
Chris Mason 已提交
319 320 321 322 323 324
/*
 * we can't consider a given block up to date unless the transid of the
 * block matches the transid in the parent node's pointer.  This is how we
 * detect blocks that either didn't get written at all or got written
 * in the wrong place.
 */
325 326 327
static int verify_parent_transid(struct extent_io_tree *io_tree,
				 struct extent_buffer *eb, u64 parent_transid)
{
328
	struct extent_state *cached_state = NULL;
329 330 331 332 333
	int ret;

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

334 335 336
	lock_extent_bits(io_tree, eb->start, eb->start + eb->len - 1,
			 0, &cached_state, GFP_NOFS);
	if (extent_buffer_uptodate(io_tree, eb, cached_state) &&
337 338 339 340
	    btrfs_header_generation(eb) == parent_transid) {
		ret = 0;
		goto out;
	}
341
	printk_ratelimited("parent transid verify failed on %llu wanted %llu "
C
Chris Mason 已提交
342 343 344 345
		       "found %llu\n",
		       (unsigned long long)eb->start,
		       (unsigned long long)parent_transid,
		       (unsigned long long)btrfs_header_generation(eb));
346
	ret = 1;
347
	clear_extent_buffer_uptodate(io_tree, eb, &cached_state);
C
Chris Mason 已提交
348
out:
349 350
	unlock_extent_cached(io_tree, eb->start, eb->start + eb->len - 1,
			     &cached_state, GFP_NOFS);
351 352 353
	return ret;
}

C
Chris Mason 已提交
354 355 356 357
/*
 * helper to read a given tree block, doing retries as required when
 * the checksums don't match and we have alternate mirrors to try.
 */
358 359
static int btree_read_extent_buffer_pages(struct btrfs_root *root,
					  struct extent_buffer *eb,
360
					  u64 start, u64 parent_transid)
361 362 363 364 365 366
{
	struct extent_io_tree *io_tree;
	int ret;
	int num_copies = 0;
	int mirror_num = 0;

367
	clear_bit(EXTENT_BUFFER_CORRUPT, &eb->bflags);
368 369
	io_tree = &BTRFS_I(root->fs_info->btree_inode)->io_tree;
	while (1) {
370 371
		ret = read_extent_buffer_pages(io_tree, eb, start,
					       WAIT_COMPLETE,
372
					       btree_get_extent, mirror_num);
373 374
		if (!ret &&
		    !verify_parent_transid(io_tree, eb, parent_transid))
375
			return ret;
C
Chris Mason 已提交
376

377 378 379 380 381 382 383 384
		/*
		 * 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))
			return ret;

385 386
		num_copies = btrfs_num_copies(&root->fs_info->mapping_tree,
					      eb->start, eb->len);
C
Chris Mason 已提交
387
		if (num_copies == 1)
388
			return ret;
C
Chris Mason 已提交
389

390
		mirror_num++;
C
Chris Mason 已提交
391
		if (mirror_num > num_copies)
392 393 394 395
			return ret;
	}
	return -EIO;
}
396

C
Chris Mason 已提交
397
/*
C
Chris Mason 已提交
398 399
 * 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
400
 */
C
Chris Mason 已提交
401

402
static int csum_dirty_buffer(struct btrfs_root *root, struct page *page)
403
{
404
	struct extent_io_tree *tree;
405
	u64 start = (u64)page->index << PAGE_CACHE_SHIFT;
406 407 408
	u64 found_start;
	unsigned long len;
	struct extent_buffer *eb;
409 410
	int ret;

411
	tree = &BTRFS_I(page->mapping->host)->io_tree;
412

C
Chris Mason 已提交
413 414
	if (page->private == EXTENT_PAGE_PRIVATE) {
		WARN_ON(1);
415
		goto out;
C
Chris Mason 已提交
416 417 418
	}
	if (!page->private) {
		WARN_ON(1);
419
		goto out;
C
Chris Mason 已提交
420
	}
421
	len = page->private >> 2;
C
Chris Mason 已提交
422 423
	WARN_ON(len == 0);

424
	eb = alloc_extent_buffer(tree, start, len, page);
T
Tsutomu Itoh 已提交
425 426 427 428
	if (eb == NULL) {
		WARN_ON(1);
		goto out;
	}
429 430
	ret = btree_read_extent_buffer_pages(root, eb, start + PAGE_CACHE_SIZE,
					     btrfs_header_generation(eb));
431
	BUG_ON(ret);
432 433
	WARN_ON(!btrfs_header_flag(eb, BTRFS_HEADER_FLAG_WRITTEN));

434 435
	found_start = btrfs_header_bytenr(eb);
	if (found_start != start) {
436 437 438 439 440 441 442 443 444 445
		WARN_ON(1);
		goto err;
	}
	if (eb->first_page != page) {
		WARN_ON(1);
		goto err;
	}
	if (!PageUptodate(page)) {
		WARN_ON(1);
		goto err;
446 447
	}
	csum_tree_block(root, eb, 0);
448
err:
449 450 451 452 453
	free_extent_buffer(eb);
out:
	return 0;
}

Y
Yan Zheng 已提交
454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472
static int check_tree_block_fsid(struct btrfs_root *root,
				 struct extent_buffer *eb)
{
	struct btrfs_fs_devices *fs_devices = root->fs_info->fs_devices;
	u8 fsid[BTRFS_UUID_SIZE];
	int ret = 1;

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

473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539
#define CORRUPT(reason, eb, root, slot)				\
	printk(KERN_CRIT "btrfs: corrupt leaf, %s: block=%llu,"	\
	       "root=%llu, slot=%d\n", reason,			\
	       (unsigned long long)btrfs_header_bytenr(eb),	\
	       (unsigned long long)root->objectid, slot)

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

	if (nritems == 0)
		return 0;

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

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

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

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

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

	return 0;
}

540
static int btree_readpage_end_io_hook(struct page *page, u64 start, u64 end,
541 542 543 544 545 546 547 548
			       struct extent_state *state)
{
	struct extent_io_tree *tree;
	u64 found_start;
	int found_level;
	unsigned long len;
	struct extent_buffer *eb;
	struct btrfs_root *root = BTRFS_I(page->mapping->host)->root;
549
	int ret = 0;
550 551 552 553 554 555

	tree = &BTRFS_I(page->mapping->host)->io_tree;
	if (page->private == EXTENT_PAGE_PRIVATE)
		goto out;
	if (!page->private)
		goto out;
C
Chris Mason 已提交
556

557
	len = page->private >> 2;
C
Chris Mason 已提交
558 559
	WARN_ON(len == 0);

560
	eb = alloc_extent_buffer(tree, start, len, page);
T
Tsutomu Itoh 已提交
561 562 563 564
	if (eb == NULL) {
		ret = -EIO;
		goto out;
	}
565

566
	found_start = btrfs_header_bytenr(eb);
567
	if (found_start != start) {
568
		printk_ratelimited(KERN_INFO "btrfs bad tree block start "
C
Chris Mason 已提交
569 570 571
			       "%llu %llu\n",
			       (unsigned long long)found_start,
			       (unsigned long long)eb->start);
572
		ret = -EIO;
573 574 575
		goto err;
	}
	if (eb->first_page != page) {
C
Chris Mason 已提交
576 577
		printk(KERN_INFO "btrfs bad first page %lu %lu\n",
		       eb->first_page->index, page->index);
578
		WARN_ON(1);
579
		ret = -EIO;
580 581
		goto err;
	}
Y
Yan Zheng 已提交
582
	if (check_tree_block_fsid(root, eb)) {
583
		printk_ratelimited(KERN_INFO "btrfs bad fsid on block %llu\n",
C
Chris Mason 已提交
584
			       (unsigned long long)eb->start);
585 586 587
		ret = -EIO;
		goto err;
	}
588 589
	found_level = btrfs_header_level(eb);

590 591
	btrfs_set_buffer_lockdep_class(btrfs_header_owner(eb),
				       eb, found_level);
592

593
	ret = csum_tree_block(root, eb, 1);
594
	if (ret) {
595
		ret = -EIO;
596 597 598 599 600 601 602 603 604 605 606 607
		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;
	}
608 609 610 611

	end = min_t(u64, eb->len, PAGE_CACHE_SIZE);
	end = eb->start + end - 1;
err:
A
Arne Jansen 已提交
612 613 614 615 616
	if (test_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags)) {
		clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags);
		btree_readahead_hook(root, eb, eb->start, ret);
	}

617 618
	free_extent_buffer(eb);
out:
619
	return ret;
620 621
}

A
Arne Jansen 已提交
622 623
static int btree_io_failed_hook(struct bio *failed_bio,
			 struct page *page, u64 start, u64 end,
624
			 int mirror_num, struct extent_state *state)
A
Arne Jansen 已提交
625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647
{
	struct extent_io_tree *tree;
	unsigned long len;
	struct extent_buffer *eb;
	struct btrfs_root *root = BTRFS_I(page->mapping->host)->root;

	tree = &BTRFS_I(page->mapping->host)->io_tree;
	if (page->private == EXTENT_PAGE_PRIVATE)
		goto out;
	if (!page->private)
		goto out;

	len = page->private >> 2;
	WARN_ON(len == 0);

	eb = alloc_extent_buffer(tree, start, len, page);
	if (eb == NULL)
		goto out;

	if (test_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags)) {
		clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags);
		btree_readahead_hook(root, eb, eb->start, -EIO);
	}
648
	free_extent_buffer(eb);
A
Arne Jansen 已提交
649 650 651 652 653

out:
	return -EIO;	/* we fixed nothing */
}

654 655 656 657 658 659 660
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;
661 662
	end_io_wq->work.func = end_workqueue_fn;
	end_io_wq->work.flags = 0;
663

664
	if (bio->bi_rw & REQ_WRITE) {
J
Josef Bacik 已提交
665
		if (end_io_wq->metadata == 1)
666 667
			btrfs_queue_worker(&fs_info->endio_meta_write_workers,
					   &end_io_wq->work);
J
Josef Bacik 已提交
668 669 670
		else if (end_io_wq->metadata == 2)
			btrfs_queue_worker(&fs_info->endio_freespace_worker,
					   &end_io_wq->work);
671 672 673
		else
			btrfs_queue_worker(&fs_info->endio_write_workers,
					   &end_io_wq->work);
674 675 676 677 678 679 680 681
	} else {
		if (end_io_wq->metadata)
			btrfs_queue_worker(&fs_info->endio_meta_workers,
					   &end_io_wq->work);
		else
			btrfs_queue_worker(&fs_info->endio_workers,
					   &end_io_wq->work);
	}
682 683
}

J
Josef Bacik 已提交
684 685 686 687 688 689 690
/*
 * For the metadata arg you want
 *
 * 0 - if data
 * 1 - if normal metadta
 * 2 - if writing to the free space cache area
 */
691 692
int btrfs_bio_wq_end_io(struct btrfs_fs_info *info, struct bio *bio,
			int metadata)
693
{
694 695 696 697 698 699 700
	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;
701
	end_io_wq->info = info;
702 703
	end_io_wq->error = 0;
	end_io_wq->bio = bio;
704
	end_io_wq->metadata = metadata;
705 706 707

	bio->bi_private = end_io_wq;
	bio->bi_end_io = end_workqueue_bio;
708 709 710
	return 0;
}

711
unsigned long btrfs_async_submit_limit(struct btrfs_fs_info *info)
712
{
713 714 715 716 717
	unsigned long limit = min_t(unsigned long,
				    info->workers.max_workers,
				    info->fs_devices->open_devices);
	return 256 * limit;
}
718

C
Chris Mason 已提交
719 720 721 722 723 724
static void run_one_async_start(struct btrfs_work *work)
{
	struct async_submit_bio *async;

	async = container_of(work, struct  async_submit_bio, work);
	async->submit_bio_start(async->inode, async->rw, async->bio,
725 726
			       async->mirror_num, async->bio_flags,
			       async->bio_offset);
C
Chris Mason 已提交
727 728 729
}

static void run_one_async_done(struct btrfs_work *work)
730 731 732
{
	struct btrfs_fs_info *fs_info;
	struct async_submit_bio *async;
733
	int limit;
734 735 736

	async = container_of(work, struct  async_submit_bio, work);
	fs_info = BTRFS_I(async->inode)->root->fs_info;
737

738
	limit = btrfs_async_submit_limit(fs_info);
739 740
	limit = limit * 2 / 3;

741
	atomic_dec(&fs_info->nr_async_submits);
742

743 744
	if (atomic_read(&fs_info->nr_async_submits) < limit &&
	    waitqueue_active(&fs_info->async_submit_wait))
745 746
		wake_up(&fs_info->async_submit_wait);

C
Chris Mason 已提交
747
	async->submit_bio_done(async->inode, async->rw, async->bio,
748 749
			       async->mirror_num, async->bio_flags,
			       async->bio_offset);
C
Chris Mason 已提交
750 751 752 753 754 755 756
}

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

	async = container_of(work, struct  async_submit_bio, work);
757 758 759
	kfree(async);
}

760 761
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
762
			unsigned long bio_flags,
763
			u64 bio_offset,
C
Chris Mason 已提交
764 765
			extent_submit_bio_hook_t *submit_bio_start,
			extent_submit_bio_hook_t *submit_bio_done)
766 767 768 769 770 771 772 773 774 775 776
{
	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
777 778 779 780 781 782 783
	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;

784
	async->work.flags = 0;
C
Chris Mason 已提交
785
	async->bio_flags = bio_flags;
786
	async->bio_offset = bio_offset;
787

788
	atomic_inc(&fs_info->nr_async_submits);
789

790
	if (rw & REQ_SYNC)
791 792
		btrfs_set_work_high_prio(&async->work);

793
	btrfs_queue_worker(&fs_info->workers, &async->work);
794

C
Chris Mason 已提交
795
	while (atomic_read(&fs_info->async_submit_draining) &&
796 797 798 799 800
	      atomic_read(&fs_info->nr_async_submits)) {
		wait_event(fs_info->async_submit_wait,
			   (atomic_read(&fs_info->nr_async_submits) == 0));
	}

