]> git.proxmox.com Git - mirror_ubuntu-eoan-kernel.git/blame - fs/ext4/ext4.h
ext4: export inline xattr functions
[mirror_ubuntu-eoan-kernel.git] / fs / ext4 / ext4.h
CommitLineData
ac27a0ec 1/*
3dcf5451 2 * ext4.h
ac27a0ec
DK
3 *
4 * Copyright (C) 1992, 1993, 1994, 1995
5 * Remy Card (card@masi.ibp.fr)
6 * Laboratoire MASI - Institut Blaise Pascal
7 * Universite Pierre et Marie Curie (Paris VI)
8 *
9 * from
10 *
11 * linux/include/linux/minix_fs.h
12 *
13 * Copyright (C) 1991, 1992 Linus Torvalds
14 */
15
3dcf5451
CH
16#ifndef _EXT4_H
17#define _EXT4_H
ac27a0ec
DK
18
19#include <linux/types.h>
3a5b2ecd 20#include <linux/blkdev.h>
ac27a0ec 21#include <linux/magic.h>
2ccb5fb9 22#include <linux/jbd2.h>
60e58e0f 23#include <linux/quota.h>
d444c3c3
TT
24#include <linux/rwsem.h>
25#include <linux/rbtree.h>
26#include <linux/seqlock.h>
27#include <linux/mutex.h>
ca0faba0
TT
28#include <linux/timer.h>
29#include <linux/wait.h>
30#include <linux/blockgroup_lock.h>
31#include <linux/percpu_counter.h>
0441984a 32#include <crypto/hash.h>
4d92dc0f
BH
33#ifdef __KERNEL__
34#include <linux/compat.h>
35#endif
c9de560d 36
ac27a0ec 37/*
8a35694e 38 * The fourth extended filesystem constants/structures
ac27a0ec
DK
39 */
40
41/*
617ba13b 42 * Define EXT4FS_DEBUG to produce debug messages
ac27a0ec 43 */
617ba13b 44#undef EXT4FS_DEBUG
ac27a0ec 45
ac27a0ec
DK
46/*
47 * Debug code
48 */
617ba13b
MC
49#ifdef EXT4FS_DEBUG
50#define ext4_debug(f, a...) \
ac27a0ec 51 do { \
af5bc92d 52 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
4db9c54a 53 __FILE__, __LINE__, __func__); \
af5bc92d 54 printk(KERN_DEBUG f, ## a); \
ac27a0ec
DK
55 } while (0)
56#else
ace36ad4 57#define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
ac27a0ec
DK
58#endif
59
4a092d73
TT
60/*
61 * Turn on EXT_DEBUG to get lots of info about extents operations.
62 */
63#define EXT_DEBUG__
64#ifdef EXT_DEBUG
65#define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__)
66#else
67#define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
68#endif
69
273df556 70#define EXT4_ERROR_INODE(inode, fmt, a...) \
c398eda0
TT
71 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
72
73#define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
74 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
273df556 75
f7c21177
TT
76#define EXT4_ERROR_FILE(file, block, fmt, a...) \
77 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
273df556 78
d444c3c3
TT
79/* data type for block offset of block group */
80typedef int ext4_grpblk_t;
81
82/* data type for filesystem-wide blocks number */
83typedef unsigned long long ext4_fsblk_t;
84
85/* data type for file logical block number */
86typedef __u32 ext4_lblk_t;
87
88/* data type for block group number */
89typedef unsigned int ext4_group_t;
90
296c355c 91/*
60e6679e 92 * Flags used in mballoc's allocation_context flags field.
296c355c
TT
93 *
94 * Also used to show what's going on for debugging purposes when the
95 * flag field is exported via the traceport interface
96 */
d444c3c3 97
c9de560d 98/* prefer goal again. length */
0ef90db9 99#define EXT4_MB_HINT_MERGE 0x0001
c9de560d 100/* blocks already reserved */
0ef90db9 101#define EXT4_MB_HINT_RESERVED 0x0002
c9de560d 102/* metadata is being allocated */
0ef90db9 103#define EXT4_MB_HINT_METADATA 0x0004
c9de560d 104/* first blocks in the file */
0ef90db9 105#define EXT4_MB_HINT_FIRST 0x0008
c9de560d 106/* search for the best chunk */
0ef90db9 107#define EXT4_MB_HINT_BEST 0x0010
c9de560d 108/* data is being allocated */
0ef90db9 109#define EXT4_MB_HINT_DATA 0x0020
c9de560d 110/* don't preallocate (for tails) */
0ef90db9 111#define EXT4_MB_HINT_NOPREALLOC 0x0040
c9de560d 112/* allocate for locality group */
0ef90db9 113#define EXT4_MB_HINT_GROUP_ALLOC 0x0080
c9de560d 114/* allocate goal blocks or none */
0ef90db9 115#define EXT4_MB_HINT_GOAL_ONLY 0x0100
c9de560d 116/* goal is meaningful */
0ef90db9 117#define EXT4_MB_HINT_TRY_GOAL 0x0200
d2a17637 118/* blocks already pre-reserved by delayed allocation */
0ef90db9 119#define EXT4_MB_DELALLOC_RESERVED 0x0400
4ba74d00
TT
120/* We are doing stream allocation */
121#define EXT4_MB_STREAM_ALLOC 0x0800
55f020db
AH
122/* Use reserved root blocks if needed */
123#define EXT4_MB_USE_ROOT_BLOCKS 0x1000
c9de560d
AT
124
125struct ext4_allocation_request {
126 /* target inode for block we're allocating */
127 struct inode *inode;
726447d8
ES
128 /* how many blocks we want to allocate */
129 unsigned int len;
c9de560d
AT
130 /* logical block in target inode */
131 ext4_lblk_t logical;
c9de560d
AT
132 /* the closest logical allocated block to the left */
133 ext4_lblk_t lleft;
c9de560d
AT
134 /* the closest logical allocated block to the right */
135 ext4_lblk_t lright;
726447d8
ES
136 /* phys. target (a hint) */
137 ext4_fsblk_t goal;
138 /* phys. block for the closest logical allocated block to the left */
139 ext4_fsblk_t pleft;
140 /* phys. block for the closest logical allocated block to the right */
c9de560d 141 ext4_fsblk_t pright;
c9de560d 142 /* flags. see above EXT4_MB_HINT_* */
498e5f24 143 unsigned int flags;
c9de560d
AT
144};
145
e35fd660
TT
146/*
147 * Logical to physical block mapping, used by ext4_map_blocks()
148 *
149 * This structure is used to pass requests into ext4_map_blocks() as
150 * well as to store the information returned by ext4_map_blocks(). It
151 * takes less room on the stack than a struct buffer_head.
152 */
153#define EXT4_MAP_NEW (1 << BH_New)
154#define EXT4_MAP_MAPPED (1 << BH_Mapped)
155#define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten)
156#define EXT4_MAP_BOUNDARY (1 << BH_Boundary)
157#define EXT4_MAP_UNINIT (1 << BH_Uninit)
7b415bf6
AK
158/* Sometimes (in the bigalloc case, from ext4_da_get_block_prep) the caller of
159 * ext4_map_blocks wants to know whether or not the underlying cluster has
160 * already been accounted for. EXT4_MAP_FROM_CLUSTER conveys to the caller that
161 * the requested mapping was from previously mapped (or delayed allocated)
162 * cluster. We use BH_AllocFromCluster only for this flag. BH_AllocFromCluster
163 * should never appear on buffer_head's state flags.
164 */
165#define EXT4_MAP_FROM_CLUSTER (1 << BH_AllocFromCluster)
e35fd660
TT
166#define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
167 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\
7b415bf6 168 EXT4_MAP_UNINIT | EXT4_MAP_FROM_CLUSTER)
e35fd660
TT
169
170struct ext4_map_blocks {
171 ext4_fsblk_t m_pblk;
172 ext4_lblk_t m_lblk;
173 unsigned int m_len;
174 unsigned int m_flags;
175};
176
b3a3ca8c
TT
177/*
178 * For delayed allocation tracking
179 */
180struct mpage_da_data {
181 struct inode *inode;
182 sector_t b_blocknr; /* start block number of extent */
183 size_t b_size; /* size of extent */
184 unsigned long b_state; /* state of the extent */
185 unsigned long first_page, next_page; /* extent of pages */
186 struct writeback_control *wbc;
187 int io_done;
188 int pages_written;
189 int retval;
190};
bd2d0210
TT
191
192/*
193 * Flags for ext4_io_end->flags
194 */
195#define EXT4_IO_END_UNWRITTEN 0x0001
196#define EXT4_IO_END_ERROR 0x0002
8c0bec21 197#define EXT4_IO_END_QUEUED 0x0004
266991b1 198#define EXT4_IO_END_DIRECT 0x0008
bd2d0210
TT
199
200struct ext4_io_page {
201 struct page *p_page;
83668e71 202 atomic_t p_count;
bd2d0210
TT
203};
204
205#define MAX_IO_PAGES 128
206
4188188b
CW
207/*
208 * For converting uninitialized extents on a work queue.
209 *
210 * 'page' is only used from the writepage() path; 'pages' is only used for
211 * buffered writes; they are used to keep page references until conversion
212 * takes place. For AIO/DIO, neither field is filled in.
