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
278 struct ext4_io_end_vec {
279 struct list_head list; /* list of io_end_vec */
280 loff_t offset; /* offset in the file */
281 ssize_t size; /* size of the extent */
282 };
283
284 /*
285 * For converting unwritten extents on a work queue. 'handle' is used for
286 * buffered writeback.
287 */
288 typedef struct ext4_io_end {
289 struct list_head list; /* per-file finished IO list */
290 handle_t *handle; /* handle reserved for extent
291 * conversion */
292 struct inode *inode; /* file being written to */
293 struct bio *bio; /* Linked list of completed
294 * bios covering the extent */
295 unsigned int flag; /* unwritten or not */
296 refcount_t count; /* reference counter */
297 struct list_head list_vec; /* list of ext4_io_end_vec */
298 } ext4_io_end_t;
299
300 struct ext4_io_submit {
301 struct writeback_control *io_wbc;
302 struct bio *io_bio;
303 ext4_io_end_t *io_end;
304 sector_t io_next_block;
305 };
306
307 /*
308 * Special inodes numbers
309 */
310 #define EXT4_BAD_INO 1 /* Bad blocks inode */
311 #define EXT4_ROOT_INO 2 /* Root inode */
312 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */
313 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
314 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
315 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
316 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
317 #define EXT4_JOURNAL_INO 8 /* Journal inode */
318
319 /* First non-reserved inode for old ext4 filesystems */
320 #define EXT4_GOOD_OLD_FIRST_INO 11
321
322 /*
323 * Maximal count of links to a file
324 */
325 #define EXT4_LINK_MAX 65000
326
327 /*
328 * Macro-instructions used to manage several block sizes
329 */
330 #define EXT4_MIN_BLOCK_SIZE 1024
331 #define EXT4_MAX_BLOCK_SIZE 65536
332 #define EXT4_MIN_BLOCK_LOG_SIZE 10
333 #define EXT4_MAX_BLOCK_LOG_SIZE 16
334 #define EXT4_MAX_CLUSTER_LOG_SIZE 30
335 #ifdef __KERNEL__
336 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
337 #else
338 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
339 #endif
340 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
341 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
342 EXT4_SB(s)->s_cluster_bits)
343 #ifdef __KERNEL__
344 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
345 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
346 #else
347 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
348 #endif
349 #ifdef __KERNEL__
350 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
351 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
352 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
353 #else
354 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
355 EXT4_GOOD_OLD_INODE_SIZE : \
356 (s)->s_inode_size)
357 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
358 EXT4_GOOD_OLD_FIRST_INO : \
359 (s)->s_first_ino)
360 #endif
361 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
362 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \
363 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \
364 blkbits))
365
366 /* Translate a block number to a cluster number */
367 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
368 /* Translate a cluster number to a block number */
369 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
370 /* Translate # of blks to # of clusters */
371 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
372 (sbi)->s_cluster_bits)
373 /* Mask out the low bits to get the starting block of the cluster */
374 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
375 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
376 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
377 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
378 /* Fill in the low bits to get the last block of the cluster */
379 #define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) | \
380 ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1))
381 /* Get the cluster offset */
382 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
383 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
384 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
385 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
386
387 /*
388 * Structure of a blocks group descriptor
389 */
390 struct ext4_group_desc
391 {
392 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
393 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
394 __le32 bg_inode_table_lo; /* Inodes table block */
395 __le16 bg_free_blocks_count_lo;/* Free blocks count */
396 __le16 bg_free_inodes_count_lo;/* Free inodes count */
397 __le16 bg_used_dirs_count_lo; /* Directories count */
398 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
399 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
400 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
401 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
402 __le16 bg_itable_unused_lo; /* Unused inodes count */
403 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
404 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
405 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
406 __le32 bg_inode_table_hi; /* Inodes table block MSB */
407 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
408 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
409 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
410 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
411 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
412 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
413 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
414 __u32 bg_reserved;
415 };
416
417 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
418 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
419 sizeof(__le16))
420 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
421 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
422 sizeof(__le16))
423
424 /*
425 * Structure of a flex block group info
426 */
427
428 struct flex_groups {
429 atomic64_t free_clusters;
430 atomic_t free_inodes;
431 atomic_t used_dirs;
432 };
433
434 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
435 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
436 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
437
438 /*
439 * Macro-instructions used to manage group descriptors
440 */
441 #define EXT4_MIN_DESC_SIZE 32
442 #define EXT4_MIN_DESC_SIZE_64BIT 64
443 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
444 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
445 #ifdef __KERNEL__
446 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
447 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
448 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
449 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
450 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
451 #else
452 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
453 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
454 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
455 #endif
456
457 /*
458 * Constants relative to the data blocks
459 */
460 #define EXT4_NDIR_BLOCKS 12
461 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
462 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
463 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
464 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
465
466 /*
467 * Inode flags
468 */
469 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
470 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
471 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
472 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
473 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
474 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
475 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
476 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
477 /* Reserved for compression usage... */
478 #define EXT4_DIRTY_FL 0x00000100
479 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
480 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
481 /* nb: was previously EXT2_ECOMPR_FL */
482 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */
483 /* End compression flags --- maybe not all used */
484 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
485 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
486 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
487 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
488 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
489 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
490 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
491 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
492 #define EXT4_VERITY_FL 0x00100000 /* Verity protected inode */
493 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
494 /* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */
495
496 #define EXT4_DAX_FL 0x02000000 /* Inode is DAX */
497
498 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
499 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
500 #define EXT4_CASEFOLD_FL 0x40000000 /* Casefolded directory */
501 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
502
503 /* User modifiable flags */
504 #define EXT4_FL_USER_MODIFIABLE (EXT4_SECRM_FL | \
505 EXT4_UNRM_FL | \
506 EXT4_COMPR_FL | \
507 EXT4_SYNC_FL | \
508 EXT4_IMMUTABLE_FL | \
509 EXT4_APPEND_FL | \
510 EXT4_NODUMP_FL | \
511 EXT4_NOATIME_FL | \
512 EXT4_JOURNAL_DATA_FL | \
513 EXT4_NOTAIL_FL | \
514 EXT4_DIRSYNC_FL | \
515 EXT4_TOPDIR_FL | \
516 EXT4_EXTENTS_FL | \
517 0x00400000 /* EXT4_EOFBLOCKS_FL */ | \
518 EXT4_DAX_FL | \
519 EXT4_PROJINHERIT_FL | \
520 EXT4_CASEFOLD_FL)
521
522 /* User visible flags */
523 #define EXT4_FL_USER_VISIBLE (EXT4_FL_USER_MODIFIABLE | \
524 EXT4_DIRTY_FL | \
525 EXT4_COMPRBLK_FL | \
526 EXT4_NOCOMPR_FL | \
527 EXT4_ENCRYPT_FL | \
528 EXT4_INDEX_FL | \
529 EXT4_VERITY_FL | \
530 EXT4_INLINE_DATA_FL)
531
532 /* Flags that should be inherited by new inodes from their parent. */
533 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
534 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
535 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
536 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
537 EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL |\
538 EXT4_DAX_FL)
539
540 /* Flags that are appropriate for regular files (all but dir-specific ones). */
541 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\
542 EXT4_PROJINHERIT_FL))
543
544 /* Flags that are appropriate for non-directories/regular files. */
545 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
546
547 /* The only flags that should be swapped */
548 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL)
549
550 /* Flags which are mutually exclusive to DAX */
551 #define EXT4_DAX_MUT_EXCL (EXT4_VERITY_FL | EXT4_ENCRYPT_FL |\
552 EXT4_JOURNAL_DATA_FL | EXT4_INLINE_DATA_FL)
553
554 /* Mask out flags that are inappropriate for the given type of inode. */
ext4_mask_flags(umode_t mode,__u32 flags)555 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
556 {
557 if (S_ISDIR(mode))
558 return flags;
559 else if (S_ISREG(mode))
560 return flags & EXT4_REG_FLMASK;
561 else
562 return flags & EXT4_OTHER_FLMASK;
563 }
564
565 /*
566 * Inode flags used for atomic set/get
567 */
568 enum {
569 EXT4_INODE_SECRM = 0, /* Secure deletion */
570 EXT4_INODE_UNRM = 1, /* Undelete */
571 EXT4_INODE_COMPR = 2, /* Compress file */
572 EXT4_INODE_SYNC = 3, /* Synchronous updates */
573 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
574 EXT4_INODE_APPEND = 5, /* writes to file may only append */
575 EXT4_INODE_NODUMP = 6, /* do not dump file */
576 EXT4_INODE_NOATIME = 7, /* do not update atime */
577 /* Reserved for compression usage... */
578 EXT4_INODE_DIRTY = 8,
579 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
580 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
581 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */
582 /* End compression flags --- maybe not all used */
583 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
584 EXT4_INODE_IMAGIC = 13, /* AFS directory */
585 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
586 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
587 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
588 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
589 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
590 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
591 EXT4_INODE_VERITY = 20, /* Verity protected inode */
592 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
593 /* 22 was formerly EXT4_INODE_EOFBLOCKS */
594 EXT4_INODE_DAX = 25, /* Inode is DAX */
595 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
596 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */
597 EXT4_INODE_CASEFOLD = 30, /* Casefolded directory */
598 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
599 };
600
601 /*
602 * Since it's pretty easy to mix up bit numbers and hex values, we use a
603 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
604 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
605 * any extra space in the compiled kernel image, otherwise, the build will fail.
606 * It's important that these values are the same, since we are using
607 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
608 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
609 * values found in ext2, ext3 and ext4 filesystems, and of course the values
610 * defined in e2fsprogs.
611 *
612 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
613 */
614 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1U << EXT4_INODE_##FLAG))
615 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
616
ext4_check_flag_values(void)617 static inline void ext4_check_flag_values(void)
618 {
619 CHECK_FLAG_VALUE(SECRM);
620 CHECK_FLAG_VALUE(UNRM);
621 CHECK_FLAG_VALUE(COMPR);
622 CHECK_FLAG_VALUE(SYNC);
623 CHECK_FLAG_VALUE(IMMUTABLE);
624 CHECK_FLAG_VALUE(APPEND);
625 CHECK_FLAG_VALUE(NODUMP);
626 CHECK_FLAG_VALUE(NOATIME);
627 CHECK_FLAG_VALUE(DIRTY);
628 CHECK_FLAG_VALUE(COMPRBLK);
629 CHECK_FLAG_VALUE(NOCOMPR);
630 CHECK_FLAG_VALUE(ENCRYPT);
631 CHECK_FLAG_VALUE(INDEX);
632 CHECK_FLAG_VALUE(IMAGIC);
633 CHECK_FLAG_VALUE(JOURNAL_DATA);
634 CHECK_FLAG_VALUE(NOTAIL);
635 CHECK_FLAG_VALUE(DIRSYNC);
636 CHECK_FLAG_VALUE(TOPDIR);
637 CHECK_FLAG_VALUE(HUGE_FILE);
638 CHECK_FLAG_VALUE(EXTENTS);
639 CHECK_FLAG_VALUE(VERITY);
640 CHECK_FLAG_VALUE(EA_INODE);
641 CHECK_FLAG_VALUE(INLINE_DATA);
642 CHECK_FLAG_VALUE(PROJINHERIT);
643 CHECK_FLAG_VALUE(CASEFOLD);
644 CHECK_FLAG_VALUE(RESERVED);
645 }
646
647 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
648 struct compat_ext4_new_group_input {
649 u32 group;
650 compat_u64 block_bitmap;
651 compat_u64 inode_bitmap;
652 compat_u64 inode_table;
653 u32 blocks_count;
654 u16 reserved_blocks;
655 u16 unused;
656 };
657 #endif
658
659 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
660 struct ext4_new_group_data {
661 __u32 group;
662 __u64 block_bitmap;
663 __u64 inode_bitmap;
664 __u64 inode_table;
665 __u32 blocks_count;
666 __u16 reserved_blocks;
667 __u16 mdata_blocks;
668 __u32 free_clusters_count;
669 };
670
671 /* Indexes used to index group tables in ext4_new_group_data */
672 enum {
673 BLOCK_BITMAP = 0, /* block bitmap */
674 INODE_BITMAP, /* inode bitmap */
675 INODE_TABLE, /* inode tables */
676 GROUP_TABLE_COUNT,
677 };
678
679 /*
680 * Flags used by ext4_map_blocks()
681 */
682 /* Allocate any needed blocks and/or convert an unwritten
683 extent to be an initialized ext4 */
684 #define EXT4_GET_BLOCKS_CREATE 0x0001
685 /* Request the creation of an unwritten extent */
686 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002
687 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\
688 EXT4_GET_BLOCKS_CREATE)
689 /* Caller is from the delayed allocation writeout path
690 * finally doing the actual allocation of delayed blocks */
691 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
692 /* caller is from the direct IO path, request to creation of an
693 unwritten extents if not allocated, split the unwritten
694 extent if blocks has been preallocated already*/
695 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
696 #define EXT4_GET_BLOCKS_CONVERT 0x0010
697 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
698 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
699 /* Convert extent to initialized after IO complete */
700 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
701 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
702 /* Eventual metadata allocation (due to growing extent tree)
703 * should not fail, so try to use reserved blocks for that.*/
704 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
705 /* Don't normalize allocation size (used for fallocate) */
706 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
707 /* Convert written extents to unwritten */
708 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100
709 /* Write zeros to newly created written extents */
710 #define EXT4_GET_BLOCKS_ZERO 0x0200
711 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\
712 EXT4_GET_BLOCKS_ZERO)
713 /* Caller will submit data before dropping transaction handle. This
714 * allows jbd2 to avoid submitting data before commit. */
715 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400
716 /* Caller is in the atomic contex, find extent if it has been cached */
717 #define EXT4_GET_BLOCKS_CACHED_NOWAIT 0x0800
718
719 /*
720 * The bit position of these flags must not overlap with any of the
721 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(),
722 * read_extent_tree_block(), ext4_split_extent_at(),
723 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
724 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
725 * caching the extents when reading from the extent tree while a
726 * truncate or punch hole operation is in progress.
727 */
728 #define EXT4_EX_NOCACHE 0x40000000
729 #define EXT4_EX_FORCE_CACHE 0x20000000
730 #define EXT4_EX_NOFAIL 0x10000000
731
732 /*
733 * Flags used by ext4_free_blocks
734 */
735 #define EXT4_FREE_BLOCKS_METADATA 0x0001
736 #define EXT4_FREE_BLOCKS_FORGET 0x0002
737 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
738 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
739 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
740 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
741 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040
742
743 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
744 /*
745 * ioctl commands in 32 bit emulation
746 */
747 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
748 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
749 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
750 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
751 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
752 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
753 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
754 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
755 #endif
756
757 /* Max physical block we can address w/o extents */
758 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
759
760 /* Max logical block we can support */
761 #define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFE
762
763 /*
764 * Structure of an inode on the disk
765 */
766 struct ext4_inode {
767 __le16 i_mode; /* File mode */
768 __le16 i_uid; /* Low 16 bits of Owner Uid */
769 __le32 i_size_lo; /* Size in bytes */
770 __le32 i_atime; /* Access time */
771 __le32 i_ctime; /* Inode Change time */
772 __le32 i_mtime; /* Modification time */
773 __le32 i_dtime; /* Deletion Time */
774 __le16 i_gid; /* Low 16 bits of Group Id */
775 __le16 i_links_count; /* Links count */
776 __le32 i_blocks_lo; /* Blocks count */
777 __le32 i_flags; /* File flags */
778 union {
779 struct {
780 __le32 l_i_version;
781 } linux1;
782 struct {
783 __u32 h_i_translator;
784 } hurd1;
785 struct {
786 __u32 m_i_reserved1;
787 } masix1;
788 } osd1; /* OS dependent 1 */
789 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
790 __le32 i_generation; /* File version (for NFS) */
791 __le32 i_file_acl_lo; /* File ACL */
792 __le32 i_size_high;
793 __le32 i_obso_faddr; /* Obsoleted fragment address */
794 union {
795 struct {
796 __le16 l_i_blocks_high; /* were l_i_reserved1 */
797 __le16 l_i_file_acl_high;
798 __le16 l_i_uid_high; /* these 2 fields */
799 __le16 l_i_gid_high; /* were reserved2[0] */
800 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
801 __le16 l_i_reserved;
802 } linux2;
803 struct {
804 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
805 __u16 h_i_mode_high;
806 __u16 h_i_uid_high;
807 __u16 h_i_gid_high;
808 __u32 h_i_author;
809 } hurd2;
810 struct {
811 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
812 __le16 m_i_file_acl_high;
813 __u32 m_i_reserved2[2];
814 } masix2;
815 } osd2; /* OS dependent 2 */
816 __le16 i_extra_isize;
817 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
818 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
819 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
820 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
821 __le32 i_crtime; /* File Creation time */
822 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
823 __le32 i_version_hi; /* high 32 bits for 64-bit version */
824 __le32 i_projid; /* Project ID */
825 };
826
827 #define EXT4_EPOCH_BITS 2
828 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
829 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
830
831 /*
832 * Extended fields will fit into an inode if the filesystem was formatted
833 * with large inodes (-I 256 or larger) and there are not currently any EAs
834 * consuming all of the available space. For new inodes we always reserve
835 * enough space for the kernel's known extended fields, but for inodes
836 * created with an old kernel this might not have been the case. None of
837 * the extended inode fields is critical for correct filesystem operation.
