]> git.ipfire.org Git - people/ms/linux.git/blame - fs/ext4/ext4.h
ext4: convert ext4_getblk() to use the ERR_PTR convention
[people/ms/linux.git] / fs / ext4 / ext4.h
CommitLineData
ac27a0ec 1/*
3dcf5451 2 * ext4.h
ac27a0ec
DK
3 *
4 * Copyright (C) 1992, 1993, 1994, 1995
5 * Remy Card (card@masi.ibp.fr)
6 * Laboratoire MASI - Institut Blaise Pascal
7 * Universite Pierre et Marie Curie (Paris VI)
8 *
9 * from
10 *
11 * linux/include/linux/minix_fs.h
12 *
13 * Copyright (C) 1991, 1992 Linus Torvalds
14 */
15
3dcf5451
CH
16#ifndef _EXT4_H
17#define _EXT4_H
ac27a0ec
DK
18
19#include <linux/types.h>
3a5b2ecd 20#include <linux/blkdev.h>
ac27a0ec 21#include <linux/magic.h>
2ccb5fb9 22#include <linux/jbd2.h>
60e58e0f 23#include <linux/quota.h>
d444c3c3
TT
24#include <linux/rwsem.h>
25#include <linux/rbtree.h>
26#include <linux/seqlock.h>
27#include <linux/mutex.h>
ca0faba0
TT
28#include <linux/timer.h>
29#include <linux/wait.h>
30#include <linux/blockgroup_lock.h>
31#include <linux/percpu_counter.h>
efbed4dc 32#include <linux/ratelimit.h>
0441984a 33#include <crypto/hash.h>
a633f5a3 34#include <linux/falloc.h>
4d92dc0f
BH
35#ifdef __KERNEL__
36#include <linux/compat.h>
37#endif
c9de560d 38
ac27a0ec 39/*
8a35694e 40 * The fourth extended filesystem constants/structures
ac27a0ec
DK
41 */
42
43/*
617ba13b 44 * Define EXT4FS_DEBUG to produce debug messages
ac27a0ec 45 */
617ba13b 46#undef EXT4FS_DEBUG
ac27a0ec 47
ac27a0ec
DK
48/*
49 * Debug code
50 */
617ba13b
MC
51#ifdef EXT4FS_DEBUG
52#define ext4_debug(f, a...) \
ac27a0ec 53 do { \
af5bc92d 54 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
4db9c54a 55 __FILE__, __LINE__, __func__); \
af5bc92d 56 printk(KERN_DEBUG f, ## a); \
ac27a0ec
DK
57 } while (0)
58#else
ace36ad4 59#define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
ac27a0ec
DK
60#endif
61
4a092d73
TT
62/*
63 * Turn on EXT_DEBUG to get lots of info about extents operations.
64 */
65#define EXT_DEBUG__
66#ifdef EXT_DEBUG
67#define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__)
68#else
69#define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
70#endif
71
273df556 72#define EXT4_ERROR_INODE(inode, fmt, a...) \
c398eda0
TT
73 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
74
75#define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
76 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
273df556 77
f7c21177
TT
78#define EXT4_ERROR_FILE(file, block, fmt, a...) \
79 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
273df556 80
d444c3c3
TT
81/* data type for block offset of block group */
82typedef int ext4_grpblk_t;
83
84/* data type for filesystem-wide blocks number */
85typedef unsigned long long ext4_fsblk_t;
86
87/* data type for file logical block number */
88typedef __u32 ext4_lblk_t;
89
90/* data type for block group number */
91typedef unsigned int ext4_group_t;
92
296c355c 93/*
60e6679e 94 * Flags used in mballoc's allocation_context flags field.
296c355c
TT
95 *
96 * Also used to show what's going on for debugging purposes when the
97 * flag field is exported via the traceport interface
98 */
d444c3c3 99
c9de560d 100/* prefer goal again. length */
0ef90db9 101#define EXT4_MB_HINT_MERGE 0x0001
c9de560d 102/* blocks already reserved */
0ef90db9 103#define EXT4_MB_HINT_RESERVED 0x0002
c9de560d 104/* metadata is being allocated */
0ef90db9 105#define EXT4_MB_HINT_METADATA 0x0004
c9de560d 106/* first blocks in the file */
0ef90db9 107#define EXT4_MB_HINT_FIRST 0x0008
c9de560d 108/* search for the best chunk */
0ef90db9 109#define EXT4_MB_HINT_BEST 0x0010
c9de560d 110/* data is being allocated */
0ef90db9 111#define EXT4_MB_HINT_DATA 0x0020
c9de560d 112/* don't preallocate (for tails) */
0ef90db9 113#define EXT4_MB_HINT_NOPREALLOC 0x0040
c9de560d 114/* allocate for locality group */
0ef90db9 115#define EXT4_MB_HINT_GROUP_ALLOC 0x0080
c9de560d 116/* allocate goal blocks or none */
0ef90db9 117#define EXT4_MB_HINT_GOAL_ONLY 0x0100
c9de560d 118/* goal is meaningful */
0ef90db9 119#define EXT4_MB_HINT_TRY_GOAL 0x0200
d2a17637 120/* blocks already pre-reserved by delayed allocation */
0ef90db9 121#define EXT4_MB_DELALLOC_RESERVED 0x0400
4ba74d00
TT
122/* We are doing stream allocation */
123#define EXT4_MB_STREAM_ALLOC 0x0800
55f020db
AH
124/* Use reserved root blocks if needed */
125#define EXT4_MB_USE_ROOT_BLOCKS 0x1000
27dd4385
LC
126/* Use blocks from reserved pool */
127#define EXT4_MB_USE_RESERVED 0x2000
c9de560d
AT
128
129struct ext4_allocation_request {
130 /* target inode for block we're allocating */
131 struct inode *inode;
726447d8
ES
132 /* how many blocks we want to allocate */
133 unsigned int len;
c9de560d
AT
134 /* logical block in target inode */
135 ext4_lblk_t logical;
c9de560d
AT
136 /* the closest logical allocated block to the left */
137 ext4_lblk_t lleft;
c9de560d
AT
138 /* the closest logical allocated block to the right */
139 ext4_lblk_t lright;
726447d8
ES
140 /* phys. target (a hint) */
141 ext4_fsblk_t goal;
142 /* phys. block for the closest logical allocated block to the left */
143 ext4_fsblk_t pleft;
144 /* phys. block for the closest logical allocated block to the right */
c9de560d 145 ext4_fsblk_t pright;
c9de560d 146 /* flags. see above EXT4_MB_HINT_* */
498e5f24 147 unsigned int flags;
c9de560d
AT
148};
149
e35fd660
TT
150/*
151 * Logical to physical block mapping, used by ext4_map_blocks()
152 *
153 * This structure is used to pass requests into ext4_map_blocks() as
154 * well as to store the information returned by ext4_map_blocks(). It
155 * takes less room on the stack than a struct buffer_head.
156 */
157#define EXT4_MAP_NEW (1 << BH_New)
158#define EXT4_MAP_MAPPED (1 << BH_Mapped)
159#define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten)
160#define EXT4_MAP_BOUNDARY (1 << BH_Boundary)
7b415bf6
AK
161/* Sometimes (in the bigalloc case, from ext4_da_get_block_prep) the caller of
162 * ext4_map_blocks wants to know whether or not the underlying cluster has
163 * already been accounted for. EXT4_MAP_FROM_CLUSTER conveys to the caller that
164 * the requested mapping was from previously mapped (or delayed allocated)
165 * cluster. We use BH_AllocFromCluster only for this flag. BH_AllocFromCluster
166 * should never appear on buffer_head's state flags.
167 */
168#define EXT4_MAP_FROM_CLUSTER (1 << BH_AllocFromCluster)
e35fd660
TT
169#define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
170 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\
090f32ee 171 EXT4_MAP_FROM_CLUSTER)
e35fd660
TT
172
173struct ext4_map_blocks {
174 ext4_fsblk_t m_pblk;
175 ext4_lblk_t m_lblk;
176 unsigned int m_len;
177 unsigned int m_flags;
178};
179
bd2d0210
TT
180/*
181 * Flags for ext4_io_end->flags
182 */
183#define EXT4_IO_END_UNWRITTEN 0x0001
bd2d0210 184
4188188b 185/*
556615dc 186 * For converting unwritten extents on a work queue. 'handle' is used for
6b523df4 187 * buffered writeback.
4188188b 188 */
0031462b 189typedef struct ext4_io_end {
5b3ff237 190 struct list_head list; /* per-file finished IO list */
6b523df4
JK
191 handle_t *handle; /* handle reserved for extent
192 * conversion */
0031462b 193 struct inode *inode; /* file being written to */
b0857d30
JK
194 struct bio *bio; /* Linked list of completed
195 * bios covering the extent */
8d5d02e6 196 unsigned int flag; /* unwritten or not */
a1de02dc
ES
197 loff_t offset; /* offset in the file */
198 ssize_t size; /* size of the extent */
97a851ed 199 atomic_t count; /* reference counter */
0031462b
MC
200} ext4_io_end_t;
201
bd2d0210
TT
202struct ext4_io_submit {
203 int io_op;
204 struct bio *io_bio;
205 ext4_io_end_t *io_end;
bd2d0210
TT
206 sector_t io_next_block;
207};
208
ac27a0ec
DK
209/*
210 * Special inodes numbers
211 */
617ba13b
MC
212#define EXT4_BAD_INO 1 /* Bad blocks inode */
213#define EXT4_ROOT_INO 2 /* Root inode */
ae812306
AK
214#define EXT4_USR_QUOTA_INO 3 /* User quota inode */
215#define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
617ba13b
MC
216#define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
217#define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
218#define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
219#define EXT4_JOURNAL_INO 8 /* Journal inode */
ac27a0ec 220
617ba13b
MC
221/* First non-reserved inode for old ext4 filesystems */
222#define EXT4_GOOD_OLD_FIRST_INO 11
ac27a0ec
DK
223
224/*
225 * Maximal count of links to a file
226 */
f8628a14 227#define EXT4_LINK_MAX 65000
ac27a0ec
DK
228
229/*
230 * Macro-instructions used to manage several block sizes
231 */
617ba13b 232#define EXT4_MIN_BLOCK_SIZE 1024
afc7cbca
TS
233#define EXT4_MAX_BLOCK_SIZE 65536
234#define EXT4_MIN_BLOCK_LOG_SIZE 10
fb1813f4 235#define EXT4_MAX_BLOCK_LOG_SIZE 16
ac27a0ec 236#ifdef __KERNEL__
617ba13b 237# define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
ac27a0ec 238#else
617ba13b 239# define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
ac27a0ec 240#endif
af5bc92d 241#define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
281b5995
TT
242#define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
243 EXT4_SB(s)->s_cluster_bits)
ac27a0ec 244#ifdef __KERNEL__
617ba13b 245# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
281b5995 246# define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
ac27a0ec 247#else
617ba13b 248# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
ac27a0ec
DK
249#endif
250#ifdef __KERNEL__
617ba13b
MC
251#define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
252#define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
253#define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
ac27a0ec 254#else
617ba13b
MC
255#define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
256 EXT4_GOOD_OLD_INODE_SIZE : \
ac27a0ec 257 (s)->s_inode_size)
617ba13b
MC
258#define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
259 EXT4_GOOD_OLD_FIRST_INO : \
ac27a0ec
DK
260 (s)->s_first_ino)
261#endif
a2df2a63 262#define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
ac27a0ec 263
d5b8f310
TT
264/* Translate a block number to a cluster number */
265#define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
266/* Translate a cluster number to a block number */
267#define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
268/* Translate # of blks to # of clusters */
269#define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
270 (sbi)->s_cluster_bits)
f5a44db5
TT
271/* Mask out the low bits to get the starting block of the cluster */
272#define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
273 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
274#define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
275 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
276/* Get the cluster offset */
277#define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
278 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
279#define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
280 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
d5b8f310 281
ac27a0ec
DK
282/*
283 * Structure of a blocks group descriptor
284 */
617ba13b 285struct ext4_group_desc
ac27a0ec 286{
3a14589c 287 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
5272f837
AK
288 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
289 __le32 bg_inode_table_lo; /* Inodes table block */
560671a0
AK
290 __le16 bg_free_blocks_count_lo;/* Free blocks count */
291 __le16 bg_free_inodes_count_lo;/* Free inodes count */
292 __le16 bg_used_dirs_count_lo; /* Directories count */
717d50e4 293 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
e6153918
DW
294 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
295 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
296 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
560671a0 297 __le16 bg_itable_unused_lo; /* Unused inodes count */
717d50e4 298 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
8fadc143
AR
299 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
300 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
301 __le32 bg_inode_table_hi; /* Inodes table block MSB */
91b51a01
CL
302 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
303 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
304 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
560671a0 305 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
e6153918
DW
306 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
307 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
308 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
309 __u32 bg_reserved;
ac27a0ec
DK
310};
311
41a246d1
DW
312#define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
313 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
314 sizeof(__le16))
315#define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
316 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
317 sizeof(__le16))
318
772cb7c8
JS
319/*
320 * Structure of a flex block group info
321 */
322
323struct flex_groups {
90ba983f
TT
324 atomic64_t free_clusters;
325 atomic_t free_inodes;
326 atomic_t used_dirs;
772cb7c8
JS
327};
328
717d50e4
AD
329#define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
330#define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
331#define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
332
ac27a0ec
DK
333/*
334 * Macro-instructions used to manage group descriptors
335 */
0d1ee42f 336#define EXT4_MIN_DESC_SIZE 32
8fadc143 337#define EXT4_MIN_DESC_SIZE_64BIT 64
0d1ee42f
AR
338#define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
339#define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
ac27a0ec 340#ifdef __KERNEL__
617ba13b 341# define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
281b5995 342# define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
617ba13b
MC
343# define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
344# define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
345# define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
ac27a0ec 346#else
617ba13b 347# define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
0d1ee42f 348# define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
617ba13b 349# define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
ac27a0ec
DK
350#endif
351
352/*
353 * Constants relative to the data blocks
354 */
617ba13b
MC
355#define EXT4_NDIR_BLOCKS 12
356#define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
357#define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
358#define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
359#define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
ac27a0ec
DK
360
361/*
362 * Inode flags
363 */
617ba13b
MC
364#define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
365#define EXT4_UNRM_FL 0x00000002 /* Undelete */
366#define EXT4_COMPR_FL 0x00000004 /* Compress file */
367#define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
368#define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
369#define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
370#define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
371#define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
ac27a0ec 372/* Reserved for compression usage... */
617ba13b
MC
373#define EXT4_DIRTY_FL 0x00000100
374#define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
375#define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
376#define EXT4_ECOMPR_FL 0x00000800 /* Compression error */
ac27a0ec 377/* End compression flags --- maybe not all used */
617ba13b
MC
378#define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
379#define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
380#define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
381#define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
382#define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
383#define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
8180a562 384#define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
a86c6181 385#define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
f710b4b9 386#define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
c8d46e41 387#define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
67cf5b09 388#define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
8180a562 389#define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
ac27a0ec 390
c8d46e41 391#define EXT4_FL_USER_VISIBLE 0x004BDFFF /* User visible flags */
996bb9fd 392#define EXT4_FL_USER_MODIFIABLE 0x004380FF /* User modifiable flags */
ac27a0ec 393
8fa43a81
DG
394/* Flags that should be inherited by new inodes from their parent. */
395#define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
1cd9f097 396 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
8fa43a81
DG
397 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
398 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
399
2dc6b0d4
DG
400/* Flags that are appropriate for regular files (all but dir-specific ones). */
401#define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
402
403/* Flags that are appropriate for non-directories/regular files. */
404#define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
405
406/* Mask out flags that are inappropriate for the given type of inode. */
407static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
408{
409 if (S_ISDIR(mode))
410 return flags;
411 else if (S_ISREG(mode))
412 return flags & EXT4_REG_FLMASK;
413 else
414 return flags & EXT4_OTHER_FLMASK;
415}
416
12e9b892
DM
417/*
418 * Inode flags used for atomic set/get
419 */
420enum {
421 EXT4_INODE_SECRM = 0, /* Secure deletion */
422 EXT4_INODE_UNRM = 1, /* Undelete */
423 EXT4_INODE_COMPR = 2, /* Compress file */
424 EXT4_INODE_SYNC = 3, /* Synchronous updates */
425 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
426 EXT4_INODE_APPEND = 5, /* writes to file may only append */
427 EXT4_INODE_NODUMP = 6, /* do not dump file */
428 EXT4_INODE_NOATIME = 7, /* do not update atime */
429/* Reserved for compression usage... */
430 EXT4_INODE_DIRTY = 8,
431 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
432 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
433 EXT4_INODE_ECOMPR = 11, /* Compression error */
434/* End compression flags --- maybe not all used */
435 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
436 EXT4_INODE_IMAGIC = 13, /* AFS directory */
437 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
438 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
439 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
440 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
441 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
442 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
443 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
444 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
67cf5b09 445 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
12e9b892
DM
446 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
447};
448
12e9b892 449/*
9a4c8019
CM
450 * Since it's pretty easy to mix up bit numbers and hex values, we use a
451 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
452 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
453 * any extra space in the compiled kernel image, otherwise, the build will fail.
