1 // SPDX-License-Identifier: GPL-2.0
2 /*
3 * ext4.h
4 *
5 * Copyright (C) 1992, 1993, 1994, 1995
6 * Remy Card (card@masi.ibp.fr)
7 * Laboratoire MASI - Institut Blaise Pascal
8 * Universite Pierre et Marie Curie (Paris VI)
9 *
10 * from
11 *
12 * linux/include/linux/minix_fs.h
13 *
14 * Copyright (C) 1991, 1992 Linus Torvalds
15 */
16
17 #ifndef _EXT4_H
18 #define _EXT4_H
19
20 #include <linux/refcount.h>
21 #include <linux/types.h>
22 #include <linux/blkdev.h>
23 #include <linux/magic.h>
24 #include <linux/jbd2.h>
25 #include <linux/quota.h>
26 #include <linux/rwsem.h>
27 #include <linux/rbtree.h>
28 #include <linux/seqlock.h>
29 #include <linux/mutex.h>
30 #include <linux/timer.h>
31 #include <linux/wait.h>
32 #include <linux/sched/signal.h>
33 #include <linux/blockgroup_lock.h>
34 #include <linux/percpu_counter.h>
35 #include <linux/ratelimit.h>
36 #include <linux/crc32c.h>
37 #include <linux/falloc.h>
38 #include <linux/percpu-rwsem.h>
39 #include <linux/fiemap.h>
40 #ifdef __KERNEL__
41 #include <linux/compat.h>
42 #endif
43 #include <uapi/linux/ext4.h>
44
45 #include <linux/fscrypt.h>
46 #include <linux/fsverity.h>
47
48 #include <linux/compiler.h>
49
50 /*
51 * The fourth extended filesystem constants/structures
52 */
53
54 /*
55 * with AGGRESSIVE_CHECK allocator runs consistency checks over
56 * structures. these checks slow things down a lot
57 */
58 #define AGGRESSIVE_CHECK__
59
60 /*
61 * with DOUBLE_CHECK defined mballoc creates persistent in-core
62 * bitmaps, maintains and uses them to check for double allocations
63 */
64 #define DOUBLE_CHECK__
65
66 /*
67 * Define EXT4FS_DEBUG to produce debug messages
68 */
69 #undef EXT4FS_DEBUG
70
71 /*
72 * Debug code
73 */
74 #ifdef EXT4FS_DEBUG
75 #define ext4_debug(f, a...) \
76 do { \
77 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
78 __FILE__, __LINE__, __func__); \
79 printk(KERN_DEBUG f, ## a); \
80 } while (0)
81 #else
82 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
83 #endif
84
85 /*
86 * Turn on EXT_DEBUG to enable ext4_ext_show_path/leaf/move in extents.c
87 */
88 #define EXT_DEBUG__
89
90 /*
91 * Dynamic printk for controlled extents debugging.
92 */
93 #ifdef CONFIG_EXT4_DEBUG
94 #define ext_debug(ino, fmt, ...) \
95 pr_debug("[%s/%d] EXT4-fs (%s): ino %lu: (%s, %d): %s:" fmt, \
96 current->comm, task_pid_nr(current), \
97 ino->i_sb->s_id, ino->i_ino, __FILE__, __LINE__, \
98 __func__, ##__VA_ARGS__)
99 #else
100 #define ext_debug(ino, fmt, ...) no_printk(fmt, ##__VA_ARGS__)
101 #endif
102
103 #define ASSERT(assert) \
104 do { \
105 if (unlikely(!(assert))) { \
106 printk(KERN_EMERG \
107 "Assertion failure in %s() at %s:%d: '%s'\n", \
108 __func__, __FILE__, __LINE__, #assert); \
109 BUG(); \
110 } \
111 } while (0)
112
113 /* data type for block offset of block group */
114 typedef int ext4_grpblk_t;
115
116 /* data type for filesystem-wide blocks number */
117 typedef unsigned long long ext4_fsblk_t;
118
119 /* data type for file logical block number */
120 typedef __u32 ext4_lblk_t;
121
122 /* data type for block group number */
123 typedef unsigned int ext4_group_t;
124
125 enum SHIFT_DIRECTION {
126 SHIFT_LEFT = 0,
127 SHIFT_RIGHT,
128 };
129
130 /*
131 * For each criteria, mballoc has slightly different way of finding
132 * the required blocks nad usually, higher the criteria the slower the
133 * allocation. We start at lower criterias and keep falling back to
134 * higher ones if we are not able to find any blocks. Lower (earlier)
135 * criteria are faster.
136 */
137 enum criteria {
138 /*
139 * Used when number of blocks needed is a power of 2. This
140 * doesn't trigger any disk IO except prefetch and is the
141 * fastest criteria.
142 */
143 CR_POWER2_ALIGNED,
144
145 /*
146 * Tries to lookup in-memory data structures to find the most
147 * suitable group that satisfies goal request. No disk IO
148 * except block prefetch.
149 */
150 CR_GOAL_LEN_FAST,
151
152 /*
153 * Same as CR_GOAL_LEN_FAST but is allowed to reduce the goal
154 * length to the best available length for faster allocation.
155 */
156 CR_BEST_AVAIL_LEN,
157
158 /*
159 * Reads each block group sequentially, performing disk IO if
160 * necessary, to find suitable block group. Tries to
161 * allocate goal length but might trim the request if nothing
162 * is found after enough tries.
163 */
164 CR_GOAL_LEN_SLOW,
165
166 /*
167 * Finds the first free set of blocks and allocates
168 * those. This is only used in rare cases when
169 * CR_GOAL_LEN_SLOW also fails to allocate anything.
170 */
171 CR_ANY_FREE,
172
173 /*
174 * Number of criterias defined.
175 */
176 EXT4_MB_NUM_CRS
177 };
178
179 /*
180 * Flags used in mballoc's allocation_context flags field.
181 *
182 * Also used to show what's going on for debugging purposes when the
183 * flag field is exported via the traceport interface
184 */
185
186 /* prefer goal again. length */
187 #define EXT4_MB_HINT_MERGE 0x0001
188 /* first blocks in the file */
189 #define EXT4_MB_HINT_FIRST 0x0008
190 /* data is being allocated */
191 #define EXT4_MB_HINT_DATA 0x0020
192 /* don't preallocate (for tails) */
193 #define EXT4_MB_HINT_NOPREALLOC 0x0040
194 /* allocate for locality group */
195 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080
196 /* allocate goal blocks or none */
197 #define EXT4_MB_HINT_GOAL_ONLY 0x0100
198 /* goal is meaningful */
199 #define EXT4_MB_HINT_TRY_GOAL 0x0200
200 /* blocks already pre-reserved by delayed allocation */
201 #define EXT4_MB_DELALLOC_RESERVED 0x0400
202 /* We are doing stream allocation */
203 #define EXT4_MB_STREAM_ALLOC 0x0800
204 /* Use reserved root blocks if needed */
205 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000
206 /* Use blocks from reserved pool */
207 #define EXT4_MB_USE_RESERVED 0x2000
208 /* Do strict check for free blocks while retrying block allocation */
209 #define EXT4_MB_STRICT_CHECK 0x4000
210
211 struct ext4_allocation_request {
212 /* target inode for block we're allocating */
213 struct inode *inode;
214 /* how many blocks we want to allocate */
215 unsigned int len;
216 /* logical block in target inode */
217 ext4_lblk_t logical;
218 /* the closest logical allocated block to the left */
219 ext4_lblk_t lleft;
220 /* the closest logical allocated block to the right */
221 ext4_lblk_t lright;
222 /* phys. target (a hint) */
223 ext4_fsblk_t goal;
224 /* phys. block for the closest logical allocated block to the left */
225 ext4_fsblk_t pleft;
226 /* phys. block for the closest logical allocated block to the right */
227 ext4_fsblk_t pright;
228 /* flags. see above EXT4_MB_HINT_* */
229 unsigned int flags;
230 };
231
232 /*
233 * Logical to physical block mapping, used by ext4_map_blocks()
234 *
235 * This structure is used to pass requests into ext4_map_blocks() as
236 * well as to store the information returned by ext4_map_blocks(). It
237 * takes less room on the stack than a struct buffer_head.
238 */
239 #define EXT4_MAP_NEW BIT(BH_New)
240 #define EXT4_MAP_MAPPED BIT(BH_Mapped)
241 #define EXT4_MAP_UNWRITTEN BIT(BH_Unwritten)
242 #define EXT4_MAP_BOUNDARY BIT(BH_Boundary)
243 #define EXT4_MAP_DELAYED BIT(BH_Delay)
244 /*
245 * This is for use in ext4_map_query_blocks() for a special case where we can
246 * have a physically and logically contiguous blocks split across two leaf
247 * nodes instead of a single extent. This is required in case of atomic writes
248 * to know whether the returned extent is last in leaf. If yes, then lookup for
249 * next in leaf block in ext4_map_query_blocks_next_in_leaf().
250 * - This is never going to be added to any buffer head state.
251 * - We use the next available bit after BH_BITMAP_UPTODATE.
252 */
253 #define EXT4_MAP_QUERY_LAST_IN_LEAF BIT(BH_BITMAP_UPTODATE + 1)
254 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
255 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\
256 EXT4_MAP_DELAYED | EXT4_MAP_QUERY_LAST_IN_LEAF)
257
258 struct ext4_map_blocks {
259 ext4_fsblk_t m_pblk;
260 ext4_lblk_t m_lblk;
261 unsigned int m_len;
262 unsigned int m_flags;
263 u64 m_seq;
264 };
265
266 /*
267 * Block validity checking, system zone rbtree.
268 */
269 struct ext4_system_blocks {
270 struct rb_root root;
271 struct rcu_head rcu;
272 };
273
274 /*
275 * Flags for ext4_io_end->flags
276 */
277 #define EXT4_IO_END_UNWRITTEN 0x0001
278 #define EXT4_IO_END_FAILED 0x0002
279
280 #define EXT4_IO_END_DEFER_COMPLETION (EXT4_IO_END_UNWRITTEN | EXT4_IO_END_FAILED)
281
282 struct ext4_io_end_vec {
283 struct list_head list; /* list of io_end_vec */
284 loff_t offset; /* offset in the file */
285 ssize_t size; /* size of the extent */
286 };
287
288 /*
289 * For converting unwritten extents on a work queue. 'handle' is used for
290 * buffered writeback.
291 */
292 typedef struct ext4_io_end {
293 struct list_head list; /* per-file finished IO list */
294 handle_t *handle; /* handle reserved for extent
295 * conversion */
296 struct inode *inode; /* file being written to */
297 struct bio *bio; /* Linked list of completed
298 * bios covering the extent */
299 unsigned int flag; /* unwritten or not */
300 refcount_t count; /* reference counter */
301 struct list_head list_vec; /* list of ext4_io_end_vec */
302 } ext4_io_end_t;
303
304 struct ext4_io_submit {
305 struct writeback_control *io_wbc;
306 struct bio *io_bio;
307 ext4_io_end_t *io_end;
308 sector_t io_next_block;
309 };
310
311 /*
312 * Special inodes numbers
313 */
314 #define EXT4_BAD_INO 1 /* Bad blocks inode */
315 #define EXT4_ROOT_INO 2 /* Root inode */
316 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */
317 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
318 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
319 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
320 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
321 #define EXT4_JOURNAL_INO 8 /* Journal inode */
322
323 /* First non-reserved inode for old ext4 filesystems */
324 #define EXT4_GOOD_OLD_FIRST_INO 11
325
326 /*
327 * Maximal count of links to a file
328 */
329 #define EXT4_LINK_MAX 65000
330
331 /*
332 * Macro-instructions used to manage several block sizes
333 */
334 #define EXT4_MIN_BLOCK_SIZE 1024
335 #define EXT4_MAX_BLOCK_SIZE 65536
336 #define EXT4_MIN_BLOCK_LOG_SIZE 10
337 #define EXT4_MAX_BLOCK_LOG_SIZE 16
338 #define EXT4_MAX_CLUSTER_LOG_SIZE 30
339 #ifdef __KERNEL__
340 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
341 #else
342 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
343 #endif
344 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
345 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
346 EXT4_SB(s)->s_cluster_bits)
347 #ifdef __KERNEL__
348 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
349 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
350 #else
351 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
352 #endif
353 #ifdef __KERNEL__
354 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
355 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
356 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
357 #else
358 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
359 EXT4_GOOD_OLD_INODE_SIZE : \
360 (s)->s_inode_size)
361 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
362 EXT4_GOOD_OLD_FIRST_INO : \
363 (s)->s_first_ino)
364 #endif
365 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
366 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \
367 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \
368 blkbits))
369 #define EXT4_B_TO_LBLK(inode, offset) \
370 (round_up((offset), i_blocksize(inode)) >> (inode)->i_blkbits)
371 #define EXT4_LBLK_TO_B(inode, lblk) ((loff_t)(lblk) << (inode)->i_blkbits)
372
373 /* Translate a block number to a page index */
374 #define EXT4_LBLK_TO_PG(inode, lblk) (EXT4_LBLK_TO_B((inode), (lblk)) >> \
375 PAGE_SHIFT)
376 /* Translate a page index to a block number */
377 #define EXT4_PG_TO_LBLK(inode, pnum) (((loff_t)(pnum) << PAGE_SHIFT) >> \
378 (inode)->i_blkbits)
379 /* Translate a block number to a cluster number */
380 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
381 /* Translate a cluster number to a block number */
382 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
383 /* Translate # of blks to # of clusters */
384 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
385 (sbi)->s_cluster_bits)
386 /* Mask out the low bits to get the starting block of the cluster */
387 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
388 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
389 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
390 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
391 /* Fill in the low bits to get the last block of the cluster */
392 #define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) | \
393 ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1))
394 /* Get the cluster offset */
395 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
396 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
397 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
398 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
399
400 /*
401 * Structure of a blocks group descriptor
402 */
403 struct ext4_group_desc
404 {
405 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
406 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
407 __le32 bg_inode_table_lo; /* Inodes table block */
408 __le16 bg_free_blocks_count_lo;/* Free blocks count */
409 __le16 bg_free_inodes_count_lo;/* Free inodes count */
410 __le16 bg_used_dirs_count_lo; /* Directories count */
411 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
412 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
413 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
414 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
415 __le16 bg_itable_unused_lo; /* Unused inodes count */
416 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
417 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
418 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
419 __le32 bg_inode_table_hi; /* Inodes table block MSB */
420 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
421 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
422 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
423 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
424 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
425 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
426 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
427 __u32 bg_reserved;
428 };
429
430 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
431 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
432 sizeof(__le16))
433 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
434 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
435 sizeof(__le16))
436
437 /*
438 * Structure of a flex block group info
439 */
440
441 struct flex_groups {
442 atomic64_t free_clusters;
443 atomic_t free_inodes;
444 atomic_t used_dirs;
445 };
446
447 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
448 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
449 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
450
451 /*
452 * Macro-instructions used to manage group descriptors
453 */
454 #define EXT4_MIN_DESC_SIZE 32
455 #define EXT4_MIN_DESC_SIZE_64BIT 64
456 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
457 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
458 #ifdef __KERNEL__
459 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
460 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
461 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
462 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
463 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
464 #else
465 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
466 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
467 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
468 #endif
469
470 /*
471 * Constants relative to the data blocks
472 */
473 #define EXT4_NDIR_BLOCKS 12
474 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
475 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
476 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
477 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
478
479 /*
480 * Inode flags
481 */
482 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
483 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
484 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
485 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
486 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
487 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
488 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
489 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
490 /* Reserved for compression usage... */
491 #define EXT4_DIRTY_FL 0x00000100
492 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
493 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
494 /* nb: was previously EXT2_ECOMPR_FL */
495 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */
496 /* End compression flags --- maybe not all used */
497 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
498 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
499 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
500 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
501 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
502 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
503 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
504 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
505 #define EXT4_VERITY_FL 0x00100000 /* Verity protected inode */
506 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
507 /* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */
508
509 #define EXT4_DAX_FL 0x02000000 /* Inode is DAX */
510
511 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
512 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
513 #define EXT4_CASEFOLD_FL 0x40000000 /* Casefolded directory */
514 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
515
516 /* User modifiable flags */
517 #define EXT4_FL_USER_MODIFIABLE (EXT4_SECRM_FL | \
518 EXT4_UNRM_FL | \
519 EXT4_COMPR_FL | \
520 EXT4_SYNC_FL | \
521 EXT4_IMMUTABLE_FL | \
522 EXT4_APPEND_FL | \
523 EXT4_NODUMP_FL | \
524 EXT4_NOATIME_FL | \
525 EXT4_JOURNAL_DATA_FL | \
526 EXT4_NOTAIL_FL | \
527 EXT4_DIRSYNC_FL | \
528 EXT4_TOPDIR_FL | \
529 EXT4_EXTENTS_FL | \
530 0x00400000 /* EXT4_EOFBLOCKS_FL */ | \
531 EXT4_DAX_FL | \
532 EXT4_PROJINHERIT_FL | \
533 EXT4_CASEFOLD_FL)
534
535 /* User visible flags */
536 #define EXT4_FL_USER_VISIBLE (EXT4_FL_USER_MODIFIABLE | \
537 EXT4_DIRTY_FL | \
538 EXT4_COMPRBLK_FL | \
539 EXT4_NOCOMPR_FL | \
540 EXT4_ENCRYPT_FL | \
541 EXT4_INDEX_FL | \
542 EXT4_VERITY_FL | \
543 EXT4_INLINE_DATA_FL)
544
545 /* Flags that should be inherited by new inodes from their parent. */
546 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
547 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
548 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
549 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
550 EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL |\
551 EXT4_DAX_FL)
552
553 /* Flags that are appropriate for regular files (all but dir-specific ones). */
554 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\
555 EXT4_PROJINHERIT_FL))
556
557 /* Flags that are appropriate for non-directories/regular files. */
558 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
559
560 /* The only flags that should be swapped */
561 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL)
562
563 /* Flags which are mutually exclusive to DAX */
564 #define EXT4_DAX_MUT_EXCL (EXT4_VERITY_FL | EXT4_ENCRYPT_FL |\
565 EXT4_JOURNAL_DATA_FL | EXT4_INLINE_DATA_FL)
566
567 /* Mask out flags that are inappropriate for the given type of inode. */
ext4_mask_flags(umode_t mode,__u32 flags)568 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
569 {
570 if (S_ISDIR(mode))
571 return flags;
572 else if (S_ISREG(mode))
573 return flags & EXT4_REG_FLMASK;
574 else
575 return flags & EXT4_OTHER_FLMASK;
576 }
577
578 /*
579 * Inode flags used for atomic set/get
580 */
581 enum {
582 EXT4_INODE_SECRM = 0, /* Secure deletion */
583 EXT4_INODE_UNRM = 1, /* Undelete */
584 EXT4_INODE_COMPR = 2, /* Compress file */
585 EXT4_INODE_SYNC = 3, /* Synchronous updates */
586 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
587 EXT4_INODE_APPEND = 5, /* writes to file may only append */
588 EXT4_INODE_NODUMP = 6, /* do not dump file */
589 EXT4_INODE_NOATIME = 7, /* do not update atime */
590 /* Reserved for compression usage... */
591 EXT4_INODE_DIRTY = 8,
592 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
593 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
594 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */
595 /* End compression flags --- maybe not all used */
596 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
597 EXT4_INODE_IMAGIC = 13, /* AFS directory */
598 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
599 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
600 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
601 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
602 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
603 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
604 EXT4_INODE_VERITY = 20, /* Verity protected inode */
605 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
606 /* 22 was formerly EXT4_INODE_EOFBLOCKS */
607 EXT4_INODE_DAX = 25, /* Inode is DAX */
608 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
609 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */
610 EXT4_INODE_CASEFOLD = 30, /* Casefolded directory */
611 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
612 };
613
614 /*
615 * Since it's pretty easy to mix up bit numbers and hex values, we use a
616 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
617 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
618 * any extra space in the compiled kernel image, otherwise, the build will fail.
