Merge tag 'devprop-5.19-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael...
[linux-block.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
c9de560d 43
734f0d24 44#include <linux/fscrypt.h>
c93d8f88 45#include <linux/fsverity.h>
734f0d24 46
072ebb3b
TT
47#include <linux/compiler.h>
48
ac27a0ec 49/*
8a35694e 50 * The fourth extended filesystem constants/structures
ac27a0ec
DK
51 */
52
a2821e34
AZ
53/*
54 * with AGGRESSIVE_CHECK allocator runs consistency checks over
55 * structures. these checks slow things down a lot
56 */
57#define AGGRESSIVE_CHECK__
58
59/*
60 * with DOUBLE_CHECK defined mballoc creates persistent in-core
61 * bitmaps, maintains and uses them to check for double allocations
62 */
63#define DOUBLE_CHECK__
64
ac27a0ec 65/*
617ba13b 66 * Define EXT4FS_DEBUG to produce debug messages
ac27a0ec 67 */
617ba13b 68#undef EXT4FS_DEBUG
ac27a0ec 69
ac27a0ec
DK
70/*
71 * Debug code
72 */
617ba13b
MC
73#ifdef EXT4FS_DEBUG
74#define ext4_debug(f, a...) \
ac27a0ec 75 do { \
af5bc92d 76 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
4db9c54a 77 __FILE__, __LINE__, __func__); \
af5bc92d 78 printk(KERN_DEBUG f, ## a); \
ac27a0ec
DK
79 } while (0)
80#else
ace36ad4 81#define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
ac27a0ec
DK
82#endif
83
70aa1554
RH
84 /*
85 * Turn on EXT_DEBUG to enable ext4_ext_show_path/leaf/move in extents.c
86 */
87#define EXT_DEBUG__
88
4a092d73 89/*
70aa1554 90 * Dynamic printk for controlled extents debugging.
4a092d73 91 */
70aa1554
RH
92#ifdef CONFIG_EXT4_DEBUG
93#define ext_debug(ino, fmt, ...) \
94 pr_debug("[%s/%d] EXT4-fs (%s): ino %lu: (%s, %d): %s:" fmt, \
95 current->comm, task_pid_nr(current), \
96 ino->i_sb->s_id, ino->i_ino, __FILE__, __LINE__, \
97 __func__, ##__VA_ARGS__)
4a092d73 98#else
70aa1554 99#define ext_debug(ino, fmt, ...) no_printk(fmt, ##__VA_ARGS__)
4a092d73
TT
100#endif
101
837c23fb
CX
102#define ASSERT(assert) \
103do { \
104 if (unlikely(!(assert))) { \
105 printk(KERN_EMERG \
106 "Assertion failure in %s() at %s:%d: '%s'\n", \
107 __func__, __FILE__, __LINE__, #assert); \
108 BUG(); \
109 } \
110} while (0)
111
d444c3c3
TT
112/* data type for block offset of block group */
113typedef int ext4_grpblk_t;
114
115/* data type for filesystem-wide blocks number */
116typedef unsigned long long ext4_fsblk_t;
117
118/* data type for file logical block number */
119typedef __u32 ext4_lblk_t;
120
121/* data type for block group number */
122typedef unsigned int ext4_group_t;
123
331573fe
NJ
124enum SHIFT_DIRECTION {
125 SHIFT_LEFT = 0,
126 SHIFT_RIGHT,
127};
128
296c355c 129/*
60e6679e 130 * Flags used in mballoc's allocation_context flags field.
296c355c
TT
131 *
132 * Also used to show what's going on for debugging purposes when the
133 * flag field is exported via the traceport interface
134 */
d444c3c3 135
c9de560d 136/* prefer goal again. length */
0ef90db9 137#define EXT4_MB_HINT_MERGE 0x0001
c9de560d 138/* blocks already reserved */
0ef90db9 139#define EXT4_MB_HINT_RESERVED 0x0002
c9de560d 140/* metadata is being allocated */
0ef90db9 141#define EXT4_MB_HINT_METADATA 0x0004
c9de560d 142/* first blocks in the file */
0ef90db9 143#define EXT4_MB_HINT_FIRST 0x0008
c9de560d 144/* search for the best chunk */
0ef90db9 145#define EXT4_MB_HINT_BEST 0x0010
c9de560d 146/* data is being allocated */
0ef90db9 147#define EXT4_MB_HINT_DATA 0x0020
c9de560d 148/* don't preallocate (for tails) */
0ef90db9 149#define EXT4_MB_HINT_NOPREALLOC 0x0040
c9de560d 150/* allocate for locality group */
0ef90db9 151#define EXT4_MB_HINT_GROUP_ALLOC 0x0080
c9de560d 152/* allocate goal blocks or none */
0ef90db9 153#define EXT4_MB_HINT_GOAL_ONLY 0x0100
c9de560d 154/* goal is meaningful */
0ef90db9 155#define EXT4_MB_HINT_TRY_GOAL 0x0200
d2a17637 156/* blocks already pre-reserved by delayed allocation */
0ef90db9 157#define EXT4_MB_DELALLOC_RESERVED 0x0400
4ba74d00
TT
158/* We are doing stream allocation */
159#define EXT4_MB_STREAM_ALLOC 0x0800
55f020db
AH
160/* Use reserved root blocks if needed */
161#define EXT4_MB_USE_ROOT_BLOCKS 0x1000
27dd4385
LC
162/* Use blocks from reserved pool */
163#define EXT4_MB_USE_RESERVED 0x2000
99377830
RH
164/* Do strict check for free blocks while retrying block allocation */
165#define EXT4_MB_STRICT_CHECK 0x4000
196e402a
HS
166/* Large fragment size list lookup succeeded at least once for cr = 0 */
167#define EXT4_MB_CR0_OPTIMIZED 0x8000
168/* Avg fragment size rb tree lookup succeeded at least once for cr = 1 */
169#define EXT4_MB_CR1_OPTIMIZED 0x00010000
170/* Perform linear traversal for one group */
171#define EXT4_MB_SEARCH_NEXT_LINEAR 0x00020000
c9de560d
AT
172struct ext4_allocation_request {
173 /* target inode for block we're allocating */
174 struct inode *inode;
726447d8
ES
175 /* how many blocks we want to allocate */
176 unsigned int len;
c9de560d
AT
177 /* logical block in target inode */
178 ext4_lblk_t logical;
c9de560d
AT
179 /* the closest logical allocated block to the left */
180 ext4_lblk_t lleft;
c9de560d
AT
181 /* the closest logical allocated block to the right */
182 ext4_lblk_t lright;
726447d8
ES
183 /* phys. target (a hint) */
184 ext4_fsblk_t goal;
185 /* phys. block for the closest logical allocated block to the left */
186 ext4_fsblk_t pleft;
187 /* phys. block for the closest logical allocated block to the right */
c9de560d 188 ext4_fsblk_t pright;
c9de560d 189 /* flags. see above EXT4_MB_HINT_* */
498e5f24 190 unsigned int flags;
c9de560d
AT
191};
192
e35fd660
TT
193/*
194 * Logical to physical block mapping, used by ext4_map_blocks()
195 *
196 * This structure is used to pass requests into ext4_map_blocks() as
197 * well as to store the information returned by ext4_map_blocks(). It
198 * takes less room on the stack than a struct buffer_head.
199 */
6db07461
RH
200#define EXT4_MAP_NEW BIT(BH_New)
201#define EXT4_MAP_MAPPED BIT(BH_Mapped)
202#define EXT4_MAP_UNWRITTEN BIT(BH_Unwritten)
203#define EXT4_MAP_BOUNDARY BIT(BH_Boundary)
e35fd660 204#define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
cbd7584e 205 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
e35fd660
TT
206
207struct ext4_map_blocks {
208 ext4_fsblk_t m_pblk;
209 ext4_lblk_t m_lblk;
210 unsigned int m_len;
211 unsigned int m_flags;
212};
213
7727ae52 214/*
215 * Block validity checking, system zone rbtree.
216 */
217struct ext4_system_blocks {
218 struct rb_root root;
219 struct rcu_head rcu;
220};
221
bd2d0210
TT
222/*
223 * Flags for ext4_io_end->flags
224 */
225#define EXT4_IO_END_UNWRITTEN 0x0001
bd2d0210 226
c8cc8816
RH
227struct ext4_io_end_vec {
228 struct list_head list; /* list of io_end_vec */
229 loff_t offset; /* offset in the file */
230 ssize_t size; /* size of the extent */
231};
232
4188188b 233/*
556615dc 234 * For converting unwritten extents on a work queue. 'handle' is used for
6b523df4 235 * buffered writeback.
4188188b 236 */
0031462b 237typedef struct ext4_io_end {
5b3ff237 238 struct list_head list; /* per-file finished IO list */
6b523df4
JK
239 handle_t *handle; /* handle reserved for extent
240 * conversion */
0031462b 241 struct inode *inode; /* file being written to */
b0857d30
JK
242 struct bio *bio; /* Linked list of completed
243 * bios covering the extent */
8d5d02e6 244 unsigned int flag; /* unwritten or not */
31d21d21 245 refcount_t count; /* reference counter */
c8cc8816 246 struct list_head list_vec; /* list of ext4_io_end_vec */
0031462b
MC
247} ext4_io_end_t;
248
bd2d0210 249struct ext4_io_submit {
5a33911f 250 struct writeback_control *io_wbc;
bd2d0210
TT
251 struct bio *io_bio;
252 ext4_io_end_t *io_end;
bd2d0210
TT
253 sector_t io_next_block;
254};
255
ac27a0ec
DK
256/*
257 * Special inodes numbers
258 */
617ba13b
MC
259#define EXT4_BAD_INO 1 /* Bad blocks inode */
260#define EXT4_ROOT_INO 2 /* Root inode */
ae812306
AK
261#define EXT4_USR_QUOTA_INO 3 /* User quota inode */
262#define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
617ba13b
MC
263#define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
264#define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
265#define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
266#define EXT4_JOURNAL_INO 8 /* Journal inode */
ac27a0ec 267
617ba13b
MC
268/* First non-reserved inode for old ext4 filesystems */
269#define EXT4_GOOD_OLD_FIRST_INO 11
ac27a0ec
DK
270
271/*
272 * Maximal count of links to a file
273 */
f8628a14 274#define EXT4_LINK_MAX 65000
ac27a0ec
DK
275
276/*
277 * Macro-instructions used to manage several block sizes
278 */
617ba13b 279#define EXT4_MIN_BLOCK_SIZE 1024
afc7cbca
TS
280#define EXT4_MAX_BLOCK_SIZE 65536
281#define EXT4_MIN_BLOCK_LOG_SIZE 10
fb1813f4 282#define EXT4_MAX_BLOCK_LOG_SIZE 16
9e47a4c9 283#define EXT4_MAX_CLUSTER_LOG_SIZE 30
ac27a0ec 284#ifdef __KERNEL__
617ba13b 285# define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
ac27a0ec 286#else
617ba13b 287# define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
ac27a0ec 288#endif
af5bc92d 289#define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
281b5995
TT
290#define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
291 EXT4_SB(s)->s_cluster_bits)
ac27a0ec 292#ifdef __KERNEL__
617ba13b 293# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
281b5995 294# define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
ac27a0ec 295#else
617ba13b 296# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
ac27a0ec
DK
297#endif
298#ifdef __KERNEL__
617ba13b
MC
299#define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
300#define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
301#define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
ac27a0ec 302#else
617ba13b
MC
303#define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
304 EXT4_GOOD_OLD_INODE_SIZE : \
ac27a0ec 305 (s)->s_inode_size)
617ba13b
MC
306#define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
307 EXT4_GOOD_OLD_FIRST_INO : \
ac27a0ec
DK
308 (s)->s_first_ino)
309#endif
a2df2a63 310#define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
518eaa63
FF
311#define EXT4_MAX_BLOCKS(size, offset, blkbits) \
312 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \
313 blkbits))
ac27a0ec 314
d5b8f310
TT
315/* Translate a block number to a cluster number */
316#define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
317/* Translate a cluster number to a block number */
318#define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
319/* Translate # of blks to # of clusters */
320#define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
321 (sbi)->s_cluster_bits)
f5a44db5
TT
322/* Mask out the low bits to get the starting block of the cluster */
323#define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
324 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
325#define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
326 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
8fcc3a58
EW
327/* Fill in the low bits to get the last block of the cluster */
328#define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) | \
329 ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1))
f5a44db5
TT
330/* Get the cluster offset */
331#define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
332 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
333#define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
334 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
d5b8f310 335
ac27a0ec
DK
336/*
337 * Structure of a blocks group descriptor
338 */
617ba13b 339struct ext4_group_desc
ac27a0ec 340{
3a14589c 341 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
5272f837
AK
342 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
343 __le32 bg_inode_table_lo; /* Inodes table block */
560671a0
AK
344 __le16 bg_free_blocks_count_lo;/* Free blocks count */
345 __le16 bg_free_inodes_count_lo;/* Free inodes count */
346 __le16 bg_used_dirs_count_lo; /* Directories count */
717d50e4 347 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
e6153918
DW
348 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
349 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
350 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
560671a0 351 __le16 bg_itable_unused_lo; /* Unused inodes count */
717d50e4 352 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
8fadc143
AR
353 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
354 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
355 __le32 bg_inode_table_hi; /* Inodes table block MSB */
91b51a01
CL
356 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
357 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
358 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
560671a0 359 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
e6153918
DW
360 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
361 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
362 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
363 __u32 bg_reserved;
ac27a0ec
DK
364};
365
41a246d1
DW
366#define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
367 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
368 sizeof(__le16))
369#define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
370 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
371 sizeof(__le16))
372
772cb7c8
JS
373/*
374 * Structure of a flex block group info
375 */
376
377struct flex_groups {
90ba983f
TT
378 atomic64_t free_clusters;
379 atomic_t free_inodes;
380 atomic_t used_dirs;
772cb7c8
JS
381};
382
717d50e4
AD
383#define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
384#define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
385#define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
386
ac27a0ec
DK
387/*
388 * Macro-instructions used to manage group descriptors
389 */
0d1ee42f 390#define EXT4_MIN_DESC_SIZE 32
8fadc143 391#define EXT4_MIN_DESC_SIZE_64BIT 64
0d1ee42f
AR
392#define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
393#define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
ac27a0ec 394#ifdef __KERNEL__
617ba13b 395# define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
281b5995 396# define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
617ba13b
MC
397# define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
398# define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
399# define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
ac27a0ec 400#else
617ba13b 401# define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
0d1ee42f 402# define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
617ba13b 403# define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
ac27a0ec
DK
404#endif
405
406/*
407 * Constants relative to the data blocks
408 */
617ba13b
MC
409#define EXT4_NDIR_BLOCKS 12
410#define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
411#define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
412#define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
413#define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
ac27a0ec
DK
414
415/*
416 * Inode flags
417 */
617ba13b
MC
418#define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
419#define EXT4_UNRM_FL 0x00000002 /* Undelete */
420#define EXT4_COMPR_FL 0x00000004 /* Compress file */
421#define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
422#define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
423#define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
424#define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
425#define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
ac27a0ec 426/* Reserved for compression usage... */
617ba13b
MC
427#define EXT4_DIRTY_FL 0x00000100
428#define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
429#define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
3edc18d8
TT
430 /* nb: was previously EXT2_ECOMPR_FL */
431#define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */
ac27a0ec 432/* End compression flags --- maybe not all used */
617ba13b
MC
433#define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
434#define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
435#define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
436#define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
437#define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
438#define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
8180a562 439#define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
a86c6181 440#define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
c93d8f88 441#define EXT4_VERITY_FL 0x00100000 /* Verity protected inode */
f710b4b9 442#define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
4337ecd1 443/* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */
b383a73f
IW
444
445#define EXT4_DAX_FL 0x02000000 /* Inode is DAX */
446
67cf5b09 447#define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
8b4953e1 448#define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
de8ff14c 449#define EXT4_CASEFOLD_FL 0x40000000 /* Casefolded directory */
8180a562 450#define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
ac27a0ec 451
2a12e147
EB
452/* User modifiable flags */
453#define EXT4_FL_USER_MODIFIABLE (EXT4_SECRM_FL | \
454 EXT4_UNRM_FL | \
455 EXT4_COMPR_FL | \
456 EXT4_SYNC_FL | \
457 EXT4_IMMUTABLE_FL | \
458 EXT4_APPEND_FL | \
459 EXT4_NODUMP_FL | \
460 EXT4_NOATIME_FL | \
461 EXT4_JOURNAL_DATA_FL | \
462 EXT4_NOTAIL_FL | \
463 EXT4_DIRSYNC_FL | \
464 EXT4_TOPDIR_FL | \
465 EXT4_EXTENTS_FL | \
466 0x00400000 /* EXT4_EOFBLOCKS_FL */ | \
467 EXT4_DAX_FL | \
468 EXT4_PROJINHERIT_FL | \
469 EXT4_CASEFOLD_FL)
470
471/* User visible flags */
472#define EXT4_FL_USER_VISIBLE (EXT4_FL_USER_MODIFIABLE | \
473 EXT4_DIRTY_FL | \
474 EXT4_COMPRBLK_FL | \
475 EXT4_NOCOMPR_FL | \
476 EXT4_ENCRYPT_FL | \
477 EXT4_INDEX_FL | \
478 EXT4_VERITY_FL | \
479 EXT4_INLINE_DATA_FL)
ac27a0ec 480
8fa43a81
DG
481/* Flags that should be inherited by new inodes from their parent. */
482#define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
1cd9f097 483 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
8fa43a81 484 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
040cb378 485 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
b383a73f
IW
486 EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL |\
487 EXT4_DAX_FL)
8fa43a81 488
2dc6b0d4 489/* Flags that are appropriate for regular files (all but dir-specific ones). */
7ddf79a1
WS
490#define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\
491 EXT4_PROJINHERIT_FL))
2dc6b0d4
DG
492
493/* Flags that are appropriate for non-directories/regular files. */
494#define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
495
abdc644e 496/* The only flags that should be swapped */
497#define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL)
498
b383a73f
IW
499/* Flags which are mutually exclusive to DAX */
500#define EXT4_DAX_MUT_EXCL (EXT4_VERITY_FL | EXT4_ENCRYPT_FL |\
aa2f7792 501 EXT4_JOURNAL_DATA_FL | EXT4_INLINE_DATA_FL)
b383a73f 502
2dc6b0d4
DG
503/* Mask out flags that are inappropriate for the given type of inode. */
504static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
505{
506 if (S_ISDIR(mode))
507 return flags;
508 else if (S_ISREG(mode))
509 return flags & EXT4_REG_FLMASK;
510 else
511 return flags & EXT4_OTHER_FLMASK;
512}
513
12e9b892
DM
514/*
515 * Inode flags used for atomic set/get
516 */
517enum {
518 EXT4_INODE_SECRM = 0, /* Secure deletion */
519 EXT4_INODE_UNRM = 1, /* Undelete */
520 EXT4_INODE_COMPR = 2, /* Compress file */
521 EXT4_INODE_SYNC = 3, /* Synchronous updates */
522 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
523 EXT4_INODE_APPEND = 5, /* writes to file may only append */
524 EXT4_INODE_NODUMP = 6, /* do not dump file */
525 EXT4_INODE_NOATIME = 7, /* do not update atime */
526/* Reserved for compression usage... */
527 EXT4_INODE_DIRTY = 8,
528 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
529 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
f542fbe8 530 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */
12e9b892
DM
531/* End compression flags --- maybe not all used */
532 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
533 EXT4_INODE_IMAGIC = 13, /* AFS directory */
534 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
535 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
536 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
537 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
538 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
539 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
c93d8f88 540 EXT4_INODE_VERITY = 20, /* Verity protected inode */
12e9b892 541 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
4337ecd1 542/* 22 was formerly EXT4_INODE_EOFBLOCKS */
b383a73f 543 EXT4_INODE_DAX = 25, /* Inode is DAX */
67cf5b09 544 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
8b4953e1 545 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */
de8ff14c 546 EXT4_INODE_CASEFOLD = 30, /* Casefolded directory */
12e9b892
DM
547 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
548};
549
12e9b892 550/*
9a4c8019
CM
551 * Since it's pretty easy to mix up bit numbers and hex values, we use a
552 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
553 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
554 * any extra space in the compiled kernel image, otherwise, the build will fail.
555 * It's important that these values are the same, since we are using
556 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
557 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
558 * values found in ext2, ext3 and ext4 filesystems, and of course the values
559 * defined in e2fsprogs.
