ctree.h 64 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 <asm/kmap_types.h>
30
#include "extent_io.h"
31
#include "extent_map.h"
32
#include "async-thread.h"
Chris Mason's avatar
Chris Mason committed
33

34
struct btrfs_trans_handle;
Chris Mason's avatar
Chris Mason committed
35
struct btrfs_transaction;
36
37
38
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
39
extern struct kmem_cache *btrfs_path_cachep;
40
struct btrfs_ordered_sum;
41

42
#define BTRFS_MAGIC "_BDRfS_M"
43

Josef Bacik's avatar
Josef Bacik committed
44
45
#define BTRFS_ACL_NOT_CACHED    ((void *)-1)

46
47
48
49
50
#ifdef CONFIG_LOCKDEP
# define BTRFS_MAX_LEVEL 7
#else
# define BTRFS_MAX_LEVEL 8
#endif
51
52

/* holds pointers to all of the tree roots */
53
#define BTRFS_ROOT_TREE_OBJECTID 1ULL
54
55

/* stores information about which extents are in use, and reference counts */
Chris Mason's avatar
Chris Mason committed
56
#define BTRFS_EXTENT_TREE_OBJECTID 2ULL
57
58
59
60
61

/*
 * chunk tree stores translations from logical -> physical block numbering
 * the super block points to the chunk tree
 */
62
#define BTRFS_CHUNK_TREE_OBJECTID 3ULL
63
64
65
66
67

/*
 * 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
 */
68
69
70
71
72
73
74
#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
75

76
77
78
/* orhpan objectid for tracking unlinked/truncated files */
#define BTRFS_ORPHAN_OBJECTID -5ULL

79
80
81
82
/* 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
83
84
85
86
/* for space balancing */
#define BTRFS_TREE_RELOC_OBJECTID -8ULL
#define BTRFS_DATA_RELOC_TREE_OBJECTID -9ULL

Zheng Yan's avatar
Zheng Yan committed
87
88
89
/* dummy objectid represents multiple objectids */
#define BTRFS_MULTIPLE_OBJECTIDS -255ULL

90
/*
91
 * All files have objectids in this range.
92
 */
93
#define BTRFS_FIRST_FREE_OBJECTID 256ULL
94
#define BTRFS_LAST_FREE_OBJECTID -256ULL
95
#define BTRFS_FIRST_CHUNK_TREE_OBJECTID 256ULL
96

97
98
99
100
101
102
103

/*
 * 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

Chris Mason's avatar
Chris Mason committed
104
105
106
107
108
109
/*
 * 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
110
111
/* 32 bytes in various csum fields */
#define BTRFS_CSUM_SIZE 32
112
113
/* four bytes for CRC32 */
#define BTRFS_CRC32_SIZE 4
114
#define BTRFS_EMPTY_DIR_SIZE 0
Chris Mason's avatar
Chris Mason committed
115

Chris Mason's avatar
Chris Mason committed
116
117
118
119
120
121
122
123
#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
124
125
#define BTRFS_FT_XATTR		8
#define BTRFS_FT_MAX		9
Chris Mason's avatar
Chris Mason committed
126

127
128
129
130
131
132
133
134
135
/*
 * the key defines the order in the tree, and so it also defines (optimal)
 * block layout.  objectid corresonds to the inode number.  The flags
 * tells us things about the object, and is a kind of stream selector.
 * so for a given inode, keys with flags of 1 might refer to the inode
 * data, flags of 2 may point to file data in the btree and flags == 3
 * may point to extents.
 *
 * offset is the starting byte offset for this key in the stream.
Chris Mason's avatar
Chris Mason committed
136
137
138
139
 *
 * 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)
140
 */
Chris Mason's avatar
Chris Mason committed
141
142
struct btrfs_disk_key {
	__le64 objectid;
143
	u8 type;
144
	__le64 offset;
Chris Mason's avatar
Chris Mason committed
145
146
147
} __attribute__ ((__packed__));

struct btrfs_key {
148
	u64 objectid;
149
	u8 type;
150
	u64 offset;
151
152
} __attribute__ ((__packed__));

153
154
155
156
struct btrfs_mapping_tree {
	struct extent_map_tree map_tree;
};

157
#define BTRFS_UUID_SIZE 16
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
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;

180
181
182
183
184
185
186
187
188
	/* 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;

189
	/* btrfs generated uuid for this device */
190
	u8 uuid[BTRFS_UUID_SIZE];
191
192
193
194
195
} __attribute__ ((__packed__));

struct btrfs_stripe {
	__le64 devid;
	__le64 offset;
196
	u8 dev_uuid[BTRFS_UUID_SIZE];
197
198
199
} __attribute__ ((__packed__));

struct btrfs_chunk {
200
201
202
203
	/* size of this chunk in bytes */
	__le64 length;

