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/refcount.h>
21 #include <linux/types.h>
22 #include <linux/blkdev.h>
23 #include <linux/magic.h>
24 #include <linux/jbd2.h>
25 #include <linux/quota.h>
26 #include <linux/rwsem.h>
27 #include <linux/rbtree.h>
28 #include <linux/seqlock.h>
29 #include <linux/mutex.h>
30 #include <linux/timer.h>
31 #include <linux/wait.h>
32 #include <linux/sched/signal.h>
33 #include <linux/blockgroup_lock.h>
34 #include <linux/percpu_counter.h>
35 #include <linux/ratelimit.h>
36 #include <crypto/hash.h>
37 #include <linux/falloc.h>
38 #include <linux/percpu-rwsem.h>
39 #include <linux/fiemap.h>
40 #ifdef __KERNEL__
41 #include <linux/compat.h>
42 #endif
43
44 #include <linux/fscrypt.h>
45 #include <linux/fsverity.h>
46
47 #include <linux/compiler.h>
48
49 /*
50 * The fourth extended filesystem constants/structures
51 */
52
53 /*
54 * with AGGRESSIVE_CHECK allocator runs consistency checks over
55 * structures. these checks slow things down a lot
56 */
57 #define AGGRESSIVE_CHECK__
58
59 /*
60 * with DOUBLE_CHECK defined mballoc creates persistent in-core
61 * bitmaps, maintains and uses them to check for double allocations
62 */
63 #define DOUBLE_CHECK__
64
65 /*
66 * Define EXT4FS_DEBUG to produce debug messages
67 */
68 #undef EXT4FS_DEBUG
69
70 /*
71 * Debug code
72 */
73 #ifdef EXT4FS_DEBUG
74 #define ext4_debug(f, a...) \
75 do { \
76 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
77 __FILE__, __LINE__, __func__); \
78 printk(KERN_DEBUG f, ## a); \
79 } while (0)
80 #else
81 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
82 #endif
83
84 /*
85 * Turn on EXT_DEBUG to enable ext4_ext_show_path/leaf/move in extents.c
86 */
87 #define EXT_DEBUG__
88
89 /*
90 * Dynamic printk for controlled extents debugging.
91 */
92 #ifdef CONFIG_EXT4_DEBUG
93 #define ext_debug(ino, fmt, ...) \
94 pr_debug("[%s/%d] EXT4-fs (%s): ino %lu: (%s, %d): %s:" fmt, \
95 current->comm, task_pid_nr(current), \
96 ino->i_sb->s_id, ino->i_ino, __FILE__, __LINE__, \
97 __func__, ##__VA_ARGS__)
98 #else
99 #define ext_debug(ino, fmt, ...) no_printk(fmt, ##__VA_ARGS__)
100 #endif
101
102 #define ASSERT(assert) \
103 do { \
104 if (unlikely(!(assert))) { \
105 printk(KERN_EMERG \
106 "Assertion failure in %s() at %s:%d: '%s'\n", \
107 __func__, __FILE__, __LINE__, #assert); \
108 BUG(); \
109 } \
110 } while (0)
111
112 /* data type for block offset of block group */
113 typedef int ext4_grpblk_t;
114
115 /* data type for filesystem-wide blocks number */
116 typedef unsigned long long ext4_fsblk_t;
117
118 /* data type for file logical block number */
119 typedef __u32 ext4_lblk_t;
120
121 /* data type for block group number */
122 typedef unsigned int ext4_group_t;
123
124 enum SHIFT_DIRECTION {
125 SHIFT_LEFT = 0,
126 SHIFT_RIGHT,
127 };
128
129 /*
130 * Flags used in mballoc's allocation_context flags field.
131 *
132 * Also used to show what's going on for debugging purposes when the
133 * flag field is exported via the traceport interface
134 */
135
136 /* prefer goal again. length */
137 #define EXT4_MB_HINT_MERGE 0x0001
138 /* blocks already reserved */
139 #define EXT4_MB_HINT_RESERVED 0x0002
140 /* metadata is being allocated */
141 #define EXT4_MB_HINT_METADATA 0x0004
142 /* first blocks in the file */
143 #define EXT4_MB_HINT_FIRST 0x0008
144 /* search for the best chunk */
145 #define EXT4_MB_HINT_BEST 0x0010
146 /* data is being allocated */
147 #define EXT4_MB_HINT_DATA 0x0020
148 /* don't preallocate (for tails) */
149 #define EXT4_MB_HINT_NOPREALLOC 0x0040
150 /* allocate for locality group */
151 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080
152 /* allocate goal blocks or none */
153 #define EXT4_MB_HINT_GOAL_ONLY 0x0100
154 /* goal is meaningful */
155 #define EXT4_MB_HINT_TRY_GOAL 0x0200
156 /* blocks already pre-reserved by delayed allocation */
157 #define EXT4_MB_DELALLOC_RESERVED 0x0400
158 /* We are doing stream allocation */
159 #define EXT4_MB_STREAM_ALLOC 0x0800
160 /* Use reserved root blocks if needed */
161 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000
162 /* Use blocks from reserved pool */
163 #define EXT4_MB_USE_RESERVED 0x2000
164 /* Do strict check for free blocks while retrying block allocation */
165 #define EXT4_MB_STRICT_CHECK 0x4000
166 /* Large fragment size list lookup succeeded at least once for cr = 0 */
167 #define EXT4_MB_CR0_OPTIMIZED 0x8000
168 /* Avg fragment size rb tree lookup succeeded at least once for cr = 1 */
169 #define EXT4_MB_CR1_OPTIMIZED 0x00010000
170 struct ext4_allocation_request {
171 /* target inode for block we're allocating */
172 struct inode *inode;
173 /* how many blocks we want to allocate */
174 unsigned int len;
175 /* logical block in target inode */
176 ext4_lblk_t logical;
177 /* the closest logical allocated block to the left */
178 ext4_lblk_t lleft;
179 /* the closest logical allocated block to the right */
180 ext4_lblk_t lright;
181 /* phys. target (a hint) */
182 ext4_fsblk_t goal;
183 /* phys. block for the closest logical allocated block to the left */
184 ext4_fsblk_t pleft;
185 /* phys. block for the closest logical allocated block to the right */
186 ext4_fsblk_t pright;
187 /* flags. see above EXT4_MB_HINT_* */
188 unsigned int flags;
189 };
190
191 /*
192 * Logical to physical block mapping, used by ext4_map_blocks()
193 *
194 * This structure is used to pass requests into ext4_map_blocks() as
195 * well as to store the information returned by ext4_map_blocks(). It
196 * takes less room on the stack than a struct buffer_head.
197 */
198 #define EXT4_MAP_NEW BIT(BH_New)
199 #define EXT4_MAP_MAPPED BIT(BH_Mapped)
200 #define EXT4_MAP_UNWRITTEN BIT(BH_Unwritten)
201 #define EXT4_MAP_BOUNDARY BIT(BH_Boundary)
202 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
203 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
204
205 struct ext4_map_blocks {
206 ext4_fsblk_t m_pblk;
207 ext4_lblk_t m_lblk;
208 unsigned int m_len;
209 unsigned int m_flags;
210 };
211
212 /*
213 * Block validity checking, system zone rbtree.
214 */
215 struct ext4_system_blocks {
216 struct rb_root root;
217 struct rcu_head rcu;
218 };
219
220 /*
221 * Flags for ext4_io_end->flags
222 */
223 #define EXT4_IO_END_UNWRITTEN 0x0001
224
225 struct ext4_io_end_vec {
226 struct list_head list; /* list of io_end_vec */
227 loff_t offset; /* offset in the file */
228 ssize_t size; /* size of the extent */
229 };
230
231 /*
232 * For converting unwritten extents on a work queue. 'handle' is used for
233 * buffered writeback.
234 */
235 typedef struct ext4_io_end {
236 struct list_head list; /* per-file finished IO list */
237 handle_t *handle; /* handle reserved for extent
238 * conversion */
239 struct inode *inode; /* file being written to */
240 struct bio *bio; /* Linked list of completed
241 * bios covering the extent */
242 unsigned int flag; /* unwritten or not */
243 refcount_t count; /* reference counter */
244 struct list_head list_vec; /* list of ext4_io_end_vec */
245 } ext4_io_end_t;
246
247 struct ext4_io_submit {
248 struct writeback_control *io_wbc;
249 struct bio *io_bio;
250 ext4_io_end_t *io_end;
251 sector_t io_next_block;
252 };
253
254 /*
255 * Special inodes numbers
256 */
257 #define EXT4_BAD_INO 1 /* Bad blocks inode */
258 #define EXT4_ROOT_INO 2 /* Root inode */
259 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */
260 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
261 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
262 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
263 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
264 #define EXT4_JOURNAL_INO 8 /* Journal inode */
265
266 /* First non-reserved inode for old ext4 filesystems */
267 #define EXT4_GOOD_OLD_FIRST_INO 11
268
269 /*
270 * Maximal count of links to a file
271 */
272 #define EXT4_LINK_MAX 65000
273
274 /*
275 * Macro-instructions used to manage several block sizes
276 */
277 #define EXT4_MIN_BLOCK_SIZE 1024
278 #define EXT4_MAX_BLOCK_SIZE 65536
279 #define EXT4_MIN_BLOCK_LOG_SIZE 10
280 #define EXT4_MAX_BLOCK_LOG_SIZE 16
281 #define EXT4_MAX_CLUSTER_LOG_SIZE 30
282 #ifdef __KERNEL__
283 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
284 #else
285 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
286 #endif
287 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
288 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
289 EXT4_SB(s)->s_cluster_bits)
290 #ifdef __KERNEL__
291 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
292 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
293 #else
294 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
295 #endif
296 #ifdef __KERNEL__
297 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
298 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
299 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
300 #else
301 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
302 EXT4_GOOD_OLD_INODE_SIZE : \
303 (s)->s_inode_size)
304 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
305 EXT4_GOOD_OLD_FIRST_INO : \
306 (s)->s_first_ino)
307 #endif
308 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
309 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \
310 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \
311 blkbits))
312
313 /* Translate a block number to a cluster number */
314 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
315 /* Translate a cluster number to a block number */
316 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
317 /* Translate # of blks to # of clusters */
318 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
319 (sbi)->s_cluster_bits)
320 /* Mask out the low bits to get the starting block of the cluster */
321 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
322 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
323 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
324 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
325 /* Fill in the low bits to get the last block of the cluster */
326 #define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) | \
327 ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1))
328 /* Get the cluster offset */
329 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
330 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
331 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
332 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
333
334 /*
335 * Structure of a blocks group descriptor
336 */
337 struct ext4_group_desc
338 {
339 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
340 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
341 __le32 bg_inode_table_lo; /* Inodes table block */
342 __le16 bg_free_blocks_count_lo;/* Free blocks count */
343 __le16 bg_free_inodes_count_lo;/* Free inodes count */
344 __le16 bg_used_dirs_count_lo; /* Directories count */
345 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
346 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
347 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
348 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
349 __le16 bg_itable_unused_lo; /* Unused inodes count */
350 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
351 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
352 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
353 __le32 bg_inode_table_hi; /* Inodes table block MSB */
354 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
355 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
356 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
357 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
358 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
359 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
360 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
361 __u32 bg_reserved;
362 };
363
364 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
365 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
366 sizeof(__le16))
367 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
368 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
369 sizeof(__le16))
370
371 /*
372 * Structure of a flex block group info
373 */
374
375 struct flex_groups {
376 atomic64_t free_clusters;
377 atomic_t free_inodes;
378 atomic_t used_dirs;
379 };
380
381 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
382 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
383 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
384
385 /*
386 * Macro-instructions used to manage group descriptors
387 */
388 #define EXT4_MIN_DESC_SIZE 32
389 #define EXT4_MIN_DESC_SIZE_64BIT 64
390 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
391 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
392 #ifdef __KERNEL__
393 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
394 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
395 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
396 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
397 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
398 #else
399 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
400 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
401 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
402 #endif
403
404 /*
405 * Constants relative to the data blocks
406 */
407 #define EXT4_NDIR_BLOCKS 12
408 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
409 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
410 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
411 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
412
413 /*
414 * Inode flags
415 */
416 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
417 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
418 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
419 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
420 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
421 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
422 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
423 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
424 /* Reserved for compression usage... */
425 #define EXT4_DIRTY_FL 0x00000100
426 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
427 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
428 /* nb: was previously EXT2_ECOMPR_FL */
429 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */
430 /* End compression flags --- maybe not all used */
431 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
432 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
433 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
434 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
435 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
436 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
437 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
438 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
439 #define EXT4_VERITY_FL 0x00100000 /* Verity protected inode */
440 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
441 /* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */
442
443 #define EXT4_DAX_FL 0x02000000 /* Inode is DAX */
444
445 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
446 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
447 #define EXT4_CASEFOLD_FL 0x40000000 /* Casefolded directory */
448 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
449
450 /* User modifiable flags */
451 #define EXT4_FL_USER_MODIFIABLE (EXT4_SECRM_FL | \
452 EXT4_UNRM_FL | \
453 EXT4_COMPR_FL | \
454 EXT4_SYNC_FL | \
455 EXT4_IMMUTABLE_FL | \
456 EXT4_APPEND_FL | \
457 EXT4_NODUMP_FL | \
458 EXT4_NOATIME_FL | \
459 EXT4_JOURNAL_DATA_FL | \
460 EXT4_NOTAIL_FL | \
461 EXT4_DIRSYNC_FL | \
462 EXT4_TOPDIR_FL | \
463 EXT4_EXTENTS_FL | \
464 0x00400000 /* EXT4_EOFBLOCKS_FL */ | \
465 EXT4_DAX_FL | \
466 EXT4_PROJINHERIT_FL | \
467 EXT4_CASEFOLD_FL)
468
469 /* User visible flags */
470 #define EXT4_FL_USER_VISIBLE (EXT4_FL_USER_MODIFIABLE | \
471 EXT4_DIRTY_FL | \
472 EXT4_COMPRBLK_FL | \
473 EXT4_NOCOMPR_FL | \
474 EXT4_ENCRYPT_FL | \
475 EXT4_INDEX_FL | \
476 EXT4_VERITY_FL | \
477 EXT4_INLINE_DATA_FL)
478
479 /* Flags that should be inherited by new inodes from their parent. */
480 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
481 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
482 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
483 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
484 EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL |\
485 EXT4_DAX_FL)
486
487 /* Flags that are appropriate for regular files (all but dir-specific ones). */
488 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\
489 EXT4_PROJINHERIT_FL))
490
491 /* Flags that are appropriate for non-directories/regular files. */
492 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
493
494 /* The only flags that should be swapped */
495 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL)
496
497 /* Flags which are mutually exclusive to DAX */
498 #define EXT4_DAX_MUT_EXCL (EXT4_VERITY_FL | EXT4_ENCRYPT_FL |\
499 EXT4_JOURNAL_DATA_FL | EXT4_INLINE_DATA_FL)
500
501 /* Mask out flags that are inappropriate for the given type of inode. */
ext4_mask_flags(umode_t mode,__u32 flags)502 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
503 {
504 if (S_ISDIR(mode))
505 return flags;
506 else if (S_ISREG(mode))
507 return flags & EXT4_REG_FLMASK;
508 else
509 return flags & EXT4_OTHER_FLMASK;
510 }
511
512 /*
513 * Inode flags used for atomic set/get
514 */
515 enum {
516 EXT4_INODE_SECRM = 0, /* Secure deletion */
517 EXT4_INODE_UNRM = 1, /* Undelete */
518 EXT4_INODE_COMPR = 2, /* Compress file */
519 EXT4_INODE_SYNC = 3, /* Synchronous updates */
520 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
521 EXT4_INODE_APPEND = 5, /* writes to file may only append */
522 EXT4_INODE_NODUMP = 6, /* do not dump file */
523 EXT4_INODE_NOATIME = 7, /* do not update atime */
524 /* Reserved for compression usage... */
525 EXT4_INODE_DIRTY = 8,
526 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
527 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
528 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */
529 /* End compression flags --- maybe not all used */
530 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
531 EXT4_INODE_IMAGIC = 13, /* AFS directory */
532 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
533 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
534 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
535 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
536 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
537 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
538 EXT4_INODE_VERITY = 20, /* Verity protected inode */
539 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
540 /* 22 was formerly EXT4_INODE_EOFBLOCKS */
541 EXT4_INODE_DAX = 25, /* Inode is DAX */
542 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
543 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */
544 EXT4_INODE_CASEFOLD = 30, /* Casefolded directory */
545 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
546 };
547
548 /*
549 * Since it's pretty easy to mix up bit numbers and hex values, we use a
550 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
551 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
552 * any extra space in the compiled kernel image, otherwise, the build will fail.
553 * It's important that these values are the same, since we are using
554 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
555 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
556 * values found in ext2, ext3 and ext4 filesystems, and of course the values
557 * defined in e2fsprogs.
