]> git.ipfire.org Git - people/ms/linux.git/blame - fs/ext4/ext4.h
ext4 crypto: add ext4 encryption facilities
[people/ms/linux.git] / fs / ext4 / ext4.h
CommitLineData
ac27a0ec 1/*
3dcf5451 2 * ext4.h
ac27a0ec
DK
3 *
4 * Copyright (C) 1992, 1993, 1994, 1995
5 * Remy Card (card@masi.ibp.fr)
6 * Laboratoire MASI - Institut Blaise Pascal
7 * Universite Pierre et Marie Curie (Paris VI)
8 *
9 * from
10 *
11 * linux/include/linux/minix_fs.h
12 *
13 * Copyright (C) 1991, 1992 Linus Torvalds
14 */
15
3dcf5451
CH
16#ifndef _EXT4_H
17#define _EXT4_H
ac27a0ec
DK
18
19#include <linux/types.h>
3a5b2ecd 20#include <linux/blkdev.h>
ac27a0ec 21#include <linux/magic.h>
2ccb5fb9 22#include <linux/jbd2.h>
60e58e0f 23#include <linux/quota.h>
d444c3c3
TT
24#include <linux/rwsem.h>
25#include <linux/rbtree.h>
26#include <linux/seqlock.h>
27#include <linux/mutex.h>
ca0faba0
TT
28#include <linux/timer.h>
29#include <linux/wait.h>
30#include <linux/blockgroup_lock.h>
31#include <linux/percpu_counter.h>
efbed4dc 32#include <linux/ratelimit.h>
0441984a 33#include <crypto/hash.h>
a633f5a3 34#include <linux/falloc.h>
4d92dc0f
BH
35#ifdef __KERNEL__
36#include <linux/compat.h>
37#endif
c9de560d 38
ac27a0ec 39/*
8a35694e 40 * The fourth extended filesystem constants/structures
ac27a0ec
DK
41 */
42
43/*
617ba13b 44 * Define EXT4FS_DEBUG to produce debug messages
ac27a0ec 45 */
617ba13b 46#undef EXT4FS_DEBUG
ac27a0ec 47
ac27a0ec
DK
48/*
49 * Debug code
50 */
617ba13b
MC
51#ifdef EXT4FS_DEBUG
52#define ext4_debug(f, a...) \
ac27a0ec 53 do { \
af5bc92d 54 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
4db9c54a 55 __FILE__, __LINE__, __func__); \
af5bc92d 56 printk(KERN_DEBUG f, ## a); \
ac27a0ec
DK
57 } while (0)
58#else
ace36ad4 59#define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
ac27a0ec
DK
60#endif
61
4a092d73
TT
62/*
63 * Turn on EXT_DEBUG to get lots of info about extents operations.
64 */
65#define EXT_DEBUG__
66#ifdef EXT_DEBUG
67#define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__)
68#else
69#define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
70#endif
71
273df556 72#define EXT4_ERROR_INODE(inode, fmt, a...) \
c398eda0
TT
73 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
74
75#define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
76 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
273df556 77
f7c21177
TT
78#define EXT4_ERROR_FILE(file, block, fmt, a...) \
79 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
273df556 80
d444c3c3
TT
81/* data type for block offset of block group */
82typedef int ext4_grpblk_t;
83
84/* data type for filesystem-wide blocks number */
85typedef unsigned long long ext4_fsblk_t;
86
87/* data type for file logical block number */
88typedef __u32 ext4_lblk_t;
89
90/* data type for block group number */
91typedef unsigned int ext4_group_t;
92
296c355c 93/*
60e6679e 94 * Flags used in mballoc's allocation_context flags field.
296c355c
TT
95 *
96 * Also used to show what's going on for debugging purposes when the
97 * flag field is exported via the traceport interface
98 */
d444c3c3 99
c9de560d 100/* prefer goal again. length */
0ef90db9 101#define EXT4_MB_HINT_MERGE 0x0001
c9de560d 102/* blocks already reserved */
0ef90db9 103#define EXT4_MB_HINT_RESERVED 0x0002
c9de560d 104/* metadata is being allocated */
0ef90db9 105#define EXT4_MB_HINT_METADATA 0x0004
c9de560d 106/* first blocks in the file */
0ef90db9 107#define EXT4_MB_HINT_FIRST 0x0008
c9de560d 108/* search for the best chunk */
0ef90db9 109#define EXT4_MB_HINT_BEST 0x0010
c9de560d 110/* data is being allocated */
0ef90db9 111#define EXT4_MB_HINT_DATA 0x0020
c9de560d 112/* don't preallocate (for tails) */
0ef90db9 113#define EXT4_MB_HINT_NOPREALLOC 0x0040
c9de560d 114/* allocate for locality group */
0ef90db9 115#define EXT4_MB_HINT_GROUP_ALLOC 0x0080
c9de560d 116/* allocate goal blocks or none */
0ef90db9 117#define EXT4_MB_HINT_GOAL_ONLY 0x0100
c9de560d 118/* goal is meaningful */
0ef90db9 119#define EXT4_MB_HINT_TRY_GOAL 0x0200
d2a17637 120/* blocks already pre-reserved by delayed allocation */
0ef90db9 121#define EXT4_MB_DELALLOC_RESERVED 0x0400
4ba74d00
TT
122/* We are doing stream allocation */
123#define EXT4_MB_STREAM_ALLOC 0x0800
55f020db
AH
124/* Use reserved root blocks if needed */
125#define EXT4_MB_USE_ROOT_BLOCKS 0x1000
27dd4385
LC
126/* Use blocks from reserved pool */
127#define EXT4_MB_USE_RESERVED 0x2000
c9de560d
AT
128
129struct ext4_allocation_request {
130 /* target inode for block we're allocating */
131 struct inode *inode;
726447d8
ES
132 /* how many blocks we want to allocate */
133 unsigned int len;
c9de560d
AT
134 /* logical block in target inode */
135 ext4_lblk_t logical;
c9de560d
AT
136 /* the closest logical allocated block to the left */
137 ext4_lblk_t lleft;
c9de560d
AT
138 /* the closest logical allocated block to the right */
139 ext4_lblk_t lright;
726447d8
ES
140 /* phys. target (a hint) */
141 ext4_fsblk_t goal;
142 /* phys. block for the closest logical allocated block to the left */
143 ext4_fsblk_t pleft;
144 /* phys. block for the closest logical allocated block to the right */
c9de560d 145 ext4_fsblk_t pright;
c9de560d 146 /* flags. see above EXT4_MB_HINT_* */
498e5f24 147 unsigned int flags;
c9de560d
AT
148};
149
e35fd660
TT
150/*
151 * Logical to physical block mapping, used by ext4_map_blocks()
152 *
153 * This structure is used to pass requests into ext4_map_blocks() as
154 * well as to store the information returned by ext4_map_blocks(). It
155 * takes less room on the stack than a struct buffer_head.
156 */
157#define EXT4_MAP_NEW (1 << BH_New)
158#define EXT4_MAP_MAPPED (1 << BH_Mapped)
159#define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten)
160#define EXT4_MAP_BOUNDARY (1 << BH_Boundary)
e35fd660 161#define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
cbd7584e 162 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
e35fd660
TT
163
164struct ext4_map_blocks {
165 ext4_fsblk_t m_pblk;
166 ext4_lblk_t m_lblk;
167 unsigned int m_len;
168 unsigned int m_flags;
169};
170
bd2d0210
TT
171/*
172 * Flags for ext4_io_end->flags
173 */
174#define EXT4_IO_END_UNWRITTEN 0x0001
bd2d0210 175
4188188b 176/*
556615dc 177 * For converting unwritten extents on a work queue. 'handle' is used for
6b523df4 178 * buffered writeback.
4188188b 179 */
0031462b 180typedef struct ext4_io_end {
5b3ff237 181 struct list_head list; /* per-file finished IO list */
6b523df4
JK
182 handle_t *handle; /* handle reserved for extent
183 * conversion */
0031462b 184 struct inode *inode; /* file being written to */
b0857d30
JK
185 struct bio *bio; /* Linked list of completed
186 * bios covering the extent */
8d5d02e6 187 unsigned int flag; /* unwritten or not */
a1de02dc
ES
188 loff_t offset; /* offset in the file */
189 ssize_t size; /* size of the extent */
97a851ed 190 atomic_t count; /* reference counter */
0031462b
MC
191} ext4_io_end_t;
192
bd2d0210
TT
193struct ext4_io_submit {
194 int io_op;
195 struct bio *io_bio;
196 ext4_io_end_t *io_end;
bd2d0210
TT
197 sector_t io_next_block;
198};
199
ac27a0ec
DK
200/*
201 * Special inodes numbers
202 */
617ba13b
MC
203#define EXT4_BAD_INO 1 /* Bad blocks inode */
204#define EXT4_ROOT_INO 2 /* Root inode */
ae812306
AK
205#define EXT4_USR_QUOTA_INO 3 /* User quota inode */
206#define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
617ba13b
MC
207#define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
208#define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
209#define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
210#define EXT4_JOURNAL_INO 8 /* Journal inode */
ac27a0ec 211
617ba13b
MC
212/* First non-reserved inode for old ext4 filesystems */
213#define EXT4_GOOD_OLD_FIRST_INO 11
ac27a0ec
DK
214
215/*
216 * Maximal count of links to a file
217 */
f8628a14 218#define EXT4_LINK_MAX 65000
ac27a0ec
DK
219
220/*
221 * Macro-instructions used to manage several block sizes
222 */
617ba13b 223#define EXT4_MIN_BLOCK_SIZE 1024
afc7cbca
TS
224#define EXT4_MAX_BLOCK_SIZE 65536
225#define EXT4_MIN_BLOCK_LOG_SIZE 10
fb1813f4 226#define EXT4_MAX_BLOCK_LOG_SIZE 16
ac27a0ec 227#ifdef __KERNEL__
617ba13b 228# define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
ac27a0ec 229#else
617ba13b 230# define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
ac27a0ec 231#endif
af5bc92d 232#define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
281b5995
TT
233#define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
234 EXT4_SB(s)->s_cluster_bits)
ac27a0ec 235#ifdef __KERNEL__
617ba13b 236# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
281b5995 237# define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
ac27a0ec 238#else
617ba13b 239# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
ac27a0ec
DK
240#endif
241#ifdef __KERNEL__
617ba13b
MC
242#define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
243#define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
244#define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
ac27a0ec 245#else
617ba13b
MC
246#define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
247 EXT4_GOOD_OLD_INODE_SIZE : \
ac27a0ec 248 (s)->s_inode_size)
617ba13b
MC
249#define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
250 EXT4_GOOD_OLD_FIRST_INO : \
ac27a0ec
DK
251 (s)->s_first_ino)
252#endif
a2df2a63 253#define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
ac27a0ec 254
d5b8f310
TT
255/* Translate a block number to a cluster number */
256#define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
257/* Translate a cluster number to a block number */
258#define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
259/* Translate # of blks to # of clusters */
260#define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
261 (sbi)->s_cluster_bits)
f5a44db5
TT
262/* Mask out the low bits to get the starting block of the cluster */
263#define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
264 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
265#define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
266 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
267/* Get the cluster offset */
268#define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
269 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
270#define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
271 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
d5b8f310 272
ac27a0ec
DK
273/*
274 * Structure of a blocks group descriptor
275 */
617ba13b 276struct ext4_group_desc
ac27a0ec 277{
3a14589c 278 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
5272f837
AK
279 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
280 __le32 bg_inode_table_lo; /* Inodes table block */
560671a0
AK
281 __le16 bg_free_blocks_count_lo;/* Free blocks count */
282 __le16 bg_free_inodes_count_lo;/* Free inodes count */
283 __le16 bg_used_dirs_count_lo; /* Directories count */
717d50e4 284 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
e6153918
DW
285 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
286 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
287 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
560671a0 288 __le16 bg_itable_unused_lo; /* Unused inodes count */
717d50e4 289 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
8fadc143
AR
290 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
291 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
292 __le32 bg_inode_table_hi; /* Inodes table block MSB */
91b51a01
CL
293 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
294 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
295 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
560671a0 296 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
e6153918
DW
297 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
298 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
299 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
300 __u32 bg_reserved;
ac27a0ec
DK
301};
302
41a246d1
DW
303#define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
304 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
305 sizeof(__le16))
306#define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
307 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
308 sizeof(__le16))
309
772cb7c8
JS
310/*
311 * Structure of a flex block group info
312 */
313
314struct flex_groups {
90ba983f
TT
315 atomic64_t free_clusters;
316 atomic_t free_inodes;
317 atomic_t used_dirs;
772cb7c8
JS
318};
319
717d50e4
AD
320#define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
321#define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
322#define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
323
ac27a0ec
DK
324/*
325 * Macro-instructions used to manage group descriptors
326 */
0d1ee42f 327#define EXT4_MIN_DESC_SIZE 32
8fadc143 328#define EXT4_MIN_DESC_SIZE_64BIT 64
0d1ee42f
AR
329#define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
330#define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
ac27a0ec 331#ifdef __KERNEL__
617ba13b 332# define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
281b5995 333# define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
617ba13b
MC
334# define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
335# define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
336# define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
ac27a0ec 337#else
617ba13b 338# define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
0d1ee42f 339# define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
617ba13b 340# define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
ac27a0ec
DK
341#endif
342
343/*
344 * Constants relative to the data blocks
345 */
617ba13b
MC
346#define EXT4_NDIR_BLOCKS 12
347#define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
348#define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
349#define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
350#define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
ac27a0ec
DK
351
352/*
353 * Inode flags
354 */
617ba13b
MC
355#define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
356#define EXT4_UNRM_FL 0x00000002 /* Undelete */
357#define EXT4_COMPR_FL 0x00000004 /* Compress file */
358#define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
359#define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
360#define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
361#define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
362#define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
ac27a0ec 363/* Reserved for compression usage... */
617ba13b
MC
364#define EXT4_DIRTY_FL 0x00000100
365#define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
366#define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
3edc18d8
TT
367 /* nb: was previously EXT2_ECOMPR_FL */
368#define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */
ac27a0ec 369/* End compression flags --- maybe not all used */
617ba13b
MC
370#define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
371#define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
372#define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
373#define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
374#define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
375#define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
8180a562 376#define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
a86c6181 377#define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
f710b4b9 378#define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
c8d46e41 379#define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
67cf5b09 380#define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
8180a562 381#define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
ac27a0ec 382
c8d46e41 383#define EXT4_FL_USER_VISIBLE 0x004BDFFF /* User visible flags */
996bb9fd 384#define EXT4_FL_USER_MODIFIABLE 0x004380FF /* User modifiable flags */
ac27a0ec 385
8fa43a81
DG
386/* Flags that should be inherited by new inodes from their parent. */
387#define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
1cd9f097 388 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
8fa43a81
DG
389 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
390 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
391
2dc6b0d4
DG
392/* Flags that are appropriate for regular files (all but dir-specific ones). */
393#define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
394
395/* Flags that are appropriate for non-directories/regular files. */
396#define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
397
398/* Mask out flags that are inappropriate for the given type of inode. */
399static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
400{
401 if (S_ISDIR(mode))
402 return flags;
403 else if (S_ISREG(mode))
404 return flags & EXT4_REG_FLMASK;
405 else
406 return flags & EXT4_OTHER_FLMASK;
407}
408
12e9b892
DM
409/*
410 * Inode flags used for atomic set/get
411 */
412enum {
413 EXT4_INODE_SECRM = 0, /* Secure deletion */
414 EXT4_INODE_UNRM = 1, /* Undelete */
415 EXT4_INODE_COMPR = 2, /* Compress file */
416 EXT4_INODE_SYNC = 3, /* Synchronous updates */
417 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
418 EXT4_INODE_APPEND = 5, /* writes to file may only append */
419 EXT4_INODE_NODUMP = 6, /* do not dump file */
420 EXT4_INODE_NOATIME = 7, /* do not update atime */
421/* Reserved for compression usage... */
422 EXT4_INODE_DIRTY = 8,
423 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
424 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
f542fbe8 425 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */
12e9b892
DM
426/* End compression flags --- maybe not all used */
427 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
428 EXT4_INODE_IMAGIC = 13, /* AFS directory */
429 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
430 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
431 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
432 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
433 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
434 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
435 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
436 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
67cf5b09 437 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
12e9b892
DM
438 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
439};
440
12e9b892 441/*
9a4c8019
CM
442 * Since it's pretty easy to mix up bit numbers and hex values, we use a
443 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
444 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
445 * any extra space in the compiled kernel image, otherwise, the build will fail.
