]> git.ipfire.org Git - thirdparty/linux.git/blame - fs/ext4/ext4.h
ext4: port block device access to file
[thirdparty/linux.git] / fs / ext4 / ext4.h
CommitLineData
f5166768 1// SPDX-License-Identifier: GPL-2.0
ac27a0ec 2/*
3dcf5451 3 * ext4.h
ac27a0ec
DK
4 *
5 * Copyright (C) 1992, 1993, 1994, 1995
6 * Remy Card (card@masi.ibp.fr)
7 * Laboratoire MASI - Institut Blaise Pascal
8 * Universite Pierre et Marie Curie (Paris VI)
9 *
10 * from
11 *
12 * linux/include/linux/minix_fs.h
13 *
14 * Copyright (C) 1991, 1992 Linus Torvalds
15 */
16
3dcf5451
CH
17#ifndef _EXT4_H
18#define _EXT4_H
ac27a0ec 19
31d21d21 20#include <linux/refcount.h>
ac27a0ec 21#include <linux/types.h>
3a5b2ecd 22#include <linux/blkdev.h>
ac27a0ec 23#include <linux/magic.h>
2ccb5fb9 24#include <linux/jbd2.h>
60e58e0f 25#include <linux/quota.h>
d444c3c3
TT
26#include <linux/rwsem.h>
27#include <linux/rbtree.h>
28#include <linux/seqlock.h>
29#include <linux/mutex.h>
ca0faba0
TT
30#include <linux/timer.h>
31#include <linux/wait.h>
174cd4b1 32#include <linux/sched/signal.h>
ca0faba0
TT
33#include <linux/blockgroup_lock.h>
34#include <linux/percpu_counter.h>
efbed4dc 35#include <linux/ratelimit.h>
0441984a 36#include <crypto/hash.h>
a633f5a3 37#include <linux/falloc.h>
c8585c6f 38#include <linux/percpu-rwsem.h>
10c5db28 39#include <linux/fiemap.h>
4d92dc0f
BH
40#ifdef __KERNEL__
41#include <linux/compat.h>
42#endif
519fe1ba 43#include <uapi/linux/ext4.h>
c9de560d 44
734f0d24 45#include <linux/fscrypt.h>
c93d8f88 46#include <linux/fsverity.h>
734f0d24 47
072ebb3b
TT
48#include <linux/compiler.h>
49
ac27a0ec 50/*
8a35694e 51 * The fourth extended filesystem constants/structures
ac27a0ec
DK
52 */
53
a2821e34
AZ
54/*
55 * with AGGRESSIVE_CHECK allocator runs consistency checks over
56 * structures. these checks slow things down a lot
57 */
58#define AGGRESSIVE_CHECK__
59
60/*
61 * with DOUBLE_CHECK defined mballoc creates persistent in-core
62 * bitmaps, maintains and uses them to check for double allocations
63 */
64#define DOUBLE_CHECK__
65
ac27a0ec 66/*
617ba13b 67 * Define EXT4FS_DEBUG to produce debug messages
ac27a0ec 68 */
617ba13b 69#undef EXT4FS_DEBUG
ac27a0ec 70
ac27a0ec
DK
71/*
72 * Debug code
73 */
617ba13b
MC
74#ifdef EXT4FS_DEBUG
75#define ext4_debug(f, a...) \
ac27a0ec 76 do { \
af5bc92d 77 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
4db9c54a 78 __FILE__, __LINE__, __func__); \
af5bc92d 79 printk(KERN_DEBUG f, ## a); \
ac27a0ec
DK
80 } while (0)
81#else
ace36ad4 82#define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
ac27a0ec
DK
83#endif
84
70aa1554
RH
85 /*
86 * Turn on EXT_DEBUG to enable ext4_ext_show_path/leaf/move in extents.c
87 */
88#define EXT_DEBUG__
89
4a092d73 90/*
70aa1554 91 * Dynamic printk for controlled extents debugging.
4a092d73 92 */
70aa1554
RH
93#ifdef CONFIG_EXT4_DEBUG
94#define ext_debug(ino, fmt, ...) \
95 pr_debug("[%s/%d] EXT4-fs (%s): ino %lu: (%s, %d): %s:" fmt, \
96 current->comm, task_pid_nr(current), \
97 ino->i_sb->s_id, ino->i_ino, __FILE__, __LINE__, \
98 __func__, ##__VA_ARGS__)
4a092d73 99#else
70aa1554 100#define ext_debug(ino, fmt, ...) no_printk(fmt, ##__VA_ARGS__)
4a092d73
TT
101#endif
102
837c23fb
CX
103#define ASSERT(assert) \
104do { \
105 if (unlikely(!(assert))) { \
106 printk(KERN_EMERG \
107 "Assertion failure in %s() at %s:%d: '%s'\n", \
108 __func__, __FILE__, __LINE__, #assert); \
109 BUG(); \
110 } \
111} while (0)
112
d444c3c3
TT
113/* data type for block offset of block group */
114typedef int ext4_grpblk_t;
115
116/* data type for filesystem-wide blocks number */
117typedef unsigned long long ext4_fsblk_t;
118
119/* data type for file logical block number */
120typedef __u32 ext4_lblk_t;
121
122/* data type for block group number */
123typedef unsigned int ext4_group_t;
124
331573fe
NJ
125enum SHIFT_DIRECTION {
126 SHIFT_LEFT = 0,
127 SHIFT_RIGHT,
128};
129
4eb7a4a1 130/*
4c0cfebd
TT
131 * For each criteria, mballoc has slightly different way of finding
132 * the required blocks nad usually, higher the criteria the slower the
133 * allocation. We start at lower criterias and keep falling back to
134 * higher ones if we are not able to find any blocks. Lower (earlier)
135 * criteria are faster.
4eb7a4a1
OM
136 */
137enum criteria {
f52f3d2b 138 /*
4c0cfebd
TT
139 * Used when number of blocks needed is a power of 2. This
140 * doesn't trigger any disk IO except prefetch and is the
141 * fastest criteria.
f52f3d2b
OM
142 */
143 CR_POWER2_ALIGNED,
144
145 /*
4c0cfebd
TT
146 * Tries to lookup in-memory data structures to find the most
147 * suitable group that satisfies goal request. No disk IO
148 * except block prefetch.
f52f3d2b
OM
149 */
150 CR_GOAL_LEN_FAST,
151
152 /*
4c0cfebd
TT
153 * Same as CR_GOAL_LEN_FAST but is allowed to reduce the goal
154 * length to the best available length for faster allocation.
f52f3d2b
OM
155 */
156 CR_BEST_AVAIL_LEN,
157
158 /*
4c0cfebd
TT
159 * Reads each block group sequentially, performing disk IO if
160 * necessary, to find find_suitable block group. Tries to
161 * allocate goal length but might trim the request if nothing
162 * is found after enough tries.
f52f3d2b
OM
163 */
164 CR_GOAL_LEN_SLOW,
165
166 /*
4c0cfebd
TT
167 * Finds the first free set of blocks and allocates
168 * those. This is only used in rare cases when
169 * CR_GOAL_LEN_SLOW also fails to allocate anything.
f52f3d2b
OM
170 */
171 CR_ANY_FREE,
4c0cfebd
TT
172
173 /*
174 * Number of criterias defined.
175 */
176 EXT4_MB_NUM_CRS
4eb7a4a1
OM
177};
178
296c355c 179/*
60e6679e 180 * Flags used in mballoc's allocation_context flags field.
296c355c
TT
181 *
182 * Also used to show what's going on for debugging purposes when the
183 * flag field is exported via the traceport interface
184 */
d444c3c3 185
c9de560d 186/* prefer goal again. length */
0ef90db9 187#define EXT4_MB_HINT_MERGE 0x0001
c9de560d 188/* blocks already reserved */
0ef90db9 189#define EXT4_MB_HINT_RESERVED 0x0002
c9de560d 190/* metadata is being allocated */
0ef90db9 191#define EXT4_MB_HINT_METADATA 0x0004
c9de560d 192/* first blocks in the file */
0ef90db9 193#define EXT4_MB_HINT_FIRST 0x0008
c9de560d 194/* search for the best chunk */
0ef90db9 195#define EXT4_MB_HINT_BEST 0x0010
c9de560d 196/* data is being allocated */
0ef90db9 197#define EXT4_MB_HINT_DATA 0x0020
c9de560d 198/* don't preallocate (for tails) */
0ef90db9 199#define EXT4_MB_HINT_NOPREALLOC 0x0040
c9de560d 200/* allocate for locality group */
0ef90db9 201#define EXT4_MB_HINT_GROUP_ALLOC 0x0080
c9de560d 202/* allocate goal blocks or none */
0ef90db9 203#define EXT4_MB_HINT_GOAL_ONLY 0x0100
c9de560d 204/* goal is meaningful */
0ef90db9 205#define EXT4_MB_HINT_TRY_GOAL 0x0200
d2a17637 206/* blocks already pre-reserved by delayed allocation */
0ef90db9 207#define EXT4_MB_DELALLOC_RESERVED 0x0400
4ba74d00
TT
208/* We are doing stream allocation */
209#define EXT4_MB_STREAM_ALLOC 0x0800
55f020db
AH
210/* Use reserved root blocks if needed */
211#define EXT4_MB_USE_ROOT_BLOCKS 0x1000
27dd4385
LC
212/* Use blocks from reserved pool */
213#define EXT4_MB_USE_RESERVED 0x2000
99377830
RH
214/* Do strict check for free blocks while retrying block allocation */
215#define EXT4_MB_STRICT_CHECK 0x4000
196e402a 216/* Large fragment size list lookup succeeded at least once for cr = 0 */
f52f3d2b 217#define EXT4_MB_CR_POWER2_ALIGNED_OPTIMIZED 0x8000
196e402a 218/* Avg fragment size rb tree lookup succeeded at least once for cr = 1 */
f52f3d2b 219#define EXT4_MB_CR_GOAL_LEN_FAST_OPTIMIZED 0x00010000
7e170922 220/* Avg fragment size rb tree lookup succeeded at least once for cr = 1.5 */
f52f3d2b 221#define EXT4_MB_CR_BEST_AVAIL_LEN_OPTIMIZED 0x00020000
7e170922 222
c9de560d
AT
223struct ext4_allocation_request {
224 /* target inode for block we're allocating */
225 struct inode *inode;
726447d8
ES
226 /* how many blocks we want to allocate */
227 unsigned int len;
c9de560d
AT
228 /* logical block in target inode */
229 ext4_lblk_t logical;
c9de560d
AT
230 /* the closest logical allocated block to the left */
231 ext4_lblk_t lleft;
c9de560d
AT
232 /* the closest logical allocated block to the right */
233 ext4_lblk_t lright;
726447d8
ES
234 /* phys. target (a hint) */
235 ext4_fsblk_t goal;
236 /* phys. block for the closest logical allocated block to the left */
237 ext4_fsblk_t pleft;
238 /* phys. block for the closest logical allocated block to the right */
c9de560d 239 ext4_fsblk_t pright;
c9de560d 240 /* flags. see above EXT4_MB_HINT_* */
498e5f24 241 unsigned int flags;
c9de560d
AT
242};
243
e35fd660
TT
244/*
245 * Logical to physical block mapping, used by ext4_map_blocks()
246 *
247 * This structure is used to pass requests into ext4_map_blocks() as
248 * well as to store the information returned by ext4_map_blocks(). It
249 * takes less room on the stack than a struct buffer_head.
250 */
6db07461
RH
251#define EXT4_MAP_NEW BIT(BH_New)
252#define EXT4_MAP_MAPPED BIT(BH_Mapped)
253#define EXT4_MAP_UNWRITTEN BIT(BH_Unwritten)
254#define EXT4_MAP_BOUNDARY BIT(BH_Boundary)
e35fd660 255#define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
cbd7584e 256 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
e35fd660
TT
257
258struct ext4_map_blocks {
259 ext4_fsblk_t m_pblk;
260 ext4_lblk_t m_lblk;
261 unsigned int m_len;
262 unsigned int m_flags;
263};
264
7727ae52 265/*
266 * Block validity checking, system zone rbtree.
267 */
268struct ext4_system_blocks {
269 struct rb_root root;
270 struct rcu_head rcu;
271};
272
bd2d0210
TT
273/*
274 * Flags for ext4_io_end->flags
275 */
276#define EXT4_IO_END_UNWRITTEN 0x0001
bd2d0210 277
c8cc8816
RH
278struct ext4_io_end_vec {
279 struct list_head list; /* list of io_end_vec */
280 loff_t offset; /* offset in the file */
281 ssize_t size; /* size of the extent */
282};
283
4188188b 284/*
556615dc 285 * For converting unwritten extents on a work queue. 'handle' is used for
6b523df4 286 * buffered writeback.
4188188b 287 */
0031462b 288typedef struct ext4_io_end {
5b3ff237 289 struct list_head list; /* per-file finished IO list */
6b523df4
JK
290 handle_t *handle; /* handle reserved for extent
291 * conversion */
0031462b 292 struct inode *inode; /* file being written to */
b0857d30
JK
293 struct bio *bio; /* Linked list of completed
294 * bios covering the extent */
8d5d02e6 295 unsigned int flag; /* unwritten or not */
31d21d21 296 refcount_t count; /* reference counter */
c8cc8816 297 struct list_head list_vec; /* list of ext4_io_end_vec */
0031462b
MC
298} ext4_io_end_t;
299
bd2d0210 300struct ext4_io_submit {
5a33911f 301 struct writeback_control *io_wbc;
bd2d0210
TT
302 struct bio *io_bio;
303 ext4_io_end_t *io_end;
bd2d0210
TT
304 sector_t io_next_block;
305};
306
ac27a0ec
DK
307/*
308 * Special inodes numbers
309 */
617ba13b
MC
310#define EXT4_BAD_INO 1 /* Bad blocks inode */
311#define EXT4_ROOT_INO 2 /* Root inode */
ae812306
AK
312#define EXT4_USR_QUOTA_INO 3 /* User quota inode */
313#define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
617ba13b
MC
314#define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
315#define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
316#define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
317#define EXT4_JOURNAL_INO 8 /* Journal inode */
ac27a0ec 318
617ba13b
MC
319/* First non-reserved inode for old ext4 filesystems */
320#define EXT4_GOOD_OLD_FIRST_INO 11
ac27a0ec
DK
321
322/*
323 * Maximal count of links to a file
324 */
f8628a14 325#define EXT4_LINK_MAX 65000
ac27a0ec
DK
326
327/*
328 * Macro-instructions used to manage several block sizes
329 */
617ba13b 330#define EXT4_MIN_BLOCK_SIZE 1024
afc7cbca
TS
331#define EXT4_MAX_BLOCK_SIZE 65536
332#define EXT4_MIN_BLOCK_LOG_SIZE 10
fb1813f4 333#define EXT4_MAX_BLOCK_LOG_SIZE 16
9e47a4c9 334#define EXT4_MAX_CLUSTER_LOG_SIZE 30
ac27a0ec 335#ifdef __KERNEL__
617ba13b 336# define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
ac27a0ec 337#else
617ba13b 338# define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
ac27a0ec 339#endif
af5bc92d 340#define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
281b5995
TT
341#define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
342 EXT4_SB(s)->s_cluster_bits)
ac27a0ec 343#ifdef __KERNEL__
617ba13b 344# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
281b5995 345# define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
ac27a0ec 346#else
617ba13b 347# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
ac27a0ec
DK
348#endif
349#ifdef __KERNEL__
617ba13b
MC
350#define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
351#define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
352#define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
ac27a0ec 353#else
617ba13b
MC
354#define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
355 EXT4_GOOD_OLD_INODE_SIZE : \
ac27a0ec 356 (s)->s_inode_size)
617ba13b
MC
357#define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
358 EXT4_GOOD_OLD_FIRST_INO : \
ac27a0ec
DK
359 (s)->s_first_ino)
360#endif
a2df2a63 361#define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
518eaa63
FF
362#define EXT4_MAX_BLOCKS(size, offset, blkbits) \
363 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \
364 blkbits))
ac27a0ec 365
d5b8f310
TT
366/* Translate a block number to a cluster number */
367#define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
368/* Translate a cluster number to a block number */
369#define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
370/* Translate # of blks to # of clusters */
371#define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
372 (sbi)->s_cluster_bits)
f5a44db5
TT
373/* Mask out the low bits to get the starting block of the cluster */
374#define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
375 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
376#define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
377 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
8fcc3a58
EW
378/* Fill in the low bits to get the last block of the cluster */
379#define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) | \
380 ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1))
f5a44db5
TT
381/* Get the cluster offset */
382#define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
383 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
384#define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
385 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
d5b8f310 386
ac27a0ec
DK
387/*
388 * Structure of a blocks group descriptor
389 */
617ba13b 390struct ext4_group_desc
ac27a0ec 391{
3a14589c 392 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
5272f837
AK
393 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
394 __le32 bg_inode_table_lo; /* Inodes table block */
560671a0
AK
395 __le16 bg_free_blocks_count_lo;/* Free blocks count */
396 __le16 bg_free_inodes_count_lo;/* Free inodes count */
397 __le16 bg_used_dirs_count_lo; /* Directories count */
717d50e4 398 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
e6153918
DW
399 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
400 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
401 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
560671a0 402 __le16 bg_itable_unused_lo; /* Unused inodes count */
717d50e4 403 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
8fadc143
AR
404 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
405 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
406 __le32 bg_inode_table_hi; /* Inodes table block MSB */
91b51a01
CL
407 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
408 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
409 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
560671a0 410 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
e6153918
DW
411 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
412 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
413 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
414 __u32 bg_reserved;
ac27a0ec
DK
415};
416
41a246d1
DW
417#define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
418 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
419 sizeof(__le16))
420#define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
421 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
422 sizeof(__le16))
423
772cb7c8
JS
424/*
425 * Structure of a flex block group info
426 */
427
428struct flex_groups {
90ba983f
TT
429 atomic64_t free_clusters;
430 atomic_t free_inodes;
431 atomic_t used_dirs;
772cb7c8
JS
432};
433
717d50e4
AD
434#define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
435#define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
436#define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
437
ac27a0ec
DK
438/*
439 * Macro-instructions used to manage group descriptors
440 */
0d1ee42f 441#define EXT4_MIN_DESC_SIZE 32
8fadc143 442#define EXT4_MIN_DESC_SIZE_64BIT 64
0d1ee42f
AR
443#define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
444#define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
ac27a0ec 445#ifdef __KERNEL__
617ba13b 446# define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
281b5995 447# define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
617ba13b
MC
448# define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
449# define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
450# define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
ac27a0ec 451#else
617ba13b 452# define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
0d1ee42f 453# define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
617ba13b 454# define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
ac27a0ec
DK
455#endif
456
457/*
458 * Constants relative to the data blocks
459 */
617ba13b
MC
460#define EXT4_NDIR_BLOCKS 12
461#define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
462#define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
463#define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
464#define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
ac27a0ec
DK
465
466/*
467 * Inode flags
468 */
617ba13b
MC
469#define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
470#define EXT4_UNRM_FL 0x00000002 /* Undelete */
471#define EXT4_COMPR_FL 0x00000004 /* Compress file */
472#define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
473#define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
474#define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
475#define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
476#define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
ac27a0ec 477/* Reserved for compression usage... */
617ba13b
MC
478#define EXT4_DIRTY_FL 0x00000100
479#define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
480#define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
3edc18d8
TT
481 /* nb: was previously EXT2_ECOMPR_FL */
482#define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */
ac27a0ec 483/* End compression flags --- maybe not all used */
617ba13b
MC
484#define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
485#define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
486#define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
487#define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
488#define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
489#define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
8180a562 490#define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
a86c6181 491#define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
c93d8f88 492#define EXT4_VERITY_FL 0x00100000 /* Verity protected inode */
f710b4b9 493#define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
4337ecd1 494/* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */
b383a73f
IW
495
496#define EXT4_DAX_FL 0x02000000 /* Inode is DAX */
497
67cf5b09 498#define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
8b4953e1 499#define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
de8ff14c 500#define EXT4_CASEFOLD_FL 0x40000000 /* Casefolded directory */
8180a562 501#define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
ac27a0ec 502
2a12e147
EB
503/* User modifiable flags */
504#define EXT4_FL_USER_MODIFIABLE (EXT4_SECRM_FL | \
505 EXT4_UNRM_FL | \
506 EXT4_COMPR_FL | \
507 EXT4_SYNC_FL | \
508 EXT4_IMMUTABLE_FL | \
509 EXT4_APPEND_FL | \
510 EXT4_NODUMP_FL | \
511 EXT4_NOATIME_FL | \
512 EXT4_JOURNAL_DATA_FL | \
513 EXT4_NOTAIL_FL | \
514 EXT4_DIRSYNC_FL | \
515 EXT4_TOPDIR_FL | \
516 EXT4_EXTENTS_FL | \
517 0x00400000 /* EXT4_EOFBLOCKS_FL */ | \
518 EXT4_DAX_FL | \
519 EXT4_PROJINHERIT_FL | \
520 EXT4_CASEFOLD_FL)
521
522/* User visible flags */
523#define EXT4_FL_USER_VISIBLE (EXT4_FL_USER_MODIFIABLE | \
524 EXT4_DIRTY_FL | \
525 EXT4_COMPRBLK_FL | \
526 EXT4_NOCOMPR_FL | \
527 EXT4_ENCRYPT_FL | \
528 EXT4_INDEX_FL | \
529 EXT4_VERITY_FL | \
530 EXT4_INLINE_DATA_FL)
ac27a0ec 531
8fa43a81
DG
532/* Flags that should be inherited by new inodes from their parent. */
533#define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
1cd9f097 534 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
8fa43a81 535 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
040cb378 536 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
b383a73f
IW
537 EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL |\
538 EXT4_DAX_FL)
8fa43a81 539
2dc6b0d4 540/* Flags that are appropriate for regular files (all but dir-specific ones). */
7ddf79a1
WS
541#define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\
542 EXT4_PROJINHERIT_FL))
2dc6b0d4
DG
543
544/* Flags that are appropriate for non-directories/regular files. */
545#define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
546
abdc644e 547/* The only flags that should be swapped */
548#define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL)
549
b383a73f
IW
550/* Flags which are mutually exclusive to DAX */
551#define EXT4_DAX_MUT_EXCL (EXT4_VERITY_FL | EXT4_ENCRYPT_FL |\
aa2f7792 552 EXT4_JOURNAL_DATA_FL | EXT4_INLINE_DATA_FL)
b383a73f 553
2dc6b0d4
DG
554/* Mask out flags that are inappropriate for the given type of inode. */
555static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
556{
557 if (S_ISDIR(mode))
558 return flags;
559 else if (S_ISREG(mode))
560 return flags & EXT4_REG_FLMASK;
561 else
562 return flags & EXT4_OTHER_FLMASK;
563}
564
12e9b892
DM
565/*
566 * Inode flags used for atomic set/get
567 */
568enum {
569 EXT4_INODE_SECRM = 0, /* Secure deletion */
570 EXT4_INODE_UNRM = 1, /* Undelete */
571 EXT4_INODE_COMPR = 2, /* Compress file */
572 EXT4_INODE_SYNC = 3, /* Synchronous updates */
573 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
574 EXT4_INODE_APPEND = 5, /* writes to file may only append */
575 EXT4_INODE_NODUMP = 6, /* do not dump file */
576 EXT4_INODE_NOATIME = 7, /* do not update atime */
577/* Reserved for compression usage... */
578 EXT4_INODE_DIRTY = 8,
579 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
580 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
f542fbe8 581 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */
12e9b892
DM
582/* End compression flags --- maybe not all used */
583 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
584 EXT4_INODE_IMAGIC = 13, /* AFS directory */
585 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
586 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
587 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
588 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
589 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
590 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
c93d8f88 591 EXT4_INODE_VERITY = 20, /* Verity protected inode */
12e9b892 592 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
4337ecd1 593/* 22 was formerly EXT4_INODE_EOFBLOCKS */
b383a73f 594 EXT4_INODE_DAX = 25, /* Inode is DAX */
67cf5b09 595 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
8b4953e1 596 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */
de8ff14c 597 EXT4_INODE_CASEFOLD = 30, /* Casefolded directory */
12e9b892
DM
598 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
599};
600
12e9b892 601/*
9a4c8019
CM
602 * Since it's pretty easy to mix up bit numbers and hex values, we use a
603 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
604 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
605 * any extra space in the compiled kernel image, otherwise, the build will fail.