558 *
559 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
560 */
561 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1U << EXT4_INODE_##FLAG))
562 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
563
ext4_check_flag_values(void)564 static inline void ext4_check_flag_values(void)
565 {
566 CHECK_FLAG_VALUE(SECRM);
567 CHECK_FLAG_VALUE(UNRM);
568 CHECK_FLAG_VALUE(COMPR);
569 CHECK_FLAG_VALUE(SYNC);
570 CHECK_FLAG_VALUE(IMMUTABLE);
571 CHECK_FLAG_VALUE(APPEND);
572 CHECK_FLAG_VALUE(NODUMP);
573 CHECK_FLAG_VALUE(NOATIME);
574 CHECK_FLAG_VALUE(DIRTY);
575 CHECK_FLAG_VALUE(COMPRBLK);
576 CHECK_FLAG_VALUE(NOCOMPR);
577 CHECK_FLAG_VALUE(ENCRYPT);
578 CHECK_FLAG_VALUE(INDEX);
579 CHECK_FLAG_VALUE(IMAGIC);
580 CHECK_FLAG_VALUE(JOURNAL_DATA);
581 CHECK_FLAG_VALUE(NOTAIL);
582 CHECK_FLAG_VALUE(DIRSYNC);
583 CHECK_FLAG_VALUE(TOPDIR);
584 CHECK_FLAG_VALUE(HUGE_FILE);
585 CHECK_FLAG_VALUE(EXTENTS);
586 CHECK_FLAG_VALUE(VERITY);
587 CHECK_FLAG_VALUE(EA_INODE);
588 CHECK_FLAG_VALUE(INLINE_DATA);
589 CHECK_FLAG_VALUE(PROJINHERIT);
590 CHECK_FLAG_VALUE(CASEFOLD);
591 CHECK_FLAG_VALUE(RESERVED);
592 }
593
594 /* Used to pass group descriptor data when online resize is done */
595 struct ext4_new_group_input {
596 __u32 group; /* Group number for this data */
597 __u64 block_bitmap; /* Absolute block number of block bitmap */
598 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
599 __u64 inode_table; /* Absolute block number of inode table start */
600 __u32 blocks_count; /* Total number of blocks in this group */
601 __u16 reserved_blocks; /* Number of reserved blocks in this group */
602 __u16 unused;
603 };
604
605 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
606 struct compat_ext4_new_group_input {
607 u32 group;
608 compat_u64 block_bitmap;
609 compat_u64 inode_bitmap;
610 compat_u64 inode_table;
611 u32 blocks_count;
612 u16 reserved_blocks;
613 u16 unused;
614 };
615 #endif
616
617 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
618 struct ext4_new_group_data {
619 __u32 group;
620 __u64 block_bitmap;
621 __u64 inode_bitmap;
622 __u64 inode_table;
623 __u32 blocks_count;
624 __u16 reserved_blocks;
625 __u16 mdata_blocks;
626 __u32 free_clusters_count;
627 };
628
629 /* Indexes used to index group tables in ext4_new_group_data */
630 enum {
631 BLOCK_BITMAP = 0, /* block bitmap */
632 INODE_BITMAP, /* inode bitmap */
633 INODE_TABLE, /* inode tables */
634 GROUP_TABLE_COUNT,
635 };
636
637 /*
638 * Flags used by ext4_map_blocks()
639 */
640 /* Allocate any needed blocks and/or convert an unwritten
641 extent to be an initialized ext4 */
642 #define EXT4_GET_BLOCKS_CREATE 0x0001
643 /* Request the creation of an unwritten extent */
644 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002
645 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\
646 EXT4_GET_BLOCKS_CREATE)
647 /* Caller is from the delayed allocation writeout path
648 * finally doing the actual allocation of delayed blocks */
649 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
650 /* caller is from the direct IO path, request to creation of an
651 unwritten extents if not allocated, split the unwritten
652 extent if blocks has been preallocated already*/
653 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
654 #define EXT4_GET_BLOCKS_CONVERT 0x0010
655 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
656 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
657 /* Convert extent to initialized after IO complete */
658 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
659 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
660 /* Eventual metadata allocation (due to growing extent tree)
661 * should not fail, so try to use reserved blocks for that.*/
662 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
663 /* Don't normalize allocation size (used for fallocate) */
664 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
665 /* Convert written extents to unwritten */
666 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100
667 /* Write zeros to newly created written extents */
668 #define EXT4_GET_BLOCKS_ZERO 0x0200
669 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\
670 EXT4_GET_BLOCKS_ZERO)
671 /* Caller will submit data before dropping transaction handle. This
672 * allows jbd2 to avoid submitting data before commit. */
673 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400
674 /* Caller is in the atomic contex, find extent if it has been cached */
675 #define EXT4_GET_BLOCKS_CACHED_NOWAIT 0x0800
676
677 /*
678 * The bit position of these flags must not overlap with any of the
679 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(),
680 * read_extent_tree_block(), ext4_split_extent_at(),
681 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
682 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
683 * caching the extents when reading from the extent tree while a
684 * truncate or punch hole operation is in progress.
685 */
686 #define EXT4_EX_NOCACHE 0x40000000
687 #define EXT4_EX_FORCE_CACHE 0x20000000
688 #define EXT4_EX_NOFAIL 0x10000000
689
690 /*
691 * Flags used by ext4_free_blocks
692 */
693 #define EXT4_FREE_BLOCKS_METADATA 0x0001
694 #define EXT4_FREE_BLOCKS_FORGET 0x0002
695 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
696 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
697 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
698 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
699 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040
700
701 /*
702 * ioctl commands
703 */
704 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
705 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
706 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
707 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
708 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
709 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
710 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
711 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
712 #define EXT4_IOC_MIGRATE _IO('f', 9)
713 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
714 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
715 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
716 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
717 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
718 #define EXT4_IOC_SWAP_BOOT _IO('f', 17)
719 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18)
720 /* ioctl codes 19--39 are reserved for fscrypt */
721 #define EXT4_IOC_CLEAR_ES_CACHE _IO('f', 40)
722 #define EXT4_IOC_GETSTATE _IOW('f', 41, __u32)
723 #define EXT4_IOC_GET_ES_CACHE _IOWR('f', 42, struct fiemap)
724 #define EXT4_IOC_CHECKPOINT _IOW('f', 43, __u32)
725 #define EXT4_IOC_GETFSUUID _IOR('f', 44, struct fsuuid)
726 #define EXT4_IOC_SETFSUUID _IOW('f', 44, struct fsuuid)
727
728 #define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32)
729
730 /*
731 * Flags for going down operation
732 */
733 #define EXT4_GOING_FLAGS_DEFAULT 0x0 /* going down */
734 #define EXT4_GOING_FLAGS_LOGFLUSH 0x1 /* flush log but not data */
735 #define EXT4_GOING_FLAGS_NOLOGFLUSH 0x2 /* don't flush log nor data */
736
737 /*
738 * Flags returned by EXT4_IOC_GETSTATE
739 *
740 * We only expose to userspace a subset of the state flags in
741 * i_state_flags
742 */
743 #define EXT4_STATE_FLAG_EXT_PRECACHED 0x00000001
744 #define EXT4_STATE_FLAG_NEW 0x00000002
745 #define EXT4_STATE_FLAG_NEWENTRY 0x00000004
746 #define EXT4_STATE_FLAG_DA_ALLOC_CLOSE 0x00000008
747
748 /* flags for ioctl EXT4_IOC_CHECKPOINT */
749 #define EXT4_IOC_CHECKPOINT_FLAG_DISCARD 0x1
750 #define EXT4_IOC_CHECKPOINT_FLAG_ZEROOUT 0x2
751 #define EXT4_IOC_CHECKPOINT_FLAG_DRY_RUN 0x4
752 #define EXT4_IOC_CHECKPOINT_FLAG_VALID (EXT4_IOC_CHECKPOINT_FLAG_DISCARD | \
753 EXT4_IOC_CHECKPOINT_FLAG_ZEROOUT | \
754 EXT4_IOC_CHECKPOINT_FLAG_DRY_RUN)
755
756 /*
757 * Structure for EXT4_IOC_GETFSUUID/EXT4_IOC_SETFSUUID
758 */
759 struct fsuuid {
760 __u32 fsu_len;
761 __u32 fsu_flags;
762 __u8 fsu_uuid[];
763 };
764
765 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
766 /*
767 * ioctl commands in 32 bit emulation
768 */
769 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
770 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
771 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
772 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
773 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
774 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
775 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
776 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
777 #endif
778
779 /*
780 * Returned by EXT4_IOC_GET_ES_CACHE as an additional possible flag.
781 * It indicates that the entry in extent status cache is for a hole.
782 */
783 #define EXT4_FIEMAP_EXTENT_HOLE 0x08000000
784
785 /* Max physical block we can address w/o extents */
786 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
787
788 /* Max logical block we can support */
789 #define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFE
790
791 /*
792 * Structure of an inode on the disk
793 */
794 struct ext4_inode {
795 __le16 i_mode; /* File mode */
796 __le16 i_uid; /* Low 16 bits of Owner Uid */
797 __le32 i_size_lo; /* Size in bytes */
798 __le32 i_atime; /* Access time */
799 __le32 i_ctime; /* Inode Change time */
800 __le32 i_mtime; /* Modification time */
801 __le32 i_dtime; /* Deletion Time */
802 __le16 i_gid; /* Low 16 bits of Group Id */
803 __le16 i_links_count; /* Links count */
804 __le32 i_blocks_lo; /* Blocks count */
805 __le32 i_flags; /* File flags */
806 union {
807 struct {
808 __le32 l_i_version;
809 } linux1;
810 struct {
811 __u32 h_i_translator;
812 } hurd1;
813 struct {
814 __u32 m_i_reserved1;
815 } masix1;
816 } osd1; /* OS dependent 1 */
817 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
818 __le32 i_generation; /* File version (for NFS) */
819 __le32 i_file_acl_lo; /* File ACL */
820 __le32 i_size_high;
821 __le32 i_obso_faddr; /* Obsoleted fragment address */
822 union {
823 struct {
824 __le16 l_i_blocks_high; /* were l_i_reserved1 */
825 __le16 l_i_file_acl_high;
826 __le16 l_i_uid_high; /* these 2 fields */
827 __le16 l_i_gid_high; /* were reserved2[0] */
828 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
829 __le16 l_i_reserved;
830 } linux2;
831 struct {
832 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
833 __u16 h_i_mode_high;
834 __u16 h_i_uid_high;
835 __u16 h_i_gid_high;
836 __u32 h_i_author;
837 } hurd2;
838 struct {
839 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
840 __le16 m_i_file_acl_high;
841 __u32 m_i_reserved2[2];
842 } masix2;
843 } osd2; /* OS dependent 2 */
844 __le16 i_extra_isize;
845 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
846 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
847 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
848 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
849 __le32 i_crtime; /* File Creation time */
850 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
851 __le32 i_version_hi; /* high 32 bits for 64-bit version */
852 __le32 i_projid; /* Project ID */
853 };
854
855 struct move_extent {
856 __u32 reserved; /* should be zero */
857 __u32 donor_fd; /* donor file descriptor */
858 __u64 orig_start; /* logical start offset in block for orig */
859 __u64 donor_start; /* logical start offset in block for donor */
860 __u64 len; /* block length to be moved */
861 __u64 moved_len; /* moved block length */
862 };
863
864 #define EXT4_EPOCH_BITS 2
865 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
866 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
867
868 /*
869 * Extended fields will fit into an inode if the filesystem was formatted
870 * with large inodes (-I 256 or larger) and there are not currently any EAs
871 * consuming all of the available space. For new inodes we always reserve
872 * enough space for the kernel's known extended fields, but for inodes
873 * created with an old kernel this might not have been the case. None of
874 * the extended inode fields is critical for correct filesystem operation.
875 * This macro checks if a certain field fits in the inode. Note that
876 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
877 */
878 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
879 ((offsetof(typeof(*ext4_inode), field) + \
880 sizeof((ext4_inode)->field)) \
881 <= (EXT4_GOOD_OLD_INODE_SIZE + \
882 (einode)->i_extra_isize)) \
883
884 /*
885 * We use an encoding that preserves the times for extra epoch "00":
886 *
887 * extra msb of adjust for signed
888 * epoch 32-bit 32-bit tv_sec to
889 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
890 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
891 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
892 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
893 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
894 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
895 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
896 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
897 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
898 *
899 * Note that previous versions of the kernel on 64-bit systems would
900 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
901 * 1970. e2fsck will correct this, assuming that it is run on the
902 * affected filesystem before 2242.
903 */
904
ext4_encode_extra_time(struct timespec64 * time)905 static inline __le32 ext4_encode_extra_time(struct timespec64 *time)
906 {
907 u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK;
908 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
909 }
910
ext4_decode_extra_time(struct timespec64 * time,__le32 extra)911 static inline void ext4_decode_extra_time(struct timespec64 *time,
912 __le32 extra)
913 {
914 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK)))
915 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
916 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
917 }
918
919 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
920 do { \
921 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {\
922 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
923 (raw_inode)->xtime ## _extra = \
924 ext4_encode_extra_time(&(inode)->xtime); \
925 } \
926 else \
927 (raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (inode)->xtime.tv_sec, S32_MIN, S32_MAX)); \
928 } while (0)
929
930 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
931 do { \
932 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
933 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
934 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
935 (raw_inode)->xtime ## _extra = \
936 ext4_encode_extra_time(&(einode)->xtime); \
937 } while (0)
938
939 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
940 do { \
941 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
942 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \
943 ext4_decode_extra_time(&(inode)->xtime, \
944 raw_inode->xtime ## _extra); \
945 } \
946 else \
947 (inode)->xtime.tv_nsec = 0; \
948 } while (0)
949
950
951 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
952 do { \
953 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
954 (einode)->xtime.tv_sec = \
955 (signed)le32_to_cpu((raw_inode)->xtime); \
956 else \
957 (einode)->xtime.tv_sec = 0; \
958 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
959 ext4_decode_extra_time(&(einode)->xtime, \
960 raw_inode->xtime ## _extra); \
961 else \
962 (einode)->xtime.tv_nsec = 0; \
963 } while (0)
964
965 #define i_disk_version osd1.linux1.l_i_version
966
967 #if defined(__KERNEL__) || defined(__linux__)
968 #define i_reserved1 osd1.linux1.l_i_reserved1
969 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
970 #define i_blocks_high osd2.linux2.l_i_blocks_high
971 #define i_uid_low i_uid
972 #define i_gid_low i_gid
973 #define i_uid_high osd2.linux2.l_i_uid_high
974 #define i_gid_high osd2.linux2.l_i_gid_high
975 #define i_checksum_lo osd2.linux2.l_i_checksum_lo
976
977 #elif defined(__GNU__)
978
979 #define i_translator osd1.hurd1.h_i_translator
980 #define i_uid_high osd2.hurd2.h_i_uid_high
981 #define i_gid_high osd2.hurd2.h_i_gid_high
982 #define i_author osd2.hurd2.h_i_author
983
984 #elif defined(__masix__)
985
986 #define i_reserved1 osd1.masix1.m_i_reserved1
987 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
988 #define i_reserved2 osd2.masix2.m_i_reserved2
989
990 #endif /* defined(__KERNEL__) || defined(__linux__) */
991
992 #include "extents_status.h"
993 #include "fast_commit.h"
994
995 /*
996 * Lock subclasses for i_data_sem in the ext4_inode_info structure.
997 *
998 * These are needed to avoid lockdep false positives when we need to
999 * allocate blocks to the quota inode during ext4_map_blocks(), while
1000 * holding i_data_sem for a normal (non-quota) inode. Since we don't
1001 * do quota tracking for the quota inode, this avoids deadlock (as
1002 * well as infinite recursion, since it isn't turtles all the way
1003 * down...)
1004 *
1005 * I_DATA_SEM_NORMAL - Used for most inodes
1006 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode
1007 * where the second inode has larger inode number
1008 * than the first
1009 * I_DATA_SEM_QUOTA - Used for quota inodes only
1010 * I_DATA_SEM_EA - Used for ea_inodes only
1011 */
1012 enum {
1013 I_DATA_SEM_NORMAL = 0,
1014 I_DATA_SEM_OTHER,
1015 I_DATA_SEM_QUOTA,
1016 I_DATA_SEM_EA
1017 };
1018
1019
1020 /*
1021 * fourth extended file system inode data in memory
1022 */
1023 struct ext4_inode_info {
1024 __le32 i_data[15]; /* unconverted */
1025 __u32 i_dtime;
1026 ext4_fsblk_t i_file_acl;
1027
1028 /*
1029 * i_block_group is the number of the block group which contains
1030 * this file's inode. Constant across the lifetime of the inode,
1031 * it is used for making block allocation decisions - we try to
1032 * place a file's data blocks near its inode block, and new inodes
1033 * near to their parent directory's inode.
1034 */
1035 ext4_group_t i_block_group;
1036 ext4_lblk_t i_dir_start_lookup;
1037 #if (BITS_PER_LONG < 64)
1038 unsigned long i_state_flags; /* Dynamic state flags */
1039 #endif
1040 unsigned long i_flags;
1041
1042 /*
1043 * Extended attributes can be read independently of the main file
1044 * data. Taking i_rwsem even when reading would cause contention
1045 * between readers of EAs and writers of regular file data, so
1046 * instead we synchronize on xattr_sem when reading or changing
1047 * EAs.
1048 */
1049 struct rw_semaphore xattr_sem;
1050
1051 /*
1052 * Inodes with EXT4_STATE_ORPHAN_FILE use i_orphan_idx. Otherwise
1053 * i_orphan is used.
