ext4: fix block reservation for bigalloc filesystems
[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)
e35fd660 161#define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
cbd7584e 162 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
e35fd660
TT
163
164struct ext4_map_blocks {
165 ext4_fsblk_t m_pblk;
166 ext4_lblk_t m_lblk;
167 unsigned int m_len;
168 unsigned int m_flags;
169};
170
bd2d0210
TT
171/*
172 * Flags for ext4_io_end->flags
173 */
174#define EXT4_IO_END_UNWRITTEN 0x0001
bd2d0210 175
4188188b 176/*
556615dc 177 * For converting unwritten extents on a work queue. 'handle' is used for
6b523df4 178 * buffered writeback.
4188188b 179 */
0031462b 180typedef struct ext4_io_end {
5b3ff237 181 struct list_head list; /* per-file finished IO list */
6b523df4
JK
182 handle_t *handle; /* handle reserved for extent
183 * conversion */
0031462b 184 struct inode *inode; /* file being written to */
b0857d30
JK
185 struct bio *bio; /* Linked list of completed
186 * bios covering the extent */
8d5d02e6 187 unsigned int flag; /* unwritten or not */
a1de02dc
ES
188 loff_t offset; /* offset in the file */
189 ssize_t size; /* size of the extent */
97a851ed 190 atomic_t count; /* reference counter */
0031462b
MC
191} ext4_io_end_t;
192
bd2d0210
TT
193struct ext4_io_submit {
194 int io_op;
195 struct bio *io_bio;
196 ext4_io_end_t *io_end;
bd2d0210
TT
197 sector_t io_next_block;
198};
199
ac27a0ec
DK
200/*
201 * Special inodes numbers
202 */
617ba13b
MC
203#define EXT4_BAD_INO 1 /* Bad blocks inode */
204#define EXT4_ROOT_INO 2 /* Root inode */
ae812306
AK
205#define EXT4_USR_QUOTA_INO 3 /* User quota inode */
206#define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
617ba13b
MC
207#define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
208#define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
209#define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
210#define EXT4_JOURNAL_INO 8 /* Journal inode */
ac27a0ec 211
617ba13b
MC
212/* First non-reserved inode for old ext4 filesystems */
213#define EXT4_GOOD_OLD_FIRST_INO 11
ac27a0ec
DK
214
215/*
216 * Maximal count of links to a file
217 */
f8628a14 218#define EXT4_LINK_MAX 65000
ac27a0ec
DK
219
220/*
221 * Macro-instructions used to manage several block sizes
222 */
617ba13b 223#define EXT4_MIN_BLOCK_SIZE 1024
afc7cbca
TS
224#define EXT4_MAX_BLOCK_SIZE 65536
225#define EXT4_MIN_BLOCK_LOG_SIZE 10
fb1813f4 226#define EXT4_MAX_BLOCK_LOG_SIZE 16
ac27a0ec 227#ifdef __KERNEL__
617ba13b 228# define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
ac27a0ec 229#else
617ba13b 230# define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
ac27a0ec 231#endif
af5bc92d 232#define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
281b5995
TT
233#define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
234 EXT4_SB(s)->s_cluster_bits)
ac27a0ec 235#ifdef __KERNEL__
617ba13b 236# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
281b5995 237# define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
ac27a0ec 238#else
617ba13b 239# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
ac27a0ec
DK
240#endif
241#ifdef __KERNEL__
617ba13b
MC
242#define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
243#define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
244#define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
ac27a0ec 245#else
617ba13b
MC
246#define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
247 EXT4_GOOD_OLD_INODE_SIZE : \
ac27a0ec 248 (s)->s_inode_size)
617ba13b
MC
249#define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
250 EXT4_GOOD_OLD_FIRST_INO : \
ac27a0ec
DK
251 (s)->s_first_ino)
252#endif
a2df2a63 253#define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
ac27a0ec 254
d5b8f310
TT
255/* Translate a block number to a cluster number */
256#define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
257/* Translate a cluster number to a block number */
258#define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
259/* Translate # of blks to # of clusters */
260#define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
261 (sbi)->s_cluster_bits)
f5a44db5
TT
262/* Mask out the low bits to get the starting block of the cluster */
263#define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
264 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
265#define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
266 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
267/* Get the cluster offset */
268#define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
269 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
270#define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
271 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
d5b8f310 272
ac27a0ec
DK
273/*
274 * Structure of a blocks group descriptor
275 */
617ba13b 276struct ext4_group_desc
ac27a0ec 277{
3a14589c 278 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
5272f837
AK
279 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
280 __le32 bg_inode_table_lo; /* Inodes table block */
560671a0
AK
281 __le16 bg_free_blocks_count_lo;/* Free blocks count */
282 __le16 bg_free_inodes_count_lo;/* Free inodes count */
283 __le16 bg_used_dirs_count_lo; /* Directories count */
717d50e4 284 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
e6153918
DW
285 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
286 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
287 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
560671a0 288 __le16 bg_itable_unused_lo; /* Unused inodes count */
717d50e4 289 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
8fadc143
AR
290 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
291 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
292 __le32 bg_inode_table_hi; /* Inodes table block MSB */
91b51a01
CL
293 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
294 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
295 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
560671a0 296 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
e6153918
DW
297 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
298 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
299 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
300 __u32 bg_reserved;
ac27a0ec
DK
301};
302
41a246d1
DW
303#define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
304 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
305 sizeof(__le16))
306#define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
307 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
308 sizeof(__le16))
309
772cb7c8
JS
310/*
311 * Structure of a flex block group info
312 */
313
314struct flex_groups {
90ba983f
TT
315 atomic64_t free_clusters;
316 atomic_t free_inodes;
317 atomic_t used_dirs;
772cb7c8
JS
318};
319
717d50e4
AD
320#define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
321#define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
322#define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
323
ac27a0ec
DK
324/*
325 * Macro-instructions used to manage group descriptors
326 */
0d1ee42f 327#define EXT4_MIN_DESC_SIZE 32
8fadc143 328#define EXT4_MIN_DESC_SIZE_64BIT 64
0d1ee42f
AR
329#define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
330#define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
ac27a0ec 331#ifdef __KERNEL__
617ba13b 332# define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
281b5995 333# define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
617ba13b
MC
334# define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
335# define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
336# define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
ac27a0ec 337#else
617ba13b 338# define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
0d1ee42f 339# define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
617ba13b 340# define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
ac27a0ec
DK
341#endif
342
343/*
344 * Constants relative to the data blocks
345 */
617ba13b
MC
346#define EXT4_NDIR_BLOCKS 12
347#define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
348#define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
349#define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
350#define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
ac27a0ec
DK
351
352/*
353 * Inode flags
354 */
617ba13b
MC
355#define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
356#define EXT4_UNRM_FL 0x00000002 /* Undelete */
357#define EXT4_COMPR_FL 0x00000004 /* Compress file */
358#define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
359#define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
360#define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
361#define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
362#define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
ac27a0ec 363/* Reserved for compression usage... */
617ba13b
MC
364#define EXT4_DIRTY_FL 0x00000100
365#define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
366#define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
367#define EXT4_ECOMPR_FL 0x00000800 /* Compression error */
ac27a0ec 368/* End compression flags --- maybe not all used */
617ba13b
MC
369#define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
370#define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
371#define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
372#define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
373#define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
374#define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
8180a562 375#define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
a86c6181 376#define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
f710b4b9 377#define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
c8d46e41 378#define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
67cf5b09 379#define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
8180a562 380#define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
ac27a0ec 381
c8d46e41 382#define EXT4_FL_USER_VISIBLE 0x004BDFFF /* User visible flags */
996bb9fd 383#define EXT4_FL_USER_MODIFIABLE 0x004380FF /* User modifiable flags */
ac27a0ec 384
8fa43a81
DG
385/* Flags that should be inherited by new inodes from their parent. */
386#define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
1cd9f097 387 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
8fa43a81
DG
388 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
389 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
390
2dc6b0d4
DG
391/* Flags that are appropriate for regular files (all but dir-specific ones). */
392#define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
393
394/* Flags that are appropriate for non-directories/regular files. */
395#define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
396
397/* Mask out flags that are inappropriate for the given type of inode. */
398static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
399{
400 if (S_ISDIR(mode))
401 return flags;
402 else if (S_ISREG(mode))
403 return flags & EXT4_REG_FLMASK;
404 else
405 return flags & EXT4_OTHER_FLMASK;
406}
407
12e9b892
DM
408/*
409 * Inode flags used for atomic set/get
410 */
411enum {
412 EXT4_INODE_SECRM = 0, /* Secure deletion */
413 EXT4_INODE_UNRM = 1, /* Undelete */
414 EXT4_INODE_COMPR = 2, /* Compress file */
415 EXT4_INODE_SYNC = 3, /* Synchronous updates */
416 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
417 EXT4_INODE_APPEND = 5, /* writes to file may only append */
418 EXT4_INODE_NODUMP = 6, /* do not dump file */
419 EXT4_INODE_NOATIME = 7, /* do not update atime */
420/* Reserved for compression usage... */
421 EXT4_INODE_DIRTY = 8,
422 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
423 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
424 EXT4_INODE_ECOMPR = 11, /* Compression error */
425/* End compression flags --- maybe not all used */
426 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
427 EXT4_INODE_IMAGIC = 13, /* AFS directory */
428 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
429 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
430 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
431 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
432 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
433 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
434 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
435 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
67cf5b09 436 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
12e9b892
DM
437 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
438};
439
12e9b892 440/*
9a4c8019
CM
441 * Since it's pretty easy to mix up bit numbers and hex values, we use a
442 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
443 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
444 * any extra space in the compiled kernel image, otherwise, the build will fail.
445 * It's important that these values are the same, since we are using
446 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
447 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
448 * values found in ext2, ext3 and ext4 filesystems, and of course the values
449 * defined in e2fsprogs.