801 802 803
	return 0;
}

804 805 806 807 808 809 810
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;

	WARN_ON(bio->bi_vcnt <= 0);
C
Chris Mason 已提交
811
	while (bio_index < bio->bi_vcnt) {
812 813 814 815 816 817 818 819
		root = BTRFS_I(bvec->bv_page->mapping->host)->root;
		csum_dirty_buffer(root, bvec->bv_page);
		bio_index++;
		bvec++;
	}
	return 0;
}

C
Chris Mason 已提交
820 821
static int __btree_submit_bio_start(struct inode *inode, int rw,
				    struct bio *bio, int mirror_num,
822 823
				    unsigned long bio_flags,
				    u64 bio_offset)
824
{
825 826
	/*
	 * when we're called for a write, we're already in the async
827
	 * submission context.  Just jump into btrfs_map_bio
828
	 */
C
Chris Mason 已提交
829 830 831
	btree_csum_one_bio(bio);
	return 0;
}
832

C
Chris Mason 已提交
833
static int __btree_submit_bio_done(struct inode *inode, int rw, struct bio *bio,
834 835
				 int mirror_num, unsigned long bio_flags,
				 u64 bio_offset)
C
Chris Mason 已提交
836
{
837
	/*
C
Chris Mason 已提交
838 839
	 * when we're called for a write, we're already in the async
	 * submission context.  Just jump into btrfs_map_bio
840 841
	 */
	return btrfs_map_bio(BTRFS_I(inode)->root, rw, bio, mirror_num, 1);
842 843
}

844
static int btree_submit_bio_hook(struct inode *inode, int rw, struct bio *bio,
845 846
				 int mirror_num, unsigned long bio_flags,
				 u64 bio_offset)
847
{
848 849 850 851 852 853
	int ret;

	ret = btrfs_bio_wq_end_io(BTRFS_I(inode)->root->fs_info,
					  bio, 1);
	BUG_ON(ret);

854
	if (!(rw & REQ_WRITE)) {
C
Chris Mason 已提交
855 856 857 858 859
		/*
		 * called for a read, do the setup so that checksum validation
		 * can happen in the async kernel threads
		 */
		return btrfs_map_bio(BTRFS_I(inode)->root, rw, bio,
860
				     mirror_num, 0);
861
	}
862

863 864 865 866
	/*
	 * kthread helpers are used to submit writes so that checksumming
	 * can happen in parallel across all CPUs
	 */
867
	return btrfs_wq_submit_bio(BTRFS_I(inode)->root->fs_info,
C
Chris Mason 已提交
868
				   inode, rw, bio, mirror_num, 0,
869
				   bio_offset,
C
Chris Mason 已提交
870 871
				   __btree_submit_bio_start,
				   __btree_submit_bio_done);
872 873
}

J
Jan Beulich 已提交
874
#ifdef CONFIG_MIGRATION
875
static int btree_migratepage(struct address_space *mapping,
876 877
			struct page *newpage, struct page *page,
			enum migrate_mode mode)
878 879 880 881 882 883 884 885 886 887 888 889 890 891
{
	/*
	 * 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;
892
	return migrate_page(mapping, newpage, page, mode);
893
}
J
Jan Beulich 已提交
894
#endif
895

896 897
static int btree_writepage(struct page *page, struct writeback_control *wbc)
{
898
	struct extent_io_tree *tree;
899 900 901 902
	struct btrfs_root *root = BTRFS_I(page->mapping->host)->root;
	struct extent_buffer *eb;
	int was_dirty;

903
	tree = &BTRFS_I(page->mapping->host)->io_tree;
904 905 906 907
	if (!(current->flags & PF_MEMALLOC)) {
		return extent_write_full_page(tree, page,
					      btree_get_extent, wbc);
	}
908

909
	redirty_page_for_writepage(wbc, page);
910
	eb = btrfs_find_tree_block(root, page_offset(page), PAGE_CACHE_SIZE);
911 912 913 914 915 916 917
	WARN_ON(!eb);

	was_dirty = test_and_set_bit(EXTENT_BUFFER_DIRTY, &eb->bflags);
	if (!was_dirty) {
		spin_lock(&root->fs_info->delalloc_lock);
		root->fs_info->dirty_metadata_bytes += PAGE_CACHE_SIZE;
		spin_unlock(&root->fs_info->delalloc_lock);
918
	}
919 920 921 922
	free_extent_buffer(eb);

	unlock_page(page);
	return 0;
923
}
924 925 926 927

static int btree_writepages(struct address_space *mapping,
			    struct writeback_control *wbc)
{
928 929
	struct extent_io_tree *tree;
	tree = &BTRFS_I(mapping->host)->io_tree;
930
	if (wbc->sync_mode == WB_SYNC_NONE) {
931
		struct btrfs_root *root = BTRFS_I(mapping->host)->root;
932
		u64 num_dirty;
933
		unsigned long thresh = 32 * 1024 * 1024;
934 935 936 937

		if (wbc->for_kupdate)
			return 0;

938 939
		/* this is a bit racy, but that's ok */
		num_dirty = root->fs_info->dirty_metadata_bytes;
C
Chris Mason 已提交
940
		if (num_dirty < thresh)
941 942
			return 0;
	}
943 944 945
	return extent_writepages(tree, mapping, btree_get_extent, wbc);
}

946
static int btree_readpage(struct file *file, struct page *page)
947
{
948 949
	struct extent_io_tree *tree;
	tree = &BTRFS_I(page->mapping->host)->io_tree;
950
	return extent_read_full_page(tree, page, btree_get_extent, 0);
951
}
C
Chris Mason 已提交
952

953
static int btree_releasepage(struct page *page, gfp_t gfp_flags)
954
{
955 956
	struct extent_io_tree *tree;
	struct extent_map_tree *map;
957
	int ret;
958

959
	if (PageWriteback(page) || PageDirty(page))
C
Chris Mason 已提交
960
		return 0;
961

962 963
	tree = &BTRFS_I(page->mapping->host)->io_tree;
	map = &BTRFS_I(page->mapping->host)->extent_tree;
964

965 966 967 968 969 970 971
	/*
	 * We need to mask out eg. __GFP_HIGHMEM and __GFP_DMA32 as we're doing
	 * slab allocation from alloc_extent_state down the callchain where
	 * it'd hit a BUG_ON as those flags are not allowed.
	 */
	gfp_flags &= ~GFP_SLAB_BUG_MASK;

972
	ret = try_release_extent_state(map, tree, page, gfp_flags);
C
Chris Mason 已提交
973
	if (!ret)
974 975 976
		return 0;

	ret = try_release_extent_buffer(tree, page);
977 978 979 980 981
	if (ret == 1) {
		ClearPagePrivate(page);
		set_page_private(page, 0);
		page_cache_release(page);
	}
982

983 984 985
	return ret;
}

986
static void btree_invalidatepage(struct page *page, unsigned long offset)
987
{
988 989
	struct extent_io_tree *tree;
	tree = &BTRFS_I(page->mapping->host)->io_tree;
990 991
	extent_invalidatepage(tree, page, offset);
	btree_releasepage(page, GFP_NOFS);
992
	if (PagePrivate(page)) {
C
Chris Mason 已提交
993 994
		printk(KERN_WARNING "btrfs warning page private not zero "
		       "on page %llu\n", (unsigned long long)page_offset(page));
995 996 997 998
		ClearPagePrivate(page);
		set_page_private(page, 0);
		page_cache_release(page);
	}
999 1000
}

1001
static const struct address_space_operations btree_aops = {
1002 1003
	.readpage	= btree_readpage,
	.writepage	= btree_writepage,
1004
	.writepages	= btree_writepages,
1005 1006
	.releasepage	= btree_releasepage,
	.invalidatepage = btree_invalidatepage,
1007
#ifdef CONFIG_MIGRATION
1008
	.migratepage	= btree_migratepage,
1009
#endif
1010 1011
};

1012 1013
int readahead_tree_block(struct btrfs_root *root, u64 bytenr, u32 blocksize,
			 u64 parent_transid)
C
Chris Mason 已提交
1014
{
1015 1016
	struct extent_buffer *buf = NULL;
	struct inode *btree_inode = root->fs_info->btree_inode;
1017
	int ret = 0;
C
Chris Mason 已提交
1018

1019
	buf = btrfs_find_create_tree_block(root, bytenr, blocksize);
1020
	if (!buf)
C
Chris Mason 已提交
1021
		return 0;
1022
	read_extent_buffer_pages(&BTRFS_I(btree_inode)->io_tree,
1023
				 buf, 0, WAIT_NONE, btree_get_extent, 0);
1024
	free_extent_buffer(buf);
1025
	return ret;
C
Chris Mason 已提交
1026 1027
}

1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059
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;
	} else if (extent_buffer_uptodate(io_tree, buf, NULL)) {
		*eb = buf;
	} else {
		free_extent_buffer(buf);
	}
	return 0;
}

1060 1061 1062 1063 1064 1065
struct extent_buffer *btrfs_find_tree_block(struct btrfs_root *root,
					    u64 bytenr, u32 blocksize)
{
	struct inode *btree_inode = root->fs_info->btree_inode;
	struct extent_buffer *eb;
	eb = find_extent_buffer(&BTRFS_I(btree_inode)->io_tree,
1066
				bytenr, blocksize);
1067 1068 1069 1070 1071 1072 1073 1074 1075 1076
	return eb;
}

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

	eb = alloc_extent_buffer(&BTRFS_I(btree_inode)->io_tree,
1077
				 bytenr, blocksize, NULL);
1078 1079 1080 1081
	return eb;
}


1082 1083
int btrfs_write_tree_block(struct extent_buffer *buf)
{
1084 1085
	return filemap_fdatawrite_range(buf->first_page->mapping, buf->start,
					buf->start + buf->len - 1);
1086 1087 1088 1089
}

int btrfs_wait_tree_block_writeback(struct extent_buffer *buf)
{
1090 1091
	return filemap_fdatawait_range(buf->first_page->mapping,
				       buf->start, buf->start + buf->len - 1);
1092 1093
}

1094
struct extent_buffer *read_tree_block(struct btrfs_root *root, u64 bytenr,
1095
				      u32 blocksize, u64 parent_transid)
1096 1097 1098 1099 1100 1101 1102 1103
{
	struct extent_buffer *buf = NULL;
	int ret;

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

1104
	ret = btree_read_extent_buffer_pages(root, buf, 0, parent_transid);
1105

C
Chris Mason 已提交
1106
	if (ret == 0)
1107
		set_bit(EXTENT_BUFFER_UPTODATE, &buf->bflags);
1108
	return buf;
1109

1110 1111
}

1112 1113
void clean_tree_block(struct btrfs_trans_handle *trans, struct btrfs_root *root,
		      struct extent_buffer *buf)
1114
{
1115
	struct inode *btree_inode = root->fs_info->btree_inode;
1116
	if (btrfs_header_generation(buf) ==
1117
	    root->fs_info->running_transaction->transid) {
1118
		btrfs_assert_tree_locked(buf);
1119

1120 1121 1122 1123
		if (test_and_clear_bit(EXTENT_BUFFER_DIRTY, &buf->bflags)) {
			spin_lock(&root->fs_info->delalloc_lock);
			if (root->fs_info->dirty_metadata_bytes >= buf->len)
				root->fs_info->dirty_metadata_bytes -= buf->len;
1124 1125 1126 1127 1128 1129 1130 1131
			else {
				spin_unlock(&root->fs_info->delalloc_lock);
				btrfs_panic(root->fs_info, -EOVERFLOW,
					  "Can't clear %lu bytes from "
					  " dirty_mdatadata_bytes (%lu)",
					  buf->len,
					  root->fs_info->dirty_metadata_bytes);
			}
1132 1133
			spin_unlock(&root->fs_info->delalloc_lock);
		}
1134

1135 1136
		/* ugh, clear_extent_buffer_dirty needs to lock the page */
		btrfs_set_lock_blocking(buf);
1137
		clear_extent_buffer_dirty(&BTRFS_I(btree_inode)->io_tree,
1138
					  buf);
1139
	}
1140 1141
}

1142
static int __setup_root(u32 nodesize, u32 leafsize, u32 sectorsize,
1143
			u32 stripesize, struct btrfs_root *root,
1144
			struct btrfs_fs_info *fs_info,
C
Chris Mason 已提交
1145
			u64 objectid)
1146
{
C
Chris Mason 已提交
1147
	root->node = NULL;
1148
	root->commit_root = NULL;
1149 1150 1151
	root->sectorsize = sectorsize;
	root->nodesize = nodesize;
	root->leafsize = leafsize;
1152
	root->stripesize = stripesize;
C
Chris Mason 已提交
1153
	root->ref_cows = 0;
1154
	root->track_dirty = 0;
1155
	root->in_radix = 0;
1156 1157
	root->orphan_item_inserted = 0;
	root->orphan_cleanup_state = 0;
1158

1159 1160
	root->objectid = objectid;
	root->last_trans = 0;
1161
	root->highest_objectid = 0;
1162
	root->name = NULL;
1163
	root->inode_tree = RB_ROOT;
1164
	INIT_RADIX_TREE(&root->delayed_nodes_tree, GFP_ATOMIC);
1165
	root->block_rsv = NULL;
1166
	root->orphan_block_rsv = NULL;
1167 1168

	INIT_LIST_HEAD(&root->dirty_list);
1169
	INIT_LIST_HEAD(&root->orphan_list);
1170
	INIT_LIST_HEAD(&root->root_list);
1171
	spin_lock_init(&root->orphan_lock);
1172
	spin_lock_init(&root->inode_lock);
1173
	spin_lock_init(&root->accounting_lock);
1174
	mutex_init(&root->objectid_mutex);
1175
	mutex_init(&root->log_mutex);
Y
Yan Zheng 已提交
1176 1177 1178 1179 1180 1181 1182 1183
	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);
	root->log_batch = 0;
	root->log_transid = 0;
1184
	root->last_log_commit = 0;
1185
	extent_io_tree_init(&root->dirty_log_pages,
1186
			     fs_info->btree_inode->i_mapping);
C
Chris Mason 已提交
1187

