1 // SPDX-License-Identifier: GPL-2.0
5 * Copyright (C) 1992, 1993, 1994, 1995
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/refcount.h>
21 #include <linux/types.h>
22 #include <linux/blkdev.h>
23 #include <linux/magic.h>
24 #include <linux/jbd2.h>
25 #include <linux/quota.h>
26 #include <linux/rwsem.h>
27 #include <linux/rbtree.h>
28 #include <linux/seqlock.h>
29 #include <linux/mutex.h>
30 #include <linux/timer.h>
31 #include <linux/wait.h>
32 #include <linux/sched/signal.h>
33 #include <linux/blockgroup_lock.h>
34 #include <linux/percpu_counter.h>
35 #include <linux/ratelimit.h>
36 #include <crypto/hash.h>
37 #include <linux/falloc.h>
38 #include <linux/percpu-rwsem.h>
39 #include <linux/fiemap.h>
41 #include <linux/compat.h>
43 #include <uapi/linux/ext4.h>
45 #include <linux/fscrypt.h>
46 #include <linux/fsverity.h>
48 #include <linux/compiler.h>
51 * The fourth extended filesystem constants/structures
55 * with AGGRESSIVE_CHECK allocator runs consistency checks over
56 * structures. these checks slow things down a lot
58 #define AGGRESSIVE_CHECK__
61 * with DOUBLE_CHECK defined mballoc creates persistent in-core
62 * bitmaps, maintains and uses them to check for double allocations
64 #define DOUBLE_CHECK__
67 * Define EXT4FS_DEBUG to produce debug messages
75 #define ext4_debug(f, a...) \
77 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
78 __FILE__, __LINE__, __func__); \
79 printk(KERN_DEBUG f, ## a); \
82 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
86 * Turn on EXT_DEBUG to enable ext4_ext_show_path/leaf/move in extents.c
91 * Dynamic printk for controlled extents debugging.
93 #ifdef CONFIG_EXT4_DEBUG
94 #define ext_debug(ino, fmt, ...) \
95 pr_debug("[%s/%d] EXT4-fs (%s): ino %lu: (%s, %d): %s:" fmt, \
96 current->comm, task_pid_nr(current), \
97 ino->i_sb->s_id, ino->i_ino, __FILE__, __LINE__, \
98 __func__, ##__VA_ARGS__)
100 #define ext_debug(ino, fmt, ...) no_printk(fmt, ##__VA_ARGS__)
103 #define ASSERT(assert) \
105 if (unlikely(!(assert))) { \
107 "Assertion failure in %s() at %s:%d: '%s'\n", \
108 __func__, __FILE__, __LINE__, #assert); \
113 /* data type for block offset of block group */
114 typedef int ext4_grpblk_t;
116 /* data type for filesystem-wide blocks number */
117 typedef unsigned long long ext4_fsblk_t;
119 /* data type for file logical block number */
120 typedef __u32 ext4_lblk_t;
122 /* data type for block group number */
123 typedef unsigned int ext4_group_t;
125 enum SHIFT_DIRECTION {
131 * For each criteria, mballoc has slightly different way of finding
132 * the required blocks nad usually, higher the criteria the slower the
133 * allocation. We start at lower criterias and keep falling back to
134 * higher ones if we are not able to find any blocks. Lower (earlier)
135 * criteria are faster.
139 * Used when number of blocks needed is a power of 2. This
140 * doesn't trigger any disk IO except prefetch and is the
146 * Tries to lookup in-memory data structures to find the most
147 * suitable group that satisfies goal request. No disk IO
148 * except block prefetch.
153 * Same as CR_GOAL_LEN_FAST but is allowed to reduce the goal
154 * length to the best available length for faster allocation.
159 * Reads each block group sequentially, performing disk IO if
160 * necessary, to find find_suitable block group. Tries to
161 * allocate goal length but might trim the request if nothing
162 * is found after enough tries.
167 * Finds the first free set of blocks and allocates
168 * those. This is only used in rare cases when
169 * CR_GOAL_LEN_SLOW also fails to allocate anything.
174 * Number of criterias defined.
179 /* criteria below which we use fast block scanning and avoid unnecessary IO */
180 #define CR_FAST CR_GOAL_LEN_SLOW
183 * Flags used in mballoc's allocation_context flags field.
185 * Also used to show what's going on for debugging purposes when the
186 * flag field is exported via the traceport interface
189 /* prefer goal again. length */
190 #define EXT4_MB_HINT_MERGE 0x0001
191 /* blocks already reserved */
192 #define EXT4_MB_HINT_RESERVED 0x0002
193 /* metadata is being allocated */
194 #define EXT4_MB_HINT_METADATA 0x0004
195 /* first blocks in the file */
196 #define EXT4_MB_HINT_FIRST 0x0008
197 /* search for the best chunk */
198 #define EXT4_MB_HINT_BEST 0x0010
199 /* data is being allocated */
200 #define EXT4_MB_HINT_DATA 0x0020
201 /* don't preallocate (for tails) */
202 #define EXT4_MB_HINT_NOPREALLOC 0x0040
203 /* allocate for locality group */
204 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080
205 /* allocate goal blocks or none */
206 #define EXT4_MB_HINT_GOAL_ONLY 0x0100
207 /* goal is meaningful */
208 #define EXT4_MB_HINT_TRY_GOAL 0x0200
209 /* blocks already pre-reserved by delayed allocation */
210 #define EXT4_MB_DELALLOC_RESERVED 0x0400
211 /* We are doing stream allocation */
212 #define EXT4_MB_STREAM_ALLOC 0x0800
213 /* Use reserved root blocks if needed */
214 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000
215 /* Use blocks from reserved pool */
216 #define EXT4_MB_USE_RESERVED 0x2000
217 /* Do strict check for free blocks while retrying block allocation */
218 #define EXT4_MB_STRICT_CHECK 0x4000
219 /* Large fragment size list lookup succeeded at least once for cr = 0 */
220 #define EXT4_MB_CR_POWER2_ALIGNED_OPTIMIZED 0x8000
221 /* Avg fragment size rb tree lookup succeeded at least once for cr = 1 */
222 #define EXT4_MB_CR_GOAL_LEN_FAST_OPTIMIZED 0x00010000
223 /* Avg fragment size rb tree lookup succeeded at least once for cr = 1.5 */
224 #define EXT4_MB_CR_BEST_AVAIL_LEN_OPTIMIZED 0x00020000
226 struct ext4_allocation_request {
227 /* target inode for block we're allocating */
229 /* how many blocks we want to allocate */
231 /* logical block in target inode */
233 /* the closest logical allocated block to the left */
235 /* the closest logical allocated block to the right */
237 /* phys. target (a hint) */
239 /* phys. block for the closest logical allocated block to the left */
241 /* phys. block for the closest logical allocated block to the right */
243 /* flags. see above EXT4_MB_HINT_* */
248 * Logical to physical block mapping, used by ext4_map_blocks()
250 * This structure is used to pass requests into ext4_map_blocks() as
251 * well as to store the information returned by ext4_map_blocks(). It
252 * takes less room on the stack than a struct buffer_head.
254 #define EXT4_MAP_NEW BIT(BH_New)
255 #define EXT4_MAP_MAPPED BIT(BH_Mapped)
256 #define EXT4_MAP_UNWRITTEN BIT(BH_Unwritten)
257 #define EXT4_MAP_BOUNDARY BIT(BH_Boundary)
258 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
259 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
261 struct ext4_map_blocks {
265 unsigned int m_flags;
269 * Block validity checking, system zone rbtree.
271 struct ext4_system_blocks {
277 * Flags for ext4_io_end->flags
279 #define EXT4_IO_END_UNWRITTEN 0x0001
281 struct ext4_io_end_vec {
282 struct list_head list; /* list of io_end_vec */
283 loff_t offset; /* offset in the file */
284 ssize_t size; /* size of the extent */
288 * For converting unwritten extents on a work queue. 'handle' is used for
289 * buffered writeback.
291 typedef struct ext4_io_end {
292 struct list_head list; /* per-file finished IO list */
293 handle_t *handle; /* handle reserved for extent
295 struct inode *inode; /* file being written to */
296 struct bio *bio; /* Linked list of completed
297 * bios covering the extent */
298 unsigned int flag; /* unwritten or not */
299 refcount_t count; /* reference counter */
300 struct list_head list_vec; /* list of ext4_io_end_vec */
303 struct ext4_io_submit {
304 struct writeback_control *io_wbc;
306 ext4_io_end_t *io_end;
307 sector_t io_next_block;
311 * Special inodes numbers
313 #define EXT4_BAD_INO 1 /* Bad blocks inode */
314 #define EXT4_ROOT_INO 2 /* Root inode */
315 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */
316 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
317 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
318 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
319 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
320 #define EXT4_JOURNAL_INO 8 /* Journal inode */
322 /* First non-reserved inode for old ext4 filesystems */
323 #define EXT4_GOOD_OLD_FIRST_INO 11
326 * Maximal count of links to a file
328 #define EXT4_LINK_MAX 65000
331 * Macro-instructions used to manage several block sizes
333 #define EXT4_MIN_BLOCK_SIZE 1024
334 #define EXT4_MAX_BLOCK_SIZE 65536
335 #define EXT4_MIN_BLOCK_LOG_SIZE 10
336 #define EXT4_MAX_BLOCK_LOG_SIZE 16
337 #define EXT4_MAX_CLUSTER_LOG_SIZE 30
339 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
341 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
343 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
344 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
345 EXT4_SB(s)->s_cluster_bits)
347 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
348 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
350 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
353 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
354 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
355 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
357 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
358 EXT4_GOOD_OLD_INODE_SIZE : \
360 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
361 EXT4_GOOD_OLD_FIRST_INO : \
364 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
365 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \
366 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \
369 /* Translate a block number to a cluster number */
370 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
371 /* Translate a cluster number to a block number */
372 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
373 /* Translate # of blks to # of clusters */
374 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
375 (sbi)->s_cluster_bits)
376 /* Mask out the low bits to get the starting block of the cluster */
377 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
378 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
379 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
380 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
381 /* Fill in the low bits to get the last block of the cluster */
382 #define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) | \
383 ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1))
384 /* Get the cluster offset */
385 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
386 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
387 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
388 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
391 * Structure of a blocks group descriptor
393 struct ext4_group_desc
395 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
396 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
397 __le32 bg_inode_table_lo; /* Inodes table block */
398 __le16 bg_free_blocks_count_lo;/* Free blocks count */
399 __le16 bg_free_inodes_count_lo;/* Free inodes count */
400 __le16 bg_used_dirs_count_lo; /* Directories count */
401 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
402 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
403 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
404 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
405 __le16 bg_itable_unused_lo; /* Unused inodes count */
406 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
407 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
408 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
409 __le32 bg_inode_table_hi; /* Inodes table block MSB */
410 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
411 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
412 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
413 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
414 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
415 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
416 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
420 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
421 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
423 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
424 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
428 * Structure of a flex block group info
432 atomic64_t free_clusters;
433 atomic_t free_inodes;
437 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
438 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
439 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
442 * Macro-instructions used to manage group descriptors
444 #define EXT4_MIN_DESC_SIZE 32
445 #define EXT4_MIN_DESC_SIZE_64BIT 64
446 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
447 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
449 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
450 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
451 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
452 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
453 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
455 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
456 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
457 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
461 * Constants relative to the data blocks
463 #define EXT4_NDIR_BLOCKS 12
464 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
465 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
466 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
467 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
472 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
473 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
474 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
475 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
476 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
477 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
478 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
479 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
480 /* Reserved for compression usage... */
481 #define EXT4_DIRTY_FL 0x00000100
482 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
483 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
484 /* nb: was previously EXT2_ECOMPR_FL */
485 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */
486 /* End compression flags --- maybe not all used */
487 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
488 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
489 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
490 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
491 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
492 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
493 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
494 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
495 #define EXT4_VERITY_FL 0x00100000 /* Verity protected inode */
496 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
497 /* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */
499 #define EXT4_DAX_FL 0x02000000 /* Inode is DAX */
501 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
502 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
503 #define EXT4_CASEFOLD_FL 0x40000000 /* Casefolded directory */
504 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
506 /* User modifiable flags */
507 #define EXT4_FL_USER_MODIFIABLE (EXT4_SECRM_FL | \
511 EXT4_IMMUTABLE_FL | \
515 EXT4_JOURNAL_DATA_FL | \
520 0x00400000 /* EXT4_EOFBLOCKS_FL */ | \
522 EXT4_PROJINHERIT_FL | \
525 /* User visible flags */
526 #define EXT4_FL_USER_VISIBLE (EXT4_FL_USER_MODIFIABLE | \
535 /* Flags that should be inherited by new inodes from their parent. */
536 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
537 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
538 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
539 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
540 EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL |\
543 /* Flags that are appropriate for regular files (all but dir-specific ones). */
544 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\
545 EXT4_PROJINHERIT_FL))
547 /* Flags that are appropriate for non-directories/regular files. */
548 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
550 /* The only flags that should be swapped */
551 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL)
553 /* Flags which are mutually exclusive to DAX */
554 #define EXT4_DAX_MUT_EXCL (EXT4_VERITY_FL | EXT4_ENCRYPT_FL |\
555 EXT4_JOURNAL_DATA_FL | EXT4_INLINE_DATA_FL)
557 /* Mask out flags that are inappropriate for the given type of inode. */
558 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
562 else if (S_ISREG(mode))
563 return flags & EXT4_REG_FLMASK;
565 return flags & EXT4_OTHER_FLMASK;
569 * Inode flags used for atomic set/get
572 EXT4_INODE_SECRM = 0, /* Secure deletion */
573 EXT4_INODE_UNRM = 1, /* Undelete */
574 EXT4_INODE_COMPR = 2, /* Compress file */
575 EXT4_INODE_SYNC = 3, /* Synchronous updates */
576 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
577 EXT4_INODE_APPEND = 5, /* writes to file may only append */
578 EXT4_INODE_NODUMP = 6, /* do not dump file */
579 EXT4_INODE_NOATIME = 7, /* do not update atime */
580 /* Reserved for compression usage... */
581 EXT4_INODE_DIRTY = 8,
582 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
583 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
584 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */
585 /* End compression flags --- maybe not all used */
586 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
587 EXT4_INODE_IMAGIC = 13, /* AFS directory */
588 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
589 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
590 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
591 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
592 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
593 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
594 EXT4_INODE_VERITY = 20, /* Verity protected inode */
595 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
596 /* 22 was formerly EXT4_INODE_EOFBLOCKS */
597 EXT4_INODE_DAX = 25, /* Inode is DAX */
598 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
599 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */
600 EXT4_INODE_CASEFOLD = 30, /* Casefolded directory */
601 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
605 * Since it's pretty easy to mix up bit numbers and hex values, we use a
606 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
607 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
608 * any extra space in the compiled kernel image, otherwise, the build will fail.
