1 // SPDX-License-Identifier: GPL-2.0
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)
12 * linux/include/linux/minix_fs.h
14 * Copyright (C) 1991, 1992 Linus Torvalds
20 #include <linux/types.h>
21 #include <linux/blkdev.h>
22 #include <linux/magic.h>
23 #include <linux/jbd2.h>
24 #include <linux/quota.h>
25 #include <linux/rwsem.h>
26 #include <linux/rbtree.h>
27 #include <linux/seqlock.h>
28 #include <linux/mutex.h>
29 #include <linux/timer.h>
30 #include <linux/wait.h>
31 #include <linux/sched/signal.h>
32 #include <linux/blockgroup_lock.h>
33 #include <linux/percpu_counter.h>
34 #include <linux/ratelimit.h>
35 #include <crypto/hash.h>
36 #include <linux/falloc.h>
37 #include <linux/percpu-rwsem.h>
38 #include <linux/fiemap.h>
40 #include <linux/compat.h>
43 #include <linux/fscrypt.h>
44 #include <linux/fsverity.h>
46 #include <linux/compiler.h>
49 * The fourth extended filesystem constants/structures
53 * with AGGRESSIVE_CHECK allocator runs consistency checks over
54 * structures. these checks slow things down a lot
56 #define AGGRESSIVE_CHECK__
59 * with DOUBLE_CHECK defined mballoc creates persistent in-core
60 * bitmaps, maintains and uses them to check for double allocations
62 #define DOUBLE_CHECK__
65 * Define EXT4FS_DEBUG to produce debug messages
73 #define ext4_debug(f, a...) \
75 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
76 __FILE__, __LINE__, __func__); \
77 printk(KERN_DEBUG f, ## a); \
80 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
84 * Turn on EXT_DEBUG to enable ext4_ext_show_path/leaf/move in extents.c
89 * Dynamic printk for controlled extents debugging.
91 #ifdef CONFIG_EXT4_DEBUG
92 #define ext_debug(ino, fmt, ...) \
93 pr_debug("[%s/%d] EXT4-fs (%s): ino %lu: (%s, %d): %s:" fmt, \
94 current->comm, task_pid_nr(current), \
95 ino->i_sb->s_id, ino->i_ino, __FILE__, __LINE__, \
96 __func__, ##__VA_ARGS__)
98 #define ext_debug(ino, fmt, ...) no_printk(fmt, ##__VA_ARGS__)
101 #define ASSERT(assert) \
103 if (unlikely(!(assert))) { \
105 "Assertion failure in %s() at %s:%d: '%s'\n", \
106 __func__, __FILE__, __LINE__, #assert); \
111 /* data type for block offset of block group */
112 typedef int ext4_grpblk_t;
114 /* data type for filesystem-wide blocks number */
115 typedef unsigned long long ext4_fsblk_t;
117 /* data type for file logical block number */
118 typedef __u32 ext4_lblk_t;
120 /* data type for block group number */
121 typedef unsigned int ext4_group_t;
123 enum SHIFT_DIRECTION {
129 * Flags used in mballoc's allocation_context flags field.
131 * Also used to show what's going on for debugging purposes when the
132 * flag field is exported via the traceport interface
135 /* prefer goal again. length */
136 #define EXT4_MB_HINT_MERGE 0x0001
137 /* blocks already reserved */
138 #define EXT4_MB_HINT_RESERVED 0x0002
139 /* metadata is being allocated */
140 #define EXT4_MB_HINT_METADATA 0x0004
141 /* first blocks in the file */
142 #define EXT4_MB_HINT_FIRST 0x0008
143 /* search for the best chunk */
144 #define EXT4_MB_HINT_BEST 0x0010
145 /* data is being allocated */
146 #define EXT4_MB_HINT_DATA 0x0020
147 /* don't preallocate (for tails) */
148 #define EXT4_MB_HINT_NOPREALLOC 0x0040
149 /* allocate for locality group */
150 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080
151 /* allocate goal blocks or none */
152 #define EXT4_MB_HINT_GOAL_ONLY 0x0100
153 /* goal is meaningful */
154 #define EXT4_MB_HINT_TRY_GOAL 0x0200
155 /* blocks already pre-reserved by delayed allocation */
156 #define EXT4_MB_DELALLOC_RESERVED 0x0400
157 /* We are doing stream allocation */
158 #define EXT4_MB_STREAM_ALLOC 0x0800
159 /* Use reserved root blocks if needed */
160 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000
161 /* Use blocks from reserved pool */
162 #define EXT4_MB_USE_RESERVED 0x2000
163 /* Do strict check for free blocks while retrying block allocation */
164 #define EXT4_MB_STRICT_CHECK 0x4000
165 /* Large fragment size list lookup succeeded at least once for cr = 0 */
166 #define EXT4_MB_CR0_OPTIMIZED 0x8000
167 /* Avg fragment size rb tree lookup succeeded at least once for cr = 1 */
168 #define EXT4_MB_CR1_OPTIMIZED 0x00010000
169 /* Perform linear traversal for one group */
170 #define EXT4_MB_SEARCH_NEXT_LINEAR 0x00020000
171 struct ext4_allocation_request {
172 /* target inode for block we're allocating */
174 /* how many blocks we want to allocate */
176 /* logical block in target inode */
178 /* the closest logical allocated block to the left */
180 /* the closest logical allocated block to the right */
182 /* phys. target (a hint) */
184 /* phys. block for the closest logical allocated block to the left */
186 /* phys. block for the closest logical allocated block to the right */
188 /* flags. see above EXT4_MB_HINT_* */
193 * Logical to physical block mapping, used by ext4_map_blocks()
195 * This structure is used to pass requests into ext4_map_blocks() as
196 * well as to store the information returned by ext4_map_blocks(). It
197 * takes less room on the stack than a struct buffer_head.
199 #define EXT4_MAP_NEW BIT(BH_New)
200 #define EXT4_MAP_MAPPED BIT(BH_Mapped)
201 #define EXT4_MAP_UNWRITTEN BIT(BH_Unwritten)
202 #define EXT4_MAP_BOUNDARY BIT(BH_Boundary)
203 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
204 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
206 struct ext4_map_blocks {
210 unsigned int m_flags;
214 * Block validity checking, system zone rbtree.
216 struct ext4_system_blocks {
222 * Flags for ext4_io_end->flags
224 #define EXT4_IO_END_UNWRITTEN 0x0001
226 struct ext4_io_end_vec {
227 struct list_head list; /* list of io_end_vec */
228 loff_t offset; /* offset in the file */
229 ssize_t size; /* size of the extent */
233 * For converting unwritten extents on a work queue. 'handle' is used for
234 * buffered writeback.
236 typedef struct ext4_io_end {
237 struct list_head list; /* per-file finished IO list */
238 handle_t *handle; /* handle reserved for extent
240 struct inode *inode; /* file being written to */
241 struct bio *bio; /* Linked list of completed
242 * bios covering the extent */
243 unsigned int flag; /* unwritten or not */
244 atomic_t count; /* reference counter */
245 struct list_head list_vec; /* list of ext4_io_end_vec */
248 struct ext4_io_submit {
249 struct writeback_control *io_wbc;
251 ext4_io_end_t *io_end;
252 sector_t io_next_block;
256 * Special inodes numbers
258 #define EXT4_BAD_INO 1 /* Bad blocks inode */
259 #define EXT4_ROOT_INO 2 /* Root inode */
260 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */
261 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
262 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
263 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
264 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
265 #define EXT4_JOURNAL_INO 8 /* Journal inode */
267 /* First non-reserved inode for old ext4 filesystems */
268 #define EXT4_GOOD_OLD_FIRST_INO 11
271 * Maximal count of links to a file
273 #define EXT4_LINK_MAX 65000
276 * Macro-instructions used to manage several block sizes
278 #define EXT4_MIN_BLOCK_SIZE 1024
279 #define EXT4_MAX_BLOCK_SIZE 65536
280 #define EXT4_MIN_BLOCK_LOG_SIZE 10
281 #define EXT4_MAX_BLOCK_LOG_SIZE 16
282 #define EXT4_MAX_CLUSTER_LOG_SIZE 30
284 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
286 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
288 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
289 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
290 EXT4_SB(s)->s_cluster_bits)
292 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
293 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
295 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
298 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
299 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
300 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
302 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
303 EXT4_GOOD_OLD_INODE_SIZE : \
305 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
306 EXT4_GOOD_OLD_FIRST_INO : \
309 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
310 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \
311 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \
314 /* Translate a block number to a cluster number */
315 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
316 /* Translate a cluster number to a block number */
317 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
318 /* Translate # of blks to # of clusters */
319 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
320 (sbi)->s_cluster_bits)
321 /* Mask out the low bits to get the starting block of the cluster */
322 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
323 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
324 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
325 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
326 /* Fill in the low bits to get the last block of the cluster */
327 #define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) | \
328 ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1))
329 /* Get the cluster offset */
330 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
331 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
332 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
333 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
336 * Structure of a blocks group descriptor
338 struct ext4_group_desc
340 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
341 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
342 __le32 bg_inode_table_lo; /* Inodes table block */
343 __le16 bg_free_blocks_count_lo;/* Free blocks count */
344 __le16 bg_free_inodes_count_lo;/* Free inodes count */
345 __le16 bg_used_dirs_count_lo; /* Directories count */
346 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
347 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
348 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
349 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
350 __le16 bg_itable_unused_lo; /* Unused inodes count */
351 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
352 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
353 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
354 __le32 bg_inode_table_hi; /* Inodes table block MSB */
355 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
356 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
357 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
358 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
359 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
360 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
361 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
365 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
366 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
368 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
369 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
373 * Structure of a flex block group info
377 atomic64_t free_clusters;
378 atomic_t free_inodes;
382 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
383 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
384 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
387 * Macro-instructions used to manage group descriptors
389 #define EXT4_MIN_DESC_SIZE 32
390 #define EXT4_MIN_DESC_SIZE_64BIT 64
391 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
392 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
394 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
395 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
396 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
397 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
398 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
400 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
401 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
402 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
406 * Constants relative to the data blocks
408 #define EXT4_NDIR_BLOCKS 12
409 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
410 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
411 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
412 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
417 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
418 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
419 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
420 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
421 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
422 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
423 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
424 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
425 /* Reserved for compression usage... */
426 #define EXT4_DIRTY_FL 0x00000100
427 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
428 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
429 /* nb: was previously EXT2_ECOMPR_FL */
430 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */
431 /* End compression flags --- maybe not all used */
432 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
433 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
434 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
435 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
436 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
437 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
438 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
439 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
440 #define EXT4_VERITY_FL 0x00100000 /* Verity protected inode */
441 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
442 /* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */
444 #define EXT4_DAX_FL 0x02000000 /* Inode is DAX */
446 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
447 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
448 #define EXT4_CASEFOLD_FL 0x40000000 /* Casefolded directory */
449 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
451 /* User modifiable flags */
452 #define EXT4_FL_USER_MODIFIABLE (EXT4_SECRM_FL | \
456 EXT4_IMMUTABLE_FL | \
460 EXT4_JOURNAL_DATA_FL | \
465 0x00400000 /* EXT4_EOFBLOCKS_FL */ | \
467 EXT4_PROJINHERIT_FL | \
470 /* User visible flags */
471 #define EXT4_FL_USER_VISIBLE (EXT4_FL_USER_MODIFIABLE | \
480 /* Flags we can manipulate with through FS_IOC_FSSETXATTR */
481 #define EXT4_FL_XFLAG_VISIBLE (EXT4_SYNC_FL | \
482 EXT4_IMMUTABLE_FL | \
486 EXT4_PROJINHERIT_FL | \
489 /* Flags that should be inherited by new inodes from their parent. */
490 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
491 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
492 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
493 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
494 EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL |\
497 /* Flags that are appropriate for regular files (all but dir-specific ones). */
498 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\
499 EXT4_PROJINHERIT_FL))
501 /* Flags that are appropriate for non-directories/regular files. */
502 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
504 /* The only flags that should be swapped */
505 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL)
507 /* Flags which are mutually exclusive to DAX */
508 #define EXT4_DAX_MUT_EXCL (EXT4_VERITY_FL | EXT4_ENCRYPT_FL |\
509 EXT4_JOURNAL_DATA_FL | EXT4_INLINE_DATA_FL)
511 /* Mask out flags that are inappropriate for the given type of inode. */
512 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
516 else if (S_ISREG(mode))
517 return flags & EXT4_REG_FLMASK;
519 return flags & EXT4_OTHER_FLMASK;
523 * Inode flags used for atomic set/get
526 EXT4_INODE_SECRM = 0, /* Secure deletion */
527 EXT4_INODE_UNRM = 1, /* Undelete */
528 EXT4_INODE_COMPR = 2, /* Compress file */
529 EXT4_INODE_SYNC = 3, /* Synchronous updates */
530 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
531 EXT4_INODE_APPEND = 5, /* writes to file may only append */
532 EXT4_INODE_NODUMP = 6, /* do not dump file */
533 EXT4_INODE_NOATIME = 7, /* do not update atime */
534 /* Reserved for compression usage... */
535 EXT4_INODE_DIRTY = 8,
536 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
537 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
538 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */
539 /* End compression flags --- maybe not all used */
540 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
541 EXT4_INODE_IMAGIC = 13, /* AFS directory */
542 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
543 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
544 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
545 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
546 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
547 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
548 EXT4_INODE_VERITY = 20, /* Verity protected inode */
549 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
550 /* 22 was formerly EXT4_INODE_EOFBLOCKS */
551 EXT4_INODE_DAX = 25, /* Inode is DAX */
552 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
553 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */
554 EXT4_INODE_CASEFOLD = 30, /* Casefolded directory */
555 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
559 * Since it's pretty easy to mix up bit numbers and hex values, we use a
560 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
561 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
562 * any extra space in the compiled kernel image, otherwise, the build will fail.