213 */
0031462b 214typedef struct ext4_io_end {
5b3ff237 215 struct list_head list; /* per-file finished IO list */
0031462b 216 struct inode *inode; /* file being written to */
8d5d02e6 217 unsigned int flag; /* unwritten or not */
4188188b 218 struct page *page; /* for writepage() path */
a1de02dc
ES
219 loff_t offset; /* offset in the file */
220 ssize_t size; /* size of the extent */
0031462b 221 struct work_struct work; /* data work queue */
5b3ff237
JZ
222 struct kiocb *iocb; /* iocb struct for AIO */
223 int result; /* error value for AIO */
4188188b
CW
224 int num_io_pages; /* for writepages() */
225 struct ext4_io_page *pages[MAX_IO_PAGES]; /* for writepages() */
0031462b
MC
226} ext4_io_end_t;
227
bd2d0210
TT
228struct ext4_io_submit {
229 int io_op;
230 struct bio *io_bio;
231 ext4_io_end_t *io_end;
232 struct ext4_io_page *io_page;
233 sector_t io_next_block;
234};
235
ac27a0ec
DK
236/*
237 * Special inodes numbers
238 */
617ba13b
MC
239#define EXT4_BAD_INO 1 /* Bad blocks inode */
240#define EXT4_ROOT_INO 2 /* Root inode */
ae812306
AK
241#define EXT4_USR_QUOTA_INO 3 /* User quota inode */
242#define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
617ba13b
MC
243#define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
244#define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
245#define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
246#define EXT4_JOURNAL_INO 8 /* Journal inode */
ac27a0ec 247
617ba13b
MC
248/* First non-reserved inode for old ext4 filesystems */
249#define EXT4_GOOD_OLD_FIRST_INO 11
ac27a0ec
DK
250
251/*
252 * Maximal count of links to a file
253 */
f8628a14 254#define EXT4_LINK_MAX 65000
ac27a0ec
DK
255
256/*
257 * Macro-instructions used to manage several block sizes
258 */
617ba13b 259#define EXT4_MIN_BLOCK_SIZE 1024
afc7cbca
TS
260#define EXT4_MAX_BLOCK_SIZE 65536
261#define EXT4_MIN_BLOCK_LOG_SIZE 10
fb1813f4 262#define EXT4_MAX_BLOCK_LOG_SIZE 16
ac27a0ec 263#ifdef __KERNEL__
617ba13b 264# define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
ac27a0ec 265#else
617ba13b 266# define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
ac27a0ec 267#endif
af5bc92d 268#define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
281b5995
TT
269#define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
270 EXT4_SB(s)->s_cluster_bits)
ac27a0ec 271#ifdef __KERNEL__
617ba13b 272# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
281b5995 273# define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
ac27a0ec 274#else
617ba13b 275# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
ac27a0ec
DK
276#endif
277#ifdef __KERNEL__
617ba13b
MC
278#define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
279#define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
280#define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
ac27a0ec 281#else
617ba13b
MC
282#define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
283 EXT4_GOOD_OLD_INODE_SIZE : \
ac27a0ec 284 (s)->s_inode_size)
617ba13b
MC
285#define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
286 EXT4_GOOD_OLD_FIRST_INO : \
ac27a0ec
DK
287 (s)->s_first_ino)
288#endif
a2df2a63 289#define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
ac27a0ec 290
d5b8f310
TT
291/* Translate a block number to a cluster number */
292#define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
293/* Translate a cluster number to a block number */
294#define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
295/* Translate # of blks to # of clusters */
296#define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
297 (sbi)->s_cluster_bits)
298
ac27a0ec
DK
299/*
300 * Structure of a blocks group descriptor
301 */
617ba13b 302struct ext4_group_desc
ac27a0ec 303{
3a14589c 304 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
5272f837
AK
305 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
306 __le32 bg_inode_table_lo; /* Inodes table block */
560671a0
AK
307 __le16 bg_free_blocks_count_lo;/* Free blocks count */
308 __le16 bg_free_inodes_count_lo;/* Free inodes count */
309 __le16 bg_used_dirs_count_lo; /* Directories count */
717d50e4 310 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
e6153918
DW
311 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
312 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
313 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
560671a0 314 __le16 bg_itable_unused_lo; /* Unused inodes count */
717d50e4 315 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
8fadc143
AR
316 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
317 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
318 __le32 bg_inode_table_hi; /* Inodes table block MSB */
91b51a01
CL
319 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
320 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
321 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
560671a0 322 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
e6153918
DW
323 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
324 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
325 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
326 __u32 bg_reserved;
ac27a0ec
DK
327};
328
41a246d1
DW
329#define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
330 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
331 sizeof(__le16))
332#define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
333 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
334 sizeof(__le16))
335
772cb7c8
JS
336/*
337 * Structure of a flex block group info
338 */
339
340struct flex_groups {
9f24e420 341 atomic_t free_inodes;
24aaa8ef 342 atomic_t free_clusters;
7d39db14 343 atomic_t used_dirs;
772cb7c8
JS
344};
345
717d50e4
AD
346#define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
347#define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
348#define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
349
ac27a0ec
DK
350/*
351 * Macro-instructions used to manage group descriptors
352 */
0d1ee42f 353#define EXT4_MIN_DESC_SIZE 32
8fadc143 354#define EXT4_MIN_DESC_SIZE_64BIT 64
0d1ee42f
AR
355#define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
356#define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
ac27a0ec 357#ifdef __KERNEL__
617ba13b 358# define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
281b5995 359# define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
617ba13b
MC
360# define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
361# define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
362# define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
ac27a0ec 363#else
617ba13b 364# define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
0d1ee42f 365# define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
617ba13b 366# define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
ac27a0ec
DK
367#endif
368
369/*
370 * Constants relative to the data blocks
371 */
617ba13b
MC
372#define EXT4_NDIR_BLOCKS 12
373#define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
374#define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
375#define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
376#define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
ac27a0ec
DK
377
378/*
379 * Inode flags
380 */
617ba13b
MC
381#define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
382#define EXT4_UNRM_FL 0x00000002 /* Undelete */
383#define EXT4_COMPR_FL 0x00000004 /* Compress file */
384#define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
385#define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
386#define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
387#define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
388#define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
ac27a0ec 389/* Reserved for compression usage... */
617ba13b
MC
390#define EXT4_DIRTY_FL 0x00000100
391#define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
392#define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
393#define EXT4_ECOMPR_FL 0x00000800 /* Compression error */
ac27a0ec 394/* End compression flags --- maybe not all used */
617ba13b
MC
395#define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
396#define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
397#define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
398#define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
399#define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
400#define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
8180a562 401#define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
a86c6181 402#define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
f710b4b9 403#define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
c8d46e41 404#define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
8180a562 405#define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
ac27a0ec 406
c8d46e41
JZ
407#define EXT4_FL_USER_VISIBLE 0x004BDFFF /* User visible flags */
408#define EXT4_FL_USER_MODIFIABLE 0x004B80FF /* User modifiable flags */
ac27a0ec 409
8fa43a81
DG
410/* Flags that should be inherited by new inodes from their parent. */
411#define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
1cd9f097 412 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
8fa43a81
DG
413 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
414 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
415
2dc6b0d4
DG
416/* Flags that are appropriate for regular files (all but dir-specific ones). */
417#define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
418
419/* Flags that are appropriate for non-directories/regular files. */
420#define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
421
422/* Mask out flags that are inappropriate for the given type of inode. */
423static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
424{
425 if (S_ISDIR(mode))
426 return flags;
427 else if (S_ISREG(mode))
428 return flags & EXT4_REG_FLMASK;
429 else
430 return flags & EXT4_OTHER_FLMASK;
431}
432
12e9b892
DM
433/*
434 * Inode flags used for atomic set/get
435 */
436enum {
437 EXT4_INODE_SECRM = 0, /* Secure deletion */
438 EXT4_INODE_UNRM = 1, /* Undelete */
439 EXT4_INODE_COMPR = 2, /* Compress file */
440 EXT4_INODE_SYNC = 3, /* Synchronous updates */
441 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
442 EXT4_INODE_APPEND = 5, /* writes to file may only append */
443 EXT4_INODE_NODUMP = 6, /* do not dump file */
444 EXT4_INODE_NOATIME = 7, /* do not update atime */
445/* Reserved for compression usage... */
446 EXT4_INODE_DIRTY = 8,
447 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
448 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
449 EXT4_INODE_ECOMPR = 11, /* Compression error */
450/* End compression flags --- maybe not all used */
451 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
452 EXT4_INODE_IMAGIC = 13, /* AFS directory */
453 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
454 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
455 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
456 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
457 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
458 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
459 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
460 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
461 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
462};
463
464#define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
465#define CHECK_FLAG_VALUE(FLAG) if (!TEST_FLAG_VALUE(FLAG)) { \
466 printk(KERN_EMERG "EXT4 flag fail: " #FLAG ": %d %d\n", \
467 EXT4_##FLAG##_FL, EXT4_INODE_##FLAG); BUG_ON(1); }
468
469/*
470 * Since it's pretty easy to mix up bit numbers and hex values, and we
471 * can't do a compile-time test for ENUM values, we use a run-time
472 * test to make sure that EXT4_XXX_FL is consistent with respect to
473 * EXT4_INODE_XXX. If all is well the printk and BUG_ON will all drop
474 * out so it won't cost any extra space in the compiled kernel image.
475 * But it's important that these values are the same, since we are
476 * using EXT4_INODE_XXX to test for the flag values, but EXT4_XX_FL
477 * must be consistent with the values of FS_XXX_FL defined in
478 * include/linux/fs.h and the on-disk values found in ext2, ext3, and
479 * ext4 filesystems, and of course the values defined in e2fsprogs.
480 *
481 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
482 */
483static inline void ext4_check_flag_values(void)
484{
485 CHECK_FLAG_VALUE(SECRM);
486 CHECK_FLAG_VALUE(UNRM);
487 CHECK_FLAG_VALUE(COMPR);
488 CHECK_FLAG_VALUE(SYNC);
489 CHECK_FLAG_VALUE(IMMUTABLE);
490 CHECK_FLAG_VALUE(APPEND);
491 CHECK_FLAG_VALUE(NODUMP);
492 CHECK_FLAG_VALUE(NOATIME);
493 CHECK_FLAG_VALUE(DIRTY);
494 CHECK_FLAG_VALUE(COMPRBLK);
495 CHECK_FLAG_VALUE(NOCOMPR);
496 CHECK_FLAG_VALUE(ECOMPR);
497 CHECK_FLAG_VALUE(INDEX);
498 CHECK_FLAG_VALUE(IMAGIC);
499 CHECK_FLAG_VALUE(JOURNAL_DATA);
500 CHECK_FLAG_VALUE(NOTAIL);
501 CHECK_FLAG_VALUE(DIRSYNC);
502 CHECK_FLAG_VALUE(TOPDIR);
503 CHECK_FLAG_VALUE(HUGE_FILE);
504 CHECK_FLAG_VALUE(EXTENTS);
505 CHECK_FLAG_VALUE(EA_INODE);
506 CHECK_FLAG_VALUE(EOFBLOCKS);
507 CHECK_FLAG_VALUE(RESERVED);
508}
509
ac27a0ec 510/* Used to pass group descriptor data when online resize is done */
617ba13b 511struct ext4_new_group_input {
8c55e204
DK
512 __u32 group; /* Group number for this data */
513 __u64 block_bitmap; /* Absolute block number of block bitmap */
514 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
515 __u64 inode_table; /* Absolute block number of inode table start */
516 __u32 blocks_count; /* Total number of blocks in this group */
517 __u16 reserved_blocks; /* Number of reserved blocks in this group */
ac27a0ec
DK
518 __u16 unused;
519};
520
4d92dc0f
BH
521#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
522struct compat_ext4_new_group_input {
523 u32 group;
524 compat_u64 block_bitmap;
525 compat_u64 inode_bitmap;
526 compat_u64 inode_table;
527 u32 blocks_count;
528 u16 reserved_blocks;
529 u16 unused;
530};
531#endif
532
617ba13b
MC
533/* The struct ext4_new_group_input in kernel space, with free_blocks_count */
534struct ext4_new_group_data {
ac27a0ec 535 __u32 group;
bd81d8ee
LV
536 __u64 block_bitmap;
537 __u64 inode_bitmap;
538 __u64 inode_table;
ac27a0ec
DK
539 __u32 blocks_count;
540 __u16 reserved_blocks;
541 __u16 unused;
542 __u32 free_blocks_count;
543};
544
33afdcc5
YY
545/* Indexes used to index group tables in ext4_new_group_data */
546enum {
547 BLOCK_BITMAP = 0, /* block bitmap */
548 INODE_BITMAP, /* inode bitmap */
549 INODE_TABLE, /* inode tables */
550 GROUP_TABLE_COUNT,
551};
552
a2df2a63 553/*
79e83036 554 * Flags used by ext4_map_blocks()
a2df2a63 555 */
c2177057
TT
556 /* Allocate any needed blocks and/or convert an unitialized
557 extent to be an initialized ext4 */
2ac3b6e0 558#define EXT4_GET_BLOCKS_CREATE 0x0001
c2177057 559 /* Request the creation of an unitialized extent */
2ac3b6e0 560#define EXT4_GET_BLOCKS_UNINIT_EXT 0x0002
c2177057
TT
561#define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT (EXT4_GET_BLOCKS_UNINIT_EXT|\
562 EXT4_GET_BLOCKS_CREATE)
c2177057 563 /* Caller is from the delayed allocation writeout path,
60e6679e 564 so set the magic i_delalloc_reserve_flag after taking the
2ac3b6e0 565 inode allocation semaphore for */
03f5d8bc 566#define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
0031462b
MC
567 /* caller is from the direct IO path, request to creation of an
568 unitialized extents if not allocated, split the uninitialized
569 extent if blocks has been preallocated already*/
c7064ef1 570#define EXT4_GET_BLOCKS_PRE_IO 0x0008
1296cc85 571#define EXT4_GET_BLOCKS_CONVERT 0x0010
c7064ef1 572#define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
0031462b 573 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
c7064ef1
JZ
574 /* Convert extent to initialized after IO complete */
575#define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
0031462b 576 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
55f020db
AH
577 /* Punch out blocks of an extent */
578#define EXT4_GET_BLOCKS_PUNCH_OUT_EXT 0x0020
556b27ab
VH
579 /* Don't normalize allocation size (used for fallocate) */
580#define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
a4e5d88b
DM
581 /* Request will not result in inode size update (user for fallocate) */
582#define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080
729f52c6
ZL
583 /* Do not take i_data_sem locking in ext4_map_blocks */
584#define EXT4_GET_BLOCKS_NO_LOCK 0x0100
ac27a0ec 585
e6362609
TT
586/*
587 * Flags used by ext4_free_blocks
588 */
589#define EXT4_FREE_BLOCKS_METADATA 0x0001
590#define EXT4_FREE_BLOCKS_FORGET 0x0002
1f2acb60 591#define EXT4_FREE_BLOCKS_VALIDATED 0x0004
7132de74 592#define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
84130193
TT
593#define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
594#define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
e6362609 595
4e96b2db
AH
596/*
597 * Flags used by ext4_discard_partial_page_buffers
598 */
599#define EXT4_DISCARD_PARTIAL_PG_ZERO_UNMAPPED 0x0001
e6362609 600
ac27a0ec
DK
601/*
602 * ioctl commands
603 */
617ba13b
MC
604#define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
605#define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
606#define EXT4_IOC_GETVERSION _IOR('f', 3, long)
607#define EXT4_IOC_SETVERSION _IOW('f', 4, long)
617ba13b
MC
608#define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
609#define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
617ba13b
MC
610#define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
611#define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
8eea80d5
TT
612#define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
613#define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
614#define EXT4_IOC_MIGRATE _IO('f', 9)
ccd2506b 615 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
8eea80d5 616 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
ccd2506b 617#define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
748de673 618#define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
19c5246d 619#define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
ac27a0ec 620
899ad0ce 621#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
ac27a0ec
DK
622/*
623 * ioctl commands in 32 bit emulation
624 */
617ba13b
MC
625#define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
626#define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
627#define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
628#define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
629#define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
630#define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
631#define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
4d92dc0f 632#define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
617ba13b
MC
633#define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
634#define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
899ad0ce 635#endif
ac27a0ec 636
b595076a 637/* Max physical block we can address w/o extents */
fb0a387d
ES
638#define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
639
ac27a0ec
DK
640/*
641 * Structure of an inode on the disk
642 */
617ba13b 643struct ext4_inode {
ac27a0ec
DK
644 __le16 i_mode; /* File mode */
645 __le16 i_uid; /* Low 16 bits of Owner Uid */
a48380f7 646 __le32 i_size_lo; /* Size in bytes */
ac27a0ec 647 __le32 i_atime; /* Access time */
ef7f3835 648 __le32 i_ctime; /* Inode Change time */
ac27a0ec
DK
649 __le32 i_mtime; /* Modification time */
650 __le32 i_dtime; /* Deletion Time */
651 __le16 i_gid; /* Low 16 bits of Group Id */
652 __le16 i_links_count; /* Links count */
0fc1b451 653 __le32 i_blocks_lo; /* Blocks count */
ac27a0ec
DK
654 __le32 i_flags; /* File flags */
655 union {
656 struct {
25ec56b5 657 __le32 l_i_version;
ac27a0ec
DK
658 } linux1;
659 struct {
660 __u32 h_i_translator;
661 } hurd1;
662 struct {
663 __u32 m_i_reserved1;
664 } masix1;
665 } osd1; /* OS dependent 1 */
617ba13b 666 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
ac27a0ec 667 __le32 i_generation; /* File version (for NFS) */
7973c0c1 668 __le32 i_file_acl_lo; /* File ACL */
a48380f7 669 __le32 i_size_high;
f077d0d7 670 __le32 i_obso_faddr; /* Obsoleted fragment address */
ac27a0ec
DK
671 union {
672 struct {
0fc1b451 673 __le16 l_i_blocks_high; /* were l_i_reserved1 */
a1ddeb7e 674 __le16 l_i_file_acl_high;
8c55e204 675 __le16 l_i_uid_high; /* these 2 fields */
ac27a0ec 676 __le16 l_i_gid_high; /* were reserved2[0] */
e6153918
DW
677 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
678 __le16 l_i_reserved;
ac27a0ec
DK
679 } linux2;
680 struct {
f077d0d7 681 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
ac27a0ec
DK
682 __u16 h_i_mode_high;
683 __u16 h_i_uid_high;
684 __u16 h_i_gid_high;
685 __u32 h_i_author;
686 } hurd2;
687 struct {
f077d0d7 688 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
a1ddeb7e 689 __le16 m_i_file_acl_high;
ac27a0ec
DK
690 __u32 m_i_reserved2[2];
691 } masix2;
692 } osd2; /* OS dependent 2 */
693 __le16 i_extra_isize;
e6153918 694 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
ef7f3835
KS
695 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
696 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
697 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
698 __le32 i_crtime; /* File Creation time */
699 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
25ec56b5 700 __le32 i_version_hi; /* high 32 bits for 64-bit version */
ac27a0ec
DK
701};
702
748de673
AF
703struct move_extent {
704 __u32 reserved; /* should be zero */
705 __u32 donor_fd; /* donor file descriptor */
706 __u64 orig_start; /* logical start offset in block for orig */
707 __u64 donor_start; /* logical start offset in block for donor */
708 __u64 len; /* block length to be moved */
709 __u64 moved_len; /* moved block length */
710};
ac27a0ec 711
ef7f3835
KS
712#define EXT4_EPOCH_BITS 2
713#define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
714#define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
715
716/*
717 * Extended fields will fit into an inode if the filesystem was formatted
718 * with large inodes (-I 256 or larger) and there are not currently any EAs
719 * consuming all of the available space. For new inodes we always reserve
720 * enough space for the kernel's known extended fields, but for inodes
721 * created with an old kernel this might not have been the case. None of
722 * the extended inode fields is critical for correct filesystem operation.
