1 // SPDX-License-Identifier: GPL-2.0
5 * Copyright (C) 1992, 1993, 1994, 1995
6 * Remy Card (card@masi.ibp.fr)
7 * Laboratoire MASI - Institut Blaise Pascal
8 * Universite Pierre et Marie Curie (Paris VI)
12 * linux/include/linux/minix_fs.h
14 * Copyright (C) 1991, 1992 Linus Torvalds
20 #include <linux/types.h>
21 #include <linux/blkdev.h>
22 #include <linux/magic.h>
23 #include <linux/jbd2.h>
24 #include <linux/quota.h>
25 #include <linux/rwsem.h>
26 #include <linux/rbtree.h>
27 #include <linux/seqlock.h>
28 #include <linux/mutex.h>
29 #include <linux/timer.h>
30 #include <linux/version.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>
44 #include <linux/fscrypt.h>
45 #include <linux/fsverity.h>
47 #include <linux/compiler.h>
50 * The fourth extended filesystem constants/structures
54 * with AGGRESSIVE_CHECK allocator runs consistency checks over
55 * structures. these checks slow things down a lot
57 #define AGGRESSIVE_CHECK__
60 * with DOUBLE_CHECK defined mballoc creates persistent in-core
61 * bitmaps, maintains and uses them to check for double allocations
63 #define DOUBLE_CHECK__
66 * Define EXT4FS_DEBUG to produce debug messages
74 #define ext4_debug(f, a...) \
76 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
77 __FILE__, __LINE__, __func__); \
78 printk(KERN_DEBUG f, ## a); \
81 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
85 * Turn on EXT_DEBUG to enable ext4_ext_show_path/leaf/move in extents.c
90 * Dynamic printk for controlled extents debugging.
92 #ifdef CONFIG_EXT4_DEBUG
93 #define ext_debug(ino, fmt, ...) \
94 pr_debug("[%s/%d] EXT4-fs (%s): ino %lu: (%s, %d): %s:" fmt, \
95 current->comm, task_pid_nr(current), \
96 ino->i_sb->s_id, ino->i_ino, __FILE__, __LINE__, \
97 __func__, ##__VA_ARGS__)
99 #define ext_debug(ino, fmt, ...) no_printk(fmt, ##__VA_ARGS__)
102 /* data type for block offset of block group */
103 typedef int ext4_grpblk_t;
105 /* data type for filesystem-wide blocks number */
106 typedef unsigned long long ext4_fsblk_t;
108 /* data type for file logical block number */
109 typedef __u32 ext4_lblk_t;
111 /* data type for block group number */
112 typedef unsigned int ext4_group_t;
114 enum SHIFT_DIRECTION {
120 * Flags used in mballoc's allocation_context flags field.
122 * Also used to show what's going on for debugging purposes when the
123 * flag field is exported via the traceport interface
126 /* prefer goal again. length */
127 #define EXT4_MB_HINT_MERGE 0x0001
128 /* blocks already reserved */
129 #define EXT4_MB_HINT_RESERVED 0x0002
130 /* metadata is being allocated */
131 #define EXT4_MB_HINT_METADATA 0x0004
132 /* first blocks in the file */
133 #define EXT4_MB_HINT_FIRST 0x0008
134 /* search for the best chunk */
135 #define EXT4_MB_HINT_BEST 0x0010
136 /* data is being allocated */
137 #define EXT4_MB_HINT_DATA 0x0020
138 /* don't preallocate (for tails) */
139 #define EXT4_MB_HINT_NOPREALLOC 0x0040
140 /* allocate for locality group */
141 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080
142 /* allocate goal blocks or none */
143 #define EXT4_MB_HINT_GOAL_ONLY 0x0100
144 /* goal is meaningful */
145 #define EXT4_MB_HINT_TRY_GOAL 0x0200
146 /* blocks already pre-reserved by delayed allocation */
147 #define EXT4_MB_DELALLOC_RESERVED 0x0400
148 /* We are doing stream allocation */
149 #define EXT4_MB_STREAM_ALLOC 0x0800
150 /* Use reserved root blocks if needed */
151 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000
152 /* Use blocks from reserved pool */
153 #define EXT4_MB_USE_RESERVED 0x2000
154 /* Do strict check for free blocks while retrying block allocation */
155 #define EXT4_MB_STRICT_CHECK 0x4000
157 struct ext4_allocation_request {
158 /* target inode for block we're allocating */
160 /* how many blocks we want to allocate */
162 /* logical block in target inode */
164 /* the closest logical allocated block to the left */
166 /* the closest logical allocated block to the right */
168 /* phys. target (a hint) */
170 /* phys. block for the closest logical allocated block to the left */
172 /* phys. block for the closest logical allocated block to the right */
174 /* flags. see above EXT4_MB_HINT_* */
179 * Logical to physical block mapping, used by ext4_map_blocks()
181 * This structure is used to pass requests into ext4_map_blocks() as
182 * well as to store the information returned by ext4_map_blocks(). It
183 * takes less room on the stack than a struct buffer_head.
185 #define EXT4_MAP_NEW BIT(BH_New)
186 #define EXT4_MAP_MAPPED BIT(BH_Mapped)
187 #define EXT4_MAP_UNWRITTEN BIT(BH_Unwritten)
188 #define EXT4_MAP_BOUNDARY BIT(BH_Boundary)
189 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
190 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
192 struct ext4_map_blocks {
196 unsigned int m_flags;
200 * Block validity checking, system zone rbtree.
202 struct ext4_system_blocks {
208 * Flags for ext4_io_end->flags
210 #define EXT4_IO_END_UNWRITTEN 0x0001
212 struct ext4_io_end_vec {
213 struct list_head list; /* list of io_end_vec */
214 loff_t offset; /* offset in the file */
215 ssize_t size; /* size of the extent */
219 * For converting unwritten extents on a work queue. 'handle' is used for
220 * buffered writeback.
222 typedef struct ext4_io_end {
223 struct list_head list; /* per-file finished IO list */
224 handle_t *handle; /* handle reserved for extent
226 struct inode *inode; /* file being written to */
227 struct bio *bio; /* Linked list of completed
228 * bios covering the extent */
229 unsigned int flag; /* unwritten or not */
230 atomic_t count; /* reference counter */
231 struct list_head list_vec; /* list of ext4_io_end_vec */
234 struct ext4_io_submit {
235 struct writeback_control *io_wbc;
237 ext4_io_end_t *io_end;
238 sector_t io_next_block;
242 * Special inodes numbers
244 #define EXT4_BAD_INO 1 /* Bad blocks inode */
245 #define EXT4_ROOT_INO 2 /* Root inode */
246 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */
247 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
248 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
249 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
250 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
251 #define EXT4_JOURNAL_INO 8 /* Journal inode */
253 /* First non-reserved inode for old ext4 filesystems */
254 #define EXT4_GOOD_OLD_FIRST_INO 11
257 * Maximal count of links to a file
259 #define EXT4_LINK_MAX 65000
262 * Macro-instructions used to manage several block sizes
264 #define EXT4_MIN_BLOCK_SIZE 1024
265 #define EXT4_MAX_BLOCK_SIZE 65536
266 #define EXT4_MIN_BLOCK_LOG_SIZE 10
267 #define EXT4_MAX_BLOCK_LOG_SIZE 16
268 #define EXT4_MAX_CLUSTER_LOG_SIZE 30
270 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
272 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
274 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
275 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
276 EXT4_SB(s)->s_cluster_bits)
278 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
279 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
281 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
284 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
285 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
286 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
288 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
289 EXT4_GOOD_OLD_INODE_SIZE : \
291 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
292 EXT4_GOOD_OLD_FIRST_INO : \
295 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
296 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \
297 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \
300 /* Translate a block number to a cluster number */
301 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
302 /* Translate a cluster number to a block number */
303 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
304 /* Translate # of blks to # of clusters */
305 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
306 (sbi)->s_cluster_bits)
307 /* Mask out the low bits to get the starting block of the cluster */
308 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
309 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
310 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
311 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
312 /* Fill in the low bits to get the last block of the cluster */
313 #define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) | \
314 ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1))
315 /* Get the cluster offset */
316 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
317 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
318 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
319 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
322 * Structure of a blocks group descriptor
324 struct ext4_group_desc
326 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
327 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
328 __le32 bg_inode_table_lo; /* Inodes table block */
329 __le16 bg_free_blocks_count_lo;/* Free blocks count */
330 __le16 bg_free_inodes_count_lo;/* Free inodes count */
331 __le16 bg_used_dirs_count_lo; /* Directories count */
332 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
333 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
334 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
335 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
336 __le16 bg_itable_unused_lo; /* Unused inodes count */
337 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
338 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
339 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
340 __le32 bg_inode_table_hi; /* Inodes table block MSB */
341 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
342 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
343 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
344 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
345 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
346 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
347 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
351 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
352 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
354 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
355 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
359 * Structure of a flex block group info
363 atomic64_t free_clusters;
364 atomic_t free_inodes;
368 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
369 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
370 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
373 * Macro-instructions used to manage group descriptors
375 #define EXT4_MIN_DESC_SIZE 32
376 #define EXT4_MIN_DESC_SIZE_64BIT 64
377 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
378 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
380 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
381 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
382 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
383 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
384 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
386 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
387 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
388 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
392 * Constants relative to the data blocks
394 #define EXT4_NDIR_BLOCKS 12
395 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
396 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
397 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
398 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
403 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
404 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
405 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
406 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
407 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
408 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
409 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
410 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
411 /* Reserved for compression usage... */
412 #define EXT4_DIRTY_FL 0x00000100
413 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
414 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
415 /* nb: was previously EXT2_ECOMPR_FL */
416 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */
417 /* End compression flags --- maybe not all used */
418 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
419 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
420 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
421 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
422 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
423 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
424 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
425 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
426 #define EXT4_VERITY_FL 0x00100000 /* Verity protected inode */
427 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
428 /* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */
429 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
430 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
431 #define EXT4_CASEFOLD_FL 0x40000000 /* Casefolded directory */
432 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
434 #define EXT4_FL_USER_VISIBLE 0x705BDFFF /* User visible flags */
435 #define EXT4_FL_USER_MODIFIABLE 0x604BC0FF /* User modifiable flags */
437 /* Flags we can manipulate with through EXT4_IOC_FSSETXATTR */
438 #define EXT4_FL_XFLAG_VISIBLE (EXT4_SYNC_FL | \
439 EXT4_IMMUTABLE_FL | \
445 /* Flags that should be inherited by new inodes from their parent. */
446 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
447 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
448 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
449 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
450 EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL)
452 /* Flags that are appropriate for regular files (all but dir-specific ones). */
453 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\
454 EXT4_PROJINHERIT_FL))
456 /* Flags that are appropriate for non-directories/regular files. */
457 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
459 /* The only flags that should be swapped */
460 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL)
462 /* Mask out flags that are inappropriate for the given type of inode. */
463 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
467 else if (S_ISREG(mode))
468 return flags & EXT4_REG_FLMASK;
470 return flags & EXT4_OTHER_FLMASK;
474 * Inode flags used for atomic set/get
477 EXT4_INODE_SECRM = 0, /* Secure deletion */
478 EXT4_INODE_UNRM = 1, /* Undelete */
479 EXT4_INODE_COMPR = 2, /* Compress file */
480 EXT4_INODE_SYNC = 3, /* Synchronous updates */
481 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
482 EXT4_INODE_APPEND = 5, /* writes to file may only append */
483 EXT4_INODE_NODUMP = 6, /* do not dump file */
484 EXT4_INODE_NOATIME = 7, /* do not update atime */
485 /* Reserved for compression usage... */
486 EXT4_INODE_DIRTY = 8,
487 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
488 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
489 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */
490 /* End compression flags --- maybe not all used */
491 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
492 EXT4_INODE_IMAGIC = 13, /* AFS directory */
493 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
494 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
495 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
496 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
497 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
498 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
499 EXT4_INODE_VERITY = 20, /* Verity protected inode */
500 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
501 /* 22 was formerly EXT4_INODE_EOFBLOCKS */
502 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
503 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */
504 EXT4_INODE_CASEFOLD = 30, /* Casefolded directory */
505 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
509 * Since it's pretty easy to mix up bit numbers and hex values, we use a
510 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
511 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
512 * any extra space in the compiled kernel image, otherwise, the build will fail.