12e9b892
DM
560 *
561 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
562 */
9a4c8019
CM
563#define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
564#define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
565
12e9b892
DM
566static inline void ext4_check_flag_values(void)
567{
568 CHECK_FLAG_VALUE(SECRM);
569 CHECK_FLAG_VALUE(UNRM);
570 CHECK_FLAG_VALUE(COMPR);
571 CHECK_FLAG_VALUE(SYNC);
572 CHECK_FLAG_VALUE(IMMUTABLE);
573 CHECK_FLAG_VALUE(APPEND);
574 CHECK_FLAG_VALUE(NODUMP);
575 CHECK_FLAG_VALUE(NOATIME);
576 CHECK_FLAG_VALUE(DIRTY);
577 CHECK_FLAG_VALUE(COMPRBLK);
578 CHECK_FLAG_VALUE(NOCOMPR);
3edc18d8 579 CHECK_FLAG_VALUE(ENCRYPT);
12e9b892
DM
580 CHECK_FLAG_VALUE(INDEX);
581 CHECK_FLAG_VALUE(IMAGIC);
582 CHECK_FLAG_VALUE(JOURNAL_DATA);
583 CHECK_FLAG_VALUE(NOTAIL);
584 CHECK_FLAG_VALUE(DIRSYNC);
585 CHECK_FLAG_VALUE(TOPDIR);
586 CHECK_FLAG_VALUE(HUGE_FILE);
587 CHECK_FLAG_VALUE(EXTENTS);
c93d8f88 588 CHECK_FLAG_VALUE(VERITY);
12e9b892 589 CHECK_FLAG_VALUE(EA_INODE);
67cf5b09 590 CHECK_FLAG_VALUE(INLINE_DATA);
8b4953e1 591 CHECK_FLAG_VALUE(PROJINHERIT);
de8ff14c 592 CHECK_FLAG_VALUE(CASEFOLD);
12e9b892
DM
593 CHECK_FLAG_VALUE(RESERVED);
594}
595
ac27a0ec 596/* Used to pass group descriptor data when online resize is done */
617ba13b 597struct ext4_new_group_input {
8c55e204
DK
598 __u32 group; /* Group number for this data */
599 __u64 block_bitmap; /* Absolute block number of block bitmap */
600 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
601 __u64 inode_table; /* Absolute block number of inode table start */
602 __u32 blocks_count; /* Total number of blocks in this group */
603 __u16 reserved_blocks; /* Number of reserved blocks in this group */
ac27a0ec
DK
604 __u16 unused;
605};
606
4d92dc0f
BH
607#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
608struct compat_ext4_new_group_input {
609 u32 group;
610 compat_u64 block_bitmap;
611 compat_u64 inode_bitmap;
612 compat_u64 inode_table;
613 u32 blocks_count;
614 u16 reserved_blocks;
615 u16 unused;
616};
617#endif
618
617ba13b
MC
619/* The struct ext4_new_group_input in kernel space, with free_blocks_count */
620struct ext4_new_group_data {
ac27a0ec 621 __u32 group;
bd81d8ee
LV
622 __u64 block_bitmap;
623 __u64 inode_bitmap;
624 __u64 inode_table;
ac27a0ec
DK
625 __u32 blocks_count;
626 __u16 reserved_blocks;
d77147ff 627 __u16 mdata_blocks;
628 __u32 free_clusters_count;
ac27a0ec
DK
629};
630
33afdcc5
YY
631/* Indexes used to index group tables in ext4_new_group_data */
632enum {
633 BLOCK_BITMAP = 0, /* block bitmap */
634 INODE_BITMAP, /* inode bitmap */
635 INODE_TABLE, /* inode tables */
636 GROUP_TABLE_COUNT,
637};
638
a2df2a63 639/*
79e83036 640 * Flags used by ext4_map_blocks()
a2df2a63 641 */
556615dc 642 /* Allocate any needed blocks and/or convert an unwritten
c2177057 643 extent to be an initialized ext4 */
2ac3b6e0 644#define EXT4_GET_BLOCKS_CREATE 0x0001
556615dc
LC
645 /* Request the creation of an unwritten extent */
646#define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002
647#define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\
c2177057 648 EXT4_GET_BLOCKS_CREATE)
27dd4385
LC
649 /* Caller is from the delayed allocation writeout path
650 * finally doing the actual allocation of delayed blocks */
03f5d8bc 651#define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
0031462b 652 /* caller is from the direct IO path, request to creation of an
556615dc 653 unwritten extents if not allocated, split the unwritten
0031462b 654 extent if blocks has been preallocated already*/
c7064ef1 655#define EXT4_GET_BLOCKS_PRE_IO 0x0008
1296cc85 656#define EXT4_GET_BLOCKS_CONVERT 0x0010
c7064ef1 657#define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
556615dc 658 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
c7064ef1
JZ
659 /* Convert extent to initialized after IO complete */
660#define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
556615dc 661 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
27dd4385
LC
662 /* Eventual metadata allocation (due to growing extent tree)
663 * should not fail, so try to use reserved blocks for that.*/
664#define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
556b27ab
VH
665 /* Don't normalize allocation size (used for fallocate) */
666#define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
b8a86845 667 /* Convert written extents to unwritten */
2dcba478 668#define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100
c86d8db3
JK
669 /* Write zeros to newly created written extents */
670#define EXT4_GET_BLOCKS_ZERO 0x0200
671#define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\
672 EXT4_GET_BLOCKS_ZERO)
ee0876bc
JK
673 /* Caller will submit data before dropping transaction handle. This
674 * allows jbd2 to avoid submitting data before commit. */
675#define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400
ac27a0ec 676
107a7bd3 677/*
7869a4a6 678 * The bit position of these flags must not overlap with any of the
ed8a1a76 679 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(),
107a7bd3 680 * read_extent_tree_block(), ext4_split_extent_at(),
7869a4a6
TT
681 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
682 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
683 * caching the extents when reading from the extent tree while a
684 * truncate or punch hole operation is in progress.
107a7bd3 685 */
d26e2c4d
TT
686#define EXT4_EX_NOCACHE 0x40000000
687#define EXT4_EX_FORCE_CACHE 0x20000000
73c384c0 688#define EXT4_EX_NOFAIL 0x10000000
107a7bd3 689
e6362609
TT
690/*
691 * Flags used by ext4_free_blocks
692 */
e7093f0d
PK
693#define EXT4_FREE_BLOCKS_METADATA 0x0001
694#define EXT4_FREE_BLOCKS_FORGET 0x0002
695#define EXT4_FREE_BLOCKS_VALIDATED 0x0004
696#define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
84130193
TT
697#define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
698#define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
9fe67149 699#define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040
e6362609 700
ac27a0ec
DK
701/*
702 * ioctl commands
703 */
617ba13b
MC
704#define EXT4_IOC_GETVERSION _IOR('f', 3, long)
705#define EXT4_IOC_SETVERSION _IOW('f', 4, long)
617ba13b
MC
706#define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
707#define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
617ba13b
MC
708#define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
709#define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
8eea80d5
TT
710#define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
711#define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
712#define EXT4_IOC_MIGRATE _IO('f', 9)
ccd2506b 713 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
8eea80d5 714 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
ccd2506b 715#define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
748de673 716#define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
19c5246d 717#define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
393d1d1d 718#define EXT4_IOC_SWAP_BOOT _IO('f', 17)
7869a4a6 719#define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18)
b0c013e2
TT
720/* ioctl codes 19--39 are reserved for fscrypt */
721#define EXT4_IOC_CLEAR_ES_CACHE _IO('f', 40)
1ad3ea6e 722#define EXT4_IOC_GETSTATE _IOW('f', 41, __u32)
bb5835ed 723#define EXT4_IOC_GET_ES_CACHE _IOWR('f', 42, struct fiemap)
351a0a3f 724#define EXT4_IOC_CHECKPOINT _IOW('f', 43, __u32)
ac27a0ec 725
e9be2ac7 726#define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32)
783d9485
TT
727
728/*
729 * Flags for going down operation
730 */
731#define EXT4_GOING_FLAGS_DEFAULT 0x0 /* going down */
732#define EXT4_GOING_FLAGS_LOGFLUSH 0x1 /* flush log but not data */
733#define EXT4_GOING_FLAGS_NOLOGFLUSH 0x2 /* don't flush log nor data */
734
1ad3ea6e
TT
735/*
736 * Flags returned by EXT4_IOC_GETSTATE
737 *
738 * We only expose to userspace a subset of the state flags in
739 * i_state_flags
740 */
741#define EXT4_STATE_FLAG_EXT_PRECACHED 0x00000001
742#define EXT4_STATE_FLAG_NEW 0x00000002
743#define EXT4_STATE_FLAG_NEWENTRY 0x00000004
744#define EXT4_STATE_FLAG_DA_ALLOC_CLOSE 0x00000008
783d9485 745
351a0a3f
LR
746/* flags for ioctl EXT4_IOC_CHECKPOINT */
747#define EXT4_IOC_CHECKPOINT_FLAG_DISCARD 0x1
748#define EXT4_IOC_CHECKPOINT_FLAG_ZEROOUT 0x2
749#define EXT4_IOC_CHECKPOINT_FLAG_DRY_RUN 0x4
750#define EXT4_IOC_CHECKPOINT_FLAG_VALID (EXT4_IOC_CHECKPOINT_FLAG_DISCARD | \
751 EXT4_IOC_CHECKPOINT_FLAG_ZEROOUT | \
752 EXT4_IOC_CHECKPOINT_FLAG_DRY_RUN)
753
899ad0ce 754#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
ac27a0ec
DK
755/*
756 * ioctl commands in 32 bit emulation
757 */
617ba13b
MC
758#define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
759#define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
760#define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
761#define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
762#define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
4d92dc0f 763#define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
617ba13b
MC
764#define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
765#define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
899ad0ce 766#endif
ac27a0ec 767
bb5835ed
TT
768/*
769 * Returned by EXT4_IOC_GET_ES_CACHE as an additional possible flag.
770 * It indicates that the entry in extent status cache is for a hole.
771 */
772#define EXT4_FIEMAP_EXTENT_HOLE 0x08000000
773
b595076a 774/* Max physical block we can address w/o extents */
fb0a387d
ES
775#define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
776
bcd8e91f 777/* Max logical block we can support */
9f44eda1 778#define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFE
bcd8e91f 779
ac27a0ec
DK
780/*
781 * Structure of an inode on the disk
782 */
617ba13b 783struct ext4_inode {
ac27a0ec
DK
784 __le16 i_mode; /* File mode */
785 __le16 i_uid; /* Low 16 bits of Owner Uid */
a48380f7 786 __le32 i_size_lo; /* Size in bytes */
ac27a0ec 787 __le32 i_atime; /* Access time */
ef7f3835 788 __le32 i_ctime; /* Inode Change time */
ac27a0ec
DK
789 __le32 i_mtime; /* Modification time */
790 __le32 i_dtime; /* Deletion Time */
791 __le16 i_gid; /* Low 16 bits of Group Id */
792 __le16 i_links_count; /* Links count */
0fc1b451 793 __le32 i_blocks_lo; /* Blocks count */
ac27a0ec
DK
794 __le32 i_flags; /* File flags */
795 union {
796 struct {
25ec56b5 797 __le32 l_i_version;
ac27a0ec
DK
798 } linux1;
799 struct {
800 __u32 h_i_translator;
801 } hurd1;
802 struct {
803 __u32 m_i_reserved1;
804 } masix1;
805 } osd1; /* OS dependent 1 */
617ba13b 806 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
ac27a0ec 807 __le32 i_generation; /* File version (for NFS) */
7973c0c1 808 __le32 i_file_acl_lo; /* File ACL */
a48380f7 809 __le32 i_size_high;
f077d0d7 810 __le32 i_obso_faddr; /* Obsoleted fragment address */
ac27a0ec
DK
811 union {
812 struct {
0fc1b451 813 __le16 l_i_blocks_high; /* were l_i_reserved1 */
a1ddeb7e 814 __le16 l_i_file_acl_high;
8c55e204 815 __le16 l_i_uid_high; /* these 2 fields */
ac27a0ec 816 __le16 l_i_gid_high; /* were reserved2[0] */
e6153918
DW
817 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
818 __le16 l_i_reserved;
ac27a0ec
DK
819 } linux2;
820 struct {
f077d0d7 821 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
ac27a0ec
DK
822 __u16 h_i_mode_high;
823 __u16 h_i_uid_high;
824 __u16 h_i_gid_high;
825 __u32 h_i_author;
826 } hurd2;
827 struct {
f077d0d7 828 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
a1ddeb7e 829 __le16 m_i_file_acl_high;
ac27a0ec
DK
830 __u32 m_i_reserved2[2];
831 } masix2;
832 } osd2; /* OS dependent 2 */
833 __le16 i_extra_isize;
e6153918 834 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
ef7f3835
KS
835 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
836 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
837 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
838 __le32 i_crtime; /* File Creation time */
839 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
25ec56b5 840 __le32 i_version_hi; /* high 32 bits for 64-bit version */
8b4953e1 841 __le32 i_projid; /* Project ID */
ac27a0ec
DK
842};
843
748de673
AF
844struct move_extent {
845 __u32 reserved; /* should be zero */
846 __u32 donor_fd; /* donor file descriptor */
847 __u64 orig_start; /* logical start offset in block for orig */
848 __u64 donor_start; /* logical start offset in block for donor */
849 __u64 len; /* block length to be moved */
850 __u64 moved_len; /* moved block length */
851};
ac27a0ec 852
ef7f3835
KS
853#define EXT4_EPOCH_BITS 2
854#define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
855#define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
856
857/*
858 * Extended fields will fit into an inode if the filesystem was formatted
859 * with large inodes (-I 256 or larger) and there are not currently any EAs
860 * consuming all of the available space. For new inodes we always reserve
861 * enough space for the kernel's known extended fields, but for inodes
862 * created with an old kernel this might not have been the case. None of
863 * the extended inode fields is critical for correct filesystem operation.
864 * This macro checks if a certain field fits in the inode. Note that
865 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
866 */
867#define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
868 ((offsetof(typeof(*ext4_inode), field) + \
869 sizeof((ext4_inode)->field)) \
870 <= (EXT4_GOOD_OLD_INODE_SIZE + \
871 (einode)->i_extra_isize)) \
872
a4dad1ae
DT
873/*
874 * We use an encoding that preserves the times for extra epoch "00":
875 *
876 * extra msb of adjust for signed
877 * epoch 32-bit 32-bit tv_sec to
878 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
879 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
880 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
881 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
882 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
883 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
884 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
885 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
886 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
887 *
888 * Note that previous versions of the kernel on 64-bit systems would
889 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
890 * 1970. e2fsck will correct this, assuming that it is run on the
891 * affected filesystem before 2242.
892 */
893
7b62b293 894static inline __le32 ext4_encode_extra_time(struct timespec64 *time)
ef7f3835 895{
7b62b293 896 u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK;
a4dad1ae 897 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
ef7f3835
KS
898}
899
7b62b293
AB
900static inline void ext4_decode_extra_time(struct timespec64 *time,
901 __le32 extra)
ef7f3835 902{
cd2d9922 903 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK)))
a4dad1ae 904 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
a4dad1ae 905 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
ef7f3835
KS
906}
907
95582b00
DD
908#define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
909do { \
95582b00 910 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {\
4881c497 911 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
95582b00 912 (raw_inode)->xtime ## _extra = \
7b62b293 913 ext4_encode_extra_time(&(inode)->xtime); \
95582b00 914 } \
cba465b4 915 else \
4881c497 916 (raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (inode)->xtime.tv_sec, S32_MIN, S32_MAX)); \
ef7f3835
KS
917} while (0)
918
919#define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
920do { \
921 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
922 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
923 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
924 (raw_inode)->xtime ## _extra = \
925 ext4_encode_extra_time(&(einode)->xtime); \
926} while (0)
927
95582b00
DD
928#define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
929do { \
930 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
931 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \
7b62b293 932 ext4_decode_extra_time(&(inode)->xtime, \
95582b00 933 raw_inode->xtime ## _extra); \
95582b00
DD
934 } \
935 else \
936 (inode)->xtime.tv_nsec = 0; \
ef7f3835
KS
937} while (0)
938
95582b00 939
ef7f3835
KS
940#define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
941do { \
942 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
943 (einode)->xtime.tv_sec = \
944 (signed)le32_to_cpu((raw_inode)->xtime); \
19ea8060
TT
945 else \
946 (einode)->xtime.tv_sec = 0; \
ef7f3835
KS
947 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
948 ext4_decode_extra_time(&(einode)->xtime, \
949 raw_inode->xtime ## _extra); \
af0b44a1
ES
950 else \
951 (einode)->xtime.tv_nsec = 0; \
ef7f3835
KS
952} while (0)
953
25ec56b5
JNC
954#define i_disk_version osd1.linux1.l_i_version
955
ac27a0ec
DK
956#if defined(__KERNEL__) || defined(__linux__)
957#define i_reserved1 osd1.linux1.l_i_reserved1
a1ddeb7e 958#define i_file_acl_high osd2.linux2.l_i_file_acl_high
0fc1b451 959#define i_blocks_high osd2.linux2.l_i_blocks_high
ac27a0ec
DK
960#define i_uid_low i_uid
961#define i_gid_low i_gid
962#define i_uid_high osd2.linux2.l_i_uid_high
963#define i_gid_high osd2.linux2.l_i_gid_high
e6153918 964#define i_checksum_lo osd2.linux2.l_i_checksum_lo
ac27a0ec
DK
965
966#elif defined(__GNU__)
967
968#define i_translator osd1.hurd1.h_i_translator
ac27a0ec
DK
969#define i_uid_high osd2.hurd2.h_i_uid_high
970#define i_gid_high osd2.hurd2.h_i_gid_high
971#define i_author osd2.hurd2.h_i_author
972
973#elif defined(__masix__)
974
975#define i_reserved1 osd1.masix1.m_i_reserved1
a1ddeb7e 976#define i_file_acl_high osd2.masix2.m_i_file_acl_high
ac27a0ec
DK
977#define i_reserved2 osd2.masix2.m_i_reserved2
978
979#endif /* defined(__KERNEL__) || defined(__linux__) */
980
c0677e6d 981#include "extents_status.h"
6866d7b3 982#include "fast_commit.h"
c0677e6d 983
daf647d2
TT
984/*
985 * Lock subclasses for i_data_sem in the ext4_inode_info structure.
986 *
987 * These are needed to avoid lockdep false positives when we need to
988 * allocate blocks to the quota inode during ext4_map_blocks(), while
989 * holding i_data_sem for a normal (non-quota) inode. Since we don't
990 * do quota tracking for the quota inode, this avoids deadlock (as
991 * well as infinite recursion, since it isn't turtles all the way
992 * down...)
993 *
994 * I_DATA_SEM_NORMAL - Used for most inodes
995 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode
996 * where the second inode has larger inode number
997 * than the first
998 * I_DATA_SEM_QUOTA - Used for quota inodes only
999 */
1000enum {
1001 I_DATA_SEM_NORMAL = 0,
1002 I_DATA_SEM_OTHER,
1003 I_DATA_SEM_QUOTA,
1004};
1005
1006
d444c3c3
TT
1007/*
1008 * fourth extended file system inode data in memory
1009 */
1010struct ext4_inode_info {
1011 __le32 i_data[15]; /* unconverted */
d444c3c3 1012 __u32 i_dtime;
12e9b892 1013 ext4_fsblk_t i_file_acl;
d444c3c3
TT
1014
1015 /*
1016 * i_block_group is the number of the block group which contains
1017 * this file's inode. Constant across the lifetime of the inode,
a627b0a7 1018 * it is used for making block allocation decisions - we try to
d444c3c3
TT
1019 * place a file's data blocks near its inode block, and new inodes
1020 * near to their parent directory's inode.
1021 */
1022 ext4_group_t i_block_group;
8a2005d3 1023 ext4_lblk_t i_dir_start_lookup;
353eb83c 1024#if (BITS_PER_LONG < 64)
19f5fb7a 1025 unsigned long i_state_flags; /* Dynamic state flags */
353eb83c 1026#endif
12e9b892 1027 unsigned long i_flags;
d444c3c3 1028
d444c3c3
TT
1029 /*
1030 * Extended attributes can be read independently of the main file
f340b3d9 1031 * data. Taking i_rwsem even when reading would cause contention
d444c3c3
TT
1032 * between readers of EAs and writers of regular file data, so
1033 * instead we synchronize on xattr_sem when reading or changing
1034 * EAs.