1054 */
1055 union {
1056 struct list_head i_orphan; /* unlinked but open inodes */
1057 unsigned int i_orphan_idx; /* Index in orphan file */
1058 };
1059
1060 /* Fast commit related info */
1061
1062 /* For tracking dentry create updates */
1063 struct list_head i_fc_dilist;
1064 struct list_head i_fc_list; /*
1065 * inodes that need fast commit
1066 * protected by sbi->s_fc_lock.
1067 */
1068
1069 /* Start of lblk range that needs to be committed in this fast commit */
1070 ext4_lblk_t i_fc_lblk_start;
1071
1072 /* End of lblk range that needs to be committed in this fast commit */
1073 ext4_lblk_t i_fc_lblk_len;
1074
1075 /* Number of ongoing updates on this inode */
1076 atomic_t i_fc_updates;
1077
1078 /* Fast commit wait queue for this inode */
1079 wait_queue_head_t i_fc_wait;
1080
1081 /* Protect concurrent accesses on i_fc_lblk_start, i_fc_lblk_len */
1082 struct mutex i_fc_lock;
1083
1084 /*
1085 * i_disksize keeps track of what the inode size is ON DISK, not
1086 * in memory. During truncate, i_size is set to the new size by
1087 * the VFS prior to calling ext4_truncate(), but the filesystem won't
1088 * set i_disksize to 0 until the truncate is actually under way.
1089 *
1090 * The intent is that i_disksize always represents the blocks which
1091 * are used by this file. This allows recovery to restart truncate
1092 * on orphans if we crash during truncate. We actually write i_disksize
1093 * into the on-disk inode when writing inodes out, instead of i_size.
1094 *
1095 * The only time when i_disksize and i_size may be different is when
1096 * a truncate is in progress. The only things which change i_disksize
1097 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
1098 */
1099 loff_t i_disksize;
1100
1101 /*
1102 * i_data_sem is for serialising ext4_truncate() against
1103 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
1104 * data tree are chopped off during truncate. We can't do that in
1105 * ext4 because whenever we perform intermediate commits during
1106 * truncate, the inode and all the metadata blocks *must* be in a
1107 * consistent state which allows truncation of the orphans to restart
1108 * during recovery. Hence we must fix the get_block-vs-truncate race
1109 * by other means, so we have i_data_sem.
1110 */
1111 struct rw_semaphore i_data_sem;
1112 struct inode vfs_inode;
1113 struct jbd2_inode *jinode;
1114
1115 spinlock_t i_raw_lock; /* protects updates to the raw inode */
1116
1117 /*
1118 * File creation time. Its function is same as that of
1119 * struct timespec64 i_{a,c,m}time in the generic inode.
1120 */
1121 struct timespec64 i_crtime;
1122
1123 /* mballoc */
1124 atomic_t i_prealloc_active;
1125 struct list_head i_prealloc_list;
1126 spinlock_t i_prealloc_lock;
1127
1128 /* extents status tree */
1129 struct ext4_es_tree i_es_tree;
1130 rwlock_t i_es_lock;
1131 struct list_head i_es_list;
1132 unsigned int i_es_all_nr; /* protected by i_es_lock */
1133 unsigned int i_es_shk_nr; /* protected by i_es_lock */
1134 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for
1135 extents to shrink. Protected by
1136 i_es_lock */
1137
1138 /* ialloc */
1139 ext4_group_t i_last_alloc_group;
1140
1141 /* allocation reservation info for delalloc */
1142 /* In case of bigalloc, this refer to clusters rather than blocks */
1143 unsigned int i_reserved_data_blocks;
1144
1145 /* pending cluster reservations for bigalloc file systems */
1146 struct ext4_pending_tree i_pending_tree;
1147
1148 /* on-disk additional length */
1149 __u16 i_extra_isize;
1150
1151 /* Indicate the inline data space. */
1152 u16 i_inline_off;
1153 u16 i_inline_size;
1154
1155 #ifdef CONFIG_QUOTA
1156 /* quota space reservation, managed internally by quota code */
1157 qsize_t i_reserved_quota;
1158 #endif
1159
1160 /* Lock protecting lists below */
1161 spinlock_t i_completed_io_lock;
1162 /*
1163 * Completed IOs that need unwritten extents handling and have
1164 * transaction reserved
1165 */
1166 struct list_head i_rsv_conversion_list;
1167 struct work_struct i_rsv_conversion_work;
1168 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1169
1170 spinlock_t i_block_reservation_lock;
1171
1172 /*
1173 * Transactions that contain inode's metadata needed to complete
1174 * fsync and fdatasync, respectively.
1175 */
1176 tid_t i_sync_tid;
1177 tid_t i_datasync_tid;
1178
1179 #ifdef CONFIG_QUOTA
1180 struct dquot *i_dquot[MAXQUOTAS];
1181 #endif
1182
1183 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1184 __u32 i_csum_seed;
1185
1186 kprojid_t i_projid;
1187 };
1188
1189 /*
1190 * File system states
1191 */
1192 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
1193 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
1194 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
1195 #define EXT4_FC_REPLAY 0x0020 /* Fast commit replay ongoing */
1196
1197 /*
1198 * Misc. filesystem flags
1199 */
1200 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
1201 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
1202 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
1203
1204 /*
1205 * Mount flags set via mount options or defaults
1206 */
1207 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */
1208 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
1209 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
1210 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
1211 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
1212 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
1213 #define EXT4_MOUNT_ERRORS_MASK 0x00070
1214 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
1215 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
1216 #ifdef CONFIG_FS_DAX
1217 #define EXT4_MOUNT_DAX_ALWAYS 0x00200 /* Direct Access */
1218 #else
1219 #define EXT4_MOUNT_DAX_ALWAYS 0
1220 #endif
1221 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
1222 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
1223 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
1224 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
1225 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
1226 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
1227 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
1228 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
1229 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
1230 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
1231 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */
1232 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota,
1233 * enable enforcement for hidden
1234 * quota files */
1235 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable
1236 * enforcement for hidden quota
1237 * files */
1238 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota
1239 * enforcement */
1240 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
1241 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
1242 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1243 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */
1244 #define EXT4_MOUNT_NO_PREFETCH_BLOCK_BITMAPS 0x4000000
1245 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
1246 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
1247 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
1248 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
1249 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
1250
1251 /*
1252 * Mount flags set either automatically (could not be set by mount option)
1253 * based on per file system feature or property or in special cases such as
1254 * distinguishing between explicit mount option definition and default.
1255 */
1256 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1257 specified delalloc */
1258 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1259 size of blocksize * 8
1260 blocks */
1261 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1262 file systems */
1263 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly
1264 specified journal checksum */
1265
1266 #define EXT4_MOUNT2_JOURNAL_FAST_COMMIT 0x00000010 /* Journal fast commit */
1267 #define EXT4_MOUNT2_DAX_NEVER 0x00000020 /* Do not allow Direct Access */
1268 #define EXT4_MOUNT2_DAX_INODE 0x00000040 /* For printing options only */
1269 #define EXT4_MOUNT2_MB_OPTIMIZE_SCAN 0x00000080 /* Optimize group
1270 * scanning in mballoc
1271 */
1272
1273 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1274 ~EXT4_MOUNT_##opt
1275 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1276 EXT4_MOUNT_##opt
1277 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1278 EXT4_MOUNT_##opt)
1279
1280 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1281 ~EXT4_MOUNT2_##opt
1282 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1283 EXT4_MOUNT2_##opt
1284 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1285 EXT4_MOUNT2_##opt)
1286
1287 #define ext4_test_and_set_bit __test_and_set_bit_le
1288 #define ext4_set_bit __set_bit_le
1289 #define ext4_set_bit_atomic ext2_set_bit_atomic
1290 #define ext4_test_and_clear_bit __test_and_clear_bit_le
1291 #define ext4_clear_bit __clear_bit_le
1292 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
1293 #define ext4_test_bit test_bit_le
1294 #define ext4_find_next_zero_bit find_next_zero_bit_le
1295 #define ext4_find_next_bit find_next_bit_le
1296
1297 extern void mb_set_bits(void *bm, int cur, int len);
1298
1299 /*
1300 * Maximal mount counts between two filesystem checks
1301 */
1302 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1303 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1304
1305 /*
1306 * Behaviour when detecting errors
1307 */
1308 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1309 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1310 #define EXT4_ERRORS_PANIC 3 /* Panic */
1311 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1312
1313 /* Metadata checksum algorithm codes */
1314 #define EXT4_CRC32C_CHKSUM 1
1315
1316 #define EXT4_LABEL_MAX 16
1317
1318 /*
1319 * Structure of the super block
1320 */
1321 struct ext4_super_block {
1322 /*00*/ __le32 s_inodes_count; /* Inodes count */
1323 __le32 s_blocks_count_lo; /* Blocks count */
1324 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1325 __le32 s_free_blocks_count_lo; /* Free blocks count */
1326 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1327 __le32 s_first_data_block; /* First Data Block */
1328 __le32 s_log_block_size; /* Block size */
1329 __le32 s_log_cluster_size; /* Allocation cluster size */
1330 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1331 __le32 s_clusters_per_group; /* # Clusters per group */
1332 __le32 s_inodes_per_group; /* # Inodes per group */
1333 __le32 s_mtime; /* Mount time */
1334 /*30*/ __le32 s_wtime; /* Write time */
1335 __le16 s_mnt_count; /* Mount count */
1336 __le16 s_max_mnt_count; /* Maximal mount count */
1337 __le16 s_magic; /* Magic signature */
1338 __le16 s_state; /* File system state */
1339 __le16 s_errors; /* Behaviour when detecting errors */
1340 __le16 s_minor_rev_level; /* minor revision level */
1341 /*40*/ __le32 s_lastcheck; /* time of last check */
1342 __le32 s_checkinterval; /* max. time between checks */
1343 __le32 s_creator_os; /* OS */
1344 __le32 s_rev_level; /* Revision level */
1345 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1346 __le16 s_def_resgid; /* Default gid for reserved blocks */
1347 /*
1348 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1349 *
1350 * Note: the difference between the compatible feature set and
1351 * the incompatible feature set is that if there is a bit set
1352 * in the incompatible feature set that the kernel doesn't
1353 * know about, it should refuse to mount the filesystem.
1354 *
1355 * e2fsck's requirements are more strict; if it doesn't know
1356 * about a feature in either the compatible or incompatible
1357 * feature set, it must abort and not try to meddle with
1358 * things it doesn't understand...
1359 */
1360 __le32 s_first_ino; /* First non-reserved inode */
1361 __le16 s_inode_size; /* size of inode structure */
1362 __le16 s_block_group_nr; /* block group # of this superblock */
1363 __le32 s_feature_compat; /* compatible feature set */
1364 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1365 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1366 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1367 /*78*/ char s_volume_name[EXT4_LABEL_MAX]; /* volume name */
1368 /*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */
1369 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1370 /*
1371 * Performance hints. Directory preallocation should only
1372 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1373 */
1374 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1375 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1376 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1377 /*
1378 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1379 */
1380 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1381 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1382 __le32 s_journal_dev; /* device number of journal file */
1383 __le32 s_last_orphan; /* start of list of inodes to delete */
1384 __le32 s_hash_seed[4]; /* HTREE hash seed */
1385 __u8 s_def_hash_version; /* Default hash version to use */
1386 __u8 s_jnl_backup_type;
1387 __le16 s_desc_size; /* size of group descriptor */
1388 /*100*/ __le32 s_default_mount_opts;
1389 __le32 s_first_meta_bg; /* First metablock block group */
1390 __le32 s_mkfs_time; /* When the filesystem was created */
1391 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1392 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1393 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1394 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1395 __le32 s_free_blocks_count_hi; /* Free blocks count */
1396 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1397 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1398 __le32 s_flags; /* Miscellaneous flags */
1399 __le16 s_raid_stride; /* RAID stride */
1400 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1401 __le64 s_mmp_block; /* Block for multi-mount protection */
1402 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1403 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1404 __u8 s_checksum_type; /* metadata checksum algorithm used */
1405 __u8 s_encryption_level; /* versioning level for encryption */
1406 __u8 s_reserved_pad; /* Padding to next 32bits */
1407 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1408 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1409 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1410 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1411 snapshot's future use */
1412 __le32 s_snapshot_list; /* inode number of the head of the
1413 on-disk snapshot list */
1414 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1415 __le32 s_error_count; /* number of fs errors */
1416 __le32 s_first_error_time; /* first time an error happened */
1417 __le32 s_first_error_ino; /* inode involved in first error */
1418 __le64 s_first_error_block; /* block involved of first error */
1419 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */
1420 __le32 s_first_error_line; /* line number where error happened */
1421 __le32 s_last_error_time; /* most recent time of an error */
1422 __le32 s_last_error_ino; /* inode involved in last error */
1423 __le32 s_last_error_line; /* line number where error happened */
1424 __le64 s_last_error_block; /* block involved of last error */
1425 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */
1426 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1427 __u8 s_mount_opts[64];
1428 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1429 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1430 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1431 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
1432 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */
1433 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */
1434 __le32 s_lpf_ino; /* Location of the lost+found inode */
1435 __le32 s_prj_quota_inum; /* inode for tracking project quota */
1436 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */
1437 __u8 s_wtime_hi;
1438 __u8 s_mtime_hi;
1439 __u8 s_mkfs_time_hi;
1440 __u8 s_lastcheck_hi;
1441 __u8 s_first_error_time_hi;
1442 __u8 s_last_error_time_hi;
1443 __u8 s_first_error_errcode;
1444 __u8 s_last_error_errcode;
1445 __le16 s_encoding; /* Filename charset encoding */
1446 __le16 s_encoding_flags; /* Filename charset encoding flags */
1447 __le32 s_orphan_file_inum; /* Inode for tracking orphan inodes */
1448 __le32 s_reserved[94]; /* Padding to the end of the block */
1449 __le32 s_checksum; /* crc32c(superblock) */
1450 };
1451
1452 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1453
1454 #ifdef __KERNEL__
1455
1456 /* Number of quota types we support */
1457 #define EXT4_MAXQUOTAS 3
1458
1459 #define EXT4_ENC_UTF8_12_1 1
1460
1461 /* Types of ext4 journal triggers */
1462 enum ext4_journal_trigger_type {
1463 EXT4_JTR_ORPHAN_FILE,
1464 EXT4_JTR_NONE /* This must be the last entry for indexing to work! */
1465 };
1466
1467 #define EXT4_JOURNAL_TRIGGER_COUNT EXT4_JTR_NONE
1468
1469 struct ext4_journal_trigger {
1470 struct jbd2_buffer_trigger_type tr_triggers;
1471 struct super_block *sb;
1472 };
1473
EXT4_TRIGGER(struct jbd2_buffer_trigger_type * trigger)1474 static inline struct ext4_journal_trigger *EXT4_TRIGGER(
1475 struct jbd2_buffer_trigger_type *trigger)
1476 {
1477 return container_of(trigger, struct ext4_journal_trigger, tr_triggers);
1478 }
1479
1480 #define EXT4_ORPHAN_BLOCK_MAGIC 0x0b10ca04
1481
1482 /* Structure at the tail of orphan block */
1483 struct ext4_orphan_block_tail {
1484 __le32 ob_magic;
1485 __le32 ob_checksum;
1486 };
1487
ext4_inodes_per_orphan_block(struct super_block * sb)1488 static inline int ext4_inodes_per_orphan_block(struct super_block *sb)
1489 {
1490 return (sb->s_blocksize - sizeof(struct ext4_orphan_block_tail)) /
1491 sizeof(u32);
1492 }
1493
1494 struct ext4_orphan_block {
1495 atomic_t ob_free_entries; /* Number of free orphan entries in block */
1496 struct buffer_head *ob_bh; /* Buffer for orphan block */
1497 };
1498
1499 /*
1500 * Info about orphan file.
