1 // SPDX-License-Identifier: GPL-2.0
2 /*
3 * ext4.h
4 *
5 * Copyright (C) 1992, 1993, 1994, 1995
6 * Remy Card (card@masi.ibp.fr)
7 * Laboratoire MASI - Institut Blaise Pascal
8 * Universite Pierre et Marie Curie (Paris VI)
9 *
10 * from
11 *
12 * linux/include/linux/minix_fs.h
13 *
14 * Copyright (C) 1991, 1992 Linus Torvalds
15 */
16
17 #ifndef _EXT4_H
18 #define _EXT4_H
19
20 #include <linux/refcount.h>
21 #include <linux/types.h>
22 #include <linux/blkdev.h>
23 #include <linux/magic.h>
24 #include <linux/jbd2.h>
25 #include <linux/quota.h>
26 #include <linux/rwsem.h>
27 #include <linux/rbtree.h>
28 #include <linux/seqlock.h>
29 #include <linux/mutex.h>
30 #include <linux/timer.h>
31 #include <linux/wait.h>
32 #include <linux/sched/signal.h>
33 #include <linux/blockgroup_lock.h>
34 #include <linux/percpu_counter.h>
35 #include <linux/ratelimit.h>
36 #include <crypto/hash.h>
37 #include <linux/falloc.h>
38 #include <linux/percpu-rwsem.h>
39 #include <linux/fiemap.h>
40 #ifdef __KERNEL__
41 #include <linux/compat.h>
42 #endif
43 #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 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 /* blocks already reserved */
189 #define EXT4_MB_HINT_RESERVED 0x0002
190 /* metadata is being allocated */
191 #define EXT4_MB_HINT_METADATA 0x0004
192 /* first blocks in the file */
193 #define EXT4_MB_HINT_FIRST 0x0008
194 /* search for the best chunk */
195 #define EXT4_MB_HINT_BEST 0x0010
196 /* data is being allocated */
197 #define EXT4_MB_HINT_DATA 0x0020
198 /* don't preallocate (for tails) */
199 #define EXT4_MB_HINT_NOPREALLOC 0x0040
200 /* allocate for locality group */
201 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080
202 /* allocate goal blocks or none */
203 #define EXT4_MB_HINT_GOAL_ONLY 0x0100
204 /* goal is meaningful */
205 #define EXT4_MB_HINT_TRY_GOAL 0x0200
206 /* blocks already pre-reserved by delayed allocation */
207 #define EXT4_MB_DELALLOC_RESERVED 0x0400
208 /* We are doing stream allocation */
209 #define EXT4_MB_STREAM_ALLOC 0x0800
210 /* Use reserved root blocks if needed */
211 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000
212 /* Use blocks from reserved pool */
213 #define EXT4_MB_USE_RESERVED 0x2000
214 /* Do strict check for free blocks while retrying block allocation */
215 #define EXT4_MB_STRICT_CHECK 0x4000
216 /* Large fragment size list lookup succeeded at least once for cr = 0 */
217 #define EXT4_MB_CR_POWER2_ALIGNED_OPTIMIZED 0x8000
218 /* Avg fragment size rb tree lookup succeeded at least once for cr = 1 */
219 #define EXT4_MB_CR_GOAL_LEN_FAST_OPTIMIZED 0x00010000
220 /* Avg fragment size rb tree lookup succeeded at least once for cr = 1.5 */
221 #define EXT4_MB_CR_BEST_AVAIL_LEN_OPTIMIZED 0x00020000
222
223 struct ext4_allocation_request {
224 /* target inode for block we're allocating */
225 struct inode *inode;
226 /* how many blocks we want to allocate */
227 unsigned int len;
228 /* logical block in target inode */
229 ext4_lblk_t logical;
230 /* the closest logical allocated block to the left */
231 ext4_lblk_t lleft;
232 /* the closest logical allocated block to the right */
233 ext4_lblk_t lright;
234 /* phys. target (a hint) */
235 ext4_fsblk_t goal;
236 /* phys. block for the closest logical allocated block to the left */
237 ext4_fsblk_t pleft;
238 /* phys. block for the closest logical allocated block to the right */
239 ext4_fsblk_t pright;
240 /* flags. see above EXT4_MB_HINT_* */
241 unsigned int flags;
242 };
243
244 /*
245 * Logical to physical block mapping, used by ext4_map_blocks()
246 *
247 * This structure is used to pass requests into ext4_map_blocks() as
248 * well as to store the information returned by ext4_map_blocks(). It
249 * takes less room on the stack than a struct buffer_head.
250 */
251 #define EXT4_MAP_NEW BIT(BH_New)
252 #define EXT4_MAP_MAPPED BIT(BH_Mapped)
253 #define EXT4_MAP_UNWRITTEN BIT(BH_Unwritten)
254 #define EXT4_MAP_BOUNDARY BIT(BH_Boundary)
255 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
256 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
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 };
264
265 /*
266 * Block validity checking, system zone rbtree.
267 */
268 struct ext4_system_blocks {
269 struct rb_root root;
270 struct rcu_head rcu;
271 };
272
273 /*
274 * Flags for ext4_io_end->flags
275 */
276 #define EXT4_IO_END_UNWRITTEN 0x0001
277 #define EXT4_IO_END_FAILED 0x0002
278
279 struct ext4_io_end_vec {
280 struct list_head list; /* list of io_end_vec */
281 loff_t offset; /* offset in the file */
282 ssize_t size; /* size of the extent */
283 };
284
285 /*
286 * For converting unwritten extents on a work queue. 'handle' is used for
287 * buffered writeback.
288 */
289 typedef struct ext4_io_end {
290 struct list_head list; /* per-file finished IO list */
291 handle_t *handle; /* handle reserved for extent
292 * conversion */
293 struct inode *inode; /* file being written to */
294 struct bio *bio; /* Linked list of completed
295 * bios covering the extent */
296 unsigned int flag; /* unwritten or not */
297 refcount_t count; /* reference counter */
298 struct list_head list_vec; /* list of ext4_io_end_vec */
299 } ext4_io_end_t;
300
301 struct ext4_io_submit {
302 struct writeback_control *io_wbc;
303 struct bio *io_bio;
304 ext4_io_end_t *io_end;
305 sector_t io_next_block;
306 };
307
308 /*
309 * Special inodes numbers
310 */
311 #define EXT4_BAD_INO 1 /* Bad blocks inode */
312 #define EXT4_ROOT_INO 2 /* Root inode */
313 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */
314 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
315 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
316 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
317 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
318 #define EXT4_JOURNAL_INO 8 /* Journal inode */
319
320 /* First non-reserved inode for old ext4 filesystems */
321 #define EXT4_GOOD_OLD_FIRST_INO 11
322
323 /*
324 * Maximal count of links to a file
325 */
326 #define EXT4_LINK_MAX 65000
327
328 /*
329 * Macro-instructions used to manage several block sizes
330 */
331 #define EXT4_MIN_BLOCK_SIZE 1024
332 #define EXT4_MAX_BLOCK_SIZE 65536
333 #define EXT4_MIN_BLOCK_LOG_SIZE 10
334 #define EXT4_MAX_BLOCK_LOG_SIZE 16
335 #define EXT4_MAX_CLUSTER_LOG_SIZE 30
336 #ifdef __KERNEL__
337 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
338 #else
339 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
340 #endif
341 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
342 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
343 EXT4_SB(s)->s_cluster_bits)
344 #ifdef __KERNEL__
345 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
346 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
347 #else
348 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
349 #endif
350 #ifdef __KERNEL__
351 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
352 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
353 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
354 #else
355 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
356 EXT4_GOOD_OLD_INODE_SIZE : \
357 (s)->s_inode_size)
358 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
359 EXT4_GOOD_OLD_FIRST_INO : \
360 (s)->s_first_ino)
361 #endif
362 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
363 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \
364 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \
365 blkbits))
366
367 /* Translate a block number to a cluster number */
368 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
369 /* Translate a cluster number to a block number */
370 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
371 /* Translate # of blks to # of clusters */
372 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
373 (sbi)->s_cluster_bits)
374 /* Mask out the low bits to get the starting block of the cluster */
375 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
376 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
377 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
378 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
379 /* Fill in the low bits to get the last block of the cluster */
380 #define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) | \
381 ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1))
382 /* Get the cluster offset */
383 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
384 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
385 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
386 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
387
388 /*
389 * Structure of a blocks group descriptor
390 */
391 struct ext4_group_desc
392 {
393 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
394 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
395 __le32 bg_inode_table_lo; /* Inodes table block */
396 __le16 bg_free_blocks_count_lo;/* Free blocks count */
397 __le16 bg_free_inodes_count_lo;/* Free inodes count */
398 __le16 bg_used_dirs_count_lo; /* Directories count */
399 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
400 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
401 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
402 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
403 __le16 bg_itable_unused_lo; /* Unused inodes count */
404 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
405 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
406 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
407 __le32 bg_inode_table_hi; /* Inodes table block MSB */
408 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
409 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
410 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
411 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
412 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
413 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
414 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
415 __u32 bg_reserved;
416 };
417
418 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
419 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
420 sizeof(__le16))
421 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
422 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
423 sizeof(__le16))
424
425 /*
426 * Structure of a flex block group info
427 */
428
429 struct flex_groups {
430 atomic64_t free_clusters;
431 atomic_t free_inodes;
432 atomic_t used_dirs;
433 };
434
435 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
436 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
437 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
438
439 /*
440 * Macro-instructions used to manage group descriptors
441 */
442 #define EXT4_MIN_DESC_SIZE 32
443 #define EXT4_MIN_DESC_SIZE_64BIT 64
444 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
445 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
446 #ifdef __KERNEL__
447 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
448 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
449 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
450 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
451 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
452 #else
453 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
454 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
455 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
456 #endif
457
458 /*
459 * Constants relative to the data blocks
460 */
461 #define EXT4_NDIR_BLOCKS 12
462 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
463 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
464 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
465 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
466
467 /*
468 * Inode flags
469 */
470 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
471 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
472 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
473 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
474 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
475 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
476 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
477 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
478 /* Reserved for compression usage... */
479 #define EXT4_DIRTY_FL 0x00000100
480 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
481 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
482 /* nb: was previously EXT2_ECOMPR_FL */
483 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */
484 /* End compression flags --- maybe not all used */
485 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
486 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
487 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
488 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
489 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
490 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
491 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
492 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
493 #define EXT4_VERITY_FL 0x00100000 /* Verity protected inode */
494 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
495 /* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */
496
497 #define EXT4_DAX_FL 0x02000000 /* Inode is DAX */
498
499 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
500 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
501 #define EXT4_CASEFOLD_FL 0x40000000 /* Casefolded directory */
502 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
503
504 /* User modifiable flags */
505 #define EXT4_FL_USER_MODIFIABLE (EXT4_SECRM_FL | \
506 EXT4_UNRM_FL | \
507 EXT4_COMPR_FL | \
508 EXT4_SYNC_FL | \
509 EXT4_IMMUTABLE_FL | \
510 EXT4_APPEND_FL | \
511 EXT4_NODUMP_FL | \
512 EXT4_NOATIME_FL | \
513 EXT4_JOURNAL_DATA_FL | \
514 EXT4_NOTAIL_FL | \
515 EXT4_DIRSYNC_FL | \
516 EXT4_TOPDIR_FL | \
517 EXT4_EXTENTS_FL | \
518 0x00400000 /* EXT4_EOFBLOCKS_FL */ | \
519 EXT4_DAX_FL | \
520 EXT4_PROJINHERIT_FL | \
521 EXT4_CASEFOLD_FL)
522
523 /* User visible flags */
524 #define EXT4_FL_USER_VISIBLE (EXT4_FL_USER_MODIFIABLE | \
525 EXT4_DIRTY_FL | \
526 EXT4_COMPRBLK_FL | \
527 EXT4_NOCOMPR_FL | \
528 EXT4_ENCRYPT_FL | \
529 EXT4_INDEX_FL | \
530 EXT4_VERITY_FL | \
531 EXT4_INLINE_DATA_FL)
532
533 /* Flags that should be inherited by new inodes from their parent. */
534 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
535 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
536 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
537 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
538 EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL |\
539 EXT4_DAX_FL)
540
541 /* Flags that are appropriate for regular files (all but dir-specific ones). */
542 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\
543 EXT4_PROJINHERIT_FL))
544
545 /* Flags that are appropriate for non-directories/regular files. */
546 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
547
548 /* The only flags that should be swapped */
549 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL)
550
551 /* Flags which are mutually exclusive to DAX */
552 #define EXT4_DAX_MUT_EXCL (EXT4_VERITY_FL | EXT4_ENCRYPT_FL |\
553 EXT4_JOURNAL_DATA_FL | EXT4_INLINE_DATA_FL)
554
555 /* Mask out flags that are inappropriate for the given type of inode. */
ext4_mask_flags(umode_t mode,__u32 flags)556 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
557 {
558 if (S_ISDIR(mode))
559 return flags;
560 else if (S_ISREG(mode))
561 return flags & EXT4_REG_FLMASK;
562 else
563 return flags & EXT4_OTHER_FLMASK;
564 }
565
566 /*
567 * Inode flags used for atomic set/get
568 */
569 enum {
570 EXT4_INODE_SECRM = 0, /* Secure deletion */
571 EXT4_INODE_UNRM = 1, /* Undelete */
572 EXT4_INODE_COMPR = 2, /* Compress file */
573 EXT4_INODE_SYNC = 3, /* Synchronous updates */
574 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
575 EXT4_INODE_APPEND = 5, /* writes to file may only append */
576 EXT4_INODE_NODUMP = 6, /* do not dump file */
577 EXT4_INODE_NOATIME = 7, /* do not update atime */
578 /* Reserved for compression usage... */
579 EXT4_INODE_DIRTY = 8,
580 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
581 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
582 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */
583 /* End compression flags --- maybe not all used */
584 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
585 EXT4_INODE_IMAGIC = 13, /* AFS directory */
586 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
587 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
588 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
589 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
590 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
591 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
592 EXT4_INODE_VERITY = 20, /* Verity protected inode */
593 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
594 /* 22 was formerly EXT4_INODE_EOFBLOCKS */
595 EXT4_INODE_DAX = 25, /* Inode is DAX */
596 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
597 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */
598 EXT4_INODE_CASEFOLD = 30, /* Casefolded directory */
599 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
600 };
601
602 /*
603 * Since it's pretty easy to mix up bit numbers and hex values, we use a
604 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
605 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
606 * any extra space in the compiled kernel image, otherwise, the build will fail.
607 * It's important that these values are the same, since we are using
608 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
609 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
610 * values found in ext2, ext3 and ext4 filesystems, and of course the values
611 * defined in e2fsprogs.
