]> git.proxmox.com Git - mirror_ubuntu-jammy-kernel.git/blob - fs/ext4/ext4.h
Merge tag 'ext4_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso...
[mirror_ubuntu-jammy-kernel.git] / fs / ext4 / ext4.h
1 // SPDX-License-Identifier: GPL-2.0
2 /*
3 * ext4.h
4 *
5 * Copyright (C) 1992, 1993, 1994, 1995
6 * Remy Card (card@masi.ibp.fr)
7 * Laboratoire MASI - Institut Blaise Pascal
8 * Universite Pierre et Marie Curie (Paris VI)
9 *
10 * from
11 *
12 * linux/include/linux/minix_fs.h
13 *
14 * Copyright (C) 1991, 1992 Linus Torvalds
15 */
16
17 #ifndef _EXT4_H
18 #define _EXT4_H
19
20 #include <linux/types.h>
21 #include <linux/blkdev.h>
22 #include <linux/magic.h>
23 #include <linux/jbd2.h>
24 #include <linux/quota.h>
25 #include <linux/rwsem.h>
26 #include <linux/rbtree.h>
27 #include <linux/seqlock.h>
28 #include <linux/mutex.h>
29 #include <linux/timer.h>
30 #include <linux/wait.h>
31 #include <linux/sched/signal.h>
32 #include <linux/blockgroup_lock.h>
33 #include <linux/percpu_counter.h>
34 #include <linux/ratelimit.h>
35 #include <crypto/hash.h>
36 #include <linux/falloc.h>
37 #include <linux/percpu-rwsem.h>
38 #include <linux/fiemap.h>
39 #ifdef __KERNEL__
40 #include <linux/compat.h>
41 #endif
42
43 #include <linux/fscrypt.h>
44 #include <linux/fsverity.h>
45
46 #include <linux/compiler.h>
47
48 /*
49 * The fourth extended filesystem constants/structures
50 */
51
52 /*
53 * with AGGRESSIVE_CHECK allocator runs consistency checks over
54 * structures. these checks slow things down a lot
55 */
56 #define AGGRESSIVE_CHECK__
57
58 /*
59 * with DOUBLE_CHECK defined mballoc creates persistent in-core
60 * bitmaps, maintains and uses them to check for double allocations
61 */
62 #define DOUBLE_CHECK__
63
64 /*
65 * Define EXT4FS_DEBUG to produce debug messages
66 */
67 #undef EXT4FS_DEBUG
68
69 /*
70 * Debug code
71 */
72 #ifdef EXT4FS_DEBUG
73 #define ext4_debug(f, a...) \
74 do { \
75 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
76 __FILE__, __LINE__, __func__); \
77 printk(KERN_DEBUG f, ## a); \
78 } while (0)
79 #else
80 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
81 #endif
82
83 /*
84 * Turn on EXT_DEBUG to enable ext4_ext_show_path/leaf/move in extents.c
85 */
86 #define EXT_DEBUG__
87
88 /*
89 * Dynamic printk for controlled extents debugging.
90 */
91 #ifdef CONFIG_EXT4_DEBUG
92 #define ext_debug(ino, fmt, ...) \
93 pr_debug("[%s/%d] EXT4-fs (%s): ino %lu: (%s, %d): %s:" fmt, \
94 current->comm, task_pid_nr(current), \
95 ino->i_sb->s_id, ino->i_ino, __FILE__, __LINE__, \
96 __func__, ##__VA_ARGS__)
97 #else
98 #define ext_debug(ino, fmt, ...) no_printk(fmt, ##__VA_ARGS__)
99 #endif
100
101 #define ASSERT(assert) \
102 do { \
103 if (unlikely(!(assert))) { \
104 printk(KERN_EMERG \
105 "Assertion failure in %s() at %s:%d: '%s'\n", \
106 __func__, __FILE__, __LINE__, #assert); \
107 BUG(); \
108 } \
109 } while (0)
110
111 /* data type for block offset of block group */
112 typedef int ext4_grpblk_t;
113
114 /* data type for filesystem-wide blocks number */
115 typedef unsigned long long ext4_fsblk_t;
116
117 /* data type for file logical block number */
118 typedef __u32 ext4_lblk_t;
119
120 /* data type for block group number */
121 typedef unsigned int ext4_group_t;
122
123 enum SHIFT_DIRECTION {
124 SHIFT_LEFT = 0,
125 SHIFT_RIGHT,
126 };
127
128 /*
129 * Flags used in mballoc's allocation_context flags field.
130 *
131 * Also used to show what's going on for debugging purposes when the
132 * flag field is exported via the traceport interface
133 */
134
135 /* prefer goal again. length */
136 #define EXT4_MB_HINT_MERGE 0x0001
137 /* blocks already reserved */
138 #define EXT4_MB_HINT_RESERVED 0x0002
139 /* metadata is being allocated */
140 #define EXT4_MB_HINT_METADATA 0x0004
141 /* first blocks in the file */
142 #define EXT4_MB_HINT_FIRST 0x0008
143 /* search for the best chunk */
144 #define EXT4_MB_HINT_BEST 0x0010
145 /* data is being allocated */
146 #define EXT4_MB_HINT_DATA 0x0020
147 /* don't preallocate (for tails) */
148 #define EXT4_MB_HINT_NOPREALLOC 0x0040
149 /* allocate for locality group */
150 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080
151 /* allocate goal blocks or none */
152 #define EXT4_MB_HINT_GOAL_ONLY 0x0100
153 /* goal is meaningful */
154 #define EXT4_MB_HINT_TRY_GOAL 0x0200
155 /* blocks already pre-reserved by delayed allocation */
156 #define EXT4_MB_DELALLOC_RESERVED 0x0400
157 /* We are doing stream allocation */
158 #define EXT4_MB_STREAM_ALLOC 0x0800
159 /* Use reserved root blocks if needed */
160 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000
161 /* Use blocks from reserved pool */
162 #define EXT4_MB_USE_RESERVED 0x2000
163 /* Do strict check for free blocks while retrying block allocation */
164 #define EXT4_MB_STRICT_CHECK 0x4000
165 /* Large fragment size list lookup succeeded at least once for cr = 0 */
166 #define EXT4_MB_CR0_OPTIMIZED 0x8000
167 /* Avg fragment size rb tree lookup succeeded at least once for cr = 1 */
168 #define EXT4_MB_CR1_OPTIMIZED 0x00010000
169 /* Perform linear traversal for one group */
170 #define EXT4_MB_SEARCH_NEXT_LINEAR 0x00020000
171 struct ext4_allocation_request {
172 /* target inode for block we're allocating */
173 struct inode *inode;
174 /* how many blocks we want to allocate */
175 unsigned int len;
176 /* logical block in target inode */
177 ext4_lblk_t logical;
178 /* the closest logical allocated block to the left */
179 ext4_lblk_t lleft;
180 /* the closest logical allocated block to the right */
181 ext4_lblk_t lright;
182 /* phys. target (a hint) */
183 ext4_fsblk_t goal;
184 /* phys. block for the closest logical allocated block to the left */
185 ext4_fsblk_t pleft;
186 /* phys. block for the closest logical allocated block to the right */
187 ext4_fsblk_t pright;
188 /* flags. see above EXT4_MB_HINT_* */
189 unsigned int flags;
190 };
191
192 /*
193 * Logical to physical block mapping, used by ext4_map_blocks()
194 *
195 * This structure is used to pass requests into ext4_map_blocks() as
196 * well as to store the information returned by ext4_map_blocks(). It
197 * takes less room on the stack than a struct buffer_head.
198 */
199 #define EXT4_MAP_NEW BIT(BH_New)
200 #define EXT4_MAP_MAPPED BIT(BH_Mapped)
201 #define EXT4_MAP_UNWRITTEN BIT(BH_Unwritten)
202 #define EXT4_MAP_BOUNDARY BIT(BH_Boundary)
203 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
204 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
205
206 struct ext4_map_blocks {
207 ext4_fsblk_t m_pblk;
208 ext4_lblk_t m_lblk;
209 unsigned int m_len;
210 unsigned int m_flags;
211 };
212
213 /*
214 * Block validity checking, system zone rbtree.
215 */
216 struct ext4_system_blocks {
217 struct rb_root root;
218 struct rcu_head rcu;
219 };
220
221 /*
222 * Flags for ext4_io_end->flags
223 */
224 #define EXT4_IO_END_UNWRITTEN 0x0001
225
226 struct ext4_io_end_vec {
227 struct list_head list; /* list of io_end_vec */
228 loff_t offset; /* offset in the file */
229 ssize_t size; /* size of the extent */
230 };
231
232 /*
233 * For converting unwritten extents on a work queue. 'handle' is used for
234 * buffered writeback.
235 */
236 typedef struct ext4_io_end {
237 struct list_head list; /* per-file finished IO list */
238 handle_t *handle; /* handle reserved for extent
239 * conversion */
240 struct inode *inode; /* file being written to */
241 struct bio *bio; /* Linked list of completed
242 * bios covering the extent */
243 unsigned int flag; /* unwritten or not */
244 atomic_t count; /* reference counter */
245 struct list_head list_vec; /* list of ext4_io_end_vec */
246 } ext4_io_end_t;
247
248 struct ext4_io_submit {
249 struct writeback_control *io_wbc;
250 struct bio *io_bio;
251 ext4_io_end_t *io_end;
252 sector_t io_next_block;
253 };
254
255 /*
256 * Special inodes numbers
257 */
258 #define EXT4_BAD_INO 1 /* Bad blocks inode */
259 #define EXT4_ROOT_INO 2 /* Root inode */
260 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */
261 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
262 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
263 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
264 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
265 #define EXT4_JOURNAL_INO 8 /* Journal inode */
266
267 /* First non-reserved inode for old ext4 filesystems */
268 #define EXT4_GOOD_OLD_FIRST_INO 11
269
270 /*
271 * Maximal count of links to a file
272 */
273 #define EXT4_LINK_MAX 65000
274
275 /*
276 * Macro-instructions used to manage several block sizes
277 */
278 #define EXT4_MIN_BLOCK_SIZE 1024
279 #define EXT4_MAX_BLOCK_SIZE 65536
280 #define EXT4_MIN_BLOCK_LOG_SIZE 10
281 #define EXT4_MAX_BLOCK_LOG_SIZE 16
282 #define EXT4_MAX_CLUSTER_LOG_SIZE 30
283 #ifdef __KERNEL__
284 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
285 #else
286 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
287 #endif
288 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
289 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
290 EXT4_SB(s)->s_cluster_bits)
291 #ifdef __KERNEL__
292 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
293 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
294 #else
295 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
296 #endif
297 #ifdef __KERNEL__
298 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
299 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
300 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
301 #else
302 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
303 EXT4_GOOD_OLD_INODE_SIZE : \
304 (s)->s_inode_size)
305 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
306 EXT4_GOOD_OLD_FIRST_INO : \
307 (s)->s_first_ino)
308 #endif
309 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
310 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \
311 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \
312 blkbits))
313
314 /* Translate a block number to a cluster number */
315 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
316 /* Translate a cluster number to a block number */
317 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
318 /* Translate # of blks to # of clusters */
319 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
320 (sbi)->s_cluster_bits)
321 /* Mask out the low bits to get the starting block of the cluster */
322 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
323 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
324 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
325 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
326 /* Fill in the low bits to get the last block of the cluster */
327 #define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) | \
328 ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1))
329 /* Get the cluster offset */
330 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
331 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
332 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
333 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
334
335 /*
336 * Structure of a blocks group descriptor
337 */
338 struct ext4_group_desc
339 {
340 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
341 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
342 __le32 bg_inode_table_lo; /* Inodes table block */
343 __le16 bg_free_blocks_count_lo;/* Free blocks count */
344 __le16 bg_free_inodes_count_lo;/* Free inodes count */
345 __le16 bg_used_dirs_count_lo; /* Directories count */
346 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
347 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
348 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
349 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
350 __le16 bg_itable_unused_lo; /* Unused inodes count */
351 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
352 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
353 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
354 __le32 bg_inode_table_hi; /* Inodes table block MSB */
355 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
356 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
357 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
358 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
359 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
360 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
361 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
362 __u32 bg_reserved;
363 };
364
365 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
366 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
367 sizeof(__le16))
368 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
369 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
370 sizeof(__le16))
371
372 /*
373 * Structure of a flex block group info
374 */
375
376 struct flex_groups {
377 atomic64_t free_clusters;
378 atomic_t free_inodes;
379 atomic_t used_dirs;
380 };
381
382 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
383 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
384 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
385
386 /*
387 * Macro-instructions used to manage group descriptors
388 */
389 #define EXT4_MIN_DESC_SIZE 32
390 #define EXT4_MIN_DESC_SIZE_64BIT 64
391 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
392 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
393 #ifdef __KERNEL__
394 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
395 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
396 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
397 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
398 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
399 #else
400 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
401 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
402 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
403 #endif
404
405 /*
406 * Constants relative to the data blocks
407 */
408 #define EXT4_NDIR_BLOCKS 12
409 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
410 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
411 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
412 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
413
414 /*
415 * Inode flags
416 */
417 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
418 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
419 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
420 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
421 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
422 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
423 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
424 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
425 /* Reserved for compression usage... */
426 #define EXT4_DIRTY_FL 0x00000100
427 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
428 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
429 /* nb: was previously EXT2_ECOMPR_FL */
430 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */
431 /* End compression flags --- maybe not all used */
432 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
433 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
434 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
435 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
436 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
437 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
438 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
439 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
440 #define EXT4_VERITY_FL 0x00100000 /* Verity protected inode */
441 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
442 /* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */
443
444 #define EXT4_DAX_FL 0x02000000 /* Inode is DAX */
445
446 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
447 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
448 #define EXT4_CASEFOLD_FL 0x40000000 /* Casefolded directory */
449 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
450
451 /* User modifiable flags */
452 #define EXT4_FL_USER_MODIFIABLE (EXT4_SECRM_FL | \
453 EXT4_UNRM_FL | \
454 EXT4_COMPR_FL | \
455 EXT4_SYNC_FL | \
456 EXT4_IMMUTABLE_FL | \
457 EXT4_APPEND_FL | \
458 EXT4_NODUMP_FL | \
459 EXT4_NOATIME_FL | \
460 EXT4_JOURNAL_DATA_FL | \
461 EXT4_NOTAIL_FL | \
462 EXT4_DIRSYNC_FL | \
463 EXT4_TOPDIR_FL | \
464 EXT4_EXTENTS_FL | \
465 0x00400000 /* EXT4_EOFBLOCKS_FL */ | \
466 EXT4_DAX_FL | \
467 EXT4_PROJINHERIT_FL | \
468 EXT4_CASEFOLD_FL)
469
470 /* User visible flags */
471 #define EXT4_FL_USER_VISIBLE (EXT4_FL_USER_MODIFIABLE | \
472 EXT4_DIRTY_FL | \
473 EXT4_COMPRBLK_FL | \
474 EXT4_NOCOMPR_FL | \
475 EXT4_ENCRYPT_FL | \
476 EXT4_INDEX_FL | \
477 EXT4_VERITY_FL | \
478 EXT4_INLINE_DATA_FL)
479
480 /* Flags that should be inherited by new inodes from their parent. */
481 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
482 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
483 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
484 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
485 EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL |\
486 EXT4_DAX_FL)
487
488 /* Flags that are appropriate for regular files (all but dir-specific ones). */
489 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\
490 EXT4_PROJINHERIT_FL))
491
492 /* Flags that are appropriate for non-directories/regular files. */
493 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
494
495 /* The only flags that should be swapped */
496 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL)
497
498 /* Flags which are mutually exclusive to DAX */
499 #define EXT4_DAX_MUT_EXCL (EXT4_VERITY_FL | EXT4_ENCRYPT_FL |\
500 EXT4_JOURNAL_DATA_FL | EXT4_INLINE_DATA_FL)
501
502 /* Mask out flags that are inappropriate for the given type of inode. */
503 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
504 {
505 if (S_ISDIR(mode))
506 return flags;
507 else if (S_ISREG(mode))
508 return flags & EXT4_REG_FLMASK;
509 else
510 return flags & EXT4_OTHER_FLMASK;
511 }
512
513 /*
514 * Inode flags used for atomic set/get
515 */
516 enum {
517 EXT4_INODE_SECRM = 0, /* Secure deletion */
518 EXT4_INODE_UNRM = 1, /* Undelete */
519 EXT4_INODE_COMPR = 2, /* Compress file */
520 EXT4_INODE_SYNC = 3, /* Synchronous updates */
521 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
522 EXT4_INODE_APPEND = 5, /* writes to file may only append */
523 EXT4_INODE_NODUMP = 6, /* do not dump file */
524 EXT4_INODE_NOATIME = 7, /* do not update atime */
525 /* Reserved for compression usage... */
526 EXT4_INODE_DIRTY = 8,
527 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
528 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
529 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */
530 /* End compression flags --- maybe not all used */
531 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
532 EXT4_INODE_IMAGIC = 13, /* AFS directory */
533 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
534 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
535 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
536 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
537 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
538 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
539 EXT4_INODE_VERITY = 20, /* Verity protected inode */
540 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
541 /* 22 was formerly EXT4_INODE_EOFBLOCKS */
542 EXT4_INODE_DAX = 25, /* Inode is DAX */
543 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
544 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */
545 EXT4_INODE_CASEFOLD = 30, /* Casefolded directory */
546 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
547 };
548
549 /*
550 * Since it's pretty easy to mix up bit numbers and hex values, we use a
551 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
552 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
553 * any extra space in the compiled kernel image, otherwise, the build will fail.