446 * It's important that these values are the same, since we are using
447 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
448 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
449 * values found in ext2, ext3 and ext4 filesystems, and of course the values
450 * defined in e2fsprogs.
12e9b892
DM
451 *
452 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
453 */
9a4c8019
CM
454#define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
455#define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
456
12e9b892
DM
457static inline void ext4_check_flag_values(void)
458{
459 CHECK_FLAG_VALUE(SECRM);
460 CHECK_FLAG_VALUE(UNRM);
461 CHECK_FLAG_VALUE(COMPR);
462 CHECK_FLAG_VALUE(SYNC);
463 CHECK_FLAG_VALUE(IMMUTABLE);
464 CHECK_FLAG_VALUE(APPEND);
465 CHECK_FLAG_VALUE(NODUMP);
466 CHECK_FLAG_VALUE(NOATIME);
467 CHECK_FLAG_VALUE(DIRTY);
468 CHECK_FLAG_VALUE(COMPRBLK);
469 CHECK_FLAG_VALUE(NOCOMPR);
3edc18d8 470 CHECK_FLAG_VALUE(ENCRYPT);
12e9b892
DM
471 CHECK_FLAG_VALUE(INDEX);
472 CHECK_FLAG_VALUE(IMAGIC);
473 CHECK_FLAG_VALUE(JOURNAL_DATA);
474 CHECK_FLAG_VALUE(NOTAIL);
475 CHECK_FLAG_VALUE(DIRSYNC);
476 CHECK_FLAG_VALUE(TOPDIR);
477 CHECK_FLAG_VALUE(HUGE_FILE);
478 CHECK_FLAG_VALUE(EXTENTS);
479 CHECK_FLAG_VALUE(EA_INODE);
480 CHECK_FLAG_VALUE(EOFBLOCKS);
67cf5b09 481 CHECK_FLAG_VALUE(INLINE_DATA);
12e9b892
DM
482 CHECK_FLAG_VALUE(RESERVED);
483}
484
ac27a0ec 485/* Used to pass group descriptor data when online resize is done */
617ba13b 486struct ext4_new_group_input {
8c55e204
DK
487 __u32 group; /* Group number for this data */
488 __u64 block_bitmap; /* Absolute block number of block bitmap */
489 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
490 __u64 inode_table; /* Absolute block number of inode table start */
491 __u32 blocks_count; /* Total number of blocks in this group */
492 __u16 reserved_blocks; /* Number of reserved blocks in this group */
ac27a0ec
DK
493 __u16 unused;
494};
495
4d92dc0f
BH
496#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
497struct compat_ext4_new_group_input {
498 u32 group;
499 compat_u64 block_bitmap;
500 compat_u64 inode_bitmap;
501 compat_u64 inode_table;
502 u32 blocks_count;
503 u16 reserved_blocks;
504 u16 unused;
505};
506#endif
507
617ba13b
MC
508/* The struct ext4_new_group_input in kernel space, with free_blocks_count */
509struct ext4_new_group_data {
ac27a0ec 510 __u32 group;
bd81d8ee
LV
511 __u64 block_bitmap;
512 __u64 inode_bitmap;
513 __u64 inode_table;
ac27a0ec
DK
514 __u32 blocks_count;
515 __u16 reserved_blocks;
516 __u16 unused;
517 __u32 free_blocks_count;
518};
519
33afdcc5
YY
520/* Indexes used to index group tables in ext4_new_group_data */
521enum {
522 BLOCK_BITMAP = 0, /* block bitmap */
523 INODE_BITMAP, /* inode bitmap */
524 INODE_TABLE, /* inode tables */
525 GROUP_TABLE_COUNT,
526};
527
a2df2a63 528/*
79e83036 529 * Flags used by ext4_map_blocks()
a2df2a63 530 */
556615dc 531 /* Allocate any needed blocks and/or convert an unwritten
c2177057 532 extent to be an initialized ext4 */
2ac3b6e0 533#define EXT4_GET_BLOCKS_CREATE 0x0001
556615dc
LC
534 /* Request the creation of an unwritten extent */
535#define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002
536#define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\
c2177057 537 EXT4_GET_BLOCKS_CREATE)
27dd4385
LC
538 /* Caller is from the delayed allocation writeout path
539 * finally doing the actual allocation of delayed blocks */
03f5d8bc 540#define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
0031462b 541 /* caller is from the direct IO path, request to creation of an
556615dc 542 unwritten extents if not allocated, split the unwritten
0031462b 543 extent if blocks has been preallocated already*/
c7064ef1 544#define EXT4_GET_BLOCKS_PRE_IO 0x0008
1296cc85 545#define EXT4_GET_BLOCKS_CONVERT 0x0010
c7064ef1 546#define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
556615dc 547 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
c7064ef1
JZ
548 /* Convert extent to initialized after IO complete */
549#define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
556615dc 550 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
27dd4385
LC
551 /* Eventual metadata allocation (due to growing extent tree)
552 * should not fail, so try to use reserved blocks for that.*/
553#define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
556b27ab
VH
554 /* Don't normalize allocation size (used for fallocate) */
555#define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
a4e5d88b
DM
556 /* Request will not result in inode size update (user for fallocate) */
557#define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080
729f52c6
ZL
558 /* Do not take i_data_sem locking in ext4_map_blocks */
559#define EXT4_GET_BLOCKS_NO_LOCK 0x0100
b8a86845 560 /* Convert written extents to unwritten */
2f8e0a7c 561#define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0200
ac27a0ec 562
107a7bd3 563/*
7869a4a6 564 * The bit position of these flags must not overlap with any of the
ed8a1a76 565 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(),
107a7bd3 566 * read_extent_tree_block(), ext4_split_extent_at(),
7869a4a6
TT
567 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
568 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
569 * caching the extents when reading from the extent tree while a
570 * truncate or punch hole operation is in progress.
107a7bd3 571 */
d26e2c4d
TT
572#define EXT4_EX_NOCACHE 0x40000000
573#define EXT4_EX_FORCE_CACHE 0x20000000
107a7bd3 574
e6362609
TT
575/*
576 * Flags used by ext4_free_blocks
577 */
578#define EXT4_FREE_BLOCKS_METADATA 0x0001
579#define EXT4_FREE_BLOCKS_FORGET 0x0002
1f2acb60 580#define EXT4_FREE_BLOCKS_VALIDATED 0x0004
7132de74 581#define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
84130193
TT
582#define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
583#define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
e6362609 584
f542fbe8
TT
585/* Encryption algorithms */
586#define EXT4_ENCRYPTION_MODE_INVALID 0
587#define EXT4_ENCRYPTION_MODE_AES_256_XTS 1
588#define EXT4_ENCRYPTION_MODE_AES_256_GCM 2
589#define EXT4_ENCRYPTION_MODE_AES_256_CBC 3
590#define EXT4_ENCRYPTION_MODE_AES_256_CTS 4
591
9bd8212f
MH
592#include "ext4_crypto.h"
593
ac27a0ec
DK
594/*
595 * ioctl commands
596 */
617ba13b
MC
597#define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
598#define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
599#define EXT4_IOC_GETVERSION _IOR('f', 3, long)
600#define EXT4_IOC_SETVERSION _IOW('f', 4, long)
617ba13b
MC
601#define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
602#define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
617ba13b
MC
603#define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
604#define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
8eea80d5
TT
605#define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
606#define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
607#define EXT4_IOC_MIGRATE _IO('f', 9)
ccd2506b 608 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
8eea80d5 609 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
ccd2506b 610#define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
748de673 611#define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
19c5246d 612#define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
393d1d1d 613#define EXT4_IOC_SWAP_BOOT _IO('f', 17)
7869a4a6 614#define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18)
9bd8212f
MH
615#define EXT4_IOC_SET_ENCRYPTION_POLICY _IOR('f', 19, struct ext4_encryption_policy)
616#define EXT4_IOC_GET_ENCRYPTION_PWSALT _IOW('f', 20, __u8[16])
617#define EXT4_IOC_GET_ENCRYPTION_POLICY _IOW('f', 21, struct ext4_encryption_policy)
ac27a0ec 618
899ad0ce 619#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
ac27a0ec
DK
620/*
621 * ioctl commands in 32 bit emulation
622 */
617ba13b
MC
623#define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
624#define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
625#define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
626#define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
627#define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
628#define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
629#define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
4d92dc0f 630#define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
617ba13b
MC
631#define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
632#define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
899ad0ce 633#endif
ac27a0ec 634
b595076a 635/* Max physical block we can address w/o extents */
fb0a387d
ES
636#define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
637
ac27a0ec
DK
638/*
639 * Structure of an inode on the disk
640 */
617ba13b 641struct ext4_inode {
ac27a0ec
DK
642 __le16 i_mode; /* File mode */
643 __le16 i_uid; /* Low 16 bits of Owner Uid */
a48380f7 644 __le32 i_size_lo; /* Size in bytes */
ac27a0ec 645 __le32 i_atime; /* Access time */
ef7f3835 646 __le32 i_ctime; /* Inode Change time */
ac27a0ec
DK
647 __le32 i_mtime; /* Modification time */
648 __le32 i_dtime; /* Deletion Time */
649 __le16 i_gid; /* Low 16 bits of Group Id */
650 __le16 i_links_count; /* Links count */
0fc1b451 651 __le32 i_blocks_lo; /* Blocks count */
ac27a0ec
DK
652 __le32 i_flags; /* File flags */
653 union {
654 struct {
25ec56b5 655 __le32 l_i_version;
ac27a0ec
DK
656 } linux1;
657 struct {
658 __u32 h_i_translator;
659 } hurd1;
660 struct {
661 __u32 m_i_reserved1;
662 } masix1;
663 } osd1; /* OS dependent 1 */
617ba13b 664 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
ac27a0ec 665 __le32 i_generation; /* File version (for NFS) */
7973c0c1 666 __le32 i_file_acl_lo; /* File ACL */
a48380f7 667 __le32 i_size_high;
f077d0d7 668 __le32 i_obso_faddr; /* Obsoleted fragment address */
ac27a0ec
DK
669 union {
670 struct {
0fc1b451 671 __le16 l_i_blocks_high; /* were l_i_reserved1 */
a1ddeb7e 672 __le16 l_i_file_acl_high;
8c55e204 673 __le16 l_i_uid_high; /* these 2 fields */
ac27a0ec 674 __le16 l_i_gid_high; /* were reserved2[0] */
e6153918
DW
675 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
676 __le16 l_i_reserved;
ac27a0ec
DK
677 } linux2;
678 struct {
f077d0d7 679 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
ac27a0ec
DK
680 __u16 h_i_mode_high;
681 __u16 h_i_uid_high;
682 __u16 h_i_gid_high;
683 __u32 h_i_author;
684 } hurd2;
685 struct {
f077d0d7 686 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
a1ddeb7e 687 __le16 m_i_file_acl_high;
ac27a0ec
DK
688 __u32 m_i_reserved2[2];
689 } masix2;
690 } osd2; /* OS dependent 2 */
691 __le16 i_extra_isize;
e6153918 692 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
ef7f3835
KS
693 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
694 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
695 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
696 __le32 i_crtime; /* File Creation time */
697 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
25ec56b5 698 __le32 i_version_hi; /* high 32 bits for 64-bit version */
ac27a0ec
DK
699};
700
748de673
AF
701struct move_extent {
702 __u32 reserved; /* should be zero */
703 __u32 donor_fd; /* donor file descriptor */
704 __u64 orig_start; /* logical start offset in block for orig */
705 __u64 donor_start; /* logical start offset in block for donor */
706 __u64 len; /* block length to be moved */
707 __u64 moved_len; /* moved block length */
708};
ac27a0ec 709
ef7f3835
KS
710#define EXT4_EPOCH_BITS 2
711#define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
712#define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
713
714/*
715 * Extended fields will fit into an inode if the filesystem was formatted
716 * with large inodes (-I 256 or larger) and there are not currently any EAs
717 * consuming all of the available space. For new inodes we always reserve
718 * enough space for the kernel's known extended fields, but for inodes
719 * created with an old kernel this might not have been the case. None of
720 * the extended inode fields is critical for correct filesystem operation.