609 * It's important that these values are the same, since we are using
610 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
611 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
612 * values found in ext2, ext3 and ext4 filesystems, and of course the values
613 * defined in e2fsprogs.
615 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
617 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1U << EXT4_INODE_##FLAG))
618 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
620 static inline void ext4_check_flag_values(void)
622 CHECK_FLAG_VALUE(SECRM);
623 CHECK_FLAG_VALUE(UNRM);
624 CHECK_FLAG_VALUE(COMPR);
625 CHECK_FLAG_VALUE(SYNC);
626 CHECK_FLAG_VALUE(IMMUTABLE);
627 CHECK_FLAG_VALUE(APPEND);
628 CHECK_FLAG_VALUE(NODUMP);
629 CHECK_FLAG_VALUE(NOATIME);
630 CHECK_FLAG_VALUE(DIRTY);
631 CHECK_FLAG_VALUE(COMPRBLK);
632 CHECK_FLAG_VALUE(NOCOMPR);
633 CHECK_FLAG_VALUE(ENCRYPT);
634 CHECK_FLAG_VALUE(INDEX);
635 CHECK_FLAG_VALUE(IMAGIC);
636 CHECK_FLAG_VALUE(JOURNAL_DATA);
637 CHECK_FLAG_VALUE(NOTAIL);
638 CHECK_FLAG_VALUE(DIRSYNC);
639 CHECK_FLAG_VALUE(TOPDIR);
640 CHECK_FLAG_VALUE(HUGE_FILE);
641 CHECK_FLAG_VALUE(EXTENTS);
642 CHECK_FLAG_VALUE(VERITY);
643 CHECK_FLAG_VALUE(EA_INODE);
644 CHECK_FLAG_VALUE(INLINE_DATA);
645 CHECK_FLAG_VALUE(PROJINHERIT);
646 CHECK_FLAG_VALUE(CASEFOLD);
647 CHECK_FLAG_VALUE(RESERVED);
650 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
651 struct compat_ext4_new_group_input {
653 compat_u64 block_bitmap;
654 compat_u64 inode_bitmap;
655 compat_u64 inode_table;
662 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
663 struct ext4_new_group_data {
669 __u16 reserved_blocks;
671 __u32 free_clusters_count;
674 /* Indexes used to index group tables in ext4_new_group_data */
676 BLOCK_BITMAP = 0, /* block bitmap */
677 INODE_BITMAP, /* inode bitmap */
678 INODE_TABLE, /* inode tables */
683 * Flags used by ext4_map_blocks()
685 /* Allocate any needed blocks and/or convert an unwritten
686 extent to be an initialized ext4 */
687 #define EXT4_GET_BLOCKS_CREATE 0x0001
688 /* Request the creation of an unwritten extent */
689 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002
690 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\
691 EXT4_GET_BLOCKS_CREATE)
692 /* Caller is from the delayed allocation writeout path
693 * finally doing the actual allocation of delayed blocks */
694 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
695 /* caller is from the direct IO path, request to creation of an
696 unwritten extents if not allocated, split the unwritten
697 extent if blocks has been preallocated already*/
698 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
699 #define EXT4_GET_BLOCKS_CONVERT 0x0010
700 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
701 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
702 /* Convert extent to initialized after IO complete */
703 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
704 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
705 /* Eventual metadata allocation (due to growing extent tree)
706 * should not fail, so try to use reserved blocks for that.*/
707 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
708 /* Don't normalize allocation size (used for fallocate) */
709 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
710 /* Convert written extents to unwritten */
711 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100
712 /* Write zeros to newly created written extents */
713 #define EXT4_GET_BLOCKS_ZERO 0x0200
714 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\
715 EXT4_GET_BLOCKS_ZERO)
716 /* Caller will submit data before dropping transaction handle. This
717 * allows jbd2 to avoid submitting data before commit. */
718 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400
719 /* Caller is in the atomic contex, find extent if it has been cached */
720 #define EXT4_GET_BLOCKS_CACHED_NOWAIT 0x0800
723 * The bit position of these flags must not overlap with any of the
724 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(),
725 * read_extent_tree_block(), ext4_split_extent_at(),
726 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
727 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
728 * caching the extents when reading from the extent tree while a
729 * truncate or punch hole operation is in progress.
731 #define EXT4_EX_NOCACHE 0x40000000
732 #define EXT4_EX_FORCE_CACHE 0x20000000
733 #define EXT4_EX_NOFAIL 0x10000000
736 * Flags used by ext4_free_blocks
738 #define EXT4_FREE_BLOCKS_METADATA 0x0001
739 #define EXT4_FREE_BLOCKS_FORGET 0x0002
740 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
741 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
742 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
743 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
744 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040
746 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
748 * ioctl commands in 32 bit emulation
750 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
751 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
752 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
753 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
754 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
755 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
756 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
757 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
760 /* Max physical block we can address w/o extents */
761 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
763 /* Max logical block we can support */
764 #define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFE
767 * Structure of an inode on the disk
770 __le16 i_mode; /* File mode */
771 __le16 i_uid; /* Low 16 bits of Owner Uid */
772 __le32 i_size_lo; /* Size in bytes */
773 __le32 i_atime; /* Access time */
774 __le32 i_ctime; /* Inode Change time */
775 __le32 i_mtime; /* Modification time */
776 __le32 i_dtime; /* Deletion Time */
777 __le16 i_gid; /* Low 16 bits of Group Id */
778 __le16 i_links_count; /* Links count */
779 __le32 i_blocks_lo; /* Blocks count */
780 __le32 i_flags; /* File flags */
786 __u32 h_i_translator;
791 } osd1; /* OS dependent 1 */
792 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
793 __le32 i_generation; /* File version (for NFS) */
794 __le32 i_file_acl_lo; /* File ACL */
796 __le32 i_obso_faddr; /* Obsoleted fragment address */
799 __le16 l_i_blocks_high; /* were l_i_reserved1 */
800 __le16 l_i_file_acl_high;
801 __le16 l_i_uid_high; /* these 2 fields */
802 __le16 l_i_gid_high; /* were reserved2[0] */
803 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
807 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
814 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
815 __le16 m_i_file_acl_high;
816 __u32 m_i_reserved2[2];
818 } osd2; /* OS dependent 2 */
819 __le16 i_extra_isize;
820 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
821 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
822 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
823 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
824 __le32 i_crtime; /* File Creation time */
825 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
826 __le32 i_version_hi; /* high 32 bits for 64-bit version */
827 __le32 i_projid; /* Project ID */
830 #define EXT4_EPOCH_BITS 2
831 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
832 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
835 * Extended fields will fit into an inode if the filesystem was formatted
836 * with large inodes (-I 256 or larger) and there are not currently any EAs
837 * consuming all of the available space. For new inodes we always reserve
838 * enough space for the kernel's known extended fields, but for inodes
839 * created with an old kernel this might not have been the case. None of
840 * the extended inode fields is critical for correct filesystem operation.
841 * This macro checks if a certain field fits in the inode. Note that
842 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
844 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
845 ((offsetof(typeof(*ext4_inode), field) + \
846 sizeof((ext4_inode)->field)) \
847 <= (EXT4_GOOD_OLD_INODE_SIZE + \
848 (einode)->i_extra_isize)) \
851 * We use an encoding that preserves the times for extra epoch "00":
853 * extra msb of adjust for signed
854 * epoch 32-bit 32-bit tv_sec to
855 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
856 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
857 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
858 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
859 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
860 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
861 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
862 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
863 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
865 * Note that previous versions of the kernel on 64-bit systems would
866 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
867 * 1970. e2fsck will correct this, assuming that it is run on the
868 * affected filesystem before 2242.
871 static inline __le32 ext4_encode_extra_time(struct timespec64 *time)
873 u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK;
874 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
877 static inline void ext4_decode_extra_time(struct timespec64 *time,
880 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK)))
881 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
882 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
885 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
887 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {\
888 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
889 (raw_inode)->xtime ## _extra = \
890 ext4_encode_extra_time(&(inode)->xtime); \
893 (raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (inode)->xtime.tv_sec, S32_MIN, S32_MAX)); \
896 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
898 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
899 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
900 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
901 (raw_inode)->xtime ## _extra = \
902 ext4_encode_extra_time(&(einode)->xtime); \
905 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
907 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
908 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \
909 ext4_decode_extra_time(&(inode)->xtime, \
910 raw_inode->xtime ## _extra); \
913 (inode)->xtime.tv_nsec = 0; \
917 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
919 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
920 (einode)->xtime.tv_sec = \
921 (signed)le32_to_cpu((raw_inode)->xtime); \
923 (einode)->xtime.tv_sec = 0; \
924 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
925 ext4_decode_extra_time(&(einode)->xtime, \
926 raw_inode->xtime ## _extra); \
928 (einode)->xtime.tv_nsec = 0; \
931 #define i_disk_version osd1.linux1.l_i_version
933 #if defined(__KERNEL__) || defined(__linux__)
934 #define i_reserved1 osd1.linux1.l_i_reserved1
935 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
936 #define i_blocks_high osd2.linux2.l_i_blocks_high
937 #define i_uid_low i_uid
938 #define i_gid_low i_gid
939 #define i_uid_high osd2.linux2.l_i_uid_high
940 #define i_gid_high osd2.linux2.l_i_gid_high
941 #define i_checksum_lo osd2.linux2.l_i_checksum_lo
943 #elif defined(__GNU__)
945 #define i_translator osd1.hurd1.h_i_translator
946 #define i_uid_high osd2.hurd2.h_i_uid_high
947 #define i_gid_high osd2.hurd2.h_i_gid_high
948 #define i_author osd2.hurd2.h_i_author
950 #elif defined(__masix__)
952 #define i_reserved1 osd1.masix1.m_i_reserved1
953 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
954 #define i_reserved2 osd2.masix2.m_i_reserved2
956 #endif /* defined(__KERNEL__) || defined(__linux__) */
958 #include "extents_status.h"
959 #include "fast_commit.h"
962 * Lock subclasses for i_data_sem in the ext4_inode_info structure.
964 * These are needed to avoid lockdep false positives when we need to
965 * allocate blocks to the quota inode during ext4_map_blocks(), while
966 * holding i_data_sem for a normal (non-quota) inode. Since we don't
967 * do quota tracking for the quota inode, this avoids deadlock (as
968 * well as infinite recursion, since it isn't turtles all the way
971 * I_DATA_SEM_NORMAL - Used for most inodes
972 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode
973 * where the second inode has larger inode number
975 * I_DATA_SEM_QUOTA - Used for quota inodes only
976 * I_DATA_SEM_EA - Used for ea_inodes only
979 I_DATA_SEM_NORMAL = 0,
987 * fourth extended file system inode data in memory
989 struct ext4_inode_info {
990 __le32 i_data[15]; /* unconverted */
992 ext4_fsblk_t i_file_acl;
995 * i_block_group is the number of the block group which contains
996 * this file's inode. Constant across the lifetime of the inode,
997 * it is used for making block allocation decisions - we try to
998 * place a file's data blocks near its inode block, and new inodes
999 * near to their parent directory's inode.
1001 ext4_group_t i_block_group;
1002 ext4_lblk_t i_dir_start_lookup;
1003 #if (BITS_PER_LONG < 64)
1004 unsigned long i_state_flags; /* Dynamic state flags */
1006 unsigned long i_flags;
1009 * Extended attributes can be read independently of the main file
1010 * data. Taking i_rwsem even when reading would cause contention
1011 * between readers of EAs and writers of regular file data, so
1012 * instead we synchronize on xattr_sem when reading or changing
1015 struct rw_semaphore xattr_sem;
1018 * Inodes with EXT4_STATE_ORPHAN_FILE use i_orphan_idx. Otherwise
1022 struct list_head i_orphan; /* unlinked but open inodes */
1023 unsigned int i_orphan_idx; /* Index in orphan file */
1026 /* Fast commit related info */
1028 /* For tracking dentry create updates */
1029 struct list_head i_fc_dilist;
1030 struct list_head i_fc_list; /*
1031 * inodes that need fast commit
1032 * protected by sbi->s_fc_lock.
1035 /* Start of lblk range that needs to be committed in this fast commit */
1036 ext4_lblk_t i_fc_lblk_start;
1038 /* End of lblk range that needs to be committed in this fast commit */
1039 ext4_lblk_t i_fc_lblk_len;
1041 /* Number of ongoing updates on this inode */
1042 atomic_t i_fc_updates;
1044 /* Fast commit wait queue for this inode */
1045 wait_queue_head_t i_fc_wait;
1047 /* Protect concurrent accesses on i_fc_lblk_start, i_fc_lblk_len */
1048 struct mutex i_fc_lock;
1051 * i_disksize keeps track of what the inode size is ON DISK, not
1052 * in memory. During truncate, i_size is set to the new size by
1053 * the VFS prior to calling ext4_truncate(), but the filesystem won't
1054 * set i_disksize to 0 until the truncate is actually under way.
1056 * The intent is that i_disksize always represents the blocks which
1057 * are used by this file. This allows recovery to restart truncate
1058 * on orphans if we crash during truncate. We actually write i_disksize
1059 * into the on-disk inode when writing inodes out, instead of i_size.
1061 * The only time when i_disksize and i_size may be different is when
1062 * a truncate is in progress. The only things which change i_disksize
1063 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
1068 * i_data_sem is for serialising ext4_truncate() against
1069 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
1070 * data tree are chopped off during truncate. We can't do that in
1071 * ext4 because whenever we perform intermediate commits during
1072 * truncate, the inode and all the metadata blocks *must* be in a
1073 * consistent state which allows truncation of the orphans to restart
1074 * during recovery. Hence we must fix the get_block-vs-truncate race
1075 * by other means, so we have i_data_sem.
