1 // SPDX-License-Identifier: GPL-2.0
2 /*
3 * ext4.h
4 *
5 * Copyright (C) 1992, 1993, 1994, 1995
6 * Remy Card (card@masi.ibp.fr)
7 * Laboratoire MASI - Institut Blaise Pascal
8 * Universite Pierre et Marie Curie (Paris VI)
9 *
10 * from
11 *
12 * linux/include/linux/minix_fs.h
13 *
14 * Copyright (C) 1991, 1992 Linus Torvalds
15 */
16
17 #ifndef _EXT4_H
18 #define _EXT4_H
19
20 #include <linux/types.h>
21 #include <linux/blkdev.h>
22 #include <linux/magic.h>
23 #include <linux/jbd2.h>
24 #include <linux/quota.h>
25 #include <linux/rwsem.h>
26 #include <linux/rbtree.h>
27 #include <linux/seqlock.h>
28 #include <linux/mutex.h>
29 #include <linux/timer.h>
30 #include <linux/wait.h>
31 #include <linux/sched/signal.h>
32 #include <linux/blockgroup_lock.h>
33 #include <linux/percpu_counter.h>
34 #include <linux/ratelimit.h>
35 #include <crypto/hash.h>
36 #include <linux/falloc.h>
37 #include <linux/percpu-rwsem.h>
38 #include <linux/fiemap.h>
39 #ifdef __KERNEL__
40 #include <linux/compat.h>
41 #endif
42
43 #include <linux/fscrypt.h>
44 #include <linux/fsverity.h>
45
46 #include <linux/compiler.h>
47
48 /*
49 * The fourth extended filesystem constants/structures
50 */
51
52 /*
53 * with AGGRESSIVE_CHECK allocator runs consistency checks over
54 * structures. these checks slow things down a lot
55 */
56 #define AGGRESSIVE_CHECK__
57
58 /*
59 * with DOUBLE_CHECK defined mballoc creates persistent in-core
60 * bitmaps, maintains and uses them to check for double allocations
61 */
62 #define DOUBLE_CHECK__
63
64 /*
65 * Define EXT4FS_DEBUG to produce debug messages
66 */
67 #undef EXT4FS_DEBUG
68
69 /*
70 * Debug code
71 */
72 #ifdef EXT4FS_DEBUG
73 #define ext4_debug(f, a...) \
74 do { \
75 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
76 __FILE__, __LINE__, __func__); \
77 printk(KERN_DEBUG f, ## a); \
78 } while (0)
79 #else
80 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
81 #endif
82
83 /*
84 * Turn on EXT_DEBUG to enable ext4_ext_show_path/leaf/move in extents.c
85 */
86 #define EXT_DEBUG__
87
88 /*
89 * Dynamic printk for controlled extents debugging.
90 */
91 #ifdef CONFIG_EXT4_DEBUG
92 #define ext_debug(ino, fmt, ...) \
93 pr_debug("[%s/%d] EXT4-fs (%s): ino %lu: (%s, %d): %s:" fmt, \
94 current->comm, task_pid_nr(current), \
95 ino->i_sb->s_id, ino->i_ino, __FILE__, __LINE__, \
96 __func__, ##__VA_ARGS__)
97 #else
98 #define ext_debug(ino, fmt, ...) no_printk(fmt, ##__VA_ARGS__)
99 #endif
100
101 /* data type for block offset of block group */
102 typedef int ext4_grpblk_t;
103
104 /* data type for filesystem-wide blocks number */
105 typedef unsigned long long ext4_fsblk_t;
106
107 /* data type for file logical block number */
108 typedef __u32 ext4_lblk_t;
109
110 /* data type for block group number */
111 typedef unsigned int ext4_group_t;
112
113 enum SHIFT_DIRECTION {
114 SHIFT_LEFT = 0,
115 SHIFT_RIGHT,
116 };
117
118 /*
119 * Flags used in mballoc's allocation_context flags field.
120 *
121 * Also used to show what's going on for debugging purposes when the
122 * flag field is exported via the traceport interface
123 */
124
125 /* prefer goal again. length */
126 #define EXT4_MB_HINT_MERGE 0x0001
127 /* blocks already reserved */
128 #define EXT4_MB_HINT_RESERVED 0x0002
129 /* metadata is being allocated */
130 #define EXT4_MB_HINT_METADATA 0x0004
131 /* first blocks in the file */
132 #define EXT4_MB_HINT_FIRST 0x0008
133 /* search for the best chunk */
134 #define EXT4_MB_HINT_BEST 0x0010
135 /* data is being allocated */
136 #define EXT4_MB_HINT_DATA 0x0020
137 /* don't preallocate (for tails) */
138 #define EXT4_MB_HINT_NOPREALLOC 0x0040
139 /* allocate for locality group */
140 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080
141 /* allocate goal blocks or none */
142 #define EXT4_MB_HINT_GOAL_ONLY 0x0100
143 /* goal is meaningful */
144 #define EXT4_MB_HINT_TRY_GOAL 0x0200
145 /* blocks already pre-reserved by delayed allocation */
146 #define EXT4_MB_DELALLOC_RESERVED 0x0400
147 /* We are doing stream allocation */
148 #define EXT4_MB_STREAM_ALLOC 0x0800
149 /* Use reserved root blocks if needed */
150 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000
151 /* Use blocks from reserved pool */
152 #define EXT4_MB_USE_RESERVED 0x2000
153 /* Do strict check for free blocks while retrying block allocation */
154 #define EXT4_MB_STRICT_CHECK 0x4000
155
156 struct ext4_allocation_request {
157 /* target inode for block we're allocating */
158 struct inode *inode;
159 /* how many blocks we want to allocate */
160 unsigned int len;
161 /* logical block in target inode */
162 ext4_lblk_t logical;
163 /* the closest logical allocated block to the left */
164 ext4_lblk_t lleft;
165 /* the closest logical allocated block to the right */
166 ext4_lblk_t lright;
167 /* phys. target (a hint) */
168 ext4_fsblk_t goal;
169 /* phys. block for the closest logical allocated block to the left */
170 ext4_fsblk_t pleft;
171 /* phys. block for the closest logical allocated block to the right */
172 ext4_fsblk_t pright;
173 /* flags. see above EXT4_MB_HINT_* */
174 unsigned int flags;
175 };
176
177 /*
178 * Logical to physical block mapping, used by ext4_map_blocks()
179 *
180 * This structure is used to pass requests into ext4_map_blocks() as
181 * well as to store the information returned by ext4_map_blocks(). It
182 * takes less room on the stack than a struct buffer_head.
183 */
184 #define EXT4_MAP_NEW BIT(BH_New)
185 #define EXT4_MAP_MAPPED BIT(BH_Mapped)
186 #define EXT4_MAP_UNWRITTEN BIT(BH_Unwritten)
187 #define EXT4_MAP_BOUNDARY BIT(BH_Boundary)
188 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
189 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
190
191 struct ext4_map_blocks {
192 ext4_fsblk_t m_pblk;
193 ext4_lblk_t m_lblk;
194 unsigned int m_len;
195 unsigned int m_flags;
196 };
197
198 /*
199 * Block validity checking, system zone rbtree.
200 */
201 struct ext4_system_blocks {
202 struct rb_root root;
203 struct rcu_head rcu;
204 };
205
206 /*
207 * Flags for ext4_io_end->flags
208 */
209 #define EXT4_IO_END_UNWRITTEN 0x0001
210
211 struct ext4_io_end_vec {
212 struct list_head list; /* list of io_end_vec */
213 loff_t offset; /* offset in the file */
214 ssize_t size; /* size of the extent */
215 };
216
217 /*
218 * For converting unwritten extents on a work queue. 'handle' is used for
219 * buffered writeback.
220 */
221 typedef struct ext4_io_end {
222 struct list_head list; /* per-file finished IO list */
223 handle_t *handle; /* handle reserved for extent
224 * conversion */
225 struct inode *inode; /* file being written to */
226 struct bio *bio; /* Linked list of completed
227 * bios covering the extent */
228 unsigned int flag; /* unwritten or not */
229 atomic_t count; /* reference counter */
230 struct list_head list_vec; /* list of ext4_io_end_vec */
231 } ext4_io_end_t;
232
233 struct ext4_io_submit {
234 struct writeback_control *io_wbc;
235 struct bio *io_bio;
236 ext4_io_end_t *io_end;
237 sector_t io_next_block;
238 };
239
240 /*
241 * Special inodes numbers
242 */
243 #define EXT4_BAD_INO 1 /* Bad blocks inode */
244 #define EXT4_ROOT_INO 2 /* Root inode */
245 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */
246 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
247 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
248 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
249 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
250 #define EXT4_JOURNAL_INO 8 /* Journal inode */
251
252 /* First non-reserved inode for old ext4 filesystems */
253 #define EXT4_GOOD_OLD_FIRST_INO 11
254
255 /*
256 * Maximal count of links to a file
257 */
258 #define EXT4_LINK_MAX 65000
259
260 /*
261 * Macro-instructions used to manage several block sizes
262 */
263 #define EXT4_MIN_BLOCK_SIZE 1024
264 #define EXT4_MAX_BLOCK_SIZE 65536
265 #define EXT4_MIN_BLOCK_LOG_SIZE 10
266 #define EXT4_MAX_BLOCK_LOG_SIZE 16
267 #define EXT4_MAX_CLUSTER_LOG_SIZE 30
268 #ifdef __KERNEL__
269 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
270 #else
271 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
272 #endif
273 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
274 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
275 EXT4_SB(s)->s_cluster_bits)
276 #ifdef __KERNEL__
277 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
278 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
279 #else
280 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
281 #endif
282 #ifdef __KERNEL__
283 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
284 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
285 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
286 #else
287 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
288 EXT4_GOOD_OLD_INODE_SIZE : \
289 (s)->s_inode_size)
290 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
291 EXT4_GOOD_OLD_FIRST_INO : \
292 (s)->s_first_ino)
293 #endif
294 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
295 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \
296 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \
297 blkbits))
298
299 /* Translate a block number to a cluster number */
300 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
301 /* Translate a cluster number to a block number */
302 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
303 /* Translate # of blks to # of clusters */
304 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
305 (sbi)->s_cluster_bits)
306 /* Mask out the low bits to get the starting block of the cluster */
307 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
308 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
309 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
310 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
311 /* Fill in the low bits to get the last block of the cluster */
312 #define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) | \
313 ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1))
314 /* Get the cluster offset */
315 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
316 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
317 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
318 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
319
320 /*
321 * Structure of a blocks group descriptor
322 */
323 struct ext4_group_desc
324 {
325 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
326 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
327 __le32 bg_inode_table_lo; /* Inodes table block */
328 __le16 bg_free_blocks_count_lo;/* Free blocks count */
329 __le16 bg_free_inodes_count_lo;/* Free inodes count */
330 __le16 bg_used_dirs_count_lo; /* Directories count */
331 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
332 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
333 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
334 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
335 __le16 bg_itable_unused_lo; /* Unused inodes count */
336 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
337 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
338 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
339 __le32 bg_inode_table_hi; /* Inodes table block MSB */
340 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
341 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
342 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
343 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
344 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
345 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
346 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
347 __u32 bg_reserved;
348 };
349
350 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
351 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
352 sizeof(__le16))
353 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
354 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
355 sizeof(__le16))
356
357 /*
358 * Structure of a flex block group info
359 */
360
361 struct flex_groups {
362 atomic64_t free_clusters;
363 atomic_t free_inodes;
364 atomic_t used_dirs;
365 };
366
367 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
368 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
369 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
370
371 /*
372 * Macro-instructions used to manage group descriptors
373 */
374 #define EXT4_MIN_DESC_SIZE 32
375 #define EXT4_MIN_DESC_SIZE_64BIT 64
376 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
377 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
378 #ifdef __KERNEL__
379 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
380 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
381 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
382 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
383 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
384 #else
385 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
386 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
387 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
388 #endif
389
390 /*
391 * Constants relative to the data blocks
392 */
393 #define EXT4_NDIR_BLOCKS 12
394 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
395 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
396 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
397 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
398
399 /*
400 * Inode flags
401 */
402 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
403 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
404 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
405 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
406 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
407 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
408 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
409 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
410 /* Reserved for compression usage... */
411 #define EXT4_DIRTY_FL 0x00000100
412 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
413 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
414 /* nb: was previously EXT2_ECOMPR_FL */
415 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */
416 /* End compression flags --- maybe not all used */
417 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
418 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
419 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
420 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
421 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
422 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
423 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
424 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
425 #define EXT4_VERITY_FL 0x00100000 /* Verity protected inode */
426 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
427 /* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */
428
429 #define EXT4_DAX_FL 0x02000000 /* Inode is DAX */
430
431 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
432 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
433 #define EXT4_CASEFOLD_FL 0x40000000 /* Casefolded directory */
434 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
435
436 /* User modifiable flags */
437 #define EXT4_FL_USER_MODIFIABLE (EXT4_SECRM_FL | \
438 EXT4_UNRM_FL | \
439 EXT4_COMPR_FL | \
440 EXT4_SYNC_FL | \
441 EXT4_IMMUTABLE_FL | \
442 EXT4_APPEND_FL | \
443 EXT4_NODUMP_FL | \
444 EXT4_NOATIME_FL | \
445 EXT4_JOURNAL_DATA_FL | \
446 EXT4_NOTAIL_FL | \
447 EXT4_DIRSYNC_FL | \
448 EXT4_TOPDIR_FL | \
449 EXT4_EXTENTS_FL | \
450 0x00400000 /* EXT4_EOFBLOCKS_FL */ | \
451 EXT4_DAX_FL | \
452 EXT4_PROJINHERIT_FL | \
453 EXT4_CASEFOLD_FL)
454
455 /* User visible flags */
456 #define EXT4_FL_USER_VISIBLE (EXT4_FL_USER_MODIFIABLE | \
457 EXT4_DIRTY_FL | \
458 EXT4_COMPRBLK_FL | \
459 EXT4_NOCOMPR_FL | \
460 EXT4_ENCRYPT_FL | \
461 EXT4_INDEX_FL | \
462 EXT4_VERITY_FL | \
463 EXT4_INLINE_DATA_FL)
464
465 /* Flags we can manipulate with through FS_IOC_FSSETXATTR */
466 #define EXT4_FL_XFLAG_VISIBLE (EXT4_SYNC_FL | \
467 EXT4_IMMUTABLE_FL | \
468 EXT4_APPEND_FL | \
469 EXT4_NODUMP_FL | \
470 EXT4_NOATIME_FL | \
471 EXT4_PROJINHERIT_FL | \
472 EXT4_DAX_FL)
473
474 /* Flags that should be inherited by new inodes from their parent. */
475 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
476 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
477 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
478 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
479 EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL |\
480 EXT4_DAX_FL)
481
482 /* Flags that are appropriate for regular files (all but dir-specific ones). */
483 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\
484 EXT4_PROJINHERIT_FL))
485
486 /* Flags that are appropriate for non-directories/regular files. */
487 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
488
489 /* The only flags that should be swapped */
490 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL)
491
492 /* Flags which are mutually exclusive to DAX */
493 #define EXT4_DAX_MUT_EXCL (EXT4_VERITY_FL | EXT4_ENCRYPT_FL |\
494 EXT4_JOURNAL_DATA_FL | EXT4_INLINE_DATA_FL)
495
496 /* Mask out flags that are inappropriate for the given type of inode. */
ext4_mask_flags(umode_t mode,__u32 flags)497 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
498 {
499 if (S_ISDIR(mode))
500 return flags;
501 else if (S_ISREG(mode))
502 return flags & EXT4_REG_FLMASK;
503 else
504 return flags & EXT4_OTHER_FLMASK;
505 }
506
507 /*
508 * Inode flags used for atomic set/get
509 */
510 enum {
511 EXT4_INODE_SECRM = 0, /* Secure deletion */
512 EXT4_INODE_UNRM = 1, /* Undelete */
513 EXT4_INODE_COMPR = 2, /* Compress file */
514 EXT4_INODE_SYNC = 3, /* Synchronous updates */
515 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
516 EXT4_INODE_APPEND = 5, /* writes to file may only append */
517 EXT4_INODE_NODUMP = 6, /* do not dump file */
518 EXT4_INODE_NOATIME = 7, /* do not update atime */
519 /* Reserved for compression usage... */
520 EXT4_INODE_DIRTY = 8,
521 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
522 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
523 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */
524 /* End compression flags --- maybe not all used */
525 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
526 EXT4_INODE_IMAGIC = 13, /* AFS directory */
527 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
528 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
529 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
530 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
531 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
532 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
533 EXT4_INODE_VERITY = 20, /* Verity protected inode */
534 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
535 /* 22 was formerly EXT4_INODE_EOFBLOCKS */
536 EXT4_INODE_DAX = 25, /* Inode is DAX */
537 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
538 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */
539 EXT4_INODE_CASEFOLD = 30, /* Casefolded directory */
540 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
541 };
542
543 /*
544 * Since it's pretty easy to mix up bit numbers and hex values, we use a
545 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
546 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
547 * any extra space in the compiled kernel image, otherwise, the build will fail.
