ctree.h 80 KB
Newer Older
Chris Mason's avatar
Chris Mason committed
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.
 */

Chris Mason's avatar
Chris Mason committed
19
20
#ifndef __BTRFS_CTREE__
#define __BTRFS_CTREE__
21

22
#include <linux/version.h>
23
24
#include <linux/mm.h>
#include <linux/highmem.h>
Chris Mason's avatar
Chris Mason committed
25
#include <linux/fs.h>
26
#include <linux/completion.h>
Chris Mason's avatar
Chris Mason committed
27
#include <linux/backing-dev.h>
28
#include <linux/wait.h>
29
#include <linux/slab.h>
30
#include <asm/kmap_types.h>
31
#include "extent_io.h"
32
#include "extent_map.h"
33
#include "async-thread.h"
Chris Mason's avatar
Chris Mason committed
34

35
struct btrfs_trans_handle;
Chris Mason's avatar
Chris Mason committed
36
struct btrfs_transaction;
37
struct btrfs_pending_snapshot;
38
39
40
extern struct kmem_cache *btrfs_trans_handle_cachep;
extern struct kmem_cache *btrfs_transaction_cachep;
extern struct kmem_cache *btrfs_bit_radix_cachep;
Chris Mason's avatar
Chris Mason committed
41
extern struct kmem_cache *btrfs_path_cachep;
42
struct btrfs_ordered_sum;
43

44
#define BTRFS_MAGIC "_BHRfS_M"
45

46
#define BTRFS_MAX_LEVEL 8
47

48
49
#define BTRFS_COMPAT_EXTENT_TREE_V0

50
51
52
53
54
55
56
/*
 * files bigger than this get some pre-flushing when they are added
 * to the ordered operations list.  That way we limit the total
 * work done by the commit
 */
#define BTRFS_ORDERED_OPERATIONS_FLUSH_LIMIT (8 * 1024 * 1024)

57
/* holds pointers to all of the tree roots */
58
#define BTRFS_ROOT_TREE_OBJECTID 1ULL
59
60

/* stores information about which extents are in use, and reference counts */
Chris Mason's avatar
Chris Mason committed
61
#define BTRFS_EXTENT_TREE_OBJECTID 2ULL
62
63
64
65
66

/*
 * chunk tree stores translations from logical -> physical block numbering
 * the super block points to the chunk tree
 */
67
#define BTRFS_CHUNK_TREE_OBJECTID 3ULL
68
69
70
71
72

/*
 * stores information about which areas of a given device are in use.
 * one per device.  The tree of tree roots points to the device tree
 */
73
74
75
76
77
78
79
#define BTRFS_DEV_TREE_OBJECTID 4ULL

/* one per subvolume, storing files and directories */
#define BTRFS_FS_TREE_OBJECTID 5ULL

/* directory objectid inside the root tree */
#define BTRFS_ROOT_TREE_DIR_OBJECTID 6ULL
80

81
82
83
/* holds checksums of all the data extents */
#define BTRFS_CSUM_TREE_OBJECTID 7ULL

84
85
86
/* orhpan objectid for tracking unlinked/truncated files */
#define BTRFS_ORPHAN_OBJECTID -5ULL

87
88
89
90
/* does write ahead logging to speed up fsyncs */
#define BTRFS_TREE_LOG_OBJECTID -6ULL
#define BTRFS_TREE_LOG_FIXUP_OBJECTID -7ULL

Zheng Yan's avatar
Zheng Yan committed
91
92
93
94
/* for space balancing */
#define BTRFS_TREE_RELOC_OBJECTID -8ULL
#define BTRFS_DATA_RELOC_TREE_OBJECTID -9ULL

95
96
97
98
99
100
101
/*
 * extent checksums all have this objectid
 * this allows them to share the logging tree
 * for fsyncs
 */
#define BTRFS_EXTENT_CSUM_OBJECTID -10ULL

Zheng Yan's avatar
Zheng Yan committed
102
103
104
/* dummy objectid represents multiple objectids */
#define BTRFS_MULTIPLE_OBJECTIDS -255ULL

105
/*
106
 * All files have objectids in this range.
107
 */
108
#define BTRFS_FIRST_FREE_OBJECTID 256ULL
109
#define BTRFS_LAST_FREE_OBJECTID -256ULL
110
#define BTRFS_FIRST_CHUNK_TREE_OBJECTID 256ULL
111

112
113
114
115
116
117
118

/*
 * the device items go into the chunk tree.  The key is in the form
 * [ 1 BTRFS_DEV_ITEM_KEY device_id ]
 */
#define BTRFS_DEV_ITEMS_OBJECTID 1ULL

119
120
121
122
#define BTRFS_BTREE_INODE_OBJECTID 1

#define BTRFS_EMPTY_SUBVOL_DIR_OBJECTID 2

Chris Mason's avatar
Chris Mason committed
123
124
125
126
127
128
/*
 * we can actually store much bigger names, but lets not confuse the rest
 * of linux
 */
#define BTRFS_NAME_LEN 255