612 *
613 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
614 */
615 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1U << EXT4_INODE_##FLAG))
616 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
617
ext4_check_flag_values(void)618 static inline void ext4_check_flag_values(void)
619 {
620 CHECK_FLAG_VALUE(SECRM);
621 CHECK_FLAG_VALUE(UNRM);
622 CHECK_FLAG_VALUE(COMPR);
623 CHECK_FLAG_VALUE(SYNC);
624 CHECK_FLAG_VALUE(IMMUTABLE);
625 CHECK_FLAG_VALUE(APPEND);
626 CHECK_FLAG_VALUE(NODUMP);
627 CHECK_FLAG_VALUE(NOATIME);
628 CHECK_FLAG_VALUE(DIRTY);
629 CHECK_FLAG_VALUE(COMPRBLK);
630 CHECK_FLAG_VALUE(NOCOMPR);
631 CHECK_FLAG_VALUE(ENCRYPT);
632 CHECK_FLAG_VALUE(INDEX);
633 CHECK_FLAG_VALUE(IMAGIC);
634 CHECK_FLAG_VALUE(JOURNAL_DATA);
635 CHECK_FLAG_VALUE(NOTAIL);
636 CHECK_FLAG_VALUE(DIRSYNC);
637 CHECK_FLAG_VALUE(TOPDIR);
638 CHECK_FLAG_VALUE(HUGE_FILE);
639 CHECK_FLAG_VALUE(EXTENTS);
640 CHECK_FLAG_VALUE(VERITY);
641 CHECK_FLAG_VALUE(EA_INODE);
642 CHECK_FLAG_VALUE(INLINE_DATA);
643 CHECK_FLAG_VALUE(PROJINHERIT);
644 CHECK_FLAG_VALUE(CASEFOLD);
645 CHECK_FLAG_VALUE(RESERVED);
646 }
647
648 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
649 struct compat_ext4_new_group_input {
650 u32 group;
651 compat_u64 block_bitmap;
652 compat_u64 inode_bitmap;
653 compat_u64 inode_table;
654 u32 blocks_count;
655 u16 reserved_blocks;
656 u16 unused;
657 };
658 #endif
659
660 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
661 struct ext4_new_group_data {
662 __u32 group;
663 __u64 block_bitmap;
664 __u64 inode_bitmap;
665 __u64 inode_table;
666 __u32 blocks_count;
667 __u16 reserved_blocks;
668 __u16 mdata_blocks;
669 __u32 free_clusters_count;
670 };
671
672 /* Indexes used to index group tables in ext4_new_group_data */
673 enum {
674 BLOCK_BITMAP = 0, /* block bitmap */
675 INODE_BITMAP, /* inode bitmap */
676 INODE_TABLE, /* inode tables */
677 GROUP_TABLE_COUNT,
678 };
679
680 /*
681 * Flags used by ext4_map_blocks()
682 */
683 /* Allocate any needed blocks and/or convert an unwritten
684 extent to be an initialized ext4 */
685 #define EXT4_GET_BLOCKS_CREATE 0x0001
686 /* Request the creation of an unwritten extent */
687 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002
688 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\
689 EXT4_GET_BLOCKS_CREATE)
690 /* Caller is from the delayed allocation writeout path
691 * finally doing the actual allocation of delayed blocks */
692 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
693 /* caller is from the direct IO path, request to creation of an
694 unwritten extents if not allocated, split the unwritten
695 extent if blocks has been preallocated already*/
696 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
697 #define EXT4_GET_BLOCKS_CONVERT 0x0010
698 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
699 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
700 /* Convert extent to initialized after IO complete */
701 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
702 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
703 /* Eventual metadata allocation (due to growing extent tree)
704 * should not fail, so try to use reserved blocks for that.*/
705 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
706 /* Don't normalize allocation size (used for fallocate) */
707 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
708 /* Convert written extents to unwritten */
709 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100
710 /* Write zeros to newly created written extents */
711 #define EXT4_GET_BLOCKS_ZERO 0x0200
712 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\
713 EXT4_GET_BLOCKS_ZERO)
714 /* Caller will submit data before dropping transaction handle. This
715 * allows jbd2 to avoid submitting data before commit. */
716 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400
717 /* Caller is in the atomic contex, find extent if it has been cached */
718 #define EXT4_GET_BLOCKS_CACHED_NOWAIT 0x0800
719
720 /*
721 * The bit position of these flags must not overlap with any of the
722 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(),
723 * read_extent_tree_block(), ext4_split_extent_at(),
724 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
725 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
726 * caching the extents when reading from the extent tree while a
727 * truncate or punch hole operation is in progress.
728 */
729 #define EXT4_EX_NOCACHE 0x40000000
730 #define EXT4_EX_FORCE_CACHE 0x20000000
731 #define EXT4_EX_NOFAIL 0x10000000
732
733 /*
734 * Flags used by ext4_free_blocks
735 */
736 #define EXT4_FREE_BLOCKS_METADATA 0x0001
737 #define EXT4_FREE_BLOCKS_FORGET 0x0002
738 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
739 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
740 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
741 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
742 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040
743
744 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
745 /*
746 * ioctl commands in 32 bit emulation
747 */
748 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
749 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
750 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
751 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
752 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
753 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
754 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
755 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
756 #endif
757
758 /* Max physical block we can address w/o extents */
759 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
760
761 /* Max logical block we can support */
762 #define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFE
763
764 /*
765 * Structure of an inode on the disk
766 */
767 struct ext4_inode {
768 __le16 i_mode; /* File mode */
769 __le16 i_uid; /* Low 16 bits of Owner Uid */
770 __le32 i_size_lo; /* Size in bytes */
771 __le32 i_atime; /* Access time */
772 __le32 i_ctime; /* Inode Change time */
773 __le32 i_mtime; /* Modification time */
774 __le32 i_dtime; /* Deletion Time */
775 __le16 i_gid; /* Low 16 bits of Group Id */
776 __le16 i_links_count; /* Links count */
777 __le32 i_blocks_lo; /* Blocks count */
778 __le32 i_flags; /* File flags */
779 union {
780 struct {
781 __le32 l_i_version;
782 } linux1;
783 struct {
784 __u32 h_i_translator;
785 } hurd1;
786 struct {
787 __u32 m_i_reserved1;
788 } masix1;
789 } osd1; /* OS dependent 1 */
790 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
791 __le32 i_generation; /* File version (for NFS) */
792 __le32 i_file_acl_lo; /* File ACL */
793 __le32 i_size_high;
794 __le32 i_obso_faddr; /* Obsoleted fragment address */
795 union {
796 struct {
797 __le16 l_i_blocks_high; /* were l_i_reserved1 */
798 __le16 l_i_file_acl_high;
799 __le16 l_i_uid_high; /* these 2 fields */
800 __le16 l_i_gid_high; /* were reserved2[0] */
801 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
802 __le16 l_i_reserved;
803 } linux2;
804 struct {
805 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
806 __u16 h_i_mode_high;
807 __u16 h_i_uid_high;
808 __u16 h_i_gid_high;
809 __u32 h_i_author;
810 } hurd2;
811 struct {
812 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
813 __le16 m_i_file_acl_high;
814 __u32 m_i_reserved2[2];
815 } masix2;
816 } osd2; /* OS dependent 2 */
817 __le16 i_extra_isize;
818 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
819 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
820 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
821 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
822 __le32 i_crtime; /* File Creation time */
823 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
824 __le32 i_version_hi; /* high 32 bits for 64-bit version */
825 __le32 i_projid; /* Project ID */
826 };
827
828 #define EXT4_EPOCH_BITS 2
829 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
830 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
831
832 /*
833 * Extended fields will fit into an inode if the filesystem was formatted
834 * with large inodes (-I 256 or larger) and there are not currently any EAs
835 * consuming all of the available space. For new inodes we always reserve
836 * enough space for the kernel's known extended fields, but for inodes
837 * created with an old kernel this might not have been the case. None of
838 * the extended inode fields is critical for correct filesystem operation.
839 * This macro checks if a certain field fits in the inode. Note that
840 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
841 */
842 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
843 ((offsetof(typeof(*ext4_inode), field) + \
844 sizeof((ext4_inode)->field)) \
845 <= (EXT4_GOOD_OLD_INODE_SIZE + \
846 (einode)->i_extra_isize)) \
847
848 /*
849 * We use an encoding that preserves the times for extra epoch "00":
850 *
851 * extra msb of adjust for signed
852 * epoch 32-bit 32-bit tv_sec to
853 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
854 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
855 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
856 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
857 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
858 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
859 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
860 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
861 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
862 *
863 * Note that previous versions of the kernel on 64-bit systems would
864 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
865 * 1970. e2fsck will correct this, assuming that it is run on the
866 * affected filesystem before 2242.
867 */
868
ext4_encode_extra_time(struct timespec64 ts)869 static inline __le32 ext4_encode_extra_time(struct timespec64 ts)
870 {
871 u32 extra = ((ts.tv_sec - (s32)ts.tv_sec) >> 32) & EXT4_EPOCH_MASK;
872 return cpu_to_le32(extra | (ts.tv_nsec << EXT4_EPOCH_BITS));
873 }
874
ext4_decode_extra_time(__le32 base,__le32 extra)875 static inline struct timespec64 ext4_decode_extra_time(__le32 base,
876 __le32 extra)
877 {
878 struct timespec64 ts = { .tv_sec = (signed)le32_to_cpu(base) };
879
880 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK)))
881 ts.tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
882 ts.tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
883 return ts;
884 }
885
886 #define EXT4_INODE_SET_XTIME_VAL(xtime, inode, raw_inode, ts) \
887 do { \
888 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \
889 (raw_inode)->xtime = cpu_to_le32((ts).tv_sec); \
890 (raw_inode)->xtime ## _extra = ext4_encode_extra_time(ts); \
891 } else \
892 (raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (ts).tv_sec, S32_MIN, S32_MAX)); \
893 } while (0)
894
895 #define EXT4_INODE_SET_ATIME(inode, raw_inode) \
896 EXT4_INODE_SET_XTIME_VAL(i_atime, inode, raw_inode, inode_get_atime(inode))
897
898 #define EXT4_INODE_SET_MTIME(inode, raw_inode) \
899 EXT4_INODE_SET_XTIME_VAL(i_mtime, inode, raw_inode, inode_get_mtime(inode))
900
901 #define EXT4_INODE_SET_CTIME(inode, raw_inode) \
902 EXT4_INODE_SET_XTIME_VAL(i_ctime, inode, raw_inode, inode_get_ctime(inode))
903
904 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
905 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
906 EXT4_INODE_SET_XTIME_VAL(xtime, &((einode)->vfs_inode), \
907 raw_inode, (einode)->xtime)
908
909 #define EXT4_INODE_GET_XTIME_VAL(xtime, inode, raw_inode) \
910 (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra) ? \
911 ext4_decode_extra_time((raw_inode)->xtime, \
912 (raw_inode)->xtime ## _extra) : \
913 (struct timespec64) { \
914 .tv_sec = (signed)le32_to_cpu((raw_inode)->xtime) \
915 })
916
917 #define EXT4_INODE_GET_ATIME(inode, raw_inode) \
918 do { \
919 inode_set_atime_to_ts(inode, \
920 EXT4_INODE_GET_XTIME_VAL(i_atime, inode, raw_inode)); \
921 } while (0)
922
923 #define EXT4_INODE_GET_MTIME(inode, raw_inode) \
924 do { \
925 inode_set_mtime_to_ts(inode, \
926 EXT4_INODE_GET_XTIME_VAL(i_mtime, inode, raw_inode)); \
927 } while (0)
928
929 #define EXT4_INODE_GET_CTIME(inode, raw_inode) \
930 do { \
931 inode_set_ctime_to_ts(inode, \
932 EXT4_INODE_GET_XTIME_VAL(i_ctime, inode, raw_inode)); \
933 } while (0)
934
935 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
936 do { \
937 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
938 (einode)->xtime = \
939 EXT4_INODE_GET_XTIME_VAL(xtime, &(einode->vfs_inode), \
940 raw_inode); \
941 else \
942 (einode)->xtime = (struct timespec64){0, 0}; \
943 } while (0)
944
945 #define i_disk_version osd1.linux1.l_i_version
946
947 #if defined(__KERNEL__) || defined(__linux__)
948 #define i_reserved1 osd1.linux1.l_i_reserved1
949 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
950 #define i_blocks_high osd2.linux2.l_i_blocks_high
951 #define i_uid_low i_uid
952 #define i_gid_low i_gid
953 #define i_uid_high osd2.linux2.l_i_uid_high
954 #define i_gid_high osd2.linux2.l_i_gid_high
955 #define i_checksum_lo osd2.linux2.l_i_checksum_lo
956
957 #elif defined(__GNU__)
958
959 #define i_translator osd1.hurd1.h_i_translator
960 #define i_uid_high osd2.hurd2.h_i_uid_high
961 #define i_gid_high osd2.hurd2.h_i_gid_high
962 #define i_author osd2.hurd2.h_i_author
963
964 #elif defined(__masix__)
965
966 #define i_reserved1 osd1.masix1.m_i_reserved1
967 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
968 #define i_reserved2 osd2.masix2.m_i_reserved2
969
970 #endif /* defined(__KERNEL__) || defined(__linux__) */
971
972 #include "extents_status.h"
973 #include "fast_commit.h"
974
975 /*
976 * Lock subclasses for i_data_sem in the ext4_inode_info structure.
977 *
978 * These are needed to avoid lockdep false positives when we need to
979 * allocate blocks to the quota inode during ext4_map_blocks(), while
980 * holding i_data_sem for a normal (non-quota) inode. Since we don't
981 * do quota tracking for the quota inode, this avoids deadlock (as
982 * well as infinite recursion, since it isn't turtles all the way
983 * down...)
984 *
985 * I_DATA_SEM_NORMAL - Used for most inodes
986 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode
987 * where the second inode has larger inode number
988 * than the first
989 * I_DATA_SEM_QUOTA - Used for quota inodes only
990 * I_DATA_SEM_EA - Used for ea_inodes only
991 */
992 enum {
993 I_DATA_SEM_NORMAL = 0,
994 I_DATA_SEM_OTHER,
995 I_DATA_SEM_QUOTA,
996 I_DATA_SEM_EA
997 };
998
999
1000 /*
1001 * fourth extended file system inode data in memory
1002 */
1003 struct ext4_inode_info {
1004 __le32 i_data[15]; /* unconverted */
1005 __u32 i_dtime;
1006 ext4_fsblk_t i_file_acl;
1007
1008 /*
1009 * i_block_group is the number of the block group which contains
1010 * this file's inode. Constant across the lifetime of the inode,
1011 * it is used for making block allocation decisions - we try to
1012 * place a file's data blocks near its inode block, and new inodes
1013 * near to their parent directory's inode.
1014 */
1015 ext4_group_t i_block_group;
1016 ext4_lblk_t i_dir_start_lookup;
1017 #if (BITS_PER_LONG < 64)
1018 unsigned long i_state_flags; /* Dynamic state flags */
1019 #endif
1020 unsigned long i_flags;
1021
1022 /*
1023 * Extended attributes can be read independently of the main file
1024 * data. Taking i_rwsem even when reading would cause contention
1025 * between readers of EAs and writers of regular file data, so
1026 * instead we synchronize on xattr_sem when reading or changing
1027 * EAs.
1028 */
1029 struct rw_semaphore xattr_sem;
1030
1031 /*
1032 * Inodes with EXT4_STATE_ORPHAN_FILE use i_orphan_idx. Otherwise
1033 * i_orphan is used.
1034 */
1035 union {
1036 struct list_head i_orphan; /* unlinked but open inodes */
1037 unsigned int i_orphan_idx; /* Index in orphan file */
1038 };
1039
1040 /* Fast commit related info */
1041
1042 /* For tracking dentry create updates */
1043 struct list_head i_fc_dilist;
1044 struct list_head i_fc_list; /*
1045 * inodes that need fast commit
1046 * protected by sbi->s_fc_lock.
1047 */
1048
1049 /* Start of lblk range that needs to be committed in this fast commit */
1050 ext4_lblk_t i_fc_lblk_start;
1051
1052 /* End of lblk range that needs to be committed in this fast commit */
1053 ext4_lblk_t i_fc_lblk_len;
1054
1055 /* Number of ongoing updates on this inode */
1056 atomic_t i_fc_updates;
1057
1058 /* Fast commit wait queue for this inode */
1059 wait_queue_head_t i_fc_wait;
1060
1061 /* Protect concurrent accesses on i_fc_lblk_start, i_fc_lblk_len */
1062 struct mutex i_fc_lock;
1063
1064 /*
1065 * i_disksize keeps track of what the inode size is ON DISK, not
1066 * in memory. During truncate, i_size is set to the new size by
1067 * the VFS prior to calling ext4_truncate(), but the filesystem won't
1068 * set i_disksize to 0 until the truncate is actually under way.
1069 *
1070 * The intent is that i_disksize always represents the blocks which
1071 * are used by this file. This allows recovery to restart truncate
1072 * on orphans if we crash during truncate. We actually write i_disksize
1073 * into the on-disk inode when writing inodes out, instead of i_size.
1074 *
1075 * The only time when i_disksize and i_size may be different is when
1076 * a truncate is in progress. The only things which change i_disksize
1077 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
1078 */
1079 loff_t i_disksize;
1080
1081 /*
1082 * i_data_sem is for serialising ext4_truncate() against
1083 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
1084 * data tree are chopped off during truncate. We can't do that in
1085 * ext4 because whenever we perform intermediate commits during
1086 * truncate, the inode and all the metadata blocks *must* be in a
1087 * consistent state which allows truncation of the orphans to restart
1088 * during recovery. Hence we must fix the get_block-vs-truncate race
1089 * by other means, so we have i_data_sem.
1090 */
1091 struct rw_semaphore i_data_sem;
1092 struct inode vfs_inode;
1093 struct jbd2_inode *jinode;
1094
1095 spinlock_t i_raw_lock; /* protects updates to the raw inode */
1096
1097 /*
1098 * File creation time. Its function is same as that of
1099 * struct timespec64 i_{a,c,m}time in the generic inode.
1100 */
1101 struct timespec64 i_crtime;
1102
1103 /* mballoc */
1104 atomic_t i_prealloc_active;
1105 struct rb_root i_prealloc_node;
1106 rwlock_t i_prealloc_lock;
1107
1108 /* extents status tree */
1109 struct ext4_es_tree i_es_tree;
1110 rwlock_t i_es_lock;
1111 struct list_head i_es_list;
1112 unsigned int i_es_all_nr; /* protected by i_es_lock */
1113 unsigned int i_es_shk_nr; /* protected by i_es_lock */
1114 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for
1115 extents to shrink. Protected by
1116 i_es_lock */
1117
1118 /* ialloc */
1119 ext4_group_t i_last_alloc_group;
1120
1121 /* allocation reservation info for delalloc */
1122 /* In case of bigalloc, this refer to clusters rather than blocks */
1123 unsigned int i_reserved_data_blocks;
1124
1125 /* pending cluster reservations for bigalloc file systems */
1126 struct ext4_pending_tree i_pending_tree;
1127
1128 /* on-disk additional length */
1129 __u16 i_extra_isize;
1130
1131 /* Indicate the inline data space. */
1132 u16 i_inline_off;
1133 u16 i_inline_size;
1134
1135 #ifdef CONFIG_QUOTA
1136 /* quota space reservation, managed internally by quota code */
1137 qsize_t i_reserved_quota;
1138 #endif
1139
1140 /* Lock protecting lists below */
1141 spinlock_t i_completed_io_lock;
1142 /*
1143 * Completed IOs that need unwritten extents handling and have
1144 * transaction reserved
1145 */
1146 struct list_head i_rsv_conversion_list;
1147 struct work_struct i_rsv_conversion_work;
1148 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1149
1150 spinlock_t i_block_reservation_lock;
1151
1152 /*
1153 * Transactions that contain inode's metadata needed to complete
1154 * fsync and fdatasync, respectively.
