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>
40 #include <linux/compat.h>
43 #include <linux/fscrypt.h>
44 #include <linux/fsverity.h>
46 #include <linux/compiler.h>
49 * The fourth extended filesystem constants/structures
53 * with AGGRESSIVE_CHECK allocator runs consistency checks over
54 * structures. these checks slow things down a lot
56 #define AGGRESSIVE_CHECK__
59 * with DOUBLE_CHECK defined mballoc creates persistent in-core
60 * bitmaps, maintains and uses them to check for double allocations
62 #define DOUBLE_CHECK__
65 * Define EXT4FS_DEBUG to produce debug messages
73 #define ext4_debug(f, a...) \
75 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
76 __FILE__, __LINE__, __func__); \
77 printk(KERN_DEBUG f, ## a); \
80 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
84 * Turn on EXT_DEBUG to get lots of info about extents operations.
88 #define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__)
90 #define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
93 /* data type for block offset of block group */
94 typedef int ext4_grpblk_t;
96 /* data type for filesystem-wide blocks number */
97 typedef unsigned long long ext4_fsblk_t;
99 /* data type for file logical block number */
100 typedef __u32 ext4_lblk_t;
102 /* data type for block group number */
103 typedef unsigned int ext4_group_t;
105 enum SHIFT_DIRECTION {
111 * Flags used in mballoc's allocation_context flags field.
113 * Also used to show what's going on for debugging purposes when the
114 * flag field is exported via the traceport interface
117 /* prefer goal again. length */
118 #define EXT4_MB_HINT_MERGE 0x0001
119 /* blocks already reserved */
120 #define EXT4_MB_HINT_RESERVED 0x0002
121 /* metadata is being allocated */
122 #define EXT4_MB_HINT_METADATA 0x0004
123 /* first blocks in the file */
124 #define EXT4_MB_HINT_FIRST 0x0008
125 /* search for the best chunk */
126 #define EXT4_MB_HINT_BEST 0x0010
127 /* data is being allocated */
128 #define EXT4_MB_HINT_DATA 0x0020
129 /* don't preallocate (for tails) */
130 #define EXT4_MB_HINT_NOPREALLOC 0x0040
131 /* allocate for locality group */
132 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080
133 /* allocate goal blocks or none */
134 #define EXT4_MB_HINT_GOAL_ONLY 0x0100
135 /* goal is meaningful */
136 #define EXT4_MB_HINT_TRY_GOAL 0x0200
137 /* blocks already pre-reserved by delayed allocation */
138 #define EXT4_MB_DELALLOC_RESERVED 0x0400
139 /* We are doing stream allocation */
140 #define EXT4_MB_STREAM_ALLOC 0x0800
141 /* Use reserved root blocks if needed */
142 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000
143 /* Use blocks from reserved pool */
144 #define EXT4_MB_USE_RESERVED 0x2000
146 struct ext4_allocation_request {
147 /* target inode for block we're allocating */
149 /* how many blocks we want to allocate */
151 /* logical block in target inode */
153 /* the closest logical allocated block to the left */
155 /* the closest logical allocated block to the right */
157 /* phys. target (a hint) */
159 /* phys. block for the closest logical allocated block to the left */
161 /* phys. block for the closest logical allocated block to the right */
163 /* flags. see above EXT4_MB_HINT_* */
168 * Logical to physical block mapping, used by ext4_map_blocks()
170 * This structure is used to pass requests into ext4_map_blocks() as
171 * well as to store the information returned by ext4_map_blocks(). It
172 * takes less room on the stack than a struct buffer_head.
174 #define EXT4_MAP_NEW (1 << BH_New)
175 #define EXT4_MAP_MAPPED (1 << BH_Mapped)
176 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten)
177 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary)
178 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
179 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
181 struct ext4_map_blocks {
185 unsigned int m_flags;
189 * Block validity checking, system zone rbtree.
191 struct ext4_system_blocks {
197 * Flags for ext4_io_end->flags
199 #define EXT4_IO_END_UNWRITTEN 0x0001
201 struct ext4_io_end_vec {
202 struct list_head list; /* list of io_end_vec */
203 loff_t offset; /* offset in the file */
204 ssize_t size; /* size of the extent */
208 * For converting unwritten extents on a work queue. 'handle' is used for
209 * buffered writeback.
211 typedef struct ext4_io_end {
212 struct list_head list; /* per-file finished IO list */
213 handle_t *handle; /* handle reserved for extent
215 struct inode *inode; /* file being written to */
216 struct bio *bio; /* Linked list of completed
217 * bios covering the extent */
218 unsigned int flag; /* unwritten or not */
219 atomic_t count; /* reference counter */
220 struct list_head list_vec; /* list of ext4_io_end_vec */
223 struct ext4_io_submit {
224 struct writeback_control *io_wbc;
226 ext4_io_end_t *io_end;
227 sector_t io_next_block;
231 * Special inodes numbers
233 #define EXT4_BAD_INO 1 /* Bad blocks inode */
234 #define EXT4_ROOT_INO 2 /* Root inode */
235 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */
236 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
237 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
238 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
239 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
240 #define EXT4_JOURNAL_INO 8 /* Journal inode */
242 /* First non-reserved inode for old ext4 filesystems */
243 #define EXT4_GOOD_OLD_FIRST_INO 11
246 * Maximal count of links to a file
248 #define EXT4_LINK_MAX 65000
251 * Macro-instructions used to manage several block sizes
253 #define EXT4_MIN_BLOCK_SIZE 1024
254 #define EXT4_MAX_BLOCK_SIZE 65536
255 #define EXT4_MIN_BLOCK_LOG_SIZE 10
256 #define EXT4_MAX_BLOCK_LOG_SIZE 16
257 #define EXT4_MAX_CLUSTER_LOG_SIZE 30
259 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
261 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
263 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
264 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
265 EXT4_SB(s)->s_cluster_bits)
267 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
268 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
270 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
273 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
274 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
275 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
277 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
278 EXT4_GOOD_OLD_INODE_SIZE : \
280 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
281 EXT4_GOOD_OLD_FIRST_INO : \
284 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
285 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \
286 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \
289 /* Translate a block number to a cluster number */
290 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
291 /* Translate a cluster number to a block number */
292 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
293 /* Translate # of blks to # of clusters */
294 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
295 (sbi)->s_cluster_bits)
296 /* Mask out the low bits to get the starting block of the cluster */
297 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
298 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
299 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
300 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
301 /* Fill in the low bits to get the last block of the cluster */
302 #define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) | \
303 ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1))
304 /* Get the cluster offset */
305 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
306 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
307 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
308 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
311 * Structure of a blocks group descriptor
313 struct ext4_group_desc
315 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
316 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
317 __le32 bg_inode_table_lo; /* Inodes table block */
318 __le16 bg_free_blocks_count_lo;/* Free blocks count */
319 __le16 bg_free_inodes_count_lo;/* Free inodes count */
320 __le16 bg_used_dirs_count_lo; /* Directories count */
321 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
322 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
323 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
324 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
325 __le16 bg_itable_unused_lo; /* Unused inodes count */
326 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
327 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
328 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
329 __le32 bg_inode_table_hi; /* Inodes table block MSB */
330 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
331 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
332 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
333 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
334 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
335 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
336 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
340 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
341 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
343 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
344 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
348 * Structure of a flex block group info
352 atomic64_t free_clusters;
353 atomic_t free_inodes;
357 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
358 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
359 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
362 * Macro-instructions used to manage group descriptors
364 #define EXT4_MIN_DESC_SIZE 32
365 #define EXT4_MIN_DESC_SIZE_64BIT 64
366 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
367 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
369 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
370 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
371 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
372 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
373 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
375 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
376 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
377 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
381 * Constants relative to the data blocks
383 #define EXT4_NDIR_BLOCKS 12
384 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
385 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
386 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
387 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
392 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
393 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
394 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
395 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
396 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
397 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
398 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
399 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
400 /* Reserved for compression usage... */
401 #define EXT4_DIRTY_FL 0x00000100
402 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
403 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
404 /* nb: was previously EXT2_ECOMPR_FL */
405 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */
406 /* End compression flags --- maybe not all used */
407 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
408 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
409 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
410 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
411 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
412 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
413 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
414 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
415 #define EXT4_VERITY_FL 0x00100000 /* Verity protected inode */
416 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
417 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
418 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
419 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
420 #define EXT4_CASEFOLD_FL 0x40000000 /* Casefolded file */
421 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
423 #define EXT4_FL_USER_VISIBLE 0x705BDFFF /* User visible flags */
424 #define EXT4_FL_USER_MODIFIABLE 0x604BC0FF /* User modifiable flags */
426 /* Flags we can manipulate with through EXT4_IOC_FSSETXATTR */
427 #define EXT4_FL_XFLAG_VISIBLE (EXT4_SYNC_FL | \
428 EXT4_IMMUTABLE_FL | \
434 /* Flags that should be inherited by new inodes from their parent. */
435 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
436 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
437 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
438 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
439 EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL)
441 /* Flags that are appropriate for regular files (all but dir-specific ones). */
442 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\
443 EXT4_PROJINHERIT_FL))
445 /* Flags that are appropriate for non-directories/regular files. */
446 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
448 /* The only flags that should be swapped */
449 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL)
451 /* Mask out flags that are inappropriate for the given type of inode. */
452 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
456 else if (S_ISREG(mode))
457 return flags & EXT4_REG_FLMASK;
459 return flags & EXT4_OTHER_FLMASK;
463 * Inode flags used for atomic set/get
466 EXT4_INODE_SECRM = 0, /* Secure deletion */
467 EXT4_INODE_UNRM = 1, /* Undelete */
468 EXT4_INODE_COMPR = 2, /* Compress file */
469 EXT4_INODE_SYNC = 3, /* Synchronous updates */
470 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
471 EXT4_INODE_APPEND = 5, /* writes to file may only append */
472 EXT4_INODE_NODUMP = 6, /* do not dump file */
473 EXT4_INODE_NOATIME = 7, /* do not update atime */
474 /* Reserved for compression usage... */
475 EXT4_INODE_DIRTY = 8,
476 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
477 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
478 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */
479 /* End compression flags --- maybe not all used */
480 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
481 EXT4_INODE_IMAGIC = 13, /* AFS directory */
482 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
483 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
484 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
485 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
486 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
487 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
488 EXT4_INODE_VERITY = 20, /* Verity protected inode */
489 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
490 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
491 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
492 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */
493 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
497 * Since it's pretty easy to mix up bit numbers and hex values, we use a
498 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
499 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
500 * any extra space in the compiled kernel image, otherwise, the build will fail.
501 * It's important that these values are the same, since we are using
502 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
503 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
504 * values found in ext2, ext3 and ext4 filesystems, and of course the values
505 * defined in e2fsprogs.