838 * This macro checks if a certain field fits in the inode. Note that
839 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
840 */
841 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
842 ((offsetof(typeof(*ext4_inode), field) + \
843 sizeof((ext4_inode)->field)) \
844 <= (EXT4_GOOD_OLD_INODE_SIZE + \
845 (einode)->i_extra_isize)) \
846
847 /*
848 * We use an encoding that preserves the times for extra epoch "00":
849 *
850 * extra msb of adjust for signed
851 * epoch 32-bit 32-bit tv_sec to
852 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
853 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
854 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
855 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
856 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
857 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
858 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
859 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
860 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
861 *
862 * Note that previous versions of the kernel on 64-bit systems would
863 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
864 * 1970. e2fsck will correct this, assuming that it is run on the
865 * affected filesystem before 2242.
866 */
867
ext4_encode_extra_time(struct timespec64 ts)868 static inline __le32 ext4_encode_extra_time(struct timespec64 ts)
869 {
870 u32 extra = ((ts.tv_sec - (s32)ts.tv_sec) >> 32) & EXT4_EPOCH_MASK;
871 return cpu_to_le32(extra | (ts.tv_nsec << EXT4_EPOCH_BITS));
872 }
873
ext4_decode_extra_time(__le32 base,__le32 extra)874 static inline struct timespec64 ext4_decode_extra_time(__le32 base,
875 __le32 extra)
876 {
877 struct timespec64 ts = { .tv_sec = (signed)le32_to_cpu(base) };
878
879 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK)))
880 ts.tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
881 ts.tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
882 return ts;
883 }
884
885 #define EXT4_INODE_SET_XTIME_VAL(xtime, inode, raw_inode, ts) \
886 do { \
887 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \
888 (raw_inode)->xtime = cpu_to_le32((ts).tv_sec); \
889 (raw_inode)->xtime ## _extra = ext4_encode_extra_time(ts); \
890 } else \
891 (raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (ts).tv_sec, S32_MIN, S32_MAX)); \
892 } while (0)
893
894 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
895 EXT4_INODE_SET_XTIME_VAL(xtime, inode, raw_inode, (inode)->xtime)
896
897 #define EXT4_INODE_SET_CTIME(inode, raw_inode) \
898 EXT4_INODE_SET_XTIME_VAL(i_ctime, inode, raw_inode, inode_get_ctime(inode))
899
900 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
901 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
902 EXT4_INODE_SET_XTIME_VAL(xtime, &((einode)->vfs_inode), \
903 raw_inode, (einode)->xtime)
904
905 #define EXT4_INODE_GET_XTIME_VAL(xtime, inode, raw_inode) \
906 (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra) ? \
907 ext4_decode_extra_time((raw_inode)->xtime, \
908 (raw_inode)->xtime ## _extra) : \
909 (struct timespec64) { \
910 .tv_sec = (signed)le32_to_cpu((raw_inode)->xtime) \
911 })
912
913 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
914 do { \
915 (inode)->xtime = EXT4_INODE_GET_XTIME_VAL(xtime, inode, raw_inode); \
916 } while (0)
917
918 #define EXT4_INODE_GET_CTIME(inode, raw_inode) \
919 do { \
920 inode_set_ctime_to_ts(inode, \
921 EXT4_INODE_GET_XTIME_VAL(i_ctime, inode, raw_inode)); \
922 } while (0)
923
924 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
925 do { \
926 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
927 (einode)->xtime = \
928 EXT4_INODE_GET_XTIME_VAL(xtime, &(einode->vfs_inode), \
929 raw_inode); \
930 else \
931 (einode)->xtime = (struct timespec64){0, 0}; \
932 } while (0)
933
934 #define i_disk_version osd1.linux1.l_i_version
935
936 #if defined(__KERNEL__) || defined(__linux__)
937 #define i_reserved1 osd1.linux1.l_i_reserved1
938 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
939 #define i_blocks_high osd2.linux2.l_i_blocks_high
940 #define i_uid_low i_uid
941 #define i_gid_low i_gid
942 #define i_uid_high osd2.linux2.l_i_uid_high
943 #define i_gid_high osd2.linux2.l_i_gid_high
944 #define i_checksum_lo osd2.linux2.l_i_checksum_lo
945
946 #elif defined(__GNU__)
947
948 #define i_translator osd1.hurd1.h_i_translator
949 #define i_uid_high osd2.hurd2.h_i_uid_high
950 #define i_gid_high osd2.hurd2.h_i_gid_high
951 #define i_author osd2.hurd2.h_i_author
952
953 #elif defined(__masix__)
954
955 #define i_reserved1 osd1.masix1.m_i_reserved1
956 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
957 #define i_reserved2 osd2.masix2.m_i_reserved2
958
959 #endif /* defined(__KERNEL__) || defined(__linux__) */
960
961 #include "extents_status.h"
962 #include "fast_commit.h"
963
964 /*
965 * Lock subclasses for i_data_sem in the ext4_inode_info structure.
966 *
967 * These are needed to avoid lockdep false positives when we need to
968 * allocate blocks to the quota inode during ext4_map_blocks(), while
969 * holding i_data_sem for a normal (non-quota) inode. Since we don't
970 * do quota tracking for the quota inode, this avoids deadlock (as
971 * well as infinite recursion, since it isn't turtles all the way
972 * down...)
973 *
974 * I_DATA_SEM_NORMAL - Used for most inodes
975 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode
976 * where the second inode has larger inode number
977 * than the first
978 * I_DATA_SEM_QUOTA - Used for quota inodes only
979 * I_DATA_SEM_EA - Used for ea_inodes only
980 */
981 enum {
982 I_DATA_SEM_NORMAL = 0,
983 I_DATA_SEM_OTHER,
984 I_DATA_SEM_QUOTA,
985 I_DATA_SEM_EA
986 };
987
988
989 /*
990 * fourth extended file system inode data in memory
991 */
992 struct ext4_inode_info {
993 __le32 i_data[15]; /* unconverted */
994 __u32 i_dtime;
995 ext4_fsblk_t i_file_acl;
996
997 /*
998 * i_block_group is the number of the block group which contains
999 * this file's inode. Constant across the lifetime of the inode,
1000 * it is used for making block allocation decisions - we try to
1001 * place a file's data blocks near its inode block, and new inodes
1002 * near to their parent directory's inode.
1003 */
1004 ext4_group_t i_block_group;
1005 ext4_lblk_t i_dir_start_lookup;
1006 #if (BITS_PER_LONG < 64)
1007 unsigned long i_state_flags; /* Dynamic state flags */
1008 #endif
1009 unsigned long i_flags;
1010
1011 /*
1012 * Extended attributes can be read independently of the main file
1013 * data. Taking i_rwsem even when reading would cause contention
1014 * between readers of EAs and writers of regular file data, so
1015 * instead we synchronize on xattr_sem when reading or changing
1016 * EAs.
1017 */
1018 struct rw_semaphore xattr_sem;
1019
1020 /*
1021 * Inodes with EXT4_STATE_ORPHAN_FILE use i_orphan_idx. Otherwise
1022 * i_orphan is used.
1023 */
1024 union {
1025 struct list_head i_orphan; /* unlinked but open inodes */
1026 unsigned int i_orphan_idx; /* Index in orphan file */
1027 };
1028
1029 /* Fast commit related info */
1030
1031 /* For tracking dentry create updates */
1032 struct list_head i_fc_dilist;
1033 struct list_head i_fc_list; /*
1034 * inodes that need fast commit
1035 * protected by sbi->s_fc_lock.
1036 */
1037
1038 /* Start of lblk range that needs to be committed in this fast commit */
1039 ext4_lblk_t i_fc_lblk_start;
1040
1041 /* End of lblk range that needs to be committed in this fast commit */
1042 ext4_lblk_t i_fc_lblk_len;
1043
1044 /* Number of ongoing updates on this inode */
1045 atomic_t i_fc_updates;
1046
1047 /* Fast commit wait queue for this inode */
1048 wait_queue_head_t i_fc_wait;
1049
1050 /* Protect concurrent accesses on i_fc_lblk_start, i_fc_lblk_len */
1051 struct mutex i_fc_lock;
1052
1053 /*
1054 * i_disksize keeps track of what the inode size is ON DISK, not
1055 * in memory. During truncate, i_size is set to the new size by
1056 * the VFS prior to calling ext4_truncate(), but the filesystem won't
1057 * set i_disksize to 0 until the truncate is actually under way.
1058 *
1059 * The intent is that i_disksize always represents the blocks which
1060 * are used by this file. This allows recovery to restart truncate
1061 * on orphans if we crash during truncate. We actually write i_disksize
1062 * into the on-disk inode when writing inodes out, instead of i_size.
1063 *
1064 * The only time when i_disksize and i_size may be different is when
1065 * a truncate is in progress. The only things which change i_disksize
1066 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
1067 */
1068 loff_t i_disksize;
1069
1070 /*
1071 * i_data_sem is for serialising ext4_truncate() against
1072 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
1073 * data tree are chopped off during truncate. We can't do that in
1074 * ext4 because whenever we perform intermediate commits during
1075 * truncate, the inode and all the metadata blocks *must* be in a
1076 * consistent state which allows truncation of the orphans to restart
1077 * during recovery. Hence we must fix the get_block-vs-truncate race
1078 * by other means, so we have i_data_sem.
1079 */
1080 struct rw_semaphore i_data_sem;
1081 struct inode vfs_inode;
1082 struct jbd2_inode *jinode;
1083
1084 spinlock_t i_raw_lock; /* protects updates to the raw inode */
1085
1086 /*
1087 * File creation time. Its function is same as that of
1088 * struct timespec64 i_{a,c,m}time in the generic inode.
1089 */
1090 struct timespec64 i_crtime;
1091
1092 /* mballoc */
1093 atomic_t i_prealloc_active;
1094 struct rb_root i_prealloc_node;
1095 rwlock_t i_prealloc_lock;
1096
1097 /* extents status tree */
1098 struct ext4_es_tree i_es_tree;
1099 rwlock_t i_es_lock;
1100 struct list_head i_es_list;
1101 unsigned int i_es_all_nr; /* protected by i_es_lock */
1102 unsigned int i_es_shk_nr; /* protected by i_es_lock */
1103 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for
1104 extents to shrink. Protected by
1105 i_es_lock */
1106
1107 /* ialloc */
1108 ext4_group_t i_last_alloc_group;
1109
1110 /* allocation reservation info for delalloc */
1111 /* In case of bigalloc, this refer to clusters rather than blocks */
1112 unsigned int i_reserved_data_blocks;
1113
1114 /* pending cluster reservations for bigalloc file systems */
1115 struct ext4_pending_tree i_pending_tree;
1116
1117 /* on-disk additional length */
1118 __u16 i_extra_isize;
1119
1120 /* Indicate the inline data space. */
1121 u16 i_inline_off;
1122 u16 i_inline_size;
1123
1124 #ifdef CONFIG_QUOTA
1125 /* quota space reservation, managed internally by quota code */
1126 qsize_t i_reserved_quota;
1127 #endif
1128
1129 /* Lock protecting lists below */
1130 spinlock_t i_completed_io_lock;
1131 /*
1132 * Completed IOs that need unwritten extents handling and have
1133 * transaction reserved
1134 */
1135 struct list_head i_rsv_conversion_list;
1136 struct work_struct i_rsv_conversion_work;
1137 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1138
1139 spinlock_t i_block_reservation_lock;
1140
1141 /*
1142 * Transactions that contain inode's metadata needed to complete
1143 * fsync and fdatasync, respectively.
1144 */
1145 tid_t i_sync_tid;
1146 tid_t i_datasync_tid;
1147
1148 #ifdef CONFIG_QUOTA
1149 struct dquot *i_dquot[MAXQUOTAS];
1150 #endif
1151
1152 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1153 __u32 i_csum_seed;
1154
1155 kprojid_t i_projid;
1156 };
1157
1158 /*
1159 * File system states
1160 */
1161 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
1162 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
1163 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
1164 #define EXT4_FC_REPLAY 0x0020 /* Fast commit replay ongoing */
1165
1166 /*
1167 * Misc. filesystem flags
1168 */
1169 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
1170 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
1171 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
1172
1173 /*
1174 * Mount flags set via mount options or defaults
1175 */
1176 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */
1177 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
1178 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
1179 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
1180 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
1181 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
1182 #define EXT4_MOUNT_ERRORS_MASK 0x00070
1183 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
1184 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
1185 #ifdef CONFIG_FS_DAX
1186 #define EXT4_MOUNT_DAX_ALWAYS 0x00200 /* Direct Access */
1187 #else
1188 #define EXT4_MOUNT_DAX_ALWAYS 0
1189 #endif
1190 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
1191 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
1192 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
1193 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
1194 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
1195 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
1196 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
1197 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
1198 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
1199 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
1200 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */
1201 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota,
1202 * enable enforcement for hidden
1203 * quota files */
1204 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable
1205 * enforcement for hidden quota
1206 * files */
1207 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota
1208 * enforcement */
1209 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
1210 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
1211 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1212 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */
1213 #define EXT4_MOUNT_NO_PREFETCH_BLOCK_BITMAPS 0x4000000
1214 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
1215 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
1216 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
1217 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
1218 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
1219
1220 /*
1221 * Mount flags set either automatically (could not be set by mount option)
1222 * based on per file system feature or property or in special cases such as
1223 * distinguishing between explicit mount option definition and default.