1035 */
1036 struct rw_semaphore xattr_sem;
d444c3c3 1037
02f310fc
JK
1038 /*
1039 * Inodes with EXT4_STATE_ORPHAN_FILE use i_orphan_idx. Otherwise
1040 * i_orphan is used.
1041 */
1042 union {
1043 struct list_head i_orphan; /* unlinked but open inodes */
1044 unsigned int i_orphan_idx; /* Index in orphan file */
1045 };
d444c3c3 1046
aa75f4d3
HS
1047 /* Fast commit related info */
1048
b3998b3b
RH
1049 /* For tracking dentry create updates */
1050 struct list_head i_fc_dilist;
aa75f4d3
HS
1051 struct list_head i_fc_list; /*
1052 * inodes that need fast commit
1053 * protected by sbi->s_fc_lock.
1054 */
1055
aa75f4d3
HS
1056 /* Start of lblk range that needs to be committed in this fast commit */
1057 ext4_lblk_t i_fc_lblk_start;
1058
1059 /* End of lblk range that needs to be committed in this fast commit */
1060 ext4_lblk_t i_fc_lblk_len;
1061
1062 /* Number of ongoing updates on this inode */
1063 atomic_t i_fc_updates;
1064
1065 /* Fast commit wait queue for this inode */
1066 wait_queue_head_t i_fc_wait;
1067
1068 /* Protect concurrent accesses on i_fc_lblk_start, i_fc_lblk_len */
1069 struct mutex i_fc_lock;
1070
d444c3c3
TT
1071 /*
1072 * i_disksize keeps track of what the inode size is ON DISK, not
1073 * in memory. During truncate, i_size is set to the new size by
1074 * the VFS prior to calling ext4_truncate(), but the filesystem won't
1075 * set i_disksize to 0 until the truncate is actually under way.
1076 *
1077 * The intent is that i_disksize always represents the blocks which
1078 * are used by this file. This allows recovery to restart truncate
1079 * on orphans if we crash during truncate. We actually write i_disksize
1080 * into the on-disk inode when writing inodes out, instead of i_size.
1081 *
1082 * The only time when i_disksize and i_size may be different is when
1083 * a truncate is in progress. The only things which change i_disksize
1084 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
1085 */
1086 loff_t i_disksize;
1087
1088 /*
1089 * i_data_sem is for serialising ext4_truncate() against
1090 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
1091 * data tree are chopped off during truncate. We can't do that in
1092 * ext4 because whenever we perform intermediate commits during
1093 * truncate, the inode and all the metadata blocks *must* be in a
1094 * consistent state which allows truncation of the orphans to restart
1095 * during recovery. Hence we must fix the get_block-vs-truncate race
1096 * by other means, so we have i_data_sem.
1097 */
1098 struct rw_semaphore i_data_sem;
1099 struct inode vfs_inode;
8aefcd55 1100 struct jbd2_inode *jinode;
d444c3c3 1101
202ee5df
TT
1102 spinlock_t i_raw_lock; /* protects updates to the raw inode */
1103
d444c3c3
TT
1104 /*
1105 * File creation time. Its function is same as that of
7b62b293 1106 * struct timespec64 i_{a,c,m}time in the generic inode.
d444c3c3 1107 */
7b62b293 1108 struct timespec64 i_crtime;
d444c3c3
TT
1109
1110 /* mballoc */
27bc446e 1111 atomic_t i_prealloc_active;
d444c3c3
TT
1112 struct list_head i_prealloc_list;
1113 spinlock_t i_prealloc_lock;
1114
c0677e6d
ZL
1115 /* extents status tree */
1116 struct ext4_es_tree i_es_tree;
1117 rwlock_t i_es_lock;
edaa53ca 1118 struct list_head i_es_list;
eb68d0e2 1119 unsigned int i_es_all_nr; /* protected by i_es_lock */
edaa53ca 1120 unsigned int i_es_shk_nr; /* protected by i_es_lock */
dd475925
JK
1121 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for
1122 extents to shrink. Protected by
1123 i_es_lock */
c0677e6d 1124
d444c3c3
TT
1125 /* ialloc */
1126 ext4_group_t i_last_alloc_group;
1127
1128 /* allocation reservation info for delalloc */
a627b0a7 1129 /* In case of bigalloc, this refer to clusters rather than blocks */
d444c3c3 1130 unsigned int i_reserved_data_blocks;
d444c3c3 1131
1dc0aa46
EW
1132 /* pending cluster reservations for bigalloc file systems */
1133 struct ext4_pending_tree i_pending_tree;
1134
d444c3c3
TT
1135 /* on-disk additional length */
1136 __u16 i_extra_isize;
1137
67cf5b09
TM
1138 /* Indicate the inline data space. */
1139 u16 i_inline_off;
1140 u16 i_inline_size;
1141
a9e7f447
DM
1142#ifdef CONFIG_QUOTA
1143 /* quota space reservation, managed internally by quota code */
1144 qsize_t i_reserved_quota;
1145#endif
8d5d02e6 1146
2e8fa54e 1147 /* Lock protecting lists below */
744692dc 1148 spinlock_t i_completed_io_lock;
2e8fa54e
JK
1149 /*
1150 * Completed IOs that need unwritten extents handling and have
1151 * transaction reserved
1152 */
1153 struct list_head i_rsv_conversion_list;
2e8fa54e 1154 struct work_struct i_rsv_conversion_work;
600be30a 1155 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
8a2005d3
TT
1156
1157 spinlock_t i_block_reservation_lock;
b436b9be
JK
1158
1159 /*
1160 * Transactions that contain inode's metadata needed to complete
1161 * fsync and fdatasync, respectively.
1162 */
1163 tid_t i_sync_tid;
1164 tid_t i_datasync_tid;
814525f4 1165
96c7e0d9
JK
1166#ifdef CONFIG_QUOTA
1167 struct dquot *i_dquot[MAXQUOTAS];
1168#endif
1169
814525f4
DW
1170 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1171 __u32 i_csum_seed;
b30ab0e0 1172
040cb378 1173 kprojid_t i_projid;
d444c3c3
TT
1174};
1175
ac27a0ec
DK
1176/*
1177 * File system states
1178 */
617ba13b
MC
1179#define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
1180#define EXT4_ERROR_FS 0x0002 /* Errors detected */
1181#define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
8016e29f 1182#define EXT4_FC_REPLAY 0x0020 /* Fast commit replay ongoing */
ac27a0ec 1183
469108ff
TT
1184/*
1185 * Misc. filesystem flags
1186 */
1187#define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
1188#define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
1189#define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
1190
ac27a0ec 1191/*
68911009 1192 * Mount flags set via mount options or defaults
ac27a0ec 1193 */
cdb7ee4c 1194#define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */
617ba13b
MC
1195#define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
1196#define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
1197#define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
1198#define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
1199#define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
39ef17f1 1200#define EXT4_MOUNT_ERRORS_MASK 0x00070
617ba13b
MC
1201#define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
1202#define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
923ae0ff 1203#ifdef CONFIG_FS_DAX
fc626fe3 1204#define EXT4_MOUNT_DAX_ALWAYS 0x00200 /* Direct Access */
923ae0ff 1205#else
fc626fe3 1206#define EXT4_MOUNT_DAX_ALWAYS 0
923ae0ff 1207#endif
617ba13b
MC
1208#define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
1209#define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
1210#define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
1211#define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
1212#define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
1213#define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
1214#define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
1215#define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
afd4672d 1216#define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
617ba13b 1217#define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
49da9392
JK
1218#define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */
1219#define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota,
1220 * enable enforcement for hidden
1221 * quota files */
1222#define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable
1223 * enforcement for hidden quota
1224 * files */
1225#define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota
1226 * enforcement */
744692dc 1227#define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
d4da6c9c 1228#define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
818d276c 1229#define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
327eaf73 1230#define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */
21175ca4 1231#define EXT4_MOUNT_NO_PREFETCH_BLOCK_BITMAPS 0x4000000
64769240 1232#define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
5bf5683a 1233#define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
6fd058f7 1234#define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
5328e635 1235#define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
bfff6873 1236#define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
5bf5683a 1237
68911009
LC
1238/*
1239 * Mount flags set either automatically (could not be set by mount option)
1240 * based on per file system feature or property or in special cases such as
1241 * distinguishing between explicit mount option definition and default.
1242 */
56889787
TT
1243#define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1244 specified delalloc */
68911009
LC
1245#define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1246 size of blocksize * 8
1247 blocks */
ed3654eb
TT
1248#define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1249 file systems */
1e381f60
DM
1250#define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly
1251 specified journal checksum */
1252
995a3ed6 1253#define EXT4_MOUNT2_JOURNAL_FAST_COMMIT 0x00000010 /* Journal fast commit */
a72b38ee
HS
1254#define EXT4_MOUNT2_DAX_NEVER 0x00000020 /* Do not allow Direct Access */
1255#define EXT4_MOUNT2_DAX_INODE 0x00000040 /* For printing options only */
196e402a
HS
1256#define EXT4_MOUNT2_MB_OPTIMIZE_SCAN 0x00000080 /* Optimize group
1257 * scanning in mballoc
1258 */
995a3ed6 1259
fd8c37ec
TT
1260#define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1261 ~EXT4_MOUNT_##opt
1262#define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1263 EXT4_MOUNT_##opt
617ba13b
MC
1264#define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1265 EXT4_MOUNT_##opt)
ac27a0ec 1266
a2595b8a
TT
1267#define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1268 ~EXT4_MOUNT2_##opt
1269#define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1270 EXT4_MOUNT2_##opt
1271#define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1272 EXT4_MOUNT2_##opt)
1273
597d508c
AM
1274#define ext4_test_and_set_bit __test_and_set_bit_le
1275#define ext4_set_bit __set_bit_le
617ba13b 1276#define ext4_set_bit_atomic ext2_set_bit_atomic
597d508c
AM
1277#define ext4_test_and_clear_bit __test_and_clear_bit_le
1278#define ext4_clear_bit __clear_bit_le
617ba13b 1279#define ext4_clear_bit_atomic ext2_clear_bit_atomic
50e0168c 1280#define ext4_test_bit test_bit_le
50e0168c
AM
1281#define ext4_find_next_zero_bit find_next_zero_bit_le
1282#define ext4_find_next_bit find_next_bit_le
ac27a0ec 1283
123e3016 1284extern void mb_set_bits(void *bm, int cur, int len);
c3e94d1d 1285
ac27a0ec
DK
1286/*
1287 * Maximal mount counts between two filesystem checks
1288 */
617ba13b
MC
1289#define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1290#define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
ac27a0ec
DK
1291
1292/*
1293 * Behaviour when detecting errors
1294 */
617ba13b
MC
1295#define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1296#define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1297#define EXT4_ERRORS_PANIC 3 /* Panic */
1298#define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
ac27a0ec 1299
e6153918
DW
1300/* Metadata checksum algorithm codes */
1301#define EXT4_CRC32C_CHKSUM 1
1302
bbc605cd
LC
1303#define EXT4_LABEL_MAX 16
1304
ac27a0ec
DK
1305/*
1306 * Structure of the super block
1307 */
617ba13b 1308struct ext4_super_block {
ac27a0ec 1309/*00*/ __le32 s_inodes_count; /* Inodes count */
6bc9feff 1310 __le32 s_blocks_count_lo; /* Blocks count */
308ba3ec
AK
1311 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1312 __le32 s_free_blocks_count_lo; /* Free blocks count */
ac27a0ec
DK
1313/*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1314 __le32 s_first_data_block; /* First Data Block */
1315 __le32 s_log_block_size; /* Block size */
281b5995 1316 __le32 s_log_cluster_size; /* Allocation cluster size */
ac27a0ec 1317/*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
281b5995 1318 __le32 s_clusters_per_group; /* # Clusters per group */
ac27a0ec
DK
1319 __le32 s_inodes_per_group; /* # Inodes per group */
1320 __le32 s_mtime; /* Mount time */
1321/*30*/ __le32 s_wtime; /* Write time */
1322 __le16 s_mnt_count; /* Mount count */
1323 __le16 s_max_mnt_count; /* Maximal mount count */
1324 __le16 s_magic; /* Magic signature */
1325 __le16 s_state; /* File system state */
1326 __le16 s_errors; /* Behaviour when detecting errors */
1327 __le16 s_minor_rev_level; /* minor revision level */
1328/*40*/ __le32 s_lastcheck; /* time of last check */
1329 __le32 s_checkinterval; /* max. time between checks */
1330 __le32 s_creator_os; /* OS */
1331 __le32 s_rev_level; /* Revision level */
1332/*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1333 __le16 s_def_resgid; /* Default gid for reserved blocks */
1334 /*
617ba13b 1335 * These fields are for EXT4_DYNAMIC_REV superblocks only.
ac27a0ec
DK
1336 *
1337 * Note: the difference between the compatible feature set and
1338 * the incompatible feature set is that if there is a bit set
1339 * in the incompatible feature set that the kernel doesn't
1340 * know about, it should refuse to mount the filesystem.
1341 *
1342 * e2fsck's requirements are more strict; if it doesn't know
1343 * about a feature in either the compatible or incompatible
1344 * feature set, it must abort and not try to meddle with
1345 * things it doesn't understand...
1346 */
1347 __le32 s_first_ino; /* First non-reserved inode */
0d1ee42f 1348 __le16 s_inode_size; /* size of inode structure */
ac27a0ec
DK
1349 __le16 s_block_group_nr; /* block group # of this superblock */
1350 __le32 s_feature_compat; /* compatible feature set */
1351/*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1352 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1353/*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
bbc605cd 1354/*78*/ char s_volume_name[EXT4_LABEL_MAX]; /* volume name */
072ebb3b 1355/*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */
ac27a0ec
DK
1356/*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1357 /*
1358 * Performance hints. Directory preallocation should only
617ba13b 1359 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
ac27a0ec
DK
1360 */
1361 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1362 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1363 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1364 /*
617ba13b 1365 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
ac27a0ec
DK
1366 */
1367/*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1368/*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1369 __le32 s_journal_dev; /* device number of journal file */
1370 __le32 s_last_orphan; /* start of list of inodes to delete */
1371 __le32 s_hash_seed[4]; /* HTREE hash seed */
1372 __u8 s_def_hash_version; /* Default hash version to use */
89eeddf0 1373 __u8 s_jnl_backup_type;
0d1ee42f 1374 __le16 s_desc_size; /* size of group descriptor */
bd81d8ee 1375/*100*/ __le32 s_default_mount_opts;
ac27a0ec 1376 __le32 s_first_meta_bg; /* First metablock block group */
bd81d8ee
LV
1377 __le32 s_mkfs_time; /* When the filesystem was created */
1378 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1379 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1380/*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1381 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1382 __le32 s_free_blocks_count_hi; /* Free blocks count */
c1bddad9
AK
1383 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1384 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1385 __le32 s_flags; /* Miscellaneous flags */
1386 __le16 s_raid_stride; /* RAID stride */
c5e06d10 1387 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
c1bddad9
AK
1388 __le64 s_mmp_block; /* Block for multi-mount protection */
1389 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
772cb7c8 1390 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
e6153918 1391 __u8 s_checksum_type; /* metadata checksum algorithm used */
f542fbe8
TT
1392 __u8 s_encryption_level; /* versioning level for encryption */
1393 __u8 s_reserved_pad; /* Padding to next 32bits */
afc32f7e 1394 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
4a9cdec7
TT
1395 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1396 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1397 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1398 snapshot's future use */
1399 __le32 s_snapshot_list; /* inode number of the head of the
1400 on-disk snapshot list */
1c13d5c0
TT
1401#define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1402 __le32 s_error_count; /* number of fs errors */
1403 __le32 s_first_error_time; /* first time an error happened */
1404 __le32 s_first_error_ino; /* inode involved in first error */
1405 __le64 s_first_error_block; /* block involved of first error */
072ebb3b 1406 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */
1c13d5c0
TT
1407 __le32 s_first_error_line; /* line number where error happened */
1408 __le32 s_last_error_time; /* most recent time of an error */
1409 __le32 s_last_error_ino; /* inode involved in last error */
1410 __le32 s_last_error_line; /* line number where error happened */
1411 __le64 s_last_error_block; /* block involved of last error */
072ebb3b 1412 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */
8b67f04a
TT
1413#define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1414 __u8 s_mount_opts[64];
281b5995
TT
1415 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1416 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1417 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1beeef1b 1418 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
3edc18d8 1419 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */
f542fbe8
TT
1420 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */
1421 __le32 s_lpf_ino; /* Location of the lost+found inode */
8b4953e1 1422 __le32 s_prj_quota_inum; /* inode for tracking project quota */
8c81bd8f 1423 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */
6a0678a7
AB
1424 __u8 s_wtime_hi;
1425 __u8 s_mtime_hi;
1426 __u8 s_mkfs_time_hi;
1427 __u8 s_lastcheck_hi;
1428 __u8 s_first_error_time_hi;
1429 __u8 s_last_error_time_hi;
878520ac
TT
1430 __u8 s_first_error_errcode;
1431 __u8 s_last_error_errcode;
c83ad55e
GKB
1432 __le16 s_encoding; /* Filename charset encoding */
1433 __le16 s_encoding_flags; /* Filename charset encoding flags */
02f310fc
JK
1434 __le32 s_orphan_file_inum; /* Inode for tracking orphan inodes */
1435 __le32 s_reserved[94]; /* Padding to the end of the block */
e6153918 1436 __le32 s_checksum; /* crc32c(superblock) */
ac27a0ec
DK
1437};
1438
1c13d5c0
TT
1439#define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1440
ac27a0ec 1441#ifdef __KERNEL__
bc0b0d6d 1442
643fa961 1443#ifdef CONFIG_FS_ENCRYPTION
ac4acb1f 1444#define DUMMY_ENCRYPTION_ENABLED(sbi) ((sbi)->s_dummy_enc_policy.policy != NULL)
6ddb2447
TT
1445#else
1446#define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1447#endif
bc0b0d6d 1448
a2d4a646 1449/* Number of quota types we support */
689c958c 1450#define EXT4_MAXQUOTAS 3
a2d4a646 1451
c83ad55e
GKB
1452#define EXT4_ENC_UTF8_12_1 1
1453
188c299e
JK
1454/* Types of ext4 journal triggers */
1455enum ext4_journal_trigger_type {
02f310fc 1456 EXT4_JTR_ORPHAN_FILE,
188c299e
JK
1457 EXT4_JTR_NONE /* This must be the last entry for indexing to work! */
1458};
1459
1460#define EXT4_JOURNAL_TRIGGER_COUNT EXT4_JTR_NONE
1461
1462struct ext4_journal_trigger {
1463 struct jbd2_buffer_trigger_type tr_triggers;
1464 struct super_block *sb;
1465};
1466
1467static inline struct ext4_journal_trigger *EXT4_TRIGGER(
1468 struct jbd2_buffer_trigger_type *trigger)
1469{
1470 return container_of(trigger, struct ext4_journal_trigger, tr_triggers);
1471}
1472
02f310fc
JK
1473#define EXT4_ORPHAN_BLOCK_MAGIC 0x0b10ca04
1474
1475/* Structure at the tail of orphan block */
1476struct ext4_orphan_block_tail {
1477 __le32 ob_magic;
1478 __le32 ob_checksum;
1479};
1480
1481static inline int ext4_inodes_per_orphan_block(struct super_block *sb)
1482{
1483 return (sb->s_blocksize - sizeof(struct ext4_orphan_block_tail)) /
1484 sizeof(u32);
1485}
1486
1487struct ext4_orphan_block {
4a79a98c 1488 atomic_t ob_free_entries; /* Number of free orphan entries in block */
02f310fc
JK
1489 struct buffer_head *ob_bh; /* Buffer for orphan block */
1490};
1491
1492/*
1493 * Info about orphan file.