563 * It's important that these values are the same, since we are using
564 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
565 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
566 * values found in ext2, ext3 and ext4 filesystems, and of course the values
567 * defined in e2fsprogs.
569 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
571 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
572 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
574 static inline void ext4_check_flag_values(void)
576 CHECK_FLAG_VALUE(SECRM);
577 CHECK_FLAG_VALUE(UNRM);
578 CHECK_FLAG_VALUE(COMPR);
579 CHECK_FLAG_VALUE(SYNC);
580 CHECK_FLAG_VALUE(IMMUTABLE);
581 CHECK_FLAG_VALUE(APPEND);
582 CHECK_FLAG_VALUE(NODUMP);
583 CHECK_FLAG_VALUE(NOATIME);
584 CHECK_FLAG_VALUE(DIRTY);
585 CHECK_FLAG_VALUE(COMPRBLK);
586 CHECK_FLAG_VALUE(NOCOMPR);
587 CHECK_FLAG_VALUE(ENCRYPT);
588 CHECK_FLAG_VALUE(INDEX);
589 CHECK_FLAG_VALUE(IMAGIC);
590 CHECK_FLAG_VALUE(JOURNAL_DATA);
591 CHECK_FLAG_VALUE(NOTAIL);
592 CHECK_FLAG_VALUE(DIRSYNC);
593 CHECK_FLAG_VALUE(TOPDIR);
594 CHECK_FLAG_VALUE(HUGE_FILE);
595 CHECK_FLAG_VALUE(EXTENTS);
596 CHECK_FLAG_VALUE(VERITY);
597 CHECK_FLAG_VALUE(EA_INODE);
598 CHECK_FLAG_VALUE(INLINE_DATA);
599 CHECK_FLAG_VALUE(PROJINHERIT);
600 CHECK_FLAG_VALUE(CASEFOLD);
601 CHECK_FLAG_VALUE(RESERVED);
604 /* Used to pass group descriptor data when online resize is done */
605 struct ext4_new_group_input {
606 __u32 group; /* Group number for this data */
607 __u64 block_bitmap; /* Absolute block number of block bitmap */
608 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
609 __u64 inode_table; /* Absolute block number of inode table start */
610 __u32 blocks_count; /* Total number of blocks in this group */
611 __u16 reserved_blocks; /* Number of reserved blocks in this group */
615 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
616 struct compat_ext4_new_group_input {
618 compat_u64 block_bitmap;
619 compat_u64 inode_bitmap;
620 compat_u64 inode_table;
627 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
628 struct ext4_new_group_data {
634 __u16 reserved_blocks;
636 __u32 free_clusters_count;
639 /* Indexes used to index group tables in ext4_new_group_data */
641 BLOCK_BITMAP = 0, /* block bitmap */
642 INODE_BITMAP, /* inode bitmap */
643 INODE_TABLE, /* inode tables */
648 * Flags used by ext4_map_blocks()
650 /* Allocate any needed blocks and/or convert an unwritten
651 extent to be an initialized ext4 */
652 #define EXT4_GET_BLOCKS_CREATE 0x0001
653 /* Request the creation of an unwritten extent */
654 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002
655 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\
656 EXT4_GET_BLOCKS_CREATE)
657 /* Caller is from the delayed allocation writeout path
658 * finally doing the actual allocation of delayed blocks */
659 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
660 /* caller is from the direct IO path, request to creation of an
661 unwritten extents if not allocated, split the unwritten
662 extent if blocks has been preallocated already*/
663 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
664 #define EXT4_GET_BLOCKS_CONVERT 0x0010
665 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
666 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
667 /* Convert extent to initialized after IO complete */
668 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
669 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
670 /* Eventual metadata allocation (due to growing extent tree)
671 * should not fail, so try to use reserved blocks for that.*/
672 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
673 /* Don't normalize allocation size (used for fallocate) */
674 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
675 /* Convert written extents to unwritten */
676 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100
677 /* Write zeros to newly created written extents */
678 #define EXT4_GET_BLOCKS_ZERO 0x0200
679 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\
680 EXT4_GET_BLOCKS_ZERO)
681 /* Caller will submit data before dropping transaction handle. This
682 * allows jbd2 to avoid submitting data before commit. */
683 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400
686 * The bit position of these flags must not overlap with any of the
687 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(),
688 * read_extent_tree_block(), ext4_split_extent_at(),
689 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
690 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
691 * caching the extents when reading from the extent tree while a
692 * truncate or punch hole operation is in progress.
694 #define EXT4_EX_NOCACHE 0x40000000
695 #define EXT4_EX_FORCE_CACHE 0x20000000
696 #define EXT4_EX_NOFAIL 0x10000000
699 * Flags used by ext4_free_blocks
701 #define EXT4_FREE_BLOCKS_METADATA 0x0001
702 #define EXT4_FREE_BLOCKS_FORGET 0x0002
703 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
704 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
705 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
706 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
707 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040
712 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
713 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
714 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
715 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
716 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
717 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
718 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
719 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
720 #define EXT4_IOC_MIGRATE _IO('f', 9)
721 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
722 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
723 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
724 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
725 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
726 #define EXT4_IOC_SWAP_BOOT _IO('f', 17)
727 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18)
728 /* ioctl codes 19--39 are reserved for fscrypt */
729 #define EXT4_IOC_CLEAR_ES_CACHE _IO('f', 40)
730 #define EXT4_IOC_GETSTATE _IOW('f', 41, __u32)
731 #define EXT4_IOC_GET_ES_CACHE _IOWR('f', 42, struct fiemap)
733 #define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32)
736 * Flags for going down operation
738 #define EXT4_GOING_FLAGS_DEFAULT 0x0 /* going down */
739 #define EXT4_GOING_FLAGS_LOGFLUSH 0x1 /* flush log but not data */
740 #define EXT4_GOING_FLAGS_NOLOGFLUSH 0x2 /* don't flush log nor data */
743 * Flags returned by EXT4_IOC_GETSTATE
745 * We only expose to userspace a subset of the state flags in
748 #define EXT4_STATE_FLAG_EXT_PRECACHED 0x00000001
749 #define EXT4_STATE_FLAG_NEW 0x00000002
750 #define EXT4_STATE_FLAG_NEWENTRY 0x00000004
751 #define EXT4_STATE_FLAG_DA_ALLOC_CLOSE 0x00000008
753 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
755 * ioctl commands in 32 bit emulation
757 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
758 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
759 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
760 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
761 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
762 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
763 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
764 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
768 * Returned by EXT4_IOC_GET_ES_CACHE as an additional possible flag.
769 * It indicates that the entry in extent status cache is for a hole.
771 #define EXT4_FIEMAP_EXTENT_HOLE 0x08000000
773 /* Max physical block we can address w/o extents */
774 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
776 /* Max logical block we can support */
777 #define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFE
780 * Structure of an inode on the disk
783 __le16 i_mode; /* File mode */
784 __le16 i_uid; /* Low 16 bits of Owner Uid */
785 __le32 i_size_lo; /* Size in bytes */
786 __le32 i_atime; /* Access time */
787 __le32 i_ctime; /* Inode Change time */
788 __le32 i_mtime; /* Modification time */
789 __le32 i_dtime; /* Deletion Time */
790 __le16 i_gid; /* Low 16 bits of Group Id */
791 __le16 i_links_count; /* Links count */
792 __le32 i_blocks_lo; /* Blocks count */
793 __le32 i_flags; /* File flags */
799 __u32 h_i_translator;
804 } osd1; /* OS dependent 1 */
805 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
806 __le32 i_generation; /* File version (for NFS) */
807 __le32 i_file_acl_lo; /* File ACL */
809 __le32 i_obso_faddr; /* Obsoleted fragment address */
812 __le16 l_i_blocks_high; /* were l_i_reserved1 */
813 __le16 l_i_file_acl_high;
814 __le16 l_i_uid_high; /* these 2 fields */
815 __le16 l_i_gid_high; /* were reserved2[0] */
816 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
820 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
827 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
828 __le16 m_i_file_acl_high;
829 __u32 m_i_reserved2[2];
831 } osd2; /* OS dependent 2 */
832 __le16 i_extra_isize;
833 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
834 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
835 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
836 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
837 __le32 i_crtime; /* File Creation time */
838 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
839 __le32 i_version_hi; /* high 32 bits for 64-bit version */
840 __le32 i_projid; /* Project ID */
844 __u32 reserved; /* should be zero */
845 __u32 donor_fd; /* donor file descriptor */
846 __u64 orig_start; /* logical start offset in block for orig */
847 __u64 donor_start; /* logical start offset in block for donor */
848 __u64 len; /* block length to be moved */
849 __u64 moved_len; /* moved block length */
852 #define EXT4_EPOCH_BITS 2
853 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
854 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
857 * Extended fields will fit into an inode if the filesystem was formatted
858 * with large inodes (-I 256 or larger) and there are not currently any EAs
859 * consuming all of the available space. For new inodes we always reserve
860 * enough space for the kernel's known extended fields, but for inodes
861 * created with an old kernel this might not have been the case. None of
862 * the extended inode fields is critical for correct filesystem operation.
863 * This macro checks if a certain field fits in the inode. Note that
864 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
866 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
867 ((offsetof(typeof(*ext4_inode), field) + \
868 sizeof((ext4_inode)->field)) \
869 <= (EXT4_GOOD_OLD_INODE_SIZE + \
870 (einode)->i_extra_isize)) \
873 * We use an encoding that preserves the times for extra epoch "00":
875 * extra msb of adjust for signed
876 * epoch 32-bit 32-bit tv_sec to
877 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
878 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
879 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
880 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
881 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
882 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
883 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
884 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
885 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
887 * Note that previous versions of the kernel on 64-bit systems would
888 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
889 * 1970. e2fsck will correct this, assuming that it is run on the
890 * affected filesystem before 2242.
893 static inline __le32 ext4_encode_extra_time(struct timespec64 *time)
895 u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK;
896 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
899 static inline void ext4_decode_extra_time(struct timespec64 *time,
902 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK)))
903 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
904 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
907 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
909 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {\
910 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
911 (raw_inode)->xtime ## _extra = \
912 ext4_encode_extra_time(&(inode)->xtime); \
915 (raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (inode)->xtime.tv_sec, S32_MIN, S32_MAX)); \
918 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
920 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
921 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
922 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
923 (raw_inode)->xtime ## _extra = \
924 ext4_encode_extra_time(&(einode)->xtime); \
927 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
929 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
930 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \
931 ext4_decode_extra_time(&(inode)->xtime, \
932 raw_inode->xtime ## _extra); \
935 (inode)->xtime.tv_nsec = 0; \
939 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
941 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
942 (einode)->xtime.tv_sec = \
943 (signed)le32_to_cpu((raw_inode)->xtime); \
945 (einode)->xtime.tv_sec = 0; \
946 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
947 ext4_decode_extra_time(&(einode)->xtime, \
948 raw_inode->xtime ## _extra); \
950 (einode)->xtime.tv_nsec = 0; \
953 #define i_disk_version osd1.linux1.l_i_version
955 #if defined(__KERNEL__) || defined(__linux__)
956 #define i_reserved1 osd1.linux1.l_i_reserved1
957 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
958 #define i_blocks_high osd2.linux2.l_i_blocks_high
959 #define i_uid_low i_uid
960 #define i_gid_low i_gid
961 #define i_uid_high osd2.linux2.l_i_uid_high
962 #define i_gid_high osd2.linux2.l_i_gid_high
963 #define i_checksum_lo osd2.linux2.l_i_checksum_lo
965 #elif defined(__GNU__)
967 #define i_translator osd1.hurd1.h_i_translator
968 #define i_uid_high osd2.hurd2.h_i_uid_high
969 #define i_gid_high osd2.hurd2.h_i_gid_high
970 #define i_author osd2.hurd2.h_i_author
972 #elif defined(__masix__)
974 #define i_reserved1 osd1.masix1.m_i_reserved1
975 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
976 #define i_reserved2 osd2.masix2.m_i_reserved2
978 #endif /* defined(__KERNEL__) || defined(__linux__) */
980 #include "extents_status.h"
981 #include "fast_commit.h"
984 * Lock subclasses for i_data_sem in the ext4_inode_info structure.
986 * These are needed to avoid lockdep false positives when we need to
987 * allocate blocks to the quota inode during ext4_map_blocks(), while
988 * holding i_data_sem for a normal (non-quota) inode. Since we don't
989 * do quota tracking for the quota inode, this avoids deadlock (as
990 * well as infinite recursion, since it isn't turtles all the way
993 * I_DATA_SEM_NORMAL - Used for most inodes
994 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode
995 * where the second inode has larger inode number
997 * I_DATA_SEM_QUOTA - Used for quota inodes only
1000 I_DATA_SEM_NORMAL = 0,
1007 * fourth extended file system inode data in memory
1009 struct ext4_inode_info {
1010 __le32 i_data[15]; /* unconverted */
1012 ext4_fsblk_t i_file_acl;
1015 * i_block_group is the number of the block group which contains
1016 * this file's inode. Constant across the lifetime of the inode,
1017 * it is used for making block allocation decisions - we try to
1018 * place a file's data blocks near its inode block, and new inodes
1019 * near to their parent directory's inode.
1021 ext4_group_t i_block_group;
1022 ext4_lblk_t i_dir_start_lookup;
1023 #if (BITS_PER_LONG < 64)
1024 unsigned long i_state_flags; /* Dynamic state flags */
1026 unsigned long i_flags;
1029 * Extended attributes can be read independently of the main file
1030 * data. Taking i_mutex even when reading would cause contention
1031 * between readers of EAs and writers of regular file data, so
1032 * instead we synchronize on xattr_sem when reading or changing
1035 struct rw_semaphore xattr_sem;
1037 struct list_head i_orphan; /* unlinked but open inodes */
1039 /* Fast commit related info */
1041 struct list_head i_fc_list; /*
1042 * inodes that need fast commit
1043 * protected by sbi->s_fc_lock.