Chris Mason's avatar
Chris Mason committed
129
130
/* 32 bytes in various csum fields */
#define BTRFS_CSUM_SIZE 32
131
132
133
134
135
136

/* csum types */
#define BTRFS_CSUM_TYPE_CRC32	0

static int btrfs_csum_sizes[] = { 4, 0 };

137
/* four bytes for CRC32 */
138
#define BTRFS_EMPTY_DIR_SIZE 0
Chris Mason's avatar
Chris Mason committed
139

Chris Mason's avatar
Chris Mason committed
140
141
142
143
144
145
146
147
#define BTRFS_FT_UNKNOWN	0
#define BTRFS_FT_REG_FILE	1
#define BTRFS_FT_DIR		2
#define BTRFS_FT_CHRDEV		3
#define BTRFS_FT_BLKDEV		4
#define BTRFS_FT_FIFO		5
#define BTRFS_FT_SOCK		6
#define BTRFS_FT_SYMLINK	7
Josef Bacik's avatar
Josef Bacik committed
148
149
#define BTRFS_FT_XATTR		8
#define BTRFS_FT_MAX		9
Chris Mason's avatar
Chris Mason committed
150

151
/*
Wu Fengguang's avatar
Wu Fengguang committed
152
153
154
155
156
157
158
159
160
 * The key defines the order in the tree, and so it also defines (optimal)
 * block layout.
 *
 * objectid corresponds to the inode number.
 *
 * type tells us things about the object, and is a kind of stream selector.
 * so for a given inode, keys with type of 1 might refer to the inode data,
 * type of 2 may point to file data in the btree and type == 3 may point to
 * extents.
161
162
 *
 * offset is the starting byte offset for this key in the stream.
Chris Mason's avatar
Chris Mason committed
163
164
165
166
 *
 * btrfs_disk_key is in disk byte order.  struct btrfs_key is always
 * in cpu native order.  Otherwise they are identical and their sizes
 * should be the same (ie both packed)
167
 */
Chris Mason's avatar
Chris Mason committed
168
169
struct btrfs_disk_key {
	__le64 objectid;
170
	u8 type;
171
	__le64 offset;
Chris Mason's avatar
Chris Mason committed
172
173
174
} __attribute__ ((__packed__));

struct btrfs_key {
175
	u64 objectid;
176
	u8 type;
177
	u64 offset;
178
179
} __attribute__ ((__packed__));

180
181
182
183
struct btrfs_mapping_tree {
	struct extent_map_tree map_tree;
};

184
#define BTRFS_UUID_SIZE 16
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
struct btrfs_dev_item {
	/* the internal btrfs device id */
	__le64 devid;

	/* size of the device */
	__le64 total_bytes;

	/* bytes used */
	__le64 bytes_used;

	/* optimal io alignment for this device */
	__le32 io_align;

	/* optimal io width for this device */
	__le32 io_width;

	/* minimal io size for this device */
	__le32 sector_size;

	/* type and info about this device */
	__le64 type;

Yan Zheng's avatar
Yan Zheng committed
207
208
209
	/* expected generation for this device */
	__le64 generation;

210
211
	/*
	 * starting byte of this partition on the device,
Wu Fengguang's avatar
Wu Fengguang committed
212
	 * to allow for stripe alignment in the future
213
214
215
	 */
	__le64 start_offset;

216
217
218
219
220
221
222
223
224
	/* grouping information for allocation decisions */
	__le32 dev_group;

	/* seek speed 0-100 where 100 is fastest */
	u8 seek_speed;

	/* bandwidth 0-100 where 100 is fastest */
	u8 bandwidth;

225
	/* btrfs generated uuid for this device */
226
	u8 uuid[BTRFS_UUID_SIZE];
Yan Zheng's avatar
Yan Zheng committed
227
228
229

	/* uuid of FS who owns this device */
	u8 fsid[BTRFS_UUID_SIZE];
230
231
232
233
234
} __attribute__ ((__packed__));

struct btrfs_stripe {
	__le64 devid;
	__le64 offset;
235
	u8 dev_uuid[BTRFS_UUID_SIZE];
236
237
238
} __attribute__ ((__packed__));

struct btrfs_chunk {
239
240
241
242
	/* size of this chunk in bytes */
	__le64 length;

	/* objectid of the root referencing this chunk */
243
	__le64 owner;
244

245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
	__le64 stripe_len;
	__le64 type;

	/* optimal io alignment for this chunk */
	__le32 io_align;

	/* optimal io width for this chunk */
	__le32 io_width;

	/* minimal io size for this chunk */
	__le32 sector_size;

	/* 2^16 stripes is quite a lot, a second limit is the size of a single
	 * item in the btree
	 */
	__le16 num_stripes;
Chris Mason's avatar
Chris Mason committed
261
262
263

