ext4: reuse path object in ext4_move_extents()
[linux-2.6-block.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
bd30d702 572/* DO NOT ASSIGN ADDITIONAL FLAG VALUES WITHOUT ADJUSTING THE FLAGS BELOW */
ac27a0ec 573
107a7bd3 574/*
7869a4a6
TT
575 * The bit position of these flags must not overlap with any of the
576 * EXT4_GET_BLOCKS_*. They are used by ext4_ext_find_extent(),
107a7bd3 577 * read_extent_tree_block(), ext4_split_extent_at(),
7869a4a6
TT
578 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
579 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
580 * caching the extents when reading from the extent tree while a
581 * truncate or punch hole operation is in progress.
107a7bd3 582 */
bd30d702
TT
583#define EXT4_EX_NOCACHE 0x0800
584#define EXT4_EX_FORCE_CACHE 0x1000
107a7bd3 585
e6362609
TT
586/*
587 * Flags used by ext4_free_blocks
588 */
589#define EXT4_FREE_BLOCKS_METADATA 0x0001
590#define EXT4_FREE_BLOCKS_FORGET 0x0002
1f2acb60 591#define EXT4_FREE_BLOCKS_VALIDATED 0x0004
7132de74 592#define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
84130193
TT
593#define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
594#define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
e6362609 595
ac27a0ec
DK
596/*
597 * ioctl commands
598 */
617ba13b
MC
599#define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
600#define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
601#define EXT4_IOC_GETVERSION _IOR('f', 3, long)
602#define EXT4_IOC_SETVERSION _IOW('f', 4, long)
617ba13b
MC
603#define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
604#define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
617ba13b
MC
605#define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
606#define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
8eea80d5
TT
607#define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
608#define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
609#define EXT4_IOC_MIGRATE _IO('f', 9)
ccd2506b 610 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
8eea80d5 611 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
ccd2506b 612#define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
748de673 613#define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
19c5246d 614#define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
393d1d1d 615#define EXT4_IOC_SWAP_BOOT _IO('f', 17)
7869a4a6 616#define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18)
ac27a0ec 617
899ad0ce 618#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
ac27a0ec
DK
619/*
620 * ioctl commands in 32 bit emulation
621 */
617ba13b
MC
622#define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
623#define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
624#define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
625#define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
626#define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
627#define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
628#define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
4d92dc0f 629#define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
617ba13b
MC
630#define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
631#define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
899ad0ce 632#endif
ac27a0ec 633
b595076a 634/* Max physical block we can address w/o extents */
fb0a387d
ES
635#define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
636
ac27a0ec
DK
637/*
638 * Structure of an inode on the disk
639 */
617ba13b 640struct ext4_inode {
ac27a0ec
DK
641 __le16 i_mode; /* File mode */
642 __le16 i_uid; /* Low 16 bits of Owner Uid */
a48380f7 643 __le32 i_size_lo; /* Size in bytes */
ac27a0ec 644 __le32 i_atime; /* Access time */
ef7f3835 645 __le32 i_ctime; /* Inode Change time */
ac27a0ec
DK
646 __le32 i_mtime; /* Modification time */
647 __le32 i_dtime; /* Deletion Time */
648 __le16 i_gid; /* Low 16 bits of Group Id */
649 __le16 i_links_count; /* Links count */
0fc1b451 650 __le32 i_blocks_lo; /* Blocks count */
ac27a0ec
DK
651 __le32 i_flags; /* File flags */
652 union {
653 struct {
25ec56b5 654 __le32 l_i_version;
ac27a0ec
DK
655 } linux1;
656 struct {
657 __u32 h_i_translator;
658 } hurd1;
659 struct {
660 __u32 m_i_reserved1;
661 } masix1;
662 } osd1; /* OS dependent 1 */
617ba13b 663 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
ac27a0ec 664 __le32 i_generation; /* File version (for NFS) */
7973c0c1 665 __le32 i_file_acl_lo; /* File ACL */
a48380f7 666 __le32 i_size_high;
f077d0d7 667 __le32 i_obso_faddr; /* Obsoleted fragment address */
ac27a0ec
DK
668 union {
669 struct {
0fc1b451 670 __le16 l_i_blocks_high; /* were l_i_reserved1 */
a1ddeb7e 671 __le16 l_i_file_acl_high;
8c55e204 672 __le16 l_i_uid_high; /* these 2 fields */
ac27a0ec 673 __le16 l_i_gid_high; /* were reserved2[0] */
e6153918
DW
674 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
675 __le16 l_i_reserved;
ac27a0ec
DK
676 } linux2;
677 struct {
f077d0d7 678 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
ac27a0ec
DK
679 __u16 h_i_mode_high;
680 __u16 h_i_uid_high;
681 __u16 h_i_gid_high;
682 __u32 h_i_author;
683 } hurd2;
684 struct {
f077d0d7 685 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
a1ddeb7e 686 __le16 m_i_file_acl_high;
ac27a0ec
DK
687 __u32 m_i_reserved2[2];
688 } masix2;
689 } osd2; /* OS dependent 2 */
690 __le16 i_extra_isize;
e6153918 691 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
ef7f3835
KS
692 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
693 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
694 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
695 __le32 i_crtime; /* File Creation time */
696 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
25ec56b5 697 __le32 i_version_hi; /* high 32 bits for 64-bit version */
ac27a0ec
DK
698};
699
748de673
AF
700struct move_extent {
701 __u32 reserved; /* should be zero */
702 __u32 donor_fd; /* donor file descriptor */
703 __u64 orig_start; /* logical start offset in block for orig */
704 __u64 donor_start; /* logical start offset in block for donor */
705 __u64 len; /* block length to be moved */
706 __u64 moved_len; /* moved block length */
707};
ac27a0ec 708
ef7f3835
KS
709#define EXT4_EPOCH_BITS 2
710#define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
711#define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
712
713/*
714 * Extended fields will fit into an inode if the filesystem was formatted
715 * with large inodes (-I 256 or larger) and there are not currently any EAs
716 * consuming all of the available space. For new inodes we always reserve
717 * enough space for the kernel's known extended fields, but for inodes
718 * created with an old kernel this might not have been the case. None of
719 * the extended inode fields is critical for correct filesystem operation.
720 * This macro checks if a certain field fits in the inode. Note that
721 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
722 */
723#define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
724 ((offsetof(typeof(*ext4_inode), field) + \
725 sizeof((ext4_inode)->field)) \
726 <= (EXT4_GOOD_OLD_INODE_SIZE + \
727 (einode)->i_extra_isize)) \
728
729static inline __le32 ext4_encode_extra_time(struct timespec *time)
730{
731 return cpu_to_le32((sizeof(time->tv_sec) > 4 ?