1046 /* Start of lblk range that needs to be committed in this fast commit */
1047 ext4_lblk_t i_fc_lblk_start;
1049 /* End of lblk range that needs to be committed in this fast commit */
1050 ext4_lblk_t i_fc_lblk_len;
1052 /* Number of ongoing updates on this inode */
1053 atomic_t i_fc_updates;
1055 /* Fast commit wait queue for this inode */
1056 wait_queue_head_t i_fc_wait;
1058 /* Protect concurrent accesses on i_fc_lblk_start, i_fc_lblk_len */
1059 struct mutex i_fc_lock;
1062 * i_disksize keeps track of what the inode size is ON DISK, not
1063 * in memory. During truncate, i_size is set to the new size by
1064 * the VFS prior to calling ext4_truncate(), but the filesystem won't
1065 * set i_disksize to 0 until the truncate is actually under way.
1067 * The intent is that i_disksize always represents the blocks which
1068 * are used by this file. This allows recovery to restart truncate
1069 * on orphans if we crash during truncate. We actually write i_disksize
1070 * into the on-disk inode when writing inodes out, instead of i_size.
1072 * The only time when i_disksize and i_size may be different is when
1073 * a truncate is in progress. The only things which change i_disksize
1074 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
1079 * i_data_sem is for serialising ext4_truncate() against
1080 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
1081 * data tree are chopped off during truncate. We can't do that in
1082 * ext4 because whenever we perform intermediate commits during
1083 * truncate, the inode and all the metadata blocks *must* be in a
1084 * consistent state which allows truncation of the orphans to restart
1085 * during recovery. Hence we must fix the get_block-vs-truncate race
1086 * by other means, so we have i_data_sem.
1088 struct rw_semaphore i_data_sem;
1090 * i_mmap_sem is for serializing page faults with truncate / punch hole
1091 * operations. We have to make sure that new page cannot be faulted in
1092 * a section of the inode that is being punched. We cannot easily use
1093 * i_data_sem for this since we need protection for the whole punch
1094 * operation and i_data_sem ranks below transaction start so we have
1095 * to occasionally drop it.
1097 struct rw_semaphore i_mmap_sem;
1098 struct inode vfs_inode;
1099 struct jbd2_inode *jinode;
1101 spinlock_t i_raw_lock; /* protects updates to the raw inode */
1104 * File creation time. Its function is same as that of
1105 * struct timespec64 i_{a,c,m}time in the generic inode.
1107 struct timespec64 i_crtime;
1110 atomic_t i_prealloc_active;
1111 struct list_head i_prealloc_list;
1112 spinlock_t i_prealloc_lock;
1114 /* extents status tree */
1115 struct ext4_es_tree i_es_tree;
1117 struct list_head i_es_list;
1118 unsigned int i_es_all_nr; /* protected by i_es_lock */
1119 unsigned int i_es_shk_nr; /* protected by i_es_lock */
1120 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for
1121 extents to shrink. Protected by
1125 ext4_group_t i_last_alloc_group;
1127 /* allocation reservation info for delalloc */
1128 /* In case of bigalloc, this refer to clusters rather than blocks */
1129 unsigned int i_reserved_data_blocks;
1131 /* pending cluster reservations for bigalloc file systems */
1132 struct ext4_pending_tree i_pending_tree;
1134 /* on-disk additional length */
1135 __u16 i_extra_isize;
1137 /* Indicate the inline data space. */
1142 /* quota space reservation, managed internally by quota code */
1143 qsize_t i_reserved_quota;
1146 /* Lock protecting lists below */
1147 spinlock_t i_completed_io_lock;
1149 * Completed IOs that need unwritten extents handling and have
1150 * transaction reserved
1152 struct list_head i_rsv_conversion_list;
1153 struct work_struct i_rsv_conversion_work;
1154 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1156 spinlock_t i_block_reservation_lock;
1159 * Transactions that contain inode's metadata needed to complete
1160 * fsync and fdatasync, respectively.
1163 tid_t i_datasync_tid;
1166 struct dquot *i_dquot[MAXQUOTAS];
1169 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1176 * File system states
1178 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
1179 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
1180 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
1181 #define EXT4_FC_REPLAY 0x0020 /* Fast commit replay ongoing */
1184 * Misc. filesystem flags
1186 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
1187 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
1188 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
1191 * Mount flags set via mount options or defaults
1193 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */
1194 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
1195 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
1196 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
1197 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
1198 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
1199 #define EXT4_MOUNT_ERRORS_MASK 0x00070
1200 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
1201 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
1202 #ifdef CONFIG_FS_DAX
1203 #define EXT4_MOUNT_DAX_ALWAYS 0x00200 /* Direct Access */
1205 #define EXT4_MOUNT_DAX_ALWAYS 0
1207 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
1208 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
1209 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
1210 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
1211 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
1212 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
1213 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
1214 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
1215 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
1216 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
1217 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */
1218 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota,
1219 * enable enforcement for hidden
1221 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable
1222 * enforcement for hidden quota
1224 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota
1226 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
1227 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
1228 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1229 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */
1230 #define EXT4_MOUNT_NO_PREFETCH_BLOCK_BITMAPS 0x4000000
1231 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
1232 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
1233 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
1234 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
1235 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
1238 * Mount flags set either automatically (could not be set by mount option)
1239 * based on per file system feature or property or in special cases such as
1240 * distinguishing between explicit mount option definition and default.
1242 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1243 specified delalloc */
1244 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1245 size of blocksize * 8
1247 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1249 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly
1250 specified journal checksum */
1252 #define EXT4_MOUNT2_JOURNAL_FAST_COMMIT 0x00000010 /* Journal fast commit */
1253 #define EXT4_MOUNT2_DAX_NEVER 0x00000020 /* Do not allow Direct Access */
1254 #define EXT4_MOUNT2_DAX_INODE 0x00000040 /* For printing options only */
1255 #define EXT4_MOUNT2_MB_OPTIMIZE_SCAN 0x00000080 /* Optimize group
1256 * scanning in mballoc
1259 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1261 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1263 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1266 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1268 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1270 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1273 #define ext4_test_and_set_bit __test_and_set_bit_le
1274 #define ext4_set_bit __set_bit_le
1275 #define ext4_set_bit_atomic ext2_set_bit_atomic
1276 #define ext4_test_and_clear_bit __test_and_clear_bit_le
1277 #define ext4_clear_bit __clear_bit_le
1278 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
1279 #define ext4_test_bit test_bit_le
1280 #define ext4_find_next_zero_bit find_next_zero_bit_le
1281 #define ext4_find_next_bit find_next_bit_le
1283 extern void ext4_set_bits(void *bm, int cur, int len);
1286 * Maximal mount counts between two filesystem checks
1288 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1289 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1292 * Behaviour when detecting errors
1294 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1295 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1296 #define EXT4_ERRORS_PANIC 3 /* Panic */
1297 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1299 /* Metadata checksum algorithm codes */
1300 #define EXT4_CRC32C_CHKSUM 1
1303 * Structure of the super block
1305 struct ext4_super_block {
1306 /*00*/ __le32 s_inodes_count; /* Inodes count */
1307 __le32 s_blocks_count_lo; /* Blocks count */
1308 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1309 __le32 s_free_blocks_count_lo; /* Free blocks count */
1310 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1311 __le32 s_first_data_block; /* First Data Block */
1312 __le32 s_log_block_size; /* Block size */
1313 __le32 s_log_cluster_size; /* Allocation cluster size */
1314 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1315 __le32 s_clusters_per_group; /* # Clusters per group */
1316 __le32 s_inodes_per_group; /* # Inodes per group */
1317 __le32 s_mtime; /* Mount time */
1318 /*30*/ __le32 s_wtime; /* Write time */
1319 __le16 s_mnt_count; /* Mount count */
1320 __le16 s_max_mnt_count; /* Maximal mount count */
1321 __le16 s_magic; /* Magic signature */
1322 __le16 s_state; /* File system state */
1323 __le16 s_errors; /* Behaviour when detecting errors */
1324 __le16 s_minor_rev_level; /* minor revision level */
1325 /*40*/ __le32 s_lastcheck; /* time of last check */
1326 __le32 s_checkinterval; /* max. time between checks */
1327 __le32 s_creator_os; /* OS */
1328 __le32 s_rev_level; /* Revision level */
1329 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1330 __le16 s_def_resgid; /* Default gid for reserved blocks */
1332 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1334 * Note: the difference between the compatible feature set and
1335 * the incompatible feature set is that if there is a bit set
1336 * in the incompatible feature set that the kernel doesn't
1337 * know about, it should refuse to mount the filesystem.
1339 * e2fsck's requirements are more strict; if it doesn't know
1340 * about a feature in either the compatible or incompatible
1341 * feature set, it must abort and not try to meddle with
1342 * things it doesn't understand...
1344 __le32 s_first_ino; /* First non-reserved inode */
1345 __le16 s_inode_size; /* size of inode structure */
1346 __le16 s_block_group_nr; /* block group # of this superblock */
1347 __le32 s_feature_compat; /* compatible feature set */
1348 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1349 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1350 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1351 /*78*/ char s_volume_name[16]; /* volume name */
1352 /*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */
1353 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1355 * Performance hints. Directory preallocation should only
1356 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1358 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1359 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1360 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1362 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1364 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1365 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1366 __le32 s_journal_dev; /* device number of journal file */
1367 __le32 s_last_orphan; /* start of list of inodes to delete */
1368 __le32 s_hash_seed[4]; /* HTREE hash seed */
1369 __u8 s_def_hash_version; /* Default hash version to use */
1370 __u8 s_jnl_backup_type;
1371 __le16 s_desc_size; /* size of group descriptor */
1372 /*100*/ __le32 s_default_mount_opts;
1373 __le32 s_first_meta_bg; /* First metablock block group */
1374 __le32 s_mkfs_time; /* When the filesystem was created */
1375 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1376 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1377 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1378 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1379 __le32 s_free_blocks_count_hi; /* Free blocks count */
1380 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1381 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1382 __le32 s_flags; /* Miscellaneous flags */
1383 __le16 s_raid_stride; /* RAID stride */
1384 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1385 __le64 s_mmp_block; /* Block for multi-mount protection */
1386 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1387 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1388 __u8 s_checksum_type; /* metadata checksum algorithm used */
1389 __u8 s_encryption_level; /* versioning level for encryption */
1390 __u8 s_reserved_pad; /* Padding to next 32bits */
1391 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1392 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1393 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1394 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1395 snapshot's future use */
1396 __le32 s_snapshot_list; /* inode number of the head of the
1397 on-disk snapshot list */
1398 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1399 __le32 s_error_count; /* number of fs errors */
1400 __le32 s_first_error_time; /* first time an error happened */
1401 __le32 s_first_error_ino; /* inode involved in first error */
1402 __le64 s_first_error_block; /* block involved of first error */
1403 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */
1404 __le32 s_first_error_line; /* line number where error happened */
1405 __le32 s_last_error_time; /* most recent time of an error */
1406 __le32 s_last_error_ino; /* inode involved in last error */
1407 __le32 s_last_error_line; /* line number where error happened */
1408 __le64 s_last_error_block; /* block involved of last error */
1409 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */
1410 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1411 __u8 s_mount_opts[64];
1412 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1413 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1414 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1415 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
1416 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */
1417 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */
1418 __le32 s_lpf_ino; /* Location of the lost+found inode */
1419 __le32 s_prj_quota_inum; /* inode for tracking project quota */
1420 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */
1423 __u8 s_mkfs_time_hi;
1424 __u8 s_lastcheck_hi;
1425 __u8 s_first_error_time_hi;
1426 __u8 s_last_error_time_hi;
1427 __u8 s_first_error_errcode;
1428 __u8 s_last_error_errcode;
1429 __le16 s_encoding; /* Filename charset encoding */
1430 __le16 s_encoding_flags; /* Filename charset encoding flags */
1431 __le32 s_reserved[95]; /* Padding to the end of the block */
1432 __le32 s_checksum; /* crc32c(superblock) */
1435 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1439 #ifdef CONFIG_FS_ENCRYPTION
1440 #define DUMMY_ENCRYPTION_ENABLED(sbi) ((sbi)->s_dummy_enc_policy.policy != NULL)