548 * It's important that these values are the same, since we are using
549 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
550 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
551 * values found in ext2, ext3 and ext4 filesystems, and of course the values
552 * defined in e2fsprogs.
553 *
554 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
555 */
556 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1U << EXT4_INODE_##FLAG))
557 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
558
ext4_check_flag_values(void)559 static inline void ext4_check_flag_values(void)
560 {
561 CHECK_FLAG_VALUE(SECRM);
562 CHECK_FLAG_VALUE(UNRM);
563 CHECK_FLAG_VALUE(COMPR);
564 CHECK_FLAG_VALUE(SYNC);
565 CHECK_FLAG_VALUE(IMMUTABLE);
566 CHECK_FLAG_VALUE(APPEND);
567 CHECK_FLAG_VALUE(NODUMP);
568 CHECK_FLAG_VALUE(NOATIME);
569 CHECK_FLAG_VALUE(DIRTY);
570 CHECK_FLAG_VALUE(COMPRBLK);
571 CHECK_FLAG_VALUE(NOCOMPR);
572 CHECK_FLAG_VALUE(ENCRYPT);
573 CHECK_FLAG_VALUE(INDEX);
574 CHECK_FLAG_VALUE(IMAGIC);
575 CHECK_FLAG_VALUE(JOURNAL_DATA);
576 CHECK_FLAG_VALUE(NOTAIL);
577 CHECK_FLAG_VALUE(DIRSYNC);
578 CHECK_FLAG_VALUE(TOPDIR);
579 CHECK_FLAG_VALUE(HUGE_FILE);
580 CHECK_FLAG_VALUE(EXTENTS);
581 CHECK_FLAG_VALUE(VERITY);
582 CHECK_FLAG_VALUE(EA_INODE);
583 CHECK_FLAG_VALUE(INLINE_DATA);
584 CHECK_FLAG_VALUE(PROJINHERIT);
585 CHECK_FLAG_VALUE(CASEFOLD);
586 CHECK_FLAG_VALUE(RESERVED);
587 }
588
589 /* Used to pass group descriptor data when online resize is done */
590 struct ext4_new_group_input {
591 __u32 group; /* Group number for this data */
592 __u64 block_bitmap; /* Absolute block number of block bitmap */
593 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
594 __u64 inode_table; /* Absolute block number of inode table start */
595 __u32 blocks_count; /* Total number of blocks in this group */
596 __u16 reserved_blocks; /* Number of reserved blocks in this group */
597 __u16 unused;
598 };
599
600 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
601 struct compat_ext4_new_group_input {
602 u32 group;
603 compat_u64 block_bitmap;
604 compat_u64 inode_bitmap;
605 compat_u64 inode_table;
606 u32 blocks_count;
607 u16 reserved_blocks;
608 u16 unused;
609 };
610 #endif
611
612 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
613 struct ext4_new_group_data {
614 __u32 group;
615 __u64 block_bitmap;
616 __u64 inode_bitmap;
617 __u64 inode_table;
618 __u32 blocks_count;
619 __u16 reserved_blocks;
620 __u16 mdata_blocks;
621 __u32 free_clusters_count;
622 };
623
624 /* Indexes used to index group tables in ext4_new_group_data */
625 enum {
626 BLOCK_BITMAP = 0, /* block bitmap */
627 INODE_BITMAP, /* inode bitmap */
628 INODE_TABLE, /* inode tables */
629 GROUP_TABLE_COUNT,
630 };
631
632 /*
633 * Flags used by ext4_map_blocks()
634 */
635 /* Allocate any needed blocks and/or convert an unwritten
636 extent to be an initialized ext4 */
637 #define EXT4_GET_BLOCKS_CREATE 0x0001
638 /* Request the creation of an unwritten extent */
639 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002
640 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\
641 EXT4_GET_BLOCKS_CREATE)
642 /* Caller is from the delayed allocation writeout path
643 * finally doing the actual allocation of delayed blocks */
644 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
645 /* caller is from the direct IO path, request to creation of an
646 unwritten extents if not allocated, split the unwritten
647 extent if blocks has been preallocated already*/
648 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
649 #define EXT4_GET_BLOCKS_CONVERT 0x0010
650 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
651 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
652 /* Convert extent to initialized after IO complete */
653 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
654 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
655 /* Eventual metadata allocation (due to growing extent tree)
656 * should not fail, so try to use reserved blocks for that.*/
657 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
658 /* Don't normalize allocation size (used for fallocate) */
659 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
660 /* Convert written extents to unwritten */
661 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100
662 /* Write zeros to newly created written extents */
663 #define EXT4_GET_BLOCKS_ZERO 0x0200
664 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\
665 EXT4_GET_BLOCKS_ZERO)
666 /* Caller will submit data before dropping transaction handle. This
667 * allows jbd2 to avoid submitting data before commit. */
668 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400
669
670 /*
671 * The bit position of these flags must not overlap with any of the
672 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(),
673 * read_extent_tree_block(), ext4_split_extent_at(),
674 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
675 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
676 * caching the extents when reading from the extent tree while a
677 * truncate or punch hole operation is in progress.
678 */
679 #define EXT4_EX_NOCACHE 0x40000000
680 #define EXT4_EX_FORCE_CACHE 0x20000000
681 #define EXT4_EX_NOFAIL 0x10000000
682
683 /*
684 * Flags used by ext4_free_blocks
685 */
686 #define EXT4_FREE_BLOCKS_METADATA 0x0001
687 #define EXT4_FREE_BLOCKS_FORGET 0x0002
688 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
689 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
690 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
691 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
692 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040
693
694 /*
695 * ioctl commands
696 */
697 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
698 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
699 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
700 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
701 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
702 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
703 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
704 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
705 #define EXT4_IOC_MIGRATE _IO('f', 9)
706 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
707 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
708 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
709 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
710 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
711 #define EXT4_IOC_SWAP_BOOT _IO('f', 17)
712 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18)
713 /* ioctl codes 19--39 are reserved for fscrypt */
714 #define EXT4_IOC_CLEAR_ES_CACHE _IO('f', 40)
715 #define EXT4_IOC_GETSTATE _IOW('f', 41, __u32)
716 #define EXT4_IOC_GET_ES_CACHE _IOWR('f', 42, struct fiemap)
717
718 #define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32)
719
720 /*
721 * Flags for going down operation
722 */
723 #define EXT4_GOING_FLAGS_DEFAULT 0x0 /* going down */
724 #define EXT4_GOING_FLAGS_LOGFLUSH 0x1 /* flush log but not data */
725 #define EXT4_GOING_FLAGS_NOLOGFLUSH 0x2 /* don't flush log nor data */
726
727 /*
728 * Flags returned by EXT4_IOC_GETSTATE
729 *
730 * We only expose to userspace a subset of the state flags in
731 * i_state_flags
732 */
733 #define EXT4_STATE_FLAG_EXT_PRECACHED 0x00000001
734 #define EXT4_STATE_FLAG_NEW 0x00000002
735 #define EXT4_STATE_FLAG_NEWENTRY 0x00000004
736 #define EXT4_STATE_FLAG_DA_ALLOC_CLOSE 0x00000008
737
738 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
739 /*
740 * ioctl commands in 32 bit emulation
741 */
742 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
743 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
744 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
745 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
746 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
747 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
748 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
749 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
750 #endif
751
752 /*
753 * Returned by EXT4_IOC_GET_ES_CACHE as an additional possible flag.
754 * It indicates that the entry in extent status cache is for a hole.
755 */
756 #define EXT4_FIEMAP_EXTENT_HOLE 0x08000000
757
758 /* Max physical block we can address w/o extents */
759 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
760
761 /* Max logical block we can support */
762 #define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFE
763
764 /*
765 * Structure of an inode on the disk
766 */
767 struct ext4_inode {
768 __le16 i_mode; /* File mode */
769 __le16 i_uid; /* Low 16 bits of Owner Uid */
770 __le32 i_size_lo; /* Size in bytes */
771 __le32 i_atime; /* Access time */
772 __le32 i_ctime; /* Inode Change time */
773 __le32 i_mtime; /* Modification time */
774 __le32 i_dtime; /* Deletion Time */
775 __le16 i_gid; /* Low 16 bits of Group Id */
776 __le16 i_links_count; /* Links count */
777 __le32 i_blocks_lo; /* Blocks count */
778 __le32 i_flags; /* File flags */
779 union {
780 struct {
781 __le32 l_i_version;
782 } linux1;
783 struct {
784 __u32 h_i_translator;
785 } hurd1;
786 struct {
787 __u32 m_i_reserved1;
788 } masix1;
789 } osd1; /* OS dependent 1 */
790 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
791 __le32 i_generation; /* File version (for NFS) */
792 __le32 i_file_acl_lo; /* File ACL */
793 __le32 i_size_high;
794 __le32 i_obso_faddr; /* Obsoleted fragment address */
795 union {
796 struct {
797 __le16 l_i_blocks_high; /* were l_i_reserved1 */
798 __le16 l_i_file_acl_high;
799 __le16 l_i_uid_high; /* these 2 fields */
800 __le16 l_i_gid_high; /* were reserved2[0] */
801 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
802 __le16 l_i_reserved;
803 } linux2;
804 struct {
805 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
806 __u16 h_i_mode_high;
807 __u16 h_i_uid_high;
808 __u16 h_i_gid_high;
809 __u32 h_i_author;
810 } hurd2;
811 struct {
812 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
813 __le16 m_i_file_acl_high;
814 __u32 m_i_reserved2[2];
815 } masix2;
816 } osd2; /* OS dependent 2 */
817 __le16 i_extra_isize;
818 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
819 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
820 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
821 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
822 __le32 i_crtime; /* File Creation time */
823 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
824 __le32 i_version_hi; /* high 32 bits for 64-bit version */
825 __le32 i_projid; /* Project ID */
826 };
827
828 struct move_extent {
829 __u32 reserved; /* should be zero */
830 __u32 donor_fd; /* donor file descriptor */
831 __u64 orig_start; /* logical start offset in block for orig */
832 __u64 donor_start; /* logical start offset in block for donor */
833 __u64 len; /* block length to be moved */
834 __u64 moved_len; /* moved block length */
835 };
836
837 #define EXT4_EPOCH_BITS 2
838 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
839 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
840
841 /*
842 * Extended fields will fit into an inode if the filesystem was formatted
843 * with large inodes (-I 256 or larger) and there are not currently any EAs
844 * consuming all of the available space. For new inodes we always reserve
845 * enough space for the kernel's known extended fields, but for inodes
846 * created with an old kernel this might not have been the case. None of
847 * the extended inode fields is critical for correct filesystem operation.
848 * This macro checks if a certain field fits in the inode. Note that
849 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
850 */
851 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
852 ((offsetof(typeof(*ext4_inode), field) + \
853 sizeof((ext4_inode)->field)) \
854 <= (EXT4_GOOD_OLD_INODE_SIZE + \
855 (einode)->i_extra_isize)) \
856
857 /*
858 * We use an encoding that preserves the times for extra epoch "00":
859 *
860 * extra msb of adjust for signed
861 * epoch 32-bit 32-bit tv_sec to
862 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
863 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
864 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
865 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
866 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
867 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
868 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
869 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
870 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
871 *
872 * Note that previous versions of the kernel on 64-bit systems would
873 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
874 * 1970. e2fsck will correct this, assuming that it is run on the
875 * affected filesystem before 2242.
876 */
877
ext4_encode_extra_time(struct timespec64 * time)878 static inline __le32 ext4_encode_extra_time(struct timespec64 *time)
879 {
880 u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK;
881 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
882 }
883
ext4_decode_extra_time(struct timespec64 * time,__le32 extra)884 static inline void ext4_decode_extra_time(struct timespec64 *time,
885 __le32 extra)
886 {
887 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK)))
888 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
889 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
890 }
891
892 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
893 do { \
894 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {\
895 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
896 (raw_inode)->xtime ## _extra = \
897 ext4_encode_extra_time(&(inode)->xtime); \
898 } \
899 else \
900 (raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (inode)->xtime.tv_sec, S32_MIN, S32_MAX)); \
901 } while (0)
902
903 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
904 do { \
905 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
906 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
907 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
908 (raw_inode)->xtime ## _extra = \
909 ext4_encode_extra_time(&(einode)->xtime); \
910 } while (0)
911
912 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
913 do { \
914 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
915 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \
916 ext4_decode_extra_time(&(inode)->xtime, \
917 raw_inode->xtime ## _extra); \
918 } \
919 else \
920 (inode)->xtime.tv_nsec = 0; \
921 } while (0)
922
923
924 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
925 do { \
926 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
927 (einode)->xtime.tv_sec = \
928 (signed)le32_to_cpu((raw_inode)->xtime); \
929 else \
930 (einode)->xtime.tv_sec = 0; \
931 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
932 ext4_decode_extra_time(&(einode)->xtime, \
933 raw_inode->xtime ## _extra); \
934 else \
935 (einode)->xtime.tv_nsec = 0; \
936 } while (0)
937
938 #define i_disk_version osd1.linux1.l_i_version
939
940 #if defined(__KERNEL__) || defined(__linux__)
941 #define i_reserved1 osd1.linux1.l_i_reserved1
942 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
943 #define i_blocks_high osd2.linux2.l_i_blocks_high
944 #define i_uid_low i_uid
945 #define i_gid_low i_gid
946 #define i_uid_high osd2.linux2.l_i_uid_high
947 #define i_gid_high osd2.linux2.l_i_gid_high
948 #define i_checksum_lo osd2.linux2.l_i_checksum_lo
949
950 #elif defined(__GNU__)
951
952 #define i_translator osd1.hurd1.h_i_translator
953 #define i_uid_high osd2.hurd2.h_i_uid_high
954 #define i_gid_high osd2.hurd2.h_i_gid_high
955 #define i_author osd2.hurd2.h_i_author
956
957 #elif defined(__masix__)
958
959 #define i_reserved1 osd1.masix1.m_i_reserved1
960 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
961 #define i_reserved2 osd2.masix2.m_i_reserved2
962
963 #endif /* defined(__KERNEL__) || defined(__linux__) */
964
965 #include "extents_status.h"
966 #include "fast_commit.h"
967
968 /*
969 * Lock subclasses for i_data_sem in the ext4_inode_info structure.
970 *
971 * These are needed to avoid lockdep false positives when we need to
972 * allocate blocks to the quota inode during ext4_map_blocks(), while
973 * holding i_data_sem for a normal (non-quota) inode. Since we don't
974 * do quota tracking for the quota inode, this avoids deadlock (as
975 * well as infinite recursion, since it isn't turtles all the way
976 * down...)
977 *
978 * I_DATA_SEM_NORMAL - Used for most inodes
979 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode
980 * where the second inode has larger inode number
981 * than the first
982 * I_DATA_SEM_QUOTA - Used for quota inodes only
983 * I_DATA_SEM_EA - Used for ea_inodes only
984 */
985 enum {
986 I_DATA_SEM_NORMAL = 0,
987 I_DATA_SEM_OTHER,
988 I_DATA_SEM_QUOTA,
989 I_DATA_SEM_EA
990 };
991
992
993 /*
994 * fourth extended file system inode data in memory
995 */
996 struct ext4_inode_info {
997 __le32 i_data[15]; /* unconverted */
998 __u32 i_dtime;
999 ext4_fsblk_t i_file_acl;
1000
1001 /*
1002 * i_block_group is the number of the block group which contains
1003 * this file's inode. Constant across the lifetime of the inode,
1004 * it is used for making block allocation decisions - we try to
1005 * place a file's data blocks near its inode block, and new inodes
1006 * near to their parent directory's inode.