619 * It's important that these values are the same, since we are using
620 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
621 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
622 * values found in ext2, ext3 and ext4 filesystems, and of course the values
623 * defined in e2fsprogs.
624 *
625 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
626 */
627 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1U << EXT4_INODE_##FLAG))
628 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
629
ext4_check_flag_values(void)630 static inline void ext4_check_flag_values(void)
631 {
632 CHECK_FLAG_VALUE(SECRM);
633 CHECK_FLAG_VALUE(UNRM);
634 CHECK_FLAG_VALUE(COMPR);
635 CHECK_FLAG_VALUE(SYNC);
636 CHECK_FLAG_VALUE(IMMUTABLE);
637 CHECK_FLAG_VALUE(APPEND);
638 CHECK_FLAG_VALUE(NODUMP);
639 CHECK_FLAG_VALUE(NOATIME);
640 CHECK_FLAG_VALUE(DIRTY);
641 CHECK_FLAG_VALUE(COMPRBLK);
642 CHECK_FLAG_VALUE(NOCOMPR);
643 CHECK_FLAG_VALUE(ENCRYPT);
644 CHECK_FLAG_VALUE(INDEX);
645 CHECK_FLAG_VALUE(IMAGIC);
646 CHECK_FLAG_VALUE(JOURNAL_DATA);
647 CHECK_FLAG_VALUE(NOTAIL);
648 CHECK_FLAG_VALUE(DIRSYNC);
649 CHECK_FLAG_VALUE(TOPDIR);
650 CHECK_FLAG_VALUE(HUGE_FILE);
651 CHECK_FLAG_VALUE(EXTENTS);
652 CHECK_FLAG_VALUE(VERITY);
653 CHECK_FLAG_VALUE(EA_INODE);
654 CHECK_FLAG_VALUE(INLINE_DATA);
655 CHECK_FLAG_VALUE(PROJINHERIT);
656 CHECK_FLAG_VALUE(CASEFOLD);
657 CHECK_FLAG_VALUE(RESERVED);
658 }
659
660 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
661 struct compat_ext4_new_group_input {
662 u32 group;
663 compat_u64 block_bitmap;
664 compat_u64 inode_bitmap;
665 compat_u64 inode_table;
666 u32 blocks_count;
667 u16 reserved_blocks;
668 u16 unused;
669 };
670 #endif
671
672 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
673 struct ext4_new_group_data {
674 __u32 group;
675 __u64 block_bitmap;
676 __u64 inode_bitmap;
677 __u64 inode_table;
678 __u32 blocks_count;
679 __u16 reserved_blocks;
680 __u16 mdata_blocks;
681 __u32 free_clusters_count;
682 };
683
684 /* Indexes used to index group tables in ext4_new_group_data */
685 enum {
686 BLOCK_BITMAP = 0, /* block bitmap */
687 INODE_BITMAP, /* inode bitmap */
688 INODE_TABLE, /* inode tables */
689 GROUP_TABLE_COUNT,
690 };
691
692 /*
693 * Flags used by ext4_map_blocks()
694 */
695 /* Allocate any needed blocks and/or convert an unwritten
696 extent to be an initialized ext4 */
697 #define EXT4_GET_BLOCKS_CREATE 0x0001
698 /* Request the creation of an unwritten extent */
699 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002
700 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\
701 EXT4_GET_BLOCKS_CREATE)
702 /* Caller is from the delayed allocation writeout path
703 * finally doing the actual allocation of delayed blocks */
704 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
705 /*
706 * This means that we cannot merge newly allocated extents, and if we
707 * found an unwritten extent, we need to split it.
708 */
709 #define EXT4_GET_BLOCKS_SPLIT_NOMERGE 0x0008
710 /* Convert unwritten extent to initialized. */
711 #define EXT4_GET_BLOCKS_CONVERT 0x0010
712 /* Eventual metadata allocation (due to growing extent tree)
713 * should not fail, so try to use reserved blocks for that.*/
714 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
715 /* Don't normalize allocation size (used for fallocate) */
716 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
717 /* Convert written extents to unwritten */
718 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100
719 /* Write zeros to newly created written extents */
720 #define EXT4_GET_BLOCKS_ZERO 0x0200
721 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\
722 EXT4_GET_BLOCKS_ZERO)
723 /* Caller is in the context of data submission, such as writeback,
724 * fsync, etc. Especially, in the generic writeback path, caller will
725 * submit data before dropping transaction handle. This allows jbd2
726 * to avoid submitting data before commit. */
727 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400
728 /* Convert extent to initialized after IO complete */
729 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT |\
730 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT |\
731 EXT4_GET_BLOCKS_IO_SUBMIT)
732 /* Caller is in the atomic contex, find extent if it has been cached */
733 #define EXT4_GET_BLOCKS_CACHED_NOWAIT 0x0800
734 /*
735 * Atomic write caller needs this to query in the slow path of mixed mapping
736 * case, when a contiguous extent can be split across two adjacent leaf nodes.
737 * Look EXT4_MAP_QUERY_LAST_IN_LEAF.
738 */
739 #define EXT4_GET_BLOCKS_QUERY_LAST_IN_LEAF 0x1000
740
741 /*
742 * The bit position of these flags must not overlap with any of the
743 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(),
744 * read_extent_tree_block(), ext4_split_extent_at(),
745 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
746 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
747 * caching the extents when reading from the extent tree while a
748 * truncate or punch hole operation is in progress.
749 */
750 #define EXT4_EX_NOCACHE 0x40000000
751 #define EXT4_EX_FORCE_CACHE 0x20000000
752 #define EXT4_EX_NOFAIL 0x10000000
753 /*
754 * ext4_map_query_blocks() uses this filter mask to filter the flags needed to
755 * pass while lookup/querying of on disk extent tree.
756 */
757 #define EXT4_EX_QUERY_FILTER (EXT4_EX_NOCACHE | EXT4_EX_FORCE_CACHE |\
758 EXT4_EX_NOFAIL |\
759 EXT4_GET_BLOCKS_QUERY_LAST_IN_LEAF)
760
761 /*
762 * Flags used by ext4_free_blocks
763 */
764 #define EXT4_FREE_BLOCKS_METADATA 0x0001
765 #define EXT4_FREE_BLOCKS_FORGET 0x0002
766 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
767 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
768 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
769 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
770 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040
771
772 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
773 /*
774 * ioctl commands in 32 bit emulation
775 */
776 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
777 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
778 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
779 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
780 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
781 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
782 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
783 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
784 #endif
785
786 /* Max physical block we can address w/o extents */
787 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
788
789 /* Max logical block we can support */
790 #define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFE
791
792 /*
793 * Structure of an inode on the disk
794 */
795 struct ext4_inode {
796 __le16 i_mode; /* File mode */
797 __le16 i_uid; /* Low 16 bits of Owner Uid */
798 __le32 i_size_lo; /* Size in bytes */
799 __le32 i_atime; /* Access time */
800 __le32 i_ctime; /* Inode Change time */
801 __le32 i_mtime; /* Modification time */
802 __le32 i_dtime; /* Deletion Time */
803 __le16 i_gid; /* Low 16 bits of Group Id */
804 __le16 i_links_count; /* Links count */
805 __le32 i_blocks_lo; /* Blocks count */
806 __le32 i_flags; /* File flags */
807 union {
808 struct {
809 __le32 l_i_version;
810 } linux1;
811 struct {
812 __u32 h_i_translator;
813 } hurd1;
814 struct {
815 __u32 m_i_reserved1;
816 } masix1;
817 } osd1; /* OS dependent 1 */
818 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
819 __le32 i_generation; /* File version (for NFS) */
820 __le32 i_file_acl_lo; /* File ACL */
821 __le32 i_size_high;
822 __le32 i_obso_faddr; /* Obsoleted fragment address */
823 union {
824 struct {
825 __le16 l_i_blocks_high; /* were l_i_reserved1 */
826 __le16 l_i_file_acl_high;
827 __le16 l_i_uid_high; /* these 2 fields */
828 __le16 l_i_gid_high; /* were reserved2[0] */
829 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
830 __le16 l_i_reserved;
831 } linux2;
832 struct {
833 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
834 __u16 h_i_mode_high;
835 __u16 h_i_uid_high;
836 __u16 h_i_gid_high;
837 __u32 h_i_author;
838 } hurd2;
839 struct {
840 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
841 __le16 m_i_file_acl_high;
842 __u32 m_i_reserved2[2];
843 } masix2;
844 } osd2; /* OS dependent 2 */
845 __le16 i_extra_isize;
846 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
847 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
848 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
849 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
850 __le32 i_crtime; /* File Creation time */
851 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
852 __le32 i_version_hi; /* high 32 bits for 64-bit version */
853 __le32 i_projid; /* Project ID */
854 };
855
856 #define EXT4_EPOCH_BITS 2
857 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
858 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
859
860 /*
861 * Extended fields will fit into an inode if the filesystem was formatted
862 * with large inodes (-I 256 or larger) and there are not currently any EAs
863 * consuming all of the available space. For new inodes we always reserve
864 * enough space for the kernel's known extended fields, but for inodes
865 * created with an old kernel this might not have been the case. None of
866 * the extended inode fields is critical for correct filesystem operation.
867 * This macro checks if a certain field fits in the inode. Note that
868 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
869 */
870 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
871 ((offsetof(typeof(*ext4_inode), field) + \
872 sizeof((ext4_inode)->field)) \
873 <= (EXT4_GOOD_OLD_INODE_SIZE + \
874 (einode)->i_extra_isize)) \
875
876 /*
877 * We use an encoding that preserves the times for extra epoch "00":
878 *
879 * extra msb of adjust for signed
880 * epoch 32-bit 32-bit tv_sec to
881 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
882 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
883 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
884 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
885 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
886 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
887 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
888 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
889 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
890 *
891 * Note that previous versions of the kernel on 64-bit systems would
892 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
893 * 1970. e2fsck will correct this, assuming that it is run on the
894 * affected filesystem before 2242.
895 */
896
ext4_encode_extra_time(struct timespec64 ts)897 static inline __le32 ext4_encode_extra_time(struct timespec64 ts)
898 {
899 u32 extra = ((ts.tv_sec - (s32)ts.tv_sec) >> 32) & EXT4_EPOCH_MASK;
900 return cpu_to_le32(extra | (ts.tv_nsec << EXT4_EPOCH_BITS));
901 }
902
ext4_decode_extra_time(__le32 base,__le32 extra)903 static inline struct timespec64 ext4_decode_extra_time(__le32 base,
904 __le32 extra)
905 {
906 struct timespec64 ts = { .tv_sec = (signed)le32_to_cpu(base) };
907
908 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK)))
909 ts.tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
910 ts.tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
911 return ts;
912 }
913
914 #define EXT4_INODE_SET_XTIME_VAL(xtime, inode, raw_inode, ts) \
915 do { \
916 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \
917 (raw_inode)->xtime = cpu_to_le32((ts).tv_sec); \
918 (raw_inode)->xtime ## _extra = ext4_encode_extra_time(ts); \
919 } else \
920 (raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (ts).tv_sec, S32_MIN, S32_MAX)); \
921 } while (0)
922
923 #define EXT4_INODE_SET_ATIME(inode, raw_inode) \
924 EXT4_INODE_SET_XTIME_VAL(i_atime, inode, raw_inode, inode_get_atime(inode))
925
926 #define EXT4_INODE_SET_MTIME(inode, raw_inode) \
927 EXT4_INODE_SET_XTIME_VAL(i_mtime, inode, raw_inode, inode_get_mtime(inode))
928
929 #define EXT4_INODE_SET_CTIME(inode, raw_inode) \
930 EXT4_INODE_SET_XTIME_VAL(i_ctime, inode, raw_inode, inode_get_ctime(inode))
931
932 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
933 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
934 EXT4_INODE_SET_XTIME_VAL(xtime, &((einode)->vfs_inode), \
935 raw_inode, (einode)->xtime)
936
937 #define EXT4_INODE_GET_XTIME_VAL(xtime, inode, raw_inode) \
938 (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra) ? \
939 ext4_decode_extra_time((raw_inode)->xtime, \
940 (raw_inode)->xtime ## _extra) : \
941 (struct timespec64) { \
942 .tv_sec = (signed)le32_to_cpu((raw_inode)->xtime) \
943 })
944
945 #define EXT4_INODE_GET_ATIME(inode, raw_inode) \
946 do { \
947 inode_set_atime_to_ts(inode, \
948 EXT4_INODE_GET_XTIME_VAL(i_atime, inode, raw_inode)); \
949 } while (0)
950
951 #define EXT4_INODE_GET_MTIME(inode, raw_inode) \
952 do { \
953 inode_set_mtime_to_ts(inode, \
954 EXT4_INODE_GET_XTIME_VAL(i_mtime, inode, raw_inode)); \
955 } while (0)
956
957 #define EXT4_INODE_GET_CTIME(inode, raw_inode) \
958 do { \
959 inode_set_ctime_to_ts(inode, \
960 EXT4_INODE_GET_XTIME_VAL(i_ctime, inode, raw_inode)); \
961 } while (0)
962
963 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
964 do { \
965 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
966 (einode)->xtime = \
967 EXT4_INODE_GET_XTIME_VAL(xtime, &(einode->vfs_inode), \
968 raw_inode); \
969 else \
970 (einode)->xtime = (struct timespec64){0, 0}; \
971 } while (0)
972
973 #define i_disk_version osd1.linux1.l_i_version
974
975 #if defined(__KERNEL__) || defined(__linux__)
976 #define i_reserved1 osd1.linux1.l_i_reserved1
977 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
978 #define i_blocks_high osd2.linux2.l_i_blocks_high
979 #define i_uid_low i_uid
980 #define i_gid_low i_gid
981 #define i_uid_high osd2.linux2.l_i_uid_high
982 #define i_gid_high osd2.linux2.l_i_gid_high
983 #define i_checksum_lo osd2.linux2.l_i_checksum_lo
984
985 #elif defined(__GNU__)
986
987 #define i_translator osd1.hurd1.h_i_translator
988 #define i_uid_high osd2.hurd2.h_i_uid_high
989 #define i_gid_high osd2.hurd2.h_i_gid_high
990 #define i_author osd2.hurd2.h_i_author
991
992 #elif defined(__masix__)
993
994 #define i_reserved1 osd1.masix1.m_i_reserved1
995 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
996 #define i_reserved2 osd2.masix2.m_i_reserved2
997
998 #endif /* defined(__KERNEL__) || defined(__linux__) */
999
1000 #include "extents_status.h"
1001 #include "fast_commit.h"
1002
1003 /*
1004 * Lock subclasses for i_data_sem in the ext4_inode_info structure.
1005 *
1006 * These are needed to avoid lockdep false positives when we need to
1007 * allocate blocks to the quota inode during ext4_map_blocks(), while
1008 * holding i_data_sem for a normal (non-quota) inode. Since we don't
1009 * do quota tracking for the quota inode, this avoids deadlock (as
1010 * well as infinite recursion, since it isn't turtles all the way
1011 * down...)
1012 *
1013 * I_DATA_SEM_NORMAL - Used for most inodes
1014 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode
1015 * where the second inode has larger inode number
1016 * than the first
1017 * I_DATA_SEM_QUOTA - Used for quota inodes only
1018 * I_DATA_SEM_EA - Used for ea_inodes only
1019 */
1020 enum {
1021 I_DATA_SEM_NORMAL = 0,
1022 I_DATA_SEM_OTHER,
1023 I_DATA_SEM_QUOTA,
1024 I_DATA_SEM_EA
1025 };
1026
1027
1028 /*
1029 * fourth extended file system inode data in memory
1030 */
1031 struct ext4_inode_info {
1032 __le32 i_data[15]; /* unconverted */
1033 __u32 i_dtime;
1034 ext4_fsblk_t i_file_acl;
1035
1036 /*
1037 * i_block_group is the number of the block group which contains
1038 * this file's inode. Constant across the lifetime of the inode,
1039 * it is used for making block allocation decisions - we try to
1040 * place a file's data blocks near its inode block, and new inodes
1041 * near to their parent directory's inode.
1042 */
1043 ext4_group_t i_block_group;
1044 ext4_lblk_t i_dir_start_lookup;
1045 #if (BITS_PER_LONG < 64)
1046 unsigned long i_state_flags; /* Dynamic state flags */
1047 #endif
1048 unsigned long i_flags;
1049
1050 /*
1051 * Extended attributes can be read independently of the main file
1052 * data. Taking i_rwsem even when reading would cause contention
1053 * between readers of EAs and writers of regular file data, so
1054 * instead we synchronize on xattr_sem when reading or changing
1055 * EAs.
1056 */
1057 struct rw_semaphore xattr_sem;
1058
1059 /*
1060 * Inodes with EXT4_STATE_ORPHAN_FILE use i_orphan_idx. Otherwise
1061 * i_orphan is used.