1188 1189
	memset(&root->root_key, 0, sizeof(root->root_key));
	memset(&root->root_item, 0, sizeof(root->root_item));
1190
	memset(&root->defrag_progress, 0, sizeof(root->defrag_progress));
1191
	memset(&root->root_kobj, 0, sizeof(root->root_kobj));
1192
	root->defrag_trans_start = fs_info->generation;
1193
	init_completion(&root->kobj_unregister);
1194
	root->defrag_running = 0;
1195
	root->root_key.objectid = objectid;
1196
	root->anon_dev = 0;
1197 1198 1199
	return 0;
}

1200
static int find_and_setup_root(struct btrfs_root *tree_root,
1201 1202
			       struct btrfs_fs_info *fs_info,
			       u64 objectid,
C
Chris Mason 已提交
1203
			       struct btrfs_root *root)
1204 1205
{
	int ret;
1206
	u32 blocksize;
1207
	u64 generation;
1208

1209
	__setup_root(tree_root->nodesize, tree_root->leafsize,
1210 1211
		     tree_root->sectorsize, tree_root->stripesize,
		     root, fs_info, objectid);
1212 1213
	ret = btrfs_find_last_root(tree_root, objectid,
				   &root->root_item, &root->root_key);
1214 1215
	if (ret > 0)
		return -ENOENT;
1216 1217
	BUG_ON(ret);

1218
	generation = btrfs_root_generation(&root->root_item);
1219
	blocksize = btrfs_level_size(root, btrfs_root_level(&root->root_item));
C
Chris Mason 已提交
1220
	root->commit_root = NULL;
1221
	root->node = read_tree_block(root, btrfs_root_bytenr(&root->root_item),
1222
				     blocksize, generation);
1223 1224
	if (!root->node || !btrfs_buffer_uptodate(root->node, generation)) {
		free_extent_buffer(root->node);
C
Chris Mason 已提交
1225
		root->node = NULL;
1226 1227
		return -EIO;
	}
1228
	root->commit_root = btrfs_root_node(root);
1229 1230 1231
	return 0;
}

1232
static struct btrfs_root *btrfs_alloc_root(struct btrfs_fs_info *fs_info)
A
Al Viro 已提交
1233 1234 1235 1236 1237 1238 1239
{
	struct btrfs_root *root = kzalloc(sizeof(*root), GFP_NOFS);
	if (root)
		root->fs_info = fs_info;
	return root;
}

Y
Yan Zheng 已提交
1240 1241
static struct btrfs_root *alloc_log_tree(struct btrfs_trans_handle *trans,
					 struct btrfs_fs_info *fs_info)
1242 1243 1244
{
	struct btrfs_root *root;
	struct btrfs_root *tree_root = fs_info->tree_root;
Y
Yan Zheng 已提交
1245
	struct extent_buffer *leaf;
1246

A
Al Viro 已提交
1247
	root = btrfs_alloc_root(fs_info);
1248
	if (!root)
Y
Yan Zheng 已提交
1249
		return ERR_PTR(-ENOMEM);
1250 1251 1252 1253 1254 1255 1256 1257

	__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
1258 1259 1260 1261 1262 1263
	/*
	 * 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).
	 */
1264 1265
	root->ref_cows = 0;

1266
	leaf = btrfs_alloc_free_block(trans, root, root->leafsize, 0,
A
Arne Jansen 已提交
1267 1268
				      BTRFS_TREE_LOG_OBJECTID, NULL,
				      0, 0, 0, 0);
Y
Yan Zheng 已提交
1269 1270 1271 1272
	if (IS_ERR(leaf)) {
		kfree(root);
		return ERR_CAST(leaf);
	}
1273

1274 1275 1276 1277 1278
	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
1279
	root->node = leaf;
1280 1281 1282 1283 1284 1285

	write_extent_buffer(root->node, root->fs_info->fsid,
			    (unsigned long)btrfs_header_fsid(root->node),
			    BTRFS_FSID_SIZE);
	btrfs_mark_buffer_dirty(root->node);
	btrfs_tree_unlock(root->node);
Y
Yan Zheng 已提交
1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321
	return root;
}

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

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

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

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

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

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

1322
	btrfs_set_root_node(&log_root->root_item, log_root->node);
Y
Yan Zheng 已提交
1323 1324 1325 1326

	WARN_ON(root->log_root);
	root->log_root = log_root;
	root->log_transid = 0;
1327
	root->last_log_commit = 0;
1328 1329 1330 1331 1332 1333 1334 1335
	return 0;
}

struct btrfs_root *btrfs_read_fs_root_no_radix(struct btrfs_root *tree_root,
					       struct btrfs_key *location)
{
	struct btrfs_root *root;
	struct btrfs_fs_info *fs_info = tree_root->fs_info;
1336
	struct btrfs_path *path;
1337
	struct extent_buffer *l;
1338
	u64 generation;
1339
	u32 blocksize;
1340 1341
	int ret = 0;

A
Al Viro 已提交
1342
	root = btrfs_alloc_root(fs_info);
C
Chris Mason 已提交
1343
	if (!root)
1344 1345
		return ERR_PTR(-ENOMEM);
	if (location->offset == (u64)-1) {
1346
		ret = find_and_setup_root(tree_root, fs_info,
1347 1348 1349 1350 1351
					  location->objectid, root);
		if (ret) {
			kfree(root);
			return ERR_PTR(ret);
		}
1352
		goto out;
1353 1354
	}

1355
	__setup_root(tree_root->nodesize, tree_root->leafsize,
1356 1357
		     tree_root->sectorsize, tree_root->stripesize,
		     root, fs_info, location->objectid);
1358 1359

	path = btrfs_alloc_path();
T
Tsutomu Itoh 已提交
1360 1361 1362 1363
	if (!path) {
		kfree(root);
		return ERR_PTR(-ENOMEM);
	}
1364
	ret = btrfs_search_slot(NULL, tree_root, location, path, 0, 0);
1365 1366 1367 1368 1369 1370
	if (ret == 0) {
		l = path->nodes[0];
		read_extent_buffer(l, &root->root_item,
				btrfs_item_ptr_offset(l, path->slots[0]),
				sizeof(root->root_item));
		memcpy(&root->root_key, location, sizeof(*location));
1371 1372 1373
	}
	btrfs_free_path(path);
	if (ret) {
1374
		kfree(root);
1375 1376
		if (ret > 0)
			ret = -ENOENT;
1377 1378
		return ERR_PTR(ret);
	}
1379

1380
	generation = btrfs_root_generation(&root->root_item);
1381 1382
	blocksize = btrfs_level_size(root, btrfs_root_level(&root->root_item));
	root->node = read_tree_block(root, btrfs_root_bytenr(&root->root_item),
1383
				     blocksize, generation);
1384
	root->commit_root = btrfs_root_node(root);
1385
	BUG_ON(!root->node);
1386
out:
1387
	if (location->objectid != BTRFS_TREE_LOG_OBJECTID) {
1388
		root->ref_cows = 1;
1389 1390
		btrfs_check_and_init_root_item(&root->root_item);
	}
1391

1392 1393 1394
	return root;
}

1395 1396
struct btrfs_root *btrfs_read_fs_root_no_name(struct btrfs_fs_info *fs_info,
					      struct btrfs_key *location)
1397 1398 1399 1400
{
	struct btrfs_root *root;
	int ret;

1401 1402 1403 1404
	if (location->objectid == BTRFS_ROOT_TREE_OBJECTID)
		return fs_info->tree_root;
	if (location->objectid == BTRFS_EXTENT_TREE_OBJECTID)
		return fs_info->extent_root;
1405 1406 1407 1408
	if (location->objectid == BTRFS_CHUNK_TREE_OBJECTID)
		return fs_info->chunk_root;
	if (location->objectid == BTRFS_DEV_TREE_OBJECTID)
		return fs_info->dev_root;
1409 1410
	if (location->objectid == BTRFS_CSUM_TREE_OBJECTID)
		return fs_info->csum_root;
1411 1412
again:
	spin_lock(&fs_info->fs_roots_radix_lock);
1413 1414
	root = radix_tree_lookup(&fs_info->fs_roots_radix,
				 (unsigned long)location->objectid);
1415
	spin_unlock(&fs_info->fs_roots_radix_lock);
1416 1417 1418
	if (root)
		return root;

1419
	root = btrfs_read_fs_root_no_radix(fs_info->tree_root, location);
1420 1421
	if (IS_ERR(root))
		return root;
1422

1423 1424 1425
	root->free_ino_ctl = kzalloc(sizeof(*root->free_ino_ctl), GFP_NOFS);
	root->free_ino_pinned = kzalloc(sizeof(*root->free_ino_pinned),
					GFP_NOFS);
1426 1427
	if (!root->free_ino_pinned || !root->free_ino_ctl) {
		ret = -ENOMEM;
1428
		goto fail;
1429
	}
1430 1431 1432 1433 1434 1435

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

1436
	ret = get_anon_bdev(&root->anon_dev);
1437 1438
	if (ret)
		goto fail;
1439

1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450
	if (btrfs_root_refs(&root->root_item) == 0) {
		ret = -ENOENT;
		goto fail;
	}

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

1451 1452 1453 1454 1455
	ret = radix_tree_preload(GFP_NOFS & ~__GFP_HIGHMEM);
	if (ret)
		goto fail;

	spin_lock(&fs_info->fs_roots_radix_lock);
C
Chris Mason 已提交
1456 1457
	ret = radix_tree_insert(&fs_info->fs_roots_radix,
				(unsigned long)root->root_key.objectid,
1458
				root);
1459
	if (ret == 0)
1460
		root->in_radix = 1;
1461

1462 1463
	spin_unlock(&fs_info->fs_roots_radix_lock);
	radix_tree_preload_end();
1464
	if (ret) {
1465 1466 1467 1468 1469
		if (ret == -EEXIST) {
			free_fs_root(root);
			goto again;
		}
		goto fail;
1470
	}
1471 1472 1473 1474

	ret = btrfs_find_dead_roots(fs_info->tree_root,
				    root->root_key.objectid);
	WARN_ON(ret);
1475
	return root;
1476 1477 1478
fail:
	free_fs_root(root);
	return ERR_PTR(ret);
1479 1480
}

C
Chris Mason 已提交
1481 1482 1483 1484 1485 1486
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
1487

1488 1489
	rcu_read_lock();
	list_for_each_entry_rcu(device, &info->fs_devices->devices, dev_list) {
1490 1491
		if (!device->bdev)
			continue;
C
Chris Mason 已提交
1492 1493 1494 1495 1496 1497
		bdi = blk_get_backing_dev_info(device->bdev);
		if (bdi && bdi_congested(bdi, bdi_bits)) {
			ret = 1;
			break;
		}
	}
1498
	rcu_read_unlock();
C
Chris Mason 已提交
1499 1500 1501
	return ret;
}

1502 1503 1504 1505
/*
 * If this fails, caller must call bdi_destroy() to get rid of the
 * bdi again.
 */
C
Chris Mason 已提交
1506 1507
static int setup_bdi(struct btrfs_fs_info *info, struct backing_dev_info *bdi)
{
1508 1509 1510
	int err;

	bdi->capabilities = BDI_CAP_MAP_COPY;
1511
	err = bdi_setup_and_register(bdi, "btrfs", BDI_CAP_MAP_COPY);
1512 1513 1514
	if (err)
		return err;

1515
	bdi->ra_pages	= default_backing_dev_info.ra_pages;
C
Chris Mason 已提交
1516 1517 1518 1519 1520
	bdi->congested_fn	= btrfs_congested_fn;
	bdi->congested_data	= info;
	return 0;
}

1521 1522 1523 1524 1525 1526 1527 1528 1529 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
static int bio_ready_for_csum(struct bio *bio)
{
	u64 length = 0;
	u64 buf_len = 0;
	u64 start = 0;
	struct page *page;
	struct extent_io_tree *io_tree = NULL;
	struct bio_vec *bvec;
	int i;
	int ret;

	bio_for_each_segment(bvec, bio, i) {
		page = bvec->bv_page;
		if (page->private == EXTENT_PAGE_PRIVATE) {
			length += bvec->bv_len;
			continue;
		}
		if (!page->private) {
			length += bvec->bv_len;
			continue;
		}
		length = bvec->bv_len;
		buf_len = page->private >> 2;
		start = page_offset(page) + bvec->bv_offset;
		io_tree = &BTRFS_I(page->mapping->host)->io_tree;
	}
	/* are we fully contained in this bio? */
	if (buf_len <= length)
		return 1;

	ret = extent_range_uptodate(io_tree, start + length,
				    start + buf_len - 1);
	return ret;
}

1556 1557 1558 1559 1560
/*
 * 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)
1561 1562
{
	struct bio *bio;
1563 1564
	struct end_io_wq *end_io_wq;
	struct btrfs_fs_info *fs_info;
1565 1566
	int error;

1567 1568 1569
	end_io_wq = container_of(work, struct end_io_wq, work);
	bio = end_io_wq->bio;
	fs_info = end_io_wq->info;
1570

1571
	/* metadata bio reads are special because the whole tree block must
1572 1573 1574 1575
	 * be checksummed at once.  This makes sure the entire block is in
	 * ram and up to date before trying to verify things.  For
	 * blocksize <= pagesize, it is basically a noop
	 */
1576
	if (!(bio->bi_rw & REQ_WRITE) && end_io_wq->metadata &&
1577
	    !bio_ready_for_csum(bio)) {
1578
		btrfs_queue_worker(&fs_info->endio_meta_workers,
1579 1580 1581 1582 1583 1584 1585 1586
				   &end_io_wq->work);
		return;
	}
	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);
1587 1588
}