c1fccc06
TT
732 (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) |
733 ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK));
ef7f3835
KS
734}
735
736static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
737{
738 if (sizeof(time->tv_sec) > 4)
739 time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK)
740 << 32;
c1fccc06 741 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
ef7f3835
KS
742}
743
744#define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
745do { \
746 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
747 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
748 (raw_inode)->xtime ## _extra = \
749 ext4_encode_extra_time(&(inode)->xtime); \
750} while (0)
751
752#define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
753do { \
754 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
755 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
756 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
757 (raw_inode)->xtime ## _extra = \
758 ext4_encode_extra_time(&(einode)->xtime); \
759} while (0)
760
761#define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
762do { \
763 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
764 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
765 ext4_decode_extra_time(&(inode)->xtime, \
766 raw_inode->xtime ## _extra); \
af0b44a1
ES
767 else \
768 (inode)->xtime.tv_nsec = 0; \
ef7f3835
KS
769} while (0)
770
771#define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
772do { \
773 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
774 (einode)->xtime.tv_sec = \
775 (signed)le32_to_cpu((raw_inode)->xtime); \
19ea8060
TT
776 else \
777 (einode)->xtime.tv_sec = 0; \
ef7f3835
KS
778 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
779 ext4_decode_extra_time(&(einode)->xtime, \
780 raw_inode->xtime ## _extra); \
af0b44a1
ES
781 else \
782 (einode)->xtime.tv_nsec = 0; \
ef7f3835
KS
783} while (0)
784
25ec56b5
JNC
785#define i_disk_version osd1.linux1.l_i_version
786
ac27a0ec
DK
787#if defined(__KERNEL__) || defined(__linux__)
788#define i_reserved1 osd1.linux1.l_i_reserved1
a1ddeb7e 789#define i_file_acl_high osd2.linux2.l_i_file_acl_high
0fc1b451 790#define i_blocks_high osd2.linux2.l_i_blocks_high
ac27a0ec
DK
791#define i_uid_low i_uid
792#define i_gid_low i_gid
793#define i_uid_high osd2.linux2.l_i_uid_high
794#define i_gid_high osd2.linux2.l_i_gid_high
e6153918 795#define i_checksum_lo osd2.linux2.l_i_checksum_lo
ac27a0ec
DK
796
797#elif defined(__GNU__)
798
799#define i_translator osd1.hurd1.h_i_translator
ac27a0ec
DK
800#define i_uid_high osd2.hurd2.h_i_uid_high
801#define i_gid_high osd2.hurd2.h_i_gid_high
802#define i_author osd2.hurd2.h_i_author
803
804#elif defined(__masix__)
805
806#define i_reserved1 osd1.masix1.m_i_reserved1
a1ddeb7e 807#define i_file_acl_high osd2.masix2.m_i_file_acl_high
ac27a0ec
DK
808#define i_reserved2 osd2.masix2.m_i_reserved2
809
810#endif /* defined(__KERNEL__) || defined(__linux__) */
811
c0677e6d
ZL
812#include "extents_status.h"
813
d444c3c3
TT
814/*
815 * fourth extended file system inode data in memory
816 */
817struct ext4_inode_info {
818 __le32 i_data[15]; /* unconverted */
d444c3c3 819 __u32 i_dtime;
12e9b892 820 ext4_fsblk_t i_file_acl;
d444c3c3
TT
821
822 /*
823 * i_block_group is the number of the block group which contains
824 * this file's inode. Constant across the lifetime of the inode,
825 * it is ued for making block allocation decisions - we try to
826 * place a file's data blocks near its inode block, and new inodes
827 * near to their parent directory's inode.
828 */
829 ext4_group_t i_block_group;
8a2005d3 830 ext4_lblk_t i_dir_start_lookup;
353eb83c 831#if (BITS_PER_LONG < 64)
19f5fb7a 832 unsigned long i_state_flags; /* Dynamic state flags */
353eb83c 833#endif
12e9b892 834 unsigned long i_flags;
d444c3c3 835
d444c3c3
TT
836 /*
837 * Extended attributes can be read independently of the main file
838 * data. Taking i_mutex even when reading would cause contention
839 * between readers of EAs and writers of regular file data, so
840 * instead we synchronize on xattr_sem when reading or changing
841 * EAs.
842 */
843 struct rw_semaphore xattr_sem;
d444c3c3
TT
844
845 struct list_head i_orphan; /* unlinked but open inodes */
846
847 /*
848 * i_disksize keeps track of what the inode size is ON DISK, not
849 * in memory. During truncate, i_size is set to the new size by
850 * the VFS prior to calling ext4_truncate(), but the filesystem won't
851 * set i_disksize to 0 until the truncate is actually under way.
852 *
853 * The intent is that i_disksize always represents the blocks which
854 * are used by this file. This allows recovery to restart truncate
855 * on orphans if we crash during truncate. We actually write i_disksize
856 * into the on-disk inode when writing inodes out, instead of i_size.
857 *
858 * The only time when i_disksize and i_size may be different is when
859 * a truncate is in progress. The only things which change i_disksize
860 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
861 */
862 loff_t i_disksize;
863
864 /*
865 * i_data_sem is for serialising ext4_truncate() against
866 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
867 * data tree are chopped off during truncate. We can't do that in
868 * ext4 because whenever we perform intermediate commits during
869 * truncate, the inode and all the metadata blocks *must* be in a
870 * consistent state which allows truncation of the orphans to restart
871 * during recovery. Hence we must fix the get_block-vs-truncate race
872 * by other means, so we have i_data_sem.
873 */
874 struct rw_semaphore i_data_sem;
875 struct inode vfs_inode;
8aefcd55 876 struct jbd2_inode *jinode;
d444c3c3 877
202ee5df
TT
878 spinlock_t i_raw_lock; /* protects updates to the raw inode */
879
d444c3c3
TT
880 /*
881 * File creation time. Its function is same as that of
882 * struct timespec i_{a,c,m}time in the generic inode.
883 */
884 struct timespec i_crtime;
885
886 /* mballoc */
887 struct list_head i_prealloc_list;
888 spinlock_t i_prealloc_lock;
889
c0677e6d
ZL
890 /* extents status tree */
891 struct ext4_es_tree i_es_tree;
892 rwlock_t i_es_lock;
74cd15cd
ZL
893 struct list_head i_es_lru;
894 unsigned int i_es_lru_nr; /* protected by i_es_lock */
d3922a77 895 unsigned long i_touch_when; /* jiffies of last accessing */
c0677e6d 896
d444c3c3
TT
897 /* ialloc */
898 ext4_group_t i_last_alloc_group;
899
900 /* allocation reservation info for delalloc */
57042651 901 /* In case of bigalloc, these refer to clusters rather than blocks */
d444c3c3
TT
902 unsigned int i_reserved_data_blocks;
903 unsigned int i_reserved_meta_blocks;
904 unsigned int i_allocated_meta_blocks;
01f49d0b 905 ext4_lblk_t i_da_metadata_calc_last_lblock;
9d0be502 906 int i_da_metadata_calc_len;
d444c3c3
TT
907
908 /* on-disk additional length */
909 __u16 i_extra_isize;
910
67cf5b09
TM
911 /* Indicate the inline data space. */
912 u16 i_inline_off;
913 u16 i_inline_size;
914
a9e7f447
DM
915#ifdef CONFIG_QUOTA
916 /* quota space reservation, managed internally by quota code */
917 qsize_t i_reserved_quota;
918#endif
8d5d02e6 919
2e8fa54e 920 /* Lock protecting lists below */
744692dc 921 spinlock_t i_completed_io_lock;
2e8fa54e
JK
922 /*
923 * Completed IOs that need unwritten extents handling and have
924 * transaction reserved
925 */
926 struct list_head i_rsv_conversion_list;
927 /*
928 * Completed IOs that need unwritten extents handling and don't have
929 * transaction reserved
930 */
8a2005d3 931 atomic_t i_ioend_count; /* Number of outstanding io_end structs */
e27f41e1 932 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
2e8fa54e 933 struct work_struct i_rsv_conversion_work;
8a2005d3
TT
934
935 spinlock_t i_block_reservation_lock;
b436b9be
JK
936
937 /*
938 * Transactions that contain inode's metadata needed to complete
939 * fsync and fdatasync, respectively.
940 */
941 tid_t i_sync_tid;
942 tid_t i_datasync_tid;
814525f4
DW
943
944 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
945 __u32 i_csum_seed;
d444c3c3
TT
946};
947
ac27a0ec
DK
948/*
949 * File system states
950 */
617ba13b
MC
951#define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
952#define EXT4_ERROR_FS 0x0002 /* Errors detected */
953#define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
ac27a0ec 954
469108ff
TT
955/*
956 * Misc. filesystem flags
957 */
958#define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
959#define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
960#define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
961
ac27a0ec 962/*
68911009 963 * Mount flags set via mount options or defaults
ac27a0ec 964 */
617ba13b
MC
965#define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
966#define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
967#define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
968#define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
969#define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
39ef17f1 970#define EXT4_MOUNT_ERRORS_MASK 0x00070
617ba13b
MC
971#define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
972#define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
617ba13b
MC
973#define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
974#define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
975#define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
976#define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
977#define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
978#define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
979#define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
980#define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
afd4672d 981#define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
617ba13b 982#define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
617ba13b
MC
983#define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
984#define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
985#define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
744692dc 986#define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
d4da6c9c 987#define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
818d276c 988#define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
64769240 989#define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
5bf5683a 990#define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
6fd058f7 991#define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
5328e635 992#define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
bfff6873 993#define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
5bf5683a 994
68911009
LC
995/*
996 * Mount flags set either automatically (could not be set by mount option)
997 * based on per file system feature or property or in special cases such as
998 * distinguishing between explicit mount option definition and default.