1501 */
1502 struct ext4_orphan_info {
1503 int of_blocks; /* Number of orphan blocks in a file */
1504 __u32 of_csum_seed; /* Checksum seed for orphan file */
1505 struct ext4_orphan_block *of_binfo; /* Array with info about orphan
1506 * file blocks */
1507 };
1508
1509 /*
1510 * fourth extended-fs super-block data in memory
1511 */
1512 struct ext4_sb_info {
1513 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1514 unsigned long s_inodes_per_block;/* Number of inodes per block */
1515 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1516 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1517 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1518 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1519 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1520 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1521 ext4_group_t s_groups_count; /* Number of groups in the fs */
1522 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1523 unsigned long s_overhead; /* # of fs overhead clusters */
1524 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1525 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1526 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1527 struct buffer_head * s_sbh; /* Buffer containing the super block */
1528 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1529 struct buffer_head * __rcu *s_group_desc;
1530 unsigned int s_mount_opt;
1531 unsigned int s_mount_opt2;
1532 unsigned long s_mount_flags;
1533 unsigned int s_def_mount_opt;
1534 unsigned int s_def_mount_opt2;
1535 ext4_fsblk_t s_sb_block;
1536 atomic64_t s_resv_clusters;
1537 kuid_t s_resuid;
1538 kgid_t s_resgid;
1539 unsigned short s_mount_state;
1540 unsigned short s_pad;
1541 int s_addr_per_block_bits;
1542 int s_desc_per_block_bits;
1543 int s_inode_size;
1544 int s_first_ino;
1545 unsigned int s_inode_readahead_blks;
1546 unsigned int s_inode_goal;
1547 u32 s_hash_seed[4];
1548 int s_def_hash_version;
1549 int s_hash_unsigned; /* 3 if hash should be unsigned, 0 if not */
1550 struct percpu_counter s_freeclusters_counter;
1551 struct percpu_counter s_freeinodes_counter;
1552 struct percpu_counter s_dirs_counter;
1553 struct percpu_counter s_dirtyclusters_counter;
1554 struct percpu_counter s_sra_exceeded_retry_limit;
1555 struct blockgroup_lock *s_blockgroup_lock;
1556 struct proc_dir_entry *s_proc;
1557 struct kobject s_kobj;
1558 struct completion s_kobj_unregister;
1559 struct super_block *s_sb;
1560 struct buffer_head *s_mmp_bh;
1561
1562 /* Journaling */
1563 struct journal_s *s_journal;
1564 unsigned long s_ext4_flags; /* Ext4 superblock flags */
1565 struct mutex s_orphan_lock; /* Protects on disk list changes */
1566 struct list_head s_orphan; /* List of orphaned inodes in on disk
1567 list */
1568 struct ext4_orphan_info s_orphan_info;
1569 unsigned long s_commit_interval;
1570 u32 s_max_batch_time;
1571 u32 s_min_batch_time;
1572 struct block_device *s_journal_bdev;
1573 #ifdef CONFIG_QUOTA
1574 /* Names of quota files with journalled quota */
1575 char __rcu *s_qf_names[EXT4_MAXQUOTAS];
1576 int s_jquota_fmt; /* Format of quota to use */
1577 #endif
1578 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1579 struct ext4_system_blocks __rcu *s_system_blks;
1580
1581 #ifdef EXTENTS_STATS
1582 /* ext4 extents stats */
1583 unsigned long s_ext_min;
1584 unsigned long s_ext_max;
1585 unsigned long s_depth_max;
1586 spinlock_t s_ext_stats_lock;
1587 unsigned long s_ext_blocks;
1588 unsigned long s_ext_extents;
1589 #endif
1590
1591 /* for buddy allocator */
1592 struct ext4_group_info ** __rcu *s_group_info;
1593 struct inode *s_buddy_cache;
1594 spinlock_t s_md_lock;
1595 unsigned short *s_mb_offsets;
1596 unsigned int *s_mb_maxs;
1597 unsigned int s_group_info_size;
1598 unsigned int s_mb_free_pending;
1599 struct list_head s_freed_data_list; /* List of blocks to be freed
1600 after commit completed */
1601 struct list_head s_discard_list;
1602 struct work_struct s_discard_work;
1603 atomic_t s_retry_alloc_pending;
1604 struct list_head *s_mb_avg_fragment_size;
1605 rwlock_t *s_mb_avg_fragment_size_locks;
1606 struct list_head *s_mb_largest_free_orders;
1607 rwlock_t *s_mb_largest_free_orders_locks;
1608
1609 /* tunables */
1610 unsigned long s_stripe;
1611 unsigned int s_mb_max_linear_groups;
1612 unsigned int s_mb_stream_request;
1613 unsigned int s_mb_max_to_scan;
1614 unsigned int s_mb_min_to_scan;
1615 unsigned int s_mb_stats;
1616 unsigned int s_mb_order2_reqs;
1617 unsigned int s_mb_group_prealloc;
1618 unsigned int s_mb_max_inode_prealloc;
1619 unsigned int s_max_dir_size_kb;
1620 /* where last allocation was done - for stream allocation */
1621 unsigned long s_mb_last_group;
1622 unsigned long s_mb_last_start;
1623 unsigned int s_mb_prefetch;
1624 unsigned int s_mb_prefetch_limit;
1625
1626 /* stats for buddy allocator */
1627 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1628 atomic_t s_bal_success; /* we found long enough chunks */
1629 atomic_t s_bal_allocated; /* in blocks */
1630 atomic_t s_bal_ex_scanned; /* total extents scanned */
1631 atomic_t s_bal_groups_scanned; /* number of groups scanned */
1632 atomic_t s_bal_goals; /* goal hits */
1633 atomic_t s_bal_breaks; /* too long searches */
1634 atomic_t s_bal_2orders; /* 2^order hits */
1635 atomic_t s_bal_cr0_bad_suggestions;
1636 atomic_t s_bal_cr1_bad_suggestions;
1637 atomic64_t s_bal_cX_groups_considered[4];
1638 atomic64_t s_bal_cX_hits[4];
1639 atomic64_t s_bal_cX_failed[4]; /* cX loop didn't find blocks */
1640 atomic_t s_mb_buddies_generated; /* number of buddies generated */
1641 atomic64_t s_mb_generation_time;
1642 atomic_t s_mb_lost_chunks;
1643 atomic_t s_mb_preallocated;
1644 atomic_t s_mb_discarded;
1645 atomic_t s_lock_busy;
1646
1647 /* locality groups */
1648 struct ext4_locality_group __percpu *s_locality_groups;
1649
1650 /* for write statistics */
1651 unsigned long s_sectors_written_start;
1652 u64 s_kbytes_written;
1653
1654 /* the size of zero-out chunk */
1655 unsigned int s_extent_max_zeroout_kb;
1656
1657 unsigned int s_log_groups_per_flex;
1658 struct flex_groups * __rcu *s_flex_groups;
1659 ext4_group_t s_flex_groups_allocated;
1660
1661 /* workqueue for reserved extent conversions (buffered io) */
1662 struct workqueue_struct *rsv_conversion_wq;
1663
1664 /* timer for periodic error stats printing */
1665 struct timer_list s_err_report;
1666
1667 /* Lazy inode table initialization info */
1668 struct ext4_li_request *s_li_request;
1669 /* Wait multiplier for lazy initialization thread */
1670 unsigned int s_li_wait_mult;
1671
1672 /* Kernel thread for multiple mount protection */
1673 struct task_struct *s_mmp_tsk;
1674
1675 /* record the last minlen when FITRIM is called. */
1676 unsigned long s_last_trim_minblks;
1677
1678 /* Reference to checksum algorithm driver via cryptoapi */
1679 struct crypto_shash *s_chksum_driver;
1680
1681 /* Precomputed FS UUID checksum for seeding other checksums */
1682 __u32 s_csum_seed;
1683
1684 /* Reclaim extents from extent status tree */
1685 struct shrinker s_es_shrinker;
1686 struct list_head s_es_list; /* List of inodes with reclaimable extents */
1687 long s_es_nr_inode;
1688 struct ext4_es_stats s_es_stats;
1689 struct mb_cache *s_ea_block_cache;
1690 struct mb_cache *s_ea_inode_cache;
1691 spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1692
1693 /* Journal triggers for checksum computation */
1694 struct ext4_journal_trigger s_journal_triggers[EXT4_JOURNAL_TRIGGER_COUNT];
1695
1696 /* Ratelimit ext4 messages. */
1697 struct ratelimit_state s_err_ratelimit_state;
1698 struct ratelimit_state s_warning_ratelimit_state;
1699 struct ratelimit_state s_msg_ratelimit_state;
1700 atomic_t s_warning_count;
1701 atomic_t s_msg_count;
1702
1703 /* Encryption policy for '-o test_dummy_encryption' */
1704 struct fscrypt_dummy_policy s_dummy_enc_policy;
1705
1706 /*
1707 * Barrier between writepages ops and changing any inode's JOURNAL_DATA
1708 * or EXTENTS flag.
1709 */
1710 struct percpu_rw_semaphore s_writepages_rwsem;
1711 struct dax_device *s_daxdev;
1712 u64 s_dax_part_off;
1713 #ifdef CONFIG_EXT4_DEBUG
1714 unsigned long s_simulate_fail;
1715 #endif
1716 /* Record the errseq of the backing block device */
1717 errseq_t s_bdev_wb_err;
1718 spinlock_t s_bdev_wb_lock;
1719
1720 /* Information about errors that happened during this mount */
1721 spinlock_t s_error_lock;
1722 int s_add_error_count;
1723 int s_first_error_code;
1724 __u32 s_first_error_line;
1725 __u32 s_first_error_ino;
1726 __u64 s_first_error_block;
1727 const char *s_first_error_func;
1728 time64_t s_first_error_time;
1729 int s_last_error_code;
1730 __u32 s_last_error_line;
1731 __u32 s_last_error_ino;
1732 __u64 s_last_error_block;
1733 const char *s_last_error_func;
1734 time64_t s_last_error_time;
1735 /*
1736 * If we are in a context where we cannot update error information in
1737 * the on-disk superblock, we queue this work to do it.
1738 */
1739 struct work_struct s_error_work;
1740
1741 /* Ext4 fast commit sub transaction ID */
1742 atomic_t s_fc_subtid;
1743
1744 /*
1745 * After commit starts, the main queue gets locked, and the further
1746 * updates get added in the staging queue.
1747 */
1748 #define FC_Q_MAIN 0
1749 #define FC_Q_STAGING 1
1750 struct list_head s_fc_q[2]; /* Inodes staged for fast commit
1751 * that have data changes in them.
1752 */
1753 struct list_head s_fc_dentry_q[2]; /* directory entry updates */
1754 unsigned int s_fc_bytes;
1755 /*
1756 * Main fast commit lock. This lock protects accesses to the
1757 * following fields:
1758 * ei->i_fc_list, s_fc_dentry_q, s_fc_q, s_fc_bytes, s_fc_bh.
1759 */
1760 spinlock_t s_fc_lock;
1761 struct buffer_head *s_fc_bh;
1762 struct ext4_fc_stats s_fc_stats;
1763 tid_t s_fc_ineligible_tid;
1764 #ifdef CONFIG_EXT4_DEBUG
1765 int s_fc_debug_max_replay;
1766 #endif
1767 struct ext4_fc_replay_state s_fc_replay_state;
1768 };
1769
EXT4_SB(struct super_block * sb)1770 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1771 {
1772 return sb->s_fs_info;
1773 }
EXT4_I(struct inode * inode)1774 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1775 {
1776 return container_of(inode, struct ext4_inode_info, vfs_inode);
1777 }
1778
ext4_valid_inum(struct super_block * sb,unsigned long ino)1779 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1780 {
1781 return ino == EXT4_ROOT_INO ||
1782 (ino >= EXT4_FIRST_INO(sb) &&
1783 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1784 }
1785
1786 /*
1787 * Returns: sbi->field[index]
1788 * Used to access an array element from the following sbi fields which require
1789 * rcu protection to avoid dereferencing an invalid pointer due to reassignment
1790 * - s_group_desc
1791 * - s_group_info
1792 * - s_flex_group
1793 */
1794 #define sbi_array_rcu_deref(sbi, field, index) \
1795 ({ \
1796 typeof(*((sbi)->field)) _v; \
1797 rcu_read_lock(); \
1798 _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index]; \
1799 rcu_read_unlock(); \
1800 _v; \
1801 })
1802
1803 /*
1804 * run-time mount flags
1805 */
1806 enum {
1807 EXT4_MF_MNTDIR_SAMPLED,
1808 EXT4_MF_FS_ABORTED, /* Fatal error detected */
1809 EXT4_MF_FC_INELIGIBLE /* Fast commit ineligible */
1810 };
1811
ext4_set_mount_flag(struct super_block * sb,int bit)1812 static inline void ext4_set_mount_flag(struct super_block *sb, int bit)
1813 {
1814 set_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1815 }
1816
ext4_clear_mount_flag(struct super_block * sb,int bit)1817 static inline void ext4_clear_mount_flag(struct super_block *sb, int bit)
1818 {
1819 clear_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1820 }
1821
ext4_test_mount_flag(struct super_block * sb,int bit)1822 static inline int ext4_test_mount_flag(struct super_block *sb, int bit)
1823 {
1824 return test_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1825 }
1826
1827
1828 /*
1829 * Simulate_fail codes
1830 */
1831 #define EXT4_SIM_BBITMAP_EIO 1
1832 #define EXT4_SIM_BBITMAP_CRC 2
1833 #define EXT4_SIM_IBITMAP_EIO 3
1834 #define EXT4_SIM_IBITMAP_CRC 4
1835 #define EXT4_SIM_INODE_EIO 5
1836 #define EXT4_SIM_INODE_CRC 6
1837 #define EXT4_SIM_DIRBLOCK_EIO 7
1838 #define EXT4_SIM_DIRBLOCK_CRC 8
1839
ext4_simulate_fail(struct super_block * sb,unsigned long code)1840 static inline bool ext4_simulate_fail(struct super_block *sb,
1841 unsigned long code)
1842 {
1843 #ifdef CONFIG_EXT4_DEBUG
1844 struct ext4_sb_info *sbi = EXT4_SB(sb);
1845
1846 if (unlikely(sbi->s_simulate_fail == code)) {
1847 sbi->s_simulate_fail = 0;
1848 return true;
1849 }
1850 #endif
1851 return false;
1852 }
1853
ext4_simulate_fail_bh(struct super_block * sb,struct buffer_head * bh,unsigned long code)1854 static inline void ext4_simulate_fail_bh(struct super_block *sb,
1855 struct buffer_head *bh,
1856 unsigned long code)
1857 {
1858 if (!IS_ERR(bh) && ext4_simulate_fail(sb, code))
1859 clear_buffer_uptodate(bh);
1860 }
1861
1862 /*
1863 * Error number codes for s_{first,last}_error_errno
1864 *
1865 * Linux errno numbers are architecture specific, so we need to translate
1866 * them into something which is architecture independent. We don't define
1867 * codes for all errno's; just the ones which are most likely to be the cause
1868 * of an ext4_error() call.
1869 */
1870 #define EXT4_ERR_UNKNOWN 1
1871 #define EXT4_ERR_EIO 2
1872 #define EXT4_ERR_ENOMEM 3
1873 #define EXT4_ERR_EFSBADCRC 4
1874 #define EXT4_ERR_EFSCORRUPTED 5
1875 #define EXT4_ERR_ENOSPC 6
1876 #define EXT4_ERR_ENOKEY 7
1877 #define EXT4_ERR_EROFS 8
1878 #define EXT4_ERR_EFBIG 9
1879 #define EXT4_ERR_EEXIST 10
1880 #define EXT4_ERR_ERANGE 11
1881 #define EXT4_ERR_EOVERFLOW 12
1882 #define EXT4_ERR_EBUSY 13
1883 #define EXT4_ERR_ENOTDIR 14
1884 #define EXT4_ERR_ENOTEMPTY 15
1885 #define EXT4_ERR_ESHUTDOWN 16
1886 #define EXT4_ERR_EFAULT 17
1887
1888 /*
1889 * Inode dynamic state flags
1890 */
1891 enum {
1892 EXT4_STATE_JDATA, /* journaled data exists */
1893 EXT4_STATE_NEW, /* inode is newly created */
1894 EXT4_STATE_XATTR, /* has in-inode xattrs */
1895 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1896 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1897 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1898 EXT4_STATE_NEWENTRY, /* File just added to dir */
1899 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1900 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
1901 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */
1902 EXT4_STATE_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */
1903 EXT4_STATE_FC_COMMITTING, /* Fast commit ongoing */
1904 EXT4_STATE_ORPHAN_FILE, /* Inode orphaned in orphan file */
1905 };
1906
1907 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1908 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1909 { \
1910 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1911 } \
1912 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1913 { \
1914 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1915 } \
1916 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1917 { \
1918 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1919 }
1920
1921 /* Add these declarations here only so that these functions can be
1922 * found by name. Otherwise, they are very hard to locate. */
1923 static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1924 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1925 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1926 EXT4_INODE_BIT_FNS(flag, flags, 0)
1927
1928 /* Add these declarations here only so that these functions can be
1929 * found by name. Otherwise, they are very hard to locate. */
1930 static inline int ext4_test_inode_state(struct inode *inode, int bit);
1931 static inline void ext4_set_inode_state(struct inode *inode, int bit);
1932 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1933 #if (BITS_PER_LONG < 64)
1934 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1935
ext4_clear_state_flags(struct ext4_inode_info * ei)1936 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1937 {
1938 (ei)->i_state_flags = 0;
1939 }
1940 #else
1941 EXT4_INODE_BIT_FNS(state, flags, 32)
1942
ext4_clear_state_flags(struct ext4_inode_info * ei)1943 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1944 {
1945 /* We depend on the fact that callers will set i_flags */
1946 }
1947 #endif
1948 #else
1949 /* Assume that user mode programs are passing in an ext4fs superblock, not
1950 * a kernel struct super_block. This will allow us to call the feature-test
1951 * macros from user land. */
1952 #define EXT4_SB(sb) (sb)
1953 #endif
1954
ext4_verity_in_progress(struct inode * inode)1955 static inline bool ext4_verity_in_progress(struct inode *inode)
1956 {
1957 return IS_ENABLED(CONFIG_FS_VERITY) &&
1958 ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS);
1959 }
1960
1961 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1962
1963 /*
1964 * Codes for operating systems
1965 */
1966 #define EXT4_OS_LINUX 0
1967 #define EXT4_OS_HURD 1
1968 #define EXT4_OS_MASIX 2
1969 #define EXT4_OS_FREEBSD 3
1970 #define EXT4_OS_LITES 4
1971
1972 /*
1973 * Revision levels
1974 */
1975 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1976 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1977
1978 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1979
1980 #define EXT4_GOOD_OLD_INODE_SIZE 128
1981
1982 #define EXT4_EXTRA_TIMESTAMP_MAX (((s64)1 << 34) - 1 + S32_MIN)
1983 #define EXT4_NON_EXTRA_TIMESTAMP_MAX S32_MAX
1984 #define EXT4_TIMESTAMP_MIN S32_MIN
1985
1986 /*
1987 * Feature set definitions
1988 */
1989
1990 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1991 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1992 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1993 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1994 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1995 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1996 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
1997 /*
1998 * The reason why "FAST_COMMIT" is a compat feature is that, FS becomes
1999 * incompatible only if fast commit blocks are present in the FS. Since we
2000 * clear the journal (and thus the fast commit blocks), we don't mark FS as
2001 * incompatible. We also have a JBD2 incompat feature, which gets set when
2002 * there are fast commit blocks present in the journal.