454 * It's important that these values are the same, since we are using
455 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
456 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
457 * values found in ext2, ext3 and ext4 filesystems, and of course the values
458 * defined in e2fsprogs.
12e9b892
DM
459 *
460 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
461 */
9a4c8019
CM
462#define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
463#define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
464
12e9b892
DM
465static inline void ext4_check_flag_values(void)
466{
467 CHECK_FLAG_VALUE(SECRM);
468 CHECK_FLAG_VALUE(UNRM);
469 CHECK_FLAG_VALUE(COMPR);
470 CHECK_FLAG_VALUE(SYNC);
471 CHECK_FLAG_VALUE(IMMUTABLE);
472 CHECK_FLAG_VALUE(APPEND);
473 CHECK_FLAG_VALUE(NODUMP);
474 CHECK_FLAG_VALUE(NOATIME);
475 CHECK_FLAG_VALUE(DIRTY);
476 CHECK_FLAG_VALUE(COMPRBLK);
477 CHECK_FLAG_VALUE(NOCOMPR);
478 CHECK_FLAG_VALUE(ECOMPR);
479 CHECK_FLAG_VALUE(INDEX);
480 CHECK_FLAG_VALUE(IMAGIC);
481 CHECK_FLAG_VALUE(JOURNAL_DATA);
482 CHECK_FLAG_VALUE(NOTAIL);
483 CHECK_FLAG_VALUE(DIRSYNC);
484 CHECK_FLAG_VALUE(TOPDIR);
485 CHECK_FLAG_VALUE(HUGE_FILE);
486 CHECK_FLAG_VALUE(EXTENTS);
487 CHECK_FLAG_VALUE(EA_INODE);
488 CHECK_FLAG_VALUE(EOFBLOCKS);
67cf5b09 489 CHECK_FLAG_VALUE(INLINE_DATA);
12e9b892
DM
490 CHECK_FLAG_VALUE(RESERVED);
491}
492
ac27a0ec 493/* Used to pass group descriptor data when online resize is done */
617ba13b 494struct ext4_new_group_input {
8c55e204
DK
495 __u32 group; /* Group number for this data */
496 __u64 block_bitmap; /* Absolute block number of block bitmap */
497 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
498 __u64 inode_table; /* Absolute block number of inode table start */
499 __u32 blocks_count; /* Total number of blocks in this group */
500 __u16 reserved_blocks; /* Number of reserved blocks in this group */
ac27a0ec
DK
501 __u16 unused;
502};
503
4d92dc0f
BH
504#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
505struct compat_ext4_new_group_input {
506 u32 group;
507 compat_u64 block_bitmap;
508 compat_u64 inode_bitmap;
509 compat_u64 inode_table;
510 u32 blocks_count;
511 u16 reserved_blocks;
512 u16 unused;
513};
514#endif
515
617ba13b
MC
516/* The struct ext4_new_group_input in kernel space, with free_blocks_count */
517struct ext4_new_group_data {
ac27a0ec 518 __u32 group;
bd81d8ee
LV
519 __u64 block_bitmap;
520 __u64 inode_bitmap;
521 __u64 inode_table;
ac27a0ec
DK
522 __u32 blocks_count;
523 __u16 reserved_blocks;
524 __u16 unused;
525 __u32 free_blocks_count;
526};
527
33afdcc5
YY
528/* Indexes used to index group tables in ext4_new_group_data */
529enum {
530 BLOCK_BITMAP = 0, /* block bitmap */
531 INODE_BITMAP, /* inode bitmap */
532 INODE_TABLE, /* inode tables */
533 GROUP_TABLE_COUNT,
534};
535
a2df2a63 536/*
79e83036 537 * Flags used by ext4_map_blocks()
a2df2a63 538 */
556615dc 539 /* Allocate any needed blocks and/or convert an unwritten
c2177057 540 extent to be an initialized ext4 */
2ac3b6e0 541#define EXT4_GET_BLOCKS_CREATE 0x0001
556615dc
LC
542 /* Request the creation of an unwritten extent */
543#define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002
544#define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\
c2177057 545 EXT4_GET_BLOCKS_CREATE)
27dd4385
LC
546 /* Caller is from the delayed allocation writeout path
547 * finally doing the actual allocation of delayed blocks */
03f5d8bc 548#define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
0031462b 549 /* caller is from the direct IO path, request to creation of an
556615dc 550 unwritten extents if not allocated, split the unwritten
0031462b 551 extent if blocks has been preallocated already*/
c7064ef1 552#define EXT4_GET_BLOCKS_PRE_IO 0x0008
1296cc85 553#define EXT4_GET_BLOCKS_CONVERT 0x0010
c7064ef1 554#define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
556615dc 555 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
c7064ef1
JZ
556 /* Convert extent to initialized after IO complete */
557#define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
556615dc 558 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
27dd4385
LC
559 /* Eventual metadata allocation (due to growing extent tree)
560 * should not fail, so try to use reserved blocks for that.*/
561#define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
556b27ab
VH
562 /* Don't normalize allocation size (used for fallocate) */
563#define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
a4e5d88b
DM
564 /* Request will not result in inode size update (user for fallocate) */
565#define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080
729f52c6
ZL
566 /* Do not take i_data_sem locking in ext4_map_blocks */
567#define EXT4_GET_BLOCKS_NO_LOCK 0x0100
d100eef2
ZL
568 /* Do not put hole in extent cache */
569#define EXT4_GET_BLOCKS_NO_PUT_HOLE 0x0200
b8a86845
LC
570 /* Convert written extents to unwritten */
571#define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0400
ac27a0ec 572
107a7bd3 573/*
7869a4a6
TT
574 * The bit position of these flags must not overlap with any of the
575 * EXT4_GET_BLOCKS_*. They are used by ext4_ext_find_extent(),
107a7bd3 576 * read_extent_tree_block(), ext4_split_extent_at(),
7869a4a6
TT
577 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
578 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
579 * caching the extents when reading from the extent tree while a
580 * truncate or punch hole operation is in progress.
107a7bd3
TT
581 */
582#define EXT4_EX_NOCACHE 0x0400
7869a4a6 583#define EXT4_EX_FORCE_CACHE 0x0800
107a7bd3 584
e6362609
TT
585/*
586 * Flags used by ext4_free_blocks
587 */
588#define EXT4_FREE_BLOCKS_METADATA 0x0001
589#define EXT4_FREE_BLOCKS_FORGET 0x0002
1f2acb60 590#define EXT4_FREE_BLOCKS_VALIDATED 0x0004
7132de74 591#define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
84130193
TT
592#define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
593#define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
e6362609 594
ac27a0ec
DK
595/*
596 * ioctl commands
597 */
617ba13b
MC
598#define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
599#define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
600#define EXT4_IOC_GETVERSION _IOR('f', 3, long)
601#define EXT4_IOC_SETVERSION _IOW('f', 4, long)
617ba13b
MC
602#define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
603#define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
617ba13b
MC
604#define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
605#define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
8eea80d5
TT
606#define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
607#define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
608#define EXT4_IOC_MIGRATE _IO('f', 9)
ccd2506b 609 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
8eea80d5 610 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
ccd2506b 611#define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
748de673 612#define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
19c5246d 613#define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
393d1d1d 614#define EXT4_IOC_SWAP_BOOT _IO('f', 17)
7869a4a6 615#define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18)
ac27a0ec 616
899ad0ce 617#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
ac27a0ec
DK
618/*
619 * ioctl commands in 32 bit emulation
620 */
617ba13b
MC
621#define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
622#define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
623#define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
624#define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
625#define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
626#define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
627#define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
4d92dc0f 628#define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
617ba13b
MC
629#define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
630#define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
899ad0ce 631#endif
ac27a0ec 632
b595076a 633/* Max physical block we can address w/o extents */
fb0a387d
ES
634#define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
635
ac27a0ec
DK
636/*
637 * Structure of an inode on the disk
638 */
617ba13b 639struct ext4_inode {
ac27a0ec
DK
640 __le16 i_mode; /* File mode */
641 __le16 i_uid; /* Low 16 bits of Owner Uid */
a48380f7 642 __le32 i_size_lo; /* Size in bytes */
ac27a0ec 643 __le32 i_atime; /* Access time */
ef7f3835 644 __le32 i_ctime; /* Inode Change time */
ac27a0ec
DK
645 __le32 i_mtime; /* Modification time */
646 __le32 i_dtime; /* Deletion Time */
647 __le16 i_gid; /* Low 16 bits of Group Id */
648 __le16 i_links_count; /* Links count */
0fc1b451 649 __le32 i_blocks_lo; /* Blocks count */
ac27a0ec
DK
650 __le32 i_flags; /* File flags */
651 union {
652 struct {
25ec56b5 653 __le32 l_i_version;
ac27a0ec
DK
654 } linux1;
655 struct {
656 __u32 h_i_translator;
657 } hurd1;
658 struct {
659 __u32 m_i_reserved1;
660 } masix1;
661 } osd1; /* OS dependent 1 */
617ba13b 662 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
ac27a0ec 663 __le32 i_generation; /* File version (for NFS) */
7973c0c1 664 __le32 i_file_acl_lo; /* File ACL */
a48380f7 665 __le32 i_size_high;
f077d0d7 666 __le32 i_obso_faddr; /* Obsoleted fragment address */
ac27a0ec
DK
667 union {
668 struct {
0fc1b451 669 __le16 l_i_blocks_high; /* were l_i_reserved1 */
a1ddeb7e 670 __le16 l_i_file_acl_high;
8c55e204 671 __le16 l_i_uid_high; /* these 2 fields */
ac27a0ec 672 __le16 l_i_gid_high; /* were reserved2[0] */
e6153918
DW
673 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
674 __le16 l_i_reserved;
ac27a0ec
DK
675 } linux2;
676 struct {
f077d0d7 677 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
ac27a0ec
DK
678 __u16 h_i_mode_high;
679 __u16 h_i_uid_high;
680 __u16 h_i_gid_high;
681 __u32 h_i_author;
682 } hurd2;
683 struct {
f077d0d7 684 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
a1ddeb7e 685 __le16 m_i_file_acl_high;
ac27a0ec
DK
686 __u32 m_i_reserved2[2];
687 } masix2;
688 } osd2; /* OS dependent 2 */
689 __le16 i_extra_isize;
e6153918 690 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
ef7f3835
KS
691 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
692 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
693 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
694 __le32 i_crtime; /* File Creation time */
695 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
25ec56b5 696 __le32 i_version_hi; /* high 32 bits for 64-bit version */
ac27a0ec
DK
697};
698
748de673
AF
699struct move_extent {
700 __u32 reserved; /* should be zero */
701 __u32 donor_fd; /* donor file descriptor */
702 __u64 orig_start; /* logical start offset in block for orig */
703 __u64 donor_start; /* logical start offset in block for donor */
704 __u64 len; /* block length to be moved */
705 __u64 moved_len; /* moved block length */
706};
ac27a0ec 707
ef7f3835
KS
708#define EXT4_EPOCH_BITS 2
709#define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
710#define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
711
712/*
713 * Extended fields will fit into an inode if the filesystem was formatted
714 * with large inodes (-I 256 or larger) and there are not currently any EAs
715 * consuming all of the available space. For new inodes we always reserve
716 * enough space for the kernel's known extended fields, but for inodes
717 * created with an old kernel this might not have been the case. None of
718 * the extended inode fields is critical for correct filesystem operation.
