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 /* data type for block offset of block group */
102 typedef int ext4_grpblk_t;
104 /* data type for filesystem-wide blocks number */
105 typedef unsigned long long ext4_fsblk_t;
107 /* data type for file logical block number */
108 typedef __u32 ext4_lblk_t;
110 /* data type for block group number */
111 typedef unsigned int ext4_group_t;
113 enum SHIFT_DIRECTION {
119 * Flags used in mballoc's allocation_context flags field.
121 * Also used to show what's going on for debugging purposes when the
122 * flag field is exported via the traceport interface
125 /* prefer goal again. length */
126 #define EXT4_MB_HINT_MERGE 0x0001
127 /* blocks already reserved */
128 #define EXT4_MB_HINT_RESERVED 0x0002
129 /* metadata is being allocated */
130 #define EXT4_MB_HINT_METADATA 0x0004
131 /* first blocks in the file */
132 #define EXT4_MB_HINT_FIRST 0x0008
133 /* search for the best chunk */
134 #define EXT4_MB_HINT_BEST 0x0010
135 /* data is being allocated */
136 #define EXT4_MB_HINT_DATA 0x0020
137 /* don't preallocate (for tails) */
138 #define EXT4_MB_HINT_NOPREALLOC 0x0040
139 /* allocate for locality group */
140 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080
141 /* allocate goal blocks or none */
142 #define EXT4_MB_HINT_GOAL_ONLY 0x0100
143 /* goal is meaningful */
144 #define EXT4_MB_HINT_TRY_GOAL 0x0200
145 /* blocks already pre-reserved by delayed allocation */
146 #define EXT4_MB_DELALLOC_RESERVED 0x0400
147 /* We are doing stream allocation */
148 #define EXT4_MB_STREAM_ALLOC 0x0800
149 /* Use reserved root blocks if needed */
150 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000
151 /* Use blocks from reserved pool */
152 #define EXT4_MB_USE_RESERVED 0x2000
153 /* Do strict check for free blocks while retrying block allocation */
154 #define EXT4_MB_STRICT_CHECK 0x4000
156 struct ext4_allocation_request {
157 /* target inode for block we're allocating */
159 /* how many blocks we want to allocate */
161 /* logical block in target inode */
163 /* the closest logical allocated block to the left */
165 /* the closest logical allocated block to the right */
167 /* phys. target (a hint) */
169 /* phys. block for the closest logical allocated block to the left */
171 /* phys. block for the closest logical allocated block to the right */
173 /* flags. see above EXT4_MB_HINT_* */
178 * Logical to physical block mapping, used by ext4_map_blocks()
180 * This structure is used to pass requests into ext4_map_blocks() as
181 * well as to store the information returned by ext4_map_blocks(). It
182 * takes less room on the stack than a struct buffer_head.
184 #define EXT4_MAP_NEW BIT(BH_New)
185 #define EXT4_MAP_MAPPED BIT(BH_Mapped)
186 #define EXT4_MAP_UNWRITTEN BIT(BH_Unwritten)
187 #define EXT4_MAP_BOUNDARY BIT(BH_Boundary)
188 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
189 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
191 struct ext4_map_blocks {
195 unsigned int m_flags;
199 * Block validity checking, system zone rbtree.
201 struct ext4_system_blocks {
207 * Flags for ext4_io_end->flags
209 #define EXT4_IO_END_UNWRITTEN 0x0001
211 struct ext4_io_end_vec {
212 struct list_head list; /* list of io_end_vec */
213 loff_t offset; /* offset in the file */
214 ssize_t size; /* size of the extent */
218 * For converting unwritten extents on a work queue. 'handle' is used for
219 * buffered writeback.
221 typedef struct ext4_io_end {
222 struct list_head list; /* per-file finished IO list */
223 handle_t *handle; /* handle reserved for extent
225 struct inode *inode; /* file being written to */
226 struct bio *bio; /* Linked list of completed
227 * bios covering the extent */
228 unsigned int flag; /* unwritten or not */
229 atomic_t count; /* reference counter */
230 struct list_head list_vec; /* list of ext4_io_end_vec */
233 struct ext4_io_submit {
234 struct writeback_control *io_wbc;
236 ext4_io_end_t *io_end;
237 sector_t io_next_block;
241 * Special inodes numbers
243 #define EXT4_BAD_INO 1 /* Bad blocks inode */
244 #define EXT4_ROOT_INO 2 /* Root inode */
245 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */
246 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
247 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
248 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
249 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
250 #define EXT4_JOURNAL_INO 8 /* Journal inode */
252 /* First non-reserved inode for old ext4 filesystems */
253 #define EXT4_GOOD_OLD_FIRST_INO 11
256 * Maximal count of links to a file
258 #define EXT4_LINK_MAX 65000
261 * Macro-instructions used to manage several block sizes
263 #define EXT4_MIN_BLOCK_SIZE 1024
264 #define EXT4_MAX_BLOCK_SIZE 65536
265 #define EXT4_MIN_BLOCK_LOG_SIZE 10
266 #define EXT4_MAX_BLOCK_LOG_SIZE 16
267 #define EXT4_MAX_CLUSTER_LOG_SIZE 30
269 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
271 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
273 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
274 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
275 EXT4_SB(s)->s_cluster_bits)
277 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
278 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
280 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
283 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
284 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
285 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
287 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
288 EXT4_GOOD_OLD_INODE_SIZE : \
290 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
291 EXT4_GOOD_OLD_FIRST_INO : \
294 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
295 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \
296 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \
299 /* Translate a block number to a cluster number */
300 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
301 /* Translate a cluster number to a block number */
302 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
303 /* Translate # of blks to # of clusters */
304 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
305 (sbi)->s_cluster_bits)
306 /* Mask out the low bits to get the starting block of the cluster */
307 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
308 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
309 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
310 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
311 /* Fill in the low bits to get the last block of the cluster */
312 #define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) | \
313 ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1))
314 /* Get the cluster offset */
315 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
316 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
317 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
318 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
321 * Structure of a blocks group descriptor
323 struct ext4_group_desc
325 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
326 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
327 __le32 bg_inode_table_lo; /* Inodes table block */
328 __le16 bg_free_blocks_count_lo;/* Free blocks count */
329 __le16 bg_free_inodes_count_lo;/* Free inodes count */
330 __le16 bg_used_dirs_count_lo; /* Directories count */
331 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
332 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
333 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
334 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
335 __le16 bg_itable_unused_lo; /* Unused inodes count */
336 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
337 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
338 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
339 __le32 bg_inode_table_hi; /* Inodes table block MSB */
340 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
341 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
342 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
343 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
344 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
345 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
346 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
350 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
351 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
353 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
354 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
358 * Structure of a flex block group info
362 atomic64_t free_clusters;
363 atomic_t free_inodes;
367 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
368 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
369 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
372 * Macro-instructions used to manage group descriptors
374 #define EXT4_MIN_DESC_SIZE 32
375 #define EXT4_MIN_DESC_SIZE_64BIT 64
376 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
377 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
379 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
380 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
381 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
382 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
383 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
385 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
386 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
387 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
391 * Constants relative to the data blocks
393 #define EXT4_NDIR_BLOCKS 12
394 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
395 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
396 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
397 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
402 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
403 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
404 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
405 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
406 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
407 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
408 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
409 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
410 /* Reserved for compression usage... */
411 #define EXT4_DIRTY_FL 0x00000100
412 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
413 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
414 /* nb: was previously EXT2_ECOMPR_FL */
415 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */
416 /* End compression flags --- maybe not all used */
417 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
418 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
419 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
420 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
421 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
422 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
423 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
424 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
425 #define EXT4_VERITY_FL 0x00100000 /* Verity protected inode */
426 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
427 /* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */
429 #define EXT4_DAX_FL 0x02000000 /* Inode is DAX */
431 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
432 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
433 #define EXT4_CASEFOLD_FL 0x40000000 /* Casefolded directory */
434 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
436 /* User modifiable flags */
437 #define EXT4_FL_USER_MODIFIABLE (EXT4_SECRM_FL | \
441 EXT4_IMMUTABLE_FL | \
445 EXT4_JOURNAL_DATA_FL | \
450 0x00400000 /* EXT4_EOFBLOCKS_FL */ | \
452 EXT4_PROJINHERIT_FL | \
455 /* User visible flags */
456 #define EXT4_FL_USER_VISIBLE (EXT4_FL_USER_MODIFIABLE | \
465 /* Flags we can manipulate with through FS_IOC_FSSETXATTR */
466 #define EXT4_FL_XFLAG_VISIBLE (EXT4_SYNC_FL | \
467 EXT4_IMMUTABLE_FL | \
471 EXT4_PROJINHERIT_FL | \
474 /* Flags that should be inherited by new inodes from their parent. */
475 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
476 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
477 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
478 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
479 EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL |\
482 /* Flags that are appropriate for regular files (all but dir-specific ones). */
483 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\
484 EXT4_PROJINHERIT_FL))
486 /* Flags that are appropriate for non-directories/regular files. */
487 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
489 /* The only flags that should be swapped */
490 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL)
492 /* Flags which are mutually exclusive to DAX */
493 #define EXT4_DAX_MUT_EXCL (EXT4_VERITY_FL | EXT4_ENCRYPT_FL |\
494 EXT4_JOURNAL_DATA_FL | EXT4_INLINE_DATA_FL)
496 /* Mask out flags that are inappropriate for the given type of inode. */
497 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
501 else if (S_ISREG(mode))
502 return flags & EXT4_REG_FLMASK;
504 return flags & EXT4_OTHER_FLMASK;
508 * Inode flags used for atomic set/get
511 EXT4_INODE_SECRM = 0, /* Secure deletion */
512 EXT4_INODE_UNRM = 1, /* Undelete */
513 EXT4_INODE_COMPR = 2, /* Compress file */
514 EXT4_INODE_SYNC = 3, /* Synchronous updates */
515 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
516 EXT4_INODE_APPEND = 5, /* writes to file may only append */
517 EXT4_INODE_NODUMP = 6, /* do not dump file */
518 EXT4_INODE_NOATIME = 7, /* do not update atime */
519 /* Reserved for compression usage... */
520 EXT4_INODE_DIRTY = 8,
521 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
522 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
523 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */
524 /* End compression flags --- maybe not all used */
525 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
526 EXT4_INODE_IMAGIC = 13, /* AFS directory */
527 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
528 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
529 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
530 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
531 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
532 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
533 EXT4_INODE_VERITY = 20, /* Verity protected inode */
534 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
535 /* 22 was formerly EXT4_INODE_EOFBLOCKS */
536 EXT4_INODE_DAX = 25, /* Inode is DAX */
537 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
538 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */
539 EXT4_INODE_CASEFOLD = 30, /* Casefolded directory */
540 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
544 * Since it's pretty easy to mix up bit numbers and hex values, we use a
545 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
546 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
547 * any extra space in the compiled kernel image, otherwise, the build will fail.
548 * It's important that these values are the same, since we are using
549 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
550 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
551 * values found in ext2, ext3 and ext4 filesystems, and of course the values
552 * defined in e2fsprogs.