1224 */
1225 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1226 specified delalloc */
1227 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1228 size of blocksize * 8
1229 blocks */
1230 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1231 file systems */
1232 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly
1233 specified journal checksum */
1234
1235 #define EXT4_MOUNT2_JOURNAL_FAST_COMMIT 0x00000010 /* Journal fast commit */
1236 #define EXT4_MOUNT2_DAX_NEVER 0x00000020 /* Do not allow Direct Access */
1237 #define EXT4_MOUNT2_DAX_INODE 0x00000040 /* For printing options only */
1238 #define EXT4_MOUNT2_MB_OPTIMIZE_SCAN 0x00000080 /* Optimize group
1239 * scanning in mballoc
1240 */
1241 #define EXT4_MOUNT2_ABORT 0x00000100 /* Abort filesystem */
1242
1243 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1244 ~EXT4_MOUNT_##opt
1245 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1246 EXT4_MOUNT_##opt
1247 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1248 EXT4_MOUNT_##opt)
1249
1250 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1251 ~EXT4_MOUNT2_##opt
1252 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1253 EXT4_MOUNT2_##opt
1254 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1255 EXT4_MOUNT2_##opt)
1256
1257 #define ext4_test_and_set_bit __test_and_set_bit_le
1258 #define ext4_set_bit __set_bit_le
1259 #define ext4_test_and_clear_bit __test_and_clear_bit_le
1260 #define ext4_clear_bit __clear_bit_le
1261 #define ext4_test_bit test_bit_le
1262 #define ext4_find_next_zero_bit find_next_zero_bit_le
1263 #define ext4_find_next_bit find_next_bit_le
1264
1265 extern void mb_set_bits(void *bm, int cur, int len);
1266
1267 /*
1268 * Maximal mount counts between two filesystem checks
1269 */
1270 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1271 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1272
1273 /*
1274 * Behaviour when detecting errors
1275 */
1276 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1277 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1278 #define EXT4_ERRORS_PANIC 3 /* Panic */
1279 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1280
1281 /* Metadata checksum algorithm codes */
1282 #define EXT4_CRC32C_CHKSUM 1
1283
1284 #define EXT4_LABEL_MAX 16
1285
1286 /*
1287 * Structure of the super block
1288 */
1289 struct ext4_super_block {
1290 /*00*/ __le32 s_inodes_count; /* Inodes count */
1291 __le32 s_blocks_count_lo; /* Blocks count */
1292 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1293 __le32 s_free_blocks_count_lo; /* Free blocks count */
1294 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1295 __le32 s_first_data_block; /* First Data Block */
1296 __le32 s_log_block_size; /* Block size */
1297 __le32 s_log_cluster_size; /* Allocation cluster size */
1298 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1299 __le32 s_clusters_per_group; /* # Clusters per group */
1300 __le32 s_inodes_per_group; /* # Inodes per group */
1301 __le32 s_mtime; /* Mount time */
1302 /*30*/ __le32 s_wtime; /* Write time */
1303 __le16 s_mnt_count; /* Mount count */
1304 __le16 s_max_mnt_count; /* Maximal mount count */
1305 __le16 s_magic; /* Magic signature */
1306 __le16 s_state; /* File system state */
1307 __le16 s_errors; /* Behaviour when detecting errors */
1308 __le16 s_minor_rev_level; /* minor revision level */
1309 /*40*/ __le32 s_lastcheck; /* time of last check */
1310 __le32 s_checkinterval; /* max. time between checks */
1311 __le32 s_creator_os; /* OS */
1312 __le32 s_rev_level; /* Revision level */
1313 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1314 __le16 s_def_resgid; /* Default gid for reserved blocks */
1315 /*
1316 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1317 *
1318 * Note: the difference between the compatible feature set and
1319 * the incompatible feature set is that if there is a bit set
1320 * in the incompatible feature set that the kernel doesn't
1321 * know about, it should refuse to mount the filesystem.
1322 *
1323 * e2fsck's requirements are more strict; if it doesn't know
1324 * about a feature in either the compatible or incompatible
1325 * feature set, it must abort and not try to meddle with
1326 * things it doesn't understand...
1327 */
1328 __le32 s_first_ino; /* First non-reserved inode */
1329 __le16 s_inode_size; /* size of inode structure */
1330 __le16 s_block_group_nr; /* block group # of this superblock */
1331 __le32 s_feature_compat; /* compatible feature set */
1332 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1333 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1334 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1335 /*78*/ char s_volume_name[EXT4_LABEL_MAX]; /* volume name */
1336 /*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */
1337 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1338 /*
1339 * Performance hints. Directory preallocation should only
1340 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1341 */
1342 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1343 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1344 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1345 /*
1346 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1347 */
1348 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1349 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1350 __le32 s_journal_dev; /* device number of journal file */
1351 __le32 s_last_orphan; /* start of list of inodes to delete */
1352 __le32 s_hash_seed[4]; /* HTREE hash seed */
1353 __u8 s_def_hash_version; /* Default hash version to use */
1354 __u8 s_jnl_backup_type;
1355 __le16 s_desc_size; /* size of group descriptor */
1356 /*100*/ __le32 s_default_mount_opts;
1357 __le32 s_first_meta_bg; /* First metablock block group */
1358 __le32 s_mkfs_time; /* When the filesystem was created */
1359 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1360 /* 64bit support valid if EXT4_FEATURE_INCOMPAT_64BIT */
1361 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1362 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1363 __le32 s_free_blocks_count_hi; /* Free blocks count */
1364 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1365 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1366 __le32 s_flags; /* Miscellaneous flags */
1367 __le16 s_raid_stride; /* RAID stride */
1368 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1369 __le64 s_mmp_block; /* Block for multi-mount protection */
1370 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1371 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1372 __u8 s_checksum_type; /* metadata checksum algorithm used */
1373 __u8 s_encryption_level; /* versioning level for encryption */
1374 __u8 s_reserved_pad; /* Padding to next 32bits */
1375 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1376 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1377 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1378 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1379 snapshot's future use */
1380 __le32 s_snapshot_list; /* inode number of the head of the
1381 on-disk snapshot list */
1382 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1383 __le32 s_error_count; /* number of fs errors */
1384 __le32 s_first_error_time; /* first time an error happened */
1385 __le32 s_first_error_ino; /* inode involved in first error */
1386 __le64 s_first_error_block; /* block involved of first error */
1387 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */
1388 __le32 s_first_error_line; /* line number where error happened */
1389 __le32 s_last_error_time; /* most recent time of an error */
1390 __le32 s_last_error_ino; /* inode involved in last error */
1391 __le32 s_last_error_line; /* line number where error happened */
1392 __le64 s_last_error_block; /* block involved of last error */
1393 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */
1394 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1395 __u8 s_mount_opts[64];
1396 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1397 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1398 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1399 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
1400 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */
1401 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */
1402 __le32 s_lpf_ino; /* Location of the lost+found inode */
1403 __le32 s_prj_quota_inum; /* inode for tracking project quota */
1404 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */
1405 __u8 s_wtime_hi;
1406 __u8 s_mtime_hi;
1407 __u8 s_mkfs_time_hi;
1408 __u8 s_lastcheck_hi;
1409 __u8 s_first_error_time_hi;
1410 __u8 s_last_error_time_hi;
1411 __u8 s_first_error_errcode;
1412 __u8 s_last_error_errcode;
1413 __le16 s_encoding; /* Filename charset encoding */
1414 __le16 s_encoding_flags; /* Filename charset encoding flags */
1415 __le32 s_orphan_file_inum; /* Inode for tracking orphan inodes */
1416 __le32 s_reserved[94]; /* Padding to the end of the block */
1417 __le32 s_checksum; /* crc32c(superblock) */
1418 };
1419
1420 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1421
1422 #ifdef __KERNEL__
1423
1424 /* Number of quota types we support */
1425 #define EXT4_MAXQUOTAS 3
1426
1427 #define EXT4_ENC_UTF8_12_1 1
1428
1429 /* Types of ext4 journal triggers */
1430 enum ext4_journal_trigger_type {
1431 EXT4_JTR_ORPHAN_FILE,
1432 EXT4_JTR_NONE /* This must be the last entry for indexing to work! */
1433 };
1434
1435 #define EXT4_JOURNAL_TRIGGER_COUNT EXT4_JTR_NONE
1436
1437 struct ext4_journal_trigger {
1438 struct jbd2_buffer_trigger_type tr_triggers;
1439 struct super_block *sb;
1440 };
1441
EXT4_TRIGGER(struct jbd2_buffer_trigger_type * trigger)1442 static inline struct ext4_journal_trigger *EXT4_TRIGGER(
1443 struct jbd2_buffer_trigger_type *trigger)
1444 {
1445 return container_of(trigger, struct ext4_journal_trigger, tr_triggers);
1446 }
1447
1448 #define EXT4_ORPHAN_BLOCK_MAGIC 0x0b10ca04
1449
1450 /* Structure at the tail of orphan block */
1451 struct ext4_orphan_block_tail {
1452 __le32 ob_magic;
1453 __le32 ob_checksum;
1454 };
1455
ext4_inodes_per_orphan_block(struct super_block * sb)1456 static inline int ext4_inodes_per_orphan_block(struct super_block *sb)
1457 {
1458 return (sb->s_blocksize - sizeof(struct ext4_orphan_block_tail)) /
1459 sizeof(u32);
1460 }
1461
1462 struct ext4_orphan_block {
1463 atomic_t ob_free_entries; /* Number of free orphan entries in block */
1464 struct buffer_head *ob_bh; /* Buffer for orphan block */
1465 };
1466
1467 /*
1468 * Info about orphan file.
1469 */
1470 struct ext4_orphan_info {
1471 int of_blocks; /* Number of orphan blocks in a file */
1472 __u32 of_csum_seed; /* Checksum seed for orphan file */
1473 struct ext4_orphan_block *of_binfo; /* Array with info about orphan
1474 * file blocks */
1475 };
1476
1477 /*
1478 * fourth extended-fs super-block data in memory
1479 */
1480 struct ext4_sb_info {
1481 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1482 unsigned long s_inodes_per_block;/* Number of inodes per block */
1483 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1484 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1485 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1486 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1487 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1488 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1489 ext4_group_t s_groups_count; /* Number of groups in the fs */
1490 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1491 unsigned long s_overhead; /* # of fs overhead clusters */
1492 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1493 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1494 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1495 struct buffer_head * s_sbh; /* Buffer containing the super block */
1496 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1497 struct buffer_head * __rcu *s_group_desc;
1498 unsigned int s_mount_opt;
1499 unsigned int s_mount_opt2;
1500 unsigned long s_mount_flags;
1501 unsigned int s_def_mount_opt;
1502 unsigned int s_def_mount_opt2;
1503 ext4_fsblk_t s_sb_block;
1504 atomic64_t s_resv_clusters;
1505 kuid_t s_resuid;
1506 kgid_t s_resgid;
1507 unsigned short s_mount_state;
1508 unsigned short s_pad;
1509 int s_addr_per_block_bits;
1510 int s_desc_per_block_bits;
1511 int s_inode_size;
1512 int s_first_ino;
1513 unsigned int s_inode_readahead_blks;
1514 unsigned int s_inode_goal;
1515 u32 s_hash_seed[4];
1516 int s_def_hash_version;
1517 int s_hash_unsigned; /* 3 if hash should be unsigned, 0 if not */
1518 struct percpu_counter s_freeclusters_counter;
1519 struct percpu_counter s_freeinodes_counter;
1520 struct percpu_counter s_dirs_counter;
1521 struct percpu_counter s_dirtyclusters_counter;
1522 struct percpu_counter s_sra_exceeded_retry_limit;
1523 struct blockgroup_lock *s_blockgroup_lock;
1524 struct proc_dir_entry *s_proc;
1525 struct kobject s_kobj;
1526 struct completion s_kobj_unregister;
1527 struct super_block *s_sb;
1528 struct buffer_head *s_mmp_bh;
1529
1530 /* Journaling */
1531 struct journal_s *s_journal;
1532 unsigned long s_ext4_flags; /* Ext4 superblock flags */
1533 struct mutex s_orphan_lock; /* Protects on disk list changes */
1534 struct list_head s_orphan; /* List of orphaned inodes in on disk
1535 list */
1536 struct ext4_orphan_info s_orphan_info;
1537 unsigned long s_commit_interval;
1538 u32 s_max_batch_time;
1539 u32 s_min_batch_time;
1540 struct block_device *s_journal_bdev;
1541 #ifdef CONFIG_QUOTA
1542 /* Names of quota files with journalled quota */
1543 char __rcu *s_qf_names[EXT4_MAXQUOTAS];
1544 int s_jquota_fmt; /* Format of quota to use */
1545 #endif
1546 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1547 struct ext4_system_blocks __rcu *s_system_blks;
1548
1549 #ifdef EXTENTS_STATS
1550 /* ext4 extents stats */
1551 unsigned long s_ext_min;
1552 unsigned long s_ext_max;
1553 unsigned long s_depth_max;
1554 spinlock_t s_ext_stats_lock;
1555 unsigned long s_ext_blocks;
1556 unsigned long s_ext_extents;
1557 #endif
1558
1559 /* for buddy allocator */
1560 struct ext4_group_info ** __rcu *s_group_info;
1561 struct inode *s_buddy_cache;
1562 spinlock_t s_md_lock;
1563 unsigned short *s_mb_offsets;
1564 unsigned int *s_mb_maxs;
1565 unsigned int s_group_info_size;
1566 unsigned int s_mb_free_pending;
1567 struct list_head s_freed_data_list; /* List of blocks to be freed
1568 after commit completed */
1569 struct list_head s_discard_list;
1570 struct work_struct s_discard_work;
1571 atomic_t s_retry_alloc_pending;
1572 struct list_head *s_mb_avg_fragment_size;
1573 rwlock_t *s_mb_avg_fragment_size_locks;
1574 struct list_head *s_mb_largest_free_orders;
1575 rwlock_t *s_mb_largest_free_orders_locks;
1576
1577 /* tunables */
1578 unsigned long s_stripe;
1579 unsigned int s_mb_max_linear_groups;
1580 unsigned int s_mb_stream_request;
1581 unsigned int s_mb_max_to_scan;
1582 unsigned int s_mb_min_to_scan;
1583 unsigned int s_mb_stats;
1584 unsigned int s_mb_order2_reqs;
1585 unsigned int s_mb_group_prealloc;
1586 unsigned int s_max_dir_size_kb;
1587 /* where last allocation was done - for stream allocation */
1588 unsigned long s_mb_last_group;
1589 unsigned long s_mb_last_start;
1590 unsigned int s_mb_prefetch;
1591 unsigned int s_mb_prefetch_limit;
1592 unsigned int s_mb_best_avail_max_trim_order;
1593
1594 /* stats for buddy allocator */
1595 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1596 atomic_t s_bal_success; /* we found long enough chunks */
1597 atomic_t s_bal_allocated; /* in blocks */
1598 atomic_t s_bal_ex_scanned; /* total extents scanned */
1599 atomic_t s_bal_cX_ex_scanned[EXT4_MB_NUM_CRS]; /* total extents scanned */
1600 atomic_t s_bal_groups_scanned; /* number of groups scanned */
1601 atomic_t s_bal_goals; /* goal hits */
1602 atomic_t s_bal_len_goals; /* len goal hits */
1603 atomic_t s_bal_breaks; /* too long searches */
1604 atomic_t s_bal_2orders; /* 2^order hits */
1605 atomic_t s_bal_p2_aligned_bad_suggestions;
1606 atomic_t s_bal_goal_fast_bad_suggestions;
1607 atomic_t s_bal_best_avail_bad_suggestions;
1608 atomic64_t s_bal_cX_groups_considered[EXT4_MB_NUM_CRS];
1609 atomic64_t s_bal_cX_hits[EXT4_MB_NUM_CRS];
1610 atomic64_t s_bal_cX_failed[EXT4_MB_NUM_CRS]; /* cX loop didn't find blocks */
1611 atomic_t s_mb_buddies_generated; /* number of buddies generated */
1612 atomic64_t s_mb_generation_time;
1613 atomic_t s_mb_lost_chunks;
1614 atomic_t s_mb_preallocated;
1615 atomic_t s_mb_discarded;
1616 atomic_t s_lock_busy;
1617
1618 /* locality groups */
1619 struct ext4_locality_group __percpu *s_locality_groups;
1620
1621 /* for write statistics */
1622 unsigned long s_sectors_written_start;
1623 u64 s_kbytes_written;
1624
1625 /* the size of zero-out chunk */
1626 unsigned int s_extent_max_zeroout_kb;
1627
1628 unsigned int s_log_groups_per_flex;
1629 struct flex_groups * __rcu *s_flex_groups;
1630 ext4_group_t s_flex_groups_allocated;
1631
1632 /* workqueue for reserved extent conversions (buffered io) */
1633 struct workqueue_struct *rsv_conversion_wq;
1634
1635 /* timer for periodic error stats printing */
1636 struct timer_list s_err_report;
1637
1638 /* Lazy inode table initialization info */
1639 struct ext4_li_request *s_li_request;
1640 /* Wait multiplier for lazy initialization thread */
1641 unsigned int s_li_wait_mult;
1642
1643 /* Kernel thread for multiple mount protection */
1644 struct task_struct *s_mmp_tsk;
1645
1646 /* record the last minlen when FITRIM is called. */
1647 unsigned long s_last_trim_minblks;
1648
1649 /* Reference to checksum algorithm driver via cryptoapi */
1650 struct crypto_shash *s_chksum_driver;
1651
1652 /* Precomputed FS UUID checksum for seeding other checksums */
1653 __u32 s_csum_seed;
1654
1655 /* Reclaim extents from extent status tree */
1656 struct shrinker s_es_shrinker;
1657 struct list_head s_es_list; /* List of inodes with reclaimable extents */
1658 long s_es_nr_inode;
1659 struct ext4_es_stats s_es_stats;
1660 struct mb_cache *s_ea_block_cache;
1661 struct mb_cache *s_ea_inode_cache;
1662 spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1663
1664 /* Journal triggers for checksum computation */
1665 struct ext4_journal_trigger s_journal_triggers[EXT4_JOURNAL_TRIGGER_COUNT];
1666
1667 /* Ratelimit ext4 messages. */
1668 struct ratelimit_state s_err_ratelimit_state;
1669 struct ratelimit_state s_warning_ratelimit_state;
1670 struct ratelimit_state s_msg_ratelimit_state;
1671 atomic_t s_warning_count;
1672 atomic_t s_msg_count;
1673
1674 /* Encryption policy for '-o test_dummy_encryption' */
1675 struct fscrypt_dummy_policy s_dummy_enc_policy;
1676
1677 /*
1678 * Barrier between writepages ops and changing any inode's JOURNAL_DATA
1679 * or EXTENTS flag.