1155 */
1156 tid_t i_sync_tid;
1157 tid_t i_datasync_tid;
1158
1159 #ifdef CONFIG_QUOTA
1160 struct dquot __rcu *i_dquot[MAXQUOTAS];
1161 #endif
1162
1163 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1164 __u32 i_csum_seed;
1165
1166 kprojid_t i_projid;
1167 };
1168
1169 /*
1170 * File system states
1171 */
1172 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
1173 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
1174 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
1175 #define EXT4_FC_REPLAY 0x0020 /* Fast commit replay ongoing */
1176
1177 /*
1178 * Misc. filesystem flags
1179 */
1180 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
1181 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
1182 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
1183
1184 /*
1185 * Mount flags set via mount options or defaults
1186 */
1187 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */
1188 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
1189 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
1190 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
1191 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
1192 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
1193 #define EXT4_MOUNT_ERRORS_MASK 0x00070
1194 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
1195 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
1196 #ifdef CONFIG_FS_DAX
1197 #define EXT4_MOUNT_DAX_ALWAYS 0x00200 /* Direct Access */
1198 #else
1199 #define EXT4_MOUNT_DAX_ALWAYS 0
1200 #endif
1201 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
1202 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
1203 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
1204 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
1205 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
1206 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
1207 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
1208 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
1209 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
1210 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
1211 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */
1212 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota,
1213 * enable enforcement for hidden
1214 * quota files */
1215 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable
1216 * enforcement for hidden quota
1217 * files */
1218 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota
1219 * enforcement */
1220 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
1221 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
1222 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1223 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */
1224 #define EXT4_MOUNT_NO_PREFETCH_BLOCK_BITMAPS 0x4000000
1225 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
1226 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
1227 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
1228 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
1229 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
1230
1231 /*
1232 * Mount flags set either automatically (could not be set by mount option)
1233 * based on per file system feature or property or in special cases such as
1234 * distinguishing between explicit mount option definition and default.
1235 */
1236 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1237 specified delalloc */
1238 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1239 size of blocksize * 8
1240 blocks */
1241 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1242 file systems */
1243 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly
1244 specified journal checksum */
1245
1246 #define EXT4_MOUNT2_JOURNAL_FAST_COMMIT 0x00000010 /* Journal fast commit */
1247 #define EXT4_MOUNT2_DAX_NEVER 0x00000020 /* Do not allow Direct Access */
1248 #define EXT4_MOUNT2_DAX_INODE 0x00000040 /* For printing options only */
1249 #define EXT4_MOUNT2_MB_OPTIMIZE_SCAN 0x00000080 /* Optimize group
1250 * scanning in mballoc
1251 */
1252 #define EXT4_MOUNT2_ABORT 0x00000100 /* Abort filesystem */
1253
1254 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1255 ~EXT4_MOUNT_##opt
1256 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1257 EXT4_MOUNT_##opt
1258 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1259 EXT4_MOUNT_##opt)
1260
1261 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1262 ~EXT4_MOUNT2_##opt
1263 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1264 EXT4_MOUNT2_##opt
1265 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1266 EXT4_MOUNT2_##opt)
1267
1268 #define ext4_test_and_set_bit __test_and_set_bit_le
1269 #define ext4_set_bit __set_bit_le
1270 #define ext4_test_and_clear_bit __test_and_clear_bit_le
1271 #define ext4_clear_bit __clear_bit_le
1272 #define ext4_test_bit test_bit_le
1273 #define ext4_find_next_zero_bit find_next_zero_bit_le
1274 #define ext4_find_next_bit find_next_bit_le
1275
1276 extern void mb_set_bits(void *bm, int cur, int len);
1277
1278 /*
1279 * Maximal mount counts between two filesystem checks
1280 */
1281 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1282 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1283
1284 /*
1285 * Behaviour when detecting errors
1286 */
1287 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1288 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1289 #define EXT4_ERRORS_PANIC 3 /* Panic */
1290 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1291
1292 /* Metadata checksum algorithm codes */
1293 #define EXT4_CRC32C_CHKSUM 1
1294
1295 #define EXT4_LABEL_MAX 16
1296
1297 /*
1298 * Structure of the super block
1299 */
1300 struct ext4_super_block {
1301 /*00*/ __le32 s_inodes_count; /* Inodes count */
1302 __le32 s_blocks_count_lo; /* Blocks count */
1303 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1304 __le32 s_free_blocks_count_lo; /* Free blocks count */
1305 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1306 __le32 s_first_data_block; /* First Data Block */
1307 __le32 s_log_block_size; /* Block size */
1308 __le32 s_log_cluster_size; /* Allocation cluster size */
1309 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1310 __le32 s_clusters_per_group; /* # Clusters per group */
1311 __le32 s_inodes_per_group; /* # Inodes per group */
1312 __le32 s_mtime; /* Mount time */
1313 /*30*/ __le32 s_wtime; /* Write time */
1314 __le16 s_mnt_count; /* Mount count */
1315 __le16 s_max_mnt_count; /* Maximal mount count */
1316 __le16 s_magic; /* Magic signature */
1317 __le16 s_state; /* File system state */
1318 __le16 s_errors; /* Behaviour when detecting errors */
1319 __le16 s_minor_rev_level; /* minor revision level */
1320 /*40*/ __le32 s_lastcheck; /* time of last check */
1321 __le32 s_checkinterval; /* max. time between checks */
1322 __le32 s_creator_os; /* OS */
1323 __le32 s_rev_level; /* Revision level */
1324 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1325 __le16 s_def_resgid; /* Default gid for reserved blocks */
1326 /*
1327 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1328 *
1329 * Note: the difference between the compatible feature set and
1330 * the incompatible feature set is that if there is a bit set
1331 * in the incompatible feature set that the kernel doesn't
1332 * know about, it should refuse to mount the filesystem.
1333 *
1334 * e2fsck's requirements are more strict; if it doesn't know
1335 * about a feature in either the compatible or incompatible
1336 * feature set, it must abort and not try to meddle with
1337 * things it doesn't understand...
1338 */
1339 __le32 s_first_ino; /* First non-reserved inode */
1340 __le16 s_inode_size; /* size of inode structure */
1341 __le16 s_block_group_nr; /* block group # of this superblock */
1342 __le32 s_feature_compat; /* compatible feature set */
1343 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1344 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1345 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1346 /*78*/ char s_volume_name[EXT4_LABEL_MAX]; /* volume name */
1347 /*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */
1348 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1349 /*
1350 * Performance hints. Directory preallocation should only
1351 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1352 */
1353 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1354 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1355 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1356 /*
1357 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1358 */
1359 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1360 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1361 __le32 s_journal_dev; /* device number of journal file */
1362 __le32 s_last_orphan; /* start of list of inodes to delete */
1363 __le32 s_hash_seed[4]; /* HTREE hash seed */
1364 __u8 s_def_hash_version; /* Default hash version to use */
1365 __u8 s_jnl_backup_type;
1366 __le16 s_desc_size; /* size of group descriptor */
1367 /*100*/ __le32 s_default_mount_opts;
1368 __le32 s_first_meta_bg; /* First metablock block group */
1369 __le32 s_mkfs_time; /* When the filesystem was created */
1370 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1371 /* 64bit support valid if EXT4_FEATURE_INCOMPAT_64BIT */
1372 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1373 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1374 __le32 s_free_blocks_count_hi; /* Free blocks count */
1375 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1376 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1377 __le32 s_flags; /* Miscellaneous flags */
1378 __le16 s_raid_stride; /* RAID stride */
1379 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1380 __le64 s_mmp_block; /* Block for multi-mount protection */
1381 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1382 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1383 __u8 s_checksum_type; /* metadata checksum algorithm used */
1384 __u8 s_encryption_level; /* versioning level for encryption */
1385 __u8 s_reserved_pad; /* Padding to next 32bits */
1386 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1387 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1388 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1389 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1390 snapshot's future use */
1391 __le32 s_snapshot_list; /* inode number of the head of the
1392 on-disk snapshot list */
1393 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1394 __le32 s_error_count; /* number of fs errors */
1395 __le32 s_first_error_time; /* first time an error happened */
1396 __le32 s_first_error_ino; /* inode involved in first error */
1397 __le64 s_first_error_block; /* block involved of first error */
1398 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */
1399 __le32 s_first_error_line; /* line number where error happened */
1400 __le32 s_last_error_time; /* most recent time of an error */
1401 __le32 s_last_error_ino; /* inode involved in last error */
1402 __le32 s_last_error_line; /* line number where error happened */
1403 __le64 s_last_error_block; /* block involved of last error */
1404 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */
1405 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1406 __u8 s_mount_opts[64];
1407 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1408 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1409 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1410 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
1411 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */
1412 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */
1413 __le32 s_lpf_ino; /* Location of the lost+found inode */
1414 __le32 s_prj_quota_inum; /* inode for tracking project quota */
1415 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */
1416 __u8 s_wtime_hi;
1417 __u8 s_mtime_hi;
1418 __u8 s_mkfs_time_hi;
1419 __u8 s_lastcheck_hi;
1420 __u8 s_first_error_time_hi;
1421 __u8 s_last_error_time_hi;
1422 __u8 s_first_error_errcode;
1423 __u8 s_last_error_errcode;
1424 __le16 s_encoding; /* Filename charset encoding */
1425 __le16 s_encoding_flags; /* Filename charset encoding flags */
1426 __le32 s_orphan_file_inum; /* Inode for tracking orphan inodes */
1427 __le32 s_reserved[94]; /* Padding to the end of the block */
1428 __le32 s_checksum; /* crc32c(superblock) */
1429 };
1430
1431 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1432
1433 #ifdef __KERNEL__
1434
1435 /* Number of quota types we support */
1436 #define EXT4_MAXQUOTAS 3
1437
1438 #define EXT4_ENC_UTF8_12_1 1
1439
1440 /* Types of ext4 journal triggers */
1441 enum ext4_journal_trigger_type {
1442 EXT4_JTR_ORPHAN_FILE,
1443 EXT4_JTR_NONE /* This must be the last entry for indexing to work! */
1444 };
1445
1446 #define EXT4_JOURNAL_TRIGGER_COUNT EXT4_JTR_NONE
1447
1448 struct ext4_journal_trigger {
1449 struct jbd2_buffer_trigger_type tr_triggers;
1450 struct super_block *sb;
1451 };
1452
EXT4_TRIGGER(struct jbd2_buffer_trigger_type * trigger)1453 static inline struct ext4_journal_trigger *EXT4_TRIGGER(
1454 struct jbd2_buffer_trigger_type *trigger)
1455 {
1456 return container_of(trigger, struct ext4_journal_trigger, tr_triggers);
1457 }
1458
1459 #define EXT4_ORPHAN_BLOCK_MAGIC 0x0b10ca04
1460
1461 /* Structure at the tail of orphan block */
1462 struct ext4_orphan_block_tail {
1463 __le32 ob_magic;
1464 __le32 ob_checksum;
1465 };
1466
ext4_inodes_per_orphan_block(struct super_block * sb)1467 static inline int ext4_inodes_per_orphan_block(struct super_block *sb)
1468 {
1469 return (sb->s_blocksize - sizeof(struct ext4_orphan_block_tail)) /
1470 sizeof(u32);
1471 }
1472
1473 struct ext4_orphan_block {
1474 atomic_t ob_free_entries; /* Number of free orphan entries in block */
1475 struct buffer_head *ob_bh; /* Buffer for orphan block */
1476 };
1477
1478 /*
1479 * Info about orphan file.