2003 */
2004 #define EXT4_FEATURE_COMPAT_FAST_COMMIT 0x0400
2005 #define EXT4_FEATURE_COMPAT_STABLE_INODES 0x0800
2006 #define EXT4_FEATURE_COMPAT_ORPHAN_FILE 0x1000 /* Orphan file exists */
2007
2008 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
2009 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
2010 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
2011 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
2012 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
2013 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
2014 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
2015 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
2016 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
2017 /*
2018 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
2019 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
2020 * all other data structures' checksums. However, the METADATA_CSUM and
2021 * GDT_CSUM bits are mutually exclusive.
2022 */
2023 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
2024 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000
2025 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000
2026 #define EXT4_FEATURE_RO_COMPAT_VERITY 0x8000
2027 #define EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT 0x10000 /* Orphan file may be
2028 non-empty */
2029
2030 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
2031 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
2032 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
2033 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
2034 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
2035 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
2036 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
2037 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
2038 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
2039 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
2040 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
2041 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000
2042 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
2043 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
2044 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000
2045 #define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000
2046
2047 extern void ext4_update_dynamic_rev(struct super_block *sb);
2048
2049 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
2050 static inline bool ext4_has_feature_##name(struct super_block *sb) \
2051 { \
2052 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
2053 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
2054 } \
2055 static inline void ext4_set_feature_##name(struct super_block *sb) \
2056 { \
2057 ext4_update_dynamic_rev(sb); \
2058 EXT4_SB(sb)->s_es->s_feature_compat |= \
2059 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
2060 } \
2061 static inline void ext4_clear_feature_##name(struct super_block *sb) \
2062 { \
2063 EXT4_SB(sb)->s_es->s_feature_compat &= \
2064 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
2065 }
2066
2067 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
2068 static inline bool ext4_has_feature_##name(struct super_block *sb) \
2069 { \
2070 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
2071 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
2072 } \
2073 static inline void ext4_set_feature_##name(struct super_block *sb) \
2074 { \
2075 ext4_update_dynamic_rev(sb); \
2076 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
2077 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
2078 } \
2079 static inline void ext4_clear_feature_##name(struct super_block *sb) \
2080 { \
2081 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
2082 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
2083 }
2084
2085 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
2086 static inline bool ext4_has_feature_##name(struct super_block *sb) \
2087 { \
2088 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
2089 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
2090 } \
2091 static inline void ext4_set_feature_##name(struct super_block *sb) \
2092 { \
2093 ext4_update_dynamic_rev(sb); \
2094 EXT4_SB(sb)->s_es->s_feature_incompat |= \
2095 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
2096 } \
2097 static inline void ext4_clear_feature_##name(struct super_block *sb) \
2098 { \
2099 EXT4_SB(sb)->s_es->s_feature_incompat &= \
2100 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
2101 }
2102
EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc,DIR_PREALLOC)2103 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC)
2104 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES)
2105 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL)
2106 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR)
2107 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE)
2108 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX)
2109 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2)
2110 EXT4_FEATURE_COMPAT_FUNCS(fast_commit, FAST_COMMIT)
2111 EXT4_FEATURE_COMPAT_FUNCS(stable_inodes, STABLE_INODES)
2112 EXT4_FEATURE_COMPAT_FUNCS(orphan_file, ORPHAN_FILE)
2113
2114 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER)
2115 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE)
2116 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR)
2117 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE)
2118 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM)
2119 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK)
2120 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE)
2121 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA)
2122 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC)
2123 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM)
2124 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY)
2125 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT)
2126 EXT4_FEATURE_RO_COMPAT_FUNCS(verity, VERITY)
2127 EXT4_FEATURE_RO_COMPAT_FUNCS(orphan_present, ORPHAN_PRESENT)
2128
2129 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION)
2130 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE)
2131 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER)
2132 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV)
2133 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG)
2134 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS)
2135 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT)
2136 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP)
2137 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG)
2138 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE)
2139 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA)
2140 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED)
2141 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR)
2142 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA)
2143 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT)
2144 EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD)
2145
2146 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
2147 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
2148 EXT4_FEATURE_INCOMPAT_META_BG)
2149 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2150 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2151 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
2152
2153 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
2154 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
2155 EXT4_FEATURE_INCOMPAT_RECOVER| \
2156 EXT4_FEATURE_INCOMPAT_META_BG)
2157 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2158 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2159 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
2160
2161 #define EXT4_FEATURE_COMPAT_SUPP (EXT4_FEATURE_COMPAT_EXT_ATTR| \
2162 EXT4_FEATURE_COMPAT_ORPHAN_FILE)
2163 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
2164 EXT4_FEATURE_INCOMPAT_RECOVER| \
2165 EXT4_FEATURE_INCOMPAT_META_BG| \
2166 EXT4_FEATURE_INCOMPAT_EXTENTS| \
2167 EXT4_FEATURE_INCOMPAT_64BIT| \
2168 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
2169 EXT4_FEATURE_INCOMPAT_EA_INODE| \
2170 EXT4_FEATURE_INCOMPAT_MMP | \
2171 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
2172 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
2173 EXT4_FEATURE_INCOMPAT_CASEFOLD | \
2174 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \
2175 EXT4_FEATURE_INCOMPAT_LARGEDIR)
2176 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2177 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2178 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
2179 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
2180 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
2181 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
2182 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
2183 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
2184 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
2185 EXT4_FEATURE_RO_COMPAT_QUOTA |\
2186 EXT4_FEATURE_RO_COMPAT_PROJECT |\
2187 EXT4_FEATURE_RO_COMPAT_VERITY |\
2188 EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT)
2189
2190 #define EXTN_FEATURE_FUNCS(ver) \
2191 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
2192 { \
2193 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
2194 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
2195 } \
2196 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
2197 { \
2198 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
2199 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
2200 } \
2201 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
2202 { \
2203 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
2204 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
2205 }
2206
2207 EXTN_FEATURE_FUNCS(2)
2208 EXTN_FEATURE_FUNCS(3)
2209 EXTN_FEATURE_FUNCS(4)
2210
2211 static inline bool ext4_has_compat_features(struct super_block *sb)
2212 {
2213 return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
2214 }
ext4_has_ro_compat_features(struct super_block * sb)2215 static inline bool ext4_has_ro_compat_features(struct super_block *sb)
2216 {
2217 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
2218 }
ext4_has_incompat_features(struct super_block * sb)2219 static inline bool ext4_has_incompat_features(struct super_block *sb)
2220 {
2221 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
2222 }
2223
2224 extern int ext4_feature_set_ok(struct super_block *sb, int readonly);
2225
2226 /*
2227 * Superblock flags
2228 */
2229 #define EXT4_FLAGS_RESIZING 0
2230 #define EXT4_FLAGS_SHUTDOWN 1
2231 #define EXT4_FLAGS_BDEV_IS_DAX 2
2232
ext4_forced_shutdown(struct ext4_sb_info * sbi)2233 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi)
2234 {
2235 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags);
2236 }
2237
2238 /*
2239 * Default values for user and/or group using reserved blocks
2240 */
2241 #define EXT4_DEF_RESUID 0
2242 #define EXT4_DEF_RESGID 0
2243
2244 /*
2245 * Default project ID
2246 */
2247 #define EXT4_DEF_PROJID 0
2248
2249 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
2250
2251 /*
2252 * Default mount options
2253 */
2254 #define EXT4_DEFM_DEBUG 0x0001
2255 #define EXT4_DEFM_BSDGROUPS 0x0002
2256 #define EXT4_DEFM_XATTR_USER 0x0004
2257 #define EXT4_DEFM_ACL 0x0008
2258 #define EXT4_DEFM_UID16 0x0010
2259 #define EXT4_DEFM_JMODE 0x0060
2260 #define EXT4_DEFM_JMODE_DATA 0x0020
2261 #define EXT4_DEFM_JMODE_ORDERED 0x0040
2262 #define EXT4_DEFM_JMODE_WBACK 0x0060
2263 #define EXT4_DEFM_NOBARRIER 0x0100
2264 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
2265 #define EXT4_DEFM_DISCARD 0x0400
2266 #define EXT4_DEFM_NODELALLOC 0x0800
2267
2268 /*
2269 * Default journal batch times
2270 */
2271 #define EXT4_DEF_MIN_BATCH_TIME 0
2272 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
2273
2274 /*
2275 * Minimum number of groups in a flexgroup before we separate out
2276 * directories into the first block group of a flexgroup
2277 */
2278 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
2279
2280 /*
2281 * Structure of a directory entry
2282 */
2283 #define EXT4_NAME_LEN 255
2284 /*
2285 * Base length of the ext4 directory entry excluding the name length
2286 */
2287 #define EXT4_BASE_DIR_LEN (sizeof(struct ext4_dir_entry_2) - EXT4_NAME_LEN)
2288
2289 struct ext4_dir_entry {
2290 __le32 inode; /* Inode number */
2291 __le16 rec_len; /* Directory entry length */
2292 __le16 name_len; /* Name length */
2293 char name[EXT4_NAME_LEN]; /* File name */
2294 };
2295
2296
2297 /*
2298 * Encrypted Casefolded entries require saving the hash on disk. This structure
2299 * followed ext4_dir_entry_2's name[name_len] at the next 4 byte aligned
2300 * boundary.
2301 */
2302 struct ext4_dir_entry_hash {
2303 __le32 hash;
2304 __le32 minor_hash;
2305 };
2306
2307 /*
2308 * The new version of the directory entry. Since EXT4 structures are
2309 * stored in intel byte order, and the name_len field could never be
2310 * bigger than 255 chars, it's safe to reclaim the extra byte for the
2311 * file_type field.
2312 */
2313 struct ext4_dir_entry_2 {
2314 __le32 inode; /* Inode number */
2315 __le16 rec_len; /* Directory entry length */
2316 __u8 name_len; /* Name length */
2317 __u8 file_type; /* See file type macros EXT4_FT_* below */
2318 char name[EXT4_NAME_LEN]; /* File name */
2319 };
2320
2321 /*
2322 * Access the hashes at the end of ext4_dir_entry_2
2323 */
2324 #define EXT4_DIRENT_HASHES(entry) \
2325 ((struct ext4_dir_entry_hash *) \
2326 (((void *)(entry)) + \
2327 ((8 + (entry)->name_len + EXT4_DIR_ROUND) & ~EXT4_DIR_ROUND)))
2328 #define EXT4_DIRENT_HASH(entry) le32_to_cpu(EXT4_DIRENT_HASHES(de)->hash)
2329 #define EXT4_DIRENT_MINOR_HASH(entry) \
2330 le32_to_cpu(EXT4_DIRENT_HASHES(de)->minor_hash)
2331
ext4_hash_in_dirent(const struct inode * inode)2332 static inline bool ext4_hash_in_dirent(const struct inode *inode)
2333 {
2334 return IS_CASEFOLDED(inode) && IS_ENCRYPTED(inode);
2335 }
2336
2337 /*
2338 * This is a bogus directory entry at the end of each leaf block that
2339 * records checksums.
2340 */
2341 struct ext4_dir_entry_tail {
2342 __le32 det_reserved_zero1; /* Pretend to be unused */
2343 __le16 det_rec_len; /* 12 */
2344 __u8 det_reserved_zero2; /* Zero name length */
2345 __u8 det_reserved_ft; /* 0xDE, fake file type */
2346 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
2347 };
2348
2349 #define EXT4_DIRENT_TAIL(block, blocksize) \
2350 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
2351 ((blocksize) - \
2352 sizeof(struct ext4_dir_entry_tail))))
2353
2354 /*
2355 * Ext4 directory file types. Only the low 3 bits are used. The
2356 * other bits are reserved for now.
2357 */
2358 #define EXT4_FT_UNKNOWN 0
2359 #define EXT4_FT_REG_FILE 1
2360 #define EXT4_FT_DIR 2
2361 #define EXT4_FT_CHRDEV 3
2362 #define EXT4_FT_BLKDEV 4
2363 #define EXT4_FT_FIFO 5
2364 #define EXT4_FT_SOCK 6
2365 #define EXT4_FT_SYMLINK 7
2366
2367 #define EXT4_FT_MAX 8
2368
2369 #define EXT4_FT_DIR_CSUM 0xDE
2370
2371 /*
2372 * EXT4_DIR_PAD defines the directory entries boundaries
2373 *
2374 * NOTE: It must be a multiple of 4
2375 */
2376 #define EXT4_DIR_PAD 4
2377 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
2378 #define EXT4_MAX_REC_LEN ((1<<16)-1)
2379
2380 /*
2381 * The rec_len is dependent on the type of directory. Directories that are
2382 * casefolded and encrypted need to store the hash as well, so we add room for
2383 * ext4_extended_dir_entry_2. For all entries related to '.' or '..' you should
2384 * pass NULL for dir, as those entries do not use the extra fields.
2385 */
ext4_dir_rec_len(__u8 name_len,const struct inode * dir)2386 static inline unsigned int ext4_dir_rec_len(__u8 name_len,
2387 const struct inode *dir)
2388 {
2389 int rec_len = (name_len + 8 + EXT4_DIR_ROUND);
2390
2391 if (dir && ext4_hash_in_dirent(dir))
2392 rec_len += sizeof(struct ext4_dir_entry_hash);
2393 return (rec_len & ~EXT4_DIR_ROUND);
2394 }
2395
2396 /*
2397 * If we ever get support for fs block sizes > page_size, we'll need
2398 * to remove the #if statements in the next two functions...
2399 */
2400 static inline unsigned int
ext4_rec_len_from_disk(__le16 dlen,unsigned blocksize)2401 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
2402 {
2403 unsigned len = le16_to_cpu(dlen);
2404
2405 #if (PAGE_SIZE >= 65536)
2406 if (len == EXT4_MAX_REC_LEN || len == 0)
2407 return blocksize;
2408 return (len & 65532) | ((len & 3) << 16);
2409 #else
2410 return len;
2411 #endif
2412 }
2413
ext4_rec_len_to_disk(unsigned len,unsigned blocksize)2414 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
2415 {
2416 BUG_ON((len > blocksize) || (blocksize > (1 << 18)) || (len & 3));
2417 #if (PAGE_SIZE >= 65536)
2418 if (len < 65536)
2419 return cpu_to_le16(len);
2420 if (len == blocksize) {
2421 if (blocksize == 65536)
2422 return cpu_to_le16(EXT4_MAX_REC_LEN);
2423 else
2424 return cpu_to_le16(0);
2425 }
2426 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
2427 #else
2428 return cpu_to_le16(len);
2429 #endif
2430 }
2431
2432 /*
2433 * Hash Tree Directory indexing
2434 * (c) Daniel Phillips, 2001
2435 */
2436
2437 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
2438 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2439 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \
2440 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir)))
2441 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2442
2443 /* Legal values for the dx_root hash_version field: */
2444
2445 #define DX_HASH_LEGACY 0
2446 #define DX_HASH_HALF_MD4 1
2447 #define DX_HASH_TEA 2
2448 #define DX_HASH_LEGACY_UNSIGNED 3
2449 #define DX_HASH_HALF_MD4_UNSIGNED 4
2450 #define DX_HASH_TEA_UNSIGNED 5
2451 #define DX_HASH_SIPHASH 6
2452
ext4_chksum(struct ext4_sb_info * sbi,u32 crc,const void * address,unsigned int length)2453 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2454 const void *address, unsigned int length)
2455 {
2456 struct {
2457 struct shash_desc shash;
2458 char ctx[4];
2459 } desc;
2460
2461 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2462
2463 desc.shash.tfm = sbi->s_chksum_driver;
2464 *(u32 *)desc.ctx = crc;
2465
2466 BUG_ON(crypto_shash_update(&desc.shash, address, length));
2467
2468 return *(u32 *)desc.ctx;
2469 }
2470
2471 #ifdef __KERNEL__
2472
2473 /* hash info structure used by the directory hash */
2474 struct dx_hash_info
2475 {
2476 u32 hash;
2477 u32 minor_hash;
2478 int hash_version;
2479 u32 *seed;
2480 };
2481
2482
2483 /* 32 and 64 bit signed EOF for dx directories */
2484 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
2485 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
2486
2487
2488 /*
2489 * Control parameters used by ext4_htree_next_block
2490 */
2491 #define HASH_NB_ALWAYS 1
2492
2493 struct ext4_filename {
2494 const struct qstr *usr_fname;
2495 struct fscrypt_str disk_name;
2496 struct dx_hash_info hinfo;
2497 #ifdef CONFIG_FS_ENCRYPTION
2498 struct fscrypt_str crypto_buf;
2499 #endif
2500 #if IS_ENABLED(CONFIG_UNICODE)
2501 struct fscrypt_str cf_name;
2502 #endif
2503 };
2504
2505 #define fname_name(p) ((p)->disk_name.name)
2506 #define fname_usr_name(p) ((p)->usr_fname->name)
2507 #define fname_len(p) ((p)->disk_name.len)
2508
2509 /*
2510 * Describe an inode's exact location on disk and in memory
2511 */
2512 struct ext4_iloc
2513 {
2514 struct buffer_head *bh;
2515 unsigned long offset;
2516 ext4_group_t block_group;
2517 };
2518
ext4_raw_inode(struct ext4_iloc * iloc)2519 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2520 {
2521 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2522 }
2523
ext4_is_quota_file(struct inode * inode)2524 static inline bool ext4_is_quota_file(struct inode *inode)
2525 {
2526 return IS_NOQUOTA(inode) &&
2527 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL);
2528 }
2529
2530 /*
2531 * This structure is stuffed into the struct file's private_data field
2532 * for directories. It is where we put information so that we can do
2533 * readdir operations in hash tree order.