513 * It's important that these values are the same, since we are using
514 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
515 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
516 * values found in ext2, ext3 and ext4 filesystems, and of course the values
517 * defined in e2fsprogs.
519 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
521 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
522 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
524 static inline void ext4_check_flag_values(void)
526 CHECK_FLAG_VALUE(SECRM);
527 CHECK_FLAG_VALUE(UNRM);
528 CHECK_FLAG_VALUE(COMPR);
529 CHECK_FLAG_VALUE(SYNC);
530 CHECK_FLAG_VALUE(IMMUTABLE);
531 CHECK_FLAG_VALUE(APPEND);
532 CHECK_FLAG_VALUE(NODUMP);
533 CHECK_FLAG_VALUE(NOATIME);
534 CHECK_FLAG_VALUE(DIRTY);
535 CHECK_FLAG_VALUE(COMPRBLK);
536 CHECK_FLAG_VALUE(NOCOMPR);
537 CHECK_FLAG_VALUE(ENCRYPT);
538 CHECK_FLAG_VALUE(INDEX);
539 CHECK_FLAG_VALUE(IMAGIC);
540 CHECK_FLAG_VALUE(JOURNAL_DATA);
541 CHECK_FLAG_VALUE(NOTAIL);
542 CHECK_FLAG_VALUE(DIRSYNC);
543 CHECK_FLAG_VALUE(TOPDIR);
544 CHECK_FLAG_VALUE(HUGE_FILE);
545 CHECK_FLAG_VALUE(EXTENTS);
546 CHECK_FLAG_VALUE(VERITY);
547 CHECK_FLAG_VALUE(EA_INODE);
548 CHECK_FLAG_VALUE(INLINE_DATA);
549 CHECK_FLAG_VALUE(PROJINHERIT);
550 CHECK_FLAG_VALUE(CASEFOLD);
551 CHECK_FLAG_VALUE(RESERVED);
554 /* Used to pass group descriptor data when online resize is done */
555 struct ext4_new_group_input {
556 __u32 group; /* Group number for this data */
557 __u64 block_bitmap; /* Absolute block number of block bitmap */
558 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
559 __u64 inode_table; /* Absolute block number of inode table start */
560 __u32 blocks_count; /* Total number of blocks in this group */
561 __u16 reserved_blocks; /* Number of reserved blocks in this group */
565 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
566 struct compat_ext4_new_group_input {
568 compat_u64 block_bitmap;
569 compat_u64 inode_bitmap;
570 compat_u64 inode_table;
577 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
578 struct ext4_new_group_data {
584 __u16 reserved_blocks;
586 __u32 free_clusters_count;
589 /* Indexes used to index group tables in ext4_new_group_data */
591 BLOCK_BITMAP = 0, /* block bitmap */
592 INODE_BITMAP, /* inode bitmap */
593 INODE_TABLE, /* inode tables */
598 * Flags used by ext4_map_blocks()
600 /* Allocate any needed blocks and/or convert an unwritten
601 extent to be an initialized ext4 */
602 #define EXT4_GET_BLOCKS_CREATE 0x0001
603 /* Request the creation of an unwritten extent */
604 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002
605 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\
606 EXT4_GET_BLOCKS_CREATE)
607 /* Caller is from the delayed allocation writeout path
608 * finally doing the actual allocation of delayed blocks */
609 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
610 /* caller is from the direct IO path, request to creation of an
611 unwritten extents if not allocated, split the unwritten
612 extent if blocks has been preallocated already*/
613 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
614 #define EXT4_GET_BLOCKS_CONVERT 0x0010
615 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
616 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
617 /* Convert extent to initialized after IO complete */
618 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
619 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
620 /* Eventual metadata allocation (due to growing extent tree)
621 * should not fail, so try to use reserved blocks for that.*/
622 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
623 /* Don't normalize allocation size (used for fallocate) */
624 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
625 /* Convert written extents to unwritten */
626 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100
627 /* Write zeros to newly created written extents */
628 #define EXT4_GET_BLOCKS_ZERO 0x0200
629 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\
630 EXT4_GET_BLOCKS_ZERO)
631 /* Caller will submit data before dropping transaction handle. This
632 * allows jbd2 to avoid submitting data before commit. */
633 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400
636 * The bit position of these flags must not overlap with any of the
637 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(),
638 * read_extent_tree_block(), ext4_split_extent_at(),
639 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
640 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
641 * caching the extents when reading from the extent tree while a
642 * truncate or punch hole operation is in progress.
644 #define EXT4_EX_NOCACHE 0x40000000
645 #define EXT4_EX_FORCE_CACHE 0x20000000
646 #define EXT4_EX_NOFAIL 0x10000000
649 * Flags used by ext4_free_blocks
651 #define EXT4_FREE_BLOCKS_METADATA 0x0001
652 #define EXT4_FREE_BLOCKS_FORGET 0x0002
653 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
654 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
655 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
656 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
657 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040
662 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
663 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
664 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
665 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
666 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
667 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
668 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
669 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
670 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
671 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
672 #define EXT4_IOC_MIGRATE _IO('f', 9)
673 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
674 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
675 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
676 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
677 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
678 #define EXT4_IOC_SWAP_BOOT _IO('f', 17)
679 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18)
680 #define EXT4_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY
681 #define EXT4_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT
682 #define EXT4_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY
683 /* ioctl codes 19--39 are reserved for fscrypt */
684 #define EXT4_IOC_CLEAR_ES_CACHE _IO('f', 40)
685 #define EXT4_IOC_GETSTATE _IOW('f', 41, __u32)
686 #define EXT4_IOC_GET_ES_CACHE _IOWR('f', 42, struct fiemap)
688 #define EXT4_IOC_FSGETXATTR FS_IOC_FSGETXATTR
689 #define EXT4_IOC_FSSETXATTR FS_IOC_FSSETXATTR
691 #define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32)
694 * Flags for going down operation
696 #define EXT4_GOING_FLAGS_DEFAULT 0x0 /* going down */
697 #define EXT4_GOING_FLAGS_LOGFLUSH 0x1 /* flush log but not data */
698 #define EXT4_GOING_FLAGS_NOLOGFLUSH 0x2 /* don't flush log nor data */
701 * Flags returned by EXT4_IOC_GETSTATE
703 * We only expose to userspace a subset of the state flags in
706 #define EXT4_STATE_FLAG_EXT_PRECACHED 0x00000001
707 #define EXT4_STATE_FLAG_NEW 0x00000002
708 #define EXT4_STATE_FLAG_NEWENTRY 0x00000004
709 #define EXT4_STATE_FLAG_DA_ALLOC_CLOSE 0x00000008
711 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
713 * ioctl commands in 32 bit emulation
715 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
716 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
717 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
718 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
719 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
720 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
721 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
722 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
723 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
724 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
728 * Returned by EXT4_IOC_GET_ES_CACHE as an additional possible flag.
729 * It indicates that the entry in extent status cache is for a hole.
731 #define EXT4_FIEMAP_EXTENT_HOLE 0x08000000
733 /* Max physical block we can address w/o extents */
734 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
736 /* Max logical block we can support */
737 #define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFE
740 * Structure of an inode on the disk
743 __le16 i_mode; /* File mode */
744 __le16 i_uid; /* Low 16 bits of Owner Uid */
745 __le32 i_size_lo; /* Size in bytes */
746 __le32 i_atime; /* Access time */
747 __le32 i_ctime; /* Inode Change time */
748 __le32 i_mtime; /* Modification time */
749 __le32 i_dtime; /* Deletion Time */
750 __le16 i_gid; /* Low 16 bits of Group Id */
751 __le16 i_links_count; /* Links count */
752 __le32 i_blocks_lo; /* Blocks count */
753 __le32 i_flags; /* File flags */
759 __u32 h_i_translator;
764 } osd1; /* OS dependent 1 */
765 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
766 __le32 i_generation; /* File version (for NFS) */
767 __le32 i_file_acl_lo; /* File ACL */
769 __le32 i_obso_faddr; /* Obsoleted fragment address */
772 __le16 l_i_blocks_high; /* were l_i_reserved1 */
773 __le16 l_i_file_acl_high;
774 __le16 l_i_uid_high; /* these 2 fields */
775 __le16 l_i_gid_high; /* were reserved2[0] */
776 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
780 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
787 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
788 __le16 m_i_file_acl_high;
789 __u32 m_i_reserved2[2];
791 } osd2; /* OS dependent 2 */
792 __le16 i_extra_isize;
793 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
794 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
795 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
796 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
797 __le32 i_crtime; /* File Creation time */
798 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
799 __le32 i_version_hi; /* high 32 bits for 64-bit version */
800 __le32 i_projid; /* Project ID */
804 __u32 reserved; /* should be zero */
805 __u32 donor_fd; /* donor file descriptor */
806 __u64 orig_start; /* logical start offset in block for orig */
807 __u64 donor_start; /* logical start offset in block for donor */
808 __u64 len; /* block length to be moved */
809 __u64 moved_len; /* moved block length */
812 #define EXT4_EPOCH_BITS 2
813 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
814 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
817 * Extended fields will fit into an inode if the filesystem was formatted
818 * with large inodes (-I 256 or larger) and there are not currently any EAs
819 * consuming all of the available space. For new inodes we always reserve
820 * enough space for the kernel's known extended fields, but for inodes
821 * created with an old kernel this might not have been the case. None of
822 * the extended inode fields is critical for correct filesystem operation.
823 * This macro checks if a certain field fits in the inode. Note that
824 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
826 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
827 ((offsetof(typeof(*ext4_inode), field) + \
828 sizeof((ext4_inode)->field)) \
829 <= (EXT4_GOOD_OLD_INODE_SIZE + \
830 (einode)->i_extra_isize)) \
833 * We use an encoding that preserves the times for extra epoch "00":
835 * extra msb of adjust for signed
836 * epoch 32-bit 32-bit tv_sec to
837 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
838 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
839 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
840 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
841 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
842 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
843 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
844 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
845 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
847 * Note that previous versions of the kernel on 64-bit systems would
848 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
849 * 1970. e2fsck will correct this, assuming that it is run on the
850 * affected filesystem before 2242.
853 static inline __le32 ext4_encode_extra_time(struct timespec64 *time)
855 u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK;
856 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
859 static inline void ext4_decode_extra_time(struct timespec64 *time,
862 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK)))
863 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
864 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
867 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
869 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {\
870 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
871 (raw_inode)->xtime ## _extra = \
872 ext4_encode_extra_time(&(inode)->xtime); \
875 (raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (inode)->xtime.tv_sec, S32_MIN, S32_MAX)); \
878 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
880 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
881 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
882 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
883 (raw_inode)->xtime ## _extra = \
884 ext4_encode_extra_time(&(einode)->xtime); \
887 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
889 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
890 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \
891 ext4_decode_extra_time(&(inode)->xtime, \
892 raw_inode->xtime ## _extra); \
895 (inode)->xtime.tv_nsec = 0; \
899 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
901 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
902 (einode)->xtime.tv_sec = \
903 (signed)le32_to_cpu((raw_inode)->xtime); \
905 (einode)->xtime.tv_sec = 0; \
906 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
907 ext4_decode_extra_time(&(einode)->xtime, \
908 raw_inode->xtime ## _extra); \
910 (einode)->xtime.tv_nsec = 0; \
913 #define i_disk_version osd1.linux1.l_i_version
915 #if defined(__KERNEL__) || defined(__linux__)
916 #define i_reserved1 osd1.linux1.l_i_reserved1
917 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
918 #define i_blocks_high osd2.linux2.l_i_blocks_high
919 #define i_uid_low i_uid
920 #define i_gid_low i_gid
921 #define i_uid_high osd2.linux2.l_i_uid_high
922 #define i_gid_high osd2.linux2.l_i_gid_high
923 #define i_checksum_lo osd2.linux2.l_i_checksum_lo
925 #elif defined(__GNU__)
927 #define i_translator osd1.hurd1.h_i_translator
928 #define i_uid_high osd2.hurd2.h_i_uid_high
929 #define i_gid_high osd2.hurd2.h_i_gid_high
930 #define i_author osd2.hurd2.h_i_author
932 #elif defined(__masix__)
934 #define i_reserved1 osd1.masix1.m_i_reserved1
935 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
936 #define i_reserved2 osd2.masix2.m_i_reserved2
938 #endif /* defined(__KERNEL__) || defined(__linux__) */
940 #include "extents_status.h"
943 * Lock subclasses for i_data_sem in the ext4_inode_info structure.