1077 struct rw_semaphore i_data_sem;
1078 struct inode vfs_inode;
1079 struct jbd2_inode *jinode;
1081 spinlock_t i_raw_lock; /* protects updates to the raw inode */
1084 * File creation time. Its function is same as that of
1085 * struct timespec64 i_{a,c,m}time in the generic inode.
1087 struct timespec64 i_crtime;
1090 atomic_t i_prealloc_active;
1091 struct rb_root i_prealloc_node;
1092 rwlock_t i_prealloc_lock;
1094 /* extents status tree */
1095 struct ext4_es_tree i_es_tree;
1097 struct list_head i_es_list;
1098 unsigned int i_es_all_nr; /* protected by i_es_lock */
1099 unsigned int i_es_shk_nr; /* protected by i_es_lock */
1100 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for
1101 extents to shrink. Protected by
1105 ext4_group_t i_last_alloc_group;
1107 /* allocation reservation info for delalloc */
1108 /* In case of bigalloc, this refer to clusters rather than blocks */
1109 unsigned int i_reserved_data_blocks;
1111 /* pending cluster reservations for bigalloc file systems */
1112 struct ext4_pending_tree i_pending_tree;
1114 /* on-disk additional length */
1115 __u16 i_extra_isize;
1117 /* Indicate the inline data space. */
1122 /* quota space reservation, managed internally by quota code */
1123 qsize_t i_reserved_quota;
1126 /* Lock protecting lists below */
1127 spinlock_t i_completed_io_lock;
1129 * Completed IOs that need unwritten extents handling and have
1130 * transaction reserved
1132 struct list_head i_rsv_conversion_list;
1133 struct work_struct i_rsv_conversion_work;
1134 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1136 spinlock_t i_block_reservation_lock;
1139 * Transactions that contain inode's metadata needed to complete
1140 * fsync and fdatasync, respectively.
1143 tid_t i_datasync_tid;
1146 struct dquot *i_dquot[MAXQUOTAS];
1149 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1156 * File system states
1158 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
1159 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
1160 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
1161 #define EXT4_FC_REPLAY 0x0020 /* Fast commit replay ongoing */
1164 * Misc. filesystem flags
1166 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
1167 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
1168 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
1171 * Mount flags set via mount options or defaults
1173 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */
1174 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
1175 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
1176 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
1177 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
1178 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
1179 #define EXT4_MOUNT_ERRORS_MASK 0x00070
1180 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
1181 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
1182 #ifdef CONFIG_FS_DAX
1183 #define EXT4_MOUNT_DAX_ALWAYS 0x00200 /* Direct Access */
1185 #define EXT4_MOUNT_DAX_ALWAYS 0
1187 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
1188 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
1189 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
1190 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
1191 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
1192 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
1193 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
1194 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
1195 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
1196 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
1197 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */
1198 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota,
1199 * enable enforcement for hidden
1201 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable
1202 * enforcement for hidden quota
1204 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota
1206 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
1207 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
1208 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1209 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */
1210 #define EXT4_MOUNT_NO_PREFETCH_BLOCK_BITMAPS 0x4000000
1211 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
1212 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
1213 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
1214 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
1215 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
1218 * Mount flags set either automatically (could not be set by mount option)
1219 * based on per file system feature or property or in special cases such as
1220 * distinguishing between explicit mount option definition and default.
1222 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1223 specified delalloc */
1224 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1225 size of blocksize * 8
1227 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1229 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly
1230 specified journal checksum */
1232 #define EXT4_MOUNT2_JOURNAL_FAST_COMMIT 0x00000010 /* Journal fast commit */
1233 #define EXT4_MOUNT2_DAX_NEVER 0x00000020 /* Do not allow Direct Access */
1234 #define EXT4_MOUNT2_DAX_INODE 0x00000040 /* For printing options only */
1235 #define EXT4_MOUNT2_MB_OPTIMIZE_SCAN 0x00000080 /* Optimize group
1236 * scanning in mballoc
1239 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1241 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1243 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1246 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1248 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1250 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1253 #define ext4_test_and_set_bit __test_and_set_bit_le
1254 #define ext4_set_bit __set_bit_le
1255 #define ext4_set_bit_atomic ext2_set_bit_atomic
1256 #define ext4_test_and_clear_bit __test_and_clear_bit_le
1257 #define ext4_clear_bit __clear_bit_le
1258 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
1259 #define ext4_test_bit test_bit_le
1260 #define ext4_find_next_zero_bit find_next_zero_bit_le
1261 #define ext4_find_next_bit find_next_bit_le
1263 extern void mb_set_bits(void *bm, int cur, int len);
1266 * Maximal mount counts between two filesystem checks
1268 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1269 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1272 * Behaviour when detecting errors
1274 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1275 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1276 #define EXT4_ERRORS_PANIC 3 /* Panic */
1277 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1279 /* Metadata checksum algorithm codes */
1280 #define EXT4_CRC32C_CHKSUM 1
1282 #define EXT4_LABEL_MAX 16
1285 * Structure of the super block
1287 struct ext4_super_block {
1288 /*00*/ __le32 s_inodes_count; /* Inodes count */
1289 __le32 s_blocks_count_lo; /* Blocks count */
1290 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1291 __le32 s_free_blocks_count_lo; /* Free blocks count */
1292 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1293 __le32 s_first_data_block; /* First Data Block */
1294 __le32 s_log_block_size; /* Block size */
1295 __le32 s_log_cluster_size; /* Allocation cluster size */
1296 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1297 __le32 s_clusters_per_group; /* # Clusters per group */
1298 __le32 s_inodes_per_group; /* # Inodes per group */
1299 __le32 s_mtime; /* Mount time */
1300 /*30*/ __le32 s_wtime; /* Write time */
1301 __le16 s_mnt_count; /* Mount count */
1302 __le16 s_max_mnt_count; /* Maximal mount count */
1303 __le16 s_magic; /* Magic signature */
1304 __le16 s_state; /* File system state */
1305 __le16 s_errors; /* Behaviour when detecting errors */
1306 __le16 s_minor_rev_level; /* minor revision level */
1307 /*40*/ __le32 s_lastcheck; /* time of last check */
1308 __le32 s_checkinterval; /* max. time between checks */
1309 __le32 s_creator_os; /* OS */
1310 __le32 s_rev_level; /* Revision level */
1311 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1312 __le16 s_def_resgid; /* Default gid for reserved blocks */
1314 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1316 * Note: the difference between the compatible feature set and
1317 * the incompatible feature set is that if there is a bit set
1318 * in the incompatible feature set that the kernel doesn't
1319 * know about, it should refuse to mount the filesystem.
1321 * e2fsck's requirements are more strict; if it doesn't know
1322 * about a feature in either the compatible or incompatible
1323 * feature set, it must abort and not try to meddle with
1324 * things it doesn't understand...
1326 __le32 s_first_ino; /* First non-reserved inode */
1327 __le16 s_inode_size; /* size of inode structure */
1328 __le16 s_block_group_nr; /* block group # of this superblock */
1329 __le32 s_feature_compat; /* compatible feature set */
1330 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1331 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1332 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1333 /*78*/ char s_volume_name[EXT4_LABEL_MAX]; /* volume name */
1334 /*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */
1335 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1337 * Performance hints. Directory preallocation should only
1338 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1340 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1341 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1342 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1344 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1346 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1347 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1348 __le32 s_journal_dev; /* device number of journal file */
1349 __le32 s_last_orphan; /* start of list of inodes to delete */
1350 __le32 s_hash_seed[4]; /* HTREE hash seed */
1351 __u8 s_def_hash_version; /* Default hash version to use */
1352 __u8 s_jnl_backup_type;
1353 __le16 s_desc_size; /* size of group descriptor */
1354 /*100*/ __le32 s_default_mount_opts;
1355 __le32 s_first_meta_bg; /* First metablock block group */
1356 __le32 s_mkfs_time; /* When the filesystem was created */
1357 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1358 /* 64bit support valid if EXT4_FEATURE_INCOMPAT_64BIT */
1359 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1360 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1361 __le32 s_free_blocks_count_hi; /* Free blocks count */
1362 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1363 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1364 __le32 s_flags; /* Miscellaneous flags */
1365 __le16 s_raid_stride; /* RAID stride */
1366 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1367 __le64 s_mmp_block; /* Block for multi-mount protection */
1368 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1369 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1370 __u8 s_checksum_type; /* metadata checksum algorithm used */
1371 __u8 s_encryption_level; /* versioning level for encryption */
1372 __u8 s_reserved_pad; /* Padding to next 32bits */
1373 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1374 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1375 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1376 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1377 snapshot's future use */
1378 __le32 s_snapshot_list; /* inode number of the head of the
1379 on-disk snapshot list */
1380 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1381 __le32 s_error_count; /* number of fs errors */
1382 __le32 s_first_error_time; /* first time an error happened */
1383 __le32 s_first_error_ino; /* inode involved in first error */
1384 __le64 s_first_error_block; /* block involved of first error */
1385 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */
1386 __le32 s_first_error_line; /* line number where error happened */
1387 __le32 s_last_error_time; /* most recent time of an error */
1388 __le32 s_last_error_ino; /* inode involved in last error */
1389 __le32 s_last_error_line; /* line number where error happened */
1390 __le64 s_last_error_block; /* block involved of last error */
1391 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */
1392 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1393 __u8 s_mount_opts[64];
1394 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1395 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1396 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1397 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
1398 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */
1399 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */
1400 __le32 s_lpf_ino; /* Location of the lost+found inode */
1401 __le32 s_prj_quota_inum; /* inode for tracking project quota */
1402 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */
1405 __u8 s_mkfs_time_hi;
1406 __u8 s_lastcheck_hi;
1407 __u8 s_first_error_time_hi;
1408 __u8 s_last_error_time_hi;
1409 __u8 s_first_error_errcode;
1410 __u8 s_last_error_errcode;
1411 __le16 s_encoding; /* Filename charset encoding */
1412 __le16 s_encoding_flags; /* Filename charset encoding flags */
1413 __le32 s_orphan_file_inum; /* Inode for tracking orphan inodes */
1414 __le32 s_reserved[94]; /* Padding to the end of the block */
1415 __le32 s_checksum; /* crc32c(superblock) */
1418 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1422 /* Number of quota types we support */
1423 #define EXT4_MAXQUOTAS 3
1425 #define EXT4_ENC_UTF8_12_1 1
1427 /* Types of ext4 journal triggers */
1428 enum ext4_journal_trigger_type {
1429 EXT4_JTR_ORPHAN_FILE,
1430 EXT4_JTR_NONE /* This must be the last entry for indexing to work! */
1433 #define EXT4_JOURNAL_TRIGGER_COUNT EXT4_JTR_NONE
1435 struct ext4_journal_trigger {
1436 struct jbd2_buffer_trigger_type tr_triggers;
1437 struct super_block *sb;
1440 static inline struct ext4_journal_trigger *EXT4_TRIGGER(
1441 struct jbd2_buffer_trigger_type *trigger)
1443 return container_of(trigger, struct ext4_journal_trigger, tr_triggers);
1446 #define EXT4_ORPHAN_BLOCK_MAGIC 0x0b10ca04
1448 /* Structure at the tail of orphan block */
1449 struct ext4_orphan_block_tail {
1454 static inline int ext4_inodes_per_orphan_block(struct super_block *sb)
1456 return (sb->s_blocksize - sizeof(struct ext4_orphan_block_tail)) /
1460 struct ext4_orphan_block {
1461 atomic_t ob_free_entries; /* Number of free orphan entries in block */
1462 struct buffer_head *ob_bh; /* Buffer for orphan block */
1466 * Info about orphan file.