999 */
56889787
TT
1000#define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1001 specified delalloc */
68911009
LC
1002#define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1003 size of blocksize * 8
1004 blocks */
ed3654eb
TT
1005#define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1006 file systems */
56889787 1007
fd8c37ec
TT
1008#define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1009 ~EXT4_MOUNT_##opt
1010#define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1011 EXT4_MOUNT_##opt
617ba13b
MC
1012#define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1013 EXT4_MOUNT_##opt)
ac27a0ec 1014
a2595b8a
TT
1015#define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1016 ~EXT4_MOUNT2_##opt
1017#define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1018 EXT4_MOUNT2_##opt
1019#define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1020 EXT4_MOUNT2_##opt)
1021
597d508c
AM
1022#define ext4_test_and_set_bit __test_and_set_bit_le
1023#define ext4_set_bit __set_bit_le
617ba13b 1024#define ext4_set_bit_atomic ext2_set_bit_atomic
597d508c
AM
1025#define ext4_test_and_clear_bit __test_and_clear_bit_le
1026#define ext4_clear_bit __clear_bit_le
617ba13b 1027#define ext4_clear_bit_atomic ext2_clear_bit_atomic
50e0168c 1028#define ext4_test_bit test_bit_le
50e0168c
AM
1029#define ext4_find_next_zero_bit find_next_zero_bit_le
1030#define ext4_find_next_bit find_next_bit_le
ac27a0ec 1031
c3e94d1d
YY
1032extern void ext4_set_bits(void *bm, int cur, int len);
1033
ac27a0ec
DK
1034/*
1035 * Maximal mount counts between two filesystem checks
1036 */
617ba13b
MC
1037#define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1038#define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
ac27a0ec
DK
1039
1040/*
1041 * Behaviour when detecting errors
1042 */
617ba13b
MC
1043#define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1044#define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1045#define EXT4_ERRORS_PANIC 3 /* Panic */
1046#define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
ac27a0ec 1047
e6153918
DW
1048/* Metadata checksum algorithm codes */
1049#define EXT4_CRC32C_CHKSUM 1
1050
ac27a0ec
DK
1051/*
1052 * Structure of the super block
1053 */
617ba13b 1054struct ext4_super_block {
ac27a0ec 1055/*00*/ __le32 s_inodes_count; /* Inodes count */
6bc9feff 1056 __le32 s_blocks_count_lo; /* Blocks count */
308ba3ec
AK
1057 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1058 __le32 s_free_blocks_count_lo; /* Free blocks count */
ac27a0ec
DK
1059/*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1060 __le32 s_first_data_block; /* First Data Block */
1061 __le32 s_log_block_size; /* Block size */
281b5995 1062 __le32 s_log_cluster_size; /* Allocation cluster size */
ac27a0ec 1063/*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
281b5995 1064 __le32 s_clusters_per_group; /* # Clusters per group */
ac27a0ec
DK
1065 __le32 s_inodes_per_group; /* # Inodes per group */
1066 __le32 s_mtime; /* Mount time */
1067/*30*/ __le32 s_wtime; /* Write time */
1068 __le16 s_mnt_count; /* Mount count */
1069 __le16 s_max_mnt_count; /* Maximal mount count */
1070 __le16 s_magic; /* Magic signature */
1071 __le16 s_state; /* File system state */
1072 __le16 s_errors; /* Behaviour when detecting errors */
1073 __le16 s_minor_rev_level; /* minor revision level */
1074/*40*/ __le32 s_lastcheck; /* time of last check */
1075 __le32 s_checkinterval; /* max. time between checks */
1076 __le32 s_creator_os; /* OS */
1077 __le32 s_rev_level; /* Revision level */
1078/*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1079 __le16 s_def_resgid; /* Default gid for reserved blocks */
1080 /*
617ba13b 1081 * These fields are for EXT4_DYNAMIC_REV superblocks only.
ac27a0ec
DK
1082 *
1083 * Note: the difference between the compatible feature set and
1084 * the incompatible feature set is that if there is a bit set
1085 * in the incompatible feature set that the kernel doesn't
1086 * know about, it should refuse to mount the filesystem.
1087 *
1088 * e2fsck's requirements are more strict; if it doesn't know
1089 * about a feature in either the compatible or incompatible
1090 * feature set, it must abort and not try to meddle with
1091 * things it doesn't understand...
1092 */
1093 __le32 s_first_ino; /* First non-reserved inode */
0d1ee42f 1094 __le16 s_inode_size; /* size of inode structure */
ac27a0ec
DK
1095 __le16 s_block_group_nr; /* block group # of this superblock */
1096 __le32 s_feature_compat; /* compatible feature set */
1097/*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1098 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1099/*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1100/*78*/ char s_volume_name[16]; /* volume name */
1101/*88*/ char s_last_mounted[64]; /* directory where last mounted */
1102/*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1103 /*
1104 * Performance hints. Directory preallocation should only
617ba13b 1105 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
ac27a0ec
DK
1106 */
1107 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1108 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1109 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1110 /*
617ba13b 1111 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
ac27a0ec
DK
1112 */
1113/*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1114/*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1115 __le32 s_journal_dev; /* device number of journal file */
1116 __le32 s_last_orphan; /* start of list of inodes to delete */
1117 __le32 s_hash_seed[4]; /* HTREE hash seed */
1118 __u8 s_def_hash_version; /* Default hash version to use */
89eeddf0 1119 __u8 s_jnl_backup_type;
0d1ee42f 1120 __le16 s_desc_size; /* size of group descriptor */
bd81d8ee 1121/*100*/ __le32 s_default_mount_opts;
ac27a0ec 1122 __le32 s_first_meta_bg; /* First metablock block group */
bd81d8ee
LV
1123 __le32 s_mkfs_time; /* When the filesystem was created */
1124 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1125 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1126/*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1127 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1128 __le32 s_free_blocks_count_hi; /* Free blocks count */
c1bddad9
AK
1129 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1130 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1131 __le32 s_flags; /* Miscellaneous flags */
1132 __le16 s_raid_stride; /* RAID stride */
c5e06d10 1133 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
c1bddad9
AK
1134 __le64 s_mmp_block; /* Block for multi-mount protection */
1135 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
772cb7c8 1136 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
e6153918 1137 __u8 s_checksum_type; /* metadata checksum algorithm used */
772cb7c8 1138 __le16 s_reserved_pad;
afc32f7e 1139 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
4a9cdec7
TT
1140 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1141 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1142 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1143 snapshot's future use */
1144 __le32 s_snapshot_list; /* inode number of the head of the
1145 on-disk snapshot list */
1c13d5c0
TT
1146#define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1147 __le32 s_error_count; /* number of fs errors */
1148 __le32 s_first_error_time; /* first time an error happened */
1149 __le32 s_first_error_ino; /* inode involved in first error */
1150 __le64 s_first_error_block; /* block involved of first error */
1151 __u8 s_first_error_func[32]; /* function where the error happened */
1152 __le32 s_first_error_line; /* line number where error happened */
1153 __le32 s_last_error_time; /* most recent time of an error */