554 * It's important that these values are the same, since we are using
555 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
556 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
557 * values found in ext2, ext3 and ext4 filesystems, and of course the values
558 * defined in e2fsprogs.
559 *
560 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
561 */
562 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
563 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
564
565 static inline void ext4_check_flag_values(void)
566 {
567 CHECK_FLAG_VALUE(SECRM);
568 CHECK_FLAG_VALUE(UNRM);
569 CHECK_FLAG_VALUE(COMPR);
570 CHECK_FLAG_VALUE(SYNC);
571 CHECK_FLAG_VALUE(IMMUTABLE);
572 CHECK_FLAG_VALUE(APPEND);
573 CHECK_FLAG_VALUE(NODUMP);
574 CHECK_FLAG_VALUE(NOATIME);
575 CHECK_FLAG_VALUE(DIRTY);
576 CHECK_FLAG_VALUE(COMPRBLK);
577 CHECK_FLAG_VALUE(NOCOMPR);
578 CHECK_FLAG_VALUE(ENCRYPT);
579 CHECK_FLAG_VALUE(INDEX);
580 CHECK_FLAG_VALUE(IMAGIC);
581 CHECK_FLAG_VALUE(JOURNAL_DATA);
582 CHECK_FLAG_VALUE(NOTAIL);
583 CHECK_FLAG_VALUE(DIRSYNC);
584 CHECK_FLAG_VALUE(TOPDIR);
585 CHECK_FLAG_VALUE(HUGE_FILE);
586 CHECK_FLAG_VALUE(EXTENTS);
587 CHECK_FLAG_VALUE(VERITY);
588 CHECK_FLAG_VALUE(EA_INODE);
589 CHECK_FLAG_VALUE(INLINE_DATA);
590 CHECK_FLAG_VALUE(PROJINHERIT);
591 CHECK_FLAG_VALUE(CASEFOLD);
592 CHECK_FLAG_VALUE(RESERVED);
593 }
594
595 /* Used to pass group descriptor data when online resize is done */
596 struct ext4_new_group_input {
597 __u32 group; /* Group number for this data */
598 __u64 block_bitmap; /* Absolute block number of block bitmap */
599 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
600 __u64 inode_table; /* Absolute block number of inode table start */
601 __u32 blocks_count; /* Total number of blocks in this group */
602 __u16 reserved_blocks; /* Number of reserved blocks in this group */
603 __u16 unused;
604 };
605
606 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
607 struct compat_ext4_new_group_input {
608 u32 group;
609 compat_u64 block_bitmap;
610 compat_u64 inode_bitmap;
611 compat_u64 inode_table;
612 u32 blocks_count;
613 u16 reserved_blocks;
614 u16 unused;
615 };
616 #endif
617
618 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
619 struct ext4_new_group_data {
620 __u32 group;
621 __u64 block_bitmap;
622 __u64 inode_bitmap;
623 __u64 inode_table;
624 __u32 blocks_count;
625 __u16 reserved_blocks;
626 __u16 mdata_blocks;
627 __u32 free_clusters_count;
628 };
629
630 /* Indexes used to index group tables in ext4_new_group_data */
631 enum {
632 BLOCK_BITMAP = 0, /* block bitmap */
633 INODE_BITMAP, /* inode bitmap */
634 INODE_TABLE, /* inode tables */
635 GROUP_TABLE_COUNT,
636 };
637
638 /*
639 * Flags used by ext4_map_blocks()
640 */
641 /* Allocate any needed blocks and/or convert an unwritten
642 extent to be an initialized ext4 */
643 #define EXT4_GET_BLOCKS_CREATE 0x0001
644 /* Request the creation of an unwritten extent */
645 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002
646 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\
647 EXT4_GET_BLOCKS_CREATE)
648 /* Caller is from the delayed allocation writeout path
649 * finally doing the actual allocation of delayed blocks */
650 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
651 /* caller is from the direct IO path, request to creation of an
652 unwritten extents if not allocated, split the unwritten
653 extent if blocks has been preallocated already*/
654 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
655 #define EXT4_GET_BLOCKS_CONVERT 0x0010
656 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
657 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
658 /* Convert extent to initialized after IO complete */
659 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
660 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
661 /* Eventual metadata allocation (due to growing extent tree)
662 * should not fail, so try to use reserved blocks for that.*/
663 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
664 /* Don't normalize allocation size (used for fallocate) */
665 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
666 /* Convert written extents to unwritten */
667 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100
668 /* Write zeros to newly created written extents */
669 #define EXT4_GET_BLOCKS_ZERO 0x0200
670 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\
671 EXT4_GET_BLOCKS_ZERO)
672 /* Caller will submit data before dropping transaction handle. This
673 * allows jbd2 to avoid submitting data before commit. */
674 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400
675
676 /*
677 * The bit position of these flags must not overlap with any of the
678 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(),
679 * read_extent_tree_block(), ext4_split_extent_at(),
680 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
681 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
682 * caching the extents when reading from the extent tree while a
683 * truncate or punch hole operation is in progress.
684 */
685 #define EXT4_EX_NOCACHE 0x40000000
686 #define EXT4_EX_FORCE_CACHE 0x20000000
687 #define EXT4_EX_NOFAIL 0x10000000
688
689 /*
690 * Flags used by ext4_free_blocks
691 */
692 #define EXT4_FREE_BLOCKS_METADATA 0x0001
693 #define EXT4_FREE_BLOCKS_FORGET 0x0002
694 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
695 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
696 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
697 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
698 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040
699
700 /*
701 * ioctl commands
702 */
703 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
704 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
705 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
706 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
707 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
708 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
709 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
710 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
711 #define EXT4_IOC_MIGRATE _IO('f', 9)
712 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
713 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
714 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
715 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
716 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
717 #define EXT4_IOC_SWAP_BOOT _IO('f', 17)
718 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18)
719 /* ioctl codes 19--39 are reserved for fscrypt */
720 #define EXT4_IOC_CLEAR_ES_CACHE _IO('f', 40)
721 #define EXT4_IOC_GETSTATE _IOW('f', 41, __u32)
722 #define EXT4_IOC_GET_ES_CACHE _IOWR('f', 42, struct fiemap)
723 #define EXT4_IOC_CHECKPOINT _IOW('f', 43, __u32)
724
725 #define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32)
726
727 /*
728 * Flags for going down operation
729 */
730 #define EXT4_GOING_FLAGS_DEFAULT 0x0 /* going down */
731 #define EXT4_GOING_FLAGS_LOGFLUSH 0x1 /* flush log but not data */
732 #define EXT4_GOING_FLAGS_NOLOGFLUSH 0x2 /* don't flush log nor data */
733
734 /*
735 * Flags returned by EXT4_IOC_GETSTATE
736 *
737 * We only expose to userspace a subset of the state flags in
738 * i_state_flags
739 */
740 #define EXT4_STATE_FLAG_EXT_PRECACHED 0x00000001
741 #define EXT4_STATE_FLAG_NEW 0x00000002
742 #define EXT4_STATE_FLAG_NEWENTRY 0x00000004
743 #define EXT4_STATE_FLAG_DA_ALLOC_CLOSE 0x00000008
744
745 /* flags for ioctl EXT4_IOC_CHECKPOINT */
746 #define EXT4_IOC_CHECKPOINT_FLAG_DISCARD 0x1
747 #define EXT4_IOC_CHECKPOINT_FLAG_ZEROOUT 0x2
748 #define EXT4_IOC_CHECKPOINT_FLAG_DRY_RUN 0x4
749 #define EXT4_IOC_CHECKPOINT_FLAG_VALID (EXT4_IOC_CHECKPOINT_FLAG_DISCARD | \
750 EXT4_IOC_CHECKPOINT_FLAG_ZEROOUT | \
751 EXT4_IOC_CHECKPOINT_FLAG_DRY_RUN)
752
753 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
754 /*
755 * ioctl commands in 32 bit emulation
756 */
757 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
758 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
759 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
760 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
761 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
762 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
763 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
764 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
765 #endif
766
767 /*
768 * Returned by EXT4_IOC_GET_ES_CACHE as an additional possible flag.
769 * It indicates that the entry in extent status cache is for a hole.
770 */
771 #define EXT4_FIEMAP_EXTENT_HOLE 0x08000000
772
773 /* Max physical block we can address w/o extents */
774 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
775
776 /* Max logical block we can support */
777 #define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFE
778
779 /*
780 * Structure of an inode on the disk
781 */
782 struct ext4_inode {
783 __le16 i_mode; /* File mode */
784 __le16 i_uid; /* Low 16 bits of Owner Uid */
785 __le32 i_size_lo; /* Size in bytes */
786 __le32 i_atime; /* Access time */
787 __le32 i_ctime; /* Inode Change time */
788 __le32 i_mtime; /* Modification time */
789 __le32 i_dtime; /* Deletion Time */
790 __le16 i_gid; /* Low 16 bits of Group Id */
791 __le16 i_links_count; /* Links count */
792 __le32 i_blocks_lo; /* Blocks count */
793 __le32 i_flags; /* File flags */
794 union {
795 struct {
796 __le32 l_i_version;
797 } linux1;
798 struct {
799 __u32 h_i_translator;
800 } hurd1;
801 struct {
802 __u32 m_i_reserved1;
803 } masix1;
804 } osd1; /* OS dependent 1 */
805 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
806 __le32 i_generation; /* File version (for NFS) */
807 __le32 i_file_acl_lo; /* File ACL */
808 __le32 i_size_high;
809 __le32 i_obso_faddr; /* Obsoleted fragment address */
810 union {
811 struct {
812 __le16 l_i_blocks_high; /* were l_i_reserved1 */
813 __le16 l_i_file_acl_high;
814 __le16 l_i_uid_high; /* these 2 fields */
815 __le16 l_i_gid_high; /* were reserved2[0] */
816 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
817 __le16 l_i_reserved;
818 } linux2;
819 struct {
820 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
821 __u16 h_i_mode_high;
822 __u16 h_i_uid_high;
823 __u16 h_i_gid_high;
824 __u32 h_i_author;
825 } hurd2;
826 struct {
827 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
828 __le16 m_i_file_acl_high;
829 __u32 m_i_reserved2[2];
830 } masix2;
831 } osd2; /* OS dependent 2 */
832 __le16 i_extra_isize;
833 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
834 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
835 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
836 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
837 __le32 i_crtime; /* File Creation time */
838 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
839 __le32 i_version_hi; /* high 32 bits for 64-bit version */
840 __le32 i_projid; /* Project ID */
841 };
842
843 struct move_extent {
844 __u32 reserved; /* should be zero */
845 __u32 donor_fd; /* donor file descriptor */
846 __u64 orig_start; /* logical start offset in block for orig */
847 __u64 donor_start; /* logical start offset in block for donor */
848 __u64 len; /* block length to be moved */
849 __u64 moved_len; /* moved block length */
850 };
851
852 #define EXT4_EPOCH_BITS 2
853 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
854 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
855
856 /*
857 * Extended fields will fit into an inode if the filesystem was formatted
858 * with large inodes (-I 256 or larger) and there are not currently any EAs
859 * consuming all of the available space. For new inodes we always reserve
860 * enough space for the kernel's known extended fields, but for inodes
861 * created with an old kernel this might not have been the case. None of
862 * the extended inode fields is critical for correct filesystem operation.
863 * This macro checks if a certain field fits in the inode. Note that
864 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
865 */
866 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
867 ((offsetof(typeof(*ext4_inode), field) + \
868 sizeof((ext4_inode)->field)) \
869 <= (EXT4_GOOD_OLD_INODE_SIZE + \
870 (einode)->i_extra_isize)) \
871
872 /*
873 * We use an encoding that preserves the times for extra epoch "00":
874 *
875 * extra msb of adjust for signed
876 * epoch 32-bit 32-bit tv_sec to
877 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
878 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
879 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
880 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
881 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
882 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
883 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
884 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
885 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
886 *
887 * Note that previous versions of the kernel on 64-bit systems would
888 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
889 * 1970. e2fsck will correct this, assuming that it is run on the
890 * affected filesystem before 2242.
891 */
892
893 static inline __le32 ext4_encode_extra_time(struct timespec64 *time)
894 {
895 u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK;
896 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
897 }
898
899 static inline void ext4_decode_extra_time(struct timespec64 *time,
900 __le32 extra)
901 {
902 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK)))
903 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
904 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
905 }
906
907 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
908 do { \
909 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {\
910 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
911 (raw_inode)->xtime ## _extra = \
912 ext4_encode_extra_time(&(inode)->xtime); \
913 } \
914 else \
915 (raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (inode)->xtime.tv_sec, S32_MIN, S32_MAX)); \
916 } while (0)
917
918 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
919 do { \
920 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
921 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
922 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
923 (raw_inode)->xtime ## _extra = \
924 ext4_encode_extra_time(&(einode)->xtime); \
925 } while (0)
926
927 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
928 do { \
929 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
930 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \
931 ext4_decode_extra_time(&(inode)->xtime, \
932 raw_inode->xtime ## _extra); \
933 } \
934 else \
935 (inode)->xtime.tv_nsec = 0; \
936 } while (0)
937
938
939 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
940 do { \
941 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
942 (einode)->xtime.tv_sec = \
943 (signed)le32_to_cpu((raw_inode)->xtime); \
944 else \
945 (einode)->xtime.tv_sec = 0; \
946 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
947 ext4_decode_extra_time(&(einode)->xtime, \
948 raw_inode->xtime ## _extra); \
949 else \
950 (einode)->xtime.tv_nsec = 0; \
951 } while (0)
952
953 #define i_disk_version osd1.linux1.l_i_version
954
955 #if defined(__KERNEL__) || defined(__linux__)
956 #define i_reserved1 osd1.linux1.l_i_reserved1
957 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
958 #define i_blocks_high osd2.linux2.l_i_blocks_high
959 #define i_uid_low i_uid
960 #define i_gid_low i_gid
961 #define i_uid_high osd2.linux2.l_i_uid_high
962 #define i_gid_high osd2.linux2.l_i_gid_high
963 #define i_checksum_lo osd2.linux2.l_i_checksum_lo
964
965 #elif defined(__GNU__)
966
967 #define i_translator osd1.hurd1.h_i_translator
968 #define i_uid_high osd2.hurd2.h_i_uid_high
969 #define i_gid_high osd2.hurd2.h_i_gid_high
970 #define i_author osd2.hurd2.h_i_author
971
972 #elif defined(__masix__)
973
974 #define i_reserved1 osd1.masix1.m_i_reserved1
975 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
976 #define i_reserved2 osd2.masix2.m_i_reserved2
977
978 #endif /* defined(__KERNEL__) || defined(__linux__) */
979
980 #include "extents_status.h"
981 #include "fast_commit.h"
982
983 /*
984 * Lock subclasses for i_data_sem in the ext4_inode_info structure.
985 *
986 * These are needed to avoid lockdep false positives when we need to
987 * allocate blocks to the quota inode during ext4_map_blocks(), while
988 * holding i_data_sem for a normal (non-quota) inode. Since we don't
989 * do quota tracking for the quota inode, this avoids deadlock (as
990 * well as infinite recursion, since it isn't turtles all the way
991 * down...)
992 *
993 * I_DATA_SEM_NORMAL - Used for most inodes
994 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode
995 * where the second inode has larger inode number
996 * than the first
997 * I_DATA_SEM_QUOTA - Used for quota inodes only
998 */
999 enum {
1000 I_DATA_SEM_NORMAL = 0,
1001 I_DATA_SEM_OTHER,
1002 I_DATA_SEM_QUOTA,
1003 };
1004
1005
1006 /*
1007 * fourth extended file system inode data in memory
1008 */
1009 struct ext4_inode_info {
1010 __le32 i_data[15]; /* unconverted */
1011 __u32 i_dtime;
1012 ext4_fsblk_t i_file_acl;
1013
1014 /*
1015 * i_block_group is the number of the block group which contains
1016 * this file's inode. Constant across the lifetime of the inode,
1017 * it is used for making block allocation decisions - we try to
1018 * place a file's data blocks near its inode block, and new inodes
1019 * near to their parent directory's inode.