507 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
509 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
510 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
512 static inline void ext4_check_flag_values(void)
514 CHECK_FLAG_VALUE(SECRM);
515 CHECK_FLAG_VALUE(UNRM);
516 CHECK_FLAG_VALUE(COMPR);
517 CHECK_FLAG_VALUE(SYNC);
518 CHECK_FLAG_VALUE(IMMUTABLE);
519 CHECK_FLAG_VALUE(APPEND);
520 CHECK_FLAG_VALUE(NODUMP);
521 CHECK_FLAG_VALUE(NOATIME);
522 CHECK_FLAG_VALUE(DIRTY);
523 CHECK_FLAG_VALUE(COMPRBLK);
524 CHECK_FLAG_VALUE(NOCOMPR);
525 CHECK_FLAG_VALUE(ENCRYPT);
526 CHECK_FLAG_VALUE(INDEX);
527 CHECK_FLAG_VALUE(IMAGIC);
528 CHECK_FLAG_VALUE(JOURNAL_DATA);
529 CHECK_FLAG_VALUE(NOTAIL);
530 CHECK_FLAG_VALUE(DIRSYNC);
531 CHECK_FLAG_VALUE(TOPDIR);
532 CHECK_FLAG_VALUE(HUGE_FILE);
533 CHECK_FLAG_VALUE(EXTENTS);
534 CHECK_FLAG_VALUE(VERITY);
535 CHECK_FLAG_VALUE(EA_INODE);
536 CHECK_FLAG_VALUE(EOFBLOCKS);
537 CHECK_FLAG_VALUE(INLINE_DATA);
538 CHECK_FLAG_VALUE(PROJINHERIT);
539 CHECK_FLAG_VALUE(RESERVED);
542 /* Used to pass group descriptor data when online resize is done */
543 struct ext4_new_group_input {
544 __u32 group; /* Group number for this data */
545 __u64 block_bitmap; /* Absolute block number of block bitmap */
546 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
547 __u64 inode_table; /* Absolute block number of inode table start */
548 __u32 blocks_count; /* Total number of blocks in this group */
549 __u16 reserved_blocks; /* Number of reserved blocks in this group */
553 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
554 struct compat_ext4_new_group_input {
556 compat_u64 block_bitmap;
557 compat_u64 inode_bitmap;
558 compat_u64 inode_table;
565 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
566 struct ext4_new_group_data {
572 __u16 reserved_blocks;
574 __u32 free_clusters_count;
577 /* Indexes used to index group tables in ext4_new_group_data */
579 BLOCK_BITMAP = 0, /* block bitmap */
580 INODE_BITMAP, /* inode bitmap */
581 INODE_TABLE, /* inode tables */
586 * Flags used by ext4_map_blocks()
588 /* Allocate any needed blocks and/or convert an unwritten
589 extent to be an initialized ext4 */
590 #define EXT4_GET_BLOCKS_CREATE 0x0001
591 /* Request the creation of an unwritten extent */
592 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002
593 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\
594 EXT4_GET_BLOCKS_CREATE)
595 /* Caller is from the delayed allocation writeout path
596 * finally doing the actual allocation of delayed blocks */
597 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
598 /* caller is from the direct IO path, request to creation of an
599 unwritten extents if not allocated, split the unwritten
600 extent if blocks has been preallocated already*/
601 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
602 #define EXT4_GET_BLOCKS_CONVERT 0x0010
603 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
604 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
605 /* Convert extent to initialized after IO complete */
606 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
607 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
608 /* Eventual metadata allocation (due to growing extent tree)
609 * should not fail, so try to use reserved blocks for that.*/
610 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
611 /* Don't normalize allocation size (used for fallocate) */
612 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
613 /* Request will not result in inode size update (user for fallocate) */
614 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080
615 /* Convert written extents to unwritten */
616 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100
617 /* Write zeros to newly created written extents */
618 #define EXT4_GET_BLOCKS_ZERO 0x0200
619 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\
620 EXT4_GET_BLOCKS_ZERO)
621 /* Caller will submit data before dropping transaction handle. This
622 * allows jbd2 to avoid submitting data before commit. */
623 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400
626 * The bit position of these flags must not overlap with any of the
627 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(),
628 * read_extent_tree_block(), ext4_split_extent_at(),
629 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
630 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
631 * caching the extents when reading from the extent tree while a
632 * truncate or punch hole operation is in progress.
634 #define EXT4_EX_NOCACHE 0x40000000
635 #define EXT4_EX_FORCE_CACHE 0x20000000
638 * Flags used by ext4_free_blocks
640 #define EXT4_FREE_BLOCKS_METADATA 0x0001
641 #define EXT4_FREE_BLOCKS_FORGET 0x0002
642 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
643 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
644 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
645 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
646 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040
651 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
652 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
653 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
654 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
655 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
656 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
657 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
658 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
659 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
660 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
661 #define EXT4_IOC_MIGRATE _IO('f', 9)
662 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
663 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
664 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
665 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
666 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
667 #define EXT4_IOC_SWAP_BOOT _IO('f', 17)
668 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18)
669 #define EXT4_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY
670 #define EXT4_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT
671 #define EXT4_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY
672 /* ioctl codes 19--39 are reserved for fscrypt */
673 #define EXT4_IOC_CLEAR_ES_CACHE _IO('f', 40)
674 #define EXT4_IOC_GETSTATE _IOW('f', 41, __u32)
675 #define EXT4_IOC_GET_ES_CACHE _IOWR('f', 42, struct fiemap)
677 #define EXT4_IOC_FSGETXATTR FS_IOC_FSGETXATTR
678 #define EXT4_IOC_FSSETXATTR FS_IOC_FSSETXATTR
680 #define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32)
683 * Flags for going down operation
685 #define EXT4_GOING_FLAGS_DEFAULT 0x0 /* going down */
686 #define EXT4_GOING_FLAGS_LOGFLUSH 0x1 /* flush log but not data */
687 #define EXT4_GOING_FLAGS_NOLOGFLUSH 0x2 /* don't flush log nor data */
690 * Flags returned by EXT4_IOC_GETSTATE
692 * We only expose to userspace a subset of the state flags in
695 #define EXT4_STATE_FLAG_EXT_PRECACHED 0x00000001
696 #define EXT4_STATE_FLAG_NEW 0x00000002
697 #define EXT4_STATE_FLAG_NEWENTRY 0x00000004
698 #define EXT4_STATE_FLAG_DA_ALLOC_CLOSE 0x00000008
700 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
702 * ioctl commands in 32 bit emulation
704 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
705 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
706 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
707 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
708 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
709 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
710 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
711 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
712 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
713 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
717 * Returned by EXT4_IOC_GET_ES_CACHE as an additional possible flag.
718 * It indicates that the entry in extent status cache is for a hole.
720 #define EXT4_FIEMAP_EXTENT_HOLE 0x08000000
722 /* Max physical block we can address w/o extents */
723 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
725 /* Max logical block we can support */
726 #define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFF
729 * Structure of an inode on the disk
732 __le16 i_mode; /* File mode */
733 __le16 i_uid; /* Low 16 bits of Owner Uid */
734 __le32 i_size_lo; /* Size in bytes */
735 __le32 i_atime; /* Access time */
736 __le32 i_ctime; /* Inode Change time */
737 __le32 i_mtime; /* Modification time */
738 __le32 i_dtime; /* Deletion Time */
739 __le16 i_gid; /* Low 16 bits of Group Id */
740 __le16 i_links_count; /* Links count */
741 __le32 i_blocks_lo; /* Blocks count */
742 __le32 i_flags; /* File flags */
748 __u32 h_i_translator;
753 } osd1; /* OS dependent 1 */
754 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
755 __le32 i_generation; /* File version (for NFS) */
756 __le32 i_file_acl_lo; /* File ACL */
758 __le32 i_obso_faddr; /* Obsoleted fragment address */
761 __le16 l_i_blocks_high; /* were l_i_reserved1 */
762 __le16 l_i_file_acl_high;
763 __le16 l_i_uid_high; /* these 2 fields */
764 __le16 l_i_gid_high; /* were reserved2[0] */
765 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
769 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
776 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
777 __le16 m_i_file_acl_high;
778 __u32 m_i_reserved2[2];
780 } osd2; /* OS dependent 2 */
781 __le16 i_extra_isize;
782 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
783 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
784 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
785 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
786 __le32 i_crtime; /* File Creation time */
787 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
788 __le32 i_version_hi; /* high 32 bits for 64-bit version */
789 __le32 i_projid; /* Project ID */
793 __u32 reserved; /* should be zero */
794 __u32 donor_fd; /* donor file descriptor */
795 __u64 orig_start; /* logical start offset in block for orig */
796 __u64 donor_start; /* logical start offset in block for donor */
797 __u64 len; /* block length to be moved */
798 __u64 moved_len; /* moved block length */
801 #define EXT4_EPOCH_BITS 2
802 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
803 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
806 * Extended fields will fit into an inode if the filesystem was formatted
807 * with large inodes (-I 256 or larger) and there are not currently any EAs
808 * consuming all of the available space. For new inodes we always reserve
809 * enough space for the kernel's known extended fields, but for inodes
810 * created with an old kernel this might not have been the case. None of
811 * the extended inode fields is critical for correct filesystem operation.
812 * This macro checks if a certain field fits in the inode. Note that
813 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
815 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
816 ((offsetof(typeof(*ext4_inode), field) + \
817 sizeof((ext4_inode)->field)) \
818 <= (EXT4_GOOD_OLD_INODE_SIZE + \
819 (einode)->i_extra_isize)) \
822 * We use an encoding that preserves the times for extra epoch "00":
824 * extra msb of adjust for signed
825 * epoch 32-bit 32-bit tv_sec to
826 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
827 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
828 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
829 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
830 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
831 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
832 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
833 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
834 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
836 * Note that previous versions of the kernel on 64-bit systems would
837 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
838 * 1970. e2fsck will correct this, assuming that it is run on the
839 * affected filesystem before 2242.
842 static inline __le32 ext4_encode_extra_time(struct timespec64 *time)
844 u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK;
845 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
848 static inline void ext4_decode_extra_time(struct timespec64 *time,
851 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK)))
852 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
853 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
856 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
858 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {\
859 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
860 (raw_inode)->xtime ## _extra = \
861 ext4_encode_extra_time(&(inode)->xtime); \
864 (raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (inode)->xtime.tv_sec, S32_MIN, S32_MAX)); \
867 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
869 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
870 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
871 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
872 (raw_inode)->xtime ## _extra = \
873 ext4_encode_extra_time(&(einode)->xtime); \
876 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
878 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
879 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \
880 ext4_decode_extra_time(&(inode)->xtime, \
881 raw_inode->xtime ## _extra); \
884 (inode)->xtime.tv_nsec = 0; \
888 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
890 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
891 (einode)->xtime.tv_sec = \
892 (signed)le32_to_cpu((raw_inode)->xtime); \
894 (einode)->xtime.tv_sec = 0; \
895 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
896 ext4_decode_extra_time(&(einode)->xtime, \
897 raw_inode->xtime ## _extra); \
899 (einode)->xtime.tv_nsec = 0; \
902 #define i_disk_version osd1.linux1.l_i_version
904 #if defined(__KERNEL__) || defined(__linux__)
905 #define i_reserved1 osd1.linux1.l_i_reserved1
906 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
907 #define i_blocks_high osd2.linux2.l_i_blocks_high
908 #define i_uid_low i_uid
909 #define i_gid_low i_gid
910 #define i_uid_high osd2.linux2.l_i_uid_high
911 #define i_gid_high osd2.linux2.l_i_gid_high
912 #define i_checksum_lo osd2.linux2.l_i_checksum_lo
914 #elif defined(__GNU__)
916 #define i_translator osd1.hurd1.h_i_translator
917 #define i_uid_high osd2.hurd2.h_i_uid_high
918 #define i_gid_high osd2.hurd2.h_i_gid_high
919 #define i_author osd2.hurd2.h_i_author
921 #elif defined(__masix__)
923 #define i_reserved1 osd1.masix1.m_i_reserved1
924 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
925 #define i_reserved2 osd2.masix2.m_i_reserved2
927 #endif /* defined(__KERNEL__) || defined(__linux__) */
929 #include "extents_status.h"
932 * Lock subclasses for i_data_sem in the ext4_inode_info structure.
