1/* SPDX-License-Identifier: GPL-2.0 WITH Linux-syscall-note */
2#ifndef _BTRFS_CTREE_H_
3#define _BTRFS_CTREE_H_
4
5#include <linux/btrfs.h>
6#include <linux/types.h>
7#ifdef __KERNEL__
8#include <linux/stddef.h>
9#else
10#include <stddef.h>
11#endif
12
13/* ASCII for _BHRfS_M, no terminating nul */
14#define BTRFS_MAGIC 0x4D5F53665248425FULL
15
16#define BTRFS_MAX_LEVEL 8
17
18/*
19 * We can actually store much bigger names, but lets not confuse the rest of
20 * linux.
21 */
22#define BTRFS_NAME_LEN 255
23
24/*
25 * Theoretical limit is larger, but we keep this down to a sane value. That
26 * should limit greatly the possibility of collisions on inode ref items.
27 */
28#define BTRFS_LINK_MAX 65535U
29
30/*
31 * This header contains the structure definitions and constants used
32 * by file system objects that can be retrieved using
33 * the BTRFS_IOC_SEARCH_TREE ioctl. That means basically anything that
34 * is needed to describe a leaf node's key or item contents.
35 */
36
37/* holds pointers to all of the tree roots */
38#define BTRFS_ROOT_TREE_OBJECTID 1ULL
39
40/* stores information about which extents are in use, and reference counts */
41#define BTRFS_EXTENT_TREE_OBJECTID 2ULL
42
43/*
44 * chunk tree stores translations from logical -> physical block numbering
45 * the super block points to the chunk tree
46 */
47#define BTRFS_CHUNK_TREE_OBJECTID 3ULL
48
49/*
50 * stores information about which areas of a given device are in use.
51 * one per device. The tree of tree roots points to the device tree
52 */
53#define BTRFS_DEV_TREE_OBJECTID 4ULL
54
55/* one per subvolume, storing files and directories */
56#define BTRFS_FS_TREE_OBJECTID 5ULL
57
58/* directory objectid inside the root tree */
59#define BTRFS_ROOT_TREE_DIR_OBJECTID 6ULL
60
61/* holds checksums of all the data extents */
62#define BTRFS_CSUM_TREE_OBJECTID 7ULL
63
64/* holds quota configuration and tracking */
65#define BTRFS_QUOTA_TREE_OBJECTID 8ULL
66
67/* for storing items that use the BTRFS_UUID_KEY* types */
68#define BTRFS_UUID_TREE_OBJECTID 9ULL
69
70/* tracks free space in block groups. */
71#define BTRFS_FREE_SPACE_TREE_OBJECTID 10ULL
72
73/* Holds the block group items for extent tree v2. */
74#define BTRFS_BLOCK_GROUP_TREE_OBJECTID 11ULL
75
76/* Tracks RAID stripes in block groups. */
77#define BTRFS_RAID_STRIPE_TREE_OBJECTID 12ULL
78
79/* device stats in the device tree */
80#define BTRFS_DEV_STATS_OBJECTID 0ULL
81
82/* for storing balance parameters in the root tree */
83#define BTRFS_BALANCE_OBJECTID -4ULL
84
85/* orphan objectid for tracking unlinked/truncated files */
86#define BTRFS_ORPHAN_OBJECTID -5ULL
87
88/* does write ahead logging to speed up fsyncs */
89#define BTRFS_TREE_LOG_OBJECTID -6ULL
90#define BTRFS_TREE_LOG_FIXUP_OBJECTID -7ULL
91
92/* for space balancing */
93#define BTRFS_TREE_RELOC_OBJECTID -8ULL
94#define BTRFS_DATA_RELOC_TREE_OBJECTID -9ULL
95
96/*
97 * extent checksums all have this objectid
98 * this allows them to share the logging tree
99 * for fsyncs
100 */
101#define BTRFS_EXTENT_CSUM_OBJECTID -10ULL
102
103/* For storing free space cache */
104#define BTRFS_FREE_SPACE_OBJECTID -11ULL
105
106/*
107 * The inode number assigned to the special inode for storing
108 * free ino cache
109 */
110#define BTRFS_FREE_INO_OBJECTID -12ULL
111
112/* dummy objectid represents multiple objectids */
113#define BTRFS_MULTIPLE_OBJECTIDS -255ULL
114
115/*
116 * All files have objectids in this range.
117 */
118#define BTRFS_FIRST_FREE_OBJECTID 256ULL
119#define BTRFS_LAST_FREE_OBJECTID -256ULL
120#define BTRFS_FIRST_CHUNK_TREE_OBJECTID 256ULL
121
122
123/*
124 * the device items go into the chunk tree. The key is in the form
125 * [ 1 BTRFS_DEV_ITEM_KEY device_id ]
126 */
127#define BTRFS_DEV_ITEMS_OBJECTID 1ULL
128
129#define BTRFS_BTREE_INODE_OBJECTID 1
130
131#define BTRFS_EMPTY_SUBVOL_DIR_OBJECTID 2
132
133#define BTRFS_DEV_REPLACE_DEVID 0ULL
134
135/*
136 * inode items have the data typically returned from stat and store other
137 * info about object characteristics. There is one for every file and dir in
138 * the FS
139 */
140#define BTRFS_INODE_ITEM_KEY 1
141#define BTRFS_INODE_REF_KEY 12
142#define BTRFS_INODE_EXTREF_KEY 13
143#define BTRFS_XATTR_ITEM_KEY 24
144
145/*
146 * fs verity items are stored under two different key types on disk.
147 * The descriptor items:
148 * [ inode objectid, BTRFS_VERITY_DESC_ITEM_KEY, offset ]
149 *
150 * At offset 0, we store a btrfs_verity_descriptor_item which tracks the size
151 * of the descriptor item and some extra data for encryption.
152 * Starting at offset 1, these hold the generic fs verity descriptor. The
153 * latter are opaque to btrfs, we just read and write them as a blob for the
154 * higher level verity code. The most common descriptor size is 256 bytes.