1442 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1445 /* Number of quota types we support */
1446 #define EXT4_MAXQUOTAS 3
1448 #define EXT4_ENC_UTF8_12_1 1
1451 * fourth extended-fs super-block data in memory
1453 struct ext4_sb_info {
1454 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1455 unsigned long s_inodes_per_block;/* Number of inodes per block */
1456 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1457 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1458 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1459 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1460 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1461 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1462 ext4_group_t s_groups_count; /* Number of groups in the fs */
1463 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1464 unsigned long s_overhead; /* # of fs overhead clusters */
1465 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1466 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1467 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1468 struct buffer_head * s_sbh; /* Buffer containing the super block */
1469 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1470 struct buffer_head * __rcu *s_group_desc;
1471 unsigned int s_mount_opt;
1472 unsigned int s_mount_opt2;
1473 unsigned long s_mount_flags;
1474 unsigned int s_def_mount_opt;
1475 ext4_fsblk_t s_sb_block;
1476 atomic64_t s_resv_clusters;
1479 unsigned short s_mount_state;
1480 unsigned short s_pad;
1481 int s_addr_per_block_bits;
1482 int s_desc_per_block_bits;
1485 unsigned int s_inode_readahead_blks;
1486 unsigned int s_inode_goal;
1488 int s_def_hash_version;
1489 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1490 struct percpu_counter s_freeclusters_counter;
1491 struct percpu_counter s_freeinodes_counter;
1492 struct percpu_counter s_dirs_counter;
1493 struct percpu_counter s_dirtyclusters_counter;
1494 struct percpu_counter s_sra_exceeded_retry_limit;
1495 struct blockgroup_lock *s_blockgroup_lock;
1496 struct proc_dir_entry *s_proc;
1497 struct kobject s_kobj;
1498 struct completion s_kobj_unregister;
1499 struct super_block *s_sb;
1502 struct journal_s *s_journal;
1503 struct list_head s_orphan;
1504 struct mutex s_orphan_lock;
1505 unsigned long s_ext4_flags; /* Ext4 superblock flags */
1506 unsigned long s_commit_interval;
1507 u32 s_max_batch_time;
1508 u32 s_min_batch_time;
1509 struct block_device *s_journal_bdev;
1511 /* Names of quota files with journalled quota */
1512 char __rcu *s_qf_names[EXT4_MAXQUOTAS];
1513 int s_jquota_fmt; /* Format of quota to use */
1515 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1516 struct ext4_system_blocks __rcu *s_system_blks;
1518 #ifdef EXTENTS_STATS
1519 /* ext4 extents stats */
1520 unsigned long s_ext_min;
1521 unsigned long s_ext_max;
1522 unsigned long s_depth_max;
1523 spinlock_t s_ext_stats_lock;
1524 unsigned long s_ext_blocks;
1525 unsigned long s_ext_extents;
1528 /* for buddy allocator */
1529 struct ext4_group_info ** __rcu *s_group_info;
1530 struct inode *s_buddy_cache;
1531 spinlock_t s_md_lock;
1532 unsigned short *s_mb_offsets;
1533 unsigned int *s_mb_maxs;
1534 unsigned int s_group_info_size;
1535 unsigned int s_mb_free_pending;
1536 struct list_head s_freed_data_list; /* List of blocks to be freed
1537 after commit completed */
1538 struct rb_root s_mb_avg_fragment_size_root;
1539 rwlock_t s_mb_rb_lock;
1540 struct list_head *s_mb_largest_free_orders;
1541 rwlock_t *s_mb_largest_free_orders_locks;
1544 unsigned long s_stripe;
1545 unsigned int s_mb_max_linear_groups;
1546 unsigned int s_mb_stream_request;
1547 unsigned int s_mb_max_to_scan;
1548 unsigned int s_mb_min_to_scan;
1549 unsigned int s_mb_stats;
1550 unsigned int s_mb_order2_reqs;
1551 unsigned int s_mb_group_prealloc;
1552 unsigned int s_mb_max_inode_prealloc;
1553 unsigned int s_max_dir_size_kb;
1554 /* where last allocation was done - for stream allocation */
1555 unsigned long s_mb_last_group;
1556 unsigned long s_mb_last_start;
1557 unsigned int s_mb_prefetch;
1558 unsigned int s_mb_prefetch_limit;
1560 /* stats for buddy allocator */
1561 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1562 atomic_t s_bal_success; /* we found long enough chunks */
1563 atomic_t s_bal_allocated; /* in blocks */
1564 atomic_t s_bal_ex_scanned; /* total extents scanned */
1565 atomic_t s_bal_groups_scanned; /* number of groups scanned */
1566 atomic_t s_bal_goals; /* goal hits */
1567 atomic_t s_bal_breaks; /* too long searches */
1568 atomic_t s_bal_2orders; /* 2^order hits */
1569 atomic_t s_bal_cr0_bad_suggestions;
1570 atomic_t s_bal_cr1_bad_suggestions;
1571 atomic64_t s_bal_cX_groups_considered[4];
1572 atomic64_t s_bal_cX_hits[4];
1573 atomic64_t s_bal_cX_failed[4]; /* cX loop didn't find blocks */
1574 atomic_t s_mb_buddies_generated; /* number of buddies generated */
1575 atomic64_t s_mb_generation_time;
1576 atomic_t s_mb_lost_chunks;
1577 atomic_t s_mb_preallocated;
1578 atomic_t s_mb_discarded;
1579 atomic_t s_lock_busy;
1581 /* locality groups */
1582 struct ext4_locality_group __percpu *s_locality_groups;
1584 /* for write statistics */
1585 unsigned long s_sectors_written_start;
1586 u64 s_kbytes_written;
1588 /* the size of zero-out chunk */
1589 unsigned int s_extent_max_zeroout_kb;
1591 unsigned int s_log_groups_per_flex;
1592 struct flex_groups * __rcu *s_flex_groups;
1593 ext4_group_t s_flex_groups_allocated;
1595 /* workqueue for reserved extent conversions (buffered io) */
1596 struct workqueue_struct *rsv_conversion_wq;
1598 /* timer for periodic error stats printing */
1599 struct timer_list s_err_report;
1601 /* Lazy inode table initialization info */
1602 struct ext4_li_request *s_li_request;
1603 /* Wait multiplier for lazy initialization thread */
1604 unsigned int s_li_wait_mult;
1606 /* Kernel thread for multiple mount protection */
1607 struct task_struct *s_mmp_tsk;
1609 /* record the last minlen when FITRIM is called. */
1610 atomic_t s_last_trim_minblks;
1612 /* Reference to checksum algorithm driver via cryptoapi */
1613 struct crypto_shash *s_chksum_driver;
1615 /* Precomputed FS UUID checksum for seeding other checksums */
1618 /* Reclaim extents from extent status tree */
1619 struct shrinker s_es_shrinker;
1620 struct list_head s_es_list; /* List of inodes with reclaimable extents */
1622 struct ext4_es_stats s_es_stats;
1623 struct mb_cache *s_ea_block_cache;
1624 struct mb_cache *s_ea_inode_cache;
1625 spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1627 /* Ratelimit ext4 messages. */
1628 struct ratelimit_state s_err_ratelimit_state;
1629 struct ratelimit_state s_warning_ratelimit_state;
1630 struct ratelimit_state s_msg_ratelimit_state;
1631 atomic_t s_warning_count;
1632 atomic_t s_msg_count;
1634 /* Encryption policy for '-o test_dummy_encryption' */
1635 struct fscrypt_dummy_policy s_dummy_enc_policy;
1638 * Barrier between writepages ops and changing any inode's JOURNAL_DATA
1641 struct percpu_rw_semaphore s_writepages_rwsem;
1642 struct dax_device *s_daxdev;
1643 #ifdef CONFIG_EXT4_DEBUG
1644 unsigned long s_simulate_fail;
1646 /* Record the errseq of the backing block device */
1647 errseq_t s_bdev_wb_err;
1648 spinlock_t s_bdev_wb_lock;
1650 /* Information about errors that happened during this mount */
1651 spinlock_t s_error_lock;
1652 int s_add_error_count;
1653 int s_first_error_code;
1654 __u32 s_first_error_line;
1655 __u32 s_first_error_ino;
1656 __u64 s_first_error_block;
1657 const char *s_first_error_func;
1658 time64_t s_first_error_time;
1659 int s_last_error_code;
1660 __u32 s_last_error_line;
1661 __u32 s_last_error_ino;
1662 __u64 s_last_error_block;
1663 const char *s_last_error_func;
1664 time64_t s_last_error_time;
1666 * If we are in a context where we cannot update error information in
1667 * the on-disk superblock, we queue this work to do it.
1669 struct work_struct s_error_work;
1671 /* Ext4 fast commit stuff */
1672 atomic_t s_fc_subtid;
1673 atomic_t s_fc_ineligible_updates;
1675 * After commit starts, the main queue gets locked, and the further
1676 * updates get added in the staging queue.
1679 #define FC_Q_STAGING 1
1680 struct list_head s_fc_q[2]; /* Inodes staged for fast commit
1681 * that have data changes in them.
1683 struct list_head s_fc_dentry_q[2]; /* directory entry updates */
1684 unsigned int s_fc_bytes;
1686 * Main fast commit lock. This lock protects accesses to the
1688 * ei->i_fc_list, s_fc_dentry_q, s_fc_q, s_fc_bytes, s_fc_bh.
1690 spinlock_t s_fc_lock;
1691 struct buffer_head *s_fc_bh;
1692 struct ext4_fc_stats s_fc_stats;
1693 u64 s_fc_avg_commit_time;
1694 #ifdef CONFIG_EXT4_DEBUG
1695 int s_fc_debug_max_replay;
1697 struct ext4_fc_replay_state s_fc_replay_state;
1700 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1702 return sb->s_fs_info;
1704 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1706 return container_of(inode, struct ext4_inode_info, vfs_inode);
1709 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1711 return ino == EXT4_ROOT_INO ||
1712 (ino >= EXT4_FIRST_INO(sb) &&
1713 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1717 * Returns: sbi->field[index]
1718 * Used to access an array element from the following sbi fields which require
1719 * rcu protection to avoid dereferencing an invalid pointer due to reassignment
1724 #define sbi_array_rcu_deref(sbi, field, index) \
1726 typeof(*((sbi)->field)) _v; \
1728 _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index]; \
1729 rcu_read_unlock(); \
1734 * run-time mount flags
1737 EXT4_MF_MNTDIR_SAMPLED,
1738 EXT4_MF_FS_ABORTED, /* Fatal error detected */
1739 EXT4_MF_FC_INELIGIBLE, /* Fast commit ineligible */
1740 EXT4_MF_FC_COMMITTING /* File system underoing a fast
1745 static inline void ext4_set_mount_flag(struct super_block *sb, int bit)
1747 set_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1750 static inline void ext4_clear_mount_flag(struct super_block *sb, int bit)
1752 clear_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1755 static inline int ext4_test_mount_flag(struct super_block *sb, int bit)
1757 return test_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1762 * Simulate_fail codes
1764 #define EXT4_SIM_BBITMAP_EIO 1
1765 #define EXT4_SIM_BBITMAP_CRC 2
1766 #define EXT4_SIM_IBITMAP_EIO 3
1767 #define EXT4_SIM_IBITMAP_CRC 4
1768 #define EXT4_SIM_INODE_EIO 5
1769 #define EXT4_SIM_INODE_CRC 6
1770 #define EXT4_SIM_DIRBLOCK_EIO 7
1771 #define EXT4_SIM_DIRBLOCK_CRC 8
1773 static inline bool ext4_simulate_fail(struct super_block *sb,
1776 #ifdef CONFIG_EXT4_DEBUG
1777 struct ext4_sb_info *sbi = EXT4_SB(sb);
1779 if (unlikely(sbi->s_simulate_fail == code)) {
1780 sbi->s_simulate_fail = 0;
1787 static inline void ext4_simulate_fail_bh(struct super_block *sb,
1788 struct buffer_head *bh,
1791 if (!IS_ERR(bh) && ext4_simulate_fail(sb, code))
1792 clear_buffer_uptodate(bh);
1796 * Error number codes for s_{first,last}_error_errno
1798 * Linux errno numbers are architecture specific, so we need to translate
1799 * them into something which is architecture independent. We don't define
1800 * codes for all errno's; just the ones which are most likely to be the cause
1801 * of an ext4_error() call.
1803 #define EXT4_ERR_UNKNOWN 1
1804 #define EXT4_ERR_EIO 2
1805 #define EXT4_ERR_ENOMEM 3
1806 #define EXT4_ERR_EFSBADCRC 4
1807 #define EXT4_ERR_EFSCORRUPTED 5
1808 #define EXT4_ERR_ENOSPC 6
1809 #define EXT4_ERR_ENOKEY 7
1810 #define EXT4_ERR_EROFS 8
1811 #define EXT4_ERR_EFBIG 9
1812 #define EXT4_ERR_EEXIST 10
1813 #define EXT4_ERR_ERANGE 11
1814 #define EXT4_ERR_EOVERFLOW 12
1815 #define EXT4_ERR_EBUSY 13
1816 #define EXT4_ERR_ENOTDIR 14
1817 #define EXT4_ERR_ENOTEMPTY 15
1818 #define EXT4_ERR_ESHUTDOWN 16
1819 #define EXT4_ERR_EFAULT 17
1822 * Inode dynamic state flags
1825 EXT4_STATE_JDATA, /* journaled data exists */
1826 EXT4_STATE_NEW, /* inode is newly created */
1827 EXT4_STATE_XATTR, /* has in-inode xattrs */
1828 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1829 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1830 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1831 EXT4_STATE_NEWENTRY, /* File just added to dir */
1832 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1833 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
1834 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */
1835 EXT4_STATE_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */
1836 EXT4_STATE_FC_COMMITTING, /* Fast commit ongoing */
1839 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1840 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1842 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1844 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1846 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1848 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1850 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1853 /* Add these declarations here only so that these functions can be
1854 * found by name. Otherwise, they are very hard to locate. */
1855 static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1856 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1857 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1858 EXT4_INODE_BIT_FNS(flag, flags, 0)
1860 /* Add these declarations here only so that these functions can be
1861 * found by name. Otherwise, they are very hard to locate. */
1862 static inline int ext4_test_inode_state(struct inode *inode, int bit);
1863 static inline void ext4_set_inode_state(struct inode *inode, int bit);
1864 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1865 #if (BITS_PER_LONG < 64)
1866 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1868 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1870 (ei)->i_state_flags = 0;
1873 EXT4_INODE_BIT_FNS(state, flags, 32)
1875 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1877 /* We depend on the fact that callers will set i_flags */
1881 /* Assume that user mode programs are passing in an ext4fs superblock, not
1882 * a kernel struct super_block. This will allow us to call the feature-test
1883 * macros from user land. */
1884 #define EXT4_SB(sb) (sb)
1887 static inline bool ext4_verity_in_progress(struct inode *inode)
1889 return IS_ENABLED(CONFIG_FS_VERITY) &&
1890 ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS);
1893 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1896 * Codes for operating systems
1898 #define EXT4_OS_LINUX 0
1899 #define EXT4_OS_HURD 1
1900 #define EXT4_OS_MASIX 2
1901 #define EXT4_OS_FREEBSD 3
1902 #define EXT4_OS_LITES 4
1907 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1908 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1910 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1912 #define EXT4_GOOD_OLD_INODE_SIZE 128
1914 #define EXT4_EXTRA_TIMESTAMP_MAX (((s64)1 << 34) - 1 + S32_MIN)
1915 #define EXT4_NON_EXTRA_TIMESTAMP_MAX S32_MAX
1916 #define EXT4_TIMESTAMP_MIN S32_MIN
1919 * Feature set definitions
1922 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1923 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1924 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1925 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1926 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1927 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1928 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
1930 * The reason why "FAST_COMMIT" is a compat feature is that, FS becomes
1931 * incompatible only if fast commit blocks are present in the FS. Since we
1932 * clear the journal (and thus the fast commit blocks), we don't mark FS as
1933 * incompatible. We also have a JBD2 incompat feature, which gets set when
1934 * there are fast commit blocks present in the journal.