2534 */
2535 struct dir_private_info {
2536 struct rb_root root;
2537 struct rb_node *curr_node;
2538 struct fname *extra_fname;
2539 loff_t last_pos;
2540 __u32 curr_hash;
2541 __u32 curr_minor_hash;
2542 __u32 next_hash;
2543 };
2544
2545 /* calculate the first block number of the group */
2546 static inline ext4_fsblk_t
ext4_group_first_block_no(struct super_block * sb,ext4_group_t group_no)2547 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2548 {
2549 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2550 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2551 }
2552
2553 /*
2554 * Special error return code only used by dx_probe() and its callers.
2555 */
2556 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
2557
2558 /* htree levels for ext4 */
2559 #define EXT4_HTREE_LEVEL_COMPAT 2
2560 #define EXT4_HTREE_LEVEL 3
2561
ext4_dir_htree_level(struct super_block * sb)2562 static inline int ext4_dir_htree_level(struct super_block *sb)
2563 {
2564 return ext4_has_feature_largedir(sb) ?
2565 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT;
2566 }
2567
2568 /*
2569 * Timeout and state flag for lazy initialization inode thread.
2570 */
2571 #define EXT4_DEF_LI_WAIT_MULT 10
2572 #define EXT4_DEF_LI_MAX_START_DELAY 5
2573 #define EXT4_LAZYINIT_QUIT 0x0001
2574 #define EXT4_LAZYINIT_RUNNING 0x0002
2575
2576 /*
2577 * Lazy inode table initialization info
2578 */
2579 struct ext4_lazy_init {
2580 unsigned long li_state;
2581 struct list_head li_request_list;
2582 struct mutex li_list_mtx;
2583 };
2584
2585 enum ext4_li_mode {
2586 EXT4_LI_MODE_PREFETCH_BBITMAP,
2587 EXT4_LI_MODE_ITABLE,
2588 };
2589
2590 struct ext4_li_request {
2591 struct super_block *lr_super;
2592 enum ext4_li_mode lr_mode;
2593 ext4_group_t lr_first_not_zeroed;
2594 ext4_group_t lr_next_group;
2595 struct list_head lr_request;
2596 unsigned long lr_next_sched;
2597 unsigned long lr_timeout;
2598 };
2599
2600 struct ext4_features {
2601 struct kobject f_kobj;
2602 struct completion f_kobj_unregister;
2603 };
2604
2605 /*
2606 * This structure will be used for multiple mount protection. It will be
2607 * written into the block number saved in the s_mmp_block field in the
2608 * superblock. Programs that check MMP should assume that if
2609 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2610 * to use the filesystem, regardless of how old the timestamp is.
2611 */
2612 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
2613 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2614 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
2615 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
2616
2617 struct mmp_struct {
2618 __le32 mmp_magic; /* Magic number for MMP */
2619 __le32 mmp_seq; /* Sequence no. updated periodically */
2620
2621 /*
2622 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2623 * purposes and do not affect the correctness of the algorithm
2624 */
2625 __le64 mmp_time; /* Time last updated */
2626 char mmp_nodename[64]; /* Node which last updated MMP block */
2627 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
2628
2629 /*
2630 * mmp_check_interval is used to verify if the MMP block has been
2631 * updated on the block device. The value is updated based on the
2632 * maximum time to write the MMP block during an update cycle.
2633 */
2634 __le16 mmp_check_interval;
2635
2636 __le16 mmp_pad1;
2637 __le32 mmp_pad2[226];
2638 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
2639 };
2640
2641 /* arguments passed to the mmp thread */
2642 struct mmpd_data {
2643 struct buffer_head *bh; /* bh from initial read_mmp_block() */
2644 struct super_block *sb; /* super block of the fs */
2645 };
2646
2647 /*
2648 * Check interval multiplier
2649 * The MMP block is written every update interval and initially checked every
2650 * update interval x the multiplier (the value is then adapted based on the
2651 * write latency). The reason is that writes can be delayed under load and we
2652 * don't want readers to incorrectly assume that the filesystem is no longer
2653 * in use.
2654 */
2655 #define EXT4_MMP_CHECK_MULT 2UL
2656
2657 /*
2658 * Minimum interval for MMP checking in seconds.
2659 */
2660 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
2661
2662 /*
2663 * Maximum interval for MMP checking in seconds.
2664 */
2665 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
2666
2667 /*
2668 * Function prototypes
2669 */
2670
2671 /*
2672 * Ok, these declarations are also in <linux/kernel.h> but none of the
2673 * ext4 source programs needs to include it so they are duplicated here.
2674 */
2675 # define NORET_TYPE /**/
2676 # define ATTRIB_NORET __attribute__((noreturn))
2677 # define NORET_AND noreturn,
2678
2679 /* bitmap.c */
2680 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2681 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2682 struct ext4_group_desc *gdp,
2683 struct buffer_head *bh, int sz);
2684 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2685 struct ext4_group_desc *gdp,
2686 struct buffer_head *bh, int sz);
2687 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2688 struct ext4_group_desc *gdp,
2689 struct buffer_head *bh);
2690 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2691 struct ext4_group_desc *gdp,
2692 struct buffer_head *bh);
2693
2694 /* balloc.c */
2695 extern void ext4_get_group_no_and_offset(struct super_block *sb,
2696 ext4_fsblk_t blocknr,
2697 ext4_group_t *blockgrpp,
2698 ext4_grpblk_t *offsetp);
2699 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2700 ext4_fsblk_t block);
2701
2702 extern unsigned int ext4_block_group(struct super_block *sb,
2703 ext4_fsblk_t blocknr);
2704 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2705 ext4_fsblk_t blocknr);
2706 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2707 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2708 ext4_group_t group);
2709 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2710 ext4_fsblk_t goal,
2711 unsigned int flags,
2712 unsigned long *count,
2713 int *errp);
2714 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2715 s64 nclusters, unsigned int flags);
2716 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2717 extern void ext4_check_blocks_bitmap(struct super_block *);
2718 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2719 ext4_group_t block_group,
2720 struct buffer_head ** bh);
2721 extern struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2722 ext4_group_t group);
2723 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2724
2725 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2726 ext4_group_t block_group,
2727 bool ignore_locked);
2728 extern int ext4_wait_block_bitmap(struct super_block *sb,
2729 ext4_group_t block_group,
2730 struct buffer_head *bh);
2731 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2732 ext4_group_t block_group);
2733 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2734 ext4_group_t block_group,
2735 struct ext4_group_desc *gdp);
2736 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2737
2738 #if IS_ENABLED(CONFIG_UNICODE)
2739 extern int ext4_fname_setup_ci_filename(struct inode *dir,
2740 const struct qstr *iname,
2741 struct ext4_filename *fname);
2742 #endif
2743
2744 /* ext4 encryption related stuff goes here crypto.c */
2745 #ifdef CONFIG_FS_ENCRYPTION
2746 extern const struct fscrypt_operations ext4_cryptops;
2747
2748 int ext4_fname_setup_filename(struct inode *dir, const struct qstr *iname,
2749 int lookup, struct ext4_filename *fname);
2750
2751 int ext4_fname_prepare_lookup(struct inode *dir, struct dentry *dentry,
2752 struct ext4_filename *fname);
2753
2754 void ext4_fname_free_filename(struct ext4_filename *fname);
2755
2756 int ext4_ioctl_get_encryption_pwsalt(struct file *filp, void __user *arg);
2757
2758 #else /* !CONFIG_FS_ENCRYPTION */
ext4_fname_setup_filename(struct inode * dir,const struct qstr * iname,int lookup,struct ext4_filename * fname)2759 static inline int ext4_fname_setup_filename(struct inode *dir,
2760 const struct qstr *iname,
2761 int lookup,
2762 struct ext4_filename *fname)
2763 {
2764 int err = 0;
2765 fname->usr_fname = iname;
2766 fname->disk_name.name = (unsigned char *) iname->name;
2767 fname->disk_name.len = iname->len;
2768
2769 #if IS_ENABLED(CONFIG_UNICODE)
2770 err = ext4_fname_setup_ci_filename(dir, iname, fname);
2771 #endif
2772
2773 return err;
2774 }
2775
ext4_fname_prepare_lookup(struct inode * dir,struct dentry * dentry,struct ext4_filename * fname)2776 static inline int ext4_fname_prepare_lookup(struct inode *dir,
2777 struct dentry *dentry,
2778 struct ext4_filename *fname)
2779 {
2780 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname);
2781 }
2782
ext4_fname_free_filename(struct ext4_filename * fname)2783 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2784 {
2785 #if IS_ENABLED(CONFIG_UNICODE)
2786 kfree(fname->cf_name.name);
2787 fname->cf_name.name = NULL;
2788 #endif
2789 }
2790
ext4_ioctl_get_encryption_pwsalt(struct file * filp,void __user * arg)2791 static inline int ext4_ioctl_get_encryption_pwsalt(struct file *filp,
2792 void __user *arg)
2793 {
2794 return -EOPNOTSUPP;
2795 }
2796 #endif /* !CONFIG_FS_ENCRYPTION */
2797
2798 /* dir.c */
2799 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2800 struct file *,
2801 struct ext4_dir_entry_2 *,
2802 struct buffer_head *, char *, int,
2803 unsigned int);
2804 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2805 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2806 (de), (bh), (buf), (size), (offset)))
2807 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2808 __u32 minor_hash,
2809 struct ext4_dir_entry_2 *dirent,
2810 struct fscrypt_str *ent_name);
2811 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2812 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2813 struct buffer_head *bh,
2814 void *buf, int buf_size,
2815 struct ext4_filename *fname,
2816 struct ext4_dir_entry_2 **dest_de);
2817 void ext4_insert_dentry(struct inode *dir, struct inode *inode,
2818 struct ext4_dir_entry_2 *de,
2819 int buf_size,
2820 struct ext4_filename *fname);
ext4_update_dx_flag(struct inode * inode)2821 static inline void ext4_update_dx_flag(struct inode *inode)
2822 {
2823 if (!ext4_has_feature_dir_index(inode->i_sb) &&
2824 ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) {
2825 /* ext4_iget() should have caught this... */
2826 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb));
2827 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2828 }
2829 }
2830 static const unsigned char ext4_filetype_table[] = {
2831 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2832 };
2833
get_dtype(struct super_block * sb,int filetype)2834 static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2835 {
2836 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2837 return DT_UNKNOWN;
2838
2839 return ext4_filetype_table[filetype];
2840 }
2841 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2842 void *buf, int buf_size);
2843
2844 /* fsync.c */
2845 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2846
2847 /* hash.c */
2848 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len,
2849 struct dx_hash_info *hinfo);
2850
2851 /* ialloc.c */
2852 extern int ext4_mark_inode_used(struct super_block *sb, int ino);
2853 extern struct inode *__ext4_new_inode(struct user_namespace *, handle_t *,
2854 struct inode *, umode_t,
2855 const struct qstr *qstr, __u32 goal,
2856 uid_t *owner, __u32 i_flags,
2857 int handle_type, unsigned int line_no,
2858 int nblocks);
2859
2860 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \
2861 __ext4_new_inode(&init_user_ns, (handle), (dir), (mode), (qstr), \
2862 (goal), (owner), i_flags, 0, 0, 0)
2863 #define ext4_new_inode_start_handle(mnt_userns, dir, mode, qstr, goal, owner, \
2864 type, nblocks) \
2865 __ext4_new_inode((mnt_userns), NULL, (dir), (mode), (qstr), (goal), (owner), \
2866 0, (type), __LINE__, (nblocks))
2867
2868
2869 extern void ext4_free_inode(handle_t *, struct inode *);
2870 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2871 extern unsigned long ext4_count_free_inodes(struct super_block *);
2872 extern unsigned long ext4_count_dirs(struct super_block *);
2873 extern void ext4_check_inodes_bitmap(struct super_block *);
2874 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2875 extern int ext4_init_inode_table(struct super_block *sb,
2876 ext4_group_t group, int barrier);
2877 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2878
2879 /* fast_commit.c */
2880 int ext4_fc_info_show(struct seq_file *seq, void *v);
2881 void ext4_fc_init(struct super_block *sb, journal_t *journal);
2882 void ext4_fc_init_inode(struct inode *inode);
2883 void ext4_fc_track_range(handle_t *handle, struct inode *inode, ext4_lblk_t start,
2884 ext4_lblk_t end);
2885 void __ext4_fc_track_unlink(handle_t *handle, struct inode *inode,
2886 struct dentry *dentry);
2887 void __ext4_fc_track_link(handle_t *handle, struct inode *inode,
2888 struct dentry *dentry);
2889 void ext4_fc_track_unlink(handle_t *handle, struct dentry *dentry);
2890 void ext4_fc_track_link(handle_t *handle, struct dentry *dentry);
2891 void __ext4_fc_track_create(handle_t *handle, struct inode *inode,
2892 struct dentry *dentry);
2893 void ext4_fc_track_create(handle_t *handle, struct dentry *dentry);
2894 void ext4_fc_track_inode(handle_t *handle, struct inode *inode);
2895 void ext4_fc_mark_ineligible(struct super_block *sb, int reason, handle_t *handle);
2896 void ext4_fc_start_update(struct inode *inode);
2897 void ext4_fc_stop_update(struct inode *inode);
2898 void ext4_fc_del(struct inode *inode);
2899 bool ext4_fc_replay_check_excluded(struct super_block *sb, ext4_fsblk_t block);
2900 void ext4_fc_replay_cleanup(struct super_block *sb);
2901 int ext4_fc_commit(journal_t *journal, tid_t commit_tid);
2902 int __init ext4_fc_init_dentry_cache(void);
2903 void ext4_fc_destroy_dentry_cache(void);
2904 int ext4_fc_record_regions(struct super_block *sb, int ino,
2905 ext4_lblk_t lblk, ext4_fsblk_t pblk,
2906 int len, int replay);
2907
2908 /* mballoc.c */
2909 extern const struct seq_operations ext4_mb_seq_groups_ops;
2910 extern const struct seq_operations ext4_mb_seq_structs_summary_ops;
2911 extern long ext4_mb_stats;
2912 extern long ext4_mb_max_to_scan;
2913 extern int ext4_seq_mb_stats_show(struct seq_file *seq, void *offset);
2914 extern int ext4_mb_init(struct super_block *);
2915 extern int ext4_mb_release(struct super_block *);
2916 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2917 struct ext4_allocation_request *, int *);
2918 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2919 extern void ext4_discard_preallocations(struct inode *, unsigned int);
2920 extern int __init ext4_init_mballoc(void);
2921 extern void ext4_exit_mballoc(void);
2922 extern ext4_group_t ext4_mb_prefetch(struct super_block *sb,
2923 ext4_group_t group,
2924 unsigned int nr, int *cnt);
2925 extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group,
2926 unsigned int nr);
2927
2928 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2929 struct buffer_head *bh, ext4_fsblk_t block,
2930 unsigned long count, int flags);
2931 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2932 ext4_group_t ngroups);
2933 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2934 ext4_group_t i, struct ext4_group_desc *desc);
2935 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2936 ext4_fsblk_t block, unsigned long count);
2937 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2938 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid);
2939 extern void ext4_mb_mark_bb(struct super_block *sb, ext4_fsblk_t block,
2940 int len, int state);
2941
2942 /* inode.c */
2943 void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw,
2944 struct ext4_inode_info *ei);
2945 int ext4_inode_is_fast_symlink(struct inode *inode);
2946 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2947 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2948 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count,
2949 bool wait, struct buffer_head **bhs);
2950 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2951 struct buffer_head *bh_result, int create);
2952 int ext4_get_block(struct inode *inode, sector_t iblock,
2953 struct buffer_head *bh_result, int create);
2954 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2955 struct buffer_head *bh, int create);
2956 int ext4_walk_page_buffers(handle_t *handle,
2957 struct inode *inode,
2958 struct buffer_head *head,
2959 unsigned from,
2960 unsigned to,
2961 int *partial,
2962 int (*fn)(handle_t *handle, struct inode *inode,
2963 struct buffer_head *bh));
2964 int do_journal_get_write_access(handle_t *handle, struct inode *inode,
2965 struct buffer_head *bh);
2966 #define FALL_BACK_TO_NONDELALLOC 1
2967 #define CONVERT_INLINE_DATA 2
2968
2969 typedef enum {
2970 EXT4_IGET_NORMAL = 0,
2971 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */
2972 EXT4_IGET_HANDLE = 0x0002, /* Inode # is from a handle */
2973 EXT4_IGET_BAD = 0x0004, /* Allow to iget a bad inode */
2974 EXT4_IGET_EA_INODE = 0x0008 /* Inode should contain an EA value */
2975 } ext4_iget_flags;
2976
2977 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
2978 ext4_iget_flags flags, const char *function,
2979 unsigned int line);
2980
2981 #define ext4_iget(sb, ino, flags) \
2982 __ext4_iget((sb), (ino), (flags), __func__, __LINE__)
2983
2984 extern int ext4_write_inode(struct inode *, struct writeback_control *);
2985 extern int ext4_setattr(struct user_namespace *, struct dentry *,
2986 struct iattr *);
2987 extern u32 ext4_dio_alignment(struct inode *inode);
2988 extern int ext4_getattr(struct user_namespace *, const struct path *,
2989 struct kstat *, u32, unsigned int);
2990 extern void ext4_evict_inode(struct inode *);