1154 __le32 s_last_error_ino; /* inode involved in last error */
1155 __le32 s_last_error_line; /* line number where error happened */
1156 __le64 s_last_error_block; /* block involved of last error */
1157 __u8 s_last_error_func[32]; /* function where the error happened */
8b67f04a
TT
1158#define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1159 __u8 s_mount_opts[64];
281b5995
TT
1160 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1161 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1162 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1beeef1b
DW
1163 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
1164 __le32 s_reserved[106]; /* Padding to the end of the block */
e6153918 1165 __le32 s_checksum; /* crc32c(superblock) */
ac27a0ec
DK
1166};
1167
1c13d5c0
TT
1168#define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1169
ac27a0ec 1170#ifdef __KERNEL__
bc0b0d6d
TT
1171
1172/*
4ab2f15b 1173 * run-time mount flags
bc0b0d6d
TT
1174 */
1175#define EXT4_MF_MNTDIR_SAMPLED 0x0001
4ab2f15b 1176#define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
bc0b0d6d 1177
ca0faba0
TT
1178/*
1179 * fourth extended-fs super-block data in memory
1180 */
1181struct ext4_sb_info {
1182 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1183 unsigned long s_inodes_per_block;/* Number of inodes per block */
1184 unsigned long s_blocks_per_group;/* Number of blocks in a group */
281b5995 1185 unsigned long s_clusters_per_group; /* Number of clusters in a group */
ca0faba0
TT
1186 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1187 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1188 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1189 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1190 ext4_group_t s_groups_count; /* Number of groups in the fs */
fb0a387d 1191 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
952fc18e 1192 unsigned long s_overhead; /* # of fs overhead clusters */
281b5995
TT
1193 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1194 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
ca0faba0
TT
1195 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1196 struct buffer_head * s_sbh; /* Buffer containing the super block */
1197 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1198 struct buffer_head **s_group_desc;
7f4520cc 1199 unsigned int s_mount_opt;
a2595b8a 1200 unsigned int s_mount_opt2;
bc0b0d6d 1201 unsigned int s_mount_flags;
5a916be1 1202 unsigned int s_def_mount_opt;
ca0faba0 1203 ext4_fsblk_t s_sb_block;
27dd4385 1204 atomic64_t s_resv_clusters;
08cefc7a
EB
1205 kuid_t s_resuid;
1206 kgid_t s_resgid;
ca0faba0
TT
1207 unsigned short s_mount_state;
1208 unsigned short s_pad;
1209 int s_addr_per_block_bits;
1210 int s_desc_per_block_bits;
1211 int s_inode_size;
1212 int s_first_ino;
1213 unsigned int s_inode_readahead_blks;
11013911 1214 unsigned int s_inode_goal;
ca0faba0
TT
1215 spinlock_t s_next_gen_lock;
1216 u32 s_next_generation;
1217 u32 s_hash_seed[4];
1218 int s_def_hash_version;
1219 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
57042651 1220 struct percpu_counter s_freeclusters_counter;
ca0faba0
TT
1221 struct percpu_counter s_freeinodes_counter;
1222 struct percpu_counter s_dirs_counter;
57042651 1223 struct percpu_counter s_dirtyclusters_counter;
ca0faba0
TT
1224 struct blockgroup_lock *s_blockgroup_lock;
1225 struct proc_dir_entry *s_proc;
1226 struct kobject s_kobj;
1227 struct completion s_kobj_unregister;
2c0544b2 1228 struct super_block *s_sb;
ca0faba0
TT
1229
1230 /* Journaling */
ca0faba0
TT
1231 struct journal_s *s_journal;
1232 struct list_head s_orphan;
1233 struct mutex s_orphan_lock;
8f82f840
YY
1234 unsigned long s_resize_flags; /* Flags indicating if there
1235 is a resizer */
ca0faba0
TT
1236 unsigned long s_commit_interval;
1237 u32 s_max_batch_time;
1238 u32 s_min_batch_time;
1239 struct block_device *journal_bdev;
ca0faba0
TT
1240#ifdef CONFIG_QUOTA
1241 char *s_qf_names[MAXQUOTAS]; /* Names of quota files with journalled quota */
1242 int s_jquota_fmt; /* Format of quota to use */
1243#endif
1244 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
6fd058f7 1245 struct rb_root system_blks;
ca0faba0
TT
1246
1247#ifdef EXTENTS_STATS
1248 /* ext4 extents stats */
1249 unsigned long s_ext_min;
1250 unsigned long s_ext_max;
1251 unsigned long s_depth_max;
1252 spinlock_t s_ext_stats_lock;
1253 unsigned long s_ext_blocks;
1254 unsigned long s_ext_extents;
1255#endif
1256
1257 /* for buddy allocator */
1258 struct ext4_group_info ***s_group_info;
1259 struct inode *s_buddy_cache;
ca0faba0 1260 spinlock_t s_md_lock;
ca0faba0
TT
1261 unsigned short *s_mb_offsets;
1262 unsigned int *s_mb_maxs;
28623c2f 1263 unsigned int s_group_info_size;
ca0faba0
TT
1264
1265 /* tunables */
1266 unsigned long s_stripe;
1267 unsigned int s_mb_stream_request;
1268 unsigned int s_mb_max_to_scan;
1269 unsigned int s_mb_min_to_scan;
1270 unsigned int s_mb_stats;
1271 unsigned int s_mb_order2_reqs;
1272 unsigned int s_mb_group_prealloc;
df981d03 1273 unsigned int s_max_dir_size_kb;
ca0faba0
TT
1274 /* where last allocation was done - for stream allocation */
1275 unsigned long s_mb_last_group;
1276 unsigned long s_mb_last_start;
1277
ca0faba0 1278 /* stats for buddy allocator */
ca0faba0
TT
1279 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1280 atomic_t s_bal_success; /* we found long enough chunks */
1281 atomic_t s_bal_allocated; /* in blocks */
1282 atomic_t s_bal_ex_scanned; /* total extents scanned */
1283 atomic_t s_bal_goals; /* goal hits */
1284 atomic_t s_bal_breaks; /* too long searches */
1285 atomic_t s_bal_2orders; /* 2^order hits */
1286 spinlock_t s_bal_lock;
1287 unsigned long s_mb_buddies_generated;
1288 unsigned long long s_mb_generation_time;
1289 atomic_t s_mb_lost_chunks;
1290 atomic_t s_mb_preallocated;
1291 atomic_t s_mb_discarded;
50797481 1292 atomic_t s_lock_busy;
ca0faba0
TT
1293
1294 /* locality groups */
003cb608 1295 struct ext4_locality_group __percpu *s_locality_groups;
ca0faba0
TT
1296
1297 /* for write statistics */
1298 unsigned long s_sectors_written_start;
1299 u64 s_kbytes_written;
1300
67a5da56
ZL
1301 /* the size of zero-out chunk */
1302 unsigned int s_extent_max_zeroout_kb;
1303
ca0faba0
TT
1304 unsigned int s_log_groups_per_flex;
1305 struct flex_groups *s_flex_groups;
117fff10 1306 ext4_group_t s_flex_groups_allocated;
4c0425ff 1307
2e8fa54e
JK
1308 /* workqueue for reserved extent conversions (buffered io) */
1309 struct workqueue_struct *rsv_conversion_wq;
66e61a9e
TT
1310
1311 /* timer for periodic error stats printing */
1312 struct timer_list s_err_report;
bfff6873
LC
1313
1314 /* Lazy inode table initialization info */
1315 struct ext4_li_request *s_li_request;
1316 /* Wait multiplier for lazy initialization thread */
1317 unsigned int s_li_wait_mult;
c5e06d10
JL
1318
1319 /* Kernel thread for multiple mount protection */
1320 struct task_struct *s_mmp_tsk;
3d56b8d2
TM
1321
1322 /* record the last minlen when FITRIM is called. */
1323 atomic_t s_last_trim_minblks;
0441984a
DW
1324
1325 /* Reference to checksum algorithm driver via cryptoapi */
1326 struct crypto_shash *s_chksum_driver;
a9c47317
DW
1327
1328 /* Precomputed FS UUID checksum for seeding other checksums */
1329 __u32 s_csum_seed;
74cd15cd
ZL
1330
1331 /* Reclaim extents from extent status tree */
1332 struct shrinker s_es_shrinker;
1333 struct list_head s_es_lru;
d3922a77 1334 unsigned long s_es_last_sorted;
1ac6466f 1335 struct percpu_counter s_extent_cache_cnt;
9c191f70 1336 struct mb_cache *s_mb_cache;
74cd15cd 1337 spinlock_t s_es_lru_lock ____cacheline_aligned_in_smp;
efbed4dc
TT
1338
1339 /* Ratelimit ext4 messages. */
1340 struct ratelimit_state s_err_ratelimit_state;
1341 struct ratelimit_state s_warning_ratelimit_state;
1342 struct ratelimit_state s_msg_ratelimit_state;