1062 */
1063 union {
1064 struct list_head i_orphan; /* unlinked but open inodes */
1065 unsigned int i_orphan_idx; /* Index in orphan file */
1066 };
1067
1068 /* Fast commit related info */
1069
1070 /* For tracking dentry create updates */
1071 struct list_head i_fc_dilist;
1072 struct list_head i_fc_list; /*
1073 * inodes that need fast commit
1074 * protected by sbi->s_fc_lock.
1075 */
1076
1077 /* Start of lblk range that needs to be committed in this fast commit */
1078 ext4_lblk_t i_fc_lblk_start;
1079
1080 /* End of lblk range that needs to be committed in this fast commit */
1081 ext4_lblk_t i_fc_lblk_len;
1082
1083 spinlock_t i_raw_lock; /* protects updates to the raw inode */
1084
1085 /* Fast commit wait queue for this inode */
1086 wait_queue_head_t i_fc_wait;
1087
1088 /*
1089 * Protect concurrent accesses on i_fc_lblk_start, i_fc_lblk_len
1090 * and inode's EXT4_FC_STATE_COMMITTING state bit.
1091 */
1092 spinlock_t i_fc_lock;
1093
1094 /*
1095 * i_disksize keeps track of what the inode size is ON DISK, not
1096 * in memory. During truncate, i_size is set to the new size by
1097 * the VFS prior to calling ext4_truncate(), but the filesystem won't
1098 * set i_disksize to 0 until the truncate is actually under way.
1099 *
1100 * The intent is that i_disksize always represents the blocks which
1101 * are used by this file. This allows recovery to restart truncate
1102 * on orphans if we crash during truncate. We actually write i_disksize
1103 * into the on-disk inode when writing inodes out, instead of i_size.
1104 *
1105 * The only time when i_disksize and i_size may be different is when
1106 * a truncate is in progress. The only things which change i_disksize
1107 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
1108 */
1109 loff_t i_disksize;
1110
1111 /*
1112 * i_data_sem is for serialising ext4_truncate() against
1113 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
1114 * data tree are chopped off during truncate. We can't do that in
1115 * ext4 because whenever we perform intermediate commits during
1116 * truncate, the inode and all the metadata blocks *must* be in a
1117 * consistent state which allows truncation of the orphans to restart
1118 * during recovery. Hence we must fix the get_block-vs-truncate race
1119 * by other means, so we have i_data_sem.
1120 */
1121 struct rw_semaphore i_data_sem;
1122 struct inode vfs_inode;
1123 struct jbd2_inode *jinode;
1124
1125 /*
1126 * File creation time. Its function is same as that of
1127 * struct timespec64 i_{a,c,m}time in the generic inode.
1128 */
1129 struct timespec64 i_crtime;
1130
1131 /* mballoc */
1132 atomic_t i_prealloc_active;
1133
1134 /* allocation reservation info for delalloc */
1135 /* In case of bigalloc, this refer to clusters rather than blocks */
1136 unsigned int i_reserved_data_blocks;
1137 struct rb_root i_prealloc_node;
1138 rwlock_t i_prealloc_lock;
1139
1140 /* extents status tree */
1141 struct ext4_es_tree i_es_tree;
1142 rwlock_t i_es_lock;
1143 struct list_head i_es_list;
1144 unsigned int i_es_all_nr; /* protected by i_es_lock */
1145 unsigned int i_es_shk_nr; /* protected by i_es_lock */
1146 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for
1147 extents to shrink. Protected by
1148 i_es_lock */
1149 u64 i_es_seq; /* Change counter for extents.
1150 Protected by i_es_lock */
1151
1152 /* ialloc */
1153 ext4_group_t i_last_alloc_group;
1154
1155 /* pending cluster reservations for bigalloc file systems */
1156 struct ext4_pending_tree i_pending_tree;
1157
1158 /* on-disk additional length */
1159 __u16 i_extra_isize;
1160
1161 /* Indicate the inline data space. */
1162 u16 i_inline_off;
1163 u16 i_inline_size;
1164
1165 #ifdef CONFIG_QUOTA
1166 /* quota space reservation, managed internally by quota code */
1167 qsize_t i_reserved_quota;
1168 #endif
1169 spinlock_t i_block_reservation_lock;
1170
1171 /* Lock protecting lists below */
1172 spinlock_t i_completed_io_lock;
1173 /*
1174 * Completed IOs that need unwritten extents handling and have
1175 * transaction reserved
1176 */
1177 struct list_head i_rsv_conversion_list;
1178 struct work_struct i_rsv_conversion_work;
1179
1180 /*
1181 * Transactions that contain inode's metadata needed to complete
1182 * fsync and fdatasync, respectively.
1183 */
1184 tid_t i_sync_tid;
1185 tid_t i_datasync_tid;
1186
1187 #ifdef CONFIG_QUOTA
1188 struct dquot __rcu *i_dquot[MAXQUOTAS];
1189 #endif
1190
1191 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1192 __u32 i_csum_seed;
1193
1194 kprojid_t i_projid;
1195
1196 #ifdef CONFIG_FS_ENCRYPTION
1197 struct fscrypt_inode_info *i_crypt_info;
1198 #endif
1199 };
1200
1201 /*
1202 * File system states
1203 */
1204 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
1205 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
1206 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
1207 #define EXT4_FC_REPLAY 0x0020 /* Fast commit replay ongoing */
1208
1209 /*
1210 * Misc. filesystem flags
1211 */
1212 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
1213 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
1214 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
1215
1216 /*
1217 * Mount flags set via mount options or defaults
1218 */
1219 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */
1220 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
1221 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
1222 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
1223 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
1224 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
1225 #define EXT4_MOUNT_ERRORS_MASK 0x00070
1226 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
1227 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
1228 #ifdef CONFIG_FS_DAX
1229 #define EXT4_MOUNT_DAX_ALWAYS 0x00200 /* Direct Access */
1230 #else
1231 #define EXT4_MOUNT_DAX_ALWAYS 0
1232 #endif
1233 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
1234 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
1235 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
1236 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
1237 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
1238 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
1239 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
1240 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
1241 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
1242 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
1243 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */
1244 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota,
1245 * enable enforcement for hidden
1246 * quota files */
1247 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable
1248 * enforcement for hidden quota
1249 * files */
1250 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota
1251 * enforcement */
1252 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
1253 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
1254 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1255 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */
1256 #define EXT4_MOUNT_NO_PREFETCH_BLOCK_BITMAPS 0x4000000
1257 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
1258 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
1259 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
1260 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
1261 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
1262
1263 /*
1264 * Mount flags set either automatically (could not be set by mount option)
1265 * based on per file system feature or property or in special cases such as
1266 * distinguishing between explicit mount option definition and default.
1267 */
1268 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1269 specified delalloc */
1270 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1271 size of blocksize * 8
1272 blocks */
1273 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1274 file systems */
1275 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly
1276 specified journal checksum */
1277
1278 #define EXT4_MOUNT2_JOURNAL_FAST_COMMIT 0x00000010 /* Journal fast commit */
1279 #define EXT4_MOUNT2_DAX_NEVER 0x00000020 /* Do not allow Direct Access */
1280 #define EXT4_MOUNT2_DAX_INODE 0x00000040 /* For printing options only */
1281 #define EXT4_MOUNT2_MB_OPTIMIZE_SCAN 0x00000080 /* Optimize group
1282 * scanning in mballoc
1283 */
1284 #define EXT4_MOUNT2_ABORT 0x00000100 /* Abort filesystem */
1285
1286 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1287 ~EXT4_MOUNT_##opt
1288 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1289 EXT4_MOUNT_##opt
1290 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1291 EXT4_MOUNT_##opt)
1292
1293 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1294 ~EXT4_MOUNT2_##opt
1295 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1296 EXT4_MOUNT2_##opt
1297 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1298 EXT4_MOUNT2_##opt)
1299
1300 #define ext4_test_and_set_bit __test_and_set_bit_le
1301 #define ext4_set_bit __set_bit_le
1302 #define ext4_test_and_clear_bit __test_and_clear_bit_le
1303 #define ext4_clear_bit __clear_bit_le
1304 #define ext4_test_bit test_bit_le
1305 #define ext4_find_next_zero_bit find_next_zero_bit_le
1306 #define ext4_find_next_bit find_next_bit_le
1307
1308 extern void mb_set_bits(void *bm, int cur, int len);
1309
1310 /*
1311 * Maximal mount counts between two filesystem checks
1312 */
1313 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1314 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1315
1316 /*
1317 * Behaviour when detecting errors
1318 */
1319 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1320 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1321 #define EXT4_ERRORS_PANIC 3 /* Panic */
1322 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1323
1324 /* Metadata checksum algorithm codes */
1325 #define EXT4_CRC32C_CHKSUM 1
1326
1327 #define EXT4_LABEL_MAX 16
1328
1329 /*
1330 * Structure of the super block
1331 */
1332 struct ext4_super_block {
1333 /*00*/ __le32 s_inodes_count; /* Inodes count */
1334 __le32 s_blocks_count_lo; /* Blocks count */
1335 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1336 __le32 s_free_blocks_count_lo; /* Free blocks count */
1337 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1338 __le32 s_first_data_block; /* First Data Block */
1339 __le32 s_log_block_size; /* Block size */
1340 __le32 s_log_cluster_size; /* Allocation cluster size */
1341 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1342 __le32 s_clusters_per_group; /* # Clusters per group */
1343 __le32 s_inodes_per_group; /* # Inodes per group */
1344 __le32 s_mtime; /* Mount time */
1345 /*30*/ __le32 s_wtime; /* Write time */
1346 __le16 s_mnt_count; /* Mount count */
1347 __le16 s_max_mnt_count; /* Maximal mount count */
1348 __le16 s_magic; /* Magic signature */
1349 __le16 s_state; /* File system state */
1350 __le16 s_errors; /* Behaviour when detecting errors */
1351 __le16 s_minor_rev_level; /* minor revision level */
1352 /*40*/ __le32 s_lastcheck; /* time of last check */
1353 __le32 s_checkinterval; /* max. time between checks */
1354 __le32 s_creator_os; /* OS */
1355 __le32 s_rev_level; /* Revision level */
1356 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1357 __le16 s_def_resgid; /* Default gid for reserved blocks */
1358 /*
1359 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1360 *
1361 * Note: the difference between the compatible feature set and
1362 * the incompatible feature set is that if there is a bit set
1363 * in the incompatible feature set that the kernel doesn't
1364 * know about, it should refuse to mount the filesystem.
1365 *
1366 * e2fsck's requirements are more strict; if it doesn't know
1367 * about a feature in either the compatible or incompatible
1368 * feature set, it must abort and not try to meddle with
1369 * things it doesn't understand...
1370 */
1371 __le32 s_first_ino; /* First non-reserved inode */
1372 __le16 s_inode_size; /* size of inode structure */
1373 __le16 s_block_group_nr; /* block group # of this superblock */
1374 __le32 s_feature_compat; /* compatible feature set */
1375 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1376 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1377 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1378 /*78*/ char s_volume_name[EXT4_LABEL_MAX] __nonstring; /* volume name */
1379 /*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */
1380 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1381 /*
1382 * Performance hints. Directory preallocation should only
1383 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1384 */
1385 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1386 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1387 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1388 /*
1389 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1390 */
1391 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1392 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1393 __le32 s_journal_dev; /* device number of journal file */
1394 __le32 s_last_orphan; /* start of list of inodes to delete */
1395 __le32 s_hash_seed[4]; /* HTREE hash seed */
1396 __u8 s_def_hash_version; /* Default hash version to use */
1397 __u8 s_jnl_backup_type;
1398 __le16 s_desc_size; /* size of group descriptor */
1399 /*100*/ __le32 s_default_mount_opts;
1400 __le32 s_first_meta_bg; /* First metablock block group */
1401 __le32 s_mkfs_time; /* When the filesystem was created */
1402 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1403 /* 64bit support valid if EXT4_FEATURE_INCOMPAT_64BIT */
1404 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1405 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1406 __le32 s_free_blocks_count_hi; /* Free blocks count */
1407 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1408 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1409 __le32 s_flags; /* Miscellaneous flags */
1410 __le16 s_raid_stride; /* RAID stride */
1411 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1412 __le64 s_mmp_block; /* Block for multi-mount protection */
1413 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1414 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1415 __u8 s_checksum_type; /* metadata checksum algorithm used */
1416 __u8 s_encryption_level; /* versioning level for encryption */
1417 __u8 s_reserved_pad; /* Padding to next 32bits */
1418 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1419 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1420 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1421 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1422 snapshot's future use */
1423 __le32 s_snapshot_list; /* inode number of the head of the
1424 on-disk snapshot list */
1425 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1426 __le32 s_error_count; /* number of fs errors */
1427 __le32 s_first_error_time; /* first time an error happened */
1428 __le32 s_first_error_ino; /* inode involved in first error */
1429 __le64 s_first_error_block; /* block involved of first error */
1430 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */
1431 __le32 s_first_error_line; /* line number where error happened */
1432 __le32 s_last_error_time; /* most recent time of an error */
1433 __le32 s_last_error_ino; /* inode involved in last error */
1434 __le32 s_last_error_line; /* line number where error happened */
1435 __le64 s_last_error_block; /* block involved of last error */
1436 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */
1437 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1438 __u8 s_mount_opts[64];
1439 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1440 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1441 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1442 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
1443 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */
1444 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */
1445 __le32 s_lpf_ino; /* Location of the lost+found inode */
1446 __le32 s_prj_quota_inum; /* inode for tracking project quota */
1447 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */
1448 __u8 s_wtime_hi;
1449 __u8 s_mtime_hi;
1450 __u8 s_mkfs_time_hi;
1451 __u8 s_lastcheck_hi;
1452 __u8 s_first_error_time_hi;
1453 __u8 s_last_error_time_hi;
1454 __u8 s_first_error_errcode;
1455 __u8 s_last_error_errcode;
1456 __le16 s_encoding; /* Filename charset encoding */
1457 __le16 s_encoding_flags; /* Filename charset encoding flags */
1458 __le32 s_orphan_file_inum; /* Inode for tracking orphan inodes */
1459 __le16 s_def_resuid_hi;
1460 __le16 s_def_resgid_hi;
1461 __le32 s_reserved[93]; /* Padding to the end of the block */
1462 __le32 s_checksum; /* crc32c(superblock) */
1463 };
1464
1465 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1466
1467 #ifdef __KERNEL__
1468
1469 /* Number of quota types we support */
1470 #define EXT4_MAXQUOTAS 3
1471
1472 #define EXT4_ENC_UTF8_12_1 1
1473
1474 /* Types of ext4 journal triggers */
1475 enum ext4_journal_trigger_type {
1476 EXT4_JTR_ORPHAN_FILE,
1477 EXT4_JTR_NONE /* This must be the last entry for indexing to work! */
1478 };
1479
1480 #define EXT4_JOURNAL_TRIGGER_COUNT EXT4_JTR_NONE
1481
1482 struct ext4_journal_trigger {
1483 struct jbd2_buffer_trigger_type tr_triggers;
1484 struct super_block *sb;
1485 };
1486
EXT4_TRIGGER(struct jbd2_buffer_trigger_type * trigger)1487 static inline struct ext4_journal_trigger *EXT4_TRIGGER(
1488 struct jbd2_buffer_trigger_type *trigger)
1489 {
1490 return container_of(trigger, struct ext4_journal_trigger, tr_triggers);
1491 }
1492
1493 #define EXT4_ORPHAN_BLOCK_MAGIC 0x0b10ca04
1494
1495 /* Structure at the tail of orphan block */
1496 struct ext4_orphan_block_tail {
1497 __le32 ob_magic;
1498 __le32 ob_checksum;
1499 };
1500
ext4_inodes_per_orphan_block(struct super_block * sb)1501 static inline int ext4_inodes_per_orphan_block(struct super_block *sb)
1502 {
1503 return (sb->s_blocksize - sizeof(struct ext4_orphan_block_tail)) /
1504 sizeof(u32);
1505 }
1506
1507 struct ext4_orphan_block {
1508 atomic_t ob_free_entries; /* Number of free orphan entries in block */
1509 struct buffer_head *ob_bh; /* Buffer for orphan block */
1510 };
1511
1512 /*
1513 * Info about orphan file.