1020 */
1021 ext4_group_t i_block_group;
1022 ext4_lblk_t i_dir_start_lookup;
1023 #if (BITS_PER_LONG < 64)
1024 unsigned long i_state_flags; /* Dynamic state flags */
1025 #endif
1026 unsigned long i_flags;
1027
1028 /*
1029 * Extended attributes can be read independently of the main file
1030 * data. Taking i_mutex even when reading would cause contention
1031 * between readers of EAs and writers of regular file data, so
1032 * instead we synchronize on xattr_sem when reading or changing
1033 * EAs.
1034 */
1035 struct rw_semaphore xattr_sem;
1036
1037 /*
1038 * Inodes with EXT4_STATE_ORPHAN_FILE use i_orphan_idx. Otherwise
1039 * i_orphan is used.
1040 */
1041 union {
1042 struct list_head i_orphan; /* unlinked but open inodes */
1043 unsigned int i_orphan_idx; /* Index in orphan file */
1044 };
1045
1046 /* Fast commit related info */
1047
1048 struct list_head i_fc_list; /*
1049 * inodes that need fast commit
1050 * protected by sbi->s_fc_lock.
1051 */
1052
1053 /* Start of lblk range that needs to be committed in this fast commit */
1054 ext4_lblk_t i_fc_lblk_start;
1055
1056 /* End of lblk range that needs to be committed in this fast commit */
1057 ext4_lblk_t i_fc_lblk_len;
1058
1059 /* Number of ongoing updates on this inode */
1060 atomic_t i_fc_updates;
1061
1062 /* Fast commit wait queue for this inode */
1063 wait_queue_head_t i_fc_wait;
1064
1065 /* Protect concurrent accesses on i_fc_lblk_start, i_fc_lblk_len */
1066 struct mutex i_fc_lock;
1067
1068 /*
1069 * i_disksize keeps track of what the inode size is ON DISK, not
1070 * in memory. During truncate, i_size is set to the new size by
1071 * the VFS prior to calling ext4_truncate(), but the filesystem won't
1072 * set i_disksize to 0 until the truncate is actually under way.
1073 *
1074 * The intent is that i_disksize always represents the blocks which
1075 * are used by this file. This allows recovery to restart truncate
1076 * on orphans if we crash during truncate. We actually write i_disksize
1077 * into the on-disk inode when writing inodes out, instead of i_size.
1078 *
1079 * The only time when i_disksize and i_size may be different is when
1080 * a truncate is in progress. The only things which change i_disksize
1081 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
1082 */
1083 loff_t i_disksize;
1084
1085 /*
1086 * i_data_sem is for serialising ext4_truncate() against
1087 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
1088 * data tree are chopped off during truncate. We can't do that in
1089 * ext4 because whenever we perform intermediate commits during
1090 * truncate, the inode and all the metadata blocks *must* be in a
1091 * consistent state which allows truncation of the orphans to restart
1092 * during recovery. Hence we must fix the get_block-vs-truncate race
1093 * by other means, so we have i_data_sem.
1094 */
1095 struct rw_semaphore i_data_sem;
1096 struct inode vfs_inode;
1097 struct jbd2_inode *jinode;
1098
1099 spinlock_t i_raw_lock; /* protects updates to the raw inode */
1100
1101 /*
1102 * File creation time. Its function is same as that of
1103 * struct timespec64 i_{a,c,m}time in the generic inode.
1104 */
1105 struct timespec64 i_crtime;
1106
1107 /* mballoc */
1108 atomic_t i_prealloc_active;
1109 struct list_head i_prealloc_list;
1110 spinlock_t i_prealloc_lock;
1111
1112 /* extents status tree */
1113 struct ext4_es_tree i_es_tree;
1114 rwlock_t i_es_lock;
1115 struct list_head i_es_list;
1116 unsigned int i_es_all_nr; /* protected by i_es_lock */
1117 unsigned int i_es_shk_nr; /* protected by i_es_lock */
1118 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for
1119 extents to shrink. Protected by
1120 i_es_lock */
1121
1122 /* ialloc */
1123 ext4_group_t i_last_alloc_group;
1124
1125 /* allocation reservation info for delalloc */
1126 /* In case of bigalloc, this refer to clusters rather than blocks */
1127 unsigned int i_reserved_data_blocks;
1128
1129 /* pending cluster reservations for bigalloc file systems */
1130 struct ext4_pending_tree i_pending_tree;
1131
1132 /* on-disk additional length */
1133 __u16 i_extra_isize;
1134
1135 /* Indicate the inline data space. */
1136 u16 i_inline_off;
1137 u16 i_inline_size;
1138
1139 #ifdef CONFIG_QUOTA
1140 /* quota space reservation, managed internally by quota code */
1141 qsize_t i_reserved_quota;
1142 #endif
1143
1144 /* Lock protecting lists below */
1145 spinlock_t i_completed_io_lock;
1146 /*
1147 * Completed IOs that need unwritten extents handling and have
1148 * transaction reserved
1149 */
1150 struct list_head i_rsv_conversion_list;
1151 struct work_struct i_rsv_conversion_work;
1152 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1153
1154 spinlock_t i_block_reservation_lock;
1155
1156 /*
1157 * Transactions that contain inode's metadata needed to complete
1158 * fsync and fdatasync, respectively.
1159 */
1160 tid_t i_sync_tid;
1161 tid_t i_datasync_tid;
1162
1163 #ifdef CONFIG_QUOTA
1164 struct dquot *i_dquot[MAXQUOTAS];
1165 #endif
1166
1167 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1168 __u32 i_csum_seed;
1169
1170 kprojid_t i_projid;
1171 };
1172
1173 /*
1174 * File system states
1175 */
1176 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
1177 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
1178 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
1179 #define EXT4_FC_REPLAY 0x0020 /* Fast commit replay ongoing */
1180
1181 /*
1182 * Misc. filesystem flags
1183 */
1184 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
1185 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
1186 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
1187
1188 /*
1189 * Mount flags set via mount options or defaults
1190 */
1191 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */
1192 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
1193 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
1194 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
1195 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
1196 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
1197 #define EXT4_MOUNT_ERRORS_MASK 0x00070
1198 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
1199 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
1200 #ifdef CONFIG_FS_DAX
1201 #define EXT4_MOUNT_DAX_ALWAYS 0x00200 /* Direct Access */
1202 #else
1203 #define EXT4_MOUNT_DAX_ALWAYS 0
1204 #endif
1205 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
1206 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
1207 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
1208 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
1209 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
1210 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
1211 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
1212 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
1213 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
1214 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
1215 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */
1216 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota,
1217 * enable enforcement for hidden
1218 * quota files */
1219 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable
1220 * enforcement for hidden quota
1221 * files */
1222 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota
1223 * enforcement */
1224 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
1225 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
1226 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1227 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */
1228 #define EXT4_MOUNT_NO_PREFETCH_BLOCK_BITMAPS 0x4000000
1229 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
1230 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
1231 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
1232 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
1233 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
1234
1235 /*
1236 * Mount flags set either automatically (could not be set by mount option)
1237 * based on per file system feature or property or in special cases such as
1238 * distinguishing between explicit mount option definition and default.
1239 */
1240 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1241 specified delalloc */
1242 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1243 size of blocksize * 8
1244 blocks */
1245 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1246 file systems */
1247 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly
1248 specified journal checksum */
1249
1250 #define EXT4_MOUNT2_JOURNAL_FAST_COMMIT 0x00000010 /* Journal fast commit */
1251 #define EXT4_MOUNT2_DAX_NEVER 0x00000020 /* Do not allow Direct Access */
1252 #define EXT4_MOUNT2_DAX_INODE 0x00000040 /* For printing options only */
1253 #define EXT4_MOUNT2_MB_OPTIMIZE_SCAN 0x00000080 /* Optimize group
1254 * scanning in mballoc
1255 */
1256
1257 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1258 ~EXT4_MOUNT_##opt
1259 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1260 EXT4_MOUNT_##opt
1261 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1262 EXT4_MOUNT_##opt)
1263
1264 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1265 ~EXT4_MOUNT2_##opt
1266 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1267 EXT4_MOUNT2_##opt
1268 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1269 EXT4_MOUNT2_##opt)
1270
1271 #define ext4_test_and_set_bit __test_and_set_bit_le
1272 #define ext4_set_bit __set_bit_le
1273 #define ext4_set_bit_atomic ext2_set_bit_atomic
1274 #define ext4_test_and_clear_bit __test_and_clear_bit_le
1275 #define ext4_clear_bit __clear_bit_le
1276 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
1277 #define ext4_test_bit test_bit_le
1278 #define ext4_find_next_zero_bit find_next_zero_bit_le
1279 #define ext4_find_next_bit find_next_bit_le
1280
1281 extern void ext4_set_bits(void *bm, int cur, int len);
1282
1283 /*
1284 * Maximal mount counts between two filesystem checks
1285 */
1286 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1287 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1288
1289 /*
1290 * Behaviour when detecting errors
1291 */
1292 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1293 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1294 #define EXT4_ERRORS_PANIC 3 /* Panic */
1295 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1296
1297 /* Metadata checksum algorithm codes */
1298 #define EXT4_CRC32C_CHKSUM 1
1299
1300 /*
1301 * Structure of the super block
1302 */
1303 struct ext4_super_block {
1304 /*00*/ __le32 s_inodes_count; /* Inodes count */
1305 __le32 s_blocks_count_lo; /* Blocks count */
1306 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1307 __le32 s_free_blocks_count_lo; /* Free blocks count */
1308 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1309 __le32 s_first_data_block; /* First Data Block */
1310 __le32 s_log_block_size; /* Block size */
1311 __le32 s_log_cluster_size; /* Allocation cluster size */
1312 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1313 __le32 s_clusters_per_group; /* # Clusters per group */
1314 __le32 s_inodes_per_group; /* # Inodes per group */
1315 __le32 s_mtime; /* Mount time */
1316 /*30*/ __le32 s_wtime; /* Write time */
1317 __le16 s_mnt_count; /* Mount count */
1318 __le16 s_max_mnt_count; /* Maximal mount count */
1319 __le16 s_magic; /* Magic signature */
1320 __le16 s_state; /* File system state */
1321 __le16 s_errors; /* Behaviour when detecting errors */
1322 __le16 s_minor_rev_level; /* minor revision level */
1323 /*40*/ __le32 s_lastcheck; /* time of last check */
1324 __le32 s_checkinterval; /* max. time between checks */
1325 __le32 s_creator_os; /* OS */
1326 __le32 s_rev_level; /* Revision level */
1327 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1328 __le16 s_def_resgid; /* Default gid for reserved blocks */
1329 /*
1330 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1331 *
1332 * Note: the difference between the compatible feature set and
1333 * the incompatible feature set is that if there is a bit set
1334 * in the incompatible feature set that the kernel doesn't
1335 * know about, it should refuse to mount the filesystem.
1336 *
1337 * e2fsck's requirements are more strict; if it doesn't know
1338 * about a feature in either the compatible or incompatible
1339 * feature set, it must abort and not try to meddle with
1340 * things it doesn't understand...
1341 */
1342 __le32 s_first_ino; /* First non-reserved inode */
1343 __le16 s_inode_size; /* size of inode structure */
1344 __le16 s_block_group_nr; /* block group # of this superblock */
1345 __le32 s_feature_compat; /* compatible feature set */
1346 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1347 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1348 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1349 /*78*/ char s_volume_name[16]; /* volume name */
1350 /*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */
1351 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1352 /*
1353 * Performance hints. Directory preallocation should only
1354 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1355 */
1356 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1357 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1358 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1359 /*
1360 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1361 */
1362 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1363 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1364 __le32 s_journal_dev; /* device number of journal file */
1365 __le32 s_last_orphan; /* start of list of inodes to delete */
1366 __le32 s_hash_seed[4]; /* HTREE hash seed */
1367 __u8 s_def_hash_version; /* Default hash version to use */
1368 __u8 s_jnl_backup_type;
1369 __le16 s_desc_size; /* size of group descriptor */
1370 /*100*/ __le32 s_default_mount_opts;
1371 __le32 s_first_meta_bg; /* First metablock block group */
1372 __le32 s_mkfs_time; /* When the filesystem was created */
1373 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1374 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1375 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1376 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1377 __le32 s_free_blocks_count_hi; /* Free blocks count */
1378 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1379 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1380 __le32 s_flags; /* Miscellaneous flags */
1381 __le16 s_raid_stride; /* RAID stride */
1382 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1383 __le64 s_mmp_block; /* Block for multi-mount protection */
1384 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1385 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1386 __u8 s_checksum_type; /* metadata checksum algorithm used */
1387 __u8 s_encryption_level; /* versioning level for encryption */
1388 __u8 s_reserved_pad; /* Padding to next 32bits */
1389 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1390 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1391 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1392 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1393 snapshot's future use */
1394 __le32 s_snapshot_list; /* inode number of the head of the
1395 on-disk snapshot list */
1396 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1397 __le32 s_error_count; /* number of fs errors */
1398 __le32 s_first_error_time; /* first time an error happened */
1399 __le32 s_first_error_ino; /* inode involved in first error */
1400 __le64 s_first_error_block; /* block involved of first error */
1401 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */
1402 __le32 s_first_error_line; /* line number where error happened */
1403 __le32 s_last_error_time; /* most recent time of an error */
1404 __le32 s_last_error_ino; /* inode involved in last error */
1405 __le32 s_last_error_line; /* line number where error happened */
1406 __le64 s_last_error_block; /* block involved of last error */
1407 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */
1408 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1409 __u8 s_mount_opts[64];
1410 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1411 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1412 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1413 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
1414 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */
1415 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */
1416 __le32 s_lpf_ino; /* Location of the lost+found inode */
1417 __le32 s_prj_quota_inum; /* inode for tracking project quota */
1418 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */
1419 __u8 s_wtime_hi;
1420 __u8 s_mtime_hi;
1421 __u8 s_mkfs_time_hi;
1422 __u8 s_lastcheck_hi;
1423 __u8 s_first_error_time_hi;
1424 __u8 s_last_error_time_hi;
1425 __u8 s_first_error_errcode;
1426 __u8 s_last_error_errcode;
1427 __le16 s_encoding; /* Filename charset encoding */
1428 __le16 s_encoding_flags; /* Filename charset encoding flags */
1429 __le32 s_orphan_file_inum; /* Inode for tracking orphan inodes */
1430 __le32 s_reserved[94]; /* Padding to the end of the block */
1431 __le32 s_checksum; /* crc32c(superblock) */
1432 };
1433
1434 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1435
1436 #ifdef __KERNEL__
1437
1438 #ifdef CONFIG_FS_ENCRYPTION
1439 #define DUMMY_ENCRYPTION_ENABLED(sbi) ((sbi)->s_dummy_enc_policy.policy != NULL)
1440 #else
1441 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1442 #endif
1443
1444 /* Number of quota types we support */
1445 #define EXT4_MAXQUOTAS 3
1446
1447 #define EXT4_ENC_UTF8_12_1 1
1448
1449 /* Types of ext4 journal triggers */
1450 enum ext4_journal_trigger_type {
1451 EXT4_JTR_ORPHAN_FILE,
1452 EXT4_JTR_NONE /* This must be the last entry for indexing to work! */
1453 };
1454
1455 #define EXT4_JOURNAL_TRIGGER_COUNT EXT4_JTR_NONE
1456
1457 struct ext4_journal_trigger {
1458 struct jbd2_buffer_trigger_type tr_triggers;
1459 struct super_block *sb;
1460 };
1461
1462 static inline struct ext4_journal_trigger *EXT4_TRIGGER(
1463 struct jbd2_buffer_trigger_type *trigger)
1464 {
1465 return container_of(trigger, struct ext4_journal_trigger, tr_triggers);
1466 }
1467
1468 #define EXT4_ORPHAN_BLOCK_MAGIC 0x0b10ca04
1469
1470 /* Structure at the tail of orphan block */
1471 struct ext4_orphan_block_tail {
1472 __le32 ob_magic;
1473 __le32 ob_checksum;
1474 };
1475
1476 static inline int ext4_inodes_per_orphan_block(struct super_block *sb)
1477 {
1478 return (sb->s_blocksize - sizeof(struct ext4_orphan_block_tail)) /
1479 sizeof(u32);
1480 }
1481
1482 struct ext4_orphan_block {
1483 atomic_t ob_free_entries; /* Number of free orphan entries in block */
1484 struct buffer_head *ob_bh; /* Buffer for orphan block */
1485 };
1486
1487 /*
1488 * Info about orphan file.