155 *
156 * The merkle tree items:
157 * [ inode objectid, BTRFS_VERITY_MERKLE_ITEM_KEY, offset ]
158 *
159 * These also start at offset 0, and correspond to the merkle tree bytes. When
160 * fsverity asks for page 0 of the merkle tree, we pull up one page starting at
161 * offset 0 for this key type. These are also opaque to btrfs, we're blindly
162 * storing whatever fsverity sends down.
163 */
164#define BTRFS_VERITY_DESC_ITEM_KEY 36
165#define BTRFS_VERITY_MERKLE_ITEM_KEY 37
166
167#define BTRFS_ORPHAN_ITEM_KEY 48
168/* reserve 2-15 close to the inode for later flexibility */
169
170/*
171 * dir items are the name -> inode pointers in a directory. There is one
172 * for every name in a directory. BTRFS_DIR_LOG_ITEM_KEY is no longer used
173 * but it's still defined here for documentation purposes and to help avoid
174 * having its numerical value reused in the future.
175 */
176#define BTRFS_DIR_LOG_ITEM_KEY 60
177#define BTRFS_DIR_LOG_INDEX_KEY 72
178#define BTRFS_DIR_ITEM_KEY 84
179#define BTRFS_DIR_INDEX_KEY 96
180/*
181 * extent data is for file data
182 */
183#define BTRFS_EXTENT_DATA_KEY 108
184
185/*
186 * extent csums are stored in a separate tree and hold csums for
187 * an entire extent on disk.
188 */
189#define BTRFS_EXTENT_CSUM_KEY 128
190
191/*
192 * root items point to tree roots. They are typically in the root
193 * tree used by the super block to find all the other trees
194 */
195#define BTRFS_ROOT_ITEM_KEY 132
196
197/*
198 * root backrefs tie subvols and snapshots to the directory entries that
199 * reference them
200 */
201#define BTRFS_ROOT_BACKREF_KEY 144
202
203/*
204 * root refs make a fast index for listing all of the snapshots and
205 * subvolumes referenced by a given root. They point directly to the
206 * directory item in the root that references the subvol
207 */
208#define BTRFS_ROOT_REF_KEY 156
209
210/*
211 * extent items are in the extent map tree. These record which blocks
212 * are used, and how many references there are to each block
213 */
214#define BTRFS_EXTENT_ITEM_KEY 168
215
216/*
217 * The same as the BTRFS_EXTENT_ITEM_KEY, except it's metadata we already know
218 * the length, so we save the level in key->offset instead of the length.
219 */
220#define BTRFS_METADATA_ITEM_KEY 169
221
222#define BTRFS_TREE_BLOCK_REF_KEY 176
223
224#define BTRFS_EXTENT_DATA_REF_KEY 178
225
226/*
227 * Obsolete key. Defintion removed in 6.6, value may be reused in the future.
228 *
229 * #define BTRFS_EXTENT_REF_V0_KEY 180
230 */
231
232#define BTRFS_SHARED_BLOCK_REF_KEY 182
233
234#define BTRFS_SHARED_DATA_REF_KEY 184
235
236/*
237 * Special inline ref key which stores the id of the subvolume which originally
238 * created the extent. This subvolume owns the extent permanently from the
239 * perspective of simple quotas. Needed to know which subvolume to free quota
240 * usage from when the extent is deleted.
241 */
242#define BTRFS_EXTENT_OWNER_REF_KEY 188
243
244/*
245 * block groups give us hints into the extent allocation trees. Which
246 * blocks are free etc etc
247 */
248#define BTRFS_BLOCK_GROUP_ITEM_KEY 192
249
250/*
251 * Every block group is represented in the free space tree by a free space info
252 * item, which stores some accounting information. It is keyed on
253 * (block_group_start, FREE_SPACE_INFO, block_group_length).
254 */
255#define BTRFS_FREE_SPACE_INFO_KEY 198
256
257/*
258 * A free space extent tracks an extent of space that is free in a block group.
259 * It is keyed on (start, FREE_SPACE_EXTENT, length).
260 */
261#define BTRFS_FREE_SPACE_EXTENT_KEY 199
262
263/*
264 * When a block group becomes very fragmented, we convert it to use bitmaps
265 * instead of extents. A free space bitmap is keyed on
266 * (start, FREE_SPACE_BITMAP, length); the corresponding item is a bitmap with
267 * (length / sectorsize) bits.
268 */
269#define BTRFS_FREE_SPACE_BITMAP_KEY 200
270
271#define BTRFS_DEV_EXTENT_KEY 204
272#define BTRFS_DEV_ITEM_KEY 216
273#define BTRFS_CHUNK_ITEM_KEY 228
274
275#define BTRFS_RAID_STRIPE_KEY 230
276
277/*
278 * Records the overall state of the qgroups.
279 * There's only one instance of this key present,
280 * (0, BTRFS_QGROUP_STATUS_KEY, 0)
281 */
282#define BTRFS_QGROUP_STATUS_KEY 240
283/*
284 * Records the currently used space of the qgroup.
285 * One key per qgroup, (0, BTRFS_QGROUP_INFO_KEY, qgroupid).
286 */
287#define BTRFS_QGROUP_INFO_KEY 242
288/*
289 * Contains the user configured limits for the qgroup.
290 * One key per qgroup, (0, BTRFS_QGROUP_LIMIT_KEY, qgroupid).
291 */
292#define BTRFS_QGROUP_LIMIT_KEY 244
293/*
294 * Records the child-parent relationship of qgroups. For
295 * each relation, 2 keys are present:
296 * (childid, BTRFS_QGROUP_RELATION_KEY, parentid)
297 * (parentid, BTRFS_QGROUP_RELATION_KEY, childid)
298 */
299#define BTRFS_QGROUP_RELATION_KEY 246
300
301/*
302 * Obsolete name, see BTRFS_TEMPORARY_ITEM_KEY.
303 */
304#define BTRFS_BALANCE_ITEM_KEY 248
305
306/*
307 * The key type for tree items that are stored persistently, but do not need to
308 * exist for extended period of time. The items can exist in any tree.