1468 struct ext4_orphan_info {
1469 int of_blocks; /* Number of orphan blocks in a file */
1470 __u32 of_csum_seed; /* Checksum seed for orphan file */
1471 struct ext4_orphan_block *of_binfo; /* Array with info about orphan
1476 * fourth extended-fs super-block data in memory
1478 struct ext4_sb_info {
1479 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1480 unsigned long s_inodes_per_block;/* Number of inodes per block */
1481 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1482 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1483 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1484 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1485 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1486 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1487 ext4_group_t s_groups_count; /* Number of groups in the fs */
1488 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1489 unsigned long s_overhead; /* # of fs overhead clusters */
1490 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1491 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1492 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1493 struct buffer_head * s_sbh; /* Buffer containing the super block */
1494 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1495 struct buffer_head * __rcu *s_group_desc;
1496 unsigned int s_mount_opt;
1497 unsigned int s_mount_opt2;
1498 unsigned long s_mount_flags;
1499 unsigned int s_def_mount_opt;
1500 unsigned int s_def_mount_opt2;
1501 ext4_fsblk_t s_sb_block;
1502 atomic64_t s_resv_clusters;
1505 unsigned short s_mount_state;
1506 unsigned short s_pad;
1507 int s_addr_per_block_bits;
1508 int s_desc_per_block_bits;
1511 unsigned int s_inode_readahead_blks;
1512 unsigned int s_inode_goal;
1514 int s_def_hash_version;
1515 int s_hash_unsigned; /* 3 if hash should be unsigned, 0 if not */
1516 struct percpu_counter s_freeclusters_counter;
1517 struct percpu_counter s_freeinodes_counter;
1518 struct percpu_counter s_dirs_counter;
1519 struct percpu_counter s_dirtyclusters_counter;
1520 struct percpu_counter s_sra_exceeded_retry_limit;
1521 struct blockgroup_lock *s_blockgroup_lock;
1522 struct proc_dir_entry *s_proc;
1523 struct kobject s_kobj;
1524 struct completion s_kobj_unregister;
1525 struct super_block *s_sb;
1526 struct buffer_head *s_mmp_bh;
1529 struct journal_s *s_journal;
1530 unsigned long s_ext4_flags; /* Ext4 superblock flags */
1531 struct mutex s_orphan_lock; /* Protects on disk list changes */
1532 struct list_head s_orphan; /* List of orphaned inodes in on disk
1534 struct ext4_orphan_info s_orphan_info;
1535 unsigned long s_commit_interval;
1536 u32 s_max_batch_time;
1537 u32 s_min_batch_time;
1538 struct block_device *s_journal_bdev;
1540 /* Names of quota files with journalled quota */
1541 char __rcu *s_qf_names[EXT4_MAXQUOTAS];
1542 int s_jquota_fmt; /* Format of quota to use */
1544 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1545 struct ext4_system_blocks __rcu *s_system_blks;
1547 #ifdef EXTENTS_STATS
1548 /* ext4 extents stats */
1549 unsigned long s_ext_min;
1550 unsigned long s_ext_max;
1551 unsigned long s_depth_max;
1552 spinlock_t s_ext_stats_lock;
1553 unsigned long s_ext_blocks;
1554 unsigned long s_ext_extents;
1557 /* for buddy allocator */
1558 struct ext4_group_info ** __rcu *s_group_info;
1559 struct inode *s_buddy_cache;
1560 spinlock_t s_md_lock;
1561 unsigned short *s_mb_offsets;
1562 unsigned int *s_mb_maxs;
1563 unsigned int s_group_info_size;
1564 unsigned int s_mb_free_pending;
1565 struct list_head s_freed_data_list; /* List of blocks to be freed
1566 after commit completed */
1567 struct list_head s_discard_list;
1568 struct work_struct s_discard_work;
1569 atomic_t s_retry_alloc_pending;
1570 struct list_head *s_mb_avg_fragment_size;
1571 rwlock_t *s_mb_avg_fragment_size_locks;
1572 struct list_head *s_mb_largest_free_orders;
1573 rwlock_t *s_mb_largest_free_orders_locks;
1576 unsigned long s_stripe;
1577 unsigned int s_mb_max_linear_groups;
1578 unsigned int s_mb_stream_request;
1579 unsigned int s_mb_max_to_scan;
1580 unsigned int s_mb_min_to_scan;
1581 unsigned int s_mb_stats;
1582 unsigned int s_mb_order2_reqs;
1583 unsigned int s_mb_group_prealloc;
1584 unsigned int s_max_dir_size_kb;
1585 /* where last allocation was done - for stream allocation */
1586 unsigned long s_mb_last_group;
1587 unsigned long s_mb_last_start;
1588 unsigned int s_mb_prefetch;
1589 unsigned int s_mb_prefetch_limit;
1590 unsigned int s_mb_best_avail_max_trim_order;
1592 /* stats for buddy allocator */
1593 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1594 atomic_t s_bal_success; /* we found long enough chunks */
1595 atomic_t s_bal_allocated; /* in blocks */
1596 atomic_t s_bal_ex_scanned; /* total extents scanned */
1597 atomic_t s_bal_cX_ex_scanned[EXT4_MB_NUM_CRS]; /* total extents scanned */
1598 atomic_t s_bal_groups_scanned; /* number of groups scanned */
1599 atomic_t s_bal_goals; /* goal hits */
1600 atomic_t s_bal_len_goals; /* len goal hits */
1601 atomic_t s_bal_breaks; /* too long searches */
1602 atomic_t s_bal_2orders; /* 2^order hits */
1603 atomic_t s_bal_p2_aligned_bad_suggestions;
1604 atomic_t s_bal_goal_fast_bad_suggestions;
1605 atomic_t s_bal_best_avail_bad_suggestions;
1606 atomic64_t s_bal_cX_groups_considered[EXT4_MB_NUM_CRS];
1607 atomic64_t s_bal_cX_hits[EXT4_MB_NUM_CRS];
1608 atomic64_t s_bal_cX_failed[EXT4_MB_NUM_CRS]; /* cX loop didn't find blocks */
1609 atomic_t s_mb_buddies_generated; /* number of buddies generated */
1610 atomic64_t s_mb_generation_time;
1611 atomic_t s_mb_lost_chunks;
1612 atomic_t s_mb_preallocated;
1613 atomic_t s_mb_discarded;
1614 atomic_t s_lock_busy;
1616 /* locality groups */
1617 struct ext4_locality_group __percpu *s_locality_groups;
1619 /* for write statistics */
1620 unsigned long s_sectors_written_start;
1621 u64 s_kbytes_written;
1623 /* the size of zero-out chunk */
1624 unsigned int s_extent_max_zeroout_kb;
1626 unsigned int s_log_groups_per_flex;
1627 struct flex_groups * __rcu *s_flex_groups;
1628 ext4_group_t s_flex_groups_allocated;
1630 /* workqueue for reserved extent conversions (buffered io) */
1631 struct workqueue_struct *rsv_conversion_wq;
1633 /* timer for periodic error stats printing */
1634 struct timer_list s_err_report;
1636 /* Lazy inode table initialization info */
1637 struct ext4_li_request *s_li_request;
1638 /* Wait multiplier for lazy initialization thread */
1639 unsigned int s_li_wait_mult;
1641 /* Kernel thread for multiple mount protection */
1642 struct task_struct *s_mmp_tsk;
1644 /* record the last minlen when FITRIM is called. */
1645 unsigned long s_last_trim_minblks;
1647 /* Reference to checksum algorithm driver via cryptoapi */
1648 struct crypto_shash *s_chksum_driver;
1650 /* Precomputed FS UUID checksum for seeding other checksums */
1653 /* Reclaim extents from extent status tree */
1654 struct shrinker s_es_shrinker;
1655 struct list_head s_es_list; /* List of inodes with reclaimable extents */
1657 struct ext4_es_stats s_es_stats;
1658 struct mb_cache *s_ea_block_cache;
1659 struct mb_cache *s_ea_inode_cache;
1660 spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1662 /* Journal triggers for checksum computation */
1663 struct ext4_journal_trigger s_journal_triggers[EXT4_JOURNAL_TRIGGER_COUNT];
1665 /* Ratelimit ext4 messages. */
1666 struct ratelimit_state s_err_ratelimit_state;
1667 struct ratelimit_state s_warning_ratelimit_state;
1668 struct ratelimit_state s_msg_ratelimit_state;
1669 atomic_t s_warning_count;
1670 atomic_t s_msg_count;
1672 /* Encryption policy for '-o test_dummy_encryption' */
1673 struct fscrypt_dummy_policy s_dummy_enc_policy;
1676 * Barrier between writepages ops and changing any inode's JOURNAL_DATA
1679 struct percpu_rw_semaphore s_writepages_rwsem;
1680 struct dax_device *s_daxdev;
1682 #ifdef CONFIG_EXT4_DEBUG
1683 unsigned long s_simulate_fail;
1685 /* Record the errseq of the backing block device */
1686 errseq_t s_bdev_wb_err;
1687 spinlock_t s_bdev_wb_lock;
1689 /* Information about errors that happened during this mount */
1690 spinlock_t s_error_lock;
1691 int s_add_error_count;
1692 int s_first_error_code;
1693 __u32 s_first_error_line;
1694 __u32 s_first_error_ino;
1695 __u64 s_first_error_block;
1696 const char *s_first_error_func;
1697 time64_t s_first_error_time;
1698 int s_last_error_code;
1699 __u32 s_last_error_line;
1700 __u32 s_last_error_ino;
1701 __u64 s_last_error_block;
1702 const char *s_last_error_func;
1703 time64_t s_last_error_time;
1705 * If we are in a context where we cannot update error information in
1706 * the on-disk superblock, we queue this work to do it.
1708 struct work_struct s_error_work;
1710 /* Ext4 fast commit sub transaction ID */
1711 atomic_t s_fc_subtid;
1714 * After commit starts, the main queue gets locked, and the further
1715 * updates get added in the staging queue.
1718 #define FC_Q_STAGING 1
1719 struct list_head s_fc_q[2]; /* Inodes staged for fast commit
1720 * that have data changes in them.
1722 struct list_head s_fc_dentry_q[2]; /* directory entry updates */
1723 unsigned int s_fc_bytes;
1725 * Main fast commit lock. This lock protects accesses to the
1727 * ei->i_fc_list, s_fc_dentry_q, s_fc_q, s_fc_bytes, s_fc_bh.
1729 spinlock_t s_fc_lock;
1730 struct buffer_head *s_fc_bh;
1731 struct ext4_fc_stats s_fc_stats;
1732 tid_t s_fc_ineligible_tid;
1733 #ifdef CONFIG_EXT4_DEBUG
1734 int s_fc_debug_max_replay;
1736 struct ext4_fc_replay_state s_fc_replay_state;
1739 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1741 return sb->s_fs_info;
1743 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1745 return container_of(inode, struct ext4_inode_info, vfs_inode);
1748 static inline int ext4_writepages_down_read(struct super_block *sb)
1750 percpu_down_read(&EXT4_SB(sb)->s_writepages_rwsem);
1751 return memalloc_nofs_save();
1754 static inline void ext4_writepages_up_read(struct super_block *sb, int ctx)
1756 memalloc_nofs_restore(ctx);
1757 percpu_up_read(&EXT4_SB(sb)->s_writepages_rwsem);
1760 static inline int ext4_writepages_down_write(struct super_block *sb)
1762 percpu_down_write(&EXT4_SB(sb)->s_writepages_rwsem);
1763 return memalloc_nofs_save();
1766 static inline void ext4_writepages_up_write(struct super_block *sb, int ctx)
1768 memalloc_nofs_restore(ctx);
1769 percpu_up_write(&EXT4_SB(sb)->s_writepages_rwsem);
1772 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1774 return ino == EXT4_ROOT_INO ||
1775 (ino >= EXT4_FIRST_INO(sb) &&
1776 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1780 * Returns: sbi->field[index]
1781 * Used to access an array element from the following sbi fields which require
1782 * rcu protection to avoid dereferencing an invalid pointer due to reassignment
1787 #define sbi_array_rcu_deref(sbi, field, index) \
1789 typeof(*((sbi)->field)) _v; \
1791 _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index]; \
1792 rcu_read_unlock(); \
1797 * run-time mount flags
1800 EXT4_MF_MNTDIR_SAMPLED,
1801 EXT4_MF_FS_ABORTED, /* Fatal error detected */
1802 EXT4_MF_FC_INELIGIBLE /* Fast commit ineligible */
1805 static inline void ext4_set_mount_flag(struct super_block *sb, int bit)
1807 set_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1810 static inline void ext4_clear_mount_flag(struct super_block *sb, int bit)
1812 clear_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1815 static inline int ext4_test_mount_flag(struct super_block *sb, int bit)
1817 return test_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1822 * Simulate_fail codes
1824 #define EXT4_SIM_BBITMAP_EIO 1
1825 #define EXT4_SIM_BBITMAP_CRC 2
1826 #define EXT4_SIM_IBITMAP_EIO 3
1827 #define EXT4_SIM_IBITMAP_CRC 4
1828 #define EXT4_SIM_INODE_EIO 5
1829 #define EXT4_SIM_INODE_CRC 6
1830 #define EXT4_SIM_DIRBLOCK_EIO 7
1831 #define EXT4_SIM_DIRBLOCK_CRC 8
1833 static inline bool ext4_simulate_fail(struct super_block *sb,
1836 #ifdef CONFIG_EXT4_DEBUG
1837 struct ext4_sb_info *sbi = EXT4_SB(sb);
1839 if (unlikely(sbi->s_simulate_fail == code)) {
1840 sbi->s_simulate_fail = 0;
1847 static inline void ext4_simulate_fail_bh(struct super_block *sb,
1848 struct buffer_head *bh,
1851 if (!IS_ERR(bh) && ext4_simulate_fail(sb, code))
1852 clear_buffer_uptodate(bh);
1856 * Error number codes for s_{first,last}_error_errno
1858 * Linux errno numbers are architecture specific, so we need to translate
1859 * them into something which is architecture independent. We don't define
1860 * codes for all errno's; just the ones which are most likely to be the cause
1861 * of an ext4_error() call.
1863 #define EXT4_ERR_UNKNOWN 1
1864 #define EXT4_ERR_EIO 2
1865 #define EXT4_ERR_ENOMEM 3
1866 #define EXT4_ERR_EFSBADCRC 4
1867 #define EXT4_ERR_EFSCORRUPTED 5
1868 #define EXT4_ERR_ENOSPC 6
1869 #define EXT4_ERR_ENOKEY 7
1870 #define EXT4_ERR_EROFS 8
1871 #define EXT4_ERR_EFBIG 9
1872 #define EXT4_ERR_EEXIST 10
1873 #define EXT4_ERR_ERANGE 11
1874 #define EXT4_ERR_EOVERFLOW 12
1875 #define EXT4_ERR_EBUSY 13
1876 #define EXT4_ERR_ENOTDIR 14
1877 #define EXT4_ERR_ENOTEMPTY 15
1878 #define EXT4_ERR_ESHUTDOWN 16
1879 #define EXT4_ERR_EFAULT 17
1882 * Inode dynamic state flags
1885 EXT4_STATE_NEW, /* inode is newly created */
1886 EXT4_STATE_XATTR, /* has in-inode xattrs */
1887 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1888 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1889 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1890 EXT4_STATE_NEWENTRY, /* File just added to dir */
1891 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1892 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
1893 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */
1894 EXT4_STATE_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */
1895 EXT4_STATE_FC_COMMITTING, /* Fast commit ongoing */
1896 EXT4_STATE_ORPHAN_FILE, /* Inode orphaned in orphan file */
1899 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1900 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1902 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1904 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1906 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1908 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1910 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1913 /* Add these declarations here only so that these functions can be
1914 * found by name. Otherwise, they are very hard to locate. */
1915 static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1916 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1917 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1918 EXT4_INODE_BIT_FNS(flag, flags, 0)
1920 /* Add these declarations here only so that these functions can be
1921 * found by name. Otherwise, they are very hard to locate. */
1922 static inline int ext4_test_inode_state(struct inode *inode, int bit);
1923 static inline void ext4_set_inode_state(struct inode *inode, int bit);
1924 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1925 #if (BITS_PER_LONG < 64)
1926 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1928 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1930 (ei)->i_state_flags = 0;
1933 EXT4_INODE_BIT_FNS(state, flags, 32)
1935 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1937 /* We depend on the fact that callers will set i_flags */
1941 /* Assume that user mode programs are passing in an ext4fs superblock, not
1942 * a kernel struct super_block. This will allow us to call the feature-test
1943 * macros from user land. */
1944 #define EXT4_SB(sb) (sb)
1947 static inline bool ext4_verity_in_progress(struct inode *inode)
1949 return IS_ENABLED(CONFIG_FS_VERITY) &&
1950 ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS);
1953 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1956 * Codes for operating systems
1958 #define EXT4_OS_LINUX 0
1959 #define EXT4_OS_HURD 1
1960 #define EXT4_OS_MASIX 2
1961 #define EXT4_OS_FREEBSD 3
1962 #define EXT4_OS_LITES 4
1967 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1968 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1970 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1972 #define EXT4_GOOD_OLD_INODE_SIZE 128
1974 #define EXT4_EXTRA_TIMESTAMP_MAX (((s64)1 << 34) - 1 + S32_MIN)
1975 #define EXT4_NON_EXTRA_TIMESTAMP_MAX S32_MAX
1976 #define EXT4_TIMESTAMP_MIN S32_MIN
1979 * Feature set definitions
1982 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1983 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1984 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1985 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1986 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1987 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1988 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
1990 * The reason why "FAST_COMMIT" is a compat feature is that, FS becomes
1991 * incompatible only if fast commit blocks are present in the FS. Since we
1992 * clear the journal (and thus the fast commit blocks), we don't mark FS as
1993 * incompatible. We also have a JBD2 incompat feature, which gets set when
1994 * there are fast commit blocks present in the journal.