	/* objectid of the root referencing this chunk */
204
	__le64 owner;
205

206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
	__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
222
223
224

	/* sub stripes only matter for raid10 */
	__le16 sub_stripes;
225
226
227
228
229
230
231
232
233
234
235
	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);
}

236
#define BTRFS_FSID_SIZE 16
237
238
#define BTRFS_HEADER_FLAG_WRITTEN (1 << 0)

239
240
241
/*
 * every tree block (leaf or node) starts with this header.
 */
242
struct btrfs_header {
243
	/* these first four must match the super block */
Chris Mason's avatar
Chris Mason committed
244
	u8 csum[BTRFS_CSUM_SIZE];
245
	u8 fsid[BTRFS_FSID_SIZE]; /* FS specific uuid */
246
	__le64 bytenr; /* which block this node is supposed to live in */
247
	__le64 flags;
248
249
250

	/* allowed to be different from the super from here on down */
	u8 chunk_tree_uuid[BTRFS_UUID_SIZE];
251
	__le64 generation;
252
	__le64 owner;
253
	__le32 nritems;
254
	u8 level;
255
256
} __attribute__ ((__packed__));

257
#define BTRFS_NODEPTRS_PER_BLOCK(r) (((r)->nodesize - \
258
			        sizeof(struct btrfs_header)) / \
259
			        sizeof(struct btrfs_key_ptr))
260
#define __BTRFS_LEAF_DATA_SIZE(bs) ((bs) - sizeof(struct btrfs_header))
261
#define BTRFS_LEAF_DATA_SIZE(r) (__BTRFS_LEAF_DATA_SIZE(r->leafsize))
262
263
264
#define BTRFS_MAX_INLINE_DATA_SIZE(r) (BTRFS_LEAF_DATA_SIZE(r) - \
					sizeof(struct btrfs_item) - \
					sizeof(struct btrfs_file_extent_item))
265

266
267
268
269
270
271

/*
 * 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
272
#define BTRFS_LABEL_SIZE 256
273

274
275
276
277
/*
 * the super block basically lists the main trees of the FS
 * it currently lacks any block count etc etc
 */
278
struct btrfs_super_block {
Chris Mason's avatar
Chris Mason committed
279
	u8 csum[BTRFS_CSUM_SIZE];
280
	/* the first 4 fields must match struct btrfs_header */
281
	u8 fsid[16];    /* FS specific uuid */
282
	__le64 bytenr; /* this block number */
283
	__le64 flags;
284
285

	/* allowed to be different from the btrfs_header from here own down */
286
287
288
	__le64 magic;
	__le64 generation;
	__le64 root;
289
	__le64 chunk_root;
290
	__le64 log_root;
291
292
	__le64 total_bytes;
	__le64 bytes_used;
293
	__le64 root_dir_objectid;
294
	__le64 num_devices;
295
296
297
	__le32 sectorsize;
	__le32 nodesize;
	__le32 leafsize;
298
	__le32 stripesize;
299
	__le32 sys_chunk_array_size;
300
	__le64 chunk_root_generation;
301
	u8 root_level;
302
	u8 chunk_root_level;
303
	u8 log_root_level;
304
	struct btrfs_dev_item dev_item;
305
	char label[BTRFS_LABEL_SIZE];
306
	u8 sys_chunk_array[BTRFS_SYSTEM_CHUNK_ARRAY_SIZE];
Chris Mason's avatar
Chris Mason committed
307
308
} __attribute__ ((__packed__));

309
/*
310
 * A leaf is full of items. offset and size tell us where to find
311
312
 * the item in the leaf (relative to the start of the data area)
 */
Chris Mason's avatar
Chris Mason committed
313
struct btrfs_item {
Chris Mason's avatar
Chris Mason committed
314
	struct btrfs_disk_key key;
315
	__le32 offset;
316
	__le32 size;
317
318
} __attribute__ ((__packed__));

319
320
321
322
323
324
325
/*
 * 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.
 */
326
struct btrfs_leaf {
327
	struct btrfs_header header;
328
	struct btrfs_item items[];
329
330
} __attribute__ ((__packed__));

331
332
333
334
/*
 * all non-leaf blocks are nodes, they hold only keys and pointers to
 * other blocks
 */
335
336
337
struct btrfs_key_ptr {
	struct btrfs_disk_key key;
	__le64 blockptr;
338
	__le64 generation;
339
340
} __attribute__ ((__packed__));

341
struct btrfs_node {
342
	struct btrfs_header header;
343
	struct btrfs_key_ptr ptrs[];
344
345
} __attribute__ ((__packed__));