719 * This macro checks if a certain field fits in the inode. Note that
720 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
721 */
722#define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
723 ((offsetof(typeof(*ext4_inode), field) + \
724 sizeof((ext4_inode)->field)) \
725 <= (EXT4_GOOD_OLD_INODE_SIZE + \
726 (einode)->i_extra_isize)) \
727
728static inline __le32 ext4_encode_extra_time(struct timespec *time)
729{
730 return cpu_to_le32((sizeof(time->tv_sec) > 4 ?
c1fccc06
TT
731 (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) |
732 ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK));
ef7f3835
KS
733}
734
735static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
736{
737 if (sizeof(time->tv_sec) > 4)
738 time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK)
739 << 32;
c1fccc06 740 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
ef7f3835
KS
741}
742
743#define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
744do { \
745 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
746 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
747 (raw_inode)->xtime ## _extra = \
748 ext4_encode_extra_time(&(inode)->xtime); \
749} while (0)
750
751#define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
752do { \
753 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
754 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
755 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
756 (raw_inode)->xtime ## _extra = \
757 ext4_encode_extra_time(&(einode)->xtime); \
758} while (0)
759
760#define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
761do { \
762 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
763 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
764 ext4_decode_extra_time(&(inode)->xtime, \
765 raw_inode->xtime ## _extra); \
af0b44a1
ES
766 else \
767 (inode)->xtime.tv_nsec = 0; \
ef7f3835
KS
768} while (0)
769
770#define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
771do { \
772 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
773 (einode)->xtime.tv_sec = \
774 (signed)le32_to_cpu((raw_inode)->xtime); \
19ea8060
TT
775 else \
776 (einode)->xtime.tv_sec = 0; \
ef7f3835
KS
777 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
778 ext4_decode_extra_time(&(einode)->xtime, \
779 raw_inode->xtime ## _extra); \
af0b44a1
ES
780 else \
781 (einode)->xtime.tv_nsec = 0; \
ef7f3835
KS
782} while (0)
783
25ec56b5
JNC
784#define i_disk_version osd1.linux1.l_i_version
785
ac27a0ec
DK
786#if defined(__KERNEL__) || defined(__linux__)
787#define i_reserved1 osd1.linux1.l_i_reserved1
a1ddeb7e 788#define i_file_acl_high osd2.linux2.l_i_file_acl_high
0fc1b451 789#define i_blocks_high osd2.linux2.l_i_blocks_high
ac27a0ec
DK
790#define i_uid_low i_uid
791#define i_gid_low i_gid
792#define i_uid_high osd2.linux2.l_i_uid_high
793#define i_gid_high osd2.linux2.l_i_gid_high
e6153918 794#define i_checksum_lo osd2.linux2.l_i_checksum_lo
ac27a0ec
DK
795
796#elif defined(__GNU__)
797
798#define i_translator osd1.hurd1.h_i_translator
ac27a0ec
DK
799#define i_uid_high osd2.hurd2.h_i_uid_high
800#define i_gid_high osd2.hurd2.h_i_gid_high
801#define i_author osd2.hurd2.h_i_author
802
803#elif defined(__masix__)
804
805#define i_reserved1 osd1.masix1.m_i_reserved1
a1ddeb7e 806#define i_file_acl_high osd2.masix2.m_i_file_acl_high
ac27a0ec
DK
807#define i_reserved2 osd2.masix2.m_i_reserved2
808
809#endif /* defined(__KERNEL__) || defined(__linux__) */
810
c0677e6d
ZL
811#include "extents_status.h"
812
d444c3c3
TT
813/*
814 * fourth extended file system inode data in memory
815 */
816struct ext4_inode_info {
817 __le32 i_data[15]; /* unconverted */
d444c3c3 818 __u32 i_dtime;
12e9b892 819 ext4_fsblk_t i_file_acl;
d444c3c3
TT
820
821 /*
822 * i_block_group is the number of the block group which contains
823 * this file's inode. Constant across the lifetime of the inode,
824 * it is ued for making block allocation decisions - we try to
825 * place a file's data blocks near its inode block, and new inodes
826 * near to their parent directory's inode.
827 */
828 ext4_group_t i_block_group;
8a2005d3 829 ext4_lblk_t i_dir_start_lookup;
353eb83c 830#if (BITS_PER_LONG < 64)
19f5fb7a 831 unsigned long i_state_flags; /* Dynamic state flags */
353eb83c 832#endif
12e9b892 833 unsigned long i_flags;
d444c3c3 834
d444c3c3
TT
835 /*
836 * Extended attributes can be read independently of the main file
837 * data. Taking i_mutex even when reading would cause contention
838 * between readers of EAs and writers of regular file data, so
839 * instead we synchronize on xattr_sem when reading or changing
840 * EAs.
841 */
842 struct rw_semaphore xattr_sem;
d444c3c3
TT
843
844 struct list_head i_orphan; /* unlinked but open inodes */
845
846 /*
847 * i_disksize keeps track of what the inode size is ON DISK, not
848 * in memory. During truncate, i_size is set to the new size by
849 * the VFS prior to calling ext4_truncate(), but the filesystem won't
850 * set i_disksize to 0 until the truncate is actually under way.
851 *
852 * The intent is that i_disksize always represents the blocks which
853 * are used by this file. This allows recovery to restart truncate
854 * on orphans if we crash during truncate. We actually write i_disksize
855 * into the on-disk inode when writing inodes out, instead of i_size.
856 *
857 * The only time when i_disksize and i_size may be different is when
858 * a truncate is in progress. The only things which change i_disksize
859 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
860 */
861 loff_t i_disksize;
862
863 /*
864 * i_data_sem is for serialising ext4_truncate() against
865 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
866 * data tree are chopped off during truncate. We can't do that in
867 * ext4 because whenever we perform intermediate commits during
868 * truncate, the inode and all the metadata blocks *must* be in a
869 * consistent state which allows truncation of the orphans to restart
870 * during recovery. Hence we must fix the get_block-vs-truncate race
871 * by other means, so we have i_data_sem.
872 */
873 struct rw_semaphore i_data_sem;
874 struct inode vfs_inode;
8aefcd55 875 struct jbd2_inode *jinode;
d444c3c3 876
202ee5df
TT
877 spinlock_t i_raw_lock; /* protects updates to the raw inode */
878
d444c3c3
TT
879 /*
880 * File creation time. Its function is same as that of
881 * struct timespec i_{a,c,m}time in the generic inode.
882 */
883 struct timespec i_crtime;
884
885 /* mballoc */
886 struct list_head i_prealloc_list;
887 spinlock_t i_prealloc_lock;
888
c0677e6d
ZL
889 /* extents status tree */
890 struct ext4_es_tree i_es_tree;
891 rwlock_t i_es_lock;
74cd15cd
ZL
892 struct list_head i_es_lru;
893 unsigned int i_es_lru_nr; /* protected by i_es_lock */
d3922a77 894 unsigned long i_touch_when; /* jiffies of last accessing */
c0677e6d 895
d444c3c3
TT
896 /* ialloc */
897 ext4_group_t i_last_alloc_group;
898
899 /* allocation reservation info for delalloc */
57042651 900 /* In case of bigalloc, these refer to clusters rather than blocks */
d444c3c3
TT
901 unsigned int i_reserved_data_blocks;
902 unsigned int i_reserved_meta_blocks;
903 unsigned int i_allocated_meta_blocks;
01f49d0b 904 ext4_lblk_t i_da_metadata_calc_last_lblock;
9d0be502 905 int i_da_metadata_calc_len;
d444c3c3
TT
906
907 /* on-disk additional length */
908 __u16 i_extra_isize;
909
67cf5b09
TM
910 /* Indicate the inline data space. */
911 u16 i_inline_off;
912 u16 i_inline_size;
913
a9e7f447
DM
914#ifdef CONFIG_QUOTA
915 /* quota space reservation, managed internally by quota code */
916 qsize_t i_reserved_quota;
917#endif
8d5d02e6 918
2e8fa54e 919 /* Lock protecting lists below */
744692dc 920 spinlock_t i_completed_io_lock;
2e8fa54e
JK
921 /*
922 * Completed IOs that need unwritten extents handling and have
923 * transaction reserved
924 */
925 struct list_head i_rsv_conversion_list;
926 /*
927 * Completed IOs that need unwritten extents handling and don't have
928 * transaction reserved
929 */
8a2005d3 930 atomic_t i_ioend_count; /* Number of outstanding io_end structs */
e27f41e1 931 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
2e8fa54e 932 struct work_struct i_rsv_conversion_work;
8a2005d3
TT
933
934 spinlock_t i_block_reservation_lock;
b436b9be
JK
935
936 /*
937 * Transactions that contain inode's metadata needed to complete
938 * fsync and fdatasync, respectively.
939 */
940 tid_t i_sync_tid;
941 tid_t i_datasync_tid;
814525f4
DW
942
943 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
944 __u32 i_csum_seed;
d444c3c3
TT
945};
946
ac27a0ec
DK
947/*
948 * File system states
949 */
617ba13b
MC
950#define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
951#define EXT4_ERROR_FS 0x0002 /* Errors detected */
952#define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
ac27a0ec 953
469108ff
TT
954/*
955 * Misc. filesystem flags
956 */
957#define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
958#define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
959#define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
960
ac27a0ec 961/*
68911009 962 * Mount flags set via mount options or defaults
ac27a0ec 963 */
617ba13b
MC
964#define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
965#define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
966#define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
967#define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
968#define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
39ef17f1 969#define EXT4_MOUNT_ERRORS_MASK 0x00070
617ba13b
MC
970#define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
971#define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
617ba13b
MC
972#define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
973#define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
974#define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
975#define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
976#define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
977#define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
978#define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
979#define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
afd4672d 980#define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
617ba13b 981#define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
617ba13b
MC
982#define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
983#define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
984#define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
744692dc 985#define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
d4da6c9c 986#define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
818d276c 987#define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
64769240 988#define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
5bf5683a 989#define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
6fd058f7 990#define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
5328e635 991#define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
bfff6873 992#define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
5bf5683a 993
68911009
LC
994/*
995 * Mount flags set either automatically (could not be set by mount option)
996 * based on per file system feature or property or in special cases such as
997 * distinguishing between explicit mount option definition and default.
998 */
56889787
TT
999#define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1000 specified delalloc */
68911009
LC
1001#define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1002 size of blocksize * 8
1003 blocks */
ed3654eb
TT
1004#define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1005 file systems */
56889787 1006
fd8c37ec
TT
1007#define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1008 ~EXT4_MOUNT_##opt
1009#define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1010 EXT4_MOUNT_##opt
617ba13b
MC
1011#define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1012 EXT4_MOUNT_##opt)
ac27a0ec 1013
a2595b8a
TT
1014#define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1015 ~EXT4_MOUNT2_##opt
1016#define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1017 EXT4_MOUNT2_##opt
1018#define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1019 EXT4_MOUNT2_##opt)
1020
597d508c
AM
1021#define ext4_test_and_set_bit __test_and_set_bit_le
1022#define ext4_set_bit __set_bit_le
617ba13b 1023#define ext4_set_bit_atomic ext2_set_bit_atomic
597d508c
AM
1024#define ext4_test_and_clear_bit __test_and_clear_bit_le
1025#define ext4_clear_bit __clear_bit_le
617ba13b 1026#define ext4_clear_bit_atomic ext2_clear_bit_atomic
50e0168c 1027#define ext4_test_bit test_bit_le
50e0168c
AM
1028#define ext4_find_next_zero_bit find_next_zero_bit_le
1029#define ext4_find_next_bit find_next_bit_le
ac27a0ec 1030
c3e94d1d
YY
1031extern void ext4_set_bits(void *bm, int cur, int len);
1032
ac27a0ec
DK
1033/*
1034 * Maximal mount counts between two filesystem checks
1035 */
617ba13b
MC
1036#define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1037#define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
ac27a0ec
DK
1038
1039/*
1040 * Behaviour when detecting errors
1041 */
617ba13b
MC
1042#define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1043#define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1044#define EXT4_ERRORS_PANIC 3 /* Panic */
1045#define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
ac27a0ec 1046
e6153918
DW
1047/* Metadata checksum algorithm codes */
1048#define EXT4_CRC32C_CHKSUM 1
1049
ac27a0ec
DK
1050/*
1051 * Structure of the super block
1052 */
617ba13b 1053struct ext4_super_block {
ac27a0ec 1054/*00*/ __le32 s_inodes_count; /* Inodes count */
6bc9feff 1055 __le32 s_blocks_count_lo; /* Blocks count */
308ba3ec
AK
1056 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1057 __le32 s_free_blocks_count_lo; /* Free blocks count */
ac27a0ec
DK
1058/*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1059 __le32 s_first_data_block; /* First Data Block */
1060 __le32 s_log_block_size; /* Block size */
281b5995 1061 __le32 s_log_cluster_size; /* Allocation cluster size */
ac27a0ec 1062/*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
281b5995 1063 __le32 s_clusters_per_group; /* # Clusters per group */
ac27a0ec
DK
1064 __le32 s_inodes_per_group; /* # Inodes per group */
1065 __le32 s_mtime; /* Mount time */
1066/*30*/ __le32 s_wtime; /* Write time */
1067 __le16 s_mnt_count; /* Mount count */
1068 __le16 s_max_mnt_count; /* Maximal mount count */
1069 __le16 s_magic; /* Magic signature */
1070 __le16 s_state; /* File system state */
1071 __le16 s_errors; /* Behaviour when detecting errors */
1072 __le16 s_minor_rev_level; /* minor revision level */
1073/*40*/ __le32 s_lastcheck; /* time of last check */
1074 __le32 s_checkinterval; /* max. time between checks */
1075 __le32 s_creator_os; /* OS */
1076 __le32 s_rev_level; /* Revision level */
1077/*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1078 __le16 s_def_resgid; /* Default gid for reserved blocks */
1079 /*
617ba13b 1080 * These fields are for EXT4_DYNAMIC_REV superblocks only.
ac27a0ec
DK
1081 *
1082 * Note: the difference between the compatible feature set and
1083 * the incompatible feature set is that if there is a bit set
1084 * in the incompatible feature set that the kernel doesn't
1085 * know about, it should refuse to mount the filesystem.