934 * These are needed to avoid lockdep false positives when we need to
935 * allocate blocks to the quota inode during ext4_map_blocks(), while
936 * holding i_data_sem for a normal (non-quota) inode. Since we don't
937 * do quota tracking for the quota inode, this avoids deadlock (as
938 * well as infinite recursion, since it isn't turtles all the way
941 * I_DATA_SEM_NORMAL - Used for most inodes
942 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode
943 * where the second inode has larger inode number
945 * I_DATA_SEM_QUOTA - Used for quota inodes only
948 I_DATA_SEM_NORMAL = 0,
955 * fourth extended file system inode data in memory
957 struct ext4_inode_info {
958 __le32 i_data[15]; /* unconverted */
960 ext4_fsblk_t i_file_acl;
963 * i_block_group is the number of the block group which contains
964 * this file's inode. Constant across the lifetime of the inode,
965 * it is used for making block allocation decisions - we try to
966 * place a file's data blocks near its inode block, and new inodes
967 * near to their parent directory's inode.
969 ext4_group_t i_block_group;
970 ext4_lblk_t i_dir_start_lookup;
971 #if (BITS_PER_LONG < 64)
972 unsigned long i_state_flags; /* Dynamic state flags */
974 unsigned long i_flags;
977 * Extended attributes can be read independently of the main file
978 * data. Taking i_mutex even when reading would cause contention
979 * between readers of EAs and writers of regular file data, so
980 * instead we synchronize on xattr_sem when reading or changing
983 struct rw_semaphore xattr_sem;
985 struct list_head i_orphan; /* unlinked but open inodes */
988 * i_disksize keeps track of what the inode size is ON DISK, not
989 * in memory. During truncate, i_size is set to the new size by
990 * the VFS prior to calling ext4_truncate(), but the filesystem won't
991 * set i_disksize to 0 until the truncate is actually under way.
993 * The intent is that i_disksize always represents the blocks which
994 * are used by this file. This allows recovery to restart truncate
995 * on orphans if we crash during truncate. We actually write i_disksize
996 * into the on-disk inode when writing inodes out, instead of i_size.
998 * The only time when i_disksize and i_size may be different is when
999 * a truncate is in progress. The only things which change i_disksize
1000 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
1005 * i_data_sem is for serialising ext4_truncate() against
1006 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
1007 * data tree are chopped off during truncate. We can't do that in
1008 * ext4 because whenever we perform intermediate commits during
1009 * truncate, the inode and all the metadata blocks *must* be in a
1010 * consistent state which allows truncation of the orphans to restart
1011 * during recovery. Hence we must fix the get_block-vs-truncate race
1012 * by other means, so we have i_data_sem.
1014 struct rw_semaphore i_data_sem;
1016 * i_mmap_sem is for serializing page faults with truncate / punch hole
1017 * operations. We have to make sure that new page cannot be faulted in
1018 * a section of the inode that is being punched. We cannot easily use
1019 * i_data_sem for this since we need protection for the whole punch
1020 * operation and i_data_sem ranks below transaction start so we have
1021 * to occasionally drop it.
1023 struct rw_semaphore i_mmap_sem;
1024 struct inode vfs_inode;
1025 struct jbd2_inode *jinode;
1027 spinlock_t i_raw_lock; /* protects updates to the raw inode */
1030 * File creation time. Its function is same as that of
1031 * struct timespec64 i_{a,c,m}time in the generic inode.
1033 struct timespec64 i_crtime;
1036 struct list_head i_prealloc_list;
1037 spinlock_t i_prealloc_lock;
1039 /* extents status tree */
1040 struct ext4_es_tree i_es_tree;
1042 struct list_head i_es_list;
1043 unsigned int i_es_all_nr; /* protected by i_es_lock */
1044 unsigned int i_es_shk_nr; /* protected by i_es_lock */
1045 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for
1046 extents to shrink. Protected by
1050 ext4_group_t i_last_alloc_group;
1052 /* allocation reservation info for delalloc */
1053 /* In case of bigalloc, this refer to clusters rather than blocks */
1054 unsigned int i_reserved_data_blocks;
1056 /* pending cluster reservations for bigalloc file systems */
1057 struct ext4_pending_tree i_pending_tree;
1059 /* on-disk additional length */
1060 __u16 i_extra_isize;
1062 /* Indicate the inline data space. */
1067 /* quota space reservation, managed internally by quota code */
1068 qsize_t i_reserved_quota;
1071 /* Lock protecting lists below */
1072 spinlock_t i_completed_io_lock;
1074 * Completed IOs that need unwritten extents handling and have
1075 * transaction reserved
1077 struct list_head i_rsv_conversion_list;
1078 struct work_struct i_rsv_conversion_work;
1079 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1081 spinlock_t i_block_reservation_lock;
1084 * Transactions that contain inode's metadata needed to complete
1085 * fsync and fdatasync, respectively.
1088 tid_t i_datasync_tid;
1091 struct dquot *i_dquot[MAXQUOTAS];
1094 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1101 * File system states
1103 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
1104 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
1105 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
1108 * Misc. filesystem flags
1110 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
1111 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
1112 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
1115 * Mount flags set via mount options or defaults
1117 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */
1118 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
1119 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
1120 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
1121 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
1122 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
1123 #define EXT4_MOUNT_ERRORS_MASK 0x00070
1124 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
1125 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
1126 #ifdef CONFIG_FS_DAX
1127 #define EXT4_MOUNT_DAX 0x00200 /* Direct Access */
1129 #define EXT4_MOUNT_DAX 0
1131 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
1132 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
1133 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
1134 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
1135 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
1136 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
1137 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
1138 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
1139 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
1140 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
1141 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */
1142 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota,
1143 * enable enforcement for hidden
1145 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable
1146 * enforcement for hidden quota
1148 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota
1150 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
1151 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
1152 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1153 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */
1154 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
1155 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
1156 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
1157 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
1158 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
1161 * Mount flags set either automatically (could not be set by mount option)
1162 * based on per file system feature or property or in special cases such as
1163 * distinguishing between explicit mount option definition and default.
1165 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1166 specified delalloc */
1167 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1168 size of blocksize * 8
1170 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1173 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly
1174 specified journal checksum */
1176 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1178 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1180 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1183 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1185 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1187 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1190 #define ext4_test_and_set_bit __test_and_set_bit_le
1191 #define ext4_set_bit __set_bit_le
1192 #define ext4_set_bit_atomic ext2_set_bit_atomic
1193 #define ext4_test_and_clear_bit __test_and_clear_bit_le
1194 #define ext4_clear_bit __clear_bit_le
1195 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
1196 #define ext4_test_bit test_bit_le
1197 #define ext4_find_next_zero_bit find_next_zero_bit_le
1198 #define ext4_find_next_bit find_next_bit_le
1200 extern void ext4_set_bits(void *bm, int cur, int len);
1203 * Maximal mount counts between two filesystem checks
1205 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1206 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1209 * Behaviour when detecting errors
1211 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1212 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1213 #define EXT4_ERRORS_PANIC 3 /* Panic */
1214 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1216 /* Metadata checksum algorithm codes */
1217 #define EXT4_CRC32C_CHKSUM 1
1220 * Structure of the super block
1222 struct ext4_super_block {
1223 /*00*/ __le32 s_inodes_count; /* Inodes count */
1224 __le32 s_blocks_count_lo; /* Blocks count */
1225 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1226 __le32 s_free_blocks_count_lo; /* Free blocks count */
1227 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1228 __le32 s_first_data_block; /* First Data Block */
1229 __le32 s_log_block_size; /* Block size */
1230 __le32 s_log_cluster_size; /* Allocation cluster size */
1231 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1232 __le32 s_clusters_per_group; /* # Clusters per group */
1233 __le32 s_inodes_per_group; /* # Inodes per group */
1234 __le32 s_mtime; /* Mount time */
1235 /*30*/ __le32 s_wtime; /* Write time */
1236 __le16 s_mnt_count; /* Mount count */
1237 __le16 s_max_mnt_count; /* Maximal mount count */
1238 __le16 s_magic; /* Magic signature */
1239 __le16 s_state; /* File system state */
1240 __le16 s_errors; /* Behaviour when detecting errors */
1241 __le16 s_minor_rev_level; /* minor revision level */
1242 /*40*/ __le32 s_lastcheck; /* time of last check */
1243 __le32 s_checkinterval; /* max. time between checks */
1244 __le32 s_creator_os; /* OS */
1245 __le32 s_rev_level; /* Revision level */
1246 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1247 __le16 s_def_resgid; /* Default gid for reserved blocks */
1249 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1251 * Note: the difference between the compatible feature set and
1252 * the incompatible feature set is that if there is a bit set
1253 * in the incompatible feature set that the kernel doesn't
1254 * know about, it should refuse to mount the filesystem.
1256 * e2fsck's requirements are more strict; if it doesn't know
1257 * about a feature in either the compatible or incompatible
1258 * feature set, it must abort and not try to meddle with
1259 * things it doesn't understand...
