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