	/* sub stripes only matter for raid10 */
	__le16 sub_stripes;
264
265
266
267
268
269
270
271
272
273
274
	struct btrfs_stripe stripe;
	/* additional stripes go here */
} __attribute__ ((__packed__));

static inline unsigned long btrfs_chunk_item_size(int num_stripes)
{
	BUG_ON(num_stripes == 0);
	return sizeof(struct btrfs_chunk) +
		sizeof(struct btrfs_stripe) * (num_stripes - 1);
}

275
#define BTRFS_FSID_SIZE 16
276
277
278
279
280
281
282
283
284
285
286
287
#define BTRFS_HEADER_FLAG_WRITTEN	(1ULL << 0)
#define BTRFS_HEADER_FLAG_RELOC		(1ULL << 1)
#define BTRFS_SUPER_FLAG_SEEDING	(1ULL << 32)
#define BTRFS_SUPER_FLAG_METADUMP	(1ULL << 33)

#define BTRFS_BACKREF_REV_MAX		256
#define BTRFS_BACKREF_REV_SHIFT		56
#define BTRFS_BACKREF_REV_MASK		(((u64)BTRFS_BACKREF_REV_MAX - 1) << \
					 BTRFS_BACKREF_REV_SHIFT)

#define BTRFS_OLD_BACKREF_REV		0
#define BTRFS_MIXED_BACKREF_REV		1
288

289
290
291
/*
 * every tree block (leaf or node) starts with this header.
 */
292
struct btrfs_header {
293
	/* these first four must match the super block */
Chris Mason's avatar
Chris Mason committed
294
	u8 csum[BTRFS_CSUM_SIZE];
295
	u8 fsid[BTRFS_FSID_SIZE]; /* FS specific uuid */
296
	__le64 bytenr; /* which block this node is supposed to live in */
297
	__le64 flags;
298
299
300

	/* allowed to be different from the super from here on down */
	u8 chunk_tree_uuid[BTRFS_UUID_SIZE];
301
	__le64 generation;
302
	__le64 owner;
303
	__le32 nritems;
304
	u8 level;
305
306
} __attribute__ ((__packed__));

307
#define BTRFS_NODEPTRS_PER_BLOCK(r) (((r)->nodesize - \
308
309
				      sizeof(struct btrfs_header)) / \
				     sizeof(struct btrfs_key_ptr))
310
#define __BTRFS_LEAF_DATA_SIZE(bs) ((bs) - sizeof(struct btrfs_header))
311
#define BTRFS_LEAF_DATA_SIZE(r) (__BTRFS_LEAF_DATA_SIZE(r->leafsize))
312
313
314
#define BTRFS_MAX_INLINE_DATA_SIZE(r) (BTRFS_LEAF_DATA_SIZE(r) - \
					sizeof(struct btrfs_item) - \
					sizeof(struct btrfs_file_extent_item))
315
316
317
#define BTRFS_MAX_XATTR_SIZE(r)	(BTRFS_LEAF_DATA_SIZE(r) - \
				 sizeof(struct btrfs_item) -\
				 sizeof(struct btrfs_dir_item))
318

319
320
321
322
323
324

/*
 * this is a very generous portion of the super block, giving us
 * room to translate 14 chunks with 3 stripes each.
 */
#define BTRFS_SYSTEM_CHUNK_ARRAY_SIZE 2048
325
#define BTRFS_LABEL_SIZE 256
326

327
328
329
330
/*
 * the super block basically lists the main trees of the FS
 * it currently lacks any block count etc etc
 */
331
struct btrfs_super_block {
Chris Mason's avatar
Chris Mason committed
332
	u8 csum[BTRFS_CSUM_SIZE];
333
	/* the first 4 fields must match struct btrfs_header */
Yan Zheng's avatar
Yan Zheng committed
334
	u8 fsid[BTRFS_FSID_SIZE];    /* FS specific uuid */
335
	__le64 bytenr; /* this block number */
336
	__le64 flags;
337
338

	/* allowed to be different from the btrfs_header from here own down */
339
340
341
	__le64 magic;
	__le64 generation;
	__le64 root;
342
	__le64 chunk_root;
343
	__le64 log_root;
344
345
346

	/* this will help find the new super based on the log root */
	__le64 log_root_transid;
347
348
	__le64 total_bytes;
	__le64 bytes_used;
349
	__le64 root_dir_objectid;
350
	__le64 num_devices;
351
352
353
	__le32 sectorsize;
	__le32 nodesize;
	__le32 leafsize;
354
	__le32 stripesize;
355
	__le32 sys_chunk_array_size;
356
	__le64 chunk_root_generation;
357
358
359
	__le64 compat_flags;
	__le64 compat_ro_flags;
	__le64 incompat_flags;
360
	__le16 csum_type;
361
	u8 root_level;
362
	u8 chunk_root_level;
363
	u8 log_root_level;
364
	struct btrfs_dev_item dev_item;
365