1007 */
1008 ext4_group_t i_block_group;
1009 ext4_lblk_t i_dir_start_lookup;
1010 #if (BITS_PER_LONG < 64)
1011 unsigned long i_state_flags; /* Dynamic state flags */
1012 #endif
1013 unsigned long i_flags;
1014
1015 /*
1016 * Extended attributes can be read independently of the main file
1017 * data. Taking i_mutex even when reading would cause contention
1018 * between readers of EAs and writers of regular file data, so
1019 * instead we synchronize on xattr_sem when reading or changing
1020 * EAs.
1021 */
1022 struct rw_semaphore xattr_sem;
1023
1024 struct list_head i_orphan; /* unlinked but open inodes */
1025
1026 /* Fast commit related info */
1027
1028 struct list_head i_fc_list; /*
1029 * inodes that need fast commit
1030 * protected by sbi->s_fc_lock.
1031 */
1032
1033 /* Start of lblk range that needs to be committed in this fast commit */
1034 ext4_lblk_t i_fc_lblk_start;
1035
1036 /* End of lblk range that needs to be committed in this fast commit */
1037 ext4_lblk_t i_fc_lblk_len;
1038
1039 /* Number of ongoing updates on this inode */
1040 atomic_t i_fc_updates;
1041
1042 /* Fast commit wait queue for this inode */
1043 wait_queue_head_t i_fc_wait;
1044
1045 /* Protect concurrent accesses on i_fc_lblk_start, i_fc_lblk_len */
1046 struct mutex i_fc_lock;
1047
1048 /*
1049 * i_disksize keeps track of what the inode size is ON DISK, not
1050 * in memory. During truncate, i_size is set to the new size by
1051 * the VFS prior to calling ext4_truncate(), but the filesystem won't
1052 * set i_disksize to 0 until the truncate is actually under way.
1053 *
1054 * The intent is that i_disksize always represents the blocks which
1055 * are used by this file. This allows recovery to restart truncate
1056 * on orphans if we crash during truncate. We actually write i_disksize
1057 * into the on-disk inode when writing inodes out, instead of i_size.
1058 *
1059 * The only time when i_disksize and i_size may be different is when
1060 * a truncate is in progress. The only things which change i_disksize
1061 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
1062 */
1063 loff_t i_disksize;
1064
1065 /*
1066 * i_data_sem is for serialising ext4_truncate() against
1067 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
1068 * data tree are chopped off during truncate. We can't do that in
1069 * ext4 because whenever we perform intermediate commits during
1070 * truncate, the inode and all the metadata blocks *must* be in a
1071 * consistent state which allows truncation of the orphans to restart
1072 * during recovery. Hence we must fix the get_block-vs-truncate race
1073 * by other means, so we have i_data_sem.
1074 */
1075 struct rw_semaphore i_data_sem;
1076 /*
1077 * i_mmap_sem is for serializing page faults with truncate / punch hole
1078 * operations. We have to make sure that new page cannot be faulted in
1079 * a section of the inode that is being punched. We cannot easily use
1080 * i_data_sem for this since we need protection for the whole punch
1081 * operation and i_data_sem ranks below transaction start so we have
1082 * to occasionally drop it.
1083 */
1084 struct rw_semaphore i_mmap_sem;
1085 struct inode vfs_inode;
1086 struct jbd2_inode *jinode;
1087
1088 spinlock_t i_raw_lock; /* protects updates to the raw inode */
1089
1090 /*
1091 * File creation time. Its function is same as that of
1092 * struct timespec64 i_{a,c,m}time in the generic inode.
1093 */
1094 struct timespec64 i_crtime;
1095
1096 /* mballoc */
1097 atomic_t i_prealloc_active;
1098 struct list_head i_prealloc_list;
1099 spinlock_t i_prealloc_lock;
1100
1101 /* extents status tree */
1102 struct ext4_es_tree i_es_tree;
1103 rwlock_t i_es_lock;
1104 struct list_head i_es_list;
1105 unsigned int i_es_all_nr; /* protected by i_es_lock */
1106 unsigned int i_es_shk_nr; /* protected by i_es_lock */
1107 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for
1108 extents to shrink. Protected by
1109 i_es_lock */
1110
1111 /* ialloc */
1112 ext4_group_t i_last_alloc_group;
1113
1114 /* allocation reservation info for delalloc */
1115 /* In case of bigalloc, this refer to clusters rather than blocks */
1116 unsigned int i_reserved_data_blocks;
1117
1118 /* pending cluster reservations for bigalloc file systems */
1119 struct ext4_pending_tree i_pending_tree;
1120
1121 /* on-disk additional length */
1122 __u16 i_extra_isize;
1123
1124 /* Indicate the inline data space. */
1125 u16 i_inline_off;
1126 u16 i_inline_size;
1127
1128 #ifdef CONFIG_QUOTA
1129 /* quota space reservation, managed internally by quota code */
1130 qsize_t i_reserved_quota;
1131 #endif
1132
1133 /* Lock protecting lists below */
1134 spinlock_t i_completed_io_lock;
1135 /*
1136 * Completed IOs that need unwritten extents handling and have
1137 * transaction reserved
1138 */
1139 struct list_head i_rsv_conversion_list;
1140 struct work_struct i_rsv_conversion_work;
1141 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1142
1143 spinlock_t i_block_reservation_lock;
1144
1145 /*
1146 * Transactions that contain inode's metadata needed to complete
1147 * fsync and fdatasync, respectively.
1148 */
1149 tid_t i_sync_tid;
1150 tid_t i_datasync_tid;
1151
1152 #ifdef CONFIG_QUOTA
1153 struct dquot *i_dquot[MAXQUOTAS];
1154 #endif
1155
1156 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1157 __u32 i_csum_seed;
1158
1159 kprojid_t i_projid;
1160 };
1161
1162 /*
1163 * File system states
1164 */
1165 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
1166 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
1167 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
1168 #define EXT4_FC_REPLAY 0x0020 /* Fast commit replay ongoing */
1169
1170 /*
1171 * Misc. filesystem flags
1172 */
1173 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
1174 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
1175 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
1176
1177 /*
1178 * Mount flags set via mount options or defaults
1179 */
1180 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */
1181 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
1182 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
1183 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
1184 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
1185 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
1186 #define EXT4_MOUNT_ERRORS_MASK 0x00070
1187 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
1188 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
1189 #ifdef CONFIG_FS_DAX
1190 #define EXT4_MOUNT_DAX_ALWAYS 0x00200 /* Direct Access */
1191 #else
1192 #define EXT4_MOUNT_DAX_ALWAYS 0
1193 #endif
1194 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
1195 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
1196 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
1197 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
1198 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
1199 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
1200 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
1201 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
1202 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
1203 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
1204 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */
1205 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota,
1206 * enable enforcement for hidden
1207 * quota files */
1208 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable
1209 * enforcement for hidden quota
1210 * files */
1211 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota
1212 * enforcement */
1213 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
1214 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
1215 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1216 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */
1217 #define EXT4_MOUNT_PREFETCH_BLOCK_BITMAPS 0x4000000
1218 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
1219 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
1220 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
1221 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
1222 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
1223
1224 /*
1225 * Mount flags set either automatically (could not be set by mount option)
1226 * based on per file system feature or property or in special cases such as
1227 * distinguishing between explicit mount option definition and default.
1228 */
1229 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1230 specified delalloc */
1231 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1232 size of blocksize * 8
1233 blocks */
1234 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1235 file systems */
1236 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly
1237 specified journal checksum */
1238
1239 #define EXT4_MOUNT2_JOURNAL_FAST_COMMIT 0x00000010 /* Journal fast commit */
1240 #define EXT4_MOUNT2_DAX_NEVER 0x00000020 /* Do not allow Direct Access */
1241 #define EXT4_MOUNT2_DAX_INODE 0x00000040 /* For printing options only */
1242
1243
1244 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1245 ~EXT4_MOUNT_##opt
1246 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1247 EXT4_MOUNT_##opt
1248 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1249 EXT4_MOUNT_##opt)
1250
1251 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1252 ~EXT4_MOUNT2_##opt
1253 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1254 EXT4_MOUNT2_##opt
1255 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1256 EXT4_MOUNT2_##opt)
1257
1258 #define ext4_test_and_set_bit __test_and_set_bit_le
1259 #define ext4_set_bit __set_bit_le
1260 #define ext4_set_bit_atomic ext2_set_bit_atomic
1261 #define ext4_test_and_clear_bit __test_and_clear_bit_le
1262 #define ext4_clear_bit __clear_bit_le
1263 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
1264 #define ext4_test_bit test_bit_le
1265 #define ext4_find_next_zero_bit find_next_zero_bit_le
1266 #define ext4_find_next_bit find_next_bit_le
1267
1268 extern void ext4_set_bits(void *bm, int cur, int len);
1269
1270 /*
1271 * Maximal mount counts between two filesystem checks
1272 */
1273 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1274 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1275
1276 /*
1277 * Behaviour when detecting errors
1278 */
1279 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1280 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1281 #define EXT4_ERRORS_PANIC 3 /* Panic */
1282 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1283
1284 /* Metadata checksum algorithm codes */
1285 #define EXT4_CRC32C_CHKSUM 1
1286
1287 /*
1288 * Structure of the super block
1289 */
1290 struct ext4_super_block {
1291 /*00*/ __le32 s_inodes_count; /* Inodes count */
1292 __le32 s_blocks_count_lo; /* Blocks count */
1293 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1294 __le32 s_free_blocks_count_lo; /* Free blocks count */
1295 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1296 __le32 s_first_data_block; /* First Data Block */
1297 __le32 s_log_block_size; /* Block size */
1298 __le32 s_log_cluster_size; /* Allocation cluster size */
1299 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1300 __le32 s_clusters_per_group; /* # Clusters per group */
1301 __le32 s_inodes_per_group; /* # Inodes per group */
1302 __le32 s_mtime; /* Mount time */
1303 /*30*/ __le32 s_wtime; /* Write time */
1304 __le16 s_mnt_count; /* Mount count */
1305 __le16 s_max_mnt_count; /* Maximal mount count */
1306 __le16 s_magic; /* Magic signature */
1307 __le16 s_state; /* File system state */
1308 __le16 s_errors; /* Behaviour when detecting errors */
1309 __le16 s_minor_rev_level; /* minor revision level */
1310 /*40*/ __le32 s_lastcheck; /* time of last check */
1311 __le32 s_checkinterval; /* max. time between checks */
1312 __le32 s_creator_os; /* OS */
1313 __le32 s_rev_level; /* Revision level */
1314 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1315 __le16 s_def_resgid; /* Default gid for reserved blocks */
1316 /*
1317 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1318 *
1319 * Note: the difference between the compatible feature set and
1320 * the incompatible feature set is that if there is a bit set
1321 * in the incompatible feature set that the kernel doesn't
1322 * know about, it should refuse to mount the filesystem.
1323 *
1324 * e2fsck's requirements are more strict; if it doesn't know
1325 * about a feature in either the compatible or incompatible
1326 * feature set, it must abort and not try to meddle with
1327 * things it doesn't understand...
1328 */
1329 __le32 s_first_ino; /* First non-reserved inode */
1330 __le16 s_inode_size; /* size of inode structure */
1331 __le16 s_block_group_nr; /* block group # of this superblock */
1332 __le32 s_feature_compat; /* compatible feature set */
1333 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1334 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1335 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1336 /*78*/ char s_volume_name[16]; /* volume name */
1337 /*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */
1338 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1339 /*
1340 * Performance hints. Directory preallocation should only
1341 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1342 */
1343 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1344 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1345 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1346 /*
1347 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1348 */
1349 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1350 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1351 __le32 s_journal_dev; /* device number of journal file */
1352 __le32 s_last_orphan; /* start of list of inodes to delete */
1353 __le32 s_hash_seed[4]; /* HTREE hash seed */
1354 __u8 s_def_hash_version; /* Default hash version to use */
1355 __u8 s_jnl_backup_type;
1356 __le16 s_desc_size; /* size of group descriptor */
1357 /*100*/ __le32 s_default_mount_opts;
1358 __le32 s_first_meta_bg; /* First metablock block group */
1359 __le32 s_mkfs_time; /* When the filesystem was created */
1360 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1361 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1362 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1363 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1364 __le32 s_free_blocks_count_hi; /* Free blocks count */
1365 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1366 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1367 __le32 s_flags; /* Miscellaneous flags */
1368 __le16 s_raid_stride; /* RAID stride */
1369 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1370 __le64 s_mmp_block; /* Block for multi-mount protection */
1371 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1372 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1373 __u8 s_checksum_type; /* metadata checksum algorithm used */
1374 __u8 s_encryption_level; /* versioning level for encryption */
1375 __u8 s_reserved_pad; /* Padding to next 32bits */
1376 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1377 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1378 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1379 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1380 snapshot's future use */
1381 __le32 s_snapshot_list; /* inode number of the head of the
1382 on-disk snapshot list */
1383 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1384 __le32 s_error_count; /* number of fs errors */
1385 __le32 s_first_error_time; /* first time an error happened */
1386 __le32 s_first_error_ino; /* inode involved in first error */
1387 __le64 s_first_error_block; /* block involved of first error */
1388 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */
1389 __le32 s_first_error_line; /* line number where error happened */
1390 __le32 s_last_error_time; /* most recent time of an error */
1391 __le32 s_last_error_ino; /* inode involved in last error */
1392 __le32 s_last_error_line; /* line number where error happened */
1393 __le64 s_last_error_block; /* block involved of last error */
1394 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */
1395 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1396 __u8 s_mount_opts[64];
1397 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1398 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1399 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1400 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
1401 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */
1402 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */
1403 __le32 s_lpf_ino; /* Location of the lost+found inode */
1404 __le32 s_prj_quota_inum; /* inode for tracking project quota */
1405 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */
1406 __u8 s_wtime_hi;
1407 __u8 s_mtime_hi;
1408 __u8 s_mkfs_time_hi;
1409 __u8 s_lastcheck_hi;
1410 __u8 s_first_error_time_hi;
1411 __u8 s_last_error_time_hi;
1412 __u8 s_first_error_errcode;
1413 __u8 s_last_error_errcode;
1414 __le16 s_encoding; /* Filename charset encoding */
1415 __le16 s_encoding_flags; /* Filename charset encoding flags */
1416 __le32 s_reserved[95]; /* Padding to the end of the block */
1417 __le32 s_checksum; /* crc32c(superblock) */
1418 };
1419
1420 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1421
1422 #ifdef __KERNEL__
1423
1424 /* Number of quota types we support */