ca0faba0
TT
1343};
1344
af5bc92d 1345static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
ac27a0ec
DK
1346{
1347 return sb->s_fs_info;
1348}
617ba13b 1349static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
ac27a0ec 1350{
617ba13b 1351 return container_of(inode, struct ext4_inode_info, vfs_inode);
ac27a0ec
DK
1352}
1353
ef7f3835
KS
1354static inline struct timespec ext4_current_time(struct inode *inode)
1355{
1356 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1357 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1358}
1359
617ba13b 1360static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
ac27a0ec 1361{
617ba13b 1362 return ino == EXT4_ROOT_INO ||
7c319d32
AK
1363 ino == EXT4_USR_QUOTA_INO ||
1364 ino == EXT4_GRP_QUOTA_INO ||
393d1d1d 1365 ino == EXT4_BOOT_LOADER_INO ||
617ba13b
MC
1366 ino == EXT4_JOURNAL_INO ||
1367 ino == EXT4_RESIZE_INO ||
1368 (ino >= EXT4_FIRST_INO(sb) &&
1369 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
ac27a0ec 1370}
19f5fb7a 1371
0edeb71d
TM
1372static inline void ext4_set_io_unwritten_flag(struct inode *inode,
1373 struct ext4_io_end *io_end)
1374{
1375 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
1376 io_end->flag |= EXT4_IO_END_UNWRITTEN;
e27f41e1 1377 atomic_inc(&EXT4_I(inode)->i_unwritten);
0edeb71d
TM
1378 }
1379}
1380
f45ee3a1
DM
1381static inline ext4_io_end_t *ext4_inode_aio(struct inode *inode)
1382{
1383 return inode->i_private;
1384}
1385
1386static inline void ext4_inode_aio_set(struct inode *inode, ext4_io_end_t *io)
1387{
1388 inode->i_private = io;
1389}
1390
19f5fb7a
TT
1391/*
1392 * Inode dynamic state flags
1393 */
1394enum {
1395 EXT4_STATE_JDATA, /* journaled data exists */
1396 EXT4_STATE_NEW, /* inode is newly created */
1397 EXT4_STATE_XATTR, /* has in-inode xattrs */
1398 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1399 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1400 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1401 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
14ece102 1402 EXT4_STATE_NEWENTRY, /* File just added to dir */
f2321097 1403 EXT4_STATE_DELALLOC_RESERVED, /* blks already reserved for delalloc */
17335dcc
DM
1404 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read
1405 nolocking */
67cf5b09 1406 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
74d553aa 1407 EXT4_STATE_ORDERED_MODE, /* data=ordered mode */
7869a4a6 1408 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
19f5fb7a
TT
1409};
1410
353eb83c 1411#define EXT4_INODE_BIT_FNS(name, field, offset) \
12e9b892
DM
1412static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1413{ \
353eb83c 1414 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
12e9b892
DM
1415} \
1416static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1417{ \
353eb83c 1418 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
12e9b892
DM
1419} \
1420static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1421{ \
353eb83c 1422 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
19f5fb7a
TT
1423}
1424
3f61c0cc
AD
1425/* Add these declarations here only so that these functions can be
1426 * found by name. Otherwise, they are very hard to locate. */
1427static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1428static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1429static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
353eb83c 1430EXT4_INODE_BIT_FNS(flag, flags, 0)
3f61c0cc
AD
1431
1432/* Add these declarations here only so that these functions can be
1433 * found by name. Otherwise, they are very hard to locate. */
1434static inline int ext4_test_inode_state(struct inode *inode, int bit);
1435static inline void ext4_set_inode_state(struct inode *inode, int bit);
1436static inline void ext4_clear_inode_state(struct inode *inode, int bit);
353eb83c
TT
1437#if (BITS_PER_LONG < 64)
1438EXT4_INODE_BIT_FNS(state, state_flags, 0)
1439
1440static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1441{
1442 (ei)->i_state_flags = 0;
1443}
1444#else
1445EXT4_INODE_BIT_FNS(state, flags, 32)
1446
1447static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1448{
1449 /* We depend on the fact that callers will set i_flags */
1450}
1451#endif
ac27a0ec 1452#else
617ba13b 1453/* Assume that user mode programs are passing in an ext4fs superblock, not
ac27a0ec
DK
1454 * a kernel struct super_block. This will allow us to call the feature-test
1455 * macros from user land. */
617ba13b 1456#define EXT4_SB(sb) (sb)
ac27a0ec
DK
1457#endif
1458
617ba13b 1459#define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
ac27a0ec
DK
1460
1461/*
1462 * Codes for operating systems
1463 */
617ba13b
MC
1464#define EXT4_OS_LINUX 0
1465#define EXT4_OS_HURD 1
1466#define EXT4_OS_MASIX 2
1467#define EXT4_OS_FREEBSD 3
1468#define EXT4_OS_LITES 4
ac27a0ec
DK
1469
1470/*
1471 * Revision levels
1472 */
617ba13b
MC
1473#define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1474#define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
ac27a0ec 1475
617ba13b
MC
1476#define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1477#define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
ac27a0ec 1478
617ba13b 1479#define EXT4_GOOD_OLD_INODE_SIZE 128
ac27a0ec
DK
1480
1481/*
1482 * Feature set definitions
1483 */
1484
617ba13b 1485#define EXT4_HAS_COMPAT_FEATURE(sb,mask) \
3a06d778 1486 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
617ba13b 1487#define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \
3a06d778 1488 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
617ba13b 1489#define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \
3a06d778 1490 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
617ba13b
MC
1491#define EXT4_SET_COMPAT_FEATURE(sb,mask) \
1492 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1493#define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \
1494 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1495#define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \
1496 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1497#define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \
1498 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1499#define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
1500 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1501#define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \
1502 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1503
1504#define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1505#define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1506#define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1507#define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1508#define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1509#define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1beeef1b 1510#define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
617ba13b
MC
1511
1512#define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1513#define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1514#define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
0fc1b451 1515#define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
717d50e4 1516#define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
f8628a14 1517#define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
ef7f3835 1518#define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
ae812306 1519#define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
281b5995 1520#define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
e6153918
DW
1521/*
1522 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1523 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1524 * all other data structures' checksums. However, the METADATA_CSUM and
1525 * GDT_CSUM bits are mutually exclusive.