1489 */
1490 struct ext4_orphan_info {
1491 int of_blocks; /* Number of orphan blocks in a file */
1492 __u32 of_csum_seed; /* Checksum seed for orphan file */
1493 struct ext4_orphan_block *of_binfo; /* Array with info about orphan
1494 * file blocks */
1495 };
1496
1497 /*
1498 * fourth extended-fs super-block data in memory
1499 */
1500 struct ext4_sb_info {
1501 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1502 unsigned long s_inodes_per_block;/* Number of inodes per block */
1503 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1504 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1505 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1506 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1507 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1508 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1509 ext4_group_t s_groups_count; /* Number of groups in the fs */
1510 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1511 unsigned long s_overhead; /* # of fs overhead clusters */
1512 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1513 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1514 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1515 struct buffer_head * s_sbh; /* Buffer containing the super block */
1516 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1517 struct buffer_head * __rcu *s_group_desc;
1518 unsigned int s_mount_opt;
1519 unsigned int s_mount_opt2;
1520 unsigned long s_mount_flags;
1521 unsigned int s_def_mount_opt;
1522 ext4_fsblk_t s_sb_block;
1523 atomic64_t s_resv_clusters;
1524 kuid_t s_resuid;
1525 kgid_t s_resgid;
1526 unsigned short s_mount_state;
1527 unsigned short s_pad;
1528 int s_addr_per_block_bits;
1529 int s_desc_per_block_bits;
1530 int s_inode_size;
1531 int s_first_ino;
1532 unsigned int s_inode_readahead_blks;
1533 unsigned int s_inode_goal;
1534 u32 s_hash_seed[4];
1535 int s_def_hash_version;
1536 int s_hash_unsigned; /* 3 if hash should be unsigned, 0 if not */
1537 struct percpu_counter s_freeclusters_counter;
1538 struct percpu_counter s_freeinodes_counter;
1539 struct percpu_counter s_dirs_counter;
1540 struct percpu_counter s_dirtyclusters_counter;
1541 struct percpu_counter s_sra_exceeded_retry_limit;
1542 struct blockgroup_lock *s_blockgroup_lock;
1543 struct proc_dir_entry *s_proc;
1544 struct kobject s_kobj;
1545 struct completion s_kobj_unregister;
1546 struct super_block *s_sb;
1547 struct buffer_head *s_mmp_bh;
1548
1549 /* Journaling */
1550 struct journal_s *s_journal;
1551 unsigned long s_ext4_flags; /* Ext4 superblock flags */
1552 struct mutex s_orphan_lock; /* Protects on disk list changes */
1553 struct list_head s_orphan; /* List of orphaned inodes in on disk
1554 list */
1555 struct ext4_orphan_info s_orphan_info;
1556 unsigned long s_commit_interval;
1557 u32 s_max_batch_time;
1558 u32 s_min_batch_time;
1559 struct block_device *s_journal_bdev;
1560 #ifdef CONFIG_QUOTA
1561 /* Names of quota files with journalled quota */
1562 char __rcu *s_qf_names[EXT4_MAXQUOTAS];
1563 int s_jquota_fmt; /* Format of quota to use */
1564 #endif
1565 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1566 struct ext4_system_blocks __rcu *s_system_blks;
1567
1568 #ifdef EXTENTS_STATS
1569 /* ext4 extents stats */
1570 unsigned long s_ext_min;
1571 unsigned long s_ext_max;
1572 unsigned long s_depth_max;
1573 spinlock_t s_ext_stats_lock;
1574 unsigned long s_ext_blocks;
1575 unsigned long s_ext_extents;
1576 #endif
1577
1578 /* for buddy allocator */
1579 struct ext4_group_info ** __rcu *s_group_info;
1580 struct inode *s_buddy_cache;
1581 spinlock_t s_md_lock;
1582 unsigned short *s_mb_offsets;
1583 unsigned int *s_mb_maxs;
1584 unsigned int s_group_info_size;
1585 unsigned int s_mb_free_pending;
1586 struct list_head s_freed_data_list; /* List of blocks to be freed
1587 after commit completed */
1588 struct list_head s_discard_list;
1589 struct work_struct s_discard_work;
1590 atomic_t s_retry_alloc_pending;
1591 struct rb_root s_mb_avg_fragment_size_root;
1592 rwlock_t s_mb_rb_lock;
1593 struct list_head *s_mb_largest_free_orders;
1594 rwlock_t *s_mb_largest_free_orders_locks;
1595
1596 /* tunables */
1597 unsigned long s_stripe;
1598 unsigned int s_mb_max_linear_groups;
1599 unsigned int s_mb_stream_request;
1600 unsigned int s_mb_max_to_scan;
1601 unsigned int s_mb_min_to_scan;
1602 unsigned int s_mb_stats;
1603 unsigned int s_mb_order2_reqs;
1604 unsigned int s_mb_group_prealloc;
1605 unsigned int s_mb_max_inode_prealloc;
1606 unsigned int s_max_dir_size_kb;
1607 /* where last allocation was done - for stream allocation */
1608 unsigned long s_mb_last_group;
1609 unsigned long s_mb_last_start;
1610 unsigned int s_mb_prefetch;
1611 unsigned int s_mb_prefetch_limit;
1612
1613 /* stats for buddy allocator */
1614 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1615 atomic_t s_bal_success; /* we found long enough chunks */
1616 atomic_t s_bal_allocated; /* in blocks */
1617 atomic_t s_bal_ex_scanned; /* total extents scanned */
1618 atomic_t s_bal_groups_scanned; /* number of groups scanned */
1619 atomic_t s_bal_goals; /* goal hits */
1620 atomic_t s_bal_breaks; /* too long searches */
1621 atomic_t s_bal_2orders; /* 2^order hits */
1622 atomic_t s_bal_cr0_bad_suggestions;
1623 atomic_t s_bal_cr1_bad_suggestions;
1624 atomic64_t s_bal_cX_groups_considered[4];
1625 atomic64_t s_bal_cX_hits[4];
1626 atomic64_t s_bal_cX_failed[4]; /* cX loop didn't find blocks */
1627 atomic_t s_mb_buddies_generated; /* number of buddies generated */
1628 atomic64_t s_mb_generation_time;
1629 atomic_t s_mb_lost_chunks;
1630 atomic_t s_mb_preallocated;
1631 atomic_t s_mb_discarded;
1632 atomic_t s_lock_busy;
1633
1634 /* locality groups */
1635 struct ext4_locality_group __percpu *s_locality_groups;
1636
1637 /* for write statistics */
1638 unsigned long s_sectors_written_start;
1639 u64 s_kbytes_written;
1640
1641 /* the size of zero-out chunk */
1642 unsigned int s_extent_max_zeroout_kb;
1643
1644 unsigned int s_log_groups_per_flex;
1645 struct flex_groups * __rcu *s_flex_groups;
1646 ext4_group_t s_flex_groups_allocated;
1647
1648 /* workqueue for reserved extent conversions (buffered io) */
1649 struct workqueue_struct *rsv_conversion_wq;
1650
1651 /* timer for periodic error stats printing */
1652 struct timer_list s_err_report;
1653
1654 /* Lazy inode table initialization info */
1655 struct ext4_li_request *s_li_request;
1656 /* Wait multiplier for lazy initialization thread */
1657 unsigned int s_li_wait_mult;
1658
1659 /* Kernel thread for multiple mount protection */
1660 struct task_struct *s_mmp_tsk;
1661
1662 /* record the last minlen when FITRIM is called. */
1663 atomic_t s_last_trim_minblks;
1664
1665 /* Reference to checksum algorithm driver via cryptoapi */
1666 struct crypto_shash *s_chksum_driver;
1667
1668 /* Precomputed FS UUID checksum for seeding other checksums */
1669 __u32 s_csum_seed;
1670
1671 /* Reclaim extents from extent status tree */
1672 struct shrinker s_es_shrinker;
1673 struct list_head s_es_list; /* List of inodes with reclaimable extents */
1674 long s_es_nr_inode;
1675 struct ext4_es_stats s_es_stats;
1676 struct mb_cache *s_ea_block_cache;
1677 struct mb_cache *s_ea_inode_cache;
1678 spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1679
1680 /* Journal triggers for checksum computation */
1681 struct ext4_journal_trigger s_journal_triggers[EXT4_JOURNAL_TRIGGER_COUNT];
1682
1683 /* Ratelimit ext4 messages. */
1684 struct ratelimit_state s_err_ratelimit_state;
1685 struct ratelimit_state s_warning_ratelimit_state;
1686 struct ratelimit_state s_msg_ratelimit_state;
1687 atomic_t s_warning_count;
1688 atomic_t s_msg_count;
1689
1690 /* Encryption policy for '-o test_dummy_encryption' */
1691 struct fscrypt_dummy_policy s_dummy_enc_policy;
1692
1693 /*
1694 * Barrier between writepages ops and changing any inode's JOURNAL_DATA
1695 * or EXTENTS flag.
1696 */
1697 struct percpu_rw_semaphore s_writepages_rwsem;
1698 struct dax_device *s_daxdev;
1699 #ifdef CONFIG_EXT4_DEBUG
1700 unsigned long s_simulate_fail;
1701 #endif
1702 /* Record the errseq of the backing block device */
1703 errseq_t s_bdev_wb_err;
1704 spinlock_t s_bdev_wb_lock;
1705
1706 /* Information about errors that happened during this mount */
1707 spinlock_t s_error_lock;
1708 int s_add_error_count;
1709 int s_first_error_code;
1710 __u32 s_first_error_line;
1711 __u32 s_first_error_ino;
1712 __u64 s_first_error_block;
1713 const char *s_first_error_func;
1714 time64_t s_first_error_time;
1715 int s_last_error_code;
1716 __u32 s_last_error_line;
1717 __u32 s_last_error_ino;
1718 __u64 s_last_error_block;
1719 const char *s_last_error_func;
1720 time64_t s_last_error_time;
1721 /*
1722 * If we are in a context where we cannot update error information in
1723 * the on-disk superblock, we queue this work to do it.
1724 */
1725 struct work_struct s_error_work;
1726
1727 /* Ext4 fast commit stuff */
1728 atomic_t s_fc_subtid;
1729 atomic_t s_fc_ineligible_updates;
1730 /*
1731 * After commit starts, the main queue gets locked, and the further
1732 * updates get added in the staging queue.
1733 */
1734 #define FC_Q_MAIN 0
1735 #define FC_Q_STAGING 1
1736 struct list_head s_fc_q[2]; /* Inodes staged for fast commit
1737 * that have data changes in them.
1738 */
1739 struct list_head s_fc_dentry_q[2]; /* directory entry updates */
1740 unsigned int s_fc_bytes;
1741 /*
1742 * Main fast commit lock. This lock protects accesses to the
1743 * following fields:
1744 * ei->i_fc_list, s_fc_dentry_q, s_fc_q, s_fc_bytes, s_fc_bh.
1745 */
1746 spinlock_t s_fc_lock;
1747 struct buffer_head *s_fc_bh;
1748 struct ext4_fc_stats s_fc_stats;
1749 u64 s_fc_avg_commit_time;
1750 #ifdef CONFIG_EXT4_DEBUG
1751 int s_fc_debug_max_replay;
1752 #endif
1753 struct ext4_fc_replay_state s_fc_replay_state;
1754 };
1755
1756 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1757 {
1758 return sb->s_fs_info;
1759 }
1760 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1761 {
1762 return container_of(inode, struct ext4_inode_info, vfs_inode);
1763 }
1764
1765 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1766 {
1767 return ino == EXT4_ROOT_INO ||
1768 (ino >= EXT4_FIRST_INO(sb) &&
1769 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1770 }
1771
1772 /*
1773 * Returns: sbi->field[index]
1774 * Used to access an array element from the following sbi fields which require
1775 * rcu protection to avoid dereferencing an invalid pointer due to reassignment
1776 * - s_group_desc
1777 * - s_group_info
1778 * - s_flex_group
1779 */
1780 #define sbi_array_rcu_deref(sbi, field, index) \
1781 ({ \
1782 typeof(*((sbi)->field)) _v; \
1783 rcu_read_lock(); \
1784 _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index]; \
1785 rcu_read_unlock(); \
1786 _v; \
1787 })
1788
1789 /*
1790 * run-time mount flags
1791 */
1792 enum {
1793 EXT4_MF_MNTDIR_SAMPLED,
1794 EXT4_MF_FS_ABORTED, /* Fatal error detected */
1795 EXT4_MF_FC_INELIGIBLE, /* Fast commit ineligible */
1796 EXT4_MF_FC_COMMITTING /* File system underoing a fast
1797 * commit.
1798 */
1799 };
1800
1801 static inline void ext4_set_mount_flag(struct super_block *sb, int bit)
1802 {
1803 set_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1804 }
1805
1806 static inline void ext4_clear_mount_flag(struct super_block *sb, int bit)
1807 {
1808 clear_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1809 }
1810
1811 static inline int ext4_test_mount_flag(struct super_block *sb, int bit)
1812 {
1813 return test_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1814 }
1815
1816
1817 /*
1818 * Simulate_fail codes
1819 */
1820 #define EXT4_SIM_BBITMAP_EIO 1
1821 #define EXT4_SIM_BBITMAP_CRC 2
1822 #define EXT4_SIM_IBITMAP_EIO 3
1823 #define EXT4_SIM_IBITMAP_CRC 4
1824 #define EXT4_SIM_INODE_EIO 5
1825 #define EXT4_SIM_INODE_CRC 6
1826 #define EXT4_SIM_DIRBLOCK_EIO 7
1827 #define EXT4_SIM_DIRBLOCK_CRC 8
1828
1829 static inline bool ext4_simulate_fail(struct super_block *sb,
1830 unsigned long code)
1831 {
1832 #ifdef CONFIG_EXT4_DEBUG
1833 struct ext4_sb_info *sbi = EXT4_SB(sb);
1834
1835 if (unlikely(sbi->s_simulate_fail == code)) {
1836 sbi->s_simulate_fail = 0;
1837 return true;
1838 }
1839 #endif
1840 return false;
1841 }
1842
1843 static inline void ext4_simulate_fail_bh(struct super_block *sb,
1844 struct buffer_head *bh,
1845 unsigned long code)
1846 {
1847 if (!IS_ERR(bh) && ext4_simulate_fail(sb, code))
1848 clear_buffer_uptodate(bh);
1849 }
1850
1851 /*
1852 * Error number codes for s_{first,last}_error_errno
1853 *
1854 * Linux errno numbers are architecture specific, so we need to translate
1855 * them into something which is architecture independent. We don't define
1856 * codes for all errno's; just the ones which are most likely to be the cause
1857 * of an ext4_error() call.
1858 */
1859 #define EXT4_ERR_UNKNOWN 1
1860 #define EXT4_ERR_EIO 2
1861 #define EXT4_ERR_ENOMEM 3
1862 #define EXT4_ERR_EFSBADCRC 4
1863 #define EXT4_ERR_EFSCORRUPTED 5
1864 #define EXT4_ERR_ENOSPC 6
1865 #define EXT4_ERR_ENOKEY 7
1866 #define EXT4_ERR_EROFS 8
1867 #define EXT4_ERR_EFBIG 9
1868 #define EXT4_ERR_EEXIST 10
1869 #define EXT4_ERR_ERANGE 11
1870 #define EXT4_ERR_EOVERFLOW 12
1871 #define EXT4_ERR_EBUSY 13
1872 #define EXT4_ERR_ENOTDIR 14
1873 #define EXT4_ERR_ENOTEMPTY 15
1874 #define EXT4_ERR_ESHUTDOWN 16
1875 #define EXT4_ERR_EFAULT 17
1876
1877 /*
1878 * Inode dynamic state flags
1879 */
1880 enum {
1881 EXT4_STATE_JDATA, /* journaled data exists */
1882 EXT4_STATE_NEW, /* inode is newly created */
1883 EXT4_STATE_XATTR, /* has in-inode xattrs */
1884 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1885 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1886 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1887 EXT4_STATE_NEWENTRY, /* File just added to dir */
1888 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1889 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
1890 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */
1891 EXT4_STATE_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */
1892 EXT4_STATE_FC_COMMITTING, /* Fast commit ongoing */
1893 EXT4_STATE_ORPHAN_FILE, /* Inode orphaned in orphan file */
1894 };
1895
1896 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1897 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1898 { \
1899 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1900 } \
1901 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1902 { \
1903 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1904 } \
1905 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1906 { \
1907 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1908 }
1909
1910 /* Add these declarations here only so that these functions can be
1911 * found by name. Otherwise, they are very hard to locate. */
1912 static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1913 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1914 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1915 EXT4_INODE_BIT_FNS(flag, flags, 0)
1916
1917 /* Add these declarations here only so that these functions can be
1918 * found by name. Otherwise, they are very hard to locate. */
1919 static inline int ext4_test_inode_state(struct inode *inode, int bit);
1920 static inline void ext4_set_inode_state(struct inode *inode, int bit);
1921 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1922 #if (BITS_PER_LONG < 64)
1923 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1924
1925 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1926 {
1927 (ei)->i_state_flags = 0;
1928 }
1929 #else
1930 EXT4_INODE_BIT_FNS(state, flags, 32)
1931
1932 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1933 {
1934 /* We depend on the fact that callers will set i_flags */
1935 }
1936 #endif
1937 #else
1938 /* Assume that user mode programs are passing in an ext4fs superblock, not
1939 * a kernel struct super_block. This will allow us to call the feature-test
1940 * macros from user land. */
1941 #define EXT4_SB(sb) (sb)
1942 #endif
1943
1944 static inline bool ext4_verity_in_progress(struct inode *inode)
1945 {
1946 return IS_ENABLED(CONFIG_FS_VERITY) &&
1947 ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS);
1948 }
1949
1950 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1951
1952 /*
1953 * Codes for operating systems
1954 */
1955 #define EXT4_OS_LINUX 0
1956 #define EXT4_OS_HURD 1
1957 #define EXT4_OS_MASIX 2
1958 #define EXT4_OS_FREEBSD 3
1959 #define EXT4_OS_LITES 4
1960
1961 /*
1962 * Revision levels
1963 */
1964 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1965 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1966
1967 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1968
1969 #define EXT4_GOOD_OLD_INODE_SIZE 128
1970
1971 #define EXT4_EXTRA_TIMESTAMP_MAX (((s64)1 << 34) - 1 + S32_MIN)
1972 #define EXT4_NON_EXTRA_TIMESTAMP_MAX S32_MAX
1973 #define EXT4_TIMESTAMP_MIN S32_MIN
1974
1975 /*
1976 * Feature set definitions
1977 */
1978
1979 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1980 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1981 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1982 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1983 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1984 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1985 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
1986 /*
1987 * The reason why "FAST_COMMIT" is a compat feature is that, FS becomes
1988 * incompatible only if fast commit blocks are present in the FS. Since we
1989 * clear the journal (and thus the fast commit blocks), we don't mark FS as
1990 * incompatible. We also have a JBD2 incompat feature, which gets set when
1991 * there are fast commit blocks present in the journal.