1425 #define EXT4_MAXQUOTAS 3
1426
1427 #define EXT4_ENC_UTF8_12_1 1
1428
1429 /*
1430 * fourth extended-fs super-block data in memory
1431 */
1432 struct ext4_sb_info {
1433 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1434 unsigned long s_inodes_per_block;/* Number of inodes per block */
1435 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1436 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1437 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1438 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1439 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1440 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1441 ext4_group_t s_groups_count; /* Number of groups in the fs */
1442 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1443 unsigned long s_overhead; /* # of fs overhead clusters */
1444 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1445 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1446 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1447 struct buffer_head * s_sbh; /* Buffer containing the super block */
1448 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1449 struct buffer_head * __rcu *s_group_desc;
1450 unsigned int s_mount_opt;
1451 unsigned int s_mount_opt2;
1452 unsigned long s_mount_flags;
1453 unsigned int s_def_mount_opt;
1454 ext4_fsblk_t s_sb_block;
1455 atomic64_t s_resv_clusters;
1456 kuid_t s_resuid;
1457 kgid_t s_resgid;
1458 unsigned short s_mount_state;
1459 unsigned short s_pad;
1460 int s_addr_per_block_bits;
1461 int s_desc_per_block_bits;
1462 int s_inode_size;
1463 int s_first_ino;
1464 unsigned int s_inode_readahead_blks;
1465 unsigned int s_inode_goal;
1466 u32 s_hash_seed[4];
1467 int s_def_hash_version;
1468 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1469 struct percpu_counter s_freeclusters_counter;
1470 struct percpu_counter s_freeinodes_counter;
1471 struct percpu_counter s_dirs_counter;
1472 struct percpu_counter s_dirtyclusters_counter;
1473 struct percpu_counter s_sra_exceeded_retry_limit;
1474 struct blockgroup_lock *s_blockgroup_lock;
1475 struct proc_dir_entry *s_proc;
1476 struct kobject s_kobj;
1477 struct completion s_kobj_unregister;
1478 struct super_block *s_sb;
1479 struct buffer_head *s_mmp_bh;
1480
1481 /* Journaling */
1482 struct journal_s *s_journal;
1483 struct list_head s_orphan;
1484 struct mutex s_orphan_lock;
1485 unsigned long s_ext4_flags; /* Ext4 superblock flags */
1486 unsigned long s_commit_interval;
1487 u32 s_max_batch_time;
1488 u32 s_min_batch_time;
1489 struct block_device *s_journal_bdev;
1490 #ifdef CONFIG_QUOTA
1491 /* Names of quota files with journalled quota */
1492 char __rcu *s_qf_names[EXT4_MAXQUOTAS];
1493 int s_jquota_fmt; /* Format of quota to use */
1494 #endif
1495 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1496 struct ext4_system_blocks __rcu *s_system_blks;
1497
1498 #ifdef EXTENTS_STATS
1499 /* ext4 extents stats */
1500 unsigned long s_ext_min;
1501 unsigned long s_ext_max;
1502 unsigned long s_depth_max;
1503 spinlock_t s_ext_stats_lock;
1504 unsigned long s_ext_blocks;
1505 unsigned long s_ext_extents;
1506 #endif
1507
1508 /* for buddy allocator */
1509 struct ext4_group_info ** __rcu *s_group_info;
1510 struct inode *s_buddy_cache;
1511 spinlock_t s_md_lock;
1512 unsigned short *s_mb_offsets;
1513 unsigned int *s_mb_maxs;
1514 unsigned int s_group_info_size;
1515 unsigned int s_mb_free_pending;
1516 struct list_head s_freed_data_list; /* List of blocks to be freed
1517 after commit completed */
1518
1519 /* tunables */
1520 unsigned long s_stripe;
1521 unsigned int s_mb_stream_request;
1522 unsigned int s_mb_max_to_scan;
1523 unsigned int s_mb_min_to_scan;
1524 unsigned int s_mb_stats;
1525 unsigned int s_mb_order2_reqs;
1526 unsigned int s_mb_group_prealloc;
1527 unsigned int s_mb_max_inode_prealloc;
1528 unsigned int s_max_dir_size_kb;
1529 /* where last allocation was done - for stream allocation */
1530 unsigned long s_mb_last_group;
1531 unsigned long s_mb_last_start;
1532 unsigned int s_mb_prefetch;
1533 unsigned int s_mb_prefetch_limit;
1534
1535 /* stats for buddy allocator */
1536 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1537 atomic_t s_bal_success; /* we found long enough chunks */
1538 atomic_t s_bal_allocated; /* in blocks */
1539 atomic_t s_bal_ex_scanned; /* total extents scanned */
1540 atomic_t s_bal_groups_scanned; /* number of groups scanned */
1541 atomic_t s_bal_goals; /* goal hits */
1542 atomic_t s_bal_breaks; /* too long searches */
1543 atomic_t s_bal_2orders; /* 2^order hits */
1544 atomic64_t s_bal_cX_groups_considered[4];
1545 atomic64_t s_bal_cX_hits[4];
1546 atomic64_t s_bal_cX_failed[4]; /* cX loop didn't find blocks */
1547 atomic_t s_mb_buddies_generated; /* number of buddies generated */
1548 atomic64_t s_mb_generation_time;
1549 atomic_t s_mb_lost_chunks;
1550 atomic_t s_mb_preallocated;
1551 atomic_t s_mb_discarded;
1552 atomic_t s_lock_busy;
1553
1554 /* locality groups */
1555 struct ext4_locality_group __percpu *s_locality_groups;
1556
1557 /* for write statistics */
1558 unsigned long s_sectors_written_start;
1559 u64 s_kbytes_written;
1560
1561 /* the size of zero-out chunk */
1562 unsigned int s_extent_max_zeroout_kb;
1563
1564 unsigned int s_log_groups_per_flex;
1565 struct flex_groups * __rcu *s_flex_groups;
1566 ext4_group_t s_flex_groups_allocated;
1567
1568 /* workqueue for reserved extent conversions (buffered io) */
1569 struct workqueue_struct *rsv_conversion_wq;
1570
1571 /* timer for periodic error stats printing */
1572 struct timer_list s_err_report;
1573
1574 /* Lazy inode table initialization info */
1575 struct ext4_li_request *s_li_request;
1576 /* Wait multiplier for lazy initialization thread */
1577 unsigned int s_li_wait_mult;
1578
1579 /* Kernel thread for multiple mount protection */
1580 struct task_struct *s_mmp_tsk;
1581
1582 /* record the last minlen when FITRIM is called. */
1583 unsigned long s_last_trim_minblks;
1584
1585 /* Reference to checksum algorithm driver via cryptoapi */
1586 struct crypto_shash *s_chksum_driver;
1587
1588 /* Precomputed FS UUID checksum for seeding other checksums */
1589 __u32 s_csum_seed;
1590
1591 /* Reclaim extents from extent status tree */
1592 struct shrinker s_es_shrinker;
1593 struct list_head s_es_list; /* List of inodes with reclaimable extents */
1594 long s_es_nr_inode;
1595 struct ext4_es_stats s_es_stats;
1596 struct mb_cache *s_ea_block_cache;
1597 struct mb_cache *s_ea_inode_cache;
1598 spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1599
1600 /* Ratelimit ext4 messages. */
1601 struct ratelimit_state s_err_ratelimit_state;
1602 struct ratelimit_state s_warning_ratelimit_state;
1603 struct ratelimit_state s_msg_ratelimit_state;
1604 atomic_t s_warning_count;
1605 atomic_t s_msg_count;
1606
1607 /* Encryption policy for '-o test_dummy_encryption' */
1608 struct fscrypt_dummy_policy s_dummy_enc_policy;
1609
1610 /*
1611 * Barrier between writepages ops and changing any inode's JOURNAL_DATA
1612 * or EXTENTS flag.
1613 */
1614 struct percpu_rw_semaphore s_writepages_rwsem;
1615 struct dax_device *s_daxdev;
1616 #ifdef CONFIG_EXT4_DEBUG
1617 unsigned long s_simulate_fail;
1618 #endif
1619 /* Record the errseq of the backing block device */
1620 errseq_t s_bdev_wb_err;
1621 spinlock_t s_bdev_wb_lock;
1622
1623 /* Information about errors that happened during this mount */
1624 spinlock_t s_error_lock;
1625 int s_add_error_count;
1626 int s_first_error_code;
1627 __u32 s_first_error_line;
1628 __u32 s_first_error_ino;
1629 __u64 s_first_error_block;
1630 const char *s_first_error_func;
1631 time64_t s_first_error_time;
1632 int s_last_error_code;
1633 __u32 s_last_error_line;
1634 __u32 s_last_error_ino;
1635 __u64 s_last_error_block;
1636 const char *s_last_error_func;
1637 time64_t s_last_error_time;
1638 /*
1639 * If we are in a context where we cannot update error information in
1640 * the on-disk superblock, we queue this work to do it.
1641 */
1642 struct work_struct s_error_work;
1643
1644 /* Ext4 fast commit stuff */
1645 atomic_t s_fc_subtid;
1646 atomic_t s_fc_ineligible_updates;
1647 /*
1648 * After commit starts, the main queue gets locked, and the further
1649 * updates get added in the staging queue.
1650 */
1651 #define FC_Q_MAIN 0
1652 #define FC_Q_STAGING 1
1653 struct list_head s_fc_q[2]; /* Inodes staged for fast commit
1654 * that have data changes in them.
1655 */
1656 struct list_head s_fc_dentry_q[2]; /* directory entry updates */
1657 unsigned int s_fc_bytes;
1658 /*
1659 * Main fast commit lock. This lock protects accesses to the
1660 * following fields:
1661 * ei->i_fc_list, s_fc_dentry_q, s_fc_q, s_fc_bytes, s_fc_bh.
1662 */
1663 spinlock_t s_fc_lock;
1664 struct buffer_head *s_fc_bh;
1665 struct ext4_fc_stats s_fc_stats;
1666 u64 s_fc_avg_commit_time;
1667 #ifdef CONFIG_EXT4_DEBUG
1668 int s_fc_debug_max_replay;
1669 #endif
1670 struct ext4_fc_replay_state s_fc_replay_state;
1671 };
1672
EXT4_SB(struct super_block * sb)1673 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1674 {
1675 return sb->s_fs_info;
1676 }
EXT4_I(struct inode * inode)1677 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1678 {
1679 return container_of(inode, struct ext4_inode_info, vfs_inode);
1680 }
1681
ext4_valid_inum(struct super_block * sb,unsigned long ino)1682 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1683 {
1684 return ino == EXT4_ROOT_INO ||
1685 (ino >= EXT4_FIRST_INO(sb) &&
1686 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1687 }
1688
1689 /*
1690 * Returns: sbi->field[index]
1691 * Used to access an array element from the following sbi fields which require
1692 * rcu protection to avoid dereferencing an invalid pointer due to reassignment
1693 * - s_group_desc
1694 * - s_group_info
1695 * - s_flex_group
1696 */
1697 #define sbi_array_rcu_deref(sbi, field, index) \
1698 ({ \
1699 typeof(*((sbi)->field)) _v; \
1700 rcu_read_lock(); \
1701 _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index]; \
1702 rcu_read_unlock(); \
1703 _v; \
1704 })
1705
1706 /*
1707 * run-time mount flags
1708 */
1709 enum {
1710 EXT4_MF_MNTDIR_SAMPLED,
1711 EXT4_MF_FS_ABORTED, /* Fatal error detected */
1712 EXT4_MF_FC_INELIGIBLE, /* Fast commit ineligible */
1713 EXT4_MF_FC_COMMITTING, /* File system underoing a fast
1714 * commit.
1715 */
1716 EXT4_MF_JOURNAL_DESTROY /* Journal is in process of destroying */
1717 };
1718
ext4_set_mount_flag(struct super_block * sb,int bit)1719 static inline void ext4_set_mount_flag(struct super_block *sb, int bit)
1720 {
1721 set_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1722 }
1723
ext4_clear_mount_flag(struct super_block * sb,int bit)1724 static inline void ext4_clear_mount_flag(struct super_block *sb, int bit)
1725 {
1726 clear_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1727 }
1728
ext4_test_mount_flag(struct super_block * sb,int bit)1729 static inline int ext4_test_mount_flag(struct super_block *sb, int bit)
1730 {
1731 return test_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1732 }
1733
1734
1735 /*
1736 * Simulate_fail codes
1737 */
1738 #define EXT4_SIM_BBITMAP_EIO 1
1739 #define EXT4_SIM_BBITMAP_CRC 2
1740 #define EXT4_SIM_IBITMAP_EIO 3
1741 #define EXT4_SIM_IBITMAP_CRC 4
1742 #define EXT4_SIM_INODE_EIO 5
1743 #define EXT4_SIM_INODE_CRC 6
1744 #define EXT4_SIM_DIRBLOCK_EIO 7
1745 #define EXT4_SIM_DIRBLOCK_CRC 8
1746
ext4_simulate_fail(struct super_block * sb,unsigned long code)1747 static inline bool ext4_simulate_fail(struct super_block *sb,
1748 unsigned long code)
1749 {
1750 #ifdef CONFIG_EXT4_DEBUG
1751 struct ext4_sb_info *sbi = EXT4_SB(sb);
1752
1753 if (unlikely(sbi->s_simulate_fail == code)) {
1754 sbi->s_simulate_fail = 0;
1755 return true;
1756 }
1757 #endif
1758 return false;
1759 }
1760
ext4_simulate_fail_bh(struct super_block * sb,struct buffer_head * bh,unsigned long code)1761 static inline void ext4_simulate_fail_bh(struct super_block *sb,
1762 struct buffer_head *bh,
1763 unsigned long code)
1764 {
1765 if (!IS_ERR(bh) && ext4_simulate_fail(sb, code))
1766 clear_buffer_uptodate(bh);
1767 }
1768
1769 /*
1770 * Error number codes for s_{first,last}_error_errno
1771 *
1772 * Linux errno numbers are architecture specific, so we need to translate
1773 * them into something which is architecture independent. We don't define
1774 * codes for all errno's; just the ones which are most likely to be the cause
1775 * of an ext4_error() call.
1776 */
1777 #define EXT4_ERR_UNKNOWN 1
1778 #define EXT4_ERR_EIO 2
1779 #define EXT4_ERR_ENOMEM 3
1780 #define EXT4_ERR_EFSBADCRC 4
1781 #define EXT4_ERR_EFSCORRUPTED 5
1782 #define EXT4_ERR_ENOSPC 6
1783 #define EXT4_ERR_ENOKEY 7
1784 #define EXT4_ERR_EROFS 8
1785 #define EXT4_ERR_EFBIG 9
1786 #define EXT4_ERR_EEXIST 10
1787 #define EXT4_ERR_ERANGE 11
1788 #define EXT4_ERR_EOVERFLOW 12
1789 #define EXT4_ERR_EBUSY 13
1790 #define EXT4_ERR_ENOTDIR 14
1791 #define EXT4_ERR_ENOTEMPTY 15
1792 #define EXT4_ERR_ESHUTDOWN 16
1793 #define EXT4_ERR_EFAULT 17
1794
1795 /*
1796 * Inode dynamic state flags
1797 */
1798 enum {
1799 EXT4_STATE_JDATA, /* journaled data exists */
1800 EXT4_STATE_NEW, /* inode is newly created */
1801 EXT4_STATE_XATTR, /* has in-inode xattrs */
1802 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1803 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1804 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1805 EXT4_STATE_NEWENTRY, /* File just added to dir */
1806 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1807 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
1808 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */
1809 EXT4_STATE_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */
1810 EXT4_STATE_FC_COMMITTING, /* Fast commit ongoing */
1811 };
1812
1813 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1814 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1815 { \
1816 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1817 } \
1818 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1819 { \
1820 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1821 } \
1822 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1823 { \
1824 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1825 }
1826
1827 /* Add these declarations here only so that these functions can be
1828 * found by name. Otherwise, they are very hard to locate. */
1829 static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1830 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1831 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1832 EXT4_INODE_BIT_FNS(flag, flags, 0)
1833
1834 /* Add these declarations here only so that these functions can be
1835 * found by name. Otherwise, they are very hard to locate. */
1836 static inline int ext4_test_inode_state(struct inode *inode, int bit);
1837 static inline void ext4_set_inode_state(struct inode *inode, int bit);
1838 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1839 #if (BITS_PER_LONG < 64)
1840 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1841
ext4_clear_state_flags(struct ext4_inode_info * ei)1842 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1843 {
1844 (ei)->i_state_flags = 0;
1845 }
1846 #else
1847 EXT4_INODE_BIT_FNS(state, flags, 32)
1848
ext4_clear_state_flags(struct ext4_inode_info * ei)1849 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1850 {
1851 /* We depend on the fact that callers will set i_flags */
1852 }
1853 #endif
1854 #else
1855 /* Assume that user mode programs are passing in an ext4fs superblock, not
1856 * a kernel struct super_block. This will allow us to call the feature-test
1857 * macros from user land. */
1858 #define EXT4_SB(sb) (sb)
1859 #endif
1860
ext4_verity_in_progress(struct inode * inode)1861 static inline bool ext4_verity_in_progress(struct inode *inode)
1862 {
1863 return IS_ENABLED(CONFIG_FS_VERITY) &&
1864 ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS);
1865 }
1866
1867 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1868
1869 /*
1870 * Codes for operating systems
1871 */
1872 #define EXT4_OS_LINUX 0
1873 #define EXT4_OS_HURD 1
1874 #define EXT4_OS_MASIX 2
1875 #define EXT4_OS_FREEBSD 3
1876 #define EXT4_OS_LITES 4
1877
1878 /*
1879 * Revision levels
1880 */
1881 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1882 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1883
1884 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1885 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1886
1887 #define EXT4_GOOD_OLD_INODE_SIZE 128
1888
1889 #define EXT4_EXTRA_TIMESTAMP_MAX (((s64)1 << 34) - 1 + S32_MIN)
1890 #define EXT4_NON_EXTRA_TIMESTAMP_MAX S32_MAX
1891 #define EXT4_TIMESTAMP_MIN S32_MIN
1892
1893 /*
1894 * Feature set definitions
1895 */
1896
1897 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1898 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1899 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1900 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1901 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1902 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1903 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
1904 /*
1905 * The reason why "FAST_COMMIT" is a compat feature is that, FS becomes
1906 * incompatible only if fast commit blocks are present in the FS. Since we
1907 * clear the journal (and thus the fast commit blocks), we don't mark FS as
1908 * incompatible. We also have a JBD2 incompat feature, which gets set when
1909 * there are fast commit blocks present in the journal.