1480 */
1481 struct ext4_orphan_info {
1482 int of_blocks; /* Number of orphan blocks in a file */
1483 __u32 of_csum_seed; /* Checksum seed for orphan file */
1484 struct ext4_orphan_block *of_binfo; /* Array with info about orphan
1485 * file blocks */
1486 };
1487
1488 /*
1489 * fourth extended-fs super-block data in memory
1490 */
1491 struct ext4_sb_info {
1492 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1493 unsigned long s_inodes_per_block;/* Number of inodes per block */
1494 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1495 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1496 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1497 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1498 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1499 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1500 ext4_group_t s_groups_count; /* Number of groups in the fs */
1501 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1502 unsigned long s_overhead; /* # of fs overhead clusters */
1503 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1504 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1505 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1506 struct buffer_head * s_sbh; /* Buffer containing the super block */
1507 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1508 struct buffer_head * __rcu *s_group_desc;
1509 unsigned int s_mount_opt;
1510 unsigned int s_mount_opt2;
1511 unsigned long s_mount_flags;
1512 unsigned int s_def_mount_opt;
1513 unsigned int s_def_mount_opt2;
1514 ext4_fsblk_t s_sb_block;
1515 atomic64_t s_resv_clusters;
1516 kuid_t s_resuid;
1517 kgid_t s_resgid;
1518 unsigned short s_mount_state;
1519 unsigned short s_pad;
1520 int s_addr_per_block_bits;
1521 int s_desc_per_block_bits;
1522 int s_inode_size;
1523 int s_first_ino;
1524 unsigned int s_inode_readahead_blks;
1525 unsigned int s_inode_goal;
1526 u32 s_hash_seed[4];
1527 int s_def_hash_version;
1528 int s_hash_unsigned; /* 3 if hash should be unsigned, 0 if not */
1529 struct percpu_counter s_freeclusters_counter;
1530 struct percpu_counter s_freeinodes_counter;
1531 struct percpu_counter s_dirs_counter;
1532 struct percpu_counter s_dirtyclusters_counter;
1533 struct percpu_counter s_sra_exceeded_retry_limit;
1534 struct blockgroup_lock *s_blockgroup_lock;
1535 struct proc_dir_entry *s_proc;
1536 struct kobject s_kobj;
1537 struct completion s_kobj_unregister;
1538 struct super_block *s_sb;
1539 struct buffer_head *s_mmp_bh;
1540
1541 /* Journaling */
1542 struct journal_s *s_journal;
1543 unsigned long s_ext4_flags; /* Ext4 superblock flags */
1544 struct mutex s_orphan_lock; /* Protects on disk list changes */
1545 struct list_head s_orphan; /* List of orphaned inodes in on disk
1546 list */
1547 struct ext4_orphan_info s_orphan_info;
1548 unsigned long s_commit_interval;
1549 u32 s_max_batch_time;
1550 u32 s_min_batch_time;
1551 struct block_device *s_journal_bdev;
1552 #ifdef CONFIG_QUOTA
1553 /* Names of quota files with journalled quota */
1554 char __rcu *s_qf_names[EXT4_MAXQUOTAS];
1555 int s_jquota_fmt; /* Format of quota to use */
1556 #endif
1557 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1558 struct ext4_system_blocks __rcu *s_system_blks;
1559
1560 #ifdef EXTENTS_STATS
1561 /* ext4 extents stats */
1562 unsigned long s_ext_min;
1563 unsigned long s_ext_max;
1564 unsigned long s_depth_max;
1565 spinlock_t s_ext_stats_lock;
1566 unsigned long s_ext_blocks;
1567 unsigned long s_ext_extents;
1568 #endif
1569
1570 /* for buddy allocator */
1571 struct ext4_group_info ** __rcu *s_group_info;
1572 struct inode *s_buddy_cache;
1573 spinlock_t s_md_lock;
1574 unsigned short *s_mb_offsets;
1575 unsigned int *s_mb_maxs;
1576 unsigned int s_group_info_size;
1577 unsigned int s_mb_free_pending;
1578 struct list_head s_freed_data_list; /* List of blocks to be freed
1579 after commit completed */
1580 struct list_head s_discard_list;
1581 struct work_struct s_discard_work;
1582 atomic_t s_retry_alloc_pending;
1583 struct list_head *s_mb_avg_fragment_size;
1584 rwlock_t *s_mb_avg_fragment_size_locks;
1585 struct list_head *s_mb_largest_free_orders;
1586 rwlock_t *s_mb_largest_free_orders_locks;
1587
1588 /* tunables */
1589 unsigned long s_stripe;
1590 unsigned int s_mb_max_linear_groups;
1591 unsigned int s_mb_stream_request;
1592 unsigned int s_mb_max_to_scan;
1593 unsigned int s_mb_min_to_scan;
1594 unsigned int s_mb_stats;
1595 unsigned int s_mb_order2_reqs;
1596 unsigned int s_mb_group_prealloc;
1597 unsigned int s_max_dir_size_kb;
1598 /* where last allocation was done - for stream allocation */
1599 unsigned long s_mb_last_group;
1600 unsigned long s_mb_last_start;
1601 unsigned int s_mb_prefetch;
1602 unsigned int s_mb_prefetch_limit;
1603 unsigned int s_mb_best_avail_max_trim_order;
1604
1605 /* stats for buddy allocator */
1606 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1607 atomic_t s_bal_success; /* we found long enough chunks */
1608 atomic_t s_bal_allocated; /* in blocks */
1609 atomic_t s_bal_ex_scanned; /* total extents scanned */
1610 atomic_t s_bal_cX_ex_scanned[EXT4_MB_NUM_CRS]; /* total extents scanned */
1611 atomic_t s_bal_groups_scanned; /* number of groups scanned */
1612 atomic_t s_bal_goals; /* goal hits */
1613 atomic_t s_bal_len_goals; /* len goal hits */
1614 atomic_t s_bal_breaks; /* too long searches */
1615 atomic_t s_bal_2orders; /* 2^order hits */
1616 atomic_t s_bal_p2_aligned_bad_suggestions;
1617 atomic_t s_bal_goal_fast_bad_suggestions;
1618 atomic_t s_bal_best_avail_bad_suggestions;
1619 atomic64_t s_bal_cX_groups_considered[EXT4_MB_NUM_CRS];
1620 atomic64_t s_bal_cX_hits[EXT4_MB_NUM_CRS];
1621 atomic64_t s_bal_cX_failed[EXT4_MB_NUM_CRS]; /* cX loop didn't find blocks */
1622 atomic_t s_mb_buddies_generated; /* number of buddies generated */
1623 atomic64_t s_mb_generation_time;
1624 atomic_t s_mb_lost_chunks;
1625 atomic_t s_mb_preallocated;
1626 atomic_t s_mb_discarded;
1627 atomic_t s_lock_busy;
1628
1629 /* locality groups */
1630 struct ext4_locality_group __percpu *s_locality_groups;
1631
1632 /* for write statistics */
1633 unsigned long s_sectors_written_start;
1634 u64 s_kbytes_written;
1635
1636 /* the size of zero-out chunk */
1637 unsigned int s_extent_max_zeroout_kb;
1638
1639 unsigned int s_log_groups_per_flex;
1640 struct flex_groups * __rcu *s_flex_groups;
1641 ext4_group_t s_flex_groups_allocated;
1642
1643 /* workqueue for reserved extent conversions (buffered io) */
1644 struct workqueue_struct *rsv_conversion_wq;
1645
1646 /* timer for periodic error stats printing */
1647 struct timer_list s_err_report;
1648
1649 /* Lazy inode table initialization info */
1650 struct ext4_li_request *s_li_request;
1651 /* Wait multiplier for lazy initialization thread */
1652 unsigned int s_li_wait_mult;
1653
1654 /* Kernel thread for multiple mount protection */
1655 struct task_struct *s_mmp_tsk;
1656
1657 /* record the last minlen when FITRIM is called. */
1658 unsigned long s_last_trim_minblks;
1659
1660 /* Reference to checksum algorithm driver via cryptoapi */
1661 struct crypto_shash *s_chksum_driver;
1662
1663 /* Precomputed FS UUID checksum for seeding other checksums */
1664 __u32 s_csum_seed;
1665
1666 /* Reclaim extents from extent status tree */
1667 struct shrinker s_es_shrinker;
1668 struct list_head s_es_list; /* List of inodes with reclaimable extents */
1669 long s_es_nr_inode;
1670 struct ext4_es_stats s_es_stats;
1671 struct mb_cache *s_ea_block_cache;
1672 struct mb_cache *s_ea_inode_cache;
1673 spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1674
1675 /* Journal triggers for checksum computation */
1676 struct ext4_journal_trigger s_journal_triggers[EXT4_JOURNAL_TRIGGER_COUNT];
1677
1678 /* Ratelimit ext4 messages. */
1679 struct ratelimit_state s_err_ratelimit_state;
1680 struct ratelimit_state s_warning_ratelimit_state;
1681 struct ratelimit_state s_msg_ratelimit_state;
1682 atomic_t s_warning_count;
1683 atomic_t s_msg_count;
1684
1685 /* Encryption policy for '-o test_dummy_encryption' */
1686 struct fscrypt_dummy_policy s_dummy_enc_policy;
1687
1688 /*
1689 * Barrier between writepages ops and changing any inode's JOURNAL_DATA
1690 * or EXTENTS flag or between writepages ops and changing DELALLOC or
1691 * DIOREAD_NOLOCK mount options on remount.
1692 */
1693 struct percpu_rw_semaphore s_writepages_rwsem;
1694 struct dax_device *s_daxdev;
1695 u64 s_dax_part_off;
1696 #ifdef CONFIG_EXT4_DEBUG
1697 unsigned long s_simulate_fail;
1698 #endif
1699 /* Record the errseq of the backing block device */
1700 errseq_t s_bdev_wb_err;
1701 spinlock_t s_bdev_wb_lock;
1702
1703 /* Information about errors that happened during this mount */
1704 spinlock_t s_error_lock;
1705 int s_add_error_count;
1706 int s_first_error_code;
1707 __u32 s_first_error_line;
1708 __u32 s_first_error_ino;
1709 __u64 s_first_error_block;
1710 const char *s_first_error_func;
1711 time64_t s_first_error_time;
1712 int s_last_error_code;
1713 __u32 s_last_error_line;
1714 __u32 s_last_error_ino;
1715 __u64 s_last_error_block;
1716 const char *s_last_error_func;
1717 time64_t s_last_error_time;
1718 /*
1719 * If we are in a context where we cannot update the on-disk
1720 * superblock, we queue the work here. This is used to update
1721 * the error information in the superblock, and for periodic
1722 * updates of the superblock called from the commit callback
1723 * function.
1724 */
1725 struct work_struct s_sb_upd_work;
1726
1727 /* Ext4 fast commit sub transaction ID */
1728 atomic_t s_fc_subtid;
1729
1730 /*
1731 * After commit starts, the main queue gets locked, and the further
1732 * updates get added in the staging queue.
1733 */
1734 #define FC_Q_MAIN 0
1735 #define FC_Q_STAGING 1
1736 struct list_head s_fc_q[2]; /* Inodes staged for fast commit
1737 * that have data changes in them.
1738 */
1739 struct list_head s_fc_dentry_q[2]; /* directory entry updates */
1740 unsigned int s_fc_bytes;
1741 /*
1742 * Main fast commit lock. This lock protects accesses to the
1743 * following fields:
1744 * ei->i_fc_list, s_fc_dentry_q, s_fc_q, s_fc_bytes, s_fc_bh.
1745 */
1746 spinlock_t s_fc_lock;
1747 struct buffer_head *s_fc_bh;
1748 struct ext4_fc_stats s_fc_stats;
1749 tid_t s_fc_ineligible_tid;
1750 #ifdef CONFIG_EXT4_DEBUG
1751 int s_fc_debug_max_replay;
1752 #endif
1753 struct ext4_fc_replay_state s_fc_replay_state;
1754 };
1755
EXT4_SB(struct super_block * sb)1756 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1757 {
1758 return sb->s_fs_info;
1759 }
EXT4_I(struct inode * inode)1760 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1761 {
1762 return container_of(inode, struct ext4_inode_info, vfs_inode);
1763 }
1764
ext4_writepages_down_read(struct super_block * sb)1765 static inline int ext4_writepages_down_read(struct super_block *sb)
1766 {
1767 percpu_down_read(&EXT4_SB(sb)->s_writepages_rwsem);
1768 return memalloc_nofs_save();
1769 }
1770
ext4_writepages_up_read(struct super_block * sb,int ctx)1771 static inline void ext4_writepages_up_read(struct super_block *sb, int ctx)
1772 {
1773 memalloc_nofs_restore(ctx);
1774 percpu_up_read(&EXT4_SB(sb)->s_writepages_rwsem);
1775 }
1776
ext4_writepages_down_write(struct super_block * sb)1777 static inline int ext4_writepages_down_write(struct super_block *sb)
1778 {
1779 percpu_down_write(&EXT4_SB(sb)->s_writepages_rwsem);
1780 return memalloc_nofs_save();
1781 }
1782
ext4_writepages_up_write(struct super_block * sb,int ctx)1783 static inline void ext4_writepages_up_write(struct super_block *sb, int ctx)
1784 {
1785 memalloc_nofs_restore(ctx);
1786 percpu_up_write(&EXT4_SB(sb)->s_writepages_rwsem);
1787 }
1788
ext4_valid_inum(struct super_block * sb,unsigned long ino)1789 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1790 {
1791 return ino == EXT4_ROOT_INO ||
1792 (ino >= EXT4_FIRST_INO(sb) &&
1793 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1794 }
1795
1796 /*
1797 * Returns: sbi->field[index]
1798 * Used to access an array element from the following sbi fields which require
1799 * rcu protection to avoid dereferencing an invalid pointer due to reassignment
1800 * - s_group_desc
1801 * - s_group_info
1802 * - s_flex_group
1803 */
1804 #define sbi_array_rcu_deref(sbi, field, index) \
1805 ({ \
1806 typeof(*((sbi)->field)) _v; \
1807 rcu_read_lock(); \
1808 _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index]; \
1809 rcu_read_unlock(); \
1810 _v; \
1811 })
1812
1813 /*
1814 * run-time mount flags
1815 */
1816 enum {
1817 EXT4_MF_MNTDIR_SAMPLED,
1818 EXT4_MF_FC_INELIGIBLE /* Fast commit ineligible */
1819 };
1820
ext4_set_mount_flag(struct super_block * sb,int bit)1821 static inline void ext4_set_mount_flag(struct super_block *sb, int bit)
1822 {
1823 set_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1824 }
1825
ext4_clear_mount_flag(struct super_block * sb,int bit)1826 static inline void ext4_clear_mount_flag(struct super_block *sb, int bit)
1827 {
1828 clear_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1829 }
1830
ext4_test_mount_flag(struct super_block * sb,int bit)1831 static inline int ext4_test_mount_flag(struct super_block *sb, int bit)
1832 {
1833 return test_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1834 }
1835
1836
1837 /*
1838 * Simulate_fail codes
1839 */
1840 #define EXT4_SIM_BBITMAP_EIO 1
1841 #define EXT4_SIM_BBITMAP_CRC 2
1842 #define EXT4_SIM_IBITMAP_EIO 3
1843 #define EXT4_SIM_IBITMAP_CRC 4
1844 #define EXT4_SIM_INODE_EIO 5
1845 #define EXT4_SIM_INODE_CRC 6
1846 #define EXT4_SIM_DIRBLOCK_EIO 7
1847 #define EXT4_SIM_DIRBLOCK_CRC 8
1848
ext4_simulate_fail(struct super_block * sb,unsigned long code)1849 static inline bool ext4_simulate_fail(struct super_block *sb,
1850 unsigned long code)
1851 {
1852 #ifdef CONFIG_EXT4_DEBUG
1853 struct ext4_sb_info *sbi = EXT4_SB(sb);
1854
1855 if (unlikely(sbi->s_simulate_fail == code)) {
1856 sbi->s_simulate_fail = 0;
1857 return true;
1858 }
1859 #endif
1860 return false;
1861 }
1862
1863 /*
1864 * Error number codes for s_{first,last}_error_errno
1865 *
1866 * Linux errno numbers are architecture specific, so we need to translate
1867 * them into something which is architecture independent. We don't define
1868 * codes for all errno's; just the ones which are most likely to be the cause
1869 * of an ext4_error() call.
1870 */
1871 #define EXT4_ERR_UNKNOWN 1
1872 #define EXT4_ERR_EIO 2
1873 #define EXT4_ERR_ENOMEM 3
1874 #define EXT4_ERR_EFSBADCRC 4
1875 #define EXT4_ERR_EFSCORRUPTED 5
1876 #define EXT4_ERR_ENOSPC 6
1877 #define EXT4_ERR_ENOKEY 7
1878 #define EXT4_ERR_EROFS 8
1879 #define EXT4_ERR_EFBIG 9
1880 #define EXT4_ERR_EEXIST 10
1881 #define EXT4_ERR_ERANGE 11
1882 #define EXT4_ERR_EOVERFLOW 12
1883 #define EXT4_ERR_EBUSY 13
1884 #define EXT4_ERR_ENOTDIR 14
1885 #define EXT4_ERR_ENOTEMPTY 15
1886 #define EXT4_ERR_ESHUTDOWN 16
1887 #define EXT4_ERR_EFAULT 17
1888
1889 /*
1890 * Inode dynamic state flags
1891 */
1892 enum {
1893 EXT4_STATE_NEW, /* inode is newly created */
1894 EXT4_STATE_XATTR, /* has in-inode xattrs */
1895 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1896 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1897 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1898 EXT4_STATE_NEWENTRY, /* File just added to dir */
1899 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1900 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
1901 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */
1902 EXT4_STATE_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */
1903 EXT4_STATE_FC_COMMITTING, /* Fast commit ongoing */
1904 EXT4_STATE_ORPHAN_FILE, /* Inode orphaned in orphan file */
1905 };
1906
1907 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1908 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1909 { \
1910 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1911 } \
1912 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1913 { \
1914 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1915 } \
1916 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1917 { \
1918 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1919 }
1920
1921 /* Add these declarations here only so that these functions can be
1922 * found by name. Otherwise, they are very hard to locate. */
1923 static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1924 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1925 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1926 EXT4_INODE_BIT_FNS(flag, flags, 0)
1927
1928 /* Add these declarations here only so that these functions can be
1929 * found by name. Otherwise, they are very hard to locate. */
1930 static inline int ext4_test_inode_state(struct inode *inode, int bit);
1931 static inline void ext4_set_inode_state(struct inode *inode, int bit);
1932 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1933 #if (BITS_PER_LONG < 64)
1934 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1935
ext4_clear_state_flags(struct ext4_inode_info * ei)1936 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1937 {
1938 (ei)->i_state_flags = 0;
1939 }
1940 #else
1941 EXT4_INODE_BIT_FNS(state, flags, 32)
1942
ext4_clear_state_flags(struct ext4_inode_info * ei)1943 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1944 {
1945 /* We depend on the fact that callers will set i_flags */
1946 }
1947 #endif
1948 #else
1949 /* Assume that user mode programs are passing in an ext4fs superblock, not
1950 * a kernel struct super_block. This will allow us to call the feature-test
1951 * macros from user land. */
1952 #define EXT4_SB(sb) (sb)
1953 #endif
1954
ext4_verity_in_progress(struct inode * inode)1955 static inline bool ext4_verity_in_progress(struct inode *inode)
1956 {
1957 return IS_ENABLED(CONFIG_FS_VERITY) &&
1958 ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS);
1959 }
1960
1961 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1962
1963 /*
1964 * Codes for operating systems
1965 */
1966 #define EXT4_OS_LINUX 0
1967 #define EXT4_OS_HURD 1
1968 #define EXT4_OS_MASIX 2
1969 #define EXT4_OS_FREEBSD 3
1970 #define EXT4_OS_LITES 4
1971
1972 /*
1973 * Revision levels
1974 */
1975 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1976 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1977
1978 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1979
1980 #define EXT4_GOOD_OLD_INODE_SIZE 128
1981
1982 #define EXT4_EXTRA_TIMESTAMP_MAX (((s64)1 << 34) - 1 + S32_MIN)
1983 #define EXT4_NON_EXTRA_TIMESTAMP_MAX S32_MAX
1984 #define EXT4_TIMESTAMP_MIN S32_MIN
1985
1986 /*
1987 * Feature set definitions
1988 */
1989
1990 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1991 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1992 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1993 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1994 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1995 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1996 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
1997 /*
1998 * The reason why "FAST_COMMIT" is a compat feature is that, FS becomes
1999 * incompatible only if fast commit blocks are present in the FS. Since we
2000 * clear the journal (and thus the fast commit blocks), we don't mark FS as
2001 * incompatible. We also have a JBD2 incompat feature, which gets set when
2002 * there are fast commit blocks present in the journal.
2003 */
2004 #define EXT4_FEATURE_COMPAT_FAST_COMMIT 0x0400
2005 #define EXT4_FEATURE_COMPAT_STABLE_INODES 0x0800
2006 #define EXT4_FEATURE_COMPAT_ORPHAN_FILE 0x1000 /* Orphan file exists */
2007
2008 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
2009 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
2010 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
2011 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
2012 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
2013 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
2014 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
2015 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
2016 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
2017 /*
2018 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
2019 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
2020 * all other data structures' checksums. However, the METADATA_CSUM and
2021 * GDT_CSUM bits are mutually exclusive.