721 * This macro checks if a certain field fits in the inode. Note that
722 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
723 */
724#define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
725 ((offsetof(typeof(*ext4_inode), field) + \
726 sizeof((ext4_inode)->field)) \
727 <= (EXT4_GOOD_OLD_INODE_SIZE + \
728 (einode)->i_extra_isize)) \
729
730static inline __le32 ext4_encode_extra_time(struct timespec *time)
731{
732 return cpu_to_le32((sizeof(time->tv_sec) > 4 ?
c1fccc06
TT
733 (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) |
734 ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK));
ef7f3835
KS
735}
736
737static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
738{
739 if (sizeof(time->tv_sec) > 4)
740 time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK)
741 << 32;
c1fccc06 742 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
ef7f3835
KS
743}
744
745#define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
746do { \
747 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
748 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
749 (raw_inode)->xtime ## _extra = \
750 ext4_encode_extra_time(&(inode)->xtime); \
751} while (0)
752
753#define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
754do { \
755 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
756 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
757 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
758 (raw_inode)->xtime ## _extra = \
759 ext4_encode_extra_time(&(einode)->xtime); \
760} while (0)
761
762#define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
763do { \
764 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
765 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
766 ext4_decode_extra_time(&(inode)->xtime, \
767 raw_inode->xtime ## _extra); \
af0b44a1
ES
768 else \
769 (inode)->xtime.tv_nsec = 0; \
ef7f3835
KS
770} while (0)
771
772#define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
773do { \
774 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
775 (einode)->xtime.tv_sec = \
776 (signed)le32_to_cpu((raw_inode)->xtime); \
19ea8060
TT
777 else \
778 (einode)->xtime.tv_sec = 0; \
ef7f3835
KS
779 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
780 ext4_decode_extra_time(&(einode)->xtime, \
781 raw_inode->xtime ## _extra); \
af0b44a1
ES
782 else \
783 (einode)->xtime.tv_nsec = 0; \
ef7f3835
KS
784} while (0)
785
25ec56b5
JNC
786#define i_disk_version osd1.linux1.l_i_version
787
ac27a0ec
DK
788#if defined(__KERNEL__) || defined(__linux__)
789#define i_reserved1 osd1.linux1.l_i_reserved1
a1ddeb7e 790#define i_file_acl_high osd2.linux2.l_i_file_acl_high
0fc1b451 791#define i_blocks_high osd2.linux2.l_i_blocks_high
ac27a0ec
DK
792#define i_uid_low i_uid
793#define i_gid_low i_gid
794#define i_uid_high osd2.linux2.l_i_uid_high
795#define i_gid_high osd2.linux2.l_i_gid_high
e6153918 796#define i_checksum_lo osd2.linux2.l_i_checksum_lo
ac27a0ec
DK
797
798#elif defined(__GNU__)
799
800#define i_translator osd1.hurd1.h_i_translator
ac27a0ec
DK
801#define i_uid_high osd2.hurd2.h_i_uid_high
802#define i_gid_high osd2.hurd2.h_i_gid_high
803#define i_author osd2.hurd2.h_i_author
804
805#elif defined(__masix__)
806
807#define i_reserved1 osd1.masix1.m_i_reserved1
a1ddeb7e 808#define i_file_acl_high osd2.masix2.m_i_file_acl_high
ac27a0ec
DK
809#define i_reserved2 osd2.masix2.m_i_reserved2
810
811#endif /* defined(__KERNEL__) || defined(__linux__) */
812
c0677e6d
ZL
813#include "extents_status.h"
814
d444c3c3
TT
815/*
816 * fourth extended file system inode data in memory
817 */
818struct ext4_inode_info {
819 __le32 i_data[15]; /* unconverted */
d444c3c3 820 __u32 i_dtime;
12e9b892 821 ext4_fsblk_t i_file_acl;
d444c3c3
TT
822
823 /*
824 * i_block_group is the number of the block group which contains
825 * this file's inode. Constant across the lifetime of the inode,
826 * it is ued for making block allocation decisions - we try to
827 * place a file's data blocks near its inode block, and new inodes
828 * near to their parent directory's inode.
829 */
830 ext4_group_t i_block_group;
8a2005d3 831 ext4_lblk_t i_dir_start_lookup;
353eb83c 832#if (BITS_PER_LONG < 64)
19f5fb7a 833 unsigned long i_state_flags; /* Dynamic state flags */
353eb83c 834#endif
12e9b892 835 unsigned long i_flags;
d444c3c3 836
d444c3c3
TT
837 /*
838 * Extended attributes can be read independently of the main file
839 * data. Taking i_mutex even when reading would cause contention
840 * between readers of EAs and writers of regular file data, so
841 * instead we synchronize on xattr_sem when reading or changing
842 * EAs.
843 */
844 struct rw_semaphore xattr_sem;
d444c3c3
TT
845
846 struct list_head i_orphan; /* unlinked but open inodes */
847
848 /*
849 * i_disksize keeps track of what the inode size is ON DISK, not
850 * in memory. During truncate, i_size is set to the new size by
851 * the VFS prior to calling ext4_truncate(), but the filesystem won't
852 * set i_disksize to 0 until the truncate is actually under way.
853 *
854 * The intent is that i_disksize always represents the blocks which
855 * are used by this file. This allows recovery to restart truncate
856 * on orphans if we crash during truncate. We actually write i_disksize
857 * into the on-disk inode when writing inodes out, instead of i_size.
858 *
859 * The only time when i_disksize and i_size may be different is when
860 * a truncate is in progress. The only things which change i_disksize
861 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
862 */
863 loff_t i_disksize;
864
865 /*
866 * i_data_sem is for serialising ext4_truncate() against
867 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
868 * data tree are chopped off during truncate. We can't do that in
869 * ext4 because whenever we perform intermediate commits during
870 * truncate, the inode and all the metadata blocks *must* be in a
871 * consistent state which allows truncation of the orphans to restart
872 * during recovery. Hence we must fix the get_block-vs-truncate race
873 * by other means, so we have i_data_sem.
874 */
875 struct rw_semaphore i_data_sem;
876 struct inode vfs_inode;
8aefcd55 877 struct jbd2_inode *jinode;
d444c3c3 878
202ee5df
TT
879 spinlock_t i_raw_lock; /* protects updates to the raw inode */
880
d444c3c3
TT
881 /*
882 * File creation time. Its function is same as that of
883 * struct timespec i_{a,c,m}time in the generic inode.
884 */
885 struct timespec i_crtime;
886
887 /* mballoc */
888 struct list_head i_prealloc_list;
889 spinlock_t i_prealloc_lock;
890
c0677e6d
ZL
891 /* extents status tree */
892 struct ext4_es_tree i_es_tree;
893 rwlock_t i_es_lock;
edaa53ca 894 struct list_head i_es_list;
eb68d0e2 895 unsigned int i_es_all_nr; /* protected by i_es_lock */
edaa53ca 896 unsigned int i_es_shk_nr; /* protected by i_es_lock */
dd475925
JK
897 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for
898 extents to shrink. Protected by
899 i_es_lock */
c0677e6d 900
d444c3c3
TT
901 /* ialloc */
902 ext4_group_t i_last_alloc_group;
903
904 /* allocation reservation info for delalloc */
57042651 905 /* In case of bigalloc, these refer to clusters rather than blocks */
d444c3c3
TT
906 unsigned int i_reserved_data_blocks;
907 unsigned int i_reserved_meta_blocks;
908 unsigned int i_allocated_meta_blocks;
01f49d0b 909 ext4_lblk_t i_da_metadata_calc_last_lblock;
9d0be502 910 int i_da_metadata_calc_len;
d444c3c3
TT
911
912 /* on-disk additional length */
913 __u16 i_extra_isize;
914
67cf5b09
TM
915 /* Indicate the inline data space. */
916 u16 i_inline_off;
917 u16 i_inline_size;
918
a9e7f447
DM
919#ifdef CONFIG_QUOTA
920 /* quota space reservation, managed internally by quota code */
921 qsize_t i_reserved_quota;
922#endif
8d5d02e6 923
2e8fa54e 924 /* Lock protecting lists below */
744692dc 925 spinlock_t i_completed_io_lock;
2e8fa54e
JK
926 /*
927 * Completed IOs that need unwritten extents handling and have
928 * transaction reserved
929 */
930 struct list_head i_rsv_conversion_list;
931 /*
932 * Completed IOs that need unwritten extents handling and don't have
933 * transaction reserved
934 */
8a2005d3 935 atomic_t i_ioend_count; /* Number of outstanding io_end structs */
e27f41e1 936 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
2e8fa54e 937 struct work_struct i_rsv_conversion_work;
8a2005d3
TT
938
939 spinlock_t i_block_reservation_lock;
b436b9be
JK
940
941 /*
942 * Transactions that contain inode's metadata needed to complete
943 * fsync and fdatasync, respectively.
944 */
945 tid_t i_sync_tid;
946 tid_t i_datasync_tid;
814525f4 947
96c7e0d9
JK
948#ifdef CONFIG_QUOTA
949 struct dquot *i_dquot[MAXQUOTAS];
950#endif
951
814525f4
DW
952 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
953 __u32 i_csum_seed;
b30ab0e0
MH
954
955#ifdef CONFIG_EXT4_FS_ENCRYPTION
956 /* Encryption params */
957 struct ext4_encryption_key i_encryption_key;
958#endif
d444c3c3
TT
959};
960
ac27a0ec
DK
961/*
962 * File system states
963 */
617ba13b
MC
964#define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
965#define EXT4_ERROR_FS 0x0002 /* Errors detected */
966#define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
ac27a0ec 967
469108ff
TT
968/*
969 * Misc. filesystem flags
970 */
971#define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
972#define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
973#define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
974
ac27a0ec 975/*
68911009 976 * Mount flags set via mount options or defaults
ac27a0ec 977 */
617ba13b
MC
978#define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
979#define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
980#define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
981#define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
982#define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
39ef17f1 983#define EXT4_MOUNT_ERRORS_MASK 0x00070
617ba13b
MC
984#define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
985#define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
923ae0ff
RZ
986#ifdef CONFIG_FS_DAX
987#define EXT4_MOUNT_DAX 0x00200 /* Direct Access */
988#else
989#define EXT4_MOUNT_DAX 0
990#endif
617ba13b
MC
991#define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
992#define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
993#define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
994#define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
995#define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
996#define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
997#define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
998#define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
afd4672d 999#define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
617ba13b 1000#define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
617ba13b
MC
1001#define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
1002#define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
1003#define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
744692dc 1004#define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
d4da6c9c 1005#define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
818d276c 1006#define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
64769240 1007#define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
5bf5683a 1008#define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
6fd058f7 1009#define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
5328e635 1010#define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
bfff6873 1011#define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
5bf5683a 1012
68911009
LC
1013/*
1014 * Mount flags set either automatically (could not be set by mount option)
1015 * based on per file system feature or property or in special cases such as
1016 * distinguishing between explicit mount option definition and default.
1017 */
56889787
TT
1018#define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1019 specified delalloc */
68911009
LC
1020#define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1021 size of blocksize * 8
1022 blocks */
ed3654eb
TT
1023#define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1024 file systems */
56889787 1025
fd8c37ec
TT
1026#define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1027 ~EXT4_MOUNT_##opt
1028#define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1029 EXT4_MOUNT_##opt
617ba13b
MC
1030#define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1031 EXT4_MOUNT_##opt)
ac27a0ec 1032
a2595b8a
TT
1033#define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1034 ~EXT4_MOUNT2_##opt
1035#define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1036 EXT4_MOUNT2_##opt
1037#define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1038 EXT4_MOUNT2_##opt)
1039
597d508c
AM
1040#define ext4_test_and_set_bit __test_and_set_bit_le
1041#define ext4_set_bit __set_bit_le
617ba13b 1042#define ext4_set_bit_atomic ext2_set_bit_atomic
597d508c
AM
1043#define ext4_test_and_clear_bit __test_and_clear_bit_le
1044#define ext4_clear_bit __clear_bit_le
617ba13b 1045#define ext4_clear_bit_atomic ext2_clear_bit_atomic
50e0168c 1046#define ext4_test_bit test_bit_le
50e0168c
AM
1047#define ext4_find_next_zero_bit find_next_zero_bit_le
1048#define ext4_find_next_bit find_next_bit_le
ac27a0ec 1049
c3e94d1d
YY
1050extern void ext4_set_bits(void *bm, int cur, int len);
1051
ac27a0ec
DK
1052/*
1053 * Maximal mount counts between two filesystem checks
1054 */
617ba13b
MC
1055#define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1056#define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
ac27a0ec
DK
1057
1058/*
1059 * Behaviour when detecting errors
1060 */
617ba13b
MC
1061#define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1062#define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1063#define EXT4_ERRORS_PANIC 3 /* Panic */
1064#define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
ac27a0ec 1065
e6153918
DW
1066/* Metadata checksum algorithm codes */
1067#define EXT4_CRC32C_CHKSUM 1
1068
3edc18d8
TT
1069/* Encryption algorithms */
1070#define EXT4_ENCRYPTION_MODE_INVALID 0
1071#define EXT4_ENCRYPTION_MODE_AES_256_XTS 1
1072#define EXT4_ENCRYPTION_MODE_AES_256_GCM 2
1073#define EXT4_ENCRYPTION_MODE_AES_256_CBC 3
1074
ac27a0ec
DK
1075/*
1076 * Structure of the super block
1077 */
617ba13b 1078struct ext4_super_block {
ac27a0ec 1079/*00*/ __le32 s_inodes_count; /* Inodes count */
6bc9feff 1080 __le32 s_blocks_count_lo; /* Blocks count */
308ba3ec
AK
1081 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1082 __le32 s_free_blocks_count_lo; /* Free blocks count */
ac27a0ec
DK
1083/*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1084 __le32 s_first_data_block; /* First Data Block */
1085 __le32 s_log_block_size; /* Block size */
281b5995 1086 __le32 s_log_cluster_size; /* Allocation cluster size */
ac27a0ec 1087/*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
281b5995 1088 __le32 s_clusters_per_group; /* # Clusters per group */
ac27a0ec
DK
1089 __le32 s_inodes_per_group; /* # Inodes per group */
1090 __le32 s_mtime; /* Mount time */
1091/*30*/ __le32 s_wtime; /* Write time */
1092 __le16 s_mnt_count; /* Mount count */
1093 __le16 s_max_mnt_count; /* Maximal mount count */
1094 __le16 s_magic; /* Magic signature */
1095 __le16 s_state; /* File system state */
1096 __le16 s_errors; /* Behaviour when detecting errors */
1097 __le16 s_minor_rev_level; /* minor revision level */
1098/*40*/ __le32 s_lastcheck; /* time of last check */
1099 __le32 s_checkinterval; /* max. time between checks */
1100 __le32 s_creator_os; /* OS */
1101 __le32 s_rev_level; /* Revision level */
1102/*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1103 __le16 s_def_resgid; /* Default gid for reserved blocks */
1104 /*
617ba13b 1105 * These fields are for EXT4_DYNAMIC_REV superblocks only.
ac27a0ec
DK
1106 *
1107 * Note: the difference between the compatible feature set and
1108 * the incompatible feature set is that if there is a bit set
1109 * in the incompatible feature set that the kernel doesn't
1110 * know about, it should refuse to mount the filesystem.