1514 */
1515 struct ext4_orphan_info {
1516 int of_blocks; /* Number of orphan blocks in a file */
1517 __u32 of_csum_seed; /* Checksum seed for orphan file */
1518 struct ext4_orphan_block *of_binfo; /* Array with info about orphan
1519 * file blocks */
1520 };
1521
1522 /*
1523 * fourth extended-fs super-block data in memory
1524 */
1525 struct ext4_sb_info {
1526 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1527 unsigned long s_inodes_per_block;/* Number of inodes per block */
1528 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1529 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1530 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1531 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1532 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1533 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1534 ext4_group_t s_groups_count; /* Number of groups in the fs */
1535 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1536 unsigned long s_overhead; /* # of fs overhead clusters */
1537 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1538 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1539 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1540 struct buffer_head * s_sbh; /* Buffer containing the super block */
1541 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1542 /* Array of bh's for the block group descriptors */
1543 struct buffer_head * __rcu *s_group_desc;
1544 unsigned int s_mount_opt;
1545 unsigned int s_mount_opt2;
1546 unsigned long s_mount_flags;
1547 unsigned int s_def_mount_opt;
1548 unsigned int s_def_mount_opt2;
1549 ext4_fsblk_t s_sb_block;
1550 atomic64_t s_resv_clusters;
1551 kuid_t s_resuid;
1552 kgid_t s_resgid;
1553 unsigned short s_mount_state;
1554 unsigned short s_pad;
1555 int s_addr_per_block_bits;
1556 int s_desc_per_block_bits;
1557 int s_inode_size;
1558 int s_first_ino;
1559 unsigned int s_inode_readahead_blks;
1560 unsigned int s_inode_goal;
1561 u32 s_hash_seed[4];
1562 int s_def_hash_version;
1563 int s_hash_unsigned; /* 3 if hash should be unsigned, 0 if not */
1564 struct percpu_counter s_freeclusters_counter;
1565 struct percpu_counter s_freeinodes_counter;
1566 struct percpu_counter s_dirs_counter;
1567 struct percpu_counter s_dirtyclusters_counter;
1568 struct percpu_counter s_sra_exceeded_retry_limit;
1569 struct blockgroup_lock *s_blockgroup_lock;
1570 struct proc_dir_entry *s_proc;
1571 struct kobject s_kobj;
1572 struct completion s_kobj_unregister;
1573 struct mutex s_error_notify_mutex; /* protects sysfs_notify vs kobject_del */
1574 struct super_block *s_sb;
1575 struct buffer_head *s_mmp_bh;
1576
1577 /* Journaling */
1578 struct journal_s *s_journal;
1579 unsigned long s_ext4_flags; /* Ext4 superblock flags */
1580 struct mutex s_orphan_lock; /* Protects on disk list changes */
1581 struct list_head s_orphan; /* List of orphaned inodes in on disk
1582 list */
1583 struct ext4_orphan_info s_orphan_info;
1584 unsigned long s_commit_interval;
1585 u32 s_max_batch_time;
1586 u32 s_min_batch_time;
1587 struct file *s_journal_bdev_file;
1588 #ifdef CONFIG_QUOTA
1589 /* Names of quota files with journalled quota */
1590 char __rcu *s_qf_names[EXT4_MAXQUOTAS];
1591 int s_jquota_fmt; /* Format of quota to use */
1592 #endif
1593 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1594 struct ext4_system_blocks __rcu *s_system_blks;
1595
1596 #ifdef EXTENTS_STATS
1597 /* ext4 extents stats */
1598 unsigned long s_ext_min;
1599 unsigned long s_ext_max;
1600 unsigned long s_depth_max;
1601 spinlock_t s_ext_stats_lock;
1602 unsigned long s_ext_blocks;
1603 unsigned long s_ext_extents;
1604 #endif
1605
1606 /* for buddy allocator */
1607 struct ext4_group_info ** __rcu *s_group_info;
1608 struct inode *s_buddy_cache;
1609 spinlock_t s_md_lock;
1610 unsigned short *s_mb_offsets;
1611 unsigned int *s_mb_maxs;
1612 unsigned int s_group_info_size;
1613 atomic_t s_mb_free_pending;
1614 struct list_head s_freed_data_list[2]; /* List of blocks to be freed
1615 after commit completed */
1616 struct list_head s_discard_list;
1617 struct work_struct s_discard_work;
1618 atomic_t s_retry_alloc_pending;
1619 struct xarray *s_mb_avg_fragment_size;
1620 struct xarray *s_mb_largest_free_orders;
1621
1622 /* tunables */
1623 unsigned long s_stripe;
1624 unsigned int s_mb_max_linear_groups;
1625 unsigned int s_mb_stream_request;
1626 unsigned int s_mb_max_to_scan;
1627 unsigned int s_mb_min_to_scan;
1628 unsigned int s_mb_stats;
1629 unsigned int s_mb_order2_reqs;
1630 unsigned int s_mb_group_prealloc;
1631 unsigned int s_max_dir_size_kb;
1632 unsigned int s_mb_prefetch;
1633 unsigned int s_mb_prefetch_limit;
1634 unsigned int s_mb_best_avail_max_trim_order;
1635 unsigned int s_sb_update_sec;
1636 unsigned int s_sb_update_kb;
1637
1638 /* where last allocation was done - for stream allocation */
1639 ext4_group_t *s_mb_last_groups;
1640 unsigned int s_mb_nr_global_goals;
1641
1642 /* stats for buddy allocator */
1643 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1644 atomic_t s_bal_success; /* we found long enough chunks */
1645 atomic_t s_bal_allocated; /* in blocks */
1646 atomic_t s_bal_ex_scanned; /* total extents scanned */
1647 atomic_t s_bal_cX_ex_scanned[EXT4_MB_NUM_CRS]; /* total extents scanned */
1648 atomic_t s_bal_groups_scanned; /* number of groups scanned */
1649 atomic_t s_bal_goals; /* goal hits */
1650 atomic_t s_bal_stream_goals; /* stream allocation global goal hits */
1651 atomic_t s_bal_len_goals; /* len goal hits */
1652 atomic_t s_bal_breaks; /* too long searches */
1653 atomic_t s_bal_2orders; /* 2^order hits */
1654 atomic64_t s_bal_cX_groups_considered[EXT4_MB_NUM_CRS];
1655 atomic64_t s_bal_cX_hits[EXT4_MB_NUM_CRS];
1656 atomic64_t s_bal_cX_failed[EXT4_MB_NUM_CRS]; /* cX loop didn't find blocks */
1657 atomic_t s_mb_buddies_generated; /* number of buddies generated */
1658 atomic64_t s_mb_generation_time;
1659 atomic_t s_mb_lost_chunks;
1660 atomic_t s_mb_preallocated;
1661 atomic_t s_mb_discarded;
1662 atomic_t s_lock_busy;
1663
1664 /* locality groups */
1665 struct ext4_locality_group __percpu *s_locality_groups;
1666
1667 /* for write statistics */
1668 unsigned long s_sectors_written_start;
1669 u64 s_kbytes_written;
1670
1671 /* the size of zero-out chunk */
1672 unsigned int s_extent_max_zeroout_kb;
1673
1674 unsigned int s_log_groups_per_flex;
1675 struct flex_groups * __rcu *s_flex_groups;
1676 ext4_group_t s_flex_groups_allocated;
1677
1678 /* workqueue for reserved extent conversions (buffered io) */
1679 struct workqueue_struct *rsv_conversion_wq;
1680
1681 /* timer for periodic error stats printing */
1682 struct timer_list s_err_report;
1683 /* timeout in seconds for s_err_report; 0 disables the timer. */
1684 unsigned long s_err_report_sec;
1685
1686 /* Lazy inode table initialization info */
1687 struct ext4_li_request *s_li_request;
1688 /* Wait multiplier for lazy initialization thread */
1689 unsigned int s_li_wait_mult;
1690
1691 /* Kernel thread for multiple mount protection */
1692 struct task_struct *s_mmp_tsk;
1693
1694 /* record the last minlen when FITRIM is called. */
1695 unsigned long s_last_trim_minblks;
1696
1697 /* minimum folio order of a page cache allocation */
1698 u16 s_min_folio_order;
1699 /* supported maximum folio order, 0 means not supported */
1700 u16 s_max_folio_order;
1701
1702 /* Precomputed FS UUID checksum for seeding other checksums */
1703 __u32 s_csum_seed;
1704
1705 /* Reclaim extents from extent status tree */
1706 struct shrinker *s_es_shrinker;
1707 struct list_head s_es_list; /* List of inodes with reclaimable extents */
1708 long s_es_nr_inode;
1709 struct ext4_es_stats s_es_stats;
1710 struct mb_cache *s_ea_block_cache;
1711 struct mb_cache *s_ea_inode_cache;
1712 spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1713
1714 /* Journal triggers for checksum computation */
1715 struct ext4_journal_trigger s_journal_triggers[EXT4_JOURNAL_TRIGGER_COUNT];
1716
1717 /* Ratelimit ext4 messages. */
1718 struct ratelimit_state s_err_ratelimit_state;
1719 struct ratelimit_state s_warning_ratelimit_state;
1720 struct ratelimit_state s_msg_ratelimit_state;
1721 atomic_t s_warning_count;
1722 atomic_t s_msg_count;
1723
1724 /* Encryption policy for '-o test_dummy_encryption' */
1725 struct fscrypt_dummy_policy s_dummy_enc_policy;
1726
1727 /*
1728 * Barrier between writepages ops and changing any inode's JOURNAL_DATA
1729 * or EXTENTS flag or between writepages ops and changing DELALLOC or
1730 * DIOREAD_NOLOCK mount options on remount.
1731 */
1732 struct percpu_rw_semaphore s_writepages_rwsem;
1733 struct dax_device *s_daxdev;
1734 u64 s_dax_part_off;
1735 #ifdef CONFIG_EXT4_DEBUG
1736 unsigned long s_simulate_fail;
1737 #endif
1738 /* Record the errseq of the backing block device */
1739 errseq_t s_bdev_wb_err;
1740 spinlock_t s_bdev_wb_lock;
1741
1742 /* Information about errors that happened during this mount */
1743 spinlock_t s_error_lock;
1744 int s_add_error_count;
1745 int s_first_error_code;
1746 __u32 s_first_error_line;
1747 __u32 s_first_error_ino;
1748 __u64 s_first_error_block;
1749 const char *s_first_error_func;
1750 time64_t s_first_error_time;
1751 int s_last_error_code;
1752 __u32 s_last_error_line;
1753 __u32 s_last_error_ino;
1754 __u64 s_last_error_block;
1755 const char *s_last_error_func;
1756 time64_t s_last_error_time;
1757 /*
1758 * If we are in a context where we cannot update the on-disk
1759 * superblock, we queue the work here. This is used to update
1760 * the error information in the superblock, and for periodic
1761 * updates of the superblock called from the commit callback
1762 * function.
1763 */
1764 struct work_struct s_sb_upd_work;
1765
1766 /* Atomic write unit values in bytes */
1767 unsigned int s_awu_min;
1768 unsigned int s_awu_max;
1769
1770 /* Ext4 fast commit sub transaction ID */
1771 atomic_t s_fc_subtid;
1772
1773 /*
1774 * After commit starts, the main queue gets locked, and the further
1775 * updates get added in the staging queue.
1776 */
1777 #define FC_Q_MAIN 0
1778 #define FC_Q_STAGING 1
1779 struct list_head s_fc_q[2]; /* Inodes staged for fast commit
1780 * that have data changes in them.
1781 */
1782 struct list_head s_fc_dentry_q[2]; /* directory entry updates */
1783 unsigned int s_fc_bytes;
1784 /*
1785 * Main fast commit lock. This lock protects accesses to the
1786 * following fields:
1787 * ei->i_fc_list, s_fc_dentry_q, s_fc_q, s_fc_bytes, s_fc_bh.
1788 *
1789 * s_fc_lock can be taken from reclaim context (inode eviction) and is
1790 * thus reclaim unsafe. Use ext4_fc_lock()/ext4_fc_unlock() helpers
1791 * when acquiring / releasing the lock.
1792 */
1793 struct mutex s_fc_lock;
1794 struct buffer_head *s_fc_bh;
1795 struct ext4_fc_stats s_fc_stats;
1796 tid_t s_fc_ineligible_tid;
1797 #ifdef CONFIG_EXT4_DEBUG
1798 int s_fc_debug_max_replay;
1799 #endif
1800 struct ext4_fc_replay_state s_fc_replay_state;
1801 };
1802
EXT4_SB(struct super_block * sb)1803 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1804 {
1805 return sb->s_fs_info;
1806 }
EXT4_I(struct inode * inode)1807 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1808 {
1809 return container_of(inode, struct ext4_inode_info, vfs_inode);
1810 }
1811
ext4_writepages_down_read(struct super_block * sb)1812 static inline int ext4_writepages_down_read(struct super_block *sb)
1813 {
1814 percpu_down_read(&EXT4_SB(sb)->s_writepages_rwsem);
1815 return memalloc_nofs_save();
1816 }
1817
ext4_writepages_up_read(struct super_block * sb,int ctx)1818 static inline void ext4_writepages_up_read(struct super_block *sb, int ctx)
1819 {
1820 memalloc_nofs_restore(ctx);
1821 percpu_up_read(&EXT4_SB(sb)->s_writepages_rwsem);
1822 }
1823
ext4_writepages_down_write(struct super_block * sb)1824 static inline int ext4_writepages_down_write(struct super_block *sb)
1825 {
1826 percpu_down_write(&EXT4_SB(sb)->s_writepages_rwsem);
1827 return memalloc_nofs_save();
1828 }
1829
ext4_writepages_up_write(struct super_block * sb,int ctx)1830 static inline void ext4_writepages_up_write(struct super_block *sb, int ctx)
1831 {
1832 memalloc_nofs_restore(ctx);
1833 percpu_up_write(&EXT4_SB(sb)->s_writepages_rwsem);
1834 }
1835
ext4_fc_lock(struct super_block * sb)1836 static inline int ext4_fc_lock(struct super_block *sb)
1837 {
1838 mutex_lock(&EXT4_SB(sb)->s_fc_lock);
1839 return memalloc_nofs_save();
1840 }
1841
ext4_fc_unlock(struct super_block * sb,int ctx)1842 static inline void ext4_fc_unlock(struct super_block *sb, int ctx)
1843 {
1844 memalloc_nofs_restore(ctx);
1845 mutex_unlock(&EXT4_SB(sb)->s_fc_lock);
1846 }
1847
ext4_valid_inum(struct super_block * sb,unsigned long ino)1848 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1849 {
1850 return ino == EXT4_ROOT_INO ||
1851 (ino >= EXT4_FIRST_INO(sb) &&
1852 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1853 }
1854
ext4_get_resuid(struct ext4_super_block * es)1855 static inline int ext4_get_resuid(struct ext4_super_block *es)
1856 {
1857 return le16_to_cpu(es->s_def_resuid) |
1858 le16_to_cpu(es->s_def_resuid_hi) << 16;
1859 }
1860
ext4_get_resgid(struct ext4_super_block * es)1861 static inline int ext4_get_resgid(struct ext4_super_block *es)
1862 {
1863 return le16_to_cpu(es->s_def_resgid) |
1864 le16_to_cpu(es->s_def_resgid_hi) << 16;
1865 }
1866
1867 /*
1868 * Returns: sbi->field[index]
1869 * Used to access an array element from the following sbi fields which require
1870 * rcu protection to avoid dereferencing an invalid pointer due to reassignment
1871 * - s_group_desc
1872 * - s_group_info
1873 * - s_flex_group
1874 */
1875 #define sbi_array_rcu_deref(sbi, field, index) \
1876 ({ \
1877 typeof(*((sbi)->field)) _v; \
1878 rcu_read_lock(); \
1879 _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index]; \
1880 rcu_read_unlock(); \
1881 _v; \
1882 })
1883
1884 /*
1885 * run-time mount flags
1886 */
1887 enum {
1888 EXT4_MF_MNTDIR_SAMPLED,
1889 EXT4_MF_FC_INELIGIBLE, /* Fast commit ineligible */
1890 EXT4_MF_JOURNAL_DESTROY /* Journal is in process of destroying */
1891 };
1892
ext4_set_mount_flag(struct super_block * sb,int bit)1893 static inline void ext4_set_mount_flag(struct super_block *sb, int bit)
1894 {
1895 set_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1896 }
1897
ext4_clear_mount_flag(struct super_block * sb,int bit)1898 static inline void ext4_clear_mount_flag(struct super_block *sb, int bit)
1899 {
1900 clear_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1901 }
1902
ext4_test_mount_flag(struct super_block * sb,int bit)1903 static inline int ext4_test_mount_flag(struct super_block *sb, int bit)
1904 {
1905 return test_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1906 }
1907
1908
1909 /*
1910 * Simulate_fail codes
1911 */
1912 #define EXT4_SIM_BBITMAP_EIO 1
1913 #define EXT4_SIM_BBITMAP_CRC 2
1914 #define EXT4_SIM_IBITMAP_EIO 3
1915 #define EXT4_SIM_IBITMAP_CRC 4
1916 #define EXT4_SIM_INODE_EIO 5
1917 #define EXT4_SIM_INODE_CRC 6
1918 #define EXT4_SIM_DIRBLOCK_EIO 7
1919 #define EXT4_SIM_DIRBLOCK_CRC 8
1920
ext4_simulate_fail(struct super_block * sb,unsigned long code)1921 static inline bool ext4_simulate_fail(struct super_block *sb,
1922 unsigned long code)
1923 {
1924 #ifdef CONFIG_EXT4_DEBUG
1925 struct ext4_sb_info *sbi = EXT4_SB(sb);
1926
1927 if (unlikely(sbi->s_simulate_fail == code)) {
1928 sbi->s_simulate_fail = 0;
1929 return true;
1930 }
1931 #endif
1932 return false;
1933 }
1934
1935 /*
1936 * Error number codes for s_{first,last}_error_errno
1937 *
1938 * Linux errno numbers are architecture specific, so we need to translate
1939 * them into something which is architecture independent. We don't define
1940 * codes for all errno's; just the ones which are most likely to be the cause
1941 * of an ext4_error() call.
1942 */
1943 #define EXT4_ERR_UNKNOWN 1
1944 #define EXT4_ERR_EIO 2
1945 #define EXT4_ERR_ENOMEM 3
1946 #define EXT4_ERR_EFSBADCRC 4
1947 #define EXT4_ERR_EFSCORRUPTED 5
1948 #define EXT4_ERR_ENOSPC 6
1949 #define EXT4_ERR_ENOKEY 7
1950 #define EXT4_ERR_EROFS 8
1951 #define EXT4_ERR_EFBIG 9
1952 #define EXT4_ERR_EEXIST 10
1953 #define EXT4_ERR_ERANGE 11
1954 #define EXT4_ERR_EOVERFLOW 12
1955 #define EXT4_ERR_EBUSY 13
1956 #define EXT4_ERR_ENOTDIR 14
1957 #define EXT4_ERR_ENOTEMPTY 15
1958 #define EXT4_ERR_ESHUTDOWN 16
1959 #define EXT4_ERR_EFAULT 17
1960
1961 /*
1962 * Inode dynamic state flags
1963 */
1964 enum {
1965 EXT4_STATE_NEW, /* inode is newly created */
1966 EXT4_STATE_XATTR, /* has in-inode xattrs */
1967 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1968 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1969 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1970 EXT4_STATE_NEWENTRY, /* File just added to dir */
1971 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1972 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
1973 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */
1974 EXT4_STATE_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */
1975 EXT4_STATE_FC_COMMITTING, /* Fast commit ongoing */
1976 EXT4_STATE_FC_FLUSHING_DATA, /* Fast commit flushing data */
1977 EXT4_STATE_ORPHAN_FILE, /* Inode orphaned in orphan file */
1978 };
1979
1980 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1981 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1982 { \
1983 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1984 } \
1985 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1986 { \
1987 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1988 } \
1989 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1990 { \
1991 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1992 }
1993
1994 /* Add these declarations here only so that these functions can be
1995 * found by name. Otherwise, they are very hard to locate. */
1996 static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1997 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1998 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1999 EXT4_INODE_BIT_FNS(flag, flags, 0)
2000
2001 /* Add these declarations here only so that these functions can be
2002 * found by name. Otherwise, they are very hard to locate. */
2003 static inline int ext4_test_inode_state(struct inode *inode, int bit);
2004 static inline void ext4_set_inode_state(struct inode *inode, int bit);
2005 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
2006 #if (BITS_PER_LONG < 64)
2007 EXT4_INODE_BIT_FNS(state, state_flags, 0)
2008
ext4_clear_state_flags(struct ext4_inode_info * ei)2009 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
2010 {
2011 (ei)->i_state_flags = 0;
2012 }
2013 #else
2014 EXT4_INODE_BIT_FNS(state, flags, 32)
2015
ext4_clear_state_flags(struct ext4_inode_info * ei)2016 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
2017 {
2018 /* We depend on the fact that callers will set i_flags */
2019 }
2020 #endif
2021 #else
2022 /* Assume that user mode programs are passing in an ext4fs superblock, not
2023 * a kernel struct super_block. This will allow us to call the feature-test
2024 * macros from user land. */
2025 #define EXT4_SB(sb) (sb)
2026 #endif
2027
ext4_verity_in_progress(struct inode * inode)2028 static inline bool ext4_verity_in_progress(struct inode *inode)
2029 {
2030 return IS_ENABLED(CONFIG_FS_VERITY) &&
2031 ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS);
2032 }
2033
2034 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
2035
2036 /*
2037 * Check whether the inode is tracked as orphan (either in orphan file or
2038 * orphan list).