1996 #define EXT4_FEATURE_COMPAT_FAST_COMMIT 0x0400
1997 #define EXT4_FEATURE_COMPAT_STABLE_INODES 0x0800
1998 #define EXT4_FEATURE_COMPAT_ORPHAN_FILE 0x1000 /* Orphan file exists */
2000 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
2001 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
2002 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
2003 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
2004 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
2005 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
2006 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
2007 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
2008 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
2010 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
2011 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
2012 * all other data structures' checksums. However, the METADATA_CSUM and
2013 * GDT_CSUM bits are mutually exclusive.
2015 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
2016 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000
2017 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000
2018 #define EXT4_FEATURE_RO_COMPAT_VERITY 0x8000
2019 #define EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT 0x10000 /* Orphan file may be
2022 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
2023 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
2024 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
2025 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
2026 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
2027 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
2028 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
2029 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
2030 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
2031 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
2032 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
2033 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000
2034 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
2035 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
2036 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000
2037 #define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000
2039 extern void ext4_update_dynamic_rev(struct super_block *sb);
2041 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
2042 static inline bool ext4_has_feature_##name(struct super_block *sb) \
2044 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
2045 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
2047 static inline void ext4_set_feature_##name(struct super_block *sb) \
2049 ext4_update_dynamic_rev(sb); \
2050 EXT4_SB(sb)->s_es->s_feature_compat |= \
2051 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
2053 static inline void ext4_clear_feature_##name(struct super_block *sb) \
2055 EXT4_SB(sb)->s_es->s_feature_compat &= \
2056 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
2059 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
2060 static inline bool ext4_has_feature_##name(struct super_block *sb) \
2062 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
2063 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
2065 static inline void ext4_set_feature_##name(struct super_block *sb) \
2067 ext4_update_dynamic_rev(sb); \
2068 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
2069 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
2071 static inline void ext4_clear_feature_##name(struct super_block *sb) \
2073 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
2074 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
2077 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
2078 static inline bool ext4_has_feature_##name(struct super_block *sb) \
2080 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
2081 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
2083 static inline void ext4_set_feature_##name(struct super_block *sb) \
2085 ext4_update_dynamic_rev(sb); \
2086 EXT4_SB(sb)->s_es->s_feature_incompat |= \
2087 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
2089 static inline void ext4_clear_feature_##name(struct super_block *sb) \
2091 EXT4_SB(sb)->s_es->s_feature_incompat &= \
2092 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
2095 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC)
2096 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES)
2097 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL)
2098 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR)
2099 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE)
2100 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX)
2101 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2)
2102 EXT4_FEATURE_COMPAT_FUNCS(fast_commit, FAST_COMMIT)
2103 EXT4_FEATURE_COMPAT_FUNCS(stable_inodes, STABLE_INODES)
2104 EXT4_FEATURE_COMPAT_FUNCS(orphan_file, ORPHAN_FILE)
2106 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER)
2107 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE)
2108 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR)
2109 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE)
2110 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM)
2111 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK)
2112 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE)
2113 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA)
2114 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC)
2115 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM)
2116 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY)
2117 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT)
2118 EXT4_FEATURE_RO_COMPAT_FUNCS(verity, VERITY)
2119 EXT4_FEATURE_RO_COMPAT_FUNCS(orphan_present, ORPHAN_PRESENT)
2121 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION)
2122 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE)
2123 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER)
2124 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV)
2125 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG)
2126 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS)
2127 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT)
2128 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP)
2129 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG)
2130 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE)
2131 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA)
2132 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED)
2133 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR)
2134 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA)
2135 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT)
2136 EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD)
2138 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
2139 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
2140 EXT4_FEATURE_INCOMPAT_META_BG)
2141 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2142 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2143 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
2145 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
2146 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
2147 EXT4_FEATURE_INCOMPAT_RECOVER| \
2148 EXT4_FEATURE_INCOMPAT_META_BG)
2149 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2150 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2151 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
2153 #define EXT4_FEATURE_COMPAT_SUPP (EXT4_FEATURE_COMPAT_EXT_ATTR| \
2154 EXT4_FEATURE_COMPAT_ORPHAN_FILE)
2155 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
2156 EXT4_FEATURE_INCOMPAT_RECOVER| \
2157 EXT4_FEATURE_INCOMPAT_META_BG| \
2158 EXT4_FEATURE_INCOMPAT_EXTENTS| \
2159 EXT4_FEATURE_INCOMPAT_64BIT| \
2160 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
2161 EXT4_FEATURE_INCOMPAT_EA_INODE| \
2162 EXT4_FEATURE_INCOMPAT_MMP | \
2163 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
2164 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
2165 EXT4_FEATURE_INCOMPAT_CASEFOLD | \
2166 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \
2167 EXT4_FEATURE_INCOMPAT_LARGEDIR)
2168 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2169 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2170 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
2171 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
2172 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
2173 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
2174 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
2175 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
2176 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
2177 EXT4_FEATURE_RO_COMPAT_QUOTA |\
2178 EXT4_FEATURE_RO_COMPAT_PROJECT |\
2179 EXT4_FEATURE_RO_COMPAT_VERITY |\
2180 EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT)
2182 #define EXTN_FEATURE_FUNCS(ver) \
2183 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
2185 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
2186 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
2188 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
2190 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
2191 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
2193 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
2195 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
2196 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
2199 EXTN_FEATURE_FUNCS(2)
2200 EXTN_FEATURE_FUNCS(3)
2201 EXTN_FEATURE_FUNCS(4)
2203 static inline bool ext4_has_compat_features(struct super_block *sb)
2205 return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
2207 static inline bool ext4_has_ro_compat_features(struct super_block *sb)
2209 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
2211 static inline bool ext4_has_incompat_features(struct super_block *sb)
2213 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
2216 extern int ext4_feature_set_ok(struct super_block *sb, int readonly);
2221 #define EXT4_FLAGS_RESIZING 0
2222 #define EXT4_FLAGS_SHUTDOWN 1
2223 #define EXT4_FLAGS_BDEV_IS_DAX 2
2225 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi)
2227 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags);
2231 * Default values for user and/or group using reserved blocks
2233 #define EXT4_DEF_RESUID 0
2234 #define EXT4_DEF_RESGID 0
2237 * Default project ID
2239 #define EXT4_DEF_PROJID 0
2241 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
2244 * Default mount options
2246 #define EXT4_DEFM_DEBUG 0x0001
2247 #define EXT4_DEFM_BSDGROUPS 0x0002
2248 #define EXT4_DEFM_XATTR_USER 0x0004
2249 #define EXT4_DEFM_ACL 0x0008
2250 #define EXT4_DEFM_UID16 0x0010
2251 #define EXT4_DEFM_JMODE 0x0060
2252 #define EXT4_DEFM_JMODE_DATA 0x0020
2253 #define EXT4_DEFM_JMODE_ORDERED 0x0040
2254 #define EXT4_DEFM_JMODE_WBACK 0x0060
2255 #define EXT4_DEFM_NOBARRIER 0x0100
2256 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
2257 #define EXT4_DEFM_DISCARD 0x0400
2258 #define EXT4_DEFM_NODELALLOC 0x0800
2261 * Default journal batch times
2263 #define EXT4_DEF_MIN_BATCH_TIME 0
2264 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
2267 * Minimum number of groups in a flexgroup before we separate out
2268 * directories into the first block group of a flexgroup
2270 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
2273 * Structure of a directory entry
2275 #define EXT4_NAME_LEN 255
2277 * Base length of the ext4 directory entry excluding the name length
2279 #define EXT4_BASE_DIR_LEN (sizeof(struct ext4_dir_entry_2) - EXT4_NAME_LEN)
2281 struct ext4_dir_entry {
2282 __le32 inode; /* Inode number */
2283 __le16 rec_len; /* Directory entry length */
2284 __le16 name_len; /* Name length */
2285 char name[EXT4_NAME_LEN]; /* File name */
2290 * Encrypted Casefolded entries require saving the hash on disk. This structure
2291 * followed ext4_dir_entry_2's name[name_len] at the next 4 byte aligned
2294 struct ext4_dir_entry_hash {
2300 * The new version of the directory entry. Since EXT4 structures are
2301 * stored in intel byte order, and the name_len field could never be
2302 * bigger than 255 chars, it's safe to reclaim the extra byte for the
2305 struct ext4_dir_entry_2 {
2306 __le32 inode; /* Inode number */
2307 __le16 rec_len; /* Directory entry length */
2308 __u8 name_len; /* Name length */
2309 __u8 file_type; /* See file type macros EXT4_FT_* below */
2310 char name[EXT4_NAME_LEN]; /* File name */
2314 * Access the hashes at the end of ext4_dir_entry_2
2316 #define EXT4_DIRENT_HASHES(entry) \
2317 ((struct ext4_dir_entry_hash *) \
2318 (((void *)(entry)) + \
2319 ((8 + (entry)->name_len + EXT4_DIR_ROUND) & ~EXT4_DIR_ROUND)))
2320 #define EXT4_DIRENT_HASH(entry) le32_to_cpu(EXT4_DIRENT_HASHES(de)->hash)
2321 #define EXT4_DIRENT_MINOR_HASH(entry) \
2322 le32_to_cpu(EXT4_DIRENT_HASHES(de)->minor_hash)
2324 static inline bool ext4_hash_in_dirent(const struct inode *inode)
2326 return IS_CASEFOLDED(inode) && IS_ENCRYPTED(inode);
2330 * This is a bogus directory entry at the end of each leaf block that
2331 * records checksums.
2333 struct ext4_dir_entry_tail {
2334 __le32 det_reserved_zero1; /* Pretend to be unused */
2335 __le16 det_rec_len; /* 12 */
2336 __u8 det_reserved_zero2; /* Zero name length */
2337 __u8 det_reserved_ft; /* 0xDE, fake file type */
2338 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
2341 #define EXT4_DIRENT_TAIL(block, blocksize) \
2342 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
2344 sizeof(struct ext4_dir_entry_tail))))
2347 * Ext4 directory file types. Only the low 3 bits are used. The
2348 * other bits are reserved for now.
2350 #define EXT4_FT_UNKNOWN 0
2351 #define EXT4_FT_REG_FILE 1
2352 #define EXT4_FT_DIR 2
2353 #define EXT4_FT_CHRDEV 3
2354 #define EXT4_FT_BLKDEV 4
2355 #define EXT4_FT_FIFO 5
2356 #define EXT4_FT_SOCK 6
2357 #define EXT4_FT_SYMLINK 7
2359 #define EXT4_FT_MAX 8
2361 #define EXT4_FT_DIR_CSUM 0xDE
2364 * EXT4_DIR_PAD defines the directory entries boundaries
2366 * NOTE: It must be a multiple of 4
2368 #define EXT4_DIR_PAD 4
2369 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
2370 #define EXT4_MAX_REC_LEN ((1<<16)-1)
2373 * The rec_len is dependent on the type of directory. Directories that are
2374 * casefolded and encrypted need to store the hash as well, so we add room for
2375 * ext4_extended_dir_entry_2. For all entries related to '.' or '..' you should
2376 * pass NULL for dir, as those entries do not use the extra fields.
2378 static inline unsigned int ext4_dir_rec_len(__u8 name_len,
2379 const struct inode *dir)
2381 int rec_len = (name_len + 8 + EXT4_DIR_ROUND);
2383 if (dir && ext4_hash_in_dirent(dir))
2384 rec_len += sizeof(struct ext4_dir_entry_hash);
2385 return (rec_len & ~EXT4_DIR_ROUND);
2389 * If we ever get support for fs block sizes > page_size, we'll need
2390 * to remove the #if statements in the next two functions...