606 * It's important that these values are the same, since we are using
607 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
608 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
609 * values found in ext2, ext3 and ext4 filesystems, and of course the values
610 * defined in e2fsprogs.
12e9b892
DM
611 *
612 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
613 */
3bf678a0 614#define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1U << EXT4_INODE_##FLAG))
9a4c8019
CM
615#define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
616
12e9b892
DM
617static inline void ext4_check_flag_values(void)
618{
619 CHECK_FLAG_VALUE(SECRM);
620 CHECK_FLAG_VALUE(UNRM);
621 CHECK_FLAG_VALUE(COMPR);
622 CHECK_FLAG_VALUE(SYNC);
623 CHECK_FLAG_VALUE(IMMUTABLE);
624 CHECK_FLAG_VALUE(APPEND);
625 CHECK_FLAG_VALUE(NODUMP);
626 CHECK_FLAG_VALUE(NOATIME);
627 CHECK_FLAG_VALUE(DIRTY);
628 CHECK_FLAG_VALUE(COMPRBLK);
629 CHECK_FLAG_VALUE(NOCOMPR);
3edc18d8 630 CHECK_FLAG_VALUE(ENCRYPT);
12e9b892
DM
631 CHECK_FLAG_VALUE(INDEX);
632 CHECK_FLAG_VALUE(IMAGIC);
633 CHECK_FLAG_VALUE(JOURNAL_DATA);
634 CHECK_FLAG_VALUE(NOTAIL);
635 CHECK_FLAG_VALUE(DIRSYNC);
636 CHECK_FLAG_VALUE(TOPDIR);
637 CHECK_FLAG_VALUE(HUGE_FILE);
638 CHECK_FLAG_VALUE(EXTENTS);
c93d8f88 639 CHECK_FLAG_VALUE(VERITY);
12e9b892 640 CHECK_FLAG_VALUE(EA_INODE);
67cf5b09 641 CHECK_FLAG_VALUE(INLINE_DATA);
8b4953e1 642 CHECK_FLAG_VALUE(PROJINHERIT);
de8ff14c 643 CHECK_FLAG_VALUE(CASEFOLD);
12e9b892
DM
644 CHECK_FLAG_VALUE(RESERVED);
645}
646
4d92dc0f
BH
647#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
648struct compat_ext4_new_group_input {
649 u32 group;
650 compat_u64 block_bitmap;
651 compat_u64 inode_bitmap;
652 compat_u64 inode_table;
653 u32 blocks_count;
654 u16 reserved_blocks;
655 u16 unused;
656};
657#endif
658
617ba13b
MC
659/* The struct ext4_new_group_input in kernel space, with free_blocks_count */
660struct ext4_new_group_data {
ac27a0ec 661 __u32 group;
bd81d8ee
LV
662 __u64 block_bitmap;
663 __u64 inode_bitmap;
664 __u64 inode_table;
ac27a0ec
DK
665 __u32 blocks_count;
666 __u16 reserved_blocks;
d77147ff 667 __u16 mdata_blocks;
668 __u32 free_clusters_count;
ac27a0ec
DK
669};
670
33afdcc5
YY
671/* Indexes used to index group tables in ext4_new_group_data */
672enum {
673 BLOCK_BITMAP = 0, /* block bitmap */
674 INODE_BITMAP, /* inode bitmap */
675 INODE_TABLE, /* inode tables */
676 GROUP_TABLE_COUNT,
677};
678
a2df2a63 679/*
79e83036 680 * Flags used by ext4_map_blocks()
a2df2a63 681 */
556615dc 682 /* Allocate any needed blocks and/or convert an unwritten
c2177057 683 extent to be an initialized ext4 */
2ac3b6e0 684#define EXT4_GET_BLOCKS_CREATE 0x0001
556615dc
LC
685 /* Request the creation of an unwritten extent */
686#define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002
687#define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\
c2177057 688 EXT4_GET_BLOCKS_CREATE)
27dd4385
LC
689 /* Caller is from the delayed allocation writeout path
690 * finally doing the actual allocation of delayed blocks */
03f5d8bc 691#define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
0031462b 692 /* caller is from the direct IO path, request to creation of an
556615dc 693 unwritten extents if not allocated, split the unwritten
0031462b 694 extent if blocks has been preallocated already*/
c7064ef1 695#define EXT4_GET_BLOCKS_PRE_IO 0x0008
1296cc85 696#define EXT4_GET_BLOCKS_CONVERT 0x0010
c7064ef1 697#define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
556615dc 698 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
c7064ef1
JZ
699 /* Convert extent to initialized after IO complete */
700#define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
556615dc 701 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
27dd4385
LC
702 /* Eventual metadata allocation (due to growing extent tree)
703 * should not fail, so try to use reserved blocks for that.*/
704#define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
556b27ab
VH
705 /* Don't normalize allocation size (used for fallocate) */
706#define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
b8a86845 707 /* Convert written extents to unwritten */
2dcba478 708#define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100
c86d8db3
JK
709 /* Write zeros to newly created written extents */
710#define EXT4_GET_BLOCKS_ZERO 0x0200
711#define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\
712 EXT4_GET_BLOCKS_ZERO)
ee0876bc
JK
713 /* Caller will submit data before dropping transaction handle. This
714 * allows jbd2 to avoid submitting data before commit. */
715#define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400
9558cf14
ZY
716 /* Caller is in the atomic contex, find extent if it has been cached */
717#define EXT4_GET_BLOCKS_CACHED_NOWAIT 0x0800
ac27a0ec 718
107a7bd3 719/*
7869a4a6 720 * The bit position of these flags must not overlap with any of the
ed8a1a76 721 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(),
107a7bd3 722 * read_extent_tree_block(), ext4_split_extent_at(),
7869a4a6
TT
723 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
724 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
725 * caching the extents when reading from the extent tree while a
726 * truncate or punch hole operation is in progress.
107a7bd3 727 */
d26e2c4d
TT
728#define EXT4_EX_NOCACHE 0x40000000
729#define EXT4_EX_FORCE_CACHE 0x20000000
73c384c0 730#define EXT4_EX_NOFAIL 0x10000000
107a7bd3 731
e6362609
TT
732/*
733 * Flags used by ext4_free_blocks
734 */
e7093f0d
PK
735#define EXT4_FREE_BLOCKS_METADATA 0x0001
736#define EXT4_FREE_BLOCKS_FORGET 0x0002
737#define EXT4_FREE_BLOCKS_VALIDATED 0x0004
738#define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
84130193
TT
739#define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
740#define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
9fe67149 741#define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040
e6362609 742
899ad0ce 743#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
ac27a0ec
DK
744/*
745 * ioctl commands in 32 bit emulation
746 */
617ba13b
MC
747#define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
748#define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
749#define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
750#define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
751#define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
4d92dc0f 752#define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
617ba13b
MC
753#define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
754#define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
899ad0ce 755#endif
ac27a0ec 756
b595076a 757/* Max physical block we can address w/o extents */
fb0a387d
ES
758#define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
759
bcd8e91f 760/* Max logical block we can support */
9f44eda1 761#define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFE
bcd8e91f 762
ac27a0ec
DK
763/*
764 * Structure of an inode on the disk
765 */
617ba13b 766struct ext4_inode {
ac27a0ec
DK
767 __le16 i_mode; /* File mode */
768 __le16 i_uid; /* Low 16 bits of Owner Uid */
a48380f7 769 __le32 i_size_lo; /* Size in bytes */
ac27a0ec 770 __le32 i_atime; /* Access time */
ef7f3835 771 __le32 i_ctime; /* Inode Change time */
ac27a0ec
DK
772 __le32 i_mtime; /* Modification time */
773 __le32 i_dtime; /* Deletion Time */
774 __le16 i_gid; /* Low 16 bits of Group Id */
775 __le16 i_links_count; /* Links count */
0fc1b451 776 __le32 i_blocks_lo; /* Blocks count */
ac27a0ec
DK
777 __le32 i_flags; /* File flags */
778 union {
779 struct {
25ec56b5 780 __le32 l_i_version;
ac27a0ec
DK
781 } linux1;
782 struct {
783 __u32 h_i_translator;
784 } hurd1;
785 struct {
786 __u32 m_i_reserved1;
787 } masix1;
788 } osd1; /* OS dependent 1 */
617ba13b 789 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
ac27a0ec 790 __le32 i_generation; /* File version (for NFS) */
7973c0c1 791 __le32 i_file_acl_lo; /* File ACL */
a48380f7 792 __le32 i_size_high;
f077d0d7 793 __le32 i_obso_faddr; /* Obsoleted fragment address */
ac27a0ec
DK
794 union {
795 struct {
0fc1b451 796 __le16 l_i_blocks_high; /* were l_i_reserved1 */
a1ddeb7e 797 __le16 l_i_file_acl_high;
8c55e204 798 __le16 l_i_uid_high; /* these 2 fields */
ac27a0ec 799 __le16 l_i_gid_high; /* were reserved2[0] */
e6153918
DW
800 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
801 __le16 l_i_reserved;
ac27a0ec
DK
802 } linux2;
803 struct {
f077d0d7 804 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
ac27a0ec
DK
805 __u16 h_i_mode_high;
806 __u16 h_i_uid_high;
807 __u16 h_i_gid_high;
808 __u32 h_i_author;
809 } hurd2;
810 struct {
f077d0d7 811 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
a1ddeb7e 812 __le16 m_i_file_acl_high;
ac27a0ec
DK
813 __u32 m_i_reserved2[2];
814 } masix2;
815 } osd2; /* OS dependent 2 */
816 __le16 i_extra_isize;
e6153918 817 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
ef7f3835
KS
818 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
819 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
820 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
821 __le32 i_crtime; /* File Creation time */
822 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
25ec56b5 823 __le32 i_version_hi; /* high 32 bits for 64-bit version */
8b4953e1 824 __le32 i_projid; /* Project ID */
ac27a0ec
DK
825};
826
ef7f3835
KS
827#define EXT4_EPOCH_BITS 2
828#define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
829#define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
830
831/*
832 * Extended fields will fit into an inode if the filesystem was formatted
833 * with large inodes (-I 256 or larger) and there are not currently any EAs
834 * consuming all of the available space. For new inodes we always reserve
835 * enough space for the kernel's known extended fields, but for inodes
836 * created with an old kernel this might not have been the case. None of
837 * the extended inode fields is critical for correct filesystem operation.
838 * This macro checks if a certain field fits in the inode. Note that
839 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
840 */
841#define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
842 ((offsetof(typeof(*ext4_inode), field) + \
843 sizeof((ext4_inode)->field)) \
844 <= (EXT4_GOOD_OLD_INODE_SIZE + \
845 (einode)->i_extra_isize)) \
846
a4dad1ae
DT
847/*
848 * We use an encoding that preserves the times for extra epoch "00":
849 *
850 * extra msb of adjust for signed
851 * epoch 32-bit 32-bit tv_sec to
852 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
853 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
854 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
855 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
856 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
857 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
858 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
859 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
860 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
861 *
862 * Note that previous versions of the kernel on 64-bit systems would
863 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
864 * 1970. e2fsck will correct this, assuming that it is run on the
865 * affected filesystem before 2242.
866 */
867
1bc33893 868static inline __le32 ext4_encode_extra_time(struct timespec64 ts)
ef7f3835 869{
1bc33893
JL
870 u32 extra = ((ts.tv_sec - (s32)ts.tv_sec) >> 32) & EXT4_EPOCH_MASK;
871 return cpu_to_le32(extra | (ts.tv_nsec << EXT4_EPOCH_BITS));
ef7f3835
KS
872}
873
1bc33893
JL
874static inline struct timespec64 ext4_decode_extra_time(__le32 base,
875 __le32 extra)
ef7f3835 876{
1bc33893
JL
877 struct timespec64 ts = { .tv_sec = (signed)le32_to_cpu(base) };
878
cd2d9922 879 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK)))
1bc33893
JL
880 ts.tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
881 ts.tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
882 return ts;
ef7f3835
KS
883}
884
1bc33893 885#define EXT4_INODE_SET_XTIME_VAL(xtime, inode, raw_inode, ts) \
95582b00 886do { \
1bc33893
JL
887 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \
888 (raw_inode)->xtime = cpu_to_le32((ts).tv_sec); \
889 (raw_inode)->xtime ## _extra = ext4_encode_extra_time(ts); \
890 } else \
891 (raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (ts).tv_sec, S32_MIN, S32_MAX)); \
ef7f3835
KS
892} while (0)
893
b898ab23
JL
894#define EXT4_INODE_SET_ATIME(inode, raw_inode) \
895 EXT4_INODE_SET_XTIME_VAL(i_atime, inode, raw_inode, inode_get_atime(inode))
1bc33893 896
b898ab23
JL
897#define EXT4_INODE_SET_MTIME(inode, raw_inode) \
898 EXT4_INODE_SET_XTIME_VAL(i_mtime, inode, raw_inode, inode_get_mtime(inode))
899
900#define EXT4_INODE_SET_CTIME(inode, raw_inode) \
1bc33893
JL
901 EXT4_INODE_SET_XTIME_VAL(i_ctime, inode, raw_inode, inode_get_ctime(inode))
902
903#define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
904 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
905 EXT4_INODE_SET_XTIME_VAL(xtime, &((einode)->vfs_inode), \
906 raw_inode, (einode)->xtime)
907
908#define EXT4_INODE_GET_XTIME_VAL(xtime, inode, raw_inode) \
909 (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra) ? \
910 ext4_decode_extra_time((raw_inode)->xtime, \
911 (raw_inode)->xtime ## _extra) : \
912 (struct timespec64) { \
913 .tv_sec = (signed)le32_to_cpu((raw_inode)->xtime) \
914 })
ef7f3835 915
b898ab23
JL
916#define EXT4_INODE_GET_ATIME(inode, raw_inode) \
917do { \
918 inode_set_atime_to_ts(inode, \
919 EXT4_INODE_GET_XTIME_VAL(i_atime, inode, raw_inode)); \
920} while (0)
921
922#define EXT4_INODE_GET_MTIME(inode, raw_inode) \
95582b00 923do { \
b898ab23
JL
924 inode_set_mtime_to_ts(inode, \
925 EXT4_INODE_GET_XTIME_VAL(i_mtime, inode, raw_inode)); \
ef7f3835
KS
926} while (0)
927
1bc33893
JL
928#define EXT4_INODE_GET_CTIME(inode, raw_inode) \
929do { \
930 inode_set_ctime_to_ts(inode, \
931 EXT4_INODE_GET_XTIME_VAL(i_ctime, inode, raw_inode)); \
932} while (0)
95582b00 933
1bc33893
JL
934#define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
935do { \
936 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
937 (einode)->xtime = \
938 EXT4_INODE_GET_XTIME_VAL(xtime, &(einode->vfs_inode), \
939 raw_inode); \
940 else \
941 (einode)->xtime = (struct timespec64){0, 0}; \
ef7f3835
KS
942} while (0)
943
25ec56b5
JNC
944#define i_disk_version osd1.linux1.l_i_version
945
ac27a0ec
DK
946#if defined(__KERNEL__) || defined(__linux__)
947#define i_reserved1 osd1.linux1.l_i_reserved1
a1ddeb7e 948#define i_file_acl_high osd2.linux2.l_i_file_acl_high
0fc1b451 949#define i_blocks_high osd2.linux2.l_i_blocks_high
ac27a0ec
DK
950#define i_uid_low i_uid
951#define i_gid_low i_gid
952#define i_uid_high osd2.linux2.l_i_uid_high
953#define i_gid_high osd2.linux2.l_i_gid_high
e6153918 954#define i_checksum_lo osd2.linux2.l_i_checksum_lo
ac27a0ec
DK
955
956#elif defined(__GNU__)
957
958#define i_translator osd1.hurd1.h_i_translator
ac27a0ec
DK
959#define i_uid_high osd2.hurd2.h_i_uid_high
960#define i_gid_high osd2.hurd2.h_i_gid_high
961#define i_author osd2.hurd2.h_i_author
962
963#elif defined(__masix__)
964
965#define i_reserved1 osd1.masix1.m_i_reserved1
a1ddeb7e 966#define i_file_acl_high osd2.masix2.m_i_file_acl_high
ac27a0ec
DK
967#define i_reserved2 osd2.masix2.m_i_reserved2
968
969#endif /* defined(__KERNEL__) || defined(__linux__) */
970
c0677e6d 971#include "extents_status.h"
6866d7b3 972#include "fast_commit.h"
c0677e6d 973
daf647d2
TT
974/*
975 * Lock subclasses for i_data_sem in the ext4_inode_info structure.
976 *
977 * These are needed to avoid lockdep false positives when we need to
978 * allocate blocks to the quota inode during ext4_map_blocks(), while
979 * holding i_data_sem for a normal (non-quota) inode. Since we don't
980 * do quota tracking for the quota inode, this avoids deadlock (as
981 * well as infinite recursion, since it isn't turtles all the way
982 * down...)