2039 */
ext4_inode_orphan_tracked(struct inode * inode)2040 static inline bool ext4_inode_orphan_tracked(struct inode *inode)
2041 {
2042 return ext4_test_inode_state(inode, EXT4_STATE_ORPHAN_FILE) ||
2043 !list_empty(&EXT4_I(inode)->i_orphan);
2044 }
2045
2046 /*
2047 * Codes for operating systems
2048 */
2049 #define EXT4_OS_LINUX 0
2050 #define EXT4_OS_HURD 1
2051 #define EXT4_OS_MASIX 2
2052 #define EXT4_OS_FREEBSD 3
2053 #define EXT4_OS_LITES 4
2054
2055 /*
2056 * Revision levels
2057 */
2058 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
2059 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
2060
2061 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
2062
2063 #define EXT4_GOOD_OLD_INODE_SIZE 128
2064
2065 #define EXT4_EXTRA_TIMESTAMP_MAX (((s64)1 << 34) - 1 + S32_MIN)
2066 #define EXT4_NON_EXTRA_TIMESTAMP_MAX S32_MAX
2067 #define EXT4_TIMESTAMP_MIN S32_MIN
2068
2069 /*
2070 * Feature set definitions
2071 */
2072
2073 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
2074 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
2075 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
2076 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
2077 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
2078 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
2079 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
2080 /*
2081 * The reason why "FAST_COMMIT" is a compat feature is that, FS becomes
2082 * incompatible only if fast commit blocks are present in the FS. Since we
2083 * clear the journal (and thus the fast commit blocks), we don't mark FS as
2084 * incompatible. We also have a JBD2 incompat feature, which gets set when
2085 * there are fast commit blocks present in the journal.
2086 */
2087 #define EXT4_FEATURE_COMPAT_FAST_COMMIT 0x0400
2088 #define EXT4_FEATURE_COMPAT_STABLE_INODES 0x0800
2089 #define EXT4_FEATURE_COMPAT_ORPHAN_FILE 0x1000 /* Orphan file exists */
2090
2091 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
2092 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
2093 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
2094 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
2095 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
2096 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
2097 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
2098 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
2099 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
2100 /*
2101 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
2102 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
2103 * all other data structures' checksums. However, the METADATA_CSUM and
2104 * GDT_CSUM bits are mutually exclusive.
2105 */
2106 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
2107 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000
2108 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000
2109 #define EXT4_FEATURE_RO_COMPAT_VERITY 0x8000
2110 #define EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT 0x10000 /* Orphan file may be
2111 non-empty */
2112
2113 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
2114 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
2115 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
2116 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
2117 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
2118 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
2119 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
2120 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
2121 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
2122 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
2123 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
2124 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000
2125 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
2126 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
2127 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000
2128 #define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000
2129
2130 extern void ext4_update_dynamic_rev(struct super_block *sb);
2131
2132 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
2133 static inline bool ext4_has_feature_##name(struct super_block *sb) \
2134 { \
2135 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
2136 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
2137 } \
2138 static inline void ext4_set_feature_##name(struct super_block *sb) \
2139 { \
2140 ext4_update_dynamic_rev(sb); \
2141 EXT4_SB(sb)->s_es->s_feature_compat |= \
2142 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
2143 } \
2144 static inline void ext4_clear_feature_##name(struct super_block *sb) \
2145 { \
2146 EXT4_SB(sb)->s_es->s_feature_compat &= \
2147 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
2148 }
2149
2150 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
2151 static inline bool ext4_has_feature_##name(struct super_block *sb) \
2152 { \
2153 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
2154 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
2155 } \
2156 static inline void ext4_set_feature_##name(struct super_block *sb) \
2157 { \
2158 ext4_update_dynamic_rev(sb); \
2159 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
2160 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
2161 } \
2162 static inline void ext4_clear_feature_##name(struct super_block *sb) \
2163 { \
2164 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
2165 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
2166 }
2167
2168 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
2169 static inline bool ext4_has_feature_##name(struct super_block *sb) \
2170 { \
2171 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
2172 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
2173 } \
2174 static inline void ext4_set_feature_##name(struct super_block *sb) \
2175 { \
2176 ext4_update_dynamic_rev(sb); \
2177 EXT4_SB(sb)->s_es->s_feature_incompat |= \
2178 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
2179 } \
2180 static inline void ext4_clear_feature_##name(struct super_block *sb) \
2181 { \
2182 EXT4_SB(sb)->s_es->s_feature_incompat &= \
2183 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
2184 }
2185
EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc,DIR_PREALLOC)2186 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC)
2187 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES)
2188 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL)
2189 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR)
2190 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE)
2191 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX)
2192 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2)
2193 EXT4_FEATURE_COMPAT_FUNCS(fast_commit, FAST_COMMIT)
2194 EXT4_FEATURE_COMPAT_FUNCS(stable_inodes, STABLE_INODES)
2195 EXT4_FEATURE_COMPAT_FUNCS(orphan_file, ORPHAN_FILE)
2196
2197 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER)
2198 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE)
2199 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR)
2200 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE)
2201 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM)
2202 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK)
2203 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE)
2204 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA)
2205 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC)
2206 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM)
2207 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY)
2208 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT)
2209 EXT4_FEATURE_RO_COMPAT_FUNCS(verity, VERITY)
2210 EXT4_FEATURE_RO_COMPAT_FUNCS(orphan_present, ORPHAN_PRESENT)
2211
2212 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION)
2213 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE)
2214 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER)
2215 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV)
2216 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG)
2217 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS)
2218 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT)
2219 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP)
2220 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG)
2221 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE)
2222 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA)
2223 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED)
2224 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR)
2225 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA)
2226 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT)
2227 EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD)
2228
2229 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
2230 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
2231 EXT4_FEATURE_INCOMPAT_META_BG)
2232 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2233 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2234 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
2235
2236 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
2237 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
2238 EXT4_FEATURE_INCOMPAT_RECOVER| \
2239 EXT4_FEATURE_INCOMPAT_META_BG)
2240 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2241 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2242 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
2243
2244 #define EXT4_FEATURE_COMPAT_SUPP (EXT4_FEATURE_COMPAT_EXT_ATTR| \
2245 EXT4_FEATURE_COMPAT_ORPHAN_FILE)
2246 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
2247 EXT4_FEATURE_INCOMPAT_RECOVER| \
2248 EXT4_FEATURE_INCOMPAT_META_BG| \
2249 EXT4_FEATURE_INCOMPAT_EXTENTS| \
2250 EXT4_FEATURE_INCOMPAT_64BIT| \
2251 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
2252 EXT4_FEATURE_INCOMPAT_EA_INODE| \
2253 EXT4_FEATURE_INCOMPAT_MMP | \
2254 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
2255 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
2256 EXT4_FEATURE_INCOMPAT_CASEFOLD | \
2257 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \
2258 EXT4_FEATURE_INCOMPAT_LARGEDIR)
2259 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2260 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2261 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
2262 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
2263 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
2264 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
2265 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
2266 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
2267 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
2268 EXT4_FEATURE_RO_COMPAT_QUOTA |\
2269 EXT4_FEATURE_RO_COMPAT_PROJECT |\
2270 EXT4_FEATURE_RO_COMPAT_VERITY |\
2271 EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT)
2272
2273 #define EXTN_FEATURE_FUNCS(ver) \
2274 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
2275 { \
2276 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
2277 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
2278 } \
2279 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
2280 { \
2281 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
2282 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
2283 } \
2284 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
2285 { \
2286 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
2287 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
2288 }
2289
2290 EXTN_FEATURE_FUNCS(2)
2291 EXTN_FEATURE_FUNCS(3)
2292 EXTN_FEATURE_FUNCS(4)
2293
2294 static inline bool ext4_has_compat_features(struct super_block *sb)
2295 {
2296 return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
2297 }
ext4_has_ro_compat_features(struct super_block * sb)2298 static inline bool ext4_has_ro_compat_features(struct super_block *sb)
2299 {
2300 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
2301 }
ext4_has_incompat_features(struct super_block * sb)2302 static inline bool ext4_has_incompat_features(struct super_block *sb)
2303 {
2304 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
2305 }
2306
2307 extern int ext4_feature_set_ok(struct super_block *sb, int readonly);
2308
2309 /*
2310 * Superblock flags
2311 */
2312 enum {
2313 EXT4_FLAGS_RESIZING, /* Avoid superblock update and resize race */
2314 EXT4_FLAGS_SHUTDOWN, /* Prevent access to the file system */
2315 EXT4_FLAGS_BDEV_IS_DAX, /* Current block device support DAX */
2316 EXT4_FLAGS_EMERGENCY_RO,/* Emergency read-only due to fs errors */
2317 };
2318
ext4_forced_shutdown(struct super_block * sb)2319 static inline int ext4_forced_shutdown(struct super_block *sb)
2320 {
2321 return test_bit(EXT4_FLAGS_SHUTDOWN, &EXT4_SB(sb)->s_ext4_flags);
2322 }
2323
ext4_emergency_ro(struct super_block * sb)2324 static inline int ext4_emergency_ro(struct super_block *sb)
2325 {
2326 return test_bit(EXT4_FLAGS_EMERGENCY_RO, &EXT4_SB(sb)->s_ext4_flags);
2327 }
2328
ext4_emergency_state(struct super_block * sb)2329 static inline int ext4_emergency_state(struct super_block *sb)
2330 {
2331 if (unlikely(ext4_forced_shutdown(sb)))
2332 return -EIO;
2333 if (unlikely(ext4_emergency_ro(sb)))
2334 return -EROFS;
2335 return 0;
2336 }
2337
2338 /*
2339 * Default values for user and/or group using reserved blocks
2340 */
2341 #define EXT4_DEF_RESUID 0
2342 #define EXT4_DEF_RESGID 0
2343
2344 /*
2345 * Default project ID
2346 */
2347 #define EXT4_DEF_PROJID 0
2348
2349 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
2350
2351 /*
2352 * Default mount options
2353 */
2354 #define EXT4_DEFM_DEBUG 0x0001
2355 #define EXT4_DEFM_BSDGROUPS 0x0002
2356 #define EXT4_DEFM_XATTR_USER 0x0004
2357 #define EXT4_DEFM_ACL 0x0008
2358 #define EXT4_DEFM_UID16 0x0010
2359 #define EXT4_DEFM_JMODE 0x0060
2360 #define EXT4_DEFM_JMODE_DATA 0x0020
2361 #define EXT4_DEFM_JMODE_ORDERED 0x0040
2362 #define EXT4_DEFM_JMODE_WBACK 0x0060
2363 #define EXT4_DEFM_NOBARRIER 0x0100
2364 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
2365 #define EXT4_DEFM_DISCARD 0x0400
2366 #define EXT4_DEFM_NODELALLOC 0x0800
2367
2368 /*
2369 * Default journal batch times and ioprio.
2370 */
2371 #define EXT4_DEF_MIN_BATCH_TIME 0
2372 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
2373 #define EXT4_DEF_JOURNAL_IOPRIO (IOPRIO_PRIO_VALUE(IOPRIO_CLASS_BE, 3))
2374
2375
2376 /*
2377 * Default values for superblock update
2378 */
2379 #define EXT4_DEF_SB_UPDATE_INTERVAL_SEC (3600) /* seconds (1 hour) */
2380 #define EXT4_DEF_SB_UPDATE_INTERVAL_KB (16384) /* kilobytes (16MB) */
2381
2382 /*
2383 * Minimum number of groups in a flexgroup before we separate out
2384 * directories into the first block group of a flexgroup
2385 */
2386 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
2387
2388 /*
2389 * Structure of a directory entry
2390 */
2391 #define EXT4_NAME_LEN 255
2392 /*
2393 * Base length of the ext4 directory entry excluding the name length
2394 */
2395 #define EXT4_BASE_DIR_LEN (sizeof(struct ext4_dir_entry_2) - EXT4_NAME_LEN)
2396
2397 struct ext4_dir_entry {
2398 __le32 inode; /* Inode number */
2399 __le16 rec_len; /* Directory entry length */
2400 __le16 name_len; /* Name length */
2401 char name[EXT4_NAME_LEN]; /* File name */
2402 };
2403
2404
2405 /*
2406 * Encrypted Casefolded entries require saving the hash on disk. This structure
2407 * followed ext4_dir_entry_2's name[name_len] at the next 4 byte aligned
2408 * boundary.
2409 */
2410 struct ext4_dir_entry_hash {
2411 __le32 hash;
2412 __le32 minor_hash;
2413 };
2414
2415 /*
2416 * The new version of the directory entry. Since EXT4 structures are
2417 * stored in intel byte order, and the name_len field could never be
2418 * bigger than 255 chars, it's safe to reclaim the extra byte for the
2419 * file_type field.
2420 */
2421 struct ext4_dir_entry_2 {
2422 __le32 inode; /* Inode number */
2423 __le16 rec_len; /* Directory entry length */
2424 __u8 name_len; /* Name length */
2425 __u8 file_type; /* See file type macros EXT4_FT_* below */
2426 char name[EXT4_NAME_LEN]; /* File name */
2427 };
2428
2429 /*
2430 * Access the hashes at the end of ext4_dir_entry_2
2431 */
2432 #define EXT4_DIRENT_HASHES(entry) \
2433 ((struct ext4_dir_entry_hash *) \
2434 (((void *)(entry)) + \
2435 ((8 + (entry)->name_len + EXT4_DIR_ROUND) & ~EXT4_DIR_ROUND)))
2436 #define EXT4_DIRENT_HASH(entry) le32_to_cpu(EXT4_DIRENT_HASHES(entry)->hash)
2437 #define EXT4_DIRENT_MINOR_HASH(entry) \
2438 le32_to_cpu(EXT4_DIRENT_HASHES(entry)->minor_hash)
2439
ext4_hash_in_dirent(const struct inode * inode)2440 static inline bool ext4_hash_in_dirent(const struct inode *inode)
2441 {
2442 return IS_CASEFOLDED(inode) && IS_ENCRYPTED(inode);
2443 }
2444
2445 /*
2446 * This is a bogus directory entry at the end of each leaf block that
2447 * records checksums.
2448 */
2449 struct ext4_dir_entry_tail {
2450 __le32 det_reserved_zero1; /* Pretend to be unused */
2451 __le16 det_rec_len; /* 12 */
2452 __u8 det_reserved_zero2; /* Zero name length */
2453 __u8 det_reserved_ft; /* 0xDE, fake file type */
2454 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
2455 };
2456
2457 #define EXT4_DIRENT_TAIL(block, blocksize) \
2458 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
2459 ((blocksize) - \
2460 sizeof(struct ext4_dir_entry_tail))))
2461
2462 /*
2463 * Ext4 directory file types. Only the low 3 bits are used. The
2464 * other bits are reserved for now.
2465 */
2466 #define EXT4_FT_UNKNOWN 0
2467 #define EXT4_FT_REG_FILE 1
2468 #define EXT4_FT_DIR 2
2469 #define EXT4_FT_CHRDEV 3
2470 #define EXT4_FT_BLKDEV 4
2471 #define EXT4_FT_FIFO 5
2472 #define EXT4_FT_SOCK 6
2473 #define EXT4_FT_SYMLINK 7
2474
2475 #define EXT4_FT_MAX 8
2476
2477 #define EXT4_FT_DIR_CSUM 0xDE
2478
2479 /*
2480 * EXT4_DIR_PAD defines the directory entries boundaries
2481 *
2482 * NOTE: It must be a multiple of 4
2483 */
2484 #define EXT4_DIR_PAD 4
2485 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
2486 #define EXT4_MAX_REC_LEN ((1<<16)-1)
2487
2488 /*
2489 * The rec_len is dependent on the type of directory. Directories that are
2490 * casefolded and encrypted need to store the hash as well, so we add room for
2491 * ext4_extended_dir_entry_2. For all entries related to '.' or '..' you should
2492 * pass NULL for dir, as those entries do not use the extra fields.