366
	char label[BTRFS_LABEL_SIZE];
367
368
369

	/* future expansion */
	__le64 reserved[32];
370
	u8 sys_chunk_array[BTRFS_SYSTEM_CHUNK_ARRAY_SIZE];
Chris Mason's avatar
Chris Mason committed
371
372
} __attribute__ ((__packed__));

373
374
375
376
/*
 * Compat flags that we support.  If any incompat flags are set other than the
 * ones specified below then we will fail to mount
 */
377
#define BTRFS_FEATURE_INCOMPAT_MIXED_BACKREF	(1ULL << 0)
378
#define BTRFS_FEATURE_INCOMPAT_DEFAULT_SUBVOL	(2ULL << 0)
379
380
381
382

#define BTRFS_FEATURE_COMPAT_SUPP		0ULL
#define BTRFS_FEATURE_COMPAT_RO_SUPP		0ULL
#define BTRFS_FEATURE_INCOMPAT_SUPP		\
383
384
	(BTRFS_FEATURE_INCOMPAT_MIXED_BACKREF |	\
	 BTRFS_FEATURE_INCOMPAT_DEFAULT_SUBVOL)
385

386
/*
387
 * A leaf is full of items. offset and size tell us where to find
388
389
 * the item in the leaf (relative to the start of the data area)
 */
Chris Mason's avatar
Chris Mason committed
390
struct btrfs_item {
Chris Mason's avatar
Chris Mason committed
391
	struct btrfs_disk_key key;
392
	__le32 offset;
393
	__le32 size;
394
395
} __attribute__ ((__packed__));

396
397
398
399
400
401
402
/*
 * leaves have an item area and a data area:
 * [item0, item1....itemN] [free space] [dataN...data1, data0]
 *
 * The data is separate from the items to get the keys closer together
 * during searches.
 */
403
struct btrfs_leaf {
404
	struct btrfs_header header;
405
	struct btrfs_item items[];
406
407
} __attribute__ ((__packed__));

408
409
410
411
/*
 * all non-leaf blocks are nodes, they hold only keys and pointers to
 * other blocks
 */
412
413
414
struct btrfs_key_ptr {
	struct btrfs_disk_key key;
	__le64 blockptr;
415
	__le64 generation;
416
417
} __attribute__ ((__packed__));

418
struct btrfs_node {
419
	struct btrfs_header header;
420
	struct btrfs_key_ptr ptrs[];
421
422
} __attribute__ ((__packed__));

423
/*
424
425
 * btrfs_paths remember the path taken from the root down to the leaf.
 * level 0 is always the leaf, and nodes[1...BTRFS_MAX_LEVEL] will point
426
427
428
429
430
 * to any other levels that are present.
 *
 * The slots array records the index of the item or block pointer
 * used while walking the tree.
 */
431
struct btrfs_path {
432
	struct extent_buffer *nodes[BTRFS_MAX_LEVEL];
433
	int slots[BTRFS_MAX_LEVEL];
434
435
	/* if there is real range locking, this locks field will change */
	int locks[BTRFS_MAX_LEVEL];
436
	int reada;
437
	/* keep some upper locks as we walk down */
438
	int lowest_level;
439
440
441
442
443

	/*
	 * set by btrfs_split_item, tells search_slot to keep all locks
	 * and to force calls to keep space in the nodes
	 */
444
445
446
447
	unsigned int search_for_split:1;
	unsigned int keep_locks:1;
	unsigned int skip_locking:1;
	unsigned int leave_spinning:1;
448
	unsigned int search_commit_root:1;
449
};
Chris Mason's avatar
Chris Mason committed
450

451
452
453
454
/*
 * items in the extent btree are used to record the objectid of the
 * owner of the block and the number of references
 */
455

456
struct btrfs_extent_item {
457
458
459
460
461
462
	__le64 refs;
	__le64 generation;
	__le64 flags;
} __attribute__ ((__packed__));

struct btrfs_extent_item_v0 {
463
	__le32 refs;
464
465
} __attribute__ ((__packed__));

466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
#define BTRFS_MAX_EXTENT_ITEM_SIZE(r) ((BTRFS_LEAF_DATA_SIZE(r) >> 4) - \
					sizeof(struct btrfs_item))

#define BTRFS_EXTENT_FLAG_DATA		(1ULL << 0)
#define BTRFS_EXTENT_FLAG_TREE_BLOCK	(1ULL << 1)

/* following flags only apply to tree blocks */

/* use full backrefs for extent pointers in the block */
#define BTRFS_BLOCK_FLAG_FULL_BACKREF	(1ULL << 8)

struct btrfs_tree_block_info {
	struct btrfs_disk_key key;
	u8 level;
} __attribute__ ((__packed__));

struct btrfs_extent_data_ref {
	__le64 root;
	__le64 objectid;
	__le64 offset;
	__le32 count;
} __attribute__ ((__packed__));

struct btrfs_shared_data_ref {
	__le32 count;
} __attribute__ ((__packed__));

struct btrfs_extent_inline_ref {
	u8 type;
495
	__le64 offset;
496
497
498
499
} __attribute__ ((__packed__));