554 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
556 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
557 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
559 static inline void ext4_check_flag_values(void)
561 CHECK_FLAG_VALUE(SECRM);
562 CHECK_FLAG_VALUE(UNRM);
563 CHECK_FLAG_VALUE(COMPR);
564 CHECK_FLAG_VALUE(SYNC);
565 CHECK_FLAG_VALUE(IMMUTABLE);
566 CHECK_FLAG_VALUE(APPEND);
567 CHECK_FLAG_VALUE(NODUMP);
568 CHECK_FLAG_VALUE(NOATIME);
569 CHECK_FLAG_VALUE(DIRTY);
570 CHECK_FLAG_VALUE(COMPRBLK);
571 CHECK_FLAG_VALUE(NOCOMPR);
572 CHECK_FLAG_VALUE(ENCRYPT);
573 CHECK_FLAG_VALUE(INDEX);
574 CHECK_FLAG_VALUE(IMAGIC);
575 CHECK_FLAG_VALUE(JOURNAL_DATA);
576 CHECK_FLAG_VALUE(NOTAIL);
577 CHECK_FLAG_VALUE(DIRSYNC);
578 CHECK_FLAG_VALUE(TOPDIR);
579 CHECK_FLAG_VALUE(HUGE_FILE);
580 CHECK_FLAG_VALUE(EXTENTS);
581 CHECK_FLAG_VALUE(VERITY);
582 CHECK_FLAG_VALUE(EA_INODE);
583 CHECK_FLAG_VALUE(INLINE_DATA);
584 CHECK_FLAG_VALUE(PROJINHERIT);
585 CHECK_FLAG_VALUE(CASEFOLD);
586 CHECK_FLAG_VALUE(RESERVED);
589 /* Used to pass group descriptor data when online resize is done */
590 struct ext4_new_group_input {
591 __u32 group; /* Group number for this data */
592 __u64 block_bitmap; /* Absolute block number of block bitmap */
593 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
594 __u64 inode_table; /* Absolute block number of inode table start */
595 __u32 blocks_count; /* Total number of blocks in this group */
596 __u16 reserved_blocks; /* Number of reserved blocks in this group */
600 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
601 struct compat_ext4_new_group_input {
603 compat_u64 block_bitmap;
604 compat_u64 inode_bitmap;
605 compat_u64 inode_table;
612 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
613 struct ext4_new_group_data {
619 __u16 reserved_blocks;
621 __u32 free_clusters_count;
624 /* Indexes used to index group tables in ext4_new_group_data */
626 BLOCK_BITMAP = 0, /* block bitmap */
627 INODE_BITMAP, /* inode bitmap */
628 INODE_TABLE, /* inode tables */
633 * Flags used by ext4_map_blocks()
635 /* Allocate any needed blocks and/or convert an unwritten
636 extent to be an initialized ext4 */
637 #define EXT4_GET_BLOCKS_CREATE 0x0001
638 /* Request the creation of an unwritten extent */
639 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002
640 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\
641 EXT4_GET_BLOCKS_CREATE)
642 /* Caller is from the delayed allocation writeout path
643 * finally doing the actual allocation of delayed blocks */
644 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
645 /* caller is from the direct IO path, request to creation of an
646 unwritten extents if not allocated, split the unwritten
647 extent if blocks has been preallocated already*/
648 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
649 #define EXT4_GET_BLOCKS_CONVERT 0x0010
650 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
651 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
652 /* Convert extent to initialized after IO complete */
653 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
654 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
655 /* Eventual metadata allocation (due to growing extent tree)
656 * should not fail, so try to use reserved blocks for that.*/
657 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
658 /* Don't normalize allocation size (used for fallocate) */
659 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
660 /* Convert written extents to unwritten */
661 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100
662 /* Write zeros to newly created written extents */
663 #define EXT4_GET_BLOCKS_ZERO 0x0200
664 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\
665 EXT4_GET_BLOCKS_ZERO)
666 /* Caller will submit data before dropping transaction handle. This
667 * allows jbd2 to avoid submitting data before commit. */
668 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400
671 * The bit position of these flags must not overlap with any of the
672 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(),
673 * read_extent_tree_block(), ext4_split_extent_at(),
674 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
675 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
676 * caching the extents when reading from the extent tree while a
677 * truncate or punch hole operation is in progress.
679 #define EXT4_EX_NOCACHE 0x40000000
680 #define EXT4_EX_FORCE_CACHE 0x20000000
681 #define EXT4_EX_NOFAIL 0x10000000
684 * Flags used by ext4_free_blocks
686 #define EXT4_FREE_BLOCKS_METADATA 0x0001
687 #define EXT4_FREE_BLOCKS_FORGET 0x0002
688 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
689 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
690 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
691 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
692 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040
697 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
698 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
699 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
700 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
701 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
702 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
703 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
704 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
705 #define EXT4_IOC_MIGRATE _IO('f', 9)
706 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
707 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
708 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
709 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
710 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
711 #define EXT4_IOC_SWAP_BOOT _IO('f', 17)
712 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18)
713 /* ioctl codes 19--39 are reserved for fscrypt */
714 #define EXT4_IOC_CLEAR_ES_CACHE _IO('f', 40)
715 #define EXT4_IOC_GETSTATE _IOW('f', 41, __u32)
716 #define EXT4_IOC_GET_ES_CACHE _IOWR('f', 42, struct fiemap)
718 #define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32)
721 * Flags for going down operation
723 #define EXT4_GOING_FLAGS_DEFAULT 0x0 /* going down */
724 #define EXT4_GOING_FLAGS_LOGFLUSH 0x1 /* flush log but not data */
725 #define EXT4_GOING_FLAGS_NOLOGFLUSH 0x2 /* don't flush log nor data */
728 * Flags returned by EXT4_IOC_GETSTATE
730 * We only expose to userspace a subset of the state flags in
733 #define EXT4_STATE_FLAG_EXT_PRECACHED 0x00000001
734 #define EXT4_STATE_FLAG_NEW 0x00000002
735 #define EXT4_STATE_FLAG_NEWENTRY 0x00000004
736 #define EXT4_STATE_FLAG_DA_ALLOC_CLOSE 0x00000008
738 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
740 * ioctl commands in 32 bit emulation
742 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
743 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
744 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
745 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
746 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
747 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
748 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
749 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
753 * Returned by EXT4_IOC_GET_ES_CACHE as an additional possible flag.
754 * It indicates that the entry in extent status cache is for a hole.
756 #define EXT4_FIEMAP_EXTENT_HOLE 0x08000000
758 /* Max physical block we can address w/o extents */
759 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
761 /* Max logical block we can support */
762 #define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFE
765 * Structure of an inode on the disk
768 __le16 i_mode; /* File mode */
769 __le16 i_uid; /* Low 16 bits of Owner Uid */
770 __le32 i_size_lo; /* Size in bytes */
771 __le32 i_atime; /* Access time */
772 __le32 i_ctime; /* Inode Change time */
773 __le32 i_mtime; /* Modification time */
774 __le32 i_dtime; /* Deletion Time */
775 __le16 i_gid; /* Low 16 bits of Group Id */
776 __le16 i_links_count; /* Links count */
777 __le32 i_blocks_lo; /* Blocks count */
778 __le32 i_flags; /* File flags */
784 __u32 h_i_translator;
789 } osd1; /* OS dependent 1 */
790 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
791 __le32 i_generation; /* File version (for NFS) */
792 __le32 i_file_acl_lo; /* File ACL */
794 __le32 i_obso_faddr; /* Obsoleted fragment address */
797 __le16 l_i_blocks_high; /* were l_i_reserved1 */
798 __le16 l_i_file_acl_high;
799 __le16 l_i_uid_high; /* these 2 fields */
800 __le16 l_i_gid_high; /* were reserved2[0] */
801 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
805 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
812 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
813 __le16 m_i_file_acl_high;
814 __u32 m_i_reserved2[2];
816 } osd2; /* OS dependent 2 */
817 __le16 i_extra_isize;
818 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
819 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
820 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
821 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
822 __le32 i_crtime; /* File Creation time */
823 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
824 __le32 i_version_hi; /* high 32 bits for 64-bit version */
825 __le32 i_projid; /* Project ID */
829 __u32 reserved; /* should be zero */
830 __u32 donor_fd; /* donor file descriptor */
831 __u64 orig_start; /* logical start offset in block for orig */
832 __u64 donor_start; /* logical start offset in block for donor */
833 __u64 len; /* block length to be moved */
834 __u64 moved_len; /* moved block length */
837 #define EXT4_EPOCH_BITS 2
838 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
839 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
842 * Extended fields will fit into an inode if the filesystem was formatted
843 * with large inodes (-I 256 or larger) and there are not currently any EAs
844 * consuming all of the available space. For new inodes we always reserve
845 * enough space for the kernel's known extended fields, but for inodes
846 * created with an old kernel this might not have been the case. None of
847 * the extended inode fields is critical for correct filesystem operation.
848 * This macro checks if a certain field fits in the inode. Note that
849 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
851 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
852 ((offsetof(typeof(*ext4_inode), field) + \
853 sizeof((ext4_inode)->field)) \
854 <= (EXT4_GOOD_OLD_INODE_SIZE + \
855 (einode)->i_extra_isize)) \
858 * We use an encoding that preserves the times for extra epoch "00":
860 * extra msb of adjust for signed
861 * epoch 32-bit 32-bit tv_sec to
862 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
863 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
864 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
865 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
866 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
867 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
868 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
869 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
870 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
872 * Note that previous versions of the kernel on 64-bit systems would
873 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
874 * 1970. e2fsck will correct this, assuming that it is run on the
875 * affected filesystem before 2242.
878 static inline __le32 ext4_encode_extra_time(struct timespec64 *time)
880 u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK;
881 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
884 static inline void ext4_decode_extra_time(struct timespec64 *time,
887 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK)))
888 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
889 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
892 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
894 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {\
895 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
896 (raw_inode)->xtime ## _extra = \
897 ext4_encode_extra_time(&(inode)->xtime); \
900 (raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (inode)->xtime.tv_sec, S32_MIN, S32_MAX)); \
903 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
905 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
906 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
907 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
908 (raw_inode)->xtime ## _extra = \
909 ext4_encode_extra_time(&(einode)->xtime); \
912 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
914 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
915 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \
916 ext4_decode_extra_time(&(inode)->xtime, \
917 raw_inode->xtime ## _extra); \
920 (inode)->xtime.tv_nsec = 0; \
924 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
926 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
927 (einode)->xtime.tv_sec = \
928 (signed)le32_to_cpu((raw_inode)->xtime); \
930 (einode)->xtime.tv_sec = 0; \
931 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
932 ext4_decode_extra_time(&(einode)->xtime, \
933 raw_inode->xtime ## _extra); \
935 (einode)->xtime.tv_nsec = 0; \
938 #define i_disk_version osd1.linux1.l_i_version
940 #if defined(__KERNEL__) || defined(__linux__)
941 #define i_reserved1 osd1.linux1.l_i_reserved1
942 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
943 #define i_blocks_high osd2.linux2.l_i_blocks_high
944 #define i_uid_low i_uid
945 #define i_gid_low i_gid
946 #define i_uid_high osd2.linux2.l_i_uid_high
947 #define i_gid_high osd2.linux2.l_i_gid_high
948 #define i_checksum_lo osd2.linux2.l_i_checksum_lo
950 #elif defined(__GNU__)
952 #define i_translator osd1.hurd1.h_i_translator
953 #define i_uid_high osd2.hurd2.h_i_uid_high
954 #define i_gid_high osd2.hurd2.h_i_gid_high
955 #define i_author osd2.hurd2.h_i_author
957 #elif defined(__masix__)
959 #define i_reserved1 osd1.masix1.m_i_reserved1
960 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
961 #define i_reserved2 osd2.masix2.m_i_reserved2
963 #endif /* defined(__KERNEL__) || defined(__linux__) */
965 #include "extents_status.h"
966 #include "fast_commit.h"
969 * Lock subclasses for i_data_sem in the ext4_inode_info structure.
971 * These are needed to avoid lockdep false positives when we need to
972 * allocate blocks to the quota inode during ext4_map_blocks(), while
973 * holding i_data_sem for a normal (non-quota) inode. Since we don't
974 * do quota tracking for the quota inode, this avoids deadlock (as
975 * well as infinite recursion, since it isn't turtles all the way
978 * I_DATA_SEM_NORMAL - Used for most inodes
979 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode
980 * where the second inode has larger inode number
982 * I_DATA_SEM_QUOTA - Used for quota inodes only
985 I_DATA_SEM_NORMAL = 0,
992 * fourth extended file system inode data in memory
994 struct ext4_inode_info {
995 __le32 i_data[15]; /* unconverted */
997 ext4_fsblk_t i_file_acl;
1000 * i_block_group is the number of the block group which contains
1001 * this file's inode. Constant across the lifetime of the inode,
1002 * it is used for making block allocation decisions - we try to
1003 * place a file's data blocks near its inode block, and new inodes
1004 * near to their parent directory's inode.
1006 ext4_group_t i_block_group;
1007 ext4_lblk_t i_dir_start_lookup;
1008 #if (BITS_PER_LONG < 64)
1009 unsigned long i_state_flags; /* Dynamic state flags */
1011 unsigned long i_flags;
1014 * Extended attributes can be read independently of the main file
1015 * data. Taking i_mutex even when reading would cause contention
1016 * between readers of EAs and writers of regular file data, so
1017 * instead we synchronize on xattr_sem when reading or changing
1020 struct rw_semaphore xattr_sem;
1022 struct list_head i_orphan; /* unlinked but open inodes */
1024 /* Fast commit related info */
1026 struct list_head i_fc_list; /*
1027 * inodes that need fast commit
1028 * protected by sbi->s_fc_lock.
1031 /* Fast commit subtid when this inode was committed */
1032 unsigned int i_fc_committed_subtid;
1034 /* Start of lblk range that needs to be committed in this fast commit */
1035 ext4_lblk_t i_fc_lblk_start;
1037 /* End of lblk range that needs to be committed in this fast commit */
1038 ext4_lblk_t i_fc_lblk_len;
1040 /* Number of ongoing updates on this inode */
1041 atomic_t i_fc_updates;
1043 /* Fast commit wait queue for this inode */
1044 wait_queue_head_t i_fc_wait;
1046 /* Protect concurrent accesses on i_fc_lblk_start, i_fc_lblk_len */
1047 struct mutex i_fc_lock;
1050 * i_disksize keeps track of what the inode size is ON DISK, not
1051 * in memory. During truncate, i_size is set to the new size by
1052 * the VFS prior to calling ext4_truncate(), but the filesystem won't
1053 * set i_disksize to 0 until the truncate is actually under way.