723 * This macro checks if a certain field fits in the inode. Note that
724 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
725 */
726#define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
727 ((offsetof(typeof(*ext4_inode), field) + \
728 sizeof((ext4_inode)->field)) \
729 <= (EXT4_GOOD_OLD_INODE_SIZE + \
730 (einode)->i_extra_isize)) \
731
732static inline __le32 ext4_encode_extra_time(struct timespec *time)
733{
734 return cpu_to_le32((sizeof(time->tv_sec) > 4 ?
c1fccc06
TT
735 (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) |
736 ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK));
ef7f3835
KS
737}
738
739static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
740{
741 if (sizeof(time->tv_sec) > 4)
742 time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK)
743 << 32;
c1fccc06 744 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
ef7f3835
KS
745}
746
747#define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
748do { \
749 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
750 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
751 (raw_inode)->xtime ## _extra = \
752 ext4_encode_extra_time(&(inode)->xtime); \
753} while (0)
754
755#define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
756do { \
757 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
758 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
759 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
760 (raw_inode)->xtime ## _extra = \
761 ext4_encode_extra_time(&(einode)->xtime); \
762} while (0)
763
764#define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
765do { \
766 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
767 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
768 ext4_decode_extra_time(&(inode)->xtime, \
769 raw_inode->xtime ## _extra); \
af0b44a1
ES
770 else \
771 (inode)->xtime.tv_nsec = 0; \
ef7f3835
KS
772} while (0)
773
774#define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
775do { \
776 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
777 (einode)->xtime.tv_sec = \
778 (signed)le32_to_cpu((raw_inode)->xtime); \
779 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
780 ext4_decode_extra_time(&(einode)->xtime, \
781 raw_inode->xtime ## _extra); \
af0b44a1
ES
782 else \
783 (einode)->xtime.tv_nsec = 0; \
ef7f3835
KS
784} while (0)
785
25ec56b5
JNC
786#define i_disk_version osd1.linux1.l_i_version
787
ac27a0ec
DK
788#if defined(__KERNEL__) || defined(__linux__)
789#define i_reserved1 osd1.linux1.l_i_reserved1
a1ddeb7e 790#define i_file_acl_high osd2.linux2.l_i_file_acl_high
0fc1b451 791#define i_blocks_high osd2.linux2.l_i_blocks_high
ac27a0ec
DK
792#define i_uid_low i_uid
793#define i_gid_low i_gid
794#define i_uid_high osd2.linux2.l_i_uid_high
795#define i_gid_high osd2.linux2.l_i_gid_high
e6153918 796#define i_checksum_lo osd2.linux2.l_i_checksum_lo
ac27a0ec
DK
797
798#elif defined(__GNU__)
799
800#define i_translator osd1.hurd1.h_i_translator
ac27a0ec
DK
801#define i_uid_high osd2.hurd2.h_i_uid_high
802#define i_gid_high osd2.hurd2.h_i_gid_high
803#define i_author osd2.hurd2.h_i_author
804
805#elif defined(__masix__)
806
807#define i_reserved1 osd1.masix1.m_i_reserved1
a1ddeb7e 808#define i_file_acl_high osd2.masix2.m_i_file_acl_high
ac27a0ec
DK
809#define i_reserved2 osd2.masix2.m_i_reserved2
810
811#endif /* defined(__KERNEL__) || defined(__linux__) */
812
d444c3c3
TT
813/*
814 * storage for cached extent
b05e6ae5
TT
815 * If ec_len == 0, then the cache is invalid.
816 * If ec_start == 0, then the cache represents a gap (null mapping)
d444c3c3
TT
817 */
818struct ext4_ext_cache {
819 ext4_fsblk_t ec_start;
820 ext4_lblk_t ec_block;
821 __u32 ec_len; /* must be 32bit to return holes */
d444c3c3
TT
822};
823
c0677e6d
ZL
824#include "extents_status.h"
825
d444c3c3
TT
826/*
827 * fourth extended file system inode data in memory
828 */
829struct ext4_inode_info {
830 __le32 i_data[15]; /* unconverted */
d444c3c3 831 __u32 i_dtime;
12e9b892 832 ext4_fsblk_t i_file_acl;
d444c3c3
TT
833
834 /*
835 * i_block_group is the number of the block group which contains
836 * this file's inode. Constant across the lifetime of the inode,
837 * it is ued for making block allocation decisions - we try to
838 * place a file's data blocks near its inode block, and new inodes
839 * near to their parent directory's inode.
840 */
841 ext4_group_t i_block_group;
8a2005d3 842 ext4_lblk_t i_dir_start_lookup;
353eb83c 843#if (BITS_PER_LONG < 64)
19f5fb7a 844 unsigned long i_state_flags; /* Dynamic state flags */
353eb83c 845#endif
12e9b892 846 unsigned long i_flags;
d444c3c3 847
d444c3c3
TT
848#ifdef CONFIG_EXT4_FS_XATTR
849 /*
850 * Extended attributes can be read independently of the main file
851 * data. Taking i_mutex even when reading would cause contention
852 * between readers of EAs and writers of regular file data, so
853 * instead we synchronize on xattr_sem when reading or changing
854 * EAs.
855 */
856 struct rw_semaphore xattr_sem;
857#endif
d444c3c3
TT
858
859 struct list_head i_orphan; /* unlinked but open inodes */
860
861 /*
862 * i_disksize keeps track of what the inode size is ON DISK, not
863 * in memory. During truncate, i_size is set to the new size by
864 * the VFS prior to calling ext4_truncate(), but the filesystem won't
865 * set i_disksize to 0 until the truncate is actually under way.
866 *
867 * The intent is that i_disksize always represents the blocks which
868 * are used by this file. This allows recovery to restart truncate
869 * on orphans if we crash during truncate. We actually write i_disksize
870 * into the on-disk inode when writing inodes out, instead of i_size.
871 *
872 * The only time when i_disksize and i_size may be different is when
873 * a truncate is in progress. The only things which change i_disksize
874 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
875 */
876 loff_t i_disksize;
877
878 /*
879 * i_data_sem is for serialising ext4_truncate() against
880 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
881 * data tree are chopped off during truncate. We can't do that in
882 * ext4 because whenever we perform intermediate commits during
883 * truncate, the inode and all the metadata blocks *must* be in a
884 * consistent state which allows truncation of the orphans to restart
885 * during recovery. Hence we must fix the get_block-vs-truncate race
886 * by other means, so we have i_data_sem.
887 */
888 struct rw_semaphore i_data_sem;
889 struct inode vfs_inode;
8aefcd55 890 struct jbd2_inode *jinode;
d444c3c3
TT
891
892 struct ext4_ext_cache i_cached_extent;
893 /*
894 * File creation time. Its function is same as that of
895 * struct timespec i_{a,c,m}time in the generic inode.
896 */
897 struct timespec i_crtime;
898
899 /* mballoc */
900 struct list_head i_prealloc_list;
901 spinlock_t i_prealloc_lock;
902
c0677e6d
ZL
903 /* extents status tree */
904 struct ext4_es_tree i_es_tree;
905 rwlock_t i_es_lock;
906
d444c3c3
TT
907 /* ialloc */
908 ext4_group_t i_last_alloc_group;
909
910 /* allocation reservation info for delalloc */
57042651 911 /* In case of bigalloc, these refer to clusters rather than blocks */
d444c3c3
TT
912 unsigned int i_reserved_data_blocks;
913 unsigned int i_reserved_meta_blocks;
914 unsigned int i_allocated_meta_blocks;
01f49d0b 915 ext4_lblk_t i_da_metadata_calc_last_lblock;
9d0be502 916 int i_da_metadata_calc_len;
d444c3c3
TT
917
918 /* on-disk additional length */
919 __u16 i_extra_isize;
920
a9e7f447
DM
921#ifdef CONFIG_QUOTA
922 /* quota space reservation, managed internally by quota code */
923 qsize_t i_reserved_quota;
924#endif
8d5d02e6 925
c7064ef1
JZ
926 /* completed IOs that might need unwritten extents handling */
927 struct list_head i_completed_io_list;
744692dc 928 spinlock_t i_completed_io_lock;
8a2005d3 929 atomic_t i_ioend_count; /* Number of outstanding io_end structs */
e27f41e1 930 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
8a2005d3
TT
931
932 spinlock_t i_block_reservation_lock;
b436b9be
JK
933
934 /*
935 * Transactions that contain inode's metadata needed to complete
936 * fsync and fdatasync, respectively.