346
/*
347
348
 * 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
349
350
351
352
353
 * to any other levels that are present.
 *
 * The slots array records the index of the item or block pointer
 * used while walking the tree.
 */
354
struct btrfs_path {
355
	struct extent_buffer *nodes[BTRFS_MAX_LEVEL];
356
	int slots[BTRFS_MAX_LEVEL];
357
358
	/* if there is real range locking, this locks field will change */
	int locks[BTRFS_MAX_LEVEL];
359
	int reada;
360
361
	/* keep some upper locks as we walk down */
	int keep_locks;
362
	int skip_locking;
363
	int lowest_level;
364
};
Chris Mason's avatar
Chris Mason committed
365

366
367
368
369
370
371
/*
 * items in the extent btree are used to record the objectid of the
 * owner of the block and the number of references
 */
struct btrfs_extent_item {
	__le32 refs;
372
373
374
375
376
377
} __attribute__ ((__packed__));

struct btrfs_extent_ref {
	__le64 root;
	__le64 generation;
	__le64 objectid;
Zheng Yan's avatar
Zheng Yan committed
378
	__le32 num_refs;
379
380
} __attribute__ ((__packed__));

381
382
/* dev extents record free space on individual devices.  The owner
 * field points back to the chunk allocation mapping tree that allocated
383
 * the extent.  The chunk tree uuid field is a way to double check the owner
384
385
 */
struct btrfs_dev_extent {
386
387
388
	__le64 chunk_tree;
	__le64 chunk_objectid;
	__le64 chunk_offset;
389
	__le64 length;
390
	u8 chunk_tree_uuid[BTRFS_UUID_SIZE];
391
392
} __attribute__ ((__packed__));

393
struct btrfs_inode_ref {
394
	__le64 index;
395
396
397
398
	__le16 name_len;
	/* name goes here */
} __attribute__ ((__packed__));

399
struct btrfs_timespec {
Chris Mason's avatar
Chris Mason committed
400
	__le64 sec;
Chris Mason's avatar
Chris Mason committed
401
402
403
	__le32 nsec;
} __attribute__ ((__packed__));

404
405
406
407
408
409
410
411
412
413
414
415
typedef enum {
	BTRFS_COMPRESS_NONE = 0,
	BTRFS_COMPRESS_ZLIB = 1,
	BTRFS_COMPRESS_LAST = 2,
} btrfs_compression_type;

/* we don't understand any encryption methods right now */
typedef enum {
	BTRFS_ENCRYPTION_NONE = 0,
	BTRFS_ENCRYPTION_LAST = 1,
} btrfs_encryption_type;

Chris Mason's avatar
Chris Mason committed
416
struct btrfs_inode_item {
417
	/* nfs style generation number */
Chris Mason's avatar
Chris Mason committed
418
	__le64 generation;
419
420
	/* transid that last touched this inode */
	__le64 transid;
Chris Mason's avatar
Chris Mason committed
421
	__le64 size;
422
	__le64 nbytes;
423
	__le64 block_group;
Chris Mason's avatar
Chris Mason committed
424
425
426
427
	__le32 nlink;
	__le32 uid;
	__le32 gid;
	__le32 mode;
428
	__le64 rdev;
Chris Mason's avatar
Chris Mason committed
429
430
	__le16 flags;
	__le16 compat_flags;
431

432
433
434
435
	struct btrfs_timespec atime;
	struct btrfs_timespec ctime;
	struct btrfs_timespec mtime;
	struct btrfs_timespec otime;
Chris Mason's avatar
Chris Mason committed
436
437
} __attribute__ ((__packed__));

438
439
440
441
struct btrfs_dir_log_item {
	__le64 end;
} __attribute__ ((__packed__));

442
struct btrfs_dir_item {
443
	struct btrfs_disk_key location;
444
	__le64 transid;
Josef Bacik's avatar
Josef Bacik committed
445
	__le16 data_len;
446
	__le16 name_len;
447
448
449
450
	u8 type;
} __attribute__ ((__packed__));

struct btrfs_root_item {
451
	struct btrfs_inode_item inode;
452
	__le64 generation;
453
	__le64 root_dirid;
454
455
456
	__le64 bytenr;
	__le64 byte_limit;
	__le64 bytes_used;
Yan Zheng's avatar
Yan Zheng committed
457
	__le64 last_snapshot;
458
	__le32 flags;
459
	__le32 refs;
460
461
	struct btrfs_disk_key drop_progress;
	u8 drop_level;
462
	u8 level;
463
} __attribute__ ((__packed__));
464

Yan Zheng's avatar
Yan Zheng committed
465
466
467
#define BTRFS_FILE_EXTENT_INLINE 0
#define BTRFS_FILE_EXTENT_REG 1
#define BTRFS_FILE_EXTENT_PREALLOC 2
468

469
struct btrfs_file_extent_item {
470
471
472
	/*
	 * transaction id that created this extent
	 */
473
	__le64 generation;
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
	/*
	 * 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? */
495
	u8 type;
496