2493 */
ext4_dir_rec_len(__u8 name_len,const struct inode * dir)2494 static inline unsigned int ext4_dir_rec_len(__u8 name_len,
2495 const struct inode *dir)
2496 {
2497 int rec_len = (name_len + 8 + EXT4_DIR_ROUND);
2498
2499 if (dir && ext4_hash_in_dirent(dir))
2500 rec_len += sizeof(struct ext4_dir_entry_hash);
2501 return (rec_len & ~EXT4_DIR_ROUND);
2502 }
2503
2504 static inline unsigned int
ext4_rec_len_from_disk(__le16 dlen,unsigned blocksize)2505 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
2506 {
2507 unsigned len = le16_to_cpu(dlen);
2508
2509 if (len == EXT4_MAX_REC_LEN || len == 0)
2510 return blocksize;
2511 return (len & 65532) | ((len & 3) << 16);
2512 }
2513
ext4_rec_len_to_disk(unsigned len,unsigned blocksize)2514 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
2515 {
2516 BUG_ON((len > blocksize) || (blocksize > (1 << 18)) || (len & 3));
2517 if (len < 65536)
2518 return cpu_to_le16(len);
2519 if (len == blocksize) {
2520 if (blocksize == 65536)
2521 return cpu_to_le16(EXT4_MAX_REC_LEN);
2522 else
2523 return cpu_to_le16(0);
2524 }
2525 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
2526 }
2527
2528 /*
2529 * Hash Tree Directory indexing
2530 * (c) Daniel Phillips, 2001
2531 */
2532
2533 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
2534 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2535 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \
2536 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir)))
2537 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2538
2539 /* Legal values for the dx_root hash_version field: */
2540
2541 #define DX_HASH_LEGACY 0
2542 #define DX_HASH_HALF_MD4 1
2543 #define DX_HASH_TEA 2
2544 #define DX_HASH_LEGACY_UNSIGNED 3
2545 #define DX_HASH_HALF_MD4_UNSIGNED 4
2546 #define DX_HASH_TEA_UNSIGNED 5
2547 #define DX_HASH_SIPHASH 6
2548 #define DX_HASH_LAST DX_HASH_SIPHASH
2549
ext4_chksum(u32 crc,const void * address,unsigned int length)2550 static inline u32 ext4_chksum(u32 crc, const void *address, unsigned int length)
2551 {
2552 return crc32c(crc, address, length);
2553 }
2554
2555 #ifdef __KERNEL__
2556
2557 /* hash info structure used by the directory hash */
2558 struct dx_hash_info
2559 {
2560 u32 hash;
2561 u32 minor_hash;
2562 int hash_version;
2563 u32 *seed;
2564 };
2565
2566
2567 /* 32 and 64 bit signed EOF for dx directories */
2568 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
2569 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
2570
2571
2572 /*
2573 * Control parameters used by ext4_htree_next_block
2574 */
2575 #define HASH_NB_ALWAYS 1
2576
2577 struct ext4_filename {
2578 const struct qstr *usr_fname;
2579 struct fscrypt_str disk_name;
2580 struct dx_hash_info hinfo;
2581 #ifdef CONFIG_FS_ENCRYPTION
2582 struct fscrypt_str crypto_buf;
2583 #endif
2584 #if IS_ENABLED(CONFIG_UNICODE)
2585 struct qstr cf_name;
2586 #endif
2587 };
2588
2589 #define fname_name(p) ((p)->disk_name.name)
2590 #define fname_usr_name(p) ((p)->usr_fname->name)
2591 #define fname_len(p) ((p)->disk_name.len)
2592
2593 /*
2594 * Describe an inode's exact location on disk and in memory
2595 */
2596 struct ext4_iloc
2597 {
2598 struct buffer_head *bh;
2599 unsigned long offset;
2600 ext4_group_t block_group;
2601 };
2602
ext4_raw_inode(struct ext4_iloc * iloc)2603 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2604 {
2605 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2606 }
2607
ext4_is_quota_file(struct inode * inode)2608 static inline bool ext4_is_quota_file(struct inode *inode)
2609 {
2610 return IS_NOQUOTA(inode) &&
2611 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL);
2612 }
2613
2614 /*
2615 * This structure is stuffed into the struct file's private_data field
2616 * for directories. It is where we put information so that we can do
2617 * readdir operations in hash tree order.
2618 */
2619 struct dir_private_info {
2620 struct rb_root root;
2621 struct rb_node *curr_node;
2622 struct fname *extra_fname;
2623 loff_t last_pos;
2624 __u32 curr_hash;
2625 __u32 curr_minor_hash;
2626 __u32 next_hash;
2627 u64 cookie;
2628 bool initialized;
2629 };
2630
2631 /* calculate the first block number of the group */
2632 static inline ext4_fsblk_t
ext4_group_first_block_no(struct super_block * sb,ext4_group_t group_no)2633 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2634 {
2635 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2636 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2637 }
2638
2639 /*
2640 * Special error return code only used by dx_probe() and its callers.
2641 */
2642 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
2643
2644 /* htree levels for ext4 */
2645 #define EXT4_HTREE_LEVEL_COMPAT 2
2646 #define EXT4_HTREE_LEVEL 3
2647
ext4_dir_htree_level(struct super_block * sb)2648 static inline int ext4_dir_htree_level(struct super_block *sb)
2649 {
2650 return ext4_has_feature_largedir(sb) ?
2651 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT;
2652 }
2653
2654 /*
2655 * Timeout and state flag for lazy initialization inode thread.
2656 */
2657 #define EXT4_DEF_LI_WAIT_MULT 10
2658 #define EXT4_DEF_LI_MAX_START_DELAY 5
2659 #define EXT4_LAZYINIT_QUIT 0x0001
2660 #define EXT4_LAZYINIT_RUNNING 0x0002
2661
2662 /*
2663 * Lazy inode table initialization info
2664 */
2665 struct ext4_lazy_init {
2666 unsigned long li_state;
2667 struct list_head li_request_list;
2668 struct mutex li_list_mtx;
2669 };
2670
2671 enum ext4_li_mode {
2672 EXT4_LI_MODE_PREFETCH_BBITMAP,
2673 EXT4_LI_MODE_ITABLE,
2674 };
2675
2676 struct ext4_li_request {
2677 struct super_block *lr_super;
2678 enum ext4_li_mode lr_mode;
2679 ext4_group_t lr_first_not_zeroed;
2680 ext4_group_t lr_next_group;
2681 struct list_head lr_request;
2682 unsigned long lr_next_sched;
2683 unsigned long lr_timeout;
2684 };
2685
2686 struct ext4_features {
2687 struct kobject f_kobj;
2688 struct completion f_kobj_unregister;
2689 };
2690
2691 /*
2692 * This structure will be used for multiple mount protection. It will be
2693 * written into the block number saved in the s_mmp_block field in the
2694 * superblock. Programs that check MMP should assume that if
2695 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2696 * to use the filesystem, regardless of how old the timestamp is.
2697 */
2698 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
2699 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2700 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
2701 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
2702
2703 struct mmp_struct {
2704 __le32 mmp_magic; /* Magic number for MMP */
2705 __le32 mmp_seq; /* Sequence no. updated periodically */
2706
2707 /*
2708 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2709 * purposes and do not affect the correctness of the algorithm
2710 */
2711 __le64 mmp_time; /* Time last updated */
2712 char mmp_nodename[64]; /* Node which last updated MMP block */
2713 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
2714
2715 /*
2716 * mmp_check_interval is used to verify if the MMP block has been
2717 * updated on the block device. The value is updated based on the
2718 * maximum time to write the MMP block during an update cycle.
2719 */
2720 __le16 mmp_check_interval;
2721
2722 __le16 mmp_pad1;
2723 __le32 mmp_pad2[226];
2724 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
2725 };
2726
2727 /* arguments passed to the mmp thread */
2728 struct mmpd_data {
2729 struct buffer_head *bh; /* bh from initial read_mmp_block() */
2730 struct super_block *sb; /* super block of the fs */
2731 };
2732
2733 /*
2734 * Check interval multiplier
2735 * The MMP block is written every update interval and initially checked every
2736 * update interval x the multiplier (the value is then adapted based on the
2737 * write latency). The reason is that writes can be delayed under load and we
2738 * don't want readers to incorrectly assume that the filesystem is no longer
2739 * in use.
2740 */
2741 #define EXT4_MMP_CHECK_MULT 2UL
2742
2743 /*
2744 * Minimum interval for MMP checking in seconds.
2745 */
2746 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
2747
2748 /*
2749 * Maximum interval for MMP checking in seconds.
2750 */
2751 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
2752
2753 /*
2754 * Function prototypes
2755 */
2756
2757 /*
2758 * Ok, these declarations are also in <linux/kernel.h> but none of the
2759 * ext4 source programs needs to include it so they are duplicated here.
2760 */
2761 # define NORET_TYPE /**/
2762 # define ATTRIB_NORET __attribute__((noreturn))
2763 # define NORET_AND noreturn,
2764
2765 /* bitmap.c */
2766 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2767 void ext4_inode_bitmap_csum_set(struct super_block *sb,
2768 struct ext4_group_desc *gdp,
2769 struct buffer_head *bh);
2770 int ext4_inode_bitmap_csum_verify(struct super_block *sb,
2771 struct ext4_group_desc *gdp,
2772 struct buffer_head *bh);
2773 void ext4_block_bitmap_csum_set(struct super_block *sb,
2774 struct ext4_group_desc *gdp,
2775 struct buffer_head *bh);
2776 int ext4_block_bitmap_csum_verify(struct super_block *sb,
2777 struct ext4_group_desc *gdp,
2778 struct buffer_head *bh);
2779
2780 /* balloc.c */
2781 extern void ext4_get_group_no_and_offset(struct super_block *sb,
2782 ext4_fsblk_t blocknr,
2783 ext4_group_t *blockgrpp,
2784 ext4_grpblk_t *offsetp);
2785 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2786 ext4_fsblk_t block);
2787
2788 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2789 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2790 ext4_group_t group);
2791 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2792 ext4_fsblk_t goal,
2793 unsigned int flags,
2794 unsigned long *count,
2795 int *errp);
2796 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2797 s64 nclusters, unsigned int flags);
2798 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2799 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2800 ext4_group_t block_group,
2801 struct buffer_head ** bh);
2802 extern struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2803 ext4_group_t group);
2804 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2805
2806 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2807 ext4_group_t block_group,
2808 bool ignore_locked);
2809 extern int ext4_wait_block_bitmap(struct super_block *sb,
2810 ext4_group_t block_group,
2811 struct buffer_head *bh);
2812 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2813 ext4_group_t block_group);
2814 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2815 ext4_group_t block_group,
2816 struct ext4_group_desc *gdp);
2817 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2818
2819 #if IS_ENABLED(CONFIG_UNICODE)
2820 extern int ext4_fname_setup_ci_filename(struct inode *dir,
2821 const struct qstr *iname,
2822 struct ext4_filename *fname);
2823
ext4_fname_free_ci_filename(struct ext4_filename * fname)2824 static inline void ext4_fname_free_ci_filename(struct ext4_filename *fname)
2825 {
2826 kfree(fname->cf_name.name);
2827 fname->cf_name.name = NULL;
2828 }
2829 #else
ext4_fname_setup_ci_filename(struct inode * dir,const struct qstr * iname,struct ext4_filename * fname)2830 static inline int ext4_fname_setup_ci_filename(struct inode *dir,
2831 const struct qstr *iname,
2832 struct ext4_filename *fname)
2833 {
2834 return 0;
2835 }
2836
ext4_fname_free_ci_filename(struct ext4_filename * fname)2837 static inline void ext4_fname_free_ci_filename(struct ext4_filename *fname)
2838 {
2839 }
2840 #endif
2841
2842 /* ext4 encryption related stuff goes here crypto.c */
2843 #ifdef CONFIG_FS_ENCRYPTION
2844 extern const struct fscrypt_operations ext4_cryptops;
2845
2846 int ext4_fname_setup_filename(struct inode *dir, const struct qstr *iname,
2847 int lookup, struct ext4_filename *fname);
2848
2849 int ext4_fname_prepare_lookup(struct inode *dir, struct dentry *dentry,
2850 struct ext4_filename *fname);
2851
2852 void ext4_fname_free_filename(struct ext4_filename *fname);
2853
2854 int ext4_ioctl_get_encryption_pwsalt(struct file *filp, void __user *arg);
2855
2856 #else /* !CONFIG_FS_ENCRYPTION */
ext4_fname_setup_filename(struct inode * dir,const struct qstr * iname,int lookup,struct ext4_filename * fname)2857 static inline int ext4_fname_setup_filename(struct inode *dir,
2858 const struct qstr *iname,
2859 int lookup,
2860 struct ext4_filename *fname)
2861 {
2862 fname->usr_fname = iname;
2863 fname->disk_name.name = (unsigned char *) iname->name;
2864 fname->disk_name.len = iname->len;
2865
2866 return ext4_fname_setup_ci_filename(dir, iname, fname);
2867 }
2868
ext4_fname_prepare_lookup(struct inode * dir,struct dentry * dentry,struct ext4_filename * fname)2869 static inline int ext4_fname_prepare_lookup(struct inode *dir,
2870 struct dentry *dentry,
2871 struct ext4_filename *fname)
2872 {
2873 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname);
2874 }
2875
ext4_fname_free_filename(struct ext4_filename * fname)2876 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2877 {
2878 ext4_fname_free_ci_filename(fname);
2879 }
2880
ext4_ioctl_get_encryption_pwsalt(struct file * filp,void __user * arg)2881 static inline int ext4_ioctl_get_encryption_pwsalt(struct file *filp,
2882 void __user *arg)
2883 {
2884 return -EOPNOTSUPP;
2885 }
2886 #endif /* !CONFIG_FS_ENCRYPTION */
2887
2888 /* dir.c */
2889 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2890 struct file *,
2891 struct ext4_dir_entry_2 *,
2892 struct buffer_head *, char *, int,
2893 unsigned int);
2894 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2895 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2896 (de), (bh), (buf), (size), (offset)))
2897 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2898 __u32 minor_hash,
2899 struct ext4_dir_entry_2 *dirent,
2900 struct fscrypt_str *ent_name);
2901 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2902 extern int ext4_find_dest_de(struct inode *dir, struct buffer_head *bh,
2903 void *buf, int buf_size,
2904 struct ext4_filename *fname,
2905 struct ext4_dir_entry_2 **dest_de);
2906 void ext4_insert_dentry(struct inode *dir, struct inode *inode,
2907 struct ext4_dir_entry_2 *de,
2908 int buf_size,
2909 struct ext4_filename *fname);
ext4_update_dx_flag(struct inode * inode)2910 static inline void ext4_update_dx_flag(struct inode *inode)
2911 {
2912 if (!ext4_has_feature_dir_index(inode->i_sb) &&
2913 ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) {
2914 /* ext4_iget() should have caught this... */
2915 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb));
2916 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2917 }
2918 }
2919 static const unsigned char ext4_filetype_table[] = {
2920 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2921 };
2922
get_dtype(struct super_block * sb,int filetype)2923 static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2924 {
2925 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2926 return DT_UNKNOWN;
2927
2928 return ext4_filetype_table[filetype];
2929 }
2930 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2931 void *buf, int buf_size);
2932
2933 /* fsync.c */
2934 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2935
2936 /* hash.c */
2937 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len,
2938 struct dx_hash_info *hinfo);
2939
2940 /* ialloc.c */
2941 extern int ext4_mark_inode_used(struct super_block *sb, int ino);
2942 extern struct inode *__ext4_new_inode(struct mnt_idmap *, handle_t *,
2943 struct inode *, umode_t,
2944 const struct qstr *qstr, __u32 goal,
2945 uid_t *owner, __u32 i_flags,
2946 int handle_type, unsigned int line_no,
2947 int nblocks);
2948
2949 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \
2950 __ext4_new_inode(&nop_mnt_idmap, (handle), (dir), (mode), (qstr), \
2951 (goal), (owner), i_flags, 0, 0, 0)
2952 #define ext4_new_inode_start_handle(idmap, dir, mode, qstr, goal, owner, \
2953 type, nblocks) \
2954 __ext4_new_inode((idmap), NULL, (dir), (mode), (qstr), (goal), (owner), \
2955 0, (type), __LINE__, (nblocks))
2956
2957
2958 extern void ext4_free_inode(handle_t *, struct inode *);
2959 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2960 extern unsigned long ext4_count_free_inodes(struct super_block *);
2961 extern unsigned long ext4_count_dirs(struct super_block *);
2962 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2963 extern int ext4_init_inode_table(struct super_block *sb,
2964 ext4_group_t group, int barrier);
2965 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2966
2967 /* fast_commit.c */
2968 int ext4_fc_info_show(struct seq_file *seq, void *v);
2969 void ext4_fc_init(struct super_block *sb, journal_t *journal);
2970 void ext4_fc_init_inode(struct inode *inode);
2971 void ext4_fc_track_range(handle_t *handle, struct inode *inode, ext4_lblk_t start,
2972 ext4_lblk_t end);
2973 void __ext4_fc_track_unlink(handle_t *handle, struct inode *inode,
2974 struct dentry *dentry);
2975 void __ext4_fc_track_link(handle_t *handle, struct inode *inode,
2976 struct dentry *dentry);
2977 void ext4_fc_track_unlink(handle_t *handle, struct dentry *dentry);
2978 void ext4_fc_track_link(handle_t *handle, struct dentry *dentry);
2979 void __ext4_fc_track_create(handle_t *handle, struct inode *inode,
2980 struct dentry *dentry);
2981 void ext4_fc_track_create(handle_t *handle, struct dentry *dentry);
2982 void ext4_fc_track_inode(handle_t *handle, struct inode *inode);
2983 void ext4_fc_mark_ineligible(struct super_block *sb, int reason, handle_t *handle);
2984 void ext4_fc_del(struct inode *inode);
2985 bool ext4_fc_replay_check_excluded(struct super_block *sb, ext4_fsblk_t block);
2986 void ext4_fc_replay_cleanup(struct super_block *sb);
2987 int ext4_fc_commit(journal_t *journal, tid_t commit_tid);
2988 int __init ext4_fc_init_dentry_cache(void);
2989 void ext4_fc_destroy_dentry_cache(void);
2990 int ext4_fc_record_regions(struct super_block *sb, int ino,
2991 ext4_lblk_t lblk, ext4_fsblk_t pblk,
2992 int len, int replay);
2993
2994 /* mballoc.c */
2995 extern const struct seq_operations ext4_mb_seq_groups_ops;
2996 extern const struct seq_operations ext4_mb_seq_structs_summary_ops;
2997 extern int ext4_seq_mb_stats_show(struct seq_file *seq, void *offset);
2998 extern int ext4_mb_init(struct super_block *);
2999 extern void ext4_mb_release(struct super_block *);
3000 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
3001 struct ext4_allocation_request *, int *);
3002 extern void ext4_discard_preallocations(struct inode *);
3003 extern int __init ext4_init_mballoc(void);
3004 extern void ext4_exit_mballoc(void);
3005 extern ext4_group_t ext4_mb_prefetch(struct super_block *sb,
3006 ext4_group_t group,
3007 unsigned int nr, int *cnt);
3008 extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group,
3009 unsigned int nr);
3010
3011 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
3012 struct buffer_head *bh, ext4_fsblk_t block,
3013 unsigned long count, int flags);
3014 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
3015 ext4_group_t ngroups);
3016 extern int ext4_mb_add_groupinfo(struct super_block *sb,
3017 ext4_group_t i, struct ext4_group_desc *desc);
3018 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
3019 ext4_fsblk_t block, unsigned long count);
3020 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
3021 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid);
3022 extern void ext4_mb_mark_bb(struct super_block *sb, ext4_fsblk_t block,
3023 int len, bool state);
ext4_mb_cr_expensive(enum criteria cr)3024 static inline bool ext4_mb_cr_expensive(enum criteria cr)
3025 {
3026 return cr >= CR_GOAL_LEN_SLOW;
3027 }
3028
3029 /* inode.c */
3030 void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw,
3031 struct ext4_inode_info *ei);
3032 int ext4_inode_is_fast_symlink(struct inode *inode);
3033 void ext4_check_map_extents_env(struct inode *inode);
3034 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
3035 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
3036 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count,
3037 bool wait, struct buffer_head **bhs);
3038 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
3039 struct buffer_head *bh_result, int create);