2392 static inline unsigned int
2393 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
2395 unsigned len = le16_to_cpu(dlen);
2397 #if (PAGE_SIZE >= 65536)
2398 if (len == EXT4_MAX_REC_LEN || len == 0)
2400 return (len & 65532) | ((len & 3) << 16);
2406 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
2408 BUG_ON((len > blocksize) || (blocksize > (1 << 18)) || (len & 3));
2409 #if (PAGE_SIZE >= 65536)
2411 return cpu_to_le16(len);
2412 if (len == blocksize) {
2413 if (blocksize == 65536)
2414 return cpu_to_le16(EXT4_MAX_REC_LEN);
2416 return cpu_to_le16(0);
2418 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
2420 return cpu_to_le16(len);
2425 * Hash Tree Directory indexing
2426 * (c) Daniel Phillips, 2001
2429 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
2430 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2431 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \
2432 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir)))
2433 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2435 /* Legal values for the dx_root hash_version field: */
2437 #define DX_HASH_LEGACY 0
2438 #define DX_HASH_HALF_MD4 1
2439 #define DX_HASH_TEA 2
2440 #define DX_HASH_LEGACY_UNSIGNED 3
2441 #define DX_HASH_HALF_MD4_UNSIGNED 4
2442 #define DX_HASH_TEA_UNSIGNED 5
2443 #define DX_HASH_SIPHASH 6
2445 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2446 const void *address, unsigned int length)
2449 struct shash_desc shash;
2453 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2455 desc.shash.tfm = sbi->s_chksum_driver;
2456 *(u32 *)desc.ctx = crc;
2458 BUG_ON(crypto_shash_update(&desc.shash, address, length));
2460 return *(u32 *)desc.ctx;
2465 /* hash info structure used by the directory hash */
2475 /* 32 and 64 bit signed EOF for dx directories */
2476 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
2477 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
2481 * Control parameters used by ext4_htree_next_block
2483 #define HASH_NB_ALWAYS 1
2485 struct ext4_filename {
2486 const struct qstr *usr_fname;
2487 struct fscrypt_str disk_name;
2488 struct dx_hash_info hinfo;
2489 #ifdef CONFIG_FS_ENCRYPTION
2490 struct fscrypt_str crypto_buf;
2492 #if IS_ENABLED(CONFIG_UNICODE)
2493 struct fscrypt_str cf_name;
2497 #define fname_name(p) ((p)->disk_name.name)
2498 #define fname_usr_name(p) ((p)->usr_fname->name)
2499 #define fname_len(p) ((p)->disk_name.len)
2502 * Describe an inode's exact location on disk and in memory
2506 struct buffer_head *bh;
2507 unsigned long offset;
2508 ext4_group_t block_group;
2511 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2513 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2516 static inline bool ext4_is_quota_file(struct inode *inode)
2518 return IS_NOQUOTA(inode) &&
2519 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL);
2523 * This structure is stuffed into the struct file's private_data field
2524 * for directories. It is where we put information so that we can do
2525 * readdir operations in hash tree order.
2527 struct dir_private_info {
2528 struct rb_root root;
2529 struct rb_node *curr_node;
2530 struct fname *extra_fname;
2533 __u32 curr_minor_hash;
2537 /* calculate the first block number of the group */
2538 static inline ext4_fsblk_t
2539 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2541 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2542 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2546 * Special error return code only used by dx_probe() and its callers.
2548 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
2550 /* htree levels for ext4 */
2551 #define EXT4_HTREE_LEVEL_COMPAT 2
2552 #define EXT4_HTREE_LEVEL 3
2554 static inline int ext4_dir_htree_level(struct super_block *sb)
2556 return ext4_has_feature_largedir(sb) ?
2557 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT;
2561 * Timeout and state flag for lazy initialization inode thread.
2563 #define EXT4_DEF_LI_WAIT_MULT 10
2564 #define EXT4_DEF_LI_MAX_START_DELAY 5
2565 #define EXT4_LAZYINIT_QUIT 0x0001
2566 #define EXT4_LAZYINIT_RUNNING 0x0002
2569 * Lazy inode table initialization info
2571 struct ext4_lazy_init {
2572 unsigned long li_state;
2573 struct list_head li_request_list;
2574 struct mutex li_list_mtx;
2578 EXT4_LI_MODE_PREFETCH_BBITMAP,
2579 EXT4_LI_MODE_ITABLE,
2582 struct ext4_li_request {
2583 struct super_block *lr_super;
2584 enum ext4_li_mode lr_mode;
2585 ext4_group_t lr_first_not_zeroed;
2586 ext4_group_t lr_next_group;
2587 struct list_head lr_request;
2588 unsigned long lr_next_sched;
2589 unsigned long lr_timeout;
2592 struct ext4_features {
2593 struct kobject f_kobj;
2594 struct completion f_kobj_unregister;
2598 * This structure will be used for multiple mount protection. It will be
2599 * written into the block number saved in the s_mmp_block field in the
2600 * superblock. Programs that check MMP should assume that if
2601 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2602 * to use the filesystem, regardless of how old the timestamp is.
2604 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
2605 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2606 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
2607 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
2610 __le32 mmp_magic; /* Magic number for MMP */
2611 __le32 mmp_seq; /* Sequence no. updated periodically */
2614 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2615 * purposes and do not affect the correctness of the algorithm
2617 __le64 mmp_time; /* Time last updated */
2618 char mmp_nodename[64]; /* Node which last updated MMP block */
2619 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
2622 * mmp_check_interval is used to verify if the MMP block has been
2623 * updated on the block device. The value is updated based on the
2624 * maximum time to write the MMP block during an update cycle.
2626 __le16 mmp_check_interval;
2629 __le32 mmp_pad2[226];
2630 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
2633 /* arguments passed to the mmp thread */
2635 struct buffer_head *bh; /* bh from initial read_mmp_block() */
2636 struct super_block *sb; /* super block of the fs */
2640 * Check interval multiplier
2641 * The MMP block is written every update interval and initially checked every
2642 * update interval x the multiplier (the value is then adapted based on the
2643 * write latency). The reason is that writes can be delayed under load and we
2644 * don't want readers to incorrectly assume that the filesystem is no longer
2647 #define EXT4_MMP_CHECK_MULT 2UL
2650 * Minimum interval for MMP checking in seconds.
2652 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
2655 * Maximum interval for MMP checking in seconds.
2657 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
2660 * Function prototypes
2664 * Ok, these declarations are also in <linux/kernel.h> but none of the
2665 * ext4 source programs needs to include it so they are duplicated here.
2667 # define NORET_TYPE /**/
2668 # define ATTRIB_NORET __attribute__((noreturn))
2669 # define NORET_AND noreturn,
2672 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2673 void ext4_inode_bitmap_csum_set(struct super_block *sb,
2674 struct ext4_group_desc *gdp,
2675 struct buffer_head *bh, int sz);
2676 int ext4_inode_bitmap_csum_verify(struct super_block *sb,
2677 struct ext4_group_desc *gdp,
2678 struct buffer_head *bh, int sz);
2679 void ext4_block_bitmap_csum_set(struct super_block *sb,
2680 struct ext4_group_desc *gdp,
2681 struct buffer_head *bh);
2682 int ext4_block_bitmap_csum_verify(struct super_block *sb,
2683 struct ext4_group_desc *gdp,
2684 struct buffer_head *bh);
2687 extern void ext4_get_group_no_and_offset(struct super_block *sb,
2688 ext4_fsblk_t blocknr,
2689 ext4_group_t *blockgrpp,
2690 ext4_grpblk_t *offsetp);
2691 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2692 ext4_fsblk_t block);
2694 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2695 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2696 ext4_group_t group);
2697 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2700 unsigned long *count,
2702 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2703 s64 nclusters, unsigned int flags);
2704 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2705 extern void ext4_check_blocks_bitmap(struct super_block *);
2706 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2707 ext4_group_t block_group,
2708 struct buffer_head ** bh);
2709 extern struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2710 ext4_group_t group);
2711 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2713 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2714 ext4_group_t block_group,
2715 bool ignore_locked);
2716 extern int ext4_wait_block_bitmap(struct super_block *sb,
2717 ext4_group_t block_group,
2718 struct buffer_head *bh);
2719 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2720 ext4_group_t block_group);
2721 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2722 ext4_group_t block_group,
2723 struct ext4_group_desc *gdp);
2724 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2726 #if IS_ENABLED(CONFIG_UNICODE)
2727 extern int ext4_fname_setup_ci_filename(struct inode *dir,
2728 const struct qstr *iname,
2729 struct ext4_filename *fname);
2732 /* ext4 encryption related stuff goes here crypto.c */
2733 #ifdef CONFIG_FS_ENCRYPTION
2734 extern const struct fscrypt_operations ext4_cryptops;
2736 int ext4_fname_setup_filename(struct inode *dir, const struct qstr *iname,
2737 int lookup, struct ext4_filename *fname);
2739 int ext4_fname_prepare_lookup(struct inode *dir, struct dentry *dentry,
2740 struct ext4_filename *fname);
2742 void ext4_fname_free_filename(struct ext4_filename *fname);
2744 int ext4_ioctl_get_encryption_pwsalt(struct file *filp, void __user *arg);
2746 #else /* !CONFIG_FS_ENCRYPTION */
2747 static inline int ext4_fname_setup_filename(struct inode *dir,
2748 const struct qstr *iname,
2750 struct ext4_filename *fname)
2753 fname->usr_fname = iname;
2754 fname->disk_name.name = (unsigned char *) iname->name;
2755 fname->disk_name.len = iname->len;
2757 #if IS_ENABLED(CONFIG_UNICODE)
2758 err = ext4_fname_setup_ci_filename(dir, iname, fname);
2764 static inline int ext4_fname_prepare_lookup(struct inode *dir,
2765 struct dentry *dentry,
2766 struct ext4_filename *fname)
2768 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname);
2771 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2773 #if IS_ENABLED(CONFIG_UNICODE)
2774 kfree(fname->cf_name.name);
2775 fname->cf_name.name = NULL;
2779 static inline int ext4_ioctl_get_encryption_pwsalt(struct file *filp,
2784 #endif /* !CONFIG_FS_ENCRYPTION */
2787 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2789 struct ext4_dir_entry_2 *,
2790 struct buffer_head *, char *, int,
2792 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2793 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2794 (de), (bh), (buf), (size), (offset)))
2795 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2797 struct ext4_dir_entry_2 *dirent,
2798 struct fscrypt_str *ent_name);
2799 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2800 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2801 struct buffer_head *bh,
2802 void *buf, int buf_size,
2803 struct ext4_filename *fname,
2804 struct ext4_dir_entry_2 **dest_de);
2805 void ext4_insert_dentry(struct inode *dir, struct inode *inode,
2806 struct ext4_dir_entry_2 *de,
2808 struct ext4_filename *fname);
2809 static inline void ext4_update_dx_flag(struct inode *inode)
2811 if (!ext4_has_feature_dir_index(inode->i_sb) &&
2812 ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) {
2813 /* ext4_iget() should have caught this... */
2814 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb));
2815 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2818 static const unsigned char ext4_filetype_table[] = {
2819 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2822 static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2824 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2827 return ext4_filetype_table[filetype];
2829 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2830 void *buf, int buf_size);
2833 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2836 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len,
2837 struct dx_hash_info *hinfo);
2840 extern int ext4_mark_inode_used(struct super_block *sb, int ino);
2841 extern struct inode *__ext4_new_inode(struct mnt_idmap *, handle_t *,
2842 struct inode *, umode_t,
2843 const struct qstr *qstr, __u32 goal,
2844 uid_t *owner, __u32 i_flags,
2845 int handle_type, unsigned int line_no,
2848 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \
2849 __ext4_new_inode(&nop_mnt_idmap, (handle), (dir), (mode), (qstr), \
2850 (goal), (owner), i_flags, 0, 0, 0)
2851 #define ext4_new_inode_start_handle(idmap, dir, mode, qstr, goal, owner, \
2853 __ext4_new_inode((idmap), NULL, (dir), (mode), (qstr), (goal), (owner), \
2854 0, (type), __LINE__, (nblocks))
2857 extern void ext4_free_inode(handle_t *, struct inode *);
2858 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2859 extern unsigned long ext4_count_free_inodes(struct super_block *);
2860 extern unsigned long ext4_count_dirs(struct super_block *);
2861 extern void ext4_check_inodes_bitmap(struct super_block *);
2862 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2863 extern int ext4_init_inode_table(struct super_block *sb,
2864 ext4_group_t group, int barrier);
2865 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2868 int ext4_fc_info_show(struct seq_file *seq, void *v);
2869 void ext4_fc_init(struct super_block *sb, journal_t *journal);
2870 void ext4_fc_init_inode(struct inode *inode);
2871 void ext4_fc_track_range(handle_t *handle, struct inode *inode, ext4_lblk_t start,
2873 void __ext4_fc_track_unlink(handle_t *handle, struct inode *inode,
2874 struct dentry *dentry);
2875 void __ext4_fc_track_link(handle_t *handle, struct inode *inode,
2876 struct dentry *dentry);
2877 void ext4_fc_track_unlink(handle_t *handle, struct dentry *dentry);
2878 void ext4_fc_track_link(handle_t *handle, struct dentry *dentry);
2879 void __ext4_fc_track_create(handle_t *handle, struct inode *inode,
2880 struct dentry *dentry);
2881 void ext4_fc_track_create(handle_t *handle, struct dentry *dentry);
2882 void ext4_fc_track_inode(handle_t *handle, struct inode *inode);
2883 void ext4_fc_mark_ineligible(struct super_block *sb, int reason, handle_t *handle);
2884 void ext4_fc_start_update(struct inode *inode);
2885 void ext4_fc_stop_update(struct inode *inode);
2886 void ext4_fc_del(struct inode *inode);
2887 bool ext4_fc_replay_check_excluded(struct super_block *sb, ext4_fsblk_t block);
2888 void ext4_fc_replay_cleanup(struct super_block *sb);
2889 int ext4_fc_commit(journal_t *journal, tid_t commit_tid);
2890 int __init ext4_fc_init_dentry_cache(void);
2891 void ext4_fc_destroy_dentry_cache(void);
2892 int ext4_fc_record_regions(struct super_block *sb, int ino,
2893 ext4_lblk_t lblk, ext4_fsblk_t pblk,
2894 int len, int replay);
2897 extern const struct seq_operations ext4_mb_seq_groups_ops;
2898 extern const struct seq_operations ext4_mb_seq_structs_summary_ops;
2899 extern int ext4_seq_mb_stats_show(struct seq_file *seq, void *offset);
2900 extern int ext4_mb_init(struct super_block *);
2901 extern int ext4_mb_release(struct super_block *);
2902 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2903 struct ext4_allocation_request *, int *);
2904 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2905 extern void ext4_discard_preallocations(struct inode *, unsigned int);
2906 extern int __init ext4_init_mballoc(void);
2907 extern void ext4_exit_mballoc(void);
2908 extern ext4_group_t ext4_mb_prefetch(struct super_block *sb,
2910 unsigned int nr, int *cnt);
2911 extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group,
2914 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2915 struct buffer_head *bh, ext4_fsblk_t block,
2916 unsigned long count, int flags);
2917 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2918 ext4_group_t ngroups);
2919 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2920 ext4_group_t i, struct ext4_group_desc *desc);
2921 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2922 ext4_fsblk_t block, unsigned long count);
2923 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2924 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid);
2925 extern void ext4_mb_mark_bb(struct super_block *sb, ext4_fsblk_t block,
2926 int len, int state);
2929 void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw,
2930 struct ext4_inode_info *ei);
2931 int ext4_inode_is_fast_symlink(struct inode *inode);
2932 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2933 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2934 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count,