2022 */
2023 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
2024 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000
2025 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000
2026 #define EXT4_FEATURE_RO_COMPAT_VERITY 0x8000
2027 #define EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT 0x10000 /* Orphan file may be
2028 non-empty */
2029
2030 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
2031 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
2032 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
2033 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
2034 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
2035 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
2036 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
2037 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
2038 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
2039 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
2040 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
2041 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000
2042 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
2043 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
2044 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000
2045 #define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000
2046
2047 extern void ext4_update_dynamic_rev(struct super_block *sb);
2048
2049 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
2050 static inline bool ext4_has_feature_##name(struct super_block *sb) \
2051 { \
2052 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
2053 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
2054 } \
2055 static inline void ext4_set_feature_##name(struct super_block *sb) \
2056 { \
2057 ext4_update_dynamic_rev(sb); \
2058 EXT4_SB(sb)->s_es->s_feature_compat |= \
2059 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
2060 } \
2061 static inline void ext4_clear_feature_##name(struct super_block *sb) \
2062 { \
2063 EXT4_SB(sb)->s_es->s_feature_compat &= \
2064 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
2065 }
2066
2067 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
2068 static inline bool ext4_has_feature_##name(struct super_block *sb) \
2069 { \
2070 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
2071 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
2072 } \
2073 static inline void ext4_set_feature_##name(struct super_block *sb) \
2074 { \
2075 ext4_update_dynamic_rev(sb); \
2076 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
2077 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
2078 } \
2079 static inline void ext4_clear_feature_##name(struct super_block *sb) \
2080 { \
2081 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
2082 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
2083 }
2084
2085 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
2086 static inline bool ext4_has_feature_##name(struct super_block *sb) \
2087 { \
2088 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
2089 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
2090 } \
2091 static inline void ext4_set_feature_##name(struct super_block *sb) \
2092 { \
2093 ext4_update_dynamic_rev(sb); \
2094 EXT4_SB(sb)->s_es->s_feature_incompat |= \
2095 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
2096 } \
2097 static inline void ext4_clear_feature_##name(struct super_block *sb) \
2098 { \
2099 EXT4_SB(sb)->s_es->s_feature_incompat &= \
2100 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
2101 }
2102
EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc,DIR_PREALLOC)2103 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC)
2104 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES)
2105 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL)
2106 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR)
2107 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE)
2108 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX)
2109 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2)
2110 EXT4_FEATURE_COMPAT_FUNCS(fast_commit, FAST_COMMIT)
2111 EXT4_FEATURE_COMPAT_FUNCS(stable_inodes, STABLE_INODES)
2112 EXT4_FEATURE_COMPAT_FUNCS(orphan_file, ORPHAN_FILE)
2113
2114 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER)
2115 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE)
2116 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR)
2117 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE)
2118 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM)
2119 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK)
2120 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE)
2121 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA)
2122 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC)
2123 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM)
2124 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY)
2125 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT)
2126 EXT4_FEATURE_RO_COMPAT_FUNCS(verity, VERITY)
2127 EXT4_FEATURE_RO_COMPAT_FUNCS(orphan_present, ORPHAN_PRESENT)
2128
2129 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION)
2130 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE)
2131 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER)
2132 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV)
2133 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG)
2134 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS)
2135 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT)
2136 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP)
2137 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG)
2138 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE)
2139 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA)
2140 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED)
2141 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR)
2142 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA)
2143 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT)
2144 EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD)
2145
2146 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
2147 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
2148 EXT4_FEATURE_INCOMPAT_META_BG)
2149 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2150 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2151 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
2152
2153 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
2154 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
2155 EXT4_FEATURE_INCOMPAT_RECOVER| \
2156 EXT4_FEATURE_INCOMPAT_META_BG)
2157 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2158 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2159 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
2160
2161 #define EXT4_FEATURE_COMPAT_SUPP (EXT4_FEATURE_COMPAT_EXT_ATTR| \
2162 EXT4_FEATURE_COMPAT_ORPHAN_FILE)
2163 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
2164 EXT4_FEATURE_INCOMPAT_RECOVER| \
2165 EXT4_FEATURE_INCOMPAT_META_BG| \
2166 EXT4_FEATURE_INCOMPAT_EXTENTS| \
2167 EXT4_FEATURE_INCOMPAT_64BIT| \
2168 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
2169 EXT4_FEATURE_INCOMPAT_EA_INODE| \
2170 EXT4_FEATURE_INCOMPAT_MMP | \
2171 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
2172 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
2173 EXT4_FEATURE_INCOMPAT_CASEFOLD | \
2174 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \
2175 EXT4_FEATURE_INCOMPAT_LARGEDIR)
2176 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2177 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2178 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
2179 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
2180 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
2181 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
2182 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
2183 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
2184 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
2185 EXT4_FEATURE_RO_COMPAT_QUOTA |\
2186 EXT4_FEATURE_RO_COMPAT_PROJECT |\
2187 EXT4_FEATURE_RO_COMPAT_VERITY |\
2188 EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT)
2189
2190 #define EXTN_FEATURE_FUNCS(ver) \
2191 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
2192 { \
2193 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
2194 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
2195 } \
2196 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
2197 { \
2198 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
2199 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
2200 } \
2201 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
2202 { \
2203 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
2204 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
2205 }
2206
2207 EXTN_FEATURE_FUNCS(2)
2208 EXTN_FEATURE_FUNCS(3)
2209 EXTN_FEATURE_FUNCS(4)
2210
2211 static inline bool ext4_has_compat_features(struct super_block *sb)
2212 {
2213 return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
2214 }
ext4_has_ro_compat_features(struct super_block * sb)2215 static inline bool ext4_has_ro_compat_features(struct super_block *sb)
2216 {
2217 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
2218 }
ext4_has_incompat_features(struct super_block * sb)2219 static inline bool ext4_has_incompat_features(struct super_block *sb)
2220 {
2221 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
2222 }
2223
2224 extern int ext4_feature_set_ok(struct super_block *sb, int readonly);
2225
2226 /*
2227 * Superblock flags
2228 */
2229 #define EXT4_FLAGS_RESIZING 0
2230 #define EXT4_FLAGS_SHUTDOWN 1
2231 #define EXT4_FLAGS_BDEV_IS_DAX 2
2232
ext4_forced_shutdown(struct super_block * sb)2233 static inline int ext4_forced_shutdown(struct super_block *sb)
2234 {
2235 return test_bit(EXT4_FLAGS_SHUTDOWN, &EXT4_SB(sb)->s_ext4_flags);
2236 }
2237
2238 /*
2239 * Default values for user and/or group using reserved blocks
2240 */
2241 #define EXT4_DEF_RESUID 0
2242 #define EXT4_DEF_RESGID 0
2243
2244 /*
2245 * Default project ID
2246 */
2247 #define EXT4_DEF_PROJID 0
2248
2249 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
2250
2251 /*
2252 * Default mount options
2253 */
2254 #define EXT4_DEFM_DEBUG 0x0001
2255 #define EXT4_DEFM_BSDGROUPS 0x0002
2256 #define EXT4_DEFM_XATTR_USER 0x0004
2257 #define EXT4_DEFM_ACL 0x0008
2258 #define EXT4_DEFM_UID16 0x0010
2259 #define EXT4_DEFM_JMODE 0x0060
2260 #define EXT4_DEFM_JMODE_DATA 0x0020
2261 #define EXT4_DEFM_JMODE_ORDERED 0x0040
2262 #define EXT4_DEFM_JMODE_WBACK 0x0060
2263 #define EXT4_DEFM_NOBARRIER 0x0100
2264 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
2265 #define EXT4_DEFM_DISCARD 0x0400
2266 #define EXT4_DEFM_NODELALLOC 0x0800
2267
2268 /*
2269 * Default journal batch times
2270 */
2271 #define EXT4_DEF_MIN_BATCH_TIME 0
2272 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
2273
2274 /*
2275 * Minimum number of groups in a flexgroup before we separate out
2276 * directories into the first block group of a flexgroup
2277 */
2278 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
2279
2280 /*
2281 * Structure of a directory entry
2282 */
2283 #define EXT4_NAME_LEN 255
2284 /*
2285 * Base length of the ext4 directory entry excluding the name length
2286 */
2287 #define EXT4_BASE_DIR_LEN (sizeof(struct ext4_dir_entry_2) - EXT4_NAME_LEN)
2288
2289 struct ext4_dir_entry {
2290 __le32 inode; /* Inode number */
2291 __le16 rec_len; /* Directory entry length */
2292 __le16 name_len; /* Name length */
2293 char name[EXT4_NAME_LEN]; /* File name */
2294 };
2295
2296
2297 /*
2298 * Encrypted Casefolded entries require saving the hash on disk. This structure
2299 * followed ext4_dir_entry_2's name[name_len] at the next 4 byte aligned
2300 * boundary.
2301 */
2302 struct ext4_dir_entry_hash {
2303 __le32 hash;
2304 __le32 minor_hash;
2305 };
2306
2307 /*
2308 * The new version of the directory entry. Since EXT4 structures are
2309 * stored in intel byte order, and the name_len field could never be
2310 * bigger than 255 chars, it's safe to reclaim the extra byte for the
2311 * file_type field.
2312 */
2313 struct ext4_dir_entry_2 {
2314 __le32 inode; /* Inode number */
2315 __le16 rec_len; /* Directory entry length */
2316 __u8 name_len; /* Name length */
2317 __u8 file_type; /* See file type macros EXT4_FT_* below */
2318 char name[EXT4_NAME_LEN]; /* File name */
2319 };
2320
2321 /*
2322 * Access the hashes at the end of ext4_dir_entry_2
2323 */
2324 #define EXT4_DIRENT_HASHES(entry) \
2325 ((struct ext4_dir_entry_hash *) \
2326 (((void *)(entry)) + \
2327 ((8 + (entry)->name_len + EXT4_DIR_ROUND) & ~EXT4_DIR_ROUND)))
2328 #define EXT4_DIRENT_HASH(entry) le32_to_cpu(EXT4_DIRENT_HASHES(de)->hash)
2329 #define EXT4_DIRENT_MINOR_HASH(entry) \
2330 le32_to_cpu(EXT4_DIRENT_HASHES(de)->minor_hash)
2331
ext4_hash_in_dirent(const struct inode * inode)2332 static inline bool ext4_hash_in_dirent(const struct inode *inode)
2333 {
2334 return IS_CASEFOLDED(inode) && IS_ENCRYPTED(inode);
2335 }
2336
2337 /*
2338 * This is a bogus directory entry at the end of each leaf block that
2339 * records checksums.
2340 */
2341 struct ext4_dir_entry_tail {
2342 __le32 det_reserved_zero1; /* Pretend to be unused */
2343 __le16 det_rec_len; /* 12 */
2344 __u8 det_reserved_zero2; /* Zero name length */
2345 __u8 det_reserved_ft; /* 0xDE, fake file type */
2346 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
2347 };
2348
2349 #define EXT4_DIRENT_TAIL(block, blocksize) \
2350 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
2351 ((blocksize) - \
2352 sizeof(struct ext4_dir_entry_tail))))
2353
2354 /*
2355 * Ext4 directory file types. Only the low 3 bits are used. The
2356 * other bits are reserved for now.
2357 */
2358 #define EXT4_FT_UNKNOWN 0
2359 #define EXT4_FT_REG_FILE 1
2360 #define EXT4_FT_DIR 2
2361 #define EXT4_FT_CHRDEV 3
2362 #define EXT4_FT_BLKDEV 4
2363 #define EXT4_FT_FIFO 5
2364 #define EXT4_FT_SOCK 6
2365 #define EXT4_FT_SYMLINK 7
2366
2367 #define EXT4_FT_MAX 8
2368
2369 #define EXT4_FT_DIR_CSUM 0xDE
2370
2371 /*
2372 * EXT4_DIR_PAD defines the directory entries boundaries
2373 *
2374 * NOTE: It must be a multiple of 4
2375 */
2376 #define EXT4_DIR_PAD 4
2377 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
2378 #define EXT4_MAX_REC_LEN ((1<<16)-1)
2379
2380 /*
2381 * The rec_len is dependent on the type of directory. Directories that are
2382 * casefolded and encrypted need to store the hash as well, so we add room for
2383 * ext4_extended_dir_entry_2. For all entries related to '.' or '..' you should
2384 * pass NULL for dir, as those entries do not use the extra fields.
2385 */
ext4_dir_rec_len(__u8 name_len,const struct inode * dir)2386 static inline unsigned int ext4_dir_rec_len(__u8 name_len,
2387 const struct inode *dir)
2388 {
2389 int rec_len = (name_len + 8 + EXT4_DIR_ROUND);
2390
2391 if (dir && ext4_hash_in_dirent(dir))
2392 rec_len += sizeof(struct ext4_dir_entry_hash);
2393 return (rec_len & ~EXT4_DIR_ROUND);
2394 }
2395
2396 /*
2397 * If we ever get support for fs block sizes > page_size, we'll need
2398 * to remove the #if statements in the next two functions...
2399 */
2400 static inline unsigned int
ext4_rec_len_from_disk(__le16 dlen,unsigned blocksize)2401 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
2402 {
2403 unsigned len = le16_to_cpu(dlen);
2404
2405 #if (PAGE_SIZE >= 65536)
2406 if (len == EXT4_MAX_REC_LEN || len == 0)
2407 return blocksize;
2408 return (len & 65532) | ((len & 3) << 16);
2409 #else
2410 return len;
2411 #endif
2412 }
2413
ext4_rec_len_to_disk(unsigned len,unsigned blocksize)2414 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
2415 {
2416 BUG_ON((len > blocksize) || (blocksize > (1 << 18)) || (len & 3));
2417 #if (PAGE_SIZE >= 65536)
2418 if (len < 65536)
2419 return cpu_to_le16(len);
2420 if (len == blocksize) {
2421 if (blocksize == 65536)
2422 return cpu_to_le16(EXT4_MAX_REC_LEN);
2423 else
2424 return cpu_to_le16(0);
2425 }
2426 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
2427 #else
2428 return cpu_to_le16(len);
2429 #endif
2430 }
2431
2432 /*
2433 * Hash Tree Directory indexing
2434 * (c) Daniel Phillips, 2001
2435 */
2436
2437 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
2438 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2439 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \
2440 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir)))
2441 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2442
2443 /* Legal values for the dx_root hash_version field: */
2444
2445 #define DX_HASH_LEGACY 0
2446 #define DX_HASH_HALF_MD4 1
2447 #define DX_HASH_TEA 2
2448 #define DX_HASH_LEGACY_UNSIGNED 3
2449 #define DX_HASH_HALF_MD4_UNSIGNED 4
2450 #define DX_HASH_TEA_UNSIGNED 5
2451 #define DX_HASH_SIPHASH 6
2452
ext4_chksum(struct ext4_sb_info * sbi,u32 crc,const void * address,unsigned int length)2453 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2454 const void *address, unsigned int length)
2455 {
2456 struct {
2457 struct shash_desc shash;
2458 char ctx[4];
2459 } desc;
2460
2461 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2462
2463 desc.shash.tfm = sbi->s_chksum_driver;
2464 *(u32 *)desc.ctx = crc;
2465
2466 BUG_ON(crypto_shash_update(&desc.shash, address, length));
2467
2468 return *(u32 *)desc.ctx;
2469 }
2470
2471 #ifdef __KERNEL__
2472
2473 /* hash info structure used by the directory hash */
2474 struct dx_hash_info
2475 {
2476 u32 hash;
2477 u32 minor_hash;
2478 int hash_version;
2479 u32 *seed;
2480 };
2481
2482
2483 /* 32 and 64 bit signed EOF for dx directories */
2484 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
2485 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
2486
2487
2488 /*
2489 * Control parameters used by ext4_htree_next_block
2490 */
2491 #define HASH_NB_ALWAYS 1
2492
2493 struct ext4_filename {
2494 const struct qstr *usr_fname;
2495 struct fscrypt_str disk_name;
2496 struct dx_hash_info hinfo;
2497 #ifdef CONFIG_FS_ENCRYPTION
2498 struct fscrypt_str crypto_buf;
2499 #endif
2500 #if IS_ENABLED(CONFIG_UNICODE)
2501 struct fscrypt_str cf_name;
2502 #endif
2503 };
2504
2505 #define fname_name(p) ((p)->disk_name.name)
2506 #define fname_usr_name(p) ((p)->usr_fname->name)
2507 #define fname_len(p) ((p)->disk_name.len)
2508
2509 /*
2510 * Describe an inode's exact location on disk and in memory
2511 */
2512 struct ext4_iloc
2513 {
2514 struct buffer_head *bh;
2515 unsigned long offset;
2516 ext4_group_t block_group;
2517 };
2518
ext4_raw_inode(struct ext4_iloc * iloc)2519 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2520 {
2521 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2522 }
2523
ext4_is_quota_file(struct inode * inode)2524 static inline bool ext4_is_quota_file(struct inode *inode)
2525 {
2526 return IS_NOQUOTA(inode) &&
2527 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL);
2528 }
2529
2530 /*
2531 * This structure is stuffed into the struct file's private_data field
2532 * for directories. It is where we put information so that we can do
2533 * readdir operations in hash tree order.