1526 */
9b90e5e0 1527#define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
617ba13b
MC
1528
1529#define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1530#define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1531#define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1532#define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1533#define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
a86c6181 1534#define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
bd81d8ee 1535#define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
0fc1b451 1536#define EXT4_FEATURE_INCOMPAT_MMP 0x0100
ce421581 1537#define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
f710b4b9
TT
1538#define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1539#define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
856cbcf9 1540#define EXT4_FEATURE_INCOMPAT_BG_USE_META_CSUM 0x2000 /* use crc32c for bg */
9b90e5e0 1541#define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
67cf5b09 1542#define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
617ba13b 1543
2035e776
TT
1544#define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1545#define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1546 EXT4_FEATURE_INCOMPAT_META_BG)
1547#define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1548 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1549 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1550
1551#define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1552#define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1553 EXT4_FEATURE_INCOMPAT_RECOVER| \
1554 EXT4_FEATURE_INCOMPAT_META_BG)
1555#define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1556 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1557 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1558
617ba13b
MC
1559#define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR
1560#define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1561 EXT4_FEATURE_INCOMPAT_RECOVER| \
a86c6181 1562 EXT4_FEATURE_INCOMPAT_META_BG| \
bd81d8ee 1563 EXT4_FEATURE_INCOMPAT_EXTENTS| \
ce421581 1564 EXT4_FEATURE_INCOMPAT_64BIT| \
c5e06d10 1565 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
f08225d1
TM
1566 EXT4_FEATURE_INCOMPAT_MMP | \
1567 EXT4_FEATURE_INCOMPAT_INLINE_DATA)
617ba13b
MC
1568#define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1569 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
717d50e4 1570 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
f8628a14 1571 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
ef7f3835 1572 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
0fc1b451 1573 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
6f16b606 1574 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
e93376c2 1575 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
7c319d32
AK
1576 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1577 EXT4_FEATURE_RO_COMPAT_QUOTA)
ac27a0ec
DK
1578
1579/*
1580 * Default values for user and/or group using reserved blocks
1581 */
617ba13b
MC
1582#define EXT4_DEF_RESUID 0
1583#define EXT4_DEF_RESGID 0
ac27a0ec 1584
240799cd
TT
1585#define EXT4_DEF_INODE_READAHEAD_BLKS 32
1586
ac27a0ec
DK
1587/*
1588 * Default mount options
1589 */
617ba13b
MC
1590#define EXT4_DEFM_DEBUG 0x0001
1591#define EXT4_DEFM_BSDGROUPS 0x0002
1592#define EXT4_DEFM_XATTR_USER 0x0004
1593#define EXT4_DEFM_ACL 0x0008
1594#define EXT4_DEFM_UID16 0x0010
1595#define EXT4_DEFM_JMODE 0x0060
1596#define EXT4_DEFM_JMODE_DATA 0x0020
1597#define EXT4_DEFM_JMODE_ORDERED 0x0040
1598#define EXT4_DEFM_JMODE_WBACK 0x0060
8b67f04a
TT
1599#define EXT4_DEFM_NOBARRIER 0x0100
1600#define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1601#define EXT4_DEFM_DISCARD 0x0400
1602#define EXT4_DEFM_NODELALLOC 0x0800
ac27a0ec 1603
30773840
TT
1604/*
1605 * Default journal batch times
1606 */
1607#define EXT4_DEF_MIN_BATCH_TIME 0
1608#define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1609
a4912123
TT
1610/*
1611 * Minimum number of groups in a flexgroup before we separate out
1612 * directories into the first block group of a flexgroup
1613 */
1614#define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1615
ac27a0ec
DK
1616/*
1617 * Structure of a directory entry
1618 */
617ba13b 1619#define EXT4_NAME_LEN 255
ac27a0ec 1620
617ba13b 1621struct ext4_dir_entry {
ac27a0ec
DK
1622 __le32 inode; /* Inode number */
1623 __le16 rec_len; /* Directory entry length */
1624 __le16 name_len; /* Name length */
617ba13b 1625 char name[EXT4_NAME_LEN]; /* File name */
ac27a0ec
DK
1626};
1627
1628/*
617ba13b 1629 * The new version of the directory entry. Since EXT4 structures are
ac27a0ec
DK
1630 * stored in intel byte order, and the name_len field could never be
1631 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1632 * file_type field.
1633 */
617ba13b 1634struct ext4_dir_entry_2 {
ac27a0ec
DK
1635 __le32 inode; /* Inode number */
1636 __le16 rec_len; /* Directory entry length */
1637 __u8 name_len; /* Name length */
1638 __u8 file_type;
617ba13b 1639 char name[EXT4_NAME_LEN]; /* File name */
ac27a0ec
DK
1640};
1641
e6153918
DW
1642/*
1643 * This is a bogus directory entry at the end of each leaf block that
1644 * records checksums.
1645 */
1646struct ext4_dir_entry_tail {
1647 __le32 det_reserved_zero1; /* Pretend to be unused */
1648 __le16 det_rec_len; /* 12 */
1649 __u8 det_reserved_zero2; /* Zero name length */
1650 __u8 det_reserved_ft; /* 0xDE, fake file type */
1651 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
1652};
1653
3c47d541
TM
1654#define EXT4_DIRENT_TAIL(block, blocksize) \
1655 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1656 ((blocksize) - \
1657 sizeof(struct ext4_dir_entry_tail))))
1658
ac27a0ec 1659/*
617ba13b 1660 * Ext4 directory file types. Only the low 3 bits are used. The
ac27a0ec
DK
1661 * other bits are reserved for now.
1662 */
617ba13b
MC
1663#define EXT4_FT_UNKNOWN 0
1664#define EXT4_FT_REG_FILE 1
1665#define EXT4_FT_DIR 2
1666#define EXT4_FT_CHRDEV 3
1667#define EXT4_FT_BLKDEV 4
1668#define EXT4_FT_FIFO 5
1669#define EXT4_FT_SOCK 6
1670#define EXT4_FT_SYMLINK 7
ac27a0ec 1671
617ba13b 1672#define EXT4_FT_MAX 8
ac27a0ec 1673
e6153918
DW
1674#define EXT4_FT_DIR_CSUM 0xDE
1675
ac27a0ec 1676/*
617ba13b 1677 * EXT4_DIR_PAD defines the directory entries boundaries
ac27a0ec
DK
1678 *
1679 * NOTE: It must be a multiple of 4
1680 */
617ba13b
MC
1681#define EXT4_DIR_PAD 4
1682#define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1683#define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1684 ~EXT4_DIR_ROUND)
a72d7f83
JK
1685#define EXT4_MAX_REC_LEN ((1<<16)-1)
1686
0cfc9255
ES
1687/*
1688 * If we ever get support for fs block sizes > page_size, we'll need
1689 * to remove the #if statements in the next two functions...
1690 */
1691static inline unsigned int
1692ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1693{
1694 unsigned len = le16_to_cpu(dlen);
1695
1696#if (PAGE_CACHE_SIZE >= 65536)
1697 if (len == EXT4_MAX_REC_LEN || len == 0)
1698 return blocksize;
1699 return (len & 65532) | ((len & 3) << 16);
1700#else
1701 return len;
1702#endif
1703}
1704
1705static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1706{
1707 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1708 BUG();
1709#if (PAGE_CACHE_SIZE >= 65536)
1710 if (len < 65536)
1711 return cpu_to_le16(len);
1712 if (len == blocksize) {
1713 if (blocksize == 65536)
1714 return cpu_to_le16(EXT4_MAX_REC_LEN);
1715 else
1716 return cpu_to_le16(0);
1717 }
1718 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1719#else
1720 return cpu_to_le16(len);
1721#endif
1722}
1723
ac27a0ec
DK
1724/*
1725 * Hash Tree Directory indexing
1726 * (c) Daniel Phillips, 2001
1727 */
1728
4074fe37
ES
1729#define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1730 EXT4_FEATURE_COMPAT_DIR_INDEX) && \
12e9b892 1731 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
617ba13b
MC
1732#define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1733#define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
ac27a0ec
DK
1734
1735/* Legal values for the dx_root hash_version field: */
1736
1737#define DX_HASH_LEGACY 0
1738#define DX_HASH_HALF_MD4 1
1739#define DX_HASH_TEA 2
f99b2589
TT
1740#define DX_HASH_LEGACY_UNSIGNED 3
1741#define DX_HASH_HALF_MD4_UNSIGNED 4
1742#define DX_HASH_TEA_UNSIGNED 5
ac27a0ec 1743
0441984a
DW
1744static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
1745 const void *address, unsigned int length)
1746{
1747 struct {
1748 struct shash_desc shash;
3108b54b 1749 char ctx[4];
0441984a
DW
1750 } desc;
1751 int err;
1752
3108b54b
TT
1753 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
1754
0441984a
DW
1755 desc.shash.tfm = sbi->s_chksum_driver;
1756 desc.shash.flags = 0;
1757 *(u32 *)desc.ctx = crc;
1758
1759 err = crypto_shash_update(&desc.shash, address, length);
1760 BUG_ON(err);
1761
1762 return *(u32 *)desc.ctx;
1763}
1764
ac27a0ec
DK
1765#ifdef __KERNEL__
1766
1767/* hash info structure used by the directory hash */
1768struct dx_hash_info
1769{
1770 u32 hash;
1771 u32 minor_hash;
1772 int hash_version;
1773 u32 *seed;
1774};
1775
d1f5273e
FY
1776
1777/* 32 and 64 bit signed EOF for dx directories */
1778#define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
1779#define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
1780
ac27a0ec
DK
1781
1782/*
617ba13b 1783 * Control parameters used by ext4_htree_next_block
ac27a0ec
DK
1784 */
1785#define HASH_NB_ALWAYS 1
1786
1787
1788/*
1789 * Describe an inode's exact location on disk and in memory
1790 */
617ba13b 1791struct ext4_iloc
ac27a0ec
DK
1792{
1793 struct buffer_head *bh;
1794 unsigned long offset;
fd2d4291 1795 ext4_group_t block_group;
ac27a0ec
DK
1796};
1797
617ba13b 1798static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
ac27a0ec 1799{
617ba13b 1800 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
ac27a0ec
DK
1801}
1802
1803/*
1804 * This structure is stuffed into the struct file's private_data field
1805 * for directories. It is where we put information so that we can do
1806 * readdir operations in hash tree order.