1055 * The intent is that i_disksize always represents the blocks which
1056 * are used by this file. This allows recovery to restart truncate
1057 * on orphans if we crash during truncate. We actually write i_disksize
1058 * into the on-disk inode when writing inodes out, instead of i_size.
1060 * The only time when i_disksize and i_size may be different is when
1061 * a truncate is in progress. The only things which change i_disksize
1062 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
1067 * i_data_sem is for serialising ext4_truncate() against
1068 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
1069 * data tree are chopped off during truncate. We can't do that in
1070 * ext4 because whenever we perform intermediate commits during
1071 * truncate, the inode and all the metadata blocks *must* be in a
1072 * consistent state which allows truncation of the orphans to restart
1073 * during recovery. Hence we must fix the get_block-vs-truncate race
1074 * by other means, so we have i_data_sem.
1076 struct rw_semaphore i_data_sem;
1078 * i_mmap_sem is for serializing page faults with truncate / punch hole
1079 * operations. We have to make sure that new page cannot be faulted in
1080 * a section of the inode that is being punched. We cannot easily use
1081 * i_data_sem for this since we need protection for the whole punch
1082 * operation and i_data_sem ranks below transaction start so we have
1083 * to occasionally drop it.
1085 struct rw_semaphore i_mmap_sem;
1086 struct inode vfs_inode;
1087 struct jbd2_inode *jinode;
1089 spinlock_t i_raw_lock; /* protects updates to the raw inode */
1092 * File creation time. Its function is same as that of
1093 * struct timespec64 i_{a,c,m}time in the generic inode.
1095 struct timespec64 i_crtime;
1098 atomic_t i_prealloc_active;
1099 struct list_head i_prealloc_list;
1100 spinlock_t i_prealloc_lock;
1102 /* extents status tree */
1103 struct ext4_es_tree i_es_tree;
1105 struct list_head i_es_list;
1106 unsigned int i_es_all_nr; /* protected by i_es_lock */
1107 unsigned int i_es_shk_nr; /* protected by i_es_lock */
1108 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for
1109 extents to shrink. Protected by
1113 ext4_group_t i_last_alloc_group;
1115 /* allocation reservation info for delalloc */
1116 /* In case of bigalloc, this refer to clusters rather than blocks */
1117 unsigned int i_reserved_data_blocks;
1119 /* pending cluster reservations for bigalloc file systems */
1120 struct ext4_pending_tree i_pending_tree;
1122 /* on-disk additional length */
1123 __u16 i_extra_isize;
1125 /* Indicate the inline data space. */
1130 /* quota space reservation, managed internally by quota code */
1131 qsize_t i_reserved_quota;
1134 /* Lock protecting lists below */
1135 spinlock_t i_completed_io_lock;
1137 * Completed IOs that need unwritten extents handling and have
1138 * transaction reserved
1140 struct list_head i_rsv_conversion_list;
1141 struct work_struct i_rsv_conversion_work;
1142 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1144 spinlock_t i_block_reservation_lock;
1147 * Transactions that contain inode's metadata needed to complete
1148 * fsync and fdatasync, respectively.
1151 tid_t i_datasync_tid;
1154 struct dquot *i_dquot[MAXQUOTAS];
1157 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1164 * File system states
1166 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
1167 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
1168 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
1169 #define EXT4_FC_REPLAY 0x0020 /* Fast commit replay ongoing */
1172 * Misc. filesystem flags
1174 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
1175 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
1176 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
1179 * Mount flags set via mount options or defaults
1181 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */
1182 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
1183 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
1184 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
1185 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
1186 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
1187 #define EXT4_MOUNT_ERRORS_MASK 0x00070
1188 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
1189 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
1190 #ifdef CONFIG_FS_DAX
1191 #define EXT4_MOUNT_DAX_ALWAYS 0x00200 /* Direct Access */
1193 #define EXT4_MOUNT_DAX_ALWAYS 0
1195 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
1196 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
1197 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
1198 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
1199 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
1200 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
1201 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
1202 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
1203 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
1204 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
1205 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */
1206 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota,
1207 * enable enforcement for hidden
1209 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable
1210 * enforcement for hidden quota
1212 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota
1214 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
1215 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
1216 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1217 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */
1218 #define EXT4_MOUNT_PREFETCH_BLOCK_BITMAPS 0x4000000
1219 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
1220 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
1221 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
1222 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
1223 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
1226 * Mount flags set either automatically (could not be set by mount option)
1227 * based on per file system feature or property or in special cases such as
1228 * distinguishing between explicit mount option definition and default.
1230 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1231 specified delalloc */
1232 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1233 size of blocksize * 8
1235 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1237 #define EXT4_MOUNT2_DAX_NEVER 0x00000008 /* Do not allow Direct Access */
1238 #define EXT4_MOUNT2_DAX_INODE 0x00000010 /* For printing options only */
1240 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly
1241 specified journal checksum */
1243 #define EXT4_MOUNT2_JOURNAL_FAST_COMMIT 0x00000010 /* Journal fast commit */
1245 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1247 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1249 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1252 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1254 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1256 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1259 #define ext4_test_and_set_bit __test_and_set_bit_le
1260 #define ext4_set_bit __set_bit_le
1261 #define ext4_set_bit_atomic ext2_set_bit_atomic
1262 #define ext4_test_and_clear_bit __test_and_clear_bit_le
1263 #define ext4_clear_bit __clear_bit_le
1264 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
1265 #define ext4_test_bit test_bit_le
1266 #define ext4_find_next_zero_bit find_next_zero_bit_le
1267 #define ext4_find_next_bit find_next_bit_le
1269 extern void ext4_set_bits(void *bm, int cur, int len);
1272 * Maximal mount counts between two filesystem checks
1274 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1275 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1278 * Behaviour when detecting errors
1280 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1281 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1282 #define EXT4_ERRORS_PANIC 3 /* Panic */
1283 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1285 /* Metadata checksum algorithm codes */
1286 #define EXT4_CRC32C_CHKSUM 1
1289 * Structure of the super block
1291 struct ext4_super_block {
1292 /*00*/ __le32 s_inodes_count; /* Inodes count */
1293 __le32 s_blocks_count_lo; /* Blocks count */
1294 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1295 __le32 s_free_blocks_count_lo; /* Free blocks count */
1296 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1297 __le32 s_first_data_block; /* First Data Block */
1298 __le32 s_log_block_size; /* Block size */
1299 __le32 s_log_cluster_size; /* Allocation cluster size */
1300 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1301 __le32 s_clusters_per_group; /* # Clusters per group */
1302 __le32 s_inodes_per_group; /* # Inodes per group */
1303 __le32 s_mtime; /* Mount time */
1304 /*30*/ __le32 s_wtime; /* Write time */
1305 __le16 s_mnt_count; /* Mount count */
1306 __le16 s_max_mnt_count; /* Maximal mount count */
1307 __le16 s_magic; /* Magic signature */
1308 __le16 s_state; /* File system state */
1309 __le16 s_errors; /* Behaviour when detecting errors */
1310 __le16 s_minor_rev_level; /* minor revision level */
1311 /*40*/ __le32 s_lastcheck; /* time of last check */
1312 __le32 s_checkinterval; /* max. time between checks */
1313 __le32 s_creator_os; /* OS */
1314 __le32 s_rev_level; /* Revision level */
1315 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1316 __le16 s_def_resgid; /* Default gid for reserved blocks */
1318 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1320 * Note: the difference between the compatible feature set and
1321 * the incompatible feature set is that if there is a bit set
1322 * in the incompatible feature set that the kernel doesn't
1323 * know about, it should refuse to mount the filesystem.
1325 * e2fsck's requirements are more strict; if it doesn't know
1326 * about a feature in either the compatible or incompatible
1327 * feature set, it must abort and not try to meddle with
1328 * things it doesn't understand...
1330 __le32 s_first_ino; /* First non-reserved inode */
1331 __le16 s_inode_size; /* size of inode structure */
1332 __le16 s_block_group_nr; /* block group # of this superblock */
1333 __le32 s_feature_compat; /* compatible feature set */
1334 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1335 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1336 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1337 /*78*/ char s_volume_name[16]; /* volume name */
1338 /*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */
1339 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1341 * Performance hints. Directory preallocation should only
1342 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1344 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1345 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1346 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1348 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1350 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1351 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1352 __le32 s_journal_dev; /* device number of journal file */
1353 __le32 s_last_orphan; /* start of list of inodes to delete */
1354 __le32 s_hash_seed[4]; /* HTREE hash seed */
1355 __u8 s_def_hash_version; /* Default hash version to use */
1356 __u8 s_jnl_backup_type;
1357 __le16 s_desc_size; /* size of group descriptor */
1358 /*100*/ __le32 s_default_mount_opts;
1359 __le32 s_first_meta_bg; /* First metablock block group */
1360 __le32 s_mkfs_time; /* When the filesystem was created */
1361 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1362 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1363 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1364 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1365 __le32 s_free_blocks_count_hi; /* Free blocks count */
1366 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1367 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1368 __le32 s_flags; /* Miscellaneous flags */
1369 __le16 s_raid_stride; /* RAID stride */
1370 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1371 __le64 s_mmp_block; /* Block for multi-mount protection */
1372 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1373 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1374 __u8 s_checksum_type; /* metadata checksum algorithm used */
1375 __u8 s_encryption_level; /* versioning level for encryption */
1376 __u8 s_reserved_pad; /* Padding to next 32bits */
1377 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1378 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1379 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1380 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1381 snapshot's future use */
1382 __le32 s_snapshot_list; /* inode number of the head of the
1383 on-disk snapshot list */
1384 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1385 __le32 s_error_count; /* number of fs errors */
1386 __le32 s_first_error_time; /* first time an error happened */
1387 __le32 s_first_error_ino; /* inode involved in first error */
1388 __le64 s_first_error_block; /* block involved of first error */
1389 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */
1390 __le32 s_first_error_line; /* line number where error happened */
1391 __le32 s_last_error_time; /* most recent time of an error */
1392 __le32 s_last_error_ino; /* inode involved in last error */
1393 __le32 s_last_error_line; /* line number where error happened */
1394 __le64 s_last_error_block; /* block involved of last error */
1395 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */
1396 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1397 __u8 s_mount_opts[64];
1398 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1399 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1400 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1401 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
1402 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */
1403 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */
1404 __le32 s_lpf_ino; /* Location of the lost+found inode */
1405 __le32 s_prj_quota_inum; /* inode for tracking project quota */
1406 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */
1409 __u8 s_mkfs_time_hi;
1410 __u8 s_lastcheck_hi;
1411 __u8 s_first_error_time_hi;
1412 __u8 s_last_error_time_hi;
1413 __u8 s_first_error_errcode;
1414 __u8 s_last_error_errcode;
1415 __le16 s_encoding; /* Filename charset encoding */
1416 __le16 s_encoding_flags; /* Filename charset encoding flags */
1417 __le32 s_reserved[95]; /* Padding to the end of the block */
1418 __le32 s_checksum; /* crc32c(superblock) */
1421 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1426 * run-time mount flags