1261 __le32 s_first_ino; /* First non-reserved inode */
1262 __le16 s_inode_size; /* size of inode structure */
1263 __le16 s_block_group_nr; /* block group # of this superblock */
1264 __le32 s_feature_compat; /* compatible feature set */
1265 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1266 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1267 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1268 /*78*/ char s_volume_name[16]; /* volume name */
1269 /*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */
1270 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1272 * Performance hints. Directory preallocation should only
1273 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1275 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1276 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1277 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1279 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1281 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1282 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1283 __le32 s_journal_dev; /* device number of journal file */
1284 __le32 s_last_orphan; /* start of list of inodes to delete */
1285 __le32 s_hash_seed[4]; /* HTREE hash seed */
1286 __u8 s_def_hash_version; /* Default hash version to use */
1287 __u8 s_jnl_backup_type;
1288 __le16 s_desc_size; /* size of group descriptor */
1289 /*100*/ __le32 s_default_mount_opts;
1290 __le32 s_first_meta_bg; /* First metablock block group */
1291 __le32 s_mkfs_time; /* When the filesystem was created */
1292 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1293 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1294 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1295 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1296 __le32 s_free_blocks_count_hi; /* Free blocks count */
1297 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1298 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1299 __le32 s_flags; /* Miscellaneous flags */
1300 __le16 s_raid_stride; /* RAID stride */
1301 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1302 __le64 s_mmp_block; /* Block for multi-mount protection */
1303 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1304 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1305 __u8 s_checksum_type; /* metadata checksum algorithm used */
1306 __u8 s_encryption_level; /* versioning level for encryption */
1307 __u8 s_reserved_pad; /* Padding to next 32bits */
1308 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1309 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1310 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1311 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1312 snapshot's future use */
1313 __le32 s_snapshot_list; /* inode number of the head of the
1314 on-disk snapshot list */
1315 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1316 __le32 s_error_count; /* number of fs errors */
1317 __le32 s_first_error_time; /* first time an error happened */
1318 __le32 s_first_error_ino; /* inode involved in first error */
1319 __le64 s_first_error_block; /* block involved of first error */
1320 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */
1321 __le32 s_first_error_line; /* line number where error happened */
1322 __le32 s_last_error_time; /* most recent time of an error */
1323 __le32 s_last_error_ino; /* inode involved in last error */
1324 __le32 s_last_error_line; /* line number where error happened */
1325 __le64 s_last_error_block; /* block involved of last error */
1326 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */
1327 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1328 __u8 s_mount_opts[64];
1329 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1330 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1331 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1332 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
1333 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */
1334 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */
1335 __le32 s_lpf_ino; /* Location of the lost+found inode */
1336 __le32 s_prj_quota_inum; /* inode for tracking project quota */
1337 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */
1340 __u8 s_mkfs_time_hi;
1341 __u8 s_lastcheck_hi;
1342 __u8 s_first_error_time_hi;
1343 __u8 s_last_error_time_hi;
1344 __u8 s_first_error_errcode;
1345 __u8 s_last_error_errcode;
1346 __le16 s_encoding; /* Filename charset encoding */
1347 __le16 s_encoding_flags; /* Filename charset encoding flags */
1348 __le32 s_reserved[95]; /* Padding to the end of the block */
1349 __le32 s_checksum; /* crc32c(superblock) */
1352 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1357 * run-time mount flags
1359 #define EXT4_MF_MNTDIR_SAMPLED 0x0001
1360 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1361 #define EXT4_MF_TEST_DUMMY_ENCRYPTION 0x0004
1363 #ifdef CONFIG_FS_ENCRYPTION
1364 #define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \
1365 EXT4_MF_TEST_DUMMY_ENCRYPTION))
1367 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1370 /* Number of quota types we support */
1371 #define EXT4_MAXQUOTAS 3
1373 #define EXT4_ENC_UTF8_12_1 1
1376 * Flags for ext4_sb_info.s_encoding_flags.
1378 #define EXT4_ENC_STRICT_MODE_FL (1 << 0)
1380 #define ext4_has_strict_mode(sbi) \
1381 (sbi->s_encoding_flags & EXT4_ENC_STRICT_MODE_FL)
1384 * fourth extended-fs super-block data in memory
1386 struct ext4_sb_info {
1387 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1388 unsigned long s_inodes_per_block;/* Number of inodes per block */
1389 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1390 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1391 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1392 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1393 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1394 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1395 ext4_group_t s_groups_count; /* Number of groups in the fs */
1396 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1397 unsigned long s_overhead; /* # of fs overhead clusters */
1398 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1399 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1400 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1401 struct buffer_head * s_sbh; /* Buffer containing the super block */
1402 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1403 struct buffer_head * __rcu *s_group_desc;
1404 unsigned int s_mount_opt;
1405 unsigned int s_mount_opt2;
1406 unsigned int s_mount_flags;
1407 unsigned int s_def_mount_opt;
1408 ext4_fsblk_t s_sb_block;
1409 atomic64_t s_resv_clusters;
1412 unsigned short s_mount_state;
1413 unsigned short s_pad;
1414 int s_addr_per_block_bits;
1415 int s_desc_per_block_bits;
1418 unsigned int s_inode_readahead_blks;
1419 unsigned int s_inode_goal;
1421 int s_def_hash_version;
1422 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1423 struct percpu_counter s_freeclusters_counter;
1424 struct percpu_counter s_freeinodes_counter;
1425 struct percpu_counter s_dirs_counter;
1426 struct percpu_counter s_dirtyclusters_counter;
1427 struct blockgroup_lock *s_blockgroup_lock;
1428 struct proc_dir_entry *s_proc;
1429 struct kobject s_kobj;
1430 struct completion s_kobj_unregister;
1431 struct super_block *s_sb;
1432 #ifdef CONFIG_UNICODE
1433 struct unicode_map *s_encoding;
1434 __u16 s_encoding_flags;
1438 struct journal_s *s_journal;
1439 struct list_head s_orphan;
1440 struct mutex s_orphan_lock;
1441 unsigned long s_ext4_flags; /* Ext4 superblock flags */
1442 unsigned long s_commit_interval;
1443 u32 s_max_batch_time;
1444 u32 s_min_batch_time;
1445 struct block_device *journal_bdev;
1447 /* Names of quota files with journalled quota */
1448 char __rcu *s_qf_names[EXT4_MAXQUOTAS];
1449 int s_jquota_fmt; /* Format of quota to use */
1451 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1452 struct ext4_system_blocks __rcu *system_blks;
1454 #ifdef EXTENTS_STATS
1455 /* ext4 extents stats */
1456 unsigned long s_ext_min;
1457 unsigned long s_ext_max;
1458 unsigned long s_depth_max;
1459 spinlock_t s_ext_stats_lock;
1460 unsigned long s_ext_blocks;
1461 unsigned long s_ext_extents;
1464 /* for buddy allocator */
1465 struct ext4_group_info ** __rcu *s_group_info;
1466 struct inode *s_buddy_cache;
1467 spinlock_t s_md_lock;
1468 unsigned short *s_mb_offsets;
1469 unsigned int *s_mb_maxs;
1470 unsigned int s_group_info_size;
1471 unsigned int s_mb_free_pending;
1472 struct list_head s_freed_data_list; /* List of blocks to be freed
1473 after commit completed */
1476 unsigned long s_stripe;
1477 unsigned int s_mb_stream_request;
1478 unsigned int s_mb_max_to_scan;
1479 unsigned int s_mb_min_to_scan;
1480 unsigned int s_mb_stats;
1481 unsigned int s_mb_order2_reqs;
1482 unsigned int s_mb_group_prealloc;
1483 unsigned int s_max_dir_size_kb;
1484 /* where last allocation was done - for stream allocation */
1485 unsigned long s_mb_last_group;
1486 unsigned long s_mb_last_start;
1488 /* stats for buddy allocator */
1489 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1490 atomic_t s_bal_success; /* we found long enough chunks */
1491 atomic_t s_bal_allocated; /* in blocks */
1492 atomic_t s_bal_ex_scanned; /* total extents scanned */
1493 atomic_t s_bal_goals; /* goal hits */
1494 atomic_t s_bal_breaks; /* too long searches */
1495 atomic_t s_bal_2orders; /* 2^order hits */
1496 spinlock_t s_bal_lock;
1497 unsigned long s_mb_buddies_generated;
1498 unsigned long long s_mb_generation_time;
1499 atomic_t s_mb_lost_chunks;
1500 atomic_t s_mb_preallocated;
1501 atomic_t s_mb_discarded;
1502 atomic_t s_lock_busy;
1504 /* locality groups */
1505 struct ext4_locality_group __percpu *s_locality_groups;
1507 /* for write statistics */
1508 unsigned long s_sectors_written_start;
1509 u64 s_kbytes_written;
1511 /* the size of zero-out chunk */
1512 unsigned int s_extent_max_zeroout_kb;
1514 unsigned int s_log_groups_per_flex;
1515 struct flex_groups * __rcu *s_flex_groups;
1516 ext4_group_t s_flex_groups_allocated;
1518 /* workqueue for reserved extent conversions (buffered io) */
1519 struct workqueue_struct *rsv_conversion_wq;
1521 /* timer for periodic error stats printing */
1522 struct timer_list s_err_report;
1524 /* Lazy inode table initialization info */
1525 struct ext4_li_request *s_li_request;
1526 /* Wait multiplier for lazy initialization thread */
1527 unsigned int s_li_wait_mult;
1529 /* Kernel thread for multiple mount protection */
1530 struct task_struct *s_mmp_tsk;
1532 /* record the last minlen when FITRIM is called. */
1533 atomic_t s_last_trim_minblks;
1535 /* Reference to checksum algorithm driver via cryptoapi */
1536 struct crypto_shash *s_chksum_driver;
1538 /* Precomputed FS UUID checksum for seeding other checksums */
1541 /* Reclaim extents from extent status tree */
1542 struct shrinker s_es_shrinker;
1543 struct list_head s_es_list; /* List of inodes with reclaimable extents */
1545 struct ext4_es_stats s_es_stats;
1546 struct mb_cache *s_ea_block_cache;
1547 struct mb_cache *s_ea_inode_cache;
1548 spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1550 /* Ratelimit ext4 messages. */
1551 struct ratelimit_state s_err_ratelimit_state;
1552 struct ratelimit_state s_warning_ratelimit_state;
1553 struct ratelimit_state s_msg_ratelimit_state;
1555 /* Barrier between changing inodes' journal flags and writepages ops. */
1556 struct percpu_rw_semaphore s_journal_flag_rwsem;
1557 struct dax_device *s_daxdev;
1558 #ifdef CONFIG_EXT4_DEBUG
1559 unsigned long s_simulate_fail;
1563 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1565 return sb->s_fs_info;
1567 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1569 return container_of(inode, struct ext4_inode_info, vfs_inode);
1572 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1574 return ino == EXT4_ROOT_INO ||
1575 (ino >= EXT4_FIRST_INO(sb) &&
1576 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1580 * Returns: sbi->field[index]
1581 * Used to access an array element from the following sbi fields which require
1582 * rcu protection to avoid dereferencing an invalid pointer due to reassignment
1587 #define sbi_array_rcu_deref(sbi, field, index) \
1589 typeof(*((sbi)->field)) _v; \
1591 _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index]; \
1592 rcu_read_unlock(); \
1597 * Simulate_fail codes
1599 #define EXT4_SIM_BBITMAP_EIO 1
1600 #define EXT4_SIM_BBITMAP_CRC 2
1601 #define EXT4_SIM_IBITMAP_EIO 3
1602 #define EXT4_SIM_IBITMAP_CRC 4
1603 #define EXT4_SIM_INODE_EIO 5
1604 #define EXT4_SIM_INODE_CRC 6
1605 #define EXT4_SIM_DIRBLOCK_EIO 7
1606 #define EXT4_SIM_DIRBLOCK_CRC 8
1608 static inline bool ext4_simulate_fail(struct super_block *sb,
1611 #ifdef CONFIG_EXT4_DEBUG
1612 struct ext4_sb_info *sbi = EXT4_SB(sb);
1614 if (unlikely(sbi->s_simulate_fail == code)) {
1615 sbi->s_simulate_fail = 0;
1622 static inline void ext4_simulate_fail_bh(struct super_block *sb,
1623 struct buffer_head *bh,
1626 if (!IS_ERR(bh) && ext4_simulate_fail(sb, code))
1627 clear_buffer_uptodate(bh);
1631 * Error number codes for s_{first,last}_error_errno
1633 * Linux errno numbers are architecture specific, so we need to translate
1634 * them into something which is architecture independent. We don't define
1635 * codes for all errno's; just the ones which are most likely to be the cause
1636 * of an ext4_error() call.