1086 *
1087 * e2fsck's requirements are more strict; if it doesn't know
1088 * about a feature in either the compatible or incompatible
1089 * feature set, it must abort and not try to meddle with
1090 * things it doesn't understand...
1091 */
1092 __le32 s_first_ino; /* First non-reserved inode */
0d1ee42f 1093 __le16 s_inode_size; /* size of inode structure */
ac27a0ec
DK
1094 __le16 s_block_group_nr; /* block group # of this superblock */
1095 __le32 s_feature_compat; /* compatible feature set */
1096/*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1097 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1098/*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1099/*78*/ char s_volume_name[16]; /* volume name */
1100/*88*/ char s_last_mounted[64]; /* directory where last mounted */
1101/*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1102 /*
1103 * Performance hints. Directory preallocation should only
617ba13b 1104 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
ac27a0ec
DK
1105 */
1106 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1107 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1108 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1109 /*
617ba13b 1110 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
ac27a0ec
DK
1111 */
1112/*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1113/*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1114 __le32 s_journal_dev; /* device number of journal file */
1115 __le32 s_last_orphan; /* start of list of inodes to delete */
1116 __le32 s_hash_seed[4]; /* HTREE hash seed */
1117 __u8 s_def_hash_version; /* Default hash version to use */
89eeddf0 1118 __u8 s_jnl_backup_type;
0d1ee42f 1119 __le16 s_desc_size; /* size of group descriptor */
bd81d8ee 1120/*100*/ __le32 s_default_mount_opts;
ac27a0ec 1121 __le32 s_first_meta_bg; /* First metablock block group */
bd81d8ee
LV
1122 __le32 s_mkfs_time; /* When the filesystem was created */
1123 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1124 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1125/*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1126 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1127 __le32 s_free_blocks_count_hi; /* Free blocks count */
c1bddad9
AK
1128 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1129 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1130 __le32 s_flags; /* Miscellaneous flags */
1131 __le16 s_raid_stride; /* RAID stride */
c5e06d10 1132 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
c1bddad9
AK
1133 __le64 s_mmp_block; /* Block for multi-mount protection */
1134 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
772cb7c8 1135 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
e6153918 1136 __u8 s_checksum_type; /* metadata checksum algorithm used */
772cb7c8 1137 __le16 s_reserved_pad;
afc32f7e 1138 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
4a9cdec7
TT
1139 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1140 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1141 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1142 snapshot's future use */
1143 __le32 s_snapshot_list; /* inode number of the head of the
1144 on-disk snapshot list */
1c13d5c0
TT
1145#define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1146 __le32 s_error_count; /* number of fs errors */
1147 __le32 s_first_error_time; /* first time an error happened */
1148 __le32 s_first_error_ino; /* inode involved in first error */
1149 __le64 s_first_error_block; /* block involved of first error */
1150 __u8 s_first_error_func[32]; /* function where the error happened */
1151 __le32 s_first_error_line; /* line number where error happened */
1152 __le32 s_last_error_time; /* most recent time of an error */
1153 __le32 s_last_error_ino; /* inode involved in last error */
1154 __le32 s_last_error_line; /* line number where error happened */
1155 __le64 s_last_error_block; /* block involved of last error */
1156 __u8 s_last_error_func[32]; /* function where the error happened */
8b67f04a
TT
1157#define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1158 __u8 s_mount_opts[64];
281b5995
TT
1159 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1160 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1161 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1beeef1b
DW
1162 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
1163 __le32 s_reserved[106]; /* Padding to the end of the block */
e6153918 1164 __le32 s_checksum; /* crc32c(superblock) */
ac27a0ec
DK
1165};
1166
1c13d5c0
TT
1167#define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1168
ac27a0ec 1169#ifdef __KERNEL__
bc0b0d6d
TT
1170
1171/*
4ab2f15b 1172 * run-time mount flags
bc0b0d6d
TT
1173 */
1174#define EXT4_MF_MNTDIR_SAMPLED 0x0001
4ab2f15b 1175#define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
bc0b0d6d 1176
ca0faba0
TT
1177/*
1178 * fourth extended-fs super-block data in memory
1179 */
1180struct ext4_sb_info {
1181 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1182 unsigned long s_inodes_per_block;/* Number of inodes per block */
1183 unsigned long s_blocks_per_group;/* Number of blocks in a group */
281b5995 1184 unsigned long s_clusters_per_group; /* Number of clusters in a group */
ca0faba0
TT
1185 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1186 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1187 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1188 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1189 ext4_group_t s_groups_count; /* Number of groups in the fs */
fb0a387d 1190 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
952fc18e 1191 unsigned long s_overhead; /* # of fs overhead clusters */
281b5995
TT
1192 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1193 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
ca0faba0
TT
1194 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1195 struct buffer_head * s_sbh; /* Buffer containing the super block */
1196 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1197 struct buffer_head **s_group_desc;
7f4520cc 1198 unsigned int s_mount_opt;
a2595b8a 1199 unsigned int s_mount_opt2;
bc0b0d6d 1200 unsigned int s_mount_flags;
5a916be1 1201 unsigned int s_def_mount_opt;
ca0faba0 1202 ext4_fsblk_t s_sb_block;
27dd4385 1203 atomic64_t s_resv_clusters;
08cefc7a
EB
1204 kuid_t s_resuid;
1205 kgid_t s_resgid;
ca0faba0
TT
1206 unsigned short s_mount_state;
1207 unsigned short s_pad;
1208 int s_addr_per_block_bits;
1209 int s_desc_per_block_bits;
1210 int s_inode_size;
1211 int s_first_ino;
1212 unsigned int s_inode_readahead_blks;
11013911 1213 unsigned int s_inode_goal;
ca0faba0
TT
1214 spinlock_t s_next_gen_lock;
1215 u32 s_next_generation;
1216 u32 s_hash_seed[4];
1217 int s_def_hash_version;
1218 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
57042651 1219 struct percpu_counter s_freeclusters_counter;
ca0faba0
TT
1220 struct percpu_counter s_freeinodes_counter;
1221 struct percpu_counter s_dirs_counter;
57042651 1222 struct percpu_counter s_dirtyclusters_counter;
ca0faba0
TT
1223 struct blockgroup_lock *s_blockgroup_lock;
1224 struct proc_dir_entry *s_proc;
1225 struct kobject s_kobj;
1226 struct completion s_kobj_unregister;
2c0544b2 1227 struct super_block *s_sb;
ca0faba0
TT
1228
1229 /* Journaling */
ca0faba0
TT
1230 struct journal_s *s_journal;
1231 struct list_head s_orphan;
1232 struct mutex s_orphan_lock;
8f82f840
YY
1233 unsigned long s_resize_flags; /* Flags indicating if there
1234 is a resizer */
ca0faba0
TT
1235 unsigned long s_commit_interval;
1236 u32 s_max_batch_time;
1237 u32 s_min_batch_time;
1238 struct block_device *journal_bdev;
ca0faba0
TT
1239#ifdef CONFIG_QUOTA
1240 char *s_qf_names[MAXQUOTAS]; /* Names of quota files with journalled quota */
1241 int s_jquota_fmt; /* Format of quota to use */
1242#endif
1243 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
6fd058f7 1244 struct rb_root system_blks;
ca0faba0
TT
1245
1246#ifdef EXTENTS_STATS
1247 /* ext4 extents stats */
1248 unsigned long s_ext_min;
1249 unsigned long s_ext_max;
1250 unsigned long s_depth_max;
1251 spinlock_t s_ext_stats_lock;
1252 unsigned long s_ext_blocks;
1253 unsigned long s_ext_extents;
1254#endif
1255
1256 /* for buddy allocator */
1257 struct ext4_group_info ***s_group_info;
1258 struct inode *s_buddy_cache;
ca0faba0 1259 spinlock_t s_md_lock;
ca0faba0
TT
1260 unsigned short *s_mb_offsets;
1261 unsigned int *s_mb_maxs;
28623c2f 1262 unsigned int s_group_info_size;
ca0faba0
TT
1263
1264 /* tunables */
1265 unsigned long s_stripe;
1266 unsigned int s_mb_stream_request;
1267 unsigned int s_mb_max_to_scan;
1268 unsigned int s_mb_min_to_scan;
1269 unsigned int s_mb_stats;
1270 unsigned int s_mb_order2_reqs;
1271 unsigned int s_mb_group_prealloc;
df981d03 1272 unsigned int s_max_dir_size_kb;
ca0faba0
TT
1273 /* where last allocation was done - for stream allocation */
1274 unsigned long s_mb_last_group;
1275 unsigned long s_mb_last_start;
1276
ca0faba0 1277 /* stats for buddy allocator */
ca0faba0
TT
1278 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1279 atomic_t s_bal_success; /* we found long enough chunks */
1280 atomic_t s_bal_allocated; /* in blocks */
1281 atomic_t s_bal_ex_scanned; /* total extents scanned */
1282 atomic_t s_bal_goals; /* goal hits */
1283 atomic_t s_bal_breaks; /* too long searches */
1284 atomic_t s_bal_2orders; /* 2^order hits */
1285 spinlock_t s_bal_lock;
1286 unsigned long s_mb_buddies_generated;
1287 unsigned long long s_mb_generation_time;
1288 atomic_t s_mb_lost_chunks;
1289 atomic_t s_mb_preallocated;
1290 atomic_t s_mb_discarded;
50797481 1291 atomic_t s_lock_busy;
ca0faba0
TT
1292
1293 /* locality groups */
003cb608 1294 struct ext4_locality_group __percpu *s_locality_groups;
ca0faba0
TT
1295
1296 /* for write statistics */
1297 unsigned long s_sectors_written_start;
1298 u64 s_kbytes_written;
1299
67a5da56
ZL
1300 /* the size of zero-out chunk */
1301 unsigned int s_extent_max_zeroout_kb;
1302
ca0faba0
TT
1303 unsigned int s_log_groups_per_flex;
1304 struct flex_groups *s_flex_groups;
117fff10 1305 ext4_group_t s_flex_groups_allocated;
4c0425ff 1306
2e8fa54e
JK
1307 /* workqueue for reserved extent conversions (buffered io) */
1308 struct workqueue_struct *rsv_conversion_wq;
66e61a9e
TT
1309
1310 /* timer for periodic error stats printing */
1311 struct timer_list s_err_report;
bfff6873
LC
1312
1313 /* Lazy inode table initialization info */
1314 struct ext4_li_request *s_li_request;
1315 /* Wait multiplier for lazy initialization thread */
1316 unsigned int s_li_wait_mult;
c5e06d10
JL
1317
1318 /* Kernel thread for multiple mount protection */
1319 struct task_struct *s_mmp_tsk;
3d56b8d2
TM
1320
1321 /* record the last minlen when FITRIM is called. */
1322 atomic_t s_last_trim_minblks;
0441984a
DW
1323
1324 /* Reference to checksum algorithm driver via cryptoapi */
1325 struct crypto_shash *s_chksum_driver;
a9c47317
DW
1326
1327 /* Precomputed FS UUID checksum for seeding other checksums */
1328 __u32 s_csum_seed;
74cd15cd
ZL
1329
1330 /* Reclaim extents from extent status tree */
1331 struct shrinker s_es_shrinker;
1332 struct list_head s_es_lru;
d3922a77 1333 unsigned long s_es_last_sorted;
1ac6466f 1334 struct percpu_counter s_extent_cache_cnt;
9c191f70 1335 struct mb_cache *s_mb_cache;
74cd15cd 1336 spinlock_t s_es_lru_lock ____cacheline_aligned_in_smp;
efbed4dc
TT
1337
1338 /* Ratelimit ext4 messages. */
1339 struct ratelimit_state s_err_ratelimit_state;
1340 struct ratelimit_state s_warning_ratelimit_state;
1341 struct ratelimit_state s_msg_ratelimit_state;
ca0faba0
TT
1342};
1343
af5bc92d 1344static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
ac27a0ec
DK
1345{
1346 return sb->s_fs_info;
1347}
617ba13b 1348static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
ac27a0ec 1349{
617ba13b 1350 return container_of(inode, struct ext4_inode_info, vfs_inode);
ac27a0ec
DK
1351}
1352
ef7f3835
KS
1353static inline struct timespec ext4_current_time(struct inode *inode)
1354{
1355 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1356 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1357}
1358
617ba13b 1359static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
ac27a0ec 1360{
617ba13b 1361 return ino == EXT4_ROOT_INO ||
7c319d32
AK
1362 ino == EXT4_USR_QUOTA_INO ||
1363 ino == EXT4_GRP_QUOTA_INO ||
393d1d1d 1364 ino == EXT4_BOOT_LOADER_INO ||
617ba13b
MC
1365 ino == EXT4_JOURNAL_INO ||
1366 ino == EXT4_RESIZE_INO ||
1367 (ino >= EXT4_FIRST_INO(sb) &&
1368 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
ac27a0ec 1369}
19f5fb7a 1370
0edeb71d
TM
1371static inline void ext4_set_io_unwritten_flag(struct inode *inode,
1372 struct ext4_io_end *io_end)
1373{
1374 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
1375 io_end->flag |= EXT4_IO_END_UNWRITTEN;
e27f41e1 1376 atomic_inc(&EXT4_I(inode)->i_unwritten);
0edeb71d
TM
1377 }
1378}
1379
f45ee3a1
DM
1380static inline ext4_io_end_t *ext4_inode_aio(struct inode *inode)
1381{
1382 return inode->i_private;
1383}
1384
1385static inline void ext4_inode_aio_set(struct inode *inode, ext4_io_end_t *io)
1386{
1387 inode->i_private = io;
1388}
1389
19f5fb7a
TT
1390/*
1391 * Inode dynamic state flags
1392 */
1393enum {
1394 EXT4_STATE_JDATA, /* journaled data exists */
1395 EXT4_STATE_NEW, /* inode is newly created */
1396 EXT4_STATE_XATTR, /* has in-inode xattrs */
1397 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1398 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1399 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1400 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
14ece102 1401 EXT4_STATE_NEWENTRY, /* File just added to dir */
f2321097 1402 EXT4_STATE_DELALLOC_RESERVED, /* blks already reserved for delalloc */
17335dcc
DM
1403 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read
1404 nolocking */
67cf5b09 1405 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
74d553aa 1406 EXT4_STATE_ORDERED_MODE, /* data=ordered mode */
7869a4a6 1407 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
19f5fb7a
TT
1408};
1409
353eb83c 1410#define EXT4_INODE_BIT_FNS(name, field, offset) \
12e9b892
DM
1411static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1412{ \
353eb83c 1413 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
12e9b892
DM
1414} \
1415static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1416{ \
353eb83c 1417 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
12e9b892
DM
1418} \
1419static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1420{ \
353eb83c 1421 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
19f5fb7a
TT
1422}
1423
3f61c0cc
AD
1424/* Add these declarations here only so that these functions can be
1425 * found by name. Otherwise, they are very hard to locate. */