1428 #define EXT4_MF_MNTDIR_SAMPLED 0x0001
1429 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1430 #define EXT4_MF_FC_INELIGIBLE 0x0004 /* Fast commit ineligible */
1431 #define EXT4_MF_FC_COMMITTING 0x0008 /* File system underoing a fast
1435 #ifdef CONFIG_FS_ENCRYPTION
1436 #define DUMMY_ENCRYPTION_ENABLED(sbi) ((sbi)->s_dummy_enc_policy.policy != NULL)
1438 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1441 /* Number of quota types we support */
1442 #define EXT4_MAXQUOTAS 3
1444 #define EXT4_ENC_UTF8_12_1 1
1447 * fourth extended-fs super-block data in memory
1449 struct ext4_sb_info {
1450 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1451 unsigned long s_inodes_per_block;/* Number of inodes per block */
1452 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1453 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1454 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1455 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1456 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1457 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1458 ext4_group_t s_groups_count; /* Number of groups in the fs */
1459 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1460 unsigned long s_overhead; /* # of fs overhead clusters */
1461 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1462 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1463 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1464 struct buffer_head * s_sbh; /* Buffer containing the super block */
1465 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1466 struct buffer_head * __rcu *s_group_desc;
1467 unsigned int s_mount_opt;
1468 unsigned int s_mount_opt2;
1469 unsigned int s_mount_flags;
1470 unsigned int s_def_mount_opt;
1471 ext4_fsblk_t s_sb_block;
1472 atomic64_t s_resv_clusters;
1475 unsigned short s_mount_state;
1476 unsigned short s_pad;
1477 int s_addr_per_block_bits;
1478 int s_desc_per_block_bits;
1481 unsigned int s_inode_readahead_blks;
1482 unsigned int s_inode_goal;
1484 int s_def_hash_version;
1485 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1486 struct percpu_counter s_freeclusters_counter;
1487 struct percpu_counter s_freeinodes_counter;
1488 struct percpu_counter s_dirs_counter;
1489 struct percpu_counter s_dirtyclusters_counter;
1490 struct blockgroup_lock *s_blockgroup_lock;
1491 struct proc_dir_entry *s_proc;
1492 struct kobject s_kobj;
1493 struct completion s_kobj_unregister;
1494 struct super_block *s_sb;
1497 struct journal_s *s_journal;
1498 struct list_head s_orphan;
1499 struct mutex s_orphan_lock;
1500 unsigned long s_ext4_flags; /* Ext4 superblock flags */
1501 unsigned long s_commit_interval;
1502 u32 s_max_batch_time;
1503 u32 s_min_batch_time;
1504 struct block_device *s_journal_bdev;
1506 /* Names of quota files with journalled quota */
1507 char __rcu *s_qf_names[EXT4_MAXQUOTAS];
1508 int s_jquota_fmt; /* Format of quota to use */
1510 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1511 struct ext4_system_blocks __rcu *s_system_blks;
1513 #ifdef EXTENTS_STATS
1514 /* ext4 extents stats */
1515 unsigned long s_ext_min;
1516 unsigned long s_ext_max;
1517 unsigned long s_depth_max;
1518 spinlock_t s_ext_stats_lock;
1519 unsigned long s_ext_blocks;
1520 unsigned long s_ext_extents;
1523 /* for buddy allocator */
1524 struct ext4_group_info ** __rcu *s_group_info;
1525 struct inode *s_buddy_cache;
1526 spinlock_t s_md_lock;
1527 unsigned short *s_mb_offsets;
1528 unsigned int *s_mb_maxs;
1529 unsigned int s_group_info_size;
1530 unsigned int s_mb_free_pending;
1531 struct list_head s_freed_data_list; /* List of blocks to be freed
1532 after commit completed */
1535 unsigned long s_stripe;
1536 unsigned int s_mb_stream_request;
1537 unsigned int s_mb_max_to_scan;
1538 unsigned int s_mb_min_to_scan;
1539 unsigned int s_mb_stats;
1540 unsigned int s_mb_order2_reqs;
1541 unsigned int s_mb_group_prealloc;
1542 unsigned int s_mb_max_inode_prealloc;
1543 unsigned int s_max_dir_size_kb;
1544 /* where last allocation was done - for stream allocation */
1545 unsigned long s_mb_last_group;
1546 unsigned long s_mb_last_start;
1547 unsigned int s_mb_prefetch;
1548 unsigned int s_mb_prefetch_limit;
1550 /* stats for buddy allocator */
1551 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1552 atomic_t s_bal_success; /* we found long enough chunks */
1553 atomic_t s_bal_allocated; /* in blocks */
1554 atomic_t s_bal_ex_scanned; /* total extents scanned */
1555 atomic_t s_bal_goals; /* goal hits */
1556 atomic_t s_bal_breaks; /* too long searches */
1557 atomic_t s_bal_2orders; /* 2^order hits */
1558 spinlock_t s_bal_lock;
1559 unsigned long s_mb_buddies_generated;
1560 unsigned long long s_mb_generation_time;
1561 atomic_t s_mb_lost_chunks;
1562 atomic_t s_mb_preallocated;
1563 atomic_t s_mb_discarded;
1564 atomic_t s_lock_busy;
1566 /* locality groups */
1567 struct ext4_locality_group __percpu *s_locality_groups;
1569 /* for write statistics */
1570 unsigned long s_sectors_written_start;
1571 u64 s_kbytes_written;
1573 /* the size of zero-out chunk */
1574 unsigned int s_extent_max_zeroout_kb;
1576 unsigned int s_log_groups_per_flex;
1577 struct flex_groups * __rcu *s_flex_groups;
1578 ext4_group_t s_flex_groups_allocated;
1580 /* workqueue for reserved extent conversions (buffered io) */
1581 struct workqueue_struct *rsv_conversion_wq;
1583 /* timer for periodic error stats printing */
1584 struct timer_list s_err_report;
1586 /* Lazy inode table initialization info */
1587 struct ext4_li_request *s_li_request;
1588 /* Wait multiplier for lazy initialization thread */
1589 unsigned int s_li_wait_mult;
1591 /* Kernel thread for multiple mount protection */
1592 struct task_struct *s_mmp_tsk;
1594 /* record the last minlen when FITRIM is called. */
1595 atomic_t s_last_trim_minblks;
1597 /* Reference to checksum algorithm driver via cryptoapi */
1598 struct crypto_shash *s_chksum_driver;
1600 /* Precomputed FS UUID checksum for seeding other checksums */
1603 /* Reclaim extents from extent status tree */
1604 struct shrinker s_es_shrinker;
1605 struct list_head s_es_list; /* List of inodes with reclaimable extents */
1607 struct ext4_es_stats s_es_stats;
1608 struct mb_cache *s_ea_block_cache;
1609 struct mb_cache *s_ea_inode_cache;
1610 spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1612 /* Ratelimit ext4 messages. */
1613 struct ratelimit_state s_err_ratelimit_state;
1614 struct ratelimit_state s_warning_ratelimit_state;
1615 struct ratelimit_state s_msg_ratelimit_state;
1616 atomic_t s_warning_count;
1617 atomic_t s_msg_count;
1619 /* Encryption policy for '-o test_dummy_encryption' */
1620 struct fscrypt_dummy_policy s_dummy_enc_policy;
1623 * Barrier between writepages ops and changing any inode's JOURNAL_DATA
1626 struct percpu_rw_semaphore s_writepages_rwsem;
1627 struct dax_device *s_daxdev;
1628 #ifdef CONFIG_EXT4_DEBUG
1629 unsigned long s_simulate_fail;
1631 /* Record the errseq of the backing block device */
1632 errseq_t s_bdev_wb_err;
1633 spinlock_t s_bdev_wb_lock;
1635 /* Ext4 fast commit stuff */
1636 atomic_t s_fc_subtid;
1637 atomic_t s_fc_ineligible_updates;
1639 * After commit starts, the main queue gets locked, and the further
1640 * updates get added in the staging queue.
1643 #define FC_Q_STAGING 1
1644 struct list_head s_fc_q[2]; /* Inodes staged for fast commit
1645 * that have data changes in them.
1647 struct list_head s_fc_dentry_q[2]; /* directory entry updates */
1648 unsigned int s_fc_bytes;
1650 * Main fast commit lock. This lock protects accesses to the
1652 * ei->i_fc_list, s_fc_dentry_q, s_fc_q, s_fc_bytes, s_fc_bh.
1654 spinlock_t s_fc_lock;
1655 struct buffer_head *s_fc_bh;
1656 struct ext4_fc_stats s_fc_stats;
1657 u64 s_fc_avg_commit_time;
1658 #ifdef CONFIG_EXT4_DEBUG
1659 int s_fc_debug_max_replay;
1661 struct ext4_fc_replay_state s_fc_replay_state;
1664 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1666 return sb->s_fs_info;
1668 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1670 return container_of(inode, struct ext4_inode_info, vfs_inode);
1673 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1675 return ino == EXT4_ROOT_INO ||
1676 (ino >= EXT4_FIRST_INO(sb) &&
1677 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1681 * Returns: sbi->field[index]
1682 * Used to access an array element from the following sbi fields which require
1683 * rcu protection to avoid dereferencing an invalid pointer due to reassignment
1688 #define sbi_array_rcu_deref(sbi, field, index) \
1690 typeof(*((sbi)->field)) _v; \
1692 _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index]; \
1693 rcu_read_unlock(); \
1698 * Simulate_fail codes
1700 #define EXT4_SIM_BBITMAP_EIO 1
1701 #define EXT4_SIM_BBITMAP_CRC 2
1702 #define EXT4_SIM_IBITMAP_EIO 3
1703 #define EXT4_SIM_IBITMAP_CRC 4
1704 #define EXT4_SIM_INODE_EIO 5
1705 #define EXT4_SIM_INODE_CRC 6
1706 #define EXT4_SIM_DIRBLOCK_EIO 7
1707 #define EXT4_SIM_DIRBLOCK_CRC 8
1709 static inline bool ext4_simulate_fail(struct super_block *sb,
1712 #ifdef CONFIG_EXT4_DEBUG
1713 struct ext4_sb_info *sbi = EXT4_SB(sb);
1715 if (unlikely(sbi->s_simulate_fail == code)) {
1716 sbi->s_simulate_fail = 0;
1723 static inline void ext4_simulate_fail_bh(struct super_block *sb,
1724 struct buffer_head *bh,
1727 if (!IS_ERR(bh) && ext4_simulate_fail(sb, code))
1728 clear_buffer_uptodate(bh);
1732 * Error number codes for s_{first,last}_error_errno
1734 * Linux errno numbers are architecture specific, so we need to translate
1735 * them into something which is architecture independent. We don't define
1736 * codes for all errno's; just the ones which are most likely to be the cause
1737 * of an ext4_error() call.
1739 #define EXT4_ERR_UNKNOWN 1
1740 #define EXT4_ERR_EIO 2
1741 #define EXT4_ERR_ENOMEM 3
1742 #define EXT4_ERR_EFSBADCRC 4
1743 #define EXT4_ERR_EFSCORRUPTED 5
1744 #define EXT4_ERR_ENOSPC 6
1745 #define EXT4_ERR_ENOKEY 7
1746 #define EXT4_ERR_EROFS 8
1747 #define EXT4_ERR_EFBIG 9
1748 #define EXT4_ERR_EEXIST 10
1749 #define EXT4_ERR_ERANGE 11
1750 #define EXT4_ERR_EOVERFLOW 12
1751 #define EXT4_ERR_EBUSY 13
1752 #define EXT4_ERR_ENOTDIR 14
1753 #define EXT4_ERR_ENOTEMPTY 15
1754 #define EXT4_ERR_ESHUTDOWN 16
1755 #define EXT4_ERR_EFAULT 17
1758 * Inode dynamic state flags
1761 EXT4_STATE_JDATA, /* journaled data exists */
1762 EXT4_STATE_NEW, /* inode is newly created */
1763 EXT4_STATE_XATTR, /* has in-inode xattrs */
1764 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1765 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1766 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1767 EXT4_STATE_NEWENTRY, /* File just added to dir */
1768 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1769 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
1770 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */
1771 EXT4_STATE_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */
1772 EXT4_STATE_FC_COMMITTING, /* Fast commit ongoing */
1775 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1776 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1778 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1780 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1782 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1784 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1786 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1789 /* Add these declarations here only so that these functions can be
1790 * found by name. Otherwise, they are very hard to locate. */
1791 static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1792 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1793 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1794 EXT4_INODE_BIT_FNS(flag, flags, 0)
1796 /* Add these declarations here only so that these functions can be
1797 * found by name. Otherwise, they are very hard to locate. */
1798 static inline int ext4_test_inode_state(struct inode *inode, int bit);
1799 static inline void ext4_set_inode_state(struct inode *inode, int bit);
1800 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1801 #if (BITS_PER_LONG < 64)
1802 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1804 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1806 (ei)->i_state_flags = 0;
1809 EXT4_INODE_BIT_FNS(state, flags, 32)
1811 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1813 /* We depend on the fact that callers will set i_flags */
1817 /* Assume that user mode programs are passing in an ext4fs superblock, not
1818 * a kernel struct super_block. This will allow us to call the feature-test
1819 * macros from user land. */
1820 #define EXT4_SB(sb) (sb)
1823 static inline bool ext4_verity_in_progress(struct inode *inode)
1825 return IS_ENABLED(CONFIG_FS_VERITY) &&
1826 ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS);
1829 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1832 * Codes for operating systems
1834 #define EXT4_OS_LINUX 0
1835 #define EXT4_OS_HURD 1
1836 #define EXT4_OS_MASIX 2
1837 #define EXT4_OS_FREEBSD 3
1838 #define EXT4_OS_LITES 4
1843 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1844 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1846 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1847 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1849 #define EXT4_GOOD_OLD_INODE_SIZE 128
1851 #define EXT4_EXTRA_TIMESTAMP_MAX (((s64)1 << 34) - 1 + S32_MIN)
1852 #define EXT4_NON_EXTRA_TIMESTAMP_MAX S32_MAX
1853 #define EXT4_TIMESTAMP_MIN S32_MIN
1856 * Feature set definitions
1859 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1860 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1861 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1862 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1863 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1864 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1865 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
1866 #define EXT4_FEATURE_COMPAT_FAST_COMMIT 0x0400
1867 #define EXT4_FEATURE_COMPAT_STABLE_INODES 0x0800
1869 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1870 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1871 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1872 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1873 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1874 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1875 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1876 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1877 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1879 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1880 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1881 * all other data structures' checksums. However, the METADATA_CSUM and
1882 * GDT_CSUM bits are mutually exclusive.