983 *
984 * I_DATA_SEM_NORMAL - Used for most inodes
985 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode
986 * where the second inode has larger inode number
987 * than the first
988 * I_DATA_SEM_QUOTA - Used for quota inodes only
aff3bea9 989 * I_DATA_SEM_EA - Used for ea_inodes only
daf647d2
TT
990 */
991enum {
992 I_DATA_SEM_NORMAL = 0,
993 I_DATA_SEM_OTHER,
994 I_DATA_SEM_QUOTA,
aff3bea9 995 I_DATA_SEM_EA
daf647d2
TT
996};
997
998
d444c3c3
TT
999/*
1000 * fourth extended file system inode data in memory
1001 */
1002struct ext4_inode_info {
1003 __le32 i_data[15]; /* unconverted */
d444c3c3 1004 __u32 i_dtime;
12e9b892 1005 ext4_fsblk_t i_file_acl;
d444c3c3
TT
1006
1007 /*
1008 * i_block_group is the number of the block group which contains
1009 * this file's inode. Constant across the lifetime of the inode,
a627b0a7 1010 * it is used for making block allocation decisions - we try to
d444c3c3
TT
1011 * place a file's data blocks near its inode block, and new inodes
1012 * near to their parent directory's inode.
1013 */
1014 ext4_group_t i_block_group;
8a2005d3 1015 ext4_lblk_t i_dir_start_lookup;
353eb83c 1016#if (BITS_PER_LONG < 64)
19f5fb7a 1017 unsigned long i_state_flags; /* Dynamic state flags */
353eb83c 1018#endif
12e9b892 1019 unsigned long i_flags;
d444c3c3 1020
d444c3c3
TT
1021 /*
1022 * Extended attributes can be read independently of the main file
f340b3d9 1023 * data. Taking i_rwsem even when reading would cause contention
d444c3c3
TT
1024 * between readers of EAs and writers of regular file data, so
1025 * instead we synchronize on xattr_sem when reading or changing
1026 * EAs.
1027 */
1028 struct rw_semaphore xattr_sem;
d444c3c3 1029
02f310fc
JK
1030 /*
1031 * Inodes with EXT4_STATE_ORPHAN_FILE use i_orphan_idx. Otherwise
1032 * i_orphan is used.
1033 */
1034 union {
1035 struct list_head i_orphan; /* unlinked but open inodes */
1036 unsigned int i_orphan_idx; /* Index in orphan file */
1037 };
d444c3c3 1038
aa75f4d3
HS
1039 /* Fast commit related info */
1040
b3998b3b
RH
1041 /* For tracking dentry create updates */
1042 struct list_head i_fc_dilist;
aa75f4d3
HS
1043 struct list_head i_fc_list; /*
1044 * inodes that need fast commit
1045 * protected by sbi->s_fc_lock.
1046 */
1047
aa75f4d3
HS
1048 /* Start of lblk range that needs to be committed in this fast commit */
1049 ext4_lblk_t i_fc_lblk_start;
1050
1051 /* End of lblk range that needs to be committed in this fast commit */
1052 ext4_lblk_t i_fc_lblk_len;
1053
1054 /* Number of ongoing updates on this inode */
1055 atomic_t i_fc_updates;
1056
1057 /* Fast commit wait queue for this inode */
1058 wait_queue_head_t i_fc_wait;
1059
1060 /* Protect concurrent accesses on i_fc_lblk_start, i_fc_lblk_len */
1061 struct mutex i_fc_lock;
1062
d444c3c3
TT
1063 /*
1064 * i_disksize keeps track of what the inode size is ON DISK, not
1065 * in memory. During truncate, i_size is set to the new size by
1066 * the VFS prior to calling ext4_truncate(), but the filesystem won't
1067 * set i_disksize to 0 until the truncate is actually under way.
1068 *
1069 * The intent is that i_disksize always represents the blocks which
1070 * are used by this file. This allows recovery to restart truncate
1071 * on orphans if we crash during truncate. We actually write i_disksize
1072 * into the on-disk inode when writing inodes out, instead of i_size.
1073 *
1074 * The only time when i_disksize and i_size may be different is when
1075 * a truncate is in progress. The only things which change i_disksize
1076 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
1077 */
1078 loff_t i_disksize;
1079
1080 /*
1081 * i_data_sem is for serialising ext4_truncate() against
1082 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
1083 * data tree are chopped off during truncate. We can't do that in
1084 * ext4 because whenever we perform intermediate commits during
1085 * truncate, the inode and all the metadata blocks *must* be in a
1086 * consistent state which allows truncation of the orphans to restart
1087 * during recovery. Hence we must fix the get_block-vs-truncate race
1088 * by other means, so we have i_data_sem.
1089 */
1090 struct rw_semaphore i_data_sem;
1091 struct inode vfs_inode;
8aefcd55 1092 struct jbd2_inode *jinode;
d444c3c3 1093
202ee5df
TT
1094 spinlock_t i_raw_lock; /* protects updates to the raw inode */
1095
d444c3c3
TT
1096 /*
1097 * File creation time. Its function is same as that of
7b62b293 1098 * struct timespec64 i_{a,c,m}time in the generic inode.
d444c3c3 1099 */
7b62b293 1100 struct timespec64 i_crtime;
d444c3c3
TT
1101
1102 /* mballoc */
27bc446e 1103 atomic_t i_prealloc_active;
38727786
OM
1104 struct rb_root i_prealloc_node;
1105 rwlock_t i_prealloc_lock;
d444c3c3 1106
c0677e6d
ZL
1107 /* extents status tree */
1108 struct ext4_es_tree i_es_tree;
1109 rwlock_t i_es_lock;
edaa53ca 1110 struct list_head i_es_list;
eb68d0e2 1111 unsigned int i_es_all_nr; /* protected by i_es_lock */
edaa53ca 1112 unsigned int i_es_shk_nr; /* protected by i_es_lock */
dd475925
JK
1113 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for
1114 extents to shrink. Protected by
1115 i_es_lock */
c0677e6d 1116
d444c3c3
TT
1117 /* ialloc */
1118 ext4_group_t i_last_alloc_group;
1119
1120 /* allocation reservation info for delalloc */
a627b0a7 1121 /* In case of bigalloc, this refer to clusters rather than blocks */
d444c3c3 1122 unsigned int i_reserved_data_blocks;
d444c3c3 1123
1dc0aa46
EW
1124 /* pending cluster reservations for bigalloc file systems */
1125 struct ext4_pending_tree i_pending_tree;
1126
d444c3c3
TT
1127 /* on-disk additional length */
1128 __u16 i_extra_isize;
1129
67cf5b09
TM
1130 /* Indicate the inline data space. */
1131 u16 i_inline_off;
1132 u16 i_inline_size;
1133
a9e7f447
DM
1134#ifdef CONFIG_QUOTA
1135 /* quota space reservation, managed internally by quota code */
1136 qsize_t i_reserved_quota;
1137#endif
8d5d02e6 1138
2e8fa54e 1139 /* Lock protecting lists below */
744692dc 1140 spinlock_t i_completed_io_lock;
2e8fa54e
JK
1141 /*
1142 * Completed IOs that need unwritten extents handling and have
1143 * transaction reserved
1144 */
1145 struct list_head i_rsv_conversion_list;
2e8fa54e 1146 struct work_struct i_rsv_conversion_work;
600be30a 1147 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
8a2005d3
TT
1148
1149 spinlock_t i_block_reservation_lock;
b436b9be
JK
1150
1151 /*
1152 * Transactions that contain inode's metadata needed to complete
1153 * fsync and fdatasync, respectively.
1154 */
1155 tid_t i_sync_tid;
1156 tid_t i_datasync_tid;
814525f4 1157
96c7e0d9
JK
1158#ifdef CONFIG_QUOTA
1159 struct dquot *i_dquot[MAXQUOTAS];
1160#endif
1161
814525f4
DW
1162 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1163 __u32 i_csum_seed;
b30ab0e0 1164
040cb378 1165 kprojid_t i_projid;
d444c3c3
TT
1166};
1167
ac27a0ec
DK
1168/*
1169 * File system states
1170 */
617ba13b
MC
1171#define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
1172#define EXT4_ERROR_FS 0x0002 /* Errors detected */
1173#define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
8016e29f 1174#define EXT4_FC_REPLAY 0x0020 /* Fast commit replay ongoing */
ac27a0ec 1175
469108ff
TT
1176/*
1177 * Misc. filesystem flags
1178 */
1179#define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
1180#define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
1181#define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
1182
ac27a0ec 1183/*
68911009 1184 * Mount flags set via mount options or defaults
ac27a0ec 1185 */
cdb7ee4c 1186#define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */
617ba13b
MC
1187#define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
1188#define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
1189#define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
1190#define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
1191#define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
39ef17f1 1192#define EXT4_MOUNT_ERRORS_MASK 0x00070
617ba13b
MC
1193#define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
1194#define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
923ae0ff 1195#ifdef CONFIG_FS_DAX
fc626fe3 1196#define EXT4_MOUNT_DAX_ALWAYS 0x00200 /* Direct Access */
923ae0ff 1197#else
fc626fe3 1198#define EXT4_MOUNT_DAX_ALWAYS 0
923ae0ff 1199#endif
617ba13b
MC
1200#define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
1201#define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
1202#define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
1203#define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
1204#define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
1205#define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
1206#define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
1207#define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
afd4672d 1208#define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
617ba13b 1209#define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
49da9392
JK
1210#define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */
1211#define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota,
1212 * enable enforcement for hidden
1213 * quota files */
1214#define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable
1215 * enforcement for hidden quota
1216 * files */
1217#define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota
1218 * enforcement */
744692dc 1219#define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
d4da6c9c 1220#define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
818d276c 1221#define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
327eaf73 1222#define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */
21175ca4 1223#define EXT4_MOUNT_NO_PREFETCH_BLOCK_BITMAPS 0x4000000
64769240 1224#define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
5bf5683a 1225#define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
6fd058f7 1226#define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
5328e635 1227#define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
bfff6873 1228#define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
5bf5683a 1229
68911009
LC
1230/*
1231 * Mount flags set either automatically (could not be set by mount option)
1232 * based on per file system feature or property or in special cases such as
1233 * distinguishing between explicit mount option definition and default.
1234 */
56889787
TT
1235#define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1236 specified delalloc */
68911009
LC
1237#define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1238 size of blocksize * 8
1239 blocks */
ed3654eb
TT
1240#define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1241 file systems */
1e381f60
DM
1242#define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly
1243 specified journal checksum */
1244
995a3ed6 1245#define EXT4_MOUNT2_JOURNAL_FAST_COMMIT 0x00000010 /* Journal fast commit */
a72b38ee
HS
1246#define EXT4_MOUNT2_DAX_NEVER 0x00000020 /* Do not allow Direct Access */
1247#define EXT4_MOUNT2_DAX_INODE 0x00000040 /* For printing options only */
196e402a
HS
1248#define EXT4_MOUNT2_MB_OPTIMIZE_SCAN 0x00000080 /* Optimize group
1249 * scanning in mballoc
1250 */
22b8d707 1251#define EXT4_MOUNT2_ABORT 0x00000100 /* Abort filesystem */
995a3ed6 1252
fd8c37ec
TT
1253#define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1254 ~EXT4_MOUNT_##opt
1255#define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1256 EXT4_MOUNT_##opt
617ba13b
MC
1257#define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1258 EXT4_MOUNT_##opt)
ac27a0ec 1259
a2595b8a
TT
1260#define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1261 ~EXT4_MOUNT2_##opt
1262#define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1263 EXT4_MOUNT2_##opt
1264#define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1265 EXT4_MOUNT2_##opt)
1266
597d508c
AM
1267#define ext4_test_and_set_bit __test_and_set_bit_le
1268#define ext4_set_bit __set_bit_le
597d508c
AM
1269#define ext4_test_and_clear_bit __test_and_clear_bit_le
1270#define ext4_clear_bit __clear_bit_le
50e0168c 1271#define ext4_test_bit test_bit_le
50e0168c
AM
1272#define ext4_find_next_zero_bit find_next_zero_bit_le
1273#define ext4_find_next_bit find_next_bit_le
ac27a0ec 1274
123e3016 1275extern void mb_set_bits(void *bm, int cur, int len);
c3e94d1d 1276
ac27a0ec
DK
1277/*
1278 * Maximal mount counts between two filesystem checks
1279 */
617ba13b
MC
1280#define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1281#define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
ac27a0ec
DK
1282
1283/*
1284 * Behaviour when detecting errors
1285 */
617ba13b
MC
1286#define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1287#define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1288#define EXT4_ERRORS_PANIC 3 /* Panic */
1289#define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
ac27a0ec 1290
e6153918
DW
1291/* Metadata checksum algorithm codes */
1292#define EXT4_CRC32C_CHKSUM 1
1293
bbc605cd
LC
1294#define EXT4_LABEL_MAX 16
1295
ac27a0ec
DK
1296/*
1297 * Structure of the super block
1298 */
617ba13b 1299struct ext4_super_block {
ac27a0ec 1300/*00*/ __le32 s_inodes_count; /* Inodes count */
6bc9feff 1301 __le32 s_blocks_count_lo; /* Blocks count */
308ba3ec
AK
1302 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1303 __le32 s_free_blocks_count_lo; /* Free blocks count */
ac27a0ec
DK
1304/*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1305 __le32 s_first_data_block; /* First Data Block */
1306 __le32 s_log_block_size; /* Block size */
281b5995 1307 __le32 s_log_cluster_size; /* Allocation cluster size */
ac27a0ec 1308/*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
281b5995 1309 __le32 s_clusters_per_group; /* # Clusters per group */
ac27a0ec
DK
1310 __le32 s_inodes_per_group; /* # Inodes per group */
1311 __le32 s_mtime; /* Mount time */
1312/*30*/ __le32 s_wtime; /* Write time */
1313 __le16 s_mnt_count; /* Mount count */
1314 __le16 s_max_mnt_count; /* Maximal mount count */
1315 __le16 s_magic; /* Magic signature */
1316 __le16 s_state; /* File system state */
1317 __le16 s_errors; /* Behaviour when detecting errors */
1318 __le16 s_minor_rev_level; /* minor revision level */
1319/*40*/ __le32 s_lastcheck; /* time of last check */
1320 __le32 s_checkinterval; /* max. time between checks */
1321 __le32 s_creator_os; /* OS */
1322 __le32 s_rev_level; /* Revision level */
1323/*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1324 __le16 s_def_resgid; /* Default gid for reserved blocks */
1325 /*
617ba13b 1326 * These fields are for EXT4_DYNAMIC_REV superblocks only.
ac27a0ec
DK
1327 *
1328 * Note: the difference between the compatible feature set and
1329 * the incompatible feature set is that if there is a bit set
1330 * in the incompatible feature set that the kernel doesn't
1331 * know about, it should refuse to mount the filesystem.
1332 *
1333 * e2fsck's requirements are more strict; if it doesn't know
1334 * about a feature in either the compatible or incompatible
1335 * feature set, it must abort and not try to meddle with
1336 * things it doesn't understand...
1337 */
1338 __le32 s_first_ino; /* First non-reserved inode */
0d1ee42f 1339 __le16 s_inode_size; /* size of inode structure */
ac27a0ec
DK
1340 __le16 s_block_group_nr; /* block group # of this superblock */
1341 __le32 s_feature_compat; /* compatible feature set */
1342/*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1343 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1344/*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
bbc605cd 1345/*78*/ char s_volume_name[EXT4_LABEL_MAX]; /* volume name */
072ebb3b 1346/*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */
ac27a0ec
DK
1347/*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1348 /*
1349 * Performance hints. Directory preallocation should only
617ba13b 1350 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
ac27a0ec
DK
1351 */
1352 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1353 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1354 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1355 /*
617ba13b 1356 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
ac27a0ec
DK
1357 */
1358/*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1359/*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1360 __le32 s_journal_dev; /* device number of journal file */
1361 __le32 s_last_orphan; /* start of list of inodes to delete */
1362 __le32 s_hash_seed[4]; /* HTREE hash seed */
1363 __u8 s_def_hash_version; /* Default hash version to use */
89eeddf0 1364 __u8 s_jnl_backup_type;
0d1ee42f 1365 __le16 s_desc_size; /* size of group descriptor */
bd81d8ee 1366/*100*/ __le32 s_default_mount_opts;
ac27a0ec 1367 __le32 s_first_meta_bg; /* First metablock block group */
bd81d8ee
LV
1368 __le32 s_mkfs_time; /* When the filesystem was created */
1369 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
7fc1f5c2 1370 /* 64bit support valid if EXT4_FEATURE_INCOMPAT_64BIT */
bd81d8ee
LV
1371/*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1372 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1373 __le32 s_free_blocks_count_hi; /* Free blocks count */
c1bddad9
AK
1374 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1375 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1376 __le32 s_flags; /* Miscellaneous flags */
1377 __le16 s_raid_stride; /* RAID stride */
c5e06d10 1378 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
c1bddad9
AK
1379 __le64 s_mmp_block; /* Block for multi-mount protection */
1380 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
772cb7c8 1381 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
e6153918 1382 __u8 s_checksum_type; /* metadata checksum algorithm used */
f542fbe8
TT
1383 __u8 s_encryption_level; /* versioning level for encryption */
1384 __u8 s_reserved_pad; /* Padding to next 32bits */
afc32f7e 1385 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
4a9cdec7
TT
1386 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1387 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1388 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1389 snapshot's future use */
1390 __le32 s_snapshot_list; /* inode number of the head of the
1391 on-disk snapshot list */
1c13d5c0
TT
1392#define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1393 __le32 s_error_count; /* number of fs errors */
1394 __le32 s_first_error_time; /* first time an error happened */
1395 __le32 s_first_error_ino; /* inode involved in first error */
1396 __le64 s_first_error_block; /* block involved of first error */
072ebb3b 1397 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */
1c13d5c0
TT
1398 __le32 s_first_error_line; /* line number where error happened */
1399 __le32 s_last_error_time; /* most recent time of an error */
1400 __le32 s_last_error_ino; /* inode involved in last error */
1401 __le32 s_last_error_line; /* line number where error happened */
1402 __le64 s_last_error_block; /* block involved of last error */
072ebb3b 1403 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */
8b67f04a
TT
1404#define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1405 __u8 s_mount_opts[64];
281b5995
TT
1406 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1407 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1408 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1beeef1b 1409 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
3edc18d8 1410 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */
f542fbe8
TT
1411 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */
1412 __le32 s_lpf_ino; /* Location of the lost+found inode */
8b4953e1 1413 __le32 s_prj_quota_inum; /* inode for tracking project quota */
8c81bd8f 1414 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */
6a0678a7
AB
1415 __u8 s_wtime_hi;
1416 __u8 s_mtime_hi;
1417 __u8 s_mkfs_time_hi;
1418 __u8 s_lastcheck_hi;
1419 __u8 s_first_error_time_hi;
1420 __u8 s_last_error_time_hi;
878520ac
TT
1421 __u8 s_first_error_errcode;
1422 __u8 s_last_error_errcode;
c83ad55e
GKB
1423 __le16 s_encoding; /* Filename charset encoding */
1424 __le16 s_encoding_flags; /* Filename charset encoding flags */
02f310fc
JK
1425 __le32 s_orphan_file_inum; /* Inode for tracking orphan inodes */
1426 __le32 s_reserved[94]; /* Padding to the end of the block */
e6153918 1427 __le32 s_checksum; /* crc32c(superblock) */
ac27a0ec
DK
1428};
1429
1c13d5c0
TT
1430#define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1431
ac27a0ec 1432#ifdef __KERNEL__
bc0b0d6d 1433
a2d4a646 1434/* Number of quota types we support */
689c958c 1435#define EXT4_MAXQUOTAS 3
a2d4a646 1436
c83ad55e
GKB
1437#define EXT4_ENC_UTF8_12_1 1
1438
188c299e
JK
1439/* Types of ext4 journal triggers */
1440enum ext4_journal_trigger_type {
02f310fc 1441 EXT4_JTR_ORPHAN_FILE,
188c299e
JK
1442 EXT4_JTR_NONE /* This must be the last entry for indexing to work! */
1443};
1444
1445#define EXT4_JOURNAL_TRIGGER_COUNT EXT4_JTR_NONE
1446
1447struct ext4_journal_trigger {
1448 struct jbd2_buffer_trigger_type tr_triggers;
1449 struct super_block *sb;
1450};
1451
1452static inline struct ext4_journal_trigger *EXT4_TRIGGER(
1453 struct jbd2_buffer_trigger_type *trigger)
1454{
1455 return container_of(trigger, struct ext4_journal_trigger, tr_triggers);
1456}
1457
02f310fc
JK
1458#define EXT4_ORPHAN_BLOCK_MAGIC 0x0b10ca04
1459
1460/* Structure at the tail of orphan block */
1461struct ext4_orphan_block_tail {
1462 __le32 ob_magic;
1463 __le32 ob_checksum;
1464};
1465
1466static inline int ext4_inodes_per_orphan_block(struct super_block *sb)
1467{
1468 return (sb->s_blocksize - sizeof(struct ext4_orphan_block_tail)) /
1469 sizeof(u32);
1470}
1471
1472struct ext4_orphan_block {
4a79a98c 1473 atomic_t ob_free_entries; /* Number of free orphan entries in block */
02f310fc
JK
1474 struct buffer_head *ob_bh; /* Buffer for orphan block */
1475};
1476
1477/*
1478 * Info about orphan file.