1992 */
1993 #define EXT4_FEATURE_COMPAT_FAST_COMMIT 0x0400
1994 #define EXT4_FEATURE_COMPAT_STABLE_INODES 0x0800
1995 #define EXT4_FEATURE_COMPAT_ORPHAN_FILE 0x1000 /* Orphan file exists */
1996
1997 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1998 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1999 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
2000 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
2001 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
2002 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
2003 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
2004 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
2005 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
2006 /*
2007 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
2008 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
2009 * all other data structures' checksums. However, the METADATA_CSUM and
2010 * GDT_CSUM bits are mutually exclusive.
2011 */
2012 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
2013 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000
2014 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000
2015 #define EXT4_FEATURE_RO_COMPAT_VERITY 0x8000
2016 #define EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT 0x10000 /* Orphan file may be
2017 non-empty */
2018
2019 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
2020 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
2021 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
2022 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
2023 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
2024 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
2025 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
2026 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
2027 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
2028 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
2029 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
2030 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000
2031 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
2032 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
2033 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000
2034 #define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000
2035
2036 extern void ext4_update_dynamic_rev(struct super_block *sb);
2037
2038 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
2039 static inline bool ext4_has_feature_##name(struct super_block *sb) \
2040 { \
2041 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
2042 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
2043 } \
2044 static inline void ext4_set_feature_##name(struct super_block *sb) \
2045 { \
2046 ext4_update_dynamic_rev(sb); \
2047 EXT4_SB(sb)->s_es->s_feature_compat |= \
2048 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
2049 } \
2050 static inline void ext4_clear_feature_##name(struct super_block *sb) \
2051 { \
2052 EXT4_SB(sb)->s_es->s_feature_compat &= \
2053 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
2054 }
2055
2056 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
2057 static inline bool ext4_has_feature_##name(struct super_block *sb) \
2058 { \
2059 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
2060 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
2061 } \
2062 static inline void ext4_set_feature_##name(struct super_block *sb) \
2063 { \
2064 ext4_update_dynamic_rev(sb); \
2065 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
2066 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
2067 } \
2068 static inline void ext4_clear_feature_##name(struct super_block *sb) \
2069 { \
2070 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
2071 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
2072 }
2073
2074 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
2075 static inline bool ext4_has_feature_##name(struct super_block *sb) \
2076 { \
2077 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
2078 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
2079 } \
2080 static inline void ext4_set_feature_##name(struct super_block *sb) \
2081 { \
2082 ext4_update_dynamic_rev(sb); \
2083 EXT4_SB(sb)->s_es->s_feature_incompat |= \
2084 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
2085 } \
2086 static inline void ext4_clear_feature_##name(struct super_block *sb) \
2087 { \
2088 EXT4_SB(sb)->s_es->s_feature_incompat &= \
2089 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
2090 }
2091
2092 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC)
2093 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES)
2094 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL)
2095 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR)
2096 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE)
2097 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX)
2098 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2)
2099 EXT4_FEATURE_COMPAT_FUNCS(fast_commit, FAST_COMMIT)
2100 EXT4_FEATURE_COMPAT_FUNCS(stable_inodes, STABLE_INODES)
2101 EXT4_FEATURE_COMPAT_FUNCS(orphan_file, ORPHAN_FILE)
2102
2103 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER)
2104 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE)
2105 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR)
2106 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE)
2107 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM)
2108 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK)
2109 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE)
2110 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA)
2111 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC)
2112 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM)
2113 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY)
2114 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT)
2115 EXT4_FEATURE_RO_COMPAT_FUNCS(verity, VERITY)
2116 EXT4_FEATURE_RO_COMPAT_FUNCS(orphan_present, ORPHAN_PRESENT)
2117
2118 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION)
2119 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE)
2120 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER)
2121 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV)
2122 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG)
2123 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS)
2124 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT)
2125 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP)
2126 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG)
2127 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE)
2128 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA)
2129 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED)
2130 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR)
2131 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA)
2132 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT)
2133 EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD)
2134
2135 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
2136 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
2137 EXT4_FEATURE_INCOMPAT_META_BG)
2138 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2139 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2140 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
2141
2142 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
2143 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
2144 EXT4_FEATURE_INCOMPAT_RECOVER| \
2145 EXT4_FEATURE_INCOMPAT_META_BG)
2146 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2147 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2148 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
2149
2150 #define EXT4_FEATURE_COMPAT_SUPP (EXT4_FEATURE_COMPAT_EXT_ATTR| \
2151 EXT4_FEATURE_COMPAT_ORPHAN_FILE)
2152 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
2153 EXT4_FEATURE_INCOMPAT_RECOVER| \
2154 EXT4_FEATURE_INCOMPAT_META_BG| \
2155 EXT4_FEATURE_INCOMPAT_EXTENTS| \
2156 EXT4_FEATURE_INCOMPAT_64BIT| \
2157 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
2158 EXT4_FEATURE_INCOMPAT_EA_INODE| \
2159 EXT4_FEATURE_INCOMPAT_MMP | \
2160 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
2161 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
2162 EXT4_FEATURE_INCOMPAT_CASEFOLD | \
2163 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \
2164 EXT4_FEATURE_INCOMPAT_LARGEDIR)
2165 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2166 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2167 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
2168 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
2169 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
2170 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
2171 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
2172 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
2173 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
2174 EXT4_FEATURE_RO_COMPAT_QUOTA |\
2175 EXT4_FEATURE_RO_COMPAT_PROJECT |\
2176 EXT4_FEATURE_RO_COMPAT_VERITY |\
2177 EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT)
2178
2179 #define EXTN_FEATURE_FUNCS(ver) \
2180 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
2181 { \
2182 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
2183 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
2184 } \
2185 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
2186 { \
2187 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
2188 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
2189 } \
2190 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
2191 { \
2192 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
2193 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
2194 }
2195
2196 EXTN_FEATURE_FUNCS(2)
2197 EXTN_FEATURE_FUNCS(3)
2198 EXTN_FEATURE_FUNCS(4)
2199
2200 static inline bool ext4_has_compat_features(struct super_block *sb)
2201 {
2202 return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
2203 }
2204 static inline bool ext4_has_ro_compat_features(struct super_block *sb)
2205 {
2206 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
2207 }
2208 static inline bool ext4_has_incompat_features(struct super_block *sb)
2209 {
2210 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
2211 }
2212
2213 extern int ext4_feature_set_ok(struct super_block *sb, int readonly);
2214
2215 /*
2216 * Superblock flags
2217 */
2218 #define EXT4_FLAGS_RESIZING 0
2219 #define EXT4_FLAGS_SHUTDOWN 1
2220 #define EXT4_FLAGS_BDEV_IS_DAX 2
2221
2222 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi)
2223 {
2224 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags);
2225 }
2226
2227 /*
2228 * Default values for user and/or group using reserved blocks
2229 */
2230 #define EXT4_DEF_RESUID 0
2231 #define EXT4_DEF_RESGID 0
2232
2233 /*
2234 * Default project ID
2235 */
2236 #define EXT4_DEF_PROJID 0
2237
2238 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
2239
2240 /*
2241 * Default mount options
2242 */
2243 #define EXT4_DEFM_DEBUG 0x0001
2244 #define EXT4_DEFM_BSDGROUPS 0x0002
2245 #define EXT4_DEFM_XATTR_USER 0x0004
2246 #define EXT4_DEFM_ACL 0x0008
2247 #define EXT4_DEFM_UID16 0x0010
2248 #define EXT4_DEFM_JMODE 0x0060
2249 #define EXT4_DEFM_JMODE_DATA 0x0020
2250 #define EXT4_DEFM_JMODE_ORDERED 0x0040
2251 #define EXT4_DEFM_JMODE_WBACK 0x0060
2252 #define EXT4_DEFM_NOBARRIER 0x0100
2253 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
2254 #define EXT4_DEFM_DISCARD 0x0400
2255 #define EXT4_DEFM_NODELALLOC 0x0800
2256
2257 /*
2258 * Default journal batch times
2259 */
2260 #define EXT4_DEF_MIN_BATCH_TIME 0
2261 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
2262
2263 /*
2264 * Minimum number of groups in a flexgroup before we separate out
2265 * directories into the first block group of a flexgroup
2266 */
2267 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
2268
2269 /*
2270 * Structure of a directory entry
2271 */
2272 #define EXT4_NAME_LEN 255
2273
2274 struct ext4_dir_entry {
2275 __le32 inode; /* Inode number */
2276 __le16 rec_len; /* Directory entry length */
2277 __le16 name_len; /* Name length */
2278 char name[EXT4_NAME_LEN]; /* File name */
2279 };
2280
2281
2282 /*
2283 * Encrypted Casefolded entries require saving the hash on disk. This structure
2284 * followed ext4_dir_entry_2's name[name_len] at the next 4 byte aligned
2285 * boundary.
2286 */
2287 struct ext4_dir_entry_hash {
2288 __le32 hash;
2289 __le32 minor_hash;
2290 };
2291
2292 /*
2293 * The new version of the directory entry. Since EXT4 structures are
2294 * stored in intel byte order, and the name_len field could never be
2295 * bigger than 255 chars, it's safe to reclaim the extra byte for the
2296 * file_type field.
2297 */
2298 struct ext4_dir_entry_2 {
2299 __le32 inode; /* Inode number */
2300 __le16 rec_len; /* Directory entry length */
2301 __u8 name_len; /* Name length */
2302 __u8 file_type; /* See file type macros EXT4_FT_* below */
2303 char name[EXT4_NAME_LEN]; /* File name */
2304 };
2305
2306 /*
2307 * Access the hashes at the end of ext4_dir_entry_2
2308 */
2309 #define EXT4_DIRENT_HASHES(entry) \
2310 ((struct ext4_dir_entry_hash *) \
2311 (((void *)(entry)) + \
2312 ((8 + (entry)->name_len + EXT4_DIR_ROUND) & ~EXT4_DIR_ROUND)))
2313 #define EXT4_DIRENT_HASH(entry) le32_to_cpu(EXT4_DIRENT_HASHES(de)->hash)
2314 #define EXT4_DIRENT_MINOR_HASH(entry) \
2315 le32_to_cpu(EXT4_DIRENT_HASHES(de)->minor_hash)
2316
2317 static inline bool ext4_hash_in_dirent(const struct inode *inode)
2318 {
2319 return IS_CASEFOLDED(inode) && IS_ENCRYPTED(inode);
2320 }
2321
2322 /*
2323 * This is a bogus directory entry at the end of each leaf block that
2324 * records checksums.
2325 */
2326 struct ext4_dir_entry_tail {
2327 __le32 det_reserved_zero1; /* Pretend to be unused */
2328 __le16 det_rec_len; /* 12 */
2329 __u8 det_reserved_zero2; /* Zero name length */
2330 __u8 det_reserved_ft; /* 0xDE, fake file type */
2331 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
2332 };
2333
2334 #define EXT4_DIRENT_TAIL(block, blocksize) \
2335 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
2336 ((blocksize) - \
2337 sizeof(struct ext4_dir_entry_tail))))
2338
2339 /*
2340 * Ext4 directory file types. Only the low 3 bits are used. The
2341 * other bits are reserved for now.
2342 */
2343 #define EXT4_FT_UNKNOWN 0
2344 #define EXT4_FT_REG_FILE 1
2345 #define EXT4_FT_DIR 2
2346 #define EXT4_FT_CHRDEV 3
2347 #define EXT4_FT_BLKDEV 4
2348 #define EXT4_FT_FIFO 5
2349 #define EXT4_FT_SOCK 6
2350 #define EXT4_FT_SYMLINK 7
2351
2352 #define EXT4_FT_MAX 8
2353
2354 #define EXT4_FT_DIR_CSUM 0xDE
2355
2356 /*
2357 * EXT4_DIR_PAD defines the directory entries boundaries
2358 *
2359 * NOTE: It must be a multiple of 4
2360 */
2361 #define EXT4_DIR_PAD 4
2362 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
2363 #define EXT4_MAX_REC_LEN ((1<<16)-1)
2364
2365 /*
2366 * The rec_len is dependent on the type of directory. Directories that are
2367 * casefolded and encrypted need to store the hash as well, so we add room for
2368 * ext4_extended_dir_entry_2. For all entries related to '.' or '..' you should
2369 * pass NULL for dir, as those entries do not use the extra fields.
2370 */
2371 static inline unsigned int ext4_dir_rec_len(__u8 name_len,
2372 const struct inode *dir)
2373 {
2374 int rec_len = (name_len + 8 + EXT4_DIR_ROUND);
2375
2376 if (dir && ext4_hash_in_dirent(dir))
2377 rec_len += sizeof(struct ext4_dir_entry_hash);
2378 return (rec_len & ~EXT4_DIR_ROUND);
2379 }
2380
2381 /*
2382 * If we ever get support for fs block sizes > page_size, we'll need
2383 * to remove the #if statements in the next two functions...