1638 #define EXT4_ERR_UNKNOWN 1
1639 #define EXT4_ERR_EIO 2
1640 #define EXT4_ERR_ENOMEM 3
1641 #define EXT4_ERR_EFSBADCRC 4
1642 #define EXT4_ERR_EFSCORRUPTED 5
1643 #define EXT4_ERR_ENOSPC 6
1644 #define EXT4_ERR_ENOKEY 7
1645 #define EXT4_ERR_EROFS 8
1646 #define EXT4_ERR_EFBIG 9
1647 #define EXT4_ERR_EEXIST 10
1648 #define EXT4_ERR_ERANGE 11
1649 #define EXT4_ERR_EOVERFLOW 12
1650 #define EXT4_ERR_EBUSY 13
1651 #define EXT4_ERR_ENOTDIR 14
1652 #define EXT4_ERR_ENOTEMPTY 15
1653 #define EXT4_ERR_ESHUTDOWN 16
1654 #define EXT4_ERR_EFAULT 17
1657 * Inode dynamic state flags
1660 EXT4_STATE_JDATA, /* journaled data exists */
1661 EXT4_STATE_NEW, /* inode is newly created */
1662 EXT4_STATE_XATTR, /* has in-inode xattrs */
1663 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1664 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1665 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1666 EXT4_STATE_NEWENTRY, /* File just added to dir */
1667 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1668 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
1669 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */
1670 EXT4_STATE_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */
1673 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1674 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1676 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1678 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1680 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1682 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1684 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1687 /* Add these declarations here only so that these functions can be
1688 * found by name. Otherwise, they are very hard to locate. */
1689 static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1690 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1691 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1692 EXT4_INODE_BIT_FNS(flag, flags, 0)
1694 /* Add these declarations here only so that these functions can be
1695 * found by name. Otherwise, they are very hard to locate. */
1696 static inline int ext4_test_inode_state(struct inode *inode, int bit);
1697 static inline void ext4_set_inode_state(struct inode *inode, int bit);
1698 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1699 #if (BITS_PER_LONG < 64)
1700 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1702 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1704 (ei)->i_state_flags = 0;
1707 EXT4_INODE_BIT_FNS(state, flags, 32)
1709 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1711 /* We depend on the fact that callers will set i_flags */
1715 /* Assume that user mode programs are passing in an ext4fs superblock, not
1716 * a kernel struct super_block. This will allow us to call the feature-test
1717 * macros from user land. */
1718 #define EXT4_SB(sb) (sb)
1721 static inline bool ext4_verity_in_progress(struct inode *inode)
1723 return IS_ENABLED(CONFIG_FS_VERITY) &&
1724 ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS);
1727 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1730 * Codes for operating systems
1732 #define EXT4_OS_LINUX 0
1733 #define EXT4_OS_HURD 1
1734 #define EXT4_OS_MASIX 2
1735 #define EXT4_OS_FREEBSD 3
1736 #define EXT4_OS_LITES 4
1741 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1742 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1744 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1745 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1747 #define EXT4_GOOD_OLD_INODE_SIZE 128
1749 #define EXT4_EXTRA_TIMESTAMP_MAX (((s64)1 << 34) - 1 + S32_MIN)
1750 #define EXT4_NON_EXTRA_TIMESTAMP_MAX S32_MAX
1751 #define EXT4_TIMESTAMP_MIN S32_MIN
1754 * Feature set definitions
1757 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1758 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1759 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1760 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1761 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1762 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1763 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
1764 #define EXT4_FEATURE_COMPAT_STABLE_INODES 0x0800
1766 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1767 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1768 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1769 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1770 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1771 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1772 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1773 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1774 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1776 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1777 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1778 * all other data structures' checksums. However, the METADATA_CSUM and
1779 * GDT_CSUM bits are mutually exclusive.
1781 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
1782 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000
1783 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000
1784 #define EXT4_FEATURE_RO_COMPAT_VERITY 0x8000
1786 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1787 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1788 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1789 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1790 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1791 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1792 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1793 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1794 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1795 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1796 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1797 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000
1798 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
1799 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
1800 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000
1801 #define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000
1803 extern void ext4_update_dynamic_rev(struct super_block *sb);
1805 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
1806 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1808 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1809 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
1811 static inline void ext4_set_feature_##name(struct super_block *sb) \
1813 ext4_update_dynamic_rev(sb); \
1814 EXT4_SB(sb)->s_es->s_feature_compat |= \
1815 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1817 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1819 EXT4_SB(sb)->s_es->s_feature_compat &= \
1820 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1823 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
1824 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1826 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1827 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
1829 static inline void ext4_set_feature_##name(struct super_block *sb) \
1831 ext4_update_dynamic_rev(sb); \
1832 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
1833 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1835 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1837 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
1838 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1841 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
1842 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1844 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1845 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
1847 static inline void ext4_set_feature_##name(struct super_block *sb) \
1849 ext4_update_dynamic_rev(sb); \
1850 EXT4_SB(sb)->s_es->s_feature_incompat |= \
1851 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1853 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1855 EXT4_SB(sb)->s_es->s_feature_incompat &= \
1856 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1859 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC)
1860 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES)
1861 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL)
1862 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR)
1863 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE)
1864 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX)
1865 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2)
1866 EXT4_FEATURE_COMPAT_FUNCS(stable_inodes, STABLE_INODES)
1868 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER)
1869 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE)
1870 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR)
1871 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE)
1872 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM)
1873 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK)
1874 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE)
1875 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA)
1876 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC)
1877 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM)
1878 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY)
1879 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT)
1880 EXT4_FEATURE_RO_COMPAT_FUNCS(verity, VERITY)
1882 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION)
1883 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE)
1884 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER)
1885 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV)
1886 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG)
1887 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS)
1888 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT)
1889 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP)
1890 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG)
1891 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE)
1892 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA)
1893 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED)
1894 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR)
1895 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA)
1896 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT)
1897 EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD)
1899 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1900 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1901 EXT4_FEATURE_INCOMPAT_META_BG)
1902 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1903 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1904 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1906 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1907 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1908 EXT4_FEATURE_INCOMPAT_RECOVER| \
1909 EXT4_FEATURE_INCOMPAT_META_BG)
1910 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1911 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1912 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1914 #define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1915 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1916 EXT4_FEATURE_INCOMPAT_RECOVER| \
1917 EXT4_FEATURE_INCOMPAT_META_BG| \
1918 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1919 EXT4_FEATURE_INCOMPAT_64BIT| \
1920 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1921 EXT4_FEATURE_INCOMPAT_EA_INODE| \
1922 EXT4_FEATURE_INCOMPAT_MMP | \
1923 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
1924 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
1925 EXT4_FEATURE_INCOMPAT_CASEFOLD | \
1926 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \
1927 EXT4_FEATURE_INCOMPAT_LARGEDIR)
1928 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1929 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1930 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1931 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1932 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1933 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1934 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1935 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1936 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1937 EXT4_FEATURE_RO_COMPAT_QUOTA |\
1938 EXT4_FEATURE_RO_COMPAT_PROJECT |\
1939 EXT4_FEATURE_RO_COMPAT_VERITY)
1941 #define EXTN_FEATURE_FUNCS(ver) \
1942 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
1944 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1945 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
1947 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
1949 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1950 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
1952 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
1954 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1955 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
1958 EXTN_FEATURE_FUNCS(2)
1959 EXTN_FEATURE_FUNCS(3)
1960 EXTN_FEATURE_FUNCS(4)
1962 static inline bool ext4_has_compat_features(struct super_block *sb)
1964 return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
1966 static inline bool ext4_has_ro_compat_features(struct super_block *sb)
1968 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
1970 static inline bool ext4_has_incompat_features(struct super_block *sb)
1972 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
1978 #define EXT4_FLAGS_RESIZING 0
1979 #define EXT4_FLAGS_SHUTDOWN 1
1981 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi)
1983 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags);
1988 * Default values for user and/or group using reserved blocks
1990 #define EXT4_DEF_RESUID 0
1991 #define EXT4_DEF_RESGID 0
1994 * Default project ID
1996 #define EXT4_DEF_PROJID 0
1998 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
2001 * Default mount options
2003 #define EXT4_DEFM_DEBUG 0x0001
2004 #define EXT4_DEFM_BSDGROUPS 0x0002
2005 #define EXT4_DEFM_XATTR_USER 0x0004
2006 #define EXT4_DEFM_ACL 0x0008
2007 #define EXT4_DEFM_UID16 0x0010
2008 #define EXT4_DEFM_JMODE 0x0060
2009 #define EXT4_DEFM_JMODE_DATA 0x0020
2010 #define EXT4_DEFM_JMODE_ORDERED 0x0040
2011 #define EXT4_DEFM_JMODE_WBACK 0x0060
2012 #define EXT4_DEFM_NOBARRIER 0x0100
2013 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
2014 #define EXT4_DEFM_DISCARD 0x0400
2015 #define EXT4_DEFM_NODELALLOC 0x0800
2018 * Default journal batch times
2020 #define EXT4_DEF_MIN_BATCH_TIME 0
2021 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
2024 * Minimum number of groups in a flexgroup before we separate out
2025 * directories into the first block group of a flexgroup
2027 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
2030 * Structure of a directory entry
2032 #define EXT4_NAME_LEN 255
2034 struct ext4_dir_entry {
2035 __le32 inode; /* Inode number */
2036 __le16 rec_len; /* Directory entry length */
2037 __le16 name_len; /* Name length */
2038 char name[EXT4_NAME_LEN]; /* File name */
2042 * The new version of the directory entry. Since EXT4 structures are
2043 * stored in intel byte order, and the name_len field could never be
2044 * bigger than 255 chars, it's safe to reclaim the extra byte for the
2047 struct ext4_dir_entry_2 {
2048 __le32 inode; /* Inode number */
2049 __le16 rec_len; /* Directory entry length */
2050 __u8 name_len; /* Name length */
2052 char name[EXT4_NAME_LEN]; /* File name */
2056 * This is a bogus directory entry at the end of each leaf block that
2057 * records checksums.
2059 struct ext4_dir_entry_tail {
2060 __le32 det_reserved_zero1; /* Pretend to be unused */
2061 __le16 det_rec_len; /* 12 */
2062 __u8 det_reserved_zero2; /* Zero name length */
2063 __u8 det_reserved_ft; /* 0xDE, fake file type */
2064 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
2067 #define EXT4_DIRENT_TAIL(block, blocksize) \
2068 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
2070 sizeof(struct ext4_dir_entry_tail))))
2073 * Ext4 directory file types. Only the low 3 bits are used. The
2074 * other bits are reserved for now.