945 * These are needed to avoid lockdep false positives when we need to
946 * allocate blocks to the quota inode during ext4_map_blocks(), while
947 * holding i_data_sem for a normal (non-quota) inode. Since we don't
948 * do quota tracking for the quota inode, this avoids deadlock (as
949 * well as infinite recursion, since it isn't turtles all the way
952 * I_DATA_SEM_NORMAL - Used for most inodes
953 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode
954 * where the second inode has larger inode number
956 * I_DATA_SEM_QUOTA - Used for quota inodes only
959 I_DATA_SEM_NORMAL = 0,
966 * fourth extended file system inode data in memory
968 struct ext4_inode_info {
969 __le32 i_data[15]; /* unconverted */
971 ext4_fsblk_t i_file_acl;
974 * i_block_group is the number of the block group which contains
975 * this file's inode. Constant across the lifetime of the inode,
976 * it is used for making block allocation decisions - we try to
977 * place a file's data blocks near its inode block, and new inodes
978 * near to their parent directory's inode.
980 ext4_group_t i_block_group;
981 ext4_lblk_t i_dir_start_lookup;
982 #if (BITS_PER_LONG < 64)
983 unsigned long i_state_flags; /* Dynamic state flags */
985 unsigned long i_flags;
988 * Extended attributes can be read independently of the main file
989 * data. Taking i_mutex even when reading would cause contention
990 * between readers of EAs and writers of regular file data, so
991 * instead we synchronize on xattr_sem when reading or changing
994 struct rw_semaphore xattr_sem;
996 struct list_head i_orphan; /* unlinked but open inodes */
999 * i_disksize keeps track of what the inode size is ON DISK, not
1000 * in memory. During truncate, i_size is set to the new size by
1001 * the VFS prior to calling ext4_truncate(), but the filesystem won't
1002 * set i_disksize to 0 until the truncate is actually under way.
1004 * The intent is that i_disksize always represents the blocks which
1005 * are used by this file. This allows recovery to restart truncate
1006 * on orphans if we crash during truncate. We actually write i_disksize
1007 * into the on-disk inode when writing inodes out, instead of i_size.
1009 * The only time when i_disksize and i_size may be different is when
1010 * a truncate is in progress. The only things which change i_disksize
1011 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
1016 * i_data_sem is for serialising ext4_truncate() against
1017 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
1018 * data tree are chopped off during truncate. We can't do that in
1019 * ext4 because whenever we perform intermediate commits during
1020 * truncate, the inode and all the metadata blocks *must* be in a
1021 * consistent state which allows truncation of the orphans to restart
1022 * during recovery. Hence we must fix the get_block-vs-truncate race
1023 * by other means, so we have i_data_sem.
1025 struct rw_semaphore i_data_sem;
1027 * i_mmap_sem is for serializing page faults with truncate / punch hole
1028 * operations. We have to make sure that new page cannot be faulted in
1029 * a section of the inode that is being punched. We cannot easily use
1030 * i_data_sem for this since we need protection for the whole punch
1031 * operation and i_data_sem ranks below transaction start so we have
1032 * to occasionally drop it.
1034 struct rw_semaphore i_mmap_sem;
1035 struct inode vfs_inode;
1036 struct jbd2_inode *jinode;
1038 spinlock_t i_raw_lock; /* protects updates to the raw inode */
1041 * File creation time. Its function is same as that of
1042 * struct timespec64 i_{a,c,m}time in the generic inode.
1044 struct timespec64 i_crtime;
1047 struct list_head i_prealloc_list;
1048 spinlock_t i_prealloc_lock;
1050 /* extents status tree */
1051 struct ext4_es_tree i_es_tree;
1053 struct list_head i_es_list;
1054 unsigned int i_es_all_nr; /* protected by i_es_lock */
1055 unsigned int i_es_shk_nr; /* protected by i_es_lock */
1056 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for
1057 extents to shrink. Protected by
1061 ext4_group_t i_last_alloc_group;
1063 /* allocation reservation info for delalloc */
1064 /* In case of bigalloc, this refer to clusters rather than blocks */
1065 unsigned int i_reserved_data_blocks;
1067 /* pending cluster reservations for bigalloc file systems */
1068 struct ext4_pending_tree i_pending_tree;
1070 /* on-disk additional length */
1071 __u16 i_extra_isize;
1073 /* Indicate the inline data space. */
1078 /* quota space reservation, managed internally by quota code */
1079 qsize_t i_reserved_quota;
1082 /* Lock protecting lists below */
1083 spinlock_t i_completed_io_lock;
1085 * Completed IOs that need unwritten extents handling and have
1086 * transaction reserved
1088 struct list_head i_rsv_conversion_list;
1089 struct work_struct i_rsv_conversion_work;
1090 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1092 spinlock_t i_block_reservation_lock;
1095 * Transactions that contain inode's metadata needed to complete
1096 * fsync and fdatasync, respectively.
1099 tid_t i_datasync_tid;
1102 struct dquot *i_dquot[MAXQUOTAS];
1105 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1112 * File system states
1114 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
1115 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
1116 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
1119 * Misc. filesystem flags
1121 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
1122 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
1123 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
1126 * Mount flags set via mount options or defaults
1128 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */
1129 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
1130 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
1131 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
1132 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
1133 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
1134 #define EXT4_MOUNT_ERRORS_MASK 0x00070
1135 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
1136 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
1137 #ifdef CONFIG_FS_DAX
1138 #define EXT4_MOUNT_DAX 0x00200 /* Direct Access */
1140 #define EXT4_MOUNT_DAX 0
1142 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
1143 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
1144 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
1145 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
1146 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
1147 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
1148 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
1149 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
1150 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
1151 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
1152 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */
1153 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota,
1154 * enable enforcement for hidden
1156 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable
1157 * enforcement for hidden quota
1159 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota
1161 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
1162 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
1163 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1164 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */
1165 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
1166 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
1167 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
1168 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
1169 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
1172 * Mount flags set either automatically (could not be set by mount option)
1173 * based on per file system feature or property or in special cases such as
1174 * distinguishing between explicit mount option definition and default.
1176 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1177 specified delalloc */
1178 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1179 size of blocksize * 8
1181 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1184 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly
1185 specified journal checksum */
1187 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1189 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1191 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1194 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1196 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1198 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1201 #define ext4_test_and_set_bit __test_and_set_bit_le
1202 #define ext4_set_bit __set_bit_le
1203 #define ext4_set_bit_atomic ext2_set_bit_atomic
1204 #define ext4_test_and_clear_bit __test_and_clear_bit_le
1205 #define ext4_clear_bit __clear_bit_le
1206 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
1207 #define ext4_test_bit test_bit_le
1208 #define ext4_find_next_zero_bit find_next_zero_bit_le
1209 #define ext4_find_next_bit find_next_bit_le
1211 extern void ext4_set_bits(void *bm, int cur, int len);
1214 * Maximal mount counts between two filesystem checks
1216 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1217 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1220 * Behaviour when detecting errors
1222 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1223 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1224 #define EXT4_ERRORS_PANIC 3 /* Panic */
1225 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1227 /* Metadata checksum algorithm codes */
1228 #define EXT4_CRC32C_CHKSUM 1
1231 * Structure of the super block
1233 struct ext4_super_block {
1234 /*00*/ __le32 s_inodes_count; /* Inodes count */
1235 __le32 s_blocks_count_lo; /* Blocks count */
1236 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1237 __le32 s_free_blocks_count_lo; /* Free blocks count */
1238 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1239 __le32 s_first_data_block; /* First Data Block */
1240 __le32 s_log_block_size; /* Block size */
1241 __le32 s_log_cluster_size; /* Allocation cluster size */
1242 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1243 __le32 s_clusters_per_group; /* # Clusters per group */
1244 __le32 s_inodes_per_group; /* # Inodes per group */
1245 __le32 s_mtime; /* Mount time */
1246 /*30*/ __le32 s_wtime; /* Write time */
1247 __le16 s_mnt_count; /* Mount count */
1248 __le16 s_max_mnt_count; /* Maximal mount count */
1249 __le16 s_magic; /* Magic signature */
1250 __le16 s_state; /* File system state */
1251 __le16 s_errors; /* Behaviour when detecting errors */
1252 __le16 s_minor_rev_level; /* minor revision level */
1253 /*40*/ __le32 s_lastcheck; /* time of last check */
1254 __le32 s_checkinterval; /* max. time between checks */
1255 __le32 s_creator_os; /* OS */
1256 __le32 s_rev_level; /* Revision level */
1257 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1258 __le16 s_def_resgid; /* Default gid for reserved blocks */
1260 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1262 * Note: the difference between the compatible feature set and
1263 * the incompatible feature set is that if there is a bit set
1264 * in the incompatible feature set that the kernel doesn't
1265 * know about, it should refuse to mount the filesystem.
1267 * e2fsck's requirements are more strict; if it doesn't know
1268 * about a feature in either the compatible or incompatible
1269 * feature set, it must abort and not try to meddle with
1270 * things it doesn't understand...
1272 __le32 s_first_ino; /* First non-reserved inode */
1273 __le16 s_inode_size; /* size of inode structure */
1274 __le16 s_block_group_nr; /* block group # of this superblock */
1275 __le32 s_feature_compat; /* compatible feature set */
1276 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1277 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1278 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1279 /*78*/ char s_volume_name[16]; /* volume name */
1280 /*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */
1281 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1283 * Performance hints. Directory preallocation should only
1284 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1286 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1287 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1288 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1290 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1292 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1293 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1294 __le32 s_journal_dev; /* device number of journal file */
1295 __le32 s_last_orphan; /* start of list of inodes to delete */
1296 __le32 s_hash_seed[4]; /* HTREE hash seed */
1297 __u8 s_def_hash_version; /* Default hash version to use */
1298 __u8 s_jnl_backup_type;
1299 __le16 s_desc_size; /* size of group descriptor */
1300 /*100*/ __le32 s_default_mount_opts;
1301 __le32 s_first_meta_bg; /* First metablock block group */
1302 __le32 s_mkfs_time; /* When the filesystem was created */
1303 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1304 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1305 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1306 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1307 __le32 s_free_blocks_count_hi; /* Free blocks count */
1308 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1309 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1310 __le32 s_flags; /* Miscellaneous flags */
1311 __le16 s_raid_stride; /* RAID stride */
1312 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1313 __le64 s_mmp_block; /* Block for multi-mount protection */
1314 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1315 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1316 __u8 s_checksum_type; /* metadata checksum algorithm used */
1317 __u8 s_encryption_level; /* versioning level for encryption */
1318 __u8 s_reserved_pad; /* Padding to next 32bits */
1319 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1320 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1321 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1322 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1323 snapshot's future use */
1324 __le32 s_snapshot_list; /* inode number of the head of the
1325 on-disk snapshot list */
1326 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1327 __le32 s_error_count; /* number of fs errors */
1328 __le32 s_first_error_time; /* first time an error happened */
1329 __le32 s_first_error_ino; /* inode involved in first error */
1330 __le64 s_first_error_block; /* block involved of first error */
1331 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */
1332 __le32 s_first_error_line; /* line number where error happened */
1333 __le32 s_last_error_time; /* most recent time of an error */
1334 __le32 s_last_error_ino; /* inode involved in last error */
1335 __le32 s_last_error_line; /* line number where error happened */
1336 __le64 s_last_error_block; /* block involved of last error */
1337 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */
1338 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1339 __u8 s_mount_opts[64];
1340 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1341 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1342 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1343 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
1344 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */
1345 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */
1346 __le32 s_lpf_ino; /* Location of the lost+found inode */
1347 __le32 s_prj_quota_inum; /* inode for tracking project quota */
1348 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */
1351 __u8 s_mkfs_time_hi;
1352 __u8 s_lastcheck_hi;
1353 __u8 s_first_error_time_hi;
1354 __u8 s_last_error_time_hi;
1355 __u8 s_first_error_errcode;
1356 __u8 s_last_error_errcode;
1357 __le16 s_encoding; /* Filename charset encoding */
1358 __le16 s_encoding_flags; /* Filename charset encoding flags */
1359 __le32 s_reserved[95]; /* Padding to the end of the block */
1360 __le32 s_checksum; /* crc32c(superblock) */
1363 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1368 * run-time mount flags
1370 #define EXT4_MF_MNTDIR_SAMPLED 0x0001
1371 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1373 #ifdef CONFIG_FS_ENCRYPTION
1374 #define DUMMY_ENCRYPTION_ENABLED(sbi) ((sbi)->s_dummy_enc_ctx.ctx != NULL)
1376 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1379 /* Number of quota types we support */
1380 #define EXT4_MAXQUOTAS 3
1382 #define EXT4_ENC_UTF8_12_1 1
1385 * Flags for ext4_sb_info.s_encoding_flags.