309 *
310 * [subtype, BTRFS_TEMPORARY_ITEM_KEY, data]
311 *
312 * Existing items:
313 *
314 * - balance status item
315 * (BTRFS_BALANCE_OBJECTID, BTRFS_TEMPORARY_ITEM_KEY, 0)
316 */
317#define BTRFS_TEMPORARY_ITEM_KEY 248
318
319/*
320 * Obsolete name, see BTRFS_PERSISTENT_ITEM_KEY
321 */
322#define BTRFS_DEV_STATS_KEY 249
323
324/*
325 * The key type for tree items that are stored persistently and usually exist
326 * for a long period, eg. filesystem lifetime. The item kinds can be status
327 * information, stats or preference values. The item can exist in any tree.
328 *
329 * [subtype, BTRFS_PERSISTENT_ITEM_KEY, data]
330 *
331 * Existing items:
332 *
333 * - device statistics, store IO stats in the device tree, one key for all
334 * stats
335 * (BTRFS_DEV_STATS_OBJECTID, BTRFS_DEV_STATS_KEY, 0)
336 */
337#define BTRFS_PERSISTENT_ITEM_KEY 249
338
339/*
340 * Persistently stores the device replace state in the device tree.
341 * The key is built like this: (0, BTRFS_DEV_REPLACE_KEY, 0).
342 */
343#define BTRFS_DEV_REPLACE_KEY 250
344
345/*
346 * Stores items that allow to quickly map UUIDs to something else.
347 * These items are part of the filesystem UUID tree.
348 * The key is built like this:
349 * (UUID_upper_64_bits, BTRFS_UUID_KEY*, UUID_lower_64_bits).
350 */
351#if BTRFS_UUID_SIZE != 16
352#error "UUID items require BTRFS_UUID_SIZE == 16!"
353#endif
354#define BTRFS_UUID_KEY_SUBVOL 251 /* for UUIDs assigned to subvols */
355#define BTRFS_UUID_KEY_RECEIVED_SUBVOL 252 /* for UUIDs assigned to
356 * received subvols */
357
358/*
359 * string items are for debugging. They just store a short string of
360 * data in the FS
361 */
362#define BTRFS_STRING_ITEM_KEY 253
363
364/* Maximum metadata block size (nodesize) */
365#define BTRFS_MAX_METADATA_BLOCKSIZE 65536
366
367/* 32 bytes in various csum fields */
368#define BTRFS_CSUM_SIZE 32
369
370/* csum types */
371enum btrfs_csum_type {
372 BTRFS_CSUM_TYPE_CRC32 = 0,
373 BTRFS_CSUM_TYPE_XXHASH = 1,
374 BTRFS_CSUM_TYPE_SHA256 = 2,
375 BTRFS_CSUM_TYPE_BLAKE2 = 3,
376};
377
378/*
379 * flags definitions for directory entry item type
380 *
381 * Used by:
382 * struct btrfs_dir_item.type
383 *
384 * Values 0..7 must match common file type values in fs_types.h.
385 */
386#define BTRFS_FT_UNKNOWN 0
387#define BTRFS_FT_REG_FILE 1
388#define BTRFS_FT_DIR 2
389#define BTRFS_FT_CHRDEV 3
390#define BTRFS_FT_BLKDEV 4
391#define BTRFS_FT_FIFO 5
392#define BTRFS_FT_SOCK 6
393#define BTRFS_FT_SYMLINK 7
394#define BTRFS_FT_XATTR 8
395#define BTRFS_FT_MAX 9
396/* Directory contains encrypted data */
397#define BTRFS_FT_ENCRYPTED 0x80
398
399static inline __u8 btrfs_dir_flags_to_ftype(__u8 flags)
400{
401 return flags & ~BTRFS_FT_ENCRYPTED;
402}
403
404/*
405 * Inode flags
406 */
407#define BTRFS_INODE_NODATASUM (1U << 0)
408#define BTRFS_INODE_NODATACOW (1U << 1)
409#define BTRFS_INODE_READONLY (1U << 2)
410#define BTRFS_INODE_NOCOMPRESS (1U << 3)
411#define BTRFS_INODE_PREALLOC (1U << 4)
412#define BTRFS_INODE_SYNC (1U << 5)
413#define BTRFS_INODE_IMMUTABLE (1U << 6)
414#define BTRFS_INODE_APPEND (1U << 7)
415#define BTRFS_INODE_NODUMP (1U << 8)
416#define BTRFS_INODE_NOATIME (1U << 9)
417#define BTRFS_INODE_DIRSYNC (1U << 10)
418#define BTRFS_INODE_COMPRESS (1U << 11)
419
420#define BTRFS_INODE_ROOT_ITEM_INIT (1U << 31)
421
422#define BTRFS_INODE_FLAG_MASK \
423 (BTRFS_INODE_NODATASUM | \
424 BTRFS_INODE_NODATACOW | \
425 BTRFS_INODE_READONLY | \
426 BTRFS_INODE_NOCOMPRESS | \
427 BTRFS_INODE_PREALLOC | \
428 BTRFS_INODE_SYNC | \
429 BTRFS_INODE_IMMUTABLE | \
430 BTRFS_INODE_APPEND | \
431 BTRFS_INODE_NODUMP | \
432 BTRFS_INODE_NOATIME | \
433 BTRFS_INODE_DIRSYNC | \
434 BTRFS_INODE_COMPRESS | \
435 BTRFS_INODE_ROOT_ITEM_INIT)
436
437#define BTRFS_INODE_RO_VERITY (1U << 0)
438
439#define BTRFS_INODE_RO_FLAG_MASK (BTRFS_INODE_RO_VERITY)
440
441/*
442 * The key defines the order in the tree, and so it also defines (optimal)
443 * block layout.
444 *
445 * objectid corresponds to the inode number.
446 *
447 * type tells us things about the object, and is a kind of stream selector.
448 * so for a given inode, keys with type of 1 might refer to the inode data,
449 * type of 2 may point to file data in the btree and type == 3 may point to
450 * extents.
451 *
452 * offset is the starting byte offset for this key in the stream.
453 *
454 * btrfs_disk_key is in disk byte order. struct btrfs_key is always
455 * in cpu native order. Otherwise they are identical and their sizes
456 * should be the same (ie both packed)
457 */
458struct btrfs_disk_key {
459 __le64 objectid;
460 __u8 type;
461 __le64 offset;
462} __attribute__ ((__packed__));
463
464struct btrfs_key {
465 __u64 objectid;
466 __u8 type;
467 __u64 offset;
468} __attribute__ ((__packed__));
469
470/*
471 * Every tree block (leaf or node) starts with this header.