2534 */
2535 struct dir_private_info {
2536 struct rb_root root;
2537 struct rb_node *curr_node;
2538 struct fname *extra_fname;
2539 loff_t last_pos;
2540 __u32 curr_hash;
2541 __u32 curr_minor_hash;
2542 __u32 next_hash;
2543 };
2544
2545 /* calculate the first block number of the group */
2546 static inline ext4_fsblk_t
ext4_group_first_block_no(struct super_block * sb,ext4_group_t group_no)2547 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2548 {
2549 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2550 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2551 }
2552
2553 /*
2554 * Special error return code only used by dx_probe() and its callers.
2555 */
2556 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
2557
2558 /* htree levels for ext4 */
2559 #define EXT4_HTREE_LEVEL_COMPAT 2
2560 #define EXT4_HTREE_LEVEL 3
2561
ext4_dir_htree_level(struct super_block * sb)2562 static inline int ext4_dir_htree_level(struct super_block *sb)
2563 {
2564 return ext4_has_feature_largedir(sb) ?
2565 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT;
2566 }
2567
2568 /*
2569 * Timeout and state flag for lazy initialization inode thread.
2570 */
2571 #define EXT4_DEF_LI_WAIT_MULT 10
2572 #define EXT4_DEF_LI_MAX_START_DELAY 5
2573 #define EXT4_LAZYINIT_QUIT 0x0001
2574 #define EXT4_LAZYINIT_RUNNING 0x0002
2575
2576 /*
2577 * Lazy inode table initialization info
2578 */
2579 struct ext4_lazy_init {
2580 unsigned long li_state;
2581 struct list_head li_request_list;
2582 struct mutex li_list_mtx;
2583 };
2584
2585 enum ext4_li_mode {
2586 EXT4_LI_MODE_PREFETCH_BBITMAP,
2587 EXT4_LI_MODE_ITABLE,
2588 };
2589
2590 struct ext4_li_request {
2591 struct super_block *lr_super;
2592 enum ext4_li_mode lr_mode;
2593 ext4_group_t lr_first_not_zeroed;
2594 ext4_group_t lr_next_group;
2595 struct list_head lr_request;
2596 unsigned long lr_next_sched;
2597 unsigned long lr_timeout;
2598 };
2599
2600 struct ext4_features {
2601 struct kobject f_kobj;
2602 struct completion f_kobj_unregister;
2603 };
2604
2605 /*
2606 * This structure will be used for multiple mount protection. It will be
2607 * written into the block number saved in the s_mmp_block field in the
2608 * superblock. Programs that check MMP should assume that if
2609 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2610 * to use the filesystem, regardless of how old the timestamp is.
2611 */
2612 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
2613 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2614 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
2615 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
2616
2617 struct mmp_struct {
2618 __le32 mmp_magic; /* Magic number for MMP */
2619 __le32 mmp_seq; /* Sequence no. updated periodically */
2620
2621 /*
2622 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2623 * purposes and do not affect the correctness of the algorithm
2624 */
2625 __le64 mmp_time; /* Time last updated */
2626 char mmp_nodename[64]; /* Node which last updated MMP block */
2627 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
2628
2629 /*
2630 * mmp_check_interval is used to verify if the MMP block has been
2631 * updated on the block device. The value is updated based on the
2632 * maximum time to write the MMP block during an update cycle.
2633 */
2634 __le16 mmp_check_interval;
2635
2636 __le16 mmp_pad1;
2637 __le32 mmp_pad2[226];
2638 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
2639 };
2640
2641 /* arguments passed to the mmp thread */
2642 struct mmpd_data {
2643 struct buffer_head *bh; /* bh from initial read_mmp_block() */
2644 struct super_block *sb; /* super block of the fs */
2645 };
2646
2647 /*
2648 * Check interval multiplier
2649 * The MMP block is written every update interval and initially checked every
2650 * update interval x the multiplier (the value is then adapted based on the
2651 * write latency). The reason is that writes can be delayed under load and we
2652 * don't want readers to incorrectly assume that the filesystem is no longer
2653 * in use.
2654 */
2655 #define EXT4_MMP_CHECK_MULT 2UL
2656
2657 /*
2658 * Minimum interval for MMP checking in seconds.
2659 */
2660 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
2661
2662 /*
2663 * Maximum interval for MMP checking in seconds.
2664 */
2665 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
2666
2667 /*
2668 * Function prototypes
2669 */
2670
2671 /*
2672 * Ok, these declarations are also in <linux/kernel.h> but none of the
2673 * ext4 source programs needs to include it so they are duplicated here.
2674 */
2675 # define NORET_TYPE /**/
2676 # define ATTRIB_NORET __attribute__((noreturn))
2677 # define NORET_AND noreturn,
2678
2679 /* bitmap.c */
2680 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2681 void ext4_inode_bitmap_csum_set(struct super_block *sb,
2682 struct ext4_group_desc *gdp,
2683 struct buffer_head *bh, int sz);
2684 int ext4_inode_bitmap_csum_verify(struct super_block *sb,
2685 struct ext4_group_desc *gdp,
2686 struct buffer_head *bh, int sz);
2687 void ext4_block_bitmap_csum_set(struct super_block *sb,
2688 struct ext4_group_desc *gdp,
2689 struct buffer_head *bh);
2690 int ext4_block_bitmap_csum_verify(struct super_block *sb,
2691 struct ext4_group_desc *gdp,
2692 struct buffer_head *bh);
2693
2694 /* balloc.c */
2695 extern void ext4_get_group_no_and_offset(struct super_block *sb,
2696 ext4_fsblk_t blocknr,
2697 ext4_group_t *blockgrpp,
2698 ext4_grpblk_t *offsetp);
2699 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2700 ext4_fsblk_t block);
2701
2702 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2703 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2704 ext4_group_t group);
2705 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2706 ext4_fsblk_t goal,
2707 unsigned int flags,
2708 unsigned long *count,
2709 int *errp);
2710 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2711 s64 nclusters, unsigned int flags);
2712 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2713 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2714 ext4_group_t block_group,
2715 struct buffer_head ** bh);
2716 extern struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2717 ext4_group_t group);
2718 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2719
2720 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2721 ext4_group_t block_group,
2722 bool ignore_locked);
2723 extern int ext4_wait_block_bitmap(struct super_block *sb,
2724 ext4_group_t block_group,
2725 struct buffer_head *bh);
2726 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2727 ext4_group_t block_group);
2728 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2729 ext4_group_t block_group,
2730 struct ext4_group_desc *gdp);
2731 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2732
2733 #if IS_ENABLED(CONFIG_UNICODE)
2734 extern int ext4_fname_setup_ci_filename(struct inode *dir,
2735 const struct qstr *iname,
2736 struct ext4_filename *fname);
2737 #endif
2738
2739 /* ext4 encryption related stuff goes here crypto.c */
2740 #ifdef CONFIG_FS_ENCRYPTION
2741 extern const struct fscrypt_operations ext4_cryptops;
2742
2743 int ext4_fname_setup_filename(struct inode *dir, const struct qstr *iname,
2744 int lookup, struct ext4_filename *fname);
2745
2746 int ext4_fname_prepare_lookup(struct inode *dir, struct dentry *dentry,
2747 struct ext4_filename *fname);
2748
2749 void ext4_fname_free_filename(struct ext4_filename *fname);
2750
2751 int ext4_ioctl_get_encryption_pwsalt(struct file *filp, void __user *arg);
2752
2753 #else /* !CONFIG_FS_ENCRYPTION */
ext4_fname_setup_filename(struct inode * dir,const struct qstr * iname,int lookup,struct ext4_filename * fname)2754 static inline int ext4_fname_setup_filename(struct inode *dir,
2755 const struct qstr *iname,
2756 int lookup,
2757 struct ext4_filename *fname)
2758 {
2759 int err = 0;
2760 fname->usr_fname = iname;
2761 fname->disk_name.name = (unsigned char *) iname->name;
2762 fname->disk_name.len = iname->len;
2763
2764 #if IS_ENABLED(CONFIG_UNICODE)
2765 err = ext4_fname_setup_ci_filename(dir, iname, fname);
2766 #endif
2767
2768 return err;
2769 }
2770
ext4_fname_prepare_lookup(struct inode * dir,struct dentry * dentry,struct ext4_filename * fname)2771 static inline int ext4_fname_prepare_lookup(struct inode *dir,
2772 struct dentry *dentry,
2773 struct ext4_filename *fname)
2774 {
2775 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname);
2776 }
2777
ext4_fname_free_filename(struct ext4_filename * fname)2778 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2779 {
2780 #if IS_ENABLED(CONFIG_UNICODE)
2781 kfree(fname->cf_name.name);
2782 fname->cf_name.name = NULL;
2783 #endif
2784 }
2785
ext4_ioctl_get_encryption_pwsalt(struct file * filp,void __user * arg)2786 static inline int ext4_ioctl_get_encryption_pwsalt(struct file *filp,
2787 void __user *arg)
2788 {
2789 return -EOPNOTSUPP;
2790 }
2791 #endif /* !CONFIG_FS_ENCRYPTION */
2792
2793 /* dir.c */
2794 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2795 struct file *,
2796 struct ext4_dir_entry_2 *,
2797 struct buffer_head *, char *, int,
2798 unsigned int);
2799 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2800 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2801 (de), (bh), (buf), (size), (offset)))
2802 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2803 __u32 minor_hash,
2804 struct ext4_dir_entry_2 *dirent,
2805 struct fscrypt_str *ent_name);
2806 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2807 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2808 struct buffer_head *bh,
2809 void *buf, int buf_size,
2810 struct ext4_filename *fname,
2811 struct ext4_dir_entry_2 **dest_de);
2812 void ext4_insert_dentry(struct inode *dir, struct inode *inode,
2813 struct ext4_dir_entry_2 *de,
2814 int buf_size,
2815 struct ext4_filename *fname);
ext4_update_dx_flag(struct inode * inode)2816 static inline void ext4_update_dx_flag(struct inode *inode)
2817 {
2818 if (!ext4_has_feature_dir_index(inode->i_sb) &&
2819 ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) {
2820 /* ext4_iget() should have caught this... */
2821 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb));
2822 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2823 }
2824 }
2825 static const unsigned char ext4_filetype_table[] = {
2826 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2827 };
2828
get_dtype(struct super_block * sb,int filetype)2829 static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2830 {
2831 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2832 return DT_UNKNOWN;
2833
2834 return ext4_filetype_table[filetype];
2835 }
2836 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2837 void *buf, int buf_size);
2838
2839 /* fsync.c */
2840 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2841
2842 /* hash.c */
2843 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len,
2844 struct dx_hash_info *hinfo);
2845
2846 /* ialloc.c */
2847 extern int ext4_mark_inode_used(struct super_block *sb, int ino);
2848 extern struct inode *__ext4_new_inode(struct mnt_idmap *, handle_t *,
2849 struct inode *, umode_t,
2850 const struct qstr *qstr, __u32 goal,
2851 uid_t *owner, __u32 i_flags,
2852 int handle_type, unsigned int line_no,
2853 int nblocks);
2854
2855 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \
2856 __ext4_new_inode(&nop_mnt_idmap, (handle), (dir), (mode), (qstr), \
2857 (goal), (owner), i_flags, 0, 0, 0)
2858 #define ext4_new_inode_start_handle(idmap, dir, mode, qstr, goal, owner, \
2859 type, nblocks) \
2860 __ext4_new_inode((idmap), NULL, (dir), (mode), (qstr), (goal), (owner), \
2861 0, (type), __LINE__, (nblocks))
2862
2863
2864 extern void ext4_free_inode(handle_t *, struct inode *);
2865 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2866 extern unsigned long ext4_count_free_inodes(struct super_block *);
2867 extern unsigned long ext4_count_dirs(struct super_block *);
2868 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2869 extern int ext4_init_inode_table(struct super_block *sb,
2870 ext4_group_t group, int barrier);
2871 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2872
2873 /* fast_commit.c */
2874 int ext4_fc_info_show(struct seq_file *seq, void *v);
2875 void ext4_fc_init(struct super_block *sb, journal_t *journal);
2876 void ext4_fc_init_inode(struct inode *inode);
2877 void ext4_fc_track_range(handle_t *handle, struct inode *inode, ext4_lblk_t start,
2878 ext4_lblk_t end);
2879 void __ext4_fc_track_unlink(handle_t *handle, struct inode *inode,
2880 struct dentry *dentry);
2881 void __ext4_fc_track_link(handle_t *handle, struct inode *inode,
2882 struct dentry *dentry);
2883 void ext4_fc_track_unlink(handle_t *handle, struct dentry *dentry);
2884 void ext4_fc_track_link(handle_t *handle, struct dentry *dentry);
2885 void __ext4_fc_track_create(handle_t *handle, struct inode *inode,
2886 struct dentry *dentry);
2887 void ext4_fc_track_create(handle_t *handle, struct dentry *dentry);
2888 void ext4_fc_track_inode(handle_t *handle, struct inode *inode);
2889 void ext4_fc_mark_ineligible(struct super_block *sb, int reason, handle_t *handle);
2890 void ext4_fc_start_update(struct inode *inode);
2891 void ext4_fc_stop_update(struct inode *inode);
2892 void ext4_fc_del(struct inode *inode);
2893 bool ext4_fc_replay_check_excluded(struct super_block *sb, ext4_fsblk_t block);
2894 void ext4_fc_replay_cleanup(struct super_block *sb);
2895 int ext4_fc_commit(journal_t *journal, tid_t commit_tid);
2896 int __init ext4_fc_init_dentry_cache(void);
2897 void ext4_fc_destroy_dentry_cache(void);
2898 int ext4_fc_record_regions(struct super_block *sb, int ino,
2899 ext4_lblk_t lblk, ext4_fsblk_t pblk,
2900 int len, int replay);
2901
2902 /* mballoc.c */
2903 extern const struct seq_operations ext4_mb_seq_groups_ops;
2904 extern const struct seq_operations ext4_mb_seq_structs_summary_ops;
2905 extern int ext4_seq_mb_stats_show(struct seq_file *seq, void *offset);
2906 extern int ext4_mb_init(struct super_block *);
2907 extern int ext4_mb_release(struct super_block *);
2908 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2909 struct ext4_allocation_request *, int *);
2910 extern void ext4_discard_preallocations(struct inode *, unsigned int);
2911 extern int __init ext4_init_mballoc(void);
2912 extern void ext4_exit_mballoc(void);
2913 extern ext4_group_t ext4_mb_prefetch(struct super_block *sb,
2914 ext4_group_t group,
2915 unsigned int nr, int *cnt);
2916 extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group,
2917 unsigned int nr);
2918
2919 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2920 struct buffer_head *bh, ext4_fsblk_t block,
2921 unsigned long count, int flags);
2922 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2923 ext4_group_t ngroups);
2924 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2925 ext4_group_t i, struct ext4_group_desc *desc);
2926 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2927 ext4_fsblk_t block, unsigned long count);
2928 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2929 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid);
2930 extern void ext4_mb_mark_bb(struct super_block *sb, ext4_fsblk_t block,
2931 int len, int state);
ext4_mb_cr_expensive(enum criteria cr)2932 static inline bool ext4_mb_cr_expensive(enum criteria cr)
2933 {
2934 return cr >= CR_GOAL_LEN_SLOW;
2935 }
2936
2937 /* inode.c */
2938 void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw,
2939 struct ext4_inode_info *ei);
2940 int ext4_inode_is_fast_symlink(struct inode *inode);
2941 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2942 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2943 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count,
2944 bool wait, struct buffer_head **bhs);
2945 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2946 struct buffer_head *bh_result, int create);
2947 int ext4_get_block(struct inode *inode, sector_t iblock,
2948 struct buffer_head *bh_result, int create);
2949 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2950 struct buffer_head *bh, int create);
2951 int ext4_walk_page_buffers(handle_t *handle,
2952 struct inode *inode,
2953 struct buffer_head *head,
2954 unsigned from,
2955 unsigned to,
2956 int *partial,
2957 int (*fn)(handle_t *handle, struct inode *inode,
2958 struct buffer_head *bh));
2959 int do_journal_get_write_access(handle_t *handle, struct inode *inode,
2960 struct buffer_head *bh);
2961 #define FALL_BACK_TO_NONDELALLOC 1
2962 #define CONVERT_INLINE_DATA 2
2963
2964 typedef enum {
2965 EXT4_IGET_NORMAL = 0,
2966 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */
2967 EXT4_IGET_HANDLE = 0x0002, /* Inode # is from a handle */
2968 EXT4_IGET_BAD = 0x0004, /* Allow to iget a bad inode */
2969 EXT4_IGET_EA_INODE = 0x0008 /* Inode should contain an EA value */
2970 } ext4_iget_flags;
2971
2972 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
2973 ext4_iget_flags flags, const char *function,
2974 unsigned int line);
2975
2976 #define ext4_iget(sb, ino, flags) \
2977 __ext4_iget((sb), (ino), (flags), __func__, __LINE__)
2978
2979 extern int ext4_write_inode(struct inode *, struct writeback_control *);
2980 extern int ext4_setattr(struct mnt_idmap *, struct dentry *,
2981 struct iattr *);
2982 extern u32 ext4_dio_alignment(struct inode *inode);
2983 extern int ext4_getattr(struct mnt_idmap *, const struct path *,
2984 struct kstat *, u32, unsigned int);
2985 extern void ext4_evict_inode(struct inode *);
2986 extern void ext4_clear_inode(struct inode *);
2987 extern int ext4_file_getattr(struct mnt_idmap *, const struct path *,
2988 struct kstat *, u32, unsigned int);
2989 extern void ext4_dirty_inode(struct inode *, int);