1387 #define EXT4_ENC_STRICT_MODE_FL (1 << 0)
1389 #define ext4_has_strict_mode(sbi) \
1390 (sbi->s_encoding_flags & EXT4_ENC_STRICT_MODE_FL)
1393 * fourth extended-fs super-block data in memory
1395 struct ext4_sb_info {
1396 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1397 unsigned long s_inodes_per_block;/* Number of inodes per block */
1398 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1399 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1400 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1401 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1402 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1403 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1404 ext4_group_t s_groups_count; /* Number of groups in the fs */
1405 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1406 unsigned long s_overhead; /* # of fs overhead clusters */
1407 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1408 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1409 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1410 struct buffer_head * s_sbh; /* Buffer containing the super block */
1411 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1412 struct buffer_head * __rcu *s_group_desc;
1413 unsigned int s_mount_opt;
1414 unsigned int s_mount_opt2;
1415 unsigned int s_mount_flags;
1416 unsigned int s_def_mount_opt;
1417 ext4_fsblk_t s_sb_block;
1418 atomic64_t s_resv_clusters;
1421 unsigned short s_mount_state;
1422 unsigned short s_pad;
1423 int s_addr_per_block_bits;
1424 int s_desc_per_block_bits;
1427 unsigned int s_inode_readahead_blks;
1428 unsigned int s_inode_goal;
1430 int s_def_hash_version;
1431 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1432 struct percpu_counter s_freeclusters_counter;
1433 struct percpu_counter s_freeinodes_counter;
1434 struct percpu_counter s_dirs_counter;
1435 struct percpu_counter s_dirtyclusters_counter;
1436 struct blockgroup_lock *s_blockgroup_lock;
1437 struct proc_dir_entry *s_proc;
1438 struct kobject s_kobj;
1439 struct completion s_kobj_unregister;
1440 struct super_block *s_sb;
1441 #ifdef CONFIG_UNICODE
1442 struct unicode_map *s_encoding;
1443 __u16 s_encoding_flags;
1447 struct journal_s *s_journal;
1448 struct list_head s_orphan;
1449 struct mutex s_orphan_lock;
1450 unsigned long s_ext4_flags; /* Ext4 superblock flags */
1451 unsigned long s_commit_interval;
1452 u32 s_max_batch_time;
1453 u32 s_min_batch_time;
1454 struct block_device *journal_bdev;
1456 /* Names of quota files with journalled quota */
1457 char __rcu *s_qf_names[EXT4_MAXQUOTAS];
1458 int s_jquota_fmt; /* Format of quota to use */
1460 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1461 struct ext4_system_blocks __rcu *system_blks;
1463 #ifdef EXTENTS_STATS
1464 /* ext4 extents stats */
1465 unsigned long s_ext_min;
1466 unsigned long s_ext_max;
1467 unsigned long s_depth_max;
1468 spinlock_t s_ext_stats_lock;
1469 unsigned long s_ext_blocks;
1470 unsigned long s_ext_extents;
1473 /* for buddy allocator */
1474 struct ext4_group_info ** __rcu *s_group_info;
1475 struct inode *s_buddy_cache;
1476 spinlock_t s_md_lock;
1477 unsigned short *s_mb_offsets;
1478 unsigned int *s_mb_maxs;
1479 unsigned int s_group_info_size;
1480 unsigned int s_mb_free_pending;
1481 struct list_head s_freed_data_list; /* List of blocks to be freed
1482 after commit completed */
1485 unsigned long s_stripe;
1486 unsigned int s_mb_stream_request;
1487 unsigned int s_mb_max_to_scan;
1488 unsigned int s_mb_min_to_scan;
1489 unsigned int s_mb_stats;
1490 unsigned int s_mb_order2_reqs;
1491 unsigned int s_mb_group_prealloc;
1492 unsigned int s_max_dir_size_kb;
1493 /* where last allocation was done - for stream allocation */
1494 unsigned long s_mb_last_group;
1495 unsigned long s_mb_last_start;
1497 /* stats for buddy allocator */
1498 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1499 atomic_t s_bal_success; /* we found long enough chunks */
1500 atomic_t s_bal_allocated; /* in blocks */
1501 atomic_t s_bal_ex_scanned; /* total extents scanned */
1502 atomic_t s_bal_goals; /* goal hits */
1503 atomic_t s_bal_breaks; /* too long searches */
1504 atomic_t s_bal_2orders; /* 2^order hits */
1505 spinlock_t s_bal_lock;
1506 unsigned long s_mb_buddies_generated;
1507 unsigned long long s_mb_generation_time;
1508 atomic_t s_mb_lost_chunks;
1509 atomic_t s_mb_preallocated;
1510 atomic_t s_mb_discarded;
1511 atomic_t s_lock_busy;
1513 /* locality groups */
1514 struct ext4_locality_group __percpu *s_locality_groups;
1516 /* for write statistics */
1517 unsigned long s_sectors_written_start;
1518 u64 s_kbytes_written;
1520 /* the size of zero-out chunk */
1521 unsigned int s_extent_max_zeroout_kb;
1523 unsigned int s_log_groups_per_flex;
1524 struct flex_groups * __rcu *s_flex_groups;
1525 ext4_group_t s_flex_groups_allocated;
1527 /* workqueue for reserved extent conversions (buffered io) */
1528 struct workqueue_struct *rsv_conversion_wq;
1530 /* timer for periodic error stats printing */
1531 struct timer_list s_err_report;
1533 /* Lazy inode table initialization info */
1534 struct ext4_li_request *s_li_request;
1535 /* Wait multiplier for lazy initialization thread */
1536 unsigned int s_li_wait_mult;
1538 /* Kernel thread for multiple mount protection */
1539 struct task_struct *s_mmp_tsk;
1541 /* record the last minlen when FITRIM is called. */
1542 atomic_t s_last_trim_minblks;
1544 /* Reference to checksum algorithm driver via cryptoapi */
1545 struct crypto_shash *s_chksum_driver;
1547 /* Precomputed FS UUID checksum for seeding other checksums */
1550 /* Reclaim extents from extent status tree */
1551 struct shrinker s_es_shrinker;
1552 struct list_head s_es_list; /* List of inodes with reclaimable extents */
1554 struct ext4_es_stats s_es_stats;
1555 struct mb_cache *s_ea_block_cache;
1556 struct mb_cache *s_ea_inode_cache;
1557 spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1559 /* Ratelimit ext4 messages. */
1560 struct ratelimit_state s_err_ratelimit_state;
1561 struct ratelimit_state s_warning_ratelimit_state;
1562 struct ratelimit_state s_msg_ratelimit_state;
1564 /* Encryption context for '-o test_dummy_encryption' */
1565 struct fscrypt_dummy_context s_dummy_enc_ctx;
1568 * Barrier between writepages ops and changing any inode's JOURNAL_DATA
1571 struct percpu_rw_semaphore s_writepages_rwsem;
1572 struct dax_device *s_daxdev;
1573 #ifdef CONFIG_EXT4_DEBUG
1574 unsigned long s_simulate_fail;
1578 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1580 return sb->s_fs_info;
1582 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1584 return container_of(inode, struct ext4_inode_info, vfs_inode);
1587 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1589 return ino == EXT4_ROOT_INO ||
1590 (ino >= EXT4_FIRST_INO(sb) &&
1591 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1595 * Returns: sbi->field[index]
1596 * Used to access an array element from the following sbi fields which require
1597 * rcu protection to avoid dereferencing an invalid pointer due to reassignment
1602 #define sbi_array_rcu_deref(sbi, field, index) \
1604 typeof(*((sbi)->field)) _v; \
1606 _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index]; \
1607 rcu_read_unlock(); \
1612 * Simulate_fail codes
1614 #define EXT4_SIM_BBITMAP_EIO 1
1615 #define EXT4_SIM_BBITMAP_CRC 2
1616 #define EXT4_SIM_IBITMAP_EIO 3
1617 #define EXT4_SIM_IBITMAP_CRC 4
1618 #define EXT4_SIM_INODE_EIO 5
1619 #define EXT4_SIM_INODE_CRC 6
1620 #define EXT4_SIM_DIRBLOCK_EIO 7
1621 #define EXT4_SIM_DIRBLOCK_CRC 8
1623 static inline bool ext4_simulate_fail(struct super_block *sb,
1626 #ifdef CONFIG_EXT4_DEBUG
1627 struct ext4_sb_info *sbi = EXT4_SB(sb);
1629 if (unlikely(sbi->s_simulate_fail == code)) {
1630 sbi->s_simulate_fail = 0;
1637 static inline void ext4_simulate_fail_bh(struct super_block *sb,
1638 struct buffer_head *bh,
1641 if (!IS_ERR(bh) && ext4_simulate_fail(sb, code))
1642 clear_buffer_uptodate(bh);
1646 * Error number codes for s_{first,last}_error_errno
1648 * Linux errno numbers are architecture specific, so we need to translate
1649 * them into something which is architecture independent. We don't define
1650 * codes for all errno's; just the ones which are most likely to be the cause
1651 * of an ext4_error() call.