12e9b892
DM
450 *
451 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
452 */
9a4c8019
CM
453#define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
454#define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
455
12e9b892
DM
456static inline void ext4_check_flag_values(void)
457{
458 CHECK_FLAG_VALUE(SECRM);
459 CHECK_FLAG_VALUE(UNRM);
460 CHECK_FLAG_VALUE(COMPR);
461 CHECK_FLAG_VALUE(SYNC);
462 CHECK_FLAG_VALUE(IMMUTABLE);
463 CHECK_FLAG_VALUE(APPEND);
464 CHECK_FLAG_VALUE(NODUMP);
465 CHECK_FLAG_VALUE(NOATIME);
466 CHECK_FLAG_VALUE(DIRTY);
467 CHECK_FLAG_VALUE(COMPRBLK);
468 CHECK_FLAG_VALUE(NOCOMPR);
469 CHECK_FLAG_VALUE(ECOMPR);
470 CHECK_FLAG_VALUE(INDEX);
471 CHECK_FLAG_VALUE(IMAGIC);
472 CHECK_FLAG_VALUE(JOURNAL_DATA);
473 CHECK_FLAG_VALUE(NOTAIL);
474 CHECK_FLAG_VALUE(DIRSYNC);
475 CHECK_FLAG_VALUE(TOPDIR);
476 CHECK_FLAG_VALUE(HUGE_FILE);
477 CHECK_FLAG_VALUE(EXTENTS);
478 CHECK_FLAG_VALUE(EA_INODE);
479 CHECK_FLAG_VALUE(EOFBLOCKS);
67cf5b09 480 CHECK_FLAG_VALUE(INLINE_DATA);
12e9b892
DM
481 CHECK_FLAG_VALUE(RESERVED);
482}
483
ac27a0ec 484/* Used to pass group descriptor data when online resize is done */
617ba13b 485struct ext4_new_group_input {
8c55e204
DK
486 __u32 group; /* Group number for this data */
487 __u64 block_bitmap; /* Absolute block number of block bitmap */
488 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
489 __u64 inode_table; /* Absolute block number of inode table start */
490 __u32 blocks_count; /* Total number of blocks in this group */
491 __u16 reserved_blocks; /* Number of reserved blocks in this group */
ac27a0ec
DK
492 __u16 unused;
493};
494
4d92dc0f
BH
495#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
496struct compat_ext4_new_group_input {
497 u32 group;
498 compat_u64 block_bitmap;
499 compat_u64 inode_bitmap;
500 compat_u64 inode_table;
501 u32 blocks_count;
502 u16 reserved_blocks;
503 u16 unused;
504};
505#endif
506
617ba13b
MC
507/* The struct ext4_new_group_input in kernel space, with free_blocks_count */
508struct ext4_new_group_data {
ac27a0ec 509 __u32 group;
bd81d8ee
LV
510 __u64 block_bitmap;
511 __u64 inode_bitmap;
512 __u64 inode_table;
ac27a0ec
DK
513 __u32 blocks_count;
514 __u16 reserved_blocks;
515 __u16 unused;
516 __u32 free_blocks_count;
517};
518
33afdcc5
YY
519/* Indexes used to index group tables in ext4_new_group_data */
520enum {
521 BLOCK_BITMAP = 0, /* block bitmap */
522 INODE_BITMAP, /* inode bitmap */
523 INODE_TABLE, /* inode tables */
524 GROUP_TABLE_COUNT,
525};
526
a2df2a63 527/*
79e83036 528 * Flags used by ext4_map_blocks()
a2df2a63 529 */
556615dc 530 /* Allocate any needed blocks and/or convert an unwritten
c2177057 531 extent to be an initialized ext4 */
2ac3b6e0 532#define EXT4_GET_BLOCKS_CREATE 0x0001
556615dc
LC
533 /* Request the creation of an unwritten extent */
534#define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002
535#define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\
c2177057 536 EXT4_GET_BLOCKS_CREATE)
27dd4385
LC
537 /* Caller is from the delayed allocation writeout path
538 * finally doing the actual allocation of delayed blocks */
03f5d8bc 539#define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
0031462b 540 /* caller is from the direct IO path, request to creation of an
556615dc 541 unwritten extents if not allocated, split the unwritten
0031462b 542 extent if blocks has been preallocated already*/
c7064ef1 543#define EXT4_GET_BLOCKS_PRE_IO 0x0008
1296cc85 544#define EXT4_GET_BLOCKS_CONVERT 0x0010
c7064ef1 545#define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
556615dc 546 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
c7064ef1
JZ
547 /* Convert extent to initialized after IO complete */
548#define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
556615dc 549 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
27dd4385
LC
550 /* Eventual metadata allocation (due to growing extent tree)
551 * should not fail, so try to use reserved blocks for that.*/
552#define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
556b27ab
VH
553 /* Don't normalize allocation size (used for fallocate) */
554#define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
a4e5d88b
DM
555 /* Request will not result in inode size update (user for fallocate) */
556#define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080
729f52c6
ZL
557 /* Do not take i_data_sem locking in ext4_map_blocks */
558#define EXT4_GET_BLOCKS_NO_LOCK 0x0100
d100eef2
ZL
559 /* Do not put hole in extent cache */
560#define EXT4_GET_BLOCKS_NO_PUT_HOLE 0x0200
b8a86845
LC
561 /* Convert written extents to unwritten */
562#define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0400
ac27a0ec 563
107a7bd3 564/*
7869a4a6 565 * The bit position of these flags must not overlap with any of the
ed8a1a76 566 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(),
107a7bd3 567 * read_extent_tree_block(), ext4_split_extent_at(),
7869a4a6
TT
568 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
569 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
570 * caching the extents when reading from the extent tree while a
571 * truncate or punch hole operation is in progress.
107a7bd3 572 */
d26e2c4d
TT
573#define EXT4_EX_NOCACHE 0x40000000
574#define EXT4_EX_FORCE_CACHE 0x20000000
107a7bd3 575
e6362609
TT
576/*
577 * Flags used by ext4_free_blocks
578 */
579#define EXT4_FREE_BLOCKS_METADATA 0x0001
580#define EXT4_FREE_BLOCKS_FORGET 0x0002
1f2acb60 581#define EXT4_FREE_BLOCKS_VALIDATED 0x0004
7132de74 582#define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
84130193
TT
583#define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
584#define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
e6362609 585
ac27a0ec
DK
586/*
587 * ioctl commands
588 */
617ba13b
MC
589#define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
590#define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
591#define EXT4_IOC_GETVERSION _IOR('f', 3, long)
592#define EXT4_IOC_SETVERSION _IOW('f', 4, long)
617ba13b
MC
593#define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
594#define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
617ba13b
MC
595#define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
596#define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
8eea80d5
TT
597#define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
598#define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
599#define EXT4_IOC_MIGRATE _IO('f', 9)
ccd2506b 600 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
8eea80d5 601 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
ccd2506b 602#define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
748de673 603#define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
19c5246d 604#define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
393d1d1d 605#define EXT4_IOC_SWAP_BOOT _IO('f', 17)
7869a4a6 606#define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18)
ac27a0ec 607
899ad0ce 608#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
ac27a0ec
DK
609/*
610 * ioctl commands in 32 bit emulation
611 */
617ba13b
MC
612#define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
613#define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
614#define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
615#define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
616#define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
617#define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
618#define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
4d92dc0f 619#define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
617ba13b
MC
620#define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
621#define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
899ad0ce 622#endif
ac27a0ec 623
b595076a 624/* Max physical block we can address w/o extents */
fb0a387d
ES
625#define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
626
ac27a0ec
DK
627/*
628 * Structure of an inode on the disk
629 */
617ba13b 630struct ext4_inode {
ac27a0ec
DK
631 __le16 i_mode; /* File mode */
632 __le16 i_uid; /* Low 16 bits of Owner Uid */
a48380f7 633 __le32 i_size_lo; /* Size in bytes */
ac27a0ec 634 __le32 i_atime; /* Access time */
ef7f3835 635 __le32 i_ctime; /* Inode Change time */
ac27a0ec
DK
636 __le32 i_mtime; /* Modification time */
637 __le32 i_dtime; /* Deletion Time */
638 __le16 i_gid; /* Low 16 bits of Group Id */
639 __le16 i_links_count; /* Links count */
0fc1b451 640 __le32 i_blocks_lo; /* Blocks count */
ac27a0ec
DK
641 __le32 i_flags; /* File flags */
642 union {
643 struct {
25ec56b5 644 __le32 l_i_version;
ac27a0ec
DK
645 } linux1;
646 struct {
647 __u32 h_i_translator;
648 } hurd1;
649 struct {
650 __u32 m_i_reserved1;
651 } masix1;
652 } osd1; /* OS dependent 1 */
617ba13b 653 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
ac27a0ec 654 __le32 i_generation; /* File version (for NFS) */
7973c0c1 655 __le32 i_file_acl_lo; /* File ACL */
a48380f7 656 __le32 i_size_high;
f077d0d7 657 __le32 i_obso_faddr; /* Obsoleted fragment address */
ac27a0ec
DK
658 union {
659 struct {
0fc1b451 660 __le16 l_i_blocks_high; /* were l_i_reserved1 */
a1ddeb7e 661 __le16 l_i_file_acl_high;
8c55e204 662 __le16 l_i_uid_high; /* these 2 fields */
ac27a0ec 663 __le16 l_i_gid_high; /* were reserved2[0] */
e6153918
DW
664 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
665 __le16 l_i_reserved;
ac27a0ec
DK
666 } linux2;
667 struct {
f077d0d7 668 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
ac27a0ec
DK
669 __u16 h_i_mode_high;
670 __u16 h_i_uid_high;
671 __u16 h_i_gid_high;
672 __u32 h_i_author;
673 } hurd2;
674 struct {
f077d0d7 675 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
a1ddeb7e 676 __le16 m_i_file_acl_high;
ac27a0ec
DK
677 __u32 m_i_reserved2[2];
678 } masix2;
679 } osd2; /* OS dependent 2 */
680 __le16 i_extra_isize;
e6153918 681 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
ef7f3835
KS
682 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
683 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
684 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
685 __le32 i_crtime; /* File Creation time */
686 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
25ec56b5 687 __le32 i_version_hi; /* high 32 bits for 64-bit version */
ac27a0ec
DK
688};
689
748de673
AF
690struct move_extent {
691 __u32 reserved; /* should be zero */
692 __u32 donor_fd; /* donor file descriptor */
693 __u64 orig_start; /* logical start offset in block for orig */
694 __u64 donor_start; /* logical start offset in block for donor */
695 __u64 len; /* block length to be moved */
696 __u64 moved_len; /* moved block length */
697};
ac27a0ec 698
ef7f3835
KS
699#define EXT4_EPOCH_BITS 2
700#define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
701#define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
702
703/*
704 * Extended fields will fit into an inode if the filesystem was formatted
705 * with large inodes (-I 256 or larger) and there are not currently any EAs
706 * consuming all of the available space. For new inodes we always reserve
707 * enough space for the kernel's known extended fields, but for inodes
708 * created with an old kernel this might not have been the case. None of
709 * the extended inode fields is critical for correct filesystem operation.