1589 1590 1591 1592 1593 1594
static int cleaner_kthread(void *arg)
{
	struct btrfs_root *root = arg;

	do {
		vfs_check_frozen(root->fs_info->sb, SB_FREEZE_WRITE);
1595 1596 1597

		if (!(root->fs_info->sb->s_flags & MS_RDONLY) &&
		    mutex_trylock(&root->fs_info->cleaner_mutex)) {
Y
Yan, Zheng 已提交
1598
			btrfs_run_delayed_iputs(root);
1599 1600
			btrfs_clean_old_snapshots(root);
			mutex_unlock(&root->fs_info->cleaner_mutex);
C
Chris Mason 已提交
1601
			btrfs_run_defrag_inodes(root->fs_info);
1602
		}
1603

1604
		if (!try_to_freeze()) {
1605
			set_current_state(TASK_INTERRUPTIBLE);
1606 1607
			if (!kthread_should_stop())
				schedule();
1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618
			__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;
1619
	u64 transid;
1620 1621 1622 1623 1624 1625 1626 1627 1628
	unsigned long now;
	unsigned long delay;
	int ret;

	do {
		delay = HZ * 30;
		vfs_check_frozen(root->fs_info->sb, SB_FREEZE_WRITE);
		mutex_lock(&root->fs_info->transaction_kthread_mutex);

J
Josef Bacik 已提交
1629
		spin_lock(&root->fs_info->trans_lock);
1630 1631
		cur = root->fs_info->running_transaction;
		if (!cur) {
J
Josef Bacik 已提交
1632
			spin_unlock(&root->fs_info->trans_lock);
1633 1634
			goto sleep;
		}
Y
Yan Zheng 已提交
1635

1636
		now = get_seconds();
1637 1638
		if (!cur->blocked &&
		    (now < cur->start_time || now - cur->start_time < 30)) {
J
Josef Bacik 已提交
1639
			spin_unlock(&root->fs_info->trans_lock);
1640 1641 1642
			delay = HZ * 5;
			goto sleep;
		}
1643
		transid = cur->transid;
J
Josef Bacik 已提交
1644
		spin_unlock(&root->fs_info->trans_lock);
1645

1646
		trans = btrfs_join_transaction(root);
1647
		BUG_ON(IS_ERR(trans));
1648 1649 1650 1651 1652 1653
		if (transid == trans->transid) {
			ret = btrfs_commit_transaction(trans, root);
			BUG_ON(ret);
		} else {
			btrfs_end_transaction(trans, root);
		}
1654 1655 1656 1657
sleep:
		wake_up_process(root->fs_info->cleaner_kthread);
		mutex_unlock(&root->fs_info->transaction_kthread_mutex);

1658
		if (!try_to_freeze()) {
1659
			set_current_state(TASK_INTERRUPTIBLE);
1660 1661 1662
			if (!kthread_should_stop() &&
			    !btrfs_transaction_blocked(root->fs_info))
				schedule_timeout(delay);
1663 1664 1665 1666 1667 1668
			__set_current_state(TASK_RUNNING);
		}
	} while (!kthread_should_stop());
	return 0;
}

C
Chris Mason 已提交
1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774
/*
 * 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));

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

	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;
}

/* helper to cleanup tree roots */
static void free_root_pointers(struct btrfs_fs_info *info, int chunk_root)
{
	free_extent_buffer(info->tree_root->node);
	free_extent_buffer(info->tree_root->commit_root);
	free_extent_buffer(info->dev_root->node);
	free_extent_buffer(info->dev_root->commit_root);
	free_extent_buffer(info->extent_root->node);
	free_extent_buffer(info->extent_root->commit_root);
	free_extent_buffer(info->csum_root->node);
	free_extent_buffer(info->csum_root->commit_root);

	info->tree_root->node = NULL;
	info->tree_root->commit_root = NULL;
	info->dev_root->node = NULL;
	info->dev_root->commit_root = NULL;
	info->extent_root->node = NULL;
	info->extent_root->commit_root = NULL;
	info->csum_root->node = NULL;
	info->csum_root->commit_root = NULL;

	if (chunk_root) {
		free_extent_buffer(info->chunk_root->node);
		free_extent_buffer(info->chunk_root->commit_root);
		info->chunk_root->node = NULL;
		info->chunk_root->commit_root = NULL;
	}
}


A
Al Viro 已提交
1898 1899 1900
int open_ctree(struct super_block *sb,
	       struct btrfs_fs_devices *fs_devices,
	       char *options)
1901
{
1902 1903 1904 1905
	u32 sectorsize;
	u32 nodesize;
	u32 leafsize;
	u32 blocksize;
1906
	u32 stripesize;
1907
	u64 generation;
1908
	u64 features;
1909
	struct btrfs_key location;
1910
	struct buffer_head *bh;
1911
	struct btrfs_super_block *disk_super;
1912
	struct btrfs_fs_info *fs_info = btrfs_sb(sb);
1913
	struct btrfs_root *tree_root;
1914 1915 1916 1917
	struct btrfs_root *extent_root;
	struct btrfs_root *csum_root;
	struct btrfs_root *chunk_root;
	struct btrfs_root *dev_root;
1918
	struct btrfs_root *log_tree_root;
1919
	int ret;
1920
	int err = -EINVAL;
C
Chris Mason 已提交
1921 1922
	int num_backups_tried = 0;
	int backup_index = 0;
1923

1924
	tree_root = fs_info->tree_root = btrfs_alloc_root(fs_info);
A
Al Viro 已提交
1925 1926 1927 1928
	extent_root = fs_info->extent_root = btrfs_alloc_root(fs_info);
	csum_root = fs_info->csum_root = btrfs_alloc_root(fs_info);
	chunk_root = fs_info->chunk_root = btrfs_alloc_root(fs_info);
	dev_root = fs_info->dev_root = btrfs_alloc_root(fs_info);
1929

1930 1931
	if (!tree_root || !extent_root || !csum_root ||
	    !chunk_root || !dev_root) {
C
Chris Mason 已提交
1932 1933 1934
		err = -ENOMEM;
		goto fail;
	}
1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953

	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;
	}

	fs_info->btree_inode = new_inode(sb);
	if (!fs_info->btree_inode) {
		err = -ENOMEM;
		goto fail_bdi;
	}

1954
	mapping_set_gfp_mask(fs_info->btree_inode->i_mapping, GFP_NOFS);
1955

1956
	INIT_RADIX_TREE(&fs_info->fs_roots_radix, GFP_ATOMIC);
C
Chris Mason 已提交
1957
	INIT_LIST_HEAD(&fs_info->trans_list);
1958
	INIT_LIST_HEAD(&fs_info->dead_roots);
Y
Yan, Zheng 已提交
1959
	INIT_LIST_HEAD(&fs_info->delayed_iputs);
1960
	INIT_LIST_HEAD(&fs_info->hashers);
1961
	INIT_LIST_HEAD(&fs_info->delalloc_inodes);
1962
	INIT_LIST_HEAD(&fs_info->ordered_operations);
1963
	INIT_LIST_HEAD(&fs_info->caching_block_groups);
1964
	spin_lock_init(&fs_info->delalloc_lock);
J
Josef Bacik 已提交
1965
	spin_lock_init(&fs_info->trans_lock);
Y
Yan Zheng 已提交
1966
	spin_lock_init(&fs_info->ref_cache_lock);
1967
	spin_lock_init(&fs_info->fs_roots_radix_lock);
Y
Yan, Zheng 已提交
1968
	spin_lock_init(&fs_info->delayed_iput_lock);
C
Chris Mason 已提交
1969
	spin_lock_init(&fs_info->defrag_inodes_lock);
1970
	spin_lock_init(&fs_info->free_chunk_lock);
C
Chris Mason 已提交
1971
	mutex_init(&fs_info->reloc_mutex);
1972

1973
	init_completion(&fs_info->kobj_unregister);
1974
	INIT_LIST_HEAD(&fs_info->dirty_cowonly_roots);
1975
	INIT_LIST_HEAD(&fs_info->space_info);
1976
	btrfs_mapping_init(&fs_info->mapping_tree);
1977 1978 1979 1980 1981
	btrfs_init_block_rsv(&fs_info->global_block_rsv);
	btrfs_init_block_rsv(&fs_info->delalloc_block_rsv);
	btrfs_init_block_rsv(&fs_info->trans_block_rsv);
	btrfs_init_block_rsv(&fs_info->chunk_block_rsv);
	btrfs_init_block_rsv(&fs_info->empty_block_rsv);
1982
	btrfs_init_block_rsv(&fs_info->delayed_block_rsv);
1983
	atomic_set(&fs_info->nr_async_submits, 0);
1984
	atomic_set(&fs_info->async_delalloc_pages, 0);
1985
	atomic_set(&fs_info->async_submit_draining, 0);
1986
	atomic_set(&fs_info->nr_async_bios, 0);
C
Chris Mason 已提交
1987
	atomic_set(&fs_info->defrag_running, 0);
C
Chris Mason 已提交
1988
	fs_info->sb = sb;
1989
	fs_info->max_inline = 8192 * 1024;
J
Josef Bacik 已提交
1990
	fs_info->metadata_ratio = 0;
C
Chris Mason 已提交
1991
	fs_info->defrag_inodes = RB_ROOT;
J
Josef Bacik 已提交
1992
	fs_info->trans_no_join = 0;
1993
	fs_info->free_chunk_space = 0;
C
Chris Mason 已提交
1994

1995 1996 1997
	/* 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
1998

1999 2000
	fs_info->thread_pool_size = min_t(unsigned long,
					  num_online_cpus() + 2, 8);
2001

2002 2003
	INIT_LIST_HEAD(&fs_info->ordered_extents);
	spin_lock_init(&fs_info->ordered_extent_lock);
2004 2005 2006 2007 2008 2009 2010
	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);
2011

A
Arne Jansen 已提交
2012 2013 2014 2015 2016 2017 2018 2019
	mutex_init(&fs_info->scrub_lock);
	atomic_set(&fs_info->scrubs_running, 0);
	atomic_set(&fs_info->scrub_pause_req, 0);
	atomic_set(&fs_info->scrubs_paused, 0);
	atomic_set(&fs_info->scrub_cancel_req, 0);
	init_waitqueue_head(&fs_info->scrub_pause_wait);
	init_rwsem(&fs_info->scrub_super_lock);
	fs_info->scrub_workers_refcnt = 0;
2020 2021 2022
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
	fs_info->check_integrity_print_mask = 0;
#endif
A
Arne Jansen 已提交
2023

2024 2025
	spin_lock_init(&fs_info->balance_lock);
	mutex_init(&fs_info->balance_mutex);
2026 2027
	atomic_set(&fs_info->balance_running, 0);
	atomic_set(&fs_info->balance_pause_req, 0);
2028
	atomic_set(&fs_info->balance_cancel_req, 0);
2029
	fs_info->balance_ctl = NULL;
2030
	init_waitqueue_head(&fs_info->balance_wait_q);
A
Arne Jansen 已提交
2031

2032 2033
	sb->s_blocksize = 4096;
	sb->s_blocksize_bits = blksize_bits(4096);
J
Jens Axboe 已提交
2034
	sb->s_bdi = &fs_info->bdi;
2035

2036
	fs_info->btree_inode->i_ino = BTRFS_BTREE_INODE_OBJECTID;
M
Miklos Szeredi 已提交
2037
	set_nlink(fs_info->btree_inode, 1);
2038 2039 2040 2041 2042 2043
	/*
	 * 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;
2044
	fs_info->btree_inode->i_mapping->a_ops = &btree_aops;
C
Chris Mason 已提交
2045 2046
	fs_info->btree_inode->i_mapping->backing_dev_info = &fs_info->bdi;

2047
	RB_CLEAR_NODE(&BTRFS_I(fs_info->btree_inode)->rb_node);
2048
	extent_io_tree_init(&BTRFS_I(fs_info->btree_inode)->io_tree,
2049
			     fs_info->btree_inode->i_mapping);
2050
	extent_map_tree_init(&BTRFS_I(fs_info->btree_inode)->extent_tree);
2051 2052

	BTRFS_I(fs_info->btree_inode)->io_tree.ops = &btree_extent_io_ops;
2053

2054 2055 2056 2057
	BTRFS_I(fs_info->btree_inode)->root = tree_root;
	memset(&BTRFS_I(fs_info->btree_inode)->location, 0,
	       sizeof(struct btrfs_key));
	BTRFS_I(fs_info->btree_inode)->dummy_inode = 1;
2058
	insert_inode_hash(fs_info->btree_inode);
2059

J
Josef Bacik 已提交
2060
	spin_lock_init(&fs_info->block_group_cache_lock);
2061
	fs_info->block_group_cache_tree = RB_ROOT;
J
Josef Bacik 已提交
2062

2063
	extent_io_tree_init(&fs_info->freed_extents[0],
2064
			     fs_info->btree_inode->i_mapping);
2065
	extent_io_tree_init(&fs_info->freed_extents[1],
2066
			     fs_info->btree_inode->i_mapping);
2067
	fs_info->pinned_extents = &fs_info->freed_extents[0];
2068
	fs_info->do_barriers = 1;
2069

C
Chris Mason 已提交
2070

2071
	mutex_init(&fs_info->ordered_operations_mutex);
2072
	mutex_init(&fs_info->tree_log_mutex);
2073
	mutex_init(&fs_info->chunk_mutex);
2074 2075
	mutex_init(&fs_info->transaction_kthread_mutex);
	mutex_init(&fs_info->cleaner_mutex);
2076
	mutex_init(&fs_info->volume_mutex);
2077
	init_rwsem(&fs_info->extent_commit_sem);
2078
	init_rwsem(&fs_info->cleanup_work_sem);
2079
	init_rwsem(&fs_info->subvol_sem);
2080 2081 2082 2083

	btrfs_init_free_cluster(&fs_info->meta_alloc_cluster);
	btrfs_init_free_cluster(&fs_info->data_alloc_cluster);

2084
	init_waitqueue_head(&fs_info->transaction_throttle);
2085
	init_waitqueue_head(&fs_info->transaction_wait);
S
Sage Weil 已提交
2086
	init_waitqueue_head(&fs_info->transaction_blocked_wait);
2087
	init_waitqueue_head(&fs_info->async_submit_wait);
2088