472 */
473struct btrfs_header {
474 /* These first four must match the super block */
475 __u8 csum[BTRFS_CSUM_SIZE];
476 /* FS specific uuid */
477 __u8 fsid[BTRFS_FSID_SIZE];
478 /* Which block this node is supposed to live in */
479 __le64 bytenr;
480 __le64 flags;
481
482 /* Allowed to be different from the super from here on down */
483 __u8 chunk_tree_uuid[BTRFS_UUID_SIZE];
484 __le64 generation;
485 __le64 owner;
486 __le32 nritems;
487 __u8 level;
488} __attribute__ ((__packed__));
489
490/*
491 * This is a very generous portion of the super block, giving us room to
492 * translate 14 chunks with 3 stripes each.
493 */
494#define BTRFS_SYSTEM_CHUNK_ARRAY_SIZE 2048
495
496/*
497 * Just in case we somehow lose the roots and are not able to mount, we store
498 * an array of the roots from previous transactions in the super.
499 */
500#define BTRFS_NUM_BACKUP_ROOTS 4
501struct btrfs_root_backup {
502 __le64 tree_root;
503 __le64 tree_root_gen;
504
505 __le64 chunk_root;
506 __le64 chunk_root_gen;
507
508 __le64 extent_root;
509 __le64 extent_root_gen;
510
511 __le64 fs_root;
512 __le64 fs_root_gen;
513
514 __le64 dev_root;
515 __le64 dev_root_gen;
516
517 __le64 csum_root;
518 __le64 csum_root_gen;
519
520 __le64 total_bytes;
521 __le64 bytes_used;
522 __le64 num_devices;
523 /* future */
524 __le64 unused_64[4];
525
526 __u8 tree_root_level;
527 __u8 chunk_root_level;
528 __u8 extent_root_level;
529 __u8 fs_root_level;
530 __u8 dev_root_level;
531 __u8 csum_root_level;
532 /* future and to align */
533 __u8 unused_8[10];
534} __attribute__ ((__packed__));
535
536/*
537 * A leaf is full of items. offset and size tell us where to find the item in
538 * the leaf (relative to the start of the data area)
539 */
540struct btrfs_item {
541 struct btrfs_disk_key key;
542 __le32 offset;
543 __le32 size;
544} __attribute__ ((__packed__));
545
546/*
547 * Leaves have an item area and a data area:
548 * [item0, item1....itemN] [free space] [dataN...data1, data0]
549 *
550 * The data is separate from the items to get the keys closer together during
551 * searches.
552 */
553struct btrfs_leaf {
554 struct btrfs_header header;
555 struct btrfs_item items[];
556} __attribute__ ((__packed__));
557
558/*
559 * All non-leaf blocks are nodes, they hold only keys and pointers to other
560 * blocks.
561 */
562struct btrfs_key_ptr {
563 struct btrfs_disk_key key;
564 __le64 blockptr;
565 __le64 generation;
566} __attribute__ ((__packed__));
567
568struct btrfs_node {
569 struct btrfs_header header;
570 struct btrfs_key_ptr ptrs[];
571} __attribute__ ((__packed__));
572
573struct btrfs_dev_item {
574 /* the internal btrfs device id */
575 __le64 devid;
576
577 /* size of the device */
578 __le64 total_bytes;
579
580 /* bytes used */
581 __le64 bytes_used;
582
583 /* optimal io alignment for this device */
584 __le32 io_align;
585
586 /* optimal io width for this device */
587 __le32 io_width;
588
589 /* minimal io size for this device */
590 __le32 sector_size;
591
592 /* type and info about this device */
593 __le64 type;
594
595 /* expected generation for this device */
596 __le64 generation;
597
598 /*
599 * starting byte of this partition on the device,
600 * to allow for stripe alignment in the future
601 */
602 __le64 start_offset;
603
604 /* grouping information for allocation decisions */
605 __le32 dev_group;
606
607 /* seek speed 0-100 where 100 is fastest */
608 __u8 seek_speed;
609
610 /* bandwidth 0-100 where 100 is fastest */
611 __u8 bandwidth;
612
613 /* btrfs generated uuid for this device */
614 __u8 uuid[BTRFS_UUID_SIZE];
615
616 /* uuid of FS who owns this device */
617 __u8 fsid[BTRFS_UUID_SIZE];
618} __attribute__ ((__packed__));
619
620struct btrfs_stripe {
621 __le64 devid;
622 __le64 offset;
623 __u8 dev_uuid[BTRFS_UUID_SIZE];
624} __attribute__ ((__packed__));
625
626struct btrfs_chunk {
627 /* size of this chunk in bytes */
628 __le64 length;
629
630 /* objectid of the root referencing this chunk */
631 __le64 owner;
632
633 __le64 stripe_len;
634 __le64 type;
635
636 /* optimal io alignment for this chunk */
637 __le32 io_align;
638
639 /* optimal io width for this chunk */
640 __le32 io_width;
641
642 /* minimal io size for this chunk */
643 __le32 sector_size;
644
645 /* 2^16 stripes is quite a lot, a second limit is the size of a single
646 * item in the btree
647 */
648 __le16 num_stripes;
649
650 /* sub stripes only matter for raid10 */
651 __le16 sub_stripes;
652 struct btrfs_stripe stripe;
653 /* additional stripes go here */
654} __attribute__ ((__packed__));
655
656/*
657 * The super block basically lists the main trees of the FS.
658 */
659struct btrfs_super_block {
660 /* The first 4 fields must match struct btrfs_header */
661 __u8 csum[BTRFS_CSUM_SIZE];
662 /* FS specific UUID, visible to user */
663 __u8 fsid[BTRFS_FSID_SIZE];
664 /* This block number */
665 __le64 bytenr;
666 __le64 flags;
667
668 /* Allowed to be different from the btrfs_header from here own down */
669 __le64 magic;
670 __le64 generation;
671 __le64 root;
672 __le64 chunk_root;
673 __le64 log_root;
674
675 /*
676 * This member has never been utilized since the very beginning, thus
677 * it's always 0 regardless of kernel version. We always use
678 * generation + 1 to read log tree root. So here we mark it deprecated.