1910 */
1911 #define EXT4_FEATURE_COMPAT_FAST_COMMIT 0x0400
1912 #define EXT4_FEATURE_COMPAT_STABLE_INODES 0x0800
1913
1914 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1915 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1916 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1917 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1918 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1919 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1920 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1921 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1922 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1923 /*
1924 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1925 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1926 * all other data structures' checksums. However, the METADATA_CSUM and
1927 * GDT_CSUM bits are mutually exclusive.
1928 */
1929 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
1930 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000
1931 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000
1932 #define EXT4_FEATURE_RO_COMPAT_VERITY 0x8000
1933
1934 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1935 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1936 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1937 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1938 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1939 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1940 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1941 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1942 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1943 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1944 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1945 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000
1946 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
1947 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
1948 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000
1949 #define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000
1950
1951 extern void ext4_update_dynamic_rev(struct super_block *sb);
1952
1953 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
1954 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1955 { \
1956 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1957 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
1958 } \
1959 static inline void ext4_set_feature_##name(struct super_block *sb) \
1960 { \
1961 ext4_update_dynamic_rev(sb); \
1962 EXT4_SB(sb)->s_es->s_feature_compat |= \
1963 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1964 } \
1965 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1966 { \
1967 EXT4_SB(sb)->s_es->s_feature_compat &= \
1968 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1969 }
1970
1971 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
1972 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1973 { \
1974 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1975 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
1976 } \
1977 static inline void ext4_set_feature_##name(struct super_block *sb) \
1978 { \
1979 ext4_update_dynamic_rev(sb); \
1980 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
1981 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1982 } \
1983 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1984 { \
1985 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
1986 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1987 }
1988
1989 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
1990 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1991 { \
1992 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1993 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
1994 } \
1995 static inline void ext4_set_feature_##name(struct super_block *sb) \
1996 { \
1997 ext4_update_dynamic_rev(sb); \
1998 EXT4_SB(sb)->s_es->s_feature_incompat |= \
1999 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
2000 } \
2001 static inline void ext4_clear_feature_##name(struct super_block *sb) \
2002 { \
2003 EXT4_SB(sb)->s_es->s_feature_incompat &= \
2004 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
2005 }
2006
EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc,DIR_PREALLOC)2007 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC)
2008 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES)
2009 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL)
2010 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR)
2011 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE)
2012 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX)
2013 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2)
2014 EXT4_FEATURE_COMPAT_FUNCS(fast_commit, FAST_COMMIT)
2015 EXT4_FEATURE_COMPAT_FUNCS(stable_inodes, STABLE_INODES)
2016
2017 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER)
2018 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE)
2019 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR)
2020 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE)
2021 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM)
2022 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK)
2023 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE)
2024 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA)
2025 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC)
2026 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM)
2027 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY)
2028 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT)
2029 EXT4_FEATURE_RO_COMPAT_FUNCS(verity, VERITY)
2030
2031 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION)
2032 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE)
2033 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER)
2034 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV)
2035 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG)
2036 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS)
2037 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT)
2038 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP)
2039 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG)
2040 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE)
2041 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA)
2042 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED)
2043 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR)
2044 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA)
2045 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT)
2046 EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD)
2047
2048 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
2049 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
2050 EXT4_FEATURE_INCOMPAT_META_BG)
2051 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2052 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2053 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
2054
2055 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
2056 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
2057 EXT4_FEATURE_INCOMPAT_RECOVER| \
2058 EXT4_FEATURE_INCOMPAT_META_BG)
2059 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2060 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2061 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
2062
2063 #define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
2064 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
2065 EXT4_FEATURE_INCOMPAT_RECOVER| \
2066 EXT4_FEATURE_INCOMPAT_META_BG| \
2067 EXT4_FEATURE_INCOMPAT_EXTENTS| \
2068 EXT4_FEATURE_INCOMPAT_64BIT| \
2069 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
2070 EXT4_FEATURE_INCOMPAT_EA_INODE| \
2071 EXT4_FEATURE_INCOMPAT_MMP | \
2072 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
2073 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
2074 EXT4_FEATURE_INCOMPAT_CASEFOLD | \
2075 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \
2076 EXT4_FEATURE_INCOMPAT_LARGEDIR)
2077 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2078 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2079 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
2080 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
2081 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
2082 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
2083 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
2084 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
2085 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
2086 EXT4_FEATURE_RO_COMPAT_QUOTA |\
2087 EXT4_FEATURE_RO_COMPAT_PROJECT |\
2088 EXT4_FEATURE_RO_COMPAT_VERITY)
2089
2090 #define EXTN_FEATURE_FUNCS(ver) \
2091 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
2092 { \
2093 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
2094 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
2095 } \
2096 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
2097 { \
2098 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
2099 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
2100 } \
2101 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
2102 { \
2103 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
2104 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
2105 }
2106
2107 EXTN_FEATURE_FUNCS(2)
2108 EXTN_FEATURE_FUNCS(3)
2109 EXTN_FEATURE_FUNCS(4)
2110
2111 static inline bool ext4_has_compat_features(struct super_block *sb)
2112 {
2113 return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
2114 }
ext4_has_ro_compat_features(struct super_block * sb)2115 static inline bool ext4_has_ro_compat_features(struct super_block *sb)
2116 {
2117 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
2118 }
ext4_has_incompat_features(struct super_block * sb)2119 static inline bool ext4_has_incompat_features(struct super_block *sb)
2120 {
2121 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
2122 }
2123
2124 /*
2125 * Superblock flags
2126 */
2127 #define EXT4_FLAGS_RESIZING 0
2128 #define EXT4_FLAGS_SHUTDOWN 1
2129 #define EXT4_FLAGS_BDEV_IS_DAX 2
2130
ext4_forced_shutdown(struct ext4_sb_info * sbi)2131 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi)
2132 {
2133 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags);
2134 }
2135
2136
2137 /*
2138 * Default values for user and/or group using reserved blocks
2139 */
2140 #define EXT4_DEF_RESUID 0
2141 #define EXT4_DEF_RESGID 0
2142
2143 /*
2144 * Default project ID
2145 */
2146 #define EXT4_DEF_PROJID 0
2147
2148 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
2149
2150 /*
2151 * Default mount options
2152 */
2153 #define EXT4_DEFM_DEBUG 0x0001
2154 #define EXT4_DEFM_BSDGROUPS 0x0002
2155 #define EXT4_DEFM_XATTR_USER 0x0004
2156 #define EXT4_DEFM_ACL 0x0008
2157 #define EXT4_DEFM_UID16 0x0010
2158 #define EXT4_DEFM_JMODE 0x0060
2159 #define EXT4_DEFM_JMODE_DATA 0x0020
2160 #define EXT4_DEFM_JMODE_ORDERED 0x0040
2161 #define EXT4_DEFM_JMODE_WBACK 0x0060
2162 #define EXT4_DEFM_NOBARRIER 0x0100
2163 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
2164 #define EXT4_DEFM_DISCARD 0x0400
2165 #define EXT4_DEFM_NODELALLOC 0x0800
2166
2167 /*
2168 * Default journal batch times
2169 */
2170 #define EXT4_DEF_MIN_BATCH_TIME 0
2171 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
2172
2173 /*
2174 * Minimum number of groups in a flexgroup before we separate out
2175 * directories into the first block group of a flexgroup
2176 */
2177 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
2178
2179 /*
2180 * Structure of a directory entry
2181 */
2182 #define EXT4_NAME_LEN 255
2183 /*
2184 * Base length of the ext4 directory entry excluding the name length
2185 */
2186 #define EXT4_BASE_DIR_LEN (sizeof(struct ext4_dir_entry_2) - EXT4_NAME_LEN)
2187
2188 struct ext4_dir_entry {
2189 __le32 inode; /* Inode number */
2190 __le16 rec_len; /* Directory entry length */
2191 __le16 name_len; /* Name length */
2192 char name[EXT4_NAME_LEN]; /* File name */
2193 };
2194
2195 /*
2196 * The new version of the directory entry. Since EXT4 structures are
2197 * stored in intel byte order, and the name_len field could never be
2198 * bigger than 255 chars, it's safe to reclaim the extra byte for the
2199 * file_type field.
2200 */
2201 struct ext4_dir_entry_2 {
2202 __le32 inode; /* Inode number */
2203 __le16 rec_len; /* Directory entry length */
2204 __u8 name_len; /* Name length */
2205 __u8 file_type; /* See file type macros EXT4_FT_* below */
2206 char name[EXT4_NAME_LEN]; /* File name */
2207 };
2208
2209 /*
2210 * This is a bogus directory entry at the end of each leaf block that
2211 * records checksums.
2212 */
2213 struct ext4_dir_entry_tail {
2214 __le32 det_reserved_zero1; /* Pretend to be unused */
2215 __le16 det_rec_len; /* 12 */
2216 __u8 det_reserved_zero2; /* Zero name length */
2217 __u8 det_reserved_ft; /* 0xDE, fake file type */
2218 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
2219 };
2220
2221 #define EXT4_DIRENT_TAIL(block, blocksize) \
2222 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
2223 ((blocksize) - \
2224 sizeof(struct ext4_dir_entry_tail))))
2225
2226 /*
2227 * Ext4 directory file types. Only the low 3 bits are used. The
2228 * other bits are reserved for now.
2229 */
2230 #define EXT4_FT_UNKNOWN 0
2231 #define EXT4_FT_REG_FILE 1
2232 #define EXT4_FT_DIR 2
2233 #define EXT4_FT_CHRDEV 3
2234 #define EXT4_FT_BLKDEV 4
2235 #define EXT4_FT_FIFO 5
2236 #define EXT4_FT_SOCK 6
2237 #define EXT4_FT_SYMLINK 7
2238
2239 #define EXT4_FT_MAX 8
2240
2241 #define EXT4_FT_DIR_CSUM 0xDE
2242
2243 /*
2244 * EXT4_DIR_PAD defines the directory entries boundaries
2245 *
2246 * NOTE: It must be a multiple of 4
2247 */
2248 #define EXT4_DIR_PAD 4
2249 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
2250 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
2251 ~EXT4_DIR_ROUND)
2252 #define EXT4_MAX_REC_LEN ((1<<16)-1)
2253
2254 /*
2255 * If we ever get support for fs block sizes > page_size, we'll need
2256 * to remove the #if statements in the next two functions...
2257 */
2258 static inline unsigned int
ext4_rec_len_from_disk(__le16 dlen,unsigned blocksize)2259 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
2260 {
2261 unsigned len = le16_to_cpu(dlen);
2262
2263 #if (PAGE_SIZE >= 65536)
2264 if (len == EXT4_MAX_REC_LEN || len == 0)
2265 return blocksize;
2266 return (len & 65532) | ((len & 3) << 16);
2267 #else
2268 return len;
2269 #endif
2270 }
2271
ext4_rec_len_to_disk(unsigned len,unsigned blocksize)2272 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
2273 {
2274 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
2275 BUG();
2276 #if (PAGE_SIZE >= 65536)
2277 if (len < 65536)
2278 return cpu_to_le16(len);
2279 if (len == blocksize) {
2280 if (blocksize == 65536)
2281 return cpu_to_le16(EXT4_MAX_REC_LEN);
2282 else
2283 return cpu_to_le16(0);
2284 }
2285 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
2286 #else
2287 return cpu_to_le16(len);
2288 #endif
2289 }
2290
2291 /*
2292 * Hash Tree Directory indexing
2293 * (c) Daniel Phillips, 2001
2294 */
2295
2296 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
2297 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2298 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \
2299 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir)))
2300 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2301
2302 /* Legal values for the dx_root hash_version field: */
2303
2304 #define DX_HASH_LEGACY 0
2305 #define DX_HASH_HALF_MD4 1
2306 #define DX_HASH_TEA 2
2307 #define DX_HASH_LEGACY_UNSIGNED 3
2308 #define DX_HASH_HALF_MD4_UNSIGNED 4
2309 #define DX_HASH_TEA_UNSIGNED 5
2310
ext4_chksum(struct ext4_sb_info * sbi,u32 crc,const void * address,unsigned int length)2311 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2312 const void *address, unsigned int length)
2313 {
2314 struct {
2315 struct shash_desc shash;
2316 char ctx[4];
2317 } desc;
2318
2319 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2320
2321 desc.shash.tfm = sbi->s_chksum_driver;
2322 *(u32 *)desc.ctx = crc;
2323
2324 BUG_ON(crypto_shash_update(&desc.shash, address, length));
2325
2326 return *(u32 *)desc.ctx;
2327 }
2328
2329 #ifdef __KERNEL__
2330
2331 /* hash info structure used by the directory hash */
2332 struct dx_hash_info
2333 {
2334 u32 hash;
2335 u32 minor_hash;
2336 int hash_version;
2337 u32 *seed;
2338 };
2339
2340
2341 /* 32 and 64 bit signed EOF for dx directories */
2342 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
2343 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
2344
2345
2346 /*
2347 * Control parameters used by ext4_htree_next_block
2348 */
2349 #define HASH_NB_ALWAYS 1
2350
2351 struct ext4_filename {
2352 const struct qstr *usr_fname;
2353 struct fscrypt_str disk_name;
2354 struct dx_hash_info hinfo;
2355 #ifdef CONFIG_FS_ENCRYPTION
2356 struct fscrypt_str crypto_buf;
2357 #endif
2358 #ifdef CONFIG_UNICODE
2359 struct fscrypt_str cf_name;
2360 #endif
2361 };
2362
2363 #define fname_name(p) ((p)->disk_name.name)
2364 #define fname_len(p) ((p)->disk_name.len)
2365
2366 /*
2367 * Describe an inode's exact location on disk and in memory
2368 */
2369 struct ext4_iloc
2370 {
2371 struct buffer_head *bh;
2372 unsigned long offset;
2373 ext4_group_t block_group;
2374 };
2375
ext4_raw_inode(struct ext4_iloc * iloc)2376 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2377 {
2378 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2379 }
2380
ext4_is_quota_file(struct inode * inode)2381 static inline bool ext4_is_quota_file(struct inode *inode)
2382 {
2383 return IS_NOQUOTA(inode) &&
2384 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL);
2385 }
2386
2387 /*
2388 * This structure is stuffed into the struct file's private_data field
2389 * for directories. It is where we put information so that we can do
2390 * readdir operations in hash tree order.
2391 */
2392 struct dir_private_info {
2393 struct rb_root root;
2394 struct rb_node *curr_node;
2395 struct fname *extra_fname;
2396 loff_t last_pos;
2397 __u32 curr_hash;
2398 __u32 curr_minor_hash;
2399 __u32 next_hash;
2400 };
2401
2402 /* calculate the first block number of the group */
2403 static inline ext4_fsblk_t
ext4_group_first_block_no(struct super_block * sb,ext4_group_t group_no)2404 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2405 {
2406 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2407 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2408 }
2409
2410 /*
2411 * Special error return code only used by dx_probe() and its callers.
2412 */
2413 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
2414
2415 /* htree levels for ext4 */
2416 #define EXT4_HTREE_LEVEL_COMPAT 2
2417 #define EXT4_HTREE_LEVEL 3
2418
ext4_dir_htree_level(struct super_block * sb)2419 static inline int ext4_dir_htree_level(struct super_block *sb)
2420 {
2421 return ext4_has_feature_largedir(sb) ?