1426static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1427static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1428static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
353eb83c 1429EXT4_INODE_BIT_FNS(flag, flags, 0)
3f61c0cc
AD
1430
1431/* Add these declarations here only so that these functions can be
1432 * found by name. Otherwise, they are very hard to locate. */
1433static inline int ext4_test_inode_state(struct inode *inode, int bit);
1434static inline void ext4_set_inode_state(struct inode *inode, int bit);
1435static inline void ext4_clear_inode_state(struct inode *inode, int bit);
353eb83c
TT
1436#if (BITS_PER_LONG < 64)
1437EXT4_INODE_BIT_FNS(state, state_flags, 0)
1438
1439static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1440{
1441 (ei)->i_state_flags = 0;
1442}
1443#else
1444EXT4_INODE_BIT_FNS(state, flags, 32)
1445
1446static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1447{
1448 /* We depend on the fact that callers will set i_flags */
1449}
1450#endif
ac27a0ec 1451#else
617ba13b 1452/* Assume that user mode programs are passing in an ext4fs superblock, not
ac27a0ec
DK
1453 * a kernel struct super_block. This will allow us to call the feature-test
1454 * macros from user land. */
617ba13b 1455#define EXT4_SB(sb) (sb)
ac27a0ec
DK
1456#endif
1457
617ba13b 1458#define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
ac27a0ec
DK
1459
1460/*
1461 * Codes for operating systems
1462 */
617ba13b
MC
1463#define EXT4_OS_LINUX 0
1464#define EXT4_OS_HURD 1
1465#define EXT4_OS_MASIX 2
1466#define EXT4_OS_FREEBSD 3
1467#define EXT4_OS_LITES 4
ac27a0ec
DK
1468
1469/*
1470 * Revision levels
1471 */
617ba13b
MC
1472#define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1473#define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
ac27a0ec 1474
617ba13b
MC
1475#define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1476#define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
ac27a0ec 1477
617ba13b 1478#define EXT4_GOOD_OLD_INODE_SIZE 128
ac27a0ec
DK
1479
1480/*
1481 * Feature set definitions
1482 */
1483
617ba13b 1484#define EXT4_HAS_COMPAT_FEATURE(sb,mask) \
3a06d778 1485 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
617ba13b 1486#define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \
3a06d778 1487 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
617ba13b 1488#define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \
3a06d778 1489 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
617ba13b
MC
1490#define EXT4_SET_COMPAT_FEATURE(sb,mask) \
1491 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1492#define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \
1493 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1494#define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \
1495 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1496#define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \
1497 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1498#define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
1499 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1500#define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \
1501 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1502
1503#define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1504#define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1505#define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1506#define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1507#define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1508#define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1beeef1b 1509#define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
617ba13b
MC
1510
1511#define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1512#define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1513#define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
0fc1b451 1514#define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
717d50e4 1515#define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
f8628a14 1516#define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
ef7f3835 1517#define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
ae812306 1518#define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
281b5995 1519#define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
e6153918
DW
1520/*
1521 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1522 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1523 * all other data structures' checksums. However, the METADATA_CSUM and
1524 * GDT_CSUM bits are mutually exclusive.
1525 */
9b90e5e0 1526#define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
617ba13b
MC
1527
1528#define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1529#define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1530#define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1531#define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1532#define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
a86c6181 1533#define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
bd81d8ee 1534#define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
0fc1b451 1535#define EXT4_FEATURE_INCOMPAT_MMP 0x0100
ce421581 1536#define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
f710b4b9
TT
1537#define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1538#define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
856cbcf9 1539#define EXT4_FEATURE_INCOMPAT_BG_USE_META_CSUM 0x2000 /* use crc32c for bg */
9b90e5e0 1540#define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
67cf5b09 1541#define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
617ba13b 1542
2035e776
TT
1543#define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1544#define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1545 EXT4_FEATURE_INCOMPAT_META_BG)
1546#define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1547 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1548 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1549
1550#define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1551#define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1552 EXT4_FEATURE_INCOMPAT_RECOVER| \
1553 EXT4_FEATURE_INCOMPAT_META_BG)
1554#define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1555 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1556 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1557
617ba13b
MC
1558#define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR
1559#define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1560 EXT4_FEATURE_INCOMPAT_RECOVER| \
a86c6181 1561 EXT4_FEATURE_INCOMPAT_META_BG| \
bd81d8ee 1562 EXT4_FEATURE_INCOMPAT_EXTENTS| \
ce421581 1563 EXT4_FEATURE_INCOMPAT_64BIT| \
c5e06d10 1564 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
f08225d1
TM
1565 EXT4_FEATURE_INCOMPAT_MMP | \
1566 EXT4_FEATURE_INCOMPAT_INLINE_DATA)
617ba13b
MC
1567#define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1568 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
717d50e4 1569 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
f8628a14 1570 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
ef7f3835 1571 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
0fc1b451 1572 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
6f16b606 1573 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
e93376c2 1574 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
7c319d32
AK
1575 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1576 EXT4_FEATURE_RO_COMPAT_QUOTA)
ac27a0ec
DK
1577
1578/*
1579 * Default values for user and/or group using reserved blocks
1580 */
617ba13b
MC
1581#define EXT4_DEF_RESUID 0
1582#define EXT4_DEF_RESGID 0
ac27a0ec 1583
240799cd
TT
1584#define EXT4_DEF_INODE_READAHEAD_BLKS 32
1585
ac27a0ec
DK
1586/*
1587 * Default mount options
1588 */
617ba13b
MC
1589#define EXT4_DEFM_DEBUG 0x0001
1590#define EXT4_DEFM_BSDGROUPS 0x0002
1591#define EXT4_DEFM_XATTR_USER 0x0004
1592#define EXT4_DEFM_ACL 0x0008
1593#define EXT4_DEFM_UID16 0x0010
1594#define EXT4_DEFM_JMODE 0x0060
1595#define EXT4_DEFM_JMODE_DATA 0x0020
1596#define EXT4_DEFM_JMODE_ORDERED 0x0040
1597#define EXT4_DEFM_JMODE_WBACK 0x0060
8b67f04a
TT
1598#define EXT4_DEFM_NOBARRIER 0x0100
1599#define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1600#define EXT4_DEFM_DISCARD 0x0400
1601#define EXT4_DEFM_NODELALLOC 0x0800
ac27a0ec 1602
30773840
TT
1603/*
1604 * Default journal batch times
1605 */
1606#define EXT4_DEF_MIN_BATCH_TIME 0
1607#define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1608
a4912123
TT
1609/*
1610 * Minimum number of groups in a flexgroup before we separate out
1611 * directories into the first block group of a flexgroup
1612 */
1613#define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1614
ac27a0ec
DK
1615/*
1616 * Structure of a directory entry
1617 */
617ba13b 1618#define EXT4_NAME_LEN 255
ac27a0ec 1619
617ba13b 1620struct ext4_dir_entry {
ac27a0ec
DK
1621 __le32 inode; /* Inode number */
1622 __le16 rec_len; /* Directory entry length */
1623 __le16 name_len; /* Name length */
617ba13b 1624 char name[EXT4_NAME_LEN]; /* File name */
ac27a0ec
DK
1625};
1626
1627/*
617ba13b 1628 * The new version of the directory entry. Since EXT4 structures are
ac27a0ec
DK
1629 * stored in intel byte order, and the name_len field could never be
1630 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1631 * file_type field.
1632 */
617ba13b 1633struct ext4_dir_entry_2 {
ac27a0ec
DK
1634 __le32 inode; /* Inode number */
1635 __le16 rec_len; /* Directory entry length */
1636 __u8 name_len; /* Name length */
1637 __u8 file_type;
617ba13b 1638 char name[EXT4_NAME_LEN]; /* File name */
ac27a0ec
DK
1639};
1640
e6153918
DW
1641/*
1642 * This is a bogus directory entry at the end of each leaf block that
1643 * records checksums.
1644 */
1645struct ext4_dir_entry_tail {
1646 __le32 det_reserved_zero1; /* Pretend to be unused */
1647 __le16 det_rec_len; /* 12 */
1648 __u8 det_reserved_zero2; /* Zero name length */
1649 __u8 det_reserved_ft; /* 0xDE, fake file type */
1650 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
1651};
1652
3c47d541
TM
1653#define EXT4_DIRENT_TAIL(block, blocksize) \
1654 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1655 ((blocksize) - \
1656 sizeof(struct ext4_dir_entry_tail))))
1657
ac27a0ec 1658/*
617ba13b 1659 * Ext4 directory file types. Only the low 3 bits are used. The
ac27a0ec
DK
1660 * other bits are reserved for now.
1661 */
617ba13b
MC
1662#define EXT4_FT_UNKNOWN 0
1663#define EXT4_FT_REG_FILE 1
1664#define EXT4_FT_DIR 2
1665#define EXT4_FT_CHRDEV 3
1666#define EXT4_FT_BLKDEV 4
1667#define EXT4_FT_FIFO 5
1668#define EXT4_FT_SOCK 6
1669#define EXT4_FT_SYMLINK 7
ac27a0ec 1670
617ba13b 1671#define EXT4_FT_MAX 8
ac27a0ec 1672
e6153918
DW
1673#define EXT4_FT_DIR_CSUM 0xDE
1674
ac27a0ec 1675/*
617ba13b 1676 * EXT4_DIR_PAD defines the directory entries boundaries
ac27a0ec
DK
1677 *
1678 * NOTE: It must be a multiple of 4
1679 */
617ba13b
MC
1680#define EXT4_DIR_PAD 4
1681#define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1682#define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1683 ~EXT4_DIR_ROUND)
a72d7f83
JK
1684#define EXT4_MAX_REC_LEN ((1<<16)-1)
1685
0cfc9255
ES
1686/*
1687 * If we ever get support for fs block sizes > page_size, we'll need
1688 * to remove the #if statements in the next two functions...
1689 */
1690static inline unsigned int
1691ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1692{
1693 unsigned len = le16_to_cpu(dlen);
1694
1695#if (PAGE_CACHE_SIZE >= 65536)
1696 if (len == EXT4_MAX_REC_LEN || len == 0)
1697 return blocksize;
1698 return (len & 65532) | ((len & 3) << 16);
1699#else
1700 return len;
1701#endif
1702}
1703
1704static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1705{
1706 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1707 BUG();
1708#if (PAGE_CACHE_SIZE >= 65536)
1709 if (len < 65536)
1710 return cpu_to_le16(len);
1711 if (len == blocksize) {
1712 if (blocksize == 65536)
1713 return cpu_to_le16(EXT4_MAX_REC_LEN);
1714 else
1715 return cpu_to_le16(0);
1716 }
1717 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1718#else
1719 return cpu_to_le16(len);
1720#endif
1721}
1722
ac27a0ec
DK
1723/*
1724 * Hash Tree Directory indexing
1725 * (c) Daniel Phillips, 2001
1726 */
1727
4074fe37
ES
1728#define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1729 EXT4_FEATURE_COMPAT_DIR_INDEX) && \
12e9b892 1730 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
617ba13b
MC
1731#define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1732#define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
ac27a0ec
DK
1733
1734/* Legal values for the dx_root hash_version field: */
1735
1736#define DX_HASH_LEGACY 0
1737#define DX_HASH_HALF_MD4 1
1738#define DX_HASH_TEA 2
f99b2589
TT
1739#define DX_HASH_LEGACY_UNSIGNED 3
1740#define DX_HASH_HALF_MD4_UNSIGNED 4
1741#define DX_HASH_TEA_UNSIGNED 5
ac27a0ec 1742
0441984a
DW
1743static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
1744 const void *address, unsigned int length)
1745{
1746 struct {
1747 struct shash_desc shash;
3108b54b 1748 char ctx[4];
0441984a
DW
1749 } desc;
1750 int err;
1751
3108b54b
TT
1752 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
1753
0441984a
DW
1754 desc.shash.tfm = sbi->s_chksum_driver;
1755 desc.shash.flags = 0;
1756 *(u32 *)desc.ctx = crc;
1757
1758 err = crypto_shash_update(&desc.shash, address, length);
1759 BUG_ON(err);
1760
1761 return *(u32 *)desc.ctx;
1762}
1763
ac27a0ec
DK
1764#ifdef __KERNEL__
1765
1766/* hash info structure used by the directory hash */
1767struct dx_hash_info
1768{
1769 u32 hash;
1770 u32 minor_hash;
1771 int hash_version;
1772 u32 *seed;
1773};
1774
d1f5273e
FY
1775
1776/* 32 and 64 bit signed EOF for dx directories */
1777#define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
1778#define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
1779
ac27a0ec
DK
1780
1781/*
617ba13b 1782 * Control parameters used by ext4_htree_next_block
ac27a0ec
DK
1783 */
1784#define HASH_NB_ALWAYS 1
1785
1786
1787/*
1788 * Describe an inode's exact location on disk and in memory
1789 */
617ba13b 1790struct ext4_iloc
ac27a0ec
DK
1791{
1792 struct buffer_head *bh;
1793 unsigned long offset;
fd2d4291 1794 ext4_group_t block_group;
ac27a0ec
DK
1795};
1796
617ba13b 1797static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
ac27a0ec 1798{
617ba13b 1799 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
ac27a0ec
DK
1800}
1801
1802/*
1803 * This structure is stuffed into the struct file's private_data field
1804 * for directories. It is where we put information so that we can do
1805 * readdir operations in hash tree order.