710 * This macro checks if a certain field fits in the inode. Note that
711 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
712 */
713#define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
714 ((offsetof(typeof(*ext4_inode), field) + \
715 sizeof((ext4_inode)->field)) \
716 <= (EXT4_GOOD_OLD_INODE_SIZE + \
717 (einode)->i_extra_isize)) \
718
719static inline __le32 ext4_encode_extra_time(struct timespec *time)
720{
721 return cpu_to_le32((sizeof(time->tv_sec) > 4 ?
c1fccc06
TT
722 (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) |
723 ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK));
ef7f3835
KS
724}
725
726static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
727{
728 if (sizeof(time->tv_sec) > 4)
729 time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK)
730 << 32;
c1fccc06 731 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
ef7f3835
KS
732}
733
734#define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
735do { \
736 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
737 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
738 (raw_inode)->xtime ## _extra = \
739 ext4_encode_extra_time(&(inode)->xtime); \
740} while (0)
741
742#define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
743do { \
744 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
745 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
746 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
747 (raw_inode)->xtime ## _extra = \
748 ext4_encode_extra_time(&(einode)->xtime); \
749} while (0)
750
751#define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
752do { \
753 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
754 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
755 ext4_decode_extra_time(&(inode)->xtime, \
756 raw_inode->xtime ## _extra); \
af0b44a1
ES
757 else \
758 (inode)->xtime.tv_nsec = 0; \
ef7f3835
KS
759} while (0)
760
761#define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
762do { \
763 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
764 (einode)->xtime.tv_sec = \
765 (signed)le32_to_cpu((raw_inode)->xtime); \
19ea8060
TT
766 else \
767 (einode)->xtime.tv_sec = 0; \
ef7f3835
KS
768 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
769 ext4_decode_extra_time(&(einode)->xtime, \
770 raw_inode->xtime ## _extra); \
af0b44a1
ES
771 else \
772 (einode)->xtime.tv_nsec = 0; \
ef7f3835
KS
773} while (0)
774
25ec56b5
JNC
775#define i_disk_version osd1.linux1.l_i_version
776
ac27a0ec
DK
777#if defined(__KERNEL__) || defined(__linux__)
778#define i_reserved1 osd1.linux1.l_i_reserved1
a1ddeb7e 779#define i_file_acl_high osd2.linux2.l_i_file_acl_high
0fc1b451 780#define i_blocks_high osd2.linux2.l_i_blocks_high
ac27a0ec
DK
781#define i_uid_low i_uid
782#define i_gid_low i_gid
783#define i_uid_high osd2.linux2.l_i_uid_high
784#define i_gid_high osd2.linux2.l_i_gid_high
e6153918 785#define i_checksum_lo osd2.linux2.l_i_checksum_lo
ac27a0ec
DK
786
787#elif defined(__GNU__)
788
789#define i_translator osd1.hurd1.h_i_translator
ac27a0ec
DK
790#define i_uid_high osd2.hurd2.h_i_uid_high
791#define i_gid_high osd2.hurd2.h_i_gid_high
792#define i_author osd2.hurd2.h_i_author
793
794#elif defined(__masix__)
795
796#define i_reserved1 osd1.masix1.m_i_reserved1
a1ddeb7e 797#define i_file_acl_high osd2.masix2.m_i_file_acl_high
ac27a0ec
DK
798#define i_reserved2 osd2.masix2.m_i_reserved2
799
800#endif /* defined(__KERNEL__) || defined(__linux__) */
801
c0677e6d
ZL
802#include "extents_status.h"
803
d444c3c3
TT
804/*
805 * fourth extended file system inode data in memory
806 */
807struct ext4_inode_info {
808 __le32 i_data[15]; /* unconverted */
d444c3c3 809 __u32 i_dtime;
12e9b892 810 ext4_fsblk_t i_file_acl;
d444c3c3
TT
811
812 /*
813 * i_block_group is the number of the block group which contains
814 * this file's inode. Constant across the lifetime of the inode,
815 * it is ued for making block allocation decisions - we try to
816 * place a file's data blocks near its inode block, and new inodes
817 * near to their parent directory's inode.
818 */
819 ext4_group_t i_block_group;
8a2005d3 820 ext4_lblk_t i_dir_start_lookup;
353eb83c 821#if (BITS_PER_LONG < 64)
19f5fb7a 822 unsigned long i_state_flags; /* Dynamic state flags */
353eb83c 823#endif
12e9b892 824 unsigned long i_flags;
d444c3c3 825
d444c3c3
TT
826 /*
827 * Extended attributes can be read independently of the main file
828 * data. Taking i_mutex even when reading would cause contention
829 * between readers of EAs and writers of regular file data, so
830 * instead we synchronize on xattr_sem when reading or changing
831 * EAs.
832 */
833 struct rw_semaphore xattr_sem;
d444c3c3
TT
834
835 struct list_head i_orphan; /* unlinked but open inodes */
836
837 /*
838 * i_disksize keeps track of what the inode size is ON DISK, not
839 * in memory. During truncate, i_size is set to the new size by
840 * the VFS prior to calling ext4_truncate(), but the filesystem won't
841 * set i_disksize to 0 until the truncate is actually under way.
842 *
843 * The intent is that i_disksize always represents the blocks which
844 * are used by this file. This allows recovery to restart truncate
845 * on orphans if we crash during truncate. We actually write i_disksize
846 * into the on-disk inode when writing inodes out, instead of i_size.
847 *
848 * The only time when i_disksize and i_size may be different is when
849 * a truncate is in progress. The only things which change i_disksize
850 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
851 */
852 loff_t i_disksize;
853
854 /*
855 * i_data_sem is for serialising ext4_truncate() against
856 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
857 * data tree are chopped off during truncate. We can't do that in
858 * ext4 because whenever we perform intermediate commits during
859 * truncate, the inode and all the metadata blocks *must* be in a
860 * consistent state which allows truncation of the orphans to restart
861 * during recovery. Hence we must fix the get_block-vs-truncate race
862 * by other means, so we have i_data_sem.
863 */
864 struct rw_semaphore i_data_sem;
865 struct inode vfs_inode;
8aefcd55 866 struct jbd2_inode *jinode;
d444c3c3 867
202ee5df
TT
868 spinlock_t i_raw_lock; /* protects updates to the raw inode */
869
d444c3c3
TT
870 /*
871 * File creation time. Its function is same as that of
872 * struct timespec i_{a,c,m}time in the generic inode.
873 */
874 struct timespec i_crtime;
875
876 /* mballoc */
877 struct list_head i_prealloc_list;
878 spinlock_t i_prealloc_lock;
879
c0677e6d
ZL
880 /* extents status tree */
881 struct ext4_es_tree i_es_tree;
882 rwlock_t i_es_lock;
74cd15cd 883 struct list_head i_es_lru;
eb68d0e2 884 unsigned int i_es_all_nr; /* protected by i_es_lock */
74cd15cd 885 unsigned int i_es_lru_nr; /* protected by i_es_lock */
d3922a77 886 unsigned long i_touch_when; /* jiffies of last accessing */
c0677e6d 887
d444c3c3
TT
888 /* ialloc */
889 ext4_group_t i_last_alloc_group;
890
891 /* allocation reservation info for delalloc */
57042651 892 /* In case of bigalloc, these refer to clusters rather than blocks */
d444c3c3
TT
893 unsigned int i_reserved_data_blocks;
894 unsigned int i_reserved_meta_blocks;
895 unsigned int i_allocated_meta_blocks;
01f49d0b 896 ext4_lblk_t i_da_metadata_calc_last_lblock;
9d0be502 897 int i_da_metadata_calc_len;
d444c3c3
TT
898
899 /* on-disk additional length */
900 __u16 i_extra_isize;
901
67cf5b09
TM
902 /* Indicate the inline data space. */
903 u16 i_inline_off;
904 u16 i_inline_size;
905
a9e7f447
DM
906#ifdef CONFIG_QUOTA
907 /* quota space reservation, managed internally by quota code */
908 qsize_t i_reserved_quota;
909#endif
8d5d02e6 910
2e8fa54e 911 /* Lock protecting lists below */
744692dc 912 spinlock_t i_completed_io_lock;
2e8fa54e
JK
913 /*
914 * Completed IOs that need unwritten extents handling and have
915 * transaction reserved
916 */
917 struct list_head i_rsv_conversion_list;
918 /*
919 * Completed IOs that need unwritten extents handling and don't have
920 * transaction reserved
921 */
8a2005d3 922 atomic_t i_ioend_count; /* Number of outstanding io_end structs */
e27f41e1 923 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
2e8fa54e 924 struct work_struct i_rsv_conversion_work;
8a2005d3
TT
925
926 spinlock_t i_block_reservation_lock;
b436b9be
JK
927
928 /*
929 * Transactions that contain inode's metadata needed to complete
930 * fsync and fdatasync, respectively.
931 */
932 tid_t i_sync_tid;
933 tid_t i_datasync_tid;
814525f4
DW
934
935 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
936 __u32 i_csum_seed;
d444c3c3
TT
937};
938
ac27a0ec
DK
939/*
940 * File system states
941 */
617ba13b
MC
942#define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
943#define EXT4_ERROR_FS 0x0002 /* Errors detected */
944#define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
ac27a0ec 945
469108ff
TT
946/*
947 * Misc. filesystem flags
948 */
949#define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
950#define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
951#define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
952
ac27a0ec 953/*
68911009 954 * Mount flags set via mount options or defaults
ac27a0ec 955 */
617ba13b
MC
956#define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
957#define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
958#define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
959#define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
960#define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
39ef17f1 961#define EXT4_MOUNT_ERRORS_MASK 0x00070
617ba13b
MC
962#define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
963#define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
617ba13b
MC
964#define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
965#define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
966#define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
967#define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
968#define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
969#define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
970#define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
971#define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
afd4672d 972#define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
617ba13b 973#define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
617ba13b
MC
974#define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
975#define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
976#define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
744692dc 977#define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
d4da6c9c 978#define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
818d276c 979#define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
64769240 980#define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
5bf5683a 981#define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
6fd058f7 982#define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
5328e635 983#define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
bfff6873 984#define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
5bf5683a 985
68911009
LC
986/*
987 * Mount flags set either automatically (could not be set by mount option)
988 * based on per file system feature or property or in special cases such as
989 * distinguishing between explicit mount option definition and default.
990 */
56889787
TT
991#define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
992 specified delalloc */
68911009
LC
993#define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
994 size of blocksize * 8
995 blocks */
ed3654eb
TT
996#define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
997 file systems */
56889787 998
fd8c37ec
TT
999#define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1000 ~EXT4_MOUNT_##opt
1001#define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1002 EXT4_MOUNT_##opt
617ba13b
MC
1003#define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1004 EXT4_MOUNT_##opt)
ac27a0ec 1005
a2595b8a
TT
1006#define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1007 ~EXT4_MOUNT2_##opt
1008#define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1009 EXT4_MOUNT2_##opt
1010#define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1011 EXT4_MOUNT2_##opt)
1012
597d508c
AM
1013#define ext4_test_and_set_bit __test_and_set_bit_le
1014#define ext4_set_bit __set_bit_le
617ba13b 1015#define ext4_set_bit_atomic ext2_set_bit_atomic
597d508c
AM
1016#define ext4_test_and_clear_bit __test_and_clear_bit_le
1017#define ext4_clear_bit __clear_bit_le
617ba13b 1018#define ext4_clear_bit_atomic ext2_clear_bit_atomic
50e0168c 1019#define ext4_test_bit test_bit_le
50e0168c
AM
1020#define ext4_find_next_zero_bit find_next_zero_bit_le
1021#define ext4_find_next_bit find_next_bit_le
ac27a0ec 1022
c3e94d1d
YY
1023extern void ext4_set_bits(void *bm, int cur, int len);
1024
ac27a0ec
DK
1025/*
1026 * Maximal mount counts between two filesystem checks
1027 */
617ba13b
MC
1028#define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1029#define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
ac27a0ec
DK
1030
1031/*
1032 * Behaviour when detecting errors
1033 */
617ba13b
MC
1034#define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1035#define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1036#define EXT4_ERRORS_PANIC 3 /* Panic */
1037#define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
ac27a0ec 1038
e6153918
DW
1039/* Metadata checksum algorithm codes */
1040#define EXT4_CRC32C_CHKSUM 1
1041
ac27a0ec
DK
1042/*
1043 * Structure of the super block
1044 */
617ba13b 1045struct ext4_super_block {
ac27a0ec 1046/*00*/ __le32 s_inodes_count; /* Inodes count */
6bc9feff 1047 __le32 s_blocks_count_lo; /* Blocks count */
308ba3ec
AK
1048 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1049 __le32 s_free_blocks_count_lo; /* Free blocks count */
ac27a0ec
DK
1050/*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1051 __le32 s_first_data_block; /* First Data Block */
1052 __le32 s_log_block_size; /* Block size */
281b5995 1053 __le32 s_log_cluster_size; /* Allocation cluster size */
ac27a0ec 1054/*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
281b5995 1055 __le32 s_clusters_per_group; /* # Clusters per group */
ac27a0ec
DK
1056 __le32 s_inodes_per_group; /* # Inodes per group */
1057 __le32 s_mtime; /* Mount time */
1058/*30*/ __le32 s_wtime; /* Write time */
1059 __le16 s_mnt_count; /* Mount count */
1060 __le16 s_max_mnt_count; /* Maximal mount count */
1061 __le16 s_magic; /* Magic signature */
1062 __le16 s_state; /* File system state */
1063 __le16 s_errors; /* Behaviour when detecting errors */
1064 __le16 s_minor_rev_level; /* minor revision level */
1065/*40*/ __le32 s_lastcheck; /* time of last check */
1066 __le32 s_checkinterval; /* max. time between checks */
1067 __le32 s_creator_os; /* OS */
1068 __le32 s_rev_level; /* Revision level */
1069/*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1070 __le16 s_def_resgid; /* Default gid for reserved blocks */
1071 /*
617ba13b 1072 * These fields are for EXT4_DYNAMIC_REV superblocks only.