679 */
680 __le64 __unused_log_root_transid;
681 __le64 total_bytes;
682 __le64 bytes_used;
683 __le64 root_dir_objectid;
684 __le64 num_devices;
685 __le32 sectorsize;
686 __le32 nodesize;
687 __le32 __unused_leafsize;
688 __le32 stripesize;
689 __le32 sys_chunk_array_size;
690 __le64 chunk_root_generation;
691 __le64 compat_flags;
692 __le64 compat_ro_flags;
693 __le64 incompat_flags;
694 __le16 csum_type;
695 __u8 root_level;
696 __u8 chunk_root_level;
697 __u8 log_root_level;
698 struct btrfs_dev_item dev_item;
699
700 char label[BTRFS_LABEL_SIZE];
701
702 __le64 cache_generation;
703 __le64 uuid_tree_generation;
704
705 /* The UUID written into btree blocks */
706 __u8 metadata_uuid[BTRFS_FSID_SIZE];
707
708 __u64 nr_global_roots;
709
710 /* Future expansion */
711 __le64 reserved[27];
712 __u8 sys_chunk_array[BTRFS_SYSTEM_CHUNK_ARRAY_SIZE];
713 struct btrfs_root_backup super_roots[BTRFS_NUM_BACKUP_ROOTS];
714
715 /* Padded to 4096 bytes */
716 __u8 padding[565];
717} __attribute__ ((__packed__));
718
719#define BTRFS_FREE_SPACE_EXTENT 1
720#define BTRFS_FREE_SPACE_BITMAP 2
721
722struct btrfs_free_space_entry {
723 __le64 offset;
724 __le64 bytes;
725 __u8 type;
726} __attribute__ ((__packed__));
727
728struct btrfs_free_space_header {
729 struct btrfs_disk_key location;
730 __le64 generation;
731 __le64 num_entries;
732 __le64 num_bitmaps;
733} __attribute__ ((__packed__));
734
735struct btrfs_raid_stride {
736 /* The id of device this raid extent lives on. */
737 __le64 devid;
738 /* The physical location on disk. */
739 __le64 physical;
740} __attribute__ ((__packed__));
741
742/* The stripe_extent::encoding, 1:1 mapping of enum btrfs_raid_types. */
743#define BTRFS_STRIPE_RAID0 1
744#define BTRFS_STRIPE_RAID1 2
745#define BTRFS_STRIPE_DUP 3
746#define BTRFS_STRIPE_RAID10 4
747#define BTRFS_STRIPE_RAID5 5
748#define BTRFS_STRIPE_RAID6 6
749#define BTRFS_STRIPE_RAID1C3 7
750#define BTRFS_STRIPE_RAID1C4 8
751
752struct btrfs_stripe_extent {
753 __u8 encoding;
754 __u8 reserved[7];
755 /* An array of raid strides this stripe is composed of. */
756 struct btrfs_raid_stride strides[];
757} __attribute__ ((__packed__));
758
759#define BTRFS_HEADER_FLAG_WRITTEN (1ULL << 0)
760#define BTRFS_HEADER_FLAG_RELOC (1ULL << 1)
761
762/* Super block flags */
763/* Errors detected */
764#define BTRFS_SUPER_FLAG_ERROR (1ULL << 2)
765
766#define BTRFS_SUPER_FLAG_SEEDING (1ULL << 32)
767#define BTRFS_SUPER_FLAG_METADUMP (1ULL << 33)
768#define BTRFS_SUPER_FLAG_METADUMP_V2 (1ULL << 34)
769#define BTRFS_SUPER_FLAG_CHANGING_FSID (1ULL << 35)
770#define BTRFS_SUPER_FLAG_CHANGING_FSID_V2 (1ULL << 36)
771
772
773/*
774 * items in the extent btree are used to record the objectid of the
775 * owner of the block and the number of references
776 */
777
778struct btrfs_extent_item {
779 __le64 refs;
780 __le64 generation;
781 __le64 flags;
782} __attribute__ ((__packed__));
783
784struct btrfs_extent_item_v0 {
785 __le32 refs;
786} __attribute__ ((__packed__));
787
788
789#define BTRFS_EXTENT_FLAG_DATA (1ULL << 0)
790#define BTRFS_EXTENT_FLAG_TREE_BLOCK (1ULL << 1)
791
792/* following flags only apply to tree blocks */
793
794/* use full backrefs for extent pointers in the block */
795#define BTRFS_BLOCK_FLAG_FULL_BACKREF (1ULL << 8)
796
797#define BTRFS_BACKREF_REV_MAX 256
798#define BTRFS_BACKREF_REV_SHIFT 56
799#define BTRFS_BACKREF_REV_MASK (((u64)BTRFS_BACKREF_REV_MAX - 1) << \
800 BTRFS_BACKREF_REV_SHIFT)
801
802#define BTRFS_OLD_BACKREF_REV 0
803#define BTRFS_MIXED_BACKREF_REV 1
804
805/*
806 * this flag is only used internally by scrub and may be changed at any time
807 * it is only declared here to avoid collisions
808 */
809#define BTRFS_EXTENT_FLAG_SUPER (1ULL << 48)
810
811struct btrfs_tree_block_info {
812 struct btrfs_disk_key key;
813 __u8 level;
814} __attribute__ ((__packed__));
815
816struct btrfs_extent_data_ref {
817 __le64 root;
818 __le64 objectid;
819 __le64 offset;
820 __le32 count;
821} __attribute__ ((__packed__));
822
823struct btrfs_shared_data_ref {
824 __le32 count;
825} __attribute__ ((__packed__));
826
827struct btrfs_extent_owner_ref {
828 __le64 root_id;
829} __attribute__ ((__packed__));
830
831struct btrfs_extent_inline_ref {
832 __u8 type;
833 __le64 offset;
834} __attribute__ ((__packed__));
835
836/* dev extents record free space on individual devices. The owner
837 * field points back to the chunk allocation mapping tree that allocated
838 * the extent. The chunk tree uuid field is a way to double check the owner
839 */
840struct btrfs_dev_extent {
841 __le64 chunk_tree;
842 __le64 chunk_objectid;
843 __le64 chunk_offset;
844 __le64 length;
845 __u8 chunk_tree_uuid[BTRFS_UUID_SIZE];
846} __attribute__ ((__packed__));
847
848struct btrfs_inode_ref {
849 __le64 index;
850 __le16 name_len;
851 /* name goes here */
852} __attribute__ ((__packed__));
853
854struct btrfs_inode_extref {
855 __le64 parent_objectid;
856 __le64 index;
857 __le16 name_len;
858 __u8 name[];
859 /* name goes here */
860} __attribute__ ((__packed__));
861
862struct btrfs_timespec {
863 __le64 sec;
864 __le32 nsec;
865} __attribute__ ((__packed__));
866
867struct btrfs_inode_item {
868 /* nfs style generation number */
869 __le64 generation;
870 /* transid that last touched this inode */
871 __le64 transid;
872 __le64 size;
873 __le64 nbytes;
874 __le64 block_group;
875 __le32 nlink;