/* old style backrefs item */
struct btrfs_extent_ref_v0 {
500
501
502
	__le64 root;
	__le64 generation;
	__le64 objectid;
503
	__le32 count;
504
505
} __attribute__ ((__packed__));

506

507
508
/* dev extents record free space on individual devices.  The owner
 * field points back to the chunk allocation mapping tree that allocated
509
 * the extent.  The chunk tree uuid field is a way to double check the owner
510
511
 */
struct btrfs_dev_extent {
512
513
514
	__le64 chunk_tree;
	__le64 chunk_objectid;
	__le64 chunk_offset;
515
	__le64 length;
516
	u8 chunk_tree_uuid[BTRFS_UUID_SIZE];
517
518
} __attribute__ ((__packed__));

519
struct btrfs_inode_ref {
520
	__le64 index;
521
522
523
524
	__le16 name_len;
	/* name goes here */
} __attribute__ ((__packed__));

525
struct btrfs_timespec {
Chris Mason's avatar
Chris Mason committed
526
	__le64 sec;
Chris Mason's avatar
Chris Mason committed
527
528
529
	__le32 nsec;
} __attribute__ ((__packed__));

Jan Engelhardt's avatar
Jan Engelhardt committed
530
enum btrfs_compression_type {
531
532
533
	BTRFS_COMPRESS_NONE = 0,
	BTRFS_COMPRESS_ZLIB = 1,
	BTRFS_COMPRESS_LAST = 2,
Jan Engelhardt's avatar
Jan Engelhardt committed
534
};
535

Chris Mason's avatar
Chris Mason committed
536
struct btrfs_inode_item {
537
	/* nfs style generation number */
Chris Mason's avatar
Chris Mason committed
538
	__le64 generation;
539
540
	/* transid that last touched this inode */
	__le64 transid;
Chris Mason's avatar
Chris Mason committed
541
	__le64 size;
542
	__le64 nbytes;
543
	__le64 block_group;
Chris Mason's avatar
Chris Mason committed
544
545
546
547
	__le32 nlink;
	__le32 uid;
	__le32 gid;
	__le32 mode;
548
	__le64 rdev;
549
	__le64 flags;
550

551
552
553
554
555
556
557
558
	/* modification sequence number for NFS */
	__le64 sequence;

	/*
	 * a little future expansion, for more than this we can
	 * just grow the inode item and version it
	 */
	__le64 reserved[4];
559
560
561
562
	struct btrfs_timespec atime;
	struct btrfs_timespec ctime;
	struct btrfs_timespec mtime;
	struct btrfs_timespec otime;
Chris Mason's avatar
Chris Mason committed
563
564
} __attribute__ ((__packed__));

565
566
567
568
struct btrfs_dir_log_item {
	__le64 end;
} __attribute__ ((__packed__));

569
struct btrfs_dir_item {
570
	struct btrfs_disk_key location;
571
	__le64 transid;
Josef Bacik's avatar
Josef Bacik committed
572
	__le16 data_len;
573
	__le16 name_len;
574
575
576
577
	u8 type;
} __attribute__ ((__packed__));

struct btrfs_root_item {
578
	struct btrfs_inode_item inode;
579
	__le64 generation;
580
	__le64 root_dirid;
581
582
583
	__le64 bytenr;
	__le64 byte_limit;
	__le64 bytes_used;
Yan Zheng's avatar
Yan Zheng committed
584
	__le64 last_snapshot;
585
	__le64 flags;
586
	__le32 refs;
587
588
	struct btrfs_disk_key drop_progress;
	u8 drop_level;
589
	u8 level;
590
} __attribute__ ((__packed__));
591

592
593
594
595
596
597
598
599
600
/*
 * this is used for both forward and backward root refs
 */
struct btrfs_root_ref {
	__le64 dirid;
	__le64 sequence;
	__le16 name_len;
} __attribute__ ((__packed__));

Yan Zheng's avatar
Yan Zheng committed
601
602
603
#define BTRFS_FILE_EXTENT_INLINE 0
#define BTRFS_FILE_EXTENT_REG 1
#define BTRFS_FILE_EXTENT_PREALLOC 2
604

605
struct btrfs_file_extent_item {
606
607
608
	/*
	 * transaction id that created this extent
	 */
609
	__le64 generation;
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
	/*
	 * max number of bytes to hold this extent in ram
	 * when we split a compressed extent we can't know how big
	 * each of the resulting pieces will be.  So, this is
	 * an upper limit on the size of the extent in ram instead of
	 * an exact limit.
	 */
	__le64 ram_bytes;