1680 */
1681 struct percpu_rw_semaphore s_writepages_rwsem;
1682 struct dax_device *s_daxdev;
1683 u64 s_dax_part_off;
1684 #ifdef CONFIG_EXT4_DEBUG
1685 unsigned long s_simulate_fail;
1686 #endif
1687 /* Record the errseq of the backing block device */
1688 errseq_t s_bdev_wb_err;
1689 spinlock_t s_bdev_wb_lock;
1690
1691 /* Information about errors that happened during this mount */
1692 spinlock_t s_error_lock;
1693 int s_add_error_count;
1694 int s_first_error_code;
1695 __u32 s_first_error_line;
1696 __u32 s_first_error_ino;
1697 __u64 s_first_error_block;
1698 const char *s_first_error_func;
1699 time64_t s_first_error_time;
1700 int s_last_error_code;
1701 __u32 s_last_error_line;
1702 __u32 s_last_error_ino;
1703 __u64 s_last_error_block;
1704 const char *s_last_error_func;
1705 time64_t s_last_error_time;
1706 /*
1707 * If we are in a context where we cannot update the on-disk
1708 * superblock, we queue the work here. This is used to update
1709 * the error information in the superblock, and for periodic
1710 * updates of the superblock called from the commit callback
1711 * function.
1712 */
1713 struct work_struct s_sb_upd_work;
1714
1715 /* Ext4 fast commit sub transaction ID */
1716 atomic_t s_fc_subtid;
1717
1718 /*
1719 * After commit starts, the main queue gets locked, and the further
1720 * updates get added in the staging queue.
1721 */
1722 #define FC_Q_MAIN 0
1723 #define FC_Q_STAGING 1
1724 struct list_head s_fc_q[2]; /* Inodes staged for fast commit
1725 * that have data changes in them.
1726 */
1727 struct list_head s_fc_dentry_q[2]; /* directory entry updates */
1728 unsigned int s_fc_bytes;
1729 /*
1730 * Main fast commit lock. This lock protects accesses to the
1731 * following fields:
1732 * ei->i_fc_list, s_fc_dentry_q, s_fc_q, s_fc_bytes, s_fc_bh.
1733 */
1734 spinlock_t s_fc_lock;
1735 struct buffer_head *s_fc_bh;
1736 struct ext4_fc_stats s_fc_stats;
1737 tid_t s_fc_ineligible_tid;
1738 #ifdef CONFIG_EXT4_DEBUG
1739 int s_fc_debug_max_replay;
1740 #endif
1741 struct ext4_fc_replay_state s_fc_replay_state;
1742 };
1743
EXT4_SB(struct super_block * sb)1744 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1745 {
1746 return sb->s_fs_info;
1747 }
EXT4_I(struct inode * inode)1748 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1749 {
1750 return container_of(inode, struct ext4_inode_info, vfs_inode);
1751 }
1752
ext4_writepages_down_read(struct super_block * sb)1753 static inline int ext4_writepages_down_read(struct super_block *sb)
1754 {
1755 percpu_down_read(&EXT4_SB(sb)->s_writepages_rwsem);
1756 return memalloc_nofs_save();
1757 }
1758
ext4_writepages_up_read(struct super_block * sb,int ctx)1759 static inline void ext4_writepages_up_read(struct super_block *sb, int ctx)
1760 {
1761 memalloc_nofs_restore(ctx);
1762 percpu_up_read(&EXT4_SB(sb)->s_writepages_rwsem);
1763 }
1764
ext4_writepages_down_write(struct super_block * sb)1765 static inline int ext4_writepages_down_write(struct super_block *sb)
1766 {
1767 percpu_down_write(&EXT4_SB(sb)->s_writepages_rwsem);
1768 return memalloc_nofs_save();
1769 }
1770
ext4_writepages_up_write(struct super_block * sb,int ctx)1771 static inline void ext4_writepages_up_write(struct super_block *sb, int ctx)
1772 {
1773 memalloc_nofs_restore(ctx);
1774 percpu_up_write(&EXT4_SB(sb)->s_writepages_rwsem);
1775 }
1776
ext4_valid_inum(struct super_block * sb,unsigned long ino)1777 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1778 {
1779 return ino == EXT4_ROOT_INO ||
1780 (ino >= EXT4_FIRST_INO(sb) &&
1781 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1782 }
1783
1784 /*
1785 * Returns: sbi->field[index]
1786 * Used to access an array element from the following sbi fields which require
1787 * rcu protection to avoid dereferencing an invalid pointer due to reassignment
1788 * - s_group_desc
1789 * - s_group_info
1790 * - s_flex_group
1791 */
1792 #define sbi_array_rcu_deref(sbi, field, index) \
1793 ({ \
1794 typeof(*((sbi)->field)) _v; \
1795 rcu_read_lock(); \
1796 _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index]; \
1797 rcu_read_unlock(); \
1798 _v; \
1799 })
1800
1801 /*
1802 * run-time mount flags
1803 */
1804 enum {
1805 EXT4_MF_MNTDIR_SAMPLED,
1806 EXT4_MF_FC_INELIGIBLE /* Fast commit ineligible */
1807 };
1808
ext4_set_mount_flag(struct super_block * sb,int bit)1809 static inline void ext4_set_mount_flag(struct super_block *sb, int bit)
1810 {
1811 set_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1812 }
1813
ext4_clear_mount_flag(struct super_block * sb,int bit)1814 static inline void ext4_clear_mount_flag(struct super_block *sb, int bit)
1815 {
1816 clear_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1817 }
1818
ext4_test_mount_flag(struct super_block * sb,int bit)1819 static inline int ext4_test_mount_flag(struct super_block *sb, int bit)
1820 {
1821 return test_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1822 }
1823
1824
1825 /*
1826 * Simulate_fail codes
1827 */
1828 #define EXT4_SIM_BBITMAP_EIO 1
1829 #define EXT4_SIM_BBITMAP_CRC 2
1830 #define EXT4_SIM_IBITMAP_EIO 3
1831 #define EXT4_SIM_IBITMAP_CRC 4
1832 #define EXT4_SIM_INODE_EIO 5
1833 #define EXT4_SIM_INODE_CRC 6
1834 #define EXT4_SIM_DIRBLOCK_EIO 7
1835 #define EXT4_SIM_DIRBLOCK_CRC 8
1836
ext4_simulate_fail(struct super_block * sb,unsigned long code)1837 static inline bool ext4_simulate_fail(struct super_block *sb,
1838 unsigned long code)
1839 {
1840 #ifdef CONFIG_EXT4_DEBUG
1841 struct ext4_sb_info *sbi = EXT4_SB(sb);
1842
1843 if (unlikely(sbi->s_simulate_fail == code)) {
1844 sbi->s_simulate_fail = 0;
1845 return true;
1846 }
1847 #endif
1848 return false;
1849 }
1850
ext4_simulate_fail_bh(struct super_block * sb,struct buffer_head * bh,unsigned long code)1851 static inline void ext4_simulate_fail_bh(struct super_block *sb,
1852 struct buffer_head *bh,
1853 unsigned long code)
1854 {
1855 if (!IS_ERR(bh) && ext4_simulate_fail(sb, code))
1856 clear_buffer_uptodate(bh);
1857 }
1858
1859 /*
1860 * Error number codes for s_{first,last}_error_errno
1861 *
1862 * Linux errno numbers are architecture specific, so we need to translate
1863 * them into something which is architecture independent. We don't define
1864 * codes for all errno's; just the ones which are most likely to be the cause
1865 * of an ext4_error() call.
1866 */
1867 #define EXT4_ERR_UNKNOWN 1
1868 #define EXT4_ERR_EIO 2
1869 #define EXT4_ERR_ENOMEM 3
1870 #define EXT4_ERR_EFSBADCRC 4
1871 #define EXT4_ERR_EFSCORRUPTED 5
1872 #define EXT4_ERR_ENOSPC 6
1873 #define EXT4_ERR_ENOKEY 7
1874 #define EXT4_ERR_EROFS 8
1875 #define EXT4_ERR_EFBIG 9
1876 #define EXT4_ERR_EEXIST 10
1877 #define EXT4_ERR_ERANGE 11
1878 #define EXT4_ERR_EOVERFLOW 12
1879 #define EXT4_ERR_EBUSY 13
1880 #define EXT4_ERR_ENOTDIR 14
1881 #define EXT4_ERR_ENOTEMPTY 15
1882 #define EXT4_ERR_ESHUTDOWN 16
1883 #define EXT4_ERR_EFAULT 17
1884
1885 /*
1886 * Inode dynamic state flags
1887 */
1888 enum {
1889 EXT4_STATE_NEW, /* inode is newly created */
1890 EXT4_STATE_XATTR, /* has in-inode xattrs */
1891 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1892 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1893 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1894 EXT4_STATE_NEWENTRY, /* File just added to dir */
1895 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1896 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
1897 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */
1898 EXT4_STATE_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */
1899 EXT4_STATE_FC_COMMITTING, /* Fast commit ongoing */
1900 EXT4_STATE_ORPHAN_FILE, /* Inode orphaned in orphan file */
1901 };
1902
1903 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1904 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1905 { \
1906 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1907 } \
1908 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1909 { \
1910 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1911 } \
1912 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1913 { \
1914 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1915 }
1916
1917 /* Add these declarations here only so that these functions can be
1918 * found by name. Otherwise, they are very hard to locate. */
1919 static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1920 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1921 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1922 EXT4_INODE_BIT_FNS(flag, flags, 0)
1923
1924 /* Add these declarations here only so that these functions can be
1925 * found by name. Otherwise, they are very hard to locate. */
1926 static inline int ext4_test_inode_state(struct inode *inode, int bit);
1927 static inline void ext4_set_inode_state(struct inode *inode, int bit);
1928 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1929 #if (BITS_PER_LONG < 64)
1930 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1931
ext4_clear_state_flags(struct ext4_inode_info * ei)1932 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1933 {
1934 (ei)->i_state_flags = 0;
1935 }
1936 #else
1937 EXT4_INODE_BIT_FNS(state, flags, 32)
1938
ext4_clear_state_flags(struct ext4_inode_info * ei)1939 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1940 {
1941 /* We depend on the fact that callers will set i_flags */
1942 }
1943 #endif
1944 #else
1945 /* Assume that user mode programs are passing in an ext4fs superblock, not
1946 * a kernel struct super_block. This will allow us to call the feature-test
1947 * macros from user land. */
1948 #define EXT4_SB(sb) (sb)
1949 #endif
1950
ext4_verity_in_progress(struct inode * inode)1951 static inline bool ext4_verity_in_progress(struct inode *inode)
1952 {
1953 return IS_ENABLED(CONFIG_FS_VERITY) &&
1954 ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS);
1955 }
1956
1957 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1958
1959 /*
1960 * Codes for operating systems
1961 */
1962 #define EXT4_OS_LINUX 0
1963 #define EXT4_OS_HURD 1
1964 #define EXT4_OS_MASIX 2
1965 #define EXT4_OS_FREEBSD 3
1966 #define EXT4_OS_LITES 4
1967
1968 /*
1969 * Revision levels
1970 */
1971 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1972 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1973
1974 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1975
1976 #define EXT4_GOOD_OLD_INODE_SIZE 128
1977
1978 #define EXT4_EXTRA_TIMESTAMP_MAX (((s64)1 << 34) - 1 + S32_MIN)
1979 #define EXT4_NON_EXTRA_TIMESTAMP_MAX S32_MAX
1980 #define EXT4_TIMESTAMP_MIN S32_MIN
1981
1982 /*
1983 * Feature set definitions
1984 */
1985
1986 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1987 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1988 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1989 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1990 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1991 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1992 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
1993 /*
1994 * The reason why "FAST_COMMIT" is a compat feature is that, FS becomes
1995 * incompatible only if fast commit blocks are present in the FS. Since we
1996 * clear the journal (and thus the fast commit blocks), we don't mark FS as
1997 * incompatible. We also have a JBD2 incompat feature, which gets set when
1998 * there are fast commit blocks present in the journal.
1999 */
2000 #define EXT4_FEATURE_COMPAT_FAST_COMMIT 0x0400
2001 #define EXT4_FEATURE_COMPAT_STABLE_INODES 0x0800
2002 #define EXT4_FEATURE_COMPAT_ORPHAN_FILE 0x1000 /* Orphan file exists */
2003
2004 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
2005 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
2006 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
2007 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
2008 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
2009 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
2010 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
2011 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
2012 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
2013 /*
2014 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
2015 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
2016 * all other data structures' checksums. However, the METADATA_CSUM and
2017 * GDT_CSUM bits are mutually exclusive.