1479 */
1480struct ext4_orphan_info {
02f310fc
JK
1481 int of_blocks; /* Number of orphan blocks in a file */
1482 __u32 of_csum_seed; /* Checksum seed for orphan file */
1483 struct ext4_orphan_block *of_binfo; /* Array with info about orphan
1484 * file blocks */
1485};
1486
ca0faba0
TT
1487/*
1488 * fourth extended-fs super-block data in memory
1489 */
1490struct ext4_sb_info {
1491 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1492 unsigned long s_inodes_per_block;/* Number of inodes per block */
1493 unsigned long s_blocks_per_group;/* Number of blocks in a group */
281b5995 1494 unsigned long s_clusters_per_group; /* Number of clusters in a group */
ca0faba0
TT
1495 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1496 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1497 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1498 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1499 ext4_group_t s_groups_count; /* Number of groups in the fs */
fb0a387d 1500 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
952fc18e 1501 unsigned long s_overhead; /* # of fs overhead clusters */
281b5995
TT
1502 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1503 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
ca0faba0
TT
1504 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1505 struct buffer_head * s_sbh; /* Buffer containing the super block */
1506 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
350bb48b 1507 /* Array of bh's for the block group descriptors */
1d0c3924 1508 struct buffer_head * __rcu *s_group_desc;
7f4520cc 1509 unsigned int s_mount_opt;
a2595b8a 1510 unsigned int s_mount_opt2;
9b5f6c9b 1511 unsigned long s_mount_flags;
5a916be1 1512 unsigned int s_def_mount_opt;
e3645d72 1513 unsigned int s_def_mount_opt2;
ca0faba0 1514 ext4_fsblk_t s_sb_block;
27dd4385 1515 atomic64_t s_resv_clusters;
08cefc7a
EB
1516 kuid_t s_resuid;
1517 kgid_t s_resgid;
ca0faba0
TT
1518 unsigned short s_mount_state;
1519 unsigned short s_pad;
1520 int s_addr_per_block_bits;
1521 int s_desc_per_block_bits;
1522 int s_inode_size;
1523 int s_first_ino;
1524 unsigned int s_inode_readahead_blks;
11013911 1525 unsigned int s_inode_goal;
ca0faba0
TT
1526 u32 s_hash_seed[4];
1527 int s_def_hash_version;
6d2424a8 1528 int s_hash_unsigned; /* 3 if hash should be unsigned, 0 if not */
57042651 1529 struct percpu_counter s_freeclusters_counter;
ca0faba0
TT
1530 struct percpu_counter s_freeinodes_counter;
1531 struct percpu_counter s_dirs_counter;
57042651 1532 struct percpu_counter s_dirtyclusters_counter;
efc61345 1533 struct percpu_counter s_sra_exceeded_retry_limit;
ca0faba0
TT
1534 struct blockgroup_lock *s_blockgroup_lock;
1535 struct proc_dir_entry *s_proc;
1536 struct kobject s_kobj;
1537 struct completion s_kobj_unregister;
2c0544b2 1538 struct super_block *s_sb;
618f0031 1539 struct buffer_head *s_mmp_bh;
ca0faba0
TT
1540
1541 /* Journaling */
ca0faba0 1542 struct journal_s *s_journal;
9549a168 1543 unsigned long s_ext4_flags; /* Ext4 superblock flags */
02f310fc
JK
1544 struct mutex s_orphan_lock; /* Protects on disk list changes */
1545 struct list_head s_orphan; /* List of orphaned inodes in on disk
1546 list */
1547 struct ext4_orphan_info s_orphan_info;
ca0faba0
TT
1548 unsigned long s_commit_interval;
1549 u32 s_max_batch_time;
1550 u32 s_min_batch_time;
61ead714 1551 struct file *s_journal_bdev_file;
ca0faba0 1552#ifdef CONFIG_QUOTA
33458eab
TT
1553 /* Names of quota files with journalled quota */
1554 char __rcu *s_qf_names[EXT4_MAXQUOTAS];
ca0faba0
TT
1555 int s_jquota_fmt; /* Format of quota to use */
1556#endif
1557 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
dd0db94f 1558 struct ext4_system_blocks __rcu *s_system_blks;
ca0faba0
TT
1559
1560#ifdef EXTENTS_STATS
1561 /* ext4 extents stats */
1562 unsigned long s_ext_min;
1563 unsigned long s_ext_max;
1564 unsigned long s_depth_max;
1565 spinlock_t s_ext_stats_lock;
1566 unsigned long s_ext_blocks;
1567 unsigned long s_ext_extents;
1568#endif
1569
1570 /* for buddy allocator */
df3da4ea 1571 struct ext4_group_info ** __rcu *s_group_info;
ca0faba0 1572 struct inode *s_buddy_cache;
ca0faba0 1573 spinlock_t s_md_lock;
ca0faba0
TT
1574 unsigned short *s_mb_offsets;
1575 unsigned int *s_mb_maxs;
28623c2f 1576 unsigned int s_group_info_size;
d08854f5 1577 unsigned int s_mb_free_pending;
ce774e53 1578 struct list_head s_freed_data_list[2]; /* List of blocks to be freed
a0154344 1579 after commit completed */
55cdd0af
WJ
1580 struct list_head s_discard_list;
1581 struct work_struct s_discard_work;
5036ab8d 1582 atomic_t s_retry_alloc_pending;
83e80a6e
JK
1583 struct list_head *s_mb_avg_fragment_size;
1584 rwlock_t *s_mb_avg_fragment_size_locks;
196e402a
HS
1585 struct list_head *s_mb_largest_free_orders;
1586 rwlock_t *s_mb_largest_free_orders_locks;
ca0faba0
TT
1587
1588 /* tunables */
1589 unsigned long s_stripe;
196e402a 1590 unsigned int s_mb_max_linear_groups;
ca0faba0
TT
1591 unsigned int s_mb_stream_request;
1592 unsigned int s_mb_max_to_scan;
1593 unsigned int s_mb_min_to_scan;
1594 unsigned int s_mb_stats;
1595 unsigned int s_mb_order2_reqs;
1596 unsigned int s_mb_group_prealloc;
df981d03 1597 unsigned int s_max_dir_size_kb;
ca0faba0
TT
1598 /* where last allocation was done - for stream allocation */
1599 unsigned long s_mb_last_group;
1600 unsigned long s_mb_last_start;
cfd73237
AZ
1601 unsigned int s_mb_prefetch;
1602 unsigned int s_mb_prefetch_limit;
f52f3d2b 1603 unsigned int s_mb_best_avail_max_trim_order;
ca0faba0 1604
ca0faba0 1605 /* stats for buddy allocator */
ca0faba0
TT
1606 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1607 atomic_t s_bal_success; /* we found long enough chunks */
1608 atomic_t s_bal_allocated; /* in blocks */
1609 atomic_t s_bal_ex_scanned; /* total extents scanned */
fdd9a009 1610 atomic_t s_bal_cX_ex_scanned[EXT4_MB_NUM_CRS]; /* total extents scanned */
a6c75eaf 1611 atomic_t s_bal_groups_scanned; /* number of groups scanned */
ca0faba0 1612 atomic_t s_bal_goals; /* goal hits */
3ef5d263 1613 atomic_t s_bal_len_goals; /* len goal hits */
ca0faba0
TT
1614 atomic_t s_bal_breaks; /* too long searches */
1615 atomic_t s_bal_2orders; /* 2^order hits */
f52f3d2b
OM
1616 atomic_t s_bal_p2_aligned_bad_suggestions;
1617 atomic_t s_bal_goal_fast_bad_suggestions;
1618 atomic_t s_bal_best_avail_bad_suggestions;
4eb7a4a1
OM
1619 atomic64_t s_bal_cX_groups_considered[EXT4_MB_NUM_CRS];
1620 atomic64_t s_bal_cX_hits[EXT4_MB_NUM_CRS];
1621 atomic64_t s_bal_cX_failed[EXT4_MB_NUM_CRS]; /* cX loop didn't find blocks */
67d25186
HS
1622 atomic_t s_mb_buddies_generated; /* number of buddies generated */
1623 atomic64_t s_mb_generation_time;
ca0faba0
TT
1624 atomic_t s_mb_lost_chunks;
1625 atomic_t s_mb_preallocated;
1626 atomic_t s_mb_discarded;
50797481 1627 atomic_t s_lock_busy;
ca0faba0
TT
1628
1629 /* locality groups */
003cb608 1630 struct ext4_locality_group __percpu *s_locality_groups;
ca0faba0
TT
1631
1632 /* for write statistics */
1633 unsigned long s_sectors_written_start;
1634 u64 s_kbytes_written;
1635
67a5da56
ZL
1636 /* the size of zero-out chunk */
1637 unsigned int s_extent_max_zeroout_kb;
1638
ca0faba0 1639 unsigned int s_log_groups_per_flex;
7c990728 1640 struct flex_groups * __rcu *s_flex_groups;
117fff10 1641 ext4_group_t s_flex_groups_allocated;
4c0425ff 1642
2e8fa54e
JK
1643 /* workqueue for reserved extent conversions (buffered io) */
1644 struct workqueue_struct *rsv_conversion_wq;
66e61a9e
TT
1645
1646 /* timer for periodic error stats printing */
1647 struct timer_list s_err_report;
bfff6873
LC
1648
1649 /* Lazy inode table initialization info */
1650 struct ext4_li_request *s_li_request;
1651 /* Wait multiplier for lazy initialization thread */
1652 unsigned int s_li_wait_mult;
c5e06d10
JL
1653
1654 /* Kernel thread for multiple mount protection */
1655 struct task_struct *s_mmp_tsk;
3d56b8d2
TM
1656
1657 /* record the last minlen when FITRIM is called. */
2327fb2e 1658 unsigned long s_last_trim_minblks;
0441984a
DW
1659
1660 /* Reference to checksum algorithm driver via cryptoapi */
1661 struct crypto_shash *s_chksum_driver;
a9c47317
DW
1662
1663 /* Precomputed FS UUID checksum for seeding other checksums */
1664 __u32 s_csum_seed;
74cd15cd
ZL
1665
1666 /* Reclaim extents from extent status tree */
4d09d75d 1667 struct shrinker *s_es_shrinker;
dd475925 1668 struct list_head s_es_list; /* List of inodes with reclaimable extents */
edaa53ca 1669 long s_es_nr_inode;
eb68d0e2 1670 struct ext4_es_stats s_es_stats;
47387409 1671 struct mb_cache *s_ea_block_cache;
dec214d0 1672 struct mb_cache *s_ea_inode_cache;
edaa53ca 1673 spinlock_t s_es_lock ____cacheline_aligned_in_smp;
efbed4dc 1674
188c299e
JK
1675 /* Journal triggers for checksum computation */
1676 struct ext4_journal_trigger s_journal_triggers[EXT4_JOURNAL_TRIGGER_COUNT];
1677
efbed4dc
TT
1678 /* Ratelimit ext4 messages. */
1679 struct ratelimit_state s_err_ratelimit_state;
1680 struct ratelimit_state s_warning_ratelimit_state;
1681 struct ratelimit_state s_msg_ratelimit_state;
1cf006ed
DM
1682 atomic_t s_warning_count;
1683 atomic_t s_msg_count;
c8585c6f 1684
ac4acb1f
EB
1685 /* Encryption policy for '-o test_dummy_encryption' */
1686 struct fscrypt_dummy_policy s_dummy_enc_policy;
ed318a6c 1687
cb85f4d2
EB
1688 /*
1689 * Barrier between writepages ops and changing any inode's JOURNAL_DATA
745f17a4
BL
1690 * or EXTENTS flag or between writepages ops and changing DELALLOC or
1691 * DIOREAD_NOLOCK mount options on remount.
cb85f4d2 1692 */
bbd55937 1693 struct percpu_rw_semaphore s_writepages_rwsem;
5e405595 1694 struct dax_device *s_daxdev;
cd913c76 1695 u64 s_dax_part_off;
46f870d6
TT
1696#ifdef CONFIG_EXT4_DEBUG
1697 unsigned long s_simulate_fail;
1698#endif
bc71726c 1699 /* Record the errseq of the backing block device */
1700 errseq_t s_bdev_wb_err;
1701 spinlock_t s_bdev_wb_lock;
aa75f4d3 1702
c92dc856
JK
1703 /* Information about errors that happened during this mount */
1704 spinlock_t s_error_lock;
1705 int s_add_error_count;
1706 int s_first_error_code;
1707 __u32 s_first_error_line;
1708 __u32 s_first_error_ino;
1709 __u64 s_first_error_block;
1710 const char *s_first_error_func;
1711 time64_t s_first_error_time;
1712 int s_last_error_code;
1713 __u32 s_last_error_line;
1714 __u32 s_last_error_ino;
1715 __u64 s_last_error_block;
1716 const char *s_last_error_func;
1717 time64_t s_last_error_time;
1718 /*
bb15cea2
TT
1719 * If we are in a context where we cannot update the on-disk
1720 * superblock, we queue the work here. This is used to update
1721 * the error information in the superblock, and for periodic
1722 * updates of the superblock called from the commit callback
1723 * function.
c92dc856 1724 */
bb15cea2 1725 struct work_struct s_sb_upd_work;
c92dc856 1726
7bbbe241 1727 /* Ext4 fast commit sub transaction ID */
aa75f4d3 1728 atomic_t s_fc_subtid;
7bbbe241 1729
aa75f4d3
HS
1730 /*
1731 * After commit starts, the main queue gets locked, and the further
1732 * updates get added in the staging queue.
1733 */
1734#define FC_Q_MAIN 0
1735#define FC_Q_STAGING 1
1736 struct list_head s_fc_q[2]; /* Inodes staged for fast commit
1737 * that have data changes in them.
1738 */
1739 struct list_head s_fc_dentry_q[2]; /* directory entry updates */
1740 unsigned int s_fc_bytes;
1741 /*
1742 * Main fast commit lock. This lock protects accesses to the
1743 * following fields:
1744 * ei->i_fc_list, s_fc_dentry_q, s_fc_q, s_fc_bytes, s_fc_bh.
1745 */
1746 spinlock_t s_fc_lock;
1747 struct buffer_head *s_fc_bh;
1748 struct ext4_fc_stats s_fc_stats;
e85c81ba 1749 tid_t s_fc_ineligible_tid;
8016e29f
HS
1750#ifdef CONFIG_EXT4_DEBUG
1751 int s_fc_debug_max_replay;
1752#endif
1753 struct ext4_fc_replay_state s_fc_replay_state;
ca0faba0
TT
1754};
1755
af5bc92d 1756static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
ac27a0ec
DK
1757{
1758 return sb->s_fs_info;
1759}
617ba13b 1760static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
ac27a0ec 1761{
617ba13b 1762 return container_of(inode, struct ext4_inode_info, vfs_inode);
ac27a0ec
DK
1763}
1764
00d873c1
JK
1765static inline int ext4_writepages_down_read(struct super_block *sb)
1766{
1767 percpu_down_read(&EXT4_SB(sb)->s_writepages_rwsem);
1768 return memalloc_nofs_save();
1769}
1770
1771static inline void ext4_writepages_up_read(struct super_block *sb, int ctx)
1772{
1773 memalloc_nofs_restore(ctx);
1774 percpu_up_read(&EXT4_SB(sb)->s_writepages_rwsem);
1775}
1776
1777static inline int ext4_writepages_down_write(struct super_block *sb)
1778{
1779 percpu_down_write(&EXT4_SB(sb)->s_writepages_rwsem);
1780 return memalloc_nofs_save();
1781}
1782
1783static inline void ext4_writepages_up_write(struct super_block *sb, int ctx)
1784{
1785 memalloc_nofs_restore(ctx);
1786 percpu_up_write(&EXT4_SB(sb)->s_writepages_rwsem);
1787}
1788
617ba13b 1789static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
ac27a0ec 1790{
617ba13b 1791 return ino == EXT4_ROOT_INO ||
617ba13b
MC
1792 (ino >= EXT4_FIRST_INO(sb) &&
1793 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
ac27a0ec 1794}
19f5fb7a 1795
1d0c3924
TT
1796/*
1797 * Returns: sbi->field[index]
1798 * Used to access an array element from the following sbi fields which require
1799 * rcu protection to avoid dereferencing an invalid pointer due to reassignment
1800 * - s_group_desc
1801 * - s_group_info
1802 * - s_flex_group
1803 */
1804#define sbi_array_rcu_deref(sbi, field, index) \
1805({ \
1806 typeof(*((sbi)->field)) _v; \
1807 rcu_read_lock(); \
1808 _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index]; \
1809 rcu_read_unlock(); \
1810 _v; \
1811})
1812
9b5f6c9b
HS
1813/*
1814 * run-time mount flags
1815 */
1816enum {
1817 EXT4_MF_MNTDIR_SAMPLED,
bdc8a53a 1818 EXT4_MF_FC_INELIGIBLE /* Fast commit ineligible */
9b5f6c9b
HS
1819};
1820
1821static inline void ext4_set_mount_flag(struct super_block *sb, int bit)
1822{
1823 set_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1824}
1825
1826static inline void ext4_clear_mount_flag(struct super_block *sb, int bit)
1827{
1828 clear_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1829}
1830
1831static inline int ext4_test_mount_flag(struct super_block *sb, int bit)
1832{
1833 return test_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1834}
1835
1836
46f870d6
TT
1837/*
1838 * Simulate_fail codes
1839 */
1840#define EXT4_SIM_BBITMAP_EIO 1
1841#define EXT4_SIM_BBITMAP_CRC 2
1842#define EXT4_SIM_IBITMAP_EIO 3
1843#define EXT4_SIM_IBITMAP_CRC 4
1844#define EXT4_SIM_INODE_EIO 5
1845#define EXT4_SIM_INODE_CRC 6
1846#define EXT4_SIM_DIRBLOCK_EIO 7
1847#define EXT4_SIM_DIRBLOCK_CRC 8
1848
1849static inline bool ext4_simulate_fail(struct super_block *sb,
1850 unsigned long code)
1851{
1852#ifdef CONFIG_EXT4_DEBUG
1853 struct ext4_sb_info *sbi = EXT4_SB(sb);
1854
1855 if (unlikely(sbi->s_simulate_fail == code)) {
1856 sbi->s_simulate_fail = 0;
1857 return true;
1858 }
1859#endif
1860 return false;
1861}
1862
1863static inline void ext4_simulate_fail_bh(struct super_block *sb,
1864 struct buffer_head *bh,
1865 unsigned long code)
1866{
1867 if (!IS_ERR(bh) && ext4_simulate_fail(sb, code))
1868 clear_buffer_uptodate(bh);
1869}
1870
878520ac
TT
1871/*
1872 * Error number codes for s_{first,last}_error_errno
1873 *
1874 * Linux errno numbers are architecture specific, so we need to translate
1875 * them into something which is architecture independent. We don't define
1876 * codes for all errno's; just the ones which are most likely to be the cause
1877 * of an ext4_error() call.