2089
	__setup_root(4096, 4096, 4096, 4096, tree_root,
C
Chris Mason 已提交
2090
		     fs_info, BTRFS_ROOT_TREE_OBJECTID);
2091

Y
Yan Zheng 已提交
2092
	bh = btrfs_read_dev_super(fs_devices->latest_bdev);
2093 2094
	if (!bh) {
		err = -EINVAL;
2095
		goto fail_alloc;
2096
	}
C
Chris Mason 已提交
2097

2098 2099 2100
	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));
2101
	brelse(bh);
2102

2103
	memcpy(fs_info->fsid, fs_info->super_copy->fsid, BTRFS_FSID_SIZE);
2104

2105
	disk_super = fs_info->super_copy;
2106
	if (!btrfs_super_root(disk_super))
2107
		goto fail_alloc;
2108

L
liubo 已提交
2109 2110 2111 2112 2113
	/* check FS state, whether FS is broken. */
	fs_info->fs_state |= btrfs_super_flags(disk_super);

	btrfs_check_super_valid(fs_info, sb->s_flags & MS_RDONLY);

C
Chris Mason 已提交
2114 2115 2116 2117 2118 2119 2120
	/*
	 * 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);

2121 2122 2123 2124 2125 2126
	/*
	 * 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
2127 2128 2129
	ret = btrfs_parse_options(tree_root, options);
	if (ret) {
		err = ret;
2130
		goto fail_alloc;
Y
Yan Zheng 已提交
2131
	}
2132

2133 2134 2135 2136 2137
	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",
2138
		       (unsigned long long)features);
2139
		err = -EINVAL;
2140
		goto fail_alloc;
2141 2142
	}

2143
	features = btrfs_super_incompat_flags(disk_super);
L
Li Zefan 已提交
2144 2145 2146 2147
	features |= BTRFS_FEATURE_INCOMPAT_MIXED_BACKREF;
	if (tree_root->fs_info->compress_type & BTRFS_COMPRESS_LZO)
		features |= BTRFS_FEATURE_INCOMPAT_COMPRESS_LZO;
	btrfs_set_super_incompat_flags(disk_super, features);
2148

2149 2150 2151 2152 2153
	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",
2154
		       (unsigned long long)features);
2155
		err = -EINVAL;
2156
		goto fail_alloc;
2157
	}
2158 2159 2160 2161

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

2162
	btrfs_init_workers(&fs_info->workers, "worker",
2163 2164
			   fs_info->thread_pool_size,
			   &fs_info->generic_worker);
C
Chris Mason 已提交
2165

2166
	btrfs_init_workers(&fs_info->delalloc_workers, "delalloc",
2167 2168
			   fs_info->thread_pool_size,
			   &fs_info->generic_worker);
2169

2170
	btrfs_init_workers(&fs_info->submit_workers, "submit",
2171
			   min_t(u64, fs_devices->num_devices,
2172 2173
			   fs_info->thread_pool_size),
			   &fs_info->generic_worker);
2174

2175 2176 2177
	btrfs_init_workers(&fs_info->caching_workers, "cache",
			   2, &fs_info->generic_worker);

2178 2179 2180 2181 2182
	/* 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;
2183

2184
	fs_info->workers.idle_thresh = 16;
C
Chris Mason 已提交
2185
	fs_info->workers.ordered = 1;
2186

2187 2188 2189
	fs_info->delalloc_workers.idle_thresh = 2;
	fs_info->delalloc_workers.ordered = 1;

2190 2191
	btrfs_init_workers(&fs_info->fixup_workers, "fixup", 1,
			   &fs_info->generic_worker);
2192
	btrfs_init_workers(&fs_info->endio_workers, "endio",
2193 2194
			   fs_info->thread_pool_size,
			   &fs_info->generic_worker);
2195
	btrfs_init_workers(&fs_info->endio_meta_workers, "endio-meta",
2196 2197
			   fs_info->thread_pool_size,
			   &fs_info->generic_worker);
2198
	btrfs_init_workers(&fs_info->endio_meta_write_workers,
2199 2200
			   "endio-meta-write", fs_info->thread_pool_size,
			   &fs_info->generic_worker);
2201
	btrfs_init_workers(&fs_info->endio_write_workers, "endio-write",
2202 2203
			   fs_info->thread_pool_size,
			   &fs_info->generic_worker);
J
Josef Bacik 已提交
2204 2205
	btrfs_init_workers(&fs_info->endio_freespace_worker, "freespace-write",
			   1, &fs_info->generic_worker);
2206 2207 2208
	btrfs_init_workers(&fs_info->delayed_workers, "delayed-meta",
			   fs_info->thread_pool_size,
			   &fs_info->generic_worker);
2209 2210 2211
	btrfs_init_workers(&fs_info->readahead_workers, "readahead",
			   fs_info->thread_pool_size,
			   &fs_info->generic_worker);
2212 2213 2214 2215 2216 2217

	/*
	 * endios are largely parallel and should have a very
	 * low idle thresh
	 */
	fs_info->endio_workers.idle_thresh = 4;
2218 2219
	fs_info->endio_meta_workers.idle_thresh = 4;

2220 2221
	fs_info->endio_write_workers.idle_thresh = 2;
	fs_info->endio_meta_write_workers.idle_thresh = 2;
2222
	fs_info->readahead_workers.idle_thresh = 2;
2223

2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244
	/*
	 * 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);
	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);
	if (ret) {
		ret = -ENOMEM;
		goto fail_sb_buffer;
	}
2245

2246
	fs_info->bdi.ra_pages *= btrfs_super_num_devices(disk_super);
C
Chris Mason 已提交
2247 2248
	fs_info->bdi.ra_pages = max(fs_info->bdi.ra_pages,
				    4 * 1024 * 1024 / PAGE_CACHE_SIZE);
2249

2250 2251 2252
	nodesize = btrfs_super_nodesize(disk_super);
	leafsize = btrfs_super_leafsize(disk_super);
	sectorsize = btrfs_super_sectorsize(disk_super);
2253
	stripesize = btrfs_super_stripesize(disk_super);
2254 2255 2256
	tree_root->nodesize = nodesize;
	tree_root->leafsize = leafsize;
	tree_root->sectorsize = sectorsize;
2257
	tree_root->stripesize = stripesize;
2258 2259 2260

	sb->s_blocksize = sectorsize;
	sb->s_blocksize_bits = blksize_bits(sectorsize);
2261

C
Chris Mason 已提交
2262 2263
	if (strncmp((char *)(&disk_super->magic), BTRFS_MAGIC,
		    sizeof(disk_super->magic))) {
C
Chris Mason 已提交
2264
		printk(KERN_INFO "btrfs: valid FS not found on %s\n", sb->s_id);
C
Chris Mason 已提交
2265 2266
		goto fail_sb_buffer;
	}
2267

2268 2269 2270 2271 2272 2273
	if (sectorsize < PAGE_SIZE) {
		printk(KERN_WARNING "btrfs: Incompatible sector size "
		       "found on %s\n", sb->s_id);
		goto fail_sb_buffer;
	}

2274
	mutex_lock(&fs_info->chunk_mutex);
Y
Yan Zheng 已提交
2275
	ret = btrfs_read_sys_array(tree_root);
2276
	mutex_unlock(&fs_info->chunk_mutex);
2277
	if (ret) {
C
Chris Mason 已提交
2278 2279
		printk(KERN_WARNING "btrfs: failed to read the system "
		       "array on %s\n", sb->s_id);
2280
		goto fail_sb_buffer;
2281
	}
2282 2283 2284

	blocksize = btrfs_level_size(tree_root,
				     btrfs_super_chunk_root_level(disk_super));
2285
	generation = btrfs_super_chunk_root_generation(disk_super);
2286 2287 2288 2289 2290 2291

	__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),
2292
					   blocksize, generation);
2293
	BUG_ON(!chunk_root->node);
2294 2295 2296
	if (!test_bit(EXTENT_BUFFER_UPTODATE, &chunk_root->node->bflags)) {
		printk(KERN_WARNING "btrfs: failed to read chunk root on %s\n",
		       sb->s_id);
C
Chris Mason 已提交
2297
		goto fail_tree_roots;
2298
	}
2299 2300
	btrfs_set_root_node(&chunk_root->root_item, chunk_root->node);
	chunk_root->commit_root = btrfs_root_node(chunk_root);
2301

2302
	read_extent_buffer(chunk_root->node, fs_info->chunk_tree_uuid,
C
Chris Mason 已提交
2303 2304
	   (unsigned long)btrfs_header_chunk_tree_uuid(chunk_root->node),
	   BTRFS_UUID_SIZE);
2305

2306
	ret = btrfs_read_chunk_tree(chunk_root);
Y
Yan Zheng 已提交
2307
	if (ret) {
C
Chris Mason 已提交
2308 2309
		printk(KERN_WARNING "btrfs: failed to read chunk tree on %s\n",
		       sb->s_id);
C
Chris Mason 已提交
2310
		goto fail_tree_roots;
Y
Yan Zheng 已提交
2311
	}
2312

2313 2314
	btrfs_close_extra_devices(fs_devices);

2315 2316 2317 2318 2319 2320
	if (!fs_devices->latest_bdev) {
		printk(KERN_CRIT "btrfs: failed to read devices on %s\n",
		       sb->s_id);
		goto fail_tree_roots;
	}

C
Chris Mason 已提交
2321
retry_root_backup:
2322 2323
	blocksize = btrfs_level_size(tree_root,
				     btrfs_super_root_level(disk_super));
2324
	generation = btrfs_super_generation(disk_super);
2325

C
Chris Mason 已提交
2326
	tree_root->node = read_tree_block(tree_root,
2327
					  btrfs_super_root(disk_super),
2328
					  blocksize, generation);
C
Chris Mason 已提交
2329 2330
	if (!tree_root->node ||
	    !test_bit(EXTENT_BUFFER_UPTODATE, &tree_root->node->bflags)) {
2331 2332
		printk(KERN_WARNING "btrfs: failed to read tree root on %s\n",
		       sb->s_id);
C
Chris Mason 已提交
2333 2334

		goto recovery_tree_root;
2335
	}
C
Chris Mason 已提交
2336

2337 2338
	btrfs_set_root_node(&tree_root->root_item, tree_root->node);
	tree_root->commit_root = btrfs_root_node(tree_root);
2339 2340

	ret = find_and_setup_root(tree_root, fs_info,
C
Chris Mason 已提交
2341
				  BTRFS_EXTENT_TREE_OBJECTID, extent_root);
2342
	if (ret)
C
Chris Mason 已提交
2343
		goto recovery_tree_root;
2344 2345 2346 2347 2348
	extent_root->track_dirty = 1;

	ret = find_and_setup_root(tree_root, fs_info,
				  BTRFS_DEV_TREE_OBJECTID, dev_root);
	if (ret)
C
Chris Mason 已提交
2349
		goto recovery_tree_root;
2350
	dev_root->track_dirty = 1;
2351

2352 2353 2354
	ret = find_and_setup_root(tree_root, fs_info,
				  BTRFS_CSUM_TREE_OBJECTID, csum_root);
	if (ret)
C
Chris Mason 已提交
2355
		goto recovery_tree_root;
2356 2357 2358

	csum_root->track_dirty = 1;

2359 2360 2361
	fs_info->generation = generation;
	fs_info->last_trans_committed = generation;

2362 2363 2364 2365 2366 2367
	ret = btrfs_init_space_info(fs_info);
	if (ret) {
		printk(KERN_ERR "Failed to initial space info: %d\n", ret);
		goto fail_block_groups;
	}

2368 2369 2370 2371 2372
	ret = btrfs_read_block_groups(extent_root);
	if (ret) {
		printk(KERN_ERR "Failed to read block groups: %d\n", ret);
		goto fail_block_groups;
	}
C
Chris Mason 已提交
2373

2374 2375
	fs_info->cleaner_kthread = kthread_run(cleaner_kthread, tree_root,
					       "btrfs-cleaner");
2376
	if (IS_ERR(fs_info->cleaner_kthread))
2377
		goto fail_block_groups;
2378 2379 2380 2381

	fs_info->transaction_kthread = kthread_run(transaction_kthread,
						   tree_root,
						   "btrfs-transaction");
2382
	if (IS_ERR(fs_info->transaction_kthread))
2383
		goto fail_cleaner;
2384

C
Chris Mason 已提交
2385 2386 2387 2388 2389 2390 2391 2392
	if (!btrfs_test_opt(tree_root, SSD) &&
	    !btrfs_test_opt(tree_root, NOSSD) &&
	    !fs_info->fs_devices->rotating) {
		printk(KERN_INFO "Btrfs detected SSD devices, enabling SSD "
		       "mode\n");
		btrfs_set_opt(fs_info->mount_opt, SSD);
	}

2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
	if (btrfs_test_opt(tree_root, CHECK_INTEGRITY)) {
		ret = btrfsic_mount(tree_root, fs_devices,
				    btrfs_test_opt(tree_root,
					CHECK_INTEGRITY_INCLUDING_EXTENT_DATA) ?
				    1 : 0,
				    fs_info->check_integrity_print_mask);
		if (ret)
			printk(KERN_WARNING "btrfs: failed to initialize"
			       " integrity check module %s\n", sb->s_id);
	}
#endif

L
liubo 已提交
2406 2407 2408
	/* do not make disk changes in broken FS */
	if (btrfs_super_log_root(disk_super) != 0 &&
	    !(fs_info->fs_state & BTRFS_SUPER_FLAG_ERROR)) {
2409 2410
		u64 bytenr = btrfs_super_log_root(disk_super);

2411
		if (fs_devices->rw_devices == 0) {
C
Chris Mason 已提交
2412 2413
			printk(KERN_WARNING "Btrfs log replay required "
			       "on RO media\n");
2414 2415 2416
			err = -EIO;
			goto fail_trans_kthread;
		}
2417 2418 2419
		blocksize =
		     btrfs_level_size(tree_root,
				      btrfs_super_log_root_level(disk_super));
C
Chris Mason 已提交
2420