2018 */
2019 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
2020 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000
2021 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000
2022 #define EXT4_FEATURE_RO_COMPAT_VERITY 0x8000
2023 #define EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT 0x10000 /* Orphan file may be
2024 non-empty */
2025
2026 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
2027 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
2028 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
2029 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
2030 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
2031 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
2032 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
2033 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
2034 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
2035 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
2036 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
2037 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000
2038 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
2039 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
2040 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000
2041 #define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000
2042
2043 extern void ext4_update_dynamic_rev(struct super_block *sb);
2044
2045 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
2046 static inline bool ext4_has_feature_##name(struct super_block *sb) \
2047 { \
2048 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
2049 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
2050 } \
2051 static inline void ext4_set_feature_##name(struct super_block *sb) \
2052 { \
2053 ext4_update_dynamic_rev(sb); \
2054 EXT4_SB(sb)->s_es->s_feature_compat |= \
2055 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
2056 } \
2057 static inline void ext4_clear_feature_##name(struct super_block *sb) \
2058 { \
2059 EXT4_SB(sb)->s_es->s_feature_compat &= \
2060 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
2061 }
2062
2063 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
2064 static inline bool ext4_has_feature_##name(struct super_block *sb) \
2065 { \
2066 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
2067 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
2068 } \
2069 static inline void ext4_set_feature_##name(struct super_block *sb) \
2070 { \
2071 ext4_update_dynamic_rev(sb); \
2072 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
2073 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
2074 } \
2075 static inline void ext4_clear_feature_##name(struct super_block *sb) \
2076 { \
2077 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
2078 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
2079 }
2080
2081 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
2082 static inline bool ext4_has_feature_##name(struct super_block *sb) \
2083 { \
2084 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
2085 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
2086 } \
2087 static inline void ext4_set_feature_##name(struct super_block *sb) \
2088 { \
2089 ext4_update_dynamic_rev(sb); \
2090 EXT4_SB(sb)->s_es->s_feature_incompat |= \
2091 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
2092 } \
2093 static inline void ext4_clear_feature_##name(struct super_block *sb) \
2094 { \
2095 EXT4_SB(sb)->s_es->s_feature_incompat &= \
2096 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
2097 }
2098
EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc,DIR_PREALLOC)2099 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC)
2100 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES)
2101 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL)
2102 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR)
2103 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE)
2104 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX)
2105 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2)
2106 EXT4_FEATURE_COMPAT_FUNCS(fast_commit, FAST_COMMIT)
2107 EXT4_FEATURE_COMPAT_FUNCS(stable_inodes, STABLE_INODES)
2108 EXT4_FEATURE_COMPAT_FUNCS(orphan_file, ORPHAN_FILE)
2109
2110 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER)
2111 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE)
2112 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR)
2113 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE)
2114 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM)
2115 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK)
2116 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE)
2117 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA)
2118 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC)
2119 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM)
2120 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY)
2121 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT)
2122 EXT4_FEATURE_RO_COMPAT_FUNCS(verity, VERITY)
2123 EXT4_FEATURE_RO_COMPAT_FUNCS(orphan_present, ORPHAN_PRESENT)
2124
2125 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION)
2126 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE)
2127 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER)
2128 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV)
2129 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG)
2130 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS)
2131 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT)
2132 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP)
2133 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG)
2134 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE)
2135 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA)
2136 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED)
2137 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR)
2138 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA)
2139 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT)
2140 EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD)
2141
2142 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
2143 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
2144 EXT4_FEATURE_INCOMPAT_META_BG)
2145 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2146 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2147 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
2148
2149 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
2150 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
2151 EXT4_FEATURE_INCOMPAT_RECOVER| \
2152 EXT4_FEATURE_INCOMPAT_META_BG)
2153 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2154 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2155 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
2156
2157 #define EXT4_FEATURE_COMPAT_SUPP (EXT4_FEATURE_COMPAT_EXT_ATTR| \
2158 EXT4_FEATURE_COMPAT_ORPHAN_FILE)
2159 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
2160 EXT4_FEATURE_INCOMPAT_RECOVER| \
2161 EXT4_FEATURE_INCOMPAT_META_BG| \
2162 EXT4_FEATURE_INCOMPAT_EXTENTS| \
2163 EXT4_FEATURE_INCOMPAT_64BIT| \
2164 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
2165 EXT4_FEATURE_INCOMPAT_EA_INODE| \
2166 EXT4_FEATURE_INCOMPAT_MMP | \
2167 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
2168 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
2169 EXT4_FEATURE_INCOMPAT_CASEFOLD | \
2170 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \
2171 EXT4_FEATURE_INCOMPAT_LARGEDIR)
2172 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2173 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2174 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
2175 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
2176 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
2177 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
2178 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
2179 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
2180 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
2181 EXT4_FEATURE_RO_COMPAT_QUOTA |\
2182 EXT4_FEATURE_RO_COMPAT_PROJECT |\
2183 EXT4_FEATURE_RO_COMPAT_VERITY |\
2184 EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT)
2185
2186 #define EXTN_FEATURE_FUNCS(ver) \
2187 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
2188 { \
2189 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
2190 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
2191 } \
2192 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
2193 { \
2194 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
2195 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
2196 } \
2197 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
2198 { \
2199 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
2200 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
2201 }
2202
2203 EXTN_FEATURE_FUNCS(2)
2204 EXTN_FEATURE_FUNCS(3)
2205 EXTN_FEATURE_FUNCS(4)
2206
2207 static inline bool ext4_has_compat_features(struct super_block *sb)
2208 {
2209 return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
2210 }
ext4_has_ro_compat_features(struct super_block * sb)2211 static inline bool ext4_has_ro_compat_features(struct super_block *sb)
2212 {
2213 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
2214 }
ext4_has_incompat_features(struct super_block * sb)2215 static inline bool ext4_has_incompat_features(struct super_block *sb)
2216 {
2217 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
2218 }
2219
2220 extern int ext4_feature_set_ok(struct super_block *sb, int readonly);
2221
2222 /*
2223 * Superblock flags
2224 */
2225 #define EXT4_FLAGS_RESIZING 0
2226 #define EXT4_FLAGS_SHUTDOWN 1
2227 #define EXT4_FLAGS_BDEV_IS_DAX 2
2228
ext4_forced_shutdown(struct super_block * sb)2229 static inline int ext4_forced_shutdown(struct super_block *sb)
2230 {
2231 return test_bit(EXT4_FLAGS_SHUTDOWN, &EXT4_SB(sb)->s_ext4_flags);
2232 }
2233
2234 /*
2235 * Default values for user and/or group using reserved blocks
2236 */
2237 #define EXT4_DEF_RESUID 0
2238 #define EXT4_DEF_RESGID 0
2239
2240 /*
2241 * Default project ID
2242 */
2243 #define EXT4_DEF_PROJID 0
2244
2245 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
2246
2247 /*
2248 * Default mount options
2249 */
2250 #define EXT4_DEFM_DEBUG 0x0001
2251 #define EXT4_DEFM_BSDGROUPS 0x0002
2252 #define EXT4_DEFM_XATTR_USER 0x0004
2253 #define EXT4_DEFM_ACL 0x0008
2254 #define EXT4_DEFM_UID16 0x0010
2255 #define EXT4_DEFM_JMODE 0x0060
2256 #define EXT4_DEFM_JMODE_DATA 0x0020
2257 #define EXT4_DEFM_JMODE_ORDERED 0x0040
2258 #define EXT4_DEFM_JMODE_WBACK 0x0060
2259 #define EXT4_DEFM_NOBARRIER 0x0100
2260 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
2261 #define EXT4_DEFM_DISCARD 0x0400
2262 #define EXT4_DEFM_NODELALLOC 0x0800
2263
2264 /*
2265 * Default journal batch times
2266 */
2267 #define EXT4_DEF_MIN_BATCH_TIME 0
2268 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
2269
2270 /*
2271 * Minimum number of groups in a flexgroup before we separate out
2272 * directories into the first block group of a flexgroup
2273 */
2274 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
2275
2276 /*
2277 * Structure of a directory entry
2278 */
2279 #define EXT4_NAME_LEN 255
2280 /*
2281 * Base length of the ext4 directory entry excluding the name length
2282 */
2283 #define EXT4_BASE_DIR_LEN (sizeof(struct ext4_dir_entry_2) - EXT4_NAME_LEN)
2284
2285 struct ext4_dir_entry {
2286 __le32 inode; /* Inode number */
2287 __le16 rec_len; /* Directory entry length */
2288 __le16 name_len; /* Name length */
2289 char name[EXT4_NAME_LEN]; /* File name */
2290 };
2291
2292
2293 /*
2294 * Encrypted Casefolded entries require saving the hash on disk. This structure
2295 * followed ext4_dir_entry_2's name[name_len] at the next 4 byte aligned
2296 * boundary.
2297 */
2298 struct ext4_dir_entry_hash {
2299 __le32 hash;
2300 __le32 minor_hash;
2301 };
2302
2303 /*
2304 * The new version of the directory entry. Since EXT4 structures are
2305 * stored in intel byte order, and the name_len field could never be
2306 * bigger than 255 chars, it's safe to reclaim the extra byte for the
2307 * file_type field.
2308 */
2309 struct ext4_dir_entry_2 {
2310 __le32 inode; /* Inode number */
2311 __le16 rec_len; /* Directory entry length */
2312 __u8 name_len; /* Name length */
2313 __u8 file_type; /* See file type macros EXT4_FT_* below */
2314 char name[EXT4_NAME_LEN]; /* File name */
2315 };
2316
2317 /*
2318 * Access the hashes at the end of ext4_dir_entry_2
2319 */
2320 #define EXT4_DIRENT_HASHES(entry) \
2321 ((struct ext4_dir_entry_hash *) \
2322 (((void *)(entry)) + \
2323 ((8 + (entry)->name_len + EXT4_DIR_ROUND) & ~EXT4_DIR_ROUND)))
2324 #define EXT4_DIRENT_HASH(entry) le32_to_cpu(EXT4_DIRENT_HASHES(de)->hash)
2325 #define EXT4_DIRENT_MINOR_HASH(entry) \
2326 le32_to_cpu(EXT4_DIRENT_HASHES(de)->minor_hash)
2327
ext4_hash_in_dirent(const struct inode * inode)2328 static inline bool ext4_hash_in_dirent(const struct inode *inode)
2329 {
2330 return IS_CASEFOLDED(inode) && IS_ENCRYPTED(inode);
2331 }
2332
2333 /*
2334 * This is a bogus directory entry at the end of each leaf block that
2335 * records checksums.
2336 */
2337 struct ext4_dir_entry_tail {
2338 __le32 det_reserved_zero1; /* Pretend to be unused */
2339 __le16 det_rec_len; /* 12 */
2340 __u8 det_reserved_zero2; /* Zero name length */
2341 __u8 det_reserved_ft; /* 0xDE, fake file type */
2342 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
2343 };
2344
2345 #define EXT4_DIRENT_TAIL(block, blocksize) \
2346 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
2347 ((blocksize) - \
2348 sizeof(struct ext4_dir_entry_tail))))
2349
2350 /*
2351 * Ext4 directory file types. Only the low 3 bits are used. The
2352 * other bits are reserved for now.
2353 */
2354 #define EXT4_FT_UNKNOWN 0
2355 #define EXT4_FT_REG_FILE 1
2356 #define EXT4_FT_DIR 2
2357 #define EXT4_FT_CHRDEV 3
2358 #define EXT4_FT_BLKDEV 4
2359 #define EXT4_FT_FIFO 5
2360 #define EXT4_FT_SOCK 6
2361 #define EXT4_FT_SYMLINK 7
2362
2363 #define EXT4_FT_MAX 8
2364
2365 #define EXT4_FT_DIR_CSUM 0xDE
2366
2367 /*
2368 * EXT4_DIR_PAD defines the directory entries boundaries
2369 *
2370 * NOTE: It must be a multiple of 4
2371 */
2372 #define EXT4_DIR_PAD 4
2373 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
2374 #define EXT4_MAX_REC_LEN ((1<<16)-1)
2375
2376 /*
2377 * The rec_len is dependent on the type of directory. Directories that are
2378 * casefolded and encrypted need to store the hash as well, so we add room for
2379 * ext4_extended_dir_entry_2. For all entries related to '.' or '..' you should
2380 * pass NULL for dir, as those entries do not use the extra fields.
2381 */
ext4_dir_rec_len(__u8 name_len,const struct inode * dir)2382 static inline unsigned int ext4_dir_rec_len(__u8 name_len,
2383 const struct inode *dir)
2384 {
2385 int rec_len = (name_len + 8 + EXT4_DIR_ROUND);
2386
2387 if (dir && ext4_hash_in_dirent(dir))
2388 rec_len += sizeof(struct ext4_dir_entry_hash);
2389 return (rec_len & ~EXT4_DIR_ROUND);
2390 }
2391
2392 /*
2393 * If we ever get support for fs block sizes > page_size, we'll need
2394 * to remove the #if statements in the next two functions...
2395 */
2396 static inline unsigned int
ext4_rec_len_from_disk(__le16 dlen,unsigned blocksize)2397 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
2398 {
2399 unsigned len = le16_to_cpu(dlen);
2400
2401 #if (PAGE_SIZE >= 65536)
2402 if (len == EXT4_MAX_REC_LEN || len == 0)
2403 return blocksize;
2404 return (len & 65532) | ((len & 3) << 16);
2405 #else
2406 return len;
2407 #endif
2408 }
2409
ext4_rec_len_to_disk(unsigned len,unsigned blocksize)2410 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
2411 {
2412 BUG_ON((len > blocksize) || (blocksize > (1 << 18)) || (len & 3));
2413 #if (PAGE_SIZE >= 65536)
2414 if (len < 65536)
2415 return cpu_to_le16(len);
2416 if (len == blocksize) {
2417 if (blocksize == 65536)
2418 return cpu_to_le16(EXT4_MAX_REC_LEN);
2419 else
2420 return cpu_to_le16(0);
2421 }
2422 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
2423 #else
2424 return cpu_to_le16(len);
2425 #endif
2426 }
2427
2428 /*
2429 * Hash Tree Directory indexing
2430 * (c) Daniel Phillips, 2001
2431 */
2432
2433 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
2434 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2435 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \
2436 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir)))
2437 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2438
2439 /* Legal values for the dx_root hash_version field: */
2440
2441 #define DX_HASH_LEGACY 0
2442 #define DX_HASH_HALF_MD4 1
2443 #define DX_HASH_TEA 2
2444 #define DX_HASH_LEGACY_UNSIGNED 3
2445 #define DX_HASH_HALF_MD4_UNSIGNED 4
2446 #define DX_HASH_TEA_UNSIGNED 5
2447 #define DX_HASH_SIPHASH 6
2448
ext4_chksum(struct ext4_sb_info * sbi,u32 crc,const void * address,unsigned int length)2449 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2450 const void *address, unsigned int length)
2451 {
2452 struct {
2453 struct shash_desc shash;
2454 char ctx[4];
2455 } desc;
2456
2457 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2458
2459 desc.shash.tfm = sbi->s_chksum_driver;
2460 *(u32 *)desc.ctx = crc;
2461
2462 BUG_ON(crypto_shash_update(&desc.shash, address, length));
2463
2464 return *(u32 *)desc.ctx;
2465 }
2466
2467 #ifdef __KERNEL__
2468
2469 /* hash info structure used by the directory hash */
2470 struct dx_hash_info
2471 {
2472 u32 hash;
2473 u32 minor_hash;
2474 int hash_version;
2475 u32 *seed;
2476 };
2477
2478
2479 /* 32 and 64 bit signed EOF for dx directories */
2480 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
2481 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
2482
2483
2484 /*
2485 * Control parameters used by ext4_htree_next_block
2486 */
2487 #define HASH_NB_ALWAYS 1
2488
2489 struct ext4_filename {
2490 const struct qstr *usr_fname;
2491 struct fscrypt_str disk_name;
2492 struct dx_hash_info hinfo;
2493 #ifdef CONFIG_FS_ENCRYPTION
2494 struct fscrypt_str crypto_buf;
2495 #endif
2496 #if IS_ENABLED(CONFIG_UNICODE)
2497 struct fscrypt_str cf_name;
2498 #endif
2499 };
2500
2501 #define fname_name(p) ((p)->disk_name.name)
2502 #define fname_usr_name(p) ((p)->usr_fname->name)
2503 #define fname_len(p) ((p)->disk_name.len)
2504
2505 /*
2506 * Describe an inode's exact location on disk and in memory
2507 */
2508 struct ext4_iloc
2509 {
2510 struct buffer_head *bh;
2511 unsigned long offset;
2512 ext4_group_t block_group;
2513 };
2514
ext4_raw_inode(struct ext4_iloc * iloc)2515 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2516 {
2517 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2518 }
2519
ext4_is_quota_file(struct inode * inode)2520 static inline bool ext4_is_quota_file(struct inode *inode)
2521 {
2522 return IS_NOQUOTA(inode) &&
2523 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL);
2524 }
2525
2526 /*
2527 * This structure is stuffed into the struct file's private_data field
2528 * for directories. It is where we put information so that we can do
2529 * readdir operations in hash tree order.
2530 */
2531 struct dir_private_info {
2532 struct rb_root root;
2533 struct rb_node *curr_node;
2534 struct fname *extra_fname;
2535 loff_t last_pos;
2536 __u32 curr_hash;
2537 __u32 curr_minor_hash;
2538 __u32 next_hash;
2539 };
2540
2541 /* calculate the first block number of the group */
2542 static inline ext4_fsblk_t
ext4_group_first_block_no(struct super_block * sb,ext4_group_t group_no)2543 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2544 {
2545 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2546 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2547 }
2548
2549 /*
2550 * Special error return code only used by dx_probe() and its callers.
2551 */
2552 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
2553
2554 /* htree levels for ext4 */
2555 #define EXT4_HTREE_LEVEL_COMPAT 2
2556 #define EXT4_HTREE_LEVEL 3
2557
ext4_dir_htree_level(struct super_block * sb)2558 static inline int ext4_dir_htree_level(struct super_block *sb)
2559 {
2560 return ext4_has_feature_largedir(sb) ?
2561 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT;
2562 }
2563
2564 /*
2565 * Timeout and state flag for lazy initialization inode thread.