1878 */
1879#define EXT4_ERR_UNKNOWN 1
1880#define EXT4_ERR_EIO 2
1881#define EXT4_ERR_ENOMEM 3
1882#define EXT4_ERR_EFSBADCRC 4
1883#define EXT4_ERR_EFSCORRUPTED 5
1884#define EXT4_ERR_ENOSPC 6
1885#define EXT4_ERR_ENOKEY 7
1886#define EXT4_ERR_EROFS 8
1887#define EXT4_ERR_EFBIG 9
1888#define EXT4_ERR_EEXIST 10
1889#define EXT4_ERR_ERANGE 11
1890#define EXT4_ERR_EOVERFLOW 12
1891#define EXT4_ERR_EBUSY 13
1892#define EXT4_ERR_ENOTDIR 14
1893#define EXT4_ERR_ENOTEMPTY 15
1894#define EXT4_ERR_ESHUTDOWN 16
1895#define EXT4_ERR_EFAULT 17
1896
19f5fb7a
TT
1897/*
1898 * Inode dynamic state flags
1899 */
1900enum {
19f5fb7a
TT
1901 EXT4_STATE_NEW, /* inode is newly created */
1902 EXT4_STATE_XATTR, /* has in-inode xattrs */
1903 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1904 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1905 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
14ece102 1906 EXT4_STATE_NEWENTRY, /* File just added to dir */
67cf5b09 1907 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
7869a4a6 1908 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
a6d05676 1909 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */
c93d8f88 1910 EXT4_STATE_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */
aa75f4d3 1911 EXT4_STATE_FC_COMMITTING, /* Fast commit ongoing */
02f310fc 1912 EXT4_STATE_ORPHAN_FILE, /* Inode orphaned in orphan file */
19f5fb7a
TT
1913};
1914
353eb83c 1915#define EXT4_INODE_BIT_FNS(name, field, offset) \
12e9b892
DM
1916static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1917{ \
353eb83c 1918 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
12e9b892
DM
1919} \
1920static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1921{ \
353eb83c 1922 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
12e9b892
DM
1923} \
1924static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1925{ \
353eb83c 1926 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
19f5fb7a
TT
1927}
1928
3f61c0cc
AD
1929/* Add these declarations here only so that these functions can be
1930 * found by name. Otherwise, they are very hard to locate. */
1931static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1932static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1933static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
353eb83c 1934EXT4_INODE_BIT_FNS(flag, flags, 0)
3f61c0cc
AD
1935
1936/* Add these declarations here only so that these functions can be
1937 * found by name. Otherwise, they are very hard to locate. */
1938static inline int ext4_test_inode_state(struct inode *inode, int bit);
1939static inline void ext4_set_inode_state(struct inode *inode, int bit);
1940static inline void ext4_clear_inode_state(struct inode *inode, int bit);
353eb83c
TT
1941#if (BITS_PER_LONG < 64)
1942EXT4_INODE_BIT_FNS(state, state_flags, 0)
1943
1944static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1945{
1946 (ei)->i_state_flags = 0;
1947}
1948#else
1949EXT4_INODE_BIT_FNS(state, flags, 32)
1950
1951static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1952{
1953 /* We depend on the fact that callers will set i_flags */
1954}
1955#endif
ac27a0ec 1956#else
617ba13b 1957/* Assume that user mode programs are passing in an ext4fs superblock, not
ac27a0ec
DK
1958 * a kernel struct super_block. This will allow us to call the feature-test
1959 * macros from user land. */
617ba13b 1960#define EXT4_SB(sb) (sb)
ac27a0ec
DK
1961#endif
1962
c93d8f88
EB
1963static inline bool ext4_verity_in_progress(struct inode *inode)
1964{
1965 return IS_ENABLED(CONFIG_FS_VERITY) &&
1966 ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS);
1967}
1968
617ba13b 1969#define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
ac27a0ec
DK
1970
1971/*
1972 * Codes for operating systems
1973 */
617ba13b
MC
1974#define EXT4_OS_LINUX 0
1975#define EXT4_OS_HURD 1
1976#define EXT4_OS_MASIX 2
1977#define EXT4_OS_FREEBSD 3
1978#define EXT4_OS_LITES 4
ac27a0ec
DK
1979
1980/*
1981 * Revision levels
1982 */
617ba13b
MC
1983#define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1984#define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
ac27a0ec 1985
617ba13b 1986#define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
ac27a0ec 1987
617ba13b 1988#define EXT4_GOOD_OLD_INODE_SIZE 128
ac27a0ec 1989
4881c497
DD
1990#define EXT4_EXTRA_TIMESTAMP_MAX (((s64)1 << 34) - 1 + S32_MIN)
1991#define EXT4_NON_EXTRA_TIMESTAMP_MAX S32_MAX
1992#define EXT4_TIMESTAMP_MIN S32_MIN
1993
ac27a0ec
DK
1994/*
1995 * Feature set definitions
1996 */
1997
617ba13b
MC
1998#define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1999#define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
2000#define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
2001#define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
2002#define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
2003#define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1beeef1b 2004#define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
a44ad683
HS
2005/*
2006 * The reason why "FAST_COMMIT" is a compat feature is that, FS becomes
2007 * incompatible only if fast commit blocks are present in the FS. Since we
2008 * clear the journal (and thus the fast commit blocks), we don't mark FS as
2009 * incompatible. We also have a JBD2 incompat feature, which gets set when
2010 * there are fast commit blocks present in the journal.
2011 */
995a3ed6 2012#define EXT4_FEATURE_COMPAT_FAST_COMMIT 0x0400
b925acb8 2013#define EXT4_FEATURE_COMPAT_STABLE_INODES 0x0800
02f310fc 2014#define EXT4_FEATURE_COMPAT_ORPHAN_FILE 0x1000 /* Orphan file exists */
617ba13b
MC
2015
2016#define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
2017#define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
2018#define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
0fc1b451 2019#define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
717d50e4 2020#define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
f8628a14 2021#define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
ef7f3835 2022#define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
ae812306 2023#define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
281b5995 2024#define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
e6153918
DW
2025/*
2026 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
2027 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
2028 * all other data structures' checksums. However, the METADATA_CSUM and
2029 * GDT_CSUM bits are mutually exclusive.
2030 */
9b90e5e0 2031#define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
2cb5cc8b 2032#define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000
8b4953e1 2033#define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000
c93d8f88 2034#define EXT4_FEATURE_RO_COMPAT_VERITY 0x8000
02f310fc
JK
2035#define EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT 0x10000 /* Orphan file may be
2036 non-empty */
617ba13b
MC
2037
2038#define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
2039#define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
2040#define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
2041#define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
2042#define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
a86c6181 2043#define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
bd81d8ee 2044#define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
0fc1b451 2045#define EXT4_FEATURE_INCOMPAT_MMP 0x0100
ce421581 2046#define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
f710b4b9
TT
2047#define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
2048#define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
8c81bd8f 2049#define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000
9b90e5e0 2050#define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
67cf5b09 2051#define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
3edc18d8 2052#define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000
c83ad55e 2053#define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000
617ba13b 2054
c9e716eb
AD
2055extern void ext4_update_dynamic_rev(struct super_block *sb);
2056
e2b911c5
DW
2057#define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
2058static inline bool ext4_has_feature_##name(struct super_block *sb) \
2059{ \
2060 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
2061 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
2062} \
2063static inline void ext4_set_feature_##name(struct super_block *sb) \
2064{ \
c9e716eb 2065 ext4_update_dynamic_rev(sb); \
e2b911c5
DW
2066 EXT4_SB(sb)->s_es->s_feature_compat |= \
2067 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
2068} \
2069static inline void ext4_clear_feature_##name(struct super_block *sb) \
2070{ \
2071 EXT4_SB(sb)->s_es->s_feature_compat &= \
2072 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
2073}
2074
2075#define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
2076static inline bool ext4_has_feature_##name(struct super_block *sb) \
2077{ \
2078 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
2079 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
2080} \
2081static inline void ext4_set_feature_##name(struct super_block *sb) \
2082{ \
c9e716eb 2083 ext4_update_dynamic_rev(sb); \
e2b911c5
DW
2084 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
2085 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
2086} \
2087static inline void ext4_clear_feature_##name(struct super_block *sb) \
2088{ \
2089 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
2090 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
2091}
2092
2093#define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
2094static inline bool ext4_has_feature_##name(struct super_block *sb) \
2095{ \
2096 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
2097 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
2098} \
2099static inline void ext4_set_feature_##name(struct super_block *sb) \
2100{ \
c9e716eb 2101 ext4_update_dynamic_rev(sb); \
e2b911c5
DW
2102 EXT4_SB(sb)->s_es->s_feature_incompat |= \
2103 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
2104} \
2105static inline void ext4_clear_feature_##name(struct super_block *sb) \
2106{ \
2107 EXT4_SB(sb)->s_es->s_feature_incompat &= \
2108 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
2109}
2110
2111EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC)
2112EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES)
2113EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL)
2114EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR)
2115EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE)
2116EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX)
2117EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2)
995a3ed6 2118EXT4_FEATURE_COMPAT_FUNCS(fast_commit, FAST_COMMIT)
b925acb8 2119EXT4_FEATURE_COMPAT_FUNCS(stable_inodes, STABLE_INODES)
02f310fc 2120EXT4_FEATURE_COMPAT_FUNCS(orphan_file, ORPHAN_FILE)
e2b911c5
DW
2121
2122EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER)
2123EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE)
2124EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR)
2125EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE)
2126EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM)
2127EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK)
2128EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE)
2129EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA)
2130EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC)
2131EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM)
2132EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY)
2133EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT)
c93d8f88 2134EXT4_FEATURE_RO_COMPAT_FUNCS(verity, VERITY)
02f310fc 2135EXT4_FEATURE_RO_COMPAT_FUNCS(orphan_present, ORPHAN_PRESENT)
e2b911c5
DW
2136
2137EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION)
2138EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE)
2139EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER)
2140EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV)
2141EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG)
2142EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS)
2143EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT)
2144EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP)
2145EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG)
2146EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE)
2147EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA)
2148EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED)
2149EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR)
2150EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA)
2151EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT)
c83ad55e 2152EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD)
e2b911c5 2153
2035e776
TT
2154#define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
2155#define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
2156 EXT4_FEATURE_INCOMPAT_META_BG)
2157#define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2158 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2159 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
2160
2161#define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
2162#define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
2163 EXT4_FEATURE_INCOMPAT_RECOVER| \
2164 EXT4_FEATURE_INCOMPAT_META_BG)
2165#define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2166 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2167 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
2168
02f310fc
JK
2169#define EXT4_FEATURE_COMPAT_SUPP (EXT4_FEATURE_COMPAT_EXT_ATTR| \
2170 EXT4_FEATURE_COMPAT_ORPHAN_FILE)
617ba13b
MC
2171#define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
2172 EXT4_FEATURE_INCOMPAT_RECOVER| \
a86c6181 2173 EXT4_FEATURE_INCOMPAT_META_BG| \
bd81d8ee 2174 EXT4_FEATURE_INCOMPAT_EXTENTS| \
ce421581 2175 EXT4_FEATURE_INCOMPAT_64BIT| \
c5e06d10 2176 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
e50e5129 2177 EXT4_FEATURE_INCOMPAT_EA_INODE| \
6ddb2447
TT
2178 EXT4_FEATURE_INCOMPAT_MMP | \
2179 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
8c81bd8f 2180 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
c83ad55e 2181 EXT4_FEATURE_INCOMPAT_CASEFOLD | \
e08ac99f
AB
2182 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \
2183 EXT4_FEATURE_INCOMPAT_LARGEDIR)
617ba13b
MC
2184#define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2185 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
717d50e4 2186 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
f8628a14 2187 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
ef7f3835 2188 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
0fc1b451 2189 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
6f16b606 2190 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
e93376c2 2191 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
7c319d32 2192 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
040cb378 2193 EXT4_FEATURE_RO_COMPAT_QUOTA |\
c93d8f88 2194 EXT4_FEATURE_RO_COMPAT_PROJECT |\
02f310fc
JK
2195 EXT4_FEATURE_RO_COMPAT_VERITY |\
2196 EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT)
ac27a0ec 2197
e2b911c5
DW
2198#define EXTN_FEATURE_FUNCS(ver) \
2199static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
2200{ \
2201 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
2202 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
2203} \
2204static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
2205{ \
2206 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
2207 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
2208} \
2209static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
2210{ \
2211 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
2212 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
2213}
2214
2215EXTN_FEATURE_FUNCS(2)
2216EXTN_FEATURE_FUNCS(3)
2217EXTN_FEATURE_FUNCS(4)
2218
2219static inline bool ext4_has_compat_features(struct super_block *sb)
2220{
2221 return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
2222}
2223static inline bool ext4_has_ro_compat_features(struct super_block *sb)
2224{
2225 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
2226}
2227static inline bool ext4_has_incompat_features(struct super_block *sb)
2228{
2229 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
2230}
2231
25c6d98f
JK
2232extern int ext4_feature_set_ok(struct super_block *sb, int readonly);
2233
9549a168
TT
2234/*
2235 * Superblock flags
2236 */
2237#define EXT4_FLAGS_RESIZING 0
0db1ff22 2238#define EXT4_FLAGS_SHUTDOWN 1
a8ab6d38 2239#define EXT4_FLAGS_BDEV_IS_DAX 2
0db1ff22 2240
eb8ab444 2241static inline int ext4_forced_shutdown(struct super_block *sb)
0db1ff22 2242{
eb8ab444 2243 return test_bit(EXT4_FLAGS_SHUTDOWN, &EXT4_SB(sb)->s_ext4_flags);
0db1ff22 2244}
9549a168 2245
ac27a0ec
DK
2246/*
2247 * Default values for user and/or group using reserved blocks
2248 */
617ba13b
MC
2249#define EXT4_DEF_RESUID 0
2250#define EXT4_DEF_RESGID 0
ac27a0ec 2251
040cb378
LX
2252/*
2253 * Default project ID
2254 */
2255#define EXT4_DEF_PROJID 0
2256
240799cd
TT
2257#define EXT4_DEF_INODE_READAHEAD_BLKS 32
2258
ac27a0ec
DK
2259/*
2260 * Default mount options
2261 */
617ba13b
MC
2262#define EXT4_DEFM_DEBUG 0x0001
2263#define EXT4_DEFM_BSDGROUPS 0x0002
2264#define EXT4_DEFM_XATTR_USER 0x0004
2265#define EXT4_DEFM_ACL 0x0008
2266#define EXT4_DEFM_UID16 0x0010
2267#define EXT4_DEFM_JMODE 0x0060
2268#define EXT4_DEFM_JMODE_DATA 0x0020
2269#define EXT4_DEFM_JMODE_ORDERED 0x0040
2270#define EXT4_DEFM_JMODE_WBACK 0x0060
8b67f04a
TT
2271#define EXT4_DEFM_NOBARRIER 0x0100
2272#define EXT4_DEFM_BLOCK_VALIDITY 0x0200
2273#define EXT4_DEFM_DISCARD 0x0400
2274#define EXT4_DEFM_NODELALLOC 0x0800
ac27a0ec 2275
30773840
TT
2276/*
2277 * Default journal batch times
2278 */
2279#define EXT4_DEF_MIN_BATCH_TIME 0
2280#define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
2281
a4912123
TT
2282/*
2283 * Minimum number of groups in a flexgroup before we separate out
2284 * directories into the first block group of a flexgroup
2285 */
2286#define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
2287
ac27a0ec
DK
2288/*
2289 * Structure of a directory entry
2290 */
617ba13b 2291#define EXT4_NAME_LEN 255
c186f088
YB
2292/*
2293 * Base length of the ext4 directory entry excluding the name length
2294 */
2295#define EXT4_BASE_DIR_LEN (sizeof(struct ext4_dir_entry_2) - EXT4_NAME_LEN)
ac27a0ec 2296
617ba13b 2297struct ext4_dir_entry {
ac27a0ec
DK
2298 __le32 inode; /* Inode number */
2299 __le16 rec_len; /* Directory entry length */
2300 __le16 name_len; /* Name length */
617ba13b 2301 char name[EXT4_NAME_LEN]; /* File name */
ac27a0ec
DK
2302};
2303
471fbbea
DR
2304
2305/*
2306 * Encrypted Casefolded entries require saving the hash on disk. This structure
2307 * followed ext4_dir_entry_2's name[name_len] at the next 4 byte aligned
2308 * boundary.
2309 */
2310struct ext4_dir_entry_hash {
2311 __le32 hash;
2312 __le32 minor_hash;
2313};
2314
ac27a0ec 2315/*
617ba13b 2316 * The new version of the directory entry. Since EXT4 structures are
ac27a0ec
DK
2317 * stored in intel byte order, and the name_len field could never be
2318 * bigger than 255 chars, it's safe to reclaim the extra byte for the
2319 * file_type field.
2320 */
617ba13b 2321struct ext4_dir_entry_2 {
ac27a0ec
DK
2322 __le32 inode; /* Inode number */
2323 __le16 rec_len; /* Directory entry length */
2324 __u8 name_len; /* Name length */
9f364e1d 2325 __u8 file_type; /* See file type macros EXT4_FT_* below */
617ba13b 2326 char name[EXT4_NAME_LEN]; /* File name */
ac27a0ec
DK
2327};
2328
471fbbea
DR
2329/*
2330 * Access the hashes at the end of ext4_dir_entry_2
2331 */
2332#define EXT4_DIRENT_HASHES(entry) \
2333 ((struct ext4_dir_entry_hash *) \
2334 (((void *)(entry)) + \
2335 ((8 + (entry)->name_len + EXT4_DIR_ROUND) & ~EXT4_DIR_ROUND)))
2336#define EXT4_DIRENT_HASH(entry) le32_to_cpu(EXT4_DIRENT_HASHES(de)->hash)
2337#define EXT4_DIRENT_MINOR_HASH(entry) \
2338 le32_to_cpu(EXT4_DIRENT_HASHES(de)->minor_hash)
2339
2340static inline bool ext4_hash_in_dirent(const struct inode *inode)
2341{
2342 return IS_CASEFOLDED(inode) && IS_ENCRYPTED(inode);
2343}
2344
e6153918
DW
2345/*
2346 * This is a bogus directory entry at the end of each leaf block that
2347 * records checksums.
2348 */
2349struct ext4_dir_entry_tail {
2350 __le32 det_reserved_zero1; /* Pretend to be unused */
2351 __le16 det_rec_len; /* 12 */
2352 __u8 det_reserved_zero2; /* Zero name length */
2353 __u8 det_reserved_ft; /* 0xDE, fake file type */
2354 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
2355};
2356
3c47d541
TM
2357#define EXT4_DIRENT_TAIL(block, blocksize) \
2358 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
2359 ((blocksize) - \
2360 sizeof(struct ext4_dir_entry_tail))))
2361
ac27a0ec 2362/*
617ba13b 2363 * Ext4 directory file types. Only the low 3 bits are used. The
ac27a0ec
DK
2364 * other bits are reserved for now.
2365 */
617ba13b
MC
2366#define EXT4_FT_UNKNOWN 0
2367#define EXT4_FT_REG_FILE 1
2368#define EXT4_FT_DIR 2
2369#define EXT4_FT_CHRDEV 3
2370#define EXT4_FT_BLKDEV 4
2371#define EXT4_FT_FIFO 5
2372#define EXT4_FT_SOCK 6
2373#define EXT4_FT_SYMLINK 7
ac27a0ec 2374
617ba13b 2375#define EXT4_FT_MAX 8
ac27a0ec 2376
e6153918
DW
2377#define EXT4_FT_DIR_CSUM 0xDE
2378
ac27a0ec 2379/*
617ba13b 2380 * EXT4_DIR_PAD defines the directory entries boundaries
ac27a0ec
DK
2381 *
2382 * NOTE: It must be a multiple of 4
2383 */
617ba13b
MC
2384#define EXT4_DIR_PAD 4
2385#define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
a72d7f83
JK
2386#define EXT4_MAX_REC_LEN ((1<<16)-1)
2387
471fbbea
DR
2388/*
2389 * The rec_len is dependent on the type of directory. Directories that are
2390 * casefolded and encrypted need to store the hash as well, so we add room for
2391 * ext4_extended_dir_entry_2. For all entries related to '.' or '..' you should
2392 * pass NULL for dir, as those entries do not use the extra fields.
2393 */
2394static inline unsigned int ext4_dir_rec_len(__u8 name_len,
2395 const struct inode *dir)
2396{
2397 int rec_len = (name_len + 8 + EXT4_DIR_ROUND);
2398
2399 if (dir && ext4_hash_in_dirent(dir))
2400 rec_len += sizeof(struct ext4_dir_entry_hash);
2401 return (rec_len & ~EXT4_DIR_ROUND);
2402}
2403
0cfc9255
ES
2404/*
2405 * If we ever get support for fs block sizes > page_size, we'll need
2406 * to remove the #if statements in the next two functions...