2384 */
2385 static inline unsigned int
2386 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
2387 {
2388 unsigned len = le16_to_cpu(dlen);
2389
2390 #if (PAGE_SIZE >= 65536)
2391 if (len == EXT4_MAX_REC_LEN || len == 0)
2392 return blocksize;
2393 return (len & 65532) | ((len & 3) << 16);
2394 #else
2395 return len;
2396 #endif
2397 }
2398
2399 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
2400 {
2401 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
2402 BUG();
2403 #if (PAGE_SIZE >= 65536)
2404 if (len < 65536)
2405 return cpu_to_le16(len);
2406 if (len == blocksize) {
2407 if (blocksize == 65536)
2408 return cpu_to_le16(EXT4_MAX_REC_LEN);
2409 else
2410 return cpu_to_le16(0);
2411 }
2412 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
2413 #else
2414 return cpu_to_le16(len);
2415 #endif
2416 }
2417
2418 /*
2419 * Hash Tree Directory indexing
2420 * (c) Daniel Phillips, 2001
2421 */
2422
2423 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
2424 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2425 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \
2426 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir)))
2427 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2428
2429 /* Legal values for the dx_root hash_version field: */
2430
2431 #define DX_HASH_LEGACY 0
2432 #define DX_HASH_HALF_MD4 1
2433 #define DX_HASH_TEA 2
2434 #define DX_HASH_LEGACY_UNSIGNED 3
2435 #define DX_HASH_HALF_MD4_UNSIGNED 4
2436 #define DX_HASH_TEA_UNSIGNED 5
2437 #define DX_HASH_SIPHASH 6
2438
2439 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2440 const void *address, unsigned int length)
2441 {
2442 struct {
2443 struct shash_desc shash;
2444 char ctx[4];
2445 } desc;
2446
2447 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2448
2449 desc.shash.tfm = sbi->s_chksum_driver;
2450 *(u32 *)desc.ctx = crc;
2451
2452 BUG_ON(crypto_shash_update(&desc.shash, address, length));
2453
2454 return *(u32 *)desc.ctx;
2455 }
2456
2457 #ifdef __KERNEL__
2458
2459 /* hash info structure used by the directory hash */
2460 struct dx_hash_info
2461 {
2462 u32 hash;
2463 u32 minor_hash;
2464 int hash_version;
2465 u32 *seed;
2466 };
2467
2468
2469 /* 32 and 64 bit signed EOF for dx directories */
2470 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
2471 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
2472
2473
2474 /*
2475 * Control parameters used by ext4_htree_next_block
2476 */
2477 #define HASH_NB_ALWAYS 1
2478
2479 struct ext4_filename {
2480 const struct qstr *usr_fname;
2481 struct fscrypt_str disk_name;
2482 struct dx_hash_info hinfo;
2483 #ifdef CONFIG_FS_ENCRYPTION
2484 struct fscrypt_str crypto_buf;
2485 #endif
2486 #ifdef CONFIG_UNICODE
2487 struct fscrypt_str cf_name;
2488 #endif
2489 };
2490
2491 #define fname_name(p) ((p)->disk_name.name)
2492 #define fname_usr_name(p) ((p)->usr_fname->name)
2493 #define fname_len(p) ((p)->disk_name.len)
2494
2495 /*
2496 * Describe an inode's exact location on disk and in memory
2497 */
2498 struct ext4_iloc
2499 {
2500 struct buffer_head *bh;
2501 unsigned long offset;
2502 ext4_group_t block_group;
2503 };
2504
2505 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2506 {
2507 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2508 }
2509
2510 static inline bool ext4_is_quota_file(struct inode *inode)
2511 {
2512 return IS_NOQUOTA(inode) &&
2513 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL);
2514 }
2515
2516 /*
2517 * This structure is stuffed into the struct file's private_data field
2518 * for directories. It is where we put information so that we can do
2519 * readdir operations in hash tree order.
2520 */
2521 struct dir_private_info {
2522 struct rb_root root;
2523 struct rb_node *curr_node;
2524 struct fname *extra_fname;
2525 loff_t last_pos;
2526 __u32 curr_hash;
2527 __u32 curr_minor_hash;
2528 __u32 next_hash;
2529 };
2530
2531 /* calculate the first block number of the group */
2532 static inline ext4_fsblk_t
2533 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2534 {
2535 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2536 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2537 }
2538
2539 /*
2540 * Special error return code only used by dx_probe() and its callers.
2541 */
2542 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
2543
2544 /* htree levels for ext4 */
2545 #define EXT4_HTREE_LEVEL_COMPAT 2
2546 #define EXT4_HTREE_LEVEL 3
2547
2548 static inline int ext4_dir_htree_level(struct super_block *sb)
2549 {
2550 return ext4_has_feature_largedir(sb) ?
2551 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT;
2552 }
2553
2554 /*
2555 * Timeout and state flag for lazy initialization inode thread.
2556 */
2557 #define EXT4_DEF_LI_WAIT_MULT 10
2558 #define EXT4_DEF_LI_MAX_START_DELAY 5
2559 #define EXT4_LAZYINIT_QUIT 0x0001
2560 #define EXT4_LAZYINIT_RUNNING 0x0002
2561
2562 /*
2563 * Lazy inode table initialization info
2564 */
2565 struct ext4_lazy_init {
2566 unsigned long li_state;
2567 struct list_head li_request_list;
2568 struct mutex li_list_mtx;
2569 };
2570
2571 enum ext4_li_mode {
2572 EXT4_LI_MODE_PREFETCH_BBITMAP,
2573 EXT4_LI_MODE_ITABLE,
2574 };
2575
2576 struct ext4_li_request {
2577 struct super_block *lr_super;
2578 enum ext4_li_mode lr_mode;
2579 ext4_group_t lr_first_not_zeroed;
2580 ext4_group_t lr_next_group;
2581 struct list_head lr_request;
2582 unsigned long lr_next_sched;
2583 unsigned long lr_timeout;
2584 };
2585
2586 struct ext4_features {
2587 struct kobject f_kobj;
2588 struct completion f_kobj_unregister;
2589 };
2590
2591 /*
2592 * This structure will be used for multiple mount protection. It will be
2593 * written into the block number saved in the s_mmp_block field in the
2594 * superblock. Programs that check MMP should assume that if
2595 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2596 * to use the filesystem, regardless of how old the timestamp is.
2597 */
2598 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
2599 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2600 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
2601 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
2602
2603 struct mmp_struct {
2604 __le32 mmp_magic; /* Magic number for MMP */
2605 __le32 mmp_seq; /* Sequence no. updated periodically */
2606
2607 /*
2608 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2609 * purposes and do not affect the correctness of the algorithm
2610 */
2611 __le64 mmp_time; /* Time last updated */
2612 char mmp_nodename[64]; /* Node which last updated MMP block */
2613 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
2614
2615 /*
2616 * mmp_check_interval is used to verify if the MMP block has been
2617 * updated on the block device. The value is updated based on the
2618 * maximum time to write the MMP block during an update cycle.
2619 */
2620 __le16 mmp_check_interval;
2621
2622 __le16 mmp_pad1;
2623 __le32 mmp_pad2[226];
2624 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
2625 };
2626
2627 /* arguments passed to the mmp thread */
2628 struct mmpd_data {
2629 struct buffer_head *bh; /* bh from initial read_mmp_block() */
2630 struct super_block *sb; /* super block of the fs */
2631 };
2632
2633 /*
2634 * Check interval multiplier
2635 * The MMP block is written every update interval and initially checked every
2636 * update interval x the multiplier (the value is then adapted based on the
2637 * write latency). The reason is that writes can be delayed under load and we
2638 * don't want readers to incorrectly assume that the filesystem is no longer
2639 * in use.
2640 */
2641 #define EXT4_MMP_CHECK_MULT 2UL
2642
2643 /*
2644 * Minimum interval for MMP checking in seconds.
2645 */
2646 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
2647
2648 /*
2649 * Maximum interval for MMP checking in seconds.
2650 */
2651 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
2652
2653 /*
2654 * Function prototypes
2655 */
2656
2657 /*
2658 * Ok, these declarations are also in <linux/kernel.h> but none of the
2659 * ext4 source programs needs to include it so they are duplicated here.
2660 */
2661 # define NORET_TYPE /**/
2662 # define ATTRIB_NORET __attribute__((noreturn))
2663 # define NORET_AND noreturn,
2664
2665 /* bitmap.c */
2666 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2667 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2668 struct ext4_group_desc *gdp,
2669 struct buffer_head *bh, int sz);
2670 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2671 struct ext4_group_desc *gdp,
2672 struct buffer_head *bh, int sz);
2673 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2674 struct ext4_group_desc *gdp,
2675 struct buffer_head *bh);
2676 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2677 struct ext4_group_desc *gdp,
2678 struct buffer_head *bh);
2679
2680 /* balloc.c */
2681 extern void ext4_get_group_no_and_offset(struct super_block *sb,
2682 ext4_fsblk_t blocknr,
2683 ext4_group_t *blockgrpp,
2684 ext4_grpblk_t *offsetp);
2685 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2686 ext4_fsblk_t block);
2687
2688 extern unsigned int ext4_block_group(struct super_block *sb,
2689 ext4_fsblk_t blocknr);
2690 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2691 ext4_fsblk_t blocknr);
2692 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2693 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2694 ext4_group_t group);
2695 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2696 ext4_fsblk_t goal,
2697 unsigned int flags,
2698 unsigned long *count,
2699 int *errp);
2700 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2701 s64 nclusters, unsigned int flags);
2702 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2703 extern void ext4_check_blocks_bitmap(struct super_block *);
2704 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2705 ext4_group_t block_group,
2706 struct buffer_head ** bh);
2707 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2708
2709 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2710 ext4_group_t block_group,
2711 bool ignore_locked);
2712 extern int ext4_wait_block_bitmap(struct super_block *sb,
2713 ext4_group_t block_group,
2714 struct buffer_head *bh);
2715 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2716 ext4_group_t block_group);
2717 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2718 ext4_group_t block_group,
2719 struct ext4_group_desc *gdp);
2720 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2721
2722 #ifdef CONFIG_UNICODE
2723 extern int ext4_fname_setup_ci_filename(struct inode *dir,
2724 const struct qstr *iname,
2725 struct ext4_filename *fname);
2726 #endif
2727
2728 #ifdef CONFIG_FS_ENCRYPTION
2729 static inline void ext4_fname_from_fscrypt_name(struct ext4_filename *dst,
2730 const struct fscrypt_name *src)
2731 {
2732 memset(dst, 0, sizeof(*dst));
2733
2734 dst->usr_fname = src->usr_fname;
2735 dst->disk_name = src->disk_name;
2736 dst->hinfo.hash = src->hash;
2737 dst->hinfo.minor_hash = src->minor_hash;
2738 dst->crypto_buf = src->crypto_buf;
2739 }
2740
2741 static inline int ext4_fname_setup_filename(struct inode *dir,
2742 const struct qstr *iname,
2743 int lookup,
2744 struct ext4_filename *fname)
2745 {
2746 struct fscrypt_name name;
2747 int err;
2748
2749 err = fscrypt_setup_filename(dir, iname, lookup, &name);
2750 if (err)
2751 return err;
2752
2753 ext4_fname_from_fscrypt_name(fname, &name);
2754
2755 #ifdef CONFIG_UNICODE
2756 err = ext4_fname_setup_ci_filename(dir, iname, fname);
2757 #endif
2758 return err;
2759 }
2760
2761 static inline int ext4_fname_prepare_lookup(struct inode *dir,
2762 struct dentry *dentry,
2763 struct ext4_filename *fname)
2764 {
2765 struct fscrypt_name name;
2766 int err;
2767
2768 err = fscrypt_prepare_lookup(dir, dentry, &name);
2769 if (err)
2770 return err;
2771
2772 ext4_fname_from_fscrypt_name(fname, &name);
2773
2774 #ifdef CONFIG_UNICODE
2775 err = ext4_fname_setup_ci_filename(dir, &dentry->d_name, fname);
2776 #endif
2777 return err;
2778 }
2779
2780 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2781 {
2782 struct fscrypt_name name;
2783
2784 name.crypto_buf = fname->crypto_buf;
2785 fscrypt_free_filename(&name);
2786
2787 fname->crypto_buf.name = NULL;
2788 fname->usr_fname = NULL;
2789 fname->disk_name.name = NULL;
2790
2791 #ifdef CONFIG_UNICODE
2792 kfree(fname->cf_name.name);
2793 fname->cf_name.name = NULL;
2794 #endif
2795 }
2796 #else /* !CONFIG_FS_ENCRYPTION */
2797 static inline int ext4_fname_setup_filename(struct inode *dir,
2798 const struct qstr *iname,
2799 int lookup,
2800 struct ext4_filename *fname)
2801 {
2802 int err = 0;
2803 fname->usr_fname = iname;
2804 fname->disk_name.name = (unsigned char *) iname->name;
2805 fname->disk_name.len = iname->len;
2806
2807 #ifdef CONFIG_UNICODE
2808 err = ext4_fname_setup_ci_filename(dir, iname, fname);
2809 #endif
2810
2811 return err;
2812 }
2813
2814 static inline int ext4_fname_prepare_lookup(struct inode *dir,
2815 struct dentry *dentry,
2816 struct ext4_filename *fname)
2817 {
2818 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname);
2819 }
2820
2821 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2822 {
2823 #ifdef CONFIG_UNICODE
2824 kfree(fname->cf_name.name);
2825 fname->cf_name.name = NULL;
2826 #endif
2827 }
2828 #endif /* !CONFIG_FS_ENCRYPTION */
2829
2830 /* dir.c */
2831 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2832 struct file *,
2833 struct ext4_dir_entry_2 *,
2834 struct buffer_head *, char *, int,
2835 unsigned int);
2836 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2837 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2838 (de), (bh), (buf), (size), (offset)))
2839 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2840 __u32 minor_hash,
2841 struct ext4_dir_entry_2 *dirent,
2842 struct fscrypt_str *ent_name);
2843 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2844 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2845 struct buffer_head *bh,
2846 void *buf, int buf_size,
2847 struct ext4_filename *fname,
2848 struct ext4_dir_entry_2 **dest_de);
2849 void ext4_insert_dentry(struct inode *dir, struct inode *inode,
2850 struct ext4_dir_entry_2 *de,
2851 int buf_size,
2852 struct ext4_filename *fname);
2853 static inline void ext4_update_dx_flag(struct inode *inode)
2854 {
2855 if (!ext4_has_feature_dir_index(inode->i_sb) &&
2856 ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) {
2857 /* ext4_iget() should have caught this... */
2858 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb));
2859 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2860 }
2861 }
2862 static const unsigned char ext4_filetype_table[] = {
2863 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2864 };
2865
2866 static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2867 {
2868 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2869 return DT_UNKNOWN;
2870
2871 return ext4_filetype_table[filetype];
2872 }
2873 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2874 void *buf, int buf_size);
2875
2876 /* fsync.c */
2877 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2878
2879 /* hash.c */
2880 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len,
2881 struct dx_hash_info *hinfo);
2882
2883 /* ialloc.c */
2884 extern int ext4_mark_inode_used(struct super_block *sb, int ino);
2885 extern struct inode *__ext4_new_inode(struct user_namespace *, handle_t *,
2886 struct inode *, umode_t,
2887 const struct qstr *qstr, __u32 goal,
2888 uid_t *owner, __u32 i_flags,
2889 int handle_type, unsigned int line_no,
2890 int nblocks);
2891
2892 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \
2893 __ext4_new_inode(&init_user_ns, (handle), (dir), (mode), (qstr), \
2894 (goal), (owner), i_flags, 0, 0, 0)
2895 #define ext4_new_inode_start_handle(mnt_userns, dir, mode, qstr, goal, owner, \
2896 type, nblocks) \
2897 __ext4_new_inode((mnt_userns), NULL, (dir), (mode), (qstr), (goal), (owner), \
2898 0, (type), __LINE__, (nblocks))
2899
2900
2901 extern void ext4_free_inode(handle_t *, struct inode *);
2902 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2903 extern unsigned long ext4_count_free_inodes(struct super_block *);
2904 extern unsigned long ext4_count_dirs(struct super_block *);
2905 extern void ext4_check_inodes_bitmap(struct super_block *);
2906 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2907 extern int ext4_init_inode_table(struct super_block *sb,
2908 ext4_group_t group, int barrier);
2909 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2910
2911 /* fast_commit.c */
2912 int ext4_fc_info_show(struct seq_file *seq, void *v);
2913 void ext4_fc_init(struct super_block *sb, journal_t *journal);
2914 void ext4_fc_init_inode(struct inode *inode);
2915 void ext4_fc_track_range(handle_t *handle, struct inode *inode, ext4_lblk_t start,
2916 ext4_lblk_t end);
2917 void __ext4_fc_track_unlink(handle_t *handle, struct inode *inode,
2918 struct dentry *dentry);
2919 void __ext4_fc_track_link(handle_t *handle, struct inode *inode,
2920 struct dentry *dentry);
2921 void ext4_fc_track_unlink(handle_t *handle, struct dentry *dentry);
2922 void ext4_fc_track_link(handle_t *handle, struct dentry *dentry);
2923 void __ext4_fc_track_create(handle_t *handle, struct inode *inode,
2924 struct dentry *dentry);
2925 void ext4_fc_track_create(handle_t *handle, struct dentry *dentry);
2926 void ext4_fc_track_inode(handle_t *handle, struct inode *inode);
2927 void ext4_fc_mark_ineligible(struct super_block *sb, int reason);
2928 void ext4_fc_start_ineligible(struct super_block *sb, int reason);
2929 void ext4_fc_stop_ineligible(struct super_block *sb);
2930 void ext4_fc_start_update(struct inode *inode);
2931 void ext4_fc_stop_update(struct inode *inode);
2932 void ext4_fc_del(struct inode *inode);
2933 bool ext4_fc_replay_check_excluded(struct super_block *sb, ext4_fsblk_t block);
2934 void ext4_fc_replay_cleanup(struct super_block *sb);
2935 int ext4_fc_commit(journal_t *journal, tid_t commit_tid);
2936 int __init ext4_fc_init_dentry_cache(void);
2937
2938 /* mballoc.c */
2939 extern const struct seq_operations ext4_mb_seq_groups_ops;
2940 extern const struct seq_operations ext4_mb_seq_structs_summary_ops;
2941 extern long ext4_mb_stats;
2942 extern long ext4_mb_max_to_scan;
2943 extern int ext4_seq_mb_stats_show(struct seq_file *seq, void *offset);
2944 extern int ext4_mb_init(struct super_block *);
2945 extern int ext4_mb_release(struct super_block *);
2946 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2947 struct ext4_allocation_request *, int *);
2948 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2949 extern void ext4_discard_preallocations(struct inode *, unsigned int);
2950 extern int __init ext4_init_mballoc(void);
2951 extern void ext4_exit_mballoc(void);
2952 extern ext4_group_t ext4_mb_prefetch(struct super_block *sb,
2953 ext4_group_t group,
2954 unsigned int nr, int *cnt);
2955 extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group,
2956 unsigned int nr);
2957
2958 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2959 struct buffer_head *bh, ext4_fsblk_t block,
2960 unsigned long count, int flags);
2961 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2962 ext4_group_t ngroups);