876 __le32 uid;
877 __le32 gid;
878 __le32 mode;
879 __le64 rdev;
880 __le64 flags;
881
882 /* modification sequence number for NFS */
883 __le64 sequence;
884
885 /*
886 * a little future expansion, for more than this we can
887 * just grow the inode item and version it
888 */
889 __le64 reserved[4];
890 struct btrfs_timespec atime;
891 struct btrfs_timespec ctime;
892 struct btrfs_timespec mtime;
893 struct btrfs_timespec otime;
894} __attribute__ ((__packed__));
895
896struct btrfs_dir_log_item {
897 __le64 end;
898} __attribute__ ((__packed__));
899
900struct btrfs_dir_item {
901 struct btrfs_disk_key location;
902 __le64 transid;
903 __le16 data_len;
904 __le16 name_len;
905 __u8 type;
906} __attribute__ ((__packed__));
907
908#define BTRFS_ROOT_SUBVOL_RDONLY (1ULL << 0)
909
910/*
911 * Internal in-memory flag that a subvolume has been marked for deletion but
912 * still visible as a directory
913 */
914#define BTRFS_ROOT_SUBVOL_DEAD (1ULL << 48)
915
916struct btrfs_root_item {
917 struct btrfs_inode_item inode;
918 __le64 generation;
919 __le64 root_dirid;
920 __le64 bytenr;
921 __le64 byte_limit;
922 __le64 bytes_used;
923 __le64 last_snapshot;
924 __le64 flags;
925 __le32 refs;
926 struct btrfs_disk_key drop_progress;
927 __u8 drop_level;
928 __u8 level;
929
930 /*
931 * The following fields appear after subvol_uuids+subvol_times
932 * were introduced.
933 */
934
935 /*
936 * This generation number is used to test if the new fields are valid
937 * and up to date while reading the root item. Every time the root item
938 * is written out, the "generation" field is copied into this field. If
939 * anyone ever mounted the fs with an older kernel, we will have
940 * mismatching generation values here and thus must invalidate the
941 * new fields. See btrfs_update_root and btrfs_find_last_root for
942 * details.
943 * the offset of generation_v2 is also used as the start for the memset
944 * when invalidating the fields.
945 */
946 __le64 generation_v2;
947 __u8 uuid[BTRFS_UUID_SIZE];
948 __u8 parent_uuid[BTRFS_UUID_SIZE];
949 __u8 received_uuid[BTRFS_UUID_SIZE];
950 __le64 ctransid; /* updated when an inode changes */
951 __le64 otransid; /* trans when created */
952 __le64 stransid; /* trans when sent. non-zero for received subvol */
953 __le64 rtransid; /* trans when received. non-zero for received subvol */
954 struct btrfs_timespec ctime;
955 struct btrfs_timespec otime;
956 struct btrfs_timespec stime;
957 struct btrfs_timespec rtime;
958 __le64 reserved[8]; /* for future */
959} __attribute__ ((__packed__));
960
961/*
962 * Btrfs root item used to be smaller than current size. The old format ends
963 * at where member generation_v2 is.
964 */
965static inline __u32 btrfs_legacy_root_item_size(void)
966{
967 return offsetof(struct btrfs_root_item, generation_v2);
968}
969
970/*
971 * this is used for both forward and backward root refs
972 */
973struct btrfs_root_ref {
974 __le64 dirid;
975 __le64 sequence;
976 __le16 name_len;
977} __attribute__ ((__packed__));
978
979struct btrfs_disk_balance_args {
980 /*
981 * profiles to operate on, single is denoted by
982 * BTRFS_AVAIL_ALLOC_BIT_SINGLE
983 */
984 __le64 profiles;
985
986 /*
987 * usage filter
988 * BTRFS_BALANCE_ARGS_USAGE with a single value means '0..N'
989 * BTRFS_BALANCE_ARGS_USAGE_RANGE - range syntax, min..max
990 */
991 union {
992 __le64 usage;
993 struct {
994 __le32 usage_min;
995 __le32 usage_max;
996 };
997 };
998
999 /* devid filter */
1000 __le64 devid;
1001
1002 /* devid subset filter [pstart..pend) */
1003 __le64 pstart;
1004 __le64 pend;
1005
1006 /* btrfs virtual address space subset filter [vstart..vend) */
1007 __le64 vstart;
1008 __le64 vend;
1009
1010 /*
1011 * profile to convert to, single is denoted by
1012 * BTRFS_AVAIL_ALLOC_BIT_SINGLE
1013 */
1014 __le64 target;
1015
1016 /* BTRFS_BALANCE_ARGS_* */
1017 __le64 flags;
1018
1019 /*
1020 * BTRFS_BALANCE_ARGS_LIMIT with value 'limit'
1021 * BTRFS_BALANCE_ARGS_LIMIT_RANGE - the extend version can use minimum
1022 * and maximum
1023 */
1024 union {
1025 __le64 limit;
1026 struct {
1027 __le32 limit_min;
1028 __le32 limit_max;
1029 };
1030 };
1031
1032 /*
1033 * Process chunks that cross stripes_min..stripes_max devices,
1034 * BTRFS_BALANCE_ARGS_STRIPES_RANGE
1035 */
1036 __le32 stripes_min;
1037 __le32 stripes_max;
1038
1039 __le64 unused[6];
1040} __attribute__ ((__packed__));
1041
1042/*
1043 * store balance parameters to disk so that balance can be properly
1044 * resumed after crash or unmount
1045 */
1046struct btrfs_balance_item {
1047 /* BTRFS_BALANCE_* */
1048 __le64 flags;
1049
1050 struct btrfs_disk_balance_args data;
1051 struct btrfs_disk_balance_args meta;
1052 struct btrfs_disk_balance_args sys;
1053
1054 __le64 unused[4];
1055} __attribute__ ((__packed__));
1056
1057enum {
1058 BTRFS_FILE_EXTENT_INLINE = 0,
1059 BTRFS_FILE_EXTENT_REG = 1,
1060 BTRFS_FILE_EXTENT_PREALLOC = 2,
1061 BTRFS_NR_FILE_EXTENT_TYPES = 3,
1062};
1063
1064struct btrfs_file_extent_item {
1065 /*
1066 * transaction id that created this extent
1067 */
1068 __le64 generation;
1069 /*
1070 * max number of bytes to hold this extent in ram
1071 * when we split a compressed extent we can't know how big
1072 * each of the resulting pieces will be. So, this is
1073 * an upper limit on the size of the extent in ram instead of
1074 * an exact limit.