1884 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
1885 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000
1886 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000
1887 #define EXT4_FEATURE_RO_COMPAT_VERITY 0x8000
1889 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1890 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1891 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1892 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1893 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1894 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1895 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1896 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1897 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1898 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1899 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1900 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000
1901 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
1902 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
1903 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000
1904 #define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000
1906 extern void ext4_update_dynamic_rev(struct super_block *sb);
1908 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
1909 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1911 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1912 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
1914 static inline void ext4_set_feature_##name(struct super_block *sb) \
1916 ext4_update_dynamic_rev(sb); \
1917 EXT4_SB(sb)->s_es->s_feature_compat |= \
1918 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1920 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1922 EXT4_SB(sb)->s_es->s_feature_compat &= \
1923 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1926 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
1927 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1929 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1930 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
1932 static inline void ext4_set_feature_##name(struct super_block *sb) \
1934 ext4_update_dynamic_rev(sb); \
1935 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
1936 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1938 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1940 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
1941 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1944 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
1945 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1947 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1948 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
1950 static inline void ext4_set_feature_##name(struct super_block *sb) \
1952 ext4_update_dynamic_rev(sb); \
1953 EXT4_SB(sb)->s_es->s_feature_incompat |= \
1954 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1956 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1958 EXT4_SB(sb)->s_es->s_feature_incompat &= \
1959 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1962 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC)
1963 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES)
1964 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL)
1965 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR)
1966 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE)
1967 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX)
1968 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2)
1969 EXT4_FEATURE_COMPAT_FUNCS(fast_commit, FAST_COMMIT)
1970 EXT4_FEATURE_COMPAT_FUNCS(stable_inodes, STABLE_INODES)
1972 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER)
1973 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE)
1974 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR)
1975 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE)
1976 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM)
1977 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK)
1978 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE)
1979 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA)
1980 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC)
1981 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM)
1982 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY)
1983 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT)
1984 EXT4_FEATURE_RO_COMPAT_FUNCS(verity, VERITY)
1986 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION)
1987 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE)
1988 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER)
1989 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV)
1990 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG)
1991 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS)
1992 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT)
1993 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP)
1994 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG)
1995 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE)
1996 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA)
1997 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED)
1998 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR)
1999 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA)
2000 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT)
2001 EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD)
2003 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
2004 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
2005 EXT4_FEATURE_INCOMPAT_META_BG)
2006 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2007 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2008 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
2010 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
2011 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
2012 EXT4_FEATURE_INCOMPAT_RECOVER| \
2013 EXT4_FEATURE_INCOMPAT_META_BG)
2014 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2015 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2016 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
2018 #define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
2019 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
2020 EXT4_FEATURE_INCOMPAT_RECOVER| \
2021 EXT4_FEATURE_INCOMPAT_META_BG| \
2022 EXT4_FEATURE_INCOMPAT_EXTENTS| \
2023 EXT4_FEATURE_INCOMPAT_64BIT| \
2024 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
2025 EXT4_FEATURE_INCOMPAT_EA_INODE| \
2026 EXT4_FEATURE_INCOMPAT_MMP | \
2027 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
2028 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
2029 EXT4_FEATURE_INCOMPAT_CASEFOLD | \
2030 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \
2031 EXT4_FEATURE_INCOMPAT_LARGEDIR)
2032 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2033 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2034 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
2035 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
2036 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
2037 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
2038 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
2039 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
2040 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
2041 EXT4_FEATURE_RO_COMPAT_QUOTA |\
2042 EXT4_FEATURE_RO_COMPAT_PROJECT |\
2043 EXT4_FEATURE_RO_COMPAT_VERITY)
2045 #define EXTN_FEATURE_FUNCS(ver) \
2046 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
2048 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
2049 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
2051 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
2053 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
2054 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
2056 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
2058 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
2059 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
2062 EXTN_FEATURE_FUNCS(2)
2063 EXTN_FEATURE_FUNCS(3)
2064 EXTN_FEATURE_FUNCS(4)
2066 static inline bool ext4_has_compat_features(struct super_block *sb)
2068 return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
2070 static inline bool ext4_has_ro_compat_features(struct super_block *sb)
2072 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
2074 static inline bool ext4_has_incompat_features(struct super_block *sb)
2076 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
2082 #define EXT4_FLAGS_RESIZING 0
2083 #define EXT4_FLAGS_SHUTDOWN 1
2084 #define EXT4_FLAGS_BDEV_IS_DAX 2
2086 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi)
2088 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags);
2093 * Default values for user and/or group using reserved blocks
2095 #define EXT4_DEF_RESUID 0
2096 #define EXT4_DEF_RESGID 0
2099 * Default project ID
2101 #define EXT4_DEF_PROJID 0
2103 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
2106 * Default mount options
2108 #define EXT4_DEFM_DEBUG 0x0001
2109 #define EXT4_DEFM_BSDGROUPS 0x0002
2110 #define EXT4_DEFM_XATTR_USER 0x0004
2111 #define EXT4_DEFM_ACL 0x0008
2112 #define EXT4_DEFM_UID16 0x0010
2113 #define EXT4_DEFM_JMODE 0x0060
2114 #define EXT4_DEFM_JMODE_DATA 0x0020
2115 #define EXT4_DEFM_JMODE_ORDERED 0x0040
2116 #define EXT4_DEFM_JMODE_WBACK 0x0060
2117 #define EXT4_DEFM_NOBARRIER 0x0100
2118 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
2119 #define EXT4_DEFM_DISCARD 0x0400
2120 #define EXT4_DEFM_NODELALLOC 0x0800
2123 * Default journal batch times
2125 #define EXT4_DEF_MIN_BATCH_TIME 0
2126 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
2129 * Minimum number of groups in a flexgroup before we separate out
2130 * directories into the first block group of a flexgroup
2132 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
2135 * Structure of a directory entry
2137 #define EXT4_NAME_LEN 255
2139 struct ext4_dir_entry {
2140 __le32 inode; /* Inode number */
2141 __le16 rec_len; /* Directory entry length */
2142 __le16 name_len; /* Name length */
2143 char name[EXT4_NAME_LEN]; /* File name */
2147 * The new version of the directory entry. Since EXT4 structures are
2148 * stored in intel byte order, and the name_len field could never be
2149 * bigger than 255 chars, it's safe to reclaim the extra byte for the
2152 struct ext4_dir_entry_2 {
2153 __le32 inode; /* Inode number */
2154 __le16 rec_len; /* Directory entry length */
2155 __u8 name_len; /* Name length */
2156 __u8 file_type; /* See file type macros EXT4_FT_* below */
2157 char name[EXT4_NAME_LEN]; /* File name */
2161 * This is a bogus directory entry at the end of each leaf block that
2162 * records checksums.
2164 struct ext4_dir_entry_tail {
2165 __le32 det_reserved_zero1; /* Pretend to be unused */
2166 __le16 det_rec_len; /* 12 */
2167 __u8 det_reserved_zero2; /* Zero name length */
2168 __u8 det_reserved_ft; /* 0xDE, fake file type */
2169 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
2172 #define EXT4_DIRENT_TAIL(block, blocksize) \
2173 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
2175 sizeof(struct ext4_dir_entry_tail))))
2178 * Ext4 directory file types. Only the low 3 bits are used. The
2179 * other bits are reserved for now.
2181 #define EXT4_FT_UNKNOWN 0
2182 #define EXT4_FT_REG_FILE 1
2183 #define EXT4_FT_DIR 2
2184 #define EXT4_FT_CHRDEV 3
2185 #define EXT4_FT_BLKDEV 4
2186 #define EXT4_FT_FIFO 5
2187 #define EXT4_FT_SOCK 6
2188 #define EXT4_FT_SYMLINK 7
2190 #define EXT4_FT_MAX 8
2192 #define EXT4_FT_DIR_CSUM 0xDE
2195 * EXT4_DIR_PAD defines the directory entries boundaries
2197 * NOTE: It must be a multiple of 4
2199 #define EXT4_DIR_PAD 4
2200 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
2201 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
2203 #define EXT4_MAX_REC_LEN ((1<<16)-1)
2206 * If we ever get support for fs block sizes > page_size, we'll need
2207 * to remove the #if statements in the next two functions...
2209 static inline unsigned int
2210 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
2212 unsigned len = le16_to_cpu(dlen);
2214 #if (PAGE_SIZE >= 65536)
2215 if (len == EXT4_MAX_REC_LEN || len == 0)
2217 return (len & 65532) | ((len & 3) << 16);
2223 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
2225 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
2227 #if (PAGE_SIZE >= 65536)
2229 return cpu_to_le16(len);
2230 if (len == blocksize) {
2231 if (blocksize == 65536)
2232 return cpu_to_le16(EXT4_MAX_REC_LEN);
2234 return cpu_to_le16(0);
2236 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
2238 return cpu_to_le16(len);
2243 * Hash Tree Directory indexing
2244 * (c) Daniel Phillips, 2001
2247 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
2248 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2249 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \
2250 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir)))
2251 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2253 /* Legal values for the dx_root hash_version field: */
2255 #define DX_HASH_LEGACY 0
2256 #define DX_HASH_HALF_MD4 1
2257 #define DX_HASH_TEA 2
2258 #define DX_HASH_LEGACY_UNSIGNED 3
2259 #define DX_HASH_HALF_MD4_UNSIGNED 4
2260 #define DX_HASH_TEA_UNSIGNED 5
2262 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2263 const void *address, unsigned int length)
2266 struct shash_desc shash;
2270 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2272 desc.shash.tfm = sbi->s_chksum_driver;
2273 *(u32 *)desc.ctx = crc;
2275 BUG_ON(crypto_shash_update(&desc.shash, address, length));
2277 return *(u32 *)desc.ctx;
2282 /* hash info structure used by the directory hash */
2292 /* 32 and 64 bit signed EOF for dx directories */
2293 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
2294 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
2298 * Control parameters used by ext4_htree_next_block
2300 #define HASH_NB_ALWAYS 1
2302 struct ext4_filename {
2303 const struct qstr *usr_fname;
2304 struct fscrypt_str disk_name;
2305 struct dx_hash_info hinfo;
2306 #ifdef CONFIG_FS_ENCRYPTION
2307 struct fscrypt_str crypto_buf;
2309 #ifdef CONFIG_UNICODE
2310 struct fscrypt_str cf_name;
2314 #define fname_name(p) ((p)->disk_name.name)
2315 #define fname_len(p) ((p)->disk_name.len)
2318 * Describe an inode's exact location on disk and in memory
2322 struct buffer_head *bh;
2323 unsigned long offset;
2324 ext4_group_t block_group;
2327 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2329 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2332 static inline bool ext4_is_quota_file(struct inode *inode)
2334 return IS_NOQUOTA(inode) &&
2335 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL);
2339 * This structure is stuffed into the struct file's private_data field
2340 * for directories. It is where we put information so that we can do
2341 * readdir operations in hash tree order.
2343 struct dir_private_info {
2344 struct rb_root root;
2345 struct rb_node *curr_node;
2346 struct fname *extra_fname;
2349 __u32 curr_minor_hash;
2353 /* calculate the first block number of the group */
2354 static inline ext4_fsblk_t
2355 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2357 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2358 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2362 * Special error return code only used by dx_probe() and its callers.
2364 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
2366 /* htree levels for ext4 */
2367 #define EXT4_HTREE_LEVEL_COMPAT 2
2368 #define EXT4_HTREE_LEVEL 3
2370 static inline int ext4_dir_htree_level(struct super_block *sb)
2372 return ext4_has_feature_largedir(sb) ?
2373 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT;
2377 * Timeout and state flag for lazy initialization inode thread.