3040 int ext4_get_block(struct inode *inode, sector_t iblock,
3041 struct buffer_head *bh_result, int create);
3042 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
3043 struct buffer_head *bh, int create);
3044 int ext4_walk_page_buffers(handle_t *handle,
3045 struct inode *inode,
3046 struct buffer_head *head,
3047 unsigned from,
3048 unsigned to,
3049 int *partial,
3050 int (*fn)(handle_t *handle, struct inode *inode,
3051 struct buffer_head *bh));
3052 int do_journal_get_write_access(handle_t *handle, struct inode *inode,
3053 struct buffer_head *bh);
3054 void ext4_set_inode_mapping_order(struct inode *inode);
3055 #define FALL_BACK_TO_NONDELALLOC 1
3056 #define CONVERT_INLINE_DATA 2
3057
3058 typedef enum {
3059 EXT4_IGET_NORMAL = 0,
3060 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */
3061 EXT4_IGET_HANDLE = 0x0002, /* Inode # is from a handle */
3062 EXT4_IGET_BAD = 0x0004, /* Allow to iget a bad inode */
3063 EXT4_IGET_EA_INODE = 0x0008 /* Inode should contain an EA value */
3064 } ext4_iget_flags;
3065
3066 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
3067 ext4_iget_flags flags, const char *function,
3068 unsigned int line);
3069
3070 #define ext4_iget(sb, ino, flags) \
3071 __ext4_iget((sb), (ino), (flags), __func__, __LINE__)
3072
3073 extern int ext4_write_inode(struct inode *, struct writeback_control *);
3074 extern int ext4_setattr(struct mnt_idmap *, struct dentry *,
3075 struct iattr *);
3076 extern u32 ext4_dio_alignment(struct inode *inode);
3077 extern int ext4_getattr(struct mnt_idmap *, const struct path *,
3078 struct kstat *, u32, unsigned int);
3079 extern void ext4_evict_inode(struct inode *);
3080 extern void ext4_clear_inode(struct inode *);
3081 extern int ext4_file_getattr(struct mnt_idmap *, const struct path *,
3082 struct kstat *, u32, unsigned int);
3083 extern void ext4_dirty_inode(struct inode *, int);
3084 extern int ext4_change_inode_journal_flag(struct inode *, int);
3085 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
3086 extern int ext4_get_fc_inode_loc(struct super_block *sb, unsigned long ino,
3087 struct ext4_iloc *iloc);
3088 extern int ext4_inode_attach_jinode(struct inode *inode);
3089 extern int ext4_can_truncate(struct inode *inode);
3090 extern int ext4_truncate(struct inode *);
3091 extern int ext4_break_layouts(struct inode *);
3092 extern int ext4_truncate_page_cache_block_range(struct inode *inode,
3093 loff_t start, loff_t end);
3094 extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length);
3095 extern void ext4_set_inode_flags(struct inode *, bool init);
3096 extern int ext4_alloc_da_blocks(struct inode *inode);
3097 extern void ext4_set_aops(struct inode *inode);
3098 extern int ext4_normal_submit_inode_data_buffers(struct jbd2_inode *jinode);
3099 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
3100 extern int ext4_chunk_trans_extent(struct inode *inode, int nrblocks);
3101 extern int ext4_meta_trans_blocks(struct inode *inode, int lblocks,
3102 int pextents);
3103 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
3104 loff_t lstart, loff_t lend);
3105 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf);
3106 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
3107 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
3108 extern void ext4_da_release_space(struct inode *inode, int to_free);
3109 extern void ext4_da_update_reserve_space(struct inode *inode,
3110 int used, int quota_claim);
3111 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
3112 ext4_fsblk_t pblk, ext4_lblk_t len);
3113
is_special_ino(struct super_block * sb,unsigned long ino)3114 static inline bool is_special_ino(struct super_block *sb, unsigned long ino)
3115 {
3116 struct ext4_super_block *es = EXT4_SB(sb)->s_es;
3117
3118 return (ino < EXT4_FIRST_INO(sb) && ino != EXT4_ROOT_INO) ||
3119 ino == le32_to_cpu(es->s_usr_quota_inum) ||
3120 ino == le32_to_cpu(es->s_grp_quota_inum) ||
3121 ino == le32_to_cpu(es->s_prj_quota_inum) ||
3122 ino == le32_to_cpu(es->s_orphan_file_inum);
3123 }
3124
3125 /* indirect.c */
3126 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
3127 struct ext4_map_blocks *map, int flags);
3128 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
3129 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
3130 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
3131 ext4_lblk_t start, ext4_lblk_t end);
3132
3133 /* ioctl.c */
3134 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
3135 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
3136 int ext4_fileattr_set(struct mnt_idmap *idmap,
3137 struct dentry *dentry, struct file_kattr *fa);
3138 int ext4_fileattr_get(struct dentry *dentry, struct file_kattr *fa);
3139 extern void ext4_reset_inode_seed(struct inode *inode);
3140 int ext4_update_overhead(struct super_block *sb, bool force);
3141 int ext4_force_shutdown(struct super_block *sb, u32 flags);
3142
3143 /* migrate.c */
3144 extern int ext4_ext_migrate(struct inode *);
3145 extern int ext4_ind_migrate(struct inode *inode);
3146
3147 /* namei.c */
3148 extern int ext4_init_new_dir(handle_t *handle, struct inode *dir,
3149 struct inode *inode);
3150 extern int ext4_dirblock_csum_verify(struct inode *inode,
3151 struct buffer_head *bh);
3152 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
3153 __u32 start_minor_hash, __u32 *next_hash);
3154 extern int ext4_search_dir(struct buffer_head *bh,
3155 char *search_buf,
3156 int buf_size,
3157 struct inode *dir,
3158 struct ext4_filename *fname,
3159 unsigned int offset,
3160 struct ext4_dir_entry_2 **res_dir);
3161 extern int ext4_generic_delete_entry(struct inode *dir,
3162 struct ext4_dir_entry_2 *de_del,
3163 struct buffer_head *bh,
3164 void *entry_buf,
3165 int buf_size,
3166 int csum_size);
3167 extern bool ext4_empty_dir(struct inode *inode);
3168
3169 /* resize.c */
3170 extern void ext4_kvfree_array_rcu(void *to_free);
3171 extern int ext4_group_add(struct super_block *sb,
3172 struct ext4_new_group_data *input);
3173 extern int ext4_group_extend(struct super_block *sb,
3174 struct ext4_super_block *es,
3175 ext4_fsblk_t n_blocks_count);
3176 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
3177 extern unsigned int ext4_list_backups(struct super_block *sb,
3178 unsigned int *three, unsigned int *five,
3179 unsigned int *seven);
3180
3181 /* super.c */
3182 extern struct buffer_head *ext4_sb_bread(struct super_block *sb,
3183 sector_t block, blk_opf_t op_flags);
3184 extern struct buffer_head *ext4_sb_bread_unmovable(struct super_block *sb,
3185 sector_t block);
3186 extern struct buffer_head *ext4_sb_bread_nofail(struct super_block *sb,
3187 sector_t block);
3188 extern void ext4_read_bh_nowait(struct buffer_head *bh, blk_opf_t op_flags,
3189 bh_end_io_t *end_io, bool simu_fail);
3190 extern int ext4_read_bh(struct buffer_head *bh, blk_opf_t op_flags,
3191 bh_end_io_t *end_io, bool simu_fail);
3192 extern int ext4_read_bh_lock(struct buffer_head *bh, blk_opf_t op_flags, bool wait);
3193 extern void ext4_sb_breadahead_unmovable(struct super_block *sb, sector_t block);
3194 extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
3195 extern int ext4_calculate_overhead(struct super_block *sb);
3196 extern __le32 ext4_superblock_csum(struct ext4_super_block *es);
3197 extern void ext4_superblock_csum_set(struct super_block *sb);
3198 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
3199 ext4_group_t ngroup);
3200 extern const char *ext4_decode_error(struct super_block *sb, int errno,
3201 char nbuf[16]);
3202 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb,
3203 ext4_group_t block_group,
3204 unsigned int flags);
3205 extern unsigned int ext4_num_base_meta_blocks(struct super_block *sb,
3206 ext4_group_t block_group);
3207 extern void print_daily_error_info(struct timer_list *t);
3208
3209 extern __printf(7, 8)
3210 void __ext4_error(struct super_block *, const char *, unsigned int, bool,
3211 int, __u64, const char *, ...);
3212 extern __printf(6, 7)
3213 void __ext4_error_inode(struct inode *, const char *, unsigned int,
3214 ext4_fsblk_t, int, const char *, ...);
3215 extern __printf(5, 6)
3216 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
3217 const char *, ...);
3218 extern void __ext4_std_error(struct super_block *, const char *,
3219 unsigned int, int);
3220 extern __printf(4, 5)
3221 void __ext4_warning(struct super_block *, const char *, unsigned int,
3222 const char *, ...);
3223 extern __printf(4, 5)
3224 void __ext4_warning_inode(const struct inode *inode, const char *function,
3225 unsigned int line, const char *fmt, ...);
3226 extern __printf(3, 4)
3227 void __ext4_msg(struct super_block *, const char *, const char *, ...);
3228 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
3229 const char *, unsigned int, const char *);
3230 extern __printf(7, 8)
3231 void __ext4_grp_locked_error(const char *, unsigned int,
3232 struct super_block *, ext4_group_t,
3233 unsigned long, ext4_fsblk_t,
3234 const char *, ...);
3235
3236 #define EXT4_ERROR_INODE(inode, fmt, a...) \
3237 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
3238
3239 #define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...) \
3240 __ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a)
3241
3242 #define ext4_error_inode_block(inode, block, err, fmt, a...) \
3243 __ext4_error_inode((inode), __func__, __LINE__, (block), (err), \
3244 (fmt), ## a)
3245
3246 #define EXT4_ERROR_FILE(file, block, fmt, a...) \
3247 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
3248
3249 #define ext4_abort(sb, err, fmt, a...) \
3250 __ext4_error((sb), __func__, __LINE__, true, (err), 0, (fmt), ## a)
3251
3252 #ifdef CONFIG_PRINTK
3253
3254 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
3255 __ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__)
3256 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \
3257 __ext4_error_inode((inode), (func), (line), (block), \
3258 (err), (fmt), ##__VA_ARGS__)
3259 #define ext4_error_file(file, func, line, block, fmt, ...) \
3260 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
3261 #define ext4_error(sb, fmt, ...) \
3262 __ext4_error((sb), __func__, __LINE__, false, 0, 0, (fmt), \
3263 ##__VA_ARGS__)
3264 #define ext4_error_err(sb, err, fmt, ...) \
3265 __ext4_error((sb), __func__, __LINE__, false, (err), 0, (fmt), \
3266 ##__VA_ARGS__)
3267 #define ext4_warning(sb, fmt, ...) \
3268 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
3269 #define ext4_warning_inode(inode, fmt, ...) \
3270 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
3271 #define ext4_msg(sb, level, fmt, ...) \
3272 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
3273 #define dump_mmp_msg(sb, mmp, msg) \
3274 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
3275 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
3276 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
3277 fmt, ##__VA_ARGS__)
3278
3279 #else
3280
3281 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
3282 do { \
3283 no_printk(fmt, ##__VA_ARGS__); \
3284 __ext4_error_inode(inode, "", 0, block, 0, " "); \
3285 } while (0)
3286 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \
3287 do { \
3288 no_printk(fmt, ##__VA_ARGS__); \
3289 __ext4_error_inode(inode, "", 0, block, err, " "); \
3290 } while (0)
3291 #define ext4_error_file(file, func, line, block, fmt, ...) \
3292 do { \
3293 no_printk(fmt, ##__VA_ARGS__); \
3294 __ext4_error_file(file, "", 0, block, " "); \
3295 } while (0)
3296 #define ext4_error(sb, fmt, ...) \
3297 do { \
3298 no_printk(fmt, ##__VA_ARGS__); \
3299 __ext4_error(sb, "", 0, false, 0, 0, " "); \
3300 } while (0)
3301 #define ext4_error_err(sb, err, fmt, ...) \
3302 do { \
3303 no_printk(fmt, ##__VA_ARGS__); \
3304 __ext4_error(sb, "", 0, false, err, 0, " "); \
3305 } while (0)
3306 #define ext4_warning(sb, fmt, ...) \
3307 do { \
3308 no_printk(fmt, ##__VA_ARGS__); \
3309 __ext4_warning(sb, "", 0, " "); \
3310 } while (0)
3311 #define ext4_warning_inode(inode, fmt, ...) \
3312 do { \
3313 no_printk(fmt, ##__VA_ARGS__); \
3314 __ext4_warning_inode(inode, "", 0, " "); \
3315 } while (0)
3316 #define ext4_msg(sb, level, fmt, ...) \
3317 do { \
3318 no_printk(fmt, ##__VA_ARGS__); \
3319 __ext4_msg(sb, "", " "); \
3320 } while (0)
3321 #define dump_mmp_msg(sb, mmp, msg) \
3322 __dump_mmp_msg(sb, mmp, "", 0, "")
3323 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
3324 do { \
3325 no_printk(fmt, ##__VA_ARGS__); \
3326 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
3327 } while (0)
3328
3329 #endif
3330
3331 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
3332 struct ext4_group_desc *bg);
3333 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
3334 struct ext4_group_desc *bg);
3335 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
3336 struct ext4_group_desc *bg);
3337 extern __u32 ext4_free_group_clusters(struct super_block *sb,
3338 struct ext4_group_desc *bg);
3339 extern __u32 ext4_free_inodes_count(struct super_block *sb,
3340 struct ext4_group_desc *bg);
3341 extern __u32 ext4_used_dirs_count(struct super_block *sb,
3342 struct ext4_group_desc *bg);
3343 extern __u32 ext4_itable_unused_count(struct super_block *sb,
3344 struct ext4_group_desc *bg);
3345 extern void ext4_block_bitmap_set(struct super_block *sb,
3346 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3347 extern void ext4_inode_bitmap_set(struct super_block *sb,
3348 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3349 extern void ext4_inode_table_set(struct super_block *sb,
3350 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3351 extern void ext4_free_group_clusters_set(struct super_block *sb,
3352 struct ext4_group_desc *bg,
3353 __u32 count);
3354 extern void ext4_free_inodes_set(struct super_block *sb,
3355 struct ext4_group_desc *bg, __u32 count);
3356 extern void ext4_used_dirs_set(struct super_block *sb,
3357 struct ext4_group_desc *bg, __u32 count);
3358 extern void ext4_itable_unused_set(struct super_block *sb,
3359 struct ext4_group_desc *bg, __u32 count);
3360 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
3361 struct ext4_group_desc *gdp);
3362 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
3363 struct ext4_group_desc *gdp);
3364 extern int ext4_register_li_request(struct super_block *sb,
3365 ext4_group_t first_not_zeroed);
3366
ext4_has_group_desc_csum(struct super_block * sb)3367 static inline int ext4_has_group_desc_csum(struct super_block *sb)
3368 {
3369 return ext4_has_feature_gdt_csum(sb) ||
3370 ext4_has_feature_metadata_csum(sb);
3371 }
3372
3373 #define ext4_read_incompat_64bit_val(es, name) \
3374 (((es)->s_feature_incompat & cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT) \
3375 ? (ext4_fsblk_t)le32_to_cpu(es->name##_hi) << 32 : 0) | \
3376 le32_to_cpu(es->name##_lo))
3377
ext4_blocks_count(struct ext4_super_block * es)3378 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
3379 {
3380 return ext4_read_incompat_64bit_val(es, s_blocks_count);
3381 }
3382
ext4_r_blocks_count(struct ext4_super_block * es)3383 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
3384 {
3385 return ext4_read_incompat_64bit_val(es, s_r_blocks_count);
3386 }
3387
ext4_free_blocks_count(struct ext4_super_block * es)3388 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
3389 {
3390 return ext4_read_incompat_64bit_val(es, s_free_blocks_count);
3391 }
3392
ext4_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)3393 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
3394 ext4_fsblk_t blk)
3395 {
3396 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
3397 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
3398 }
3399
ext4_free_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)3400 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
3401 ext4_fsblk_t blk)
3402 {
3403 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
3404 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
3405 }
3406
ext4_r_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)3407 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
3408 ext4_fsblk_t blk)
3409 {
3410 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
3411 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
3412 }
3413
ext4_isize(struct super_block * sb,struct ext4_inode * raw_inode)3414 static inline loff_t ext4_isize(struct super_block *sb,
3415 struct ext4_inode *raw_inode)
3416 {
3417 if (ext4_has_feature_largedir(sb) ||
3418 S_ISREG(le16_to_cpu(raw_inode->i_mode)))
3419 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
3420 le32_to_cpu(raw_inode->i_size_lo);
3421
3422 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
3423 }
3424
ext4_isize_set(struct ext4_inode * raw_inode,loff_t i_size)3425 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
3426 {
3427 raw_inode->i_size_lo = cpu_to_le32(i_size);
3428 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
3429 }
3430
3431 /*
3432 * Reading s_groups_count requires using smp_rmb() afterwards. See
3433 * the locking protocol documented in the comments of ext4_group_add()
3434 * in resize.c
3435 */
ext4_get_groups_count(struct super_block * sb)3436 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
3437 {
3438 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
3439
3440 smp_rmb();
3441 return ngroups;
3442 }
3443
ext4_flex_group(struct ext4_sb_info * sbi,ext4_group_t block_group)3444 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
3445 ext4_group_t block_group)
3446 {
3447 return block_group >> sbi->s_log_groups_per_flex;
3448 }
3449
ext4_flex_bg_size(struct ext4_sb_info * sbi)3450 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
3451 {
3452 return 1 << sbi->s_log_groups_per_flex;
3453 }
3454
ext4_get_maxbytes(struct inode * inode)3455 static inline loff_t ext4_get_maxbytes(struct inode *inode)
3456 {
3457 if (ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS))
3458 return inode->i_sb->s_maxbytes;
3459 return EXT4_SB(inode->i_sb)->s_bitmap_maxbytes;
3460 }
3461
3462 #define ext4_std_error(sb, errno) \
3463 do { \
3464 if ((errno)) \
3465 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
3466 } while (0)
3467
3468 #ifdef CONFIG_SMP
3469 /* Each CPU can accumulate percpu_counter_batch clusters in their local
3470 * counters. So we need to make sure we have free clusters more
3471 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
3472 */
3473 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
3474 #else
3475 #define EXT4_FREECLUSTERS_WATERMARK 0
3476 #endif
3477
3478 /* Update i_disksize. Requires i_rwsem to avoid races with truncate */
ext4_update_i_disksize(struct inode * inode,loff_t newsize)3479 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
3480 {
3481 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
3482 !inode_is_locked(inode));
3483 down_write(&EXT4_I(inode)->i_data_sem);
3484 if (newsize > EXT4_I(inode)->i_disksize)
3485 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize);
3486 up_write(&EXT4_I(inode)->i_data_sem);
3487 }
3488
3489 /* Update i_size, i_disksize. Requires i_rwsem to avoid races with truncate */
ext4_update_inode_size(struct inode * inode,loff_t newsize)3490 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
3491 {
3492 int changed = 0;
3493
3494 if (newsize > inode->i_size) {
3495 i_size_write(inode, newsize);
3496 changed = 1;
3497 }
3498 if (newsize > EXT4_I(inode)->i_disksize) {
3499 ext4_update_i_disksize(inode, newsize);
3500 changed |= 2;
3501 }
3502 return changed;
3503 }
3504
3505 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
3506 loff_t len);
3507
3508 struct ext4_group_info {
3509 unsigned long bb_state;
3510 #ifdef AGGRESSIVE_CHECK
3511 unsigned long bb_check_counter;
3512 #endif
3513 struct rb_root bb_free_root;
3514 ext4_grpblk_t bb_first_free; /* first free block */
3515 ext4_grpblk_t bb_free; /* total free blocks */
3516 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
3517 int bb_avg_fragment_size_order; /* order of average
3518 fragment in BG */
3519 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
3520 ext4_group_t bb_group; /* Group number */
3521 struct list_head bb_prealloc_list;
3522 #ifdef DOUBLE_CHECK
3523 void *bb_bitmap;
3524 #endif
3525 struct rw_semaphore alloc_sem;
3526 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
3527 * regions, index is order.