1075 */
1076 __le64 ram_bytes;
1077
1078 /*
1079 * 32 bits for the various ways we might encode the data,
1080 * including compression and encryption. If any of these
1081 * are set to something a given disk format doesn't understand
1082 * it is treated like an incompat flag for reading and writing,
1083 * but not for stat.
1084 */
1085 __u8 compression;
1086 __u8 encryption;
1087 __le16 other_encoding; /* spare for later use */
1088
1089 /* are we inline data or a real extent? */
1090 __u8 type;
1091
1092 /*
1093 * disk space consumed by the extent, checksum blocks are included
1094 * in these numbers
1095 *
1096 * At this offset in the structure, the inline extent data start.
1097 */
1098 __le64 disk_bytenr;
1099 __le64 disk_num_bytes;
1100 /*
1101 * the logical offset in file blocks (no csums)
1102 * this extent record is for. This allows a file extent to point
1103 * into the middle of an existing extent on disk, sharing it
1104 * between two snapshots (useful if some bytes in the middle of the
1105 * extent have changed
1106 */
1107 __le64 offset;
1108 /*
1109 * the logical number of file blocks (no csums included). This
1110 * always reflects the size uncompressed and without encoding.
1111 */
1112 __le64 num_bytes;
1113
1114} __attribute__ ((__packed__));
1115
1116struct btrfs_csum_item {
1117 __u8 csum;
1118} __attribute__ ((__packed__));
1119
1120struct btrfs_dev_stats_item {
1121 /*
1122 * grow this item struct at the end for future enhancements and keep
1123 * the existing values unchanged
1124 */
1125 __le64 values[BTRFS_DEV_STAT_VALUES_MAX];
1126} __attribute__ ((__packed__));
1127
1128#define BTRFS_DEV_REPLACE_ITEM_CONT_READING_FROM_SRCDEV_MODE_ALWAYS 0
1129#define BTRFS_DEV_REPLACE_ITEM_CONT_READING_FROM_SRCDEV_MODE_AVOID 1
1130
1131struct btrfs_dev_replace_item {
1132 /*
1133 * grow this item struct at the end for future enhancements and keep
1134 * the existing values unchanged
1135 */
1136 __le64 src_devid;
1137 __le64 cursor_left;
1138 __le64 cursor_right;
1139 __le64 cont_reading_from_srcdev_mode;
1140
1141 __le64 replace_state;
1142 __le64 time_started;
1143 __le64 time_stopped;
1144 __le64 num_write_errors;
1145 __le64 num_uncorrectable_read_errors;
1146} __attribute__ ((__packed__));
1147
1148/* different types of block groups (and chunks) */
1149#define BTRFS_BLOCK_GROUP_DATA (1ULL << 0)
1150#define BTRFS_BLOCK_GROUP_SYSTEM (1ULL << 1)
1151#define BTRFS_BLOCK_GROUP_METADATA (1ULL << 2)
1152#define BTRFS_BLOCK_GROUP_RAID0 (1ULL << 3)
1153#define BTRFS_BLOCK_GROUP_RAID1 (1ULL << 4)
1154#define BTRFS_BLOCK_GROUP_DUP (1ULL << 5)
1155#define BTRFS_BLOCK_GROUP_RAID10 (1ULL << 6)
1156#define BTRFS_BLOCK_GROUP_RAID5 (1ULL << 7)
1157#define BTRFS_BLOCK_GROUP_RAID6 (1ULL << 8)
1158#define BTRFS_BLOCK_GROUP_RAID1C3 (1ULL << 9)
1159#define BTRFS_BLOCK_GROUP_RAID1C4 (1ULL << 10)
1160#define BTRFS_BLOCK_GROUP_RESERVED (BTRFS_AVAIL_ALLOC_BIT_SINGLE | \
1161 BTRFS_SPACE_INFO_GLOBAL_RSV)
1162
1163#define BTRFS_BLOCK_GROUP_TYPE_MASK (BTRFS_BLOCK_GROUP_DATA | \
1164 BTRFS_BLOCK_GROUP_SYSTEM | \
1165 BTRFS_BLOCK_GROUP_METADATA)
1166
1167#define BTRFS_BLOCK_GROUP_PROFILE_MASK (BTRFS_BLOCK_GROUP_RAID0 | \
1168 BTRFS_BLOCK_GROUP_RAID1 | \
1169 BTRFS_BLOCK_GROUP_RAID1C3 | \
1170 BTRFS_BLOCK_GROUP_RAID1C4 | \
1171 BTRFS_BLOCK_GROUP_RAID5 | \
1172 BTRFS_BLOCK_GROUP_RAID6 | \
1173 BTRFS_BLOCK_GROUP_DUP | \
1174 BTRFS_BLOCK_GROUP_RAID10)
1175#define BTRFS_BLOCK_GROUP_RAID56_MASK (BTRFS_BLOCK_GROUP_RAID5 | \
1176 BTRFS_BLOCK_GROUP_RAID6)
1177
1178#define BTRFS_BLOCK_GROUP_RAID1_MASK (BTRFS_BLOCK_GROUP_RAID1 | \
1179 BTRFS_BLOCK_GROUP_RAID1C3 | \
1180 BTRFS_BLOCK_GROUP_RAID1C4)
1181
1182/*
1183 * We need a bit for restriper to be able to tell when chunks of type
1184 * SINGLE are available. This "extended" profile format is used in
1185 * fs_info->avail_*_alloc_bits (in-memory) and balance item fields
1186 * (on-disk). The corresponding on-disk bit in chunk.type is reserved
1187 * to avoid remappings between two formats in future.