497
498
499
500
	/*
	 * disk space consumed by the extent, checksum blocks are included
	 * in these numbers
	 */
501
502
	__le64 disk_bytenr;
	__le64 disk_num_bytes;
503
	/*
Chris Mason's avatar
Chris Mason committed
504
	 * the logical offset in file blocks (no csums)
505
506
507
508
509
510
511
	 * 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;
	/*
512
513
	 * the logical number of file blocks (no csums included).  This
	 * always reflects the size uncompressed and without encoding.
514
	 */
515
	__le64 num_bytes;
516

517
518
} __attribute__ ((__packed__));

Chris Mason's avatar
Chris Mason committed
519
struct btrfs_csum_item {
520
	u8 csum;
Chris Mason's avatar
Chris Mason committed
521
522
} __attribute__ ((__packed__));

523
524
525
526
/* 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)
527
#define BTRFS_BLOCK_GROUP_RAID0    (1 << 3)
528
#define BTRFS_BLOCK_GROUP_RAID1    (1 << 4)
529
#define BTRFS_BLOCK_GROUP_DUP	   (1 << 5)
Chris Mason's avatar
Chris Mason committed
530
#define BTRFS_BLOCK_GROUP_RAID10   (1 << 6)
Chris Mason's avatar
Chris Mason committed
531

Chris Mason's avatar
Chris Mason committed
532
533
struct btrfs_block_group_item {
	__le64 used;
534
535
	__le64 chunk_objectid;
	__le64 flags;
Chris Mason's avatar
Chris Mason committed
536
537
} __attribute__ ((__packed__));

538
539
540
541
542
struct btrfs_space_info {
	u64 flags;
	u64 total_bytes;
	u64 bytes_used;
	u64 bytes_pinned;
543
	u64 bytes_reserved;
544
	int full;
545
	int force_alloc;
546
	struct list_head list;
547
548
549
550

	/* for block groups in our same type */
	struct list_head block_groups;
	spinlock_t lock;
551
	struct rw_semaphore groups_sem;
552
553
554
555
556
557
558
};

struct btrfs_free_space {
	struct rb_node bytes_index;
	struct rb_node offset_index;
	u64 offset;
	u64 bytes;
559
560
};

Chris Mason's avatar
Chris Mason committed
561
562
563
struct btrfs_block_group_cache {
	struct btrfs_key key;
	struct btrfs_block_group_item item;
564
	spinlock_t lock;
565
	struct mutex alloc_mutex;
566
	u64 pinned;
567
	u64 reserved;
568
569
	u64 flags;
	int cached;
570
	int ro;
571
572
573
574
575
576
577
578
579
580
581
582
583
	int dirty;

	struct btrfs_space_info *space_info;

	/* free space cache stuff */
	struct rb_root free_space_bytes;
	struct rb_root free_space_offset;

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

	/* for block groups in the same raid type */
	struct list_head list;
Chris Mason's avatar
Chris Mason committed
584
};
585

Zheng Yan's avatar
Zheng Yan committed
586
587
588
589
590
591
struct btrfs_leaf_ref_tree {
	struct rb_root root;
	struct list_head list;
	spinlock_t lock;
};

592
struct btrfs_device;
593
struct btrfs_fs_devices;
594
struct btrfs_fs_info {
595
	u8 fsid[BTRFS_FSID_SIZE];
596
	u8 chunk_tree_uuid[BTRFS_UUID_SIZE];
597
598
	struct btrfs_root *extent_root;
	struct btrfs_root *tree_root;
599
600
	struct btrfs_root *chunk_root;
	struct btrfs_root *dev_root;
601
602
603

	/* the log root tree is a directory of all the other log roots */
	struct btrfs_root *log_root_tree;
604
	struct radix_tree_root fs_roots_radix;
605

606
607
608
609
	/* block group cache stuff */
	spinlock_t block_group_cache_lock;
	struct rb_root block_group_cache_tree;

610
611
612
	struct extent_io_tree pinned_extents;
	struct extent_io_tree pending_del;
	struct extent_io_tree extent_ins;
613

614
615
616
	/* logical->physical extent mapping */
	struct btrfs_mapping_tree mapping_tree;

617
	u64 generation;
618
	u64 last_trans_committed;
619
	u64 last_trans_new_blockgroup;
620
	u64 open_ioctl_trans;
621
	unsigned long mount_opt;
622
	u64 max_extent;
623
	u64 max_inline;
624
	u64 alloc_start;
Chris Mason's avatar
Chris Mason committed
625
	struct btrfs_transaction *running_transaction;
626
	wait_queue_head_t transaction_throttle;
627
	wait_queue_head_t transaction_wait;
628

629
	wait_queue_head_t async_submit_wait;
630
631
	wait_queue_head_t tree_log_wait;