ac27a0ec
DK
1073 *
1074 * Note: the difference between the compatible feature set and
1075 * the incompatible feature set is that if there is a bit set
1076 * in the incompatible feature set that the kernel doesn't
1077 * know about, it should refuse to mount the filesystem.
1078 *
1079 * e2fsck's requirements are more strict; if it doesn't know
1080 * about a feature in either the compatible or incompatible
1081 * feature set, it must abort and not try to meddle with
1082 * things it doesn't understand...
1083 */
1084 __le32 s_first_ino; /* First non-reserved inode */
0d1ee42f 1085 __le16 s_inode_size; /* size of inode structure */
ac27a0ec
DK
1086 __le16 s_block_group_nr; /* block group # of this superblock */
1087 __le32 s_feature_compat; /* compatible feature set */
1088/*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1089 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1090/*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1091/*78*/ char s_volume_name[16]; /* volume name */
1092/*88*/ char s_last_mounted[64]; /* directory where last mounted */
1093/*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1094 /*
1095 * Performance hints. Directory preallocation should only
617ba13b 1096 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
ac27a0ec
DK
1097 */
1098 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1099 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1100 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1101 /*
617ba13b 1102 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
ac27a0ec
DK
1103 */
1104/*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1105/*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1106 __le32 s_journal_dev; /* device number of journal file */
1107 __le32 s_last_orphan; /* start of list of inodes to delete */
1108 __le32 s_hash_seed[4]; /* HTREE hash seed */
1109 __u8 s_def_hash_version; /* Default hash version to use */
89eeddf0 1110 __u8 s_jnl_backup_type;
0d1ee42f 1111 __le16 s_desc_size; /* size of group descriptor */
bd81d8ee 1112/*100*/ __le32 s_default_mount_opts;
ac27a0ec 1113 __le32 s_first_meta_bg; /* First metablock block group */
bd81d8ee
LV
1114 __le32 s_mkfs_time; /* When the filesystem was created */
1115 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1116 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1117/*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1118 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1119 __le32 s_free_blocks_count_hi; /* Free blocks count */
c1bddad9
AK
1120 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1121 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1122 __le32 s_flags; /* Miscellaneous flags */
1123 __le16 s_raid_stride; /* RAID stride */
c5e06d10 1124 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
c1bddad9
AK
1125 __le64 s_mmp_block; /* Block for multi-mount protection */
1126 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
772cb7c8 1127 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
e6153918 1128 __u8 s_checksum_type; /* metadata checksum algorithm used */
772cb7c8 1129 __le16 s_reserved_pad;
afc32f7e 1130 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
4a9cdec7
TT
1131 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1132 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1133 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1134 snapshot's future use */
1135 __le32 s_snapshot_list; /* inode number of the head of the
1136 on-disk snapshot list */
1c13d5c0
TT
1137#define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1138 __le32 s_error_count; /* number of fs errors */
1139 __le32 s_first_error_time; /* first time an error happened */
1140 __le32 s_first_error_ino; /* inode involved in first error */
1141 __le64 s_first_error_block; /* block involved of first error */
1142 __u8 s_first_error_func[32]; /* function where the error happened */
1143 __le32 s_first_error_line; /* line number where error happened */
1144 __le32 s_last_error_time; /* most recent time of an error */
1145 __le32 s_last_error_ino; /* inode involved in last error */
1146 __le32 s_last_error_line; /* line number where error happened */
1147 __le64 s_last_error_block; /* block involved of last error */
1148 __u8 s_last_error_func[32]; /* function where the error happened */
8b67f04a
TT
1149#define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1150 __u8 s_mount_opts[64];
281b5995
TT
1151 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1152 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1153 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1beeef1b
DW
1154 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
1155 __le32 s_reserved[106]; /* Padding to the end of the block */
e6153918 1156 __le32 s_checksum; /* crc32c(superblock) */
ac27a0ec
DK
1157};
1158
1c13d5c0
TT
1159#define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1160
ac27a0ec 1161#ifdef __KERNEL__
bc0b0d6d
TT
1162
1163/*
4ab2f15b 1164 * run-time mount flags
bc0b0d6d
TT
1165 */
1166#define EXT4_MF_MNTDIR_SAMPLED 0x0001
4ab2f15b 1167#define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
bc0b0d6d 1168
a2d4a646
JK
1169/* Number of quota types we support */
1170#define EXT4_MAXQUOTAS 2
1171
ca0faba0
TT
1172/*
1173 * fourth extended-fs super-block data in memory
1174 */
1175struct ext4_sb_info {
1176 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1177 unsigned long s_inodes_per_block;/* Number of inodes per block */
1178 unsigned long s_blocks_per_group;/* Number of blocks in a group */
281b5995 1179 unsigned long s_clusters_per_group; /* Number of clusters in a group */
ca0faba0
TT
1180 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1181 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1182 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1183 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1184 ext4_group_t s_groups_count; /* Number of groups in the fs */
fb0a387d 1185 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
952fc18e 1186 unsigned long s_overhead; /* # of fs overhead clusters */
281b5995
TT
1187 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1188 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
ca0faba0
TT
1189 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1190 struct buffer_head * s_sbh; /* Buffer containing the super block */
1191 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1192 struct buffer_head **s_group_desc;
7f4520cc 1193 unsigned int s_mount_opt;
a2595b8a 1194 unsigned int s_mount_opt2;
bc0b0d6d 1195 unsigned int s_mount_flags;
5a916be1 1196 unsigned int s_def_mount_opt;
ca0faba0 1197 ext4_fsblk_t s_sb_block;
27dd4385 1198 atomic64_t s_resv_clusters;
08cefc7a
EB
1199 kuid_t s_resuid;
1200 kgid_t s_resgid;
ca0faba0
TT
1201 unsigned short s_mount_state;
1202 unsigned short s_pad;
1203 int s_addr_per_block_bits;
1204 int s_desc_per_block_bits;
1205 int s_inode_size;
1206 int s_first_ino;
1207 unsigned int s_inode_readahead_blks;
11013911 1208 unsigned int s_inode_goal;
ca0faba0
TT
1209 spinlock_t s_next_gen_lock;
1210 u32 s_next_generation;
1211 u32 s_hash_seed[4];
1212 int s_def_hash_version;
1213 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
57042651 1214 struct percpu_counter s_freeclusters_counter;
ca0faba0
TT
1215 struct percpu_counter s_freeinodes_counter;
1216 struct percpu_counter s_dirs_counter;
57042651 1217 struct percpu_counter s_dirtyclusters_counter;
ca0faba0
TT
1218 struct blockgroup_lock *s_blockgroup_lock;
1219 struct proc_dir_entry *s_proc;
1220 struct kobject s_kobj;
1221 struct completion s_kobj_unregister;
2c0544b2 1222 struct super_block *s_sb;
ca0faba0
TT
1223
1224 /* Journaling */
ca0faba0
TT
1225 struct journal_s *s_journal;
1226 struct list_head s_orphan;
1227 struct mutex s_orphan_lock;
8f82f840
YY
1228 unsigned long s_resize_flags; /* Flags indicating if there
1229 is a resizer */
ca0faba0
TT
1230 unsigned long s_commit_interval;
1231 u32 s_max_batch_time;
1232 u32 s_min_batch_time;
1233 struct block_device *journal_bdev;
ca0faba0 1234#ifdef CONFIG_QUOTA
a2d4a646 1235 char *s_qf_names[EXT4_MAXQUOTAS]; /* Names of quota files with journalled quota */
ca0faba0
TT
1236 int s_jquota_fmt; /* Format of quota to use */
1237#endif
1238 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
6fd058f7 1239 struct rb_root system_blks;
ca0faba0
TT
1240
1241#ifdef EXTENTS_STATS
1242 /* ext4 extents stats */
1243 unsigned long s_ext_min;
1244 unsigned long s_ext_max;
1245 unsigned long s_depth_max;
1246 spinlock_t s_ext_stats_lock;
1247 unsigned long s_ext_blocks;
1248 unsigned long s_ext_extents;
1249#endif
1250
1251 /* for buddy allocator */
1252 struct ext4_group_info ***s_group_info;
1253 struct inode *s_buddy_cache;
ca0faba0 1254 spinlock_t s_md_lock;
ca0faba0
TT
1255 unsigned short *s_mb_offsets;
1256 unsigned int *s_mb_maxs;
28623c2f 1257 unsigned int s_group_info_size;
ca0faba0
TT
1258
1259 /* tunables */
1260 unsigned long s_stripe;
1261 unsigned int s_mb_stream_request;
1262 unsigned int s_mb_max_to_scan;
1263 unsigned int s_mb_min_to_scan;
1264 unsigned int s_mb_stats;
1265 unsigned int s_mb_order2_reqs;
1266 unsigned int s_mb_group_prealloc;
df981d03 1267 unsigned int s_max_dir_size_kb;
ca0faba0
TT
1268 /* where last allocation was done - for stream allocation */
1269 unsigned long s_mb_last_group;
1270 unsigned long s_mb_last_start;
1271
ca0faba0 1272 /* stats for buddy allocator */
ca0faba0
TT
1273 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1274 atomic_t s_bal_success; /* we found long enough chunks */
1275 atomic_t s_bal_allocated; /* in blocks */
1276 atomic_t s_bal_ex_scanned; /* total extents scanned */
1277 atomic_t s_bal_goals; /* goal hits */
1278 atomic_t s_bal_breaks; /* too long searches */
1279 atomic_t s_bal_2orders; /* 2^order hits */
1280 spinlock_t s_bal_lock;
1281 unsigned long s_mb_buddies_generated;
1282 unsigned long long s_mb_generation_time;
1283 atomic_t s_mb_lost_chunks;
1284 atomic_t s_mb_preallocated;
1285 atomic_t s_mb_discarded;
50797481 1286 atomic_t s_lock_busy;
ca0faba0
TT
1287
1288 /* locality groups */
003cb608 1289 struct ext4_locality_group __percpu *s_locality_groups;
ca0faba0
TT
1290
1291 /* for write statistics */
1292 unsigned long s_sectors_written_start;
1293 u64 s_kbytes_written;
1294
67a5da56
ZL
1295 /* the size of zero-out chunk */
1296 unsigned int s_extent_max_zeroout_kb;
1297
ca0faba0
TT
1298 unsigned int s_log_groups_per_flex;
1299 struct flex_groups *s_flex_groups;
117fff10 1300 ext4_group_t s_flex_groups_allocated;
4c0425ff 1301
2e8fa54e
JK
1302 /* workqueue for reserved extent conversions (buffered io) */
1303 struct workqueue_struct *rsv_conversion_wq;
66e61a9e
TT
1304
1305 /* timer for periodic error stats printing */
1306 struct timer_list s_err_report;
bfff6873
LC
1307
1308 /* Lazy inode table initialization info */
1309 struct ext4_li_request *s_li_request;
1310 /* Wait multiplier for lazy initialization thread */
1311 unsigned int s_li_wait_mult;
c5e06d10
JL
1312
1313 /* Kernel thread for multiple mount protection */
1314 struct task_struct *s_mmp_tsk;
3d56b8d2
TM
1315
1316 /* record the last minlen when FITRIM is called. */
1317 atomic_t s_last_trim_minblks;
0441984a
DW
1318
1319 /* Reference to checksum algorithm driver via cryptoapi */
1320 struct crypto_shash *s_chksum_driver;
a9c47317
DW
1321
1322 /* Precomputed FS UUID checksum for seeding other checksums */
1323 __u32 s_csum_seed;
74cd15cd
ZL
1324
1325 /* Reclaim extents from extent status tree */
1326 struct shrinker s_es_shrinker;
1327 struct list_head s_es_lru;
eb68d0e2 1328 struct ext4_es_stats s_es_stats;
9c191f70 1329 struct mb_cache *s_mb_cache;
74cd15cd 1330 spinlock_t s_es_lru_lock ____cacheline_aligned_in_smp;
efbed4dc
TT
1331
1332 /* Ratelimit ext4 messages. */
1333 struct ratelimit_state s_err_ratelimit_state;
1334 struct ratelimit_state s_warning_ratelimit_state;
1335 struct ratelimit_state s_msg_ratelimit_state;
ca0faba0
TT
1336};
1337
af5bc92d 1338static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
ac27a0ec
DK
1339{
1340 return sb->s_fs_info;
1341}
617ba13b 1342static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
ac27a0ec 1343{