2422 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT;
2423 }
2424
2425 /*
2426 * Timeout and state flag for lazy initialization inode thread.
2427 */
2428 #define EXT4_DEF_LI_WAIT_MULT 10
2429 #define EXT4_DEF_LI_MAX_START_DELAY 5
2430 #define EXT4_LAZYINIT_QUIT 0x0001
2431 #define EXT4_LAZYINIT_RUNNING 0x0002
2432
2433 /*
2434 * Lazy inode table initialization info
2435 */
2436 struct ext4_lazy_init {
2437 unsigned long li_state;
2438 struct list_head li_request_list;
2439 struct mutex li_list_mtx;
2440 };
2441
2442 enum ext4_li_mode {
2443 EXT4_LI_MODE_PREFETCH_BBITMAP,
2444 EXT4_LI_MODE_ITABLE,
2445 };
2446
2447 struct ext4_li_request {
2448 struct super_block *lr_super;
2449 enum ext4_li_mode lr_mode;
2450 ext4_group_t lr_first_not_zeroed;
2451 ext4_group_t lr_next_group;
2452 struct list_head lr_request;
2453 unsigned long lr_next_sched;
2454 unsigned long lr_timeout;
2455 };
2456
2457 struct ext4_features {
2458 struct kobject f_kobj;
2459 struct completion f_kobj_unregister;
2460 };
2461
2462 /*
2463 * This structure will be used for multiple mount protection. It will be
2464 * written into the block number saved in the s_mmp_block field in the
2465 * superblock. Programs that check MMP should assume that if
2466 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2467 * to use the filesystem, regardless of how old the timestamp is.
2468 */
2469 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
2470 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2471 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
2472 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
2473
2474 struct mmp_struct {
2475 __le32 mmp_magic; /* Magic number for MMP */
2476 __le32 mmp_seq; /* Sequence no. updated periodically */
2477
2478 /*
2479 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2480 * purposes and do not affect the correctness of the algorithm
2481 */
2482 __le64 mmp_time; /* Time last updated */
2483 char mmp_nodename[64]; /* Node which last updated MMP block */
2484 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
2485
2486 /*
2487 * mmp_check_interval is used to verify if the MMP block has been
2488 * updated on the block device. The value is updated based on the
2489 * maximum time to write the MMP block during an update cycle.
2490 */
2491 __le16 mmp_check_interval;
2492
2493 __le16 mmp_pad1;
2494 __le32 mmp_pad2[226];
2495 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
2496 };
2497
2498 /* arguments passed to the mmp thread */
2499 struct mmpd_data {
2500 struct buffer_head *bh; /* bh from initial read_mmp_block() */
2501 struct super_block *sb; /* super block of the fs */
2502 };
2503
2504 /*
2505 * Check interval multiplier
2506 * The MMP block is written every update interval and initially checked every
2507 * update interval x the multiplier (the value is then adapted based on the
2508 * write latency). The reason is that writes can be delayed under load and we
2509 * don't want readers to incorrectly assume that the filesystem is no longer
2510 * in use.
2511 */
2512 #define EXT4_MMP_CHECK_MULT 2UL
2513
2514 /*
2515 * Minimum interval for MMP checking in seconds.
2516 */
2517 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
2518
2519 /*
2520 * Maximum interval for MMP checking in seconds.
2521 */
2522 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
2523
2524 /*
2525 * Function prototypes
2526 */
2527
2528 /*
2529 * Ok, these declarations are also in <linux/kernel.h> but none of the
2530 * ext4 source programs needs to include it so they are duplicated here.
2531 */
2532 # define NORET_TYPE /**/
2533 # define ATTRIB_NORET __attribute__((noreturn))
2534 # define NORET_AND noreturn,
2535
2536 /* bitmap.c */
2537 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2538 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2539 struct ext4_group_desc *gdp,
2540 struct buffer_head *bh, int sz);
2541 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2542 struct ext4_group_desc *gdp,
2543 struct buffer_head *bh, int sz);
2544 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2545 struct ext4_group_desc *gdp,
2546 struct buffer_head *bh);
2547 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2548 struct ext4_group_desc *gdp,
2549 struct buffer_head *bh);
2550
2551 /* balloc.c */
2552 extern void ext4_get_group_no_and_offset(struct super_block *sb,
2553 ext4_fsblk_t blocknr,
2554 ext4_group_t *blockgrpp,
2555 ext4_grpblk_t *offsetp);
2556 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2557 ext4_fsblk_t block);
2558
2559 extern unsigned int ext4_block_group(struct super_block *sb,
2560 ext4_fsblk_t blocknr);
2561 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2562 ext4_fsblk_t blocknr);
2563 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2564 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2565 ext4_group_t group);
2566 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2567 ext4_fsblk_t goal,
2568 unsigned int flags,
2569 unsigned long *count,
2570 int *errp);
2571 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2572 s64 nclusters, unsigned int flags);
2573 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2574 extern void ext4_check_blocks_bitmap(struct super_block *);
2575 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2576 ext4_group_t block_group,
2577 struct buffer_head ** bh);
2578 extern struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2579 ext4_group_t group);
2580 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2581
2582 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2583 ext4_group_t block_group,
2584 bool ignore_locked);
2585 extern int ext4_wait_block_bitmap(struct super_block *sb,
2586 ext4_group_t block_group,
2587 struct buffer_head *bh);
2588 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2589 ext4_group_t block_group);
2590 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2591 ext4_group_t block_group,
2592 struct ext4_group_desc *gdp);
2593 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2594
2595 #ifdef CONFIG_UNICODE
2596 extern void ext4_fname_setup_ci_filename(struct inode *dir,
2597 const struct qstr *iname,
2598 struct fscrypt_str *fname);
2599 #endif
2600
2601 #ifdef CONFIG_FS_ENCRYPTION
ext4_fname_from_fscrypt_name(struct ext4_filename * dst,const struct fscrypt_name * src)2602 static inline void ext4_fname_from_fscrypt_name(struct ext4_filename *dst,
2603 const struct fscrypt_name *src)
2604 {
2605 memset(dst, 0, sizeof(*dst));
2606
2607 dst->usr_fname = src->usr_fname;
2608 dst->disk_name = src->disk_name;
2609 dst->hinfo.hash = src->hash;
2610 dst->hinfo.minor_hash = src->minor_hash;
2611 dst->crypto_buf = src->crypto_buf;
2612 }
2613
ext4_fname_setup_filename(struct inode * dir,const struct qstr * iname,int lookup,struct ext4_filename * fname)2614 static inline int ext4_fname_setup_filename(struct inode *dir,
2615 const struct qstr *iname,
2616 int lookup,
2617 struct ext4_filename *fname)
2618 {
2619 struct fscrypt_name name;
2620 int err;
2621
2622 err = fscrypt_setup_filename(dir, iname, lookup, &name);
2623 if (err)
2624 return err;
2625
2626 ext4_fname_from_fscrypt_name(fname, &name);
2627
2628 #ifdef CONFIG_UNICODE
2629 ext4_fname_setup_ci_filename(dir, iname, &fname->cf_name);
2630 #endif
2631 return 0;
2632 }
2633
ext4_fname_prepare_lookup(struct inode * dir,struct dentry * dentry,struct ext4_filename * fname)2634 static inline int ext4_fname_prepare_lookup(struct inode *dir,
2635 struct dentry *dentry,
2636 struct ext4_filename *fname)
2637 {
2638 struct fscrypt_name name;
2639 int err;
2640
2641 err = fscrypt_prepare_lookup(dir, dentry, &name);
2642 if (err)
2643 return err;
2644
2645 ext4_fname_from_fscrypt_name(fname, &name);
2646
2647 #ifdef CONFIG_UNICODE
2648 ext4_fname_setup_ci_filename(dir, &dentry->d_name, &fname->cf_name);
2649 #endif
2650 return 0;
2651 }
2652
ext4_fname_free_filename(struct ext4_filename * fname)2653 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2654 {
2655 struct fscrypt_name name;
2656
2657 name.crypto_buf = fname->crypto_buf;
2658 fscrypt_free_filename(&name);
2659
2660 fname->crypto_buf.name = NULL;
2661 fname->usr_fname = NULL;
2662 fname->disk_name.name = NULL;
2663
2664 #ifdef CONFIG_UNICODE
2665 kfree(fname->cf_name.name);
2666 fname->cf_name.name = NULL;
2667 #endif
2668 }
2669 #else /* !CONFIG_FS_ENCRYPTION */
ext4_fname_setup_filename(struct inode * dir,const struct qstr * iname,int lookup,struct ext4_filename * fname)2670 static inline int ext4_fname_setup_filename(struct inode *dir,
2671 const struct qstr *iname,
2672 int lookup,
2673 struct ext4_filename *fname)
2674 {
2675 fname->usr_fname = iname;
2676 fname->disk_name.name = (unsigned char *) iname->name;
2677 fname->disk_name.len = iname->len;
2678
2679 #ifdef CONFIG_UNICODE
2680 ext4_fname_setup_ci_filename(dir, iname, &fname->cf_name);
2681 #endif
2682
2683 return 0;
2684 }
2685
ext4_fname_prepare_lookup(struct inode * dir,struct dentry * dentry,struct ext4_filename * fname)2686 static inline int ext4_fname_prepare_lookup(struct inode *dir,
2687 struct dentry *dentry,
2688 struct ext4_filename *fname)
2689 {
2690 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname);
2691 }
2692
ext4_fname_free_filename(struct ext4_filename * fname)2693 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2694 {
2695 #ifdef CONFIG_UNICODE
2696 kfree(fname->cf_name.name);
2697 fname->cf_name.name = NULL;
2698 #endif
2699 }
2700 #endif /* !CONFIG_FS_ENCRYPTION */
2701
2702 /* dir.c */
2703 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2704 struct file *,
2705 struct ext4_dir_entry_2 *,
2706 struct buffer_head *, char *, int,
2707 unsigned int);
2708 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2709 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2710 (de), (bh), (buf), (size), (offset)))
2711 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2712 __u32 minor_hash,
2713 struct ext4_dir_entry_2 *dirent,
2714 struct fscrypt_str *ent_name);
2715 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2716 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2717 struct buffer_head *bh,
2718 void *buf, int buf_size,
2719 struct ext4_filename *fname,
2720 struct ext4_dir_entry_2 **dest_de);
2721 void ext4_insert_dentry(struct inode *inode,
2722 struct ext4_dir_entry_2 *de,
2723 int buf_size,
2724 struct ext4_filename *fname);
ext4_update_dx_flag(struct inode * inode)2725 static inline void ext4_update_dx_flag(struct inode *inode)
2726 {
2727 if (!ext4_has_feature_dir_index(inode->i_sb) &&
2728 ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) {
2729 /* ext4_iget() should have caught this... */
2730 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb));
2731 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2732 }
2733 }
2734 static const unsigned char ext4_filetype_table[] = {
2735 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2736 };
2737
get_dtype(struct super_block * sb,int filetype)2738 static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2739 {
2740 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2741 return DT_UNKNOWN;
2742
2743 return ext4_filetype_table[filetype];
2744 }
2745 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2746 void *buf, int buf_size);
2747
2748 /* fsync.c */
2749 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2750
2751 /* hash.c */
2752 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len,
2753 struct dx_hash_info *hinfo);
2754
2755 /* ialloc.c */
2756 extern int ext4_mark_inode_used(struct super_block *sb, int ino);
2757 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2758 const struct qstr *qstr, __u32 goal,
2759 uid_t *owner, __u32 i_flags,
2760 int handle_type, unsigned int line_no,
2761 int nblocks);
2762
2763 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \
2764 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2765 i_flags, 0, 0, 0)
2766 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2767 type, nblocks) \
2768 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2769 0, (type), __LINE__, (nblocks))
2770
2771
2772 extern void ext4_free_inode(handle_t *, struct inode *);
2773 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2774 extern unsigned long ext4_count_free_inodes(struct super_block *);
2775 extern unsigned long ext4_count_dirs(struct super_block *);
2776 extern void ext4_check_inodes_bitmap(struct super_block *);
2777 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2778 extern int ext4_init_inode_table(struct super_block *sb,
2779 ext4_group_t group, int barrier);
2780 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2781
2782 /* fast_commit.c */
2783 int ext4_fc_info_show(struct seq_file *seq, void *v);
2784 void ext4_fc_init(struct super_block *sb, journal_t *journal);
2785 void ext4_fc_init_inode(struct inode *inode);
2786 void ext4_fc_track_range(handle_t *handle, struct inode *inode, ext4_lblk_t start,
2787 ext4_lblk_t end);
2788 void __ext4_fc_track_unlink(handle_t *handle, struct inode *inode,
2789 struct dentry *dentry);
2790 void __ext4_fc_track_link(handle_t *handle, struct inode *inode,
2791 struct dentry *dentry);
2792 void ext4_fc_track_unlink(handle_t *handle, struct dentry *dentry);
2793 void ext4_fc_track_link(handle_t *handle, struct dentry *dentry);
2794 void __ext4_fc_track_create(handle_t *handle, struct inode *inode,
2795 struct dentry *dentry);
2796 void ext4_fc_track_create(handle_t *handle, struct dentry *dentry);
2797 void ext4_fc_track_inode(handle_t *handle, struct inode *inode);
2798 void ext4_fc_mark_ineligible(struct super_block *sb, int reason);
2799 void ext4_fc_start_ineligible(struct super_block *sb, int reason);
2800 void ext4_fc_stop_ineligible(struct super_block *sb);
2801 void ext4_fc_start_update(struct inode *inode);
2802 void ext4_fc_stop_update(struct inode *inode);
2803 void ext4_fc_del(struct inode *inode);
2804 bool ext4_fc_replay_check_excluded(struct super_block *sb, ext4_fsblk_t block);
2805 void ext4_fc_replay_cleanup(struct super_block *sb);
2806 int ext4_fc_commit(journal_t *journal, tid_t commit_tid);
2807 int __init ext4_fc_init_dentry_cache(void);
2808 void ext4_fc_destroy_dentry_cache(void);
2809 int ext4_fc_record_regions(struct super_block *sb, int ino,
2810 ext4_lblk_t lblk, ext4_fsblk_t pblk,
2811 int len, int replay);
2812
2813 /* mballoc.c */
2814 extern const struct seq_operations ext4_mb_seq_groups_ops;
2815 extern long ext4_mb_stats;
2816 extern long ext4_mb_max_to_scan;
2817 extern int ext4_seq_mb_stats_show(struct seq_file *seq, void *offset);
2818 extern int ext4_mb_init(struct super_block *);
2819 extern int ext4_mb_release(struct super_block *);
2820 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2821 struct ext4_allocation_request *, int *);
2822 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2823 extern void ext4_discard_preallocations(struct inode *, unsigned int);
2824 extern int __init ext4_init_mballoc(void);
2825 extern void ext4_exit_mballoc(void);
2826 extern ext4_group_t ext4_mb_prefetch(struct super_block *sb,
2827 ext4_group_t group,
2828 unsigned int nr, int *cnt);
2829 extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group,
2830 unsigned int nr);
2831
2832 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2833 struct buffer_head *bh, ext4_fsblk_t block,
2834 unsigned long count, int flags);
2835 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2836 ext4_group_t ngroups);
2837 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2838 ext4_group_t i, struct ext4_group_desc *desc);
2839 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2840 ext4_fsblk_t block, unsigned long count);
2841 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2842 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid);
2843 extern void ext4_mb_mark_bb(struct super_block *sb, ext4_fsblk_t block,
2844 int len, int state);
2845
2846 /* inode.c */
2847 void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw,
2848 struct ext4_inode_info *ei);
2849 int ext4_inode_is_fast_symlink(struct inode *inode);
2850 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2851 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2852 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count,
2853 bool wait, struct buffer_head **bhs);
2854 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2855 struct buffer_head *bh_result, int create);
2856 int ext4_get_block(struct inode *inode, sector_t iblock,
2857 struct buffer_head *bh_result, int create);
2858 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2859 struct buffer_head *bh, int create);
2860 int ext4_walk_page_buffers(handle_t *handle,
2861 struct buffer_head *head,
2862 unsigned from,
2863 unsigned to,
2864 int *partial,
2865 int (*fn)(handle_t *handle,
2866 struct buffer_head *bh));
2867 int do_journal_get_write_access(handle_t *handle,
2868 struct buffer_head *bh);
2869 #define FALL_BACK_TO_NONDELALLOC 1
2870 #define CONVERT_INLINE_DATA 2
2871
2872 typedef enum {
2873 EXT4_IGET_NORMAL = 0,
2874 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */
2875 EXT4_IGET_HANDLE = 0x0002, /* Inode # is from a handle */
2876 EXT4_IGET_BAD = 0x0004, /* Allow to iget a bad inode */
2877 EXT4_IGET_EA_INODE = 0x0008 /* Inode should contain an EA value */
2878 } ext4_iget_flags;
2879
2880 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
2881 ext4_iget_flags flags, const char *function,
2882 unsigned int line);
2883
2884 #define ext4_iget(sb, ino, flags) \
2885 __ext4_iget((sb), (ino), (flags), __func__, __LINE__)