1653 #define EXT4_ERR_UNKNOWN 1
1654 #define EXT4_ERR_EIO 2
1655 #define EXT4_ERR_ENOMEM 3
1656 #define EXT4_ERR_EFSBADCRC 4
1657 #define EXT4_ERR_EFSCORRUPTED 5
1658 #define EXT4_ERR_ENOSPC 6
1659 #define EXT4_ERR_ENOKEY 7
1660 #define EXT4_ERR_EROFS 8
1661 #define EXT4_ERR_EFBIG 9
1662 #define EXT4_ERR_EEXIST 10
1663 #define EXT4_ERR_ERANGE 11
1664 #define EXT4_ERR_EOVERFLOW 12
1665 #define EXT4_ERR_EBUSY 13
1666 #define EXT4_ERR_ENOTDIR 14
1667 #define EXT4_ERR_ENOTEMPTY 15
1668 #define EXT4_ERR_ESHUTDOWN 16
1669 #define EXT4_ERR_EFAULT 17
1672 * Inode dynamic state flags
1675 EXT4_STATE_JDATA, /* journaled data exists */
1676 EXT4_STATE_NEW, /* inode is newly created */
1677 EXT4_STATE_XATTR, /* has in-inode xattrs */
1678 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1679 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1680 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1681 EXT4_STATE_NEWENTRY, /* File just added to dir */
1682 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1683 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
1684 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */
1685 EXT4_STATE_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */
1688 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1689 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1691 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1693 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1695 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1697 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1699 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1702 /* Add these declarations here only so that these functions can be
1703 * found by name. Otherwise, they are very hard to locate. */
1704 static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1705 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1706 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1707 EXT4_INODE_BIT_FNS(flag, flags, 0)
1709 /* Add these declarations here only so that these functions can be
1710 * found by name. Otherwise, they are very hard to locate. */
1711 static inline int ext4_test_inode_state(struct inode *inode, int bit);
1712 static inline void ext4_set_inode_state(struct inode *inode, int bit);
1713 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1714 #if (BITS_PER_LONG < 64)
1715 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1717 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1719 (ei)->i_state_flags = 0;
1722 EXT4_INODE_BIT_FNS(state, flags, 32)
1724 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1726 /* We depend on the fact that callers will set i_flags */
1730 /* Assume that user mode programs are passing in an ext4fs superblock, not
1731 * a kernel struct super_block. This will allow us to call the feature-test
1732 * macros from user land. */
1733 #define EXT4_SB(sb) (sb)
1736 static inline bool ext4_verity_in_progress(struct inode *inode)
1738 return IS_ENABLED(CONFIG_FS_VERITY) &&
1739 ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS);
1742 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1745 * Codes for operating systems
1747 #define EXT4_OS_LINUX 0
1748 #define EXT4_OS_HURD 1
1749 #define EXT4_OS_MASIX 2
1750 #define EXT4_OS_FREEBSD 3
1751 #define EXT4_OS_LITES 4
1756 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1757 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1759 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1760 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1762 #define EXT4_GOOD_OLD_INODE_SIZE 128
1764 #define EXT4_EXTRA_TIMESTAMP_MAX (((s64)1 << 34) - 1 + S32_MIN)
1765 #define EXT4_NON_EXTRA_TIMESTAMP_MAX S32_MAX
1766 #define EXT4_TIMESTAMP_MIN S32_MIN
1769 * Feature set definitions
1772 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1773 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1774 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1775 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1776 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1777 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1778 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
1779 #define EXT4_FEATURE_COMPAT_STABLE_INODES 0x0800
1781 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1782 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1783 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1784 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1785 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1786 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1787 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1788 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1789 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1791 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1792 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1793 * all other data structures' checksums. However, the METADATA_CSUM and
1794 * GDT_CSUM bits are mutually exclusive.
1796 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
1797 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000
1798 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000
1799 #define EXT4_FEATURE_RO_COMPAT_VERITY 0x8000
1801 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1802 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1803 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1804 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1805 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1806 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1807 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1808 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1809 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1810 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1811 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1812 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000
1813 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
1814 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
1815 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000
1816 #define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000
1818 extern void ext4_update_dynamic_rev(struct super_block *sb);
1820 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
1821 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1823 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1824 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
1826 static inline void ext4_set_feature_##name(struct super_block *sb) \
1828 ext4_update_dynamic_rev(sb); \
1829 EXT4_SB(sb)->s_es->s_feature_compat |= \
1830 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1832 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1834 EXT4_SB(sb)->s_es->s_feature_compat &= \
1835 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1838 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
1839 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1841 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1842 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
1844 static inline void ext4_set_feature_##name(struct super_block *sb) \
1846 ext4_update_dynamic_rev(sb); \
1847 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
1848 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1850 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1852 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
1853 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1856 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
1857 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1859 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1860 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
1862 static inline void ext4_set_feature_##name(struct super_block *sb) \
1864 ext4_update_dynamic_rev(sb); \
1865 EXT4_SB(sb)->s_es->s_feature_incompat |= \
1866 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1868 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1870 EXT4_SB(sb)->s_es->s_feature_incompat &= \
1871 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1874 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC)
1875 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES)
1876 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL)
1877 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR)
1878 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE)
1879 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX)
1880 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2)
1881 EXT4_FEATURE_COMPAT_FUNCS(stable_inodes, STABLE_INODES)
1883 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER)
1884 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE)
1885 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR)
1886 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE)
1887 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM)
1888 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK)
1889 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE)
1890 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA)
1891 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC)
1892 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM)
1893 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY)
1894 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT)
1895 EXT4_FEATURE_RO_COMPAT_FUNCS(verity, VERITY)
1897 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION)
1898 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE)
1899 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER)
1900 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV)
1901 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG)
1902 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS)
1903 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT)
1904 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP)
1905 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG)
1906 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE)
1907 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA)
1908 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED)
1909 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR)
1910 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA)
1911 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT)
1912 EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD)
1914 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1915 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1916 EXT4_FEATURE_INCOMPAT_META_BG)
1917 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1918 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1919 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1921 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1922 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1923 EXT4_FEATURE_INCOMPAT_RECOVER| \
1924 EXT4_FEATURE_INCOMPAT_META_BG)
1925 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1926 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1927 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1929 #define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1930 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1931 EXT4_FEATURE_INCOMPAT_RECOVER| \
1932 EXT4_FEATURE_INCOMPAT_META_BG| \
1933 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1934 EXT4_FEATURE_INCOMPAT_64BIT| \
1935 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1936 EXT4_FEATURE_INCOMPAT_EA_INODE| \
1937 EXT4_FEATURE_INCOMPAT_MMP | \
1938 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
1939 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
1940 EXT4_FEATURE_INCOMPAT_CASEFOLD | \
1941 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \
1942 EXT4_FEATURE_INCOMPAT_LARGEDIR)
1943 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1944 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1945 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1946 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1947 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1948 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1949 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1950 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1951 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1952 EXT4_FEATURE_RO_COMPAT_QUOTA |\
1953 EXT4_FEATURE_RO_COMPAT_PROJECT |\
1954 EXT4_FEATURE_RO_COMPAT_VERITY)
1956 #define EXTN_FEATURE_FUNCS(ver) \
1957 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
1959 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1960 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
1962 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
1964 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1965 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
1967 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
1969 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1970 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
1973 EXTN_FEATURE_FUNCS(2)
1974 EXTN_FEATURE_FUNCS(3)
1975 EXTN_FEATURE_FUNCS(4)
1977 static inline bool ext4_has_compat_features(struct super_block *sb)
1979 return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
1981 static inline bool ext4_has_ro_compat_features(struct super_block *sb)
1983 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
1985 static inline bool ext4_has_incompat_features(struct super_block *sb)
1987 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
1993 #define EXT4_FLAGS_RESIZING 0
1994 #define EXT4_FLAGS_SHUTDOWN 1
1996 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi)
1998 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags);
2003 * Default values for user and/or group using reserved blocks
2005 #define EXT4_DEF_RESUID 0
2006 #define EXT4_DEF_RESGID 0
2009 * Default project ID
2011 #define EXT4_DEF_PROJID 0
2013 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
2016 * Default mount options
2018 #define EXT4_DEFM_DEBUG 0x0001
2019 #define EXT4_DEFM_BSDGROUPS 0x0002
2020 #define EXT4_DEFM_XATTR_USER 0x0004
2021 #define EXT4_DEFM_ACL 0x0008
2022 #define EXT4_DEFM_UID16 0x0010
2023 #define EXT4_DEFM_JMODE 0x0060
2024 #define EXT4_DEFM_JMODE_DATA 0x0020
2025 #define EXT4_DEFM_JMODE_ORDERED 0x0040
2026 #define EXT4_DEFM_JMODE_WBACK 0x0060
2027 #define EXT4_DEFM_NOBARRIER 0x0100
2028 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
2029 #define EXT4_DEFM_DISCARD 0x0400
2030 #define EXT4_DEFM_NODELALLOC 0x0800
2033 * Default journal batch times
2035 #define EXT4_DEF_MIN_BATCH_TIME 0
2036 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
2039 * Minimum number of groups in a flexgroup before we separate out
2040 * directories into the first block group of a flexgroup
2042 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
2045 * Structure of a directory entry
2047 #define EXT4_NAME_LEN 255
2049 struct ext4_dir_entry {
2050 __le32 inode; /* Inode number */
2051 __le16 rec_len; /* Directory entry length */
2052 __le16 name_len; /* Name length */
2053 char name[EXT4_NAME_LEN]; /* File name */
2057 * The new version of the directory entry. Since EXT4 structures are
2058 * stored in intel byte order, and the name_len field could never be
2059 * bigger than 255 chars, it's safe to reclaim the extra byte for the
2062 struct ext4_dir_entry_2 {
2063 __le32 inode; /* Inode number */
2064 __le16 rec_len; /* Directory entry length */
2065 __u8 name_len; /* Name length */
2066 __u8 file_type; /* See file type macros EXT4_FT_* below */
2067 char name[EXT4_NAME_LEN]; /* File name */
2071 * This is a bogus directory entry at the end of each leaf block that
2072 * records checksums.
2074 struct ext4_dir_entry_tail {
2075 __le32 det_reserved_zero1; /* Pretend to be unused */
2076 __le16 det_rec_len; /* 12 */
2077 __u8 det_reserved_zero2; /* Zero name length */
2078 __u8 det_reserved_ft; /* 0xDE, fake file type */
2079 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
2082 #define EXT4_DIRENT_TAIL(block, blocksize) \
2083 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
2085 sizeof(struct ext4_dir_entry_tail))))
2088 * Ext4 directory file types. Only the low 3 bits are used. The
2089 * other bits are reserved for now.
2091 #define EXT4_FT_UNKNOWN 0
2092 #define EXT4_FT_REG_FILE 1
2093 #define EXT4_FT_DIR 2
2094 #define EXT4_FT_CHRDEV 3
2095 #define EXT4_FT_BLKDEV 4
2096 #define EXT4_FT_FIFO 5
2097 #define EXT4_FT_SOCK 6
2098 #define EXT4_FT_SYMLINK 7
2100 #define EXT4_FT_MAX 8
2102 #define EXT4_FT_DIR_CSUM 0xDE
2105 * EXT4_DIR_PAD defines the directory entries boundaries
2107 * NOTE: It must be a multiple of 4
2109 #define EXT4_DIR_PAD 4
2110 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
2111 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
2113 #define EXT4_MAX_REC_LEN ((1<<16)-1)
2116 * If we ever get support for fs block sizes > page_size, we'll need
2117 * to remove the #if statements in the next two functions...
2119 static inline unsigned int
2120 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
2122 unsigned len = le16_to_cpu(dlen);
2124 #if (PAGE_SIZE >= 65536)
2125 if (len == EXT4_MAX_REC_LEN || len == 0)
2127 return (len & 65532) | ((len & 3) << 16);
2133 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
2135 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
2137 #if (PAGE_SIZE >= 65536)
2139 return cpu_to_le16(len);
2140 if (len == blocksize) {
2141 if (blocksize == 65536)
2142 return cpu_to_le16(EXT4_MAX_REC_LEN);
2144 return cpu_to_le16(0);
2146 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
2148 return cpu_to_le16(len);
2153 * Hash Tree Directory indexing
2154 * (c) Daniel Phillips, 2001
2157 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
2158 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2159 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \
2160 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir)))
2161 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2163 /* Legal values for the dx_root hash_version field: */
2165 #define DX_HASH_LEGACY 0
2166 #define DX_HASH_HALF_MD4 1
2167 #define DX_HASH_TEA 2
2168 #define DX_HASH_LEGACY_UNSIGNED 3
2169 #define DX_HASH_HALF_MD4_UNSIGNED 4
2170 #define DX_HASH_TEA_UNSIGNED 5
2172 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2173 const void *address, unsigned int length)
2176 struct shash_desc shash;
2180 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2182 desc.shash.tfm = sbi->s_chksum_driver;
2183 *(u32 *)desc.ctx = crc;
2185 BUG_ON(crypto_shash_update(&desc.shash, address, length));
2187 return *(u32 *)desc.ctx;
2192 /* hash info structure used by the directory hash */
2202 /* 32 and 64 bit signed EOF for dx directories */
2203 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
2204 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
2208 * Control parameters used by ext4_htree_next_block
2210 #define HASH_NB_ALWAYS 1
2212 struct ext4_filename {
2213 const struct qstr *usr_fname;
2214 struct fscrypt_str disk_name;
2215 struct dx_hash_info hinfo;
2216 #ifdef CONFIG_FS_ENCRYPTION
2217 struct fscrypt_str crypto_buf;
2219 #ifdef CONFIG_UNICODE
2220 struct fscrypt_str cf_name;
2224 #define fname_name(p) ((p)->disk_name.name)
2225 #define fname_len(p) ((p)->disk_name.len)
2228 * Describe an inode's exact location on disk and in memory
2232 struct buffer_head *bh;
2233 unsigned long offset;
2234 ext4_group_t block_group;
2237 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2239 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2242 static inline bool ext4_is_quota_file(struct inode *inode)
2244 return IS_NOQUOTA(inode) &&
2245 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL);
2249 * This structure is stuffed into the struct file's private_data field
2250 * for directories. It is where we put information so that we can do
2251 * readdir operations in hash tree order.