2566 */
2567 #define EXT4_DEF_LI_WAIT_MULT 10
2568 #define EXT4_DEF_LI_MAX_START_DELAY 5
2569 #define EXT4_LAZYINIT_QUIT 0x0001
2570 #define EXT4_LAZYINIT_RUNNING 0x0002
2571
2572 /*
2573 * Lazy inode table initialization info
2574 */
2575 struct ext4_lazy_init {
2576 unsigned long li_state;
2577 struct list_head li_request_list;
2578 struct mutex li_list_mtx;
2579 };
2580
2581 enum ext4_li_mode {
2582 EXT4_LI_MODE_PREFETCH_BBITMAP,
2583 EXT4_LI_MODE_ITABLE,
2584 };
2585
2586 struct ext4_li_request {
2587 struct super_block *lr_super;
2588 enum ext4_li_mode lr_mode;
2589 ext4_group_t lr_first_not_zeroed;
2590 ext4_group_t lr_next_group;
2591 struct list_head lr_request;
2592 unsigned long lr_next_sched;
2593 unsigned long lr_timeout;
2594 };
2595
2596 struct ext4_features {
2597 struct kobject f_kobj;
2598 struct completion f_kobj_unregister;
2599 };
2600
2601 /*
2602 * This structure will be used for multiple mount protection. It will be
2603 * written into the block number saved in the s_mmp_block field in the
2604 * superblock. Programs that check MMP should assume that if
2605 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2606 * to use the filesystem, regardless of how old the timestamp is.
2607 */
2608 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
2609 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2610 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
2611 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
2612
2613 struct mmp_struct {
2614 __le32 mmp_magic; /* Magic number for MMP */
2615 __le32 mmp_seq; /* Sequence no. updated periodically */
2616
2617 /*
2618 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2619 * purposes and do not affect the correctness of the algorithm
2620 */
2621 __le64 mmp_time; /* Time last updated */
2622 char mmp_nodename[64]; /* Node which last updated MMP block */
2623 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
2624
2625 /*
2626 * mmp_check_interval is used to verify if the MMP block has been
2627 * updated on the block device. The value is updated based on the
2628 * maximum time to write the MMP block during an update cycle.
2629 */
2630 __le16 mmp_check_interval;
2631
2632 __le16 mmp_pad1;
2633 __le32 mmp_pad2[226];
2634 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
2635 };
2636
2637 /* arguments passed to the mmp thread */
2638 struct mmpd_data {
2639 struct buffer_head *bh; /* bh from initial read_mmp_block() */
2640 struct super_block *sb; /* super block of the fs */
2641 };
2642
2643 /*
2644 * Check interval multiplier
2645 * The MMP block is written every update interval and initially checked every
2646 * update interval x the multiplier (the value is then adapted based on the
2647 * write latency). The reason is that writes can be delayed under load and we
2648 * don't want readers to incorrectly assume that the filesystem is no longer
2649 * in use.
2650 */
2651 #define EXT4_MMP_CHECK_MULT 2UL
2652
2653 /*
2654 * Minimum interval for MMP checking in seconds.
2655 */
2656 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
2657
2658 /*
2659 * Maximum interval for MMP checking in seconds.
2660 */
2661 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
2662
2663 /*
2664 * Function prototypes
2665 */
2666
2667 /*
2668 * Ok, these declarations are also in <linux/kernel.h> but none of the
2669 * ext4 source programs needs to include it so they are duplicated here.
2670 */
2671 # define NORET_TYPE /**/
2672 # define ATTRIB_NORET __attribute__((noreturn))
2673 # define NORET_AND noreturn,
2674
2675 /* bitmap.c */
2676 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2677 void ext4_inode_bitmap_csum_set(struct super_block *sb,
2678 struct ext4_group_desc *gdp,
2679 struct buffer_head *bh, int sz);
2680 int ext4_inode_bitmap_csum_verify(struct super_block *sb,
2681 struct ext4_group_desc *gdp,
2682 struct buffer_head *bh, int sz);
2683 void ext4_block_bitmap_csum_set(struct super_block *sb,
2684 struct ext4_group_desc *gdp,
2685 struct buffer_head *bh);
2686 int ext4_block_bitmap_csum_verify(struct super_block *sb,
2687 struct ext4_group_desc *gdp,
2688 struct buffer_head *bh);
2689
2690 /* balloc.c */
2691 extern void ext4_get_group_no_and_offset(struct super_block *sb,
2692 ext4_fsblk_t blocknr,
2693 ext4_group_t *blockgrpp,
2694 ext4_grpblk_t *offsetp);
2695 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2696 ext4_fsblk_t block);
2697
2698 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2699 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2700 ext4_group_t group);
2701 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2702 ext4_fsblk_t goal,
2703 unsigned int flags,
2704 unsigned long *count,
2705 int *errp);
2706 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2707 s64 nclusters, unsigned int flags);
2708 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2709 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2710 ext4_group_t block_group,
2711 struct buffer_head ** bh);
2712 extern struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2713 ext4_group_t group);
2714 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2715
2716 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2717 ext4_group_t block_group,
2718 bool ignore_locked);
2719 extern int ext4_wait_block_bitmap(struct super_block *sb,
2720 ext4_group_t block_group,
2721 struct buffer_head *bh);
2722 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2723 ext4_group_t block_group);
2724 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2725 ext4_group_t block_group,
2726 struct ext4_group_desc *gdp);
2727 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2728
2729 #if IS_ENABLED(CONFIG_UNICODE)
2730 extern int ext4_fname_setup_ci_filename(struct inode *dir,
2731 const struct qstr *iname,
2732 struct ext4_filename *fname);
2733 #endif
2734
2735 /* ext4 encryption related stuff goes here crypto.c */
2736 #ifdef CONFIG_FS_ENCRYPTION
2737 extern const struct fscrypt_operations ext4_cryptops;
2738
2739 int ext4_fname_setup_filename(struct inode *dir, const struct qstr *iname,
2740 int lookup, struct ext4_filename *fname);
2741
2742 int ext4_fname_prepare_lookup(struct inode *dir, struct dentry *dentry,
2743 struct ext4_filename *fname);
2744
2745 void ext4_fname_free_filename(struct ext4_filename *fname);
2746
2747 int ext4_ioctl_get_encryption_pwsalt(struct file *filp, void __user *arg);
2748
2749 #else /* !CONFIG_FS_ENCRYPTION */
ext4_fname_setup_filename(struct inode * dir,const struct qstr * iname,int lookup,struct ext4_filename * fname)2750 static inline int ext4_fname_setup_filename(struct inode *dir,
2751 const struct qstr *iname,
2752 int lookup,
2753 struct ext4_filename *fname)
2754 {
2755 int err = 0;
2756 fname->usr_fname = iname;
2757 fname->disk_name.name = (unsigned char *) iname->name;
2758 fname->disk_name.len = iname->len;
2759
2760 #if IS_ENABLED(CONFIG_UNICODE)
2761 err = ext4_fname_setup_ci_filename(dir, iname, fname);
2762 #endif
2763
2764 return err;
2765 }
2766
ext4_fname_prepare_lookup(struct inode * dir,struct dentry * dentry,struct ext4_filename * fname)2767 static inline int ext4_fname_prepare_lookup(struct inode *dir,
2768 struct dentry *dentry,
2769 struct ext4_filename *fname)
2770 {
2771 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname);
2772 }
2773
ext4_fname_free_filename(struct ext4_filename * fname)2774 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2775 {
2776 #if IS_ENABLED(CONFIG_UNICODE)
2777 kfree(fname->cf_name.name);
2778 fname->cf_name.name = NULL;
2779 #endif
2780 }
2781
ext4_ioctl_get_encryption_pwsalt(struct file * filp,void __user * arg)2782 static inline int ext4_ioctl_get_encryption_pwsalt(struct file *filp,
2783 void __user *arg)
2784 {
2785 return -EOPNOTSUPP;
2786 }
2787 #endif /* !CONFIG_FS_ENCRYPTION */
2788
2789 /* dir.c */
2790 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2791 struct file *,
2792 struct ext4_dir_entry_2 *,
2793 struct buffer_head *, char *, int,
2794 unsigned int);
2795 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2796 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2797 (de), (bh), (buf), (size), (offset)))
2798 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2799 __u32 minor_hash,
2800 struct ext4_dir_entry_2 *dirent,
2801 struct fscrypt_str *ent_name);
2802 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2803 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2804 struct buffer_head *bh,
2805 void *buf, int buf_size,
2806 struct ext4_filename *fname,
2807 struct ext4_dir_entry_2 **dest_de);
2808 void ext4_insert_dentry(struct inode *dir, struct inode *inode,
2809 struct ext4_dir_entry_2 *de,
2810 int buf_size,
2811 struct ext4_filename *fname);
ext4_update_dx_flag(struct inode * inode)2812 static inline void ext4_update_dx_flag(struct inode *inode)
2813 {
2814 if (!ext4_has_feature_dir_index(inode->i_sb) &&
2815 ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) {
2816 /* ext4_iget() should have caught this... */
2817 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb));
2818 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2819 }
2820 }
2821 static const unsigned char ext4_filetype_table[] = {
2822 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2823 };
2824
get_dtype(struct super_block * sb,int filetype)2825 static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2826 {
2827 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2828 return DT_UNKNOWN;
2829
2830 return ext4_filetype_table[filetype];
2831 }
2832 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2833 void *buf, int buf_size);
2834
2835 /* fsync.c */
2836 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2837
2838 /* hash.c */
2839 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len,
2840 struct dx_hash_info *hinfo);
2841
2842 /* ialloc.c */
2843 extern int ext4_mark_inode_used(struct super_block *sb, int ino);
2844 extern struct inode *__ext4_new_inode(struct mnt_idmap *, handle_t *,
2845 struct inode *, umode_t,
2846 const struct qstr *qstr, __u32 goal,
2847 uid_t *owner, __u32 i_flags,
2848 int handle_type, unsigned int line_no,
2849 int nblocks);
2850
2851 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \
2852 __ext4_new_inode(&nop_mnt_idmap, (handle), (dir), (mode), (qstr), \
2853 (goal), (owner), i_flags, 0, 0, 0)
2854 #define ext4_new_inode_start_handle(idmap, dir, mode, qstr, goal, owner, \
2855 type, nblocks) \
2856 __ext4_new_inode((idmap), NULL, (dir), (mode), (qstr), (goal), (owner), \
2857 0, (type), __LINE__, (nblocks))
2858
2859
2860 extern void ext4_free_inode(handle_t *, struct inode *);
2861 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2862 extern unsigned long ext4_count_free_inodes(struct super_block *);
2863 extern unsigned long ext4_count_dirs(struct super_block *);
2864 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2865 extern int ext4_init_inode_table(struct super_block *sb,
2866 ext4_group_t group, int barrier);
2867 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2868
2869 /* fast_commit.c */
2870 int ext4_fc_info_show(struct seq_file *seq, void *v);
2871 void ext4_fc_init(struct super_block *sb, journal_t *journal);
2872 void ext4_fc_init_inode(struct inode *inode);
2873 void ext4_fc_track_range(handle_t *handle, struct inode *inode, ext4_lblk_t start,
2874 ext4_lblk_t end);
2875 void __ext4_fc_track_unlink(handle_t *handle, struct inode *inode,
2876 struct dentry *dentry);
2877 void __ext4_fc_track_link(handle_t *handle, struct inode *inode,
2878 struct dentry *dentry);
2879 void ext4_fc_track_unlink(handle_t *handle, struct dentry *dentry);
2880 void ext4_fc_track_link(handle_t *handle, struct dentry *dentry);
2881 void __ext4_fc_track_create(handle_t *handle, struct inode *inode,
2882 struct dentry *dentry);
2883 void ext4_fc_track_create(handle_t *handle, struct dentry *dentry);
2884 void ext4_fc_track_inode(handle_t *handle, struct inode *inode);
2885 void ext4_fc_mark_ineligible(struct super_block *sb, int reason, handle_t *handle);
2886 void ext4_fc_start_update(struct inode *inode);
2887 void ext4_fc_stop_update(struct inode *inode);
2888 void ext4_fc_del(struct inode *inode);
2889 bool ext4_fc_replay_check_excluded(struct super_block *sb, ext4_fsblk_t block);
2890 void ext4_fc_replay_cleanup(struct super_block *sb);
2891 int ext4_fc_commit(journal_t *journal, tid_t commit_tid);
2892 int __init ext4_fc_init_dentry_cache(void);
2893 void ext4_fc_destroy_dentry_cache(void);
2894 int ext4_fc_record_regions(struct super_block *sb, int ino,
2895 ext4_lblk_t lblk, ext4_fsblk_t pblk,
2896 int len, int replay);
2897
2898 /* mballoc.c */
2899 extern const struct seq_operations ext4_mb_seq_groups_ops;
2900 extern const struct seq_operations ext4_mb_seq_structs_summary_ops;
2901 extern int ext4_seq_mb_stats_show(struct seq_file *seq, void *offset);
2902 extern int ext4_mb_init(struct super_block *);
2903 extern int ext4_mb_release(struct super_block *);
2904 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2905 struct ext4_allocation_request *, int *);
2906 extern void ext4_discard_preallocations(struct inode *, unsigned int);
2907 extern int __init ext4_init_mballoc(void);
2908 extern void ext4_exit_mballoc(void);
2909 extern ext4_group_t ext4_mb_prefetch(struct super_block *sb,
2910 ext4_group_t group,
2911 unsigned int nr, int *cnt);
2912 extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group,
2913 unsigned int nr);
2914
2915 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2916 struct buffer_head *bh, ext4_fsblk_t block,
2917 unsigned long count, int flags);
2918 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2919 ext4_group_t ngroups);
2920 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2921 ext4_group_t i, struct ext4_group_desc *desc);
2922 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2923 ext4_fsblk_t block, unsigned long count);
2924 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2925 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid);
2926 extern void ext4_mb_mark_bb(struct super_block *sb, ext4_fsblk_t block,
2927 int len, int state);
ext4_mb_cr_expensive(enum criteria cr)2928 static inline bool ext4_mb_cr_expensive(enum criteria cr)
2929 {
2930 return cr >= CR_GOAL_LEN_SLOW;
2931 }
2932
2933 /* inode.c */
2934 void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw,
2935 struct ext4_inode_info *ei);
2936 int ext4_inode_is_fast_symlink(struct inode *inode);
2937 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2938 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2939 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count,
2940 bool wait, struct buffer_head **bhs);
2941 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2942 struct buffer_head *bh_result, int create);
2943 int ext4_get_block(struct inode *inode, sector_t iblock,
2944 struct buffer_head *bh_result, int create);
2945 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2946 struct buffer_head *bh, int create);
2947 int ext4_walk_page_buffers(handle_t *handle,
2948 struct inode *inode,
2949 struct buffer_head *head,
2950 unsigned from,
2951 unsigned to,
2952 int *partial,
2953 int (*fn)(handle_t *handle, struct inode *inode,
2954 struct buffer_head *bh));
2955 int do_journal_get_write_access(handle_t *handle, struct inode *inode,
2956 struct buffer_head *bh);
2957 #define FALL_BACK_TO_NONDELALLOC 1
2958 #define CONVERT_INLINE_DATA 2
2959
2960 typedef enum {
2961 EXT4_IGET_NORMAL = 0,
2962 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */
2963 EXT4_IGET_HANDLE = 0x0002, /* Inode # is from a handle */
2964 EXT4_IGET_BAD = 0x0004, /* Allow to iget a bad inode */
2965 EXT4_IGET_EA_INODE = 0x0008 /* Inode should contain an EA value */
2966 } ext4_iget_flags;
2967
2968 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
2969 ext4_iget_flags flags, const char *function,
2970 unsigned int line);
2971
2972 #define ext4_iget(sb, ino, flags) \
2973 __ext4_iget((sb), (ino), (flags), __func__, __LINE__)
2974
2975 extern int ext4_write_inode(struct inode *, struct writeback_control *);
2976 extern int ext4_setattr(struct mnt_idmap *, struct dentry *,
2977 struct iattr *);
2978 extern u32 ext4_dio_alignment(struct inode *inode);
2979 extern int ext4_getattr(struct mnt_idmap *, const struct path *,
2980 struct kstat *, u32, unsigned int);
2981 extern void ext4_evict_inode(struct inode *);
2982 extern void ext4_clear_inode(struct inode *);
2983 extern int ext4_file_getattr(struct mnt_idmap *, const struct path *,
2984 struct kstat *, u32, unsigned int);
2985 extern void ext4_dirty_inode(struct inode *, int);
2986 extern int ext4_change_inode_journal_flag(struct inode *, int);
2987 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2988 extern int ext4_get_fc_inode_loc(struct super_block *sb, unsigned long ino,
2989 struct ext4_iloc *iloc);
2990 extern int ext4_inode_attach_jinode(struct inode *inode);
2991 extern int ext4_can_truncate(struct inode *inode);
2992 extern int ext4_truncate(struct inode *);
2993 extern int ext4_break_layouts(struct inode *);
2994 extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length);