2407 */
2408static inline unsigned int
2409ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
2410{
2411 unsigned len = le16_to_cpu(dlen);
2412
ea1754a0 2413#if (PAGE_SIZE >= 65536)
0cfc9255
ES
2414 if (len == EXT4_MAX_REC_LEN || len == 0)
2415 return blocksize;
2416 return (len & 65532) | ((len & 3) << 16);
2417#else
2418 return len;
2419#endif
2420}
2421
2422static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
2423{
a2e3965d 2424 BUG_ON((len > blocksize) || (blocksize > (1 << 18)) || (len & 3));
ea1754a0 2425#if (PAGE_SIZE >= 65536)
0cfc9255
ES
2426 if (len < 65536)
2427 return cpu_to_le16(len);
2428 if (len == blocksize) {
2429 if (blocksize == 65536)
2430 return cpu_to_le16(EXT4_MAX_REC_LEN);
2431 else
2432 return cpu_to_le16(0);
2433 }
2434 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
2435#else
2436 return cpu_to_le16(len);
2437#endif
2438}
2439
ac27a0ec
DK
2440/*
2441 * Hash Tree Directory indexing
2442 * (c) Daniel Phillips, 2001
2443 */
2444
e2b911c5 2445#define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
12e9b892 2446 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
c7414892
AD
2447#define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \
2448 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir)))
617ba13b 2449#define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
ac27a0ec
DK
2450
2451/* Legal values for the dx_root hash_version field: */
2452
e7093f0d
PK
2453#define DX_HASH_LEGACY 0
2454#define DX_HASH_HALF_MD4 1
2455#define DX_HASH_TEA 2
2456#define DX_HASH_LEGACY_UNSIGNED 3
f99b2589
TT
2457#define DX_HASH_HALF_MD4_UNSIGNED 4
2458#define DX_HASH_TEA_UNSIGNED 5
471fbbea 2459#define DX_HASH_SIPHASH 6
ac27a0ec 2460
0441984a
DW
2461static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2462 const void *address, unsigned int length)
2463{
2464 struct {
2465 struct shash_desc shash;
3108b54b 2466 char ctx[4];
0441984a 2467 } desc;
0441984a 2468
3108b54b
TT
2469 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2470
0441984a 2471 desc.shash.tfm = sbi->s_chksum_driver;
0441984a
DW
2472 *(u32 *)desc.ctx = crc;
2473
e7093f0d 2474 BUG_ON(crypto_shash_update(&desc.shash, address, length));
0441984a
DW
2475
2476 return *(u32 *)desc.ctx;
2477}
2478
ac27a0ec
DK
2479#ifdef __KERNEL__
2480
2481/* hash info structure used by the directory hash */
2482struct dx_hash_info
2483{
2484 u32 hash;
2485 u32 minor_hash;
2486 int hash_version;
2487 u32 *seed;
2488};
2489
d1f5273e
FY
2490
2491/* 32 and 64 bit signed EOF for dx directories */
2492#define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
2493#define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
2494
ac27a0ec
DK
2495
2496/*
617ba13b 2497 * Control parameters used by ext4_htree_next_block
ac27a0ec
DK
2498 */
2499#define HASH_NB_ALWAYS 1
2500
5b643f9c
TT
2501struct ext4_filename {
2502 const struct qstr *usr_fname;
a7550b30 2503 struct fscrypt_str disk_name;
5b643f9c 2504 struct dx_hash_info hinfo;
643fa961 2505#ifdef CONFIG_FS_ENCRYPTION
a7550b30 2506 struct fscrypt_str crypto_buf;
5b643f9c 2507#endif
5298d4bf 2508#if IS_ENABLED(CONFIG_UNICODE)
3ae72562
GKB
2509 struct fscrypt_str cf_name;
2510#endif
5b643f9c
TT
2511};
2512
2513#define fname_name(p) ((p)->disk_name.name)
471fbbea 2514#define fname_usr_name(p) ((p)->usr_fname->name)
5b643f9c 2515#define fname_len(p) ((p)->disk_name.len)
ac27a0ec
DK
2516
2517/*
2518 * Describe an inode's exact location on disk and in memory
2519 */
617ba13b 2520struct ext4_iloc
ac27a0ec
DK
2521{
2522 struct buffer_head *bh;
2523 unsigned long offset;
fd2d4291 2524 ext4_group_t block_group;
ac27a0ec
DK
2525};
2526
617ba13b 2527static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
ac27a0ec 2528{
617ba13b 2529 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
ac27a0ec
DK
2530}
2531
dec214d0
TE
2532static inline bool ext4_is_quota_file(struct inode *inode)
2533{
2534 return IS_NOQUOTA(inode) &&
2535 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL);
2536}
02749a4c 2537
ac27a0ec
DK
2538/*
2539 * This structure is stuffed into the struct file's private_data field
2540 * for directories. It is where we put information so that we can do
2541 * readdir operations in hash tree order.
2542 */
2543struct dir_private_info {
2544 struct rb_root root;
2545 struct rb_node *curr_node;
2546 struct fname *extra_fname;
2547 loff_t last_pos;
2548 __u32 curr_hash;
2549 __u32 curr_minor_hash;
2550 __u32 next_hash;
2551};
2552
2553/* calculate the first block number of the group */
617ba13b 2554static inline ext4_fsblk_t
fd2d4291 2555ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
ac27a0ec 2556{
617ba13b
MC
2557 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2558 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
ac27a0ec
DK
2559}
2560
2561/*
2562 * Special error return code only used by dx_probe() and its callers.
2563 */
36de9286 2564#define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
ac27a0ec 2565
e08ac99f
AB
2566/* htree levels for ext4 */
2567#define EXT4_HTREE_LEVEL_COMPAT 2
2568#define EXT4_HTREE_LEVEL 3
2569
2570static inline int ext4_dir_htree_level(struct super_block *sb)
2571{
2572 return ext4_has_feature_largedir(sb) ?
2573 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT;
2574}
2575
bfff6873
LC
2576/*
2577 * Timeout and state flag for lazy initialization inode thread.
2578 */
2579#define EXT4_DEF_LI_WAIT_MULT 10
2580#define EXT4_DEF_LI_MAX_START_DELAY 5
2581#define EXT4_LAZYINIT_QUIT 0x0001
2582#define EXT4_LAZYINIT_RUNNING 0x0002
2583
2584/*
2585 * Lazy inode table initialization info
2586 */
2587struct ext4_lazy_init {
2588 unsigned long li_state;
bfff6873
LC
2589 struct list_head li_request_list;
2590 struct mutex li_list_mtx;
2591};
2592
3d392b26
TT
2593enum ext4_li_mode {
2594 EXT4_LI_MODE_PREFETCH_BBITMAP,
2595 EXT4_LI_MODE_ITABLE,
2596};
2597
bfff6873
LC
2598struct ext4_li_request {
2599 struct super_block *lr_super;
3d392b26
TT
2600 enum ext4_li_mode lr_mode;
2601 ext4_group_t lr_first_not_zeroed;
bfff6873
LC
2602 ext4_group_t lr_next_group;
2603 struct list_head lr_request;
2604 unsigned long lr_next_sched;
2605 unsigned long lr_timeout;
2606};
2607
857ac889
LC
2608struct ext4_features {
2609 struct kobject f_kobj;
2610 struct completion f_kobj_unregister;
2611};
2612
c5e06d10
JL
2613/*
2614 * This structure will be used for multiple mount protection. It will be
2615 * written into the block number saved in the s_mmp_block field in the
2616 * superblock. Programs that check MMP should assume that if
2617 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2618 * to use the filesystem, regardless of how old the timestamp is.
2619 */
2620#define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
2621#define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2622#define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
2623#define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
2624
2625struct mmp_struct {
2626 __le32 mmp_magic; /* Magic number for MMP */
2627 __le32 mmp_seq; /* Sequence no. updated periodically */
2628
2629 /*
2630 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2631 * purposes and do not affect the correctness of the algorithm
2632 */
2633 __le64 mmp_time; /* Time last updated */
2634 char mmp_nodename[64]; /* Node which last updated MMP block */
2635 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
2636
2637 /*
2638 * mmp_check_interval is used to verify if the MMP block has been
2639 * updated on the block device. The value is updated based on the
2640 * maximum time to write the MMP block during an update cycle.
2641 */
2642 __le16 mmp_check_interval;
2643
2644 __le16 mmp_pad1;
e6153918
DW
2645 __le32 mmp_pad2[226];
2646 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
c5e06d10
JL
2647};
2648
2649/* arguments passed to the mmp thread */
2650struct mmpd_data {
2651 struct buffer_head *bh; /* bh from initial read_mmp_block() */
2652 struct super_block *sb; /* super block of the fs */
2653};
2654
2655/*
2656 * Check interval multiplier
2657 * The MMP block is written every update interval and initially checked every
2658 * update interval x the multiplier (the value is then adapted based on the
2659 * write latency). The reason is that writes can be delayed under load and we
2660 * don't want readers to incorrectly assume that the filesystem is no longer
2661 * in use.
2662 */
2663#define EXT4_MMP_CHECK_MULT 2UL
2664
2665/*
2666 * Minimum interval for MMP checking in seconds.
2667 */
2668#define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
2669
2670/*
2671 * Maximum interval for MMP checking in seconds.
2672 */
2673#define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
2674
ac27a0ec
DK
2675/*
2676 * Function prototypes
2677 */
2678
2679/*
2680 * Ok, these declarations are also in <linux/kernel.h> but none of the
617ba13b 2681 * ext4 source programs needs to include it so they are duplicated here.
ac27a0ec 2682 */
8c55e204
DK
2683# define NORET_TYPE /**/
2684# define ATTRIB_NORET __attribute__((noreturn))
2685# define NORET_AND noreturn,
ac27a0ec 2686
498e5f24 2687/* bitmap.c */
f6fb99ca 2688extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
4fd873c8 2689void ext4_inode_bitmap_csum_set(struct super_block *sb,
41a246d1
DW
2690 struct ext4_group_desc *gdp,
2691 struct buffer_head *bh, int sz);
b83acc77 2692int ext4_inode_bitmap_csum_verify(struct super_block *sb,
41a246d1
DW
2693 struct ext4_group_desc *gdp,
2694 struct buffer_head *bh, int sz);
1df9bde4 2695void ext4_block_bitmap_csum_set(struct super_block *sb,
fa77dcfa 2696 struct ext4_group_desc *gdp,
79f1ba49 2697 struct buffer_head *bh);
82483dfe 2698int ext4_block_bitmap_csum_verify(struct super_block *sb,
fa77dcfa 2699 struct ext4_group_desc *gdp,
79f1ba49 2700 struct buffer_head *bh);
498e5f24 2701
ac27a0ec 2702/* balloc.c */
bd86298e
LC
2703extern void ext4_get_group_no_and_offset(struct super_block *sb,
2704 ext4_fsblk_t blocknr,
2705 ext4_group_t *blockgrpp,
2706 ext4_grpblk_t *offsetp);
2707extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2708 ext4_fsblk_t block);
2709
fd2d4291
AM
2710extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2711extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2712 ext4_group_t group);
7061eba7 2713extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
55f020db
AH
2714 ext4_fsblk_t goal,
2715 unsigned int flags,
2716 unsigned long *count,
2717 int *errp);
e7d5f315
TT
2718extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2719 s64 nclusters, unsigned int flags);
5dee5437 2720extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
617ba13b 2721extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
fd2d4291 2722 ext4_group_t block_group,
ac27a0ec 2723 struct buffer_head ** bh);
5354b2af
TT
2724extern struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2725 ext4_group_t group);
617ba13b 2726extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
813e5727
TT
2727
2728extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
cfd73237
AZ
2729 ext4_group_t block_group,
2730 bool ignore_locked);
813e5727
TT
2731extern int ext4_wait_block_bitmap(struct super_block *sb,
2732 ext4_group_t block_group,
2733 struct buffer_head *bh);
2734extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2735 ext4_group_t block_group);
cff1dfd7
TT
2736extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2737 ext4_group_t block_group,
2738 struct ext4_group_desc *gdp);
f86186b4 2739ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
ac27a0ec 2740
5298d4bf 2741#if IS_ENABLED(CONFIG_UNICODE)
1ae98e29 2742extern int ext4_fname_setup_ci_filename(struct inode *dir,
3ae72562 2743 const struct qstr *iname,
1ae98e29 2744 struct ext4_filename *fname);
3ae72562
GKB
2745#endif
2746
b1241c8e 2747/* ext4 encryption related stuff goes here crypto.c */
643fa961 2748#ifdef CONFIG_FS_ENCRYPTION
b1241c8e
RH
2749extern const struct fscrypt_operations ext4_cryptops;
2750
3030b59c
RH
2751int ext4_fname_setup_filename(struct inode *dir, const struct qstr *iname,
2752 int lookup, struct ext4_filename *fname);
d5d0e8c7 2753
3030b59c
RH
2754int ext4_fname_prepare_lookup(struct inode *dir, struct dentry *dentry,
2755 struct ext4_filename *fname);
b7236e21 2756
3030b59c 2757void ext4_fname_free_filename(struct ext4_filename *fname);
3ae72562 2758
72f63f4a
RH
2759int ext4_ioctl_get_encryption_pwsalt(struct file *filp, void __user *arg);
2760
b01531db 2761#else /* !CONFIG_FS_ENCRYPTION */
a7550b30 2762static inline int ext4_fname_setup_filename(struct inode *dir,
b01531db
EB
2763 const struct qstr *iname,
2764 int lookup,
2765 struct ext4_filename *fname)
b7236e21 2766{
1ae98e29 2767 int err = 0;
a7550b30
JK
2768 fname->usr_fname = iname;
2769 fname->disk_name.name = (unsigned char *) iname->name;
2770 fname->disk_name.len = iname->len;
3ae72562 2771
5298d4bf 2772#if IS_ENABLED(CONFIG_UNICODE)
1ae98e29 2773 err = ext4_fname_setup_ci_filename(dir, iname, fname);
3ae72562
GKB
2774#endif
2775
1ae98e29 2776 return err;
b7236e21 2777}
88bd6ccd 2778
b01531db
EB
2779static inline int ext4_fname_prepare_lookup(struct inode *dir,
2780 struct dentry *dentry,
2781 struct ext4_filename *fname)
2782{
2783 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname);
2784}
2785
3ae72562
GKB
2786static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2787{
5298d4bf 2788#if IS_ENABLED(CONFIG_UNICODE)
3ae72562
GKB
2789 kfree(fname->cf_name.name);
2790 fname->cf_name.name = NULL;
2791#endif
2792}
72f63f4a
RH
2793
2794static inline int ext4_ioctl_get_encryption_pwsalt(struct file *filp,
2795 void __user *arg)
2796{
2797 return -EOPNOTSUPP;
2798}
b01531db 2799#endif /* !CONFIG_FS_ENCRYPTION */
88bd6ccd 2800
ac27a0ec 2801/* dir.c */
60fd4da3 2802extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
f7c21177 2803 struct file *,
60fd4da3 2804 struct ext4_dir_entry_2 *,
226ba972
TM
2805 struct buffer_head *, char *, int,
2806 unsigned int);
471fbbea 2807#define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
f7c21177 2808 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
471fbbea 2809 (de), (bh), (buf), (size), (offset)))
617ba13b 2810extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2f61830a
TT
2811 __u32 minor_hash,
2812 struct ext4_dir_entry_2 *dirent,
a7550b30 2813 struct fscrypt_str *ent_name);
617ba13b 2814extern void ext4_htree_free_dir_info(struct dir_private_info *p);
978fef91
TM
2815extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2816 struct buffer_head *bh,
2817 void *buf, int buf_size,
5b643f9c 2818 struct ext4_filename *fname,
978fef91 2819 struct ext4_dir_entry_2 **dest_de);
471fbbea 2820void ext4_insert_dentry(struct inode *dir, struct inode *inode,
1bc0af60
EB
2821 struct ext4_dir_entry_2 *de,
2822 int buf_size,
2823 struct ext4_filename *fname);
978fef91
TM
2824static inline void ext4_update_dx_flag(struct inode *inode)
2825{
f902b216
JK
2826 if (!ext4_has_feature_dir_index(inode->i_sb) &&
2827 ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) {
48a34311
JK
2828 /* ext4_iget() should have caught this... */
2829 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb));
978fef91 2830 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
48a34311 2831 }
978fef91 2832}
d6006186 2833static const unsigned char ext4_filetype_table[] = {
65d165d9
TM
2834 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2835};
2836
2837static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2838{
e2b911c5 2839 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
65d165d9
TM
2840 return DT_UNKNOWN;
2841
2842 return ext4_filetype_table[filetype];
2843}
40b163f1
DW
2844extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2845 void *buf, int buf_size);
ac27a0ec
DK
2846
2847/* fsync.c */
02c24a82 2848extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
ac27a0ec
DK
2849
2850/* hash.c */
b886ee3e
GKB
2851extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len,
2852 struct dx_hash_info *hinfo);
ac27a0ec
DK
2853
2854/* ialloc.c */
8016e29f 2855extern int ext4_mark_inode_used(struct super_block *sb, int ino);
f2d40141 2856extern struct inode *__ext4_new_inode(struct mnt_idmap *, handle_t *,
14f3db55 2857 struct inode *, umode_t,
1139575a 2858 const struct qstr *qstr, __u32 goal,
1b917ed8
TE
2859 uid_t *owner, __u32 i_flags,
2860 int handle_type, unsigned int line_no,
2861 int nblocks);
1139575a 2862
14f3db55 2863#define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \
f2d40141 2864 __ext4_new_inode(&nop_mnt_idmap, (handle), (dir), (mode), (qstr), \
14f3db55 2865 (goal), (owner), i_flags, 0, 0, 0)
f2d40141 2866#define ext4_new_inode_start_handle(idmap, dir, mode, qstr, goal, owner, \
1139575a 2867 type, nblocks) \
f2d40141 2868 __ext4_new_inode((idmap), NULL, (dir), (mode), (qstr), (goal), (owner), \
1b917ed8 2869 0, (type), __LINE__, (nblocks))
1139575a
TT
2870
2871
af5bc92d
TT
2872extern void ext4_free_inode(handle_t *, struct inode *);
2873extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2874extern unsigned long ext4_count_free_inodes(struct super_block *);
2875extern unsigned long ext4_count_dirs(struct super_block *);
61d08673 2876extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
bfff6873
LC
2877extern int ext4_init_inode_table(struct super_block *sb,
2878 ext4_group_t group, int barrier);
813e5727 2879extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
ac27a0ec 2880
6866d7b3 2881/* fast_commit.c */
ce8c59d1 2882int ext4_fc_info_show(struct seq_file *seq, void *v);
6866d7b3 2883void ext4_fc_init(struct super_block *sb, journal_t *journal);
aa75f4d3 2884void ext4_fc_init_inode(struct inode *inode);
a80f7fcf 2885void ext4_fc_track_range(handle_t *handle, struct inode *inode, ext4_lblk_t start,
aa75f4d3 2886 ext4_lblk_t end);
a80f7fcf
HS
2887void __ext4_fc_track_unlink(handle_t *handle, struct inode *inode,
2888 struct dentry *dentry);
2889void __ext4_fc_track_link(handle_t *handle, struct inode *inode,
2890 struct dentry *dentry);
2891void ext4_fc_track_unlink(handle_t *handle, struct dentry *dentry);
2892void ext4_fc_track_link(handle_t *handle, struct dentry *dentry);
8210bb29
HS
2893void __ext4_fc_track_create(handle_t *handle, struct inode *inode,
2894 struct dentry *dentry);
a80f7fcf
HS
2895void ext4_fc_track_create(handle_t *handle, struct dentry *dentry);
2896void ext4_fc_track_inode(handle_t *handle, struct inode *inode);
e85c81ba 2897void ext4_fc_mark_ineligible(struct super_block *sb, int reason, handle_t *handle);
aa75f4d3
HS
2898void ext4_fc_start_update(struct inode *inode);
2899void ext4_fc_stop_update(struct inode *inode);
2900void ext4_fc_del(struct inode *inode);
8016e29f
HS
2901bool ext4_fc_replay_check_excluded(struct super_block *sb, ext4_fsblk_t block);
2902void ext4_fc_replay_cleanup(struct super_block *sb);
aa75f4d3
HS
2903int ext4_fc_commit(journal_t *journal, tid_t commit_tid);
2904int __init ext4_fc_init_dentry_cache(void);
ab047d51 2905void ext4_fc_destroy_dentry_cache(void);
599ea31d
XY
2906int ext4_fc_record_regions(struct super_block *sb, int ino,
2907 ext4_lblk_t lblk, ext4_fsblk_t pblk,
2908 int len, int replay);
aa75f4d3 2909
c9de560d 2910/* mballoc.c */
247dbed8 2911extern const struct seq_operations ext4_mb_seq_groups_ops;
f68f4063 2912extern const struct seq_operations ext4_mb_seq_structs_summary_ops;
a6c75eaf 2913extern int ext4_seq_mb_stats_show(struct seq_file *seq, void *offset);
9d99012f 2914extern int ext4_mb_init(struct super_block *);
c9de560d
AT
2915extern int ext4_mb_release(struct super_block *);
2916extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2917 struct ext4_allocation_request *, int *);
27bc446e 2918extern void ext4_discard_preallocations(struct inode *, unsigned int);
5dabfc78
TT
2919extern int __init ext4_init_mballoc(void);
2920extern void ext4_exit_mballoc(void);
3d392b26
TT
2921extern ext4_group_t ext4_mb_prefetch(struct super_block *sb,
2922 ext4_group_t group,
2923 unsigned int nr, int *cnt);
2924extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group,
2925 unsigned int nr);
2926
44338711 2927extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
e6362609
TT
2928 struct buffer_head *bh, ext4_fsblk_t block,
2929 unsigned long count, int flags);
28623c2f
TT
2930extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2931 ext4_group_t ngroups);
920313a7 2932extern int ext4_mb_add_groupinfo(struct super_block *sb,
5f21b0e6 2933 ext4_group_t i, struct ext4_group_desc *desc);
cc7365df 2934extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2846e820 2935 ext4_fsblk_t block, unsigned long count);
7360d173 2936extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
a0154344 2937extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid);
8016e29f 2938extern void ext4_mb_mark_bb(struct super_block *sb, ext4_fsblk_t block,
d2f7cf40 2939 int len, bool state);
304749c0
OM
2940static inline bool ext4_mb_cr_expensive(enum criteria cr)
2941{
2942 return cr >= CR_GOAL_LEN_SLOW;
2943}
7360d173 2944
ac27a0ec 2945/* inode.c */
8016e29f
HS
2946void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw,
2947 struct ext4_inode_info *ei);
f348c252 2948int ext4_inode_is_fast_symlink(struct inode *inode);
10560082 2949struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