3528 * bb_counters[3] = 5 means
3529 * 5 free 8-block regions. */
3530 };
3531
3532 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
3533 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
3534 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
3535 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
3536 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \
3537 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT)
3538 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \
3539 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT)
3540 #define EXT4_GROUP_INFO_BBITMAP_READ_BIT 4
3541
3542 #define EXT4_MB_GRP_NEED_INIT(grp) \
3543 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
3544 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
3545 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3546 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
3547 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3548
3549 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
3550 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3551 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
3552 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3553 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
3554 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3555 #define EXT4_MB_GRP_TEST_AND_SET_READ(grp) \
3556 (test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state)))
3557
3558 #define EXT4_MAX_CONTENTION 8
3559 #define EXT4_CONTENTION_THRESHOLD 2
3560
ext4_group_lock_ptr(struct super_block * sb,ext4_group_t group)3561 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
3562 ext4_group_t group)
3563 {
3564 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
3565 }
3566
3567 /*
3568 * Returns true if the filesystem is busy enough that attempts to
3569 * access the block group locks has run into contention.
3570 */
ext4_fs_is_busy(struct ext4_sb_info * sbi)3571 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
3572 {
3573 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
3574 }
3575
ext4_try_lock_group(struct super_block * sb,ext4_group_t group)3576 static inline bool ext4_try_lock_group(struct super_block *sb, ext4_group_t group)
3577 {
3578 if (!spin_trylock(ext4_group_lock_ptr(sb, group)))
3579 return false;
3580 /*
3581 * We're able to grab the lock right away, so drop the lock
3582 * contention counter.
3583 */
3584 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
3585 return true;
3586 }
3587
ext4_lock_group(struct super_block * sb,ext4_group_t group)3588 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
3589 {
3590 if (!ext4_try_lock_group(sb, group)) {
3591 /*
3592 * The lock is busy, so bump the contention counter,
3593 * and then wait on the spin lock.
3594 */
3595 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
3596 EXT4_MAX_CONTENTION);
3597 spin_lock(ext4_group_lock_ptr(sb, group));
3598 }
3599 }
3600
ext4_unlock_group(struct super_block * sb,ext4_group_t group)3601 static inline void ext4_unlock_group(struct super_block *sb,
3602 ext4_group_t group)
3603 {
3604 spin_unlock(ext4_group_lock_ptr(sb, group));
3605 }
3606
3607 #ifdef CONFIG_QUOTA
ext4_quota_capable(struct super_block * sb)3608 static inline bool ext4_quota_capable(struct super_block *sb)
3609 {
3610 return (test_opt(sb, QUOTA) || ext4_has_feature_quota(sb));
3611 }
3612
ext4_is_quota_journalled(struct super_block * sb)3613 static inline bool ext4_is_quota_journalled(struct super_block *sb)
3614 {
3615 struct ext4_sb_info *sbi = EXT4_SB(sb);
3616
3617 return (ext4_has_feature_quota(sb) ||
3618 sbi->s_qf_names[USRQUOTA] || sbi->s_qf_names[GRPQUOTA]);
3619 }
3620 int ext4_enable_quotas(struct super_block *sb);
3621 #endif
3622
3623 /*
3624 * Block validity checking
3625 */
3626 #define ext4_check_indirect_blockref(inode, bh) \
3627 ext4_check_blockref(__func__, __LINE__, inode, \
3628 (__le32 *)(bh)->b_data, \
3629 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
3630
3631 #define ext4_ind_check_inode(inode) \
3632 ext4_check_blockref(__func__, __LINE__, inode, \
3633 EXT4_I(inode)->i_data, \
3634 EXT4_NDIR_BLOCKS)
3635
3636 /*
3637 * Inodes and files operations
3638 */
3639
3640 /* dir.c */
3641 extern const struct file_operations ext4_dir_operations;
3642
3643 /* file.c */
3644 extern const struct inode_operations ext4_file_inode_operations;
3645 extern const struct file_operations ext4_file_operations;
3646 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
3647
3648 /* inline.c */
3649 extern int ext4_get_max_inline_size(struct inode *inode);
3650 extern int ext4_find_inline_data_nolock(struct inode *inode);
3651 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
3652 extern void ext4_update_final_de(void *de_buf, int old_size, int new_size);
3653
3654 int ext4_readpage_inline(struct inode *inode, struct folio *folio);
3655 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3656 struct inode *inode,
3657 loff_t pos, unsigned len,
3658 struct folio **foliop);
3659 int ext4_write_inline_data_end(struct inode *inode, loff_t pos, unsigned len,
3660 unsigned copied, struct folio *folio);
3661 extern int ext4_generic_write_inline_data(struct address_space *mapping,
3662 struct inode *inode,
3663 loff_t pos, unsigned len,
3664 struct folio **foliop,
3665 void **fsdata, bool da);
3666 extern int ext4_try_add_inline_entry(handle_t *handle,
3667 struct ext4_filename *fname,
3668 struct inode *dir, struct inode *inode);
3669 extern int ext4_try_create_inline_dir(handle_t *handle,
3670 struct inode *parent,
3671 struct inode *inode);
3672 extern int ext4_read_inline_dir(struct file *filp,
3673 struct dir_context *ctx,
3674 int *has_inline_data);
3675 extern int ext4_inlinedir_to_tree(struct file *dir_file,
3676 struct inode *dir, ext4_lblk_t block,
3677 struct dx_hash_info *hinfo,
3678 __u32 start_hash, __u32 start_minor_hash,
3679 int *has_inline_data);
3680 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3681 struct ext4_filename *fname,
3682 struct ext4_dir_entry_2 **res_dir,
3683 int *has_inline_data);
3684 extern int ext4_delete_inline_entry(handle_t *handle,
3685 struct inode *dir,
3686 struct ext4_dir_entry_2 *de_del,
3687 struct buffer_head *bh,
3688 int *has_inline_data);
3689 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3690 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3691 struct ext4_dir_entry_2 **parent_de,
3692 int *retval);
3693 extern void *ext4_read_inline_link(struct inode *inode);
3694
3695 struct iomap;
3696 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap);
3697
3698 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3699
3700 extern int ext4_convert_inline_data(struct inode *inode);
3701
ext4_has_inline_data(struct inode * inode)3702 static inline int ext4_has_inline_data(struct inode *inode)
3703 {
3704 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3705 EXT4_I(inode)->i_inline_off;
3706 }
3707
3708 /* namei.c */
3709 extern const struct inode_operations ext4_dir_inode_operations;
3710 extern const struct inode_operations ext4_special_inode_operations;
3711 extern struct dentry *ext4_get_parent(struct dentry *child);
3712 extern int ext4_init_dirblock(handle_t *handle, struct inode *inode,
3713 struct buffer_head *dir_block,
3714 unsigned int parent_ino, void *inline_buf,
3715 int inline_size);
3716 extern void ext4_initialize_dirent_tail(struct buffer_head *bh,
3717 unsigned int blocksize);
3718 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode,
3719 struct buffer_head *bh);
3720 extern int __ext4_unlink(struct inode *dir, const struct qstr *d_name,
3721 struct inode *inode, struct dentry *dentry);
3722 extern int __ext4_link(struct inode *dir, struct inode *inode,
3723 struct dentry *dentry);
3724
3725 #define S_SHIFT 12
3726 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = {
3727 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
3728 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
3729 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
3730 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
3731 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
3732 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
3733 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
3734 };
3735
ext4_set_de_type(struct super_block * sb,struct ext4_dir_entry_2 * de,umode_t mode)3736 static inline void ext4_set_de_type(struct super_block *sb,
3737 struct ext4_dir_entry_2 *de,
3738 umode_t mode) {
3739 if (ext4_has_feature_filetype(sb))
3740 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3741 }
3742
3743 /* readpages.c */
3744 int ext4_read_folio(struct file *file, struct folio *folio);
3745 void ext4_readahead(struct readahead_control *rac);
3746 extern int __init ext4_init_post_read_processing(void);
3747 extern void ext4_exit_post_read_processing(void);
3748
3749 /* symlink.c */
3750 extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3751 extern const struct inode_operations ext4_symlink_inode_operations;
3752 extern const struct inode_operations ext4_fast_symlink_inode_operations;
3753
3754 /* sysfs.c */
3755 extern void ext4_notify_error_sysfs(struct ext4_sb_info *sbi);
3756 extern int ext4_register_sysfs(struct super_block *sb);
3757 extern void ext4_unregister_sysfs(struct super_block *sb);
3758 extern int __init ext4_init_sysfs(void);
3759 extern void ext4_exit_sysfs(void);
3760
3761 /* block_validity */
3762 extern void ext4_release_system_zone(struct super_block *sb);
3763 extern int ext4_setup_system_zone(struct super_block *sb);
3764 extern int __init ext4_init_system_zone(void);
3765 extern void ext4_exit_system_zone(void);
3766 extern int ext4_inode_block_valid(struct inode *inode,
3767 ext4_fsblk_t start_blk,
3768 unsigned int count);
3769 extern int ext4_check_blockref(const char *, unsigned int,
3770 struct inode *, __le32 *, unsigned int);
3771 extern int ext4_sb_block_valid(struct super_block *sb, struct inode *inode,
3772 ext4_fsblk_t start_blk, unsigned int count);
3773
3774
3775 /* extents.c */
3776 struct ext4_ext_path;
3777 struct ext4_extent;
3778
3779 /*
3780 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3781 * __le32.
3782 */
3783 #define EXT_MAX_BLOCKS 0xffffffff
3784
3785 extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode);
3786 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3787 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3788 struct ext4_map_blocks *map, int flags);
3789 extern int ext4_ext_truncate(handle_t *, struct inode *);
3790 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3791 ext4_lblk_t end);
3792 extern void ext4_ext_init(struct super_block *);
3793 extern void ext4_ext_release(struct super_block *);
3794 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3795 loff_t len);
3796 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3797 loff_t offset, ssize_t len);
3798 extern int ext4_convert_unwritten_extents_atomic(handle_t *handle,
3799 struct inode *inode, loff_t offset, ssize_t len);
3800 extern int ext4_convert_unwritten_io_end_vec(handle_t *handle,
3801 ext4_io_end_t *io_end);
3802 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3803 struct ext4_map_blocks *map, int flags);
3804 extern int ext4_map_query_blocks(handle_t *handle, struct inode *inode,
3805 struct ext4_map_blocks *map, int flags);
3806 extern int ext4_map_create_blocks(handle_t *handle, struct inode *inode,
3807 struct ext4_map_blocks *map, int flags);
3808 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3809 int num,
3810 struct ext4_ext_path *path);
3811 extern struct ext4_ext_path *ext4_ext_insert_extent(
3812 handle_t *handle, struct inode *inode,
3813 struct ext4_ext_path *path,
3814 struct ext4_extent *newext, int gb_flags);
3815 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3816 struct ext4_ext_path *,
3817 int flags);
3818 extern void ext4_free_ext_path(struct ext4_ext_path *);
3819 extern int ext4_ext_check_inode(struct inode *inode);
3820 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3821 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3822 __u64 start, __u64 len);
3823 extern int ext4_get_es_cache(struct inode *inode,
3824 struct fiemap_extent_info *fieinfo,
3825 __u64 start, __u64 len);
3826 extern int ext4_ext_precache(struct inode *inode);
3827 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3828 struct inode *inode2, ext4_lblk_t lblk1,
3829 ext4_lblk_t lblk2, ext4_lblk_t count,
3830 int mark_unwritten,int *err);
3831 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu);
3832 extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode,
3833 int check_cred, int restart_cred,
3834 int revoke_cred);
3835 extern void ext4_ext_replay_shrink_inode(struct inode *inode, ext4_lblk_t end);
3836 extern int ext4_ext_replay_set_iblocks(struct inode *inode);
3837 extern int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start,
3838 int len, int unwritten, ext4_fsblk_t pblk);
3839 extern int ext4_ext_clear_bb(struct inode *inode);
3840
3841
3842 /* move_extent.c */
3843 extern void ext4_double_down_write_data_sem(struct inode *first,
3844 struct inode *second);
3845 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3846 struct inode *donor_inode);
3847 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3848 __u64 start_orig, __u64 start_donor,
3849 __u64 len, __u64 *moved_len);
3850
3851 /* page-io.c */
3852 extern int __init ext4_init_pageio(void);
3853 extern void ext4_exit_pageio(void);
3854 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3855 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3856 extern int ext4_put_io_end(ext4_io_end_t *io_end);
3857 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3858 extern void ext4_io_submit_init(struct ext4_io_submit *io,
3859 struct writeback_control *wbc);
3860 extern void ext4_end_io_rsv_work(struct work_struct *work);
3861 extern void ext4_io_submit(struct ext4_io_submit *io);
3862 int ext4_bio_write_folio(struct ext4_io_submit *io, struct folio *page,
3863 size_t len);
3864 extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end);
3865 extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end);
3866
3867 /* mmp.c */
3868 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3869
3870 /* mmp.c */
3871 extern void ext4_stop_mmpd(struct ext4_sb_info *sbi);
3872
3873 /* verity.c */
3874 extern const struct fsverity_operations ext4_verityops;
3875
3876 /* orphan.c */
3877 extern int ext4_orphan_add(handle_t *, struct inode *);
3878 extern int ext4_orphan_del(handle_t *, struct inode *);
3879 extern void ext4_orphan_cleanup(struct super_block *sb,
3880 struct ext4_super_block *es);
3881 extern void ext4_release_orphan_info(struct super_block *sb);
3882 extern int ext4_init_orphan_info(struct super_block *sb);
3883 extern int ext4_orphan_file_empty(struct super_block *sb);
3884 extern void ext4_orphan_file_block_trigger(
3885 struct jbd2_buffer_trigger_type *triggers,
3886 struct buffer_head *bh,
3887 void *data, size_t size);
3888
3889 /*
3890 * Add new method to test whether block and inode bitmaps are properly
3891 * initialized. With uninit_bg reading the block from disk is not enough
3892 * to mark the bitmap uptodate. We need to also zero-out the bitmap
3893 */
3894 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3895
bitmap_uptodate(struct buffer_head * bh)3896 static inline int bitmap_uptodate(struct buffer_head *bh)
3897 {
3898 return (buffer_uptodate(bh) &&
3899 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3900 }
set_bitmap_uptodate(struct buffer_head * bh)3901 static inline void set_bitmap_uptodate(struct buffer_head *bh)
3902 {
3903 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3904 }
3905
3906 extern int ext4_resize_begin(struct super_block *sb);
3907 extern int ext4_resize_end(struct super_block *sb, bool update_backups);
3908
ext4_set_io_unwritten_flag(struct ext4_io_end * io_end)3909 static inline void ext4_set_io_unwritten_flag(struct ext4_io_end *io_end)
3910 {
3911 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN))
3912 io_end->flag |= EXT4_IO_END_UNWRITTEN;
3913 }
3914
ext4_clear_io_unwritten_flag(ext4_io_end_t * io_end)3915 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3916 {
3917 if (io_end->flag & EXT4_IO_END_UNWRITTEN)
3918 io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3919 }
3920
3921 extern const struct iomap_ops ext4_iomap_ops;
3922 extern const struct iomap_ops ext4_iomap_report_ops;
3923
ext4_buffer_uptodate(struct buffer_head * bh)3924 static inline int ext4_buffer_uptodate(struct buffer_head *bh)
3925 {
3926 /*
3927 * If the buffer has the write error flag, we have failed
3928 * to write out data in the block. In this case, we don't
3929 * have to read the block because we may read the old data
3930 * successfully.
3931 */
3932 if (buffer_write_io_error(bh))
3933 set_buffer_uptodate(bh);
3934 return buffer_uptodate(bh);
3935 }
3936
ext4_inode_can_atomic_write(struct inode * inode)3937 static inline bool ext4_inode_can_atomic_write(struct inode *inode)
3938 {
3939
3940 return S_ISREG(inode->i_mode) &&
3941 ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS) &&
3942 EXT4_SB(inode->i_sb)->s_awu_min > 0;
3943 }
3944
3945 extern int ext4_block_write_begin(handle_t *handle, struct folio *folio,
3946 loff_t pos, unsigned len,
3947 get_block_t *get_block);
3948
3949 #if IS_ENABLED(CONFIG_EXT4_KUNIT_TESTS)
3950 #define EXPORT_SYMBOL_FOR_EXT4_TEST(sym) \
3951 EXPORT_SYMBOL_FOR_MODULES(sym, "ext4-test")
3952 #endif
3953 #endif /* __KERNEL__ */
3954
3955 #endif /* _EXT4_H */
3956