A
Al Viro 已提交
2421
		log_tree_root = btrfs_alloc_root(fs_info);
2422 2423 2424 2425
		if (!log_tree_root) {
			err = -ENOMEM;
			goto fail_trans_kthread;
		}
2426 2427 2428 2429 2430

		__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,
2431 2432
						      blocksize,
						      generation + 1);
2433 2434
		ret = btrfs_recover_log_trees(log_tree_root);
		BUG_ON(ret);
2435 2436 2437 2438 2439

		if (sb->s_flags & MS_RDONLY) {
			ret =  btrfs_commit_super(tree_root);
			BUG_ON(ret);
		}
2440
	}
Z
Zheng Yan 已提交
2441

2442 2443 2444
	ret = btrfs_find_orphan_roots(tree_root);
	BUG_ON(ret);

2445
	if (!(sb->s_flags & MS_RDONLY)) {
2446 2447 2448
		ret = btrfs_cleanup_fs_roots(fs_info);
		BUG_ON(ret);

2449
		ret = btrfs_recover_relocation(tree_root);
2450 2451 2452 2453 2454 2455
		if (ret < 0) {
			printk(KERN_WARNING
			       "btrfs: failed to recover relocation\n");
			err = -EINVAL;
			goto fail_trans_kthread;
		}
2456
	}
Z
Zheng Yan 已提交
2457

2458 2459 2460 2461 2462 2463
	location.objectid = BTRFS_FS_TREE_OBJECTID;
	location.type = BTRFS_ROOT_ITEM_KEY;
	location.offset = (u64)-1;

	fs_info->fs_root = btrfs_read_fs_root_no_name(fs_info, &location);
	if (!fs_info->fs_root)
2464
		goto fail_trans_kthread;
2465 2466 2467 2468
	if (IS_ERR(fs_info->fs_root)) {
		err = PTR_ERR(fs_info->fs_root);
		goto fail_trans_kthread;
	}
C
Chris Mason 已提交
2469

2470 2471
	if (!(sb->s_flags & MS_RDONLY)) {
		down_read(&fs_info->cleanup_work_sem);
2472 2473 2474
		err = btrfs_orphan_cleanup(fs_info->fs_root);
		if (!err)
			err = btrfs_orphan_cleanup(fs_info->tree_root);
2475
		up_read(&fs_info->cleanup_work_sem);
I
Ilya Dryomov 已提交
2476 2477 2478 2479

		if (!err)
			err = btrfs_recover_balance(fs_info->tree_root);

2480 2481
		if (err) {
			close_ctree(tree_root);
A
Al Viro 已提交
2482
			return err;
2483
		}
2484 2485
	}

A
Al Viro 已提交
2486
	return 0;
C
Chris Mason 已提交
2487

2488 2489
fail_trans_kthread:
	kthread_stop(fs_info->transaction_kthread);
2490
fail_cleaner:
2491
	kthread_stop(fs_info->cleaner_kthread);
2492 2493 2494 2495 2496 2497 2498 2499

	/*
	 * make sure we're done with the btree inode before we stop our
	 * kthreads
	 */
	filemap_write_and_wait(fs_info->btree_inode->i_mapping);
	invalidate_inode_pages2(fs_info->btree_inode->i_mapping);

2500 2501
fail_block_groups:
	btrfs_free_block_groups(fs_info);
C
Chris Mason 已提交
2502 2503 2504 2505

fail_tree_roots:
	free_root_pointers(fs_info, 1);

C
Chris Mason 已提交
2506
fail_sb_buffer:
2507
	btrfs_stop_workers(&fs_info->generic_worker);
2508
	btrfs_stop_workers(&fs_info->readahead_workers);
2509
	btrfs_stop_workers(&fs_info->fixup_workers);
2510
	btrfs_stop_workers(&fs_info->delalloc_workers);
2511 2512
	btrfs_stop_workers(&fs_info->workers);
	btrfs_stop_workers(&fs_info->endio_workers);
2513
	btrfs_stop_workers(&fs_info->endio_meta_workers);
2514
	btrfs_stop_workers(&fs_info->endio_meta_write_workers);
2515
	btrfs_stop_workers(&fs_info->endio_write_workers);
J
Josef Bacik 已提交
2516
	btrfs_stop_workers(&fs_info->endio_freespace_worker);
2517
	btrfs_stop_workers(&fs_info->submit_workers);
2518
	btrfs_stop_workers(&fs_info->delayed_workers);
2519
	btrfs_stop_workers(&fs_info->caching_workers);
2520
fail_alloc:
2521
fail_iput:
2522 2523
	btrfs_mapping_tree_free(&fs_info->mapping_tree);

2524
	invalidate_inode_pages2(fs_info->btree_inode->i_mapping);
2525
	iput(fs_info->btree_inode);
2526
fail_bdi:
2527
	bdi_destroy(&fs_info->bdi);
2528 2529
fail_srcu:
	cleanup_srcu_struct(&fs_info->subvol_srcu);
2530
fail:
2531
	btrfs_close_devices(fs_info->fs_devices);
A
Al Viro 已提交
2532
	return err;
C
Chris Mason 已提交
2533 2534 2535 2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550

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;
2551 2552
}

2553 2554 2555 2556 2557 2558 2559
static void btrfs_end_buffer_write_sync(struct buffer_head *bh, int uptodate)
{
	char b[BDEVNAME_SIZE];

	if (uptodate) {
		set_buffer_uptodate(bh);
	} else {
2560
		printk_ratelimited(KERN_WARNING "lost page write due to "
2561 2562
					"I/O error on %s\n",
				       bdevname(bh->b_bdev, b));
2563 2564 2565
		/* note, we dont' set_buffer_write_io_error because we have
		 * our own ways of dealing with the IO errors
		 */
2566 2567 2568 2569 2570 2571
		clear_buffer_uptodate(bh);
	}
	unlock_buffer(bh);
	put_bh(bh);
}

Y
Yan Zheng 已提交
2572 2573 2574 2575 2576 2577 2578 2579 2580 2581 2582 2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612
struct buffer_head *btrfs_read_dev_super(struct block_device *bdev)
{
	struct buffer_head *bh;
	struct buffer_head *latest = NULL;
	struct btrfs_super_block *super;
	int i;
	u64 transid = 0;
	u64 bytenr;

	/* we would like to check all the supers, but that would make
	 * a btrfs mount succeed after a mkfs from a different FS.
	 * So, we need to add a special mount option to scan for
	 * later supers, using BTRFS_SUPER_MIRROR_MAX instead
	 */
	for (i = 0; i < 1; i++) {
		bytenr = btrfs_sb_offset(i);
		if (bytenr + 4096 >= i_size_read(bdev->bd_inode))
			break;
		bh = __bread(bdev, bytenr / 4096, 4096);
		if (!bh)
			continue;

		super = (struct btrfs_super_block *)bh->b_data;
		if (btrfs_super_bytenr(super) != bytenr ||
		    strncmp((char *)(&super->magic), BTRFS_MAGIC,
			    sizeof(super->magic))) {
			brelse(bh);
			continue;
		}

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

2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623
/*
 * 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
2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646 2647
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);
			BUG_ON(!bh);
			wait_on_buffer(bh);
2648 2649 2650 2651 2652 2653 2654 2655 2656
			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
2657 2658 2659 2660 2661 2662 2663 2664 2665 2666
		} else {
			btrfs_set_super_bytenr(sb, bytenr);

			crc = ~(u32)0;
			crc = btrfs_csum_data(NULL, (char *)sb +
					      BTRFS_CSUM_SIZE, crc,
					      BTRFS_SUPER_INFO_SIZE -
					      BTRFS_CSUM_SIZE);
			btrfs_csum_final(crc, sb->csum);

2667 2668 2669 2670
			/*
			 * one reference for us, and we leave it for the
			 * caller
			 */
Y
Yan Zheng 已提交
2671 2672 2673 2674
			bh = __getblk(device->bdev, bytenr / 4096,
				      BTRFS_SUPER_INFO_SIZE);
			memcpy(bh->b_data, sb, BTRFS_SUPER_INFO_SIZE);

2675
			/* one reference for submit_bh */
Y
Yan Zheng 已提交
2676
			get_bh(bh);
2677 2678

			set_buffer_uptodate(bh);
Y
Yan Zheng 已提交
2679 2680 2681 2682
			lock_buffer(bh);
			bh->b_end_io = btrfs_end_buffer_write_sync;
		}

C
Chris Mason 已提交
2683 2684 2685 2686
		/*
		 * we fua the first super.  The others we allow
		 * to go down lazy.
		 */
2687
		ret = btrfsic_submit_bh(WRITE_FUA, bh);
2688
		if (ret)
Y
Yan Zheng 已提交
2689 2690 2691 2692 2693
			errors++;
	}
	return errors < i ? 0 : -1;
}

C
Chris Mason 已提交
2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763
/*
 * 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)) {
			printk("btrfs: disabling barriers on dev %s\n",
			       device->name);
			device->nobarriers = 1;
		}
		if (!bio_flagged(bio, BIO_UPTODATE)) {
			ret = -EIO;
		}

		/* 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
	 */
	device->flush_bio = NULL;;
	bio = bio_alloc(GFP_NOFS, 0);
	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);
2764
	btrfsic_submit_bio(WRITE_FLUSH, bio);
C
Chris Mason 已提交
2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812

	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;
	int errors = 0;
	int ret;

	/* send down all the barriers */
	head = &info->fs_devices->devices;
	list_for_each_entry_rcu(dev, head, dev_list) {
		if (!dev->bdev) {
			errors++;
			continue;
		}
		if (!dev->in_fs_metadata || !dev->writeable)
			continue;

		ret = write_dev_flush(dev, 0);
		if (ret)
			errors++;
	}

	/* wait for all the barriers */
	list_for_each_entry_rcu(dev, head, dev_list) {
		if (!dev->bdev) {
			errors++;
			continue;
		}
		if (!dev->in_fs_metadata || !dev->writeable)
			continue;

		ret = write_dev_flush(dev, 1);
		if (ret)
			errors++;
	}
	if (errors)
		return -EIO;
	return 0;
}

Y
Yan Zheng 已提交
2813
int write_all_supers(struct btrfs_root *root, int max_mirrors)
2814
{
2815
	struct list_head *head;
2816
	struct btrfs_device *dev;
2817
	struct btrfs_super_block *sb;
2818 2819 2820
	struct btrfs_dev_item *dev_item;
	int ret;
	int do_barriers;
2821 2822
	int max_errors;
	int total_errors = 0;
2823
	u64 flags;
2824

2825
	max_errors = btrfs_super_num_devices(root->fs_info->super_copy) - 1;
2826
	do_barriers = !btrfs_test_opt(root, NOBARRIER);
C
Chris Mason 已提交
2827
	backup_super_roots(root->fs_info);
2828

2829
	sb = root->fs_info->super_for_commit;
2830
	dev_item = &sb->dev_item;
2831

2832
	mutex_lock(&root->fs_info->fs_devices->device_list_mutex);
2833
	head = &root->fs_info->fs_devices->devices;
C
Chris Mason 已提交
2834 2835 2836 2837

	if (do_barriers)
		barrier_all_devices(root->fs_info);

2838
	list_for_each_entry_rcu(dev, head, dev_list) {
2839 2840 2841 2842
		if (!dev->bdev) {
			total_errors++;
			continue;
		}
Y
Yan Zheng 已提交
2843
		if (!dev->in_fs_metadata || !dev->writeable)
2844 2845
			continue;

Y
Yan Zheng 已提交
2846
		btrfs_set_stack_device_generation(dev_item, 0);
2847 2848 2849 2850 2851 2852 2853 2854
		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
2855
		memcpy(dev_item->fsid, dev->fs_devices->fsid, BTRFS_UUID_SIZE);
Y
Yan Zheng 已提交
2856

2857 2858 2859
		flags = btrfs_super_flags(sb);
		btrfs_set_super_flags(sb, flags | BTRFS_HEADER_FLAG_WRITTEN);

Y
Yan Zheng 已提交
2860
		ret = write_dev_supers(dev, sb, do_barriers, 0, max_mirrors);
2861 2862
		if (ret)
			total_errors++;
2863
	}
2864
	if (total_errors > max_errors) {
C
Chris Mason 已提交
2865 2866
		printk(KERN_ERR "btrfs: %d errors while writing supers\n",
		       total_errors);
2867 2868
		BUG();
	}
2869

Y
Yan Zheng 已提交
2870
	total_errors = 0;
2871
	list_for_each_entry_rcu(dev, head, dev_list) {
2872 2873
		if (!dev->bdev)
			continue;
Y
Yan Zheng 已提交
2874
		if (!dev->in_fs_metadata || !dev->writeable)
2875 2876
			continue;

Y
Yan Zheng 已提交
2877 2878 2879
		ret = write_dev_supers(dev, sb, do_barriers, 1, max_mirrors);
		if (ret)
			total_errors++;
2880
	}
2881
	mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);
2882
	if (total_errors > max_errors) {
C
Chris Mason 已提交
2883 2884
		printk(KERN_ERR "btrfs: %d errors while writing supers\n",
		       total_errors);
2885 2886
		BUG();
	}
2887 2888 2889
	return 0;
}

Y
Yan Zheng 已提交
2890 2891
int write_ctree_super(struct btrfs_trans_handle *trans,
		      struct btrfs_root *root, int max_mirrors)
2892
{
2893
	int ret;
2894

Y
Yan Zheng 已提交
2895
	ret = write_all_supers(root, max_mirrors);
2896
	return ret;
C
Chris Mason 已提交
2897 2898
}

2899
int btrfs_free_fs_root(struct btrfs_fs_info *fs_info, struct btrfs_root *root)
C
Chris Mason 已提交
2900
{
2901
	spin_lock(&fs_info->fs_roots_radix_lock);
C
Chris Mason 已提交
2902 2903
	radix_tree_delete(&fs_info->fs_roots_radix,
			  (unsigned long)root->root_key.objectid);
2904
	spin_unlock(&fs_info->fs_roots_radix_lock);
2905 2906 2907 2908