937 */
938 tid_t i_sync_tid;
939 tid_t i_datasync_tid;
814525f4
DW
940
941 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
942 __u32 i_csum_seed;
d444c3c3
TT
943};
944
ac27a0ec
DK
945/*
946 * File system states
947 */
617ba13b
MC
948#define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
949#define EXT4_ERROR_FS 0x0002 /* Errors detected */
950#define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
ac27a0ec 951
469108ff
TT
952/*
953 * Misc. filesystem flags
954 */
955#define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
956#define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
957#define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
958
ac27a0ec
DK
959/*
960 * Mount flags
961 */
617ba13b
MC
962#define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
963#define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
964#define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
965#define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
966#define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
39ef17f1 967#define EXT4_MOUNT_ERRORS_MASK 0x00070
617ba13b
MC
968#define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
969#define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
617ba13b
MC
970#define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
971#define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
972#define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
973#define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
974#define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
975#define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
976#define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
977#define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
afd4672d 978#define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
617ba13b 979#define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
617ba13b
MC
980#define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
981#define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
982#define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
744692dc 983#define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
d4da6c9c 984#define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
818d276c 985#define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1449032b 986#define EXT4_MOUNT_MBLK_IO_SUBMIT 0x4000000 /* multi-block io submits */
64769240 987#define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
5bf5683a 988#define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
6fd058f7 989#define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
5328e635 990#define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
bfff6873 991#define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
5bf5683a 992
56889787
TT
993#define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
994 specified delalloc */
995
fd8c37ec
TT
996#define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
997 ~EXT4_MOUNT_##opt
998#define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
999 EXT4_MOUNT_##opt
617ba13b
MC
1000#define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1001 EXT4_MOUNT_##opt)
ac27a0ec 1002
a2595b8a
TT
1003#define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1004 ~EXT4_MOUNT2_##opt
1005#define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1006 EXT4_MOUNT2_##opt
1007#define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1008 EXT4_MOUNT2_##opt)
1009
597d508c
AM
1010#define ext4_test_and_set_bit __test_and_set_bit_le
1011#define ext4_set_bit __set_bit_le
617ba13b 1012#define ext4_set_bit_atomic ext2_set_bit_atomic
597d508c
AM
1013#define ext4_test_and_clear_bit __test_and_clear_bit_le
1014#define ext4_clear_bit __clear_bit_le
617ba13b 1015#define ext4_clear_bit_atomic ext2_clear_bit_atomic
50e0168c 1016#define ext4_test_bit test_bit_le
50e0168c
AM
1017#define ext4_find_next_zero_bit find_next_zero_bit_le
1018#define ext4_find_next_bit find_next_bit_le
ac27a0ec 1019
c3e94d1d
YY
1020extern void ext4_set_bits(void *bm, int cur, int len);
1021
ac27a0ec
DK
1022/*
1023 * Maximal mount counts between two filesystem checks
1024 */
617ba13b
MC
1025#define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1026#define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
ac27a0ec
DK
1027
1028/*
1029 * Behaviour when detecting errors
1030 */
617ba13b
MC
1031#define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1032#define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1033#define EXT4_ERRORS_PANIC 3 /* Panic */
1034#define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
ac27a0ec 1035
e6153918
DW
1036/* Metadata checksum algorithm codes */
1037#define EXT4_CRC32C_CHKSUM 1
1038
ac27a0ec
DK
1039/*
1040 * Structure of the super block
1041 */
617ba13b 1042struct ext4_super_block {
ac27a0ec 1043/*00*/ __le32 s_inodes_count; /* Inodes count */
6bc9feff 1044 __le32 s_blocks_count_lo; /* Blocks count */
308ba3ec
AK
1045 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1046 __le32 s_free_blocks_count_lo; /* Free blocks count */
ac27a0ec
DK
1047/*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1048 __le32 s_first_data_block; /* First Data Block */
1049 __le32 s_log_block_size; /* Block size */
281b5995 1050 __le32 s_log_cluster_size; /* Allocation cluster size */
ac27a0ec 1051/*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
281b5995 1052 __le32 s_clusters_per_group; /* # Clusters per group */
ac27a0ec
DK
1053 __le32 s_inodes_per_group; /* # Inodes per group */
1054 __le32 s_mtime; /* Mount time */
1055/*30*/ __le32 s_wtime; /* Write time */
1056 __le16 s_mnt_count; /* Mount count */
1057 __le16 s_max_mnt_count; /* Maximal mount count */
1058 __le16 s_magic; /* Magic signature */
1059 __le16 s_state; /* File system state */
1060 __le16 s_errors; /* Behaviour when detecting errors */
1061 __le16 s_minor_rev_level; /* minor revision level */
1062/*40*/ __le32 s_lastcheck; /* time of last check */
1063 __le32 s_checkinterval; /* max. time between checks */
1064 __le32 s_creator_os; /* OS */
1065 __le32 s_rev_level; /* Revision level */
1066/*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1067 __le16 s_def_resgid; /* Default gid for reserved blocks */
1068 /*
617ba13b 1069 * These fields are for EXT4_DYNAMIC_REV superblocks only.
ac27a0ec
DK
1070 *
1071 * Note: the difference between the compatible feature set and
1072 * the incompatible feature set is that if there is a bit set
1073 * in the incompatible feature set that the kernel doesn't
1074 * know about, it should refuse to mount the filesystem.
1075 *
1076 * e2fsck's requirements are more strict; if it doesn't know
1077 * about a feature in either the compatible or incompatible
1078 * feature set, it must abort and not try to meddle with
1079 * things it doesn't understand...
1080 */
1081 __le32 s_first_ino; /* First non-reserved inode */
0d1ee42f 1082 __le16 s_inode_size; /* size of inode structure */
ac27a0ec
DK
1083 __le16 s_block_group_nr; /* block group # of this superblock */
1084 __le32 s_feature_compat; /* compatible feature set */
1085/*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1086 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1087/*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1088/*78*/ char s_volume_name[16]; /* volume name */
1089/*88*/ char s_last_mounted[64]; /* directory where last mounted */
1090/*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1091 /*
1092 * Performance hints. Directory preallocation should only
617ba13b 1093 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
ac27a0ec
DK
1094 */
1095 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1096 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1097 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1098 /*
617ba13b 1099 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
ac27a0ec
DK
1100 */
1101/*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1102/*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1103 __le32 s_journal_dev; /* device number of journal file */
1104 __le32 s_last_orphan; /* start of list of inodes to delete */
1105 __le32 s_hash_seed[4]; /* HTREE hash seed */
1106 __u8 s_def_hash_version; /* Default hash version to use */
89eeddf0 1107 __u8 s_jnl_backup_type;
0d1ee42f 1108 __le16 s_desc_size; /* size of group descriptor */
bd81d8ee 1109/*100*/ __le32 s_default_mount_opts;
ac27a0ec 1110 __le32 s_first_meta_bg; /* First metablock block group */
bd81d8ee
LV
1111 __le32 s_mkfs_time; /* When the filesystem was created */
1112 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1113 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1114/*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1115 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1116 __le32 s_free_blocks_count_hi; /* Free blocks count */
c1bddad9
AK
1117 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1118 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1119 __le32 s_flags; /* Miscellaneous flags */
1120 __le16 s_raid_stride; /* RAID stride */
c5e06d10 1121 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
c1bddad9
AK
1122 __le64 s_mmp_block; /* Block for multi-mount protection */
1123 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
772cb7c8 1124 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
e6153918 1125 __u8 s_checksum_type; /* metadata checksum algorithm used */
772cb7c8 1126 __le16 s_reserved_pad;
afc32f7e 1127 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
4a9cdec7
TT
1128 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1129 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1130 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1131 snapshot's future use */
1132 __le32 s_snapshot_list; /* inode number of the head of the
1133 on-disk snapshot list */
1c13d5c0
TT
1134#define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1135 __le32 s_error_count; /* number of fs errors */
1136 __le32 s_first_error_time; /* first time an error happened */
1137 __le32 s_first_error_ino; /* inode involved in first error */
1138 __le64 s_first_error_block; /* block involved of first error */
1139 __u8 s_first_error_func[32]; /* function where the error happened */
1140 __le32 s_first_error_line; /* line number where error happened */
1141 __le32 s_last_error_time; /* most recent time of an error */
1142 __le32 s_last_error_ino; /* inode involved in last error */
1143 __le32 s_last_error_line; /* line number where error happened */
1144 __le64 s_last_error_block; /* block involved of last error */
1145 __u8 s_last_error_func[32]; /* function where the error happened */
8b67f04a
TT
1146#define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1147 __u8 s_mount_opts[64];
281b5995
TT
1148 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1149 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1150 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
e6153918
DW
1151 __le32 s_reserved[108]; /* Padding to the end of the block */
1152 __le32 s_checksum; /* crc32c(superblock) */
ac27a0ec
DK
1153};
1154
1c13d5c0
TT
1155#define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1156
ac27a0ec 1157#ifdef __KERNEL__
bc0b0d6d
TT
1158
1159/*
4ab2f15b 1160 * run-time mount flags
bc0b0d6d
TT
1161 */
1162#define EXT4_MF_MNTDIR_SAMPLED 0x0001
4ab2f15b 1163#define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
bc0b0d6d 1164
ca0faba0
TT
1165/*
1166 * fourth extended-fs super-block data in memory
1167 */
1168struct ext4_sb_info {
1169 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1170 unsigned long s_inodes_per_block;/* Number of inodes per block */
1171 unsigned long s_blocks_per_group;/* Number of blocks in a group */
281b5995 1172 unsigned long s_clusters_per_group; /* Number of clusters in a group */
ca0faba0
TT
1173 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1174 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1175 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1176 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1177 ext4_group_t s_groups_count; /* Number of groups in the fs */
fb0a387d 1178 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
952fc18e 1179 unsigned long s_overhead; /* # of fs overhead clusters */
281b5995
TT
1180 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1181 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
ca0faba0
TT
1182 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1183 struct buffer_head * s_sbh; /* Buffer containing the super block */
1184 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1185 struct buffer_head **s_group_desc;
7f4520cc 1186 unsigned int s_mount_opt;
a2595b8a 1187 unsigned int s_mount_opt2;
bc0b0d6d 1188 unsigned int s_mount_flags;
5a916be1 1189 unsigned int s_def_mount_opt;
ca0faba0 1190 ext4_fsblk_t s_sb_block;
08cefc7a
EB
1191 kuid_t s_resuid;
1192 kgid_t s_resgid;
ca0faba0
TT
1193 unsigned short s_mount_state;
1194 unsigned short s_pad;
1195 int s_addr_per_block_bits;
1196 int s_desc_per_block_bits;
1197 int s_inode_size;
1198 int s_first_ino;
1199 unsigned int s_inode_readahead_blks;
11013911 1200 unsigned int s_inode_goal;
ca0faba0
TT
1201 spinlock_t s_next_gen_lock;
1202 u32 s_next_generation;
1203 u32 s_hash_seed[4];
1204 int s_def_hash_version;
1205 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
57042651 1206 struct percpu_counter s_freeclusters_counter;
ca0faba0
TT
1207 struct percpu_counter s_freeinodes_counter;
1208 struct percpu_counter s_dirs_counter;
57042651 1209 struct percpu_counter s_dirtyclusters_counter;
ca0faba0
TT
1210 struct blockgroup_lock *s_blockgroup_lock;
1211 struct proc_dir_entry *s_proc;
1212 struct kobject s_kobj;
1213 struct completion s_kobj_unregister;
2c0544b2 1214 struct super_block *s_sb;
ca0faba0
TT
1215
1216 /* Journaling */
ca0faba0
TT
1217 struct journal_s *s_journal;
1218 struct list_head s_orphan;
1219 struct mutex s_orphan_lock;
8f82f840
YY
1220 unsigned long s_resize_flags; /* Flags indicating if there
1221 is a resizer */
ca0faba0
TT
1222 unsigned long s_commit_interval;
1223 u32 s_max_batch_time;
1224 u32 s_min_batch_time;
1225 struct block_device *journal_bdev;
ca0faba0
TT
1226#ifdef CONFIG_QUOTA
1227 char *s_qf_names[MAXQUOTAS]; /* Names of quota files with journalled quota */
1228 int s_jquota_fmt; /* Format of quota to use */
1229#endif
1230 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
6fd058f7 1231 struct rb_root system_blks;
ca0faba0
TT
1232
1233#ifdef EXTENTS_STATS
1234 /* ext4 extents stats */
1235 unsigned long s_ext_min;