	/*
	 * 32 bits for the various ways we might encode the data,
	 * including compression and encryption.  If any of these
	 * are set to something a given disk format doesn't understand
	 * it is treated like an incompat flag for reading and writing,
	 * but not for stat.
	 */
	u8 compression;
	u8 encryption;
	__le16 other_encoding; /* spare for later use */

	/* are we inline data or a real extent? */
631
	u8 type;
632

633
634
635
636
	/*
	 * disk space consumed by the extent, checksum blocks are included
	 * in these numbers
	 */
637
638
	__le64 disk_bytenr;
	__le64 disk_num_bytes;
639
	/*
Chris Mason's avatar
Chris Mason committed
640
	 * the logical offset in file blocks (no csums)
641
642
643
644
645
646
647
	 * this extent record is for.  This allows a file extent to point
	 * into the middle of an existing extent on disk, sharing it
	 * between two snapshots (useful if some bytes in the middle of the
	 * extent have changed
	 */
	__le64 offset;
	/*
648
649
	 * the logical number of file blocks (no csums included).  This
	 * always reflects the size uncompressed and without encoding.
650
	 */
651
	__le64 num_bytes;
652

653
654
} __attribute__ ((__packed__));

Chris Mason's avatar
Chris Mason committed
655
struct btrfs_csum_item {
656
	u8 csum;
Chris Mason's avatar
Chris Mason committed
657
658
} __attribute__ ((__packed__));

659
660
661
662
/* different types of block groups (and chunks) */
#define BTRFS_BLOCK_GROUP_DATA     (1 << 0)
#define BTRFS_BLOCK_GROUP_SYSTEM   (1 << 1)
#define BTRFS_BLOCK_GROUP_METADATA (1 << 2)
663
#define BTRFS_BLOCK_GROUP_RAID0    (1 << 3)
664
#define BTRFS_BLOCK_GROUP_RAID1    (1 << 4)
665
#define BTRFS_BLOCK_GROUP_DUP	   (1 << 5)
Chris Mason's avatar
Chris Mason committed
666
#define BTRFS_BLOCK_GROUP_RAID10   (1 << 6)
667
#define BTRFS_NR_RAID_TYPES	   5
Chris Mason's avatar
Chris Mason committed
668

Chris Mason's avatar
Chris Mason committed
669
670
struct btrfs_block_group_item {
	__le64 used;
671
672
	__le64 chunk_objectid;
	__le64 flags;
Chris Mason's avatar
Chris Mason committed
673
674
} __attribute__ ((__packed__));

675
676
struct btrfs_space_info {
	u64 flags;
677
678

	u64 total_bytes;	/* total bytes in the space */
679
680
	u64 bytes_used;		/* total bytes used,
				   this does't take mirrors into account */
681
682
683
684
685
	u64 bytes_pinned;	/* total bytes pinned, will be freed when the
				   transaction finishes */
	u64 bytes_reserved;	/* total bytes the allocator has reserved for
				   current allocations */
	u64 bytes_readonly;	/* total bytes that are read only */
686

687
	u64 bytes_may_use;	/* number of bytes that may be used for
Josef Bacik's avatar
Josef Bacik committed
688
				   delalloc/allocations */
689
	u64 disk_used;		/* total bytes used on disk */
690
691
692
693
694
695

	int full;		/* indicates that we cannot allocate any more
				   chunks for this space */
	int force_alloc;	/* set if we need to force a chunk alloc for
				   this space */

696
	struct list_head list;
697
698

	/* for block groups in our same type */
699
	struct list_head block_groups[BTRFS_NR_RAID_TYPES];
700
	spinlock_t lock;
701
	struct rw_semaphore groups_sem;
Josef Bacik's avatar
Josef Bacik committed
702
	atomic_t caching_threads;
703
704
};

705
706
707
708
709
710
711
712
713
714
715
716
717
718
struct btrfs_block_rsv {
	u64 size;
	u64 reserved;
	u64 freed[2];
	struct btrfs_space_info *space_info;
	struct list_head list;
	spinlock_t lock;
	atomic_t usage;
	unsigned int priority:8;
	unsigned int durable:1;
	unsigned int refill_used:1;
	unsigned int full:1;
};

719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
/*
 * free clusters are used to claim free space in relatively large chunks,
 * allowing us to do less seeky writes.  They are used for all metadata
 * allocations and data allocations in ssd mode.
 */
struct btrfs_free_cluster {
	spinlock_t lock;
	spinlock_t refill_lock;
	struct rb_root root;

	/* largest extent in this cluster */
	u64 max_size;

	/* first extent starting offset */
	u64 window_start;

735
736
737
	/* if this cluster simply points at a bitmap in the block group */
	bool points_to_bitmap;

738
739
740
741
742
743
744
	struct btrfs_block_group_cache *block_group;
	/*
	 * when a cluster is allocated from a block group, we put the
	 * cluster onto a list in the block group so that it can
	 * be freed before the block group is freed.
	 */
	struct list_head block_group_list;
745
746
};