1936 #define EXT4_FEATURE_COMPAT_FAST_COMMIT 0x0400
1937 #define EXT4_FEATURE_COMPAT_STABLE_INODES 0x0800
1939 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1940 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1941 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1942 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1943 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1944 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1945 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1946 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1947 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1949 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1950 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1951 * all other data structures' checksums. However, the METADATA_CSUM and
1952 * GDT_CSUM bits are mutually exclusive.
1954 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
1955 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000
1956 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000
1957 #define EXT4_FEATURE_RO_COMPAT_VERITY 0x8000
1959 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1960 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1961 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1962 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1963 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1964 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1965 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1966 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1967 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1968 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1969 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1970 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000
1971 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
1972 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
1973 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000
1974 #define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000
1976 extern void ext4_update_dynamic_rev(struct super_block *sb);
1978 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
1979 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1981 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1982 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
1984 static inline void ext4_set_feature_##name(struct super_block *sb) \
1986 ext4_update_dynamic_rev(sb); \
1987 EXT4_SB(sb)->s_es->s_feature_compat |= \
1988 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1990 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1992 EXT4_SB(sb)->s_es->s_feature_compat &= \
1993 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1996 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
1997 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1999 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
2000 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
2002 static inline void ext4_set_feature_##name(struct super_block *sb) \
2004 ext4_update_dynamic_rev(sb); \
2005 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
2006 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
2008 static inline void ext4_clear_feature_##name(struct super_block *sb) \
2010 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
2011 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
2014 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
2015 static inline bool ext4_has_feature_##name(struct super_block *sb) \
2017 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
2018 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
2020 static inline void ext4_set_feature_##name(struct super_block *sb) \
2022 ext4_update_dynamic_rev(sb); \
2023 EXT4_SB(sb)->s_es->s_feature_incompat |= \
2024 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
2026 static inline void ext4_clear_feature_##name(struct super_block *sb) \
2028 EXT4_SB(sb)->s_es->s_feature_incompat &= \
2029 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
2032 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC)
2033 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES)
2034 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL)
2035 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR)
2036 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE)
2037 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX)
2038 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2)
2039 EXT4_FEATURE_COMPAT_FUNCS(fast_commit, FAST_COMMIT)
2040 EXT4_FEATURE_COMPAT_FUNCS(stable_inodes, STABLE_INODES)
2042 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER)
2043 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE)
2044 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR)
2045 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE)
2046 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM)
2047 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK)
2048 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE)
2049 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA)
2050 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC)
2051 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM)
2052 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY)
2053 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT)
2054 EXT4_FEATURE_RO_COMPAT_FUNCS(verity, VERITY)
2056 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION)
2057 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE)
2058 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER)
2059 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV)
2060 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG)
2061 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS)
2062 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT)
2063 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP)
2064 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG)
2065 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE)
2066 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA)
2067 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED)
2068 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR)
2069 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA)
2070 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT)
2071 EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD)
2073 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
2074 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
2075 EXT4_FEATURE_INCOMPAT_META_BG)
2076 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2077 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2078 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
2080 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
2081 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
2082 EXT4_FEATURE_INCOMPAT_RECOVER| \
2083 EXT4_FEATURE_INCOMPAT_META_BG)
2084 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2085 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2086 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
2088 #define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
2089 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
2090 EXT4_FEATURE_INCOMPAT_RECOVER| \
2091 EXT4_FEATURE_INCOMPAT_META_BG| \
2092 EXT4_FEATURE_INCOMPAT_EXTENTS| \
2093 EXT4_FEATURE_INCOMPAT_64BIT| \
2094 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
2095 EXT4_FEATURE_INCOMPAT_EA_INODE| \
2096 EXT4_FEATURE_INCOMPAT_MMP | \
2097 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
2098 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
2099 EXT4_FEATURE_INCOMPAT_CASEFOLD | \
2100 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \
2101 EXT4_FEATURE_INCOMPAT_LARGEDIR)
2102 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2103 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2104 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
2105 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
2106 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
2107 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
2108 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
2109 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
2110 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
2111 EXT4_FEATURE_RO_COMPAT_QUOTA |\
2112 EXT4_FEATURE_RO_COMPAT_PROJECT |\
2113 EXT4_FEATURE_RO_COMPAT_VERITY)
2115 #define EXTN_FEATURE_FUNCS(ver) \
2116 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
2118 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
2119 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
2121 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
2123 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
2124 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
2126 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
2128 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
2129 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
2132 EXTN_FEATURE_FUNCS(2)
2133 EXTN_FEATURE_FUNCS(3)
2134 EXTN_FEATURE_FUNCS(4)
2136 static inline bool ext4_has_compat_features(struct super_block *sb)
2138 return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
2140 static inline bool ext4_has_ro_compat_features(struct super_block *sb)
2142 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
2144 static inline bool ext4_has_incompat_features(struct super_block *sb)
2146 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
2152 #define EXT4_FLAGS_RESIZING 0
2153 #define EXT4_FLAGS_SHUTDOWN 1
2154 #define EXT4_FLAGS_BDEV_IS_DAX 2
2156 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi)
2158 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags);
2163 * Default values for user and/or group using reserved blocks
2165 #define EXT4_DEF_RESUID 0
2166 #define EXT4_DEF_RESGID 0
2169 * Default project ID
2171 #define EXT4_DEF_PROJID 0
2173 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
2176 * Default mount options
2178 #define EXT4_DEFM_DEBUG 0x0001
2179 #define EXT4_DEFM_BSDGROUPS 0x0002
2180 #define EXT4_DEFM_XATTR_USER 0x0004
2181 #define EXT4_DEFM_ACL 0x0008
2182 #define EXT4_DEFM_UID16 0x0010
2183 #define EXT4_DEFM_JMODE 0x0060
2184 #define EXT4_DEFM_JMODE_DATA 0x0020
2185 #define EXT4_DEFM_JMODE_ORDERED 0x0040
2186 #define EXT4_DEFM_JMODE_WBACK 0x0060
2187 #define EXT4_DEFM_NOBARRIER 0x0100
2188 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
2189 #define EXT4_DEFM_DISCARD 0x0400
2190 #define EXT4_DEFM_NODELALLOC 0x0800
2193 * Default journal batch times
2195 #define EXT4_DEF_MIN_BATCH_TIME 0
2196 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
2199 * Minimum number of groups in a flexgroup before we separate out
2200 * directories into the first block group of a flexgroup
2202 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
2205 * Structure of a directory entry
2207 #define EXT4_NAME_LEN 255
2209 struct ext4_dir_entry {
2210 __le32 inode; /* Inode number */
2211 __le16 rec_len; /* Directory entry length */
2212 __le16 name_len; /* Name length */
2213 char name[EXT4_NAME_LEN]; /* File name */
2218 * Encrypted Casefolded entries require saving the hash on disk. This structure
2219 * followed ext4_dir_entry_2's name[name_len] at the next 4 byte aligned
2222 struct ext4_dir_entry_hash {
2228 * The new version of the directory entry. Since EXT4 structures are
2229 * stored in intel byte order, and the name_len field could never be
2230 * bigger than 255 chars, it's safe to reclaim the extra byte for the
2233 struct ext4_dir_entry_2 {
2234 __le32 inode; /* Inode number */
2235 __le16 rec_len; /* Directory entry length */
2236 __u8 name_len; /* Name length */
2237 __u8 file_type; /* See file type macros EXT4_FT_* below */
2238 char name[EXT4_NAME_LEN]; /* File name */
2242 * Access the hashes at the end of ext4_dir_entry_2
2244 #define EXT4_DIRENT_HASHES(entry) \
2245 ((struct ext4_dir_entry_hash *) \
2246 (((void *)(entry)) + \
2247 ((8 + (entry)->name_len + EXT4_DIR_ROUND) & ~EXT4_DIR_ROUND)))
2248 #define EXT4_DIRENT_HASH(entry) le32_to_cpu(EXT4_DIRENT_HASHES(de)->hash)
2249 #define EXT4_DIRENT_MINOR_HASH(entry) \
2250 le32_to_cpu(EXT4_DIRENT_HASHES(de)->minor_hash)
2252 static inline bool ext4_hash_in_dirent(const struct inode *inode)
2254 return IS_CASEFOLDED(inode) && IS_ENCRYPTED(inode);
2258 * This is a bogus directory entry at the end of each leaf block that
2259 * records checksums.
2261 struct ext4_dir_entry_tail {
2262 __le32 det_reserved_zero1; /* Pretend to be unused */
2263 __le16 det_rec_len; /* 12 */
2264 __u8 det_reserved_zero2; /* Zero name length */
2265 __u8 det_reserved_ft; /* 0xDE, fake file type */
2266 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
2269 #define EXT4_DIRENT_TAIL(block, blocksize) \
2270 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
2272 sizeof(struct ext4_dir_entry_tail))))
2275 * Ext4 directory file types. Only the low 3 bits are used. The
2276 * other bits are reserved for now.
2278 #define EXT4_FT_UNKNOWN 0
2279 #define EXT4_FT_REG_FILE 1
2280 #define EXT4_FT_DIR 2
2281 #define EXT4_FT_CHRDEV 3
2282 #define EXT4_FT_BLKDEV 4
2283 #define EXT4_FT_FIFO 5
2284 #define EXT4_FT_SOCK 6
2285 #define EXT4_FT_SYMLINK 7
2287 #define EXT4_FT_MAX 8
2289 #define EXT4_FT_DIR_CSUM 0xDE
2292 * EXT4_DIR_PAD defines the directory entries boundaries
2294 * NOTE: It must be a multiple of 4
2296 #define EXT4_DIR_PAD 4
2297 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
2298 #define EXT4_MAX_REC_LEN ((1<<16)-1)
2301 * The rec_len is dependent on the type of directory. Directories that are
2302 * casefolded and encrypted need to store the hash as well, so we add room for
2303 * ext4_extended_dir_entry_2. For all entries related to '.' or '..' you should
2304 * pass NULL for dir, as those entries do not use the extra fields.
2306 static inline unsigned int ext4_dir_rec_len(__u8 name_len,
2307 const struct inode *dir)
2309 int rec_len = (name_len + 8 + EXT4_DIR_ROUND);
2311 if (dir && ext4_hash_in_dirent(dir))
2312 rec_len += sizeof(struct ext4_dir_entry_hash);
2313 return (rec_len & ~EXT4_DIR_ROUND);
2317 * If we ever get support for fs block sizes > page_size, we'll need
2318 * to remove the #if statements in the next two functions...
2320 static inline unsigned int
2321 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
2323 unsigned len = le16_to_cpu(dlen);
2325 #if (PAGE_SIZE >= 65536)
2326 if (len == EXT4_MAX_REC_LEN || len == 0)
2328 return (len & 65532) | ((len & 3) << 16);
2334 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
2336 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
2338 #if (PAGE_SIZE >= 65536)
2340 return cpu_to_le16(len);
2341 if (len == blocksize) {
2342 if (blocksize == 65536)
2343 return cpu_to_le16(EXT4_MAX_REC_LEN);
2345 return cpu_to_le16(0);
2347 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
2349 return cpu_to_le16(len);
2354 * Hash Tree Directory indexing
2355 * (c) Daniel Phillips, 2001
2358 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
2359 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2360 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \
2361 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir)))
2362 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2364 /* Legal values for the dx_root hash_version field: */
2366 #define DX_HASH_LEGACY 0
2367 #define DX_HASH_HALF_MD4 1
2368 #define DX_HASH_TEA 2
2369 #define DX_HASH_LEGACY_UNSIGNED 3
2370 #define DX_HASH_HALF_MD4_UNSIGNED 4
2371 #define DX_HASH_TEA_UNSIGNED 5
2372 #define DX_HASH_SIPHASH 6
2374 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2375 const void *address, unsigned int length)
2378 struct shash_desc shash;
2382 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2384 desc.shash.tfm = sbi->s_chksum_driver;
2385 *(u32 *)desc.ctx = crc;
2387 BUG_ON(crypto_shash_update(&desc.shash, address, length));
2389 return *(u32 *)desc.ctx;
2394 /* hash info structure used by the directory hash */
2404 /* 32 and 64 bit signed EOF for dx directories */
2405 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
2406 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
2410 * Control parameters used by ext4_htree_next_block
2412 #define HASH_NB_ALWAYS 1
2414 struct ext4_filename {
2415 const struct qstr *usr_fname;
2416 struct fscrypt_str disk_name;
2417 struct dx_hash_info hinfo;
2418 #ifdef CONFIG_FS_ENCRYPTION
2419 struct fscrypt_str crypto_buf;
2421 #ifdef CONFIG_UNICODE
2422 struct fscrypt_str cf_name;
2426 #define fname_name(p) ((p)->disk_name.name)
2427 #define fname_usr_name(p) ((p)->usr_fname->name)
2428 #define fname_len(p) ((p)->disk_name.len)
2431 * Describe an inode's exact location on disk and in memory
2435 struct buffer_head *bh;
2436 unsigned long offset;
2437 ext4_group_t block_group;
2440 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2442 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2445 static inline bool ext4_is_quota_file(struct inode *inode)
2447 return IS_NOQUOTA(inode) &&
2448 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL);
2452 * This structure is stuffed into the struct file's private_data field
2453 * for directories. It is where we put information so that we can do
2454 * readdir operations in hash tree order.