2379 #define EXT4_DEF_LI_WAIT_MULT 10
2380 #define EXT4_DEF_LI_MAX_START_DELAY 5
2381 #define EXT4_LAZYINIT_QUIT 0x0001
2382 #define EXT4_LAZYINIT_RUNNING 0x0002
2385 * Lazy inode table initialization info
2387 struct ext4_lazy_init {
2388 unsigned long li_state;
2389 struct list_head li_request_list;
2390 struct mutex li_list_mtx;
2394 EXT4_LI_MODE_PREFETCH_BBITMAP,
2395 EXT4_LI_MODE_ITABLE,
2398 struct ext4_li_request {
2399 struct super_block *lr_super;
2400 enum ext4_li_mode lr_mode;
2401 ext4_group_t lr_first_not_zeroed;
2402 ext4_group_t lr_next_group;
2403 struct list_head lr_request;
2404 unsigned long lr_next_sched;
2405 unsigned long lr_timeout;
2408 struct ext4_features {
2409 struct kobject f_kobj;
2410 struct completion f_kobj_unregister;
2414 * This structure will be used for multiple mount protection. It will be
2415 * written into the block number saved in the s_mmp_block field in the
2416 * superblock. Programs that check MMP should assume that if
2417 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2418 * to use the filesystem, regardless of how old the timestamp is.
2420 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
2421 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2422 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
2423 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
2426 __le32 mmp_magic; /* Magic number for MMP */
2427 __le32 mmp_seq; /* Sequence no. updated periodically */
2430 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2431 * purposes and do not affect the correctness of the algorithm
2433 __le64 mmp_time; /* Time last updated */
2434 char mmp_nodename[64]; /* Node which last updated MMP block */
2435 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
2438 * mmp_check_interval is used to verify if the MMP block has been
2439 * updated on the block device. The value is updated based on the
2440 * maximum time to write the MMP block during an update cycle.
2442 __le16 mmp_check_interval;
2445 __le32 mmp_pad2[226];
2446 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
2449 /* arguments passed to the mmp thread */
2451 struct buffer_head *bh; /* bh from initial read_mmp_block() */
2452 struct super_block *sb; /* super block of the fs */
2456 * Check interval multiplier
2457 * The MMP block is written every update interval and initially checked every
2458 * update interval x the multiplier (the value is then adapted based on the
2459 * write latency). The reason is that writes can be delayed under load and we
2460 * don't want readers to incorrectly assume that the filesystem is no longer
2463 #define EXT4_MMP_CHECK_MULT 2UL
2466 * Minimum interval for MMP checking in seconds.
2468 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
2471 * Maximum interval for MMP checking in seconds.
2473 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
2476 * Function prototypes
2480 * Ok, these declarations are also in <linux/kernel.h> but none of the
2481 * ext4 source programs needs to include it so they are duplicated here.
2483 # define NORET_TYPE /**/
2484 # define ATTRIB_NORET __attribute__((noreturn))
2485 # define NORET_AND noreturn,
2488 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2489 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2490 struct ext4_group_desc *gdp,
2491 struct buffer_head *bh, int sz);
2492 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2493 struct ext4_group_desc *gdp,
2494 struct buffer_head *bh, int sz);
2495 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2496 struct ext4_group_desc *gdp,
2497 struct buffer_head *bh);
2498 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2499 struct ext4_group_desc *gdp,
2500 struct buffer_head *bh);
2503 extern void ext4_get_group_no_and_offset(struct super_block *sb,
2504 ext4_fsblk_t blocknr,
2505 ext4_group_t *blockgrpp,
2506 ext4_grpblk_t *offsetp);
2507 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2508 ext4_fsblk_t block);
2510 extern unsigned int ext4_block_group(struct super_block *sb,
2511 ext4_fsblk_t blocknr);
2512 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2513 ext4_fsblk_t blocknr);
2514 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2515 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2516 ext4_group_t group);
2517 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2520 unsigned long *count,
2522 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2523 s64 nclusters, unsigned int flags);
2524 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2525 extern void ext4_check_blocks_bitmap(struct super_block *);
2526 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2527 ext4_group_t block_group,
2528 struct buffer_head ** bh);
2529 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2531 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2532 ext4_group_t block_group,
2533 bool ignore_locked);
2534 extern int ext4_wait_block_bitmap(struct super_block *sb,
2535 ext4_group_t block_group,
2536 struct buffer_head *bh);
2537 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2538 ext4_group_t block_group);
2539 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2540 ext4_group_t block_group,
2541 struct ext4_group_desc *gdp);
2542 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2544 #ifdef CONFIG_UNICODE
2545 extern void ext4_fname_setup_ci_filename(struct inode *dir,
2546 const struct qstr *iname,
2547 struct fscrypt_str *fname);
2550 #ifdef CONFIG_FS_ENCRYPTION
2551 static inline void ext4_fname_from_fscrypt_name(struct ext4_filename *dst,
2552 const struct fscrypt_name *src)
2554 memset(dst, 0, sizeof(*dst));
2556 dst->usr_fname = src->usr_fname;
2557 dst->disk_name = src->disk_name;
2558 dst->hinfo.hash = src->hash;
2559 dst->hinfo.minor_hash = src->minor_hash;
2560 dst->crypto_buf = src->crypto_buf;
2563 static inline int ext4_fname_setup_filename(struct inode *dir,
2564 const struct qstr *iname,
2566 struct ext4_filename *fname)
2568 struct fscrypt_name name;
2571 err = fscrypt_setup_filename(dir, iname, lookup, &name);
2575 ext4_fname_from_fscrypt_name(fname, &name);
2577 #ifdef CONFIG_UNICODE
2578 ext4_fname_setup_ci_filename(dir, iname, &fname->cf_name);
2583 static inline int ext4_fname_prepare_lookup(struct inode *dir,
2584 struct dentry *dentry,
2585 struct ext4_filename *fname)
2587 struct fscrypt_name name;
2590 err = fscrypt_prepare_lookup(dir, dentry, &name);
2594 ext4_fname_from_fscrypt_name(fname, &name);
2596 #ifdef CONFIG_UNICODE
2597 ext4_fname_setup_ci_filename(dir, &dentry->d_name, &fname->cf_name);
2602 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2604 struct fscrypt_name name;
2606 name.crypto_buf = fname->crypto_buf;
2607 fscrypt_free_filename(&name);
2609 fname->crypto_buf.name = NULL;
2610 fname->usr_fname = NULL;
2611 fname->disk_name.name = NULL;
2613 #ifdef CONFIG_UNICODE
2614 kfree(fname->cf_name.name);
2615 fname->cf_name.name = NULL;
2618 #else /* !CONFIG_FS_ENCRYPTION */
2619 static inline int ext4_fname_setup_filename(struct inode *dir,
2620 const struct qstr *iname,
2622 struct ext4_filename *fname)
2624 fname->usr_fname = iname;
2625 fname->disk_name.name = (unsigned char *) iname->name;
2626 fname->disk_name.len = iname->len;
2628 #ifdef CONFIG_UNICODE
2629 ext4_fname_setup_ci_filename(dir, iname, &fname->cf_name);
2635 static inline int ext4_fname_prepare_lookup(struct inode *dir,
2636 struct dentry *dentry,
2637 struct ext4_filename *fname)
2639 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname);
2642 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2644 #ifdef CONFIG_UNICODE
2645 kfree(fname->cf_name.name);
2646 fname->cf_name.name = NULL;
2649 #endif /* !CONFIG_FS_ENCRYPTION */
2652 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2654 struct ext4_dir_entry_2 *,
2655 struct buffer_head *, char *, int,
2657 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2658 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2659 (de), (bh), (buf), (size), (offset)))
2660 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2662 struct ext4_dir_entry_2 *dirent,
2663 struct fscrypt_str *ent_name);
2664 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2665 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2666 struct buffer_head *bh,
2667 void *buf, int buf_size,
2668 struct ext4_filename *fname,
2669 struct ext4_dir_entry_2 **dest_de);
2670 void ext4_insert_dentry(struct inode *inode,
2671 struct ext4_dir_entry_2 *de,
2673 struct ext4_filename *fname);
2674 static inline void ext4_update_dx_flag(struct inode *inode)
2676 if (!ext4_has_feature_dir_index(inode->i_sb)) {
2677 /* ext4_iget() should have caught this... */
2678 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb));
2679 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2682 static const unsigned char ext4_filetype_table[] = {
2683 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2686 static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2688 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2691 return ext4_filetype_table[filetype];
2693 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2694 void *buf, int buf_size);
2697 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2700 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len,
2701 struct dx_hash_info *hinfo);
2704 extern int ext4_mark_inode_used(struct super_block *sb, int ino);
2705 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2706 const struct qstr *qstr, __u32 goal,
2707 uid_t *owner, __u32 i_flags,
2708 int handle_type, unsigned int line_no,
2711 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \
2712 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2714 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2716 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2717 0, (type), __LINE__, (nblocks))
2720 extern void ext4_free_inode(handle_t *, struct inode *);
2721 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2722 extern unsigned long ext4_count_free_inodes(struct super_block *);
2723 extern unsigned long ext4_count_dirs(struct super_block *);
2724 extern void ext4_check_inodes_bitmap(struct super_block *);
2725 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2726 extern int ext4_init_inode_table(struct super_block *sb,
2727 ext4_group_t group, int barrier);
2728 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2731 int ext4_fc_info_show(struct seq_file *seq, void *v);
2732 void ext4_fc_init(struct super_block *sb, journal_t *journal);
2733 void ext4_fc_init_inode(struct inode *inode);
2734 void ext4_fc_track_range(struct inode *inode, ext4_lblk_t start,
2736 void ext4_fc_track_unlink(struct inode *inode, struct dentry *dentry);
2737 void ext4_fc_track_link(struct inode *inode, struct dentry *dentry);
2738 void ext4_fc_track_create(struct inode *inode, struct dentry *dentry);
2739 void ext4_fc_track_inode(struct inode *inode);
2740 void ext4_fc_mark_ineligible(struct super_block *sb, int reason);
2741 void ext4_fc_start_ineligible(struct super_block *sb, int reason);
2742 void ext4_fc_stop_ineligible(struct super_block *sb);
2743 void ext4_fc_start_update(struct inode *inode);
2744 void ext4_fc_stop_update(struct inode *inode);
2745 void ext4_fc_del(struct inode *inode);
2746 bool ext4_fc_replay_check_excluded(struct super_block *sb, ext4_fsblk_t block);
2747 void ext4_fc_replay_cleanup(struct super_block *sb);
2748 int ext4_fc_commit(journal_t *journal, tid_t commit_tid);
2749 int __init ext4_fc_init_dentry_cache(void);
2752 extern const struct seq_operations ext4_mb_seq_groups_ops;
2753 extern long ext4_mb_stats;
2754 extern long ext4_mb_max_to_scan;
2755 extern int ext4_mb_init(struct super_block *);
2756 extern int ext4_mb_release(struct super_block *);
2757 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2758 struct ext4_allocation_request *, int *);
2759 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2760 extern void ext4_discard_preallocations(struct inode *, unsigned int);
2761 extern int __init ext4_init_mballoc(void);
2762 extern void ext4_exit_mballoc(void);
2763 extern ext4_group_t ext4_mb_prefetch(struct super_block *sb,
2765 unsigned int nr, int *cnt);
2766 extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group,
2769 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2770 struct buffer_head *bh, ext4_fsblk_t block,
2771 unsigned long count, int flags);
2772 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2773 ext4_group_t ngroups);
2774 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2775 ext4_group_t i, struct ext4_group_desc *desc);
2776 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2777 ext4_fsblk_t block, unsigned long count);
2778 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2779 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid);
2780 extern void ext4_mb_mark_bb(struct super_block *sb, ext4_fsblk_t block,
2781 int len, int state);
2784 void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw,
2785 struct ext4_inode_info *ei);
2786 int ext4_inode_is_fast_symlink(struct inode *inode);
2787 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2788 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2789 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count,
2790 bool wait, struct buffer_head **bhs);
2791 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2792 struct buffer_head *bh_result, int create);
2793 int ext4_get_block(struct inode *inode, sector_t iblock,
2794 struct buffer_head *bh_result, int create);
2795 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2796 struct buffer_head *bh, int create);
2797 int ext4_walk_page_buffers(handle_t *handle,
2798 struct buffer_head *head,
2802 int (*fn)(handle_t *handle,
2803 struct buffer_head *bh));
2804 int do_journal_get_write_access(handle_t *handle,
2805 struct buffer_head *bh);
2806 #define FALL_BACK_TO_NONDELALLOC 1
2807 #define CONVERT_INLINE_DATA 2
2810 EXT4_IGET_NORMAL = 0,
2811 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */
2812 EXT4_IGET_HANDLE = 0x0002 /* Inode # is from a handle */
2815 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
2816 ext4_iget_flags flags, const char *function,
2819 #define ext4_iget(sb, ino, flags) \
2820 __ext4_iget((sb), (ino), (flags), __func__, __LINE__)
2822 extern int ext4_write_inode(struct inode *, struct writeback_control *);
2823 extern int ext4_setattr(struct dentry *, struct iattr *);
2824 extern int ext4_getattr(const struct path *, struct kstat *, u32, unsigned int);
2825 extern void ext4_evict_inode(struct inode *);
2826 extern void ext4_clear_inode(struct inode *);
2827 extern int ext4_file_getattr(const struct path *, struct kstat *, u32, unsigned int);
2828 extern int ext4_sync_inode(handle_t *, struct inode *);
2829 extern void ext4_dirty_inode(struct inode *, int);
2830 extern int ext4_change_inode_journal_flag(struct inode *, int);