1807 */
1808struct dir_private_info {
1809 struct rb_root root;
1810 struct rb_node *curr_node;
1811 struct fname *extra_fname;
1812 loff_t last_pos;
1813 __u32 curr_hash;
1814 __u32 curr_minor_hash;
1815 __u32 next_hash;
1816};
1817
1818/* calculate the first block number of the group */
617ba13b 1819static inline ext4_fsblk_t
fd2d4291 1820ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
ac27a0ec 1821{
617ba13b
MC
1822 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
1823 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
ac27a0ec
DK
1824}
1825
1826/*
1827 * Special error return code only used by dx_probe() and its callers.
1828 */
36de9286 1829#define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
ac27a0ec 1830
bfff6873
LC
1831/*
1832 * Timeout and state flag for lazy initialization inode thread.
1833 */
1834#define EXT4_DEF_LI_WAIT_MULT 10
1835#define EXT4_DEF_LI_MAX_START_DELAY 5
1836#define EXT4_LAZYINIT_QUIT 0x0001
1837#define EXT4_LAZYINIT_RUNNING 0x0002
1838
1839/*
1840 * Lazy inode table initialization info
1841 */
1842struct ext4_lazy_init {
1843 unsigned long li_state;
bfff6873
LC
1844 struct list_head li_request_list;
1845 struct mutex li_list_mtx;
1846};
1847
1848struct ext4_li_request {
1849 struct super_block *lr_super;
1850 struct ext4_sb_info *lr_sbi;
1851 ext4_group_t lr_next_group;
1852 struct list_head lr_request;
1853 unsigned long lr_next_sched;
1854 unsigned long lr_timeout;
1855};
1856
857ac889
LC
1857struct ext4_features {
1858 struct kobject f_kobj;
1859 struct completion f_kobj_unregister;
1860};
1861
c5e06d10
JL
1862/*
1863 * This structure will be used for multiple mount protection. It will be
1864 * written into the block number saved in the s_mmp_block field in the
1865 * superblock. Programs that check MMP should assume that if
1866 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
1867 * to use the filesystem, regardless of how old the timestamp is.
1868 */
1869#define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
1870#define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
1871#define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
1872#define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
1873
1874struct mmp_struct {
1875 __le32 mmp_magic; /* Magic number for MMP */
1876 __le32 mmp_seq; /* Sequence no. updated periodically */
1877
1878 /*
1879 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
1880 * purposes and do not affect the correctness of the algorithm
1881 */
1882 __le64 mmp_time; /* Time last updated */
1883 char mmp_nodename[64]; /* Node which last updated MMP block */
1884 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
1885
1886 /*
1887 * mmp_check_interval is used to verify if the MMP block has been
1888 * updated on the block device. The value is updated based on the
1889 * maximum time to write the MMP block during an update cycle.
1890 */
1891 __le16 mmp_check_interval;
1892
1893 __le16 mmp_pad1;
e6153918
DW
1894 __le32 mmp_pad2[226];
1895 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
c5e06d10
JL
1896};
1897
1898/* arguments passed to the mmp thread */
1899struct mmpd_data {
1900 struct buffer_head *bh; /* bh from initial read_mmp_block() */
1901 struct super_block *sb; /* super block of the fs */
1902};
1903
1904/*
1905 * Check interval multiplier
1906 * The MMP block is written every update interval and initially checked every
1907 * update interval x the multiplier (the value is then adapted based on the
1908 * write latency). The reason is that writes can be delayed under load and we
1909 * don't want readers to incorrectly assume that the filesystem is no longer
1910 * in use.
1911 */
1912#define EXT4_MMP_CHECK_MULT 2UL
1913
1914/*
1915 * Minimum interval for MMP checking in seconds.
1916 */
1917#define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
1918
1919/*
1920 * Maximum interval for MMP checking in seconds.
1921 */
1922#define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
1923
ac27a0ec
DK
1924/*
1925 * Function prototypes
1926 */
1927
1928/*
1929 * Ok, these declarations are also in <linux/kernel.h> but none of the
617ba13b 1930 * ext4 source programs needs to include it so they are duplicated here.
ac27a0ec 1931 */
8c55e204
DK
1932# define NORET_TYPE /**/
1933# define ATTRIB_NORET __attribute__((noreturn))
1934# define NORET_AND noreturn,
ac27a0ec 1935
498e5f24 1936/* bitmap.c */
f6fb99ca 1937extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
41a246d1
DW
1938void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
1939 struct ext4_group_desc *gdp,
1940 struct buffer_head *bh, int sz);
1941int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
1942 struct ext4_group_desc *gdp,
1943 struct buffer_head *bh, int sz);
fa77dcfa
DW
1944void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
1945 struct ext4_group_desc *gdp,
79f1ba49 1946 struct buffer_head *bh);
fa77dcfa
DW
1947int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
1948 struct ext4_group_desc *gdp,
79f1ba49 1949 struct buffer_head *bh);
498e5f24 1950
ac27a0ec 1951/* balloc.c */
bd86298e
LC
1952extern void ext4_get_group_no_and_offset(struct super_block *sb,
1953 ext4_fsblk_t blocknr,
1954 ext4_group_t *blockgrpp,
1955 ext4_grpblk_t *offsetp);
1956extern ext4_group_t ext4_get_group_number(struct super_block *sb,
1957 ext4_fsblk_t block);
1958
3a5b2ecd
MC
1959extern unsigned int ext4_block_group(struct super_block *sb,
1960 ext4_fsblk_t blocknr);
1961extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
1962 ext4_fsblk_t blocknr);
fd2d4291
AM
1963extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
1964extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
1965 ext4_group_t group);
7061eba7 1966extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
55f020db
AH
1967 ext4_fsblk_t goal,
1968 unsigned int flags,
1969 unsigned long *count,
1970 int *errp);
e7d5f315
TT
1971extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
1972 s64 nclusters, unsigned int flags);
5dee5437 1973extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
af5bc92d 1974extern void ext4_check_blocks_bitmap(struct super_block *);
617ba13b 1975extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
fd2d4291 1976 ext4_group_t block_group,
ac27a0ec 1977 struct buffer_head ** bh);
617ba13b 1978extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
813e5727
TT
1979
1980extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
1981 ext4_group_t block_group);
1982extern int ext4_wait_block_bitmap(struct super_block *sb,
1983 ext4_group_t block_group,
1984 struct buffer_head *bh);
1985extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
1986 ext4_group_t block_group);
cff1dfd7
TT
1987extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
1988 ext4_group_t block_group,
1989 struct ext4_group_desc *gdp);
f86186b4 1990ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
ac27a0ec
DK
1991
1992/* dir.c */
60fd4da3 1993extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
f7c21177 1994 struct file *,
60fd4da3 1995 struct ext4_dir_entry_2 *,
226ba972
TM
1996 struct buffer_head *, char *, int,
1997 unsigned int);
1998#define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
f7c21177 1999 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
226ba972 2000 (de), (bh), (buf), (size), (offset)))
617ba13b 2001extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
ac27a0ec 2002 __u32 minor_hash,
617ba13b
MC
2003 struct ext4_dir_entry_2 *dirent);
2004extern void ext4_htree_free_dir_info(struct dir_private_info *p);
978fef91
TM
2005extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2006 struct buffer_head *bh,
2007 void *buf, int buf_size,
2008 const char *name, int namelen,
2009 struct ext4_dir_entry_2 **dest_de);
2010void ext4_insert_dentry(struct inode *inode,
2011 struct ext4_dir_entry_2 *de,