2995 extern void ext4_set_inode_flags(struct inode *, bool init);
2996 extern int ext4_alloc_da_blocks(struct inode *inode);
2997 extern void ext4_set_aops(struct inode *inode);
2998 extern int ext4_writepage_trans_blocks(struct inode *);
2999 extern int ext4_normal_submit_inode_data_buffers(struct jbd2_inode *jinode);
3000 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
3001 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
3002 loff_t lstart, loff_t lend);
3003 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf);
3004 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
3005 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
3006 extern void ext4_da_release_space(struct inode *inode, int to_free);
3007 extern void ext4_da_update_reserve_space(struct inode *inode,
3008 int used, int quota_claim);
3009 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
3010 ext4_fsblk_t pblk, ext4_lblk_t len);
3011
3012 /* indirect.c */
3013 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
3014 struct ext4_map_blocks *map, int flags);
3015 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
3016 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
3017 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
3018 ext4_lblk_t start, ext4_lblk_t end);
3019
3020 /* ioctl.c */
3021 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
3022 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
3023 int ext4_fileattr_set(struct mnt_idmap *idmap,
3024 struct dentry *dentry, struct fileattr *fa);
3025 int ext4_fileattr_get(struct dentry *dentry, struct fileattr *fa);
3026 extern void ext4_reset_inode_seed(struct inode *inode);
3027 int ext4_update_overhead(struct super_block *sb, bool force);
3028 int ext4_force_shutdown(struct super_block *sb, u32 flags);
3029
3030 /* migrate.c */
3031 extern int ext4_ext_migrate(struct inode *);
3032 extern int ext4_ind_migrate(struct inode *inode);
3033
3034 /* namei.c */
3035 extern int ext4_init_new_dir(handle_t *handle, struct inode *dir,
3036 struct inode *inode);
3037 extern int ext4_dirblock_csum_verify(struct inode *inode,
3038 struct buffer_head *bh);
3039 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
3040 __u32 start_minor_hash, __u32 *next_hash);
3041 extern int ext4_search_dir(struct buffer_head *bh,
3042 char *search_buf,
3043 int buf_size,
3044 struct inode *dir,
3045 struct ext4_filename *fname,
3046 unsigned int offset,
3047 struct ext4_dir_entry_2 **res_dir);
3048 extern int ext4_generic_delete_entry(struct inode *dir,
3049 struct ext4_dir_entry_2 *de_del,
3050 struct buffer_head *bh,
3051 void *entry_buf,
3052 int buf_size,
3053 int csum_size);
3054 extern bool ext4_empty_dir(struct inode *inode);
3055
3056 /* resize.c */
3057 extern void ext4_kvfree_array_rcu(void *to_free);
3058 extern int ext4_group_add(struct super_block *sb,
3059 struct ext4_new_group_data *input);
3060 extern int ext4_group_extend(struct super_block *sb,
3061 struct ext4_super_block *es,
3062 ext4_fsblk_t n_blocks_count);
3063 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
3064 extern unsigned int ext4_list_backups(struct super_block *sb,
3065 unsigned int *three, unsigned int *five,
3066 unsigned int *seven);
3067
3068 /* super.c */
3069 extern struct buffer_head *ext4_sb_bread(struct super_block *sb,
3070 sector_t block, blk_opf_t op_flags);
3071 extern struct buffer_head *ext4_sb_bread_unmovable(struct super_block *sb,
3072 sector_t block);
3073 extern void ext4_read_bh_nowait(struct buffer_head *bh, blk_opf_t op_flags,
3074 bh_end_io_t *end_io);
3075 extern int ext4_read_bh(struct buffer_head *bh, blk_opf_t op_flags,
3076 bh_end_io_t *end_io);
3077 extern int ext4_read_bh_lock(struct buffer_head *bh, blk_opf_t op_flags, bool wait);
3078 extern void ext4_sb_breadahead_unmovable(struct super_block *sb, sector_t block);
3079 extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
3080 extern int ext4_calculate_overhead(struct super_block *sb);
3081 extern __le32 ext4_superblock_csum(struct super_block *sb,
3082 struct ext4_super_block *es);
3083 extern void ext4_superblock_csum_set(struct super_block *sb);
3084 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
3085 ext4_group_t ngroup);
3086 extern const char *ext4_decode_error(struct super_block *sb, int errno,
3087 char nbuf[16]);
3088 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb,
3089 ext4_group_t block_group,
3090 unsigned int flags);
3091 extern unsigned int ext4_num_base_meta_blocks(struct super_block *sb,
3092 ext4_group_t block_group);
3093
3094 extern __printf(7, 8)
3095 void __ext4_error(struct super_block *, const char *, unsigned int, bool,
3096 int, __u64, const char *, ...);
3097 extern __printf(6, 7)
3098 void __ext4_error_inode(struct inode *, const char *, unsigned int,
3099 ext4_fsblk_t, int, const char *, ...);
3100 extern __printf(5, 6)
3101 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
3102 const char *, ...);
3103 extern void __ext4_std_error(struct super_block *, const char *,
3104 unsigned int, int);
3105 extern __printf(4, 5)
3106 void __ext4_warning(struct super_block *, const char *, unsigned int,
3107 const char *, ...);
3108 extern __printf(4, 5)
3109 void __ext4_warning_inode(const struct inode *inode, const char *function,
3110 unsigned int line, const char *fmt, ...);
3111 extern __printf(3, 4)
3112 void __ext4_msg(struct super_block *, const char *, const char *, ...);
3113 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
3114 const char *, unsigned int, const char *);
3115 extern __printf(7, 8)
3116 void __ext4_grp_locked_error(const char *, unsigned int,
3117 struct super_block *, ext4_group_t,
3118 unsigned long, ext4_fsblk_t,
3119 const char *, ...);
3120
3121 #define EXT4_ERROR_INODE(inode, fmt, a...) \
3122 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
3123
3124 #define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...) \
3125 __ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a)
3126
3127 #define ext4_error_inode_block(inode, block, err, fmt, a...) \
3128 __ext4_error_inode((inode), __func__, __LINE__, (block), (err), \
3129 (fmt), ## a)
3130
3131 #define EXT4_ERROR_FILE(file, block, fmt, a...) \
3132 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
3133
3134 #define ext4_abort(sb, err, fmt, a...) \
3135 __ext4_error((sb), __func__, __LINE__, true, (err), 0, (fmt), ## a)
3136
3137 #ifdef CONFIG_PRINTK
3138
3139 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
3140 __ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__)
3141 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \
3142 __ext4_error_inode((inode), (func), (line), (block), \
3143 (err), (fmt), ##__VA_ARGS__)
3144 #define ext4_error_file(file, func, line, block, fmt, ...) \
3145 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
3146 #define ext4_error(sb, fmt, ...) \
3147 __ext4_error((sb), __func__, __LINE__, false, 0, 0, (fmt), \
3148 ##__VA_ARGS__)
3149 #define ext4_error_err(sb, err, fmt, ...) \
3150 __ext4_error((sb), __func__, __LINE__, false, (err), 0, (fmt), \
3151 ##__VA_ARGS__)
3152 #define ext4_warning(sb, fmt, ...) \
3153 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
3154 #define ext4_warning_inode(inode, fmt, ...) \
3155 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
3156 #define ext4_msg(sb, level, fmt, ...) \
3157 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
3158 #define dump_mmp_msg(sb, mmp, msg) \
3159 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
3160 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
3161 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
3162 fmt, ##__VA_ARGS__)
3163
3164 #else
3165
3166 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
3167 do { \
3168 no_printk(fmt, ##__VA_ARGS__); \
3169 __ext4_error_inode(inode, "", 0, block, 0, " "); \
3170 } while (0)
3171 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \
3172 do { \
3173 no_printk(fmt, ##__VA_ARGS__); \
3174 __ext4_error_inode(inode, "", 0, block, err, " "); \
3175 } while (0)
3176 #define ext4_error_file(file, func, line, block, fmt, ...) \
3177 do { \
3178 no_printk(fmt, ##__VA_ARGS__); \
3179 __ext4_error_file(file, "", 0, block, " "); \
3180 } while (0)
3181 #define ext4_error(sb, fmt, ...) \
3182 do { \
3183 no_printk(fmt, ##__VA_ARGS__); \
3184 __ext4_error(sb, "", 0, false, 0, 0, " "); \
3185 } while (0)
3186 #define ext4_error_err(sb, err, fmt, ...) \
3187 do { \
3188 no_printk(fmt, ##__VA_ARGS__); \
3189 __ext4_error(sb, "", 0, false, err, 0, " "); \
3190 } while (0)
3191 #define ext4_warning(sb, fmt, ...) \
3192 do { \
3193 no_printk(fmt, ##__VA_ARGS__); \
3194 __ext4_warning(sb, "", 0, " "); \
3195 } while (0)
3196 #define ext4_warning_inode(inode, fmt, ...) \
3197 do { \
3198 no_printk(fmt, ##__VA_ARGS__); \
3199 __ext4_warning_inode(inode, "", 0, " "); \
3200 } while (0)
3201 #define ext4_msg(sb, level, fmt, ...) \
3202 do { \
3203 no_printk(fmt, ##__VA_ARGS__); \
3204 __ext4_msg(sb, "", " "); \
3205 } while (0)
3206 #define dump_mmp_msg(sb, mmp, msg) \
3207 __dump_mmp_msg(sb, mmp, "", 0, "")
3208 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
3209 do { \
3210 no_printk(fmt, ##__VA_ARGS__); \
3211 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
3212 } while (0)
3213
3214 #endif
3215
3216 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
3217 struct ext4_group_desc *bg);
3218 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
3219 struct ext4_group_desc *bg);
3220 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
3221 struct ext4_group_desc *bg);
3222 extern __u32 ext4_free_group_clusters(struct super_block *sb,
3223 struct ext4_group_desc *bg);
3224 extern __u32 ext4_free_inodes_count(struct super_block *sb,
3225 struct ext4_group_desc *bg);
3226 extern __u32 ext4_used_dirs_count(struct super_block *sb,
3227 struct ext4_group_desc *bg);
3228 extern __u32 ext4_itable_unused_count(struct super_block *sb,
3229 struct ext4_group_desc *bg);
3230 extern void ext4_block_bitmap_set(struct super_block *sb,
3231 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3232 extern void ext4_inode_bitmap_set(struct super_block *sb,
3233 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3234 extern void ext4_inode_table_set(struct super_block *sb,
3235 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3236 extern void ext4_free_group_clusters_set(struct super_block *sb,
3237 struct ext4_group_desc *bg,
3238 __u32 count);
3239 extern void ext4_free_inodes_set(struct super_block *sb,
3240 struct ext4_group_desc *bg, __u32 count);
3241 extern void ext4_used_dirs_set(struct super_block *sb,
3242 struct ext4_group_desc *bg, __u32 count);
3243 extern void ext4_itable_unused_set(struct super_block *sb,
3244 struct ext4_group_desc *bg, __u32 count);
3245 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
3246 struct ext4_group_desc *gdp);
3247 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
3248 struct ext4_group_desc *gdp);
3249 extern int ext4_register_li_request(struct super_block *sb,
3250 ext4_group_t first_not_zeroed);
3251
ext4_has_metadata_csum(struct super_block * sb)3252 static inline int ext4_has_metadata_csum(struct super_block *sb)
3253 {
3254 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
3255 !EXT4_SB(sb)->s_chksum_driver);
3256
3257 return ext4_has_feature_metadata_csum(sb) &&
3258 (EXT4_SB(sb)->s_chksum_driver != NULL);
3259 }
3260
ext4_has_group_desc_csum(struct super_block * sb)3261 static inline int ext4_has_group_desc_csum(struct super_block *sb)
3262 {
3263 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb);
3264 }
3265
3266 #define ext4_read_incompat_64bit_val(es, name) \
3267 (((es)->s_feature_incompat & cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT) \
3268 ? (ext4_fsblk_t)le32_to_cpu(es->name##_hi) << 32 : 0) | \
3269 le32_to_cpu(es->name##_lo))
3270
ext4_blocks_count(struct ext4_super_block * es)3271 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
3272 {
3273 return ext4_read_incompat_64bit_val(es, s_blocks_count);
3274 }
3275
ext4_r_blocks_count(struct ext4_super_block * es)3276 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
3277 {
3278 return ext4_read_incompat_64bit_val(es, s_r_blocks_count);
3279 }
3280
ext4_free_blocks_count(struct ext4_super_block * es)3281 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
3282 {
3283 return ext4_read_incompat_64bit_val(es, s_free_blocks_count);
3284 }
3285
ext4_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)3286 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
3287 ext4_fsblk_t blk)
3288 {
3289 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
3290 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
3291 }
3292
ext4_free_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)3293 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
3294 ext4_fsblk_t blk)
3295 {
3296 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
3297 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
3298 }
3299
ext4_r_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)3300 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
3301 ext4_fsblk_t blk)
3302 {
3303 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
3304 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
3305 }
3306
ext4_isize(struct super_block * sb,struct ext4_inode * raw_inode)3307 static inline loff_t ext4_isize(struct super_block *sb,
3308 struct ext4_inode *raw_inode)
3309 {
3310 if (ext4_has_feature_largedir(sb) ||
3311 S_ISREG(le16_to_cpu(raw_inode->i_mode)))
3312 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
3313 le32_to_cpu(raw_inode->i_size_lo);
3314
3315 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
3316 }
3317
ext4_isize_set(struct ext4_inode * raw_inode,loff_t i_size)3318 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
3319 {
3320 raw_inode->i_size_lo = cpu_to_le32(i_size);
3321 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
3322 }
3323
3324 /*
3325 * Reading s_groups_count requires using smp_rmb() afterwards. See
3326 * the locking protocol documented in the comments of ext4_group_add()
3327 * in resize.c
3328 */
ext4_get_groups_count(struct super_block * sb)3329 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
3330 {
3331 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
3332
3333 smp_rmb();
3334 return ngroups;
3335 }
3336
ext4_flex_group(struct ext4_sb_info * sbi,ext4_group_t block_group)3337 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
3338 ext4_group_t block_group)
3339 {
3340 return block_group >> sbi->s_log_groups_per_flex;
3341 }
3342
ext4_flex_bg_size(struct ext4_sb_info * sbi)3343 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
3344 {
3345 return 1 << sbi->s_log_groups_per_flex;
3346 }
3347
3348 #define ext4_std_error(sb, errno) \
3349 do { \
3350 if ((errno)) \
3351 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
3352 } while (0)
3353
3354 #ifdef CONFIG_SMP
3355 /* Each CPU can accumulate percpu_counter_batch clusters in their local
3356 * counters. So we need to make sure we have free clusters more
3357 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
3358 */
3359 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
3360 #else
3361 #define EXT4_FREECLUSTERS_WATERMARK 0
3362 #endif
3363
3364 /* Update i_disksize. Requires i_rwsem to avoid races with truncate */
ext4_update_i_disksize(struct inode * inode,loff_t newsize)3365 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
3366 {
3367 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
3368 !inode_is_locked(inode));
3369 down_write(&EXT4_I(inode)->i_data_sem);
3370 if (newsize > EXT4_I(inode)->i_disksize)
3371 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize);
3372 up_write(&EXT4_I(inode)->i_data_sem);
3373 }
3374
3375 /* Update i_size, i_disksize. Requires i_rwsem to avoid races with truncate */
ext4_update_inode_size(struct inode * inode,loff_t newsize)3376 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
3377 {
3378 int changed = 0;
3379
3380 if (newsize > inode->i_size) {
3381 i_size_write(inode, newsize);
3382 changed = 1;
3383 }
3384 if (newsize > EXT4_I(inode)->i_disksize) {
3385 ext4_update_i_disksize(inode, newsize);
3386 changed |= 2;
3387 }
3388 return changed;
3389 }
3390
3391 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
3392 loff_t len);
3393
3394 struct ext4_group_info {
3395 unsigned long bb_state;
3396 #ifdef AGGRESSIVE_CHECK
3397 unsigned long bb_check_counter;
3398 #endif
3399 struct rb_root bb_free_root;
3400 ext4_grpblk_t bb_first_free; /* first free block */
3401 ext4_grpblk_t bb_free; /* total free blocks */
3402 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
3403 int bb_avg_fragment_size_order; /* order of average
3404 fragment in BG */
3405 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
3406 ext4_group_t bb_group; /* Group number */
3407 struct list_head bb_prealloc_list;
3408 #ifdef DOUBLE_CHECK
3409 void *bb_bitmap;
3410 #endif
3411 struct rw_semaphore alloc_sem;
3412 struct list_head bb_avg_fragment_size_node;
3413 struct list_head bb_largest_free_order_node;
3414 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
3415 * regions, index is order.