2991 extern void ext4_clear_inode(struct inode *);
2992 extern int ext4_file_getattr(struct user_namespace *, const struct path *,
2993 struct kstat *, u32, unsigned int);
2994 extern int ext4_sync_inode(handle_t *, struct inode *);
2995 extern void ext4_dirty_inode(struct inode *, int);
2996 extern int ext4_change_inode_journal_flag(struct inode *, int);
2997 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2998 extern int ext4_get_fc_inode_loc(struct super_block *sb, unsigned long ino,
2999 struct ext4_iloc *iloc);
3000 extern int ext4_inode_attach_jinode(struct inode *inode);
3001 extern int ext4_can_truncate(struct inode *inode);
3002 extern int ext4_truncate(struct inode *);
3003 extern int ext4_break_layouts(struct inode *);
3004 extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length);
3005 extern void ext4_set_inode_flags(struct inode *, bool init);
3006 extern int ext4_alloc_da_blocks(struct inode *inode);
3007 extern void ext4_set_aops(struct inode *inode);
3008 extern int ext4_writepage_trans_blocks(struct inode *);
3009 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
3010 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
3011 loff_t lstart, loff_t lend);
3012 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf);
3013 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
3014 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
3015 extern void ext4_da_release_space(struct inode *inode, int to_free);
3016 extern void ext4_da_update_reserve_space(struct inode *inode,
3017 int used, int quota_claim);
3018 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
3019 ext4_fsblk_t pblk, ext4_lblk_t len);
3020
3021 /* indirect.c */
3022 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
3023 struct ext4_map_blocks *map, int flags);
3024 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
3025 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
3026 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
3027 ext4_lblk_t start, ext4_lblk_t end);
3028
3029 /* ioctl.c */
3030 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
3031 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
3032 int ext4_fileattr_set(struct user_namespace *mnt_userns,
3033 struct dentry *dentry, struct fileattr *fa);
3034 int ext4_fileattr_get(struct dentry *dentry, struct fileattr *fa);
3035 extern void ext4_reset_inode_seed(struct inode *inode);
3036 int ext4_update_overhead(struct super_block *sb, bool force);
3037
3038 /* migrate.c */
3039 extern int ext4_ext_migrate(struct inode *);
3040 extern int ext4_ind_migrate(struct inode *inode);
3041
3042 /* namei.c */
3043 extern int ext4_init_new_dir(handle_t *handle, struct inode *dir,
3044 struct inode *inode);
3045 extern int ext4_dirblock_csum_verify(struct inode *inode,
3046 struct buffer_head *bh);
3047 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
3048 __u32 start_minor_hash, __u32 *next_hash);
3049 extern int ext4_search_dir(struct buffer_head *bh,
3050 char *search_buf,
3051 int buf_size,
3052 struct inode *dir,
3053 struct ext4_filename *fname,
3054 unsigned int offset,
3055 struct ext4_dir_entry_2 **res_dir);
3056 extern int ext4_generic_delete_entry(struct inode *dir,
3057 struct ext4_dir_entry_2 *de_del,
3058 struct buffer_head *bh,
3059 void *entry_buf,
3060 int buf_size,
3061 int csum_size);
3062 extern bool ext4_empty_dir(struct inode *inode);
3063
3064 /* resize.c */
3065 extern void ext4_kvfree_array_rcu(void *to_free);
3066 extern int ext4_group_add(struct super_block *sb,
3067 struct ext4_new_group_data *input);
3068 extern int ext4_group_extend(struct super_block *sb,
3069 struct ext4_super_block *es,
3070 ext4_fsblk_t n_blocks_count);
3071 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
3072 extern unsigned int ext4_list_backups(struct super_block *sb,
3073 unsigned int *three, unsigned int *five,
3074 unsigned int *seven);
3075
3076 /* super.c */
3077 extern struct buffer_head *ext4_sb_bread(struct super_block *sb,
3078 sector_t block, blk_opf_t op_flags);
3079 extern struct buffer_head *ext4_sb_bread_unmovable(struct super_block *sb,
3080 sector_t block);
3081 extern void ext4_read_bh_nowait(struct buffer_head *bh, blk_opf_t op_flags,
3082 bh_end_io_t *end_io);
3083 extern int ext4_read_bh(struct buffer_head *bh, blk_opf_t op_flags,
3084 bh_end_io_t *end_io);
3085 extern int ext4_read_bh_lock(struct buffer_head *bh, blk_opf_t op_flags, bool wait);
3086 extern void ext4_sb_breadahead_unmovable(struct super_block *sb, sector_t block);
3087 extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
3088 extern int ext4_calculate_overhead(struct super_block *sb);
3089 extern __le32 ext4_superblock_csum(struct super_block *sb,
3090 struct ext4_super_block *es);
3091 extern void ext4_superblock_csum_set(struct super_block *sb);
3092 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
3093 ext4_group_t ngroup);
3094 extern const char *ext4_decode_error(struct super_block *sb, int errno,
3095 char nbuf[16]);
3096 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb,
3097 ext4_group_t block_group,
3098 unsigned int flags);
3099 extern unsigned int ext4_num_base_meta_blocks(struct super_block *sb,
3100 ext4_group_t block_group);
3101
3102 extern __printf(7, 8)
3103 void __ext4_error(struct super_block *, const char *, unsigned int, bool,
3104 int, __u64, const char *, ...);
3105 extern __printf(6, 7)
3106 void __ext4_error_inode(struct inode *, const char *, unsigned int,
3107 ext4_fsblk_t, int, const char *, ...);
3108 extern __printf(5, 6)
3109 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
3110 const char *, ...);
3111 extern void __ext4_std_error(struct super_block *, const char *,
3112 unsigned int, int);
3113 extern __printf(4, 5)
3114 void __ext4_warning(struct super_block *, const char *, unsigned int,
3115 const char *, ...);
3116 extern __printf(4, 5)
3117 void __ext4_warning_inode(const struct inode *inode, const char *function,
3118 unsigned int line, const char *fmt, ...);
3119 extern __printf(3, 4)
3120 void __ext4_msg(struct super_block *, const char *, const char *, ...);
3121 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
3122 const char *, unsigned int, const char *);
3123 extern __printf(7, 8)
3124 void __ext4_grp_locked_error(const char *, unsigned int,
3125 struct super_block *, ext4_group_t,
3126 unsigned long, ext4_fsblk_t,
3127 const char *, ...);
3128
3129 #define EXT4_ERROR_INODE(inode, fmt, a...) \
3130 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
3131
3132 #define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...) \
3133 __ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a)
3134
3135 #define ext4_error_inode_block(inode, block, err, fmt, a...) \
3136 __ext4_error_inode((inode), __func__, __LINE__, (block), (err), \
3137 (fmt), ## a)
3138
3139 #define EXT4_ERROR_FILE(file, block, fmt, a...) \
3140 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
3141
3142 #define ext4_abort(sb, err, fmt, a...) \
3143 __ext4_error((sb), __func__, __LINE__, true, (err), 0, (fmt), ## a)
3144
3145 #ifdef CONFIG_PRINTK
3146
3147 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
3148 __ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__)
3149 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \
3150 __ext4_error_inode((inode), (func), (line), (block), \
3151 (err), (fmt), ##__VA_ARGS__)
3152 #define ext4_error_file(file, func, line, block, fmt, ...) \
3153 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
3154 #define ext4_error(sb, fmt, ...) \
3155 __ext4_error((sb), __func__, __LINE__, false, 0, 0, (fmt), \
3156 ##__VA_ARGS__)
3157 #define ext4_error_err(sb, err, fmt, ...) \
3158 __ext4_error((sb), __func__, __LINE__, false, (err), 0, (fmt), \
3159 ##__VA_ARGS__)
3160 #define ext4_warning(sb, fmt, ...) \
3161 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
3162 #define ext4_warning_inode(inode, fmt, ...) \
3163 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
3164 #define ext4_msg(sb, level, fmt, ...) \
3165 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
3166 #define dump_mmp_msg(sb, mmp, msg) \
3167 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
3168 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
3169 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
3170 fmt, ##__VA_ARGS__)
3171
3172 #else
3173
3174 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
3175 do { \
3176 no_printk(fmt, ##__VA_ARGS__); \
3177 __ext4_error_inode(inode, "", 0, block, 0, " "); \
3178 } while (0)
3179 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \
3180 do { \
3181 no_printk(fmt, ##__VA_ARGS__); \
3182 __ext4_error_inode(inode, "", 0, block, err, " "); \
3183 } while (0)
3184 #define ext4_error_file(file, func, line, block, fmt, ...) \
3185 do { \
3186 no_printk(fmt, ##__VA_ARGS__); \
3187 __ext4_error_file(file, "", 0, block, " "); \
3188 } while (0)
3189 #define ext4_error(sb, fmt, ...) \
3190 do { \
3191 no_printk(fmt, ##__VA_ARGS__); \
3192 __ext4_error(sb, "", 0, false, 0, 0, " "); \
3193 } while (0)
3194 #define ext4_error_err(sb, err, fmt, ...) \
3195 do { \
3196 no_printk(fmt, ##__VA_ARGS__); \
3197 __ext4_error(sb, "", 0, false, err, 0, " "); \
3198 } while (0)
3199 #define ext4_warning(sb, fmt, ...) \
3200 do { \
3201 no_printk(fmt, ##__VA_ARGS__); \
3202 __ext4_warning(sb, "", 0, " "); \
3203 } while (0)
3204 #define ext4_warning_inode(inode, fmt, ...) \
3205 do { \
3206 no_printk(fmt, ##__VA_ARGS__); \
3207 __ext4_warning_inode(inode, "", 0, " "); \
3208 } while (0)
3209 #define ext4_msg(sb, level, fmt, ...) \
3210 do { \
3211 no_printk(fmt, ##__VA_ARGS__); \
3212 __ext4_msg(sb, "", " "); \
3213 } while (0)
3214 #define dump_mmp_msg(sb, mmp, msg) \
3215 __dump_mmp_msg(sb, mmp, "", 0, "")
3216 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
3217 do { \
3218 no_printk(fmt, ##__VA_ARGS__); \
3219 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
3220 } while (0)
3221
3222 #endif
3223
3224 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
3225 struct ext4_group_desc *bg);
3226 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
3227 struct ext4_group_desc *bg);
3228 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
3229 struct ext4_group_desc *bg);
3230 extern __u32 ext4_free_group_clusters(struct super_block *sb,
3231 struct ext4_group_desc *bg);
3232 extern __u32 ext4_free_inodes_count(struct super_block *sb,
3233 struct ext4_group_desc *bg);
3234 extern __u32 ext4_used_dirs_count(struct super_block *sb,
3235 struct ext4_group_desc *bg);
3236 extern __u32 ext4_itable_unused_count(struct super_block *sb,
3237 struct ext4_group_desc *bg);
3238 extern void ext4_block_bitmap_set(struct super_block *sb,
3239 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3240 extern void ext4_inode_bitmap_set(struct super_block *sb,
3241 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3242 extern void ext4_inode_table_set(struct super_block *sb,
3243 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3244 extern void ext4_free_group_clusters_set(struct super_block *sb,
3245 struct ext4_group_desc *bg,
3246 __u32 count);
3247 extern void ext4_free_inodes_set(struct super_block *sb,
3248 struct ext4_group_desc *bg, __u32 count);
3249 extern void ext4_used_dirs_set(struct super_block *sb,
3250 struct ext4_group_desc *bg, __u32 count);
3251 extern void ext4_itable_unused_set(struct super_block *sb,
3252 struct ext4_group_desc *bg, __u32 count);
3253 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
3254 struct ext4_group_desc *gdp);
3255 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
3256 struct ext4_group_desc *gdp);
3257 extern int ext4_register_li_request(struct super_block *sb,
3258 ext4_group_t first_not_zeroed);
3259
ext4_has_metadata_csum(struct super_block * sb)3260 static inline int ext4_has_metadata_csum(struct super_block *sb)
3261 {
3262 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
3263 !EXT4_SB(sb)->s_chksum_driver);
3264
3265 return ext4_has_feature_metadata_csum(sb) &&
3266 (EXT4_SB(sb)->s_chksum_driver != NULL);
3267 }
3268
ext4_has_group_desc_csum(struct super_block * sb)3269 static inline int ext4_has_group_desc_csum(struct super_block *sb)
3270 {
3271 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb);
3272 }
3273
3274 #define ext4_read_incompat_64bit_val(es, name) \
3275 (((es)->s_feature_incompat & cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT) \
3276 ? (ext4_fsblk_t)le32_to_cpu(es->name##_hi) << 32 : 0) | \
3277 le32_to_cpu(es->name##_lo))
3278
ext4_blocks_count(struct ext4_super_block * es)3279 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
3280 {
3281 return ext4_read_incompat_64bit_val(es, s_blocks_count);
3282 }
3283
ext4_r_blocks_count(struct ext4_super_block * es)3284 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
3285 {
3286 return ext4_read_incompat_64bit_val(es, s_r_blocks_count);
3287 }
3288
ext4_free_blocks_count(struct ext4_super_block * es)3289 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
3290 {
3291 return ext4_read_incompat_64bit_val(es, s_free_blocks_count);
3292 }
3293
ext4_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)3294 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
3295 ext4_fsblk_t blk)
3296 {
3297 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
3298 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
3299 }
3300
ext4_free_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)3301 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
3302 ext4_fsblk_t blk)
3303 {
3304 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
3305 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
3306 }
3307
ext4_r_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)3308 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
3309 ext4_fsblk_t blk)
3310 {
3311 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
3312 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
3313 }
3314
ext4_isize(struct super_block * sb,struct ext4_inode * raw_inode)3315 static inline loff_t ext4_isize(struct super_block *sb,
3316 struct ext4_inode *raw_inode)
3317 {
3318 if (ext4_has_feature_largedir(sb) ||
3319 S_ISREG(le16_to_cpu(raw_inode->i_mode)))
3320 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
3321 le32_to_cpu(raw_inode->i_size_lo);
3322
3323 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
3324 }
3325
ext4_isize_set(struct ext4_inode * raw_inode,loff_t i_size)3326 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
3327 {
3328 raw_inode->i_size_lo = cpu_to_le32(i_size);
3329 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
3330 }
3331
3332 /*
3333 * Reading s_groups_count requires using smp_rmb() afterwards. See
3334 * the locking protocol documented in the comments of ext4_group_add()
3335 * in resize.c
3336 */
ext4_get_groups_count(struct super_block * sb)3337 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
3338 {
3339 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
3340
3341 smp_rmb();
3342 return ngroups;
3343 }
3344
ext4_flex_group(struct ext4_sb_info * sbi,ext4_group_t block_group)3345 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
3346 ext4_group_t block_group)
3347 {
3348 return block_group >> sbi->s_log_groups_per_flex;
3349 }
3350
ext4_flex_bg_size(struct ext4_sb_info * sbi)3351 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
3352 {
3353 return 1 << sbi->s_log_groups_per_flex;
3354 }
3355
3356 #define ext4_std_error(sb, errno) \
3357 do { \
3358 if ((errno)) \
3359 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
3360 } while (0)
3361
3362 #ifdef CONFIG_SMP
3363 /* Each CPU can accumulate percpu_counter_batch clusters in their local
3364 * counters. So we need to make sure we have free clusters more
3365 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
3366 */
3367 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
3368 #else
3369 #define EXT4_FREECLUSTERS_WATERMARK 0
3370 #endif
3371
3372 /* Update i_disksize. Requires i_rwsem to avoid races with truncate */
ext4_update_i_disksize(struct inode * inode,loff_t newsize)3373 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
3374 {
3375 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
3376 !inode_is_locked(inode));
3377 down_write(&EXT4_I(inode)->i_data_sem);
3378 if (newsize > EXT4_I(inode)->i_disksize)
3379 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize);
3380 up_write(&EXT4_I(inode)->i_data_sem);
3381 }
3382
3383 /* Update i_size, i_disksize. Requires i_rwsem to avoid races with truncate */
ext4_update_inode_size(struct inode * inode,loff_t newsize)3384 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
3385 {
3386 int changed = 0;
3387
3388 if (newsize > inode->i_size) {
3389 i_size_write(inode, newsize);
3390 changed = 1;
3391 }
3392 if (newsize > EXT4_I(inode)->i_disksize) {
3393 ext4_update_i_disksize(inode, newsize);
3394 changed |= 2;
3395 }
3396 return changed;
3397 }
3398
3399 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
3400 loff_t len);
3401
3402 struct ext4_group_info {
3403 unsigned long bb_state;
3404 #ifdef AGGRESSIVE_CHECK
3405 unsigned long bb_check_counter;
3406 #endif
3407 struct rb_root bb_free_root;
3408 ext4_grpblk_t bb_first_free; /* first free block */
3409 ext4_grpblk_t bb_free; /* total free blocks */
3410 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
3411 int bb_avg_fragment_size_order; /* order of average
3412 fragment in BG */
3413 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
3414 ext4_group_t bb_group; /* Group number */
3415 struct list_head bb_prealloc_list;
3416 #ifdef DOUBLE_CHECK
3417 void *bb_bitmap;
3418 #endif
3419 struct rw_semaphore alloc_sem;
3420 struct list_head bb_avg_fragment_size_node;
3421 struct list_head bb_largest_free_order_node;
3422 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
3423 * regions, index is order.