2012 int buf_size,
2013 const char *name, int namelen);
2014static inline void ext4_update_dx_flag(struct inode *inode)
2015{
2016 if (!EXT4_HAS_COMPAT_FEATURE(inode->i_sb,
2017 EXT4_FEATURE_COMPAT_DIR_INDEX))
2018 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2019}
65d165d9
TM
2020static unsigned char ext4_filetype_table[] = {
2021 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2022};
2023
2024static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2025{
2026 if (!EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE) ||
2027 (filetype >= EXT4_FT_MAX))
2028 return DT_UNKNOWN;
2029
2030 return ext4_filetype_table[filetype];
2031}
40b163f1
DW
2032extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2033 void *buf, int buf_size);
ac27a0ec
DK
2034
2035/* fsync.c */
02c24a82 2036extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
ac27a0ec
DK
2037
2038/* hash.c */
617ba13b 2039extern int ext4fs_dirhash(const char *name, int len, struct
ac27a0ec
DK
2040 dx_hash_info *hinfo);
2041
2042/* ialloc.c */
1139575a
TT
2043extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2044 const struct qstr *qstr, __u32 goal,
2045 uid_t *owner, int handle_type,
2046 unsigned int line_no, int nblocks);
2047
2048#define ext4_new_inode(handle, dir, mode, qstr, goal, owner) \
2049 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2050 0, 0, 0)
2051#define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2052 type, nblocks) \
2053 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2054 (type), __LINE__, (nblocks))
2055
2056
af5bc92d
TT
2057extern void ext4_free_inode(handle_t *, struct inode *);
2058extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2059extern unsigned long ext4_count_free_inodes(struct super_block *);
2060extern unsigned long ext4_count_dirs(struct super_block *);
2061extern void ext4_check_inodes_bitmap(struct super_block *);
61d08673 2062extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
bfff6873
LC
2063extern int ext4_init_inode_table(struct super_block *sb,
2064 ext4_group_t group, int barrier);
813e5727 2065extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
ac27a0ec 2066
c9de560d
AT
2067/* mballoc.c */
2068extern long ext4_mb_stats;
2069extern long ext4_mb_max_to_scan;
9d99012f 2070extern int ext4_mb_init(struct super_block *);
c9de560d
AT
2071extern int ext4_mb_release(struct super_block *);
2072extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2073 struct ext4_allocation_request *, int *);
2074extern int ext4_mb_reserve_blocks(struct super_block *, int);
c2ea3fde 2075extern void ext4_discard_preallocations(struct inode *);
5dabfc78
TT
2076extern int __init ext4_init_mballoc(void);
2077extern void ext4_exit_mballoc(void);
44338711 2078extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
e6362609
TT
2079 struct buffer_head *bh, ext4_fsblk_t block,
2080 unsigned long count, int flags);
28623c2f
TT
2081extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2082 ext4_group_t ngroups);
920313a7 2083extern int ext4_mb_add_groupinfo(struct super_block *sb,
5f21b0e6 2084 ext4_group_t i, struct ext4_group_desc *desc);
cc7365df 2085extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2846e820 2086 ext4_fsblk_t block, unsigned long count);
7360d173
LC
2087extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2088
ac27a0ec 2089/* inode.c */
10560082 2090struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
1c215028 2091struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, 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 *,
dfe50809 2733 struct ext4_ext_path **,
4a092d73
TT
2734 struct ext4_extent *, int);
2735extern struct ext4_ext_path *ext4_ext_find_extent(struct inode *, ext4_lblk_t,
705912ca 2736 struct ext4_ext_path **,
107a7bd3 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);
fcf6b1b7 2744extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3a06d778
AK
2745extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
2746 __u64 start, __u64 len);
7869a4a6 2747extern int ext4_ext_precache(struct inode *inode);
9eb79482 2748extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len);
fcf6b1b7
DM
2749extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
2750 struct inode *inode2, ext4_lblk_t lblk1,
2751 ext4_lblk_t lblk2, ext4_lblk_t count,
2752 int mark_unwritten,int *err);
4a092d73 2753
748de673 2754/* move_extent.c */
393d1d1d
DTB
2755extern void ext4_double_down_write_data_sem(struct inode *first,
2756 struct inode *second);
2757extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
2758 struct inode *donor_inode);
748de673
AF
2759extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
2760 __u64 start_orig, __u64 start_donor,
2761 __u64 len, __u64 *moved_len);
2762
bd2d0210 2763/* page-io.c */
5dabfc78
TT
2764extern int __init ext4_init_pageio(void);
2765extern void ext4_exit_pageio(void);
bd2d0210 2766extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
97a851ed
JK
2767extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
2768extern int ext4_put_io_end(ext4_io_end_t *io_end);
2769extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
2770extern void ext4_io_submit_init(struct ext4_io_submit *io,
2771 struct writeback_control *wbc);
2e8fa54e 2772extern void ext4_end_io_rsv_work(struct work_struct *work);
bd2d0210
TT
2773extern void ext4_io_submit(struct ext4_io_submit *io);
2774extern int ext4_bio_write_page(struct ext4_io_submit *io,
2775 struct page *page,
2776 int len,
1c8349a1
NJ
2777 struct writeback_control *wbc,
2778 bool keep_towrite);
3a06d778 2779
c5e06d10
JL
2780/* mmp.c */
2781extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
2782
3613d228
JK
2783/*
2784 * Note that these flags will never ever appear in a buffer_head's state flag.
2785 * See EXT4_MAP_... to see where this is used.
2786 */
744692dc 2787enum ext4_state_bits {
090f32ee 2788 BH_AllocFromCluster /* allocated blocks were part of already
3613d228 2789 * allocated cluster. */
090f32ee 2790 = BH_JBDPrivateStart
744692dc
JZ
2791};
2792
2ccb5fb9 2793/*
48fc7f7e 2794 * Add new method to test whether block and inode bitmaps are properly
2ccb5fb9
AK
2795 * initialized. With uninit_bg reading the block from disk is not enough
2796 * to mark the bitmap uptodate. We need to also zero-out the bitmap
2797 */
2798#define BH_BITMAP_UPTODATE BH_JBDPrivateStart
2799
2800static inline int bitmap_uptodate(struct buffer_head *bh)
2801{
2802 return (buffer_uptodate(bh) &&
2803 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
2804}
2805static inline void set_bitmap_uptodate(struct buffer_head *bh)
2806{
2807 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
2808}
2809
17335dcc
DM
2810/*
2811 * Disable DIO read nolock optimization, so new dioreaders will be forced
2812 * to grab i_mutex
2813 */
2814static inline void ext4_inode_block_unlocked_dio(struct inode *inode)
2815{
2816 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
2817 smp_mb();
2818}
2819static inline void ext4_inode_resume_unlocked_dio(struct inode *inode)
2820{
2821 smp_mb();
2822 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
2823}
2824
731eb1a0
AM
2825#define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
2826
e9e3bcec
ES
2827/* For ioend & aio unwritten conversion wait queues */
2828#define EXT4_WQ_HASH_SZ 37
2829#define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
2830 EXT4_WQ_HASH_SZ])
2831#define ext4_aio_mutex(v) (&ext4__aio_mutex[((unsigned long)(v)) %\
2832 EXT4_WQ_HASH_SZ])
2833extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
2834extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ];
2835
8f82f840
YY
2836#define EXT4_RESIZING 0
2837extern int ext4_resize_begin(struct super_block *sb);
2838extern void ext4_resize_end(struct super_block *sb);
2839
ac27a0ec
DK
2840#endif /* __KERNEL__ */
2841
3dcf5451 2842#endif /* _EXT4_H */