632
	struct btrfs_super_block super_copy;
633
	struct btrfs_super_block super_for_commit;
634
	struct block_device *__bdev;
Chris Mason's avatar
Chris Mason committed
635
	struct super_block *sb;
636
	struct inode *btree_inode;
Chris Mason's avatar
Chris Mason committed
637
	struct backing_dev_info bdi;
638
	spinlock_t hash_lock;
Chris Mason's avatar
Chris Mason committed
639
	struct mutex trans_mutex;
640
	struct mutex tree_log_mutex;
641
642
	struct mutex transaction_kthread_mutex;
	struct mutex cleaner_mutex;
643
644
	struct mutex extent_ins_mutex;
	struct mutex pinned_mutex;
645
	struct mutex chunk_mutex;
646
	struct mutex drop_mutex;
647
	struct mutex volume_mutex;
Zheng Yan's avatar
Zheng Yan committed
648
	struct mutex tree_reloc_mutex;
Chris Mason's avatar
Chris Mason committed
649
	struct list_head trans_list;
650
	struct list_head hashers;
651
	struct list_head dead_roots;
652

653
	atomic_t nr_async_submits;
654
	atomic_t async_submit_draining;
655
	atomic_t nr_async_bios;
656
	atomic_t async_delalloc_pages;
657
658
659
660
	atomic_t tree_log_writers;
	atomic_t tree_log_commit;
	unsigned long tree_log_batch;
	u64 tree_log_transid;
661

662
663
664
665
666
667
	/*
	 * this is used by the balancing code to wait for all the pending
	 * ordered extents
	 */
	spinlock_t ordered_extent_lock;
	struct list_head ordered_extents;
668
	struct list_head delalloc_inodes;
669

670
671
672
673
674
675
	/*
	 * 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.
676
677
678
	 *
	 * A third pool does submit_bio to avoid deadlocking with the other
	 * two
679
680
	 */
	struct btrfs_workers workers;
681
	struct btrfs_workers delalloc_workers;
682
	struct btrfs_workers endio_workers;
683
	struct btrfs_workers endio_write_workers;
684
	struct btrfs_workers submit_workers;
685
686
687
688
689
690
	/*
	 * 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;
691
692
	struct task_struct *transaction_kthread;
	struct task_struct *cleaner_kthread;
693
	int thread_pool_size;
694

Zheng Yan's avatar
Zheng Yan committed
695
696
697
	/* tree relocation relocated fields */
	struct list_head dead_reloc_roots;
	struct btrfs_leaf_ref_tree reloc_ref_tree;
Zheng Yan's avatar
Zheng Yan committed
698
699
	struct btrfs_leaf_ref_tree shared_ref_tree;

700
701
	struct kobject super_kobj;
	struct completion kobj_unregister;
702
	int do_barriers;
703
	int closing;
704
	int log_root_recovering;
705
	atomic_t throttles;
706
	atomic_t throttle_gen;
707

708
	u64 total_pinned;
709
710
	struct list_head dirty_cowonly_roots;

711
	struct btrfs_fs_devices *fs_devices;
712
	struct list_head space_info;
713
	spinlock_t delalloc_lock;
714
	spinlock_t new_trans_lock;
715
	u64 delalloc_bytes;
716
	u64 last_alloc;
717
	u64 last_data_alloc;
718

Yan Zheng's avatar
Yan Zheng committed
719
720
721
	spinlock_t ref_cache_lock;
	u64 total_ref_cache_size;

722
723
724
725
726
727
	u64 avail_data_alloc_bits;
	u64 avail_metadata_alloc_bits;
	u64 avail_system_alloc_bits;
	u64 data_alloc_profile;
	u64 metadata_alloc_profile;
	u64 system_alloc_profile;
728
729

	void *bdev_holder;
730
};
731

732
733
/*
 * in ram representation of the tree.  extent_root is used for all allocations
734
 * and for the extent tree extent_root root.
735
 */
736
struct btrfs_dirty_root;
737
struct btrfs_root {
738
	struct extent_buffer *node;
739
740
741
742

	/* the node lock is held while changing the node pointer */
	spinlock_t node_lock;

743
	struct extent_buffer *commit_root;
Yan Zheng's avatar
Yan Zheng committed
744
	struct btrfs_leaf_ref_tree *ref_tree;
745
	struct btrfs_leaf_ref_tree ref_tree_struct;
746
	struct btrfs_dirty_root *dirty_root;
747
	struct btrfs_root *log_root;
Zheng Yan's avatar
Zheng Yan committed
748
	struct btrfs_root *reloc_root;
Yan Zheng's avatar
Yan Zheng committed
749

750
751
	struct btrfs_root_item root_item;
	struct btrfs_key root_key;
752
	struct btrfs_fs_info *fs_info;
753
	struct inode *inode;
754
755
	struct extent_io_tree dirty_log_pages;