2253 struct dir_private_info {
2254 struct rb_root root;
2255 struct rb_node *curr_node;
2256 struct fname *extra_fname;
2259 __u32 curr_minor_hash;
2263 /* calculate the first block number of the group */
2264 static inline ext4_fsblk_t
2265 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2267 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2268 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2272 * Special error return code only used by dx_probe() and its callers.
2274 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
2276 /* htree levels for ext4 */
2277 #define EXT4_HTREE_LEVEL_COMPAT 2
2278 #define EXT4_HTREE_LEVEL 3
2280 static inline int ext4_dir_htree_level(struct super_block *sb)
2282 return ext4_has_feature_largedir(sb) ?
2283 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT;
2287 * Timeout and state flag for lazy initialization inode thread.
2289 #define EXT4_DEF_LI_WAIT_MULT 10
2290 #define EXT4_DEF_LI_MAX_START_DELAY 5
2291 #define EXT4_LAZYINIT_QUIT 0x0001
2292 #define EXT4_LAZYINIT_RUNNING 0x0002
2295 * Lazy inode table initialization info
2297 struct ext4_lazy_init {
2298 unsigned long li_state;
2299 struct list_head li_request_list;
2300 struct mutex li_list_mtx;
2303 struct ext4_li_request {
2304 struct super_block *lr_super;
2305 struct ext4_sb_info *lr_sbi;
2306 ext4_group_t lr_next_group;
2307 struct list_head lr_request;
2308 unsigned long lr_next_sched;
2309 unsigned long lr_timeout;
2312 struct ext4_features {
2313 struct kobject f_kobj;
2314 struct completion f_kobj_unregister;
2318 * This structure will be used for multiple mount protection. It will be
2319 * written into the block number saved in the s_mmp_block field in the
2320 * superblock. Programs that check MMP should assume that if
2321 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2322 * to use the filesystem, regardless of how old the timestamp is.
2324 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
2325 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2326 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
2327 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
2330 __le32 mmp_magic; /* Magic number for MMP */
2331 __le32 mmp_seq; /* Sequence no. updated periodically */
2334 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2335 * purposes and do not affect the correctness of the algorithm
2337 __le64 mmp_time; /* Time last updated */
2338 char mmp_nodename[64]; /* Node which last updated MMP block */
2339 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
2342 * mmp_check_interval is used to verify if the MMP block has been
2343 * updated on the block device. The value is updated based on the
2344 * maximum time to write the MMP block during an update cycle.
2346 __le16 mmp_check_interval;
2349 __le32 mmp_pad2[226];
2350 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
2353 /* arguments passed to the mmp thread */
2355 struct buffer_head *bh; /* bh from initial read_mmp_block() */
2356 struct super_block *sb; /* super block of the fs */
2360 * Check interval multiplier
2361 * The MMP block is written every update interval and initially checked every
2362 * update interval x the multiplier (the value is then adapted based on the
2363 * write latency). The reason is that writes can be delayed under load and we
2364 * don't want readers to incorrectly assume that the filesystem is no longer
2367 #define EXT4_MMP_CHECK_MULT 2UL
2370 * Minimum interval for MMP checking in seconds.
2372 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
2375 * Maximum interval for MMP checking in seconds.
2377 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
2380 * Function prototypes
2384 * Ok, these declarations are also in <linux/kernel.h> but none of the
2385 * ext4 source programs needs to include it so they are duplicated here.
2387 # define NORET_TYPE /**/
2388 # define ATTRIB_NORET __attribute__((noreturn))
2389 # define NORET_AND noreturn,
2392 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2393 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2394 struct ext4_group_desc *gdp,
2395 struct buffer_head *bh, int sz);
2396 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2397 struct ext4_group_desc *gdp,
2398 struct buffer_head *bh, int sz);
2399 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2400 struct ext4_group_desc *gdp,
2401 struct buffer_head *bh);
2402 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2403 struct ext4_group_desc *gdp,
2404 struct buffer_head *bh);
2407 extern void ext4_get_group_no_and_offset(struct super_block *sb,
2408 ext4_fsblk_t blocknr,
2409 ext4_group_t *blockgrpp,
2410 ext4_grpblk_t *offsetp);
2411 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2412 ext4_fsblk_t block);
2414 extern unsigned int ext4_block_group(struct super_block *sb,
2415 ext4_fsblk_t blocknr);
2416 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2417 ext4_fsblk_t blocknr);
2418 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2419 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2420 ext4_group_t group);
2421 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2424 unsigned long *count,
2426 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2427 s64 nclusters, unsigned int flags);
2428 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2429 extern void ext4_check_blocks_bitmap(struct super_block *);
2430 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2431 ext4_group_t block_group,
2432 struct buffer_head ** bh);
2433 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2435 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2436 ext4_group_t block_group);
2437 extern int ext4_wait_block_bitmap(struct super_block *sb,
2438 ext4_group_t block_group,
2439 struct buffer_head *bh);
2440 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2441 ext4_group_t block_group);
2442 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2443 ext4_group_t block_group,
2444 struct ext4_group_desc *gdp);
2445 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2447 #ifdef CONFIG_UNICODE
2448 extern void ext4_fname_setup_ci_filename(struct inode *dir,
2449 const struct qstr *iname,
2450 struct fscrypt_str *fname);
2453 #ifdef CONFIG_FS_ENCRYPTION
2454 static inline void ext4_fname_from_fscrypt_name(struct ext4_filename *dst,
2455 const struct fscrypt_name *src)
2457 memset(dst, 0, sizeof(*dst));
2459 dst->usr_fname = src->usr_fname;
2460 dst->disk_name = src->disk_name;
2461 dst->hinfo.hash = src->hash;
2462 dst->hinfo.minor_hash = src->minor_hash;
2463 dst->crypto_buf = src->crypto_buf;
2466 static inline int ext4_fname_setup_filename(struct inode *dir,
2467 const struct qstr *iname,
2469 struct ext4_filename *fname)
2471 struct fscrypt_name name;
2474 err = fscrypt_setup_filename(dir, iname, lookup, &name);
2478 ext4_fname_from_fscrypt_name(fname, &name);
2480 #ifdef CONFIG_UNICODE
2481 ext4_fname_setup_ci_filename(dir, iname, &fname->cf_name);
2486 static inline int ext4_fname_prepare_lookup(struct inode *dir,
2487 struct dentry *dentry,
2488 struct ext4_filename *fname)
2490 struct fscrypt_name name;
2493 err = fscrypt_prepare_lookup(dir, dentry, &name);
2497 ext4_fname_from_fscrypt_name(fname, &name);
2499 #ifdef CONFIG_UNICODE
2500 ext4_fname_setup_ci_filename(dir, &dentry->d_name, &fname->cf_name);
2505 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2507 struct fscrypt_name name;
2509 name.crypto_buf = fname->crypto_buf;
2510 fscrypt_free_filename(&name);
2512 fname->crypto_buf.name = NULL;
2513 fname->usr_fname = NULL;
2514 fname->disk_name.name = NULL;
2516 #ifdef CONFIG_UNICODE
2517 kfree(fname->cf_name.name);
2518 fname->cf_name.name = NULL;
2521 #else /* !CONFIG_FS_ENCRYPTION */
2522 static inline int ext4_fname_setup_filename(struct inode *dir,
2523 const struct qstr *iname,
2525 struct ext4_filename *fname)
2527 fname->usr_fname = iname;
2528 fname->disk_name.name = (unsigned char *) iname->name;
2529 fname->disk_name.len = iname->len;
2531 #ifdef CONFIG_UNICODE
2532 ext4_fname_setup_ci_filename(dir, iname, &fname->cf_name);
2538 static inline int ext4_fname_prepare_lookup(struct inode *dir,
2539 struct dentry *dentry,
2540 struct ext4_filename *fname)
2542 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname);
2545 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2547 #ifdef CONFIG_UNICODE
2548 kfree(fname->cf_name.name);
2549 fname->cf_name.name = NULL;
2552 #endif /* !CONFIG_FS_ENCRYPTION */
2555 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2557 struct ext4_dir_entry_2 *,
2558 struct buffer_head *, char *, int,
2560 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2561 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2562 (de), (bh), (buf), (size), (offset)))
2563 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2565 struct ext4_dir_entry_2 *dirent,
2566 struct fscrypt_str *ent_name);
2567 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2568 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2569 struct buffer_head *bh,
2570 void *buf, int buf_size,
2571 struct ext4_filename *fname,
2572 struct ext4_dir_entry_2 **dest_de);
2573 void ext4_insert_dentry(struct inode *inode,
2574 struct ext4_dir_entry_2 *de,
2576 struct ext4_filename *fname);
2577 static inline void ext4_update_dx_flag(struct inode *inode)
2579 if (!ext4_has_feature_dir_index(inode->i_sb)) {
2580 /* ext4_iget() should have caught this... */
2581 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb));
2582 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2585 static const unsigned char ext4_filetype_table[] = {
2586 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2589 static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2591 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2594 return ext4_filetype_table[filetype];
2596 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2597 void *buf, int buf_size);
2600 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2603 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len,
2604 struct dx_hash_info *hinfo);
2607 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2608 const struct qstr *qstr, __u32 goal,
2609 uid_t *owner, __u32 i_flags,
2610 int handle_type, unsigned int line_no,
2613 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \
2614 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2616 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2618 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2619 0, (type), __LINE__, (nblocks))
2622 extern void ext4_free_inode(handle_t *, struct inode *);
2623 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2624 extern unsigned long ext4_count_free_inodes(struct super_block *);
2625 extern unsigned long ext4_count_dirs(struct super_block *);
2626 extern void ext4_check_inodes_bitmap(struct super_block *);
2627 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2628 extern int ext4_init_inode_table(struct super_block *sb,
2629 ext4_group_t group, int barrier);
2630 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2633 extern const struct seq_operations ext4_mb_seq_groups_ops;
2634 extern long ext4_mb_stats;
2635 extern long ext4_mb_max_to_scan;
2636 extern int ext4_mb_init(struct super_block *);
2637 extern int ext4_mb_release(struct super_block *);
2638 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2639 struct ext4_allocation_request *, int *);
2640 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2641 extern void ext4_discard_preallocations(struct inode *);
2642 extern int __init ext4_init_mballoc(void);
2643 extern void ext4_exit_mballoc(void);
2644 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2645 struct buffer_head *bh, ext4_fsblk_t block,
2646 unsigned long count, int flags);
2647 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2648 ext4_group_t ngroups);
2649 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2650 ext4_group_t i, struct ext4_group_desc *desc);
2651 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2652 ext4_fsblk_t block, unsigned long count);
2653 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2654 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid);
2657 int ext4_inode_is_fast_symlink(struct inode *inode);
2658 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2659 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2660 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count,
2661 bool wait, struct buffer_head **bhs);
2662 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2663 struct buffer_head *bh_result, int create);
2664 int ext4_get_block(struct inode *inode, sector_t iblock,
2665 struct buffer_head *bh_result, int create);
2666 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2667 struct buffer_head *bh, int create);
2668 int ext4_walk_page_buffers(handle_t *handle,
2669 struct buffer_head *head,
2673 int (*fn)(handle_t *handle,
2674 struct buffer_head *bh));
2675 int do_journal_get_write_access(handle_t *handle,
2676 struct buffer_head *bh);
2677 #define FALL_BACK_TO_NONDELALLOC 1
2678 #define CONVERT_INLINE_DATA 2
2681 EXT4_IGET_NORMAL = 0,
2682 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */
2683 EXT4_IGET_HANDLE = 0x0002 /* Inode # is from a handle */
2686 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
2687 ext4_iget_flags flags, const char *function,
2690 #define ext4_iget(sb, ino, flags) \
2691 __ext4_iget((sb), (ino), (flags), __func__, __LINE__)
2693 extern int ext4_write_inode(struct inode *, struct writeback_control *);
2694 extern int ext4_setattr(struct dentry *, struct iattr *);
2695 extern int ext4_getattr(const struct path *, struct kstat *, u32, unsigned int);
2696 extern void ext4_evict_inode(struct inode *);
2697 extern void ext4_clear_inode(struct inode *);
2698 extern int ext4_file_getattr(const struct path *, struct kstat *, u32, unsigned int);
2699 extern int ext4_sync_inode(handle_t *, struct inode *);
2700 extern void ext4_dirty_inode(struct inode *, int);
2701 extern int ext4_change_inode_journal_flag(struct inode *, int);
2702 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2703 extern int ext4_inode_attach_jinode(struct inode *inode);