Josef Bacik's avatar
Josef Bacik committed
747
748
749
750
751
752
enum btrfs_caching_type {
	BTRFS_CACHE_NO		= 0,
	BTRFS_CACHE_STARTED	= 1,
	BTRFS_CACHE_FINISHED	= 2,
};

753
754
755
756
757
758
759
760
761
struct btrfs_caching_control {
	struct list_head list;
	struct mutex mutex;
	wait_queue_head_t wait;
	struct btrfs_block_group_cache *block_group;
	u64 progress;
	atomic_t count;
};

Chris Mason's avatar
Chris Mason committed
762
763
764
struct btrfs_block_group_cache {
	struct btrfs_key key;
	struct btrfs_block_group_item item;
Josef Bacik's avatar
Josef Bacik committed
765
	struct btrfs_fs_info *fs_info;
766
	spinlock_t lock;
767
	u64 pinned;
768
	u64 reserved;
769
	u64 reserved_pinned;
770
	u64 bytes_super;
771
	u64 flags;
772
773
774
775
	u64 sectorsize;
	int extents_thresh;
	int free_extents;
	int total_bitmaps;
776
	int ro;
777
778
	int dirty;

Josef Bacik's avatar
Josef Bacik committed
779
780
	/* cache tracking stuff */
	int cached;
781
782
	struct btrfs_caching_control *caching_ctl;
	u64 last_byte_to_unpin;
Josef Bacik's avatar
Josef Bacik committed
783

784
785
786
	struct btrfs_space_info *space_info;

	/* free space cache stuff */
787
	spinlock_t tree_lock;
788
	struct rb_root free_space_offset;
Josef Bacik's avatar
Josef Bacik committed
789
	u64 free_space;
790
791
792
793
794
795

	/* block group cache stuff */
	struct rb_node cache_node;

	/* for block groups in the same raid type */
	struct list_head list;
796
797
798

	/* usage count */
	atomic_t count;
799
800
801
802
803

	/* List of struct btrfs_free_clusters for this block group.
	 * Today it will only have one thing on it, but that may change
	 */
	struct list_head cluster_list;
Chris Mason's avatar
Chris Mason committed
804
};
805

806
struct reloc_control;
807
struct btrfs_device;
808
struct btrfs_fs_devices;
809
struct btrfs_fs_info {
810
	u8 fsid[BTRFS_FSID_SIZE];
811
	u8 chunk_tree_uuid[BTRFS_UUID_SIZE];
812
813
	struct btrfs_root *extent_root;
	struct btrfs_root *tree_root;
814
815
	struct btrfs_root *chunk_root;
	struct btrfs_root *dev_root;
816
	struct btrfs_root *fs_root;
817
	struct btrfs_root *csum_root;
818
819
820

	/* the log root tree is a directory of all the other log roots */
	struct btrfs_root *log_root_tree;
821
822

	spinlock_t fs_roots_radix_lock;
823
	struct radix_tree_root fs_roots_radix;
824

825
826
827
828
	/* block group cache stuff */
	spinlock_t block_group_cache_lock;
	struct rb_root block_group_cache_tree;

829
830
	struct extent_io_tree freed_extents[2];
	struct extent_io_tree *pinned_extents;
831

832
833
834
	/* logical->physical extent mapping */
	struct btrfs_mapping_tree mapping_tree;

835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
	/* block reservation for extent, checksum and root tree */
	struct btrfs_block_rsv global_block_rsv;
	/* block reservation for delay allocation */
	struct btrfs_block_rsv delalloc_block_rsv;
	/* block reservation for metadata operations */
	struct btrfs_block_rsv trans_block_rsv;
	/* block reservation for chunk tree */
	struct btrfs_block_rsv chunk_block_rsv;

	struct btrfs_block_rsv empty_block_rsv;

	/* list of block reservations that cross multiple transactions */
	struct list_head durable_block_rsv_list;

	struct mutex durable_block_rsv_mutex;

851
	u64 generation;
852
	u64 last_trans_committed;
853
854
855
856
857
858

	/*
	 * this is updated to the current trans every time a full commit
	 * is required instead of the faster short fsync log commits
	 */
	u64 last_trans_log_full_commit;
859
	u64 open_ioctl_trans;
860
	unsigned long mount_opt;
861
	u64 max_inline;
862
	u64 alloc_start;
Chris Mason's avatar
Chris Mason committed
863
	struct btrfs_transaction *running_transaction;
864
	wait_queue_head_t transaction_throttle;
865
	wait_queue_head_t transaction_wait;
866
	wait_queue_head_t async_submit_wait;
867