1494 */
1495struct ext4_orphan_info {
02f310fc
JK
1496 int of_blocks; /* Number of orphan blocks in a file */
1497 __u32 of_csum_seed; /* Checksum seed for orphan file */
1498 struct ext4_orphan_block *of_binfo; /* Array with info about orphan
1499 * file blocks */
1500};
1501
ca0faba0
TT
1502/*
1503 * fourth extended-fs super-block data in memory
1504 */
1505struct ext4_sb_info {
1506 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1507 unsigned long s_inodes_per_block;/* Number of inodes per block */
1508 unsigned long s_blocks_per_group;/* Number of blocks in a group */
281b5995 1509 unsigned long s_clusters_per_group; /* Number of clusters in a group */
ca0faba0
TT
1510 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1511 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1512 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1513 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1514 ext4_group_t s_groups_count; /* Number of groups in the fs */
fb0a387d 1515 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
952fc18e 1516 unsigned long s_overhead; /* # of fs overhead clusters */
281b5995
TT
1517 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1518 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
ca0faba0
TT
1519 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1520 struct buffer_head * s_sbh; /* Buffer containing the super block */
1521 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1d0c3924 1522 struct buffer_head * __rcu *s_group_desc;
7f4520cc 1523 unsigned int s_mount_opt;
a2595b8a 1524 unsigned int s_mount_opt2;
9b5f6c9b 1525 unsigned long s_mount_flags;
5a916be1 1526 unsigned int s_def_mount_opt;
ca0faba0 1527 ext4_fsblk_t s_sb_block;
27dd4385 1528 atomic64_t s_resv_clusters;
08cefc7a
EB
1529 kuid_t s_resuid;
1530 kgid_t s_resgid;
ca0faba0
TT
1531 unsigned short s_mount_state;
1532 unsigned short s_pad;
1533 int s_addr_per_block_bits;
1534 int s_desc_per_block_bits;
1535 int s_inode_size;
1536 int s_first_ino;
1537 unsigned int s_inode_readahead_blks;
11013911 1538 unsigned int s_inode_goal;
ca0faba0
TT
1539 u32 s_hash_seed[4];
1540 int s_def_hash_version;
6d2424a8 1541 int s_hash_unsigned; /* 3 if hash should be unsigned, 0 if not */
57042651 1542 struct percpu_counter s_freeclusters_counter;
ca0faba0
TT
1543 struct percpu_counter s_freeinodes_counter;
1544 struct percpu_counter s_dirs_counter;
57042651 1545 struct percpu_counter s_dirtyclusters_counter;
efc61345 1546 struct percpu_counter s_sra_exceeded_retry_limit;
ca0faba0
TT
1547 struct blockgroup_lock *s_blockgroup_lock;
1548 struct proc_dir_entry *s_proc;
1549 struct kobject s_kobj;
1550 struct completion s_kobj_unregister;
2c0544b2 1551 struct super_block *s_sb;
618f0031 1552 struct buffer_head *s_mmp_bh;
ca0faba0
TT
1553
1554 /* Journaling */
ca0faba0 1555 struct journal_s *s_journal;
9549a168 1556 unsigned long s_ext4_flags; /* Ext4 superblock flags */
02f310fc
JK
1557 struct mutex s_orphan_lock; /* Protects on disk list changes */
1558 struct list_head s_orphan; /* List of orphaned inodes in on disk
1559 list */
1560 struct ext4_orphan_info s_orphan_info;
ca0faba0
TT
1561 unsigned long s_commit_interval;
1562 u32 s_max_batch_time;
1563 u32 s_min_batch_time;
ee7ed3aa 1564 struct block_device *s_journal_bdev;
ca0faba0 1565#ifdef CONFIG_QUOTA
33458eab
TT
1566 /* Names of quota files with journalled quota */
1567 char __rcu *s_qf_names[EXT4_MAXQUOTAS];
ca0faba0
TT
1568 int s_jquota_fmt; /* Format of quota to use */
1569#endif
1570 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
dd0db94f 1571 struct ext4_system_blocks __rcu *s_system_blks;
ca0faba0
TT
1572
1573#ifdef EXTENTS_STATS
1574 /* ext4 extents stats */
1575 unsigned long s_ext_min;
1576 unsigned long s_ext_max;
1577 unsigned long s_depth_max;
1578 spinlock_t s_ext_stats_lock;
1579 unsigned long s_ext_blocks;
1580 unsigned long s_ext_extents;
1581#endif
1582
1583 /* for buddy allocator */
df3da4ea 1584 struct ext4_group_info ** __rcu *s_group_info;
ca0faba0 1585 struct inode *s_buddy_cache;
ca0faba0 1586 spinlock_t s_md_lock;
ca0faba0
TT
1587 unsigned short *s_mb_offsets;
1588 unsigned int *s_mb_maxs;
28623c2f 1589 unsigned int s_group_info_size;
d08854f5 1590 unsigned int s_mb_free_pending;
a0154344
DJ
1591 struct list_head s_freed_data_list; /* List of blocks to be freed
1592 after commit completed */
55cdd0af
WJ
1593 struct list_head s_discard_list;
1594 struct work_struct s_discard_work;
5036ab8d 1595 atomic_t s_retry_alloc_pending;
196e402a
HS
1596 struct rb_root s_mb_avg_fragment_size_root;
1597 rwlock_t s_mb_rb_lock;
1598 struct list_head *s_mb_largest_free_orders;
1599 rwlock_t *s_mb_largest_free_orders_locks;
ca0faba0
TT
1600
1601 /* tunables */
1602 unsigned long s_stripe;
196e402a 1603 unsigned int s_mb_max_linear_groups;
ca0faba0
TT
1604 unsigned int s_mb_stream_request;
1605 unsigned int s_mb_max_to_scan;
1606 unsigned int s_mb_min_to_scan;
1607 unsigned int s_mb_stats;
1608 unsigned int s_mb_order2_reqs;
1609 unsigned int s_mb_group_prealloc;
27bc446e 1610 unsigned int s_mb_max_inode_prealloc;
df981d03 1611 unsigned int s_max_dir_size_kb;
ca0faba0
TT
1612 /* where last allocation was done - for stream allocation */
1613 unsigned long s_mb_last_group;
1614 unsigned long s_mb_last_start;
cfd73237
AZ
1615 unsigned int s_mb_prefetch;
1616 unsigned int s_mb_prefetch_limit;
ca0faba0 1617
ca0faba0 1618 /* stats for buddy allocator */
ca0faba0
TT
1619 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1620 atomic_t s_bal_success; /* we found long enough chunks */
1621 atomic_t s_bal_allocated; /* in blocks */
1622 atomic_t s_bal_ex_scanned; /* total extents scanned */
a6c75eaf 1623 atomic_t s_bal_groups_scanned; /* number of groups scanned */
ca0faba0
TT
1624 atomic_t s_bal_goals; /* goal hits */
1625 atomic_t s_bal_breaks; /* too long searches */
1626 atomic_t s_bal_2orders; /* 2^order hits */
196e402a
HS
1627 atomic_t s_bal_cr0_bad_suggestions;
1628 atomic_t s_bal_cr1_bad_suggestions;
a6c75eaf
HS
1629 atomic64_t s_bal_cX_groups_considered[4];
1630 atomic64_t s_bal_cX_hits[4];
1631 atomic64_t s_bal_cX_failed[4]; /* cX loop didn't find blocks */
67d25186
HS
1632 atomic_t s_mb_buddies_generated; /* number of buddies generated */
1633 atomic64_t s_mb_generation_time;
ca0faba0
TT
1634 atomic_t s_mb_lost_chunks;
1635 atomic_t s_mb_preallocated;
1636 atomic_t s_mb_discarded;
50797481 1637 atomic_t s_lock_busy;
ca0faba0
TT
1638
1639 /* locality groups */
003cb608 1640 struct ext4_locality_group __percpu *s_locality_groups;
ca0faba0
TT
1641
1642 /* for write statistics */
1643 unsigned long s_sectors_written_start;
1644 u64 s_kbytes_written;
1645
67a5da56
ZL
1646 /* the size of zero-out chunk */
1647 unsigned int s_extent_max_zeroout_kb;
1648
ca0faba0 1649 unsigned int s_log_groups_per_flex;
7c990728 1650 struct flex_groups * __rcu *s_flex_groups;
117fff10 1651 ext4_group_t s_flex_groups_allocated;
4c0425ff 1652
2e8fa54e
JK
1653 /* workqueue for reserved extent conversions (buffered io) */
1654 struct workqueue_struct *rsv_conversion_wq;
66e61a9e
TT
1655
1656 /* timer for periodic error stats printing */
1657 struct timer_list s_err_report;
bfff6873
LC
1658
1659 /* Lazy inode table initialization info */
1660 struct ext4_li_request *s_li_request;
1661 /* Wait multiplier for lazy initialization thread */
1662 unsigned int s_li_wait_mult;
c5e06d10
JL
1663
1664 /* Kernel thread for multiple mount protection */
1665 struct task_struct *s_mmp_tsk;
3d56b8d2
TM
1666
1667 /* record the last minlen when FITRIM is called. */
2327fb2e 1668 unsigned long s_last_trim_minblks;
0441984a
DW
1669
1670 /* Reference to checksum algorithm driver via cryptoapi */
1671 struct crypto_shash *s_chksum_driver;
a9c47317
DW
1672
1673 /* Precomputed FS UUID checksum for seeding other checksums */
1674 __u32 s_csum_seed;
74cd15cd
ZL
1675
1676 /* Reclaim extents from extent status tree */
1677 struct shrinker s_es_shrinker;
dd475925 1678 struct list_head s_es_list; /* List of inodes with reclaimable extents */
edaa53ca 1679 long s_es_nr_inode;
eb68d0e2 1680 struct ext4_es_stats s_es_stats;
47387409 1681 struct mb_cache *s_ea_block_cache;
dec214d0 1682 struct mb_cache *s_ea_inode_cache;
edaa53ca 1683 spinlock_t s_es_lock ____cacheline_aligned_in_smp;
efbed4dc 1684
188c299e
JK
1685 /* Journal triggers for checksum computation */
1686 struct ext4_journal_trigger s_journal_triggers[EXT4_JOURNAL_TRIGGER_COUNT];
1687
efbed4dc
TT
1688 /* Ratelimit ext4 messages. */
1689 struct ratelimit_state s_err_ratelimit_state;
1690 struct ratelimit_state s_warning_ratelimit_state;
1691 struct ratelimit_state s_msg_ratelimit_state;
1cf006ed
DM
1692 atomic_t s_warning_count;
1693 atomic_t s_msg_count;
c8585c6f 1694
ac4acb1f
EB
1695 /* Encryption policy for '-o test_dummy_encryption' */
1696 struct fscrypt_dummy_policy s_dummy_enc_policy;
ed318a6c 1697
cb85f4d2
EB
1698 /*
1699 * Barrier between writepages ops and changing any inode's JOURNAL_DATA
1700 * or EXTENTS flag.
1701 */
bbd55937 1702 struct percpu_rw_semaphore s_writepages_rwsem;
5e405595 1703 struct dax_device *s_daxdev;
cd913c76 1704 u64 s_dax_part_off;
46f870d6
TT
1705#ifdef CONFIG_EXT4_DEBUG
1706 unsigned long s_simulate_fail;
1707#endif
bc71726c 1708 /* Record the errseq of the backing block device */
1709 errseq_t s_bdev_wb_err;
1710 spinlock_t s_bdev_wb_lock;
aa75f4d3 1711
c92dc856
JK
1712 /* Information about errors that happened during this mount */
1713 spinlock_t s_error_lock;
1714 int s_add_error_count;
1715 int s_first_error_code;
1716 __u32 s_first_error_line;
1717 __u32 s_first_error_ino;
1718 __u64 s_first_error_block;
1719 const char *s_first_error_func;
1720 time64_t s_first_error_time;
1721 int s_last_error_code;
1722 __u32 s_last_error_line;
1723 __u32 s_last_error_ino;
1724 __u64 s_last_error_block;
1725 const char *s_last_error_func;
1726 time64_t s_last_error_time;
1727 /*
1728 * If we are in a context where we cannot update error information in
1729 * the on-disk superblock, we queue this work to do it.
1730 */
1731 struct work_struct s_error_work;
1732
7bbbe241 1733 /* Ext4 fast commit sub transaction ID */
aa75f4d3 1734 atomic_t s_fc_subtid;
7bbbe241 1735
aa75f4d3
HS
1736 /*
1737 * After commit starts, the main queue gets locked, and the further
1738 * updates get added in the staging queue.
1739 */
1740#define FC_Q_MAIN 0
1741#define FC_Q_STAGING 1
1742 struct list_head s_fc_q[2]; /* Inodes staged for fast commit
1743 * that have data changes in them.
1744 */
1745 struct list_head s_fc_dentry_q[2]; /* directory entry updates */
1746 unsigned int s_fc_bytes;
1747 /*
1748 * Main fast commit lock. This lock protects accesses to the
1749 * following fields:
1750 * ei->i_fc_list, s_fc_dentry_q, s_fc_q, s_fc_bytes, s_fc_bh.
1751 */
1752 spinlock_t s_fc_lock;
1753 struct buffer_head *s_fc_bh;
1754 struct ext4_fc_stats s_fc_stats;
e85c81ba 1755 tid_t s_fc_ineligible_tid;
8016e29f
HS
1756#ifdef CONFIG_EXT4_DEBUG
1757 int s_fc_debug_max_replay;
1758#endif
1759 struct ext4_fc_replay_state s_fc_replay_state;
ca0faba0
TT
1760};
1761
af5bc92d 1762static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
ac27a0ec
DK
1763{
1764 return sb->s_fs_info;
1765}
617ba13b 1766static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
ac27a0ec 1767{
617ba13b 1768 return container_of(inode, struct ext4_inode_info, vfs_inode);
ac27a0ec
DK
1769}
1770
617ba13b 1771static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
ac27a0ec 1772{
617ba13b 1773 return ino == EXT4_ROOT_INO ||
617ba13b
MC
1774 (ino >= EXT4_FIRST_INO(sb) &&
1775 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
ac27a0ec 1776}
19f5fb7a 1777
1d0c3924
TT
1778/*
1779 * Returns: sbi->field[index]
1780 * Used to access an array element from the following sbi fields which require
1781 * rcu protection to avoid dereferencing an invalid pointer due to reassignment
1782 * - s_group_desc
1783 * - s_group_info
1784 * - s_flex_group
1785 */
1786#define sbi_array_rcu_deref(sbi, field, index) \
1787({ \
1788 typeof(*((sbi)->field)) _v; \
1789 rcu_read_lock(); \
1790 _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index]; \
1791 rcu_read_unlock(); \
1792 _v; \
1793})
1794
9b5f6c9b
HS
1795/*
1796 * run-time mount flags
1797 */
1798enum {
1799 EXT4_MF_MNTDIR_SAMPLED,
1800 EXT4_MF_FS_ABORTED, /* Fatal error detected */
bdc8a53a 1801 EXT4_MF_FC_INELIGIBLE /* Fast commit ineligible */
9b5f6c9b
HS
1802};
1803
1804static inline void ext4_set_mount_flag(struct super_block *sb, int bit)
1805{
1806 set_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1807}
1808
1809static inline void ext4_clear_mount_flag(struct super_block *sb, int bit)
1810{
1811 clear_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1812}
1813
1814static inline int ext4_test_mount_flag(struct super_block *sb, int bit)
1815{
1816 return test_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1817}
1818
1819
46f870d6
TT
1820/*
1821 * Simulate_fail codes
1822 */
1823#define EXT4_SIM_BBITMAP_EIO 1
1824#define EXT4_SIM_BBITMAP_CRC 2
1825#define EXT4_SIM_IBITMAP_EIO 3
1826#define EXT4_SIM_IBITMAP_CRC 4
1827#define EXT4_SIM_INODE_EIO 5
1828#define EXT4_SIM_INODE_CRC 6
1829#define EXT4_SIM_DIRBLOCK_EIO 7
1830#define EXT4_SIM_DIRBLOCK_CRC 8
1831
1832static inline bool ext4_simulate_fail(struct super_block *sb,
1833 unsigned long code)
1834{
1835#ifdef CONFIG_EXT4_DEBUG
1836 struct ext4_sb_info *sbi = EXT4_SB(sb);
1837
1838 if (unlikely(sbi->s_simulate_fail == code)) {
1839 sbi->s_simulate_fail = 0;
1840 return true;
1841 }
1842#endif
1843 return false;
1844}
1845
1846static inline void ext4_simulate_fail_bh(struct super_block *sb,
1847 struct buffer_head *bh,
1848 unsigned long code)
1849{
1850 if (!IS_ERR(bh) && ext4_simulate_fail(sb, code))
1851 clear_buffer_uptodate(bh);
1852}
1853
878520ac
TT
1854/*
1855 * Error number codes for s_{first,last}_error_errno
1856 *
1857 * Linux errno numbers are architecture specific, so we need to translate
1858 * them into something which is architecture independent. We don't define
1859 * codes for all errno's; just the ones which are most likely to be the cause
1860 * of an ext4_error() call.
1861 */
1862#define EXT4_ERR_UNKNOWN 1
1863#define EXT4_ERR_EIO 2
1864#define EXT4_ERR_ENOMEM 3
1865#define EXT4_ERR_EFSBADCRC 4
1866#define EXT4_ERR_EFSCORRUPTED 5
1867#define EXT4_ERR_ENOSPC 6
1868#define EXT4_ERR_ENOKEY 7
1869#define EXT4_ERR_EROFS 8
1870#define EXT4_ERR_EFBIG 9
1871#define EXT4_ERR_EEXIST 10
1872#define EXT4_ERR_ERANGE 11
1873#define EXT4_ERR_EOVERFLOW 12
1874#define EXT4_ERR_EBUSY 13
1875#define EXT4_ERR_ENOTDIR 14
1876#define EXT4_ERR_ENOTEMPTY 15
1877#define EXT4_ERR_ESHUTDOWN 16
1878#define EXT4_ERR_EFAULT 17
1879
19f5fb7a
TT
1880/*
1881 * Inode dynamic state flags
1882 */
1883enum {
1884 EXT4_STATE_JDATA, /* journaled data exists */
1885 EXT4_STATE_NEW, /* inode is newly created */
1886 EXT4_STATE_XATTR, /* has in-inode xattrs */
1887 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1888 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1889 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
14ece102 1890 EXT4_STATE_NEWENTRY, /* File just added to dir */
67cf5b09 1891 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
7869a4a6 1892 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
a6d05676 1893 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */
c93d8f88 1894 EXT4_STATE_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */
aa75f4d3 1895 EXT4_STATE_FC_COMMITTING, /* Fast commit ongoing */
02f310fc 1896 EXT4_STATE_ORPHAN_FILE, /* Inode orphaned in orphan file */
19f5fb7a
TT
1897};
1898
353eb83c 1899#define EXT4_INODE_BIT_FNS(name, field, offset) \
12e9b892
DM
1900static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1901{ \
353eb83c 1902 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
12e9b892
DM
1903} \
1904static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1905{ \
353eb83c 1906 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
12e9b892
DM
1907} \
1908static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1909{ \
353eb83c 1910 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
19f5fb7a
TT
1911}
1912
3f61c0cc
AD
1913/* Add these declarations here only so that these functions can be
1914 * found by name. Otherwise, they are very hard to locate. */
1915static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1916static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1917static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
353eb83c 1918EXT4_INODE_BIT_FNS(flag, flags, 0)
3f61c0cc
AD
1919
1920/* Add these declarations here only so that these functions can be
1921 * found by name. Otherwise, they are very hard to locate. */
1922static inline int ext4_test_inode_state(struct inode *inode, int bit);
1923static inline void ext4_set_inode_state(struct inode *inode, int bit);
1924static inline void ext4_clear_inode_state(struct inode *inode, int bit);
353eb83c
TT
1925#if (BITS_PER_LONG < 64)
1926EXT4_INODE_BIT_FNS(state, state_flags, 0)
1927
1928static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1929{
1930 (ei)->i_state_flags = 0;
1931}
1932#else
1933EXT4_INODE_BIT_FNS(state, flags, 32)
1934
1935static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1936{
1937 /* We depend on the fact that callers will set i_flags */
1938}
1939#endif
ac27a0ec 1940#else
617ba13b 1941/* Assume that user mode programs are passing in an ext4fs superblock, not
ac27a0ec
DK
1942 * a kernel struct super_block. This will allow us to call the feature-test
1943 * macros from user land. */
617ba13b 1944#define EXT4_SB(sb) (sb)
ac27a0ec
DK
1945#endif
1946
c93d8f88
EB
1947static inline bool ext4_verity_in_progress(struct inode *inode)
1948{
1949 return IS_ENABLED(CONFIG_FS_VERITY) &&
1950 ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS);
1951}
1952
617ba13b 1953#define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
ac27a0ec
DK
1954
1955/*
1956 * Codes for operating systems
1957 */
617ba13b
MC
1958#define EXT4_OS_LINUX 0
1959#define EXT4_OS_HURD 1
1960#define EXT4_OS_MASIX 2
1961#define EXT4_OS_FREEBSD 3
1962#define EXT4_OS_LITES 4
ac27a0ec
DK
1963
1964/*
1965 * Revision levels
1966 */
617ba13b
MC
1967#define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1968#define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
ac27a0ec 1969
617ba13b 1970#define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
ac27a0ec 1971
617ba13b 1972#define EXT4_GOOD_OLD_INODE_SIZE 128
ac27a0ec 1973
4881c497
DD
1974#define EXT4_EXTRA_TIMESTAMP_MAX (((s64)1 << 34) - 1 + S32_MIN)
1975#define EXT4_NON_EXTRA_TIMESTAMP_MAX S32_MAX
1976#define EXT4_TIMESTAMP_MIN S32_MIN
1977
ac27a0ec
DK
1978/*
1979 * Feature set definitions
1980 */
1981
617ba13b
MC
1982#define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1983#define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1984#define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1985#define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1986#define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1987#define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1beeef1b 1988#define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
a44ad683
HS
1989/*
1990 * The reason why "FAST_COMMIT" is a compat feature is that, FS becomes
1991 * incompatible only if fast commit blocks are present in the FS. Since we
1992 * clear the journal (and thus the fast commit blocks), we don't mark FS as
1993 * incompatible. We also have a JBD2 incompat feature, which gets set when
1994 * there are fast commit blocks present in the journal.