2963 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2964 ext4_group_t i, struct ext4_group_desc *desc);
2965 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2966 ext4_fsblk_t block, unsigned long count);
2967 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2968 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid);
2969 extern void ext4_mb_mark_bb(struct super_block *sb, ext4_fsblk_t block,
2970 int len, int state);
2971
2972 /* inode.c */
2973 void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw,
2974 struct ext4_inode_info *ei);
2975 int ext4_inode_is_fast_symlink(struct inode *inode);
2976 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2977 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2978 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count,
2979 bool wait, struct buffer_head **bhs);
2980 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2981 struct buffer_head *bh_result, int create);
2982 int ext4_get_block(struct inode *inode, sector_t iblock,
2983 struct buffer_head *bh_result, int create);
2984 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2985 struct buffer_head *bh, int create);
2986 int ext4_walk_page_buffers(handle_t *handle,
2987 struct inode *inode,
2988 struct buffer_head *head,
2989 unsigned from,
2990 unsigned to,
2991 int *partial,
2992 int (*fn)(handle_t *handle, struct inode *inode,
2993 struct buffer_head *bh));
2994 int do_journal_get_write_access(handle_t *handle, struct inode *inode,
2995 struct buffer_head *bh);
2996 #define FALL_BACK_TO_NONDELALLOC 1
2997 #define CONVERT_INLINE_DATA 2
2998
2999 typedef enum {
3000 EXT4_IGET_NORMAL = 0,
3001 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */
3002 EXT4_IGET_HANDLE = 0x0002 /* Inode # is from a handle */
3003 } ext4_iget_flags;
3004
3005 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
3006 ext4_iget_flags flags, const char *function,
3007 unsigned int line);
3008
3009 #define ext4_iget(sb, ino, flags) \
3010 __ext4_iget((sb), (ino), (flags), __func__, __LINE__)
3011
3012 extern int ext4_write_inode(struct inode *, struct writeback_control *);
3013 extern int ext4_setattr(struct user_namespace *, struct dentry *,
3014 struct iattr *);
3015 extern int ext4_getattr(struct user_namespace *, const struct path *,
3016 struct kstat *, u32, unsigned int);
3017 extern void ext4_evict_inode(struct inode *);
3018 extern void ext4_clear_inode(struct inode *);
3019 extern int ext4_file_getattr(struct user_namespace *, const struct path *,
3020 struct kstat *, u32, unsigned int);
3021 extern int ext4_sync_inode(handle_t *, struct inode *);
3022 extern void ext4_dirty_inode(struct inode *, int);
3023 extern int ext4_change_inode_journal_flag(struct inode *, int);
3024 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
3025 extern int ext4_get_fc_inode_loc(struct super_block *sb, unsigned long ino,
3026 struct ext4_iloc *iloc);
3027 extern int ext4_inode_attach_jinode(struct inode *inode);
3028 extern int ext4_can_truncate(struct inode *inode);
3029 extern int ext4_truncate(struct inode *);
3030 extern int ext4_break_layouts(struct inode *);
3031 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
3032 extern void ext4_set_inode_flags(struct inode *, bool init);
3033 extern int ext4_alloc_da_blocks(struct inode *inode);
3034 extern void ext4_set_aops(struct inode *inode);
3035 extern int ext4_writepage_trans_blocks(struct inode *);
3036 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
3037 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
3038 loff_t lstart, loff_t lend);
3039 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf);
3040 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
3041 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
3042 extern void ext4_da_release_space(struct inode *inode, int to_free);
3043 extern void ext4_da_update_reserve_space(struct inode *inode,
3044 int used, int quota_claim);
3045 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
3046 ext4_fsblk_t pblk, ext4_lblk_t len);
3047
3048 /* indirect.c */
3049 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
3050 struct ext4_map_blocks *map, int flags);
3051 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
3052 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
3053 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
3054 ext4_lblk_t start, ext4_lblk_t end);
3055
3056 /* ioctl.c */
3057 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
3058 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
3059 int ext4_fileattr_set(struct user_namespace *mnt_userns,
3060 struct dentry *dentry, struct fileattr *fa);
3061 int ext4_fileattr_get(struct dentry *dentry, struct fileattr *fa);
3062 extern void ext4_reset_inode_seed(struct inode *inode);
3063
3064 /* migrate.c */
3065 extern int ext4_ext_migrate(struct inode *);
3066 extern int ext4_ind_migrate(struct inode *inode);
3067
3068 /* namei.c */
3069 extern int ext4_init_new_dir(handle_t *handle, struct inode *dir,
3070 struct inode *inode);
3071 extern int ext4_dirblock_csum_verify(struct inode *inode,
3072 struct buffer_head *bh);
3073 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
3074 __u32 start_minor_hash, __u32 *next_hash);
3075 extern int ext4_search_dir(struct buffer_head *bh,
3076 char *search_buf,
3077 int buf_size,
3078 struct inode *dir,
3079 struct ext4_filename *fname,
3080 unsigned int offset,
3081 struct ext4_dir_entry_2 **res_dir);
3082 extern int ext4_generic_delete_entry(struct inode *dir,
3083 struct ext4_dir_entry_2 *de_del,
3084 struct buffer_head *bh,
3085 void *entry_buf,
3086 int buf_size,
3087 int csum_size);
3088 extern bool ext4_empty_dir(struct inode *inode);
3089
3090 /* resize.c */
3091 extern void ext4_kvfree_array_rcu(void *to_free);
3092 extern int ext4_group_add(struct super_block *sb,
3093 struct ext4_new_group_data *input);
3094 extern int ext4_group_extend(struct super_block *sb,
3095 struct ext4_super_block *es,
3096 ext4_fsblk_t n_blocks_count);
3097 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
3098
3099 /* super.c */
3100 extern struct buffer_head *ext4_sb_bread(struct super_block *sb,
3101 sector_t block, int op_flags);
3102 extern struct buffer_head *ext4_sb_bread_unmovable(struct super_block *sb,
3103 sector_t block);
3104 extern void ext4_read_bh_nowait(struct buffer_head *bh, int op_flags,
3105 bh_end_io_t *end_io);
3106 extern int ext4_read_bh(struct buffer_head *bh, int op_flags,
3107 bh_end_io_t *end_io);
3108 extern int ext4_read_bh_lock(struct buffer_head *bh, int op_flags, bool wait);
3109 extern void ext4_sb_breadahead_unmovable(struct super_block *sb, sector_t block);
3110 extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
3111 extern int ext4_calculate_overhead(struct super_block *sb);
3112 extern void ext4_superblock_csum_set(struct super_block *sb);
3113 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
3114 ext4_group_t ngroup);
3115 extern const char *ext4_decode_error(struct super_block *sb, int errno,
3116 char nbuf[16]);
3117 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb,
3118 ext4_group_t block_group,
3119 unsigned int flags);
3120
3121 extern __printf(7, 8)
3122 void __ext4_error(struct super_block *, const char *, unsigned int, bool,
3123 int, __u64, const char *, ...);
3124 extern __printf(6, 7)
3125 void __ext4_error_inode(struct inode *, const char *, unsigned int,
3126 ext4_fsblk_t, int, const char *, ...);
3127 extern __printf(5, 6)
3128 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
3129 const char *, ...);
3130 extern void __ext4_std_error(struct super_block *, const char *,
3131 unsigned int, int);
3132 extern __printf(4, 5)
3133 void __ext4_warning(struct super_block *, const char *, unsigned int,
3134 const char *, ...);
3135 extern __printf(4, 5)
3136 void __ext4_warning_inode(const struct inode *inode, const char *function,
3137 unsigned int line, const char *fmt, ...);
3138 extern __printf(3, 4)
3139 void __ext4_msg(struct super_block *, const char *, const char *, ...);
3140 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
3141 const char *, unsigned int, const char *);
3142 extern __printf(7, 8)
3143 void __ext4_grp_locked_error(const char *, unsigned int,
3144 struct super_block *, ext4_group_t,
3145 unsigned long, ext4_fsblk_t,
3146 const char *, ...);
3147
3148 #define EXT4_ERROR_INODE(inode, fmt, a...) \
3149 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
3150
3151 #define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...) \
3152 __ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a)
3153
3154 #define ext4_error_inode_block(inode, block, err, fmt, a...) \
3155 __ext4_error_inode((inode), __func__, __LINE__, (block), (err), \
3156 (fmt), ## a)
3157
3158 #define EXT4_ERROR_FILE(file, block, fmt, a...) \
3159 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
3160
3161 #define ext4_abort(sb, err, fmt, a...) \
3162 __ext4_error((sb), __func__, __LINE__, true, (err), 0, (fmt), ## a)
3163
3164 #ifdef CONFIG_PRINTK
3165
3166 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
3167 __ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__)
3168 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \
3169 __ext4_error_inode((inode), (func), (line), (block), \
3170 (err), (fmt), ##__VA_ARGS__)
3171 #define ext4_error_file(file, func, line, block, fmt, ...) \
3172 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
3173 #define ext4_error(sb, fmt, ...) \
3174 __ext4_error((sb), __func__, __LINE__, false, 0, 0, (fmt), \
3175 ##__VA_ARGS__)
3176 #define ext4_error_err(sb, err, fmt, ...) \
3177 __ext4_error((sb), __func__, __LINE__, false, (err), 0, (fmt), \
3178 ##__VA_ARGS__)
3179 #define ext4_warning(sb, fmt, ...) \
3180 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
3181 #define ext4_warning_inode(inode, fmt, ...) \
3182 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
3183 #define ext4_msg(sb, level, fmt, ...) \
3184 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
3185 #define dump_mmp_msg(sb, mmp, msg) \
3186 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
3187 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
3188 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
3189 fmt, ##__VA_ARGS__)
3190
3191 #else
3192
3193 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
3194 do { \
3195 no_printk(fmt, ##__VA_ARGS__); \
3196 __ext4_error_inode(inode, "", 0, block, 0, " "); \
3197 } while (0)
3198 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \
3199 do { \
3200 no_printk(fmt, ##__VA_ARGS__); \
3201 __ext4_error_inode(inode, "", 0, block, err, " "); \
3202 } while (0)
3203 #define ext4_error_file(file, func, line, block, fmt, ...) \
3204 do { \
3205 no_printk(fmt, ##__VA_ARGS__); \
3206 __ext4_error_file(file, "", 0, block, " "); \
3207 } while (0)
3208 #define ext4_error(sb, fmt, ...) \
3209 do { \
3210 no_printk(fmt, ##__VA_ARGS__); \
3211 __ext4_error(sb, "", 0, false, 0, 0, " "); \
3212 } while (0)
3213 #define ext4_error_err(sb, err, fmt, ...) \
3214 do { \
3215 no_printk(fmt, ##__VA_ARGS__); \
3216 __ext4_error(sb, "", 0, false, err, 0, " "); \
3217 } while (0)
3218 #define ext4_warning(sb, fmt, ...) \
3219 do { \
3220 no_printk(fmt, ##__VA_ARGS__); \
3221 __ext4_warning(sb, "", 0, " "); \
3222 } while (0)
3223 #define ext4_warning_inode(inode, fmt, ...) \
3224 do { \
3225 no_printk(fmt, ##__VA_ARGS__); \
3226 __ext4_warning_inode(inode, "", 0, " "); \
3227 } while (0)
3228 #define ext4_msg(sb, level, fmt, ...) \
3229 do { \
3230 no_printk(fmt, ##__VA_ARGS__); \
3231 __ext4_msg(sb, "", " "); \
3232 } while (0)
3233 #define dump_mmp_msg(sb, mmp, msg) \
3234 __dump_mmp_msg(sb, mmp, "", 0, "")
3235 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
3236 do { \
3237 no_printk(fmt, ##__VA_ARGS__); \
3238 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
3239 } while (0)
3240
3241 #endif
3242
3243 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
3244 struct ext4_group_desc *bg);
3245 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
3246 struct ext4_group_desc *bg);
3247 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
3248 struct ext4_group_desc *bg);
3249 extern __u32 ext4_free_group_clusters(struct super_block *sb,
3250 struct ext4_group_desc *bg);
3251 extern __u32 ext4_free_inodes_count(struct super_block *sb,
3252 struct ext4_group_desc *bg);
3253 extern __u32 ext4_used_dirs_count(struct super_block *sb,
3254 struct ext4_group_desc *bg);
3255 extern __u32 ext4_itable_unused_count(struct super_block *sb,
3256 struct ext4_group_desc *bg);
3257 extern void ext4_block_bitmap_set(struct super_block *sb,
3258 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3259 extern void ext4_inode_bitmap_set(struct super_block *sb,
3260 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3261 extern void ext4_inode_table_set(struct super_block *sb,
3262 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3263 extern void ext4_free_group_clusters_set(struct super_block *sb,
3264 struct ext4_group_desc *bg,
3265 __u32 count);
3266 extern void ext4_free_inodes_set(struct super_block *sb,
3267 struct ext4_group_desc *bg, __u32 count);
3268 extern void ext4_used_dirs_set(struct super_block *sb,
3269 struct ext4_group_desc *bg, __u32 count);
3270 extern void ext4_itable_unused_set(struct super_block *sb,
3271 struct ext4_group_desc *bg, __u32 count);
3272 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
3273 struct ext4_group_desc *gdp);
3274 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
3275 struct ext4_group_desc *gdp);
3276 extern int ext4_register_li_request(struct super_block *sb,
3277 ext4_group_t first_not_zeroed);
3278
3279 static inline int ext4_has_metadata_csum(struct super_block *sb)
3280 {
3281 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
3282 !EXT4_SB(sb)->s_chksum_driver);
3283
3284 return ext4_has_feature_metadata_csum(sb) &&
3285 (EXT4_SB(sb)->s_chksum_driver != NULL);
3286 }
3287
3288 static inline int ext4_has_group_desc_csum(struct super_block *sb)
3289 {
3290 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb);
3291 }
3292
3293 #define ext4_read_incompat_64bit_val(es, name) \
3294 (((es)->s_feature_incompat & cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT) \
3295 ? (ext4_fsblk_t)le32_to_cpu(es->name##_hi) << 32 : 0) | \
3296 le32_to_cpu(es->name##_lo))
3297
3298 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
3299 {
3300 return ext4_read_incompat_64bit_val(es, s_blocks_count);
3301 }
3302
3303 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
3304 {
3305 return ext4_read_incompat_64bit_val(es, s_r_blocks_count);
3306 }
3307
3308 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
3309 {
3310 return ext4_read_incompat_64bit_val(es, s_free_blocks_count);
3311 }
3312
3313 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
3314 ext4_fsblk_t blk)
3315 {
3316 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
3317 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
3318 }
3319
3320 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
3321 ext4_fsblk_t blk)
3322 {
3323 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
3324 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
3325 }
3326
3327 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
3328 ext4_fsblk_t blk)
3329 {
3330 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
3331 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
3332 }
3333
3334 static inline loff_t ext4_isize(struct super_block *sb,
3335 struct ext4_inode *raw_inode)
3336 {
3337 if (ext4_has_feature_largedir(sb) ||
3338 S_ISREG(le16_to_cpu(raw_inode->i_mode)))
3339 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
3340 le32_to_cpu(raw_inode->i_size_lo);
3341
3342 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
3343 }
3344
3345 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
3346 {
3347 raw_inode->i_size_lo = cpu_to_le32(i_size);
3348 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
3349 }
3350
3351 static inline
3352 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
3353 ext4_group_t group)
3354 {
3355 struct ext4_group_info **grp_info;
3356 long indexv, indexh;
3357 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
3358 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
3359 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
3360 grp_info = sbi_array_rcu_deref(EXT4_SB(sb), s_group_info, indexv);
3361 return grp_info[indexh];
3362 }
3363
3364 /*
3365 * Reading s_groups_count requires using smp_rmb() afterwards. See
3366 * the locking protocol documented in the comments of ext4_group_add()
3367 * in resize.c
3368 */
3369 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
3370 {
3371 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
3372
3373 smp_rmb();
3374 return ngroups;
3375 }
3376
3377 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
3378 ext4_group_t block_group)
3379 {
3380 return block_group >> sbi->s_log_groups_per_flex;
3381 }
3382
3383 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
3384 {
3385 return 1 << sbi->s_log_groups_per_flex;
3386 }
3387
3388 #define ext4_std_error(sb, errno) \
3389 do { \
3390 if ((errno)) \
3391 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
3392 } while (0)
3393
3394 #ifdef CONFIG_SMP
3395 /* Each CPU can accumulate percpu_counter_batch clusters in their local
3396 * counters. So we need to make sure we have free clusters more
3397 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
3398 */
3399 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
3400 #else
3401 #define EXT4_FREECLUSTERS_WATERMARK 0
3402 #endif
3403
3404 /* Update i_disksize. Requires i_mutex to avoid races with truncate */
3405 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
3406 {
3407 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
3408 !inode_is_locked(inode));
3409 down_write(&EXT4_I(inode)->i_data_sem);
3410 if (newsize > EXT4_I(inode)->i_disksize)
3411 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize);
3412 up_write(&EXT4_I(inode)->i_data_sem);
3413 }
3414
3415 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
3416 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
3417 {
3418 int changed = 0;
3419
3420 if (newsize > inode->i_size) {
3421 i_size_write(inode, newsize);
3422 changed = 1;
3423 }
3424 if (newsize > EXT4_I(inode)->i_disksize) {
3425 ext4_update_i_disksize(inode, newsize);
3426 changed |= 2;
3427 }
3428 return changed;
3429 }
3430
3431 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
3432 loff_t len);
3433
3434 struct ext4_group_info {
3435 unsigned long bb_state;
3436 #ifdef AGGRESSIVE_CHECK
3437 unsigned long bb_check_counter;
3438 #endif
3439 struct rb_root bb_free_root;
3440 ext4_grpblk_t bb_first_free; /* first free block */
3441 ext4_grpblk_t bb_free; /* total free blocks */
3442 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
3443 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
3444 ext4_group_t bb_group; /* Group number */
3445 struct list_head bb_prealloc_list;
3446 #ifdef DOUBLE_CHECK
3447 void *bb_bitmap;
3448 #endif
3449 struct rw_semaphore alloc_sem;
3450 struct rb_node bb_avg_fragment_size_rb;
3451 struct list_head bb_largest_free_order_node;
3452 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
3453 * regions, index is order.