1806 */
1807struct dir_private_info {
1808 struct rb_root root;
1809 struct rb_node *curr_node;
1810 struct fname *extra_fname;
1811 loff_t last_pos;
1812 __u32 curr_hash;
1813 __u32 curr_minor_hash;
1814 __u32 next_hash;
1815};
1816
1817/* calculate the first block number of the group */
617ba13b 1818static inline ext4_fsblk_t
fd2d4291 1819ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
ac27a0ec 1820{
617ba13b
MC
1821 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
1822 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
ac27a0ec
DK
1823}
1824
1825/*
1826 * Special error return code only used by dx_probe() and its callers.
1827 */
36de9286 1828#define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
ac27a0ec 1829
bfff6873
LC
1830/*
1831 * Timeout and state flag for lazy initialization inode thread.
1832 */
1833#define EXT4_DEF_LI_WAIT_MULT 10
1834#define EXT4_DEF_LI_MAX_START_DELAY 5
1835#define EXT4_LAZYINIT_QUIT 0x0001
1836#define EXT4_LAZYINIT_RUNNING 0x0002
1837
1838/*
1839 * Lazy inode table initialization info
1840 */
1841struct ext4_lazy_init {
1842 unsigned long li_state;
bfff6873
LC
1843 struct list_head li_request_list;
1844 struct mutex li_list_mtx;
1845};
1846
1847struct ext4_li_request {
1848 struct super_block *lr_super;
1849 struct ext4_sb_info *lr_sbi;
1850 ext4_group_t lr_next_group;
1851 struct list_head lr_request;
1852 unsigned long lr_next_sched;
1853 unsigned long lr_timeout;
1854};
1855
857ac889
LC
1856struct ext4_features {
1857 struct kobject f_kobj;
1858 struct completion f_kobj_unregister;
1859};
1860
c5e06d10
JL
1861/*
1862 * This structure will be used for multiple mount protection. It will be
1863 * written into the block number saved in the s_mmp_block field in the
1864 * superblock. Programs that check MMP should assume that if
1865 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
1866 * to use the filesystem, regardless of how old the timestamp is.
1867 */
1868#define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
1869#define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
1870#define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
1871#define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
1872
1873struct mmp_struct {
1874 __le32 mmp_magic; /* Magic number for MMP */
1875 __le32 mmp_seq; /* Sequence no. updated periodically */
1876
1877 /*
1878 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
1879 * purposes and do not affect the correctness of the algorithm
1880 */
1881 __le64 mmp_time; /* Time last updated */
1882 char mmp_nodename[64]; /* Node which last updated MMP block */
1883 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
1884
1885 /*
1886 * mmp_check_interval is used to verify if the MMP block has been
1887 * updated on the block device. The value is updated based on the
1888 * maximum time to write the MMP block during an update cycle.
1889 */
1890 __le16 mmp_check_interval;
1891
1892 __le16 mmp_pad1;
e6153918
DW
1893 __le32 mmp_pad2[226];
1894 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
c5e06d10
JL
1895};
1896
1897/* arguments passed to the mmp thread */
1898struct mmpd_data {
1899 struct buffer_head *bh; /* bh from initial read_mmp_block() */
1900 struct super_block *sb; /* super block of the fs */
1901};
1902
1903/*
1904 * Check interval multiplier
1905 * The MMP block is written every update interval and initially checked every
1906 * update interval x the multiplier (the value is then adapted based on the
1907 * write latency). The reason is that writes can be delayed under load and we
1908 * don't want readers to incorrectly assume that the filesystem is no longer
1909 * in use.
1910 */
1911#define EXT4_MMP_CHECK_MULT 2UL
1912
1913/*
1914 * Minimum interval for MMP checking in seconds.
1915 */
1916#define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
1917
1918/*
1919 * Maximum interval for MMP checking in seconds.
1920 */
1921#define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
1922
ac27a0ec
DK
1923/*
1924 * Function prototypes
1925 */
1926
1927/*
1928 * Ok, these declarations are also in <linux/kernel.h> but none of the
617ba13b 1929 * ext4 source programs needs to include it so they are duplicated here.
ac27a0ec 1930 */
8c55e204
DK
1931# define NORET_TYPE /**/
1932# define ATTRIB_NORET __attribute__((noreturn))
1933# define NORET_AND noreturn,
ac27a0ec 1934
498e5f24 1935/* bitmap.c */
f6fb99ca 1936extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
41a246d1
DW
1937void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
1938 struct ext4_group_desc *gdp,
1939 struct buffer_head *bh, int sz);
1940int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
1941 struct ext4_group_desc *gdp,
1942 struct buffer_head *bh, int sz);
fa77dcfa
DW
1943void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
1944 struct ext4_group_desc *gdp,
79f1ba49 1945 struct buffer_head *bh);
fa77dcfa
DW
1946int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
1947 struct ext4_group_desc *gdp,
79f1ba49 1948 struct buffer_head *bh);
498e5f24 1949
ac27a0ec 1950/* balloc.c */
bd86298e
LC
1951extern void ext4_get_group_no_and_offset(struct super_block *sb,
1952 ext4_fsblk_t blocknr,
1953 ext4_group_t *blockgrpp,
1954 ext4_grpblk_t *offsetp);
1955extern ext4_group_t ext4_get_group_number(struct super_block *sb,
1956 ext4_fsblk_t block);
1957
3a5b2ecd
MC
1958extern unsigned int ext4_block_group(struct super_block *sb,
1959 ext4_fsblk_t blocknr);
1960extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
1961 ext4_fsblk_t blocknr);
fd2d4291
AM
1962extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
1963extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
1964 ext4_group_t group);
7061eba7 1965extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
55f020db
AH
1966 ext4_fsblk_t goal,
1967 unsigned int flags,
1968 unsigned long *count,
1969 int *errp);
e7d5f315
TT
1970extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
1971 s64 nclusters, unsigned int flags);
5dee5437 1972extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
af5bc92d 1973extern void ext4_check_blocks_bitmap(struct super_block *);
617ba13b 1974extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
fd2d4291 1975 ext4_group_t block_group,
ac27a0ec 1976 struct buffer_head ** bh);
617ba13b 1977extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
813e5727
TT
1978
1979extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
1980 ext4_group_t block_group);
1981extern int ext4_wait_block_bitmap(struct super_block *sb,
1982 ext4_group_t block_group,
1983 struct buffer_head *bh);
1984extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
1985 ext4_group_t block_group);
cff1dfd7
TT
1986extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
1987 ext4_group_t block_group,
1988 struct ext4_group_desc *gdp);
f86186b4 1989ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
ac27a0ec
DK
1990
1991/* dir.c */
60fd4da3 1992extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
f7c21177 1993 struct file *,
60fd4da3 1994 struct ext4_dir_entry_2 *,
226ba972
TM
1995 struct buffer_head *, char *, int,
1996 unsigned int);
1997#define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
f7c21177 1998 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
226ba972 1999 (de), (bh), (buf), (size), (offset)))
617ba13b 2000extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
ac27a0ec 2001 __u32 minor_hash,
617ba13b
MC
2002 struct ext4_dir_entry_2 *dirent);
2003extern void ext4_htree_free_dir_info(struct dir_private_info *p);
978fef91
TM
2004extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2005 struct buffer_head *bh,
2006 void *buf, int buf_size,
2007 const char *name, int namelen,
2008 struct ext4_dir_entry_2 **dest_de);
2009void ext4_insert_dentry(struct inode *inode,
2010 struct ext4_dir_entry_2 *de,
2011 int buf_size,
2012 const char *name, int namelen);
2013static inline void ext4_update_dx_flag(struct inode *inode)
2014{
2015 if (!EXT4_HAS_COMPAT_FEATURE(inode->i_sb,
2016 EXT4_FEATURE_COMPAT_DIR_INDEX))
2017 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2018}
65d165d9
TM
2019static unsigned char ext4_filetype_table[] = {
2020 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2021};
2022
2023static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2024{
2025 if (!EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE) ||
2026 (filetype >= EXT4_FT_MAX))
2027 return DT_UNKNOWN;
2028
2029 return ext4_filetype_table[filetype];
2030}
40b163f1
DW
2031extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2032 void *buf, int buf_size);
ac27a0ec
DK
2033
2034/* fsync.c */
02c24a82 2035extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
ac27a0ec
DK
2036
2037/* hash.c */
617ba13b 2038extern int ext4fs_dirhash(const char *name, int len, struct
ac27a0ec
DK
2039 dx_hash_info *hinfo);
2040
2041/* ialloc.c */
1139575a
TT
2042extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2043 const struct qstr *qstr, __u32 goal,
2044 uid_t *owner, int handle_type,
2045 unsigned int line_no, int nblocks);
2046
2047#define ext4_new_inode(handle, dir, mode, qstr, goal, owner) \
2048 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2049 0, 0, 0)
2050#define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2051 type, nblocks) \
2052 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2053 (type), __LINE__, (nblocks))
2054
2055
af5bc92d
TT
2056extern void ext4_free_inode(handle_t *, struct inode *);
2057extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2058extern unsigned long ext4_count_free_inodes(struct super_block *);
2059extern unsigned long ext4_count_dirs(struct super_block *);
2060extern void ext4_check_inodes_bitmap(struct super_block *);
61d08673 2061extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
bfff6873
LC
2062extern int ext4_init_inode_table(struct super_block *sb,
2063 ext4_group_t group, int barrier);
813e5727 2064extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
ac27a0ec 2065
c9de560d
AT
2066/* mballoc.c */
2067extern long ext4_mb_stats;
2068extern long ext4_mb_max_to_scan;
9d99012f 2069extern int ext4_mb_init(struct super_block *);
c9de560d
AT
2070extern int ext4_mb_release(struct super_block *);
2071extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2072 struct ext4_allocation_request *, int *);
2073extern int ext4_mb_reserve_blocks(struct super_block *, int);
c2ea3fde 2074extern void ext4_discard_preallocations(struct inode *);
5dabfc78
TT
2075extern int __init ext4_init_mballoc(void);
2076extern void ext4_exit_mballoc(void);
44338711 2077extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
e6362609
TT
2078 struct buffer_head *bh, ext4_fsblk_t block,
2079 unsigned long count, int flags);
28623c2f
TT
2080extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2081 ext4_group_t ngroups);
920313a7 2082extern int ext4_mb_add_groupinfo(struct super_block *sb,
5f21b0e6 2083 ext4_group_t i, struct ext4_group_desc *desc);
cc7365df 2084extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2846e820 2085 ext4_fsblk_t block, unsigned long count);
7360d173
LC
2086extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2087
ac27a0ec 2088/* inode.c */
10560082 2089struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
725d26d3
AK
2090struct buffer_head *ext4_bread(handle_t *, struct inode *,
2091 ext4_lblk_t, int, int *);
f19d5870
TM
2092int ext4_get_block_write(struct inode *inode, sector_t iblock,
2093 struct buffer_head *bh_result, int create);
6873fa0d
ES
2094int ext4_get_block(struct inode *inode, sector_t iblock,
2095 struct buffer_head *bh_result, int create);
9c3569b5
TM
2096int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2097 struct buffer_head *bh, int create);
f19d5870
TM
2098int ext4_walk_page_buffers(handle_t *handle,
2099 struct buffer_head *head,
2100 unsigned from,
2101 unsigned to,
2102 int *partial,
2103 int (*fn)(handle_t *handle,
2104 struct buffer_head *bh));
2105int do_journal_get_write_access(handle_t *handle,
2106 struct buffer_head *bh);
9c3569b5
TM
2107#define FALL_BACK_TO_NONDELALLOC 1
2108#define CONVERT_INLINE_DATA 2
ac27a0ec 2109
1d1fe1ee 2110extern struct inode *ext4_iget(struct super_block *, unsigned long);
a9185b41 2111extern int ext4_write_inode(struct inode *, struct writeback_control *);
af5bc92d 2112extern int ext4_setattr(struct dentry *, struct iattr *);
3e3398a0
MC
2113extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
2114 struct kstat *stat);
0930fcc1
AV
2115extern void ext4_evict_inode(struct inode *);
2116extern void ext4_clear_inode(struct inode *);
af5bc92d 2117extern int ext4_sync_inode(handle_t *, struct inode *);
aa385729 2118extern void ext4_dirty_inode(struct inode *, int);
617ba13b
MC
2119extern int ext4_change_inode_journal_flag(struct inode *, int);
2120extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
a361293f 2121extern int ext4_inode_attach_jinode(struct inode *inode);
91ef4caf 2122extern int ext4_can_truncate(struct inode *inode);
af5bc92d 2123extern void ext4_truncate(struct inode *);
aeb2817a 2124extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
487caeef 2125extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
617ba13b 2126extern void ext4_set_inode_flags(struct inode *);
ff9ddf7e 2127extern void ext4_get_inode_flags(struct ext4_inode_info *);
ccd2506b 2128extern int ext4_alloc_da_blocks(struct inode *inode);
617ba13b 2129extern void ext4_set_aops(struct inode *inode);
a86c6181 2130extern int ext4_writepage_trans_blocks(struct inode *);
f3bd1f3f 2131extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
a87dd18c
LC
2132extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2133 loff_t lstart, loff_t lend);
c2ec175c 2134extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
a9e7f447 2135extern qsize_t *ext4_get_reserved_space(struct inode *inode);
5f634d06
AK
2136extern void ext4_da_update_reserve_space(struct inode *inode,