617ba13b 1344 return container_of(inode, struct ext4_inode_info, vfs_inode);
ac27a0ec
DK
1345}
1346
ef7f3835
KS
1347static inline struct timespec ext4_current_time(struct inode *inode)
1348{
1349 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1350 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1351}
1352
617ba13b 1353static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
ac27a0ec 1354{
617ba13b 1355 return ino == EXT4_ROOT_INO ||
7c319d32
AK
1356 ino == EXT4_USR_QUOTA_INO ||
1357 ino == EXT4_GRP_QUOTA_INO ||
393d1d1d 1358 ino == EXT4_BOOT_LOADER_INO ||
617ba13b
MC
1359 ino == EXT4_JOURNAL_INO ||
1360 ino == EXT4_RESIZE_INO ||
1361 (ino >= EXT4_FIRST_INO(sb) &&
1362 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
ac27a0ec 1363}
19f5fb7a 1364
0edeb71d
TM
1365static inline void ext4_set_io_unwritten_flag(struct inode *inode,
1366 struct ext4_io_end *io_end)
1367{
1368 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
1369 io_end->flag |= EXT4_IO_END_UNWRITTEN;
e27f41e1 1370 atomic_inc(&EXT4_I(inode)->i_unwritten);
0edeb71d
TM
1371 }
1372}
1373
f45ee3a1
DM
1374static inline ext4_io_end_t *ext4_inode_aio(struct inode *inode)
1375{
1376 return inode->i_private;
1377}
1378
1379static inline void ext4_inode_aio_set(struct inode *inode, ext4_io_end_t *io)
1380{
1381 inode->i_private = io;
1382}
1383
19f5fb7a
TT
1384/*
1385 * Inode dynamic state flags
1386 */
1387enum {
1388 EXT4_STATE_JDATA, /* journaled data exists */
1389 EXT4_STATE_NEW, /* inode is newly created */
1390 EXT4_STATE_XATTR, /* has in-inode xattrs */
1391 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1392 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1393 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1394 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
14ece102 1395 EXT4_STATE_NEWENTRY, /* File just added to dir */
17335dcc
DM
1396 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read
1397 nolocking */
67cf5b09 1398 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
74d553aa 1399 EXT4_STATE_ORDERED_MODE, /* data=ordered mode */
7869a4a6 1400 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
19f5fb7a
TT
1401};
1402
353eb83c 1403#define EXT4_INODE_BIT_FNS(name, field, offset) \
12e9b892
DM
1404static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1405{ \
353eb83c 1406 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
12e9b892
DM
1407} \
1408static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1409{ \
353eb83c 1410 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
12e9b892
DM
1411} \
1412static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1413{ \
353eb83c 1414 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
19f5fb7a
TT
1415}
1416
3f61c0cc
AD
1417/* Add these declarations here only so that these functions can be
1418 * found by name. Otherwise, they are very hard to locate. */
1419static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1420static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1421static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
353eb83c 1422EXT4_INODE_BIT_FNS(flag, flags, 0)
3f61c0cc
AD
1423
1424/* Add these declarations here only so that these functions can be
1425 * found by name. Otherwise, they are very hard to locate. */
1426static inline int ext4_test_inode_state(struct inode *inode, int bit);
1427static inline void ext4_set_inode_state(struct inode *inode, int bit);
1428static inline void ext4_clear_inode_state(struct inode *inode, int bit);
353eb83c
TT
1429#if (BITS_PER_LONG < 64)
1430EXT4_INODE_BIT_FNS(state, state_flags, 0)
1431
1432static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1433{
1434 (ei)->i_state_flags = 0;
1435}
1436#else
1437EXT4_INODE_BIT_FNS(state, flags, 32)
1438
1439static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1440{
1441 /* We depend on the fact that callers will set i_flags */
1442}
1443#endif
ac27a0ec 1444#else
617ba13b 1445/* Assume that user mode programs are passing in an ext4fs superblock, not
ac27a0ec
DK
1446 * a kernel struct super_block. This will allow us to call the feature-test
1447 * macros from user land. */
617ba13b 1448#define EXT4_SB(sb) (sb)
ac27a0ec
DK
1449#endif
1450
617ba13b 1451#define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
ac27a0ec
DK
1452
1453/*
1454 * Codes for operating systems
1455 */
617ba13b
MC
1456#define EXT4_OS_LINUX 0
1457#define EXT4_OS_HURD 1
1458#define EXT4_OS_MASIX 2
1459#define EXT4_OS_FREEBSD 3
1460#define EXT4_OS_LITES 4
ac27a0ec
DK
1461
1462/*
1463 * Revision levels
1464 */
617ba13b
MC
1465#define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1466#define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
ac27a0ec 1467
617ba13b
MC
1468#define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1469#define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
ac27a0ec 1470
617ba13b 1471#define EXT4_GOOD_OLD_INODE_SIZE 128
ac27a0ec
DK
1472
1473/*
1474 * Feature set definitions
1475 */
1476
617ba13b 1477#define EXT4_HAS_COMPAT_FEATURE(sb,mask) \
3a06d778 1478 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
617ba13b 1479#define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \
3a06d778 1480 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
617ba13b 1481#define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \
3a06d778 1482 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
617ba13b
MC
1483#define EXT4_SET_COMPAT_FEATURE(sb,mask) \
1484 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1485#define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \
1486 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1487#define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \
1488 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1489#define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \
1490 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1491#define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
1492 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1493#define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \
1494 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1495
1496#define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1497#define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1498#define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1499#define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1500#define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1501#define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1beeef1b 1502#define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
617ba13b
MC
1503
1504#define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1505#define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1506#define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
0fc1b451 1507#define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
717d50e4 1508#define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
f8628a14 1509#define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
ef7f3835 1510#define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
ae812306 1511#define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
281b5995 1512#define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
e6153918
DW
1513/*
1514 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1515 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1516 * all other data structures' checksums. However, the METADATA_CSUM and
1517 * GDT_CSUM bits are mutually exclusive.
1518 */
9b90e5e0 1519#define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
617ba13b
MC
1520
1521#define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1522#define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1523#define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1524#define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1525#define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
a86c6181 1526#define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
bd81d8ee 1527#define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
0fc1b451 1528#define EXT4_FEATURE_INCOMPAT_MMP 0x0100
ce421581 1529#define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
f710b4b9
TT
1530#define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1531#define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
856cbcf9 1532#define EXT4_FEATURE_INCOMPAT_BG_USE_META_CSUM 0x2000 /* use crc32c for bg */
9b90e5e0 1533#define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
67cf5b09 1534#define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
617ba13b 1535
2035e776
TT
1536#define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1537#define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1538 EXT4_FEATURE_INCOMPAT_META_BG)
1539#define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1540 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1541 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1542
1543#define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1544#define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1545 EXT4_FEATURE_INCOMPAT_RECOVER| \
1546 EXT4_FEATURE_INCOMPAT_META_BG)
1547#define EXT3_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
617ba13b
MC
1551#define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR
1552#define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1553 EXT4_FEATURE_INCOMPAT_RECOVER| \
a86c6181 1554 EXT4_FEATURE_INCOMPAT_META_BG| \
bd81d8ee 1555 EXT4_FEATURE_INCOMPAT_EXTENTS| \
ce421581 1556 EXT4_FEATURE_INCOMPAT_64BIT| \
c5e06d10 1557 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
f08225d1
TM
1558 EXT4_FEATURE_INCOMPAT_MMP | \
1559 EXT4_FEATURE_INCOMPAT_INLINE_DATA)
617ba13b
MC
1560#define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1561 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
717d50e4 1562 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
f8628a14 1563 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
ef7f3835 1564 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
0fc1b451 1565 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
6f16b606 1566 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
e93376c2 1567 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
7c319d32
AK
1568 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1569 EXT4_FEATURE_RO_COMPAT_QUOTA)
ac27a0ec
DK
1570
1571/*
1572 * Default values for user and/or group using reserved blocks
1573 */
617ba13b
MC
1574#define EXT4_DEF_RESUID 0
1575#define EXT4_DEF_RESGID 0
ac27a0ec 1576
240799cd
TT
1577#define EXT4_DEF_INODE_READAHEAD_BLKS 32
1578
ac27a0ec
DK
1579/*
1580 * Default mount options
1581 */
617ba13b
MC
1582#define EXT4_DEFM_DEBUG 0x0001
1583#define EXT4_DEFM_BSDGROUPS 0x0002
1584#define EXT4_DEFM_XATTR_USER 0x0004
1585#define EXT4_DEFM_ACL 0x0008
1586#define EXT4_DEFM_UID16 0x0010
1587#define EXT4_DEFM_JMODE 0x0060
1588#define EXT4_DEFM_JMODE_DATA 0x0020
1589#define EXT4_DEFM_JMODE_ORDERED 0x0040
1590#define EXT4_DEFM_JMODE_WBACK 0x0060
8b67f04a
TT
1591#define EXT4_DEFM_NOBARRIER 0x0100
1592#define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1593#define EXT4_DEFM_DISCARD 0x0400
1594#define EXT4_DEFM_NODELALLOC 0x0800
ac27a0ec 1595
30773840
TT
1596/*
1597 * Default journal batch times
1598 */
1599#define EXT4_DEF_MIN_BATCH_TIME 0
1600#define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1601
a4912123
TT
1602/*
1603 * Minimum number of groups in a flexgroup before we separate out
1604 * directories into the first block group of a flexgroup
1605 */
1606#define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1607
ac27a0ec
DK
1608/*
1609 * Structure of a directory entry
1610 */
617ba13b 1611#define EXT4_NAME_LEN 255
ac27a0ec 1612
617ba13b 1613struct ext4_dir_entry {
ac27a0ec
DK
1614 __le32 inode; /* Inode number */
1615 __le16 rec_len; /* Directory entry length */
1616 __le16 name_len; /* Name length */
617ba13b 1617 char name[EXT4_NAME_LEN]; /* File name */
ac27a0ec
DK
1618};
1619
1620/*
617ba13b 1621 * The new version of the directory entry. Since EXT4 structures are
ac27a0ec
DK
1622 * stored in intel byte order, and the name_len field could never be
1623 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1624 * file_type field.