2831 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2832 extern int ext4_get_fc_inode_loc(struct super_block *sb, unsigned long ino,
2833 struct ext4_iloc *iloc);
2834 extern int ext4_inode_attach_jinode(struct inode *inode);
2835 extern int ext4_can_truncate(struct inode *inode);
2836 extern int ext4_truncate(struct inode *);
2837 extern int ext4_break_layouts(struct inode *);
2838 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2839 extern void ext4_set_inode_flags(struct inode *, bool init);
2840 extern int ext4_alloc_da_blocks(struct inode *inode);
2841 extern void ext4_set_aops(struct inode *inode);
2842 extern int ext4_writepage_trans_blocks(struct inode *);
2843 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2844 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2845 loff_t lstart, loff_t lend);
2846 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf);
2847 extern vm_fault_t ext4_filemap_fault(struct vm_fault *vmf);
2848 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2849 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
2850 extern void ext4_da_release_space(struct inode *inode, int to_free);
2851 extern void ext4_da_update_reserve_space(struct inode *inode,
2852 int used, int quota_claim);
2853 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
2854 ext4_fsblk_t pblk, ext4_lblk_t len);
2857 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2858 struct ext4_map_blocks *map, int flags);
2859 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2860 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2861 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2862 ext4_lblk_t start, ext4_lblk_t end);
2865 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2866 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2867 extern void ext4_reset_inode_seed(struct inode *inode);
2870 extern int ext4_ext_migrate(struct inode *);
2871 extern int ext4_ind_migrate(struct inode *inode);
2874 extern int ext4_init_new_dir(handle_t *handle, struct inode *dir,
2875 struct inode *inode);
2876 extern int ext4_dirblock_csum_verify(struct inode *inode,
2877 struct buffer_head *bh);
2878 extern int ext4_orphan_add(handle_t *, struct inode *);
2879 extern int ext4_orphan_del(handle_t *, struct inode *);
2880 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2881 __u32 start_minor_hash, __u32 *next_hash);
2882 extern int ext4_search_dir(struct buffer_head *bh,
2886 struct ext4_filename *fname,
2887 unsigned int offset,
2888 struct ext4_dir_entry_2 **res_dir);
2889 extern int ext4_generic_delete_entry(struct inode *dir,
2890 struct ext4_dir_entry_2 *de_del,
2891 struct buffer_head *bh,
2895 extern bool ext4_empty_dir(struct inode *inode);
2898 extern void ext4_kvfree_array_rcu(void *to_free);
2899 extern int ext4_group_add(struct super_block *sb,
2900 struct ext4_new_group_data *input);
2901 extern int ext4_group_extend(struct super_block *sb,
2902 struct ext4_super_block *es,
2903 ext4_fsblk_t n_blocks_count);
2904 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2907 extern struct buffer_head *ext4_sb_bread(struct super_block *sb,
2908 sector_t block, int op_flags);
2909 extern struct buffer_head *ext4_sb_bread_unmovable(struct super_block *sb,
2911 extern void ext4_read_bh_nowait(struct buffer_head *bh, int op_flags,
2912 bh_end_io_t *end_io);
2913 extern int ext4_read_bh(struct buffer_head *bh, int op_flags,
2914 bh_end_io_t *end_io);
2915 extern int ext4_read_bh_lock(struct buffer_head *bh, int op_flags, bool wait);
2916 extern void ext4_sb_breadahead_unmovable(struct super_block *sb, sector_t block);
2917 extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
2918 extern int ext4_calculate_overhead(struct super_block *sb);
2919 extern void ext4_superblock_csum_set(struct super_block *sb);
2920 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2921 ext4_group_t ngroup);
2922 extern const char *ext4_decode_error(struct super_block *sb, int errno,
2924 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb,
2925 ext4_group_t block_group,
2926 unsigned int flags);
2928 extern __printf(6, 7)
2929 void __ext4_error(struct super_block *, const char *, unsigned int, int, __u64,
2931 extern __printf(6, 7)
2932 void __ext4_error_inode(struct inode *, const char *, unsigned int,
2933 ext4_fsblk_t, int, const char *, ...);
2934 extern __printf(5, 6)
2935 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2937 extern void __ext4_std_error(struct super_block *, const char *,
2939 extern __printf(5, 6)
2940 void __ext4_abort(struct super_block *, const char *, unsigned int, int,
2942 extern __printf(4, 5)
2943 void __ext4_warning(struct super_block *, const char *, unsigned int,
2945 extern __printf(4, 5)
2946 void __ext4_warning_inode(const struct inode *inode, const char *function,
2947 unsigned int line, const char *fmt, ...);
2948 extern __printf(3, 4)
2949 void __ext4_msg(struct super_block *, const char *, const char *, ...);
2950 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2951 const char *, unsigned int, const char *);
2952 extern __printf(7, 8)
2953 void __ext4_grp_locked_error(const char *, unsigned int,
2954 struct super_block *, ext4_group_t,
2955 unsigned long, ext4_fsblk_t,
2958 #define EXT4_ERROR_INODE(inode, fmt, a...) \
2959 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
2961 #define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...) \
2962 __ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a)
2964 #define ext4_error_inode_block(inode, block, err, fmt, a...) \
2965 __ext4_error_inode((inode), __func__, __LINE__, (block), (err), \
2968 #define EXT4_ERROR_FILE(file, block, fmt, a...) \
2969 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
2971 #ifdef CONFIG_PRINTK
2973 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2974 __ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__)
2975 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \
2976 __ext4_error_inode((inode), (func), (line), (block), \
2977 (err), (fmt), ##__VA_ARGS__)
2978 #define ext4_error_file(file, func, line, block, fmt, ...) \
2979 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2980 #define ext4_error(sb, fmt, ...) \
2981 __ext4_error((sb), __func__, __LINE__, 0, 0, (fmt), ##__VA_ARGS__)
2982 #define ext4_error_err(sb, err, fmt, ...) \
2983 __ext4_error((sb), __func__, __LINE__, (err), 0, (fmt), ##__VA_ARGS__)
2984 #define ext4_abort(sb, err, fmt, ...) \
2985 __ext4_abort((sb), __func__, __LINE__, (err), (fmt), ##__VA_ARGS__)
2986 #define ext4_warning(sb, fmt, ...) \
2987 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2988 #define ext4_warning_inode(inode, fmt, ...) \
2989 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
2990 #define ext4_msg(sb, level, fmt, ...) \
2991 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2992 #define dump_mmp_msg(sb, mmp, msg) \
2993 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2994 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2995 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
3000 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
3002 no_printk(fmt, ##__VA_ARGS__); \
3003 __ext4_error_inode(inode, "", 0, block, 0, " "); \
3005 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \
3007 no_printk(fmt, ##__VA_ARGS__); \
3008 __ext4_error_inode(inode, "", 0, block, err, " "); \
3010 #define ext4_error_file(file, func, line, block, fmt, ...) \
3012 no_printk(fmt, ##__VA_ARGS__); \
3013 __ext4_error_file(file, "", 0, block, " "); \
3015 #define ext4_error(sb, fmt, ...) \
3017 no_printk(fmt, ##__VA_ARGS__); \
3018 __ext4_error(sb, "", 0, 0, 0, " "); \
3020 #define ext4_error_err(sb, err, fmt, ...) \
3022 no_printk(fmt, ##__VA_ARGS__); \
3023 __ext4_error(sb, "", 0, err, 0, " "); \
3025 #define ext4_abort(sb, err, fmt, ...) \
3027 no_printk(fmt, ##__VA_ARGS__); \
3028 __ext4_abort(sb, "", 0, err, " "); \
3030 #define ext4_warning(sb, fmt, ...) \
3032 no_printk(fmt, ##__VA_ARGS__); \
3033 __ext4_warning(sb, "", 0, " "); \
3035 #define ext4_warning_inode(inode, fmt, ...) \
3037 no_printk(fmt, ##__VA_ARGS__); \
3038 __ext4_warning_inode(inode, "", 0, " "); \
3040 #define ext4_msg(sb, level, fmt, ...) \
3042 no_printk(fmt, ##__VA_ARGS__); \
3043 __ext4_msg(sb, "", " "); \
3045 #define dump_mmp_msg(sb, mmp, msg) \
3046 __dump_mmp_msg(sb, mmp, "", 0, "")
3047 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
3049 no_printk(fmt, ##__VA_ARGS__); \
3050 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
3055 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
3056 struct ext4_group_desc *bg);
3057 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
3058 struct ext4_group_desc *bg);
3059 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
3060 struct ext4_group_desc *bg);
3061 extern __u32 ext4_free_group_clusters(struct super_block *sb,
3062 struct ext4_group_desc *bg);
3063 extern __u32 ext4_free_inodes_count(struct super_block *sb,
3064 struct ext4_group_desc *bg);
3065 extern __u32 ext4_used_dirs_count(struct super_block *sb,
3066 struct ext4_group_desc *bg);
3067 extern __u32 ext4_itable_unused_count(struct super_block *sb,
3068 struct ext4_group_desc *bg);
3069 extern void ext4_block_bitmap_set(struct super_block *sb,
3070 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3071 extern void ext4_inode_bitmap_set(struct super_block *sb,
3072 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3073 extern void ext4_inode_table_set(struct super_block *sb,
3074 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3075 extern void ext4_free_group_clusters_set(struct super_block *sb,
3076 struct ext4_group_desc *bg,
3078 extern void ext4_free_inodes_set(struct super_block *sb,
3079 struct ext4_group_desc *bg, __u32 count);
3080 extern void ext4_used_dirs_set(struct super_block *sb,
3081 struct ext4_group_desc *bg, __u32 count);
3082 extern void ext4_itable_unused_set(struct super_block *sb,
3083 struct ext4_group_desc *bg, __u32 count);
3084 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
3085 struct ext4_group_desc *gdp);
3086 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
3087 struct ext4_group_desc *gdp);
3088 extern int ext4_register_li_request(struct super_block *sb,
3089 ext4_group_t first_not_zeroed);
3091 static inline int ext4_has_metadata_csum(struct super_block *sb)
3093 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
3094 !EXT4_SB(sb)->s_chksum_driver);
3096 return ext4_has_feature_metadata_csum(sb) &&
3097 (EXT4_SB(sb)->s_chksum_driver != NULL);
3100 static inline int ext4_has_group_desc_csum(struct super_block *sb)
3102 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb);
3105 #define ext4_read_incompat_64bit_val(es, name) \
3106 (((es)->s_feature_incompat & cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT) \
3107 ? (ext4_fsblk_t)le32_to_cpu(es->name##_hi) << 32 : 0) | \
3108 le32_to_cpu(es->name##_lo))
3110 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
3112 return ext4_read_incompat_64bit_val(es, s_blocks_count);
3115 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
3117 return ext4_read_incompat_64bit_val(es, s_r_blocks_count);
3120 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
3122 return ext4_read_incompat_64bit_val(es, s_free_blocks_count);
3125 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
3128 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
3129 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
3132 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
3135 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
3136 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
3139 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
3142 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
3143 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
3146 static inline loff_t ext4_isize(struct super_block *sb,
3147 struct ext4_inode *raw_inode)
3149 if (ext4_has_feature_largedir(sb) ||
3150 S_ISREG(le16_to_cpu(raw_inode->i_mode)))
3151 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
3152 le32_to_cpu(raw_inode->i_size_lo);
3154 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
3157 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
3159 raw_inode->i_size_lo = cpu_to_le32(i_size);
3160 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
3164 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
3167 struct ext4_group_info **grp_info;
3168 long indexv, indexh;
3169 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
3170 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
3171 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
3172 grp_info = sbi_array_rcu_deref(EXT4_SB(sb), s_group_info, indexv);
3173 return grp_info[indexh];
3177 * Reading s_groups_count requires using smp_rmb() afterwards. See
3178 * the locking protocol documented in the comments of ext4_group_add()
3181 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
3183 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
3189 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
3190 ext4_group_t block_group)
3192 return block_group >> sbi->s_log_groups_per_flex;
3195 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
3197 return 1 << sbi->s_log_groups_per_flex;
3200 #define ext4_std_error(sb, errno) \
3203 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
3207 /* Each CPU can accumulate percpu_counter_batch clusters in their local
3208 * counters. So we need to make sure we have free clusters more
3209 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
3211 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
3213 #define EXT4_FREECLUSTERS_WATERMARK 0
3216 /* Update i_disksize. Requires i_mutex to avoid races with truncate */
3217 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
3219 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
3220 !inode_is_locked(inode));
3221 down_write(&EXT4_I(inode)->i_data_sem);
3222 if (newsize > EXT4_I(inode)->i_disksize)
3223 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize);
3224 up_write(&EXT4_I(inode)->i_data_sem);
3227 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
3228 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
3232 if (newsize > inode->i_size) {
3233 i_size_write(inode, newsize);
3236 if (newsize > EXT4_I(inode)->i_disksize) {
3237 ext4_update_i_disksize(inode, newsize);
3243 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
3246 struct ext4_group_info {
3247 unsigned long bb_state;
3248 #ifdef AGGRESSIVE_CHECK
3249 unsigned long bb_check_counter;
3251 struct rb_root bb_free_root;
3252 ext4_grpblk_t bb_first_free; /* first free block */
3253 ext4_grpblk_t bb_free; /* total free blocks */
3254 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
3255 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
3256 struct list_head bb_prealloc_list;
3260 struct rw_semaphore alloc_sem;
3261 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
3262 * regions, index is order.