1c215028 2950struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
9699d4f9
TE
2951int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count,
2952 bool wait, struct buffer_head **bhs);
705965bd
JK
2953int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2954 struct buffer_head *bh_result, int create);
6873fa0d 2955int ext4_get_block(struct inode *inode, sector_t iblock,
705965bd 2956 struct buffer_head *bh_result, int create);
9c3569b5
TM
2957int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2958 struct buffer_head *bh, int create);
f19d5870 2959int ext4_walk_page_buffers(handle_t *handle,
188c299e 2960 struct inode *inode,
f19d5870
TM
2961 struct buffer_head *head,
2962 unsigned from,
2963 unsigned to,
2964 int *partial,
188c299e 2965 int (*fn)(handle_t *handle, struct inode *inode,
f19d5870 2966 struct buffer_head *bh));
188c299e 2967int do_journal_get_write_access(handle_t *handle, struct inode *inode,
f19d5870 2968 struct buffer_head *bh);
9c3569b5
TM
2969#define FALL_BACK_TO_NONDELALLOC 1
2970#define CONVERT_INLINE_DATA 2
ac27a0ec 2971
8a363970
TT
2972typedef enum {
2973 EXT4_IGET_NORMAL = 0,
2974 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */
63b1e9bc 2975 EXT4_IGET_HANDLE = 0x0002, /* Inode # is from a handle */
b3e6bcb9
TT
2976 EXT4_IGET_BAD = 0x0004, /* Allow to iget a bad inode */
2977 EXT4_IGET_EA_INODE = 0x0008 /* Inode should contain an EA value */
8a363970
TT
2978} ext4_iget_flags;
2979
2980extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
2981 ext4_iget_flags flags, const char *function,
2982 unsigned int line);
2983
2984#define ext4_iget(sb, ino, flags) \
2985 __ext4_iget((sb), (ino), (flags), __func__, __LINE__)
2986
a9185b41 2987extern int ext4_write_inode(struct inode *, struct writeback_control *);
c1632a0f 2988extern int ext4_setattr(struct mnt_idmap *, struct dentry *,
549c7297 2989 struct iattr *);
8434ef1d 2990extern u32 ext4_dio_alignment(struct inode *inode);
b74d24f7 2991extern int ext4_getattr(struct mnt_idmap *, const struct path *,
549c7297 2992 struct kstat *, u32, unsigned int);
0930fcc1
AV
2993extern void ext4_evict_inode(struct inode *);
2994extern void ext4_clear_inode(struct inode *);
b74d24f7 2995extern int ext4_file_getattr(struct mnt_idmap *, const struct path *,
549c7297 2996 struct kstat *, u32, unsigned int);
aa385729 2997extern void ext4_dirty_inode(struct inode *, int);
617ba13b
MC
2998extern int ext4_change_inode_journal_flag(struct inode *, int);
2999extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
8016e29f
HS
3000extern int ext4_get_fc_inode_loc(struct super_block *sb, unsigned long ino,
3001 struct ext4_iloc *iloc);
a361293f 3002extern int ext4_inode_attach_jinode(struct inode *inode);
91ef4caf 3003extern int ext4_can_truncate(struct inode *inode);
2c98eb5e 3004extern int ext4_truncate(struct inode *);
430657b6 3005extern int ext4_break_layouts(struct inode *);
ad5cd4f4 3006extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length);
043546e4 3007extern void ext4_set_inode_flags(struct inode *, bool init);
ccd2506b 3008extern int ext4_alloc_da_blocks(struct inode *inode);
617ba13b 3009extern void ext4_set_aops(struct inode *inode);
a86c6181 3010extern int ext4_writepage_trans_blocks(struct inode *);
59205c8d 3011extern int ext4_normal_submit_inode_data_buffers(struct jbd2_inode *jinode);
f3bd1f3f 3012extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
a87dd18c
LC
3013extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
3014 loff_t lstart, loff_t lend);
401b25aa 3015extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf);
a9e7f447 3016extern qsize_t *ext4_get_reserved_space(struct inode *inode);
040cb378 3017extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
f456767d 3018extern void ext4_da_release_space(struct inode *inode, int to_free);
5f634d06
AK
3019extern void ext4_da_update_reserve_space(struct inode *inode,
3020 int used, int quota_claim);
53085fac
JK
3021extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
3022 ext4_fsblk_t pblk, ext4_lblk_t len);
dae1e52c
AG
3023
3024/* indirect.c */
3025extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
3026 struct ext4_map_blocks *map, int flags);
fa55a0ed 3027extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
819c4920 3028extern void ext4_ind_truncate(handle_t *, struct inode *inode);
4f579ae7
LC
3029extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
3030 ext4_lblk_t start, ext4_lblk_t end);
ff9893dc 3031
ac27a0ec 3032/* ioctl.c */
5cdd7b2d 3033extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
af5bc92d 3034extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
8782a9ae 3035int ext4_fileattr_set(struct mnt_idmap *idmap,
4db5c2e6
MS
3036 struct dentry *dentry, struct fileattr *fa);
3037int ext4_fileattr_get(struct dentry *dentry, struct fileattr *fa);
8016e29f 3038extern void ext4_reset_inode_seed(struct inode *inode);
827891a3 3039int ext4_update_overhead(struct super_block *sb, bool force);
97524b45 3040int ext4_force_shutdown(struct super_block *sb, u32 flags);
ac27a0ec 3041
c14c6fd5 3042/* migrate.c */
2a43a878 3043extern int ext4_ext_migrate(struct inode *);
0d14b098 3044extern int ext4_ind_migrate(struct inode *inode);
3d0518f4 3045
ac27a0ec 3046/* namei.c */
8016e29f
HS
3047extern int ext4_init_new_dir(handle_t *handle, struct inode *dir,
3048 struct inode *inode);
f036adb3
TT
3049extern int ext4_dirblock_csum_verify(struct inode *inode,
3050 struct buffer_head *bh);
617ba13b 3051extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
ac27a0ec 3052 __u32 start_minor_hash, __u32 *next_hash);
5b643f9c
TT
3053extern int ext4_search_dir(struct buffer_head *bh,
3054 char *search_buf,
3055 int buf_size,
3056 struct inode *dir,
3057 struct ext4_filename *fname,
5b643f9c
TT
3058 unsigned int offset,
3059 struct ext4_dir_entry_2 **res_dir);
2fe34d29 3060extern int ext4_generic_delete_entry(struct inode *dir,
05019a9e
TM
3061 struct ext4_dir_entry_2 *de_del,
3062 struct buffer_head *bh,
3063 void *entry_buf,
3064 int buf_size,
3065 int csum_size);
a7550b30 3066extern bool ext4_empty_dir(struct inode *inode);
ac27a0ec
DK
3067
3068/* resize.c */
1d0c3924 3069extern void ext4_kvfree_array_rcu(void *to_free);
617ba13b
MC
3070extern int ext4_group_add(struct super_block *sb,
3071 struct ext4_new_group_data *input);
3072extern int ext4_group_extend(struct super_block *sb,
3073 struct ext4_super_block *es,
3074 ext4_fsblk_t n_blocks_count);
19c5246d 3075extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
bbc605cd
LC
3076extern unsigned int ext4_list_backups(struct super_block *sb,
3077 unsigned int *three, unsigned int *five,
3078 unsigned int *seven);
ac27a0ec
DK
3079
3080/* super.c */
fb265c9c 3081extern struct buffer_head *ext4_sb_bread(struct super_block *sb,
67c0f556 3082 sector_t block, blk_opf_t op_flags);
8394a6ab 3083extern struct buffer_head *ext4_sb_bread_unmovable(struct super_block *sb,
3084 sector_t block);
67c0f556 3085extern void ext4_read_bh_nowait(struct buffer_head *bh, blk_opf_t op_flags,
fa491b14 3086 bh_end_io_t *end_io);
67c0f556 3087extern int ext4_read_bh(struct buffer_head *bh, blk_opf_t op_flags,
fa491b14 3088 bh_end_io_t *end_io);
67c0f556 3089extern int ext4_read_bh_lock(struct buffer_head *bh, blk_opf_t op_flags, bool wait);
5df1d412 3090extern void ext4_sb_breadahead_unmovable(struct super_block *sb, sector_t block);
ebd173be 3091extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
952fc18e 3092extern int ext4_calculate_overhead(struct super_block *sb);
bbc605cd
LC
3093extern __le32 ext4_superblock_csum(struct super_block *sb,
3094 struct ext4_super_block *es);
06db49e6 3095extern void ext4_superblock_csum_set(struct super_block *sb);
117fff10
TT
3096extern int ext4_alloc_flex_bg_array(struct super_block *sb,
3097 ext4_group_t ngroup);
722887dd
TT
3098extern const char *ext4_decode_error(struct super_block *sb, int errno,
3099 char nbuf[16]);
db79e6d1
WS
3100extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb,
3101 ext4_group_t block_group,
3102 unsigned int flags);
68228da5
WJ
3103extern unsigned int ext4_num_base_meta_blocks(struct super_block *sb,
3104 ext4_group_t block_group);
e7c96e8e 3105
014c9caa
JK
3106extern __printf(7, 8)
3107void __ext4_error(struct super_block *, const char *, unsigned int, bool,
3108 int, __u64, const char *, ...);
54d3adbc
TT
3109extern __printf(6, 7)
3110void __ext4_error_inode(struct inode *, const char *, unsigned int,
3111 ext4_fsblk_t, int, const char *, ...);
b9075fa9 3112extern __printf(5, 6)
e7c96e8e 3113void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
b9075fa9 3114 const char *, ...);
c398eda0
TT
3115extern void __ext4_std_error(struct super_block *, const char *,
3116 unsigned int, int);
b9075fa9
JP
3117extern __printf(4, 5)
3118void __ext4_warning(struct super_block *, const char *, unsigned int,
3119 const char *, ...);
b03a2f7e
AD
3120extern __printf(4, 5)
3121void __ext4_warning_inode(const struct inode *inode, const char *function,
3122 unsigned int line, const char *fmt, ...);
b9075fa9 3123extern __printf(3, 4)
e7c96e8e 3124void __ext4_msg(struct super_block *, const char *, const char *, ...);
c5e06d10
JL
3125extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
3126 const char *, unsigned int, const char *);
b9075fa9
JP
3127extern __printf(7, 8)
3128void __ext4_grp_locked_error(const char *, unsigned int,
3129 struct super_block *, ext4_group_t,
3130 unsigned long, ext4_fsblk_t,
3131 const char *, ...);
e7c96e8e 3132
b03a2f7e
AD
3133#define EXT4_ERROR_INODE(inode, fmt, a...) \
3134 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
3135
54d3adbc
TT
3136#define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...) \
3137 __ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a)
3138
3139#define ext4_error_inode_block(inode, block, err, fmt, a...) \
3140 __ext4_error_inode((inode), __func__, __LINE__, (block), (err), \
3141 (fmt), ## a)
b03a2f7e
AD
3142
3143#define EXT4_ERROR_FILE(file, block, fmt, a...) \
3144 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
3145
014c9caa
JK
3146#define ext4_abort(sb, err, fmt, a...) \
3147 __ext4_error((sb), __func__, __LINE__, true, (err), 0, (fmt), ## a)
3148
e7c96e8e
JP
3149#ifdef CONFIG_PRINTK
3150
3151#define ext4_error_inode(inode, func, line, block, fmt, ...) \
54d3adbc
TT
3152 __ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__)
3153#define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \
3154 __ext4_error_inode((inode), (func), (line), (block), \
3155 (err), (fmt), ##__VA_ARGS__)
e7c96e8e
JP
3156#define ext4_error_file(file, func, line, block, fmt, ...) \
3157 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
3158#define ext4_error(sb, fmt, ...) \
014c9caa
JK
3159 __ext4_error((sb), __func__, __LINE__, false, 0, 0, (fmt), \
3160 ##__VA_ARGS__)
54d3adbc 3161#define ext4_error_err(sb, err, fmt, ...) \
014c9caa
JK
3162 __ext4_error((sb), __func__, __LINE__, false, (err), 0, (fmt), \
3163 ##__VA_ARGS__)
e7c96e8e
JP
3164#define ext4_warning(sb, fmt, ...) \
3165 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
b03a2f7e
AD
3166#define ext4_warning_inode(inode, fmt, ...) \
3167 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
e7c96e8e
JP
3168#define ext4_msg(sb, level, fmt, ...) \
3169 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
3170#define dump_mmp_msg(sb, mmp, msg) \
3171 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
3172#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
3173 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
3174 fmt, ##__VA_ARGS__)
3175
3176#else
3177
3178#define ext4_error_inode(inode, func, line, block, fmt, ...) \
3179do { \
3180 no_printk(fmt, ##__VA_ARGS__); \
54d3adbc
TT
3181 __ext4_error_inode(inode, "", 0, block, 0, " "); \
3182} while (0)
3183#define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \
3184do { \
3185 no_printk(fmt, ##__VA_ARGS__); \
3186 __ext4_error_inode(inode, "", 0, block, err, " "); \
e7c96e8e
JP
3187} while (0)
3188#define ext4_error_file(file, func, line, block, fmt, ...) \
3189do { \
3190 no_printk(fmt, ##__VA_ARGS__); \
3191 __ext4_error_file(file, "", 0, block, " "); \
3192} while (0)
3193#define ext4_error(sb, fmt, ...) \
3194do { \
3195 no_printk(fmt, ##__VA_ARGS__); \
014c9caa 3196 __ext4_error(sb, "", 0, false, 0, 0, " "); \
54d3adbc
TT
3197} while (0)
3198#define ext4_error_err(sb, err, fmt, ...) \
3199do { \
3200 no_printk(fmt, ##__VA_ARGS__); \
014c9caa 3201 __ext4_error(sb, "", 0, false, err, 0, " "); \
e7c96e8e
JP
3202} while (0)
3203#define ext4_warning(sb, fmt, ...) \
3204do { \
3205 no_printk(fmt, ##__VA_ARGS__); \
3206 __ext4_warning(sb, "", 0, " "); \
3207} while (0)
b03a2f7e
AD
3208#define ext4_warning_inode(inode, fmt, ...) \
3209do { \
3210 no_printk(fmt, ##__VA_ARGS__); \
3211 __ext4_warning_inode(inode, "", 0, " "); \
3212} while (0)
e7c96e8e
JP
3213#define ext4_msg(sb, level, fmt, ...) \
3214do { \
3215 no_printk(fmt, ##__VA_ARGS__); \
3216 __ext4_msg(sb, "", " "); \
3217} while (0)
3218#define dump_mmp_msg(sb, mmp, msg) \
3219 __dump_mmp_msg(sb, mmp, "", 0, "")
3220#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
3221do { \
3222 no_printk(fmt, ##__VA_ARGS__); \
3223 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
3224} while (0)
3225
3226#endif
3227
8fadc143
AR
3228extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
3229 struct ext4_group_desc *bg);
3230extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
3231 struct ext4_group_desc *bg);
3232extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
3233 struct ext4_group_desc *bg);
021b65bb
TT
3234extern __u32 ext4_free_group_clusters(struct super_block *sb,
3235 struct ext4_group_desc *bg);
560671a0
AK
3236extern __u32 ext4_free_inodes_count(struct super_block *sb,
3237 struct ext4_group_desc *bg);
3238extern __u32 ext4_used_dirs_count(struct super_block *sb,
3239 struct ext4_group_desc *bg);
3240extern __u32 ext4_itable_unused_count(struct super_block *sb,
3241 struct ext4_group_desc *bg);
8fadc143
AR
3242extern void ext4_block_bitmap_set(struct super_block *sb,
3243 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3244extern void ext4_inode_bitmap_set(struct super_block *sb,
3245 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3246extern void ext4_inode_table_set(struct super_block *sb,
3247 struct ext4_group_desc *bg, ext4_fsblk_t blk);
021b65bb
TT
3248extern void ext4_free_group_clusters_set(struct super_block *sb,
3249 struct ext4_group_desc *bg,
3250 __u32 count);
560671a0
AK
3251extern void ext4_free_inodes_set(struct super_block *sb,
3252 struct ext4_group_desc *bg, __u32 count);
3253extern void ext4_used_dirs_set(struct super_block *sb,
3254 struct ext4_group_desc *bg, __u32 count);
3255extern void ext4_itable_unused_set(struct super_block *sb,
3256 struct ext4_group_desc *bg, __u32 count);
feb0ab32 3257extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
bb23c20a 3258 struct ext4_group_desc *gdp);
feb0ab32
DW
3259extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
3260 struct ext4_group_desc *gdp);
7f511862
TT
3261extern int ext4_register_li_request(struct super_block *sb,
3262 ext4_group_t first_not_zeroed);
bd81d8ee 3263
9aa5d32b
DM
3264static inline int ext4_has_metadata_csum(struct super_block *sb)
3265{
e2b911c5 3266 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
9aa5d32b
DM
3267 !EXT4_SB(sb)->s_chksum_driver);
3268
dec214d0
TE
3269 return ext4_has_feature_metadata_csum(sb) &&
3270 (EXT4_SB(sb)->s_chksum_driver != NULL);
9aa5d32b 3271}
dec214d0
TE
3272
3273static inline int ext4_has_group_desc_csum(struct super_block *sb)
3274{
3275 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb);
3276}
3277
81e8c3c5
PM
3278#define ext4_read_incompat_64bit_val(es, name) \
3279 (((es)->s_feature_incompat & cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT) \
3280 ? (ext4_fsblk_t)le32_to_cpu(es->name##_hi) << 32 : 0) | \
3281 le32_to_cpu(es->name##_lo))
3282
bd81d8ee
LV
3283static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
3284{
81e8c3c5 3285 return ext4_read_incompat_64bit_val(es, s_blocks_count);
bd81d8ee
LV
3286}
3287
3288static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
3289{
81e8c3c5 3290 return ext4_read_incompat_64bit_val(es, s_r_blocks_count);
bd81d8ee
LV
3291}
3292
3293static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
3294{
81e8c3c5 3295 return ext4_read_incompat_64bit_val(es, s_free_blocks_count);
bd81d8ee
LV
3296}
3297
3298static inline void ext4_blocks_count_set(struct ext4_super_block *es,
3299 ext4_fsblk_t blk)
3300{
6bc9feff 3301 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
bd81d8ee
LV
3302 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
3303}
3304
3305static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
3306 ext4_fsblk_t blk)
3307{
308ba3ec 3308 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
bd81d8ee
LV
3309 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
3310}
3311
3312static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
3313 ext4_fsblk_t blk)
3314{
308ba3ec 3315 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
bd81d8ee
LV
3316 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
3317}
3318
e08ac99f
AB
3319static inline loff_t ext4_isize(struct super_block *sb,
3320 struct ext4_inode *raw_inode)
a48380f7 3321{
e08ac99f
AB
3322 if (ext4_has_feature_largedir(sb) ||
3323 S_ISREG(le16_to_cpu(raw_inode->i_mode)))
06a279d6
TT
3324 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
3325 le32_to_cpu(raw_inode->i_size_lo);
e08ac99f
AB
3326
3327 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
a48380f7 3328}
bd81d8ee 3329
a48380f7
AK
3330static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
3331{
3332 raw_inode->i_size_lo = cpu_to_le32(i_size);
3333 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
3334}
ac27a0ec 3335
8df9675f
TT
3336/*
3337 * Reading s_groups_count requires using smp_rmb() afterwards. See
3338 * the locking protocol documented in the comments of ext4_group_add()
3339 * in resize.c
3340 */
3341static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
3342{
3343 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
3344
3345 smp_rmb();
3346 return ngroups;
3347}
c9de560d 3348
772cb7c8
JS
3349static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
3350 ext4_group_t block_group)
3351{
3352 return block_group >> sbi->s_log_groups_per_flex;
3353}
3354
3355static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
3356{
3357 return 1 << sbi->s_log_groups_per_flex;
3358}
3359
617ba13b 3360#define ext4_std_error(sb, errno) \
ac27a0ec
DK
3361do { \
3362 if ((errno)) \
c398eda0 3363 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
ac27a0ec
DK
3364} while (0)
3365
a30d542a 3366#ifdef CONFIG_SMP
df55c99d
TT
3367/* Each CPU can accumulate percpu_counter_batch clusters in their local
3368 * counters. So we need to make sure we have free clusters more
179f7ebf 3369 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
a30d542a 3370 */
df55c99d 3371#define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
a30d542a 3372#else
df55c99d 3373#define EXT4_FREECLUSTERS_WATERMARK 0
a30d542a
AK
3374#endif
3375
f340b3d9 3376/* Update i_disksize. Requires i_rwsem to avoid races with truncate */
cf17fea6
AK
3377static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
3378{
90e775b7 3379 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
5955102c 3380 !inode_is_locked(inode));
90e775b7
JK
3381 down_write(&EXT4_I(inode)->i_data_sem);
3382 if (newsize > EXT4_I(inode)->i_disksize)
35df4299 3383 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize);
90e775b7
JK
3384 up_write(&EXT4_I(inode)->i_data_sem);
3385}
3386
f340b3d9 3387/* Update i_size, i_disksize. Requires i_rwsem to avoid races with truncate */
4631dbf6
DM
3388static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
3389{
3390 int changed = 0;
3391
3392 if (newsize > inode->i_size) {
3393 i_size_write(inode, newsize);
3394 changed = 1;
3395 }
3396 if (newsize > EXT4_I(inode)->i_disksize) {
3397 ext4_update_i_disksize(inode, newsize);
3398 changed |= 2;
3399 }
3400 return changed;
3401}
3402
01127848
JK
3403int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
3404 loff_t len);
3405
5d1b1b3f
AK
3406struct ext4_group_info {
3407 unsigned long bb_state;
addd752c
CX
3408#ifdef AGGRESSIVE_CHECK
3409 unsigned long bb_check_counter;
3410#endif
5d1b1b3f 3411 struct rb_root bb_free_root;
a36b4498
ES
3412 ext4_grpblk_t bb_first_free; /* first free block */
3413 ext4_grpblk_t bb_free; /* total free blocks */
3414 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
83e80a6e
JK
3415 int bb_avg_fragment_size_order; /* order of average
3416 fragment in BG */
8a57d9d6 3417 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
196e402a 3418 ext4_group_t bb_group; /* Group number */
5d1b1b3f
AK
3419 struct list_head bb_prealloc_list;
3420#ifdef DOUBLE_CHECK
3421 void *bb_bitmap;
3422#endif
3423 struct rw_semaphore alloc_sem;
83e80a6e 3424 struct list_head bb_avg_fragment_size_node;
196e402a 3425 struct list_head bb_largest_free_order_node;
a36b4498
ES
3426 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
3427 * regions, index is order.