1236 unsigned long s_ext_max;
1237 unsigned long s_depth_max;
1238 spinlock_t s_ext_stats_lock;
1239 unsigned long s_ext_blocks;
1240 unsigned long s_ext_extents;
1241#endif
1242
1243 /* for buddy allocator */
1244 struct ext4_group_info ***s_group_info;
1245 struct inode *s_buddy_cache;
ca0faba0 1246 spinlock_t s_md_lock;
ca0faba0
TT
1247 unsigned short *s_mb_offsets;
1248 unsigned int *s_mb_maxs;
28623c2f 1249 unsigned int s_group_info_size;
ca0faba0
TT
1250
1251 /* tunables */
1252 unsigned long s_stripe;
1253 unsigned int s_mb_stream_request;
1254 unsigned int s_mb_max_to_scan;
1255 unsigned int s_mb_min_to_scan;
1256 unsigned int s_mb_stats;
1257 unsigned int s_mb_order2_reqs;
1258 unsigned int s_mb_group_prealloc;
55138e0b 1259 unsigned int s_max_writeback_mb_bump;
df981d03 1260 unsigned int s_max_dir_size_kb;
ca0faba0
TT
1261 /* where last allocation was done - for stream allocation */
1262 unsigned long s_mb_last_group;
1263 unsigned long s_mb_last_start;
1264
ca0faba0 1265 /* stats for buddy allocator */
ca0faba0
TT
1266 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1267 atomic_t s_bal_success; /* we found long enough chunks */
1268 atomic_t s_bal_allocated; /* in blocks */
1269 atomic_t s_bal_ex_scanned; /* total extents scanned */
1270 atomic_t s_bal_goals; /* goal hits */
1271 atomic_t s_bal_breaks; /* too long searches */
1272 atomic_t s_bal_2orders; /* 2^order hits */
1273 spinlock_t s_bal_lock;
1274 unsigned long s_mb_buddies_generated;
1275 unsigned long long s_mb_generation_time;
1276 atomic_t s_mb_lost_chunks;
1277 atomic_t s_mb_preallocated;
1278 atomic_t s_mb_discarded;
50797481 1279 atomic_t s_lock_busy;
ca0faba0
TT
1280
1281 /* locality groups */
003cb608 1282 struct ext4_locality_group __percpu *s_locality_groups;
ca0faba0
TT
1283
1284 /* for write statistics */
1285 unsigned long s_sectors_written_start;
1286 u64 s_kbytes_written;
1287
67a5da56
ZL
1288 /* the size of zero-out chunk */
1289 unsigned int s_extent_max_zeroout_kb;
1290
ca0faba0
TT
1291 unsigned int s_log_groups_per_flex;
1292 struct flex_groups *s_flex_groups;
117fff10 1293 ext4_group_t s_flex_groups_allocated;
4c0425ff
MC
1294
1295 /* workqueue for dio unwritten */
1296 struct workqueue_struct *dio_unwritten_wq;
66e61a9e
TT
1297
1298 /* timer for periodic error stats printing */
1299 struct timer_list s_err_report;
bfff6873
LC
1300
1301 /* Lazy inode table initialization info */
1302 struct ext4_li_request *s_li_request;
1303 /* Wait multiplier for lazy initialization thread */
1304 unsigned int s_li_wait_mult;
c5e06d10
JL
1305
1306 /* Kernel thread for multiple mount protection */
1307 struct task_struct *s_mmp_tsk;
3d56b8d2
TM
1308
1309 /* record the last minlen when FITRIM is called. */
1310 atomic_t s_last_trim_minblks;
0441984a
DW
1311
1312 /* Reference to checksum algorithm driver via cryptoapi */
1313 struct crypto_shash *s_chksum_driver;
a9c47317
DW
1314
1315 /* Precomputed FS UUID checksum for seeding other checksums */
1316 __u32 s_csum_seed;
ca0faba0
TT
1317};
1318
af5bc92d 1319static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
ac27a0ec
DK
1320{
1321 return sb->s_fs_info;
1322}
617ba13b 1323static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
ac27a0ec 1324{
617ba13b 1325 return container_of(inode, struct ext4_inode_info, vfs_inode);
ac27a0ec
DK
1326}
1327
ef7f3835
KS
1328static inline struct timespec ext4_current_time(struct inode *inode)
1329{
1330 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1331 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1332}
1333
617ba13b 1334static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
ac27a0ec 1335{
617ba13b 1336 return ino == EXT4_ROOT_INO ||
7c319d32
AK
1337 ino == EXT4_USR_QUOTA_INO ||
1338 ino == EXT4_GRP_QUOTA_INO ||
617ba13b
MC
1339 ino == EXT4_JOURNAL_INO ||
1340 ino == EXT4_RESIZE_INO ||
1341 (ino >= EXT4_FIRST_INO(sb) &&
1342 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
ac27a0ec 1343}
19f5fb7a 1344
0edeb71d
TM
1345static inline void ext4_set_io_unwritten_flag(struct inode *inode,
1346 struct ext4_io_end *io_end)
1347{
1348 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
1349 io_end->flag |= EXT4_IO_END_UNWRITTEN;
e27f41e1 1350 atomic_inc(&EXT4_I(inode)->i_unwritten);
0edeb71d
TM
1351 }
1352}
1353
f45ee3a1
DM
1354static inline ext4_io_end_t *ext4_inode_aio(struct inode *inode)
1355{
1356 return inode->i_private;
1357}
1358
1359static inline void ext4_inode_aio_set(struct inode *inode, ext4_io_end_t *io)
1360{
1361 inode->i_private = io;
1362}
1363
19f5fb7a
TT
1364/*
1365 * Inode dynamic state flags
1366 */
1367enum {
1368 EXT4_STATE_JDATA, /* journaled data exists */
1369 EXT4_STATE_NEW, /* inode is newly created */
1370 EXT4_STATE_XATTR, /* has in-inode xattrs */
1371 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1372 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1373 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1374 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
14ece102 1375 EXT4_STATE_NEWENTRY, /* File just added to dir */
f2321097 1376 EXT4_STATE_DELALLOC_RESERVED, /* blks already reserved for delalloc */
17335dcc
DM
1377 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read
1378 nolocking */
19f5fb7a
TT
1379};
1380
353eb83c 1381#define EXT4_INODE_BIT_FNS(name, field, offset) \
12e9b892
DM
1382static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1383{ \
353eb83c 1384 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
12e9b892
DM
1385} \
1386static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1387{ \
353eb83c 1388 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
12e9b892
DM
1389} \
1390static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1391{ \
353eb83c 1392 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
19f5fb7a
TT
1393}
1394
353eb83c
TT
1395EXT4_INODE_BIT_FNS(flag, flags, 0)
1396#if (BITS_PER_LONG < 64)
1397EXT4_INODE_BIT_FNS(state, state_flags, 0)
1398
1399static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1400{
1401 (ei)->i_state_flags = 0;
1402}
1403#else
1404EXT4_INODE_BIT_FNS(state, flags, 32)
1405
1406static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1407{
1408 /* We depend on the fact that callers will set i_flags */
1409}
1410#endif
ac27a0ec 1411#else
617ba13b 1412/* Assume that user mode programs are passing in an ext4fs superblock, not
ac27a0ec
DK
1413 * a kernel struct super_block. This will allow us to call the feature-test
1414 * macros from user land. */
617ba13b 1415#define EXT4_SB(sb) (sb)
ac27a0ec
DK
1416#endif
1417
617ba13b 1418#define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
ac27a0ec
DK
1419
1420/*
1421 * Codes for operating systems
1422 */
617ba13b
MC
1423#define EXT4_OS_LINUX 0
1424#define EXT4_OS_HURD 1
1425#define EXT4_OS_MASIX 2
1426#define EXT4_OS_FREEBSD 3
1427#define EXT4_OS_LITES 4
ac27a0ec
DK
1428
1429/*
1430 * Revision levels
1431 */
617ba13b
MC
1432#define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1433#define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
ac27a0ec 1434
617ba13b
MC
1435#define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1436#define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
ac27a0ec 1437
617ba13b 1438#define EXT4_GOOD_OLD_INODE_SIZE 128
ac27a0ec
DK
1439
1440/*
1441 * Feature set definitions
1442 */
1443
617ba13b 1444#define EXT4_HAS_COMPAT_FEATURE(sb,mask) \
3a06d778 1445 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
617ba13b 1446#define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \
3a06d778 1447 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
617ba13b 1448#define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \
3a06d778 1449 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
617ba13b
MC
1450#define EXT4_SET_COMPAT_FEATURE(sb,mask) \
1451 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1452#define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \
1453 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1454#define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \
1455 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1456#define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \
1457 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1458#define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
1459 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1460#define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \
1461 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1462
1463#define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1464#define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1465#define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1466#define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1467#define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1468#define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1469
1470#define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1471#define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1472#define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
0fc1b451 1473#define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
717d50e4 1474#define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
f8628a14 1475#define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
ef7f3835 1476#define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
ae812306 1477#define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
281b5995 1478#define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
e6153918
DW
1479/*
1480 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1481 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1482 * all other data structures' checksums. However, the METADATA_CSUM and
1483 * GDT_CSUM bits are mutually exclusive.
1484 */
9b90e5e0 1485#define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
617ba13b
MC
1486
1487#define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1488#define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1489#define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1490#define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1491#define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
a86c6181 1492#define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
bd81d8ee 1493#define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
0fc1b451 1494#define EXT4_FEATURE_INCOMPAT_MMP 0x0100
ce421581 1495#define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
f710b4b9
TT
1496#define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1497#define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
856cbcf9 1498#define EXT4_FEATURE_INCOMPAT_BG_USE_META_CSUM 0x2000 /* use crc32c for bg */
9b90e5e0 1499#define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
856cbcf9 1500#define EXT4_FEATURE_INCOMPAT_INLINEDATA 0x8000 /* data in inode */
617ba13b 1501
2035e776
TT
1502#define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1503#define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1504 EXT4_FEATURE_INCOMPAT_META_BG)
1505#define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1506 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1507 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1508
1509#define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1510#define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1511 EXT4_FEATURE_INCOMPAT_RECOVER| \
1512 EXT4_FEATURE_INCOMPAT_META_BG)
1513#define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1514 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1515 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1516
617ba13b
MC
1517#define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR
1518#define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1519 EXT4_FEATURE_INCOMPAT_RECOVER| \
a86c6181 1520 EXT4_FEATURE_INCOMPAT_META_BG| \
bd81d8ee 1521 EXT4_FEATURE_INCOMPAT_EXTENTS| \
ce421581 1522 EXT4_FEATURE_INCOMPAT_64BIT| \
c5e06d10
JL
1523 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1524 EXT4_FEATURE_INCOMPAT_MMP)
617ba13b
MC
1525#define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1526 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
717d50e4 1527 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
f8628a14 1528 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
ef7f3835 1529 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
0fc1b451 1530 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
6f16b606 1531 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
e93376c2 1532 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
7c319d32
AK
1533 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1534 EXT4_FEATURE_RO_COMPAT_QUOTA)
ac27a0ec
DK
1535
1536/*
1537 * Default values for user and/or group using reserved blocks
1538 */
617ba13b
MC
1539#define EXT4_DEF_RESUID 0
1540#define EXT4_DEF_RESGID 0
ac27a0ec 1541
240799cd
TT
1542#define EXT4_DEF_INODE_READAHEAD_BLKS 32
1543
ac27a0ec
DK
1544/*
1545 * Default mount options
1546 */
617ba13b
MC
1547#define EXT4_DEFM_DEBUG 0x0001
1548#define EXT4_DEFM_BSDGROUPS 0x0002
1549#define EXT4_DEFM_XATTR_USER 0x0004
1550#define EXT4_DEFM_ACL 0x0008
1551#define EXT4_DEFM_UID16 0x0010
1552#define EXT4_DEFM_JMODE 0x0060
1553#define EXT4_DEFM_JMODE_DATA 0x0020
1554#define EXT4_DEFM_JMODE_ORDERED 0x0040
1555#define EXT4_DEFM_JMODE_WBACK 0x0060
8b67f04a
TT
1556#define EXT4_DEFM_NOBARRIER 0x0100
1557#define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1558#define EXT4_DEFM_DISCARD 0x0400
1559#define EXT4_DEFM_NODELALLOC 0x0800
ac27a0ec 1560
30773840
TT
1561/*
1562 * Default journal batch times
1563 */
1564#define EXT4_DEF_MIN_BATCH_TIME 0
1565#define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1566
a4912123
TT
1567/*
1568 * Minimum number of groups in a flexgroup before we separate out
1569 * directories into the first block group of a flexgroup
1570 */
1571#define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1572
ac27a0ec
DK
1573/*
1574 * Structure of a directory entry
1575 */
617ba13b 1576#define EXT4_NAME_LEN 255
ac27a0ec 1577
617ba13b 1578struct ext4_dir_entry {
ac27a0ec
DK
1579 __le32 inode; /* Inode number */
1580 __le16 rec_len; /* Directory entry length */
1581 __le16 name_len; /* Name length */
617ba13b 1582 char name[EXT4_NAME_LEN]; /* File name */
ac27a0ec
DK
1583};
1584
1585/*
617ba13b 1586 * The new version of the directory entry. Since EXT4 structures are
ac27a0ec
DK
1587 * stored in intel byte order, and the name_len field could never be
1588 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1589 * file_type field.