2137 int used, int quota_claim);
dae1e52c
AG
2138
2139/* indirect.c */
2140extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2141 struct ext4_map_blocks *map, int flags);
2142extern ssize_t ext4_ind_direct_IO(int rw, struct kiocb *iocb,
16b1f05d 2143 struct iov_iter *iter, loff_t offset);
dae1e52c 2144extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
fa55a0ed 2145extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
819c4920 2146extern void ext4_ind_truncate(handle_t *, struct inode *inode);
4f579ae7
LC
2147extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2148 ext4_lblk_t start, ext4_lblk_t end);
ff9893dc 2149
ac27a0ec 2150/* ioctl.c */
5cdd7b2d 2151extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
af5bc92d 2152extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
ac27a0ec 2153
c14c6fd5 2154/* migrate.c */
2a43a878 2155extern int ext4_ext_migrate(struct inode *);
0d14b098 2156extern int ext4_ind_migrate(struct inode *inode);
3d0518f4 2157
ac27a0ec 2158/* namei.c */
b0336e8d
DW
2159extern int ext4_dirent_csum_verify(struct inode *inode,
2160 struct ext4_dir_entry *dirent);
617ba13b
MC
2161extern int ext4_orphan_add(handle_t *, struct inode *);
2162extern int ext4_orphan_del(handle_t *, struct inode *);
2163extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
ac27a0ec 2164 __u32 start_minor_hash, __u32 *next_hash);
7335cd3b
TM
2165extern int search_dir(struct buffer_head *bh,
2166 char *search_buf,
2167 int buf_size,
2168 struct inode *dir,
2169 const struct qstr *d_name,
2170 unsigned int offset,
2171 struct ext4_dir_entry_2 **res_dir);
05019a9e
TM
2172extern int ext4_generic_delete_entry(handle_t *handle,
2173 struct inode *dir,
2174 struct ext4_dir_entry_2 *de_del,
2175 struct buffer_head *bh,
2176 void *entry_buf,
2177 int buf_size,
2178 int csum_size);
ac27a0ec
DK
2179
2180/* resize.c */
617ba13b
MC
2181extern int ext4_group_add(struct super_block *sb,
2182 struct ext4_new_group_data *input);
2183extern int ext4_group_extend(struct super_block *sb,
2184 struct ext4_super_block *es,
2185 ext4_fsblk_t n_blocks_count);
19c5246d 2186extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
ac27a0ec
DK
2187
2188/* super.c */
952fc18e 2189extern int ext4_calculate_overhead(struct super_block *sb);
06db49e6 2190extern void ext4_superblock_csum_set(struct super_block *sb);
9933fc0a
TT
2191extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2192extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2193extern void ext4_kvfree(void *ptr);
117fff10
TT
2194extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2195 ext4_group_t ngroup);
722887dd
TT
2196extern const char *ext4_decode_error(struct super_block *sb, int errno,
2197 char nbuf[16]);
e7c96e8e 2198
b9075fa9
JP
2199extern __printf(4, 5)
2200void __ext4_error(struct super_block *, const char *, unsigned int,
2201 const char *, ...);
b9075fa9 2202extern __printf(5, 6)
e7c96e8e 2203void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
b9075fa9
JP
2204 const char *, ...);
2205extern __printf(5, 6)
e7c96e8e 2206void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
b9075fa9 2207 const char *, ...);
c398eda0
TT
2208extern void __ext4_std_error(struct super_block *, const char *,
2209 unsigned int, int);
b9075fa9
JP
2210extern __printf(4, 5)
2211void __ext4_abort(struct super_block *, const char *, unsigned int,
2212 const char *, ...);
b9075fa9
JP
2213extern __printf(4, 5)
2214void __ext4_warning(struct super_block *, const char *, unsigned int,
2215 const char *, ...);
b9075fa9 2216extern __printf(3, 4)
e7c96e8e 2217void __ext4_msg(struct super_block *, const char *, const char *, ...);
c5e06d10
JL
2218extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2219 const char *, unsigned int, const char *);
b9075fa9
JP
2220extern __printf(7, 8)
2221void __ext4_grp_locked_error(const char *, unsigned int,
2222 struct super_block *, ext4_group_t,
2223 unsigned long, ext4_fsblk_t,
2224 const char *, ...);
e7c96e8e
JP
2225
2226#ifdef CONFIG_PRINTK
2227
2228#define ext4_error_inode(inode, func, line, block, fmt, ...) \
2229 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__)
2230#define ext4_error_file(file, func, line, block, fmt, ...) \
2231 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2232#define ext4_error(sb, fmt, ...) \
2233 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2234#define ext4_abort(sb, fmt, ...) \
2235 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2236#define ext4_warning(sb, fmt, ...) \
2237 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2238#define ext4_msg(sb, level, fmt, ...) \
2239 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2240#define dump_mmp_msg(sb, mmp, msg) \
2241 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2242#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2243 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2244 fmt, ##__VA_ARGS__)
2245
2246#else
2247
2248#define ext4_error_inode(inode, func, line, block, fmt, ...) \
2249do { \
2250 no_printk(fmt, ##__VA_ARGS__); \
2251 __ext4_error_inode(inode, "", 0, block, " "); \
2252} while (0)
2253#define ext4_error_file(file, func, line, block, fmt, ...) \
2254do { \
2255 no_printk(fmt, ##__VA_ARGS__); \
2256 __ext4_error_file(file, "", 0, block, " "); \
2257} while (0)
2258#define ext4_error(sb, fmt, ...) \
2259do { \
2260 no_printk(fmt, ##__VA_ARGS__); \
2261 __ext4_error(sb, "", 0, " "); \
2262} while (0)
2263#define ext4_abort(sb, fmt, ...) \
2264do { \
2265 no_printk(fmt, ##__VA_ARGS__); \
2266 __ext4_abort(sb, "", 0, " "); \
2267} while (0)
2268#define ext4_warning(sb, fmt, ...) \
2269do { \
2270 no_printk(fmt, ##__VA_ARGS__); \
2271 __ext4_warning(sb, "", 0, " "); \
2272} while (0)
2273#define ext4_msg(sb, level, fmt, ...) \
2274do { \
2275 no_printk(fmt, ##__VA_ARGS__); \
2276 __ext4_msg(sb, "", " "); \
2277} while (0)
2278#define dump_mmp_msg(sb, mmp, msg) \
2279 __dump_mmp_msg(sb, mmp, "", 0, "")
2280#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2281do { \
2282 no_printk(fmt, ##__VA_ARGS__); \
2283 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
2284} while (0)
2285
2286#endif
2287
af5bc92d 2288extern void ext4_update_dynamic_rev(struct super_block *sb);
99e6f829
AK
2289extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2290 __u32 compat);
2291extern int ext4_update_rocompat_feature(handle_t *handle,
2292 struct super_block *sb, __u32 rocompat);
2293extern int ext4_update_incompat_feature(handle_t *handle,
2294 struct super_block *sb, __u32 incompat);
8fadc143
AR
2295extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2296 struct ext4_group_desc *bg);
2297extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2298 struct ext4_group_desc *bg);
2299extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2300 struct ext4_group_desc *bg);
021b65bb
TT
2301extern __u32 ext4_free_group_clusters(struct super_block *sb,
2302 struct ext4_group_desc *bg);
560671a0
AK
2303extern __u32 ext4_free_inodes_count(struct super_block *sb,
2304 struct ext4_group_desc *bg);
2305extern __u32 ext4_used_dirs_count(struct super_block *sb,
2306 struct ext4_group_desc *bg);
2307extern __u32 ext4_itable_unused_count(struct super_block *sb,
2308 struct ext4_group_desc *bg);
8fadc143
AR
2309extern void ext4_block_bitmap_set(struct super_block *sb,
2310 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2311extern void ext4_inode_bitmap_set(struct super_block *sb,
2312 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2313extern void ext4_inode_table_set(struct super_block *sb,
2314 struct ext4_group_desc *bg, ext4_fsblk_t blk);
021b65bb
TT
2315extern void ext4_free_group_clusters_set(struct super_block *sb,
2316 struct ext4_group_desc *bg,
2317 __u32 count);
560671a0
AK
2318extern void ext4_free_inodes_set(struct super_block *sb,
2319 struct ext4_group_desc *bg, __u32 count);
2320extern void ext4_used_dirs_set(struct super_block *sb,
2321 struct ext4_group_desc *bg, __u32 count);
2322extern void ext4_itable_unused_set(struct super_block *sb,
2323 struct ext4_group_desc *bg, __u32 count);
feb0ab32 2324extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
bb23c20a 2325 struct ext4_group_desc *gdp);
feb0ab32
DW
2326extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2327 struct ext4_group_desc *gdp);
7f511862
TT
2328extern int ext4_register_li_request(struct super_block *sb,
2329 ext4_group_t first_not_zeroed);
bd81d8ee 2330
41a246d1
DW
2331static inline int ext4_has_group_desc_csum(struct super_block *sb)
2332{
2333 return EXT4_HAS_RO_COMPAT_FEATURE(sb,
2334 EXT4_FEATURE_RO_COMPAT_GDT_CSUM |
2335 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM);
2336}
bd81d8ee
LV
2337
2338static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2339{
2340 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
6bc9feff 2341 le32_to_cpu(es->s_blocks_count_lo);
bd81d8ee
LV
2342}
2343
2344static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2345{
2346 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
308ba3ec 2347 le32_to_cpu(es->s_r_blocks_count_lo);
bd81d8ee
LV
2348}
2349
2350static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2351{
2352 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
308ba3ec 2353 le32_to_cpu(es->s_free_blocks_count_lo);
bd81d8ee
LV
2354}
2355
2356static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2357 ext4_fsblk_t blk)
2358{
6bc9feff 2359 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
bd81d8ee
LV
2360 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2361}
2362
2363static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2364 ext4_fsblk_t blk)
2365{
308ba3ec 2366 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
bd81d8ee
LV
2367 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2368}
2369
2370static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2371 ext4_fsblk_t blk)
2372{
308ba3ec 2373 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
bd81d8ee
LV
2374 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2375}
2376
a48380f7
AK
2377static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
2378{
06a279d6
TT
2379 if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2380 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2381 le32_to_cpu(raw_inode->i_size_lo);
2382 else
2383 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
a48380f7 2384}
bd81d8ee 2385
a48380f7
AK
2386static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2387{
2388 raw_inode->i_size_lo = cpu_to_le32(i_size);
2389 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2390}
ac27a0ec 2391
c9de560d
AT
2392static inline
2393struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
af5bc92d 2394 ext4_group_t group)
c9de560d
AT
2395{
2396 struct ext4_group_info ***grp_info;
2397 long indexv, indexh;
2f2e09eb 2398 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
c9de560d
AT
2399 grp_info = EXT4_SB(sb)->s_group_info;
2400 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2401 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2402 return grp_info[indexv][indexh];
2403}
2404
8df9675f
TT
2405/*
2406 * Reading s_groups_count requires using smp_rmb() afterwards. See
2407 * the locking protocol documented in the comments of ext4_group_add()
2408 * in resize.c
2409 */
2410static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2411{
2412 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
2413
2414 smp_rmb();
2415 return ngroups;
2416}
c9de560d 2417
772cb7c8
JS
2418static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2419 ext4_group_t block_group)
2420{
2421 return block_group >> sbi->s_log_groups_per_flex;
2422}
2423
2424static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2425{
2426 return 1 << sbi->s_log_groups_per_flex;
2427}
2428
617ba13b 2429#define ext4_std_error(sb, errno) \
ac27a0ec
DK
2430do { \
2431 if ((errno)) \
c398eda0 2432 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
ac27a0ec
DK
2433} while (0)
2434
a30d542a 2435#ifdef CONFIG_SMP
df55c99d
TT
2436/* Each CPU can accumulate percpu_counter_batch clusters in their local
2437 * counters. So we need to make sure we have free clusters more
179f7ebf 2438 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
a30d542a 2439 */
df55c99d 2440#define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
a30d542a 2441#else
df55c99d 2442#define EXT4_FREECLUSTERS_WATERMARK 0
a30d542a
AK
2443#endif
2444
90e775b7 2445/* Update i_disksize. Requires i_mutex to avoid races with truncate */
cf17fea6
AK
2446static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2447{
90e775b7
JK
2448 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
2449 !mutex_is_locked(&inode->i_mutex));
2450 down_write(&EXT4_I(inode)->i_data_sem);
2451 if (newsize > EXT4_I(inode)->i_disksize)
2452 EXT4_I(inode)->i_disksize = newsize;
2453 up_write(&EXT4_I(inode)->i_data_sem);
2454}
2455
4631dbf6
DM
2456/* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
2457static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
2458{
2459 int changed = 0;
2460
2461 if (newsize > inode->i_size) {
2462 i_size_write(inode, newsize);
2463 changed = 1;
2464 }
2465 if (newsize > EXT4_I(inode)->i_disksize) {
2466 ext4_update_i_disksize(inode, newsize);
2467 changed |= 2;
2468 }
2469 return changed;
2470}
2471
5d1b1b3f
AK
2472struct ext4_group_info {
2473 unsigned long bb_state;
2474 struct rb_root bb_free_root;
a36b4498
ES
2475 ext4_grpblk_t bb_first_free; /* first free block */
2476 ext4_grpblk_t bb_free; /* total free blocks */
2477 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
8a57d9d6 2478 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
5d1b1b3f
AK
2479 struct list_head bb_prealloc_list;
2480#ifdef DOUBLE_CHECK
2481 void *bb_bitmap;
2482#endif
2483 struct rw_semaphore alloc_sem;
a36b4498
ES
2484 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
2485 * regions, index is order.