2704 extern int ext4_can_truncate(struct inode *inode);
2705 extern int ext4_truncate(struct inode *);
2706 extern int ext4_break_layouts(struct inode *);
2707 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2708 extern void ext4_set_inode_flags(struct inode *);
2709 extern int ext4_alloc_da_blocks(struct inode *inode);
2710 extern void ext4_set_aops(struct inode *inode);
2711 extern int ext4_writepage_trans_blocks(struct inode *);
2712 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2713 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2714 loff_t lstart, loff_t lend);
2715 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf);
2716 extern vm_fault_t ext4_filemap_fault(struct vm_fault *vmf);
2717 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2718 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
2719 extern void ext4_da_release_space(struct inode *inode, int to_free);
2720 extern void ext4_da_update_reserve_space(struct inode *inode,
2721 int used, int quota_claim);
2722 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
2723 ext4_fsblk_t pblk, ext4_lblk_t len);
2726 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2727 struct ext4_map_blocks *map, int flags);
2728 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2729 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2730 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2731 ext4_lblk_t start, ext4_lblk_t end);
2734 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2735 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2738 extern int ext4_ext_migrate(struct inode *);
2739 extern int ext4_ind_migrate(struct inode *inode);
2742 extern int ext4_dirblock_csum_verify(struct inode *inode,
2743 struct buffer_head *bh);
2744 extern int ext4_orphan_add(handle_t *, struct inode *);
2745 extern int ext4_orphan_del(handle_t *, struct inode *);
2746 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2747 __u32 start_minor_hash, __u32 *next_hash);
2748 extern int ext4_search_dir(struct buffer_head *bh,
2752 struct ext4_filename *fname,
2753 unsigned int offset,
2754 struct ext4_dir_entry_2 **res_dir);
2755 extern int ext4_generic_delete_entry(handle_t *handle,
2757 struct ext4_dir_entry_2 *de_del,
2758 struct buffer_head *bh,
2762 extern bool ext4_empty_dir(struct inode *inode);
2765 extern void ext4_kvfree_array_rcu(void *to_free);
2766 extern int ext4_group_add(struct super_block *sb,
2767 struct ext4_new_group_data *input);
2768 extern int ext4_group_extend(struct super_block *sb,
2769 struct ext4_super_block *es,
2770 ext4_fsblk_t n_blocks_count);
2771 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2774 extern struct buffer_head *ext4_sb_bread(struct super_block *sb,
2775 sector_t block, int op_flags);
2776 extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
2777 extern int ext4_calculate_overhead(struct super_block *sb);
2778 extern void ext4_superblock_csum_set(struct super_block *sb);
2779 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2780 ext4_group_t ngroup);
2781 extern const char *ext4_decode_error(struct super_block *sb, int errno,
2783 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb,
2784 ext4_group_t block_group,
2785 unsigned int flags);
2787 extern __printf(6, 7)
2788 void __ext4_error(struct super_block *, const char *, unsigned int, int, __u64,
2790 extern __printf(6, 7)
2791 void __ext4_error_inode(struct inode *, const char *, unsigned int,
2792 ext4_fsblk_t, int, const char *, ...);
2793 extern __printf(5, 6)
2794 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2796 extern void __ext4_std_error(struct super_block *, const char *,
2798 extern __printf(5, 6)
2799 void __ext4_abort(struct super_block *, const char *, unsigned int, int,
2801 extern __printf(4, 5)
2802 void __ext4_warning(struct super_block *, const char *, unsigned int,
2804 extern __printf(4, 5)
2805 void __ext4_warning_inode(const struct inode *inode, const char *function,
2806 unsigned int line, const char *fmt, ...);
2807 extern __printf(3, 4)
2808 void __ext4_msg(struct super_block *, const char *, const char *, ...);
2809 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2810 const char *, unsigned int, const char *);
2811 extern __printf(7, 8)
2812 void __ext4_grp_locked_error(const char *, unsigned int,
2813 struct super_block *, ext4_group_t,
2814 unsigned long, ext4_fsblk_t,
2817 #define EXT4_ERROR_INODE(inode, fmt, a...) \
2818 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
2820 #define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...) \
2821 __ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a)
2823 #define ext4_error_inode_block(inode, block, err, fmt, a...) \
2824 __ext4_error_inode((inode), __func__, __LINE__, (block), (err), \
2827 #define EXT4_ERROR_FILE(file, block, fmt, a...) \
2828 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
2830 #ifdef CONFIG_PRINTK
2832 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2833 __ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__)
2834 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \
2835 __ext4_error_inode((inode), (func), (line), (block), \
2836 (err), (fmt), ##__VA_ARGS__)
2837 #define ext4_error_file(file, func, line, block, fmt, ...) \
2838 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2839 #define ext4_error(sb, fmt, ...) \
2840 __ext4_error((sb), __func__, __LINE__, 0, 0, (fmt), ##__VA_ARGS__)
2841 #define ext4_error_err(sb, err, fmt, ...) \
2842 __ext4_error((sb), __func__, __LINE__, (err), 0, (fmt), ##__VA_ARGS__)
2843 #define ext4_abort(sb, err, fmt, ...) \
2844 __ext4_abort((sb), __func__, __LINE__, (err), (fmt), ##__VA_ARGS__)
2845 #define ext4_warning(sb, fmt, ...) \
2846 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2847 #define ext4_warning_inode(inode, fmt, ...) \
2848 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
2849 #define ext4_msg(sb, level, fmt, ...) \
2850 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2851 #define dump_mmp_msg(sb, mmp, msg) \
2852 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2853 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2854 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2859 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2861 no_printk(fmt, ##__VA_ARGS__); \
2862 __ext4_error_inode(inode, "", 0, block, 0, " "); \
2864 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \
2866 no_printk(fmt, ##__VA_ARGS__); \
2867 __ext4_error_inode(inode, "", 0, block, err, " "); \
2869 #define ext4_error_file(file, func, line, block, fmt, ...) \
2871 no_printk(fmt, ##__VA_ARGS__); \
2872 __ext4_error_file(file, "", 0, block, " "); \
2874 #define ext4_error(sb, fmt, ...) \
2876 no_printk(fmt, ##__VA_ARGS__); \
2877 __ext4_error(sb, "", 0, 0, 0, " "); \
2879 #define ext4_error_err(sb, err, fmt, ...) \
2881 no_printk(fmt, ##__VA_ARGS__); \
2882 __ext4_error(sb, "", 0, err, 0, " "); \
2884 #define ext4_abort(sb, err, fmt, ...) \
2886 no_printk(fmt, ##__VA_ARGS__); \
2887 __ext4_abort(sb, "", 0, err, " "); \
2889 #define ext4_warning(sb, fmt, ...) \
2891 no_printk(fmt, ##__VA_ARGS__); \
2892 __ext4_warning(sb, "", 0, " "); \
2894 #define ext4_warning_inode(inode, fmt, ...) \
2896 no_printk(fmt, ##__VA_ARGS__); \
2897 __ext4_warning_inode(inode, "", 0, " "); \
2899 #define ext4_msg(sb, level, fmt, ...) \
2901 no_printk(fmt, ##__VA_ARGS__); \
2902 __ext4_msg(sb, "", " "); \
2904 #define dump_mmp_msg(sb, mmp, msg) \
2905 __dump_mmp_msg(sb, mmp, "", 0, "")
2906 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2908 no_printk(fmt, ##__VA_ARGS__); \
2909 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
2914 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2916 extern int ext4_update_rocompat_feature(handle_t *handle,
2917 struct super_block *sb, __u32 rocompat);
2918 extern int ext4_update_incompat_feature(handle_t *handle,
2919 struct super_block *sb, __u32 incompat);
2920 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2921 struct ext4_group_desc *bg);
2922 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2923 struct ext4_group_desc *bg);
2924 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2925 struct ext4_group_desc *bg);
2926 extern __u32 ext4_free_group_clusters(struct super_block *sb,
2927 struct ext4_group_desc *bg);
2928 extern __u32 ext4_free_inodes_count(struct super_block *sb,
2929 struct ext4_group_desc *bg);
2930 extern __u32 ext4_used_dirs_count(struct super_block *sb,
2931 struct ext4_group_desc *bg);
2932 extern __u32 ext4_itable_unused_count(struct super_block *sb,
2933 struct ext4_group_desc *bg);
2934 extern void ext4_block_bitmap_set(struct super_block *sb,
2935 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2936 extern void ext4_inode_bitmap_set(struct super_block *sb,
2937 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2938 extern void ext4_inode_table_set(struct super_block *sb,
2939 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2940 extern void ext4_free_group_clusters_set(struct super_block *sb,
2941 struct ext4_group_desc *bg,
2943 extern void ext4_free_inodes_set(struct super_block *sb,
2944 struct ext4_group_desc *bg, __u32 count);
2945 extern void ext4_used_dirs_set(struct super_block *sb,
2946 struct ext4_group_desc *bg, __u32 count);
2947 extern void ext4_itable_unused_set(struct super_block *sb,
2948 struct ext4_group_desc *bg, __u32 count);
2949 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2950 struct ext4_group_desc *gdp);
2951 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2952 struct ext4_group_desc *gdp);
2953 extern int ext4_register_li_request(struct super_block *sb,
2954 ext4_group_t first_not_zeroed);
2956 static inline int ext4_has_metadata_csum(struct super_block *sb)
2958 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
2959 !EXT4_SB(sb)->s_chksum_driver);
2961 return ext4_has_feature_metadata_csum(sb) &&
2962 (EXT4_SB(sb)->s_chksum_driver != NULL);
2965 static inline int ext4_has_group_desc_csum(struct super_block *sb)
2967 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb);
2970 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2972 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2973 le32_to_cpu(es->s_blocks_count_lo);
2976 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2978 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2979 le32_to_cpu(es->s_r_blocks_count_lo);
2982 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2984 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2985 le32_to_cpu(es->s_free_blocks_count_lo);
2988 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2991 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2992 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2995 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2998 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2999 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
3002 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
3005 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
3006 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
3009 static inline loff_t ext4_isize(struct super_block *sb,
3010 struct ext4_inode *raw_inode)
3012 if (ext4_has_feature_largedir(sb) ||
3013 S_ISREG(le16_to_cpu(raw_inode->i_mode)))
3014 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
3015 le32_to_cpu(raw_inode->i_size_lo);
3017 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
3020 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
3022 raw_inode->i_size_lo = cpu_to_le32(i_size);
3023 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
3027 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
3030 struct ext4_group_info **grp_info;
3031 long indexv, indexh;
3032 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
3033 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
3034 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
3035 grp_info = sbi_array_rcu_deref(EXT4_SB(sb), s_group_info, indexv);
3036 return grp_info[indexh];
3040 * Reading s_groups_count requires using smp_rmb() afterwards. See
3041 * the locking protocol documented in the comments of ext4_group_add()
3044 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
3046 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
3052 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
3053 ext4_group_t block_group)
3055 return block_group >> sbi->s_log_groups_per_flex;
3058 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
3060 return 1 << sbi->s_log_groups_per_flex;
3063 #define ext4_std_error(sb, errno) \
3066 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
3070 /* Each CPU can accumulate percpu_counter_batch clusters in their local
3071 * counters. So we need to make sure we have free clusters more
3072 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
3074 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
3076 #define EXT4_FREECLUSTERS_WATERMARK 0
3079 /* Update i_disksize. Requires i_mutex to avoid races with truncate */
3080 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
3082 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
3083 !inode_is_locked(inode));
3084 down_write(&EXT4_I(inode)->i_data_sem);
3085 if (newsize > EXT4_I(inode)->i_disksize)
3086 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize);
3087 up_write(&EXT4_I(inode)->i_data_sem);
3090 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
3091 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
3095 if (newsize > inode->i_size) {
3096 i_size_write(inode, newsize);
3099 if (newsize > EXT4_I(inode)->i_disksize) {
3100 ext4_update_i_disksize(inode, newsize);
3106 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
3109 struct ext4_group_info {
3110 unsigned long bb_state;
3111 struct rb_root bb_free_root;
3112 ext4_grpblk_t bb_first_free; /* first free block */
3113 ext4_grpblk_t bb_free; /* total free blocks */
3114 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
3115 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
3116 struct list_head bb_prealloc_list;
3120 struct rw_semaphore alloc_sem;
3121 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
3122 * regions, index is order.