1111 *
1112 * e2fsck's requirements are more strict; if it doesn't know
1113 * about a feature in either the compatible or incompatible
1114 * feature set, it must abort and not try to meddle with
1115 * things it doesn't understand...
1116 */
1117 __le32 s_first_ino; /* First non-reserved inode */
0d1ee42f 1118 __le16 s_inode_size; /* size of inode structure */
ac27a0ec
DK
1119 __le16 s_block_group_nr; /* block group # of this superblock */
1120 __le32 s_feature_compat; /* compatible feature set */
1121/*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1122 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1123/*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1124/*78*/ char s_volume_name[16]; /* volume name */
1125/*88*/ char s_last_mounted[64]; /* directory where last mounted */
1126/*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1127 /*
1128 * Performance hints. Directory preallocation should only
617ba13b 1129 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
ac27a0ec
DK
1130 */
1131 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1132 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1133 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1134 /*
617ba13b 1135 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
ac27a0ec
DK
1136 */
1137/*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1138/*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1139 __le32 s_journal_dev; /* device number of journal file */
1140 __le32 s_last_orphan; /* start of list of inodes to delete */
1141 __le32 s_hash_seed[4]; /* HTREE hash seed */
1142 __u8 s_def_hash_version; /* Default hash version to use */
89eeddf0 1143 __u8 s_jnl_backup_type;
0d1ee42f 1144 __le16 s_desc_size; /* size of group descriptor */
bd81d8ee 1145/*100*/ __le32 s_default_mount_opts;
ac27a0ec 1146 __le32 s_first_meta_bg; /* First metablock block group */
bd81d8ee
LV
1147 __le32 s_mkfs_time; /* When the filesystem was created */
1148 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1149 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1150/*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1151 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1152 __le32 s_free_blocks_count_hi; /* Free blocks count */
c1bddad9
AK
1153 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1154 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1155 __le32 s_flags; /* Miscellaneous flags */
1156 __le16 s_raid_stride; /* RAID stride */
c5e06d10 1157 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
c1bddad9
AK
1158 __le64 s_mmp_block; /* Block for multi-mount protection */
1159 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
772cb7c8 1160 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
e6153918 1161 __u8 s_checksum_type; /* metadata checksum algorithm used */
f542fbe8
TT
1162 __u8 s_encryption_level; /* versioning level for encryption */
1163 __u8 s_reserved_pad; /* Padding to next 32bits */
afc32f7e 1164 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
4a9cdec7
TT
1165 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1166 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1167 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1168 snapshot's future use */
1169 __le32 s_snapshot_list; /* inode number of the head of the
1170 on-disk snapshot list */
1c13d5c0
TT
1171#define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1172 __le32 s_error_count; /* number of fs errors */
1173 __le32 s_first_error_time; /* first time an error happened */
1174 __le32 s_first_error_ino; /* inode involved in first error */
1175 __le64 s_first_error_block; /* block involved of first error */
1176 __u8 s_first_error_func[32]; /* function where the error happened */
1177 __le32 s_first_error_line; /* line number where error happened */
1178 __le32 s_last_error_time; /* most recent time of an error */
1179 __le32 s_last_error_ino; /* inode involved in last error */
1180 __le32 s_last_error_line; /* line number where error happened */
1181 __le64 s_last_error_block; /* block involved of last error */
1182 __u8 s_last_error_func[32]; /* function where the error happened */
8b67f04a
TT
1183#define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1184 __u8 s_mount_opts[64];
281b5995
TT
1185 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1186 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1187 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1beeef1b 1188 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
3edc18d8 1189 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */
f542fbe8
TT
1190 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */
1191 __le32 s_lpf_ino; /* Location of the lost+found inode */
1192 __le32 s_reserved[100]; /* Padding to the end of the block */
e6153918 1193 __le32 s_checksum; /* crc32c(superblock) */
ac27a0ec
DK
1194};
1195
1c13d5c0
TT
1196#define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1197
ac27a0ec 1198#ifdef __KERNEL__
bc0b0d6d
TT
1199
1200/*
4ab2f15b 1201 * run-time mount flags
bc0b0d6d
TT
1202 */
1203#define EXT4_MF_MNTDIR_SAMPLED 0x0001
4ab2f15b 1204#define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
bc0b0d6d 1205
a2d4a646
JK
1206/* Number of quota types we support */
1207#define EXT4_MAXQUOTAS 2
1208
ca0faba0
TT
1209/*
1210 * fourth extended-fs super-block data in memory
1211 */
1212struct ext4_sb_info {
1213 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1214 unsigned long s_inodes_per_block;/* Number of inodes per block */
1215 unsigned long s_blocks_per_group;/* Number of blocks in a group */
281b5995 1216 unsigned long s_clusters_per_group; /* Number of clusters in a group */
ca0faba0
TT
1217 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1218 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1219 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1220 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1221 ext4_group_t s_groups_count; /* Number of groups in the fs */
fb0a387d 1222 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
952fc18e 1223 unsigned long s_overhead; /* # of fs overhead clusters */
281b5995
TT
1224 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1225 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
ca0faba0
TT
1226 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1227 struct buffer_head * s_sbh; /* Buffer containing the super block */
1228 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1229 struct buffer_head **s_group_desc;
7f4520cc 1230 unsigned int s_mount_opt;
a2595b8a 1231 unsigned int s_mount_opt2;
bc0b0d6d 1232 unsigned int s_mount_flags;
5a916be1 1233 unsigned int s_def_mount_opt;
ca0faba0 1234 ext4_fsblk_t s_sb_block;
27dd4385 1235 atomic64_t s_resv_clusters;
08cefc7a
EB
1236 kuid_t s_resuid;
1237 kgid_t s_resgid;
ca0faba0
TT
1238 unsigned short s_mount_state;
1239 unsigned short s_pad;
1240 int s_addr_per_block_bits;
1241 int s_desc_per_block_bits;
1242 int s_inode_size;
1243 int s_first_ino;
1244 unsigned int s_inode_readahead_blks;
11013911 1245 unsigned int s_inode_goal;
ca0faba0
TT
1246 spinlock_t s_next_gen_lock;
1247 u32 s_next_generation;
1248 u32 s_hash_seed[4];
1249 int s_def_hash_version;
1250 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
57042651 1251 struct percpu_counter s_freeclusters_counter;
ca0faba0
TT
1252 struct percpu_counter s_freeinodes_counter;
1253 struct percpu_counter s_dirs_counter;
57042651 1254 struct percpu_counter s_dirtyclusters_counter;
ca0faba0
TT
1255 struct blockgroup_lock *s_blockgroup_lock;
1256 struct proc_dir_entry *s_proc;
1257 struct kobject s_kobj;
1258 struct completion s_kobj_unregister;
2c0544b2 1259 struct super_block *s_sb;
ca0faba0
TT
1260
1261 /* Journaling */
ca0faba0
TT
1262 struct journal_s *s_journal;
1263 struct list_head s_orphan;
1264 struct mutex s_orphan_lock;
8f82f840
YY
1265 unsigned long s_resize_flags; /* Flags indicating if there
1266 is a resizer */
ca0faba0
TT
1267 unsigned long s_commit_interval;
1268 u32 s_max_batch_time;
1269 u32 s_min_batch_time;
1270 struct block_device *journal_bdev;
ca0faba0 1271#ifdef CONFIG_QUOTA
a2d4a646 1272 char *s_qf_names[EXT4_MAXQUOTAS]; /* Names of quota files with journalled quota */
ca0faba0
TT
1273 int s_jquota_fmt; /* Format of quota to use */
1274#endif
1275 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
6fd058f7 1276 struct rb_root system_blks;
ca0faba0
TT
1277
1278#ifdef EXTENTS_STATS
1279 /* ext4 extents stats */
1280 unsigned long s_ext_min;
1281 unsigned long s_ext_max;
1282 unsigned long s_depth_max;
1283 spinlock_t s_ext_stats_lock;
1284 unsigned long s_ext_blocks;
1285 unsigned long s_ext_extents;
1286#endif
1287
1288 /* for buddy allocator */
1289 struct ext4_group_info ***s_group_info;
1290 struct inode *s_buddy_cache;
ca0faba0 1291 spinlock_t s_md_lock;
ca0faba0
TT
1292 unsigned short *s_mb_offsets;
1293 unsigned int *s_mb_maxs;
28623c2f 1294 unsigned int s_group_info_size;
ca0faba0
TT
1295
1296 /* tunables */
1297 unsigned long s_stripe;
1298 unsigned int s_mb_stream_request;
1299 unsigned int s_mb_max_to_scan;
1300 unsigned int s_mb_min_to_scan;
1301 unsigned int s_mb_stats;
1302 unsigned int s_mb_order2_reqs;
1303 unsigned int s_mb_group_prealloc;
df981d03 1304 unsigned int s_max_dir_size_kb;
ca0faba0
TT
1305 /* where last allocation was done - for stream allocation */
1306 unsigned long s_mb_last_group;
1307 unsigned long s_mb_last_start;
1308
ca0faba0 1309 /* stats for buddy allocator */
ca0faba0
TT
1310 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1311 atomic_t s_bal_success; /* we found long enough chunks */
1312 atomic_t s_bal_allocated; /* in blocks */
1313 atomic_t s_bal_ex_scanned; /* total extents scanned */
1314 atomic_t s_bal_goals; /* goal hits */
1315 atomic_t s_bal_breaks; /* too long searches */
1316 atomic_t s_bal_2orders; /* 2^order hits */
1317 spinlock_t s_bal_lock;
1318 unsigned long s_mb_buddies_generated;
1319 unsigned long long s_mb_generation_time;
1320 atomic_t s_mb_lost_chunks;
1321 atomic_t s_mb_preallocated;
1322 atomic_t s_mb_discarded;
50797481 1323 atomic_t s_lock_busy;
ca0faba0
TT
1324
1325 /* locality groups */
003cb608 1326 struct ext4_locality_group __percpu *s_locality_groups;
ca0faba0
TT
1327
1328 /* for write statistics */
1329 unsigned long s_sectors_written_start;
1330 u64 s_kbytes_written;
1331
67a5da56
ZL
1332 /* the size of zero-out chunk */
1333 unsigned int s_extent_max_zeroout_kb;
1334
ca0faba0
TT
1335 unsigned int s_log_groups_per_flex;
1336 struct flex_groups *s_flex_groups;
117fff10 1337 ext4_group_t s_flex_groups_allocated;
4c0425ff 1338
2e8fa54e
JK
1339 /* workqueue for reserved extent conversions (buffered io) */
1340 struct workqueue_struct *rsv_conversion_wq;
66e61a9e
TT
1341
1342 /* timer for periodic error stats printing */
1343 struct timer_list s_err_report;
bfff6873
LC
1344
1345 /* Lazy inode table initialization info */
1346 struct ext4_li_request *s_li_request;
1347 /* Wait multiplier for lazy initialization thread */
1348 unsigned int s_li_wait_mult;
c5e06d10
JL
1349
1350 /* Kernel thread for multiple mount protection */
1351 struct task_struct *s_mmp_tsk;
3d56b8d2
TM
1352
1353 /* record the last minlen when FITRIM is called. */
1354 atomic_t s_last_trim_minblks;
0441984a
DW
1355
1356 /* Reference to checksum algorithm driver via cryptoapi */
1357 struct crypto_shash *s_chksum_driver;
a9c47317
DW
1358
1359 /* Precomputed FS UUID checksum for seeding other checksums */
1360 __u32 s_csum_seed;
74cd15cd
ZL
1361
1362 /* Reclaim extents from extent status tree */
1363 struct shrinker s_es_shrinker;
dd475925 1364 struct list_head s_es_list; /* List of inodes with reclaimable extents */
edaa53ca 1365 long s_es_nr_inode;
eb68d0e2 1366 struct ext4_es_stats s_es_stats;
9c191f70 1367 struct mb_cache *s_mb_cache;
edaa53ca 1368 spinlock_t s_es_lock ____cacheline_aligned_in_smp;
efbed4dc
TT
1369
1370 /* Ratelimit ext4 messages. */
1371 struct ratelimit_state s_err_ratelimit_state;
1372 struct ratelimit_state s_warning_ratelimit_state;
1373 struct ratelimit_state s_msg_ratelimit_state;
b30ab0e0
MH
1374
1375#ifdef CONFIG_EXT4_FS_ENCRYPTION
1376 /* Encryption */
1377 uint32_t s_file_encryption_mode;
1378 uint32_t s_dir_encryption_mode;
1379#endif
ca0faba0
TT
1380};
1381
af5bc92d 1382static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
ac27a0ec
DK
1383{
1384 return sb->s_fs_info;
1385}
617ba13b 1386static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
ac27a0ec 1387{
617ba13b 1388 return container_of(inode, struct ext4_inode_info, vfs_inode);
ac27a0ec
DK
1389}
1390
ef7f3835
KS
1391static inline struct timespec ext4_current_time(struct inode *inode)
1392{
1393 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1394 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1395}
1396
617ba13b 1397static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
ac27a0ec 1398{
617ba13b 1399 return ino == EXT4_ROOT_INO ||
7c319d32
AK
1400 ino == EXT4_USR_QUOTA_INO ||
1401 ino == EXT4_GRP_QUOTA_INO ||
393d1d1d 1402 ino == EXT4_BOOT_LOADER_INO ||
617ba13b
MC
1403 ino == EXT4_JOURNAL_INO ||
1404 ino == EXT4_RESIZE_INO ||
1405 (ino >= EXT4_FIRST_INO(sb) &&
1406 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
ac27a0ec 1407}
19f5fb7a 1408
0edeb71d
TM
1409static inline void ext4_set_io_unwritten_flag(struct inode *inode,
1410 struct ext4_io_end *io_end)
1411{
1412 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
1413 io_end->flag |= EXT4_IO_END_UNWRITTEN;
e27f41e1 1414 atomic_inc(&EXT4_I(inode)->i_unwritten);
0edeb71d
TM
1415 }
1416}
1417
f45ee3a1
DM
1418static inline ext4_io_end_t *ext4_inode_aio(struct inode *inode)
1419{
1420 return inode->i_private;
1421}
1422
1423static inline void ext4_inode_aio_set(struct inode *inode, ext4_io_end_t *io)
1424{
1425 inode->i_private = io;
1426}
1427
19f5fb7a
TT
1428/*
1429 * Inode dynamic state flags
1430 */
1431enum {
1432 EXT4_STATE_JDATA, /* journaled data exists */
1433 EXT4_STATE_NEW, /* inode is newly created */
1434 EXT4_STATE_XATTR, /* has in-inode xattrs */
1435 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1436 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1437 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1438 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
14ece102 1439 EXT4_STATE_NEWENTRY, /* File just added to dir */
17335dcc
DM
1440 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read
1441 nolocking */
67cf5b09 1442 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
74d553aa 1443 EXT4_STATE_ORDERED_MODE, /* data=ordered mode */
7869a4a6 1444 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
19f5fb7a
TT
1445};
1446
353eb83c 1447#define EXT4_INODE_BIT_FNS(name, field, offset) \
12e9b892
DM
1448static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1449{ \
353eb83c 1450 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
12e9b892
DM
1451} \
1452static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1453{ \
353eb83c 1454 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
12e9b892
DM
1455} \
1456static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1457{ \
353eb83c 1458 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
19f5fb7a
TT
1459}
1460
3f61c0cc
AD
1461/* Add these declarations here only so that these functions can be