2935 bool wait, struct buffer_head **bhs);
2936 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2937 struct buffer_head *bh_result, int create);
2938 int ext4_get_block(struct inode *inode, sector_t iblock,
2939 struct buffer_head *bh_result, int create);
2940 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2941 struct buffer_head *bh, int create);
2942 int ext4_walk_page_buffers(handle_t *handle,
2943 struct inode *inode,
2944 struct buffer_head *head,
2948 int (*fn)(handle_t *handle, struct inode *inode,
2949 struct buffer_head *bh));
2950 int do_journal_get_write_access(handle_t *handle, struct inode *inode,
2951 struct buffer_head *bh);
2952 #define FALL_BACK_TO_NONDELALLOC 1
2953 #define CONVERT_INLINE_DATA 2
2956 EXT4_IGET_NORMAL = 0,
2957 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */
2958 EXT4_IGET_HANDLE = 0x0002, /* Inode # is from a handle */
2959 EXT4_IGET_BAD = 0x0004, /* Allow to iget a bad inode */
2960 EXT4_IGET_EA_INODE = 0x0008 /* Inode should contain an EA value */
2963 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
2964 ext4_iget_flags flags, const char *function,
2967 #define ext4_iget(sb, ino, flags) \
2968 __ext4_iget((sb), (ino), (flags), __func__, __LINE__)
2970 extern int ext4_write_inode(struct inode *, struct writeback_control *);
2971 extern int ext4_setattr(struct mnt_idmap *, struct dentry *,
2973 extern u32 ext4_dio_alignment(struct inode *inode);
2974 extern int ext4_getattr(struct mnt_idmap *, const struct path *,
2975 struct kstat *, u32, unsigned int);
2976 extern void ext4_evict_inode(struct inode *);
2977 extern void ext4_clear_inode(struct inode *);
2978 extern int ext4_file_getattr(struct mnt_idmap *, const struct path *,
2979 struct kstat *, u32, unsigned int);
2980 extern int ext4_sync_inode(handle_t *, struct inode *);
2981 extern void ext4_dirty_inode(struct inode *, int);
2982 extern int ext4_change_inode_journal_flag(struct inode *, int);
2983 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2984 extern int ext4_get_fc_inode_loc(struct super_block *sb, unsigned long ino,
2985 struct ext4_iloc *iloc);
2986 extern int ext4_inode_attach_jinode(struct inode *inode);
2987 extern int ext4_can_truncate(struct inode *inode);
2988 extern int ext4_truncate(struct inode *);
2989 extern int ext4_break_layouts(struct inode *);
2990 extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length);
2991 extern void ext4_set_inode_flags(struct inode *, bool init);
2992 extern int ext4_alloc_da_blocks(struct inode *inode);
2993 extern void ext4_set_aops(struct inode *inode);
2994 extern int ext4_writepage_trans_blocks(struct inode *);
2995 extern int ext4_normal_submit_inode_data_buffers(struct jbd2_inode *jinode);
2996 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2997 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2998 loff_t lstart, loff_t lend);
2999 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf);
3000 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
3001 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
3002 extern void ext4_da_release_space(struct inode *inode, int to_free);
3003 extern void ext4_da_update_reserve_space(struct inode *inode,
3004 int used, int quota_claim);
3005 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
3006 ext4_fsblk_t pblk, ext4_lblk_t len);
3009 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
3010 struct ext4_map_blocks *map, int flags);
3011 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
3012 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
3013 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
3014 ext4_lblk_t start, ext4_lblk_t end);
3017 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
3018 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
3019 int ext4_fileattr_set(struct mnt_idmap *idmap,
3020 struct dentry *dentry, struct fileattr *fa);
3021 int ext4_fileattr_get(struct dentry *dentry, struct fileattr *fa);
3022 extern void ext4_reset_inode_seed(struct inode *inode);
3023 int ext4_update_overhead(struct super_block *sb, bool force);
3024 int ext4_force_shutdown(struct super_block *sb, u32 flags);
3027 extern int ext4_ext_migrate(struct inode *);
3028 extern int ext4_ind_migrate(struct inode *inode);
3031 extern int ext4_init_new_dir(handle_t *handle, struct inode *dir,
3032 struct inode *inode);
3033 extern int ext4_dirblock_csum_verify(struct inode *inode,
3034 struct buffer_head *bh);
3035 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
3036 __u32 start_minor_hash, __u32 *next_hash);
3037 extern int ext4_search_dir(struct buffer_head *bh,
3041 struct ext4_filename *fname,
3042 unsigned int offset,
3043 struct ext4_dir_entry_2 **res_dir);
3044 extern int ext4_generic_delete_entry(struct inode *dir,
3045 struct ext4_dir_entry_2 *de_del,
3046 struct buffer_head *bh,
3050 extern bool ext4_empty_dir(struct inode *inode);
3053 extern void ext4_kvfree_array_rcu(void *to_free);
3054 extern int ext4_group_add(struct super_block *sb,
3055 struct ext4_new_group_data *input);
3056 extern int ext4_group_extend(struct super_block *sb,
3057 struct ext4_super_block *es,
3058 ext4_fsblk_t n_blocks_count);
3059 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
3060 extern unsigned int ext4_list_backups(struct super_block *sb,
3061 unsigned int *three, unsigned int *five,
3062 unsigned int *seven);
3065 extern struct buffer_head *ext4_sb_bread(struct super_block *sb,
3066 sector_t block, blk_opf_t op_flags);
3067 extern struct buffer_head *ext4_sb_bread_unmovable(struct super_block *sb,
3069 extern void ext4_read_bh_nowait(struct buffer_head *bh, blk_opf_t op_flags,
3070 bh_end_io_t *end_io);
3071 extern int ext4_read_bh(struct buffer_head *bh, blk_opf_t op_flags,
3072 bh_end_io_t *end_io);
3073 extern int ext4_read_bh_lock(struct buffer_head *bh, blk_opf_t op_flags, bool wait);
3074 extern void ext4_sb_breadahead_unmovable(struct super_block *sb, sector_t block);
3075 extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
3076 extern int ext4_calculate_overhead(struct super_block *sb);
3077 extern __le32 ext4_superblock_csum(struct super_block *sb,
3078 struct ext4_super_block *es);
3079 extern void ext4_superblock_csum_set(struct super_block *sb);
3080 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
3081 ext4_group_t ngroup);
3082 extern const char *ext4_decode_error(struct super_block *sb, int errno,
3084 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb,
3085 ext4_group_t block_group,
3086 unsigned int flags);
3088 extern __printf(7, 8)
3089 void __ext4_error(struct super_block *, const char *, unsigned int, bool,
3090 int, __u64, const char *, ...);
3091 extern __printf(6, 7)
3092 void __ext4_error_inode(struct inode *, const char *, unsigned int,
3093 ext4_fsblk_t, int, const char *, ...);
3094 extern __printf(5, 6)
3095 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
3097 extern void __ext4_std_error(struct super_block *, const char *,
3099 extern __printf(4, 5)
3100 void __ext4_warning(struct super_block *, const char *, unsigned int,
3102 extern __printf(4, 5)
3103 void __ext4_warning_inode(const struct inode *inode, const char *function,
3104 unsigned int line, const char *fmt, ...);
3105 extern __printf(3, 4)
3106 void __ext4_msg(struct super_block *, const char *, const char *, ...);
3107 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
3108 const char *, unsigned int, const char *);
3109 extern __printf(7, 8)
3110 void __ext4_grp_locked_error(const char *, unsigned int,
3111 struct super_block *, ext4_group_t,
3112 unsigned long, ext4_fsblk_t,
3115 #define EXT4_ERROR_INODE(inode, fmt, a...) \
3116 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
3118 #define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...) \
3119 __ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a)
3121 #define ext4_error_inode_block(inode, block, err, fmt, a...) \
3122 __ext4_error_inode((inode), __func__, __LINE__, (block), (err), \
3125 #define EXT4_ERROR_FILE(file, block, fmt, a...) \
3126 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
3128 #define ext4_abort(sb, err, fmt, a...) \
3129 __ext4_error((sb), __func__, __LINE__, true, (err), 0, (fmt), ## a)
3131 #ifdef CONFIG_PRINTK
3133 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
3134 __ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__)
3135 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \
3136 __ext4_error_inode((inode), (func), (line), (block), \
3137 (err), (fmt), ##__VA_ARGS__)
3138 #define ext4_error_file(file, func, line, block, fmt, ...) \
3139 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
3140 #define ext4_error(sb, fmt, ...) \
3141 __ext4_error((sb), __func__, __LINE__, false, 0, 0, (fmt), \
3143 #define ext4_error_err(sb, err, fmt, ...) \
3144 __ext4_error((sb), __func__, __LINE__, false, (err), 0, (fmt), \
3146 #define ext4_warning(sb, fmt, ...) \
3147 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
3148 #define ext4_warning_inode(inode, fmt, ...) \
3149 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
3150 #define ext4_msg(sb, level, fmt, ...) \
3151 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
3152 #define dump_mmp_msg(sb, mmp, msg) \
3153 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
3154 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
3155 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
3160 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
3162 no_printk(fmt, ##__VA_ARGS__); \
3163 __ext4_error_inode(inode, "", 0, block, 0, " "); \
3165 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \
3167 no_printk(fmt, ##__VA_ARGS__); \
3168 __ext4_error_inode(inode, "", 0, block, err, " "); \
3170 #define ext4_error_file(file, func, line, block, fmt, ...) \
3172 no_printk(fmt, ##__VA_ARGS__); \
3173 __ext4_error_file(file, "", 0, block, " "); \
3175 #define ext4_error(sb, fmt, ...) \
3177 no_printk(fmt, ##__VA_ARGS__); \
3178 __ext4_error(sb, "", 0, false, 0, 0, " "); \
3180 #define ext4_error_err(sb, err, fmt, ...) \
3182 no_printk(fmt, ##__VA_ARGS__); \
3183 __ext4_error(sb, "", 0, false, err, 0, " "); \
3185 #define ext4_warning(sb, fmt, ...) \
3187 no_printk(fmt, ##__VA_ARGS__); \
3188 __ext4_warning(sb, "", 0, " "); \
3190 #define ext4_warning_inode(inode, fmt, ...) \
3192 no_printk(fmt, ##__VA_ARGS__); \
3193 __ext4_warning_inode(inode, "", 0, " "); \
3195 #define ext4_msg(sb, level, fmt, ...) \
3197 no_printk(fmt, ##__VA_ARGS__); \
3198 __ext4_msg(sb, "", " "); \
3200 #define dump_mmp_msg(sb, mmp, msg) \
3201 __dump_mmp_msg(sb, mmp, "", 0, "")
3202 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
3204 no_printk(fmt, ##__VA_ARGS__); \
3205 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
3210 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
3211 struct ext4_group_desc *bg);
3212 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
3213 struct ext4_group_desc *bg);
3214 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
3215 struct ext4_group_desc *bg);
3216 extern __u32 ext4_free_group_clusters(struct super_block *sb,
3217 struct ext4_group_desc *bg);
3218 extern __u32 ext4_free_inodes_count(struct super_block *sb,
3219 struct ext4_group_desc *bg);
3220 extern __u32 ext4_used_dirs_count(struct super_block *sb,
3221 struct ext4_group_desc *bg);
3222 extern __u32 ext4_itable_unused_count(struct super_block *sb,
3223 struct ext4_group_desc *bg);
3224 extern void ext4_block_bitmap_set(struct super_block *sb,
3225 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3226 extern void ext4_inode_bitmap_set(struct super_block *sb,
3227 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3228 extern void ext4_inode_table_set(struct super_block *sb,
3229 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3230 extern void ext4_free_group_clusters_set(struct super_block *sb,
3231 struct ext4_group_desc *bg,
3233 extern void ext4_free_inodes_set(struct super_block *sb,
3234 struct ext4_group_desc *bg, __u32 count);
3235 extern void ext4_used_dirs_set(struct super_block *sb,
3236 struct ext4_group_desc *bg, __u32 count);
3237 extern void ext4_itable_unused_set(struct super_block *sb,
3238 struct ext4_group_desc *bg, __u32 count);
3239 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
3240 struct ext4_group_desc *gdp);
3241 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
3242 struct ext4_group_desc *gdp);
3243 extern int ext4_register_li_request(struct super_block *sb,
3244 ext4_group_t first_not_zeroed);
3246 static inline int ext4_has_metadata_csum(struct super_block *sb)
3248 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
3249 !EXT4_SB(sb)->s_chksum_driver);
3251 return ext4_has_feature_metadata_csum(sb) &&
3252 (EXT4_SB(sb)->s_chksum_driver != NULL);
3255 static inline int ext4_has_group_desc_csum(struct super_block *sb)
3257 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb);
3260 #define ext4_read_incompat_64bit_val(es, name) \
3261 (((es)->s_feature_incompat & cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT) \
3262 ? (ext4_fsblk_t)le32_to_cpu(es->name##_hi) << 32 : 0) | \
3263 le32_to_cpu(es->name##_lo))
3265 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
3267 return ext4_read_incompat_64bit_val(es, s_blocks_count);
3270 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
3272 return ext4_read_incompat_64bit_val(es, s_r_blocks_count);
3275 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
3277 return ext4_read_incompat_64bit_val(es, s_free_blocks_count);
3280 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
3283 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
3284 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
3287 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
3290 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
3291 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
3294 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
3297 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
3298 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
3301 static inline loff_t ext4_isize(struct super_block *sb,
3302 struct ext4_inode *raw_inode)
3304 if (ext4_has_feature_largedir(sb) ||
3305 S_ISREG(le16_to_cpu(raw_inode->i_mode)))
3306 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
3307 le32_to_cpu(raw_inode->i_size_lo);
3309 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
3312 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
3314 raw_inode->i_size_lo = cpu_to_le32(i_size);
3315 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
3319 * Reading s_groups_count requires using smp_rmb() afterwards. See
3320 * the locking protocol documented in the comments of ext4_group_add()
3323 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
3325 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
3331 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
3332 ext4_group_t block_group)
3334 return block_group >> sbi->s_log_groups_per_flex;
3337 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
3339 return 1 << sbi->s_log_groups_per_flex;
3342 #define ext4_std_error(sb, errno) \
3345 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
3349 /* Each CPU can accumulate percpu_counter_batch clusters in their local
3350 * counters. So we need to make sure we have free clusters more
3351 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
3353 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
3355 #define EXT4_FREECLUSTERS_WATERMARK 0
3358 /* Update i_disksize. Requires i_rwsem to avoid races with truncate */
3359 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
3361 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
3362 !inode_is_locked(inode));
3363 down_write(&EXT4_I(inode)->i_data_sem);
3364 if (newsize > EXT4_I(inode)->i_disksize)
3365 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize);
3366 up_write(&EXT4_I(inode)->i_data_sem);
3369 /* Update i_size, i_disksize. Requires i_rwsem to avoid races with truncate */
3370 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
3374 if (newsize > inode->i_size) {
3375 i_size_write(inode, newsize);
3378 if (newsize > EXT4_I(inode)->i_disksize) {
3379 ext4_update_i_disksize(inode, newsize);
3385 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
3388 struct ext4_group_info {
3389 unsigned long bb_state;
3390 #ifdef AGGRESSIVE_CHECK
3391 unsigned long bb_check_counter;
3393 struct rb_root bb_free_root;
3394 ext4_grpblk_t bb_first_free; /* first free block */
3395 ext4_grpblk_t bb_free; /* total free blocks */
3396 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
3397 int bb_avg_fragment_size_order; /* order of average
3399 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
3400 ext4_group_t bb_group; /* Group number */
3401 struct list_head bb_prealloc_list;
3405 struct rw_semaphore alloc_sem;
3406 struct list_head bb_avg_fragment_size_node;
3407 struct list_head bb_largest_free_order_node;
3408 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
3409 * regions, index is order.