756
757
	struct kobject root_kobj;
	struct completion kobj_unregister;
758
	struct mutex objectid_mutex;
759
	struct mutex log_mutex;
760

761
762
	u64 objectid;
	u64 last_trans;
763
764
765
766
767
768
769
770
771
772

	/* data allocations are done in sectorsize units */
	u32 sectorsize;

	/* node allocations are done in nodesize units */
	u32 nodesize;

	/* leaf allocations are done in leafsize units */
	u32 leafsize;

773
774
	u32 stripesize;

775
	u32 type;
Chris Mason's avatar
Chris Mason committed
776
777
	u64 highest_inode;
	u64 last_inode_alloc;
778
	int ref_cows;
779
	int track_dirty;
780
	u64 defrag_trans_start;
781
	struct btrfs_key defrag_progress;
782
	struct btrfs_key defrag_max;
783
784
	int defrag_running;
	int defrag_level;
785
	char *name;
786
	int in_sysfs;
787
788
789

	/* the dirty list is only used by non-reference counted roots */
	struct list_head dirty_list;
790

791
792
	spinlock_t list_lock;
	struct list_head dead_list;
793
	struct list_head orphan_list;
794
795
};

Chris Mason's avatar
Chris Mason committed
796
/*
797

Chris Mason's avatar
Chris Mason committed
798
799
800
801
 * inode items have the data typically returned from stat and store other
 * info about object characteristics.  There is one for every file and dir in
 * the FS
 */
Chris Mason's avatar
Chris Mason committed
802
#define BTRFS_INODE_ITEM_KEY		1
803
804
#define BTRFS_INODE_REF_KEY		2
#define BTRFS_XATTR_ITEM_KEY		8
805
#define BTRFS_ORPHAN_ITEM_KEY		9
Chris Mason's avatar
Chris Mason committed
806
/* reserve 2-15 close to the inode for later flexibility */
Chris Mason's avatar
Chris Mason committed
807
808
809
810
811

/*
 * dir items are the name -> inode pointers in a directory.  There is one
 * for every name in a directory.
 */
812
813
#define BTRFS_DIR_LOG_ITEM_KEY  14
#define BTRFS_DIR_LOG_INDEX_KEY 15
Chris Mason's avatar
Chris Mason committed
814
815
#define BTRFS_DIR_ITEM_KEY	16
#define BTRFS_DIR_INDEX_KEY	17
Chris Mason's avatar
Chris Mason committed
816
/*
Chris Mason's avatar
Chris Mason committed
817
 * extent data is for file data
Chris Mason's avatar
Chris Mason committed
818
 */
Chris Mason's avatar
Chris Mason committed
819
#define BTRFS_EXTENT_DATA_KEY	18
Chris Mason's avatar
Chris Mason committed
820
821
822
/*
 * csum items have the checksums for data in the extents
 */
Chris Mason's avatar
Chris Mason committed
823
824
#define BTRFS_CSUM_ITEM_KEY	19

825
826

/* reserve 21-31 for other file/dir stuff */
Chris Mason's avatar
Chris Mason committed
827

Chris Mason's avatar
Chris Mason committed
828
829
830
831
/*
 * root items point to tree roots.  There are typically in the root
 * tree used by the super block to find all the other trees
 */
Chris Mason's avatar
Chris Mason committed
832
#define BTRFS_ROOT_ITEM_KEY	32
Chris Mason's avatar
Chris Mason committed
833
834
835
836
/*
 * extent items are in the extent map tree.  These record which blocks
 * are used, and how many references there are to each block
 */
Chris Mason's avatar
Chris Mason committed
837
#define BTRFS_EXTENT_ITEM_KEY	33
838
#define BTRFS_EXTENT_REF_KEY	34
Chris Mason's avatar
Chris Mason committed
839
840
841
842
843

/*
 * block groups give us hints into the extent allocation trees.  Which
 * blocks are free etc etc
 */
844
#define BTRFS_BLOCK_GROUP_ITEM_KEY 50
845

846
847
848
849
#define BTRFS_DEV_EXTENT_KEY	75
#define BTRFS_DEV_ITEM_KEY	76
#define BTRFS_CHUNK_ITEM_KEY	77

Chris Mason's avatar
Chris Mason committed
850
851
852
853
/*
 * string items are for debugging.  They just store a short string of
 * data in the FS
 */
Chris Mason's avatar
Chris Mason committed
854
855
#define BTRFS_STRING_ITEM_KEY	253