1462 * found by name. Otherwise, they are very hard to locate. */
1463static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1464static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1465static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
353eb83c 1466EXT4_INODE_BIT_FNS(flag, flags, 0)
3f61c0cc
AD
1467
1468/* Add these declarations here only so that these functions can be
1469 * found by name. Otherwise, they are very hard to locate. */
1470static inline int ext4_test_inode_state(struct inode *inode, int bit);
1471static inline void ext4_set_inode_state(struct inode *inode, int bit);
1472static inline void ext4_clear_inode_state(struct inode *inode, int bit);
353eb83c
TT
1473#if (BITS_PER_LONG < 64)
1474EXT4_INODE_BIT_FNS(state, state_flags, 0)
1475
1476static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1477{
1478 (ei)->i_state_flags = 0;
1479}
1480#else
1481EXT4_INODE_BIT_FNS(state, flags, 32)
1482
1483static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1484{
1485 /* We depend on the fact that callers will set i_flags */
1486}
1487#endif
ac27a0ec 1488#else
617ba13b 1489/* Assume that user mode programs are passing in an ext4fs superblock, not
ac27a0ec
DK
1490 * a kernel struct super_block. This will allow us to call the feature-test
1491 * macros from user land. */
617ba13b 1492#define EXT4_SB(sb) (sb)
ac27a0ec
DK
1493#endif
1494
b30ab0e0
MH
1495/*
1496 * Returns true if the inode is inode is encrypted
1497 */
1498static inline int ext4_encrypted_inode(struct inode *inode)
1499{
1500#ifdef CONFIG_EXT4_FS_ENCRYPTION
1501 return ext4_test_inode_flag(inode, EXT4_INODE_ENCRYPT);
1502#else
1503 return 0;
1504#endif
1505}
1506
617ba13b 1507#define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
ac27a0ec
DK
1508
1509/*
1510 * Codes for operating systems
1511 */
617ba13b
MC
1512#define EXT4_OS_LINUX 0
1513#define EXT4_OS_HURD 1
1514#define EXT4_OS_MASIX 2
1515#define EXT4_OS_FREEBSD 3
1516#define EXT4_OS_LITES 4
ac27a0ec
DK
1517
1518/*
1519 * Revision levels
1520 */
617ba13b
MC
1521#define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1522#define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
ac27a0ec 1523
617ba13b
MC
1524#define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1525#define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
ac27a0ec 1526
617ba13b 1527#define EXT4_GOOD_OLD_INODE_SIZE 128
ac27a0ec
DK
1528
1529/*
1530 * Feature set definitions
1531 */
1532
617ba13b 1533#define EXT4_HAS_COMPAT_FEATURE(sb,mask) \
3a06d778 1534 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
617ba13b 1535#define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \
3a06d778 1536 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
617ba13b 1537#define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \
3a06d778 1538 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
617ba13b
MC
1539#define EXT4_SET_COMPAT_FEATURE(sb,mask) \
1540 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1541#define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \
1542 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1543#define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \
1544 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1545#define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \
1546 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1547#define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
1548 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1549#define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \
1550 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1551
1552#define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1553#define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1554#define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1555#define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1556#define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1557#define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1beeef1b 1558#define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
617ba13b
MC
1559
1560#define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1561#define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1562#define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
0fc1b451 1563#define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
717d50e4 1564#define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
f8628a14 1565#define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
ef7f3835 1566#define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
ae812306 1567#define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
281b5995 1568#define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
e6153918
DW
1569/*
1570 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1571 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1572 * all other data structures' checksums. However, the METADATA_CSUM and
1573 * GDT_CSUM bits are mutually exclusive.
1574 */
9b90e5e0 1575#define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
2cb5cc8b 1576#define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000
617ba13b
MC
1577
1578#define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1579#define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1580#define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1581#define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1582#define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
a86c6181 1583#define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
bd81d8ee 1584#define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
0fc1b451 1585#define EXT4_FEATURE_INCOMPAT_MMP 0x0100
ce421581 1586#define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
f710b4b9
TT
1587#define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1588#define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
856cbcf9 1589#define EXT4_FEATURE_INCOMPAT_BG_USE_META_CSUM 0x2000 /* use crc32c for bg */
9b90e5e0 1590#define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
67cf5b09 1591#define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
3edc18d8 1592#define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000
617ba13b 1593
2035e776
TT
1594#define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1595#define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1596 EXT4_FEATURE_INCOMPAT_META_BG)
1597#define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1598 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1599 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1600
1601#define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1602#define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1603 EXT4_FEATURE_INCOMPAT_RECOVER| \
1604 EXT4_FEATURE_INCOMPAT_META_BG)
1605#define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1606 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1607 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1608
617ba13b
MC
1609#define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR
1610#define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1611 EXT4_FEATURE_INCOMPAT_RECOVER| \
a86c6181 1612 EXT4_FEATURE_INCOMPAT_META_BG| \
bd81d8ee 1613 EXT4_FEATURE_INCOMPAT_EXTENTS| \
ce421581 1614 EXT4_FEATURE_INCOMPAT_64BIT| \
c5e06d10 1615 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
f08225d1
TM
1616 EXT4_FEATURE_INCOMPAT_MMP | \
1617 EXT4_FEATURE_INCOMPAT_INLINE_DATA)
617ba13b
MC
1618#define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1619 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
717d50e4 1620 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
f8628a14 1621 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
ef7f3835 1622 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
0fc1b451 1623 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
6f16b606 1624 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
e93376c2 1625 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
7c319d32
AK
1626 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1627 EXT4_FEATURE_RO_COMPAT_QUOTA)
ac27a0ec
DK
1628
1629/*
1630 * Default values for user and/or group using reserved blocks
1631 */
617ba13b
MC
1632#define EXT4_DEF_RESUID 0
1633#define EXT4_DEF_RESGID 0
ac27a0ec 1634
240799cd
TT
1635#define EXT4_DEF_INODE_READAHEAD_BLKS 32
1636
ac27a0ec
DK
1637/*
1638 * Default mount options
1639 */
617ba13b
MC
1640#define EXT4_DEFM_DEBUG 0x0001
1641#define EXT4_DEFM_BSDGROUPS 0x0002
1642#define EXT4_DEFM_XATTR_USER 0x0004
1643#define EXT4_DEFM_ACL 0x0008
1644#define EXT4_DEFM_UID16 0x0010
1645#define EXT4_DEFM_JMODE 0x0060
1646#define EXT4_DEFM_JMODE_DATA 0x0020
1647#define EXT4_DEFM_JMODE_ORDERED 0x0040
1648#define EXT4_DEFM_JMODE_WBACK 0x0060
8b67f04a
TT
1649#define EXT4_DEFM_NOBARRIER 0x0100
1650#define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1651#define EXT4_DEFM_DISCARD 0x0400
1652#define EXT4_DEFM_NODELALLOC 0x0800
ac27a0ec 1653
30773840
TT
1654/*
1655 * Default journal batch times
1656 */
1657#define EXT4_DEF_MIN_BATCH_TIME 0
1658#define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1659
a4912123
TT
1660/*
1661 * Minimum number of groups in a flexgroup before we separate out
1662 * directories into the first block group of a flexgroup
1663 */
1664#define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1665
ac27a0ec
DK
1666/*
1667 * Structure of a directory entry
1668 */
617ba13b 1669#define EXT4_NAME_LEN 255
ac27a0ec 1670
617ba13b 1671struct ext4_dir_entry {
ac27a0ec
DK
1672 __le32 inode; /* Inode number */
1673 __le16 rec_len; /* Directory entry length */
1674 __le16 name_len; /* Name length */
617ba13b 1675 char name[EXT4_NAME_LEN]; /* File name */
ac27a0ec
DK
1676};
1677
1678/*
617ba13b 1679 * The new version of the directory entry. Since EXT4 structures are
ac27a0ec
DK
1680 * stored in intel byte order, and the name_len field could never be
1681 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1682 * file_type field.
1683 */
617ba13b 1684struct ext4_dir_entry_2 {
ac27a0ec
DK
1685 __le32 inode; /* Inode number */
1686 __le16 rec_len; /* Directory entry length */
1687 __u8 name_len; /* Name length */
1688 __u8 file_type;
617ba13b 1689 char name[EXT4_NAME_LEN]; /* File name */
ac27a0ec
DK
1690};
1691
e6153918
DW
1692/*
1693 * This is a bogus directory entry at the end of each leaf block that
1694 * records checksums.
1695 */
1696struct ext4_dir_entry_tail {
1697 __le32 det_reserved_zero1; /* Pretend to be unused */
1698 __le16 det_rec_len; /* 12 */
1699 __u8 det_reserved_zero2; /* Zero name length */
1700 __u8 det_reserved_ft; /* 0xDE, fake file type */
1701 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
1702};
1703
3c47d541
TM
1704#define EXT4_DIRENT_TAIL(block, blocksize) \
1705 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1706 ((blocksize) - \
1707 sizeof(struct ext4_dir_entry_tail))))
1708
ac27a0ec 1709/*
617ba13b 1710 * Ext4 directory file types. Only the low 3 bits are used. The
ac27a0ec
DK
1711 * other bits are reserved for now.
1712 */
617ba13b
MC
1713#define EXT4_FT_UNKNOWN 0
1714#define EXT4_FT_REG_FILE 1
1715#define EXT4_FT_DIR 2
1716#define EXT4_FT_CHRDEV 3
1717#define EXT4_FT_BLKDEV 4
1718#define EXT4_FT_FIFO 5
1719#define EXT4_FT_SOCK 6
1720#define EXT4_FT_SYMLINK 7
ac27a0ec 1721
617ba13b 1722#define EXT4_FT_MAX 8
ac27a0ec 1723
e6153918
DW
1724#define EXT4_FT_DIR_CSUM 0xDE
1725
ac27a0ec 1726/*
617ba13b 1727 * EXT4_DIR_PAD defines the directory entries boundaries
ac27a0ec
DK
1728 *
1729 * NOTE: It must be a multiple of 4
1730 */
617ba13b
MC
1731#define EXT4_DIR_PAD 4
1732#define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1733#define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1734 ~EXT4_DIR_ROUND)
a72d7f83
JK
1735#define EXT4_MAX_REC_LEN ((1<<16)-1)
1736
0cfc9255
ES
1737/*
1738 * If we ever get support for fs block sizes > page_size, we'll need
1739 * to remove the #if statements in the next two functions...
1740 */
1741static inline unsigned int
1742ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1743{
1744 unsigned len = le16_to_cpu(dlen);
1745
1746#if (PAGE_CACHE_SIZE >= 65536)
1747 if (len == EXT4_MAX_REC_LEN || len == 0)
1748 return blocksize;
1749 return (len & 65532) | ((len & 3) << 16);
1750#else
1751 return len;
1752#endif
1753}
1754
1755static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1756{
1757 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1758 BUG();
1759#if (PAGE_CACHE_SIZE >= 65536)
1760 if (len < 65536)
1761 return cpu_to_le16(len);
1762 if (len == blocksize) {
1763 if (blocksize == 65536)
1764 return cpu_to_le16(EXT4_MAX_REC_LEN);
1765 else
1766 return cpu_to_le16(0);
1767 }
1768 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1769#else
1770 return cpu_to_le16(len);
1771#endif
1772}
1773
ac27a0ec
DK
1774/*
1775 * Hash Tree Directory indexing
1776 * (c) Daniel Phillips, 2001
1777 */
1778
4074fe37
ES
1779#define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1780 EXT4_FEATURE_COMPAT_DIR_INDEX) && \
12e9b892 1781 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
617ba13b
MC
1782#define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1783#define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
ac27a0ec
DK
1784
1785/* Legal values for the dx_root hash_version field: */
1786
1787#define DX_HASH_LEGACY 0
1788#define DX_HASH_HALF_MD4 1
1789#define DX_HASH_TEA 2
f99b2589
TT
1790#define DX_HASH_LEGACY_UNSIGNED 3
1791#define DX_HASH_HALF_MD4_UNSIGNED 4
1792#define DX_HASH_TEA_UNSIGNED 5
ac27a0ec 1793
0441984a
DW
1794static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
1795 const void *address, unsigned int length)
1796{
1797 struct {
1798 struct shash_desc shash;
3108b54b 1799 char ctx[4];
0441984a
DW
1800 } desc;
1801 int err;
1802
3108b54b
TT
1803 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
1804
0441984a
DW
1805 desc.shash.tfm = sbi->s_chksum_driver;
1806 desc.shash.flags = 0;
1807 *(u32 *)desc.ctx = crc;
1808
1809 err = crypto_shash_update(&desc.shash, address, length);
1810 BUG_ON(err);
1811
1812 return *(u32 *)desc.ctx;
1813}
1814
ac27a0ec
DK
1815#ifdef __KERNEL__
1816
1817/* hash info structure used by the directory hash */
1818struct dx_hash_info
1819{
1820 u32 hash;
1821 u32 minor_hash;
1822 int hash_version;
1823 u32 *seed;
1824};
1825
d1f5273e
FY
1826
1827/* 32 and 64 bit signed EOF for dx directories */
1828#define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
1829#define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
1830
ac27a0ec
DK
1831
1832/*
617ba13b 1833 * Control parameters used by ext4_htree_next_block
ac27a0ec
DK
1834 */
1835#define HASH_NB_ALWAYS 1
1836
1837
1838/*
1839 * Describe an inode's exact location on disk and in memory
1840 */
617ba13b 1841struct ext4_iloc
ac27a0ec
DK
1842{
1843 struct buffer_head *bh;
1844 unsigned long offset;
fd2d4291 1845 ext4_group_t block_group;
ac27a0ec
DK
1846};
1847
617ba13b 1848static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
ac27a0ec 1849{
617ba13b 1850 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
ac27a0ec
DK
1851}
1852
1853/*
1854 * This structure is stuffed into the struct file's private_data field
1855 * for directories. It is where we put information so that we can do
1856 * readdir operations in hash tree order.