1625 */
617ba13b 1626struct ext4_dir_entry_2 {
ac27a0ec
DK
1627 __le32 inode; /* Inode number */
1628 __le16 rec_len; /* Directory entry length */
1629 __u8 name_len; /* Name length */
1630 __u8 file_type;
617ba13b 1631 char name[EXT4_NAME_LEN]; /* File name */
ac27a0ec
DK
1632};
1633
e6153918
DW
1634/*
1635 * This is a bogus directory entry at the end of each leaf block that
1636 * records checksums.
1637 */
1638struct ext4_dir_entry_tail {
1639 __le32 det_reserved_zero1; /* Pretend to be unused */
1640 __le16 det_rec_len; /* 12 */
1641 __u8 det_reserved_zero2; /* Zero name length */
1642 __u8 det_reserved_ft; /* 0xDE, fake file type */
1643 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
1644};
1645
3c47d541
TM
1646#define EXT4_DIRENT_TAIL(block, blocksize) \
1647 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1648 ((blocksize) - \
1649 sizeof(struct ext4_dir_entry_tail))))
1650
ac27a0ec 1651/*
617ba13b 1652 * Ext4 directory file types. Only the low 3 bits are used. The
ac27a0ec
DK
1653 * other bits are reserved for now.
1654 */
617ba13b
MC
1655#define EXT4_FT_UNKNOWN 0
1656#define EXT4_FT_REG_FILE 1
1657#define EXT4_FT_DIR 2
1658#define EXT4_FT_CHRDEV 3
1659#define EXT4_FT_BLKDEV 4
1660#define EXT4_FT_FIFO 5
1661#define EXT4_FT_SOCK 6
1662#define EXT4_FT_SYMLINK 7
ac27a0ec 1663
617ba13b 1664#define EXT4_FT_MAX 8
ac27a0ec 1665
e6153918
DW
1666#define EXT4_FT_DIR_CSUM 0xDE
1667
ac27a0ec 1668/*
617ba13b 1669 * EXT4_DIR_PAD defines the directory entries boundaries
ac27a0ec
DK
1670 *
1671 * NOTE: It must be a multiple of 4
1672 */
617ba13b
MC
1673#define EXT4_DIR_PAD 4
1674#define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1675#define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1676 ~EXT4_DIR_ROUND)
a72d7f83
JK
1677#define EXT4_MAX_REC_LEN ((1<<16)-1)
1678
0cfc9255
ES
1679/*
1680 * If we ever get support for fs block sizes > page_size, we'll need
1681 * to remove the #if statements in the next two functions...
1682 */
1683static inline unsigned int
1684ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1685{
1686 unsigned len = le16_to_cpu(dlen);
1687
1688#if (PAGE_CACHE_SIZE >= 65536)
1689 if (len == EXT4_MAX_REC_LEN || len == 0)
1690 return blocksize;
1691 return (len & 65532) | ((len & 3) << 16);
1692#else
1693 return len;
1694#endif
1695}
1696
1697static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1698{
1699 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1700 BUG();
1701#if (PAGE_CACHE_SIZE >= 65536)
1702 if (len < 65536)
1703 return cpu_to_le16(len);
1704 if (len == blocksize) {
1705 if (blocksize == 65536)
1706 return cpu_to_le16(EXT4_MAX_REC_LEN);
1707 else
1708 return cpu_to_le16(0);
1709 }
1710 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1711#else
1712 return cpu_to_le16(len);
1713#endif
1714}
1715
ac27a0ec
DK
1716/*
1717 * Hash Tree Directory indexing
1718 * (c) Daniel Phillips, 2001
1719 */
1720
4074fe37
ES
1721#define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1722 EXT4_FEATURE_COMPAT_DIR_INDEX) && \
12e9b892 1723 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
617ba13b
MC
1724#define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1725#define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
ac27a0ec
DK
1726
1727/* Legal values for the dx_root hash_version field: */
1728
1729#define DX_HASH_LEGACY 0
1730#define DX_HASH_HALF_MD4 1
1731#define DX_HASH_TEA 2
f99b2589
TT
1732#define DX_HASH_LEGACY_UNSIGNED 3
1733#define DX_HASH_HALF_MD4_UNSIGNED 4
1734#define DX_HASH_TEA_UNSIGNED 5
ac27a0ec 1735
0441984a
DW
1736static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
1737 const void *address, unsigned int length)
1738{
1739 struct {
1740 struct shash_desc shash;
3108b54b 1741 char ctx[4];
0441984a
DW
1742 } desc;
1743 int err;
1744
3108b54b
TT
1745 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
1746
0441984a
DW
1747 desc.shash.tfm = sbi->s_chksum_driver;
1748 desc.shash.flags = 0;
1749 *(u32 *)desc.ctx = crc;
1750
1751 err = crypto_shash_update(&desc.shash, address, length);
1752 BUG_ON(err);
1753
1754 return *(u32 *)desc.ctx;
1755}
1756
ac27a0ec
DK
1757#ifdef __KERNEL__
1758
1759/* hash info structure used by the directory hash */
1760struct dx_hash_info
1761{
1762 u32 hash;
1763 u32 minor_hash;
1764 int hash_version;
1765 u32 *seed;
1766};
1767
d1f5273e
FY
1768
1769/* 32 and 64 bit signed EOF for dx directories */
1770#define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
1771#define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
1772
ac27a0ec
DK
1773
1774/*
617ba13b 1775 * Control parameters used by ext4_htree_next_block
ac27a0ec
DK
1776 */
1777#define HASH_NB_ALWAYS 1
1778
1779
1780/*
1781 * Describe an inode's exact location on disk and in memory
1782 */
617ba13b 1783struct ext4_iloc
ac27a0ec
DK
1784{
1785 struct buffer_head *bh;
1786 unsigned long offset;
fd2d4291 1787 ext4_group_t block_group;
ac27a0ec
DK
1788};
1789
617ba13b 1790static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
ac27a0ec 1791{
617ba13b 1792 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
ac27a0ec
DK
1793}
1794
1795/*
1796 * This structure is stuffed into the struct file's private_data field
1797 * for directories. It is where we put information so that we can do
1798 * readdir operations in hash tree order.
1799 */
1800struct dir_private_info {
1801 struct rb_root root;
1802 struct rb_node *curr_node;
1803 struct fname *extra_fname;
1804 loff_t last_pos;
1805 __u32 curr_hash;
1806 __u32 curr_minor_hash;
1807 __u32 next_hash;
1808};
1809
1810/* calculate the first block number of the group */
617ba13b 1811static inline ext4_fsblk_t
fd2d4291 1812ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
ac27a0ec 1813{
617ba13b
MC
1814 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
1815 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
ac27a0ec
DK
1816}
1817
1818/*
1819 * Special error return code only used by dx_probe() and its callers.
1820 */
36de9286 1821#define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
ac27a0ec 1822
bfff6873
LC
1823/*
1824 * Timeout and state flag for lazy initialization inode thread.
1825 */
1826#define EXT4_DEF_LI_WAIT_MULT 10
1827#define EXT4_DEF_LI_MAX_START_DELAY 5
1828#define EXT4_LAZYINIT_QUIT 0x0001
1829#define EXT4_LAZYINIT_RUNNING 0x0002
1830
1831/*
1832 * Lazy inode table initialization info
1833 */
1834struct ext4_lazy_init {
1835 unsigned long li_state;
bfff6873
LC
1836 struct list_head li_request_list;
1837 struct mutex li_list_mtx;
1838};
1839
1840struct ext4_li_request {
1841 struct super_block *lr_super;
1842 struct ext4_sb_info *lr_sbi;
1843 ext4_group_t lr_next_group;
1844 struct list_head lr_request;
1845 unsigned long lr_next_sched;
1846 unsigned long lr_timeout;
1847};
1848
857ac889
LC
1849struct ext4_features {
1850 struct kobject f_kobj;
1851 struct completion f_kobj_unregister;
1852};
1853
c5e06d10
JL
1854/*
1855 * This structure will be used for multiple mount protection. It will be
1856 * written into the block number saved in the s_mmp_block field in the
1857 * superblock. Programs that check MMP should assume that if
1858 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
1859 * to use the filesystem, regardless of how old the timestamp is.
1860 */
1861#define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
1862#define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
1863#define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
1864#define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
1865
1866struct mmp_struct {
1867 __le32 mmp_magic; /* Magic number for MMP */
1868 __le32 mmp_seq; /* Sequence no. updated periodically */
1869
1870 /*
1871 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
1872 * purposes and do not affect the correctness of the algorithm
1873 */
1874 __le64 mmp_time; /* Time last updated */
1875 char mmp_nodename[64]; /* Node which last updated MMP block */
1876 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
1877
1878 /*
1879 * mmp_check_interval is used to verify if the MMP block has been
1880 * updated on the block device. The value is updated based on the
1881 * maximum time to write the MMP block during an update cycle.
1882 */
1883 __le16 mmp_check_interval;
1884
1885 __le16 mmp_pad1;
e6153918
DW
1886 __le32 mmp_pad2[226];
1887 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
c5e06d10
JL
1888};
1889
1890/* arguments passed to the mmp thread */
1891struct mmpd_data {
1892 struct buffer_head *bh; /* bh from initial read_mmp_block() */
1893 struct super_block *sb; /* super block of the fs */
1894};
1895
1896/*
1897 * Check interval multiplier
1898 * The MMP block is written every update interval and initially checked every
1899 * update interval x the multiplier (the value is then adapted based on the
1900 * write latency). The reason is that writes can be delayed under load and we
1901 * don't want readers to incorrectly assume that the filesystem is no longer
1902 * in use.