2076 #define EXT4_FT_UNKNOWN 0
2077 #define EXT4_FT_REG_FILE 1
2078 #define EXT4_FT_DIR 2
2079 #define EXT4_FT_CHRDEV 3
2080 #define EXT4_FT_BLKDEV 4
2081 #define EXT4_FT_FIFO 5
2082 #define EXT4_FT_SOCK 6
2083 #define EXT4_FT_SYMLINK 7
2085 #define EXT4_FT_MAX 8
2087 #define EXT4_FT_DIR_CSUM 0xDE
2090 * EXT4_DIR_PAD defines the directory entries boundaries
2092 * NOTE: It must be a multiple of 4
2094 #define EXT4_DIR_PAD 4
2095 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
2096 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
2098 #define EXT4_MAX_REC_LEN ((1<<16)-1)
2101 * If we ever get support for fs block sizes > page_size, we'll need
2102 * to remove the #if statements in the next two functions...
2104 static inline unsigned int
2105 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
2107 unsigned len = le16_to_cpu(dlen);
2109 #if (PAGE_SIZE >= 65536)
2110 if (len == EXT4_MAX_REC_LEN || len == 0)
2112 return (len & 65532) | ((len & 3) << 16);
2118 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
2120 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
2122 #if (PAGE_SIZE >= 65536)
2124 return cpu_to_le16(len);
2125 if (len == blocksize) {
2126 if (blocksize == 65536)
2127 return cpu_to_le16(EXT4_MAX_REC_LEN);
2129 return cpu_to_le16(0);
2131 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
2133 return cpu_to_le16(len);
2138 * Hash Tree Directory indexing
2139 * (c) Daniel Phillips, 2001
2142 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
2143 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2144 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \
2145 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir)))
2146 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2148 /* Legal values for the dx_root hash_version field: */
2150 #define DX_HASH_LEGACY 0
2151 #define DX_HASH_HALF_MD4 1
2152 #define DX_HASH_TEA 2
2153 #define DX_HASH_LEGACY_UNSIGNED 3
2154 #define DX_HASH_HALF_MD4_UNSIGNED 4
2155 #define DX_HASH_TEA_UNSIGNED 5
2157 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2158 const void *address, unsigned int length)
2161 struct shash_desc shash;
2165 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2167 desc.shash.tfm = sbi->s_chksum_driver;
2168 *(u32 *)desc.ctx = crc;
2170 BUG_ON(crypto_shash_update(&desc.shash, address, length));
2172 return *(u32 *)desc.ctx;
2177 /* hash info structure used by the directory hash */
2187 /* 32 and 64 bit signed EOF for dx directories */
2188 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
2189 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
2193 * Control parameters used by ext4_htree_next_block
2195 #define HASH_NB_ALWAYS 1
2197 struct ext4_filename {
2198 const struct qstr *usr_fname;
2199 struct fscrypt_str disk_name;
2200 struct dx_hash_info hinfo;
2201 #ifdef CONFIG_FS_ENCRYPTION
2202 struct fscrypt_str crypto_buf;
2204 #ifdef CONFIG_UNICODE
2205 struct fscrypt_str cf_name;
2209 #define fname_name(p) ((p)->disk_name.name)
2210 #define fname_len(p) ((p)->disk_name.len)
2213 * Describe an inode's exact location on disk and in memory
2217 struct buffer_head *bh;
2218 unsigned long offset;
2219 ext4_group_t block_group;
2222 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2224 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2227 static inline bool ext4_is_quota_file(struct inode *inode)
2229 return IS_NOQUOTA(inode) &&
2230 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL);
2234 * This structure is stuffed into the struct file's private_data field
2235 * for directories. It is where we put information so that we can do
2236 * readdir operations in hash tree order.
2238 struct dir_private_info {
2239 struct rb_root root;
2240 struct rb_node *curr_node;
2241 struct fname *extra_fname;
2244 __u32 curr_minor_hash;
2248 /* calculate the first block number of the group */
2249 static inline ext4_fsblk_t
2250 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2252 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2253 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2257 * Special error return code only used by dx_probe() and its callers.
2259 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
2261 /* htree levels for ext4 */
2262 #define EXT4_HTREE_LEVEL_COMPAT 2
2263 #define EXT4_HTREE_LEVEL 3
2265 static inline int ext4_dir_htree_level(struct super_block *sb)
2267 return ext4_has_feature_largedir(sb) ?
2268 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT;
2272 * Timeout and state flag for lazy initialization inode thread.
2274 #define EXT4_DEF_LI_WAIT_MULT 10
2275 #define EXT4_DEF_LI_MAX_START_DELAY 5
2276 #define EXT4_LAZYINIT_QUIT 0x0001
2277 #define EXT4_LAZYINIT_RUNNING 0x0002
2280 * Lazy inode table initialization info
2282 struct ext4_lazy_init {
2283 unsigned long li_state;
2284 struct list_head li_request_list;
2285 struct mutex li_list_mtx;
2288 struct ext4_li_request {
2289 struct super_block *lr_super;
2290 struct ext4_sb_info *lr_sbi;
2291 ext4_group_t lr_next_group;
2292 struct list_head lr_request;
2293 unsigned long lr_next_sched;
2294 unsigned long lr_timeout;
2297 struct ext4_features {
2298 struct kobject f_kobj;
2299 struct completion f_kobj_unregister;
2303 * This structure will be used for multiple mount protection. It will be
2304 * written into the block number saved in the s_mmp_block field in the
2305 * superblock. Programs that check MMP should assume that if
2306 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2307 * to use the filesystem, regardless of how old the timestamp is.
2309 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
2310 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2311 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
2312 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
2315 __le32 mmp_magic; /* Magic number for MMP */
2316 __le32 mmp_seq; /* Sequence no. updated periodically */
2319 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2320 * purposes and do not affect the correctness of the algorithm
2322 __le64 mmp_time; /* Time last updated */
2323 char mmp_nodename[64]; /* Node which last updated MMP block */
2324 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
2327 * mmp_check_interval is used to verify if the MMP block has been
2328 * updated on the block device. The value is updated based on the
2329 * maximum time to write the MMP block during an update cycle.
2331 __le16 mmp_check_interval;
2334 __le32 mmp_pad2[226];
2335 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
2338 /* arguments passed to the mmp thread */
2340 struct buffer_head *bh; /* bh from initial read_mmp_block() */
2341 struct super_block *sb; /* super block of the fs */
2345 * Check interval multiplier
2346 * The MMP block is written every update interval and initially checked every
2347 * update interval x the multiplier (the value is then adapted based on the
2348 * write latency). The reason is that writes can be delayed under load and we
2349 * don't want readers to incorrectly assume that the filesystem is no longer
2352 #define EXT4_MMP_CHECK_MULT 2UL
2355 * Minimum interval for MMP checking in seconds.
2357 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
2360 * Maximum interval for MMP checking in seconds.
2362 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
2365 * Function prototypes
2369 * Ok, these declarations are also in <linux/kernel.h> but none of the
2370 * ext4 source programs needs to include it so they are duplicated here.
2372 # define NORET_TYPE /**/
2373 # define ATTRIB_NORET __attribute__((noreturn))
2374 # define NORET_AND noreturn,
2377 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2378 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2379 struct ext4_group_desc *gdp,
2380 struct buffer_head *bh, int sz);
2381 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2382 struct ext4_group_desc *gdp,
2383 struct buffer_head *bh, int sz);
2384 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2385 struct ext4_group_desc *gdp,
2386 struct buffer_head *bh);
2387 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2388 struct ext4_group_desc *gdp,
2389 struct buffer_head *bh);
2392 extern void ext4_get_group_no_and_offset(struct super_block *sb,
2393 ext4_fsblk_t blocknr,
2394 ext4_group_t *blockgrpp,
2395 ext4_grpblk_t *offsetp);
2396 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2397 ext4_fsblk_t block);
2399 extern unsigned int ext4_block_group(struct super_block *sb,
2400 ext4_fsblk_t blocknr);
2401 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2402 ext4_fsblk_t blocknr);
2403 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2404 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2405 ext4_group_t group);
2406 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2409 unsigned long *count,
2411 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2412 s64 nclusters, unsigned int flags);
2413 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2414 extern void ext4_check_blocks_bitmap(struct super_block *);
2415 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2416 ext4_group_t block_group,
2417 struct buffer_head ** bh);
2418 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2420 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2421 ext4_group_t block_group);
2422 extern int ext4_wait_block_bitmap(struct super_block *sb,
2423 ext4_group_t block_group,
2424 struct buffer_head *bh);
2425 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2426 ext4_group_t block_group);
2427 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2428 ext4_group_t block_group,
2429 struct ext4_group_desc *gdp);
2430 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2432 #ifdef CONFIG_UNICODE
2433 extern void ext4_fname_setup_ci_filename(struct inode *dir,
2434 const struct qstr *iname,
2435 struct fscrypt_str *fname);
2438 #ifdef CONFIG_FS_ENCRYPTION
2439 static inline void ext4_fname_from_fscrypt_name(struct ext4_filename *dst,
2440 const struct fscrypt_name *src)
2442 memset(dst, 0, sizeof(*dst));
2444 dst->usr_fname = src->usr_fname;
2445 dst->disk_name = src->disk_name;
2446 dst->hinfo.hash = src->hash;
2447 dst->hinfo.minor_hash = src->minor_hash;
2448 dst->crypto_buf = src->crypto_buf;
2451 static inline int ext4_fname_setup_filename(struct inode *dir,
2452 const struct qstr *iname,
2454 struct ext4_filename *fname)
2456 struct fscrypt_name name;
2459 err = fscrypt_setup_filename(dir, iname, lookup, &name);
2463 ext4_fname_from_fscrypt_name(fname, &name);
2465 #ifdef CONFIG_UNICODE
2466 ext4_fname_setup_ci_filename(dir, iname, &fname->cf_name);
2471 static inline int ext4_fname_prepare_lookup(struct inode *dir,
2472 struct dentry *dentry,
2473 struct ext4_filename *fname)
2475 struct fscrypt_name name;
2478 err = fscrypt_prepare_lookup(dir, dentry, &name);
2482 ext4_fname_from_fscrypt_name(fname, &name);
2484 #ifdef CONFIG_UNICODE
2485 ext4_fname_setup_ci_filename(dir, &dentry->d_name, &fname->cf_name);
2490 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2492 struct fscrypt_name name;
2494 name.crypto_buf = fname->crypto_buf;
2495 fscrypt_free_filename(&name);
2497 fname->crypto_buf.name = NULL;
2498 fname->usr_fname = NULL;
2499 fname->disk_name.name = NULL;
2501 #ifdef CONFIG_UNICODE
2502 kfree(fname->cf_name.name);
2503 fname->cf_name.name = NULL;
2506 #else /* !CONFIG_FS_ENCRYPTION */
2507 static inline int ext4_fname_setup_filename(struct inode *dir,
2508 const struct qstr *iname,
2510 struct ext4_filename *fname)
2512 fname->usr_fname = iname;
2513 fname->disk_name.name = (unsigned char *) iname->name;
2514 fname->disk_name.len = iname->len;
2516 #ifdef CONFIG_UNICODE
2517 ext4_fname_setup_ci_filename(dir, iname, &fname->cf_name);
2523 static inline int ext4_fname_prepare_lookup(struct inode *dir,
2524 struct dentry *dentry,
2525 struct ext4_filename *fname)
2527 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname);
2530 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2532 #ifdef CONFIG_UNICODE
2533 kfree(fname->cf_name.name);
2534 fname->cf_name.name = NULL;
2537 #endif /* !CONFIG_FS_ENCRYPTION */
2540 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2542 struct ext4_dir_entry_2 *,
2543 struct buffer_head *, char *, int,
2545 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2546 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2547 (de), (bh), (buf), (size), (offset)))
2548 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2550 struct ext4_dir_entry_2 *dirent,
2551 struct fscrypt_str *ent_name);
2552 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2553 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2554 struct buffer_head *bh,
2555 void *buf, int buf_size,
2556 struct ext4_filename *fname,
2557 struct ext4_dir_entry_2 **dest_de);
2558 void ext4_insert_dentry(struct inode *inode,
2559 struct ext4_dir_entry_2 *de,
2561 struct ext4_filename *fname);
2562 static inline void ext4_update_dx_flag(struct inode *inode)