1995 */
995a3ed6 1996#define EXT4_FEATURE_COMPAT_FAST_COMMIT 0x0400
b925acb8 1997#define EXT4_FEATURE_COMPAT_STABLE_INODES 0x0800
02f310fc 1998#define EXT4_FEATURE_COMPAT_ORPHAN_FILE 0x1000 /* Orphan file exists */
617ba13b
MC
1999
2000#define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
2001#define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
2002#define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
0fc1b451 2003#define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
717d50e4 2004#define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
f8628a14 2005#define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
ef7f3835 2006#define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
ae812306 2007#define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
281b5995 2008#define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
e6153918
DW
2009/*
2010 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
2011 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
2012 * all other data structures' checksums. However, the METADATA_CSUM and
2013 * GDT_CSUM bits are mutually exclusive.
2014 */
9b90e5e0 2015#define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
2cb5cc8b 2016#define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000
8b4953e1 2017#define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000
c93d8f88 2018#define EXT4_FEATURE_RO_COMPAT_VERITY 0x8000
02f310fc
JK
2019#define EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT 0x10000 /* Orphan file may be
2020 non-empty */
617ba13b
MC
2021
2022#define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
2023#define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
2024#define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
2025#define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
2026#define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
a86c6181 2027#define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
bd81d8ee 2028#define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
0fc1b451 2029#define EXT4_FEATURE_INCOMPAT_MMP 0x0100
ce421581 2030#define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
f710b4b9
TT
2031#define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
2032#define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
8c81bd8f 2033#define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000
9b90e5e0 2034#define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
67cf5b09 2035#define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
3edc18d8 2036#define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000
c83ad55e 2037#define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000
617ba13b 2038
c9e716eb
AD
2039extern void ext4_update_dynamic_rev(struct super_block *sb);
2040
e2b911c5
DW
2041#define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
2042static inline bool ext4_has_feature_##name(struct super_block *sb) \
2043{ \
2044 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
2045 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
2046} \
2047static inline void ext4_set_feature_##name(struct super_block *sb) \
2048{ \
c9e716eb 2049 ext4_update_dynamic_rev(sb); \
e2b911c5
DW
2050 EXT4_SB(sb)->s_es->s_feature_compat |= \
2051 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
2052} \
2053static inline void ext4_clear_feature_##name(struct super_block *sb) \
2054{ \
2055 EXT4_SB(sb)->s_es->s_feature_compat &= \
2056 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
2057}
2058
2059#define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
2060static inline bool ext4_has_feature_##name(struct super_block *sb) \
2061{ \
2062 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
2063 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
2064} \
2065static inline void ext4_set_feature_##name(struct super_block *sb) \
2066{ \
c9e716eb 2067 ext4_update_dynamic_rev(sb); \
e2b911c5
DW
2068 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
2069 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
2070} \
2071static inline void ext4_clear_feature_##name(struct super_block *sb) \
2072{ \
2073 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
2074 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
2075}
2076
2077#define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
2078static inline bool ext4_has_feature_##name(struct super_block *sb) \
2079{ \
2080 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
2081 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
2082} \
2083static inline void ext4_set_feature_##name(struct super_block *sb) \
2084{ \
c9e716eb 2085 ext4_update_dynamic_rev(sb); \
e2b911c5
DW
2086 EXT4_SB(sb)->s_es->s_feature_incompat |= \
2087 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
2088} \
2089static inline void ext4_clear_feature_##name(struct super_block *sb) \
2090{ \
2091 EXT4_SB(sb)->s_es->s_feature_incompat &= \
2092 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
2093}
2094
2095EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC)
2096EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES)
2097EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL)
2098EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR)
2099EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE)
2100EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX)
2101EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2)
995a3ed6 2102EXT4_FEATURE_COMPAT_FUNCS(fast_commit, FAST_COMMIT)
b925acb8 2103EXT4_FEATURE_COMPAT_FUNCS(stable_inodes, STABLE_INODES)
02f310fc 2104EXT4_FEATURE_COMPAT_FUNCS(orphan_file, ORPHAN_FILE)
e2b911c5
DW
2105
2106EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER)
2107EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE)
2108EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR)
2109EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE)
2110EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM)
2111EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK)
2112EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE)
2113EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA)
2114EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC)
2115EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM)
2116EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY)
2117EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT)
c93d8f88 2118EXT4_FEATURE_RO_COMPAT_FUNCS(verity, VERITY)
02f310fc 2119EXT4_FEATURE_RO_COMPAT_FUNCS(orphan_present, ORPHAN_PRESENT)
e2b911c5
DW
2120
2121EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION)
2122EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE)
2123EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER)
2124EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV)
2125EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG)
2126EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS)
2127EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT)
2128EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP)
2129EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG)
2130EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE)
2131EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA)
2132EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED)
2133EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR)
2134EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA)
2135EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT)
c83ad55e 2136EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD)
e2b911c5 2137
2035e776
TT
2138#define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
2139#define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
2140 EXT4_FEATURE_INCOMPAT_META_BG)
2141#define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2142 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2143 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
2144
2145#define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
2146#define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
2147 EXT4_FEATURE_INCOMPAT_RECOVER| \
2148 EXT4_FEATURE_INCOMPAT_META_BG)
2149#define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2150 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2151 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
2152
02f310fc
JK
2153#define EXT4_FEATURE_COMPAT_SUPP (EXT4_FEATURE_COMPAT_EXT_ATTR| \
2154 EXT4_FEATURE_COMPAT_ORPHAN_FILE)
617ba13b
MC
2155#define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
2156 EXT4_FEATURE_INCOMPAT_RECOVER| \
a86c6181 2157 EXT4_FEATURE_INCOMPAT_META_BG| \
bd81d8ee 2158 EXT4_FEATURE_INCOMPAT_EXTENTS| \
ce421581 2159 EXT4_FEATURE_INCOMPAT_64BIT| \
c5e06d10 2160 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
e50e5129 2161 EXT4_FEATURE_INCOMPAT_EA_INODE| \
6ddb2447
TT
2162 EXT4_FEATURE_INCOMPAT_MMP | \
2163 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
8c81bd8f 2164 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
c83ad55e 2165 EXT4_FEATURE_INCOMPAT_CASEFOLD | \
e08ac99f
AB
2166 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \
2167 EXT4_FEATURE_INCOMPAT_LARGEDIR)
617ba13b
MC
2168#define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2169 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
717d50e4 2170 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
f8628a14 2171 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
ef7f3835 2172 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
0fc1b451 2173 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
6f16b606 2174 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
e93376c2 2175 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
7c319d32 2176 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
040cb378 2177 EXT4_FEATURE_RO_COMPAT_QUOTA |\
c93d8f88 2178 EXT4_FEATURE_RO_COMPAT_PROJECT |\
02f310fc
JK
2179 EXT4_FEATURE_RO_COMPAT_VERITY |\
2180 EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT)
ac27a0ec 2181
e2b911c5
DW
2182#define EXTN_FEATURE_FUNCS(ver) \
2183static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
2184{ \
2185 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
2186 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
2187} \
2188static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
2189{ \
2190 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
2191 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
2192} \
2193static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
2194{ \
2195 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
2196 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
2197}
2198
2199EXTN_FEATURE_FUNCS(2)
2200EXTN_FEATURE_FUNCS(3)
2201EXTN_FEATURE_FUNCS(4)
2202
2203static inline bool ext4_has_compat_features(struct super_block *sb)
2204{
2205 return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
2206}
2207static inline bool ext4_has_ro_compat_features(struct super_block *sb)
2208{
2209 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
2210}
2211static inline bool ext4_has_incompat_features(struct super_block *sb)
2212{
2213 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
2214}
2215
25c6d98f
JK
2216extern int ext4_feature_set_ok(struct super_block *sb, int readonly);
2217
9549a168
TT
2218/*
2219 * Superblock flags
2220 */
2221#define EXT4_FLAGS_RESIZING 0
0db1ff22 2222#define EXT4_FLAGS_SHUTDOWN 1
a8ab6d38 2223#define EXT4_FLAGS_BDEV_IS_DAX 2
0db1ff22
TT
2224
2225static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi)
2226{
2227 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags);
2228}
9549a168 2229
ac27a0ec
DK
2230/*
2231 * Default values for user and/or group using reserved blocks
2232 */
617ba13b
MC
2233#define EXT4_DEF_RESUID 0
2234#define EXT4_DEF_RESGID 0
ac27a0ec 2235
040cb378
LX
2236/*
2237 * Default project ID
2238 */
2239#define EXT4_DEF_PROJID 0
2240
240799cd
TT
2241#define EXT4_DEF_INODE_READAHEAD_BLKS 32
2242
ac27a0ec
DK
2243/*
2244 * Default mount options
2245 */
617ba13b
MC
2246#define EXT4_DEFM_DEBUG 0x0001
2247#define EXT4_DEFM_BSDGROUPS 0x0002
2248#define EXT4_DEFM_XATTR_USER 0x0004
2249#define EXT4_DEFM_ACL 0x0008
2250#define EXT4_DEFM_UID16 0x0010
2251#define EXT4_DEFM_JMODE 0x0060
2252#define EXT4_DEFM_JMODE_DATA 0x0020
2253#define EXT4_DEFM_JMODE_ORDERED 0x0040
2254#define EXT4_DEFM_JMODE_WBACK 0x0060
8b67f04a
TT
2255#define EXT4_DEFM_NOBARRIER 0x0100
2256#define EXT4_DEFM_BLOCK_VALIDITY 0x0200
2257#define EXT4_DEFM_DISCARD 0x0400
2258#define EXT4_DEFM_NODELALLOC 0x0800
ac27a0ec 2259
30773840
TT
2260/*
2261 * Default journal batch times
2262 */
2263#define EXT4_DEF_MIN_BATCH_TIME 0
2264#define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
2265
a4912123
TT
2266/*
2267 * Minimum number of groups in a flexgroup before we separate out
2268 * directories into the first block group of a flexgroup
2269 */
2270#define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
2271
ac27a0ec
DK
2272/*
2273 * Structure of a directory entry
2274 */
617ba13b 2275#define EXT4_NAME_LEN 255
c186f088
YB
2276/*
2277 * Base length of the ext4 directory entry excluding the name length
2278 */
2279#define EXT4_BASE_DIR_LEN (sizeof(struct ext4_dir_entry_2) - EXT4_NAME_LEN)
ac27a0ec 2280
617ba13b 2281struct ext4_dir_entry {
ac27a0ec
DK
2282 __le32 inode; /* Inode number */
2283 __le16 rec_len; /* Directory entry length */
2284 __le16 name_len; /* Name length */
617ba13b 2285 char name[EXT4_NAME_LEN]; /* File name */
ac27a0ec
DK
2286};
2287
471fbbea
DR
2288
2289/*
2290 * Encrypted Casefolded entries require saving the hash on disk. This structure
2291 * followed ext4_dir_entry_2's name[name_len] at the next 4 byte aligned
2292 * boundary.
2293 */
2294struct ext4_dir_entry_hash {
2295 __le32 hash;
2296 __le32 minor_hash;
2297};
2298
ac27a0ec 2299/*
617ba13b 2300 * The new version of the directory entry. Since EXT4 structures are
ac27a0ec
DK
2301 * stored in intel byte order, and the name_len field could never be
2302 * bigger than 255 chars, it's safe to reclaim the extra byte for the
2303 * file_type field.
2304 */
617ba13b 2305struct ext4_dir_entry_2 {
ac27a0ec
DK
2306 __le32 inode; /* Inode number */
2307 __le16 rec_len; /* Directory entry length */
2308 __u8 name_len; /* Name length */
9f364e1d 2309 __u8 file_type; /* See file type macros EXT4_FT_* below */
617ba13b 2310 char name[EXT4_NAME_LEN]; /* File name */
ac27a0ec
DK
2311};
2312
471fbbea
DR
2313/*
2314 * Access the hashes at the end of ext4_dir_entry_2
2315 */
2316#define EXT4_DIRENT_HASHES(entry) \
2317 ((struct ext4_dir_entry_hash *) \
2318 (((void *)(entry)) + \
2319 ((8 + (entry)->name_len + EXT4_DIR_ROUND) & ~EXT4_DIR_ROUND)))
2320#define EXT4_DIRENT_HASH(entry) le32_to_cpu(EXT4_DIRENT_HASHES(de)->hash)
2321#define EXT4_DIRENT_MINOR_HASH(entry) \
2322 le32_to_cpu(EXT4_DIRENT_HASHES(de)->minor_hash)
2323
2324static inline bool ext4_hash_in_dirent(const struct inode *inode)
2325{
2326 return IS_CASEFOLDED(inode) && IS_ENCRYPTED(inode);
2327}
2328
e6153918
DW
2329/*
2330 * This is a bogus directory entry at the end of each leaf block that
2331 * records checksums.
2332 */
2333struct ext4_dir_entry_tail {
2334 __le32 det_reserved_zero1; /* Pretend to be unused */
2335 __le16 det_rec_len; /* 12 */
2336 __u8 det_reserved_zero2; /* Zero name length */
2337 __u8 det_reserved_ft; /* 0xDE, fake file type */
2338 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
2339};
2340
3c47d541
TM
2341#define EXT4_DIRENT_TAIL(block, blocksize) \
2342 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
2343 ((blocksize) - \
2344 sizeof(struct ext4_dir_entry_tail))))
2345
ac27a0ec 2346/*
617ba13b 2347 * Ext4 directory file types. Only the low 3 bits are used. The
ac27a0ec
DK
2348 * other bits are reserved for now.
2349 */
617ba13b
MC
2350#define EXT4_FT_UNKNOWN 0
2351#define EXT4_FT_REG_FILE 1
2352#define EXT4_FT_DIR 2
2353#define EXT4_FT_CHRDEV 3
2354#define EXT4_FT_BLKDEV 4
2355#define EXT4_FT_FIFO 5
2356#define EXT4_FT_SOCK 6
2357#define EXT4_FT_SYMLINK 7
ac27a0ec 2358
617ba13b 2359#define EXT4_FT_MAX 8
ac27a0ec 2360
e6153918
DW
2361#define EXT4_FT_DIR_CSUM 0xDE
2362
ac27a0ec 2363/*
617ba13b 2364 * EXT4_DIR_PAD defines the directory entries boundaries
ac27a0ec
DK
2365 *
2366 * NOTE: It must be a multiple of 4
2367 */
617ba13b
MC
2368#define EXT4_DIR_PAD 4
2369#define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
a72d7f83
JK
2370#define EXT4_MAX_REC_LEN ((1<<16)-1)
2371
471fbbea
DR
2372/*
2373 * The rec_len is dependent on the type of directory. Directories that are
2374 * casefolded and encrypted need to store the hash as well, so we add room for
2375 * ext4_extended_dir_entry_2. For all entries related to '.' or '..' you should
2376 * pass NULL for dir, as those entries do not use the extra fields.
2377 */
2378static inline unsigned int ext4_dir_rec_len(__u8 name_len,
2379 const struct inode *dir)
2380{
2381 int rec_len = (name_len + 8 + EXT4_DIR_ROUND);
2382
2383 if (dir && ext4_hash_in_dirent(dir))
2384 rec_len += sizeof(struct ext4_dir_entry_hash);
2385 return (rec_len & ~EXT4_DIR_ROUND);
2386}
2387
0cfc9255
ES
2388/*
2389 * If we ever get support for fs block sizes > page_size, we'll need
2390 * to remove the #if statements in the next two functions...
2391 */
2392static inline unsigned int
2393ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
2394{
2395 unsigned len = le16_to_cpu(dlen);
2396
ea1754a0 2397#if (PAGE_SIZE >= 65536)
0cfc9255
ES
2398 if (len == EXT4_MAX_REC_LEN || len == 0)
2399 return blocksize;
2400 return (len & 65532) | ((len & 3) << 16);
2401#else
2402 return len;
2403#endif
2404}
2405
2406static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
2407{
a2e3965d 2408 BUG_ON((len > blocksize) || (blocksize > (1 << 18)) || (len & 3));
ea1754a0 2409#if (PAGE_SIZE >= 65536)
0cfc9255
ES
2410 if (len < 65536)
2411 return cpu_to_le16(len);
2412 if (len == blocksize) {
2413 if (blocksize == 65536)
2414 return cpu_to_le16(EXT4_MAX_REC_LEN);
2415 else
2416 return cpu_to_le16(0);
2417 }
2418 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
2419#else
2420 return cpu_to_le16(len);
2421#endif
2422}
2423
ac27a0ec
DK
2424/*
2425 * Hash Tree Directory indexing
2426 * (c) Daniel Phillips, 2001
2427 */
2428
e2b911c5 2429#define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
12e9b892 2430 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
c7414892
AD
2431#define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \
2432 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir)))
617ba13b 2433#define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
ac27a0ec
DK
2434
2435/* Legal values for the dx_root hash_version field: */
2436
e7093f0d
PK
2437#define DX_HASH_LEGACY 0
2438#define DX_HASH_HALF_MD4 1
2439#define DX_HASH_TEA 2
2440#define DX_HASH_LEGACY_UNSIGNED 3
f99b2589
TT
2441#define DX_HASH_HALF_MD4_UNSIGNED 4
2442#define DX_HASH_TEA_UNSIGNED 5
471fbbea 2443#define DX_HASH_SIPHASH 6
ac27a0ec 2444
0441984a
DW
2445static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2446 const void *address, unsigned int length)
2447{
2448 struct {
2449 struct shash_desc shash;
3108b54b 2450 char ctx[4];
0441984a 2451 } desc;
0441984a 2452
3108b54b
TT
2453 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2454
0441984a 2455 desc.shash.tfm = sbi->s_chksum_driver;
0441984a
DW
2456 *(u32 *)desc.ctx = crc;
2457
e7093f0d 2458 BUG_ON(crypto_shash_update(&desc.shash, address, length));
0441984a
DW
2459
2460 return *(u32 *)desc.ctx;
2461}
2462
ac27a0ec
DK
2463#ifdef __KERNEL__
2464
2465/* hash info structure used by the directory hash */
2466struct dx_hash_info
2467{
2468 u32 hash;
2469 u32 minor_hash;
2470 int hash_version;
2471 u32 *seed;
2472};
2473
d1f5273e
FY
2474
2475/* 32 and 64 bit signed EOF for dx directories */
2476#define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
2477#define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
2478
ac27a0ec
DK
2479
2480/*
617ba13b 2481 * Control parameters used by ext4_htree_next_block
ac27a0ec
DK
2482 */
2483#define HASH_NB_ALWAYS 1
2484
5b643f9c
TT
2485struct ext4_filename {
2486 const struct qstr *usr_fname;
a7550b30 2487 struct fscrypt_str disk_name;
5b643f9c 2488 struct dx_hash_info hinfo;
643fa961 2489#ifdef CONFIG_FS_ENCRYPTION
a7550b30 2490 struct fscrypt_str crypto_buf;
5b643f9c 2491#endif
5298d4bf 2492#if IS_ENABLED(CONFIG_UNICODE)
3ae72562
GKB
2493 struct fscrypt_str cf_name;
2494#endif
5b643f9c
TT
2495};
2496
2497#define fname_name(p) ((p)->disk_name.name)
471fbbea 2498#define fname_usr_name(p) ((p)->usr_fname->name)
5b643f9c 2499#define fname_len(p) ((p)->disk_name.len)
ac27a0ec
DK
2500
2501/*
2502 * Describe an inode's exact location on disk and in memory
2503 */
617ba13b 2504struct ext4_iloc
ac27a0ec
DK
2505{
2506 struct buffer_head *bh;
2507 unsigned long offset;
fd2d4291 2508 ext4_group_t block_group;
ac27a0ec
DK
2509};
2510
617ba13b 2511static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
ac27a0ec 2512{
617ba13b 2513 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
ac27a0ec
DK
2514}
2515
dec214d0
TE
2516static inline bool ext4_is_quota_file(struct inode *inode)
2517{
2518 return IS_NOQUOTA(inode) &&
2519 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL);
2520}
02749a4c 2521
ac27a0ec
DK
2522/*
2523 * This structure is stuffed into the struct file's private_data field
2524 * for directories. It is where we put information so that we can do
2525 * readdir operations in hash tree order.