3416 * bb_counters[3] = 5 means
3417 * 5 free 8-block regions. */
3418 };
3419
3420 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
3421 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
3422 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
3423 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
3424 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \
3425 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT)
3426 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \
3427 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT)
3428 #define EXT4_GROUP_INFO_BBITMAP_READ_BIT 4
3429
3430 #define EXT4_MB_GRP_NEED_INIT(grp) \
3431 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
3432 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
3433 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3434 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
3435 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3436
3437 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
3438 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3439 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
3440 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3441 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
3442 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3443 #define EXT4_MB_GRP_TEST_AND_SET_READ(grp) \
3444 (test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state)))
3445
3446 #define EXT4_MAX_CONTENTION 8
3447 #define EXT4_CONTENTION_THRESHOLD 2
3448
ext4_group_lock_ptr(struct super_block * sb,ext4_group_t group)3449 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
3450 ext4_group_t group)
3451 {
3452 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
3453 }
3454
3455 /*
3456 * Returns true if the filesystem is busy enough that attempts to
3457 * access the block group locks has run into contention.
3458 */
ext4_fs_is_busy(struct ext4_sb_info * sbi)3459 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
3460 {
3461 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
3462 }
3463
ext4_lock_group(struct super_block * sb,ext4_group_t group)3464 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
3465 {
3466 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
3467 if (spin_trylock(lock))
3468 /*
3469 * We're able to grab the lock right away, so drop the
3470 * lock contention counter.
3471 */
3472 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
3473 else {
3474 /*
3475 * The lock is busy, so bump the contention counter,
3476 * and then wait on the spin lock.
3477 */
3478 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
3479 EXT4_MAX_CONTENTION);
3480 spin_lock(lock);
3481 }
3482 }
3483
ext4_unlock_group(struct super_block * sb,ext4_group_t group)3484 static inline void ext4_unlock_group(struct super_block *sb,
3485 ext4_group_t group)
3486 {
3487 spin_unlock(ext4_group_lock_ptr(sb, group));
3488 }
3489
3490 #ifdef CONFIG_QUOTA
ext4_quota_capable(struct super_block * sb)3491 static inline bool ext4_quota_capable(struct super_block *sb)
3492 {
3493 return (test_opt(sb, QUOTA) || ext4_has_feature_quota(sb));
3494 }
3495
ext4_is_quota_journalled(struct super_block * sb)3496 static inline bool ext4_is_quota_journalled(struct super_block *sb)
3497 {
3498 struct ext4_sb_info *sbi = EXT4_SB(sb);
3499
3500 return (ext4_has_feature_quota(sb) ||
3501 sbi->s_qf_names[USRQUOTA] || sbi->s_qf_names[GRPQUOTA]);
3502 }
3503 int ext4_enable_quotas(struct super_block *sb);
3504 #endif
3505
3506 /*
3507 * Block validity checking
3508 */
3509 #define ext4_check_indirect_blockref(inode, bh) \
3510 ext4_check_blockref(__func__, __LINE__, inode, \
3511 (__le32 *)(bh)->b_data, \
3512 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
3513
3514 #define ext4_ind_check_inode(inode) \
3515 ext4_check_blockref(__func__, __LINE__, inode, \
3516 EXT4_I(inode)->i_data, \
3517 EXT4_NDIR_BLOCKS)
3518
3519 /*
3520 * Inodes and files operations
3521 */
3522
3523 /* dir.c */
3524 extern const struct file_operations ext4_dir_operations;
3525
3526 /* file.c */
3527 extern const struct inode_operations ext4_file_inode_operations;
3528 extern const struct file_operations ext4_file_operations;
3529 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
3530
3531 /* inline.c */
3532 extern int ext4_get_max_inline_size(struct inode *inode);
3533 extern int ext4_find_inline_data_nolock(struct inode *inode);
3534 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
3535
3536 int ext4_readpage_inline(struct inode *inode, struct folio *folio);
3537 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3538 struct inode *inode,
3539 loff_t pos, unsigned len,
3540 struct page **pagep);
3541 int ext4_write_inline_data_end(struct inode *inode, loff_t pos, unsigned len,
3542 unsigned copied, struct folio *folio);
3543 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
3544 struct inode *inode,
3545 loff_t pos, unsigned len,
3546 struct page **pagep,
3547 void **fsdata);
3548 extern int ext4_try_add_inline_entry(handle_t *handle,
3549 struct ext4_filename *fname,
3550 struct inode *dir, struct inode *inode);
3551 extern int ext4_try_create_inline_dir(handle_t *handle,
3552 struct inode *parent,
3553 struct inode *inode);
3554 extern int ext4_read_inline_dir(struct file *filp,
3555 struct dir_context *ctx,
3556 int *has_inline_data);
3557 extern int ext4_inlinedir_to_tree(struct file *dir_file,
3558 struct inode *dir, ext4_lblk_t block,
3559 struct dx_hash_info *hinfo,
3560 __u32 start_hash, __u32 start_minor_hash,
3561 int *has_inline_data);
3562 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3563 struct ext4_filename *fname,
3564 struct ext4_dir_entry_2 **res_dir,
3565 int *has_inline_data);
3566 extern int ext4_delete_inline_entry(handle_t *handle,
3567 struct inode *dir,
3568 struct ext4_dir_entry_2 *de_del,
3569 struct buffer_head *bh,
3570 int *has_inline_data);
3571 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3572 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3573 struct ext4_dir_entry_2 **parent_de,
3574 int *retval);
3575 extern void *ext4_read_inline_link(struct inode *inode);
3576
3577 struct iomap;
3578 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap);
3579
3580 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3581
3582 extern int ext4_convert_inline_data(struct inode *inode);
3583
ext4_has_inline_data(struct inode * inode)3584 static inline int ext4_has_inline_data(struct inode *inode)
3585 {
3586 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3587 EXT4_I(inode)->i_inline_off;
3588 }
3589
3590 /* namei.c */
3591 extern const struct inode_operations ext4_dir_inode_operations;
3592 extern const struct inode_operations ext4_special_inode_operations;
3593 extern struct dentry *ext4_get_parent(struct dentry *child);
3594 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3595 struct ext4_dir_entry_2 *de,
3596 int blocksize, int csum_size,
3597 unsigned int parent_ino, int dotdot_real_len);
3598 extern void ext4_initialize_dirent_tail(struct buffer_head *bh,
3599 unsigned int blocksize);
3600 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode,
3601 struct buffer_head *bh);
3602 extern int __ext4_unlink(struct inode *dir, const struct qstr *d_name,
3603 struct inode *inode, struct dentry *dentry);
3604 extern int __ext4_link(struct inode *dir, struct inode *inode,
3605 struct dentry *dentry);
3606
3607 #define S_SHIFT 12
3608 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = {
3609 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
3610 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
3611 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
3612 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
3613 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
3614 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
3615 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
3616 };
3617
ext4_set_de_type(struct super_block * sb,struct ext4_dir_entry_2 * de,umode_t mode)3618 static inline void ext4_set_de_type(struct super_block *sb,
3619 struct ext4_dir_entry_2 *de,
3620 umode_t mode) {
3621 if (ext4_has_feature_filetype(sb))
3622 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3623 }
3624
3625 /* readpages.c */
3626 extern int ext4_mpage_readpages(struct inode *inode,
3627 struct readahead_control *rac, struct folio *folio);
3628 extern int __init ext4_init_post_read_processing(void);
3629 extern void ext4_exit_post_read_processing(void);
3630
3631 /* symlink.c */
3632 extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3633 extern const struct inode_operations ext4_symlink_inode_operations;
3634 extern const struct inode_operations ext4_fast_symlink_inode_operations;
3635
3636 /* sysfs.c */
3637 extern void ext4_notify_error_sysfs(struct ext4_sb_info *sbi);
3638 extern int ext4_register_sysfs(struct super_block *sb);
3639 extern void ext4_unregister_sysfs(struct super_block *sb);
3640 extern int __init ext4_init_sysfs(void);
3641 extern void ext4_exit_sysfs(void);
3642
3643 /* block_validity */
3644 extern void ext4_release_system_zone(struct super_block *sb);
3645 extern int ext4_setup_system_zone(struct super_block *sb);
3646 extern int __init ext4_init_system_zone(void);
3647 extern void ext4_exit_system_zone(void);
3648 extern int ext4_inode_block_valid(struct inode *inode,
3649 ext4_fsblk_t start_blk,
3650 unsigned int count);
3651 extern int ext4_check_blockref(const char *, unsigned int,
3652 struct inode *, __le32 *, unsigned int);
3653 extern int ext4_sb_block_valid(struct super_block *sb, struct inode *inode,
3654 ext4_fsblk_t start_blk, unsigned int count);
3655
3656
3657 /* extents.c */
3658 struct ext4_ext_path;
3659 struct ext4_extent;
3660
3661 /*
3662 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3663 * __le32.
3664 */
3665 #define EXT_MAX_BLOCKS 0xffffffff
3666
3667 extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode);
3668 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3669 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3670 struct ext4_map_blocks *map, int flags);
3671 extern int ext4_ext_truncate(handle_t *, struct inode *);
3672 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3673 ext4_lblk_t end);
3674 extern void ext4_ext_init(struct super_block *);
3675 extern void ext4_ext_release(struct super_block *);
3676 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3677 loff_t len);
3678 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3679 loff_t offset, ssize_t len);
3680 extern int ext4_convert_unwritten_io_end_vec(handle_t *handle,
3681 ext4_io_end_t *io_end);
3682 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3683 struct ext4_map_blocks *map, int flags);
3684 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3685 int num,
3686 struct ext4_ext_path *path);
3687 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3688 struct ext4_ext_path **,
3689 struct ext4_extent *, int);
3690 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3691 struct ext4_ext_path **,
3692 int flags);
3693 extern void ext4_free_ext_path(struct ext4_ext_path *);
3694 extern int ext4_ext_check_inode(struct inode *inode);
3695 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3696 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3697 __u64 start, __u64 len);
3698 extern int ext4_get_es_cache(struct inode *inode,
3699 struct fiemap_extent_info *fieinfo,
3700 __u64 start, __u64 len);
3701 extern int ext4_ext_precache(struct inode *inode);
3702 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3703 struct inode *inode2, ext4_lblk_t lblk1,
3704 ext4_lblk_t lblk2, ext4_lblk_t count,
3705 int mark_unwritten,int *err);
3706 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu);
3707 extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode,
3708 int check_cred, int restart_cred,
3709 int revoke_cred);
3710 extern void ext4_ext_replay_shrink_inode(struct inode *inode, ext4_lblk_t end);
3711 extern int ext4_ext_replay_set_iblocks(struct inode *inode);
3712 extern int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start,
3713 int len, int unwritten, ext4_fsblk_t pblk);
3714 extern int ext4_ext_clear_bb(struct inode *inode);
3715
3716
3717 /* move_extent.c */
3718 extern void ext4_double_down_write_data_sem(struct inode *first,
3719 struct inode *second);
3720 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3721 struct inode *donor_inode);
3722 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3723 __u64 start_orig, __u64 start_donor,
3724 __u64 len, __u64 *moved_len);
3725
3726 /* page-io.c */
3727 extern int __init ext4_init_pageio(void);
3728 extern void ext4_exit_pageio(void);
3729 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3730 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3731 extern int ext4_put_io_end(ext4_io_end_t *io_end);
3732 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3733 extern void ext4_io_submit_init(struct ext4_io_submit *io,
3734 struct writeback_control *wbc);
3735 extern void ext4_end_io_rsv_work(struct work_struct *work);
3736 extern void ext4_io_submit(struct ext4_io_submit *io);
3737 int ext4_bio_write_folio(struct ext4_io_submit *io, struct folio *page,
3738 size_t len);
3739 extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end);
3740 extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end);
3741
3742 /* mmp.c */
3743 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3744
3745 /* mmp.c */
3746 extern void ext4_stop_mmpd(struct ext4_sb_info *sbi);
3747
3748 /* verity.c */
3749 extern const struct fsverity_operations ext4_verityops;
3750
3751 /* orphan.c */
3752 extern int ext4_orphan_add(handle_t *, struct inode *);
3753 extern int ext4_orphan_del(handle_t *, struct inode *);
3754 extern void ext4_orphan_cleanup(struct super_block *sb,
3755 struct ext4_super_block *es);
3756 extern void ext4_release_orphan_info(struct super_block *sb);
3757 extern int ext4_init_orphan_info(struct super_block *sb);
3758 extern int ext4_orphan_file_empty(struct super_block *sb);
3759 extern void ext4_orphan_file_block_trigger(
3760 struct jbd2_buffer_trigger_type *triggers,
3761 struct buffer_head *bh,
3762 void *data, size_t size);
3763
3764 /*
3765 * Add new method to test whether block and inode bitmaps are properly
3766 * initialized. With uninit_bg reading the block from disk is not enough
3767 * to mark the bitmap uptodate. We need to also zero-out the bitmap
3768 */
3769 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3770
bitmap_uptodate(struct buffer_head * bh)3771 static inline int bitmap_uptodate(struct buffer_head *bh)
3772 {
3773 return (buffer_uptodate(bh) &&
3774 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3775 }
set_bitmap_uptodate(struct buffer_head * bh)3776 static inline void set_bitmap_uptodate(struct buffer_head *bh)
3777 {
3778 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3779 }
3780
3781 /* For ioend & aio unwritten conversion wait queues */
3782 #define EXT4_WQ_HASH_SZ 37
3783 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
3784 EXT4_WQ_HASH_SZ])
3785 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3786
3787 extern int ext4_resize_begin(struct super_block *sb);
3788 extern int ext4_resize_end(struct super_block *sb, bool update_backups);
3789
ext4_set_io_unwritten_flag(struct inode * inode,struct ext4_io_end * io_end)3790 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
3791 struct ext4_io_end *io_end)
3792 {
3793 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
3794 io_end->flag |= EXT4_IO_END_UNWRITTEN;
3795 atomic_inc(&EXT4_I(inode)->i_unwritten);
3796 }
3797 }
3798
ext4_clear_io_unwritten_flag(ext4_io_end_t * io_end)3799 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3800 {
3801 struct inode *inode = io_end->inode;
3802
3803 if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
3804 io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3805 /* Wake up anyone waiting on unwritten extent conversion */
3806 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
3807 wake_up_all(ext4_ioend_wq(inode));
3808 }
3809 }
3810
3811 extern const struct iomap_ops ext4_iomap_ops;
3812 extern const struct iomap_ops ext4_iomap_overwrite_ops;
3813 extern const struct iomap_ops ext4_iomap_report_ops;
3814
ext4_buffer_uptodate(struct buffer_head * bh)3815 static inline int ext4_buffer_uptodate(struct buffer_head *bh)
3816 {
3817 /*
3818 * If the buffer has the write error flag, we have failed
3819 * to write out data in the block. In this case, we don't
3820 * have to read the block because we may read the old data
3821 * successfully.
3822 */
3823 if (buffer_write_io_error(bh))
3824 set_buffer_uptodate(bh);
3825 return buffer_uptodate(bh);
3826 }
3827
3828 #endif /* __KERNEL__ */
3829
3830 #define EFSBADCRC EBADMSG /* Bad CRC detected */
3831 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
3832
3833 #endif /* _EXT4_H */
3834