2886
2887 extern int ext4_write_inode(struct inode *, struct writeback_control *);
2888 extern int ext4_setattr(struct dentry *, struct iattr *);
2889 extern int ext4_getattr(const struct path *, struct kstat *, u32, unsigned int);
2890 extern void ext4_evict_inode(struct inode *);
2891 extern void ext4_clear_inode(struct inode *);
2892 extern int ext4_file_getattr(const struct path *, struct kstat *, u32, unsigned int);
2893 extern int ext4_sync_inode(handle_t *, struct inode *);
2894 extern void ext4_dirty_inode(struct inode *, int);
2895 extern int ext4_change_inode_journal_flag(struct inode *, int);
2896 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2897 extern int ext4_get_fc_inode_loc(struct super_block *sb, unsigned long ino,
2898 struct ext4_iloc *iloc);
2899 extern int ext4_inode_attach_jinode(struct inode *inode);
2900 extern int ext4_can_truncate(struct inode *inode);
2901 extern int ext4_truncate(struct inode *);
2902 extern int ext4_break_layouts(struct inode *);
2903 extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length);
2904 extern void ext4_set_inode_flags(struct inode *, bool init);
2905 extern int ext4_alloc_da_blocks(struct inode *inode);
2906 extern void ext4_set_aops(struct inode *inode);
2907 extern int ext4_writepage_trans_blocks(struct inode *);
2908 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2909 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2910 loff_t lstart, loff_t lend);
2911 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf);
2912 extern vm_fault_t ext4_filemap_fault(struct vm_fault *vmf);
2913 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2914 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
2915 extern void ext4_da_release_space(struct inode *inode, int to_free);
2916 extern void ext4_da_update_reserve_space(struct inode *inode,
2917 int used, int quota_claim);
2918 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
2919 ext4_fsblk_t pblk, ext4_lblk_t len);
2920
2921 /* indirect.c */
2922 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2923 struct ext4_map_blocks *map, int flags);
2924 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2925 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2926 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2927 ext4_lblk_t start, ext4_lblk_t end);
2928
2929 /* ioctl.c */
2930 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2931 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2932 extern void ext4_reset_inode_seed(struct inode *inode);
2933
2934 /* migrate.c */
2935 extern int ext4_ext_migrate(struct inode *);
2936 extern int ext4_ind_migrate(struct inode *inode);
2937
2938 /* namei.c */
2939 extern int ext4_init_new_dir(handle_t *handle, struct inode *dir,
2940 struct inode *inode);
2941 extern int ext4_dirblock_csum_verify(struct inode *inode,
2942 struct buffer_head *bh);
2943 extern int ext4_orphan_add(handle_t *, struct inode *);
2944 extern int ext4_orphan_del(handle_t *, struct inode *);
2945 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2946 __u32 start_minor_hash, __u32 *next_hash);
2947 extern int ext4_search_dir(struct buffer_head *bh,
2948 char *search_buf,
2949 int buf_size,
2950 struct inode *dir,
2951 struct ext4_filename *fname,
2952 unsigned int offset,
2953 struct ext4_dir_entry_2 **res_dir);
2954 extern int ext4_generic_delete_entry(struct inode *dir,
2955 struct ext4_dir_entry_2 *de_del,
2956 struct buffer_head *bh,
2957 void *entry_buf,
2958 int buf_size,
2959 int csum_size);
2960 extern bool ext4_empty_dir(struct inode *inode);
2961
2962 /* resize.c */
2963 extern void ext4_kvfree_array_rcu(void *to_free);
2964 extern int ext4_group_add(struct super_block *sb,
2965 struct ext4_new_group_data *input);
2966 extern int ext4_group_extend(struct super_block *sb,
2967 struct ext4_super_block *es,
2968 ext4_fsblk_t n_blocks_count);
2969 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2970
2971 /* super.c */
2972 extern struct buffer_head *ext4_sb_bread(struct super_block *sb,
2973 sector_t block, int op_flags);
2974 extern struct buffer_head *ext4_sb_bread_unmovable(struct super_block *sb,
2975 sector_t block);
2976 extern void ext4_read_bh_nowait(struct buffer_head *bh, int op_flags,
2977 bh_end_io_t *end_io);
2978 extern int ext4_read_bh(struct buffer_head *bh, int op_flags,
2979 bh_end_io_t *end_io);
2980 extern int ext4_read_bh_lock(struct buffer_head *bh, int op_flags, bool wait);
2981 extern void ext4_sb_breadahead_unmovable(struct super_block *sb, sector_t block);
2982 extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
2983 extern int ext4_calculate_overhead(struct super_block *sb);
2984 extern void ext4_superblock_csum_set(struct super_block *sb);
2985 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2986 ext4_group_t ngroup);
2987 extern const char *ext4_decode_error(struct super_block *sb, int errno,
2988 char nbuf[16]);
2989 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb,
2990 ext4_group_t block_group,
2991 unsigned int flags);
2992 extern unsigned int ext4_num_base_meta_blocks(struct super_block *sb,
2993 ext4_group_t block_group);
2994
2995 extern __printf(7, 8)
2996 void __ext4_error(struct super_block *, const char *, unsigned int, bool,
2997 int, __u64, const char *, ...);
2998 extern __printf(6, 7)
2999 void __ext4_error_inode(struct inode *, const char *, unsigned int,
3000 ext4_fsblk_t, int, const char *, ...);
3001 extern __printf(5, 6)
3002 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
3003 const char *, ...);
3004 extern void __ext4_std_error(struct super_block *, const char *,
3005 unsigned int, int);
3006 extern __printf(4, 5)
3007 void __ext4_warning(struct super_block *, const char *, unsigned int,
3008 const char *, ...);
3009 extern __printf(4, 5)
3010 void __ext4_warning_inode(const struct inode *inode, const char *function,
3011 unsigned int line, const char *fmt, ...);
3012 extern __printf(3, 4)
3013 void __ext4_msg(struct super_block *, const char *, const char *, ...);
3014 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
3015 const char *, unsigned int, const char *);
3016 extern __printf(7, 8)
3017 void __ext4_grp_locked_error(const char *, unsigned int,
3018 struct super_block *, ext4_group_t,
3019 unsigned long, ext4_fsblk_t,
3020 const char *, ...);
3021
3022 #define EXT4_ERROR_INODE(inode, fmt, a...) \
3023 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
3024
3025 #define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...) \
3026 __ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a)
3027
3028 #define ext4_error_inode_block(inode, block, err, fmt, a...) \
3029 __ext4_error_inode((inode), __func__, __LINE__, (block), (err), \
3030 (fmt), ## a)
3031
3032 #define EXT4_ERROR_FILE(file, block, fmt, a...) \
3033 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
3034
3035 #define ext4_abort(sb, err, fmt, a...) \
3036 __ext4_error((sb), __func__, __LINE__, true, (err), 0, (fmt), ## a)
3037
3038 #ifdef CONFIG_PRINTK
3039
3040 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
3041 __ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__)
3042 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \
3043 __ext4_error_inode((inode), (func), (line), (block), \
3044 (err), (fmt), ##__VA_ARGS__)
3045 #define ext4_error_file(file, func, line, block, fmt, ...) \
3046 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
3047 #define ext4_error(sb, fmt, ...) \
3048 __ext4_error((sb), __func__, __LINE__, false, 0, 0, (fmt), \
3049 ##__VA_ARGS__)
3050 #define ext4_error_err(sb, err, fmt, ...) \
3051 __ext4_error((sb), __func__, __LINE__, false, (err), 0, (fmt), \
3052 ##__VA_ARGS__)
3053 #define ext4_warning(sb, fmt, ...) \
3054 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
3055 #define ext4_warning_inode(inode, fmt, ...) \
3056 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
3057 #define ext4_msg(sb, level, fmt, ...) \
3058 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
3059 #define dump_mmp_msg(sb, mmp, msg) \
3060 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
3061 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
3062 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
3063 fmt, ##__VA_ARGS__)
3064
3065 #else
3066
3067 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
3068 do { \
3069 no_printk(fmt, ##__VA_ARGS__); \
3070 __ext4_error_inode(inode, "", 0, block, 0, " "); \
3071 } while (0)
3072 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \
3073 do { \
3074 no_printk(fmt, ##__VA_ARGS__); \
3075 __ext4_error_inode(inode, "", 0, block, err, " "); \
3076 } while (0)
3077 #define ext4_error_file(file, func, line, block, fmt, ...) \
3078 do { \
3079 no_printk(fmt, ##__VA_ARGS__); \
3080 __ext4_error_file(file, "", 0, block, " "); \
3081 } while (0)
3082 #define ext4_error(sb, fmt, ...) \
3083 do { \
3084 no_printk(fmt, ##__VA_ARGS__); \
3085 __ext4_error(sb, "", 0, false, 0, 0, " "); \
3086 } while (0)
3087 #define ext4_error_err(sb, err, fmt, ...) \
3088 do { \
3089 no_printk(fmt, ##__VA_ARGS__); \
3090 __ext4_error(sb, "", 0, false, err, 0, " "); \
3091 } while (0)
3092 #define ext4_warning(sb, fmt, ...) \
3093 do { \
3094 no_printk(fmt, ##__VA_ARGS__); \
3095 __ext4_warning(sb, "", 0, " "); \
3096 } while (0)
3097 #define ext4_warning_inode(inode, fmt, ...) \
3098 do { \
3099 no_printk(fmt, ##__VA_ARGS__); \
3100 __ext4_warning_inode(inode, "", 0, " "); \
3101 } while (0)
3102 #define ext4_msg(sb, level, fmt, ...) \
3103 do { \
3104 no_printk(fmt, ##__VA_ARGS__); \
3105 __ext4_msg(sb, "", " "); \
3106 } while (0)
3107 #define dump_mmp_msg(sb, mmp, msg) \
3108 __dump_mmp_msg(sb, mmp, "", 0, "")
3109 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
3110 do { \
3111 no_printk(fmt, ##__VA_ARGS__); \
3112 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
3113 } while (0)
3114
3115 #endif
3116
3117 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
3118 struct ext4_group_desc *bg);
3119 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
3120 struct ext4_group_desc *bg);
3121 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
3122 struct ext4_group_desc *bg);
3123 extern __u32 ext4_free_group_clusters(struct super_block *sb,
3124 struct ext4_group_desc *bg);
3125 extern __u32 ext4_free_inodes_count(struct super_block *sb,
3126 struct ext4_group_desc *bg);
3127 extern __u32 ext4_used_dirs_count(struct super_block *sb,
3128 struct ext4_group_desc *bg);
3129 extern __u32 ext4_itable_unused_count(struct super_block *sb,
3130 struct ext4_group_desc *bg);
3131 extern void ext4_block_bitmap_set(struct super_block *sb,
3132 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3133 extern void ext4_inode_bitmap_set(struct super_block *sb,
3134 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3135 extern void ext4_inode_table_set(struct super_block *sb,
3136 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3137 extern void ext4_free_group_clusters_set(struct super_block *sb,
3138 struct ext4_group_desc *bg,
3139 __u32 count);
3140 extern void ext4_free_inodes_set(struct super_block *sb,
3141 struct ext4_group_desc *bg, __u32 count);
3142 extern void ext4_used_dirs_set(struct super_block *sb,
3143 struct ext4_group_desc *bg, __u32 count);
3144 extern void ext4_itable_unused_set(struct super_block *sb,
3145 struct ext4_group_desc *bg, __u32 count);
3146 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
3147 struct ext4_group_desc *gdp);
3148 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
3149 struct ext4_group_desc *gdp);
3150 extern int ext4_register_li_request(struct super_block *sb,
3151 ext4_group_t first_not_zeroed);
3152
ext4_has_metadata_csum(struct super_block * sb)3153 static inline int ext4_has_metadata_csum(struct super_block *sb)
3154 {
3155 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
3156 !EXT4_SB(sb)->s_chksum_driver);
3157
3158 return ext4_has_feature_metadata_csum(sb) &&
3159 (EXT4_SB(sb)->s_chksum_driver != NULL);
3160 }
3161
ext4_has_group_desc_csum(struct super_block * sb)3162 static inline int ext4_has_group_desc_csum(struct super_block *sb)
3163 {
3164 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb);
3165 }
3166
3167 #define ext4_read_incompat_64bit_val(es, name) \
3168 (((es)->s_feature_incompat & cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT) \
3169 ? (ext4_fsblk_t)le32_to_cpu(es->name##_hi) << 32 : 0) | \
3170 le32_to_cpu(es->name##_lo))
3171
ext4_blocks_count(struct ext4_super_block * es)3172 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
3173 {
3174 return ext4_read_incompat_64bit_val(es, s_blocks_count);
3175 }
3176
ext4_r_blocks_count(struct ext4_super_block * es)3177 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
3178 {
3179 return ext4_read_incompat_64bit_val(es, s_r_blocks_count);
3180 }
3181
ext4_free_blocks_count(struct ext4_super_block * es)3182 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
3183 {
3184 return ext4_read_incompat_64bit_val(es, s_free_blocks_count);
3185 }
3186
ext4_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)3187 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
3188 ext4_fsblk_t blk)
3189 {
3190 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
3191 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
3192 }
3193
ext4_free_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)3194 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
3195 ext4_fsblk_t blk)
3196 {
3197 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
3198 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
3199 }
3200
ext4_r_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)3201 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
3202 ext4_fsblk_t blk)
3203 {
3204 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
3205 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
3206 }
3207
ext4_isize(struct super_block * sb,struct ext4_inode * raw_inode)3208 static inline loff_t ext4_isize(struct super_block *sb,
3209 struct ext4_inode *raw_inode)
3210 {
3211 if (ext4_has_feature_largedir(sb) ||
3212 S_ISREG(le16_to_cpu(raw_inode->i_mode)))
3213 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
3214 le32_to_cpu(raw_inode->i_size_lo);
3215
3216 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
3217 }
3218
ext4_isize_set(struct ext4_inode * raw_inode,loff_t i_size)3219 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
3220 {
3221 raw_inode->i_size_lo = cpu_to_le32(i_size);
3222 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
3223 }
3224
3225 /*
3226 * Reading s_groups_count requires using smp_rmb() afterwards. See
3227 * the locking protocol documented in the comments of ext4_group_add()
3228 * in resize.c
3229 */
ext4_get_groups_count(struct super_block * sb)3230 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
3231 {
3232 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
3233
3234 smp_rmb();
3235 return ngroups;
3236 }
3237
ext4_flex_group(struct ext4_sb_info * sbi,ext4_group_t block_group)3238 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
3239 ext4_group_t block_group)
3240 {
3241 return block_group >> sbi->s_log_groups_per_flex;
3242 }
3243
ext4_flex_bg_size(struct ext4_sb_info * sbi)3244 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
3245 {
3246 return 1 << sbi->s_log_groups_per_flex;
3247 }
3248
3249 #define ext4_std_error(sb, errno) \
3250 do { \
3251 if ((errno)) \
3252 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
3253 } while (0)
3254
3255 #ifdef CONFIG_SMP
3256 /* Each CPU can accumulate percpu_counter_batch clusters in their local
3257 * counters. So we need to make sure we have free clusters more
3258 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
3259 */
3260 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
3261 #else
3262 #define EXT4_FREECLUSTERS_WATERMARK 0
3263 #endif
3264
3265 /* Update i_disksize. Requires i_mutex to avoid races with truncate */
ext4_update_i_disksize(struct inode * inode,loff_t newsize)3266 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
3267 {
3268 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
3269 !inode_is_locked(inode));
3270 down_write(&EXT4_I(inode)->i_data_sem);
3271 if (newsize > EXT4_I(inode)->i_disksize)
3272 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize);
3273 up_write(&EXT4_I(inode)->i_data_sem);
3274 }
3275
3276 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
ext4_update_inode_size(struct inode * inode,loff_t newsize)3277 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
3278 {
3279 int changed = 0;
3280
3281 if (newsize > inode->i_size) {
3282 i_size_write(inode, newsize);
3283 changed = 1;
3284 }
3285 if (newsize > EXT4_I(inode)->i_disksize) {
3286 ext4_update_i_disksize(inode, newsize);
3287 changed |= 2;
3288 }
3289 return changed;
3290 }
3291
3292 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
3293 loff_t len);
3294
3295 struct ext4_group_info {
3296 unsigned long bb_state;
3297 #ifdef AGGRESSIVE_CHECK
3298 unsigned long bb_check_counter;
3299 #endif
3300 struct rb_root bb_free_root;
3301 ext4_grpblk_t bb_first_free; /* first free block */
3302 ext4_grpblk_t bb_free; /* total free blocks */
3303 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
3304 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
3305 struct list_head bb_prealloc_list;
3306 #ifdef DOUBLE_CHECK
3307 void *bb_bitmap;
3308 #endif
3309 struct rw_semaphore alloc_sem;
3310 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
3311 * regions, index is order.