2526 */
2527struct dir_private_info {
2528 struct rb_root root;
2529 struct rb_node *curr_node;
2530 struct fname *extra_fname;
2531 loff_t last_pos;
2532 __u32 curr_hash;
2533 __u32 curr_minor_hash;
2534 __u32 next_hash;
2535};
2536
2537/* calculate the first block number of the group */
617ba13b 2538static inline ext4_fsblk_t
fd2d4291 2539ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
ac27a0ec 2540{
617ba13b
MC
2541 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2542 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
ac27a0ec
DK
2543}
2544
2545/*
2546 * Special error return code only used by dx_probe() and its callers.
2547 */
36de9286 2548#define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
ac27a0ec 2549
e08ac99f
AB
2550/* htree levels for ext4 */
2551#define EXT4_HTREE_LEVEL_COMPAT 2
2552#define EXT4_HTREE_LEVEL 3
2553
2554static inline int ext4_dir_htree_level(struct super_block *sb)
2555{
2556 return ext4_has_feature_largedir(sb) ?
2557 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT;
2558}
2559
bfff6873
LC
2560/*
2561 * Timeout and state flag for lazy initialization inode thread.
2562 */
2563#define EXT4_DEF_LI_WAIT_MULT 10
2564#define EXT4_DEF_LI_MAX_START_DELAY 5
2565#define EXT4_LAZYINIT_QUIT 0x0001
2566#define EXT4_LAZYINIT_RUNNING 0x0002
2567
2568/*
2569 * Lazy inode table initialization info
2570 */
2571struct ext4_lazy_init {
2572 unsigned long li_state;
bfff6873
LC
2573 struct list_head li_request_list;
2574 struct mutex li_list_mtx;
2575};
2576
3d392b26
TT
2577enum ext4_li_mode {
2578 EXT4_LI_MODE_PREFETCH_BBITMAP,
2579 EXT4_LI_MODE_ITABLE,
2580};
2581
bfff6873
LC
2582struct ext4_li_request {
2583 struct super_block *lr_super;
3d392b26
TT
2584 enum ext4_li_mode lr_mode;
2585 ext4_group_t lr_first_not_zeroed;
bfff6873
LC
2586 ext4_group_t lr_next_group;
2587 struct list_head lr_request;
2588 unsigned long lr_next_sched;
2589 unsigned long lr_timeout;
2590};
2591
857ac889
LC
2592struct ext4_features {
2593 struct kobject f_kobj;
2594 struct completion f_kobj_unregister;
2595};
2596
c5e06d10
JL
2597/*
2598 * This structure will be used for multiple mount protection. It will be
2599 * written into the block number saved in the s_mmp_block field in the
2600 * superblock. Programs that check MMP should assume that if
2601 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2602 * to use the filesystem, regardless of how old the timestamp is.
2603 */
2604#define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
2605#define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2606#define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
2607#define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
2608
2609struct mmp_struct {
2610 __le32 mmp_magic; /* Magic number for MMP */
2611 __le32 mmp_seq; /* Sequence no. updated periodically */
2612
2613 /*
2614 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2615 * purposes and do not affect the correctness of the algorithm
2616 */
2617 __le64 mmp_time; /* Time last updated */
2618 char mmp_nodename[64]; /* Node which last updated MMP block */
2619 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
2620
2621 /*
2622 * mmp_check_interval is used to verify if the MMP block has been
2623 * updated on the block device. The value is updated based on the
2624 * maximum time to write the MMP block during an update cycle.
2625 */
2626 __le16 mmp_check_interval;
2627
2628 __le16 mmp_pad1;
e6153918
DW
2629 __le32 mmp_pad2[226];
2630 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
c5e06d10
JL
2631};
2632
2633/* arguments passed to the mmp thread */
2634struct mmpd_data {
2635 struct buffer_head *bh; /* bh from initial read_mmp_block() */
2636 struct super_block *sb; /* super block of the fs */
2637};
2638
2639/*
2640 * Check interval multiplier
2641 * The MMP block is written every update interval and initially checked every
2642 * update interval x the multiplier (the value is then adapted based on the
2643 * write latency). The reason is that writes can be delayed under load and we
2644 * don't want readers to incorrectly assume that the filesystem is no longer
2645 * in use.
2646 */
2647#define EXT4_MMP_CHECK_MULT 2UL
2648
2649/*
2650 * Minimum interval for MMP checking in seconds.
2651 */
2652#define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
2653
2654/*
2655 * Maximum interval for MMP checking in seconds.
2656 */
2657#define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
2658
ac27a0ec
DK
2659/*
2660 * Function prototypes
2661 */
2662
2663/*
2664 * Ok, these declarations are also in <linux/kernel.h> but none of the
617ba13b 2665 * ext4 source programs needs to include it so they are duplicated here.
ac27a0ec 2666 */
8c55e204
DK
2667# define NORET_TYPE /**/
2668# define ATTRIB_NORET __attribute__((noreturn))
2669# define NORET_AND noreturn,
ac27a0ec 2670
498e5f24 2671/* bitmap.c */
f6fb99ca 2672extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
41a246d1
DW
2673void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2674 struct ext4_group_desc *gdp,
2675 struct buffer_head *bh, int sz);
2676int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2677 struct ext4_group_desc *gdp,
2678 struct buffer_head *bh, int sz);
fa77dcfa
DW
2679void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2680 struct ext4_group_desc *gdp,
79f1ba49 2681 struct buffer_head *bh);
fa77dcfa
DW
2682int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2683 struct ext4_group_desc *gdp,
79f1ba49 2684 struct buffer_head *bh);
498e5f24 2685
ac27a0ec 2686/* balloc.c */
bd86298e
LC
2687extern void ext4_get_group_no_and_offset(struct super_block *sb,
2688 ext4_fsblk_t blocknr,
2689 ext4_group_t *blockgrpp,
2690 ext4_grpblk_t *offsetp);
2691extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2692 ext4_fsblk_t block);
2693
3a5b2ecd
MC
2694extern unsigned int ext4_block_group(struct super_block *sb,
2695 ext4_fsblk_t blocknr);
2696extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2697 ext4_fsblk_t blocknr);
fd2d4291
AM
2698extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2699extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2700 ext4_group_t group);
7061eba7 2701extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
55f020db
AH
2702 ext4_fsblk_t goal,
2703 unsigned int flags,
2704 unsigned long *count,
2705 int *errp);
e7d5f315
TT
2706extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2707 s64 nclusters, unsigned int flags);
5dee5437 2708extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
af5bc92d 2709extern void ext4_check_blocks_bitmap(struct super_block *);
617ba13b 2710extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
fd2d4291 2711 ext4_group_t block_group,
ac27a0ec 2712 struct buffer_head ** bh);
617ba13b 2713extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
813e5727
TT
2714
2715extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
cfd73237
AZ
2716 ext4_group_t block_group,
2717 bool ignore_locked);
813e5727
TT
2718extern int ext4_wait_block_bitmap(struct super_block *sb,
2719 ext4_group_t block_group,
2720 struct buffer_head *bh);
2721extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2722 ext4_group_t block_group);
cff1dfd7
TT
2723extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2724 ext4_group_t block_group,
2725 struct ext4_group_desc *gdp);
f86186b4 2726ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
ac27a0ec 2727
5298d4bf 2728#if IS_ENABLED(CONFIG_UNICODE)
1ae98e29 2729extern int ext4_fname_setup_ci_filename(struct inode *dir,
3ae72562 2730 const struct qstr *iname,
1ae98e29 2731 struct ext4_filename *fname);
3ae72562
GKB
2732#endif
2733
643fa961 2734#ifdef CONFIG_FS_ENCRYPTION
b01531db
EB
2735static inline void ext4_fname_from_fscrypt_name(struct ext4_filename *dst,
2736 const struct fscrypt_name *src)
2737{
2738 memset(dst, 0, sizeof(*dst));
2739
2740 dst->usr_fname = src->usr_fname;
2741 dst->disk_name = src->disk_name;
2742 dst->hinfo.hash = src->hash;
2743 dst->hinfo.minor_hash = src->minor_hash;
2744 dst->crypto_buf = src->crypto_buf;
2745}
2746
5b643f9c 2747static inline int ext4_fname_setup_filename(struct inode *dir,
b01531db
EB
2748 const struct qstr *iname,
2749 int lookup,
2750 struct ext4_filename *fname)
5b643f9c 2751{
a7550b30
JK
2752 struct fscrypt_name name;
2753 int err;
d5d0e8c7 2754
a7550b30 2755 err = fscrypt_setup_filename(dir, iname, lookup, &name);
b01531db
EB
2756 if (err)
2757 return err;
b7236e21 2758
b01531db 2759 ext4_fname_from_fscrypt_name(fname, &name);
3ae72562 2760
5298d4bf 2761#if IS_ENABLED(CONFIG_UNICODE)
1ae98e29 2762 err = ext4_fname_setup_ci_filename(dir, iname, fname);
3ae72562 2763#endif
1ae98e29 2764 return err;
b01531db
EB
2765}
2766
2767static inline int ext4_fname_prepare_lookup(struct inode *dir,
2768 struct dentry *dentry,
2769 struct ext4_filename *fname)
2770{
2771 struct fscrypt_name name;
2772 int err;
2773
2774 err = fscrypt_prepare_lookup(dir, dentry, &name);
2775 if (err)
2776 return err;
2777
2778 ext4_fname_from_fscrypt_name(fname, &name);
3ae72562 2779
5298d4bf 2780#if IS_ENABLED(CONFIG_UNICODE)
1ae98e29 2781 err = ext4_fname_setup_ci_filename(dir, &dentry->d_name, fname);
3ae72562 2782#endif
1ae98e29 2783 return err;
b7236e21
TT
2784}
2785
a7550b30 2786static inline void ext4_fname_free_filename(struct ext4_filename *fname)
e709e9df 2787{
a7550b30 2788 struct fscrypt_name name;
e709e9df 2789
a7550b30
JK
2790 name.crypto_buf = fname->crypto_buf;
2791 fscrypt_free_filename(&name);
2792
2793 fname->crypto_buf.name = NULL;
2794 fname->usr_fname = NULL;
2795 fname->disk_name.name = NULL;
3ae72562 2796
5298d4bf 2797#if IS_ENABLED(CONFIG_UNICODE)
3ae72562
GKB
2798 kfree(fname->cf_name.name);
2799 fname->cf_name.name = NULL;
2800#endif
88bd6ccd 2801}
b01531db 2802#else /* !CONFIG_FS_ENCRYPTION */
a7550b30 2803static inline int ext4_fname_setup_filename(struct inode *dir,
b01531db
EB
2804 const struct qstr *iname,
2805 int lookup,
2806 struct ext4_filename *fname)
b7236e21 2807{
1ae98e29 2808 int err = 0;
a7550b30
JK
2809 fname->usr_fname = iname;
2810 fname->disk_name.name = (unsigned char *) iname->name;
2811 fname->disk_name.len = iname->len;
3ae72562 2812
5298d4bf 2813#if IS_ENABLED(CONFIG_UNICODE)
1ae98e29 2814 err = ext4_fname_setup_ci_filename(dir, iname, fname);
3ae72562
GKB
2815#endif
2816
1ae98e29 2817 return err;
b7236e21 2818}
88bd6ccd 2819
b01531db
EB
2820static inline int ext4_fname_prepare_lookup(struct inode *dir,
2821 struct dentry *dentry,
2822 struct ext4_filename *fname)
2823{
2824 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname);
2825}
2826
3ae72562
GKB
2827static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2828{
5298d4bf 2829#if IS_ENABLED(CONFIG_UNICODE)
3ae72562
GKB
2830 kfree(fname->cf_name.name);
2831 fname->cf_name.name = NULL;
2832#endif
2833}
b01531db 2834#endif /* !CONFIG_FS_ENCRYPTION */
88bd6ccd 2835
ac27a0ec 2836/* dir.c */
60fd4da3 2837extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
f7c21177 2838 struct file *,
60fd4da3 2839 struct ext4_dir_entry_2 *,
226ba972
TM
2840 struct buffer_head *, char *, int,
2841 unsigned int);
471fbbea 2842#define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
f7c21177 2843 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
471fbbea 2844 (de), (bh), (buf), (size), (offset)))
617ba13b 2845extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2f61830a
TT
2846 __u32 minor_hash,
2847 struct ext4_dir_entry_2 *dirent,
a7550b30 2848 struct fscrypt_str *ent_name);
617ba13b 2849extern void ext4_htree_free_dir_info(struct dir_private_info *p);
978fef91
TM
2850extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2851 struct buffer_head *bh,
2852 void *buf, int buf_size,
5b643f9c 2853 struct ext4_filename *fname,
978fef91 2854 struct ext4_dir_entry_2 **dest_de);
471fbbea 2855void ext4_insert_dentry(struct inode *dir, struct inode *inode,
1bc0af60
EB
2856 struct ext4_dir_entry_2 *de,
2857 int buf_size,
2858 struct ext4_filename *fname);
978fef91
TM
2859static inline void ext4_update_dx_flag(struct inode *inode)
2860{
f902b216
JK
2861 if (!ext4_has_feature_dir_index(inode->i_sb) &&
2862 ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) {
48a34311
JK
2863 /* ext4_iget() should have caught this... */
2864 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb));
978fef91 2865 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
48a34311 2866 }
978fef91 2867}
d6006186 2868static const unsigned char ext4_filetype_table[] = {
65d165d9
TM
2869 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2870};
2871
2872static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2873{
e2b911c5 2874 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
65d165d9
TM
2875 return DT_UNKNOWN;
2876
2877 return ext4_filetype_table[filetype];
2878}
40b163f1
DW
2879extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2880 void *buf, int buf_size);
ac27a0ec
DK
2881
2882/* fsync.c */
02c24a82 2883extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
ac27a0ec
DK
2884
2885/* hash.c */
b886ee3e
GKB
2886extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len,
2887 struct dx_hash_info *hinfo);
ac27a0ec
DK
2888
2889/* ialloc.c */
8016e29f 2890extern int ext4_mark_inode_used(struct super_block *sb, int ino);
14f3db55
CB
2891extern struct inode *__ext4_new_inode(struct user_namespace *, handle_t *,
2892 struct inode *, umode_t,
1139575a 2893 const struct qstr *qstr, __u32 goal,
1b917ed8
TE
2894 uid_t *owner, __u32 i_flags,
2895 int handle_type, unsigned int line_no,
2896 int nblocks);
1139575a 2897
14f3db55
CB
2898#define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \
2899 __ext4_new_inode(&init_user_ns, (handle), (dir), (mode), (qstr), \
2900 (goal), (owner), i_flags, 0, 0, 0)
2901#define ext4_new_inode_start_handle(mnt_userns, dir, mode, qstr, goal, owner, \
1139575a 2902 type, nblocks) \
14f3db55 2903 __ext4_new_inode((mnt_userns), NULL, (dir), (mode), (qstr), (goal), (owner), \
1b917ed8 2904 0, (type), __LINE__, (nblocks))
1139575a
TT
2905
2906
af5bc92d
TT
2907extern void ext4_free_inode(handle_t *, struct inode *);
2908extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2909extern unsigned long ext4_count_free_inodes(struct super_block *);
2910extern unsigned long ext4_count_dirs(struct super_block *);
2911extern void ext4_check_inodes_bitmap(struct super_block *);
61d08673 2912extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
bfff6873
LC
2913extern int ext4_init_inode_table(struct super_block *sb,
2914 ext4_group_t group, int barrier);
813e5727 2915extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
ac27a0ec 2916
6866d7b3 2917/* fast_commit.c */
ce8c59d1 2918int ext4_fc_info_show(struct seq_file *seq, void *v);
6866d7b3 2919void ext4_fc_init(struct super_block *sb, journal_t *journal);
aa75f4d3 2920void ext4_fc_init_inode(struct inode *inode);
a80f7fcf 2921void ext4_fc_track_range(handle_t *handle, struct inode *inode, ext4_lblk_t start,
aa75f4d3 2922 ext4_lblk_t end);
a80f7fcf
HS
2923void __ext4_fc_track_unlink(handle_t *handle, struct inode *inode,
2924 struct dentry *dentry);
2925void __ext4_fc_track_link(handle_t *handle, struct inode *inode,
2926 struct dentry *dentry);
2927void ext4_fc_track_unlink(handle_t *handle, struct dentry *dentry);
2928void ext4_fc_track_link(handle_t *handle, struct dentry *dentry);
8210bb29
HS
2929void __ext4_fc_track_create(handle_t *handle, struct inode *inode,
2930 struct dentry *dentry);
a80f7fcf
HS
2931void ext4_fc_track_create(handle_t *handle, struct dentry *dentry);
2932void ext4_fc_track_inode(handle_t *handle, struct inode *inode);
e85c81ba 2933void ext4_fc_mark_ineligible(struct super_block *sb, int reason, handle_t *handle);
aa75f4d3
HS
2934void ext4_fc_start_update(struct inode *inode);
2935void ext4_fc_stop_update(struct inode *inode);
2936void ext4_fc_del(struct inode *inode);
8016e29f
HS
2937bool ext4_fc_replay_check_excluded(struct super_block *sb, ext4_fsblk_t block);
2938void ext4_fc_replay_cleanup(struct super_block *sb);
aa75f4d3
HS
2939int ext4_fc_commit(journal_t *journal, tid_t commit_tid);
2940int __init ext4_fc_init_dentry_cache(void);
ab047d51 2941void ext4_fc_destroy_dentry_cache(void);
599ea31d
XY
2942int ext4_fc_record_regions(struct super_block *sb, int ino,
2943 ext4_lblk_t lblk, ext4_fsblk_t pblk,
2944 int len, int replay);
aa75f4d3 2945
c9de560d 2946/* mballoc.c */
247dbed8 2947extern const struct seq_operations ext4_mb_seq_groups_ops;
f68f4063 2948extern const struct seq_operations ext4_mb_seq_structs_summary_ops;
c9de560d
AT
2949extern long ext4_mb_stats;
2950extern long ext4_mb_max_to_scan;
a6c75eaf 2951extern int ext4_seq_mb_stats_show(struct seq_file *seq, void *offset);
9d99012f 2952extern int ext4_mb_init(struct super_block *);
c9de560d
AT
2953extern int ext4_mb_release(struct super_block *);
2954extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2955 struct ext4_allocation_request *, int *);
2956extern int ext4_mb_reserve_blocks(struct super_block *, int);
27bc446e 2957extern void ext4_discard_preallocations(struct inode *, unsigned int);
5dabfc78
TT
2958extern int __init ext4_init_mballoc(void);
2959extern void ext4_exit_mballoc(void);
3d392b26
TT
2960extern ext4_group_t ext4_mb_prefetch(struct super_block *sb,
2961 ext4_group_t group,
2962 unsigned int nr, int *cnt);
2963extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group,
2964 unsigned int nr);
2965
44338711 2966extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
e6362609
TT
2967 struct buffer_head *bh, ext4_fsblk_t block,
2968 unsigned long count, int flags);
28623c2f
TT
2969extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2970 ext4_group_t ngroups);
920313a7 2971extern int ext4_mb_add_groupinfo(struct super_block *sb,
5f21b0e6 2972 ext4_group_t i, struct ext4_group_desc *desc);
cc7365df 2973extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2846e820 2974 ext4_fsblk_t block, unsigned long count);
7360d173 2975extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
a0154344 2976extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid);
8016e29f
HS
2977extern void ext4_mb_mark_bb(struct super_block *sb, ext4_fsblk_t block,
2978 int len, int state);
7360d173 2979
ac27a0ec 2980/* inode.c */
8016e29f
HS
2981void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw,
2982 struct ext4_inode_info *ei);
f348c252 2983int ext4_inode_is_fast_symlink(struct inode *inode);
10560082 2984struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
1c215028 2985struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
9699d4f9
TE
2986int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count,
2987 bool wait, struct buffer_head **bhs);
705965bd
JK
2988int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2989 struct buffer_head *bh_result, int create);
6873fa0d 2990int ext4_get_block(struct inode *inode, sector_t iblock,
705965bd 2991 struct buffer_head *bh_result, int create);
9c3569b5
TM
2992int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2993 struct buffer_head *bh, int create);
f19d5870 2994int ext4_walk_page_buffers(handle_t *handle,
188c299e 2995 struct inode *inode,
f19d5870
TM
2996 struct buffer_head *head,
2997 unsigned from,
2998 unsigned to,
2999 int *partial,
188c299e 3000 int (*fn)(handle_t *handle, struct inode *inode,
f19d5870 3001 struct buffer_head *bh));
188c299e 3002int do_journal_get_write_access(handle_t *handle, struct inode *inode,
f19d5870 3003 struct buffer_head *bh);
9c3569b5
TM
3004#define FALL_BACK_TO_NONDELALLOC 1
3005#define CONVERT_INLINE_DATA 2
ac27a0ec 3006
8a363970
TT
3007typedef enum {
3008 EXT4_IGET_NORMAL = 0,
3009 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */
3010 EXT4_IGET_HANDLE = 0x0002 /* Inode # is from a handle */
3011} ext4_iget_flags;
3012
3013extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