1590 */
617ba13b 1591struct ext4_dir_entry_2 {
ac27a0ec
DK
1592 __le32 inode; /* Inode number */
1593 __le16 rec_len; /* Directory entry length */
1594 __u8 name_len; /* Name length */
1595 __u8 file_type;
617ba13b 1596 char name[EXT4_NAME_LEN]; /* File name */
ac27a0ec
DK
1597};
1598
e6153918
DW
1599/*
1600 * This is a bogus directory entry at the end of each leaf block that
1601 * records checksums.
1602 */
1603struct ext4_dir_entry_tail {
1604 __le32 det_reserved_zero1; /* Pretend to be unused */
1605 __le16 det_rec_len; /* 12 */
1606 __u8 det_reserved_zero2; /* Zero name length */
1607 __u8 det_reserved_ft; /* 0xDE, fake file type */
1608 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
1609};
1610
ac27a0ec 1611/*
617ba13b 1612 * Ext4 directory file types. Only the low 3 bits are used. The
ac27a0ec
DK
1613 * other bits are reserved for now.
1614 */
617ba13b
MC
1615#define EXT4_FT_UNKNOWN 0
1616#define EXT4_FT_REG_FILE 1
1617#define EXT4_FT_DIR 2
1618#define EXT4_FT_CHRDEV 3
1619#define EXT4_FT_BLKDEV 4
1620#define EXT4_FT_FIFO 5
1621#define EXT4_FT_SOCK 6
1622#define EXT4_FT_SYMLINK 7
ac27a0ec 1623
617ba13b 1624#define EXT4_FT_MAX 8
ac27a0ec 1625
e6153918
DW
1626#define EXT4_FT_DIR_CSUM 0xDE
1627
ac27a0ec 1628/*
617ba13b 1629 * EXT4_DIR_PAD defines the directory entries boundaries
ac27a0ec
DK
1630 *
1631 * NOTE: It must be a multiple of 4
1632 */
617ba13b
MC
1633#define EXT4_DIR_PAD 4
1634#define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1635#define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1636 ~EXT4_DIR_ROUND)
a72d7f83
JK
1637#define EXT4_MAX_REC_LEN ((1<<16)-1)
1638
0cfc9255
ES
1639/*
1640 * If we ever get support for fs block sizes > page_size, we'll need
1641 * to remove the #if statements in the next two functions...
1642 */
1643static inline unsigned int
1644ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1645{
1646 unsigned len = le16_to_cpu(dlen);
1647
1648#if (PAGE_CACHE_SIZE >= 65536)
1649 if (len == EXT4_MAX_REC_LEN || len == 0)
1650 return blocksize;
1651 return (len & 65532) | ((len & 3) << 16);
1652#else
1653 return len;
1654#endif
1655}
1656
1657static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1658{
1659 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1660 BUG();
1661#if (PAGE_CACHE_SIZE >= 65536)
1662 if (len < 65536)
1663 return cpu_to_le16(len);
1664 if (len == blocksize) {
1665 if (blocksize == 65536)
1666 return cpu_to_le16(EXT4_MAX_REC_LEN);
1667 else
1668 return cpu_to_le16(0);
1669 }
1670 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1671#else
1672 return cpu_to_le16(len);
1673#endif
1674}
1675
ac27a0ec
DK
1676/*
1677 * Hash Tree Directory indexing
1678 * (c) Daniel Phillips, 2001
1679 */
1680
4074fe37
ES
1681#define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1682 EXT4_FEATURE_COMPAT_DIR_INDEX) && \
12e9b892 1683 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
617ba13b
MC
1684#define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1685#define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
ac27a0ec
DK
1686
1687/* Legal values for the dx_root hash_version field: */
1688
1689#define DX_HASH_LEGACY 0
1690#define DX_HASH_HALF_MD4 1
1691#define DX_HASH_TEA 2
f99b2589
TT
1692#define DX_HASH_LEGACY_UNSIGNED 3
1693#define DX_HASH_HALF_MD4_UNSIGNED 4
1694#define DX_HASH_TEA_UNSIGNED 5
ac27a0ec 1695
0441984a
DW
1696static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
1697 const void *address, unsigned int length)
1698{
1699 struct {
1700 struct shash_desc shash;
3108b54b 1701 char ctx[4];
0441984a
DW
1702 } desc;
1703 int err;
1704
3108b54b
TT
1705 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
1706
0441984a
DW
1707 desc.shash.tfm = sbi->s_chksum_driver;
1708 desc.shash.flags = 0;
1709 *(u32 *)desc.ctx = crc;
1710
1711 err = crypto_shash_update(&desc.shash, address, length);
1712 BUG_ON(err);
1713
1714 return *(u32 *)desc.ctx;
1715}
1716
ac27a0ec
DK
1717#ifdef __KERNEL__
1718
1719/* hash info structure used by the directory hash */
1720struct dx_hash_info
1721{
1722 u32 hash;
1723 u32 minor_hash;
1724 int hash_version;
1725 u32 *seed;
1726};
1727
d1f5273e
FY
1728
1729/* 32 and 64 bit signed EOF for dx directories */
1730#define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
1731#define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
1732
ac27a0ec
DK
1733
1734/*
617ba13b 1735 * Control parameters used by ext4_htree_next_block
ac27a0ec
DK
1736 */
1737#define HASH_NB_ALWAYS 1
1738
1739
1740/*
1741 * Describe an inode's exact location on disk and in memory
1742 */
617ba13b 1743struct ext4_iloc
ac27a0ec
DK
1744{
1745 struct buffer_head *bh;
1746 unsigned long offset;
fd2d4291 1747 ext4_group_t block_group;
ac27a0ec
DK
1748};
1749
617ba13b 1750static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
ac27a0ec 1751{
617ba13b 1752 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
ac27a0ec
DK
1753}
1754
1755/*
1756 * This structure is stuffed into the struct file's private_data field
1757 * for directories. It is where we put information so that we can do
1758 * readdir operations in hash tree order.
1759 */
1760struct dir_private_info {
1761 struct rb_root root;
1762 struct rb_node *curr_node;
1763 struct fname *extra_fname;
1764 loff_t last_pos;
1765 __u32 curr_hash;
1766 __u32 curr_minor_hash;
1767 __u32 next_hash;
1768};
1769
1770/* calculate the first block number of the group */
617ba13b 1771static inline ext4_fsblk_t
fd2d4291 1772ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
ac27a0ec 1773{
617ba13b
MC
1774 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
1775 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
ac27a0ec
DK
1776}
1777
1778/*
1779 * Special error return code only used by dx_probe() and its callers.
1780 */
1781#define ERR_BAD_DX_DIR -75000
1782
72b64b59 1783void ext4_get_group_no_and_offset(struct super_block *sb, ext4_fsblk_t blocknr,
a9df9a49 1784 ext4_group_t *blockgrpp, ext4_grpblk_t *offsetp);
3a5b2ecd 1785
bfff6873
LC
1786/*
1787 * Timeout and state flag for lazy initialization inode thread.
1788 */
1789#define EXT4_DEF_LI_WAIT_MULT 10
1790#define EXT4_DEF_LI_MAX_START_DELAY 5
1791#define EXT4_LAZYINIT_QUIT 0x0001
1792#define EXT4_LAZYINIT_RUNNING 0x0002
1793
1794/*
1795 * Lazy inode table initialization info
1796 */
1797struct ext4_lazy_init {
1798 unsigned long li_state;
bfff6873
LC
1799 struct list_head li_request_list;
1800 struct mutex li_list_mtx;
1801};
1802
1803struct ext4_li_request {
1804 struct super_block *lr_super;
1805 struct ext4_sb_info *lr_sbi;
1806 ext4_group_t lr_next_group;
1807 struct list_head lr_request;
1808 unsigned long lr_next_sched;
1809 unsigned long lr_timeout;
1810};
1811
857ac889
LC
1812struct ext4_features {
1813 struct kobject f_kobj;
1814 struct completion f_kobj_unregister;
1815};
1816
c5e06d10
JL
1817/*
1818 * This structure will be used for multiple mount protection. It will be
1819 * written into the block number saved in the s_mmp_block field in the
1820 * superblock. Programs that check MMP should assume that if
1821 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
1822 * to use the filesystem, regardless of how old the timestamp is.
1823 */
1824#define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
1825#define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
1826#define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
1827#define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
1828
1829struct mmp_struct {
1830 __le32 mmp_magic; /* Magic number for MMP */
1831 __le32 mmp_seq; /* Sequence no. updated periodically */
1832
1833 /*
1834 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
1835 * purposes and do not affect the correctness of the algorithm
1836 */
1837 __le64 mmp_time; /* Time last updated */
1838 char mmp_nodename[64]; /* Node which last updated MMP block */
1839 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
1840
1841 /*
1842 * mmp_check_interval is used to verify if the MMP block has been
1843 * updated on the block device. The value is updated based on the
1844 * maximum time to write the MMP block during an update cycle.
1845 */
1846 __le16 mmp_check_interval;
1847
1848 __le16 mmp_pad1;
e6153918
DW
1849 __le32 mmp_pad2[226];
1850 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
c5e06d10
JL
1851};
1852
1853/* arguments passed to the mmp thread */
1854struct mmpd_data {
1855 struct buffer_head *bh; /* bh from initial read_mmp_block() */
1856 struct super_block *sb; /* super block of the fs */
1857};
1858
1859/*
1860 * Check interval multiplier
1861 * The MMP block is written every update interval and initially checked every
1862 * update interval x the multiplier (the value is then adapted based on the
1863 * write latency). The reason is that writes can be delayed under load and we
1864 * don't want readers to incorrectly assume that the filesystem is no longer
1865 * in use.
1866 */
1867#define EXT4_MMP_CHECK_MULT 2UL
1868
1869/*
1870 * Minimum interval for MMP checking in seconds.
1871 */
1872#define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
1873
1874/*
1875 * Maximum interval for MMP checking in seconds.
1876 */
1877#define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
1878
ac27a0ec
DK
1879/*
1880 * Function prototypes
1881 */
1882
1883/*
1884 * Ok, these declarations are also in <linux/kernel.h> but none of the
617ba13b 1885 * ext4 source programs needs to include it so they are duplicated here.