1188 */
1189#define BTRFS_AVAIL_ALLOC_BIT_SINGLE (1ULL << 48)
1190
1191/*
1192 * A fake block group type that is used to communicate global block reserve
1193 * size to userspace via the SPACE_INFO ioctl.
1194 */
1195#define BTRFS_SPACE_INFO_GLOBAL_RSV (1ULL << 49)
1196
1197#define BTRFS_EXTENDED_PROFILE_MASK (BTRFS_BLOCK_GROUP_PROFILE_MASK | \
1198 BTRFS_AVAIL_ALLOC_BIT_SINGLE)
1199
1200static inline __u64 chunk_to_extended(__u64 flags)
1201{
1202 if ((flags & BTRFS_BLOCK_GROUP_PROFILE_MASK) == 0)
1203 flags |= BTRFS_AVAIL_ALLOC_BIT_SINGLE;
1204
1205 return flags;
1206}
1207static inline __u64 extended_to_chunk(__u64 flags)
1208{
1209 return flags & ~BTRFS_AVAIL_ALLOC_BIT_SINGLE;
1210}
1211
1212struct btrfs_block_group_item {
1213 __le64 used;
1214 __le64 chunk_objectid;
1215 __le64 flags;
1216} __attribute__ ((__packed__));
1217
1218struct btrfs_free_space_info {
1219 __le32 extent_count;
1220 __le32 flags;
1221} __attribute__ ((__packed__));
1222
1223#define BTRFS_FREE_SPACE_USING_BITMAPS (1ULL << 0)
1224
1225#define BTRFS_QGROUP_LEVEL_SHIFT 48
1226static inline __u16 btrfs_qgroup_level(__u64 qgroupid)
1227{
1228 return (__u16)(qgroupid >> BTRFS_QGROUP_LEVEL_SHIFT);
1229}
1230
1231/*
1232 * is subvolume quota turned on?
1233 */
1234#define BTRFS_QGROUP_STATUS_FLAG_ON (1ULL << 0)
1235/*
1236 * RESCAN is set during the initialization phase
1237 */
1238#define BTRFS_QGROUP_STATUS_FLAG_RESCAN (1ULL << 1)
1239/*
1240 * Some qgroup entries are known to be out of date,
1241 * either because the configuration has changed in a way that
1242 * makes a rescan necessary, or because the fs has been mounted
1243 * with a non-qgroup-aware version.
1244 * Turning qouta off and on again makes it inconsistent, too.
1245 */
1246#define BTRFS_QGROUP_STATUS_FLAG_INCONSISTENT (1ULL << 2)
1247
1248/*
1249 * Whether or not this filesystem is using simple quotas. Not exactly the
1250 * incompat bit, because we support using simple quotas, disabling it, then
1251 * going back to full qgroup quotas.
1252 */
1253#define BTRFS_QGROUP_STATUS_FLAG_SIMPLE_MODE (1ULL << 3)
1254
1255#define BTRFS_QGROUP_STATUS_FLAGS_MASK (BTRFS_QGROUP_STATUS_FLAG_ON | \
1256 BTRFS_QGROUP_STATUS_FLAG_RESCAN | \
1257 BTRFS_QGROUP_STATUS_FLAG_INCONSISTENT | \
1258 BTRFS_QGROUP_STATUS_FLAG_SIMPLE_MODE)
1259
1260#define BTRFS_QGROUP_STATUS_VERSION 1
1261
1262struct btrfs_qgroup_status_item {
1263 __le64 version;
1264 /*
1265 * the generation is updated during every commit. As older
1266 * versions of btrfs are not aware of qgroups, it will be
1267 * possible to detect inconsistencies by checking the
1268 * generation on mount time
1269 */
1270 __le64 generation;
1271
1272 /* flag definitions see above */
1273 __le64 flags;
1274
1275 /*
1276 * only used during scanning to record the progress
1277 * of the scan. It contains a logical address
1278 */
1279 __le64 rescan;
1280
1281 /*
1282 * The generation when quotas were last enabled. Used by simple quotas to
1283 * avoid decrementing when freeing an extent that was written before
1284 * enable.
1285 *
1286 * Set only if flags contain BTRFS_QGROUP_STATUS_FLAG_SIMPLE_MODE.
1287 */
1288 __le64 enable_gen;
1289} __attribute__ ((__packed__));
1290
1291struct btrfs_qgroup_info_item {
1292 __le64 generation;
1293 __le64 rfer;
1294 __le64 rfer_cmpr;
1295 __le64 excl;
1296 __le64 excl_cmpr;
1297} __attribute__ ((__packed__));
1298
1299struct btrfs_qgroup_limit_item {
1300 /*
1301 * only updated when any of the other values change
1302 */
1303 __le64 flags;
1304 __le64 max_rfer;
1305 __le64 max_excl;
1306 __le64 rsv_rfer;
1307 __le64 rsv_excl;
1308} __attribute__ ((__packed__));
1309
1310struct btrfs_verity_descriptor_item {
1311 /* Size of the verity descriptor in bytes */
1312 __le64 size;
1313 /*
1314 * When we implement support for fscrypt, we will need to encrypt the
1315 * Merkle tree for encrypted verity files. These 128 bits are for the
1316 * eventual storage of an fscrypt initialization vector.
1317 */
1318 __le64 reserved[2];
1319 __u8 encryption;
1320} __attribute__ ((__packed__));
1321
1322#endif /* _BTRFS_CTREE_H_ */
1323

source code of linux/include/uapi/linux/btrfs_tree.h