2456 struct dir_private_info {
2457 struct rb_root root;
2458 struct rb_node *curr_node;
2459 struct fname *extra_fname;
2462 __u32 curr_minor_hash;
2466 /* calculate the first block number of the group */
2467 static inline ext4_fsblk_t
2468 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2470 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2471 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2475 * Special error return code only used by dx_probe() and its callers.
2477 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
2479 /* htree levels for ext4 */
2480 #define EXT4_HTREE_LEVEL_COMPAT 2
2481 #define EXT4_HTREE_LEVEL 3
2483 static inline int ext4_dir_htree_level(struct super_block *sb)
2485 return ext4_has_feature_largedir(sb) ?
2486 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT;
2490 * Timeout and state flag for lazy initialization inode thread.
2492 #define EXT4_DEF_LI_WAIT_MULT 10
2493 #define EXT4_DEF_LI_MAX_START_DELAY 5
2494 #define EXT4_LAZYINIT_QUIT 0x0001
2495 #define EXT4_LAZYINIT_RUNNING 0x0002
2498 * Lazy inode table initialization info
2500 struct ext4_lazy_init {
2501 unsigned long li_state;
2502 struct list_head li_request_list;
2503 struct mutex li_list_mtx;
2507 EXT4_LI_MODE_PREFETCH_BBITMAP,
2508 EXT4_LI_MODE_ITABLE,
2511 struct ext4_li_request {
2512 struct super_block *lr_super;
2513 enum ext4_li_mode lr_mode;
2514 ext4_group_t lr_first_not_zeroed;
2515 ext4_group_t lr_next_group;
2516 struct list_head lr_request;
2517 unsigned long lr_next_sched;
2518 unsigned long lr_timeout;
2521 struct ext4_features {
2522 struct kobject f_kobj;
2523 struct completion f_kobj_unregister;
2527 * This structure will be used for multiple mount protection. It will be
2528 * written into the block number saved in the s_mmp_block field in the
2529 * superblock. Programs that check MMP should assume that if
2530 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2531 * to use the filesystem, regardless of how old the timestamp is.
2533 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
2534 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2535 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
2536 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
2539 __le32 mmp_magic; /* Magic number for MMP */
2540 __le32 mmp_seq; /* Sequence no. updated periodically */
2543 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2544 * purposes and do not affect the correctness of the algorithm
2546 __le64 mmp_time; /* Time last updated */
2547 char mmp_nodename[64]; /* Node which last updated MMP block */
2548 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
2551 * mmp_check_interval is used to verify if the MMP block has been
2552 * updated on the block device. The value is updated based on the
2553 * maximum time to write the MMP block during an update cycle.
2555 __le16 mmp_check_interval;
2558 __le32 mmp_pad2[226];
2559 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
2562 /* arguments passed to the mmp thread */
2564 struct buffer_head *bh; /* bh from initial read_mmp_block() */
2565 struct super_block *sb; /* super block of the fs */
2569 * Check interval multiplier
2570 * The MMP block is written every update interval and initially checked every
2571 * update interval x the multiplier (the value is then adapted based on the
2572 * write latency). The reason is that writes can be delayed under load and we
2573 * don't want readers to incorrectly assume that the filesystem is no longer
2576 #define EXT4_MMP_CHECK_MULT 2UL
2579 * Minimum interval for MMP checking in seconds.
2581 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
2584 * Maximum interval for MMP checking in seconds.
2586 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
2589 * Function prototypes
2593 * Ok, these declarations are also in <linux/kernel.h> but none of the
2594 * ext4 source programs needs to include it so they are duplicated here.
2596 # define NORET_TYPE /**/
2597 # define ATTRIB_NORET __attribute__((noreturn))
2598 # define NORET_AND noreturn,
2601 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2602 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2603 struct ext4_group_desc *gdp,
2604 struct buffer_head *bh, int sz);
2605 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2606 struct ext4_group_desc *gdp,
2607 struct buffer_head *bh, int sz);
2608 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2609 struct ext4_group_desc *gdp,
2610 struct buffer_head *bh);
2611 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2612 struct ext4_group_desc *gdp,
2613 struct buffer_head *bh);
2616 extern void ext4_get_group_no_and_offset(struct super_block *sb,
2617 ext4_fsblk_t blocknr,
2618 ext4_group_t *blockgrpp,
2619 ext4_grpblk_t *offsetp);
2620 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2621 ext4_fsblk_t block);
2623 extern unsigned int ext4_block_group(struct super_block *sb,
2624 ext4_fsblk_t blocknr);
2625 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2626 ext4_fsblk_t blocknr);
2627 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2628 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2629 ext4_group_t group);
2630 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2633 unsigned long *count,
2635 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2636 s64 nclusters, unsigned int flags);
2637 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2638 extern void ext4_check_blocks_bitmap(struct super_block *);
2639 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2640 ext4_group_t block_group,
2641 struct buffer_head ** bh);
2642 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2644 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2645 ext4_group_t block_group,
2646 bool ignore_locked);
2647 extern int ext4_wait_block_bitmap(struct super_block *sb,
2648 ext4_group_t block_group,
2649 struct buffer_head *bh);
2650 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2651 ext4_group_t block_group);
2652 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2653 ext4_group_t block_group,
2654 struct ext4_group_desc *gdp);
2655 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2657 #ifdef CONFIG_UNICODE
2658 extern int ext4_fname_setup_ci_filename(struct inode *dir,
2659 const struct qstr *iname,
2660 struct ext4_filename *fname);
2663 #ifdef CONFIG_FS_ENCRYPTION
2664 static inline void ext4_fname_from_fscrypt_name(struct ext4_filename *dst,
2665 const struct fscrypt_name *src)
2667 memset(dst, 0, sizeof(*dst));
2669 dst->usr_fname = src->usr_fname;
2670 dst->disk_name = src->disk_name;
2671 dst->hinfo.hash = src->hash;
2672 dst->hinfo.minor_hash = src->minor_hash;
2673 dst->crypto_buf = src->crypto_buf;
2676 static inline int ext4_fname_setup_filename(struct inode *dir,
2677 const struct qstr *iname,
2679 struct ext4_filename *fname)
2681 struct fscrypt_name name;
2684 err = fscrypt_setup_filename(dir, iname, lookup, &name);
2688 ext4_fname_from_fscrypt_name(fname, &name);
2690 #ifdef CONFIG_UNICODE
2691 err = ext4_fname_setup_ci_filename(dir, iname, fname);
2696 static inline int ext4_fname_prepare_lookup(struct inode *dir,
2697 struct dentry *dentry,
2698 struct ext4_filename *fname)
2700 struct fscrypt_name name;
2703 err = fscrypt_prepare_lookup(dir, dentry, &name);
2707 ext4_fname_from_fscrypt_name(fname, &name);
2709 #ifdef CONFIG_UNICODE
2710 err = ext4_fname_setup_ci_filename(dir, &dentry->d_name, fname);
2715 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2717 struct fscrypt_name name;
2719 name.crypto_buf = fname->crypto_buf;
2720 fscrypt_free_filename(&name);
2722 fname->crypto_buf.name = NULL;
2723 fname->usr_fname = NULL;
2724 fname->disk_name.name = NULL;
2726 #ifdef CONFIG_UNICODE
2727 kfree(fname->cf_name.name);
2728 fname->cf_name.name = NULL;
2731 #else /* !CONFIG_FS_ENCRYPTION */
2732 static inline int ext4_fname_setup_filename(struct inode *dir,
2733 const struct qstr *iname,
2735 struct ext4_filename *fname)
2738 fname->usr_fname = iname;
2739 fname->disk_name.name = (unsigned char *) iname->name;
2740 fname->disk_name.len = iname->len;
2742 #ifdef CONFIG_UNICODE
2743 err = ext4_fname_setup_ci_filename(dir, iname, fname);
2749 static inline int ext4_fname_prepare_lookup(struct inode *dir,
2750 struct dentry *dentry,
2751 struct ext4_filename *fname)
2753 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname);
2756 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2758 #ifdef CONFIG_UNICODE
2759 kfree(fname->cf_name.name);
2760 fname->cf_name.name = NULL;
2763 #endif /* !CONFIG_FS_ENCRYPTION */
2766 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2768 struct ext4_dir_entry_2 *,
2769 struct buffer_head *, char *, int,
2771 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2772 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2773 (de), (bh), (buf), (size), (offset)))
2774 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2776 struct ext4_dir_entry_2 *dirent,
2777 struct fscrypt_str *ent_name);
2778 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2779 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2780 struct buffer_head *bh,
2781 void *buf, int buf_size,
2782 struct ext4_filename *fname,
2783 struct ext4_dir_entry_2 **dest_de);
2784 void ext4_insert_dentry(struct inode *dir, struct inode *inode,
2785 struct ext4_dir_entry_2 *de,
2787 struct ext4_filename *fname);
2788 static inline void ext4_update_dx_flag(struct inode *inode)
2790 if (!ext4_has_feature_dir_index(inode->i_sb) &&
2791 ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) {
2792 /* ext4_iget() should have caught this... */
2793 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb));
2794 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2797 static const unsigned char ext4_filetype_table[] = {
2798 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2801 static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2803 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2806 return ext4_filetype_table[filetype];
2808 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2809 void *buf, int buf_size);
2812 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2815 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len,
2816 struct dx_hash_info *hinfo);
2819 extern int ext4_mark_inode_used(struct super_block *sb, int ino);
2820 extern struct inode *__ext4_new_inode(struct user_namespace *, handle_t *,
2821 struct inode *, umode_t,
2822 const struct qstr *qstr, __u32 goal,
2823 uid_t *owner, __u32 i_flags,
2824 int handle_type, unsigned int line_no,
2827 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \
2828 __ext4_new_inode(&init_user_ns, (handle), (dir), (mode), (qstr), \
2829 (goal), (owner), i_flags, 0, 0, 0)
2830 #define ext4_new_inode_start_handle(mnt_userns, dir, mode, qstr, goal, owner, \
2832 __ext4_new_inode((mnt_userns), NULL, (dir), (mode), (qstr), (goal), (owner), \
2833 0, (type), __LINE__, (nblocks))
2836 extern void ext4_free_inode(handle_t *, struct inode *);
2837 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2838 extern unsigned long ext4_count_free_inodes(struct super_block *);
2839 extern unsigned long ext4_count_dirs(struct super_block *);
2840 extern void ext4_check_inodes_bitmap(struct super_block *);
2841 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2842 extern int ext4_init_inode_table(struct super_block *sb,
2843 ext4_group_t group, int barrier);
2844 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2847 int ext4_fc_info_show(struct seq_file *seq, void *v);
2848 void ext4_fc_init(struct super_block *sb, journal_t *journal);
2849 void ext4_fc_init_inode(struct inode *inode);
2850 void ext4_fc_track_range(handle_t *handle, struct inode *inode, ext4_lblk_t start,
2852 void __ext4_fc_track_unlink(handle_t *handle, struct inode *inode,
2853 struct dentry *dentry);
2854 void __ext4_fc_track_link(handle_t *handle, struct inode *inode,
2855 struct dentry *dentry);
2856 void ext4_fc_track_unlink(handle_t *handle, struct dentry *dentry);
2857 void ext4_fc_track_link(handle_t *handle, struct dentry *dentry);
2858 void __ext4_fc_track_create(handle_t *handle, struct inode *inode,
2859 struct dentry *dentry);
2860 void ext4_fc_track_create(handle_t *handle, struct dentry *dentry);
2861 void ext4_fc_track_inode(handle_t *handle, struct inode *inode);
2862 void ext4_fc_mark_ineligible(struct super_block *sb, int reason);
2863 void ext4_fc_start_ineligible(struct super_block *sb, int reason);
2864 void ext4_fc_stop_ineligible(struct super_block *sb);
2865 void ext4_fc_start_update(struct inode *inode);
2866 void ext4_fc_stop_update(struct inode *inode);
2867 void ext4_fc_del(struct inode *inode);
2868 bool ext4_fc_replay_check_excluded(struct super_block *sb, ext4_fsblk_t block);
2869 void ext4_fc_replay_cleanup(struct super_block *sb);
2870 int ext4_fc_commit(journal_t *journal, tid_t commit_tid);
2871 int __init ext4_fc_init_dentry_cache(void);
2874 extern const struct seq_operations ext4_mb_seq_groups_ops;
2875 extern const struct seq_operations ext4_mb_seq_structs_summary_ops;
2876 extern long ext4_mb_stats;
2877 extern long ext4_mb_max_to_scan;
2878 extern int ext4_seq_mb_stats_show(struct seq_file *seq, void *offset);
2879 extern int ext4_mb_init(struct super_block *);
2880 extern int ext4_mb_release(struct super_block *);
2881 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2882 struct ext4_allocation_request *, int *);
2883 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2884 extern void ext4_discard_preallocations(struct inode *, unsigned int);
2885 extern int __init ext4_init_mballoc(void);
2886 extern void ext4_exit_mballoc(void);
2887 extern ext4_group_t ext4_mb_prefetch(struct super_block *sb,
2889 unsigned int nr, int *cnt);
2890 extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group,
2893 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2894 struct buffer_head *bh, ext4_fsblk_t block,
2895 unsigned long count, int flags);
2896 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2897 ext4_group_t ngroups);
2898 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2899 ext4_group_t i, struct ext4_group_desc *desc);
2900 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2901 ext4_fsblk_t block, unsigned long count);
2902 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2903 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid);
2904 extern void ext4_mb_mark_bb(struct super_block *sb, ext4_fsblk_t block,
2905 int len, int state);
2908 void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw,
2909 struct ext4_inode_info *ei);
2910 int ext4_inode_is_fast_symlink(struct inode *inode);
2911 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2912 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2913 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count,
2914 bool wait, struct buffer_head **bhs);
2915 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2916 struct buffer_head *bh_result, int create);
2917 int ext4_get_block(struct inode *inode, sector_t iblock,
2918 struct buffer_head *bh_result, int create);
2919 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2920 struct buffer_head *bh, int create);
2921 int ext4_walk_page_buffers(handle_t *handle,
2922 struct buffer_head *head,
2926 int (*fn)(handle_t *handle,
2927 struct buffer_head *bh));
2928 int do_journal_get_write_access(handle_t *handle,
2929 struct buffer_head *bh);
2930 #define FALL_BACK_TO_NONDELALLOC 1
2931 #define CONVERT_INLINE_DATA 2
2934 EXT4_IGET_NORMAL = 0,
2935 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */
2936 EXT4_IGET_HANDLE = 0x0002 /* Inode # is from a handle */
2939 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