3428 * bb_counters[3] = 5 means
3429 * 5 free 8-block regions. */
5d1b1b3f
AK
3430};
3431
3d56b8d2
TM
3432#define EXT4_GROUP_INFO_NEED_INIT_BIT 0
3433#define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
163a203d 3434#define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
87a39389 3435#define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
db79e6d1
WS
3436#define EXT4_GROUP_INFO_BBITMAP_CORRUPT \
3437 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT)
3438#define EXT4_GROUP_INFO_IBITMAP_CORRUPT \
3439 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT)
cfd73237 3440#define EXT4_GROUP_INFO_BBITMAP_READ_BIT 4
5d1b1b3f
AK
3441
3442#define EXT4_MB_GRP_NEED_INIT(grp) \
3443 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
163a203d
DW
3444#define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
3445 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
87a39389
DW
3446#define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
3447 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
5d1b1b3f 3448
3d56b8d2
TM
3449#define EXT4_MB_GRP_WAS_TRIMMED(grp) \
3450 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3451#define EXT4_MB_GRP_SET_TRIMMED(grp) \
3452 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3453#define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
3454 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
cfd73237
AZ
3455#define EXT4_MB_GRP_TEST_AND_SET_READ(grp) \
3456 (test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state)))
3d56b8d2 3457
50797481
TT
3458#define EXT4_MAX_CONTENTION 8
3459#define EXT4_CONTENTION_THRESHOLD 2
3460
955ce5f5
AK
3461static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
3462 ext4_group_t group)
5d1b1b3f 3463{
955ce5f5
AK
3464 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
3465}
5d1b1b3f 3466
50797481
TT
3467/*
3468 * Returns true if the filesystem is busy enough that attempts to
3469 * access the block group locks has run into contention.
3470 */
3471static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
3472{
3473 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
3474}
3475
955ce5f5
AK
3476static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
3477{
50797481
TT
3478 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
3479 if (spin_trylock(lock))
3480 /*
3481 * We're able to grab the lock right away, so drop the
3482 * lock contention counter.
3483 */
3484 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
3485 else {
3486 /*
3487 * The lock is busy, so bump the contention counter,
3488 * and then wait on the spin lock.
3489 */
3490 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
3491 EXT4_MAX_CONTENTION);
3492 spin_lock(lock);
3493 }
5d1b1b3f
AK
3494}
3495
3496static inline void ext4_unlock_group(struct super_block *sb,
3497 ext4_group_t group)
3498{
955ce5f5 3499 spin_unlock(ext4_group_lock_ptr(sb, group));
5d1b1b3f
AK
3500}
3501
f177ee08
RA
3502#ifdef CONFIG_QUOTA
3503static inline bool ext4_quota_capable(struct super_block *sb)
3504{
3505 return (test_opt(sb, QUOTA) || ext4_has_feature_quota(sb));
3506}
3507
3508static inline bool ext4_is_quota_journalled(struct super_block *sb)
3509{
3510 struct ext4_sb_info *sbi = EXT4_SB(sb);
3511
3512 return (ext4_has_feature_quota(sb) ||
3513 sbi->s_qf_names[USRQUOTA] || sbi->s_qf_names[GRPQUOTA]);
3514}
25c6d98f 3515int ext4_enable_quotas(struct super_block *sb);
f177ee08
RA
3516#endif
3517
1f7d1e77
TT
3518/*
3519 * Block validity checking
3520 */
3521#define ext4_check_indirect_blockref(inode, bh) \
3522 ext4_check_blockref(__func__, __LINE__, inode, \
3523 (__le32 *)(bh)->b_data, \
3524 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
3525
3526#define ext4_ind_check_inode(inode) \
3527 ext4_check_blockref(__func__, __LINE__, inode, \
3528 EXT4_I(inode)->i_data, \
3529 EXT4_NDIR_BLOCKS)
3530
ac27a0ec
DK
3531/*
3532 * Inodes and files operations
3533 */
3534
3535/* dir.c */
617ba13b 3536extern const struct file_operations ext4_dir_operations;
ac27a0ec
DK
3537
3538/* file.c */
c5ef1c42 3539extern const struct inode_operations ext4_file_inode_operations;
617ba13b 3540extern const struct file_operations ext4_file_operations;
e0d10bfa 3541extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
ac27a0ec 3542
95eaefbd 3543/* inline.c */
95eaefbd
TT
3544extern int ext4_get_max_inline_size(struct inode *inode);
3545extern int ext4_find_inline_data_nolock(struct inode *inode);
95eaefbd
TT
3546extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
3547
3edde93e 3548int ext4_readpage_inline(struct inode *inode, struct folio *folio);
95eaefbd
TT
3549extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3550 struct inode *inode,
3551 loff_t pos, unsigned len,
95eaefbd 3552 struct page **pagep);
d19500da
RH
3553int ext4_write_inline_data_end(struct inode *inode, loff_t pos, unsigned len,
3554 unsigned copied, struct folio *folio);
95eaefbd
TT
3555extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
3556 struct inode *inode,
3557 loff_t pos, unsigned len,
95eaefbd
TT
3558 struct page **pagep,
3559 void **fsdata);
5b643f9c
TT
3560extern int ext4_try_add_inline_entry(handle_t *handle,
3561 struct ext4_filename *fname,
56a04915 3562 struct inode *dir, struct inode *inode);
95eaefbd
TT
3563extern int ext4_try_create_inline_dir(handle_t *handle,
3564 struct inode *parent,
3565 struct inode *inode);
3566extern int ext4_read_inline_dir(struct file *filp,
725bebb2 3567 struct dir_context *ctx,
95eaefbd 3568 int *has_inline_data);
7633b08b
TT
3569extern int ext4_inlinedir_to_tree(struct file *dir_file,
3570 struct inode *dir, ext4_lblk_t block,
3571 struct dx_hash_info *hinfo,
3572 __u32 start_hash, __u32 start_minor_hash,
3573 int *has_inline_data);
95eaefbd 3574extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
5b643f9c 3575 struct ext4_filename *fname,
95eaefbd
TT
3576 struct ext4_dir_entry_2 **res_dir,
3577 int *has_inline_data);
3578extern int ext4_delete_inline_entry(handle_t *handle,
3579 struct inode *dir,
3580 struct ext4_dir_entry_2 *de_del,
3581 struct buffer_head *bh,
3582 int *has_inline_data);
a7550b30 3583extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
95eaefbd
TT
3584extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3585 struct ext4_dir_entry_2 **parent_de,
3586 int *retval);
5a57bca9 3587extern void *ext4_read_inline_link(struct inode *inode);
7046ae35
AG
3588
3589struct iomap;
3590extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap);
3591
01daf945 3592extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline);
95eaefbd
TT
3593
3594extern int ext4_convert_inline_data(struct inode *inode);
3595
83447ccb
ZL
3596static inline int ext4_has_inline_data(struct inode *inode)
3597{
3598 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3599 EXT4_I(inode)->i_inline_off;
3600}
3601
ac27a0ec 3602/* namei.c */
c5ef1c42
AV
3603extern const struct inode_operations ext4_dir_inode_operations;
3604extern const struct inode_operations ext4_special_inode_operations;
596397b7 3605extern struct dentry *ext4_get_parent(struct dentry *child);
a774f9c2
TM
3606extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3607 struct ext4_dir_entry_2 *de,
3608 int blocksize, int csum_size,
3609 unsigned int parent_ino, int dotdot_real_len);
ddce3b94
TT
3610extern void ext4_initialize_dirent_tail(struct buffer_head *bh,
3611 unsigned int blocksize);
f036adb3
TT
3612extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode,
3613 struct buffer_head *bh);
4c0d5778
EB
3614extern int __ext4_unlink(struct inode *dir, const struct qstr *d_name,
3615 struct inode *inode, struct dentry *dentry);
8016e29f
HS
3616extern int __ext4_link(struct inode *dir, struct inode *inode,
3617 struct dentry *dentry);
b886ee3e 3618
8af0f082 3619#define S_SHIFT 12
381cebfe 3620static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = {
8af0f082
TM
3621 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
3622 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
3623 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
3624 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
3625 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
3626 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
3627 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
3628};
3629
3630static inline void ext4_set_de_type(struct super_block *sb,
3631 struct ext4_dir_entry_2 *de,
3632 umode_t mode) {
e2b911c5 3633 if (ext4_has_feature_filetype(sb))
8af0f082
TM
3634 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3635}
3636
f64e02fe 3637/* readpages.c */
a07f624b 3638extern int ext4_mpage_readpages(struct inode *inode,
c0be8e6f 3639 struct readahead_control *rac, struct folio *folio);
22cfe4b4
EB
3640extern int __init ext4_init_post_read_processing(void);
3641extern void ext4_exit_post_read_processing(void);
ac27a0ec
DK
3642
3643/* symlink.c */
a7a67e8a 3644extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
c5ef1c42
AV
3645extern const struct inode_operations ext4_symlink_inode_operations;
3646extern const struct inode_operations ext4_fast_symlink_inode_operations;
ac27a0ec 3647
b5799018 3648/* sysfs.c */
d578b994 3649extern void ext4_notify_error_sysfs(struct ext4_sb_info *sbi);
b5799018 3650extern int ext4_register_sysfs(struct super_block *sb);
ebd173be 3651extern void ext4_unregister_sysfs(struct super_block *sb);
b5799018
TT
3652extern int __init ext4_init_sysfs(void);
3653extern void ext4_exit_sysfs(void);
3654
6fd058f7
TT
3655/* block_validity */
3656extern void ext4_release_system_zone(struct super_block *sb);
3657extern int ext4_setup_system_zone(struct super_block *sb);
5dabfc78
TT
3658extern int __init ext4_init_system_zone(void);
3659extern void ext4_exit_system_zone(void);
ce9f24cc
JK
3660extern int ext4_inode_block_valid(struct inode *inode,
3661 ext4_fsblk_t start_blk,
3662 unsigned int count);
1f7d1e77
TT
3663extern int ext4_check_blockref(const char *, unsigned int,
3664 struct inode *, __le32 *, unsigned int);
6bc6c2bd
RH
3665extern int ext4_sb_block_valid(struct super_block *sb, struct inode *inode,
3666 ext4_fsblk_t start_blk, unsigned int count);
3667
6fd058f7 3668
a86c6181 3669/* extents.c */
4a092d73
TT
3670struct ext4_ext_path;
3671struct ext4_extent;
3672
d7b2a00c
ZL
3673/*
3674 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3675 * __le32.
3676 */
3677#define EXT_MAX_BLOCKS 0xffffffff
3678
4209ae12 3679extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode);
fffb2739 3680extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
e35fd660
TT
3681extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3682 struct ext4_map_blocks *map, int flags);
d0abb36d 3683extern int ext4_ext_truncate(handle_t *, struct inode *);
26a4c0c6
TT
3684extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3685 ext4_lblk_t end);
a86c6181
AT
3686extern void ext4_ext_init(struct super_block *);
3687extern void ext4_ext_release(struct super_block *);
2fe17c10 3688extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
a2df2a63 3689 loff_t len);
6b523df4
JK
3690extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3691 loff_t offset, ssize_t len);
a00713ea
RH
3692extern int ext4_convert_unwritten_io_end_vec(handle_t *handle,
3693 ext4_io_end_t *io_end);
e35fd660
TT
3694extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3695 struct ext4_map_blocks *map, int flags);
4a092d73
TT
3696extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3697 int num,
3698 struct ext4_ext_path *path);
4a092d73 3699extern int ext4_ext_insert_extent(handle_t *, struct inode *,
dfe50809 3700 struct ext4_ext_path **,
4a092d73 3701 struct ext4_extent *, int);
ed8a1a76
TT
3702extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3703 struct ext4_ext_path **,
3704 int flags);
7ff5fdda 3705extern void ext4_free_ext_path(struct ext4_ext_path *);
4a092d73 3706extern int ext4_ext_check_inode(struct inode *inode);
fcf6b1b7 3707extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3a06d778
AK
3708extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3709 __u64 start, __u64 len);
bb5835ed
TT
3710extern int ext4_get_es_cache(struct inode *inode,
3711 struct fiemap_extent_info *fieinfo,
3712 __u64 start, __u64 len);
7869a4a6 3713extern int ext4_ext_precache(struct inode *inode);
fcf6b1b7
DM
3714extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3715 struct inode *inode2, ext4_lblk_t lblk1,
3716 ext4_lblk_t lblk2, ext4_lblk_t count,
3717 int mark_unwritten,int *err);
0b02f4c0 3718extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu);
a4130367 3719extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode,
83448bdf
JK
3720 int check_cred, int restart_cred,
3721 int revoke_cred);
8016e29f
HS
3722extern void ext4_ext_replay_shrink_inode(struct inode *inode, ext4_lblk_t end);
3723extern int ext4_ext_replay_set_iblocks(struct inode *inode);
3724extern int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start,
3725 int len, int unwritten, ext4_fsblk_t pblk);
3726extern int ext4_ext_clear_bb(struct inode *inode);
a4130367 3727
4a092d73 3728
748de673 3729/* move_extent.c */
393d1d1d
DTB
3730extern void ext4_double_down_write_data_sem(struct inode *first,
3731 struct inode *second);
3732extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3733 struct inode *donor_inode);
748de673
AF
3734extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3735 __u64 start_orig, __u64 start_donor,
3736 __u64 len, __u64 *moved_len);
3737
bd2d0210 3738/* page-io.c */
5dabfc78
TT
3739extern int __init ext4_init_pageio(void);
3740extern void ext4_exit_pageio(void);
bd2d0210 3741extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
97a851ed
JK
3742extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3743extern int ext4_put_io_end(ext4_io_end_t *io_end);
3744extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3745extern void ext4_io_submit_init(struct ext4_io_submit *io,
3746 struct writeback_control *wbc);
2e8fa54e 3747extern void ext4_end_io_rsv_work(struct work_struct *work);
bd2d0210 3748extern void ext4_io_submit(struct ext4_io_submit *io);
e8d6062c
MW
3749int ext4_bio_write_folio(struct ext4_io_submit *io, struct folio *page,
3750 size_t len);
c8cc8816
RH
3751extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end);
3752extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end);
3a06d778 3753
c5e06d10
JL
3754/* mmp.c */
3755extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3756
618f0031
PS
3757/* mmp.c */
3758extern void ext4_stop_mmpd(struct ext4_sb_info *sbi);
3759
c93d8f88
EB
3760/* verity.c */
3761extern const struct fsverity_operations ext4_verityops;
3762
25c6d98f
JK
3763/* orphan.c */
3764extern int ext4_orphan_add(handle_t *, struct inode *);
3765extern int ext4_orphan_del(handle_t *, struct inode *);
3766extern void ext4_orphan_cleanup(struct super_block *sb,
3767 struct ext4_super_block *es);
02f310fc
JK
3768extern void ext4_release_orphan_info(struct super_block *sb);
3769extern int ext4_init_orphan_info(struct super_block *sb);
3770extern int ext4_orphan_file_empty(struct super_block *sb);
3771extern void ext4_orphan_file_block_trigger(
3772 struct jbd2_buffer_trigger_type *triggers,
3773 struct buffer_head *bh,
3774 void *data, size_t size);
25c6d98f 3775
2ccb5fb9 3776/*
48fc7f7e 3777 * Add new method to test whether block and inode bitmaps are properly
2ccb5fb9
AK
3778 * initialized. With uninit_bg reading the block from disk is not enough
3779 * to mark the bitmap uptodate. We need to also zero-out the bitmap
3780 */
3781#define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3782
3783static inline int bitmap_uptodate(struct buffer_head *bh)
3784{
3785 return (buffer_uptodate(bh) &&
3786 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3787}
3788static inline void set_bitmap_uptodate(struct buffer_head *bh)
3789{
3790 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3791}
3792
e9e3bcec
ES
3793/* For ioend & aio unwritten conversion wait queues */
3794#define EXT4_WQ_HASH_SZ 37
3795#define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
3796 EXT4_WQ_HASH_SZ])
e9e3bcec 3797extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
e9e3bcec 3798
8f82f840 3799extern int ext4_resize_begin(struct super_block *sb);
827891a3 3800extern int ext4_resize_end(struct super_block *sb, bool update_backups);
8f82f840 3801
74c66bcb
JK
3802static inline void ext4_set_io_unwritten_flag(struct inode *inode,
3803 struct ext4_io_end *io_end)
3804{
3805 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
3806 io_end->flag |= EXT4_IO_END_UNWRITTEN;
3807 atomic_inc(&EXT4_I(inode)->i_unwritten);
3808 }
3809}
3810
3811static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3812{
3813 struct inode *inode = io_end->inode;
3814
3815 if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
3816 io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3817 /* Wake up anyone waiting on unwritten extent conversion */
3818 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
3819 wake_up_all(ext4_ioend_wq(inode));
3820 }
3821}
3822
8ff6daa1 3823extern const struct iomap_ops ext4_iomap_ops;
8cd115bd 3824extern const struct iomap_ops ext4_iomap_overwrite_ops;
09edf4d3 3825extern const struct iomap_ops ext4_iomap_report_ops;
364443cb 3826
7963e5ac
Z
3827static inline int ext4_buffer_uptodate(struct buffer_head *bh)
3828{
3829 /*
3830 * If the buffer has the write error flag, we have failed
3831 * to write out data in the block. In this case, we don't
3832 * have to read the block because we may read the old data
3833 * successfully.
3834 */
5c680150 3835 if (buffer_write_io_error(bh))
7963e5ac
Z
3836 set_buffer_uptodate(bh);
3837 return buffer_uptodate(bh);
3838}
3839
ac27a0ec
DK
3840#endif /* __KERNEL__ */
3841
6a797d27
DW
3842#define EFSBADCRC EBADMSG /* Bad CRC detected */
3843#define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
3844
3dcf5451 3845#endif /* _EXT4_H */