2564 if (!ext4_has_feature_dir_index(inode->i_sb)) {
2565 /* ext4_iget() should have caught this... */
2566 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb));
2567 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2570 static const unsigned char ext4_filetype_table[] = {
2571 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2574 static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2576 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2579 return ext4_filetype_table[filetype];
2581 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2582 void *buf, int buf_size);
2585 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2588 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len,
2589 struct dx_hash_info *hinfo);
2592 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2593 const struct qstr *qstr, __u32 goal,
2594 uid_t *owner, __u32 i_flags,
2595 int handle_type, unsigned int line_no,
2598 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \
2599 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2601 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2603 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2604 0, (type), __LINE__, (nblocks))
2607 extern void ext4_free_inode(handle_t *, struct inode *);
2608 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2609 extern unsigned long ext4_count_free_inodes(struct super_block *);
2610 extern unsigned long ext4_count_dirs(struct super_block *);
2611 extern void ext4_check_inodes_bitmap(struct super_block *);
2612 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2613 extern int ext4_init_inode_table(struct super_block *sb,
2614 ext4_group_t group, int barrier);
2615 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2618 extern const struct seq_operations ext4_mb_seq_groups_ops;
2619 extern long ext4_mb_stats;
2620 extern long ext4_mb_max_to_scan;
2621 extern int ext4_mb_init(struct super_block *);
2622 extern int ext4_mb_release(struct super_block *);
2623 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2624 struct ext4_allocation_request *, int *);
2625 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2626 extern void ext4_discard_preallocations(struct inode *);
2627 extern int __init ext4_init_mballoc(void);
2628 extern void ext4_exit_mballoc(void);
2629 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2630 struct buffer_head *bh, ext4_fsblk_t block,
2631 unsigned long count, int flags);
2632 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2633 ext4_group_t ngroups);
2634 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2635 ext4_group_t i, struct ext4_group_desc *desc);
2636 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2637 ext4_fsblk_t block, unsigned long count);
2638 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2639 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid);
2642 int ext4_inode_is_fast_symlink(struct inode *inode);
2643 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2644 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2645 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count,
2646 bool wait, struct buffer_head **bhs);
2647 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2648 struct buffer_head *bh_result, int create);
2649 int ext4_get_block(struct inode *inode, sector_t iblock,
2650 struct buffer_head *bh_result, int create);
2651 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2652 struct buffer_head *bh, int create);
2653 int ext4_walk_page_buffers(handle_t *handle,
2654 struct buffer_head *head,
2658 int (*fn)(handle_t *handle,
2659 struct buffer_head *bh));
2660 int do_journal_get_write_access(handle_t *handle,
2661 struct buffer_head *bh);
2662 #define FALL_BACK_TO_NONDELALLOC 1
2663 #define CONVERT_INLINE_DATA 2
2666 EXT4_IGET_NORMAL = 0,
2667 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */
2668 EXT4_IGET_HANDLE = 0x0002 /* Inode # is from a handle */
2671 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
2672 ext4_iget_flags flags, const char *function,
2675 #define ext4_iget(sb, ino, flags) \
2676 __ext4_iget((sb), (ino), (flags), __func__, __LINE__)
2678 extern int ext4_write_inode(struct inode *, struct writeback_control *);
2679 extern int ext4_setattr(struct dentry *, struct iattr *);
2680 extern int ext4_getattr(const struct path *, struct kstat *, u32, unsigned int);
2681 extern void ext4_evict_inode(struct inode *);
2682 extern void ext4_clear_inode(struct inode *);
2683 extern int ext4_file_getattr(const struct path *, struct kstat *, u32, unsigned int);
2684 extern int ext4_sync_inode(handle_t *, struct inode *);
2685 extern void ext4_dirty_inode(struct inode *, int);
2686 extern int ext4_change_inode_journal_flag(struct inode *, int);
2687 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2688 extern int ext4_inode_attach_jinode(struct inode *inode);
2689 extern int ext4_can_truncate(struct inode *inode);
2690 extern int ext4_truncate(struct inode *);
2691 extern int ext4_break_layouts(struct inode *);
2692 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2693 extern void ext4_set_inode_flags(struct inode *);
2694 extern int ext4_alloc_da_blocks(struct inode *inode);
2695 extern void ext4_set_aops(struct inode *inode);
2696 extern int ext4_writepage_trans_blocks(struct inode *);
2697 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2698 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2699 loff_t lstart, loff_t lend);
2700 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf);
2701 extern vm_fault_t ext4_filemap_fault(struct vm_fault *vmf);
2702 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2703 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
2704 extern void ext4_da_release_space(struct inode *inode, int to_free);
2705 extern void ext4_da_update_reserve_space(struct inode *inode,
2706 int used, int quota_claim);
2707 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
2708 ext4_fsblk_t pblk, ext4_lblk_t len);
2711 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2712 struct ext4_map_blocks *map, int flags);
2713 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2714 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2715 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2716 ext4_lblk_t start, ext4_lblk_t end);
2719 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2720 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2723 extern int ext4_ext_migrate(struct inode *);
2724 extern int ext4_ind_migrate(struct inode *inode);
2727 extern int ext4_dirblock_csum_verify(struct inode *inode,
2728 struct buffer_head *bh);
2729 extern int ext4_orphan_add(handle_t *, struct inode *);
2730 extern int ext4_orphan_del(handle_t *, struct inode *);
2731 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2732 __u32 start_minor_hash, __u32 *next_hash);
2733 extern int ext4_search_dir(struct buffer_head *bh,
2737 struct ext4_filename *fname,
2738 unsigned int offset,
2739 struct ext4_dir_entry_2 **res_dir);
2740 extern int ext4_generic_delete_entry(handle_t *handle,
2742 struct ext4_dir_entry_2 *de_del,
2743 struct buffer_head *bh,
2747 extern bool ext4_empty_dir(struct inode *inode);
2750 extern void ext4_kvfree_array_rcu(void *to_free);
2751 extern int ext4_group_add(struct super_block *sb,
2752 struct ext4_new_group_data *input);
2753 extern int ext4_group_extend(struct super_block *sb,
2754 struct ext4_super_block *es,
2755 ext4_fsblk_t n_blocks_count);
2756 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2759 extern struct buffer_head *ext4_sb_bread(struct super_block *sb,
2760 sector_t block, int op_flags);
2761 extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
2762 extern int ext4_calculate_overhead(struct super_block *sb);
2763 extern void ext4_superblock_csum_set(struct super_block *sb);
2764 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2765 ext4_group_t ngroup);
2766 extern const char *ext4_decode_error(struct super_block *sb, int errno,
2768 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb,
2769 ext4_group_t block_group,
2770 unsigned int flags);
2771 extern void ext4_set_errno(struct super_block *sb, int err);
2773 extern __printf(4, 5)
2774 void __ext4_error(struct super_block *, const char *, unsigned int,
2776 extern __printf(5, 6)
2777 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2779 extern __printf(5, 6)
2780 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2782 extern void __ext4_std_error(struct super_block *, const char *,
2784 extern __printf(4, 5)
2785 void __ext4_abort(struct super_block *, const char *, unsigned int,
2787 extern __printf(4, 5)
2788 void __ext4_warning(struct super_block *, const char *, unsigned int,
2790 extern __printf(4, 5)
2791 void __ext4_warning_inode(const struct inode *inode, const char *function,
2792 unsigned int line, const char *fmt, ...);
2793 extern __printf(3, 4)
2794 void __ext4_msg(struct super_block *, const char *, const char *, ...);
2795 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2796 const char *, unsigned int, const char *);
2797 extern __printf(7, 8)
2798 void __ext4_grp_locked_error(const char *, unsigned int,
2799 struct super_block *, ext4_group_t,
2800 unsigned long, ext4_fsblk_t,
2803 #define EXT4_ERROR_INODE(inode, fmt, a...) \
2804 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
2806 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
2807 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
2809 #define EXT4_ERROR_FILE(file, block, fmt, a...) \
2810 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
2812 #ifdef CONFIG_PRINTK
2814 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2815 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__)
2816 #define ext4_error_file(file, func, line, block, fmt, ...) \
2817 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2818 #define ext4_error(sb, fmt, ...) \
2819 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2820 #define ext4_abort(sb, fmt, ...) \
2821 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2822 #define ext4_warning(sb, fmt, ...) \
2823 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2824 #define ext4_warning_inode(inode, fmt, ...) \
2825 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
2826 #define ext4_msg(sb, level, fmt, ...) \
2827 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2828 #define dump_mmp_msg(sb, mmp, msg) \
2829 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2830 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2831 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2836 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2838 no_printk(fmt, ##__VA_ARGS__); \
2839 __ext4_error_inode(inode, "", 0, block, " "); \
2841 #define ext4_error_file(file, func, line, block, fmt, ...) \
2843 no_printk(fmt, ##__VA_ARGS__); \
2844 __ext4_error_file(file, "", 0, block, " "); \
2846 #define ext4_error(sb, fmt, ...) \
2848 no_printk(fmt, ##__VA_ARGS__); \
2849 __ext4_error(sb, "", 0, " "); \
2851 #define ext4_abort(sb, fmt, ...) \
2853 no_printk(fmt, ##__VA_ARGS__); \
2854 __ext4_abort(sb, "", 0, " "); \
2856 #define ext4_warning(sb, fmt, ...) \
2858 no_printk(fmt, ##__VA_ARGS__); \
2859 __ext4_warning(sb, "", 0, " "); \
2861 #define ext4_warning_inode(inode, fmt, ...) \
2863 no_printk(fmt, ##__VA_ARGS__); \
2864 __ext4_warning_inode(inode, "", 0, " "); \
2866 #define ext4_msg(sb, level, fmt, ...) \
2868 no_printk(fmt, ##__VA_ARGS__); \
2869 __ext4_msg(sb, "", " "); \
2871 #define dump_mmp_msg(sb, mmp, msg) \
2872 __dump_mmp_msg(sb, mmp, "", 0, "")
2873 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2875 no_printk(fmt, ##__VA_ARGS__); \
2876 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
2881 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2883 extern int ext4_update_rocompat_feature(handle_t *handle,
2884 struct super_block *sb, __u32 rocompat);
2885 extern int ext4_update_incompat_feature(handle_t *handle,
2886 struct super_block *sb, __u32 incompat);
2887 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2888 struct ext4_group_desc *bg);
2889 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2890 struct ext4_group_desc *bg);
2891 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2892 struct ext4_group_desc *bg);
2893 extern __u32 ext4_free_group_clusters(struct super_block *sb,
2894 struct ext4_group_desc *bg);
2895 extern __u32 ext4_free_inodes_count(struct super_block *sb,
2896 struct ext4_group_desc *bg);
2897 extern __u32 ext4_used_dirs_count(struct super_block *sb,
2898 struct ext4_group_desc *bg);
2899 extern __u32 ext4_itable_unused_count(struct super_block *sb,
2900 struct ext4_group_desc *bg);
2901 extern void ext4_block_bitmap_set(struct super_block *sb,
2902 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2903 extern void ext4_inode_bitmap_set(struct super_block *sb,
2904 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2905 extern void ext4_inode_table_set(struct super_block *sb,