856
857
858
#define BTRFS_MOUNT_NODATASUM		(1 << 0)
#define BTRFS_MOUNT_NODATACOW		(1 << 1)
#define BTRFS_MOUNT_NOBARRIER		(1 << 2)
859
#define BTRFS_MOUNT_SSD			(1 << 3)
860
#define BTRFS_MOUNT_DEGRADED		(1 << 4)
861
#define BTRFS_MOUNT_COMPRESS		(1 << 5)
862
863
864
865
866

#define btrfs_clear_opt(o, opt)		((o) &= ~BTRFS_MOUNT_##opt)
#define btrfs_set_opt(o, opt)		((o) |= BTRFS_MOUNT_##opt)
#define btrfs_test_opt(root, opt)	((root)->fs_info->mount_opt & \
					 BTRFS_MOUNT_##opt)
Yan's avatar
Yan committed
867
868
869
/*
 * Inode flags
 */
Yan's avatar
Yan committed
870
871
872
#define BTRFS_INODE_NODATASUM		(1 << 0)
#define BTRFS_INODE_NODATACOW		(1 << 1)
#define BTRFS_INODE_READONLY		(1 << 2)
873
#define BTRFS_INODE_NOCOMPRESS		(1 << 3)
Yan Zheng's avatar
Yan Zheng committed
874
#define BTRFS_INODE_PREALLOC		(1 << 4)
Yan's avatar
Yan committed
875
876
877
878
879
880
#define btrfs_clear_flag(inode, flag)	(BTRFS_I(inode)->flags &= \
					 ~BTRFS_INODE_##flag)
#define btrfs_set_flag(inode, flag)	(BTRFS_I(inode)->flags |= \
					 BTRFS_INODE_##flag)
#define btrfs_test_flag(inode, flag)	(BTRFS_I(inode)->flags & \
					 BTRFS_INODE_##flag)
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
/* some macros to generate set/get funcs for the struct fields.  This
 * assumes there is a lefoo_to_cpu for every type, so lets make a simple
 * one for u8:
 */
#define le8_to_cpu(v) (v)
#define cpu_to_le8(v) (v)
#define __le8 u8

#define read_eb_member(eb, ptr, type, member, result) (			\
	read_extent_buffer(eb, (char *)(result),			\
			   ((unsigned long)(ptr)) +			\
			    offsetof(type, member),			\
			   sizeof(((type *)0)->member)))

#define write_eb_member(eb, ptr, type, member, result) (		\
	write_extent_buffer(eb, (char *)(result),			\
			   ((unsigned long)(ptr)) +			\
			    offsetof(type, member),			\
			   sizeof(((type *)0)->member)))

901
#ifndef BTRFS_SETGET_FUNCS
902
#define BTRFS_SETGET_FUNCS(name, type, member, bits)			\
903
904
905
u##bits btrfs_##name(struct extent_buffer *eb, type *s);		\
void btrfs_set_##name(struct extent_buffer *eb, type *s, u##bits val);
#endif
906
907
908
909

#define BTRFS_SETGET_HEADER_FUNCS(name, type, member, bits)		\
static inline u##bits btrfs_##name(struct extent_buffer *eb)		\
{									\
David Miller's avatar
David Miller committed
910
911
912
	type *p = kmap_atomic(eb->first_page, KM_USER0);		\
	u##bits res = le##bits##_to_cpu(p->member);			\
	kunmap_atomic(p, KM_USER0);					\
913
	return res;							\
914
915
916
917
}									\
static inline void btrfs_set_##name(struct extent_buffer *eb,		\
				    u##bits val)			\
{									\
David Miller's avatar
David Miller committed
918
919
920
	type *p = kmap_atomic(eb->first_page, KM_USER0);		\
	p->member = cpu_to_le##bits(val);				\
	kunmap_atomic(p, KM_USER0);					\
921
}
Chris Mason's avatar
Chris Mason committed
922

923
924
925
926
927
928
929
930
#define BTRFS_SETGET_STACK_FUNCS(name, type, member, bits)		\
static inline u##bits btrfs_##name(type *s)				\
{									\
	return le##bits##_to_cpu(s->member);				\
}									\
static inline void btrfs_set_##name(type *s, u##bits val)		\
{									\
	s->member = cpu_to_le##bits(val);				\
Chris Mason's avatar
Chris Mason committed
931
932
}

933
934
935
936
937
938
939
BTRFS_SETGET_FUNCS(device_type, struct btrfs_dev_item, type, 64);
BTRFS_SETGET_FUNCS(device_total_bytes, struct btrfs_dev_item, total_bytes, 64);
BTRFS_SETGET_FUNCS(device_bytes_used, struct btrfs_dev_item, bytes_used, 64);
BTRFS_SETGET_FUNCS(device_io_align, struct btrfs_dev_item, io_align, 32);
BTRFS_SETGET_FUNCS(device_io_width, struct btrfs_dev_item, io_width, 32);
BTRFS_SETGET_FUNCS(device_sector_size, struct btrfs_dev_item, sector_size, 32);
BTRFS_SETGET_FUNCS(device_id, struct btrfs_dev_item, devid, 64);
940
941
942
BTRFS_SETGET_FUNCS(device_group, struct btrfs_dev_item, dev_group, 32);
BTRFS_SETGET_FUNCS(device_seek_speed, struct btrfs_dev_item, seek_speed, 8);
BTRFS_SETGET_FUNCS(device_bandwidth, struct btrfs_dev_item, bandwidth, 8);
943