3263 * bb_counters[3] = 5 means
3264 * 5 free 8-block regions. */
3267 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
3268 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
3269 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
3270 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
3271 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \
3272 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT)
3273 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \
3274 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT)
3275 #define EXT4_GROUP_INFO_BBITMAP_READ_BIT 4
3277 #define EXT4_MB_GRP_NEED_INIT(grp) \
3278 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
3279 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
3280 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3281 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
3282 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3284 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
3285 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3286 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
3287 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3288 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
3289 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3290 #define EXT4_MB_GRP_TEST_AND_SET_READ(grp) \
3291 (test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state)))
3293 #define EXT4_MAX_CONTENTION 8
3294 #define EXT4_CONTENTION_THRESHOLD 2
3296 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
3299 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
3303 * Returns true if the filesystem is busy enough that attempts to
3304 * access the block group locks has run into contention.
3306 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
3308 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
3311 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
3313 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
3314 if (spin_trylock(lock))
3316 * We're able to grab the lock right away, so drop the
3317 * lock contention counter.
3319 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
3322 * The lock is busy, so bump the contention counter,
3323 * and then wait on the spin lock.
3325 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
3326 EXT4_MAX_CONTENTION);
3331 static inline void ext4_unlock_group(struct super_block *sb,
3334 spin_unlock(ext4_group_lock_ptr(sb, group));
3338 * Block validity checking
3340 #define ext4_check_indirect_blockref(inode, bh) \
3341 ext4_check_blockref(__func__, __LINE__, inode, \
3342 (__le32 *)(bh)->b_data, \
3343 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
3345 #define ext4_ind_check_inode(inode) \
3346 ext4_check_blockref(__func__, __LINE__, inode, \
3347 EXT4_I(inode)->i_data, \
3351 * Inodes and files operations
3355 extern const struct file_operations ext4_dir_operations;
3357 #ifdef CONFIG_UNICODE
3358 extern const struct dentry_operations ext4_dentry_ops;
3362 extern const struct inode_operations ext4_file_inode_operations;
3363 extern const struct file_operations ext4_file_operations;
3364 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
3367 extern int ext4_get_max_inline_size(struct inode *inode);
3368 extern int ext4_find_inline_data_nolock(struct inode *inode);
3369 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
3371 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
3373 extern int ext4_readpage_inline(struct inode *inode, struct page *page);
3374 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3375 struct inode *inode,
3376 loff_t pos, unsigned len,
3378 struct page **pagep);
3379 extern int ext4_write_inline_data_end(struct inode *inode,
3380 loff_t pos, unsigned len,
3383 extern struct buffer_head *
3384 ext4_journalled_write_inline_data(struct inode *inode,
3387 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
3388 struct inode *inode,
3389 loff_t pos, unsigned len,
3391 struct page **pagep,
3393 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
3394 unsigned len, unsigned copied,
3396 extern int ext4_try_add_inline_entry(handle_t *handle,
3397 struct ext4_filename *fname,
3398 struct inode *dir, struct inode *inode);
3399 extern int ext4_try_create_inline_dir(handle_t *handle,
3400 struct inode *parent,
3401 struct inode *inode);
3402 extern int ext4_read_inline_dir(struct file *filp,
3403 struct dir_context *ctx,
3404 int *has_inline_data);
3405 extern int ext4_inlinedir_to_tree(struct file *dir_file,
3406 struct inode *dir, ext4_lblk_t block,
3407 struct dx_hash_info *hinfo,
3408 __u32 start_hash, __u32 start_minor_hash,
3409 int *has_inline_data);
3410 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3411 struct ext4_filename *fname,
3412 struct ext4_dir_entry_2 **res_dir,
3413 int *has_inline_data);
3414 extern int ext4_delete_inline_entry(handle_t *handle,
3416 struct ext4_dir_entry_2 *de_del,
3417 struct buffer_head *bh,
3418 int *has_inline_data);
3419 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3420 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3421 struct ext4_dir_entry_2 **parent_de,
3423 extern int ext4_inline_data_fiemap(struct inode *inode,
3424 struct fiemap_extent_info *fieinfo,
3425 int *has_inline, __u64 start, __u64 len);
3428 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap);
3430 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3432 extern int ext4_convert_inline_data(struct inode *inode);
3434 static inline int ext4_has_inline_data(struct inode *inode)
3436 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3437 EXT4_I(inode)->i_inline_off;
3441 extern const struct inode_operations ext4_dir_inode_operations;
3442 extern const struct inode_operations ext4_special_inode_operations;
3443 extern struct dentry *ext4_get_parent(struct dentry *child);
3444 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3445 struct ext4_dir_entry_2 *de,
3446 int blocksize, int csum_size,
3447 unsigned int parent_ino, int dotdot_real_len);
3448 extern void ext4_initialize_dirent_tail(struct buffer_head *bh,
3449 unsigned int blocksize);
3450 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode,
3451 struct buffer_head *bh);
3452 extern int ext4_ci_compare(const struct inode *parent,
3453 const struct qstr *fname,
3454 const struct qstr *entry, bool quick);
3455 extern int __ext4_unlink(struct inode *dir, const struct qstr *d_name,
3456 struct inode *inode);
3457 extern int __ext4_link(struct inode *dir, struct inode *inode,
3458 struct dentry *dentry);
3461 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = {
3462 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
3463 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
3464 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
3465 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
3466 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
3467 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
3468 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
3471 static inline void ext4_set_de_type(struct super_block *sb,
3472 struct ext4_dir_entry_2 *de,
3474 if (ext4_has_feature_filetype(sb))
3475 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3479 extern int ext4_mpage_readpages(struct inode *inode,
3480 struct readahead_control *rac, struct page *page);
3481 extern int __init ext4_init_post_read_processing(void);
3482 extern void ext4_exit_post_read_processing(void);
3485 extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3486 extern const struct inode_operations ext4_symlink_inode_operations;
3487 extern const struct inode_operations ext4_fast_symlink_inode_operations;
3490 extern int ext4_register_sysfs(struct super_block *sb);
3491 extern void ext4_unregister_sysfs(struct super_block *sb);
3492 extern int __init ext4_init_sysfs(void);
3493 extern void ext4_exit_sysfs(void);
3495 /* block_validity */
3496 extern void ext4_release_system_zone(struct super_block *sb);
3497 extern int ext4_setup_system_zone(struct super_block *sb);
3498 extern int __init ext4_init_system_zone(void);
3499 extern void ext4_exit_system_zone(void);
3500 extern int ext4_inode_block_valid(struct inode *inode,
3501 ext4_fsblk_t start_blk,
3502 unsigned int count);
3503 extern int ext4_check_blockref(const char *, unsigned int,
3504 struct inode *, __le32 *, unsigned int);
3507 struct ext4_ext_path;
3511 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3514 #define EXT_MAX_BLOCKS 0xffffffff
3516 extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode);
3517 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3518 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3519 struct ext4_map_blocks *map, int flags);
3520 extern int ext4_ext_truncate(handle_t *, struct inode *);
3521 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3523 extern void ext4_ext_init(struct super_block *);
3524 extern void ext4_ext_release(struct super_block *);
3525 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3527 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3528 loff_t offset, ssize_t len);
3529 extern int ext4_convert_unwritten_io_end_vec(handle_t *handle,
3530 ext4_io_end_t *io_end);
3531 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3532 struct ext4_map_blocks *map, int flags);
3533 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3535 struct ext4_ext_path *path);
3536 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3537 struct ext4_ext_path **,
3538 struct ext4_extent *, int);
3539 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3540 struct ext4_ext_path **,
3542 extern void ext4_ext_drop_refs(struct ext4_ext_path *);
3543 extern int ext4_ext_check_inode(struct inode *inode);
3544 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3545 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3546 __u64 start, __u64 len);
3547 extern int ext4_get_es_cache(struct inode *inode,
3548 struct fiemap_extent_info *fieinfo,
3549 __u64 start, __u64 len);
3550 extern int ext4_ext_precache(struct inode *inode);
3551 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3552 struct inode *inode2, ext4_lblk_t lblk1,
3553 ext4_lblk_t lblk2, ext4_lblk_t count,
3554 int mark_unwritten,int *err);
3555 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu);
3556 extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode,
3557 int check_cred, int restart_cred,
3559 extern void ext4_ext_replay_shrink_inode(struct inode *inode, ext4_lblk_t end);
3560 extern int ext4_ext_replay_set_iblocks(struct inode *inode);
3561 extern int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start,
3562 int len, int unwritten, ext4_fsblk_t pblk);
3563 extern int ext4_ext_clear_bb(struct inode *inode);
3567 extern void ext4_double_down_write_data_sem(struct inode *first,
3568 struct inode *second);
3569 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3570 struct inode *donor_inode);
3571 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3572 __u64 start_orig, __u64 start_donor,
3573 __u64 len, __u64 *moved_len);
3576 extern int __init ext4_init_pageio(void);
3577 extern void ext4_exit_pageio(void);
3578 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3579 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3580 extern int ext4_put_io_end(ext4_io_end_t *io_end);
3581 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3582 extern void ext4_io_submit_init(struct ext4_io_submit *io,
3583 struct writeback_control *wbc);
3584 extern void ext4_end_io_rsv_work(struct work_struct *work);
3585 extern void ext4_io_submit(struct ext4_io_submit *io);
3586 extern int ext4_bio_write_page(struct ext4_io_submit *io,
3589 struct writeback_control *wbc,
3591 extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end);
3592 extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end);
3595 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3598 extern const struct fsverity_operations ext4_verityops;
3601 * Add new method to test whether block and inode bitmaps are properly
3602 * initialized. With uninit_bg reading the block from disk is not enough
3603 * to mark the bitmap uptodate. We need to also zero-out the bitmap
3605 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3607 static inline int bitmap_uptodate(struct buffer_head *bh)
3609 return (buffer_uptodate(bh) &&
3610 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3612 static inline void set_bitmap_uptodate(struct buffer_head *bh)
3614 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3617 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
3619 /* For ioend & aio unwritten conversion wait queues */
3620 #define EXT4_WQ_HASH_SZ 37
3621 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
3623 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3625 extern int ext4_resize_begin(struct super_block *sb);
3626 extern void ext4_resize_end(struct super_block *sb);
3628 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
3629 struct ext4_io_end *io_end)
3631 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
3632 io_end->flag |= EXT4_IO_END_UNWRITTEN;
3633 atomic_inc(&EXT4_I(inode)->i_unwritten);
3637 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3639 struct inode *inode = io_end->inode;
3641 if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
3642 io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3643 /* Wake up anyone waiting on unwritten extent conversion */
3644 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
3645 wake_up_all(ext4_ioend_wq(inode));
3649 extern const struct iomap_ops ext4_iomap_ops;
3650 extern const struct iomap_ops ext4_iomap_overwrite_ops;
3651 extern const struct iomap_ops ext4_iomap_report_ops;
3653 static inline int ext4_buffer_uptodate(struct buffer_head *bh)
3656 * If the buffer has the write error flag, we have failed
3657 * to write out data in the block. In this case, we don't
3658 * have to read the block because we may read the old data
3661 if (!buffer_uptodate(bh) && buffer_write_io_error(bh))
3662 set_buffer_uptodate(bh);
3663 return buffer_uptodate(bh);
3666 #endif /* __KERNEL__ */
3668 #define EFSBADCRC EBADMSG /* Bad CRC detected */
3669 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
3671 #endif /* _EXT4_H */