1857 */
1858struct dir_private_info {
1859 struct rb_root root;
1860 struct rb_node *curr_node;
1861 struct fname *extra_fname;
1862 loff_t last_pos;
1863 __u32 curr_hash;
1864 __u32 curr_minor_hash;
1865 __u32 next_hash;
1866};
1867
1868/* calculate the first block number of the group */
617ba13b 1869static inline ext4_fsblk_t
fd2d4291 1870ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
ac27a0ec 1871{
617ba13b
MC
1872 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
1873 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
ac27a0ec
DK
1874}
1875
1876/*
1877 * Special error return code only used by dx_probe() and its callers.
1878 */
36de9286 1879#define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
ac27a0ec 1880
bfff6873
LC
1881/*
1882 * Timeout and state flag for lazy initialization inode thread.
1883 */
1884#define EXT4_DEF_LI_WAIT_MULT 10
1885#define EXT4_DEF_LI_MAX_START_DELAY 5
1886#define EXT4_LAZYINIT_QUIT 0x0001
1887#define EXT4_LAZYINIT_RUNNING 0x0002
1888
1889/*
1890 * Lazy inode table initialization info
1891 */
1892struct ext4_lazy_init {
1893 unsigned long li_state;
bfff6873
LC
1894 struct list_head li_request_list;
1895 struct mutex li_list_mtx;
1896};
1897
1898struct ext4_li_request {
1899 struct super_block *lr_super;
1900 struct ext4_sb_info *lr_sbi;
1901 ext4_group_t lr_next_group;
1902 struct list_head lr_request;
1903 unsigned long lr_next_sched;
1904 unsigned long lr_timeout;
1905};
1906
857ac889
LC
1907struct ext4_features {
1908 struct kobject f_kobj;
1909 struct completion f_kobj_unregister;
1910};
1911
c5e06d10
JL
1912/*
1913 * This structure will be used for multiple mount protection. It will be
1914 * written into the block number saved in the s_mmp_block field in the
1915 * superblock. Programs that check MMP should assume that if
1916 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
1917 * to use the filesystem, regardless of how old the timestamp is.
1918 */
1919#define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
1920#define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
1921#define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
1922#define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
1923
1924struct mmp_struct {
1925 __le32 mmp_magic; /* Magic number for MMP */
1926 __le32 mmp_seq; /* Sequence no. updated periodically */
1927
1928 /*
1929 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
1930 * purposes and do not affect the correctness of the algorithm
1931 */
1932 __le64 mmp_time; /* Time last updated */
1933 char mmp_nodename[64]; /* Node which last updated MMP block */
1934 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
1935
1936 /*
1937 * mmp_check_interval is used to verify if the MMP block has been
1938 * updated on the block device. The value is updated based on the
1939 * maximum time to write the MMP block during an update cycle.
1940 */
1941 __le16 mmp_check_interval;
1942
1943 __le16 mmp_pad1;
e6153918
DW
1944 __le32 mmp_pad2[226];
1945 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
c5e06d10
JL
1946};
1947
1948/* arguments passed to the mmp thread */
1949struct mmpd_data {
1950 struct buffer_head *bh; /* bh from initial read_mmp_block() */
1951 struct super_block *sb; /* super block of the fs */
1952};
1953
1954/*
1955 * Check interval multiplier
1956 * The MMP block is written every update interval and initially checked every
1957 * update interval x the multiplier (the value is then adapted based on the
1958 * write latency). The reason is that writes can be delayed under load and we
1959 * don't want readers to incorrectly assume that the filesystem is no longer
1960 * in use.
1961 */
1962#define EXT4_MMP_CHECK_MULT 2UL
1963
1964/*
1965 * Minimum interval for MMP checking in seconds.
1966 */
1967#define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
1968
1969/*
1970 * Maximum interval for MMP checking in seconds.
1971 */
1972#define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
1973
ac27a0ec
DK
1974/*
1975 * Function prototypes
1976 */
1977
1978/*
1979 * Ok, these declarations are also in <linux/kernel.h> but none of the
617ba13b 1980 * ext4 source programs needs to include it so they are duplicated here.
ac27a0ec 1981 */
8c55e204
DK
1982# define NORET_TYPE /**/
1983# define ATTRIB_NORET __attribute__((noreturn))
1984# define NORET_AND noreturn,
ac27a0ec 1985
498e5f24 1986/* bitmap.c */
f6fb99ca 1987extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
41a246d1
DW
1988void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
1989 struct ext4_group_desc *gdp,
1990 struct buffer_head *bh, int sz);
1991int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
1992 struct ext4_group_desc *gdp,
1993 struct buffer_head *bh, int sz);
fa77dcfa
DW
1994void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
1995 struct ext4_group_desc *gdp,
79f1ba49 1996 struct buffer_head *bh);
fa77dcfa
DW
1997int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
1998 struct ext4_group_desc *gdp,
79f1ba49 1999 struct buffer_head *bh);
498e5f24 2000
ac27a0ec 2001/* balloc.c */
bd86298e
LC
2002extern void ext4_get_group_no_and_offset(struct super_block *sb,
2003 ext4_fsblk_t blocknr,
2004 ext4_group_t *blockgrpp,
2005 ext4_grpblk_t *offsetp);
2006extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2007 ext4_fsblk_t block);
2008
3a5b2ecd
MC
2009extern unsigned int ext4_block_group(struct super_block *sb,
2010 ext4_fsblk_t blocknr);
2011extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2012 ext4_fsblk_t blocknr);
fd2d4291
AM
2013extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2014extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2015 ext4_group_t group);
7061eba7 2016extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
55f020db
AH
2017 ext4_fsblk_t goal,
2018 unsigned int flags,
2019 unsigned long *count,
2020 int *errp);
e7d5f315
TT
2021extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2022 s64 nclusters, unsigned int flags);
5dee5437 2023extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
af5bc92d 2024extern void ext4_check_blocks_bitmap(struct super_block *);
617ba13b 2025extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
fd2d4291 2026 ext4_group_t block_group,
ac27a0ec 2027 struct buffer_head ** bh);
617ba13b 2028extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
813e5727
TT
2029
2030extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2031 ext4_group_t block_group);
2032extern int ext4_wait_block_bitmap(struct super_block *sb,
2033 ext4_group_t block_group,
2034 struct buffer_head *bh);
2035extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2036 ext4_group_t block_group);
cff1dfd7
TT
2037extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2038 ext4_group_t block_group,
2039 struct ext4_group_desc *gdp);
f86186b4 2040ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
ac27a0ec 2041
9bd8212f
MH
2042/* crypto_policy.c */
2043int ext4_is_child_context_consistent_with_parent(struct inode *parent,
2044 struct inode *child);
2045int ext4_inherit_context(struct inode *parent, struct inode *child);
2046void ext4_to_hex(char *dst, char *src, size_t src_size);
2047int ext4_process_policy(const struct ext4_encryption_policy *policy,
2048 struct inode *inode);
2049int ext4_get_policy(struct inode *inode,
2050 struct ext4_encryption_policy *policy);
2051
b30ab0e0
MH
2052/* crypto.c */
2053bool ext4_valid_contents_enc_mode(uint32_t mode);
2054uint32_t ext4_validate_encryption_key_size(uint32_t mode, uint32_t size);
2055extern struct workqueue_struct *ext4_read_workqueue;
2056struct ext4_crypto_ctx *ext4_get_crypto_ctx(struct inode *inode);
2057void ext4_release_crypto_ctx(struct ext4_crypto_ctx *ctx);
2058void ext4_restore_control_page(struct page *data_page);
2059struct page *ext4_encrypt(struct inode *inode,
2060 struct page *plaintext_page);
2061int ext4_decrypt(struct ext4_crypto_ctx *ctx, struct page *page);
2062int ext4_decrypt_one(struct inode *inode, struct page *page);
2063int ext4_encrypted_zeroout(struct inode *inode, struct ext4_extent *ex);
2064
2065#ifdef CONFIG_EXT4_FS_ENCRYPTION
2066int ext4_init_crypto(void);
2067void ext4_exit_crypto(void);
2068static inline int ext4_sb_has_crypto(struct super_block *sb)
2069{
2070 return EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_ENCRYPT);
2071}
2072#else
2073static inline int ext4_init_crypto(void) { return 0; }
2074static inline void ext4_exit_crypto(void) { }
2075static inline int ext4_sb_has_crypto(struct super_block *sb)
2076{
2077 return 0;
2078}
2079#endif
2080
ac27a0ec 2081/* dir.c */
60fd4da3 2082extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
f7c21177 2083 struct file *,
60fd4da3 2084 struct ext4_dir_entry_2 *,
226ba972
TM
2085 struct buffer_head *, char *, int,
2086 unsigned int);
2087#define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
f7c21177 2088 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
226ba972 2089 (de), (bh), (buf), (size), (offset)))
617ba13b 2090extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
ac27a0ec 2091 __u32 minor_hash,
617ba13b
MC
2092 struct ext4_dir_entry_2 *dirent);
2093extern void ext4_htree_free_dir_info(struct dir_private_info *p);
978fef91
TM
2094extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2095 struct buffer_head *bh,
2096 void *buf, int buf_size,
2097 const char *name, int namelen,
2098 struct ext4_dir_entry_2 **dest_de);
2099void ext4_insert_dentry(struct inode *inode,
2100 struct ext4_dir_entry_2 *de,
2101 int buf_size,
2102 const char *name, int namelen);
2103static inline void ext4_update_dx_flag(struct inode *inode)
2104{
2105 if (!EXT4_HAS_COMPAT_FEATURE(inode->i_sb,
2106 EXT4_FEATURE_COMPAT_DIR_INDEX))
2107 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2108}
65d165d9
TM
2109static unsigned char ext4_filetype_table[] = {
2110 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2111};
2112
2113static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2114{
2115 if (!EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE) ||
2116 (filetype >= EXT4_FT_MAX))
2117 return DT_UNKNOWN;
2118
2119 return ext4_filetype_table[filetype];
2120}
40b163f1
DW
2121extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2122 void *buf, int buf_size);
ac27a0ec
DK
2123
2124/* fsync.c */
02c24a82 2125extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
ac27a0ec
DK
2126
2127/* hash.c */
617ba13b 2128extern int ext4fs_dirhash(const char *name, int len, struct
ac27a0ec
DK
2129 dx_hash_info *hinfo);
2130
2131/* ialloc.c */
1139575a
TT
2132extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2133 const struct qstr *qstr, __u32 goal,
2134 uid_t *owner, int handle_type,
2135 unsigned int line_no, int nblocks);
2136
2137#define ext4_new_inode(handle, dir, mode, qstr, goal, owner) \
2138 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2139 0, 0, 0)
2140#define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2141 type, nblocks) \
2142 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2143 (type), __LINE__, (nblocks))
2144
2145
af5bc92d
TT
2146extern void ext4_free_inode(handle_t *, struct inode *);
2147extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2148extern unsigned long ext4_count_free_inodes(struct super_block *);
2149extern unsigned long ext4_count_dirs(struct super_block *);
2150extern void ext4_check_inodes_bitmap(struct super_block *);
61d08673 2151extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
bfff6873
LC
2152extern int ext4_init_inode_table(struct super_block *sb,
2153 ext4_group_t group, int barrier);
813e5727 2154extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
ac27a0ec 2155
c9de560d
AT
2156/* mballoc.c */
2157extern long ext4_mb_stats;
2158extern long ext4_mb_max_to_scan;
9d99012f 2159extern int ext4_mb_init(struct super_block *);
c9de560d
AT
2160extern int ext4_mb_release(struct super_block *);
2161extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2162 struct ext4_allocation_request *, int *);
2163extern int ext4_mb_reserve_blocks(struct super_block *, int);
c2ea3fde 2164extern void ext4_discard_preallocations(struct inode *);
5dabfc78
TT
2165extern int __init ext4_init_mballoc(void);
2166extern void ext4_exit_mballoc(void);
44338711 2167extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
e6362609
TT
2168 struct buffer_head *bh, ext4_fsblk_t block,
2169 unsigned long count, int flags);
28623c2f
TT
2170extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2171 ext4_group_t ngroups);
920313a7 2172extern int ext4_mb_add_groupinfo(struct super_block *sb,
5f21b0e6 2173 ext4_group_t i, struct ext4_group_desc *desc);
cc7365df 2174extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2846e820 2175 ext4_fsblk_t block, unsigned long count);
7360d173
LC
2176extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2177
ac27a0ec 2178/* inode.c */
10560082 2179struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
1c215028 2180struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
f19d5870
TM
2181int ext4_get_block_write(struct inode *inode, sector_t iblock,
2182 struct buffer_head *bh_result, int create);
6873fa0d
ES
2183int ext4_get_block(struct inode *inode, sector_t iblock,
2184 struct buffer_head *bh_result, int create);
9c3569b5
TM
2185int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2186 struct buffer_head *bh, int create);
f19d5870
TM
2187int ext4_walk_page_buffers(handle_t *handle,
2188 struct buffer_head *head,
2189 unsigned from,
2190 unsigned to,
2191 int *partial,
2192 int (*fn)(handle_t *handle,
2193 struct buffer_head *bh));
2194int do_journal_get_write_access(handle_t *handle,
2195 struct buffer_head *bh);
9c3569b5
TM
2196#define FALL_BACK_TO_NONDELALLOC 1
2197#define CONVERT_INLINE_DATA 2
ac27a0ec 2198
1d1fe1ee 2199extern struct inode *ext4_iget(struct super_block *, unsigned long);
f4bb2981 2200extern struct inode *ext4_iget_normal(struct super_block *, unsigned long);
a9185b41 2201extern int ext4_write_inode(struct inode *, struct writeback_control *);
af5bc92d 2202extern int ext4_setattr(struct dentry *, struct iattr *);
3e3398a0
MC
2203extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
2204 struct kstat *stat);
0930fcc1
AV
2205extern void ext4_evict_inode(struct inode *);
2206extern void ext4_clear_inode(struct inode *);
af5bc92d 2207extern int ext4_sync_inode(handle_t *, struct inode *);
aa385729 2208extern void ext4_dirty_inode(struct inode *, int);
617ba13b
MC
2209extern int ext4_change_inode_journal_flag(struct inode *, int);
2210extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
a361293f 2211extern int ext4_inode_attach_jinode(struct inode *inode);
91ef4caf 2212extern int ext4_can_truncate(struct inode *inode);
af5bc92d 2213extern void ext4_truncate(struct inode *);
aeb2817a 2214extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
487caeef 2215extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