944
945
946
947
948
949
950
951
952
953
954
955
BTRFS_SETGET_STACK_FUNCS(stack_device_type, struct btrfs_dev_item, type, 64);
BTRFS_SETGET_STACK_FUNCS(stack_device_total_bytes, struct btrfs_dev_item,
			 total_bytes, 64);
BTRFS_SETGET_STACK_FUNCS(stack_device_bytes_used, struct btrfs_dev_item,
			 bytes_used, 64);
BTRFS_SETGET_STACK_FUNCS(stack_device_io_align, struct btrfs_dev_item,
			 io_align, 32);
BTRFS_SETGET_STACK_FUNCS(stack_device_io_width, struct btrfs_dev_item,
			 io_width, 32);
BTRFS_SETGET_STACK_FUNCS(stack_device_sector_size, struct btrfs_dev_item,
			 sector_size, 32);
BTRFS_SETGET_STACK_FUNCS(stack_device_id, struct btrfs_dev_item, devid, 64);
956
957
958
959
960
961
BTRFS_SETGET_STACK_FUNCS(stack_device_group, struct btrfs_dev_item,
			 dev_group, 32);
BTRFS_SETGET_STACK_FUNCS(stack_device_seek_speed, struct btrfs_dev_item,
			 seek_speed, 8);
BTRFS_SETGET_STACK_FUNCS(stack_device_bandwidth, struct btrfs_dev_item,
			 bandwidth, 8);
962

963
964
965
966
967
static inline char *btrfs_device_uuid(struct btrfs_dev_item *d)
{
	return (char *)d + offsetof(struct btrfs_dev_item, uuid);
}

968
BTRFS_SETGET_FUNCS(chunk_length, struct btrfs_chunk, length, 64);
969
970
971
972
973
974
975
BTRFS_SETGET_FUNCS(chunk_owner, struct btrfs_chunk, owner, 64);
BTRFS_SETGET_FUNCS(chunk_stripe_len, struct btrfs_chunk, stripe_len, 64);
BTRFS_SETGET_FUNCS(chunk_io_align, struct btrfs_chunk, io_align, 32);
BTRFS_SETGET_FUNCS(chunk_io_width, struct btrfs_chunk, io_width, 32);
BTRFS_SETGET_FUNCS(chunk_sector_size, struct btrfs_chunk, sector_size, 32);
BTRFS_SETGET_FUNCS(chunk_type, struct btrfs_chunk, type, 64);
BTRFS_SETGET_FUNCS(chunk_num_stripes, struct btrfs_chunk, num_stripes, 16);
Chris Mason's avatar
Chris Mason committed
976
BTRFS_SETGET_FUNCS(chunk_sub_stripes, struct btrfs_chunk, sub_stripes, 16);
977
978
979
BTRFS_SETGET_FUNCS(stripe_devid, struct btrfs_stripe, devid, 64);
BTRFS_SETGET_FUNCS(stripe_offset, struct btrfs_stripe, offset, 64);

980
981
982
983
984
985
static inline char *btrfs_stripe_dev_uuid(struct btrfs_stripe *s)
{
	return (char *)s + offsetof(struct btrfs_stripe, dev_uuid);
}

BTRFS_SETGET_STACK_FUNCS(stack_chunk_length, struct btrfs_chunk, length, 64);
986
987
988
989
990
991
992
993
994
995
996
997
BTRFS_SETGET_STACK_FUNCS(stack_chunk_owner, struct btrfs_chunk, owner, 64);
BTRFS_SETGET_STACK_FUNCS(stack_chunk_stripe_len, struct btrfs_chunk,
			 stripe_len, 64);
BTRFS_SETGET_STACK_FUNCS(stack_chunk_io_align, struct btrfs_chunk,
			 io_align, 32);
BTRFS_SETGET_STACK_FUNCS(stack_chunk_io_width, struct btrfs_chunk,
			 io_width, 32);
BTRFS_SETGET_STACK_FUNCS(stack_chunk_sector_size, struct btrfs_chunk,
			 sector_size, 32);
BTRFS_SETGET_STACK_FUNCS(stack_chunk_type, struct btrfs_chunk, type, 64);
BTRFS_SETGET_STACK_FUNCS(stack_chunk_num_stripes, struct btrfs_chunk,
			 num_stripes, 16);