ac27a0ec 1886 */
8c55e204
DK
1887# define NORET_TYPE /**/
1888# define ATTRIB_NORET __attribute__((noreturn))
1889# define NORET_AND noreturn,
ac27a0ec 1890
498e5f24 1891/* bitmap.c */
f6fb99ca 1892extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
41a246d1
DW
1893void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
1894 struct ext4_group_desc *gdp,
1895 struct buffer_head *bh, int sz);
1896int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
1897 struct ext4_group_desc *gdp,
1898 struct buffer_head *bh, int sz);
fa77dcfa
DW
1899void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
1900 struct ext4_group_desc *gdp,
79f1ba49 1901 struct buffer_head *bh);
fa77dcfa
DW
1902int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
1903 struct ext4_group_desc *gdp,
79f1ba49 1904 struct buffer_head *bh);
498e5f24 1905
ac27a0ec 1906/* balloc.c */
fa77dcfa
DW
1907extern void ext4_validate_block_bitmap(struct super_block *sb,
1908 struct ext4_group_desc *desc,
1909 unsigned int block_group,
1910 struct buffer_head *bh);
3a5b2ecd
MC
1911extern unsigned int ext4_block_group(struct super_block *sb,
1912 ext4_fsblk_t blocknr);
1913extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
1914 ext4_fsblk_t blocknr);
fd2d4291
AM
1915extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
1916extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
1917 ext4_group_t group);
7061eba7 1918extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
55f020db
AH
1919 ext4_fsblk_t goal,
1920 unsigned int flags,
1921 unsigned long *count,
1922 int *errp);
e7d5f315
TT
1923extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
1924 s64 nclusters, unsigned int flags);
5dee5437 1925extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
af5bc92d 1926extern void ext4_check_blocks_bitmap(struct super_block *);
617ba13b 1927extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
fd2d4291 1928 ext4_group_t block_group,
ac27a0ec 1929 struct buffer_head ** bh);
617ba13b 1930extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
813e5727
TT
1931
1932extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
1933 ext4_group_t block_group);
1934extern int ext4_wait_block_bitmap(struct super_block *sb,
1935 ext4_group_t block_group,
1936 struct buffer_head *bh);
1937extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
1938 ext4_group_t block_group);
fd034a84
TT
1939extern void ext4_init_block_bitmap(struct super_block *sb,
1940 struct buffer_head *bh,
1941 ext4_group_t group,
1942 struct ext4_group_desc *desc);
cff1dfd7
TT
1943extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
1944 ext4_group_t block_group,
1945 struct ext4_group_desc *gdp);
d5b8f310
TT
1946extern unsigned ext4_num_overhead_clusters(struct super_block *sb,
1947 ext4_group_t block_group,
1948 struct ext4_group_desc *gdp);
f86186b4 1949ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
ac27a0ec
DK
1950
1951/* dir.c */
60fd4da3 1952extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
f7c21177 1953 struct file *,
60fd4da3
TT
1954 struct ext4_dir_entry_2 *,
1955 struct buffer_head *, unsigned int);
f7c21177
TT
1956#define ext4_check_dir_entry(dir, filp, de, bh, offset) \
1957 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
1958 (de), (bh), (offset)))
617ba13b 1959extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
ac27a0ec 1960 __u32 minor_hash,
617ba13b
MC
1961 struct ext4_dir_entry_2 *dirent);
1962extern void ext4_htree_free_dir_info(struct dir_private_info *p);
ac27a0ec
DK
1963
1964/* fsync.c */
02c24a82 1965extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
c278531d 1966extern int ext4_flush_unwritten_io(struct inode *);
ac27a0ec
DK
1967
1968/* hash.c */
617ba13b 1969extern int ext4fs_dirhash(const char *name, int len, struct
ac27a0ec
DK
1970 dx_hash_info *hinfo);
1971
1972/* ialloc.c */
dcca3fec 1973extern struct inode *ext4_new_inode(handle_t *, struct inode *, umode_t,
5cb81dab
DM
1974 const struct qstr *qstr, __u32 goal,
1975 uid_t *owner);
af5bc92d
TT
1976extern void ext4_free_inode(handle_t *, struct inode *);
1977extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
1978extern unsigned long ext4_count_free_inodes(struct super_block *);
1979extern unsigned long ext4_count_dirs(struct super_block *);
1980extern void ext4_check_inodes_bitmap(struct super_block *);
61d08673 1981extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
bfff6873
LC
1982extern int ext4_init_inode_table(struct super_block *sb,
1983 ext4_group_t group, int barrier);
813e5727 1984extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
ac27a0ec 1985
c9de560d
AT
1986/* mballoc.c */
1987extern long ext4_mb_stats;
1988extern long ext4_mb_max_to_scan;
9d99012f 1989extern int ext4_mb_init(struct super_block *);
c9de560d
AT
1990extern int ext4_mb_release(struct super_block *);
1991extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
1992 struct ext4_allocation_request *, int *);
1993extern int ext4_mb_reserve_blocks(struct super_block *, int);
c2ea3fde 1994extern void ext4_discard_preallocations(struct inode *);
5dabfc78
TT
1995extern int __init ext4_init_mballoc(void);
1996extern void ext4_exit_mballoc(void);
44338711 1997extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
e6362609
TT
1998 struct buffer_head *bh, ext4_fsblk_t block,
1999 unsigned long count, int flags);
28623c2f
TT
2000extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2001 ext4_group_t ngroups);
920313a7 2002extern int ext4_mb_add_groupinfo(struct super_block *sb,
5f21b0e6 2003 ext4_group_t i, struct ext4_group_desc *desc);
cc7365df 2004extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2846e820 2005 ext4_fsblk_t block, unsigned long count);
7360d173
LC
2006extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2007
ac27a0ec 2008/* inode.c */
725d26d3
AK
2009struct buffer_head *ext4_getblk(handle_t *, struct inode *,
2010 ext4_lblk_t, int, int *);
2011struct buffer_head *ext4_bread(handle_t *, struct inode *,
2012 ext4_lblk_t, int, int *);
6873fa0d
ES
2013int ext4_get_block(struct inode *inode, sector_t iblock,
2014 struct buffer_head *bh_result, int create);
ac27a0ec 2015
1d1fe1ee 2016extern struct inode *ext4_iget(struct super_block *, unsigned long);
a9185b41 2017extern int ext4_write_inode(struct inode *, struct writeback_control *);
af5bc92d 2018extern int ext4_setattr(struct dentry *, struct iattr *);
3e3398a0
MC
2019extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
2020 struct kstat *stat);
0930fcc1
AV
2021extern void ext4_evict_inode(struct inode *);
2022extern void ext4_clear_inode(struct inode *);
af5bc92d 2023extern int ext4_sync_inode(handle_t *, struct inode *);
aa385729 2024extern void ext4_dirty_inode(struct inode *, int);
617ba13b
MC
2025extern int ext4_change_inode_journal_flag(struct inode *, int);
2026extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
91ef4caf 2027extern int ext4_can_truncate(struct inode *inode);
af5bc92d 2028extern void ext4_truncate(struct inode *);
a4bb6b64 2029extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length);
487caeef 2030extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
617ba13b 2031extern void ext4_set_inode_flags(struct inode *);
ff9ddf7e 2032extern void ext4_get_inode_flags(struct ext4_inode_info *);
ccd2506b 2033extern int ext4_alloc_da_blocks(struct inode *inode);
617ba13b 2034extern void ext4_set_aops(struct inode *inode);
a86c6181 2035extern int ext4_writepage_trans_blocks(struct inode *);
f3bd1f3f 2036extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
4e96b2db
AH
2037extern int ext4_discard_partial_page_buffers(handle_t *handle,
2038 struct address_space *mapping, loff_t from,
2039 loff_t length, int flags);
c2ec175c 2040extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
a9e7f447 2041extern qsize_t *ext4_get_reserved_space(struct inode *inode);
5f634d06
AK
2042extern void ext4_da_update_reserve_space(struct inode *inode,
2043 int used, int quota_claim);
dae1e52c
AG
2044
2045/* indirect.c */
2046extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2047 struct ext4_map_blocks *map, int flags);
2048extern ssize_t ext4_ind_direct_IO(int rw, struct kiocb *iocb,
2049 const struct iovec *iov, loff_t offset,
2050 unsigned long nr_segs);
2051extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2052extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks, int chunk);
ff9893dc
AG
2053extern void ext4_ind_truncate(struct inode *inode);
2054
ac27a0ec 2055/* ioctl.c */
5cdd7b2d 2056extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
af5bc92d 2057extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
ac27a0ec 2058
c14c6fd5 2059/* migrate.c */
2a43a878 2060extern int ext4_ext_migrate(struct inode *);
3d0518f4 2061
ac27a0ec 2062/* namei.c */
b0336e8d
DW
2063extern int ext4_dirent_csum_verify(struct inode *inode,
2064 struct ext4_dir_entry *dirent);
617ba13b
MC
2065extern int ext4_orphan_add(handle_t *, struct inode *);
2066extern int ext4_orphan_del(handle_t *, struct inode *);
2067extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
ac27a0ec
DK
2068 __u32 start_minor_hash, __u32 *next_hash);
2069
2070/* resize.c */
617ba13b
MC
2071extern int ext4_group_add(struct super_block *sb,
2072 struct ext4_new_group_data *input);
2073extern int ext4_group_extend(struct super_block *sb,
2074 struct ext4_super_block *es,
2075 ext4_fsblk_t n_blocks_count);
19c5246d 2076extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
ac27a0ec
DK
2077
2078/* super.c */
952fc18e 2079extern int ext4_calculate_overhead(struct super_block *sb);
a9c47317
DW
2080extern int ext4_superblock_csum_verify(struct super_block *sb,
2081 struct ext4_super_block *es);
06db49e6 2082extern void ext4_superblock_csum_set(struct super_block *sb);
9933fc0a
TT
2083extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2084extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2085extern void ext4_kvfree(void *ptr);
117fff10
TT
2086extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2087 ext4_group_t ngroup);
b9075fa9
JP
2088extern __printf(4, 5)
2089void __ext4_error(struct super_block *, const char *, unsigned int,
2090 const char *, ...);
c398eda0
TT
2091#define ext4_error(sb, message...) __ext4_error(sb, __func__, \
2092 __LINE__, ## message)
b9075fa9
JP
2093extern __printf(5, 6)
2094void ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2095 const char *, ...);
2096extern __printf(5, 6)
2097void ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2098 const char *, ...);
c398eda0
TT
2099extern void __ext4_std_error(struct super_block *, const char *,
2100 unsigned int, int);
b9075fa9
JP
2101extern __printf(4, 5)
2102void __ext4_abort(struct super_block *, const char *, unsigned int,
2103 const char *, ...);
c67d859e 2104#define ext4_abort(sb, message...) __ext4_abort(sb, __func__, \
c398eda0 2105 __LINE__, ## message)
b9075fa9
JP
2106extern __printf(4, 5)
2107void __ext4_warning(struct super_block *, const char *, unsigned int,
2108 const char *, ...);
c398eda0
TT
2109#define ext4_warning(sb, message...) __ext4_warning(sb, __func__, \
2110 __LINE__, ## message)
b9075fa9
JP
2111extern __printf(3, 4)
2112void ext4_msg(struct super_block *, const char *, const char *, ...);
c5e06d10
JL
2113extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2114 const char *, unsigned int, const char *);
2115#define dump_mmp_msg(sb, mmp, msg) __dump_mmp_msg(sb, mmp, __func__, \
2116 __LINE__, msg)
b9075fa9
JP
2117extern __printf(7, 8)
2118void __ext4_grp_locked_error(const char *, unsigned int,
2119 struct super_block *, ext4_group_t,
2120 unsigned long, ext4_fsblk_t,
2121 const char *, ...);
e29136f8
TT
2122#define ext4_grp_locked_error(sb, grp, message...) \
2123 __ext4_grp_locked_error(__func__, __LINE__, (sb), (grp), ## message)
af5bc92d 2124extern void ext4_update_dynamic_rev(struct super_block *sb);
99e6f829
AK
2125extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2126 __u32 compat);
2127extern int ext4_update_rocompat_feature(handle_t *handle,
2128 struct super_block *sb, __u32 rocompat);
2129extern int ext4_update_incompat_feature(handle_t *handle,
2130 struct super_block *sb, __u32 incompat);
8fadc143
AR
2131extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2132 struct ext4_group_desc *bg);
2133extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2134 struct ext4_group_desc *bg);
2135extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2136 struct ext4_group_desc *bg);
021b65bb
TT
2137extern __u32 ext4_free_group_clusters(struct super_block *sb,
2138 struct ext4_group_desc *bg);
560671a0
AK
2139extern __u32 ext4_free_inodes_count(struct super_block *sb,
2140 struct ext4_group_desc *bg);
2141extern __u32 ext4_used_dirs_count(struct super_block *sb,
2142 struct ext4_group_desc *bg);
2143extern __u32 ext4_itable_unused_count(struct super_block *sb,
2144 struct ext4_group_desc *bg);
8fadc143
AR
2145extern void ext4_block_bitmap_set(struct super_block *sb,
2146 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2147extern void ext4_inode_bitmap_set(struct super_block *sb,
2148 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2149extern void ext4_inode_table_set(struct super_block *sb,
2150 struct ext4_group_desc *bg, ext4_fsblk_t blk);
021b65bb
TT
2151extern void ext4_free_group_clusters_set(struct super_block *sb,
2152 struct ext4_group_desc *bg,
2153 __u32 count);
560671a0
AK
2154extern void ext4_free_inodes_set(struct super_block *sb,
2155 struct ext4_group_desc *bg, __u32 count);
2156extern void ext4_used_dirs_set(struct super_block *sb,
2157 struct ext4_group_desc *bg, __u32 count);
2158extern void ext4_itable_unused_set(struct super_block *sb,
2159 struct ext4_group_desc *bg, __u32 count);
feb0ab32 2160extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
bb23c20a 2161 struct ext4_group_desc *gdp);
feb0ab32
DW
2162extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2163 struct ext4_group_desc *gdp);
bd81d8ee 2164
41a246d1
DW
2165static inline int ext4_has_group_desc_csum(struct super_block *sb)
2166{
2167 return EXT4_HAS_RO_COMPAT_FEATURE(sb,
2168 EXT4_FEATURE_RO_COMPAT_GDT_CSUM |
2169 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM);
2170}
bd81d8ee
LV
2171
2172static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2173{
2174 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
6bc9feff 2175 le32_to_cpu(es->s_blocks_count_lo);
bd81d8ee
LV
2176}
2177
2178static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2179{
2180 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
308ba3ec 2181 le32_to_cpu(es->s_r_blocks_count_lo);
bd81d8ee
LV
2182}
2183
2184static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2185{
2186 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
308ba3ec 2187 le32_to_cpu(es->s_free_blocks_count_lo);
bd81d8ee
LV
2188}
2189
2190static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2191 ext4_fsblk_t blk)
2192{
6bc9feff 2193 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
bd81d8ee
LV
2194 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2195}
2196
2197static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2198 ext4_fsblk_t blk)
2199{
308ba3ec 2200 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
bd81d8ee
LV
2201 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2202}
2203
2204static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2205 ext4_fsblk_t blk)
2206{
308ba3ec 2207 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
bd81d8ee
LV
2208 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2209}
2210
a48380f7
AK
2211static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
2212{
06a279d6
TT
2213 if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2214 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2215 le32_to_cpu(raw_inode->i_size_lo);
2216 else
2217 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
a48380f7 2218}
bd81d8ee 2219
a48380f7
AK
2220static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2221{
2222 raw_inode->i_size_lo = cpu_to_le32(i_size);
2223 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2224}
ac27a0ec 2225
c9de560d
AT
2226static inline
2227struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
af5bc92d 2228 ext4_group_t group)
c9de560d
AT
2229{
2230 struct ext4_group_info ***grp_info;
2231 long indexv, indexh;
2232 grp_info = EXT4_SB(sb)->s_group_info;
2233 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2234 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2235 return grp_info[indexv][indexh];
2236}
2237
8df9675f
TT
2238/*
2239 * Reading s_groups_count requires using smp_rmb() afterwards. See
2240 * the locking protocol documented in the comments of ext4_group_add()
2241 * in resize.c
2242 */
2243static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2244{
2245 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
2246
2247 smp_rmb();
2248 return ngroups;
2249}
c9de560d 2250
772cb7c8
JS
2251static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2252 ext4_group_t block_group)
2253{
2254 return block_group >> sbi->s_log_groups_per_flex;
2255}
2256
2257static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2258{
2259 return 1 << sbi->s_log_groups_per_flex;
2260}
2261
617ba13b 2262#define ext4_std_error(sb, errno) \
ac27a0ec
DK
2263do { \
2264 if ((errno)) \
c398eda0 2265 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
ac27a0ec
DK
2266} while (0)
2267
a30d542a 2268#ifdef CONFIG_SMP
df55c99d
TT
2269/* Each CPU can accumulate percpu_counter_batch clusters in their local
2270 * counters. So we need to make sure we have free clusters more
179f7ebf 2271 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
a30d542a 2272 */
df55c99d 2273#define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
a30d542a 2274#else
df55c99d 2275#define EXT4_FREECLUSTERS_WATERMARK 0
a30d542a
AK
2276#endif
2277
cf17fea6
AK
2278static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2279{
2280 /*
2281 * XXX: replace with spinlock if seen contended -bzzz
2282 */
2283 down_write(&EXT4_I(inode)->i_data_sem);
2284 if (newsize > EXT4_I(inode)->i_disksize)
2285 EXT4_I(inode)->i_disksize = newsize;
2286 up_write(&EXT4_I(inode)->i_data_sem);
2287 return ;
2288}
a30d542a 2289
5d1b1b3f
AK
2290struct ext4_group_info {
2291 unsigned long bb_state;
2292 struct rb_root bb_free_root;
a36b4498
ES
2293 ext4_grpblk_t bb_first_free; /* first free block */
2294 ext4_grpblk_t bb_free; /* total free blocks */
2295 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
8a57d9d6 2296 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
5d1b1b3f
AK
2297 struct list_head bb_prealloc_list;
2298#ifdef DOUBLE_CHECK
2299 void *bb_bitmap;
2300#endif
2301 struct rw_semaphore alloc_sem;
a36b4498
ES
2302 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
2303 * regions, index is order.