2486 * bb_counters[3] = 5 means
2487 * 5 free 8-block regions. */
5d1b1b3f
AK
2488};
2489
3d56b8d2
TM
2490#define EXT4_GROUP_INFO_NEED_INIT_BIT 0
2491#define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
163a203d 2492#define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
87a39389 2493#define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
5d1b1b3f
AK
2494
2495#define EXT4_MB_GRP_NEED_INIT(grp) \
2496 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
163a203d
DW
2497#define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
2498 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
87a39389
DW
2499#define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
2500 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
5d1b1b3f 2501
3d56b8d2
TM
2502#define EXT4_MB_GRP_WAS_TRIMMED(grp) \
2503 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2504#define EXT4_MB_GRP_SET_TRIMMED(grp) \
2505 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2506#define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
2507 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2508
50797481
TT
2509#define EXT4_MAX_CONTENTION 8
2510#define EXT4_CONTENTION_THRESHOLD 2
2511
955ce5f5
AK
2512static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2513 ext4_group_t group)
5d1b1b3f 2514{
955ce5f5
AK
2515 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2516}
5d1b1b3f 2517
50797481
TT
2518/*
2519 * Returns true if the filesystem is busy enough that attempts to
2520 * access the block group locks has run into contention.
2521 */
2522static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2523{
2524 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2525}
2526
955ce5f5
AK
2527static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2528{
50797481
TT
2529 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2530 if (spin_trylock(lock))
2531 /*
2532 * We're able to grab the lock right away, so drop the
2533 * lock contention counter.
2534 */
2535 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2536 else {
2537 /*
2538 * The lock is busy, so bump the contention counter,
2539 * and then wait on the spin lock.
2540 */
2541 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2542 EXT4_MAX_CONTENTION);
2543 spin_lock(lock);
2544 }
5d1b1b3f
AK
2545}
2546
2547static inline void ext4_unlock_group(struct super_block *sb,
2548 ext4_group_t group)
2549{
955ce5f5 2550 spin_unlock(ext4_group_lock_ptr(sb, group));
5d1b1b3f
AK
2551}
2552
1f7d1e77
TT
2553/*
2554 * Block validity checking
2555 */
2556#define ext4_check_indirect_blockref(inode, bh) \
2557 ext4_check_blockref(__func__, __LINE__, inode, \
2558 (__le32 *)(bh)->b_data, \
2559 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2560
2561#define ext4_ind_check_inode(inode) \
2562 ext4_check_blockref(__func__, __LINE__, inode, \
2563 EXT4_I(inode)->i_data, \
2564 EXT4_NDIR_BLOCKS)
2565
ac27a0ec
DK
2566/*
2567 * Inodes and files operations
2568 */
2569
2570/* dir.c */
617ba13b 2571extern const struct file_operations ext4_dir_operations;
ac27a0ec
DK
2572
2573/* file.c */
c5ef1c42 2574extern const struct inode_operations ext4_file_inode_operations;
617ba13b 2575extern const struct file_operations ext4_file_operations;
e0d10bfa 2576extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
ac27a0ec 2577
95eaefbd 2578/* inline.c */
95eaefbd
TT
2579extern int ext4_get_max_inline_size(struct inode *inode);
2580extern int ext4_find_inline_data_nolock(struct inode *inode);
95eaefbd
TT
2581extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
2582 unsigned int len);
2583extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
2584
2585extern int ext4_readpage_inline(struct inode *inode, struct page *page);
2586extern int ext4_try_to_write_inline_data(struct address_space *mapping,
2587 struct inode *inode,
2588 loff_t pos, unsigned len,
2589 unsigned flags,
2590 struct page **pagep);
2591extern int ext4_write_inline_data_end(struct inode *inode,
2592 loff_t pos, unsigned len,
2593 unsigned copied,
2594 struct page *page);
2595extern struct buffer_head *
2596ext4_journalled_write_inline_data(struct inode *inode,
2597 unsigned len,
2598 struct page *page);
2599extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
2600 struct inode *inode,
2601 loff_t pos, unsigned len,
2602 unsigned flags,
2603 struct page **pagep,
2604 void **fsdata);
2605extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
2606 unsigned len, unsigned copied,
2607 struct page *page);
2608extern int ext4_try_add_inline_entry(handle_t *handle, struct dentry *dentry,
2609 struct inode *inode);
2610extern int ext4_try_create_inline_dir(handle_t *handle,
2611 struct inode *parent,
2612 struct inode *inode);
2613extern int ext4_read_inline_dir(struct file *filp,
725bebb2 2614 struct dir_context *ctx,
95eaefbd 2615 int *has_inline_data);
8af0f082
TM
2616extern int htree_inlinedir_to_tree(struct file *dir_file,
2617 struct inode *dir, ext4_lblk_t block,
2618 struct dx_hash_info *hinfo,
2619 __u32 start_hash, __u32 start_minor_hash,
2620 int *has_inline_data);
95eaefbd
TT
2621extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
2622 const struct qstr *d_name,
2623 struct ext4_dir_entry_2 **res_dir,
2624 int *has_inline_data);
2625extern int ext4_delete_inline_entry(handle_t *handle,
2626 struct inode *dir,
2627 struct ext4_dir_entry_2 *de_del,
2628 struct buffer_head *bh,
2629 int *has_inline_data);
2630extern int empty_inline_dir(struct inode *dir, int *has_inline_data);
2631extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
2632 struct ext4_dir_entry_2 **parent_de,
2633 int *retval);
2634extern int ext4_inline_data_fiemap(struct inode *inode,
2635 struct fiemap_extent_info *fieinfo,
2636 int *has_inline);
2637extern int ext4_try_to_evict_inline_data(handle_t *handle,
2638 struct inode *inode,
2639 int needed);
2640extern void ext4_inline_data_truncate(struct inode *inode, int *has_inline);
2641
2642extern int ext4_convert_inline_data(struct inode *inode);
2643
83447ccb
ZL
2644static inline int ext4_has_inline_data(struct inode *inode)
2645{
2646 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
2647 EXT4_I(inode)->i_inline_off;
2648}
2649
ac27a0ec 2650/* namei.c */
c5ef1c42
AV
2651extern const struct inode_operations ext4_dir_inode_operations;
2652extern const struct inode_operations ext4_special_inode_operations;
596397b7 2653extern struct dentry *ext4_get_parent(struct dentry *child);
a774f9c2
TM
2654extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
2655 struct ext4_dir_entry_2 *de,
2656 int blocksize, int csum_size,
2657 unsigned int parent_ino, int dotdot_real_len);
3c47d541
TM
2658extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t,
2659 unsigned int blocksize);
2660extern int ext4_handle_dirty_dirent_node(handle_t *handle,
2661 struct inode *inode,
2662 struct buffer_head *bh);
8af0f082
TM
2663#define S_SHIFT 12
2664static unsigned char ext4_type_by_mode[S_IFMT >> S_SHIFT] = {
2665 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
2666 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
2667 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
2668 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
2669 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
2670 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
2671 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
2672};
2673
2674static inline void ext4_set_de_type(struct super_block *sb,
2675 struct ext4_dir_entry_2 *de,
2676 umode_t mode) {
2677 if (EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE))
2678 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
2679}
2680
ac27a0ec
DK
2681
2682/* symlink.c */
c5ef1c42
AV
2683extern const struct inode_operations ext4_symlink_inode_operations;
2684extern const struct inode_operations ext4_fast_symlink_inode_operations;
ac27a0ec 2685
6fd058f7
TT
2686/* block_validity */
2687extern void ext4_release_system_zone(struct super_block *sb);
2688extern int ext4_setup_system_zone(struct super_block *sb);
5dabfc78
TT
2689extern int __init ext4_init_system_zone(void);
2690extern void ext4_exit_system_zone(void);
6fd058f7
TT
2691extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
2692 ext4_fsblk_t start_blk,
2693 unsigned int count);
1f7d1e77
TT
2694extern int ext4_check_blockref(const char *, unsigned int,
2695 struct inode *, __le32 *, unsigned int);
6fd058f7 2696
a86c6181 2697/* extents.c */
4a092d73
TT
2698struct ext4_ext_path;
2699struct ext4_extent;
2700
d7b2a00c
ZL
2701/*
2702 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
2703 * __le32.
2704 */
2705#define EXT_MAX_BLOCKS 0xffffffff
2706
a86c6181
AT
2707extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
2708extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
fffb2739 2709extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
e35fd660
TT
2710extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
2711 struct ext4_map_blocks *map, int flags);
819c4920 2712extern void ext4_ext_truncate(handle_t *, struct inode *);
26a4c0c6
TT
2713extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
2714 ext4_lblk_t end);
a86c6181
AT
2715extern void ext4_ext_init(struct super_block *);
2716extern void ext4_ext_release(struct super_block *);
2fe17c10 2717extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
a2df2a63 2718 loff_t len);
6b523df4
JK
2719extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
2720 loff_t offset, ssize_t len);
e35fd660
TT
2721extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
2722 struct ext4_map_blocks *map, int flags);
4a092d73
TT
2723extern int ext4_ext_calc_metadata_amount(struct inode *inode,
2724 ext4_lblk_t lblocks);
2725extern int ext4_extent_tree_init(handle_t *, struct inode *);
2726extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
2727 int num,
2728 struct ext4_ext_path *path);
2729extern int ext4_can_extents_be_merged(struct inode *inode,
2730 struct ext4_extent *ex1,
2731 struct ext4_extent *ex2);
2732extern int ext4_ext_insert_extent(handle_t *, struct inode *,
2733 struct ext4_ext_path *,
2734 struct ext4_extent *, int);
2735extern struct ext4_ext_path *ext4_ext_find_extent(struct inode *, ext4_lblk_t,
107a7bd3
TT
2736 struct ext4_ext_path *,
2737 int flags);
4a092d73
TT
2738extern void ext4_ext_drop_refs(struct ext4_ext_path *);
2739extern int ext4_ext_check_inode(struct inode *inode);
f7fec032
ZL
2740extern int ext4_find_delalloc_range(struct inode *inode,
2741 ext4_lblk_t lblk_start,
2742 ext4_lblk_t lblk_end);
4a092d73 2743extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk);
3a06d778
AK
2744extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
2745 __u64 start, __u64 len);
7869a4a6 2746extern int ext4_ext_precache(struct inode *inode);
9eb79482 2747extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len);
4a092d73 2748
748de673 2749/* move_extent.c */
393d1d1d
DTB
2750extern void ext4_double_down_write_data_sem(struct inode *first,
2751 struct inode *second);
2752extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
2753 struct inode *donor_inode);
748de673
AF
2754extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
2755 __u64 start_orig, __u64 start_donor,
2756 __u64 len, __u64 *moved_len);
9eb79482
NJ
2757extern int mext_next_extent(struct inode *inode, struct ext4_ext_path *path,
2758 struct ext4_extent **extent);
748de673 2759
bd2d0210 2760/* page-io.c */
5dabfc78
TT
2761extern int __init ext4_init_pageio(void);
2762extern void ext4_exit_pageio(void);
bd2d0210 2763extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
97a851ed
JK
2764extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
2765extern int ext4_put_io_end(ext4_io_end_t *io_end);
2766extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
2767extern void ext4_io_submit_init(struct ext4_io_submit *io,
2768 struct writeback_control *wbc);
2e8fa54e 2769extern void ext4_end_io_rsv_work(struct work_struct *work);
bd2d0210
TT
2770extern void ext4_io_submit(struct ext4_io_submit *io);
2771extern int ext4_bio_write_page(struct ext4_io_submit *io,
2772 struct page *page,
2773 int len,
1c8349a1
NJ
2774 struct writeback_control *wbc,
2775 bool keep_towrite);
3a06d778 2776
c5e06d10
JL
2777/* mmp.c */
2778extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
2779
3613d228
JK
2780/*
2781 * Note that these flags will never ever appear in a buffer_head's state flag.
2782 * See EXT4_MAP_... to see where this is used.
2783 */
744692dc 2784enum ext4_state_bits {
090f32ee 2785 BH_AllocFromCluster /* allocated blocks were part of already
3613d228 2786 * allocated cluster. */
090f32ee 2787 = BH_JBDPrivateStart
744692dc
JZ
2788};
2789
2ccb5fb9 2790/*
48fc7f7e 2791 * Add new method to test whether block and inode bitmaps are properly
2ccb5fb9
AK
2792 * initialized. With uninit_bg reading the block from disk is not enough
2793 * to mark the bitmap uptodate. We need to also zero-out the bitmap
2794 */
2795#define BH_BITMAP_UPTODATE BH_JBDPrivateStart
2796
2797static inline int bitmap_uptodate(struct buffer_head *bh)
2798{
2799 return (buffer_uptodate(bh) &&
2800 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
2801}
2802static inline void set_bitmap_uptodate(struct buffer_head *bh)
2803{
2804 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
2805}
2806
17335dcc
DM
2807/*
2808 * Disable DIO read nolock optimization, so new dioreaders will be forced
2809 * to grab i_mutex
2810 */
2811static inline void ext4_inode_block_unlocked_dio(struct inode *inode)
2812{
2813 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
2814 smp_mb();
2815}
2816static inline void ext4_inode_resume_unlocked_dio(struct inode *inode)
2817{
2818 smp_mb();
2819 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
2820}
2821
731eb1a0
AM
2822#define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
2823
e9e3bcec
ES
2824/* For ioend & aio unwritten conversion wait queues */
2825#define EXT4_WQ_HASH_SZ 37
2826#define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
2827 EXT4_WQ_HASH_SZ])
2828#define ext4_aio_mutex(v) (&ext4__aio_mutex[((unsigned long)(v)) %\
2829 EXT4_WQ_HASH_SZ])
2830extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
2831extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ];
2832
8f82f840
YY
2833#define EXT4_RESIZING 0
2834extern int ext4_resize_begin(struct super_block *sb);
2835extern void ext4_resize_end(struct super_block *sb);
2836
ac27a0ec
DK
2837#endif /* __KERNEL__ */
2838
3dcf5451 2839#endif /* _EXT4_H */