2990 extern int ext4_change_inode_journal_flag(struct inode *, int);
2991 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2992 extern int ext4_get_fc_inode_loc(struct super_block *sb, unsigned long ino,
2993 struct ext4_iloc *iloc);
2994 extern int ext4_inode_attach_jinode(struct inode *inode);
2995 extern int ext4_can_truncate(struct inode *inode);
2996 extern int ext4_truncate(struct inode *);
2997 extern int ext4_break_layouts(struct inode *);
2998 extern int ext4_truncate_page_cache_block_range(struct inode *inode,
2999 loff_t start, loff_t end);
3000 extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length);
3001 extern void ext4_set_inode_flags(struct inode *, bool init);
3002 extern int ext4_alloc_da_blocks(struct inode *inode);
3003 extern void ext4_set_aops(struct inode *inode);
3004 extern int ext4_writepage_trans_blocks(struct inode *);
3005 extern int ext4_normal_submit_inode_data_buffers(struct jbd2_inode *jinode);
3006 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
3007 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
3008 loff_t lstart, loff_t lend);
3009 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf);
3010 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
3011 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
3012 extern void ext4_da_release_space(struct inode *inode, int to_free);
3013 extern void ext4_da_update_reserve_space(struct inode *inode,
3014 int used, int quota_claim);
3015 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
3016 ext4_fsblk_t pblk, ext4_lblk_t len);
3017
3018 /* indirect.c */
3019 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
3020 struct ext4_map_blocks *map, int flags);
3021 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
3022 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
3023 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
3024 ext4_lblk_t start, ext4_lblk_t end);
3025
3026 /* ioctl.c */
3027 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
3028 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
3029 int ext4_fileattr_set(struct mnt_idmap *idmap,
3030 struct dentry *dentry, struct fileattr *fa);
3031 int ext4_fileattr_get(struct dentry *dentry, struct fileattr *fa);
3032 extern void ext4_reset_inode_seed(struct inode *inode);
3033 int ext4_update_overhead(struct super_block *sb, bool force);
3034 int ext4_force_shutdown(struct super_block *sb, u32 flags);
3035
3036 /* migrate.c */
3037 extern int ext4_ext_migrate(struct inode *);
3038 extern int ext4_ind_migrate(struct inode *inode);
3039
3040 /* namei.c */
3041 extern int ext4_init_new_dir(handle_t *handle, struct inode *dir,
3042 struct inode *inode);
3043 extern int ext4_dirblock_csum_verify(struct inode *inode,
3044 struct buffer_head *bh);
3045 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
3046 __u32 start_minor_hash, __u32 *next_hash);
3047 extern int ext4_search_dir(struct buffer_head *bh,
3048 char *search_buf,
3049 int buf_size,
3050 struct inode *dir,
3051 struct ext4_filename *fname,
3052 unsigned int offset,
3053 struct ext4_dir_entry_2 **res_dir);
3054 extern int ext4_generic_delete_entry(struct inode *dir,
3055 struct ext4_dir_entry_2 *de_del,
3056 struct buffer_head *bh,
3057 void *entry_buf,
3058 int buf_size,
3059 int csum_size);
3060 extern bool ext4_empty_dir(struct inode *inode);
3061
3062 /* resize.c */
3063 extern void ext4_kvfree_array_rcu(void *to_free);
3064 extern int ext4_group_add(struct super_block *sb,
3065 struct ext4_new_group_data *input);
3066 extern int ext4_group_extend(struct super_block *sb,
3067 struct ext4_super_block *es,
3068 ext4_fsblk_t n_blocks_count);
3069 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
3070 extern unsigned int ext4_list_backups(struct super_block *sb,
3071 unsigned int *three, unsigned int *five,
3072 unsigned int *seven);
3073
3074 /* super.c */
3075 extern struct buffer_head *ext4_sb_bread(struct super_block *sb,
3076 sector_t block, blk_opf_t op_flags);
3077 extern struct buffer_head *ext4_sb_bread_unmovable(struct super_block *sb,
3078 sector_t block);
3079 extern void ext4_read_bh_nowait(struct buffer_head *bh, blk_opf_t op_flags,
3080 bh_end_io_t *end_io, bool simu_fail);
3081 extern int ext4_read_bh(struct buffer_head *bh, blk_opf_t op_flags,
3082 bh_end_io_t *end_io, bool simu_fail);
3083 extern int ext4_read_bh_lock(struct buffer_head *bh, blk_opf_t op_flags, bool wait);
3084 extern void ext4_sb_breadahead_unmovable(struct super_block *sb, sector_t block);
3085 extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
3086 extern int ext4_calculate_overhead(struct super_block *sb);
3087 extern __le32 ext4_superblock_csum(struct super_block *sb,
3088 struct ext4_super_block *es);
3089 extern void ext4_superblock_csum_set(struct super_block *sb);
3090 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
3091 ext4_group_t ngroup);
3092 extern const char *ext4_decode_error(struct super_block *sb, int errno,
3093 char nbuf[16]);
3094 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb,
3095 ext4_group_t block_group,
3096 unsigned int flags);
3097 extern unsigned int ext4_num_base_meta_blocks(struct super_block *sb,
3098 ext4_group_t block_group);
3099
3100 extern __printf(7, 8)
3101 void __ext4_error(struct super_block *, const char *, unsigned int, bool,
3102 int, __u64, const char *, ...);
3103 extern __printf(6, 7)
3104 void __ext4_error_inode(struct inode *, const char *, unsigned int,
3105 ext4_fsblk_t, int, const char *, ...);
3106 extern __printf(5, 6)
3107 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
3108 const char *, ...);
3109 extern void __ext4_std_error(struct super_block *, const char *,
3110 unsigned int, int);
3111 extern __printf(4, 5)
3112 void __ext4_warning(struct super_block *, const char *, unsigned int,
3113 const char *, ...);
3114 extern __printf(4, 5)
3115 void __ext4_warning_inode(const struct inode *inode, const char *function,
3116 unsigned int line, const char *fmt, ...);
3117 extern __printf(3, 4)
3118 void __ext4_msg(struct super_block *, const char *, const char *, ...);
3119 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
3120 const char *, unsigned int, const char *);
3121 extern __printf(7, 8)
3122 void __ext4_grp_locked_error(const char *, unsigned int,
3123 struct super_block *, ext4_group_t,
3124 unsigned long, ext4_fsblk_t,
3125 const char *, ...);
3126
3127 #define EXT4_ERROR_INODE(inode, fmt, a...) \
3128 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
3129
3130 #define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...) \
3131 __ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a)
3132
3133 #define ext4_error_inode_block(inode, block, err, fmt, a...) \
3134 __ext4_error_inode((inode), __func__, __LINE__, (block), (err), \
3135 (fmt), ## a)
3136
3137 #define EXT4_ERROR_FILE(file, block, fmt, a...) \
3138 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
3139
3140 #define ext4_abort(sb, err, fmt, a...) \
3141 __ext4_error((sb), __func__, __LINE__, true, (err), 0, (fmt), ## a)
3142
3143 #ifdef CONFIG_PRINTK
3144
3145 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
3146 __ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__)
3147 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \
3148 __ext4_error_inode((inode), (func), (line), (block), \
3149 (err), (fmt), ##__VA_ARGS__)
3150 #define ext4_error_file(file, func, line, block, fmt, ...) \
3151 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
3152 #define ext4_error(sb, fmt, ...) \
3153 __ext4_error((sb), __func__, __LINE__, false, 0, 0, (fmt), \
3154 ##__VA_ARGS__)
3155 #define ext4_error_err(sb, err, fmt, ...) \
3156 __ext4_error((sb), __func__, __LINE__, false, (err), 0, (fmt), \
3157 ##__VA_ARGS__)
3158 #define ext4_warning(sb, fmt, ...) \
3159 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
3160 #define ext4_warning_inode(inode, fmt, ...) \
3161 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
3162 #define ext4_msg(sb, level, fmt, ...) \
3163 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
3164 #define dump_mmp_msg(sb, mmp, msg) \
3165 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
3166 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
3167 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
3168 fmt, ##__VA_ARGS__)
3169
3170 #else
3171
3172 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
3173 do { \
3174 no_printk(fmt, ##__VA_ARGS__); \
3175 __ext4_error_inode(inode, "", 0, block, 0, " "); \
3176 } while (0)
3177 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \
3178 do { \
3179 no_printk(fmt, ##__VA_ARGS__); \
3180 __ext4_error_inode(inode, "", 0, block, err, " "); \
3181 } while (0)
3182 #define ext4_error_file(file, func, line, block, fmt, ...) \
3183 do { \
3184 no_printk(fmt, ##__VA_ARGS__); \
3185 __ext4_error_file(file, "", 0, block, " "); \
3186 } while (0)
3187 #define ext4_error(sb, fmt, ...) \
3188 do { \
3189 no_printk(fmt, ##__VA_ARGS__); \
3190 __ext4_error(sb, "", 0, false, 0, 0, " "); \
3191 } while (0)
3192 #define ext4_error_err(sb, err, fmt, ...) \
3193 do { \
3194 no_printk(fmt, ##__VA_ARGS__); \
3195 __ext4_error(sb, "", 0, false, err, 0, " "); \
3196 } while (0)
3197 #define ext4_warning(sb, fmt, ...) \
3198 do { \
3199 no_printk(fmt, ##__VA_ARGS__); \
3200 __ext4_warning(sb, "", 0, " "); \
3201 } while (0)
3202 #define ext4_warning_inode(inode, fmt, ...) \
3203 do { \
3204 no_printk(fmt, ##__VA_ARGS__); \
3205 __ext4_warning_inode(inode, "", 0, " "); \
3206 } while (0)
3207 #define ext4_msg(sb, level, fmt, ...) \
3208 do { \
3209 no_printk(fmt, ##__VA_ARGS__); \
3210 __ext4_msg(sb, "", " "); \
3211 } while (0)
3212 #define dump_mmp_msg(sb, mmp, msg) \
3213 __dump_mmp_msg(sb, mmp, "", 0, "")
3214 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
3215 do { \
3216 no_printk(fmt, ##__VA_ARGS__); \
3217 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
3218 } while (0)
3219
3220 #endif
3221
3222 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
3223 struct ext4_group_desc *bg);
3224 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
3225 struct ext4_group_desc *bg);
3226 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
3227 struct ext4_group_desc *bg);
3228 extern __u32 ext4_free_group_clusters(struct super_block *sb,
3229 struct ext4_group_desc *bg);
3230 extern __u32 ext4_free_inodes_count(struct super_block *sb,
3231 struct ext4_group_desc *bg);
3232 extern __u32 ext4_used_dirs_count(struct super_block *sb,
3233 struct ext4_group_desc *bg);
3234 extern __u32 ext4_itable_unused_count(struct super_block *sb,
3235 struct ext4_group_desc *bg);
3236 extern void ext4_block_bitmap_set(struct super_block *sb,
3237 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3238 extern void ext4_inode_bitmap_set(struct super_block *sb,
3239 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3240 extern void ext4_inode_table_set(struct super_block *sb,
3241 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3242 extern void ext4_free_group_clusters_set(struct super_block *sb,
3243 struct ext4_group_desc *bg,
3244 __u32 count);
3245 extern void ext4_free_inodes_set(struct super_block *sb,
3246 struct ext4_group_desc *bg, __u32 count);
3247 extern void ext4_used_dirs_set(struct super_block *sb,
3248 struct ext4_group_desc *bg, __u32 count);
3249 extern void ext4_itable_unused_set(struct super_block *sb,
3250 struct ext4_group_desc *bg, __u32 count);
3251 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
3252 struct ext4_group_desc *gdp);
3253 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
3254 struct ext4_group_desc *gdp);
3255 extern int ext4_register_li_request(struct super_block *sb,
3256 ext4_group_t first_not_zeroed);
3257
ext4_has_metadata_csum(struct super_block * sb)3258 static inline int ext4_has_metadata_csum(struct super_block *sb)
3259 {
3260 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
3261 !EXT4_SB(sb)->s_chksum_driver);
3262
3263 return ext4_has_feature_metadata_csum(sb) &&
3264 (EXT4_SB(sb)->s_chksum_driver != NULL);
3265 }
3266
ext4_has_group_desc_csum(struct super_block * sb)3267 static inline int ext4_has_group_desc_csum(struct super_block *sb)
3268 {
3269 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb);
3270 }
3271
3272 #define ext4_read_incompat_64bit_val(es, name) \
3273 (((es)->s_feature_incompat & cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT) \
3274 ? (ext4_fsblk_t)le32_to_cpu(es->name##_hi) << 32 : 0) | \
3275 le32_to_cpu(es->name##_lo))
3276
ext4_blocks_count(struct ext4_super_block * es)3277 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
3278 {
3279 return ext4_read_incompat_64bit_val(es, s_blocks_count);
3280 }
3281
ext4_r_blocks_count(struct ext4_super_block * es)3282 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
3283 {
3284 return ext4_read_incompat_64bit_val(es, s_r_blocks_count);
3285 }
3286
ext4_free_blocks_count(struct ext4_super_block * es)3287 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
3288 {
3289 return ext4_read_incompat_64bit_val(es, s_free_blocks_count);
3290 }
3291
ext4_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)3292 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
3293 ext4_fsblk_t blk)
3294 {
3295 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
3296 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
3297 }
3298
ext4_free_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)3299 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
3300 ext4_fsblk_t blk)
3301 {
3302 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
3303 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
3304 }
3305
ext4_r_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)3306 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
3307 ext4_fsblk_t blk)
3308 {
3309 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
3310 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
3311 }
3312
ext4_isize(struct super_block * sb,struct ext4_inode * raw_inode)3313 static inline loff_t ext4_isize(struct super_block *sb,
3314 struct ext4_inode *raw_inode)
3315 {
3316 if (ext4_has_feature_largedir(sb) ||
3317 S_ISREG(le16_to_cpu(raw_inode->i_mode)))
3318 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
3319 le32_to_cpu(raw_inode->i_size_lo);
3320
3321 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
3322 }
3323
ext4_isize_set(struct ext4_inode * raw_inode,loff_t i_size)3324 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
3325 {
3326 raw_inode->i_size_lo = cpu_to_le32(i_size);
3327 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
3328 }
3329
3330 /*
3331 * Reading s_groups_count requires using smp_rmb() afterwards. See
3332 * the locking protocol documented in the comments of ext4_group_add()
3333 * in resize.c
3334 */
ext4_get_groups_count(struct super_block * sb)3335 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
3336 {
3337 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
3338
3339 smp_rmb();
3340 return ngroups;
3341 }
3342
ext4_flex_group(struct ext4_sb_info * sbi,ext4_group_t block_group)3343 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
3344 ext4_group_t block_group)
3345 {
3346 return block_group >> sbi->s_log_groups_per_flex;
3347 }
3348
ext4_flex_bg_size(struct ext4_sb_info * sbi)3349 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
3350 {
3351 return 1 << sbi->s_log_groups_per_flex;
3352 }
3353
3354 #define ext4_std_error(sb, errno) \
3355 do { \
3356 if ((errno)) \
3357 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
3358 } while (0)
3359
3360 #ifdef CONFIG_SMP
3361 /* Each CPU can accumulate percpu_counter_batch clusters in their local
3362 * counters. So we need to make sure we have free clusters more
3363 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
3364 */
3365 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
3366 #else
3367 #define EXT4_FREECLUSTERS_WATERMARK 0
3368 #endif
3369
3370 /* Update i_disksize. Requires i_rwsem to avoid races with truncate */
ext4_update_i_disksize(struct inode * inode,loff_t newsize)3371 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
3372 {
3373 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
3374 !inode_is_locked(inode));
3375 down_write(&EXT4_I(inode)->i_data_sem);
3376 if (newsize > EXT4_I(inode)->i_disksize)
3377 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize);
3378 up_write(&EXT4_I(inode)->i_data_sem);
3379 }
3380
3381 /* Update i_size, i_disksize. Requires i_rwsem to avoid races with truncate */
ext4_update_inode_size(struct inode * inode,loff_t newsize)3382 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
3383 {
3384 int changed = 0;
3385
3386 if (newsize > inode->i_size) {
3387 i_size_write(inode, newsize);
3388 changed = 1;
3389 }
3390 if (newsize > EXT4_I(inode)->i_disksize) {
3391 ext4_update_i_disksize(inode, newsize);
3392 changed |= 2;
3393 }
3394 return changed;
3395 }
3396
3397 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
3398 loff_t len);
3399
3400 struct ext4_group_info {
3401 unsigned long bb_state;
3402 #ifdef AGGRESSIVE_CHECK
3403 unsigned long bb_check_counter;
3404 #endif
3405 struct rb_root bb_free_root;
3406 ext4_grpblk_t bb_first_free; /* first free block */
3407 ext4_grpblk_t bb_free; /* total free blocks */
3408 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
3409 int bb_avg_fragment_size_order; /* order of average
3410 fragment in BG */
3411 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
3412 ext4_group_t bb_group; /* Group number */
3413 struct list_head bb_prealloc_list;
3414 #ifdef DOUBLE_CHECK
3415 void *bb_bitmap;
3416 #endif
3417 struct rw_semaphore alloc_sem;
3418 struct list_head bb_avg_fragment_size_node;
3419 struct list_head bb_largest_free_order_node;
3420 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
3421 * regions, index is order.