3312 * bb_counters[3] = 5 means
3313 * 5 free 8-block regions. */
3314 };
3315
3316 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
3317 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
3318 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
3319 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
3320 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \
3321 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT)
3322 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \
3323 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT)
3324 #define EXT4_GROUP_INFO_BBITMAP_READ_BIT 4
3325
3326 #define EXT4_MB_GRP_NEED_INIT(grp) \
3327 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
3328 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
3329 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3330 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
3331 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3332
3333 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
3334 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3335 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
3336 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3337 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
3338 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3339 #define EXT4_MB_GRP_TEST_AND_SET_READ(grp) \
3340 (test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state)))
3341
3342 #define EXT4_MAX_CONTENTION 8
3343 #define EXT4_CONTENTION_THRESHOLD 2
3344
ext4_group_lock_ptr(struct super_block * sb,ext4_group_t group)3345 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
3346 ext4_group_t group)
3347 {
3348 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
3349 }
3350
3351 /*
3352 * Returns true if the filesystem is busy enough that attempts to
3353 * access the block group locks has run into contention.
3354 */
ext4_fs_is_busy(struct ext4_sb_info * sbi)3355 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
3356 {
3357 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
3358 }
3359
ext4_lock_group(struct super_block * sb,ext4_group_t group)3360 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
3361 {
3362 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
3363 if (spin_trylock(lock))
3364 /*
3365 * We're able to grab the lock right away, so drop the
3366 * lock contention counter.
3367 */
3368 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
3369 else {
3370 /*
3371 * The lock is busy, so bump the contention counter,
3372 * and then wait on the spin lock.
3373 */
3374 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
3375 EXT4_MAX_CONTENTION);
3376 spin_lock(lock);
3377 }
3378 }
3379
ext4_unlock_group(struct super_block * sb,ext4_group_t group)3380 static inline void ext4_unlock_group(struct super_block *sb,
3381 ext4_group_t group)
3382 {
3383 spin_unlock(ext4_group_lock_ptr(sb, group));
3384 }
3385
3386 /*
3387 * Block validity checking
3388 */
3389 #define ext4_check_indirect_blockref(inode, bh) \
3390 ext4_check_blockref(__func__, __LINE__, inode, \
3391 (__le32 *)(bh)->b_data, \
3392 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
3393
3394 #define ext4_ind_check_inode(inode) \
3395 ext4_check_blockref(__func__, __LINE__, inode, \
3396 EXT4_I(inode)->i_data, \
3397 EXT4_NDIR_BLOCKS)
3398
3399 /*
3400 * Inodes and files operations
3401 */
3402
3403 /* dir.c */
3404 extern const struct file_operations ext4_dir_operations;
3405
3406 #ifdef CONFIG_UNICODE
3407 extern const struct dentry_operations ext4_dentry_ops;
3408 #endif
3409
3410 /* file.c */
3411 extern const struct inode_operations ext4_file_inode_operations;
3412 extern const struct file_operations ext4_file_operations;
3413 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
3414
3415 /* inline.c */
3416 extern int ext4_get_max_inline_size(struct inode *inode);
3417 extern int ext4_find_inline_data_nolock(struct inode *inode);
3418 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
3419 unsigned int len);
3420 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
3421
3422 extern int ext4_readpage_inline(struct inode *inode, struct page *page);
3423 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3424 struct inode *inode,
3425 loff_t pos, unsigned len,
3426 unsigned flags,
3427 struct page **pagep);
3428 extern int ext4_write_inline_data_end(struct inode *inode,
3429 loff_t pos, unsigned len,
3430 unsigned copied,
3431 struct page *page);
3432 extern struct buffer_head *
3433 ext4_journalled_write_inline_data(struct inode *inode,
3434 unsigned len,
3435 struct page *page);
3436 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
3437 struct inode *inode,
3438 loff_t pos, unsigned len,
3439 unsigned flags,
3440 struct page **pagep,
3441 void **fsdata);
3442 extern int ext4_try_add_inline_entry(handle_t *handle,
3443 struct ext4_filename *fname,
3444 struct inode *dir, struct inode *inode);
3445 extern int ext4_try_create_inline_dir(handle_t *handle,
3446 struct inode *parent,
3447 struct inode *inode);
3448 extern int ext4_read_inline_dir(struct file *filp,
3449 struct dir_context *ctx,
3450 int *has_inline_data);
3451 extern int ext4_inlinedir_to_tree(struct file *dir_file,
3452 struct inode *dir, ext4_lblk_t block,
3453 struct dx_hash_info *hinfo,
3454 __u32 start_hash, __u32 start_minor_hash,
3455 int *has_inline_data);
3456 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3457 struct ext4_filename *fname,
3458 struct ext4_dir_entry_2 **res_dir,
3459 int *has_inline_data);
3460 extern int ext4_delete_inline_entry(handle_t *handle,
3461 struct inode *dir,
3462 struct ext4_dir_entry_2 *de_del,
3463 struct buffer_head *bh,
3464 int *has_inline_data);
3465 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3466 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3467 struct ext4_dir_entry_2 **parent_de,
3468 int *retval);
3469 extern int ext4_inline_data_fiemap(struct inode *inode,
3470 struct fiemap_extent_info *fieinfo,
3471 int *has_inline, __u64 start, __u64 len);
3472
3473 struct iomap;
3474 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap);
3475
3476 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3477
3478 extern int ext4_convert_inline_data(struct inode *inode);
3479
ext4_has_inline_data(struct inode * inode)3480 static inline int ext4_has_inline_data(struct inode *inode)
3481 {
3482 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3483 EXT4_I(inode)->i_inline_off;
3484 }
3485
3486 /* namei.c */
3487 extern const struct inode_operations ext4_dir_inode_operations;
3488 extern const struct inode_operations ext4_special_inode_operations;
3489 extern struct dentry *ext4_get_parent(struct dentry *child);
3490 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3491 struct ext4_dir_entry_2 *de,
3492 int blocksize, int csum_size,
3493 unsigned int parent_ino, int dotdot_real_len);
3494 extern void ext4_initialize_dirent_tail(struct buffer_head *bh,
3495 unsigned int blocksize);
3496 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode,
3497 struct buffer_head *bh);
3498 extern int ext4_ci_compare(const struct inode *parent,
3499 const struct qstr *fname,
3500 const struct qstr *entry, bool quick);
3501 extern int __ext4_unlink(struct inode *dir, const struct qstr *d_name,
3502 struct inode *inode, struct dentry *dentry);
3503 extern int __ext4_link(struct inode *dir, struct inode *inode,
3504 struct dentry *dentry);
3505
3506 #define S_SHIFT 12
3507 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = {
3508 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
3509 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
3510 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
3511 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
3512 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
3513 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
3514 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
3515 };
3516
ext4_set_de_type(struct super_block * sb,struct ext4_dir_entry_2 * de,umode_t mode)3517 static inline void ext4_set_de_type(struct super_block *sb,
3518 struct ext4_dir_entry_2 *de,
3519 umode_t mode) {
3520 if (ext4_has_feature_filetype(sb))
3521 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3522 }
3523
3524 /* readpages.c */
3525 extern int ext4_mpage_readpages(struct inode *inode,
3526 struct readahead_control *rac, struct page *page);
3527 extern int __init ext4_init_post_read_processing(void);
3528 extern void ext4_exit_post_read_processing(void);
3529
3530 /* symlink.c */
3531 extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3532 extern const struct inode_operations ext4_symlink_inode_operations;
3533 extern const struct inode_operations ext4_fast_symlink_inode_operations;
3534
3535 /* sysfs.c */
3536 extern int ext4_register_sysfs(struct super_block *sb);
3537 extern void ext4_unregister_sysfs(struct super_block *sb);
3538 extern int __init ext4_init_sysfs(void);
3539 extern void ext4_exit_sysfs(void);
3540
3541 /* block_validity */
3542 extern void ext4_release_system_zone(struct super_block *sb);
3543 extern int ext4_setup_system_zone(struct super_block *sb);
3544 extern int __init ext4_init_system_zone(void);
3545 extern void ext4_exit_system_zone(void);
3546 extern int ext4_inode_block_valid(struct inode *inode,
3547 ext4_fsblk_t start_blk,
3548 unsigned int count);
3549 extern int ext4_check_blockref(const char *, unsigned int,
3550 struct inode *, __le32 *, unsigned int);
3551 extern int ext4_sb_block_valid(struct super_block *sb, struct inode *inode,
3552 ext4_fsblk_t start_blk, unsigned int count);
3553
3554
3555 /* extents.c */
3556 struct ext4_ext_path;
3557 struct ext4_extent;
3558
3559 /*
3560 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3561 * __le32.
3562 */
3563 #define EXT_MAX_BLOCKS 0xffffffff
3564
3565 extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode);
3566 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3567 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3568 struct ext4_map_blocks *map, int flags);
3569 extern int ext4_ext_truncate(handle_t *, struct inode *);
3570 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3571 ext4_lblk_t end);
3572 extern void ext4_ext_init(struct super_block *);
3573 extern void ext4_ext_release(struct super_block *);
3574 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3575 loff_t len);
3576 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3577 loff_t offset, ssize_t len);
3578 extern int ext4_convert_unwritten_io_end_vec(handle_t *handle,
3579 ext4_io_end_t *io_end);
3580 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3581 struct ext4_map_blocks *map, int flags);
3582 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3583 int num,
3584 struct ext4_ext_path *path);
3585 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3586 struct ext4_ext_path **,
3587 struct ext4_extent *, int);
3588 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3589 struct ext4_ext_path **,
3590 int flags);
3591 extern void ext4_ext_drop_refs(struct ext4_ext_path *);
3592 extern int ext4_ext_check_inode(struct inode *inode);
3593 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3594 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3595 __u64 start, __u64 len);
3596 extern int ext4_get_es_cache(struct inode *inode,
3597 struct fiemap_extent_info *fieinfo,
3598 __u64 start, __u64 len);
3599 extern int ext4_ext_precache(struct inode *inode);
3600 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3601 struct inode *inode2, ext4_lblk_t lblk1,
3602 ext4_lblk_t lblk2, ext4_lblk_t count,
3603 int mark_unwritten,int *err);
3604 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu);
3605 extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode,
3606 int check_cred, int restart_cred,
3607 int revoke_cred);
3608 extern void ext4_ext_replay_shrink_inode(struct inode *inode, ext4_lblk_t end);
3609 extern int ext4_ext_replay_set_iblocks(struct inode *inode);
3610 extern int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start,
3611 int len, int unwritten, ext4_fsblk_t pblk);
3612 extern int ext4_ext_clear_bb(struct inode *inode);
3613
3614
3615 /* move_extent.c */
3616 extern void ext4_double_down_write_data_sem(struct inode *first,
3617 struct inode *second);
3618 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3619 struct inode *donor_inode);
3620 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3621 __u64 start_orig, __u64 start_donor,
3622 __u64 len, __u64 *moved_len);
3623
3624 /* page-io.c */
3625 extern int __init ext4_init_pageio(void);
3626 extern void ext4_exit_pageio(void);
3627 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3628 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3629 extern int ext4_put_io_end(ext4_io_end_t *io_end);
3630 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3631 extern void ext4_io_submit_init(struct ext4_io_submit *io,
3632 struct writeback_control *wbc);
3633 extern void ext4_end_io_rsv_work(struct work_struct *work);
3634 extern void ext4_io_submit(struct ext4_io_submit *io);
3635 extern int ext4_bio_write_page(struct ext4_io_submit *io,
3636 struct page *page,
3637 int len,
3638 struct writeback_control *wbc,
3639 bool keep_towrite);
3640 extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end);
3641 extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end);
3642
3643 /* mmp.c */
3644 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3645
3646 /* mmp.c */
3647 extern void ext4_stop_mmpd(struct ext4_sb_info *sbi);
3648
3649 /* verity.c */
3650 extern const struct fsverity_operations ext4_verityops;
3651
3652 /*
3653 * Add new method to test whether block and inode bitmaps are properly
3654 * initialized. With uninit_bg reading the block from disk is not enough
3655 * to mark the bitmap uptodate. We need to also zero-out the bitmap
3656 */
3657 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3658
bitmap_uptodate(struct buffer_head * bh)3659 static inline int bitmap_uptodate(struct buffer_head *bh)
3660 {
3661 return (buffer_uptodate(bh) &&
3662 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3663 }
set_bitmap_uptodate(struct buffer_head * bh)3664 static inline void set_bitmap_uptodate(struct buffer_head *bh)
3665 {
3666 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3667 }
3668
3669 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
3670
3671 /* For ioend & aio unwritten conversion wait queues */
3672 #define EXT4_WQ_HASH_SZ 37
3673 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
3674 EXT4_WQ_HASH_SZ])
3675 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3676
3677 extern int ext4_resize_begin(struct super_block *sb);
3678 extern void ext4_resize_end(struct super_block *sb);
3679
ext4_set_io_unwritten_flag(struct inode * inode,struct ext4_io_end * io_end)3680 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
3681 struct ext4_io_end *io_end)
3682 {
3683 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
3684 io_end->flag |= EXT4_IO_END_UNWRITTEN;
3685 atomic_inc(&EXT4_I(inode)->i_unwritten);
3686 }
3687 }
3688
ext4_clear_io_unwritten_flag(ext4_io_end_t * io_end)3689 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3690 {
3691 struct inode *inode = io_end->inode;
3692
3693 if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
3694 io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3695 /* Wake up anyone waiting on unwritten extent conversion */
3696 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
3697 wake_up_all(ext4_ioend_wq(inode));
3698 }
3699 }
3700
3701 extern const struct iomap_ops ext4_iomap_ops;
3702 extern const struct iomap_ops ext4_iomap_overwrite_ops;
3703 extern const struct iomap_ops ext4_iomap_report_ops;
3704
ext4_buffer_uptodate(struct buffer_head * bh)3705 static inline int ext4_buffer_uptodate(struct buffer_head *bh)
3706 {
3707 /*
3708 * If the buffer has the write error flag, we have failed
3709 * to write out data in the block. In this case, we don't
3710 * have to read the block because we may read the old data
3711 * successfully.
3712 */
3713 if (!buffer_uptodate(bh) && buffer_write_io_error(bh))
3714 set_buffer_uptodate(bh);
3715 return buffer_uptodate(bh);
3716 }
3717
3718 #endif /* __KERNEL__ */
3719
3720 #define EFSBADCRC EBADMSG /* Bad CRC detected */
3721 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
3722
3723 #endif /* _EXT4_H */
3724