868
	struct btrfs_super_block super_copy;
869
	struct btrfs_super_block super_for_commit;
870
	struct block_device *__bdev;
Chris Mason's avatar
Chris Mason committed
871
	struct super_block *sb;
872
	struct inode *btree_inode;
Chris Mason's avatar
Chris Mason committed
873
	struct backing_dev_info bdi;
Chris Mason's avatar
Chris Mason committed
874
	struct mutex trans_mutex;
875
	struct mutex tree_log_mutex;
876
877
	struct mutex transaction_kthread_mutex;
	struct mutex cleaner_mutex;
878
	struct mutex chunk_mutex;
879
	struct mutex volume_mutex;
880
881
882
883
884
885
886
887
	/*
	 * this protects the ordered operations list only while we are
	 * processing all of the entries on it.  This way we make
	 * sure the commit code doesn't find the list temporarily empty
	 * because another function happens to be doing non-waiting preflush
	 * before jumping into the main commit.
	 */
	struct mutex ordered_operations_mutex;
888
	struct rw_semaphore extent_commit_sem;
889

890
	struct rw_semaphore cleanup_work_sem;
891

892
	struct rw_semaphore subvol_sem;
893
894
	struct srcu_struct subvol_srcu;

Chris Mason's avatar
Chris Mason committed
895
	struct list_head trans_list;
896
	struct list_head hashers;
897
	struct list_head dead_roots;
898
	struct list_head caching_block_groups;
899

Yan, Zheng's avatar
Yan, Zheng committed
900
901
902
	spinlock_t delayed_iput_lock;
	struct list_head delayed_iputs;

903
	atomic_t nr_async_submits;
904
	atomic_t async_submit_draining;
905
	atomic_t nr_async_bios;
906
	atomic_t async_delalloc_pages;
907

908
909
910
911
912
	/*
	 * this is used by the balancing code to wait for all the pending
	 * ordered extents
	 */
	spinlock_t ordered_extent_lock;
913
914
915
916
917
918

	/*
	 * all of the data=ordered extents pending writeback
	 * these can span multiple transactions and basically include
	 * every dirty data page that isn't from nodatacow
	 */
919
	struct list_head ordered_extents;
920
921
922
923
924
925

	/*
	 * all of the inodes that have delalloc bytes.  It is possible for
	 * this list to be empty even when there is still dirty data=ordered
	 * extents waiting to finish IO.
	 */
926
	struct list_head delalloc_inodes;
927

928
929
930
931
932
933
934
	/*
	 * special rename and truncate targets that must be on disk before
	 * we're allowed to commit.  This is basically the ext3 style
	 * data=ordered list.
	 */
	struct list_head ordered_operations;

935
936
937
938
939
940
	/*
	 * there is a pool of worker threads for checksumming during writes
	 * and a pool for checksumming after reads.  This is because readers
	 * can run with FS locks held, and the writers may be waiting for
	 * those locks.  We don't want ordering in the pending list to cause
	 * deadlocks, and so the two are serviced separately.
941
942
943
	 *
	 * A third pool does submit_bio to avoid deadlocking with the other
	 * two
944
	 */
945
	struct btrfs_workers generic_worker;
946
	struct btrfs_workers workers;
947
	struct btrfs_workers delalloc_workers;
948
	struct btrfs_workers endio_workers;
949
	struct btrfs_workers endio_meta_workers;
950
	struct btrfs_workers endio_meta_write_workers;
951
	struct btrfs_workers endio_write_workers;
952
	struct btrfs_workers submit_workers;
953
954
955
956
957
958
	/*
	 * fixup workers take dirty pages that didn't properly go through
	 * the cow mechanism and make them safe to write.  It happens
	 * for the sys_munmap function call path
	 */
	struct btrfs_workers fixup_workers;
959
960
	struct task_struct *transaction_kthread;
	struct task_struct *cleaner_kthread;
961
	int thread_pool_size;
962

963
964
	struct kobject super_kobj;
	struct completion kobj_unregister;
965
	int do_barriers;
966
	int closing;
967
	int log_root_recovering;
968
	int enospc_unlink;
969

970
	u64 total_pinned;
971
972
973
974
975

	/* protected by the delalloc lock, used to keep from writing
	 * metadata until there is a nice batch
	 */
	u64 dirty_metadata_bytes;
976
977
	struct list_head dirty_cowonly_roots;

978
	struct btrfs_fs_devices *fs_devices;
979
980
981
982
983
984

	/*
	 * the space_info list is almost entirely read only.  It only changes
	 * when we add a new raid type to the FS, and that happens
	 * very rarely.  RCU is used to protect it.
	 */
985
	struct list_head space_info;
986

987
988
	struct reloc_control *reloc_ctl;

989
	spinlock_t delalloc_lock;
990
	spinlock_t new_trans_lock;
991
	u64 delalloc_bytes;
992
993
994
995
996
997

	/* data_alloc_cluster is only used in ssd mode */
	struct btrfs_free_cluster data_alloc_cluster;

	/* all metadata allocations go through this cluster */
	struct btrfs_free_cluster meta_alloc_cluster;
998

Yan Zheng's avatar
Yan Zheng committed
999
1000