3422 * bb_counters[3] = 5 means
3423 * 5 free 8-block regions. */
3424 };
3425
3426 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
3427 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
3428 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
3429 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
3430 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \
3431 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT)
3432 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \
3433 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT)
3434 #define EXT4_GROUP_INFO_BBITMAP_READ_BIT 4
3435
3436 #define EXT4_MB_GRP_NEED_INIT(grp) \
3437 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
3438 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
3439 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3440 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
3441 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3442
3443 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
3444 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3445 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
3446 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3447 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
3448 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3449 #define EXT4_MB_GRP_TEST_AND_SET_READ(grp) \
3450 (test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state)))
3451
3452 #define EXT4_MAX_CONTENTION 8
3453 #define EXT4_CONTENTION_THRESHOLD 2
3454
ext4_group_lock_ptr(struct super_block * sb,ext4_group_t group)3455 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
3456 ext4_group_t group)
3457 {
3458 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
3459 }
3460
3461 /*
3462 * Returns true if the filesystem is busy enough that attempts to
3463 * access the block group locks has run into contention.
3464 */
ext4_fs_is_busy(struct ext4_sb_info * sbi)3465 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
3466 {
3467 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
3468 }
3469
ext4_lock_group(struct super_block * sb,ext4_group_t group)3470 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
3471 {
3472 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
3473 if (spin_trylock(lock))
3474 /*
3475 * We're able to grab the lock right away, so drop the
3476 * lock contention counter.
3477 */
3478 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
3479 else {
3480 /*
3481 * The lock is busy, so bump the contention counter,
3482 * and then wait on the spin lock.
3483 */
3484 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
3485 EXT4_MAX_CONTENTION);
3486 spin_lock(lock);
3487 }
3488 }
3489
ext4_unlock_group(struct super_block * sb,ext4_group_t group)3490 static inline void ext4_unlock_group(struct super_block *sb,
3491 ext4_group_t group)
3492 {
3493 spin_unlock(ext4_group_lock_ptr(sb, group));
3494 }
3495
3496 #ifdef CONFIG_QUOTA
ext4_quota_capable(struct super_block * sb)3497 static inline bool ext4_quota_capable(struct super_block *sb)
3498 {
3499 return (test_opt(sb, QUOTA) || ext4_has_feature_quota(sb));
3500 }
3501
ext4_is_quota_journalled(struct super_block * sb)3502 static inline bool ext4_is_quota_journalled(struct super_block *sb)
3503 {
3504 struct ext4_sb_info *sbi = EXT4_SB(sb);
3505
3506 return (ext4_has_feature_quota(sb) ||
3507 sbi->s_qf_names[USRQUOTA] || sbi->s_qf_names[GRPQUOTA]);
3508 }
3509 int ext4_enable_quotas(struct super_block *sb);
3510 #endif
3511
3512 /*
3513 * Block validity checking
3514 */
3515 #define ext4_check_indirect_blockref(inode, bh) \
3516 ext4_check_blockref(__func__, __LINE__, inode, \
3517 (__le32 *)(bh)->b_data, \
3518 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
3519
3520 #define ext4_ind_check_inode(inode) \
3521 ext4_check_blockref(__func__, __LINE__, inode, \
3522 EXT4_I(inode)->i_data, \
3523 EXT4_NDIR_BLOCKS)
3524
3525 /*
3526 * Inodes and files operations
3527 */
3528
3529 /* dir.c */
3530 extern const struct file_operations ext4_dir_operations;
3531
3532 /* file.c */
3533 extern const struct inode_operations ext4_file_inode_operations;
3534 extern const struct file_operations ext4_file_operations;
3535 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
3536
3537 /* inline.c */
3538 extern int ext4_get_max_inline_size(struct inode *inode);
3539 extern int ext4_find_inline_data_nolock(struct inode *inode);
3540 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
3541
3542 int ext4_readpage_inline(struct inode *inode, struct folio *folio);
3543 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3544 struct inode *inode,
3545 loff_t pos, unsigned len,
3546 struct page **pagep);
3547 int ext4_write_inline_data_end(struct inode *inode, loff_t pos, unsigned len,
3548 unsigned copied, struct folio *folio);
3549 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
3550 struct inode *inode,
3551 loff_t pos, unsigned len,
3552 struct page **pagep,
3553 void **fsdata);
3554 extern int ext4_try_add_inline_entry(handle_t *handle,
3555 struct ext4_filename *fname,
3556 struct inode *dir, struct inode *inode);
3557 extern int ext4_try_create_inline_dir(handle_t *handle,
3558 struct inode *parent,
3559 struct inode *inode);
3560 extern int ext4_read_inline_dir(struct file *filp,
3561 struct dir_context *ctx,
3562 int *has_inline_data);
3563 extern int ext4_inlinedir_to_tree(struct file *dir_file,
3564 struct inode *dir, ext4_lblk_t block,
3565 struct dx_hash_info *hinfo,
3566 __u32 start_hash, __u32 start_minor_hash,
3567 int *has_inline_data);
3568 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3569 struct ext4_filename *fname,
3570 struct ext4_dir_entry_2 **res_dir,
3571 int *has_inline_data);
3572 extern int ext4_delete_inline_entry(handle_t *handle,
3573 struct inode *dir,
3574 struct ext4_dir_entry_2 *de_del,
3575 struct buffer_head *bh,
3576 int *has_inline_data);
3577 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3578 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3579 struct ext4_dir_entry_2 **parent_de,
3580 int *retval);
3581 extern void *ext4_read_inline_link(struct inode *inode);
3582
3583 struct iomap;
3584 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap);
3585
3586 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3587
3588 extern int ext4_convert_inline_data(struct inode *inode);
3589
ext4_has_inline_data(struct inode * inode)3590 static inline int ext4_has_inline_data(struct inode *inode)
3591 {
3592 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3593 EXT4_I(inode)->i_inline_off;
3594 }
3595
3596 /* namei.c */
3597 extern const struct inode_operations ext4_dir_inode_operations;
3598 extern const struct inode_operations ext4_special_inode_operations;
3599 extern struct dentry *ext4_get_parent(struct dentry *child);
3600 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3601 struct ext4_dir_entry_2 *de,
3602 int blocksize, int csum_size,
3603 unsigned int parent_ino, int dotdot_real_len);
3604 extern void ext4_initialize_dirent_tail(struct buffer_head *bh,
3605 unsigned int blocksize);
3606 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode,
3607 struct buffer_head *bh);
3608 extern int __ext4_unlink(struct inode *dir, const struct qstr *d_name,
3609 struct inode *inode, struct dentry *dentry);
3610 extern int __ext4_link(struct inode *dir, struct inode *inode,
3611 struct dentry *dentry);
3612
3613 #define S_SHIFT 12
3614 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = {
3615 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
3616 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
3617 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
3618 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
3619 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
3620 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
3621 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
3622 };
3623
ext4_set_de_type(struct super_block * sb,struct ext4_dir_entry_2 * de,umode_t mode)3624 static inline void ext4_set_de_type(struct super_block *sb,
3625 struct ext4_dir_entry_2 *de,
3626 umode_t mode) {
3627 if (ext4_has_feature_filetype(sb))
3628 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3629 }
3630
3631 /* readpages.c */
3632 extern int ext4_mpage_readpages(struct inode *inode,
3633 struct readahead_control *rac, struct folio *folio);
3634 extern int __init ext4_init_post_read_processing(void);
3635 extern void ext4_exit_post_read_processing(void);
3636
3637 /* symlink.c */
3638 extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3639 extern const struct inode_operations ext4_symlink_inode_operations;
3640 extern const struct inode_operations ext4_fast_symlink_inode_operations;
3641
3642 /* sysfs.c */
3643 extern void ext4_notify_error_sysfs(struct ext4_sb_info *sbi);
3644 extern int ext4_register_sysfs(struct super_block *sb);
3645 extern void ext4_unregister_sysfs(struct super_block *sb);
3646 extern int __init ext4_init_sysfs(void);
3647 extern void ext4_exit_sysfs(void);
3648
3649 /* block_validity */
3650 extern void ext4_release_system_zone(struct super_block *sb);
3651 extern int ext4_setup_system_zone(struct super_block *sb);
3652 extern int __init ext4_init_system_zone(void);
3653 extern void ext4_exit_system_zone(void);
3654 extern int ext4_inode_block_valid(struct inode *inode,
3655 ext4_fsblk_t start_blk,
3656 unsigned int count);
3657 extern int ext4_check_blockref(const char *, unsigned int,
3658 struct inode *, __le32 *, unsigned int);
3659 extern int ext4_sb_block_valid(struct super_block *sb, struct inode *inode,
3660 ext4_fsblk_t start_blk, unsigned int count);
3661
3662
3663 /* extents.c */
3664 struct ext4_ext_path;
3665 struct ext4_extent;
3666
3667 /*
3668 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3669 * __le32.
3670 */
3671 #define EXT_MAX_BLOCKS 0xffffffff
3672
3673 extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode);
3674 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3675 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3676 struct ext4_map_blocks *map, int flags);
3677 extern int ext4_ext_truncate(handle_t *, struct inode *);
3678 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3679 ext4_lblk_t end);
3680 extern void ext4_ext_init(struct super_block *);
3681 extern void ext4_ext_release(struct super_block *);
3682 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3683 loff_t len);
3684 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3685 loff_t offset, ssize_t len);
3686 extern int ext4_convert_unwritten_io_end_vec(handle_t *handle,
3687 ext4_io_end_t *io_end);
3688 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3689 struct ext4_map_blocks *map, int flags);
3690 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3691 int num,
3692 struct ext4_ext_path *path);
3693 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3694 struct ext4_ext_path **,
3695 struct ext4_extent *, int);
3696 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3697 struct ext4_ext_path **,
3698 int flags);
3699 extern void ext4_free_ext_path(struct ext4_ext_path *);
3700 extern int ext4_ext_check_inode(struct inode *inode);
3701 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3702 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3703 __u64 start, __u64 len);
3704 extern int ext4_get_es_cache(struct inode *inode,
3705 struct fiemap_extent_info *fieinfo,
3706 __u64 start, __u64 len);
3707 extern int ext4_ext_precache(struct inode *inode);
3708 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3709 struct inode *inode2, ext4_lblk_t lblk1,
3710 ext4_lblk_t lblk2, ext4_lblk_t count,
3711 int mark_unwritten,int *err);
3712 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu);
3713 extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode,
3714 int check_cred, int restart_cred,
3715 int revoke_cred);
3716 extern void ext4_ext_replay_shrink_inode(struct inode *inode, ext4_lblk_t end);
3717 extern int ext4_ext_replay_set_iblocks(struct inode *inode);
3718 extern int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start,
3719 int len, int unwritten, ext4_fsblk_t pblk);
3720 extern int ext4_ext_clear_bb(struct inode *inode);
3721
3722
3723 /* move_extent.c */
3724 extern void ext4_double_down_write_data_sem(struct inode *first,
3725 struct inode *second);
3726 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3727 struct inode *donor_inode);
3728 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3729 __u64 start_orig, __u64 start_donor,
3730 __u64 len, __u64 *moved_len);
3731
3732 /* page-io.c */
3733 extern int __init ext4_init_pageio(void);
3734 extern void ext4_exit_pageio(void);
3735 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3736 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3737 extern int ext4_put_io_end(ext4_io_end_t *io_end);
3738 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3739 extern void ext4_io_submit_init(struct ext4_io_submit *io,
3740 struct writeback_control *wbc);
3741 extern void ext4_end_io_rsv_work(struct work_struct *work);
3742 extern void ext4_io_submit(struct ext4_io_submit *io);
3743 int ext4_bio_write_folio(struct ext4_io_submit *io, struct folio *page,
3744 size_t len);
3745 extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end);
3746 extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end);
3747
3748 /* mmp.c */
3749 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3750
3751 /* mmp.c */
3752 extern void ext4_stop_mmpd(struct ext4_sb_info *sbi);
3753
3754 /* verity.c */
3755 extern const struct fsverity_operations ext4_verityops;
3756
3757 /* orphan.c */
3758 extern int ext4_orphan_add(handle_t *, struct inode *);
3759 extern int ext4_orphan_del(handle_t *, struct inode *);
3760 extern void ext4_orphan_cleanup(struct super_block *sb,
3761 struct ext4_super_block *es);
3762 extern void ext4_release_orphan_info(struct super_block *sb);
3763 extern int ext4_init_orphan_info(struct super_block *sb);
3764 extern int ext4_orphan_file_empty(struct super_block *sb);
3765 extern void ext4_orphan_file_block_trigger(
3766 struct jbd2_buffer_trigger_type *triggers,
3767 struct buffer_head *bh,
3768 void *data, size_t size);
3769
3770 /*
3771 * Add new method to test whether block and inode bitmaps are properly
3772 * initialized. With uninit_bg reading the block from disk is not enough
3773 * to mark the bitmap uptodate. We need to also zero-out the bitmap
3774 */
3775 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3776
bitmap_uptodate(struct buffer_head * bh)3777 static inline int bitmap_uptodate(struct buffer_head *bh)
3778 {
3779 return (buffer_uptodate(bh) &&
3780 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3781 }
set_bitmap_uptodate(struct buffer_head * bh)3782 static inline void set_bitmap_uptodate(struct buffer_head *bh)
3783 {
3784 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3785 }
3786
3787 /* For ioend & aio unwritten conversion wait queues */
3788 #define EXT4_WQ_HASH_SZ 37
3789 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
3790 EXT4_WQ_HASH_SZ])
3791 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3792
3793 extern int ext4_resize_begin(struct super_block *sb);
3794 extern int ext4_resize_end(struct super_block *sb, bool update_backups);
3795
ext4_set_io_unwritten_flag(struct inode * inode,struct ext4_io_end * io_end)3796 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
3797 struct ext4_io_end *io_end)
3798 {
3799 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
3800 io_end->flag |= EXT4_IO_END_UNWRITTEN;
3801 atomic_inc(&EXT4_I(inode)->i_unwritten);
3802 }
3803 }
3804
ext4_clear_io_unwritten_flag(ext4_io_end_t * io_end)3805 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3806 {
3807 struct inode *inode = io_end->inode;
3808
3809 if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
3810 io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3811 /* Wake up anyone waiting on unwritten extent conversion */
3812 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
3813 wake_up_all(ext4_ioend_wq(inode));
3814 }
3815 }
3816
3817 extern const struct iomap_ops ext4_iomap_ops;
3818 extern const struct iomap_ops ext4_iomap_overwrite_ops;
3819 extern const struct iomap_ops ext4_iomap_report_ops;
3820
ext4_buffer_uptodate(struct buffer_head * bh)3821 static inline int ext4_buffer_uptodate(struct buffer_head *bh)
3822 {
3823 /*
3824 * If the buffer has the write error flag, we have failed
3825 * to write out data in the block. In this case, we don't
3826 * have to read the block because we may read the old data
3827 * successfully.
3828 */
3829 if (buffer_write_io_error(bh))
3830 set_buffer_uptodate(bh);
3831 return buffer_uptodate(bh);
3832 }
3833
3834 #endif /* __KERNEL__ */
3835
3836 #define EFSBADCRC EBADMSG /* Bad CRC detected */
3837 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
3838
3839 #endif /* _EXT4_H */
3840