2940 ext4_iget_flags flags, const char *function,
2943 #define ext4_iget(sb, ino, flags) \
2944 __ext4_iget((sb), (ino), (flags), __func__, __LINE__)
2946 extern int ext4_write_inode(struct inode *, struct writeback_control *);
2947 extern int ext4_setattr(struct user_namespace *, struct dentry *,
2949 extern int ext4_getattr(struct user_namespace *, const struct path *,
2950 struct kstat *, u32, unsigned int);
2951 extern void ext4_evict_inode(struct inode *);
2952 extern void ext4_clear_inode(struct inode *);
2953 extern int ext4_file_getattr(struct user_namespace *, const struct path *,
2954 struct kstat *, u32, unsigned int);
2955 extern int ext4_sync_inode(handle_t *, struct inode *);
2956 extern void ext4_dirty_inode(struct inode *, int);
2957 extern int ext4_change_inode_journal_flag(struct inode *, int);
2958 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2959 extern int ext4_get_fc_inode_loc(struct super_block *sb, unsigned long ino,
2960 struct ext4_iloc *iloc);
2961 extern int ext4_inode_attach_jinode(struct inode *inode);
2962 extern int ext4_can_truncate(struct inode *inode);
2963 extern int ext4_truncate(struct inode *);
2964 extern int ext4_break_layouts(struct inode *);
2965 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2966 extern void ext4_set_inode_flags(struct inode *, bool init);
2967 extern int ext4_alloc_da_blocks(struct inode *inode);
2968 extern void ext4_set_aops(struct inode *inode);
2969 extern int ext4_writepage_trans_blocks(struct inode *);
2970 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2971 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2972 loff_t lstart, loff_t lend);
2973 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf);
2974 extern vm_fault_t ext4_filemap_fault(struct vm_fault *vmf);
2975 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2976 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
2977 extern void ext4_da_release_space(struct inode *inode, int to_free);
2978 extern void ext4_da_update_reserve_space(struct inode *inode,
2979 int used, int quota_claim);
2980 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
2981 ext4_fsblk_t pblk, ext4_lblk_t len);
2984 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2985 struct ext4_map_blocks *map, int flags);
2986 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2987 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2988 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2989 ext4_lblk_t start, ext4_lblk_t end);
2992 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2993 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2994 extern void ext4_reset_inode_seed(struct inode *inode);
2997 extern int ext4_ext_migrate(struct inode *);
2998 extern int ext4_ind_migrate(struct inode *inode);
3001 extern int ext4_init_new_dir(handle_t *handle, struct inode *dir,
3002 struct inode *inode);
3003 extern int ext4_dirblock_csum_verify(struct inode *inode,
3004 struct buffer_head *bh);
3005 extern int ext4_orphan_add(handle_t *, struct inode *);
3006 extern int ext4_orphan_del(handle_t *, struct inode *);
3007 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
3008 __u32 start_minor_hash, __u32 *next_hash);
3009 extern int ext4_search_dir(struct buffer_head *bh,
3013 struct ext4_filename *fname,
3014 unsigned int offset,
3015 struct ext4_dir_entry_2 **res_dir);
3016 extern int ext4_generic_delete_entry(struct inode *dir,
3017 struct ext4_dir_entry_2 *de_del,
3018 struct buffer_head *bh,
3022 extern bool ext4_empty_dir(struct inode *inode);
3025 extern void ext4_kvfree_array_rcu(void *to_free);
3026 extern int ext4_group_add(struct super_block *sb,
3027 struct ext4_new_group_data *input);
3028 extern int ext4_group_extend(struct super_block *sb,
3029 struct ext4_super_block *es,
3030 ext4_fsblk_t n_blocks_count);
3031 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
3034 extern struct buffer_head *ext4_sb_bread(struct super_block *sb,
3035 sector_t block, int op_flags);
3036 extern struct buffer_head *ext4_sb_bread_unmovable(struct super_block *sb,
3038 extern void ext4_read_bh_nowait(struct buffer_head *bh, int op_flags,
3039 bh_end_io_t *end_io);
3040 extern int ext4_read_bh(struct buffer_head *bh, int op_flags,
3041 bh_end_io_t *end_io);
3042 extern int ext4_read_bh_lock(struct buffer_head *bh, int op_flags, bool wait);
3043 extern void ext4_sb_breadahead_unmovable(struct super_block *sb, sector_t block);
3044 extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
3045 extern int ext4_calculate_overhead(struct super_block *sb);
3046 extern void ext4_superblock_csum_set(struct super_block *sb);
3047 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
3048 ext4_group_t ngroup);
3049 extern const char *ext4_decode_error(struct super_block *sb, int errno,
3051 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb,
3052 ext4_group_t block_group,
3053 unsigned int flags);
3055 extern __printf(7, 8)
3056 void __ext4_error(struct super_block *, const char *, unsigned int, bool,
3057 int, __u64, const char *, ...);
3058 extern __printf(6, 7)
3059 void __ext4_error_inode(struct inode *, const char *, unsigned int,
3060 ext4_fsblk_t, int, const char *, ...);
3061 extern __printf(5, 6)
3062 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
3064 extern void __ext4_std_error(struct super_block *, const char *,
3066 extern __printf(4, 5)
3067 void __ext4_warning(struct super_block *, const char *, unsigned int,
3069 extern __printf(4, 5)
3070 void __ext4_warning_inode(const struct inode *inode, const char *function,
3071 unsigned int line, const char *fmt, ...);
3072 extern __printf(3, 4)
3073 void __ext4_msg(struct super_block *, const char *, const char *, ...);
3074 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
3075 const char *, unsigned int, const char *);
3076 extern __printf(7, 8)
3077 void __ext4_grp_locked_error(const char *, unsigned int,
3078 struct super_block *, ext4_group_t,
3079 unsigned long, ext4_fsblk_t,
3082 #define EXT4_ERROR_INODE(inode, fmt, a...) \
3083 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
3085 #define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...) \
3086 __ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a)
3088 #define ext4_error_inode_block(inode, block, err, fmt, a...) \
3089 __ext4_error_inode((inode), __func__, __LINE__, (block), (err), \
3092 #define EXT4_ERROR_FILE(file, block, fmt, a...) \
3093 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
3095 #define ext4_abort(sb, err, fmt, a...) \
3096 __ext4_error((sb), __func__, __LINE__, true, (err), 0, (fmt), ## a)
3098 #ifdef CONFIG_PRINTK
3100 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
3101 __ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__)
3102 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \
3103 __ext4_error_inode((inode), (func), (line), (block), \
3104 (err), (fmt), ##__VA_ARGS__)
3105 #define ext4_error_file(file, func, line, block, fmt, ...) \
3106 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
3107 #define ext4_error(sb, fmt, ...) \
3108 __ext4_error((sb), __func__, __LINE__, false, 0, 0, (fmt), \
3110 #define ext4_error_err(sb, err, fmt, ...) \
3111 __ext4_error((sb), __func__, __LINE__, false, (err), 0, (fmt), \
3113 #define ext4_warning(sb, fmt, ...) \
3114 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
3115 #define ext4_warning_inode(inode, fmt, ...) \
3116 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
3117 #define ext4_msg(sb, level, fmt, ...) \
3118 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
3119 #define dump_mmp_msg(sb, mmp, msg) \
3120 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
3121 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
3122 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
3127 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
3129 no_printk(fmt, ##__VA_ARGS__); \
3130 __ext4_error_inode(inode, "", 0, block, 0, " "); \
3132 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \
3134 no_printk(fmt, ##__VA_ARGS__); \
3135 __ext4_error_inode(inode, "", 0, block, err, " "); \
3137 #define ext4_error_file(file, func, line, block, fmt, ...) \
3139 no_printk(fmt, ##__VA_ARGS__); \
3140 __ext4_error_file(file, "", 0, block, " "); \
3142 #define ext4_error(sb, fmt, ...) \
3144 no_printk(fmt, ##__VA_ARGS__); \
3145 __ext4_error(sb, "", 0, false, 0, 0, " "); \
3147 #define ext4_error_err(sb, err, fmt, ...) \
3149 no_printk(fmt, ##__VA_ARGS__); \
3150 __ext4_error(sb, "", 0, false, err, 0, " "); \
3152 #define ext4_warning(sb, fmt, ...) \
3154 no_printk(fmt, ##__VA_ARGS__); \
3155 __ext4_warning(sb, "", 0, " "); \
3157 #define ext4_warning_inode(inode, fmt, ...) \
3159 no_printk(fmt, ##__VA_ARGS__); \
3160 __ext4_warning_inode(inode, "", 0, " "); \
3162 #define ext4_msg(sb, level, fmt, ...) \
3164 no_printk(fmt, ##__VA_ARGS__); \
3165 __ext4_msg(sb, "", " "); \
3167 #define dump_mmp_msg(sb, mmp, msg) \
3168 __dump_mmp_msg(sb, mmp, "", 0, "")
3169 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
3171 no_printk(fmt, ##__VA_ARGS__); \
3172 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
3177 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
3178 struct ext4_group_desc *bg);
3179 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
3180 struct ext4_group_desc *bg);
3181 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
3182 struct ext4_group_desc *bg);
3183 extern __u32 ext4_free_group_clusters(struct super_block *sb,
3184 struct ext4_group_desc *bg);
3185 extern __u32 ext4_free_inodes_count(struct super_block *sb,
3186 struct ext4_group_desc *bg);
3187 extern __u32 ext4_used_dirs_count(struct super_block *sb,
3188 struct ext4_group_desc *bg);
3189 extern __u32 ext4_itable_unused_count(struct super_block *sb,
3190 struct ext4_group_desc *bg);
3191 extern void ext4_block_bitmap_set(struct super_block *sb,
3192 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3193 extern void ext4_inode_bitmap_set(struct super_block *sb,
3194 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3195 extern void ext4_inode_table_set(struct super_block *sb,
3196 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3197 extern void ext4_free_group_clusters_set(struct super_block *sb,
3198 struct ext4_group_desc *bg,
3200 extern void ext4_free_inodes_set(struct super_block *sb,
3201 struct ext4_group_desc *bg, __u32 count);
3202 extern void ext4_used_dirs_set(struct super_block *sb,
3203 struct ext4_group_desc *bg, __u32 count);
3204 extern void ext4_itable_unused_set(struct super_block *sb,
3205 struct ext4_group_desc *bg, __u32 count);
3206 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
3207 struct ext4_group_desc *gdp);
3208 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
3209 struct ext4_group_desc *gdp);
3210 extern int ext4_register_li_request(struct super_block *sb,
3211 ext4_group_t first_not_zeroed);
3213 static inline int ext4_has_metadata_csum(struct super_block *sb)
3215 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
3216 !EXT4_SB(sb)->s_chksum_driver);
3218 return ext4_has_feature_metadata_csum(sb) &&
3219 (EXT4_SB(sb)->s_chksum_driver != NULL);
3222 static inline int ext4_has_group_desc_csum(struct super_block *sb)
3224 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb);
3227 #define ext4_read_incompat_64bit_val(es, name) \
3228 (((es)->s_feature_incompat & cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT) \
3229 ? (ext4_fsblk_t)le32_to_cpu(es->name##_hi) << 32 : 0) | \
3230 le32_to_cpu(es->name##_lo))
3232 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
3234 return ext4_read_incompat_64bit_val(es, s_blocks_count);
3237 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
3239 return ext4_read_incompat_64bit_val(es, s_r_blocks_count);
3242 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
3244 return ext4_read_incompat_64bit_val(es, s_free_blocks_count);
3247 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
3250 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
3251 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
3254 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
3257 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
3258 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
3261 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
3264 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
3265 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
3268 static inline loff_t ext4_isize(struct super_block *sb,
3269 struct ext4_inode *raw_inode)
3271 if (ext4_has_feature_largedir(sb) ||
3272 S_ISREG(le16_to_cpu(raw_inode->i_mode)))
3273 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
3274 le32_to_cpu(raw_inode->i_size_lo);
3276 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
3279 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
3281 raw_inode->i_size_lo = cpu_to_le32(i_size);
3282 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
3286 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
3289 struct ext4_group_info **grp_info;
3290 long indexv, indexh;
3291 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
3292 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
3293 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
3294 grp_info = sbi_array_rcu_deref(EXT4_SB(sb), s_group_info, indexv);
3295 return grp_info[indexh];
3299 * Reading s_groups_count requires using smp_rmb() afterwards. See
3300 * the locking protocol documented in the comments of ext4_group_add()
3303 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
3305 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
3311 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
3312 ext4_group_t block_group)
3314 return block_group >> sbi->s_log_groups_per_flex;
3317 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
3319 return 1 << sbi->s_log_groups_per_flex;
3322 #define ext4_std_error(sb, errno) \
3325 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
3329 /* Each CPU can accumulate percpu_counter_batch clusters in their local
3330 * counters. So we need to make sure we have free clusters more
3331 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
3333 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
3335 #define EXT4_FREECLUSTERS_WATERMARK 0
3338 /* Update i_disksize. Requires i_mutex to avoid races with truncate */
3339 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
3341 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
3342 !inode_is_locked(inode));
3343 down_write(&EXT4_I(inode)->i_data_sem);
3344 if (newsize > EXT4_I(inode)->i_disksize)
3345 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize);
3346 up_write(&EXT4_I(inode)->i_data_sem);
3349 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
3350 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
3354 if (newsize > inode->i_size) {
3355 i_size_write(inode, newsize);
3358 if (newsize > EXT4_I(inode)->i_disksize) {
3359 ext4_update_i_disksize(inode, newsize);
3365 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
3368 struct ext4_group_info {
3369 unsigned long bb_state;
3370 #ifdef AGGRESSIVE_CHECK
3371 unsigned long bb_check_counter;
3373 struct rb_root bb_free_root;
3374 ext4_grpblk_t bb_first_free; /* first free block */
3375 ext4_grpblk_t bb_free; /* total free blocks */
3376 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
3377 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
3378 ext4_group_t bb_group; /* Group number */
3379 struct list_head bb_prealloc_list;
3383 struct rw_semaphore alloc_sem;
3384 struct rb_node bb_avg_fragment_size_rb;
3385 struct list_head bb_largest_free_order_node;
3386 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
3387 * regions, index is order.