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

2909 2910
	__btrfs_remove_free_space_cache(root->free_ino_pinned);
	__btrfs_remove_free_space_cache(root->free_ino_ctl);
2911 2912 2913 2914 2915 2916
	free_fs_root(root);
	return 0;
}

static void free_fs_root(struct btrfs_root *root)
{
2917
	iput(root->cache_inode);
2918
	WARN_ON(!RB_EMPTY_ROOT(&root->inode_tree));
2919 2920
	if (root->anon_dev)
		free_anon_bdev(root->anon_dev);
2921 2922
	free_extent_buffer(root->node);
	free_extent_buffer(root->commit_root);
2923 2924
	kfree(root->free_ino_ctl);
	kfree(root->free_ino_pinned);
C
Chris Mason 已提交
2925
	kfree(root->name);
C
Chris Mason 已提交
2926 2927 2928
	kfree(root);
}

C
Chris Mason 已提交
2929
static int del_fs_roots(struct btrfs_fs_info *fs_info)
2930 2931 2932 2933 2934
{
	int ret;
	struct btrfs_root *gang[8];
	int i;

2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946 2947 2948
	while (!list_empty(&fs_info->dead_roots)) {
		gang[0] = list_entry(fs_info->dead_roots.next,
				     struct btrfs_root, root_list);
		list_del(&gang[0]->root_list);

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

C
Chris Mason 已提交
2949
	while (1) {
2950 2951 2952 2953 2954
		ret = radix_tree_gang_lookup(&fs_info->fs_roots_radix,
					     (void **)gang, 0,
					     ARRAY_SIZE(gang));
		if (!ret)
			break;
C
Chris Mason 已提交
2955
		for (i = 0; i < ret; i++)
2956
			btrfs_free_fs_root(fs_info, gang[i]);
2957 2958 2959
	}
	return 0;
}
2960

Y
Yan Zheng 已提交
2961
int btrfs_cleanup_fs_roots(struct btrfs_fs_info *fs_info)
C
Chris Mason 已提交
2962
{
Y
Yan Zheng 已提交
2963 2964 2965
	u64 root_objectid = 0;
	struct btrfs_root *gang[8];
	int i;
2966
	int ret;
2967

Y
Yan Zheng 已提交
2968 2969 2970 2971 2972 2973
	while (1) {
		ret = radix_tree_gang_lookup(&fs_info->fs_roots_radix,
					     (void **)gang, root_objectid,
					     ARRAY_SIZE(gang));
		if (!ret)
			break;
2974 2975

		root_objectid = gang[ret - 1]->root_key.objectid + 1;
Y
Yan Zheng 已提交
2976
		for (i = 0; i < ret; i++) {
2977 2978
			int err;

Y
Yan Zheng 已提交
2979
			root_objectid = gang[i]->root_key.objectid;
2980 2981 2982
			err = btrfs_orphan_cleanup(gang[i]);
			if (err)
				return err;
Y
Yan Zheng 已提交
2983 2984 2985 2986 2987
		}
		root_objectid++;
	}
	return 0;
}
2988

Y
Yan Zheng 已提交
2989 2990 2991 2992
int btrfs_commit_super(struct btrfs_root *root)
{
	struct btrfs_trans_handle *trans;
	int ret;
2993

Y
Yan Zheng 已提交
2994
	mutex_lock(&root->fs_info->cleaner_mutex);
Y
Yan, Zheng 已提交
2995
	btrfs_run_delayed_iputs(root);
2996
	btrfs_clean_old_snapshots(root);
Y
Yan Zheng 已提交
2997
	mutex_unlock(&root->fs_info->cleaner_mutex);
2998 2999 3000 3001 3002

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

3003
	trans = btrfs_join_transaction(root);
3004 3005
	if (IS_ERR(trans))
		return PTR_ERR(trans);
3006
	ret = btrfs_commit_transaction(trans, root);
Y
Yan Zheng 已提交
3007 3008
	BUG_ON(ret);
	/* run commit again to drop the original snapshot */
3009
	trans = btrfs_join_transaction(root);
3010 3011
	if (IS_ERR(trans))
		return PTR_ERR(trans);
C
Chris Mason 已提交
3012 3013
	btrfs_commit_transaction(trans, root);
	ret = btrfs_write_and_wait_transaction(NULL, root);
3014
	BUG_ON(ret);
3015

Y
Yan Zheng 已提交
3016
	ret = write_ctree_super(NULL, root, 0);
Y
Yan Zheng 已提交
3017 3018 3019 3020 3021 3022 3023 3024 3025 3026 3027
	return ret;
}

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

	fs_info->closing = 1;
	smp_mb();

3028 3029 3030
	/* pause restriper - we want to resume on mount */
	btrfs_pause_balance(root->fs_info);

A
Arne Jansen 已提交
3031
	btrfs_scrub_cancel(root);
C
Chris Mason 已提交
3032 3033 3034 3035 3036 3037

	/* 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 */
A
Al Viro 已提交
3038
	btrfs_run_defrag_inodes(fs_info);
C
Chris Mason 已提交
3039

L
liubo 已提交
3040 3041 3042 3043 3044 3045 3046 3047 3048
	/*
	 * Here come 2 situations when btrfs is broken to flip readonly:
	 *
	 * 1. when btrfs flips readonly somewhere else before
	 * btrfs_commit_super, sb->s_flags has MS_RDONLY flag,
	 * and btrfs will skip to write sb directly to keep
	 * ERROR state on disk.
	 *
	 * 2. when btrfs flips readonly just in btrfs_commit_super,
3049
	 * and in such case, btrfs cannot write sb via btrfs_commit_super,
L
liubo 已提交
3050 3051 3052
	 * and since fs_state has been set BTRFS_SUPER_FLAG_ERROR flag,
	 * btrfs will cleanup all FS resources first and write sb then.
	 */
Y
Yan Zheng 已提交
3053
	if (!(fs_info->sb->s_flags & MS_RDONLY)) {
L
liubo 已提交
3054 3055 3056 3057 3058 3059 3060
		ret = btrfs_commit_super(root);
		if (ret)
			printk(KERN_ERR "btrfs: commit super ret %d\n", ret);
	}

	if (fs_info->fs_state & BTRFS_SUPER_FLAG_ERROR) {
		ret = btrfs_error_commit_super(root);
C
Chris Mason 已提交
3061 3062
		if (ret)
			printk(KERN_ERR "btrfs: commit super ret %d\n", ret);
Y
Yan Zheng 已提交
3063
	}
3064

3065 3066
	btrfs_put_block_group_cache(fs_info);

A
Al Viro 已提交
3067 3068
	kthread_stop(fs_info->transaction_kthread);
	kthread_stop(fs_info->cleaner_kthread);
3069

3070 3071 3072
	fs_info->closing = 2;
	smp_mb();

C
Chris Mason 已提交
3073
	if (fs_info->delalloc_bytes) {
C
Chris Mason 已提交
3074
		printk(KERN_INFO "btrfs: at unmount delalloc count %llu\n",
3075
		       (unsigned long long)fs_info->delalloc_bytes);
C
Chris Mason 已提交
3076
	}
Y
Yan Zheng 已提交
3077
	if (fs_info->total_ref_cache_size) {
C
Chris Mason 已提交
3078 3079
		printk(KERN_INFO "btrfs: at umount reference cache size %llu\n",
		       (unsigned long long)fs_info->total_ref_cache_size);
Y
Yan Zheng 已提交
3080
	}
3081

3082 3083 3084 3085
	free_extent_buffer(fs_info->extent_root->node);
	free_extent_buffer(fs_info->extent_root->commit_root);
	free_extent_buffer(fs_info->tree_root->node);
	free_extent_buffer(fs_info->tree_root->commit_root);
A
Al Viro 已提交
3086 3087 3088 3089 3090 3091
	free_extent_buffer(fs_info->chunk_root->node);
	free_extent_buffer(fs_info->chunk_root->commit_root);
	free_extent_buffer(fs_info->dev_root->node);
	free_extent_buffer(fs_info->dev_root->commit_root);
	free_extent_buffer(fs_info->csum_root->node);
	free_extent_buffer(fs_info->csum_root->commit_root);
3092

A
Al Viro 已提交
3093
	btrfs_free_block_groups(fs_info);
3094

Y
Yan Zheng 已提交
3095
	del_fs_roots(fs_info);
3096

Y
Yan Zheng 已提交
3097
	iput(fs_info->btree_inode);
3098

3099
	btrfs_stop_workers(&fs_info->generic_worker);
3100
	btrfs_stop_workers(&fs_info->fixup_workers);
3101
	btrfs_stop_workers(&fs_info->delalloc_workers);
3102 3103
	btrfs_stop_workers(&fs_info->workers);
	btrfs_stop_workers(&fs_info->endio_workers);
3104
	btrfs_stop_workers(&fs_info->endio_meta_workers);
3105
	btrfs_stop_workers(&fs_info->endio_meta_write_workers);
3106
	btrfs_stop_workers(&fs_info->endio_write_workers);
J
Josef Bacik 已提交
3107
	btrfs_stop_workers(&fs_info->endio_freespace_worker);
3108
	btrfs_stop_workers(&fs_info->submit_workers);
3109
	btrfs_stop_workers(&fs_info->delayed_workers);
3110
	btrfs_stop_workers(&fs_info->caching_workers);
3111
	btrfs_stop_workers(&fs_info->readahead_workers);
3112

3113 3114 3115 3116 3117
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
	if (btrfs_test_opt(root, CHECK_INTEGRITY))
		btrfsic_unmount(root, fs_info->fs_devices);
#endif

3118
	btrfs_close_devices(fs_info->fs_devices);
3119
	btrfs_mapping_tree_free(&fs_info->mapping_tree);
3120

C
Chris Mason 已提交
3121
	bdi_destroy(&fs_info->bdi);
3122
	cleanup_srcu_struct(&fs_info->subvol_srcu);
3123

3124 3125 3126
	return 0;
}

3127
int btrfs_buffer_uptodate(struct extent_buffer *buf, u64 parent_transid)
3128
{
3129
	int ret;
3130
	struct inode *btree_inode = buf->first_page->mapping->host;
3131

3132 3133
	ret = extent_buffer_uptodate(&BTRFS_I(btree_inode)->io_tree, buf,
				     NULL);
3134 3135 3136 3137 3138 3139
	if (!ret)
		return ret;

	ret = verify_parent_transid(&BTRFS_I(btree_inode)->io_tree, buf,
				    parent_transid);
	return !ret;
3140 3141 3142
}

int btrfs_set_buffer_uptodate(struct extent_buffer *buf)
C
Chris Mason 已提交
3143
{
3144
	struct inode *btree_inode = buf->first_page->mapping->host;
3145
	return set_extent_buffer_uptodate(&BTRFS_I(btree_inode)->io_tree,
3146 3147
					  buf);
}
3148

3149 3150
void btrfs_mark_buffer_dirty(struct extent_buffer *buf)
{
3151
	struct btrfs_root *root = BTRFS_I(buf->first_page->mapping->host)->root;
3152 3153
	u64 transid = btrfs_header_generation(buf);
	struct inode *btree_inode = root->fs_info->btree_inode;
3154
	int was_dirty;
3155

3156
	btrfs_assert_tree_locked(buf);
C
Chris Mason 已提交
3157
	if (transid != root->fs_info->generation) {
C
Chris Mason 已提交
3158 3159
		printk(KERN_CRIT "btrfs transid mismatch buffer %llu, "
		       "found %llu running %llu\n",
3160
			(unsigned long long)buf->start,
C
Chris Mason 已提交
3161 3162
			(unsigned long long)transid,
			(unsigned long long)root->fs_info->generation);
C
Chris Mason 已提交
3163 3164
		WARN_ON(1);
	}
3165 3166 3167 3168 3169 3170 3171
	was_dirty = set_extent_buffer_dirty(&BTRFS_I(btree_inode)->io_tree,
					    buf);
	if (!was_dirty) {
		spin_lock(&root->fs_info->delalloc_lock);
		root->fs_info->dirty_metadata_bytes += buf->len;
		spin_unlock(&root->fs_info->delalloc_lock);
	}
3172 3173
}

3174
void btrfs_btree_balance_dirty(struct btrfs_root *root, unsigned long nr)
3175 3176 3177 3178 3179 3180 3181 3182 3183 3184 3185 3186 3187 3188 3189 3190 3191 3192 3193 3194 3195 3196 3197
{
	/*
	 * looks as though older kernels can get into trouble with
	 * this code, they end up stuck in balance_dirty_pages forever
	 */
	u64 num_dirty;
	unsigned long thresh = 32 * 1024 * 1024;

	if (current->flags & PF_MEMALLOC)
		return;

	btrfs_balance_delayed_items(root);

	num_dirty = root->fs_info->dirty_metadata_bytes;

	if (num_dirty > thresh) {
		balance_dirty_pages_ratelimited_nr(
				   root->fs_info->btree_inode->i_mapping, 1);
	}
	return;
}

void __btrfs_btree_balance_dirty(struct btrfs_root *root, unsigned long nr)
C
Chris Mason 已提交
3198
{
3199 3200 3201 3202
	/*
	 * looks as though older kernels can get into trouble with
	 * this code, they end up stuck in balance_dirty_pages forever
	 */
3203
	u64 num_dirty;
3204
	unsigned long thresh = 32 * 1024 * 1024;
3205

3206
	if (current->flags & PF_MEMALLOC)
3207 3208
		return;

3209 3210
	num_dirty = root->fs_info->dirty_metadata_bytes;

3211 3212
	if (num_dirty > thresh) {
		balance_dirty_pages_ratelimited_nr(
C
Chris Mason 已提交
3213
				   root->fs_info->btree_inode->i_mapping, 1);
3214
	}
3215
	return;
C
Chris Mason 已提交
3216
}
3217