3014 ext4_iget_flags flags, const char *function,
3015 unsigned int line);
3016
3017#define ext4_iget(sb, ino, flags) \
3018 __ext4_iget((sb), (ino), (flags), __func__, __LINE__)
3019
a9185b41 3020extern int ext4_write_inode(struct inode *, struct writeback_control *);
549c7297
CB
3021extern int ext4_setattr(struct user_namespace *, struct dentry *,
3022 struct iattr *);
3023extern int ext4_getattr(struct user_namespace *, const struct path *,
3024 struct kstat *, u32, unsigned int);
0930fcc1
AV
3025extern void ext4_evict_inode(struct inode *);
3026extern void ext4_clear_inode(struct inode *);
549c7297
CB
3027extern int ext4_file_getattr(struct user_namespace *, const struct path *,
3028 struct kstat *, u32, unsigned int);
af5bc92d 3029extern int ext4_sync_inode(handle_t *, struct inode *);
aa385729 3030extern void ext4_dirty_inode(struct inode *, int);
617ba13b
MC
3031extern int ext4_change_inode_journal_flag(struct inode *, int);
3032extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
8016e29f
HS
3033extern int ext4_get_fc_inode_loc(struct super_block *sb, unsigned long ino,
3034 struct ext4_iloc *iloc);
a361293f 3035extern int ext4_inode_attach_jinode(struct inode *inode);
91ef4caf 3036extern int ext4_can_truncate(struct inode *inode);
2c98eb5e 3037extern int ext4_truncate(struct inode *);
430657b6 3038extern int ext4_break_layouts(struct inode *);
ad5cd4f4 3039extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length);
043546e4 3040extern void ext4_set_inode_flags(struct inode *, bool init);
ccd2506b 3041extern int ext4_alloc_da_blocks(struct inode *inode);
617ba13b 3042extern void ext4_set_aops(struct inode *inode);
a86c6181 3043extern int ext4_writepage_trans_blocks(struct inode *);
f3bd1f3f 3044extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
a87dd18c
LC
3045extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
3046 loff_t lstart, loff_t lend);
401b25aa 3047extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf);
a9e7f447 3048extern qsize_t *ext4_get_reserved_space(struct inode *inode);
040cb378 3049extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
f456767d 3050extern void ext4_da_release_space(struct inode *inode, int to_free);
5f634d06
AK
3051extern void ext4_da_update_reserve_space(struct inode *inode,
3052 int used, int quota_claim);
53085fac
JK
3053extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
3054 ext4_fsblk_t pblk, ext4_lblk_t len);
dae1e52c
AG
3055
3056/* indirect.c */
3057extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
3058 struct ext4_map_blocks *map, int flags);
fa55a0ed 3059extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
819c4920 3060extern void ext4_ind_truncate(handle_t *, struct inode *inode);
4f579ae7
LC
3061extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
3062 ext4_lblk_t start, ext4_lblk_t end);
ff9893dc 3063
ac27a0ec 3064/* ioctl.c */
5cdd7b2d 3065extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
af5bc92d 3066extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
4db5c2e6
MS
3067int ext4_fileattr_set(struct user_namespace *mnt_userns,
3068 struct dentry *dentry, struct fileattr *fa);
3069int ext4_fileattr_get(struct dentry *dentry, struct fileattr *fa);
8016e29f 3070extern void ext4_reset_inode_seed(struct inode *inode);
eb705421 3071int ext4_update_overhead(struct super_block *sb);
ac27a0ec 3072
c14c6fd5 3073/* migrate.c */
2a43a878 3074extern int ext4_ext_migrate(struct inode *);
0d14b098 3075extern int ext4_ind_migrate(struct inode *inode);
3d0518f4 3076
ac27a0ec 3077/* namei.c */
8016e29f
HS
3078extern int ext4_init_new_dir(handle_t *handle, struct inode *dir,
3079 struct inode *inode);
f036adb3
TT
3080extern int ext4_dirblock_csum_verify(struct inode *inode,
3081 struct buffer_head *bh);
617ba13b 3082extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
ac27a0ec 3083 __u32 start_minor_hash, __u32 *next_hash);
5b643f9c
TT
3084extern int ext4_search_dir(struct buffer_head *bh,
3085 char *search_buf,
3086 int buf_size,
3087 struct inode *dir,
3088 struct ext4_filename *fname,
5b643f9c
TT
3089 unsigned int offset,
3090 struct ext4_dir_entry_2 **res_dir);
2fe34d29 3091extern int ext4_generic_delete_entry(struct inode *dir,
05019a9e
TM
3092 struct ext4_dir_entry_2 *de_del,
3093 struct buffer_head *bh,
3094 void *entry_buf,
3095 int buf_size,
3096 int csum_size);
a7550b30 3097extern bool ext4_empty_dir(struct inode *inode);
ac27a0ec
DK
3098
3099/* resize.c */
1d0c3924 3100extern void ext4_kvfree_array_rcu(void *to_free);
617ba13b
MC
3101extern int ext4_group_add(struct super_block *sb,
3102 struct ext4_new_group_data *input);
3103extern int ext4_group_extend(struct super_block *sb,
3104 struct ext4_super_block *es,
3105 ext4_fsblk_t n_blocks_count);
19c5246d 3106extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
bbc605cd
LC
3107extern unsigned int ext4_list_backups(struct super_block *sb,
3108 unsigned int *three, unsigned int *five,
3109 unsigned int *seven);
ac27a0ec
DK
3110
3111/* super.c */
fb265c9c
TT
3112extern struct buffer_head *ext4_sb_bread(struct super_block *sb,
3113 sector_t block, int op_flags);
8394a6ab 3114extern struct buffer_head *ext4_sb_bread_unmovable(struct super_block *sb,
3115 sector_t block);
fa491b14 3116extern void ext4_read_bh_nowait(struct buffer_head *bh, int op_flags,
3117 bh_end_io_t *end_io);
3118extern int ext4_read_bh(struct buffer_head *bh, int op_flags,
3119 bh_end_io_t *end_io);
3120extern int ext4_read_bh_lock(struct buffer_head *bh, int op_flags, bool wait);
5df1d412 3121extern void ext4_sb_breadahead_unmovable(struct super_block *sb, sector_t block);
ebd173be 3122extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
952fc18e 3123extern int ext4_calculate_overhead(struct super_block *sb);
bbc605cd
LC
3124extern __le32 ext4_superblock_csum(struct super_block *sb,
3125 struct ext4_super_block *es);
06db49e6 3126extern void ext4_superblock_csum_set(struct super_block *sb);
117fff10
TT
3127extern int ext4_alloc_flex_bg_array(struct super_block *sb,
3128 ext4_group_t ngroup);
722887dd
TT
3129extern const char *ext4_decode_error(struct super_block *sb, int errno,
3130 char nbuf[16]);
db79e6d1
WS
3131extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb,
3132 ext4_group_t block_group,
3133 unsigned int flags);
e7c96e8e 3134
014c9caa
JK
3135extern __printf(7, 8)
3136void __ext4_error(struct super_block *, const char *, unsigned int, bool,
3137 int, __u64, const char *, ...);
54d3adbc
TT
3138extern __printf(6, 7)
3139void __ext4_error_inode(struct inode *, const char *, unsigned int,
3140 ext4_fsblk_t, int, const char *, ...);
b9075fa9 3141extern __printf(5, 6)
e7c96e8e 3142void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
b9075fa9 3143 const char *, ...);
c398eda0
TT
3144extern void __ext4_std_error(struct super_block *, const char *,
3145 unsigned int, int);
b9075fa9
JP
3146extern __printf(4, 5)
3147void __ext4_warning(struct super_block *, const char *, unsigned int,
3148 const char *, ...);
b03a2f7e
AD
3149extern __printf(4, 5)
3150void __ext4_warning_inode(const struct inode *inode, const char *function,
3151 unsigned int line, const char *fmt, ...);
b9075fa9 3152extern __printf(3, 4)
e7c96e8e 3153void __ext4_msg(struct super_block *, const char *, const char *, ...);
c5e06d10
JL
3154extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
3155 const char *, unsigned int, const char *);
b9075fa9
JP
3156extern __printf(7, 8)
3157void __ext4_grp_locked_error(const char *, unsigned int,
3158 struct super_block *, ext4_group_t,
3159 unsigned long, ext4_fsblk_t,
3160 const char *, ...);
e7c96e8e 3161
b03a2f7e
AD
3162#define EXT4_ERROR_INODE(inode, fmt, a...) \
3163 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
3164
54d3adbc
TT
3165#define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...) \
3166 __ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a)
3167
3168#define ext4_error_inode_block(inode, block, err, fmt, a...) \
3169 __ext4_error_inode((inode), __func__, __LINE__, (block), (err), \
3170 (fmt), ## a)
b03a2f7e
AD
3171
3172#define EXT4_ERROR_FILE(file, block, fmt, a...) \
3173 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
3174
014c9caa
JK
3175#define ext4_abort(sb, err, fmt, a...) \
3176 __ext4_error((sb), __func__, __LINE__, true, (err), 0, (fmt), ## a)
3177
e7c96e8e
JP
3178#ifdef CONFIG_PRINTK
3179
3180#define ext4_error_inode(inode, func, line, block, fmt, ...) \
54d3adbc
TT
3181 __ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__)
3182#define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \
3183 __ext4_error_inode((inode), (func), (line), (block), \
3184 (err), (fmt), ##__VA_ARGS__)
e7c96e8e
JP
3185#define ext4_error_file(file, func, line, block, fmt, ...) \
3186 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
3187#define ext4_error(sb, fmt, ...) \
014c9caa
JK
3188 __ext4_error((sb), __func__, __LINE__, false, 0, 0, (fmt), \
3189 ##__VA_ARGS__)
54d3adbc 3190#define ext4_error_err(sb, err, fmt, ...) \
014c9caa
JK
3191 __ext4_error((sb), __func__, __LINE__, false, (err), 0, (fmt), \
3192 ##__VA_ARGS__)
e7c96e8e
JP
3193#define ext4_warning(sb, fmt, ...) \
3194 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
b03a2f7e
AD
3195#define ext4_warning_inode(inode, fmt, ...) \
3196 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
e7c96e8e
JP
3197#define ext4_msg(sb, level, fmt, ...) \
3198 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
3199#define dump_mmp_msg(sb, mmp, msg) \
3200 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
3201#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
3202 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
3203 fmt, ##__VA_ARGS__)
3204
3205#else
3206
3207#define ext4_error_inode(inode, func, line, block, fmt, ...) \
3208do { \
3209 no_printk(fmt, ##__VA_ARGS__); \
54d3adbc
TT
3210 __ext4_error_inode(inode, "", 0, block, 0, " "); \
3211} while (0)
3212#define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \
3213do { \
3214 no_printk(fmt, ##__VA_ARGS__); \
3215 __ext4_error_inode(inode, "", 0, block, err, " "); \
e7c96e8e
JP
3216} while (0)
3217#define ext4_error_file(file, func, line, block, fmt, ...) \
3218do { \
3219 no_printk(fmt, ##__VA_ARGS__); \
3220 __ext4_error_file(file, "", 0, block, " "); \
3221} while (0)
3222#define ext4_error(sb, fmt, ...) \
3223do { \
3224 no_printk(fmt, ##__VA_ARGS__); \
014c9caa 3225 __ext4_error(sb, "", 0, false, 0, 0, " "); \
54d3adbc
TT
3226} while (0)
3227#define ext4_error_err(sb, err, fmt, ...) \
3228do { \
3229 no_printk(fmt, ##__VA_ARGS__); \
014c9caa 3230 __ext4_error(sb, "", 0, false, err, 0, " "); \
e7c96e8e
JP
3231} while (0)
3232#define ext4_warning(sb, fmt, ...) \
3233do { \
3234 no_printk(fmt, ##__VA_ARGS__); \
3235 __ext4_warning(sb, "", 0, " "); \
3236} while (0)
b03a2f7e
AD
3237#define ext4_warning_inode(inode, fmt, ...) \
3238do { \
3239 no_printk(fmt, ##__VA_ARGS__); \
3240 __ext4_warning_inode(inode, "", 0, " "); \
3241} while (0)
e7c96e8e
JP
3242#define ext4_msg(sb, level, fmt, ...) \
3243do { \
3244 no_printk(fmt, ##__VA_ARGS__); \
3245 __ext4_msg(sb, "", " "); \
3246} while (0)
3247#define dump_mmp_msg(sb, mmp, msg) \
3248 __dump_mmp_msg(sb, mmp, "", 0, "")
3249#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
3250do { \
3251 no_printk(fmt, ##__VA_ARGS__); \
3252 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
3253} while (0)
3254
3255#endif
3256
8fadc143
AR
3257extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
3258 struct ext4_group_desc *bg);
3259extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
3260 struct ext4_group_desc *bg);
3261extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
3262 struct ext4_group_desc *bg);
021b65bb
TT
3263extern __u32 ext4_free_group_clusters(struct super_block *sb,
3264 struct ext4_group_desc *bg);
560671a0
AK
3265extern __u32 ext4_free_inodes_count(struct super_block *sb,
3266 struct ext4_group_desc *bg);
3267extern __u32 ext4_used_dirs_count(struct super_block *sb,
3268 struct ext4_group_desc *bg);
3269extern __u32 ext4_itable_unused_count(struct super_block *sb,
3270 struct ext4_group_desc *bg);
8fadc143
AR
3271extern void ext4_block_bitmap_set(struct super_block *sb,
3272 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3273extern void ext4_inode_bitmap_set(struct super_block *sb,
3274 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3275extern void ext4_inode_table_set(struct super_block *sb,
3276 struct ext4_group_desc *bg, ext4_fsblk_t blk);
021b65bb
TT
3277extern void ext4_free_group_clusters_set(struct super_block *sb,
3278 struct ext4_group_desc *bg,
3279 __u32 count);
560671a0
AK
3280extern void ext4_free_inodes_set(struct super_block *sb,
3281 struct ext4_group_desc *bg, __u32 count);
3282extern void ext4_used_dirs_set(struct super_block *sb,
3283 struct ext4_group_desc *bg, __u32 count);
3284extern void ext4_itable_unused_set(struct super_block *sb,
3285 struct ext4_group_desc *bg, __u32 count);
feb0ab32 3286extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
bb23c20a 3287 struct ext4_group_desc *gdp);
feb0ab32
DW
3288extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
3289 struct ext4_group_desc *gdp);
7f511862
TT
3290extern int ext4_register_li_request(struct super_block *sb,
3291 ext4_group_t first_not_zeroed);
bd81d8ee 3292
9aa5d32b
DM
3293static inline int ext4_has_metadata_csum(struct super_block *sb)
3294{
e2b911c5 3295 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
9aa5d32b
DM
3296 !EXT4_SB(sb)->s_chksum_driver);
3297
dec214d0
TE
3298 return ext4_has_feature_metadata_csum(sb) &&
3299 (EXT4_SB(sb)->s_chksum_driver != NULL);
9aa5d32b 3300}
dec214d0
TE
3301
3302static inline int ext4_has_group_desc_csum(struct super_block *sb)
3303{
3304 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb);
3305}
3306
81e8c3c5
PM
3307#define ext4_read_incompat_64bit_val(es, name) \
3308 (((es)->s_feature_incompat & cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT) \
3309 ? (ext4_fsblk_t)le32_to_cpu(es->name##_hi) << 32 : 0) | \
3310 le32_to_cpu(es->name##_lo))
3311
bd81d8ee
LV
3312static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
3313{
81e8c3c5 3314 return ext4_read_incompat_64bit_val(es, s_blocks_count);
bd81d8ee
LV
3315}
3316
3317static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
3318{
81e8c3c5 3319 return ext4_read_incompat_64bit_val(es, s_r_blocks_count);
bd81d8ee
LV
3320}
3321
3322static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
3323{
81e8c3c5 3324 return ext4_read_incompat_64bit_val(es, s_free_blocks_count);
bd81d8ee
LV
3325}
3326
3327static inline void ext4_blocks_count_set(struct ext4_super_block *es,
3328 ext4_fsblk_t blk)
3329{
6bc9feff 3330 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
bd81d8ee
LV
3331 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
3332}
3333
3334static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
3335 ext4_fsblk_t blk)
3336{
308ba3ec 3337 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
bd81d8ee
LV
3338 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
3339}
3340
3341static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
3342 ext4_fsblk_t blk)
3343{
308ba3ec 3344 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
bd81d8ee
LV
3345 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
3346}
3347
e08ac99f
AB
3348static inline loff_t ext4_isize(struct super_block *sb,
3349 struct ext4_inode *raw_inode)
a48380f7 3350{
e08ac99f
AB
3351 if (ext4_has_feature_largedir(sb) ||
3352 S_ISREG(le16_to_cpu(raw_inode->i_mode)))
06a279d6
TT
3353 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
3354 le32_to_cpu(raw_inode->i_size_lo);
e08ac99f
AB
3355
3356 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
a48380f7 3357}
bd81d8ee 3358
a48380f7
AK
3359static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
3360{
3361 raw_inode->i_size_lo = cpu_to_le32(i_size);
3362 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
3363}
ac27a0ec 3364
c9de560d
AT
3365static inline
3366struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
af5bc92d 3367 ext4_group_t group)
c9de560d 3368{
df3da4ea 3369 struct ext4_group_info **grp_info;
c9de560d 3370 long indexv, indexh;
2f2e09eb 3371 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
c9de560d
AT
3372 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
3373 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
df3da4ea
SJS
3374 grp_info = sbi_array_rcu_deref(EXT4_SB(sb), s_group_info, indexv);
3375 return grp_info[indexh];
c9de560d
AT
3376}
3377
8df9675f
TT
3378/*
3379 * Reading s_groups_count requires using smp_rmb() afterwards. See
3380 * the locking protocol documented in the comments of ext4_group_add()
3381 * in resize.c
3382 */
3383static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
3384{
3385 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
3386
3387 smp_rmb();
3388 return ngroups;
3389}
c9de560d 3390
772cb7c8
JS
3391static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
3392 ext4_group_t block_group)
3393{
3394 return block_group >> sbi->s_log_groups_per_flex;
3395}
3396
3397static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
3398{
3399 return 1 << sbi->s_log_groups_per_flex;
3400}
3401
617ba13b 3402#define ext4_std_error(sb, errno) \
ac27a0ec
DK
3403do { \
3404 if ((errno)) \
c398eda0 3405 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
ac27a0ec
DK
3406} while (0)
3407
a30d542a 3408#ifdef CONFIG_SMP
df55c99d
TT
3409/* Each CPU can accumulate percpu_counter_batch clusters in their local
3410 * counters. So we need to make sure we have free clusters more
179f7ebf 3411 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
a30d542a 3412 */
df55c99d 3413#define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
a30d542a 3414#else
df55c99d 3415#define EXT4_FREECLUSTERS_WATERMARK 0
a30d542a
AK
3416#endif
3417
f340b3d9 3418/* Update i_disksize. Requires i_rwsem to avoid races with truncate */
cf17fea6
AK
3419static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
3420{
90e775b7 3421 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
5955102c 3422 !inode_is_locked(inode));
90e775b7
JK
3423 down_write(&EXT4_I(inode)->i_data_sem);
3424 if (newsize > EXT4_I(inode)->i_disksize)
35df4299 3425 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize);
90e775b7
JK
3426 up_write(&EXT4_I(inode)->i_data_sem);
3427}
3428
f340b3d9 3429/* Update i_size, i_disksize. Requires i_rwsem to avoid races with truncate */
4631dbf6
DM
3430static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
3431{
3432 int changed = 0;
3433
3434 if (newsize > inode->i_size) {
3435 i_size_write(inode, newsize);
3436 changed = 1;
3437 }
3438 if (newsize > EXT4_I(inode)->i_disksize) {
3439 ext4_update_i_disksize(inode, newsize);
3440 changed |= 2;
3441 }
3442 return changed;
3443}
3444
01127848
JK
3445int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
3446 loff_t len);
3447
5d1b1b3f
AK
3448struct ext4_group_info {
3449 unsigned long bb_state;
addd752c
CX
3450#ifdef AGGRESSIVE_CHECK
3451 unsigned long bb_check_counter;
3452#endif
5d1b1b3f 3453 struct rb_root bb_free_root;
a36b4498
ES
3454 ext4_grpblk_t bb_first_free; /* first free block */
3455 ext4_grpblk_t bb_free; /* total free blocks */
3456 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
8a57d9d6 3457 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
196e402a 3458 ext4_group_t bb_group; /* Group number */
5d1b1b3f
AK
3459 struct list_head bb_prealloc_list;
3460#ifdef DOUBLE_CHECK
3461 void *bb_bitmap;
3462#endif
3463 struct rw_semaphore alloc_sem;
196e402a
HS
3464 struct rb_node bb_avg_fragment_size_rb;
3465 struct list_head bb_largest_free_order_node;
a36b4498
ES
3466 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
3467 * regions, index is order.