3424 * bb_counters[3] = 5 means
3425 * 5 free 8-block regions. */
3426 };
3427
3428 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
3429 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
3430 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
3431 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
3432 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \
3433 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT)
3434 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \
3435 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT)
3436 #define EXT4_GROUP_INFO_BBITMAP_READ_BIT 4
3437
3438 #define EXT4_MB_GRP_NEED_INIT(grp) \
3439 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
3440 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
3441 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3442 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
3443 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3444
3445 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
3446 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3447 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
3448 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3449 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
3450 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3451 #define EXT4_MB_GRP_TEST_AND_SET_READ(grp) \
3452 (test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state)))
3453
3454 #define EXT4_MAX_CONTENTION 8
3455 #define EXT4_CONTENTION_THRESHOLD 2
3456
ext4_group_lock_ptr(struct super_block * sb,ext4_group_t group)3457 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
3458 ext4_group_t group)
3459 {
3460 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
3461 }
3462
3463 /*
3464 * Returns true if the filesystem is busy enough that attempts to
3465 * access the block group locks has run into contention.
3466 */
ext4_fs_is_busy(struct ext4_sb_info * sbi)3467 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
3468 {
3469 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
3470 }
3471
ext4_lock_group(struct super_block * sb,ext4_group_t group)3472 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
3473 {
3474 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
3475 if (spin_trylock(lock))
3476 /*
3477 * We're able to grab the lock right away, so drop the
3478 * lock contention counter.
3479 */
3480 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
3481 else {
3482 /*
3483 * The lock is busy, so bump the contention counter,
3484 * and then wait on the spin lock.
3485 */
3486 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
3487 EXT4_MAX_CONTENTION);
3488 spin_lock(lock);
3489 }
3490 }
3491
ext4_unlock_group(struct super_block * sb,ext4_group_t group)3492 static inline void ext4_unlock_group(struct super_block *sb,
3493 ext4_group_t group)
3494 {
3495 spin_unlock(ext4_group_lock_ptr(sb, group));
3496 }
3497
3498 #ifdef CONFIG_QUOTA
ext4_quota_capable(struct super_block * sb)3499 static inline bool ext4_quota_capable(struct super_block *sb)
3500 {
3501 return (test_opt(sb, QUOTA) || ext4_has_feature_quota(sb));
3502 }
3503
ext4_is_quota_journalled(struct super_block * sb)3504 static inline bool ext4_is_quota_journalled(struct super_block *sb)
3505 {
3506 struct ext4_sb_info *sbi = EXT4_SB(sb);
3507
3508 return (ext4_has_feature_quota(sb) ||
3509 sbi->s_qf_names[USRQUOTA] || sbi->s_qf_names[GRPQUOTA]);
3510 }
3511 int ext4_enable_quotas(struct super_block *sb);
3512 #endif
3513
3514 /*
3515 * Block validity checking
3516 */
3517 #define ext4_check_indirect_blockref(inode, bh) \
3518 ext4_check_blockref(__func__, __LINE__, inode, \
3519 (__le32 *)(bh)->b_data, \
3520 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
3521
3522 #define ext4_ind_check_inode(inode) \
3523 ext4_check_blockref(__func__, __LINE__, inode, \
3524 EXT4_I(inode)->i_data, \
3525 EXT4_NDIR_BLOCKS)
3526
3527 /*
3528 * Inodes and files operations
3529 */
3530
3531 /* dir.c */
3532 extern const struct file_operations ext4_dir_operations;
3533
3534 /* file.c */
3535 extern const struct inode_operations ext4_file_inode_operations;
3536 extern const struct file_operations ext4_file_operations;
3537 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
3538
3539 /* inline.c */
3540 extern int ext4_get_max_inline_size(struct inode *inode);
3541 extern int ext4_find_inline_data_nolock(struct inode *inode);
3542 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
3543 unsigned int len);
3544 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
3545
3546 extern int ext4_readpage_inline(struct inode *inode, struct page *page);
3547 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3548 struct inode *inode,
3549 loff_t pos, unsigned len,
3550 struct page **pagep);
3551 extern int ext4_write_inline_data_end(struct inode *inode,
3552 loff_t pos, unsigned len,
3553 unsigned copied,
3554 struct page *page);
3555 extern struct buffer_head *
3556 ext4_journalled_write_inline_data(struct inode *inode,
3557 unsigned len,
3558 struct page *page);
3559 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
3560 struct inode *inode,
3561 loff_t pos, unsigned len,
3562 struct page **pagep,
3563 void **fsdata);
3564 extern int ext4_try_add_inline_entry(handle_t *handle,
3565 struct ext4_filename *fname,
3566 struct inode *dir, struct inode *inode);
3567 extern int ext4_try_create_inline_dir(handle_t *handle,
3568 struct inode *parent,
3569 struct inode *inode);
3570 extern int ext4_read_inline_dir(struct file *filp,
3571 struct dir_context *ctx,
3572 int *has_inline_data);
3573 extern int ext4_inlinedir_to_tree(struct file *dir_file,
3574 struct inode *dir, ext4_lblk_t block,
3575 struct dx_hash_info *hinfo,
3576 __u32 start_hash, __u32 start_minor_hash,
3577 int *has_inline_data);
3578 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3579 struct ext4_filename *fname,
3580 struct ext4_dir_entry_2 **res_dir,
3581 int *has_inline_data);
3582 extern int ext4_delete_inline_entry(handle_t *handle,
3583 struct inode *dir,
3584 struct ext4_dir_entry_2 *de_del,
3585 struct buffer_head *bh,
3586 int *has_inline_data);
3587 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3588 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3589 struct ext4_dir_entry_2 **parent_de,
3590 int *retval);
3591 extern void *ext4_read_inline_link(struct inode *inode);
3592
3593 struct iomap;
3594 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap);
3595
3596 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3597
3598 extern int ext4_convert_inline_data(struct inode *inode);
3599
ext4_has_inline_data(struct inode * inode)3600 static inline int ext4_has_inline_data(struct inode *inode)
3601 {
3602 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3603 EXT4_I(inode)->i_inline_off;
3604 }
3605
3606 /* namei.c */
3607 extern const struct inode_operations ext4_dir_inode_operations;
3608 extern const struct inode_operations ext4_special_inode_operations;
3609 extern struct dentry *ext4_get_parent(struct dentry *child);
3610 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3611 struct ext4_dir_entry_2 *de,
3612 int blocksize, int csum_size,
3613 unsigned int parent_ino, int dotdot_real_len);
3614 extern void ext4_initialize_dirent_tail(struct buffer_head *bh,
3615 unsigned int blocksize);
3616 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode,
3617 struct buffer_head *bh);
3618 extern int __ext4_unlink(struct inode *dir, const struct qstr *d_name,
3619 struct inode *inode, struct dentry *dentry);
3620 extern int __ext4_link(struct inode *dir, struct inode *inode,
3621 struct dentry *dentry);
3622
3623 #define S_SHIFT 12
3624 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = {
3625 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
3626 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
3627 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
3628 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
3629 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
3630 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
3631 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
3632 };
3633
ext4_set_de_type(struct super_block * sb,struct ext4_dir_entry_2 * de,umode_t mode)3634 static inline void ext4_set_de_type(struct super_block *sb,
3635 struct ext4_dir_entry_2 *de,
3636 umode_t mode) {
3637 if (ext4_has_feature_filetype(sb))
3638 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3639 }
3640
3641 /* readpages.c */
3642 extern int ext4_mpage_readpages(struct inode *inode,
3643 struct readahead_control *rac, struct page *page);
3644 extern int __init ext4_init_post_read_processing(void);
3645 extern void ext4_exit_post_read_processing(void);
3646
3647 /* symlink.c */
3648 extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3649 extern const struct inode_operations ext4_symlink_inode_operations;
3650 extern const struct inode_operations ext4_fast_symlink_inode_operations;
3651
3652 /* sysfs.c */
3653 extern void ext4_notify_error_sysfs(struct ext4_sb_info *sbi);
3654 extern int ext4_register_sysfs(struct super_block *sb);
3655 extern void ext4_unregister_sysfs(struct super_block *sb);
3656 extern int __init ext4_init_sysfs(void);
3657 extern void ext4_exit_sysfs(void);
3658
3659 /* block_validity */
3660 extern void ext4_release_system_zone(struct super_block *sb);
3661 extern int ext4_setup_system_zone(struct super_block *sb);
3662 extern int __init ext4_init_system_zone(void);
3663 extern void ext4_exit_system_zone(void);
3664 extern int ext4_inode_block_valid(struct inode *inode,
3665 ext4_fsblk_t start_blk,
3666 unsigned int count);
3667 extern int ext4_check_blockref(const char *, unsigned int,
3668 struct inode *, __le32 *, unsigned int);
3669 extern int ext4_sb_block_valid(struct super_block *sb, struct inode *inode,
3670 ext4_fsblk_t start_blk, unsigned int count);
3671
3672
3673 /* extents.c */
3674 struct ext4_ext_path;
3675 struct ext4_extent;
3676
3677 /*
3678 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3679 * __le32.
3680 */
3681 #define EXT_MAX_BLOCKS 0xffffffff
3682
3683 extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode);
3684 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3685 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3686 struct ext4_map_blocks *map, int flags);
3687 extern int ext4_ext_truncate(handle_t *, struct inode *);
3688 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3689 ext4_lblk_t end);
3690 extern void ext4_ext_init(struct super_block *);
3691 extern void ext4_ext_release(struct super_block *);
3692 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3693 loff_t len);
3694 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3695 loff_t offset, ssize_t len);
3696 extern int ext4_convert_unwritten_io_end_vec(handle_t *handle,
3697 ext4_io_end_t *io_end);
3698 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3699 struct ext4_map_blocks *map, int flags);
3700 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3701 int num,
3702 struct ext4_ext_path *path);
3703 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3704 struct ext4_ext_path **,
3705 struct ext4_extent *, int);
3706 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3707 struct ext4_ext_path **,
3708 int flags);
3709 extern void ext4_free_ext_path(struct ext4_ext_path *);
3710 extern int ext4_ext_check_inode(struct inode *inode);
3711 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3712 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3713 __u64 start, __u64 len);
3714 extern int ext4_get_es_cache(struct inode *inode,
3715 struct fiemap_extent_info *fieinfo,
3716 __u64 start, __u64 len);
3717 extern int ext4_ext_precache(struct inode *inode);
3718 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3719 struct inode *inode2, ext4_lblk_t lblk1,
3720 ext4_lblk_t lblk2, ext4_lblk_t count,
3721 int mark_unwritten,int *err);
3722 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu);
3723 extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode,
3724 int check_cred, int restart_cred,
3725 int revoke_cred);
3726 extern void ext4_ext_replay_shrink_inode(struct inode *inode, ext4_lblk_t end);
3727 extern int ext4_ext_replay_set_iblocks(struct inode *inode);
3728 extern int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start,
3729 int len, int unwritten, ext4_fsblk_t pblk);
3730 extern int ext4_ext_clear_bb(struct inode *inode);
3731
3732
3733 /* move_extent.c */
3734 extern void ext4_double_down_write_data_sem(struct inode *first,
3735 struct inode *second);
3736 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3737 struct inode *donor_inode);
3738 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3739 __u64 start_orig, __u64 start_donor,
3740 __u64 len, __u64 *moved_len);
3741
3742 /* page-io.c */
3743 extern int __init ext4_init_pageio(void);
3744 extern void ext4_exit_pageio(void);
3745 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3746 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3747 extern int ext4_put_io_end(ext4_io_end_t *io_end);
3748 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3749 extern void ext4_io_submit_init(struct ext4_io_submit *io,
3750 struct writeback_control *wbc);
3751 extern void ext4_end_io_rsv_work(struct work_struct *work);
3752 extern void ext4_io_submit(struct ext4_io_submit *io);
3753 extern int ext4_bio_write_page(struct ext4_io_submit *io,
3754 struct page *page,
3755 int len,
3756 bool keep_towrite);
3757 extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end);
3758 extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end);
3759
3760 /* mmp.c */
3761 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3762
3763 /* mmp.c */
3764 extern void ext4_stop_mmpd(struct ext4_sb_info *sbi);
3765
3766 /* verity.c */
3767 extern const struct fsverity_operations ext4_verityops;
3768
3769 /* orphan.c */
3770 extern int ext4_orphan_add(handle_t *, struct inode *);
3771 extern int ext4_orphan_del(handle_t *, struct inode *);
3772 extern void ext4_orphan_cleanup(struct super_block *sb,
3773 struct ext4_super_block *es);
3774 extern void ext4_release_orphan_info(struct super_block *sb);
3775 extern int ext4_init_orphan_info(struct super_block *sb);
3776 extern int ext4_orphan_file_empty(struct super_block *sb);
3777 extern void ext4_orphan_file_block_trigger(
3778 struct jbd2_buffer_trigger_type *triggers,
3779 struct buffer_head *bh,
3780 void *data, size_t size);
3781
3782 /*
3783 * Add new method to test whether block and inode bitmaps are properly
3784 * initialized. With uninit_bg reading the block from disk is not enough
3785 * to mark the bitmap uptodate. We need to also zero-out the bitmap
3786 */
3787 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3788
bitmap_uptodate(struct buffer_head * bh)3789 static inline int bitmap_uptodate(struct buffer_head *bh)
3790 {
3791 return (buffer_uptodate(bh) &&
3792 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3793 }
set_bitmap_uptodate(struct buffer_head * bh)3794 static inline void set_bitmap_uptodate(struct buffer_head *bh)
3795 {
3796 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3797 }
3798
3799 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
3800
3801 /* For ioend & aio unwritten conversion wait queues */
3802 #define EXT4_WQ_HASH_SZ 37
3803 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
3804 EXT4_WQ_HASH_SZ])
3805 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3806
3807 extern int ext4_resize_begin(struct super_block *sb);
3808 extern int ext4_resize_end(struct super_block *sb, bool update_backups);
3809
ext4_set_io_unwritten_flag(struct inode * inode,struct ext4_io_end * io_end)3810 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
3811 struct ext4_io_end *io_end)
3812 {
3813 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
3814 io_end->flag |= EXT4_IO_END_UNWRITTEN;
3815 atomic_inc(&EXT4_I(inode)->i_unwritten);
3816 }
3817 }
3818
ext4_clear_io_unwritten_flag(ext4_io_end_t * io_end)3819 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3820 {
3821 struct inode *inode = io_end->inode;
3822
3823 if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
3824 io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3825 /* Wake up anyone waiting on unwritten extent conversion */
3826 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
3827 wake_up_all(ext4_ioend_wq(inode));
3828 }
3829 }
3830
3831 extern const struct iomap_ops ext4_iomap_ops;
3832 extern const struct iomap_ops ext4_iomap_overwrite_ops;
3833 extern const struct iomap_ops ext4_iomap_report_ops;
3834
ext4_buffer_uptodate(struct buffer_head * bh)3835 static inline int ext4_buffer_uptodate(struct buffer_head *bh)
3836 {
3837 /*
3838 * If the buffer has the write error flag, we have failed
3839 * to write out data in the block. In this case, we don't
3840 * have to read the block because we may read the old data
3841 * successfully.
3842 */
3843 if (buffer_write_io_error(bh))
3844 set_buffer_uptodate(bh);
3845 return buffer_uptodate(bh);
3846 }
3847
3848 #endif /* __KERNEL__ */
3849
3850 #define EFSBADCRC EBADMSG /* Bad CRC detected */
3851 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
3852
3853 #endif /* _EXT4_H */
3854