617ba13b 2216extern void ext4_set_inode_flags(struct inode *);
ff9ddf7e 2217extern void ext4_get_inode_flags(struct ext4_inode_info *);
ccd2506b 2218extern int ext4_alloc_da_blocks(struct inode *inode);
617ba13b 2219extern void ext4_set_aops(struct inode *inode);
a86c6181 2220extern int ext4_writepage_trans_blocks(struct inode *);
f3bd1f3f 2221extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
a87dd18c
LC
2222extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2223 loff_t lstart, loff_t lend);
c2ec175c 2224extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
a9e7f447 2225extern qsize_t *ext4_get_reserved_space(struct inode *inode);
5f634d06
AK
2226extern void ext4_da_update_reserve_space(struct inode *inode,
2227 int used, int quota_claim);
dae1e52c
AG
2228
2229/* indirect.c */
2230extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2231 struct ext4_map_blocks *map, int flags);
2232extern ssize_t ext4_ind_direct_IO(int rw, struct kiocb *iocb,
16b1f05d 2233 struct iov_iter *iter, loff_t offset);
dae1e52c 2234extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
fa55a0ed 2235extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
819c4920 2236extern void ext4_ind_truncate(handle_t *, struct inode *inode);
4f579ae7
LC
2237extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2238 ext4_lblk_t start, ext4_lblk_t end);
ff9893dc 2239
ac27a0ec 2240/* ioctl.c */
5cdd7b2d 2241extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
af5bc92d 2242extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
ac27a0ec 2243
c14c6fd5 2244/* migrate.c */
2a43a878 2245extern int ext4_ext_migrate(struct inode *);
0d14b098 2246extern int ext4_ind_migrate(struct inode *inode);
3d0518f4 2247
ac27a0ec 2248/* namei.c */
b0336e8d
DW
2249extern int ext4_dirent_csum_verify(struct inode *inode,
2250 struct ext4_dir_entry *dirent);
617ba13b
MC
2251extern int ext4_orphan_add(handle_t *, struct inode *);
2252extern int ext4_orphan_del(handle_t *, struct inode *);
2253extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
ac27a0ec 2254 __u32 start_minor_hash, __u32 *next_hash);
7335cd3b
TM
2255extern int search_dir(struct buffer_head *bh,
2256 char *search_buf,
2257 int buf_size,
2258 struct inode *dir,
2259 const struct qstr *d_name,
2260 unsigned int offset,
2261 struct ext4_dir_entry_2 **res_dir);
05019a9e
TM
2262extern int ext4_generic_delete_entry(handle_t *handle,
2263 struct inode *dir,
2264 struct ext4_dir_entry_2 *de_del,
2265 struct buffer_head *bh,
2266 void *entry_buf,
2267 int buf_size,
2268 int csum_size);
e875a2dd 2269extern int ext4_empty_dir(struct inode *inode);
ac27a0ec
DK
2270
2271/* resize.c */
617ba13b
MC
2272extern int ext4_group_add(struct super_block *sb,
2273 struct ext4_new_group_data *input);
2274extern int ext4_group_extend(struct super_block *sb,
2275 struct ext4_super_block *es,
2276 ext4_fsblk_t n_blocks_count);
19c5246d 2277extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
ac27a0ec
DK
2278
2279/* super.c */
952fc18e 2280extern int ext4_calculate_overhead(struct super_block *sb);
06db49e6 2281extern void ext4_superblock_csum_set(struct super_block *sb);
9933fc0a
TT
2282extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2283extern void *ext4_kvzalloc(size_t size, gfp_t flags);
117fff10
TT
2284extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2285 ext4_group_t ngroup);
722887dd
TT
2286extern const char *ext4_decode_error(struct super_block *sb, int errno,
2287 char nbuf[16]);
e7c96e8e 2288
b9075fa9
JP
2289extern __printf(4, 5)
2290void __ext4_error(struct super_block *, const char *, unsigned int,
2291 const char *, ...);
b9075fa9 2292extern __printf(5, 6)
e7c96e8e 2293void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
b9075fa9
JP
2294 const char *, ...);
2295extern __printf(5, 6)
e7c96e8e 2296void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
b9075fa9 2297 const char *, ...);
c398eda0
TT
2298extern void __ext4_std_error(struct super_block *, const char *,
2299 unsigned int, int);
b9075fa9
JP
2300extern __printf(4, 5)
2301void __ext4_abort(struct super_block *, const char *, unsigned int,
2302 const char *, ...);
b9075fa9
JP
2303extern __printf(4, 5)
2304void __ext4_warning(struct super_block *, const char *, unsigned int,
2305 const char *, ...);
b9075fa9 2306extern __printf(3, 4)
e7c96e8e 2307void __ext4_msg(struct super_block *, const char *, const char *, ...);
c5e06d10
JL
2308extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2309 const char *, unsigned int, const char *);
b9075fa9
JP
2310extern __printf(7, 8)
2311void __ext4_grp_locked_error(const char *, unsigned int,
2312 struct super_block *, ext4_group_t,
2313 unsigned long, ext4_fsblk_t,
2314 const char *, ...);
e7c96e8e
JP
2315
2316#ifdef CONFIG_PRINTK
2317
2318#define ext4_error_inode(inode, func, line, block, fmt, ...) \
2319 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__)
2320#define ext4_error_file(file, func, line, block, fmt, ...) \
2321 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2322#define ext4_error(sb, fmt, ...) \
2323 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2324#define ext4_abort(sb, fmt, ...) \
2325 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2326#define ext4_warning(sb, fmt, ...) \
2327 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2328#define ext4_msg(sb, level, fmt, ...) \
2329 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2330#define dump_mmp_msg(sb, mmp, msg) \
2331 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2332#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2333 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2334 fmt, ##__VA_ARGS__)
2335
2336#else
2337
2338#define ext4_error_inode(inode, func, line, block, fmt, ...) \
2339do { \
2340 no_printk(fmt, ##__VA_ARGS__); \
2341 __ext4_error_inode(inode, "", 0, block, " "); \
2342} while (0)
2343#define ext4_error_file(file, func, line, block, fmt, ...) \
2344do { \
2345 no_printk(fmt, ##__VA_ARGS__); \
2346 __ext4_error_file(file, "", 0, block, " "); \
2347} while (0)
2348#define ext4_error(sb, fmt, ...) \
2349do { \
2350 no_printk(fmt, ##__VA_ARGS__); \
2351 __ext4_error(sb, "", 0, " "); \
2352} while (0)
2353#define ext4_abort(sb, fmt, ...) \
2354do { \
2355 no_printk(fmt, ##__VA_ARGS__); \
2356 __ext4_abort(sb, "", 0, " "); \
2357} while (0)
2358#define ext4_warning(sb, fmt, ...) \
2359do { \
2360 no_printk(fmt, ##__VA_ARGS__); \
2361 __ext4_warning(sb, "", 0, " "); \
2362} while (0)
2363#define ext4_msg(sb, level, fmt, ...) \
2364do { \
2365 no_printk(fmt, ##__VA_ARGS__); \
2366 __ext4_msg(sb, "", " "); \
2367} while (0)
2368#define dump_mmp_msg(sb, mmp, msg) \
2369 __dump_mmp_msg(sb, mmp, "", 0, "")
2370#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2371do { \
2372 no_printk(fmt, ##__VA_ARGS__); \
2373 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
2374} while (0)
2375
2376#endif
2377
af5bc92d 2378extern void ext4_update_dynamic_rev(struct super_block *sb);
99e6f829
AK
2379extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2380 __u32 compat);
2381extern int ext4_update_rocompat_feature(handle_t *handle,
2382 struct super_block *sb, __u32 rocompat);
2383extern int ext4_update_incompat_feature(handle_t *handle,
2384 struct super_block *sb, __u32 incompat);
8fadc143
AR
2385extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2386 struct ext4_group_desc *bg);
2387extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2388 struct ext4_group_desc *bg);
2389extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2390 struct ext4_group_desc *bg);
021b65bb
TT
2391extern __u32 ext4_free_group_clusters(struct super_block *sb,
2392 struct ext4_group_desc *bg);
560671a0
AK
2393extern __u32 ext4_free_inodes_count(struct super_block *sb,
2394 struct ext4_group_desc *bg);
2395extern __u32 ext4_used_dirs_count(struct super_block *sb,
2396 struct ext4_group_desc *bg);
2397extern __u32 ext4_itable_unused_count(struct super_block *sb,
2398 struct ext4_group_desc *bg);
8fadc143
AR
2399extern void ext4_block_bitmap_set(struct super_block *sb,
2400 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2401extern void ext4_inode_bitmap_set(struct super_block *sb,
2402 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2403extern void ext4_inode_table_set(struct super_block *sb,
2404 struct ext4_group_desc *bg, ext4_fsblk_t blk);
021b65bb
TT
2405extern void ext4_free_group_clusters_set(struct super_block *sb,
2406 struct ext4_group_desc *bg,
2407 __u32 count);
560671a0
AK
2408extern void ext4_free_inodes_set(struct super_block *sb,
2409 struct ext4_group_desc *bg, __u32 count);
2410extern void ext4_used_dirs_set(struct super_block *sb,
2411 struct ext4_group_desc *bg, __u32 count);
2412extern void ext4_itable_unused_set(struct super_block *sb,
2413 struct ext4_group_desc *bg, __u32 count);
feb0ab32 2414extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
bb23c20a 2415 struct ext4_group_desc *gdp);
feb0ab32
DW
2416extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2417 struct ext4_group_desc *gdp);
7f511862
TT
2418extern int ext4_register_li_request(struct super_block *sb,
2419 ext4_group_t first_not_zeroed);
bd81d8ee 2420
41a246d1
DW
2421static inline int ext4_has_group_desc_csum(struct super_block *sb)
2422{
2423 return EXT4_HAS_RO_COMPAT_FEATURE(sb,
813d32f9
DW
2424 EXT4_FEATURE_RO_COMPAT_GDT_CSUM) ||
2425 (EXT4_SB(sb)->s_chksum_driver != NULL);
41a246d1 2426}
bd81d8ee 2427
9aa5d32b
DM
2428static inline int ext4_has_metadata_csum(struct super_block *sb)
2429{
2430 WARN_ON_ONCE(EXT4_HAS_RO_COMPAT_FEATURE(sb,
2431 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM) &&
2432 !EXT4_SB(sb)->s_chksum_driver);
2433
2434 return (EXT4_SB(sb)->s_chksum_driver != NULL);
2435}
bd81d8ee
LV
2436static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2437{
2438 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
6bc9feff 2439 le32_to_cpu(es->s_blocks_count_lo);
bd81d8ee
LV
2440}
2441
2442static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2443{
2444 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
308ba3ec 2445 le32_to_cpu(es->s_r_blocks_count_lo);
bd81d8ee
LV
2446}
2447
2448static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2449{
2450 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
308ba3ec 2451 le32_to_cpu(es->s_free_blocks_count_lo);
bd81d8ee
LV
2452}
2453
2454static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2455 ext4_fsblk_t blk)
2456{
6bc9feff 2457 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
bd81d8ee
LV
2458 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2459}
2460
2461static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2462 ext4_fsblk_t blk)
2463{
308ba3ec 2464 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
bd81d8ee
LV
2465 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2466}
2467
2468static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2469 ext4_fsblk_t blk)
2470{
308ba3ec 2471 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
bd81d8ee
LV
2472 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2473}
2474
a48380f7
AK
2475static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
2476{
06a279d6
TT
2477 if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2478 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2479 le32_to_cpu(raw_inode->i_size_lo);
2480 else
2481 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
a48380f7 2482}
bd81d8ee 2483
a48380f7
AK
2484static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2485{
2486 raw_inode->i_size_lo = cpu_to_le32(i_size);
2487 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2488}
ac27a0ec 2489
c9de560d
AT
2490static inline
2491struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
af5bc92d 2492 ext4_group_t group)
c9de560d
AT
2493{
2494 struct ext4_group_info ***grp_info;
2495 long indexv, indexh;
2f2e09eb 2496 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
c9de560d
AT
2497 grp_info = EXT4_SB(sb)->s_group_info;
2498 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2499 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2500 return grp_info[indexv][indexh];
2501}
2502
8df9675f
TT
2503/*
2504 * Reading s_groups_count requires using smp_rmb() afterwards. See
2505 * the locking protocol documented in the comments of ext4_group_add()
2506 * in resize.c
2507 */
2508static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2509{
2510 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
2511
2512 smp_rmb();
2513 return ngroups;
2514}
c9de560d 2515
772cb7c8
JS
2516static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2517 ext4_group_t block_group)
2518{
2519 return block_group >> sbi->s_log_groups_per_flex;
2520}
2521
2522static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2523{
2524 return 1 << sbi->s_log_groups_per_flex;
2525}
2526
617ba13b 2527#define ext4_std_error(sb, errno) \
ac27a0ec
DK
2528do { \
2529 if ((errno)) \
c398eda0 2530 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
ac27a0ec
DK
2531} while (0)
2532
a30d542a 2533#ifdef CONFIG_SMP
df55c99d
TT
2534/* Each CPU can accumulate percpu_counter_batch clusters in their local
2535 * counters. So we need to make sure we have free clusters more
179f7ebf 2536 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
a30d542a 2537 */
df55c99d 2538#define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
a30d542a 2539#else
df55c99d 2540#define EXT4_FREECLUSTERS_WATERMARK 0
a30d542a
AK
2541#endif
2542
90e775b7 2543/* Update i_disksize. Requires i_mutex to avoid races with truncate */
cf17fea6
AK
2544static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2545{
90e775b7
JK
2546 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
2547 !mutex_is_locked(&inode->i_mutex));
2548 down_write(&EXT4_I(inode)->i_data_sem);
2549 if (newsize > EXT4_I(inode)->i_disksize)
2550 EXT4_I(inode)->i_disksize = newsize;
2551 up_write(&EXT4_I(inode)->i_data_sem);
2552}
2553
4631dbf6
DM
2554/* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
2555static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
2556{
2557 int changed = 0;
2558
2559 if (newsize > inode->i_size) {
2560 i_size_write(inode, newsize);
2561 changed = 1;
2562 }
2563 if (newsize > EXT4_I(inode)->i_disksize) {
2564 ext4_update_i_disksize(inode, newsize);
2565 changed |= 2;
2566 }
2567 return changed;
2568}
2569
5d1b1b3f
AK
2570struct ext4_group_info {
2571 unsigned long bb_state;
2572 struct rb_root bb_free_root;
a36b4498
ES
2573 ext4_grpblk_t bb_first_free; /* first free block */
2574 ext4_grpblk_t bb_free; /* total free blocks */
2575 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
8a57d9d6 2576 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
5d1b1b3f
AK
2577 struct list_head bb_prealloc_list;
2578#ifdef DOUBLE_CHECK
2579 void *bb_bitmap;
2580#endif
2581 struct rw_semaphore alloc_sem;
a36b4498
ES
2582 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
2583 * regions, index is order.