3218
int btrfs_read_buffer(struct extent_buffer *buf, u64 parent_transid)
3219
{
3220
	struct btrfs_root *root = BTRFS_I(buf->first_page->mapping->host)->root;
3221
	int ret;
3222
	ret = btree_read_extent_buffer_pages(root, buf, 0, parent_transid);
C
Chris Mason 已提交
3223
	if (ret == 0)
3224
		set_bit(EXTENT_BUFFER_UPTODATE, &buf->bflags);
3225
	return ret;
3226
}
3227

3228 3229
static int btree_lock_page_hook(struct page *page, void *data,
				void (*flush_fn)(void *))
C
Chris Mason 已提交
3230 3231
{
	struct inode *inode = page->mapping->host;
3232
	struct btrfs_root *root = BTRFS_I(inode)->root;
C
Chris Mason 已提交
3233 3234 3235 3236 3237 3238 3239 3240 3241
	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
	struct extent_buffer *eb;
	unsigned long len;
	u64 bytenr = page_offset(page);

	if (page->private == EXTENT_PAGE_PRIVATE)
		goto out;

	len = page->private >> 2;
3242
	eb = find_extent_buffer(io_tree, bytenr, len);
C
Chris Mason 已提交
3243 3244 3245
	if (!eb)
		goto out;

3246 3247 3248 3249
	if (!btrfs_try_tree_write_lock(eb)) {
		flush_fn(data);
		btrfs_tree_lock(eb);
	}
C
Chris Mason 已提交
3250
	btrfs_set_header_flag(eb, BTRFS_HEADER_FLAG_WRITTEN);
3251 3252 3253 3254 3255 3256 3257 3258 3259 3260

	if (test_and_clear_bit(EXTENT_BUFFER_DIRTY, &eb->bflags)) {
		spin_lock(&root->fs_info->delalloc_lock);
		if (root->fs_info->dirty_metadata_bytes >= eb->len)
			root->fs_info->dirty_metadata_bytes -= eb->len;
		else
			WARN_ON(1);
		spin_unlock(&root->fs_info->delalloc_lock);
	}

C
Chris Mason 已提交
3261 3262 3263
	btrfs_tree_unlock(eb);
	free_extent_buffer(eb);
out:
3264 3265 3266 3267
	if (!trylock_page(page)) {
		flush_fn(data);
		lock_page(page);
	}
C
Chris Mason 已提交
3268 3269 3270
	return 0;
}

L
liubo 已提交
3271 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 3332 3333 3334 3335 3336 3337 3338 3339 3340 3341 3342 3343 3344 3345 3346 3347 3348 3349 3350 3351 3352 3353 3354 3355 3356 3357 3358 3359 3360 3361 3362 3363 3364 3365 3366 3367 3368 3369 3370 3371 3372 3373 3374
static void btrfs_check_super_valid(struct btrfs_fs_info *fs_info,
			      int read_only)
{
	if (read_only)
		return;

	if (fs_info->fs_state & BTRFS_SUPER_FLAG_ERROR)
		printk(KERN_WARNING "warning: mount fs with errors, "
		       "running btrfsck is recommended\n");
}

int btrfs_error_commit_super(struct btrfs_root *root)
{
	int ret;

	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);

	ret = write_ctree_super(NULL, root, 0);

	return ret;
}

static int btrfs_destroy_ordered_operations(struct btrfs_root *root)
{
	struct btrfs_inode *btrfs_inode;
	struct list_head splice;

	INIT_LIST_HEAD(&splice);

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

	list_splice_init(&root->fs_info->ordered_operations, &splice);
	while (!list_empty(&splice)) {
		btrfs_inode = list_entry(splice.next, struct btrfs_inode,
					 ordered_operations);

		list_del_init(&btrfs_inode->ordered_operations);

		btrfs_invalidate_inodes(btrfs_inode->root);
	}

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

	return 0;
}

static int btrfs_destroy_ordered_extents(struct btrfs_root *root)
{
	struct list_head splice;
	struct btrfs_ordered_extent *ordered;
	struct inode *inode;

	INIT_LIST_HEAD(&splice);

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

	list_splice_init(&root->fs_info->ordered_extents, &splice);
	while (!list_empty(&splice)) {
		ordered = list_entry(splice.next, struct btrfs_ordered_extent,
				     root_extent_list);

		list_del_init(&ordered->root_extent_list);
		atomic_inc(&ordered->refs);

		/* the inode may be getting freed (in sys_unlink path). */
		inode = igrab(ordered->inode);

		spin_unlock(&root->fs_info->ordered_extent_lock);
		if (inode)
			iput(inode);

		atomic_set(&ordered->refs, 1);
		btrfs_put_ordered_extent(ordered);

		spin_lock(&root->fs_info->ordered_extent_lock);
	}

	spin_unlock(&root->fs_info->ordered_extent_lock);

	return 0;
}

static int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans,
				      struct btrfs_root *root)
{
	struct rb_node *node;
	struct btrfs_delayed_ref_root *delayed_refs;
	struct btrfs_delayed_ref_node *ref;
	int ret = 0;

	delayed_refs = &trans->delayed_refs;

	spin_lock(&delayed_refs->lock);
	if (delayed_refs->num_entries == 0) {
3375
		spin_unlock(&delayed_refs->lock);
L
liubo 已提交
3376 3377 3378 3379 3380 3381 3382 3383 3384 3385 3386 3387 3388 3389 3390 3391 3392 3393 3394 3395 3396 3397 3398 3399 3400 3401 3402 3403 3404 3405 3406 3407 3408 3409 3410 3411 3412 3413 3414 3415 3416 3417 3418 3419 3420 3421 3422 3423 3424 3425 3426 3427 3428 3429 3430 3431 3432 3433 3434 3435 3436 3437 3438 3439 3440 3441 3442 3443 3444
		printk(KERN_INFO "delayed_refs has NO entry\n");
		return ret;
	}

	node = rb_first(&delayed_refs->root);
	while (node) {
		ref = rb_entry(node, struct btrfs_delayed_ref_node, rb_node);
		node = rb_next(node);

		ref->in_tree = 0;
		rb_erase(&ref->rb_node, &delayed_refs->root);
		delayed_refs->num_entries--;

		atomic_set(&ref->refs, 1);
		if (btrfs_delayed_ref_is_head(ref)) {
			struct btrfs_delayed_ref_head *head;

			head = btrfs_delayed_node_to_head(ref);
			mutex_lock(&head->mutex);
			kfree(head->extent_op);
			delayed_refs->num_heads--;
			if (list_empty(&head->cluster))
				delayed_refs->num_heads_ready--;
			list_del_init(&head->cluster);
			mutex_unlock(&head->mutex);
		}

		spin_unlock(&delayed_refs->lock);
		btrfs_put_delayed_ref(ref);

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

	spin_unlock(&delayed_refs->lock);

	return ret;
}

static int btrfs_destroy_pending_snapshots(struct btrfs_transaction *t)
{
	struct btrfs_pending_snapshot *snapshot;
	struct list_head splice;

	INIT_LIST_HEAD(&splice);

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

	while (!list_empty(&splice)) {
		snapshot = list_entry(splice.next,
				      struct btrfs_pending_snapshot,
				      list);

		list_del_init(&snapshot->list);

		kfree(snapshot);
	}

	return 0;
}

static int btrfs_destroy_delalloc_inodes(struct btrfs_root *root)
{
	struct btrfs_inode *btrfs_inode;
	struct list_head splice;

	INIT_LIST_HEAD(&splice);

	spin_lock(&root->fs_info->delalloc_lock);
3445
	list_splice_init(&root->fs_info->delalloc_inodes, &splice);
L
liubo 已提交
3446 3447 3448 3449 3450 3451 3452 3453 3454 3455 3456 3457 3458 3459 3460 3461 3462 3463 3464 3465 3466 3467 3468 3469 3470 3471 3472 3473 3474 3475 3476 3477 3478 3479 3480 3481 3482 3483 3484 3485 3486 3487 3488 3489 3490 3491 3492 3493 3494 3495 3496 3497 3498 3499 3500 3501 3502 3503 3504 3505 3506 3507 3508 3509 3510 3511 3512 3513 3514 3515 3516 3517 3518 3519 3520 3521 3522 3523 3524 3525 3526 3527 3528 3529 3530 3531 3532 3533 3534 3535

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

		list_del_init(&btrfs_inode->delalloc_inodes);

		btrfs_invalidate_inodes(btrfs_inode->root);
	}

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

	return 0;
}

static int btrfs_destroy_marked_extents(struct btrfs_root *root,
					struct extent_io_tree *dirty_pages,
					int mark)
{
	int ret;
	struct page *page;
	struct inode *btree_inode = root->fs_info->btree_inode;
	struct extent_buffer *eb;
	u64 start = 0;
	u64 end;
	u64 offset;
	unsigned long index;

	while (1) {
		ret = find_first_extent_bit(dirty_pages, start, &start, &end,
					    mark);
		if (ret)
			break;

		clear_extent_bits(dirty_pages, start, end, mark, GFP_NOFS);
		while (start <= end) {
			index = start >> PAGE_CACHE_SHIFT;
			start = (u64)(index + 1) << PAGE_CACHE_SHIFT;
			page = find_get_page(btree_inode->i_mapping, index);
			if (!page)
				continue;
			offset = page_offset(page);

			spin_lock(&dirty_pages->buffer_lock);
			eb = radix_tree_lookup(
			     &(&BTRFS_I(page->mapping->host)->io_tree)->buffer,
					       offset >> PAGE_CACHE_SHIFT);
			spin_unlock(&dirty_pages->buffer_lock);
			if (eb) {
				ret = test_and_clear_bit(EXTENT_BUFFER_DIRTY,
							 &eb->bflags);
				atomic_set(&eb->refs, 1);
			}
			if (PageWriteback(page))
				end_page_writeback(page);

			lock_page(page);
			if (PageDirty(page)) {
				clear_page_dirty_for_io(page);
				spin_lock_irq(&page->mapping->tree_lock);
				radix_tree_tag_clear(&page->mapping->page_tree,
							page_index(page),
							PAGECACHE_TAG_DIRTY);
				spin_unlock_irq(&page->mapping->tree_lock);
			}

			page->mapping->a_ops->invalidatepage(page, 0);
			unlock_page(page);
		}
	}

	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;

	unpin = pinned_extents;
	while (1) {
		ret = find_first_extent_bit(unpin, 0, &start, &end,
					    EXTENT_DIRTY);
		if (ret)
			break;

		/* opt_discard */
3536 3537 3538 3539
		if (btrfs_test_opt(root, DISCARD))
			ret = btrfs_error_discard_extent(root, start,
							 end + 1 - start,
							 NULL);
L
liubo 已提交
3540 3541 3542 3543 3544 3545 3546 3547 3548 3549 3550 3551 3552 3553 3554 3555 3556 3557

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

	return 0;
}

static int btrfs_cleanup_transaction(struct btrfs_root *root)
{
	struct btrfs_transaction *t;
	LIST_HEAD(list);

	WARN_ON(1);

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

J
Josef Bacik 已提交
3558
	spin_lock(&root->fs_info->trans_lock);
L
liubo 已提交
3559
	list_splice_init(&root->fs_info->trans_list, &list);
J
Josef Bacik 已提交
3560 3561 3562
	root->fs_info->trans_no_join = 1;
	spin_unlock(&root->fs_info->trans_lock);

L
liubo 已提交
3563 3564 3565 3566 3567 3568 3569 3570 3571 3572 3573 3574 3575 3576 3577 3578 3579 3580 3581 3582 3583 3584 3585 3586 3587 3588 3589 3590 3591 3592 3593 3594 3595
	while (!list_empty(&list)) {
		t = list_entry(list.next, struct btrfs_transaction, list);
		if (!t)
			break;

		btrfs_destroy_ordered_operations(root);

		btrfs_destroy_ordered_extents(root);

		btrfs_destroy_delayed_refs(t, root);

		btrfs_block_rsv_release(root,
					&root->fs_info->trans_block_rsv,
					t->dirty_pages.dirty_bytes);

		/* FIXME: cleanup wait for commit */
		t->in_commit = 1;
		t->blocked = 1;
		if (waitqueue_active(&root->fs_info->transaction_blocked_wait))
			wake_up(&root->fs_info->transaction_blocked_wait);

		t->blocked = 0;
		if (waitqueue_active(&root->fs_info->transaction_wait))
			wake_up(&root->fs_info->transaction_wait);

		t->commit_done = 1;
		if (waitqueue_active(&t->commit_wait))
			wake_up(&t->commit_wait);

		btrfs_destroy_pending_snapshots(t);

		btrfs_destroy_delalloc_inodes(root);

J
Josef Bacik 已提交
3596
		spin_lock(&root->fs_info->trans_lock);
L
liubo 已提交
3597
		root->fs_info->running_transaction = NULL;
J
Josef Bacik 已提交
3598
		spin_unlock(&root->fs_info->trans_lock);
L
liubo 已提交
3599 3600 3601 3602 3603 3604 3605

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

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

3606
		atomic_set(&t->use_count, 0);
L
liubo 已提交
3607 3608 3609 3610 3611
		list_del_init(&t->list);
		memset(t, 0, sizeof(*t));
		kmem_cache_free(btrfs_transaction_cachep, t);
	}

J
Josef Bacik 已提交
3612 3613 3614
	spin_lock(&root->fs_info->trans_lock);
	root->fs_info->trans_no_join = 0;
	spin_unlock(&root->fs_info->trans_lock);
L
liubo 已提交
3615 3616 3617 3618 3619
	mutex_unlock(&root->fs_info->transaction_kthread_mutex);

	return 0;
}

3620
static struct extent_io_ops btree_extent_io_ops = {
C
Chris Mason 已提交
3621
	.write_cache_pages_lock_hook = btree_lock_page_hook,
3622
	.readpage_end_io_hook = btree_readpage_end_io_hook,
A
Arne Jansen 已提交
3623
	.readpage_io_failed_hook = btree_io_failed_hook,
3624
	.submit_bio_hook = btree_submit_bio_hook,
3625 3626
	/* note we're sharing with inode.c for the merge bio hook */
	.merge_bio_hook = btrfs_merge_bio_hook,
3627
};