3410 * bb_counters[3] = 5 means
3411 * 5 free 8-block regions. */
3414 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
3415 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
3416 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
3417 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
3418 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \
3419 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT)
3420 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \
3421 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT)
3422 #define EXT4_GROUP_INFO_BBITMAP_READ_BIT 4
3424 #define EXT4_MB_GRP_NEED_INIT(grp) \
3425 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
3426 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
3427 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3428 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
3429 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3431 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
3432 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3433 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
3434 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3435 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
3436 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3437 #define EXT4_MB_GRP_TEST_AND_SET_READ(grp) \
3438 (test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state)))
3440 #define EXT4_MAX_CONTENTION 8
3441 #define EXT4_CONTENTION_THRESHOLD 2
3443 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
3446 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
3450 * Returns true if the filesystem is busy enough that attempts to
3451 * access the block group locks has run into contention.
3453 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
3455 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
3458 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
3460 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
3461 if (spin_trylock(lock))
3463 * We're able to grab the lock right away, so drop the
3464 * lock contention counter.
3466 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
3469 * The lock is busy, so bump the contention counter,
3470 * and then wait on the spin lock.
3472 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
3473 EXT4_MAX_CONTENTION);
3478 static inline void ext4_unlock_group(struct super_block *sb,
3481 spin_unlock(ext4_group_lock_ptr(sb, group));
3485 static inline bool ext4_quota_capable(struct super_block *sb)
3487 return (test_opt(sb, QUOTA) || ext4_has_feature_quota(sb));
3490 static inline bool ext4_is_quota_journalled(struct super_block *sb)
3492 struct ext4_sb_info *sbi = EXT4_SB(sb);
3494 return (ext4_has_feature_quota(sb) ||
3495 sbi->s_qf_names[USRQUOTA] || sbi->s_qf_names[GRPQUOTA]);
3497 int ext4_enable_quotas(struct super_block *sb);
3501 * Block validity checking
3503 #define ext4_check_indirect_blockref(inode, bh) \
3504 ext4_check_blockref(__func__, __LINE__, inode, \
3505 (__le32 *)(bh)->b_data, \
3506 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
3508 #define ext4_ind_check_inode(inode) \
3509 ext4_check_blockref(__func__, __LINE__, inode, \
3510 EXT4_I(inode)->i_data, \
3514 * Inodes and files operations
3518 extern const struct file_operations ext4_dir_operations;
3521 extern const struct inode_operations ext4_file_inode_operations;
3522 extern const struct file_operations ext4_file_operations;
3523 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
3526 extern int ext4_get_max_inline_size(struct inode *inode);
3527 extern int ext4_find_inline_data_nolock(struct inode *inode);
3528 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
3530 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
3532 int ext4_readpage_inline(struct inode *inode, struct folio *folio);
3533 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3534 struct inode *inode,
3535 loff_t pos, unsigned len,
3536 struct page **pagep);
3537 int ext4_write_inline_data_end(struct inode *inode, loff_t pos, unsigned len,
3538 unsigned copied, struct folio *folio);
3539 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
3540 struct inode *inode,
3541 loff_t pos, unsigned len,
3542 struct page **pagep,
3544 extern int ext4_try_add_inline_entry(handle_t *handle,
3545 struct ext4_filename *fname,
3546 struct inode *dir, struct inode *inode);
3547 extern int ext4_try_create_inline_dir(handle_t *handle,
3548 struct inode *parent,
3549 struct inode *inode);
3550 extern int ext4_read_inline_dir(struct file *filp,
3551 struct dir_context *ctx,
3552 int *has_inline_data);
3553 extern int ext4_inlinedir_to_tree(struct file *dir_file,
3554 struct inode *dir, ext4_lblk_t block,
3555 struct dx_hash_info *hinfo,
3556 __u32 start_hash, __u32 start_minor_hash,
3557 int *has_inline_data);
3558 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3559 struct ext4_filename *fname,
3560 struct ext4_dir_entry_2 **res_dir,
3561 int *has_inline_data);
3562 extern int ext4_delete_inline_entry(handle_t *handle,
3564 struct ext4_dir_entry_2 *de_del,
3565 struct buffer_head *bh,
3566 int *has_inline_data);
3567 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3568 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3569 struct ext4_dir_entry_2 **parent_de,
3571 extern void *ext4_read_inline_link(struct inode *inode);
3574 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap);
3576 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3578 extern int ext4_convert_inline_data(struct inode *inode);
3580 static inline int ext4_has_inline_data(struct inode *inode)
3582 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3583 EXT4_I(inode)->i_inline_off;
3587 extern const struct inode_operations ext4_dir_inode_operations;
3588 extern const struct inode_operations ext4_special_inode_operations;
3589 extern struct dentry *ext4_get_parent(struct dentry *child);
3590 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3591 struct ext4_dir_entry_2 *de,
3592 int blocksize, int csum_size,
3593 unsigned int parent_ino, int dotdot_real_len);
3594 extern void ext4_initialize_dirent_tail(struct buffer_head *bh,
3595 unsigned int blocksize);
3596 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode,
3597 struct buffer_head *bh);
3598 extern int __ext4_unlink(struct inode *dir, const struct qstr *d_name,
3599 struct inode *inode, struct dentry *dentry);
3600 extern int __ext4_link(struct inode *dir, struct inode *inode,
3601 struct dentry *dentry);
3604 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = {
3605 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
3606 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
3607 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
3608 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
3609 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
3610 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
3611 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
3614 static inline void ext4_set_de_type(struct super_block *sb,
3615 struct ext4_dir_entry_2 *de,
3617 if (ext4_has_feature_filetype(sb))
3618 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3622 extern int ext4_mpage_readpages(struct inode *inode,
3623 struct readahead_control *rac, struct folio *folio);
3624 extern int __init ext4_init_post_read_processing(void);
3625 extern void ext4_exit_post_read_processing(void);
3628 extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3629 extern const struct inode_operations ext4_symlink_inode_operations;
3630 extern const struct inode_operations ext4_fast_symlink_inode_operations;
3633 extern void ext4_notify_error_sysfs(struct ext4_sb_info *sbi);
3634 extern int ext4_register_sysfs(struct super_block *sb);
3635 extern void ext4_unregister_sysfs(struct super_block *sb);
3636 extern int __init ext4_init_sysfs(void);
3637 extern void ext4_exit_sysfs(void);
3639 /* block_validity */
3640 extern void ext4_release_system_zone(struct super_block *sb);
3641 extern int ext4_setup_system_zone(struct super_block *sb);
3642 extern int __init ext4_init_system_zone(void);
3643 extern void ext4_exit_system_zone(void);
3644 extern int ext4_inode_block_valid(struct inode *inode,
3645 ext4_fsblk_t start_blk,
3646 unsigned int count);
3647 extern int ext4_check_blockref(const char *, unsigned int,
3648 struct inode *, __le32 *, unsigned int);
3649 extern int ext4_sb_block_valid(struct super_block *sb, struct inode *inode,
3650 ext4_fsblk_t start_blk, unsigned int count);
3654 struct ext4_ext_path;
3658 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3661 #define EXT_MAX_BLOCKS 0xffffffff
3663 extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode);
3664 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3665 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3666 struct ext4_map_blocks *map, int flags);
3667 extern int ext4_ext_truncate(handle_t *, struct inode *);
3668 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3670 extern void ext4_ext_init(struct super_block *);
3671 extern void ext4_ext_release(struct super_block *);
3672 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3674 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3675 loff_t offset, ssize_t len);
3676 extern int ext4_convert_unwritten_io_end_vec(handle_t *handle,
3677 ext4_io_end_t *io_end);
3678 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3679 struct ext4_map_blocks *map, int flags);
3680 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3682 struct ext4_ext_path *path);
3683 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3684 struct ext4_ext_path **,
3685 struct ext4_extent *, int);
3686 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3687 struct ext4_ext_path **,
3689 extern void ext4_free_ext_path(struct ext4_ext_path *);
3690 extern int ext4_ext_check_inode(struct inode *inode);
3691 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3692 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3693 __u64 start, __u64 len);
3694 extern int ext4_get_es_cache(struct inode *inode,
3695 struct fiemap_extent_info *fieinfo,
3696 __u64 start, __u64 len);
3697 extern int ext4_ext_precache(struct inode *inode);
3698 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3699 struct inode *inode2, ext4_lblk_t lblk1,
3700 ext4_lblk_t lblk2, ext4_lblk_t count,
3701 int mark_unwritten,int *err);
3702 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu);
3703 extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode,
3704 int check_cred, int restart_cred,
3706 extern void ext4_ext_replay_shrink_inode(struct inode *inode, ext4_lblk_t end);
3707 extern int ext4_ext_replay_set_iblocks(struct inode *inode);
3708 extern int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start,
3709 int len, int unwritten, ext4_fsblk_t pblk);
3710 extern int ext4_ext_clear_bb(struct inode *inode);
3714 extern void ext4_double_down_write_data_sem(struct inode *first,
3715 struct inode *second);
3716 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3717 struct inode *donor_inode);
3718 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3719 __u64 start_orig, __u64 start_donor,
3720 __u64 len, __u64 *moved_len);
3723 extern int __init ext4_init_pageio(void);
3724 extern void ext4_exit_pageio(void);
3725 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3726 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3727 extern int ext4_put_io_end(ext4_io_end_t *io_end);
3728 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3729 extern void ext4_io_submit_init(struct ext4_io_submit *io,
3730 struct writeback_control *wbc);
3731 extern void ext4_end_io_rsv_work(struct work_struct *work);
3732 extern void ext4_io_submit(struct ext4_io_submit *io);
3733 int ext4_bio_write_folio(struct ext4_io_submit *io, struct folio *page,
3735 extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end);
3736 extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end);
3739 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3742 extern void ext4_stop_mmpd(struct ext4_sb_info *sbi);
3745 extern const struct fsverity_operations ext4_verityops;
3748 extern int ext4_orphan_add(handle_t *, struct inode *);
3749 extern int ext4_orphan_del(handle_t *, struct inode *);
3750 extern void ext4_orphan_cleanup(struct super_block *sb,
3751 struct ext4_super_block *es);
3752 extern void ext4_release_orphan_info(struct super_block *sb);
3753 extern int ext4_init_orphan_info(struct super_block *sb);
3754 extern int ext4_orphan_file_empty(struct super_block *sb);
3755 extern void ext4_orphan_file_block_trigger(
3756 struct jbd2_buffer_trigger_type *triggers,
3757 struct buffer_head *bh,
3758 void *data, size_t size);
3761 * Add new method to test whether block and inode bitmaps are properly
3762 * initialized. With uninit_bg reading the block from disk is not enough
3763 * to mark the bitmap uptodate. We need to also zero-out the bitmap
3765 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3767 static inline int bitmap_uptodate(struct buffer_head *bh)
3769 return (buffer_uptodate(bh) &&
3770 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3772 static inline void set_bitmap_uptodate(struct buffer_head *bh)
3774 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3777 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
3779 /* For ioend & aio unwritten conversion wait queues */
3780 #define EXT4_WQ_HASH_SZ 37
3781 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
3783 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3785 extern int ext4_resize_begin(struct super_block *sb);
3786 extern int ext4_resize_end(struct super_block *sb, bool update_backups);
3788 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
3789 struct ext4_io_end *io_end)
3791 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
3792 io_end->flag |= EXT4_IO_END_UNWRITTEN;
3793 atomic_inc(&EXT4_I(inode)->i_unwritten);
3797 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3799 struct inode *inode = io_end->inode;
3801 if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
3802 io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3803 /* Wake up anyone waiting on unwritten extent conversion */
3804 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
3805 wake_up_all(ext4_ioend_wq(inode));
3809 extern const struct iomap_ops ext4_iomap_ops;
3810 extern const struct iomap_ops ext4_iomap_overwrite_ops;
3811 extern const struct iomap_ops ext4_iomap_report_ops;
3813 static inline int ext4_buffer_uptodate(struct buffer_head *bh)
3816 * If the buffer has the write error flag, we have failed
3817 * to write out data in the block. In this case, we don't
3818 * have to read the block because we may read the old data
3821 if (buffer_write_io_error(bh))
3822 set_buffer_uptodate(bh);
3823 return buffer_uptodate(bh);
3826 #endif /* __KERNEL__ */
3828 #define EFSBADCRC EBADMSG /* Bad CRC detected */
3829 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
3831 #endif /* _EXT4_H */