2304 * bb_counters[3] = 5 means
2305 * 5 free 8-block regions. */
5d1b1b3f
AK
2306};
2307
3d56b8d2
TM
2308#define EXT4_GROUP_INFO_NEED_INIT_BIT 0
2309#define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
5d1b1b3f
AK
2310
2311#define EXT4_MB_GRP_NEED_INIT(grp) \
2312 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2313
3d56b8d2
TM
2314#define EXT4_MB_GRP_WAS_TRIMMED(grp) \
2315 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2316#define EXT4_MB_GRP_SET_TRIMMED(grp) \
2317 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2318#define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
2319 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2320
50797481
TT
2321#define EXT4_MAX_CONTENTION 8
2322#define EXT4_CONTENTION_THRESHOLD 2
2323
955ce5f5
AK
2324static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2325 ext4_group_t group)
5d1b1b3f 2326{
955ce5f5
AK
2327 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2328}
5d1b1b3f 2329
50797481
TT
2330/*
2331 * Returns true if the filesystem is busy enough that attempts to
2332 * access the block group locks has run into contention.
2333 */
2334static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2335{
2336 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2337}
2338
955ce5f5
AK
2339static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2340{
50797481
TT
2341 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2342 if (spin_trylock(lock))
2343 /*
2344 * We're able to grab the lock right away, so drop the
2345 * lock contention counter.
2346 */
2347 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2348 else {
2349 /*
2350 * The lock is busy, so bump the contention counter,
2351 * and then wait on the spin lock.
2352 */
2353 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2354 EXT4_MAX_CONTENTION);
2355 spin_lock(lock);
2356 }
5d1b1b3f
AK
2357}
2358
2359static inline void ext4_unlock_group(struct super_block *sb,
2360 ext4_group_t group)
2361{
955ce5f5 2362 spin_unlock(ext4_group_lock_ptr(sb, group));
5d1b1b3f
AK
2363}
2364
1f7d1e77
TT
2365/*
2366 * Block validity checking
2367 */
2368#define ext4_check_indirect_blockref(inode, bh) \
2369 ext4_check_blockref(__func__, __LINE__, inode, \
2370 (__le32 *)(bh)->b_data, \
2371 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2372
2373#define ext4_ind_check_inode(inode) \
2374 ext4_check_blockref(__func__, __LINE__, inode, \
2375 EXT4_I(inode)->i_data, \
2376 EXT4_NDIR_BLOCKS)
2377
ac27a0ec
DK
2378/*
2379 * Inodes and files operations
2380 */
2381
2382/* dir.c */
617ba13b 2383extern const struct file_operations ext4_dir_operations;
ac27a0ec
DK
2384
2385/* file.c */
c5ef1c42 2386extern const struct inode_operations ext4_file_inode_operations;
617ba13b 2387extern const struct file_operations ext4_file_operations;
e0d10bfa 2388extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
c278531d 2389extern void ext4_unwritten_wait(struct inode *inode);
ac27a0ec
DK
2390
2391/* namei.c */
c5ef1c42
AV
2392extern const struct inode_operations ext4_dir_inode_operations;
2393extern const struct inode_operations ext4_special_inode_operations;
596397b7 2394extern struct dentry *ext4_get_parent(struct dentry *child);
ac27a0ec
DK
2395
2396/* symlink.c */
c5ef1c42
AV
2397extern const struct inode_operations ext4_symlink_inode_operations;
2398extern const struct inode_operations ext4_fast_symlink_inode_operations;
ac27a0ec 2399
6fd058f7
TT
2400/* block_validity */
2401extern void ext4_release_system_zone(struct super_block *sb);
2402extern int ext4_setup_system_zone(struct super_block *sb);
5dabfc78
TT
2403extern int __init ext4_init_system_zone(void);
2404extern void ext4_exit_system_zone(void);
6fd058f7
TT
2405extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
2406 ext4_fsblk_t start_blk,
2407 unsigned int count);
1f7d1e77
TT
2408extern int ext4_check_blockref(const char *, unsigned int,
2409 struct inode *, __le32 *, unsigned int);
6fd058f7 2410
a86c6181 2411/* extents.c */
4a092d73
TT
2412struct ext4_ext_path;
2413struct ext4_extent;
2414
a86c6181
AT
2415extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
2416extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
a02908f1
MC
2417extern int ext4_ext_index_trans_blocks(struct inode *inode, int nrblocks,
2418 int chunk);
e35fd660
TT
2419extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
2420 struct ext4_map_blocks *map, int flags);
cf108bca 2421extern void ext4_ext_truncate(struct inode *);
a4bb6b64
AH
2422extern int ext4_ext_punch_hole(struct file *file, loff_t offset,
2423 loff_t length);
a86c6181
AT
2424extern void ext4_ext_init(struct super_block *);
2425extern void ext4_ext_release(struct super_block *);
2fe17c10 2426extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
a2df2a63 2427 loff_t len);
0031462b 2428extern int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset,
a1de02dc 2429 ssize_t len);
e35fd660
TT
2430extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
2431 struct ext4_map_blocks *map, int flags);
4a092d73
TT
2432extern int ext4_ext_calc_metadata_amount(struct inode *inode,
2433 ext4_lblk_t lblocks);
2434extern int ext4_extent_tree_init(handle_t *, struct inode *);
2435extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
2436 int num,
2437 struct ext4_ext_path *path);
2438extern int ext4_can_extents_be_merged(struct inode *inode,
2439 struct ext4_extent *ex1,
2440 struct ext4_extent *ex2);
2441extern int ext4_ext_insert_extent(handle_t *, struct inode *,
2442 struct ext4_ext_path *,
2443 struct ext4_extent *, int);
2444extern struct ext4_ext_path *ext4_ext_find_extent(struct inode *, ext4_lblk_t,
2445 struct ext4_ext_path *);
2446extern void ext4_ext_drop_refs(struct ext4_ext_path *);
2447extern int ext4_ext_check_inode(struct inode *inode);
2448extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk);
3a06d778
AK
2449extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
2450 __u64 start, __u64 len);
4a092d73
TT
2451
2452
748de673
AF
2453/* move_extent.c */
2454extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
2455 __u64 start_orig, __u64 start_donor,
2456 __u64 len, __u64 *moved_len);
2457
bd2d0210 2458/* page-io.c */
5dabfc78 2459extern int __init ext4_init_pageio(void);
28a535f9 2460extern void ext4_add_complete_io(ext4_io_end_t *io_end);
5dabfc78 2461extern void ext4_exit_pageio(void);
f7ad6d2e 2462extern void ext4_ioend_wait(struct inode *);
bd2d0210
TT
2463extern void ext4_free_io_end(ext4_io_end_t *io);
2464extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
bd2d0210
TT
2465extern void ext4_io_submit(struct ext4_io_submit *io);
2466extern int ext4_bio_write_page(struct ext4_io_submit *io,
2467 struct page *page,
2468 int len,
2469 struct writeback_control *wbc);
3a06d778 2470
c5e06d10
JL
2471/* mmp.c */
2472extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
5c359a47
DW
2473extern void ext4_mmp_csum_set(struct super_block *sb, struct mmp_struct *mmp);
2474extern int ext4_mmp_csum_verify(struct super_block *sb,
2475 struct mmp_struct *mmp);
c5e06d10 2476
744692dc
JZ
2477/* BH_Uninit flag: blocks are allocated but uninitialized on disk */
2478enum ext4_state_bits {
2479 BH_Uninit /* blocks are allocated but uninitialized on disk */
2480 = BH_JBDPrivateStart,
7b415bf6
AK
2481 BH_AllocFromCluster, /* allocated blocks were part of already
2482 * allocated cluster. Note that this flag will
2483 * never, ever appear in a buffer_head's state
2484 * flag. See EXT4_MAP_FROM_CLUSTER to see where
2485 * this is used. */
744692dc
JZ
2486};
2487
2488BUFFER_FNS(Uninit, uninit)
2489TAS_BUFFER_FNS(Uninit, uninit)
2490
2ccb5fb9
AK
2491/*
2492 * Add new method to test wether block and inode bitmaps are properly
2493 * initialized. With uninit_bg reading the block from disk is not enough
2494 * to mark the bitmap uptodate. We need to also zero-out the bitmap
2495 */
2496#define BH_BITMAP_UPTODATE BH_JBDPrivateStart
2497
2498static inline int bitmap_uptodate(struct buffer_head *bh)
2499{
2500 return (buffer_uptodate(bh) &&
2501 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
2502}
2503static inline void set_bitmap_uptodate(struct buffer_head *bh)
2504{
2505 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
2506}
2507
17335dcc
DM
2508/*
2509 * Disable DIO read nolock optimization, so new dioreaders will be forced
2510 * to grab i_mutex
2511 */
2512static inline void ext4_inode_block_unlocked_dio(struct inode *inode)
2513{
2514 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
2515 smp_mb();
2516}
2517static inline void ext4_inode_resume_unlocked_dio(struct inode *inode)
2518{
2519 smp_mb();
2520 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
2521}
2522
731eb1a0
AM
2523#define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
2524
e9e3bcec
ES
2525/* For ioend & aio unwritten conversion wait queues */
2526#define EXT4_WQ_HASH_SZ 37
2527#define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
2528 EXT4_WQ_HASH_SZ])
2529#define ext4_aio_mutex(v) (&ext4__aio_mutex[((unsigned long)(v)) %\
2530 EXT4_WQ_HASH_SZ])
2531extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
2532extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ];
2533
8f82f840
YY
2534#define EXT4_RESIZING 0
2535extern int ext4_resize_begin(struct super_block *sb);
2536extern void ext4_resize_end(struct super_block *sb);
2537
ac27a0ec
DK
2538#endif /* __KERNEL__ */
2539
3dcf5451 2540#endif /* _EXT4_H */