2584 * bb_counters[3] = 5 means
2585 * 5 free 8-block regions. */
5d1b1b3f
AK
2586};
2587
3d56b8d2
TM
2588#define EXT4_GROUP_INFO_NEED_INIT_BIT 0
2589#define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
163a203d 2590#define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
87a39389 2591#define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
5d1b1b3f
AK
2592
2593#define EXT4_MB_GRP_NEED_INIT(grp) \
2594 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
163a203d
DW
2595#define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
2596 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
87a39389
DW
2597#define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
2598 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
5d1b1b3f 2599
3d56b8d2
TM
2600#define EXT4_MB_GRP_WAS_TRIMMED(grp) \
2601 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2602#define EXT4_MB_GRP_SET_TRIMMED(grp) \
2603 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2604#define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
2605 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2606
50797481
TT
2607#define EXT4_MAX_CONTENTION 8
2608#define EXT4_CONTENTION_THRESHOLD 2
2609
955ce5f5
AK
2610static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2611 ext4_group_t group)
5d1b1b3f 2612{
955ce5f5
AK
2613 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2614}
5d1b1b3f 2615
50797481
TT
2616/*
2617 * Returns true if the filesystem is busy enough that attempts to
2618 * access the block group locks has run into contention.
2619 */
2620static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2621{
2622 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2623}
2624
955ce5f5
AK
2625static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2626{
50797481
TT
2627 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2628 if (spin_trylock(lock))
2629 /*
2630 * We're able to grab the lock right away, so drop the
2631 * lock contention counter.
2632 */
2633 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2634 else {
2635 /*
2636 * The lock is busy, so bump the contention counter,
2637 * and then wait on the spin lock.
2638 */
2639 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2640 EXT4_MAX_CONTENTION);
2641 spin_lock(lock);
2642 }
5d1b1b3f
AK
2643}
2644
2645static inline void ext4_unlock_group(struct super_block *sb,
2646 ext4_group_t group)
2647{
955ce5f5 2648 spin_unlock(ext4_group_lock_ptr(sb, group));
5d1b1b3f
AK
2649}
2650
1f7d1e77
TT
2651/*
2652 * Block validity checking
2653 */
2654#define ext4_check_indirect_blockref(inode, bh) \
2655 ext4_check_blockref(__func__, __LINE__, inode, \
2656 (__le32 *)(bh)->b_data, \
2657 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2658
2659#define ext4_ind_check_inode(inode) \
2660 ext4_check_blockref(__func__, __LINE__, inode, \
2661 EXT4_I(inode)->i_data, \
2662 EXT4_NDIR_BLOCKS)
2663
ac27a0ec
DK
2664/*
2665 * Inodes and files operations
2666 */
2667
2668/* dir.c */
617ba13b 2669extern const struct file_operations ext4_dir_operations;
ac27a0ec
DK
2670
2671/* file.c */
c5ef1c42 2672extern const struct inode_operations ext4_file_inode_operations;
617ba13b 2673extern const struct file_operations ext4_file_operations;
923ae0ff 2674extern const struct file_operations ext4_dax_file_operations;
e0d10bfa 2675extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
ac27a0ec 2676
95eaefbd 2677/* inline.c */
95eaefbd
TT
2678extern int ext4_get_max_inline_size(struct inode *inode);
2679extern int ext4_find_inline_data_nolock(struct inode *inode);
95eaefbd
TT
2680extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
2681 unsigned int len);
2682extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
2683
2684extern int ext4_readpage_inline(struct inode *inode, struct page *page);
2685extern int ext4_try_to_write_inline_data(struct address_space *mapping,
2686 struct inode *inode,
2687 loff_t pos, unsigned len,
2688 unsigned flags,
2689 struct page **pagep);
2690extern int ext4_write_inline_data_end(struct inode *inode,
2691 loff_t pos, unsigned len,
2692 unsigned copied,
2693 struct page *page);
2694extern struct buffer_head *
2695ext4_journalled_write_inline_data(struct inode *inode,
2696 unsigned len,
2697 struct page *page);
2698extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
2699 struct inode *inode,
2700 loff_t pos, unsigned len,
2701 unsigned flags,
2702 struct page **pagep,
2703 void **fsdata);
2704extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
2705 unsigned len, unsigned copied,
2706 struct page *page);
2707extern int ext4_try_add_inline_entry(handle_t *handle, struct dentry *dentry,
2708 struct inode *inode);
2709extern int ext4_try_create_inline_dir(handle_t *handle,
2710 struct inode *parent,
2711 struct inode *inode);
2712extern int ext4_read_inline_dir(struct file *filp,
725bebb2 2713 struct dir_context *ctx,
95eaefbd 2714 int *has_inline_data);
8af0f082
TM
2715extern int htree_inlinedir_to_tree(struct file *dir_file,
2716 struct inode *dir, ext4_lblk_t block,
2717 struct dx_hash_info *hinfo,
2718 __u32 start_hash, __u32 start_minor_hash,
2719 int *has_inline_data);
95eaefbd
TT
2720extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
2721 const struct qstr *d_name,
2722 struct ext4_dir_entry_2 **res_dir,
2723 int *has_inline_data);
2724extern int ext4_delete_inline_entry(handle_t *handle,
2725 struct inode *dir,
2726 struct ext4_dir_entry_2 *de_del,
2727 struct buffer_head *bh,
2728 int *has_inline_data);
2729extern int empty_inline_dir(struct inode *dir, int *has_inline_data);
2730extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
2731 struct ext4_dir_entry_2 **parent_de,
2732 int *retval);
2733extern int ext4_inline_data_fiemap(struct inode *inode,
2734 struct fiemap_extent_info *fieinfo,
d952d69e 2735 int *has_inline, __u64 start, __u64 len);
95eaefbd
TT
2736extern int ext4_try_to_evict_inline_data(handle_t *handle,
2737 struct inode *inode,
2738 int needed);
2739extern void ext4_inline_data_truncate(struct inode *inode, int *has_inline);
2740
2741extern int ext4_convert_inline_data(struct inode *inode);
2742
83447ccb
ZL
2743static inline int ext4_has_inline_data(struct inode *inode)
2744{
2745 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
2746 EXT4_I(inode)->i_inline_off;
2747}
2748
ac27a0ec 2749/* namei.c */
c5ef1c42
AV
2750extern const struct inode_operations ext4_dir_inode_operations;
2751extern const struct inode_operations ext4_special_inode_operations;
596397b7 2752extern struct dentry *ext4_get_parent(struct dentry *child);
a774f9c2
TM
2753extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
2754 struct ext4_dir_entry_2 *de,
2755 int blocksize, int csum_size,
2756 unsigned int parent_ino, int dotdot_real_len);
3c47d541
TM
2757extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t,
2758 unsigned int blocksize);
2759extern int ext4_handle_dirty_dirent_node(handle_t *handle,
2760 struct inode *inode,
2761 struct buffer_head *bh);
8af0f082
TM
2762#define S_SHIFT 12
2763static unsigned char ext4_type_by_mode[S_IFMT >> S_SHIFT] = {
2764 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
2765 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
2766 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
2767 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
2768 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
2769 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
2770 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
2771};
2772
2773static inline void ext4_set_de_type(struct super_block *sb,
2774 struct ext4_dir_entry_2 *de,
2775 umode_t mode) {
2776 if (EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE))
2777 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
2778}
2779
f64e02fe
TT
2780/* readpages.c */
2781extern int ext4_mpage_readpages(struct address_space *mapping,
2782 struct list_head *pages, struct page *page,
2783 unsigned nr_pages);
ac27a0ec
DK
2784
2785/* symlink.c */
c5ef1c42
AV
2786extern const struct inode_operations ext4_symlink_inode_operations;
2787extern const struct inode_operations ext4_fast_symlink_inode_operations;
ac27a0ec 2788
6fd058f7
TT
2789/* block_validity */
2790extern void ext4_release_system_zone(struct super_block *sb);
2791extern int ext4_setup_system_zone(struct super_block *sb);
5dabfc78
TT
2792extern int __init ext4_init_system_zone(void);
2793extern void ext4_exit_system_zone(void);
6fd058f7
TT
2794extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
2795 ext4_fsblk_t start_blk,
2796 unsigned int count);
1f7d1e77
TT
2797extern int ext4_check_blockref(const char *, unsigned int,
2798 struct inode *, __le32 *, unsigned int);
6fd058f7 2799
a86c6181 2800/* extents.c */
4a092d73
TT
2801struct ext4_ext_path;
2802struct ext4_extent;
2803
d7b2a00c
ZL
2804/*
2805 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
2806 * __le32.
2807 */
2808#define EXT_MAX_BLOCKS 0xffffffff
2809
a86c6181
AT
2810extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
2811extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
fffb2739 2812extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
e35fd660
TT
2813extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
2814 struct ext4_map_blocks *map, int flags);
819c4920 2815extern void ext4_ext_truncate(handle_t *, struct inode *);
26a4c0c6
TT
2816extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
2817 ext4_lblk_t end);
a86c6181
AT
2818extern void ext4_ext_init(struct super_block *);
2819extern void ext4_ext_release(struct super_block *);
2fe17c10 2820extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
a2df2a63 2821 loff_t len);
6b523df4
JK
2822extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
2823 loff_t offset, ssize_t len);
e35fd660
TT
2824extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
2825 struct ext4_map_blocks *map, int flags);
4a092d73
TT
2826extern int ext4_ext_calc_metadata_amount(struct inode *inode,
2827 ext4_lblk_t lblocks);
2828extern int ext4_extent_tree_init(handle_t *, struct inode *);
2829extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
2830 int num,
2831 struct ext4_ext_path *path);
2832extern int ext4_can_extents_be_merged(struct inode *inode,
2833 struct ext4_extent *ex1,
2834 struct ext4_extent *ex2);
2835extern int ext4_ext_insert_extent(handle_t *, struct inode *,
dfe50809 2836 struct ext4_ext_path **,
4a092d73 2837 struct ext4_extent *, int);
ed8a1a76
TT
2838extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
2839 struct ext4_ext_path **,
2840 int flags);
4a092d73
TT
2841extern void ext4_ext_drop_refs(struct ext4_ext_path *);
2842extern int ext4_ext_check_inode(struct inode *inode);
f7fec032
ZL
2843extern int ext4_find_delalloc_range(struct inode *inode,
2844 ext4_lblk_t lblk_start,
2845 ext4_lblk_t lblk_end);
4a092d73 2846extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk);
fcf6b1b7 2847extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3a06d778
AK
2848extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
2849 __u64 start, __u64 len);
7869a4a6 2850extern int ext4_ext_precache(struct inode *inode);
9eb79482 2851extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len);
fcf6b1b7
DM
2852extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
2853 struct inode *inode2, ext4_lblk_t lblk1,
2854 ext4_lblk_t lblk2, ext4_lblk_t count,
2855 int mark_unwritten,int *err);
4a092d73 2856
748de673 2857/* move_extent.c */
393d1d1d
DTB
2858extern void ext4_double_down_write_data_sem(struct inode *first,
2859 struct inode *second);
2860extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
2861 struct inode *donor_inode);
748de673
AF
2862extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
2863 __u64 start_orig, __u64 start_donor,
2864 __u64 len, __u64 *moved_len);
2865
bd2d0210 2866/* page-io.c */
5dabfc78
TT
2867extern int __init ext4_init_pageio(void);
2868extern void ext4_exit_pageio(void);
bd2d0210 2869extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
97a851ed
JK
2870extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
2871extern int ext4_put_io_end(ext4_io_end_t *io_end);
2872extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
2873extern void ext4_io_submit_init(struct ext4_io_submit *io,
2874 struct writeback_control *wbc);
2e8fa54e 2875extern void ext4_end_io_rsv_work(struct work_struct *work);
bd2d0210
TT
2876extern void ext4_io_submit(struct ext4_io_submit *io);
2877extern int ext4_bio_write_page(struct ext4_io_submit *io,
2878 struct page *page,
2879 int len,
1c8349a1
NJ
2880 struct writeback_control *wbc,
2881 bool keep_towrite);
3a06d778 2882
c5e06d10
JL
2883/* mmp.c */
2884extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
2885
2ccb5fb9 2886/*
48fc7f7e 2887 * Add new method to test whether block and inode bitmaps are properly
2ccb5fb9
AK
2888 * initialized. With uninit_bg reading the block from disk is not enough
2889 * to mark the bitmap uptodate. We need to also zero-out the bitmap
2890 */
2891#define BH_BITMAP_UPTODATE BH_JBDPrivateStart
2892
2893static inline int bitmap_uptodate(struct buffer_head *bh)
2894{
2895 return (buffer_uptodate(bh) &&
2896 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
2897}
2898static inline void set_bitmap_uptodate(struct buffer_head *bh)
2899{
2900 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
2901}
2902
17335dcc
DM
2903/*
2904 * Disable DIO read nolock optimization, so new dioreaders will be forced
2905 * to grab i_mutex
2906 */
2907static inline void ext4_inode_block_unlocked_dio(struct inode *inode)
2908{
2909 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
2910 smp_mb();
2911}
2912static inline void ext4_inode_resume_unlocked_dio(struct inode *inode)
2913{
2914 smp_mb();
2915 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
2916}
2917
731eb1a0
AM
2918#define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
2919
e9e3bcec
ES
2920/* For ioend & aio unwritten conversion wait queues */
2921#define EXT4_WQ_HASH_SZ 37
2922#define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
2923 EXT4_WQ_HASH_SZ])
2924#define ext4_aio_mutex(v) (&ext4__aio_mutex[((unsigned long)(v)) %\
2925 EXT4_WQ_HASH_SZ])
2926extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
2927extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ];
2928
8f82f840
YY
2929#define EXT4_RESIZING 0
2930extern int ext4_resize_begin(struct super_block *sb);
2931extern void ext4_resize_end(struct super_block *sb);
2932
ac27a0ec
DK
2933#endif /* __KERNEL__ */
2934
3dcf5451 2935#endif /* _EXT4_H */