3454 * bb_counters[3] = 5 means
3455 * 5 free 8-block regions. */
3456 };
3457
3458 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
3459 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
3460 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
3461 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
3462 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \
3463 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT)
3464 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \
3465 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT)
3466 #define EXT4_GROUP_INFO_BBITMAP_READ_BIT 4
3467
3468 #define EXT4_MB_GRP_NEED_INIT(grp) \
3469 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
3470 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
3471 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3472 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
3473 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3474
3475 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
3476 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3477 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
3478 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3479 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
3480 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3481 #define EXT4_MB_GRP_TEST_AND_SET_READ(grp) \
3482 (test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state)))
3483
3484 #define EXT4_MAX_CONTENTION 8
3485 #define EXT4_CONTENTION_THRESHOLD 2
3486
3487 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
3488 ext4_group_t group)
3489 {
3490 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
3491 }
3492
3493 /*
3494 * Returns true if the filesystem is busy enough that attempts to
3495 * access the block group locks has run into contention.
3496 */
3497 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
3498 {
3499 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
3500 }
3501
3502 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
3503 {
3504 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
3505 if (spin_trylock(lock))
3506 /*
3507 * We're able to grab the lock right away, so drop the
3508 * lock contention counter.
3509 */
3510 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
3511 else {
3512 /*
3513 * The lock is busy, so bump the contention counter,
3514 * and then wait on the spin lock.
3515 */
3516 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
3517 EXT4_MAX_CONTENTION);
3518 spin_lock(lock);
3519 }
3520 }
3521
3522 static inline void ext4_unlock_group(struct super_block *sb,
3523 ext4_group_t group)
3524 {
3525 spin_unlock(ext4_group_lock_ptr(sb, group));
3526 }
3527
3528 #ifdef CONFIG_QUOTA
3529 static inline bool ext4_quota_capable(struct super_block *sb)
3530 {
3531 return (test_opt(sb, QUOTA) || ext4_has_feature_quota(sb));
3532 }
3533
3534 static inline bool ext4_is_quota_journalled(struct super_block *sb)
3535 {
3536 struct ext4_sb_info *sbi = EXT4_SB(sb);
3537
3538 return (ext4_has_feature_quota(sb) ||
3539 sbi->s_qf_names[USRQUOTA] || sbi->s_qf_names[GRPQUOTA]);
3540 }
3541 int ext4_enable_quotas(struct super_block *sb);
3542 #endif
3543
3544 /*
3545 * Block validity checking
3546 */
3547 #define ext4_check_indirect_blockref(inode, bh) \
3548 ext4_check_blockref(__func__, __LINE__, inode, \
3549 (__le32 *)(bh)->b_data, \
3550 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
3551
3552 #define ext4_ind_check_inode(inode) \
3553 ext4_check_blockref(__func__, __LINE__, inode, \
3554 EXT4_I(inode)->i_data, \
3555 EXT4_NDIR_BLOCKS)
3556
3557 /*
3558 * Inodes and files operations
3559 */
3560
3561 /* dir.c */
3562 extern const struct file_operations ext4_dir_operations;
3563
3564 /* file.c */
3565 extern const struct inode_operations ext4_file_inode_operations;
3566 extern const struct file_operations ext4_file_operations;
3567 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
3568
3569 /* inline.c */
3570 extern int ext4_get_max_inline_size(struct inode *inode);
3571 extern int ext4_find_inline_data_nolock(struct inode *inode);
3572 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
3573 unsigned int len);
3574 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
3575
3576 extern int ext4_readpage_inline(struct inode *inode, struct page *page);
3577 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3578 struct inode *inode,
3579 loff_t pos, unsigned len,
3580 unsigned flags,
3581 struct page **pagep);
3582 extern int ext4_write_inline_data_end(struct inode *inode,
3583 loff_t pos, unsigned len,
3584 unsigned copied,
3585 struct page *page);
3586 extern struct buffer_head *
3587 ext4_journalled_write_inline_data(struct inode *inode,
3588 unsigned len,
3589 struct page *page);
3590 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
3591 struct inode *inode,
3592 loff_t pos, unsigned len,
3593 unsigned flags,
3594 struct page **pagep,
3595 void **fsdata);
3596 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
3597 unsigned len, unsigned copied,
3598 struct page *page);
3599 extern int ext4_try_add_inline_entry(handle_t *handle,
3600 struct ext4_filename *fname,
3601 struct inode *dir, struct inode *inode);
3602 extern int ext4_try_create_inline_dir(handle_t *handle,
3603 struct inode *parent,
3604 struct inode *inode);
3605 extern int ext4_read_inline_dir(struct file *filp,
3606 struct dir_context *ctx,
3607 int *has_inline_data);
3608 extern int ext4_inlinedir_to_tree(struct file *dir_file,
3609 struct inode *dir, ext4_lblk_t block,
3610 struct dx_hash_info *hinfo,
3611 __u32 start_hash, __u32 start_minor_hash,
3612 int *has_inline_data);
3613 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3614 struct ext4_filename *fname,
3615 struct ext4_dir_entry_2 **res_dir,
3616 int *has_inline_data);
3617 extern int ext4_delete_inline_entry(handle_t *handle,
3618 struct inode *dir,
3619 struct ext4_dir_entry_2 *de_del,
3620 struct buffer_head *bh,
3621 int *has_inline_data);
3622 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3623 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3624 struct ext4_dir_entry_2 **parent_de,
3625 int *retval);
3626 extern int ext4_inline_data_fiemap(struct inode *inode,
3627 struct fiemap_extent_info *fieinfo,
3628 int *has_inline, __u64 start, __u64 len);
3629
3630 struct iomap;
3631 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap);
3632
3633 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3634
3635 extern int ext4_convert_inline_data(struct inode *inode);
3636
3637 static inline int ext4_has_inline_data(struct inode *inode)
3638 {
3639 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3640 EXT4_I(inode)->i_inline_off;
3641 }
3642
3643 /* namei.c */
3644 extern const struct inode_operations ext4_dir_inode_operations;
3645 extern const struct inode_operations ext4_special_inode_operations;
3646 extern struct dentry *ext4_get_parent(struct dentry *child);
3647 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3648 struct ext4_dir_entry_2 *de,
3649 int blocksize, int csum_size,
3650 unsigned int parent_ino, int dotdot_real_len);
3651 extern void ext4_initialize_dirent_tail(struct buffer_head *bh,
3652 unsigned int blocksize);
3653 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode,
3654 struct buffer_head *bh);
3655 extern int __ext4_unlink(handle_t *handle, struct inode *dir, const struct qstr *d_name,
3656 struct inode *inode);
3657 extern int __ext4_link(struct inode *dir, struct inode *inode,
3658 struct dentry *dentry);
3659
3660 #define S_SHIFT 12
3661 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = {
3662 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
3663 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
3664 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
3665 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
3666 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
3667 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
3668 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
3669 };
3670
3671 static inline void ext4_set_de_type(struct super_block *sb,
3672 struct ext4_dir_entry_2 *de,
3673 umode_t mode) {
3674 if (ext4_has_feature_filetype(sb))
3675 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3676 }
3677
3678 /* readpages.c */
3679 extern int ext4_mpage_readpages(struct inode *inode,
3680 struct readahead_control *rac, struct page *page);
3681 extern int __init ext4_init_post_read_processing(void);
3682 extern void ext4_exit_post_read_processing(void);
3683
3684 /* symlink.c */
3685 extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3686 extern const struct inode_operations ext4_symlink_inode_operations;
3687 extern const struct inode_operations ext4_fast_symlink_inode_operations;
3688
3689 /* sysfs.c */
3690 extern void ext4_notify_error_sysfs(struct ext4_sb_info *sbi);
3691 extern int ext4_register_sysfs(struct super_block *sb);
3692 extern void ext4_unregister_sysfs(struct super_block *sb);
3693 extern int __init ext4_init_sysfs(void);
3694 extern void ext4_exit_sysfs(void);
3695
3696 /* block_validity */
3697 extern void ext4_release_system_zone(struct super_block *sb);
3698 extern int ext4_setup_system_zone(struct super_block *sb);
3699 extern int __init ext4_init_system_zone(void);
3700 extern void ext4_exit_system_zone(void);
3701 extern int ext4_inode_block_valid(struct inode *inode,
3702 ext4_fsblk_t start_blk,
3703 unsigned int count);
3704 extern int ext4_check_blockref(const char *, unsigned int,
3705 struct inode *, __le32 *, unsigned int);
3706
3707 /* extents.c */
3708 struct ext4_ext_path;
3709 struct ext4_extent;
3710
3711 /*
3712 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3713 * __le32.
3714 */
3715 #define EXT_MAX_BLOCKS 0xffffffff
3716
3717 extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode);
3718 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3719 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3720 struct ext4_map_blocks *map, int flags);
3721 extern int ext4_ext_truncate(handle_t *, struct inode *);
3722 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3723 ext4_lblk_t end);
3724 extern void ext4_ext_init(struct super_block *);
3725 extern void ext4_ext_release(struct super_block *);
3726 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3727 loff_t len);
3728 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3729 loff_t offset, ssize_t len);
3730 extern int ext4_convert_unwritten_io_end_vec(handle_t *handle,
3731 ext4_io_end_t *io_end);
3732 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3733 struct ext4_map_blocks *map, int flags);
3734 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3735 int num,
3736 struct ext4_ext_path *path);
3737 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3738 struct ext4_ext_path **,
3739 struct ext4_extent *, int);
3740 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3741 struct ext4_ext_path **,
3742 int flags);
3743 extern void ext4_ext_drop_refs(struct ext4_ext_path *);
3744 extern int ext4_ext_check_inode(struct inode *inode);
3745 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3746 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3747 __u64 start, __u64 len);
3748 extern int ext4_get_es_cache(struct inode *inode,
3749 struct fiemap_extent_info *fieinfo,
3750 __u64 start, __u64 len);
3751 extern int ext4_ext_precache(struct inode *inode);
3752 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3753 struct inode *inode2, ext4_lblk_t lblk1,
3754 ext4_lblk_t lblk2, ext4_lblk_t count,
3755 int mark_unwritten,int *err);
3756 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu);
3757 extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode,
3758 int check_cred, int restart_cred,
3759 int revoke_cred);
3760 extern void ext4_ext_replay_shrink_inode(struct inode *inode, ext4_lblk_t end);
3761 extern int ext4_ext_replay_set_iblocks(struct inode *inode);
3762 extern int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start,
3763 int len, int unwritten, ext4_fsblk_t pblk);
3764 extern int ext4_ext_clear_bb(struct inode *inode);
3765
3766
3767 /* move_extent.c */
3768 extern void ext4_double_down_write_data_sem(struct inode *first,
3769 struct inode *second);
3770 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3771 struct inode *donor_inode);
3772 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3773 __u64 start_orig, __u64 start_donor,
3774 __u64 len, __u64 *moved_len);
3775
3776 /* page-io.c */
3777 extern int __init ext4_init_pageio(void);
3778 extern void ext4_exit_pageio(void);
3779 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3780 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3781 extern int ext4_put_io_end(ext4_io_end_t *io_end);
3782 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3783 extern void ext4_io_submit_init(struct ext4_io_submit *io,
3784 struct writeback_control *wbc);
3785 extern void ext4_end_io_rsv_work(struct work_struct *work);
3786 extern void ext4_io_submit(struct ext4_io_submit *io);
3787 extern int ext4_bio_write_page(struct ext4_io_submit *io,
3788 struct page *page,
3789 int len,
3790 bool keep_towrite);
3791 extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end);
3792 extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end);
3793
3794 /* mmp.c */
3795 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3796
3797 /* mmp.c */
3798 extern void ext4_stop_mmpd(struct ext4_sb_info *sbi);
3799
3800 /* verity.c */
3801 extern const struct fsverity_operations ext4_verityops;
3802
3803 /* orphan.c */
3804 extern int ext4_orphan_add(handle_t *, struct inode *);
3805 extern int ext4_orphan_del(handle_t *, struct inode *);
3806 extern void ext4_orphan_cleanup(struct super_block *sb,
3807 struct ext4_super_block *es);
3808 extern void ext4_release_orphan_info(struct super_block *sb);
3809 extern int ext4_init_orphan_info(struct super_block *sb);
3810 extern int ext4_orphan_file_empty(struct super_block *sb);
3811 extern void ext4_orphan_file_block_trigger(
3812 struct jbd2_buffer_trigger_type *triggers,
3813 struct buffer_head *bh,
3814 void *data, size_t size);
3815
3816 /*
3817 * Add new method to test whether block and inode bitmaps are properly
3818 * initialized. With uninit_bg reading the block from disk is not enough
3819 * to mark the bitmap uptodate. We need to also zero-out the bitmap
3820 */
3821 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3822
3823 static inline int bitmap_uptodate(struct buffer_head *bh)
3824 {
3825 return (buffer_uptodate(bh) &&
3826 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3827 }
3828 static inline void set_bitmap_uptodate(struct buffer_head *bh)
3829 {
3830 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3831 }
3832
3833 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
3834
3835 /* For ioend & aio unwritten conversion wait queues */
3836 #define EXT4_WQ_HASH_SZ 37
3837 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
3838 EXT4_WQ_HASH_SZ])
3839 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3840
3841 extern int ext4_resize_begin(struct super_block *sb);
3842 extern void ext4_resize_end(struct super_block *sb);
3843
3844 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
3845 struct ext4_io_end *io_end)
3846 {
3847 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
3848 io_end->flag |= EXT4_IO_END_UNWRITTEN;
3849 atomic_inc(&EXT4_I(inode)->i_unwritten);
3850 }
3851 }
3852
3853 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3854 {
3855 struct inode *inode = io_end->inode;
3856
3857 if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
3858 io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3859 /* Wake up anyone waiting on unwritten extent conversion */
3860 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
3861 wake_up_all(ext4_ioend_wq(inode));
3862 }
3863 }
3864
3865 extern const struct iomap_ops ext4_iomap_ops;
3866 extern const struct iomap_ops ext4_iomap_overwrite_ops;
3867 extern const struct iomap_ops ext4_iomap_report_ops;
3868
3869 static inline int ext4_buffer_uptodate(struct buffer_head *bh)
3870 {
3871 /*
3872 * If the buffer has the write error flag, we have failed
3873 * to write out data in the block. In this case, we don't
3874 * have to read the block because we may read the old data
3875 * successfully.
3876 */
3877 if (buffer_write_io_error(bh))
3878 set_buffer_uptodate(bh);
3879 return buffer_uptodate(bh);
3880 }
3881
3882 #endif /* __KERNEL__ */
3883
3884 #define EFSBADCRC EBADMSG /* Bad CRC detected */
3885 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
3886
3887 #endif /* _EXT4_H */