1903 */
1904#define EXT4_MMP_CHECK_MULT 2UL
1905
1906/*
1907 * Minimum interval for MMP checking in seconds.
1908 */
1909#define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
1910
1911/*
1912 * Maximum interval for MMP checking in seconds.
1913 */
1914#define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
1915
ac27a0ec
DK
1916/*
1917 * Function prototypes
1918 */
1919
1920/*
1921 * Ok, these declarations are also in <linux/kernel.h> but none of the
617ba13b 1922 * ext4 source programs needs to include it so they are duplicated here.
ac27a0ec 1923 */
8c55e204
DK
1924# define NORET_TYPE /**/
1925# define ATTRIB_NORET __attribute__((noreturn))
1926# define NORET_AND noreturn,
ac27a0ec 1927
498e5f24 1928/* bitmap.c */
f6fb99ca 1929extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
41a246d1
DW
1930void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
1931 struct ext4_group_desc *gdp,
1932 struct buffer_head *bh, int sz);
1933int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
1934 struct ext4_group_desc *gdp,
1935 struct buffer_head *bh, int sz);
fa77dcfa
DW
1936void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
1937 struct ext4_group_desc *gdp,
79f1ba49 1938 struct buffer_head *bh);
fa77dcfa
DW
1939int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
1940 struct ext4_group_desc *gdp,
79f1ba49 1941 struct buffer_head *bh);
498e5f24 1942
ac27a0ec 1943/* balloc.c */
bd86298e
LC
1944extern void ext4_get_group_no_and_offset(struct super_block *sb,
1945 ext4_fsblk_t blocknr,
1946 ext4_group_t *blockgrpp,
1947 ext4_grpblk_t *offsetp);
1948extern ext4_group_t ext4_get_group_number(struct super_block *sb,
1949 ext4_fsblk_t block);
1950
3a5b2ecd
MC
1951extern unsigned int ext4_block_group(struct super_block *sb,
1952 ext4_fsblk_t blocknr);
1953extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
1954 ext4_fsblk_t blocknr);
fd2d4291
AM
1955extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
1956extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
1957 ext4_group_t group);
7061eba7 1958extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
55f020db
AH
1959 ext4_fsblk_t goal,
1960 unsigned int flags,
1961 unsigned long *count,
1962 int *errp);
e7d5f315
TT
1963extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
1964 s64 nclusters, unsigned int flags);
5dee5437 1965extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
af5bc92d 1966extern void ext4_check_blocks_bitmap(struct super_block *);
617ba13b 1967extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
fd2d4291 1968 ext4_group_t block_group,
ac27a0ec 1969 struct buffer_head ** bh);
617ba13b 1970extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
813e5727
TT
1971
1972extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
1973 ext4_group_t block_group);
1974extern int ext4_wait_block_bitmap(struct super_block *sb,
1975 ext4_group_t block_group,
1976 struct buffer_head *bh);
1977extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
1978 ext4_group_t block_group);
cff1dfd7
TT
1979extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
1980 ext4_group_t block_group,
1981 struct ext4_group_desc *gdp);
f86186b4 1982ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
ac27a0ec
DK
1983
1984/* dir.c */
60fd4da3 1985extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
f7c21177 1986 struct file *,
60fd4da3 1987 struct ext4_dir_entry_2 *,
226ba972
TM
1988 struct buffer_head *, char *, int,
1989 unsigned int);
1990#define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
f7c21177 1991 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
226ba972 1992 (de), (bh), (buf), (size), (offset)))
617ba13b 1993extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
ac27a0ec 1994 __u32 minor_hash,
617ba13b
MC
1995 struct ext4_dir_entry_2 *dirent);
1996extern void ext4_htree_free_dir_info(struct dir_private_info *p);
978fef91
TM
1997extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
1998 struct buffer_head *bh,
1999 void *buf, int buf_size,
2000 const char *name, int namelen,
2001 struct ext4_dir_entry_2 **dest_de);
2002void ext4_insert_dentry(struct inode *inode,
2003 struct ext4_dir_entry_2 *de,
2004 int buf_size,
2005 const char *name, int namelen);
2006static inline void ext4_update_dx_flag(struct inode *inode)
2007{
2008 if (!EXT4_HAS_COMPAT_FEATURE(inode->i_sb,
2009 EXT4_FEATURE_COMPAT_DIR_INDEX))
2010 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2011}
65d165d9
TM
2012static unsigned char ext4_filetype_table[] = {
2013 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2014};
2015
2016static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2017{
2018 if (!EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE) ||
2019 (filetype >= EXT4_FT_MAX))
2020 return DT_UNKNOWN;
2021
2022 return ext4_filetype_table[filetype];
2023}
40b163f1
DW
2024extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2025 void *buf, int buf_size);
ac27a0ec
DK
2026
2027/* fsync.c */
02c24a82 2028extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
ac27a0ec
DK
2029
2030/* hash.c */
617ba13b 2031extern int ext4fs_dirhash(const char *name, int len, struct
ac27a0ec
DK
2032 dx_hash_info *hinfo);
2033
2034/* ialloc.c */
1139575a
TT
2035extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2036 const struct qstr *qstr, __u32 goal,
2037 uid_t *owner, int handle_type,
2038 unsigned int line_no, int nblocks);
2039
2040#define ext4_new_inode(handle, dir, mode, qstr, goal, owner) \
2041 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2042 0, 0, 0)
2043#define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2044 type, nblocks) \
2045 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2046 (type), __LINE__, (nblocks))
2047
2048
af5bc92d
TT
2049extern void ext4_free_inode(handle_t *, struct inode *);
2050extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2051extern unsigned long ext4_count_free_inodes(struct super_block *);
2052extern unsigned long ext4_count_dirs(struct super_block *);
2053extern void ext4_check_inodes_bitmap(struct super_block *);
61d08673 2054extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
bfff6873
LC
2055extern int ext4_init_inode_table(struct super_block *sb,
2056 ext4_group_t group, int barrier);
813e5727 2057extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
ac27a0ec 2058
c9de560d
AT
2059/* mballoc.c */
2060extern long ext4_mb_stats;
2061extern long ext4_mb_max_to_scan;
9d99012f 2062extern int ext4_mb_init(struct super_block *);
c9de560d
AT
2063extern int ext4_mb_release(struct super_block *);
2064extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2065 struct ext4_allocation_request *, int *);
2066extern int ext4_mb_reserve_blocks(struct super_block *, int);
c2ea3fde 2067extern void ext4_discard_preallocations(struct inode *);
5dabfc78
TT
2068extern int __init ext4_init_mballoc(void);
2069extern void ext4_exit_mballoc(void);
44338711 2070extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
e6362609
TT
2071 struct buffer_head *bh, ext4_fsblk_t block,
2072 unsigned long count, int flags);
28623c2f
TT
2073extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2074 ext4_group_t ngroups);
920313a7 2075extern int ext4_mb_add_groupinfo(struct super_block *sb,
5f21b0e6 2076 ext4_group_t i, struct ext4_group_desc *desc);
cc7365df 2077extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2846e820 2078 ext4_fsblk_t block, unsigned long count);
7360d173
LC
2079extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2080
ac27a0ec 2081/* inode.c */
10560082 2082struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
1c215028 2083struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
f19d5870
TM
2084int ext4_get_block_write(struct inode *inode, sector_t iblock,
2085 struct buffer_head *bh_result, int create);
6873fa0d
ES
2086int ext4_get_block(struct inode *inode, sector_t iblock,
2087 struct buffer_head *bh_result, int create);
9c3569b5
TM
2088int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2089 struct buffer_head *bh, int create);
f19d5870
TM
2090int ext4_walk_page_buffers(handle_t *handle,
2091 struct buffer_head *head,
2092 unsigned from,
2093 unsigned to,
2094 int *partial,
2095 int (*fn)(handle_t *handle,
2096 struct buffer_head *bh));
2097int do_journal_get_write_access(handle_t *handle,
2098 struct buffer_head *bh);
9c3569b5
TM
2099#define FALL_BACK_TO_NONDELALLOC 1
2100#define CONVERT_INLINE_DATA 2
ac27a0ec 2101
1d1fe1ee 2102extern struct inode *ext4_iget(struct super_block *, unsigned long);
f4bb2981 2103extern struct inode *ext4_iget_normal(struct super_block *, unsigned long);
a9185b41 2104extern int ext4_write_inode(struct inode *, struct writeback_control *);
af5bc92d 2105extern int ext4_setattr(struct dentry *, struct iattr *);
3e3398a0
MC
2106extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
2107 struct kstat *stat);
0930fcc1
AV
2108extern void ext4_evict_inode(struct inode *);
2109extern void ext4_clear_inode(struct inode *);
af5bc92d 2110extern int ext4_sync_inode(handle_t *, struct inode *);
aa385729 2111extern void ext4_dirty_inode(struct inode *, int);
617ba13b
MC
2112extern int ext4_change_inode_journal_flag(struct inode *, int);
2113extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
a361293f 2114extern int ext4_inode_attach_jinode(struct inode *inode);
91ef4caf 2115extern int ext4_can_truncate(struct inode *inode);
af5bc92d 2116extern void ext4_truncate(struct inode *);
aeb2817a 2117extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
487caeef 2118extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
617ba13b 2119extern void ext4_set_inode_flags(struct inode *);
ff9ddf7e 2120extern void ext4_get_inode_flags(struct ext4_inode_info *);
ccd2506b 2121extern int ext4_alloc_da_blocks(struct inode *inode);
617ba13b 2122extern void ext4_set_aops(struct inode *inode);
a86c6181 2123extern int ext4_writepage_trans_blocks(struct inode *);
f3bd1f3f 2124extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
a87dd18c
LC
2125extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2126 loff_t lstart, loff_t lend);
c2ec175c 2127extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
a9e7f447 2128extern qsize_t *ext4_get_reserved_space(struct inode *inode);
5f634d06
AK
2129extern void ext4_da_update_reserve_space(struct inode *inode,
2130 int used, int quota_claim);
dae1e52c
AG
2131
2132/* indirect.c */
2133extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2134 struct ext4_map_blocks *map, int flags);
2135extern ssize_t ext4_ind_direct_IO(int rw, struct kiocb *iocb,
16b1f05d 2136 struct iov_iter *iter, loff_t offset);
dae1e52c 2137extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
fa55a0ed 2138extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
819c4920 2139extern void ext4_ind_truncate(handle_t *, struct inode *inode);
4f579ae7
LC