3388 * bb_counters[3] = 5 means
3389 * 5 free 8-block regions. */
3392 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
3393 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
3394 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
3395 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
3396 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \
3397 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT)
3398 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \
3399 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT)
3400 #define EXT4_GROUP_INFO_BBITMAP_READ_BIT 4
3402 #define EXT4_MB_GRP_NEED_INIT(grp) \
3403 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
3404 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
3405 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3406 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
3407 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3409 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
3410 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3411 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
3412 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3413 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
3414 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3415 #define EXT4_MB_GRP_TEST_AND_SET_READ(grp) \
3416 (test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state)))
3418 #define EXT4_MAX_CONTENTION 8
3419 #define EXT4_CONTENTION_THRESHOLD 2
3421 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
3424 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
3428 * Returns true if the filesystem is busy enough that attempts to
3429 * access the block group locks has run into contention.
3431 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
3433 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
3436 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
3438 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
3439 if (spin_trylock(lock))
3441 * We're able to grab the lock right away, so drop the
3442 * lock contention counter.
3444 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
3447 * The lock is busy, so bump the contention counter,
3448 * and then wait on the spin lock.
3450 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
3451 EXT4_MAX_CONTENTION);
3456 static inline void ext4_unlock_group(struct super_block *sb,
3459 spin_unlock(ext4_group_lock_ptr(sb, group));
3463 static inline bool ext4_quota_capable(struct super_block *sb)
3465 return (test_opt(sb, QUOTA) || ext4_has_feature_quota(sb));
3468 static inline bool ext4_is_quota_journalled(struct super_block *sb)
3470 struct ext4_sb_info *sbi = EXT4_SB(sb);
3472 return (ext4_has_feature_quota(sb) ||
3473 sbi->s_qf_names[USRQUOTA] || sbi->s_qf_names[GRPQUOTA]);
3478 * Block validity checking
3480 #define ext4_check_indirect_blockref(inode, bh) \
3481 ext4_check_blockref(__func__, __LINE__, inode, \
3482 (__le32 *)(bh)->b_data, \
3483 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
3485 #define ext4_ind_check_inode(inode) \
3486 ext4_check_blockref(__func__, __LINE__, inode, \
3487 EXT4_I(inode)->i_data, \
3491 * Inodes and files operations
3495 extern const struct file_operations ext4_dir_operations;
3498 extern const struct inode_operations ext4_file_inode_operations;
3499 extern const struct file_operations ext4_file_operations;
3500 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
3503 extern int ext4_get_max_inline_size(struct inode *inode);
3504 extern int ext4_find_inline_data_nolock(struct inode *inode);
3505 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
3507 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
3509 extern int ext4_readpage_inline(struct inode *inode, struct page *page);
3510 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3511 struct inode *inode,
3512 loff_t pos, unsigned len,
3514 struct page **pagep);
3515 extern int ext4_write_inline_data_end(struct inode *inode,
3516 loff_t pos, unsigned len,
3519 extern struct buffer_head *
3520 ext4_journalled_write_inline_data(struct inode *inode,
3523 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
3524 struct inode *inode,
3525 loff_t pos, unsigned len,
3527 struct page **pagep,
3529 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
3530 unsigned len, unsigned copied,
3532 extern int ext4_try_add_inline_entry(handle_t *handle,
3533 struct ext4_filename *fname,
3534 struct inode *dir, struct inode *inode);
3535 extern int ext4_try_create_inline_dir(handle_t *handle,
3536 struct inode *parent,
3537 struct inode *inode);
3538 extern int ext4_read_inline_dir(struct file *filp,
3539 struct dir_context *ctx,
3540 int *has_inline_data);
3541 extern int ext4_inlinedir_to_tree(struct file *dir_file,
3542 struct inode *dir, ext4_lblk_t block,
3543 struct dx_hash_info *hinfo,
3544 __u32 start_hash, __u32 start_minor_hash,
3545 int *has_inline_data);
3546 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3547 struct ext4_filename *fname,
3548 struct ext4_dir_entry_2 **res_dir,
3549 int *has_inline_data);
3550 extern int ext4_delete_inline_entry(handle_t *handle,
3552 struct ext4_dir_entry_2 *de_del,
3553 struct buffer_head *bh,
3554 int *has_inline_data);
3555 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3556 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3557 struct ext4_dir_entry_2 **parent_de,
3559 extern int ext4_inline_data_fiemap(struct inode *inode,
3560 struct fiemap_extent_info *fieinfo,
3561 int *has_inline, __u64 start, __u64 len);
3564 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap);
3566 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3568 extern int ext4_convert_inline_data(struct inode *inode);
3570 static inline int ext4_has_inline_data(struct inode *inode)
3572 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3573 EXT4_I(inode)->i_inline_off;
3577 extern const struct inode_operations ext4_dir_inode_operations;
3578 extern const struct inode_operations ext4_special_inode_operations;
3579 extern struct dentry *ext4_get_parent(struct dentry *child);
3580 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3581 struct ext4_dir_entry_2 *de,
3582 int blocksize, int csum_size,
3583 unsigned int parent_ino, int dotdot_real_len);
3584 extern void ext4_initialize_dirent_tail(struct buffer_head *bh,
3585 unsigned int blocksize);
3586 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode,
3587 struct buffer_head *bh);
3588 extern int __ext4_unlink(handle_t *handle, struct inode *dir, const struct qstr *d_name,
3589 struct inode *inode);
3590 extern int __ext4_link(struct inode *dir, struct inode *inode,
3591 struct dentry *dentry);
3594 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = {
3595 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
3596 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
3597 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
3598 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
3599 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
3600 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
3601 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
3604 static inline void ext4_set_de_type(struct super_block *sb,
3605 struct ext4_dir_entry_2 *de,
3607 if (ext4_has_feature_filetype(sb))
3608 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3612 extern int ext4_mpage_readpages(struct inode *inode,
3613 struct readahead_control *rac, struct page *page);
3614 extern int __init ext4_init_post_read_processing(void);
3615 extern void ext4_exit_post_read_processing(void);
3618 extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3619 extern const struct inode_operations ext4_symlink_inode_operations;
3620 extern const struct inode_operations ext4_fast_symlink_inode_operations;
3623 extern int ext4_register_sysfs(struct super_block *sb);
3624 extern void ext4_unregister_sysfs(struct super_block *sb);
3625 extern int __init ext4_init_sysfs(void);
3626 extern void ext4_exit_sysfs(void);
3628 /* block_validity */
3629 extern void ext4_release_system_zone(struct super_block *sb);
3630 extern int ext4_setup_system_zone(struct super_block *sb);
3631 extern int __init ext4_init_system_zone(void);
3632 extern void ext4_exit_system_zone(void);
3633 extern int ext4_inode_block_valid(struct inode *inode,
3634 ext4_fsblk_t start_blk,
3635 unsigned int count);
3636 extern int ext4_check_blockref(const char *, unsigned int,
3637 struct inode *, __le32 *, unsigned int);
3640 struct ext4_ext_path;
3644 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3647 #define EXT_MAX_BLOCKS 0xffffffff
3649 extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode);
3650 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3651 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3652 struct ext4_map_blocks *map, int flags);
3653 extern int ext4_ext_truncate(handle_t *, struct inode *);
3654 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3656 extern void ext4_ext_init(struct super_block *);
3657 extern void ext4_ext_release(struct super_block *);
3658 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3660 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3661 loff_t offset, ssize_t len);
3662 extern int ext4_convert_unwritten_io_end_vec(handle_t *handle,
3663 ext4_io_end_t *io_end);
3664 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3665 struct ext4_map_blocks *map, int flags);
3666 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3668 struct ext4_ext_path *path);
3669 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3670 struct ext4_ext_path **,
3671 struct ext4_extent *, int);
3672 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3673 struct ext4_ext_path **,
3675 extern void ext4_ext_drop_refs(struct ext4_ext_path *);
3676 extern int ext4_ext_check_inode(struct inode *inode);
3677 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3678 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3679 __u64 start, __u64 len);
3680 extern int ext4_get_es_cache(struct inode *inode,
3681 struct fiemap_extent_info *fieinfo,
3682 __u64 start, __u64 len);
3683 extern int ext4_ext_precache(struct inode *inode);
3684 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3685 struct inode *inode2, ext4_lblk_t lblk1,
3686 ext4_lblk_t lblk2, ext4_lblk_t count,
3687 int mark_unwritten,int *err);
3688 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu);
3689 extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode,
3690 int check_cred, int restart_cred,
3692 extern void ext4_ext_replay_shrink_inode(struct inode *inode, ext4_lblk_t end);
3693 extern int ext4_ext_replay_set_iblocks(struct inode *inode);
3694 extern int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start,
3695 int len, int unwritten, ext4_fsblk_t pblk);
3696 extern int ext4_ext_clear_bb(struct inode *inode);
3700 extern void ext4_double_down_write_data_sem(struct inode *first,
3701 struct inode *second);
3702 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3703 struct inode *donor_inode);
3704 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3705 __u64 start_orig, __u64 start_donor,
3706 __u64 len, __u64 *moved_len);
3709 extern int __init ext4_init_pageio(void);
3710 extern void ext4_exit_pageio(void);
3711 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3712 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3713 extern int ext4_put_io_end(ext4_io_end_t *io_end);
3714 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3715 extern void ext4_io_submit_init(struct ext4_io_submit *io,
3716 struct writeback_control *wbc);
3717 extern void ext4_end_io_rsv_work(struct work_struct *work);
3718 extern void ext4_io_submit(struct ext4_io_submit *io);
3719 extern int ext4_bio_write_page(struct ext4_io_submit *io,
3723 extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end);
3724 extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end);
3727 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3730 extern const struct fsverity_operations ext4_verityops;
3733 * Add new method to test whether block and inode bitmaps are properly
3734 * initialized. With uninit_bg reading the block from disk is not enough
3735 * to mark the bitmap uptodate. We need to also zero-out the bitmap
3737 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3739 static inline int bitmap_uptodate(struct buffer_head *bh)
3741 return (buffer_uptodate(bh) &&
3742 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3744 static inline void set_bitmap_uptodate(struct buffer_head *bh)
3746 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3749 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
3751 /* For ioend & aio unwritten conversion wait queues */
3752 #define EXT4_WQ_HASH_SZ 37
3753 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
3755 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3757 extern int ext4_resize_begin(struct super_block *sb);
3758 extern void ext4_resize_end(struct super_block *sb);
3760 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
3761 struct ext4_io_end *io_end)
3763 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
3764 io_end->flag |= EXT4_IO_END_UNWRITTEN;
3765 atomic_inc(&EXT4_I(inode)->i_unwritten);
3769 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3771 struct inode *inode = io_end->inode;
3773 if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
3774 io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3775 /* Wake up anyone waiting on unwritten extent conversion */
3776 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
3777 wake_up_all(ext4_ioend_wq(inode));
3781 extern const struct iomap_ops ext4_iomap_ops;
3782 extern const struct iomap_ops ext4_iomap_overwrite_ops;
3783 extern const struct iomap_ops ext4_iomap_report_ops;
3785 static inline int ext4_buffer_uptodate(struct buffer_head *bh)
3788 * If the buffer has the write error flag, we have failed
3789 * to write out data in the block. In this case, we don't
3790 * have to read the block because we may read the old data
3793 if (!buffer_uptodate(bh) && buffer_write_io_error(bh))
3794 set_buffer_uptodate(bh);
3795 return buffer_uptodate(bh);
3798 #endif /* __KERNEL__ */
3800 #define EFSBADCRC EBADMSG /* Bad CRC detected */
3801 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
3803 #endif /* _EXT4_H */