3468 * bb_counters[3] = 5 means
3469 * 5 free 8-block regions. */
5d1b1b3f
AK
3470};
3471
3d56b8d2
TM
3472#define EXT4_GROUP_INFO_NEED_INIT_BIT 0
3473#define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
163a203d 3474#define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
87a39389 3475#define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
db79e6d1
WS
3476#define EXT4_GROUP_INFO_BBITMAP_CORRUPT \
3477 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT)
3478#define EXT4_GROUP_INFO_IBITMAP_CORRUPT \
3479 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT)
cfd73237 3480#define EXT4_GROUP_INFO_BBITMAP_READ_BIT 4
5d1b1b3f
AK
3481
3482#define EXT4_MB_GRP_NEED_INIT(grp) \
3483 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
163a203d
DW
3484#define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
3485 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
87a39389
DW
3486#define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
3487 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
5d1b1b3f 3488
3d56b8d2
TM
3489#define EXT4_MB_GRP_WAS_TRIMMED(grp) \
3490 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3491#define EXT4_MB_GRP_SET_TRIMMED(grp) \
3492 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3493#define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
3494 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
cfd73237
AZ
3495#define EXT4_MB_GRP_TEST_AND_SET_READ(grp) \
3496 (test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state)))
3d56b8d2 3497
50797481
TT
3498#define EXT4_MAX_CONTENTION 8
3499#define EXT4_CONTENTION_THRESHOLD 2
3500
955ce5f5
AK
3501static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
3502 ext4_group_t group)
5d1b1b3f 3503{
955ce5f5
AK
3504 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
3505}
5d1b1b3f 3506
50797481
TT
3507/*
3508 * Returns true if the filesystem is busy enough that attempts to
3509 * access the block group locks has run into contention.
3510 */
3511static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
3512{
3513 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
3514}
3515
955ce5f5
AK
3516static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
3517{
50797481
TT
3518 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
3519 if (spin_trylock(lock))
3520 /*
3521 * We're able to grab the lock right away, so drop the
3522 * lock contention counter.
3523 */
3524 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
3525 else {
3526 /*
3527 * The lock is busy, so bump the contention counter,
3528 * and then wait on the spin lock.
3529 */
3530 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
3531 EXT4_MAX_CONTENTION);
3532 spin_lock(lock);
3533 }
5d1b1b3f
AK
3534}
3535
3536static inline void ext4_unlock_group(struct super_block *sb,
3537 ext4_group_t group)
3538{
955ce5f5 3539 spin_unlock(ext4_group_lock_ptr(sb, group));
5d1b1b3f
AK
3540}
3541
f177ee08
RA
3542#ifdef CONFIG_QUOTA
3543static inline bool ext4_quota_capable(struct super_block *sb)
3544{
3545 return (test_opt(sb, QUOTA) || ext4_has_feature_quota(sb));
3546}
3547
3548static inline bool ext4_is_quota_journalled(struct super_block *sb)
3549{
3550 struct ext4_sb_info *sbi = EXT4_SB(sb);
3551
3552 return (ext4_has_feature_quota(sb) ||
3553 sbi->s_qf_names[USRQUOTA] || sbi->s_qf_names[GRPQUOTA]);
3554}
25c6d98f 3555int ext4_enable_quotas(struct super_block *sb);
f177ee08
RA
3556#endif
3557
1f7d1e77
TT
3558/*
3559 * Block validity checking
3560 */
3561#define ext4_check_indirect_blockref(inode, bh) \
3562 ext4_check_blockref(__func__, __LINE__, inode, \
3563 (__le32 *)(bh)->b_data, \
3564 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
3565
3566#define ext4_ind_check_inode(inode) \
3567 ext4_check_blockref(__func__, __LINE__, inode, \
3568 EXT4_I(inode)->i_data, \
3569 EXT4_NDIR_BLOCKS)
3570
ac27a0ec
DK
3571/*
3572 * Inodes and files operations
3573 */
3574
3575/* dir.c */
617ba13b 3576extern const struct file_operations ext4_dir_operations;
ac27a0ec
DK
3577
3578/* file.c */
c5ef1c42 3579extern const struct inode_operations ext4_file_inode_operations;
617ba13b 3580extern const struct file_operations ext4_file_operations;
e0d10bfa 3581extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
ac27a0ec 3582
95eaefbd 3583/* inline.c */
95eaefbd
TT
3584extern int ext4_get_max_inline_size(struct inode *inode);
3585extern int ext4_find_inline_data_nolock(struct inode *inode);
95eaefbd
TT
3586extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
3587 unsigned int len);
3588extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
3589
3590extern int ext4_readpage_inline(struct inode *inode, struct page *page);
3591extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3592 struct inode *inode,
3593 loff_t pos, unsigned len,
3594 unsigned flags,
3595 struct page **pagep);
3596extern int ext4_write_inline_data_end(struct inode *inode,
3597 loff_t pos, unsigned len,
3598 unsigned copied,
3599 struct page *page);
3600extern struct buffer_head *
3601ext4_journalled_write_inline_data(struct inode *inode,
3602 unsigned len,
3603 struct page *page);
3604extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
3605 struct inode *inode,
3606 loff_t pos, unsigned len,
3607 unsigned flags,
3608 struct page **pagep,
3609 void **fsdata);
5b643f9c
TT
3610extern int ext4_try_add_inline_entry(handle_t *handle,
3611 struct ext4_filename *fname,
56a04915 3612 struct inode *dir, struct inode *inode);
95eaefbd
TT
3613extern int ext4_try_create_inline_dir(handle_t *handle,
3614 struct inode *parent,
3615 struct inode *inode);
3616extern int ext4_read_inline_dir(struct file *filp,
725bebb2 3617 struct dir_context *ctx,
95eaefbd 3618 int *has_inline_data);
7633b08b
TT
3619extern int ext4_inlinedir_to_tree(struct file *dir_file,
3620 struct inode *dir, ext4_lblk_t block,
3621 struct dx_hash_info *hinfo,
3622 __u32 start_hash, __u32 start_minor_hash,
3623 int *has_inline_data);
95eaefbd 3624extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
5b643f9c 3625 struct ext4_filename *fname,
95eaefbd
TT
3626 struct ext4_dir_entry_2 **res_dir,
3627 int *has_inline_data);
3628extern int ext4_delete_inline_entry(handle_t *handle,
3629 struct inode *dir,
3630 struct ext4_dir_entry_2 *de_del,
3631 struct buffer_head *bh,
3632 int *has_inline_data);
a7550b30 3633extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
95eaefbd
TT
3634extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3635 struct ext4_dir_entry_2 **parent_de,
3636 int *retval);
3637extern int ext4_inline_data_fiemap(struct inode *inode,
3638 struct fiemap_extent_info *fieinfo,
d952d69e 3639 int *has_inline, __u64 start, __u64 len);
7046ae35
AG
3640
3641struct iomap;
3642extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap);
3643
01daf945 3644extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline);
95eaefbd
TT
3645
3646extern int ext4_convert_inline_data(struct inode *inode);
3647
83447ccb
ZL
3648static inline int ext4_has_inline_data(struct inode *inode)
3649{
3650 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3651 EXT4_I(inode)->i_inline_off;
3652}
3653
ac27a0ec 3654/* namei.c */
c5ef1c42
AV
3655extern const struct inode_operations ext4_dir_inode_operations;
3656extern const struct inode_operations ext4_special_inode_operations;
596397b7 3657extern struct dentry *ext4_get_parent(struct dentry *child);
a774f9c2
TM
3658extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3659 struct ext4_dir_entry_2 *de,
3660 int blocksize, int csum_size,
3661 unsigned int parent_ino, int dotdot_real_len);
ddce3b94
TT
3662extern void ext4_initialize_dirent_tail(struct buffer_head *bh,
3663 unsigned int blocksize);
f036adb3
TT
3664extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode,
3665 struct buffer_head *bh);
a80f7fcf 3666extern int __ext4_unlink(handle_t *handle, struct inode *dir, const struct qstr *d_name,
8016e29f
HS
3667 struct inode *inode);
3668extern int __ext4_link(struct inode *dir, struct inode *inode,
3669 struct dentry *dentry);
b886ee3e 3670
8af0f082 3671#define S_SHIFT 12
381cebfe 3672static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = {
8af0f082
TM
3673 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
3674 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
3675 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
3676 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
3677 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
3678 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
3679 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
3680};
3681
3682static inline void ext4_set_de_type(struct super_block *sb,
3683 struct ext4_dir_entry_2 *de,
3684 umode_t mode) {
e2b911c5 3685 if (ext4_has_feature_filetype(sb))
8af0f082
TM
3686 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3687}
3688
f64e02fe 3689/* readpages.c */
a07f624b 3690extern int ext4_mpage_readpages(struct inode *inode,
6311f91f 3691 struct readahead_control *rac, struct page *page);
22cfe4b4
EB
3692extern int __init ext4_init_post_read_processing(void);
3693extern void ext4_exit_post_read_processing(void);
ac27a0ec
DK
3694
3695/* symlink.c */
a7a67e8a 3696extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
c5ef1c42
AV
3697extern const struct inode_operations ext4_symlink_inode_operations;
3698extern const struct inode_operations ext4_fast_symlink_inode_operations;
ac27a0ec 3699
b5799018 3700/* sysfs.c */
d578b994 3701extern void ext4_notify_error_sysfs(struct ext4_sb_info *sbi);
b5799018 3702extern int ext4_register_sysfs(struct super_block *sb);
ebd173be 3703extern void ext4_unregister_sysfs(struct super_block *sb);
b5799018
TT
3704extern int __init ext4_init_sysfs(void);
3705extern void ext4_exit_sysfs(void);
3706
6fd058f7
TT
3707/* block_validity */
3708extern void ext4_release_system_zone(struct super_block *sb);
3709extern int ext4_setup_system_zone(struct super_block *sb);
5dabfc78
TT
3710extern int __init ext4_init_system_zone(void);
3711extern void ext4_exit_system_zone(void);
ce9f24cc
JK
3712extern int ext4_inode_block_valid(struct inode *inode,
3713 ext4_fsblk_t start_blk,
3714 unsigned int count);
1f7d1e77
TT
3715extern int ext4_check_blockref(const char *, unsigned int,
3716 struct inode *, __le32 *, unsigned int);
6bc6c2bd
RH
3717extern int ext4_sb_block_valid(struct super_block *sb, struct inode *inode,
3718 ext4_fsblk_t start_blk, unsigned int count);
3719
6fd058f7 3720
a86c6181 3721/* extents.c */
4a092d73
TT
3722struct ext4_ext_path;
3723struct ext4_extent;
3724
d7b2a00c
ZL
3725/*
3726 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3727 * __le32.
3728 */
3729#define EXT_MAX_BLOCKS 0xffffffff
3730
4209ae12 3731extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode);
fffb2739 3732extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
e35fd660
TT
3733extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3734 struct ext4_map_blocks *map, int flags);
d0abb36d 3735extern int ext4_ext_truncate(handle_t *, struct inode *);
26a4c0c6
TT
3736extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3737 ext4_lblk_t end);
a86c6181
AT
3738extern void ext4_ext_init(struct super_block *);
3739extern void ext4_ext_release(struct super_block *);
2fe17c10 3740extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
a2df2a63 3741 loff_t len);
6b523df4
JK
3742extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3743 loff_t offset, ssize_t len);
a00713ea
RH
3744extern int ext4_convert_unwritten_io_end_vec(handle_t *handle,
3745 ext4_io_end_t *io_end);
e35fd660
TT
3746extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3747 struct ext4_map_blocks *map, int flags);
4a092d73
TT
3748extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3749 int num,
3750 struct ext4_ext_path *path);
4a092d73 3751extern int ext4_ext_insert_extent(handle_t *, struct inode *,
dfe50809 3752 struct ext4_ext_path **,
4a092d73 3753 struct ext4_extent *, int);
ed8a1a76
TT
3754extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3755 struct ext4_ext_path **,
3756 int flags);
4a092d73
TT
3757extern void ext4_ext_drop_refs(struct ext4_ext_path *);
3758extern int ext4_ext_check_inode(struct inode *inode);
fcf6b1b7 3759extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3a06d778
AK
3760extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3761 __u64 start, __u64 len);
bb5835ed
TT
3762extern int ext4_get_es_cache(struct inode *inode,
3763 struct fiemap_extent_info *fieinfo,
3764 __u64 start, __u64 len);
7869a4a6 3765extern int ext4_ext_precache(struct inode *inode);
fcf6b1b7
DM
3766extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3767 struct inode *inode2, ext4_lblk_t lblk1,
3768 ext4_lblk_t lblk2, ext4_lblk_t count,
3769 int mark_unwritten,int *err);
0b02f4c0 3770extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu);
a4130367 3771extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode,
83448bdf
JK
3772 int check_cred, int restart_cred,
3773 int revoke_cred);
8016e29f
HS
3774extern void ext4_ext_replay_shrink_inode(struct inode *inode, ext4_lblk_t end);
3775extern int ext4_ext_replay_set_iblocks(struct inode *inode);
3776extern int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start,
3777 int len, int unwritten, ext4_fsblk_t pblk);
3778extern int ext4_ext_clear_bb(struct inode *inode);
a4130367 3779
4a092d73 3780
748de673 3781/* move_extent.c */
393d1d1d
DTB
3782extern void ext4_double_down_write_data_sem(struct inode *first,
3783 struct inode *second);
3784extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3785 struct inode *donor_inode);
748de673
AF
3786extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3787 __u64 start_orig, __u64 start_donor,
3788 __u64 len, __u64 *moved_len);
3789
bd2d0210 3790/* page-io.c */
5dabfc78
TT
3791extern int __init ext4_init_pageio(void);
3792extern void ext4_exit_pageio(void);
bd2d0210 3793extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
97a851ed
JK
3794extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3795extern int ext4_put_io_end(ext4_io_end_t *io_end);
3796extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3797extern void ext4_io_submit_init(struct ext4_io_submit *io,
3798 struct writeback_control *wbc);
2e8fa54e 3799extern void ext4_end_io_rsv_work(struct work_struct *work);
bd2d0210
TT
3800extern void ext4_io_submit(struct ext4_io_submit *io);
3801extern int ext4_bio_write_page(struct ext4_io_submit *io,
3802 struct page *page,
3803 int len,
1c8349a1 3804 bool keep_towrite);
c8cc8816
RH
3805extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end);
3806extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end);
3a06d778 3807
c5e06d10
JL
3808/* mmp.c */
3809extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3810
618f0031
PS
3811/* mmp.c */
3812extern void ext4_stop_mmpd(struct ext4_sb_info *sbi);
3813
c93d8f88
EB
3814/* verity.c */
3815extern const struct fsverity_operations ext4_verityops;
3816
25c6d98f
JK
3817/* orphan.c */
3818extern int ext4_orphan_add(handle_t *, struct inode *);
3819extern int ext4_orphan_del(handle_t *, struct inode *);
3820extern void ext4_orphan_cleanup(struct super_block *sb,
3821 struct ext4_super_block *es);
02f310fc
JK
3822extern void ext4_release_orphan_info(struct super_block *sb);
3823extern int ext4_init_orphan_info(struct super_block *sb);
3824extern int ext4_orphan_file_empty(struct super_block *sb);
3825extern void ext4_orphan_file_block_trigger(
3826 struct jbd2_buffer_trigger_type *triggers,
3827 struct buffer_head *bh,
3828 void *data, size_t size);
25c6d98f 3829
2ccb5fb9 3830/*
48fc7f7e 3831 * Add new method to test whether block and inode bitmaps are properly
2ccb5fb9
AK
3832 * initialized. With uninit_bg reading the block from disk is not enough
3833 * to mark the bitmap uptodate. We need to also zero-out the bitmap
3834 */
3835#define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3836
3837static inline int bitmap_uptodate(struct buffer_head *bh)
3838{
3839 return (buffer_uptodate(bh) &&
3840 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3841}
3842static inline void set_bitmap_uptodate(struct buffer_head *bh)
3843{
3844 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3845}
3846
731eb1a0
AM
3847#define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
3848
e9e3bcec
ES
3849/* For ioend & aio unwritten conversion wait queues */
3850#define EXT4_WQ_HASH_SZ 37
3851#define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
3852 EXT4_WQ_HASH_SZ])
e9e3bcec 3853extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
e9e3bcec 3854
8f82f840
YY
3855extern int ext4_resize_begin(struct super_block *sb);
3856extern void ext4_resize_end(struct super_block *sb);
3857
74c66bcb
JK
3858static inline void ext4_set_io_unwritten_flag(struct inode *inode,
3859 struct ext4_io_end *io_end)
3860{
3861 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
3862 io_end->flag |= EXT4_IO_END_UNWRITTEN;
3863 atomic_inc(&EXT4_I(inode)->i_unwritten);
3864 }
3865}
3866
3867static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3868{
3869 struct inode *inode = io_end->inode;
3870
3871 if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
3872 io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3873 /* Wake up anyone waiting on unwritten extent conversion */
3874 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
3875 wake_up_all(ext4_ioend_wq(inode));
3876 }
3877}
3878
8ff6daa1 3879extern const struct iomap_ops ext4_iomap_ops;
8cd115bd 3880extern const struct iomap_ops ext4_iomap_overwrite_ops;
09edf4d3 3881extern const struct iomap_ops ext4_iomap_report_ops;
364443cb 3882
7963e5ac
Z
3883static inline int ext4_buffer_uptodate(struct buffer_head *bh)
3884{
3885 /*
3886 * If the buffer has the write error flag, we have failed
3887 * to write out data in the block. In this case, we don't
3888 * have to read the block because we may read the old data
3889 * successfully.
3890 */
5c680150 3891 if (buffer_write_io_error(bh))
7963e5ac
Z
3892 set_buffer_uptodate(bh);
3893 return buffer_uptodate(bh);
3894}
3895
ac27a0ec
DK
3896#endif /* __KERNEL__ */
3897
6a797d27
DW
3898#define EFSBADCRC EBADMSG /* Bad CRC detected */
3899#define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
3900
3dcf5451 3901#endif /* _EXT4_H */