2140extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2141 ext4_lblk_t start, ext4_lblk_t end);
ff9893dc 2142
ac27a0ec 2143/* ioctl.c */
5cdd7b2d 2144extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
af5bc92d 2145extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
ac27a0ec 2146
c14c6fd5 2147/* migrate.c */
2a43a878 2148extern int ext4_ext_migrate(struct inode *);
0d14b098 2149extern int ext4_ind_migrate(struct inode *inode);
3d0518f4 2150
ac27a0ec 2151/* namei.c */
b0336e8d
DW
2152extern int ext4_dirent_csum_verify(struct inode *inode,
2153 struct ext4_dir_entry *dirent);
617ba13b
MC
2154extern int ext4_orphan_add(handle_t *, struct inode *);
2155extern int ext4_orphan_del(handle_t *, struct inode *);
2156extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
ac27a0ec 2157 __u32 start_minor_hash, __u32 *next_hash);
7335cd3b
TM
2158extern int search_dir(struct buffer_head *bh,
2159 char *search_buf,
2160 int buf_size,
2161 struct inode *dir,
2162 const struct qstr *d_name,
2163 unsigned int offset,
2164 struct ext4_dir_entry_2 **res_dir);
05019a9e
TM
2165extern int ext4_generic_delete_entry(handle_t *handle,
2166 struct inode *dir,
2167 struct ext4_dir_entry_2 *de_del,
2168 struct buffer_head *bh,
2169 void *entry_buf,
2170 int buf_size,
2171 int csum_size);
ac27a0ec
DK
2172
2173/* resize.c */
617ba13b
MC
2174extern int ext4_group_add(struct super_block *sb,
2175 struct ext4_new_group_data *input);
2176extern int ext4_group_extend(struct super_block *sb,
2177 struct ext4_super_block *es,
2178 ext4_fsblk_t n_blocks_count);
19c5246d 2179extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
ac27a0ec
DK
2180
2181/* super.c */
952fc18e 2182extern int ext4_calculate_overhead(struct super_block *sb);
06db49e6 2183extern void ext4_superblock_csum_set(struct super_block *sb);
9933fc0a
TT
2184extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2185extern void *ext4_kvzalloc(size_t size, gfp_t flags);
117fff10
TT
2186extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2187 ext4_group_t ngroup);
722887dd
TT
2188extern const char *ext4_decode_error(struct super_block *sb, int errno,
2189 char nbuf[16]);
e7c96e8e 2190
b9075fa9
JP
2191extern __printf(4, 5)
2192void __ext4_error(struct super_block *, const char *, unsigned int,
2193 const char *, ...);
b9075fa9 2194extern __printf(5, 6)
e7c96e8e 2195void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
b9075fa9
JP
2196 const char *, ...);
2197extern __printf(5, 6)
e7c96e8e 2198void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
b9075fa9 2199 const char *, ...);
c398eda0
TT
2200extern void __ext4_std_error(struct super_block *, const char *,
2201 unsigned int, int);
b9075fa9
JP
2202extern __printf(4, 5)
2203void __ext4_abort(struct super_block *, const char *, unsigned int,
2204 const char *, ...);
b9075fa9
JP
2205extern __printf(4, 5)
2206void __ext4_warning(struct super_block *, const char *, unsigned int,
2207 const char *, ...);
b9075fa9 2208extern __printf(3, 4)
e7c96e8e 2209void __ext4_msg(struct super_block *, const char *, const char *, ...);
c5e06d10
JL
2210extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2211 const char *, unsigned int, const char *);
b9075fa9
JP
2212extern __printf(7, 8)
2213void __ext4_grp_locked_error(const char *, unsigned int,
2214 struct super_block *, ext4_group_t,
2215 unsigned long, ext4_fsblk_t,
2216 const char *, ...);
e7c96e8e
JP
2217
2218#ifdef CONFIG_PRINTK
2219
2220#define ext4_error_inode(inode, func, line, block, fmt, ...) \
2221 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__)
2222#define ext4_error_file(file, func, line, block, fmt, ...) \
2223 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2224#define ext4_error(sb, fmt, ...) \
2225 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2226#define ext4_abort(sb, fmt, ...) \
2227 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2228#define ext4_warning(sb, fmt, ...) \
2229 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2230#define ext4_msg(sb, level, fmt, ...) \
2231 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2232#define dump_mmp_msg(sb, mmp, msg) \
2233 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2234#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2235 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2236 fmt, ##__VA_ARGS__)
2237
2238#else
2239
2240#define ext4_error_inode(inode, func, line, block, fmt, ...) \
2241do { \
2242 no_printk(fmt, ##__VA_ARGS__); \
2243 __ext4_error_inode(inode, "", 0, block, " "); \
2244} while (0)
2245#define ext4_error_file(file, func, line, block, fmt, ...) \
2246do { \
2247 no_printk(fmt, ##__VA_ARGS__); \
2248 __ext4_error_file(file, "", 0, block, " "); \
2249} while (0)
2250#define ext4_error(sb, fmt, ...) \
2251do { \
2252 no_printk(fmt, ##__VA_ARGS__); \
2253 __ext4_error(sb, "", 0, " "); \
2254} while (0)
2255#define ext4_abort(sb, fmt, ...) \
2256do { \
2257 no_printk(fmt, ##__VA_ARGS__); \
2258 __ext4_abort(sb, "", 0, " "); \
2259} while (0)
2260#define ext4_warning(sb, fmt, ...) \
2261do { \
2262 no_printk(fmt, ##__VA_ARGS__); \
2263 __ext4_warning(sb, "", 0, " "); \
2264} while (0)
2265#define ext4_msg(sb, level, fmt, ...) \
2266do { \
2267 no_printk(fmt, ##__VA_ARGS__); \
2268 __ext4_msg(sb, "", " "); \
2269} while (0)
2270#define dump_mmp_msg(sb, mmp, msg) \
2271 __dump_mmp_msg(sb, mmp, "", 0, "")
2272#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2273do { \
2274 no_printk(fmt, ##__VA_ARGS__); \
2275 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
2276} while (0)
2277
2278#endif
2279
af5bc92d 2280extern void ext4_update_dynamic_rev(struct super_block *sb);
99e6f829
AK
2281extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2282 __u32 compat);
2283extern int ext4_update_rocompat_feature(handle_t *handle,
2284 struct super_block *sb, __u32 rocompat);
2285extern int ext4_update_incompat_feature(handle_t *handle,
2286 struct super_block *sb, __u32 incompat);
8fadc143
AR
2287extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2288 struct ext4_group_desc *bg);
2289extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2290 struct ext4_group_desc *bg);
2291extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2292 struct ext4_group_desc *bg);
021b65bb
TT
2293extern __u32 ext4_free_group_clusters(struct super_block *sb,
2294 struct ext4_group_desc *bg);
560671a0
AK
2295extern __u32 ext4_free_inodes_count(struct super_block *sb,
2296 struct ext4_group_desc *bg);
2297extern __u32 ext4_used_dirs_count(struct super_block *sb,
2298 struct ext4_group_desc *bg);
2299extern __u32 ext4_itable_unused_count(struct super_block *sb,
2300 struct ext4_group_desc *bg);
8fadc143
AR
2301extern void ext4_block_bitmap_set(struct super_block *sb,
2302 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2303extern void ext4_inode_bitmap_set(struct super_block *sb,
2304 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2305extern void ext4_inode_table_set(struct super_block *sb,
2306 struct ext4_group_desc *bg, ext4_fsblk_t blk);
021b65bb
TT
2307extern void ext4_free_group_clusters_set(struct super_block *sb,
2308 struct ext4_group_desc *bg,
2309 __u32 count);
560671a0
AK
2310extern void ext4_free_inodes_set(struct super_block *sb,
2311 struct ext4_group_desc *bg, __u32 count);
2312extern void ext4_used_dirs_set(struct super_block *sb,
2313 struct ext4_group_desc *bg, __u32 count);
2314extern void ext4_itable_unused_set(struct super_block *sb,
2315 struct ext4_group_desc *bg, __u32 count);
feb0ab32 2316extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
bb23c20a 2317 struct ext4_group_desc *gdp);
feb0ab32
DW
2318extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2319 struct ext4_group_desc *gdp);
7f511862
TT
2320extern int ext4_register_li_request(struct super_block *sb,
2321 ext4_group_t first_not_zeroed);
bd81d8ee 2322
41a246d1
DW
2323static inline int ext4_has_group_desc_csum(struct super_block *sb)
2324{
2325 return EXT4_HAS_RO_COMPAT_FEATURE(sb,
813d32f9
DW
2326 EXT4_FEATURE_RO_COMPAT_GDT_CSUM) ||
2327 (EXT4_SB(sb)->s_chksum_driver != NULL);
41a246d1 2328}
bd81d8ee 2329
9aa5d32b
DM
2330static inline int ext4_has_metadata_csum(struct super_block *sb)
2331{
2332 WARN_ON_ONCE(EXT4_HAS_RO_COMPAT_FEATURE(sb,
2333 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM) &&
2334 !EXT4_SB(sb)->s_chksum_driver);
2335
2336 return (EXT4_SB(sb)->s_chksum_driver != NULL);
2337}
bd81d8ee
LV
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 2734 struct ext4_extent *, int);
ed8a1a76
TT
2735extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
2736 struct ext4_ext_path **,
2737 int flags);
4a092d73
TT
2738extern void ext4_ext_drop_refs(struct ext4_ext_path *);
2739extern int ext4_ext_check_inode(struct inode *inode);
f7fec032
ZL
2740extern int ext4_find_delalloc_range(struct inode *inode,
2741 ext4_lblk_t lblk_start,
2742 ext4_lblk_t lblk_end);
4a092d73 2743extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk);
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
2ccb5fb9 2783/*
48fc7f7e 2784 * Add new method to test whether block and inode bitmaps are properly
2ccb5fb9
AK
2785 * initialized. With uninit_bg reading the block from disk is not enough
2786 * to mark the bitmap uptodate. We need to also zero-out the bitmap
2787 */
2788#define BH_BITMAP_UPTODATE BH_JBDPrivateStart
2789
2790static inline int bitmap_uptodate(struct buffer_head *bh)
2791{
2792 return (buffer_uptodate(bh) &&
2793 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
2794}
2795static inline void set_bitmap_uptodate(struct buffer_head *bh)
2796{
2797 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
2798}
2799
17335dcc
DM
2800/*
2801 * Disable DIO read nolock optimization, so new dioreaders will be forced
2802 * to grab i_mutex
2803 */
2804static inline void ext4_inode_block_unlocked_dio(struct inode *inode)
2805{
2806 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
2807 smp_mb();
2808}
2809static inline void ext4_inode_resume_unlocked_dio(struct inode *inode)
2810{
2811 smp_mb();
2812 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
2813}
2814
731eb1a0
AM
2815#define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
2816
e9e3bcec
ES
2817/* For ioend & aio unwritten conversion wait queues */
2818#define EXT4_WQ_HASH_SZ 37
2819#define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
2820 EXT4_WQ_HASH_SZ])
2821#define ext4_aio_mutex(v) (&ext4__aio_mutex[((unsigned long)(v)) %\
2822 EXT4_WQ_HASH_SZ])
2823extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
2824extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ];
2825
8f82f840
YY
2826#define EXT4_RESIZING 0
2827extern int ext4_resize_begin(struct super_block *sb);
2828extern void ext4_resize_end(struct super_block *sb);
2829
ac27a0ec
DK
2830#endif /* __KERNEL__ */
2831
3dcf5451 2832#endif /* _EXT4_H */