3123 * bb_counters[3] = 5 means
3124 * 5 free 8-block regions. */
3127 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
3128 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
3129 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
3130 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
3131 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \
3132 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT)
3133 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \
3134 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT)
3136 #define EXT4_MB_GRP_NEED_INIT(grp) \
3137 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
3138 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
3139 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3140 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
3141 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3143 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
3144 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3145 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
3146 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3147 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
3148 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3150 #define EXT4_MAX_CONTENTION 8
3151 #define EXT4_CONTENTION_THRESHOLD 2
3153 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
3156 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
3160 * Returns true if the filesystem is busy enough that attempts to
3161 * access the block group locks has run into contention.
3163 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
3165 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
3168 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
3170 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
3171 if (spin_trylock(lock))
3173 * We're able to grab the lock right away, so drop the
3174 * lock contention counter.
3176 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
3179 * The lock is busy, so bump the contention counter,
3180 * and then wait on the spin lock.
3182 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
3183 EXT4_MAX_CONTENTION);
3188 static inline void ext4_unlock_group(struct super_block *sb,
3191 spin_unlock(ext4_group_lock_ptr(sb, group));
3195 * Block validity checking
3197 #define ext4_check_indirect_blockref(inode, bh) \
3198 ext4_check_blockref(__func__, __LINE__, inode, \
3199 (__le32 *)(bh)->b_data, \
3200 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
3202 #define ext4_ind_check_inode(inode) \
3203 ext4_check_blockref(__func__, __LINE__, inode, \
3204 EXT4_I(inode)->i_data, \
3208 * Inodes and files operations
3212 extern const struct file_operations ext4_dir_operations;
3214 #ifdef CONFIG_UNICODE
3215 extern const struct dentry_operations ext4_dentry_ops;
3219 extern const struct inode_operations ext4_file_inode_operations;
3220 extern const struct file_operations ext4_file_operations;
3221 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
3224 extern int ext4_get_max_inline_size(struct inode *inode);
3225 extern int ext4_find_inline_data_nolock(struct inode *inode);
3226 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
3228 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
3230 extern int ext4_readpage_inline(struct inode *inode, struct page *page);
3231 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3232 struct inode *inode,
3233 loff_t pos, unsigned len,
3235 struct page **pagep);
3236 extern int ext4_write_inline_data_end(struct inode *inode,
3237 loff_t pos, unsigned len,
3240 extern struct buffer_head *
3241 ext4_journalled_write_inline_data(struct inode *inode,
3244 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
3245 struct inode *inode,
3246 loff_t pos, unsigned len,
3248 struct page **pagep,
3250 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
3251 unsigned len, unsigned copied,
3253 extern int ext4_try_add_inline_entry(handle_t *handle,
3254 struct ext4_filename *fname,
3255 struct inode *dir, struct inode *inode);
3256 extern int ext4_try_create_inline_dir(handle_t *handle,
3257 struct inode *parent,
3258 struct inode *inode);
3259 extern int ext4_read_inline_dir(struct file *filp,
3260 struct dir_context *ctx,
3261 int *has_inline_data);
3262 extern int ext4_inlinedir_to_tree(struct file *dir_file,
3263 struct inode *dir, ext4_lblk_t block,
3264 struct dx_hash_info *hinfo,
3265 __u32 start_hash, __u32 start_minor_hash,
3266 int *has_inline_data);
3267 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3268 struct ext4_filename *fname,
3269 struct ext4_dir_entry_2 **res_dir,
3270 int *has_inline_data);
3271 extern int ext4_delete_inline_entry(handle_t *handle,
3273 struct ext4_dir_entry_2 *de_del,
3274 struct buffer_head *bh,
3275 int *has_inline_data);
3276 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3277 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3278 struct ext4_dir_entry_2 **parent_de,
3280 extern int ext4_inline_data_fiemap(struct inode *inode,
3281 struct fiemap_extent_info *fieinfo,
3282 int *has_inline, __u64 start, __u64 len);
3285 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap);
3287 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3289 extern int ext4_convert_inline_data(struct inode *inode);
3291 static inline int ext4_has_inline_data(struct inode *inode)
3293 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3294 EXT4_I(inode)->i_inline_off;
3298 extern const struct inode_operations ext4_dir_inode_operations;
3299 extern const struct inode_operations ext4_special_inode_operations;
3300 extern struct dentry *ext4_get_parent(struct dentry *child);
3301 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3302 struct ext4_dir_entry_2 *de,
3303 int blocksize, int csum_size,
3304 unsigned int parent_ino, int dotdot_real_len);
3305 extern void ext4_initialize_dirent_tail(struct buffer_head *bh,
3306 unsigned int blocksize);
3307 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode,
3308 struct buffer_head *bh);
3309 extern int ext4_ci_compare(const struct inode *parent,
3310 const struct qstr *fname,
3311 const struct qstr *entry, bool quick);
3314 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = {
3315 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
3316 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
3317 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
3318 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
3319 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
3320 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
3321 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
3324 static inline void ext4_set_de_type(struct super_block *sb,
3325 struct ext4_dir_entry_2 *de,
3327 if (ext4_has_feature_filetype(sb))
3328 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3332 extern int ext4_mpage_readpages(struct inode *inode,
3333 struct readahead_control *rac, struct page *page);
3334 extern int __init ext4_init_post_read_processing(void);
3335 extern void ext4_exit_post_read_processing(void);
3338 extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3339 extern const struct inode_operations ext4_symlink_inode_operations;
3340 extern const struct inode_operations ext4_fast_symlink_inode_operations;
3343 extern int ext4_register_sysfs(struct super_block *sb);
3344 extern void ext4_unregister_sysfs(struct super_block *sb);
3345 extern int __init ext4_init_sysfs(void);
3346 extern void ext4_exit_sysfs(void);
3348 /* block_validity */
3349 extern void ext4_release_system_zone(struct super_block *sb);
3350 extern int ext4_setup_system_zone(struct super_block *sb);
3351 extern int __init ext4_init_system_zone(void);
3352 extern void ext4_exit_system_zone(void);
3353 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
3354 ext4_fsblk_t start_blk,
3355 unsigned int count);
3356 extern int ext4_check_blockref(const char *, unsigned int,
3357 struct inode *, __le32 *, unsigned int);
3360 struct ext4_ext_path;
3364 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3367 #define EXT_MAX_BLOCKS 0xffffffff
3369 extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode);
3370 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3371 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3372 struct ext4_map_blocks *map, int flags);
3373 extern int ext4_ext_truncate(handle_t *, struct inode *);
3374 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3376 extern void ext4_ext_init(struct super_block *);
3377 extern void ext4_ext_release(struct super_block *);
3378 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3380 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3381 loff_t offset, ssize_t len);
3382 extern int ext4_convert_unwritten_io_end_vec(handle_t *handle,
3383 ext4_io_end_t *io_end);
3384 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3385 struct ext4_map_blocks *map, int flags);
3386 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3388 struct ext4_ext_path *path);
3389 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3390 struct ext4_ext_path **,
3391 struct ext4_extent *, int);
3392 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3393 struct ext4_ext_path **,
3395 extern void ext4_ext_drop_refs(struct ext4_ext_path *);
3396 extern int ext4_ext_check_inode(struct inode *inode);
3397 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3398 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3399 __u64 start, __u64 len);
3400 extern int ext4_get_es_cache(struct inode *inode,
3401 struct fiemap_extent_info *fieinfo,
3402 __u64 start, __u64 len);
3403 extern int ext4_ext_precache(struct inode *inode);
3404 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3405 struct inode *inode2, ext4_lblk_t lblk1,
3406 ext4_lblk_t lblk2, ext4_lblk_t count,
3407 int mark_unwritten,int *err);
3408 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu);
3409 extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode,
3410 int check_cred, int restart_cred,
3415 extern void ext4_double_down_write_data_sem(struct inode *first,
3416 struct inode *second);
3417 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3418 struct inode *donor_inode);
3419 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3420 __u64 start_orig, __u64 start_donor,
3421 __u64 len, __u64 *moved_len);
3424 extern int __init ext4_init_pageio(void);
3425 extern void ext4_exit_pageio(void);
3426 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3427 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3428 extern int ext4_put_io_end(ext4_io_end_t *io_end);
3429 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3430 extern void ext4_io_submit_init(struct ext4_io_submit *io,
3431 struct writeback_control *wbc);
3432 extern void ext4_end_io_rsv_work(struct work_struct *work);
3433 extern void ext4_io_submit(struct ext4_io_submit *io);
3434 extern int ext4_bio_write_page(struct ext4_io_submit *io,
3437 struct writeback_control *wbc,
3439 extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end);
3440 extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end);
3443 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3446 extern const struct fsverity_operations ext4_verityops;
3449 * Add new method to test whether block and inode bitmaps are properly
3450 * initialized. With uninit_bg reading the block from disk is not enough
3451 * to mark the bitmap uptodate. We need to also zero-out the bitmap
3453 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3455 static inline int bitmap_uptodate(struct buffer_head *bh)
3457 return (buffer_uptodate(bh) &&
3458 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3460 static inline void set_bitmap_uptodate(struct buffer_head *bh)
3462 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3465 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
3467 /* For ioend & aio unwritten conversion wait queues */
3468 #define EXT4_WQ_HASH_SZ 37
3469 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
3471 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3473 extern int ext4_resize_begin(struct super_block *sb);
3474 extern void ext4_resize_end(struct super_block *sb);
3476 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
3477 struct ext4_io_end *io_end)
3479 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
3480 io_end->flag |= EXT4_IO_END_UNWRITTEN;
3481 atomic_inc(&EXT4_I(inode)->i_unwritten);
3485 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3487 struct inode *inode = io_end->inode;
3489 if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
3490 io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3491 /* Wake up anyone waiting on unwritten extent conversion */
3492 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
3493 wake_up_all(ext4_ioend_wq(inode));
3497 extern const struct iomap_ops ext4_iomap_ops;
3498 extern const struct iomap_ops ext4_iomap_overwrite_ops;
3499 extern const struct iomap_ops ext4_iomap_report_ops;
3501 static inline int ext4_buffer_uptodate(struct buffer_head *bh)
3504 * If the buffer has the write error flag, we have failed
3505 * to write out data in the block. In this case, we don't
3506 * have to read the block because we may read the old data
3509 if (!buffer_uptodate(bh) && buffer_write_io_error(bh))
3510 set_buffer_uptodate(bh);
3511 return buffer_uptodate(bh);
3514 #endif /* __KERNEL__ */
3516 #define EFSBADCRC EBADMSG /* Bad CRC detected */
3517 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
3519 #endif /* _EXT4_H */
projects / linux-block.git / blob