2906 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2907 extern void ext4_free_group_clusters_set(struct super_block *sb,
2908 struct ext4_group_desc *bg,
2910 extern void ext4_free_inodes_set(struct super_block *sb,
2911 struct ext4_group_desc *bg, __u32 count);
2912 extern void ext4_used_dirs_set(struct super_block *sb,
2913 struct ext4_group_desc *bg, __u32 count);
2914 extern void ext4_itable_unused_set(struct super_block *sb,
2915 struct ext4_group_desc *bg, __u32 count);
2916 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2917 struct ext4_group_desc *gdp);
2918 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2919 struct ext4_group_desc *gdp);
2920 extern int ext4_register_li_request(struct super_block *sb,
2921 ext4_group_t first_not_zeroed);
2923 static inline int ext4_has_metadata_csum(struct super_block *sb)
2925 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
2926 !EXT4_SB(sb)->s_chksum_driver);
2928 return ext4_has_feature_metadata_csum(sb) &&
2929 (EXT4_SB(sb)->s_chksum_driver != NULL);
2932 static inline int ext4_has_group_desc_csum(struct super_block *sb)
2934 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb);
2937 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2939 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2940 le32_to_cpu(es->s_blocks_count_lo);
2943 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2945 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2946 le32_to_cpu(es->s_r_blocks_count_lo);
2949 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2951 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2952 le32_to_cpu(es->s_free_blocks_count_lo);
2955 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2958 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2959 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2962 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2965 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2966 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2969 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2972 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2973 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2976 static inline loff_t ext4_isize(struct super_block *sb,
2977 struct ext4_inode *raw_inode)
2979 if (ext4_has_feature_largedir(sb) ||
2980 S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2981 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2982 le32_to_cpu(raw_inode->i_size_lo);
2984 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2987 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2989 raw_inode->i_size_lo = cpu_to_le32(i_size);
2990 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2994 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2997 struct ext4_group_info **grp_info;
2998 long indexv, indexh;
2999 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
3000 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
3001 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
3002 grp_info = sbi_array_rcu_deref(EXT4_SB(sb), s_group_info, indexv);
3003 return grp_info[indexh];
3007 * Reading s_groups_count requires using smp_rmb() afterwards. See
3008 * the locking protocol documented in the comments of ext4_group_add()
3011 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
3013 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
3019 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
3020 ext4_group_t block_group)
3022 return block_group >> sbi->s_log_groups_per_flex;
3025 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
3027 return 1 << sbi->s_log_groups_per_flex;
3030 #define ext4_std_error(sb, errno) \
3033 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
3037 /* Each CPU can accumulate percpu_counter_batch clusters in their local
3038 * counters. So we need to make sure we have free clusters more
3039 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
3041 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
3043 #define EXT4_FREECLUSTERS_WATERMARK 0
3046 /* Update i_disksize. Requires i_mutex to avoid races with truncate */
3047 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
3049 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
3050 !inode_is_locked(inode));
3051 down_write(&EXT4_I(inode)->i_data_sem);
3052 if (newsize > EXT4_I(inode)->i_disksize)
3053 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize);
3054 up_write(&EXT4_I(inode)->i_data_sem);
3057 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
3058 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
3062 if (newsize > inode->i_size) {
3063 i_size_write(inode, newsize);
3066 if (newsize > EXT4_I(inode)->i_disksize) {
3067 ext4_update_i_disksize(inode, newsize);
3073 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
3076 struct ext4_group_info {
3077 unsigned long bb_state;
3078 struct rb_root bb_free_root;
3079 ext4_grpblk_t bb_first_free; /* first free block */
3080 ext4_grpblk_t bb_free; /* total free blocks */
3081 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
3082 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
3083 struct list_head bb_prealloc_list;
3087 struct rw_semaphore alloc_sem;
3088 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
3089 * regions, index is order.
3090 * bb_counters[3] = 5 means
3091 * 5 free 8-block regions. */
3094 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
3095 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
3096 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
3097 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
3098 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \
3099 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT)
3100 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \
3101 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT)
3103 #define EXT4_MB_GRP_NEED_INIT(grp) \
3104 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
3105 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
3106 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3107 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
3108 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3110 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
3111 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3112 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
3113 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3114 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
3115 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3117 #define EXT4_MAX_CONTENTION 8
3118 #define EXT4_CONTENTION_THRESHOLD 2
3120 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
3123 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
3127 * Returns true if the filesystem is busy enough that attempts to
3128 * access the block group locks has run into contention.
3130 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
3132 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
3135 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
3137 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
3138 if (spin_trylock(lock))
3140 * We're able to grab the lock right away, so drop the
3141 * lock contention counter.
3143 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
3146 * The lock is busy, so bump the contention counter,
3147 * and then wait on the spin lock.
3149 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
3150 EXT4_MAX_CONTENTION);
3155 static inline void ext4_unlock_group(struct super_block *sb,
3158 spin_unlock(ext4_group_lock_ptr(sb, group));
3162 * Block validity checking
3164 #define ext4_check_indirect_blockref(inode, bh) \
3165 ext4_check_blockref(__func__, __LINE__, inode, \
3166 (__le32 *)(bh)->b_data, \
3167 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
3169 #define ext4_ind_check_inode(inode) \
3170 ext4_check_blockref(__func__, __LINE__, inode, \
3171 EXT4_I(inode)->i_data, \
3175 * Inodes and files operations
3179 extern const struct file_operations ext4_dir_operations;
3181 #ifdef CONFIG_UNICODE
3182 extern const struct dentry_operations ext4_dentry_ops;
3186 extern const struct inode_operations ext4_file_inode_operations;
3187 extern const struct file_operations ext4_file_operations;
3188 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
3191 extern int ext4_get_max_inline_size(struct inode *inode);
3192 extern int ext4_find_inline_data_nolock(struct inode *inode);
3193 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
3195 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
3197 extern int ext4_readpage_inline(struct inode *inode, struct page *page);
3198 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3199 struct inode *inode,
3200 loff_t pos, unsigned len,
3202 struct page **pagep);
3203 extern int ext4_write_inline_data_end(struct inode *inode,
3204 loff_t pos, unsigned len,
3207 extern struct buffer_head *
3208 ext4_journalled_write_inline_data(struct inode *inode,
3211 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
3212 struct inode *inode,
3213 loff_t pos, unsigned len,
3215 struct page **pagep,
3217 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
3218 unsigned len, unsigned copied,
3220 extern int ext4_try_add_inline_entry(handle_t *handle,
3221 struct ext4_filename *fname,
3222 struct inode *dir, struct inode *inode);
3223 extern int ext4_try_create_inline_dir(handle_t *handle,
3224 struct inode *parent,
3225 struct inode *inode);
3226 extern int ext4_read_inline_dir(struct file *filp,
3227 struct dir_context *ctx,
3228 int *has_inline_data);
3229 extern int ext4_inlinedir_to_tree(struct file *dir_file,
3230 struct inode *dir, ext4_lblk_t block,
3231 struct dx_hash_info *hinfo,
3232 __u32 start_hash, __u32 start_minor_hash,
3233 int *has_inline_data);
3234 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3235 struct ext4_filename *fname,
3236 struct ext4_dir_entry_2 **res_dir,
3237 int *has_inline_data);
3238 extern int ext4_delete_inline_entry(handle_t *handle,
3240 struct ext4_dir_entry_2 *de_del,
3241 struct buffer_head *bh,
3242 int *has_inline_data);
3243 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3244 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3245 struct ext4_dir_entry_2 **parent_de,
3247 extern int ext4_inline_data_fiemap(struct inode *inode,
3248 struct fiemap_extent_info *fieinfo,
3249 int *has_inline, __u64 start, __u64 len);
3252 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap);
3254 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3256 extern int ext4_convert_inline_data(struct inode *inode);
3258 static inline int ext4_has_inline_data(struct inode *inode)
3260 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3261 EXT4_I(inode)->i_inline_off;
3265 extern const struct inode_operations ext4_dir_inode_operations;
3266 extern const struct inode_operations ext4_special_inode_operations;
3267 extern struct dentry *ext4_get_parent(struct dentry *child);
3268 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3269 struct ext4_dir_entry_2 *de,
3270 int blocksize, int csum_size,
3271 unsigned int parent_ino, int dotdot_real_len);
3272 extern void ext4_initialize_dirent_tail(struct buffer_head *bh,
3273 unsigned int blocksize);
3274 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode,
3275 struct buffer_head *bh);
3276 extern int ext4_ci_compare(const struct inode *parent,
3277 const struct qstr *fname,
3278 const struct qstr *entry, bool quick);
3281 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = {
3282 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
3283 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
3284 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
3285 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
3286 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
3287 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
3288 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
3291 static inline void ext4_set_de_type(struct super_block *sb,
3292 struct ext4_dir_entry_2 *de,
3294 if (ext4_has_feature_filetype(sb))
3295 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3299 extern int ext4_mpage_readpages(struct address_space *mapping,
3300 struct list_head *pages, struct page *page,
3301 unsigned nr_pages, bool is_readahead);
3302 extern int __init ext4_init_post_read_processing(void);
3303 extern void ext4_exit_post_read_processing(void);
3306 extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3307 extern const struct inode_operations ext4_symlink_inode_operations;
3308 extern const struct inode_operations ext4_fast_symlink_inode_operations;
3311 extern int ext4_register_sysfs(struct super_block *sb);
3312 extern void ext4_unregister_sysfs(struct super_block *sb);
3313 extern int __init ext4_init_sysfs(void);
3314 extern void ext4_exit_sysfs(void);
3316 /* block_validity */
3317 extern void ext4_release_system_zone(struct super_block *sb);
3318 extern int ext4_setup_system_zone(struct super_block *sb);
3319 extern int __init ext4_init_system_zone(void);
3320 extern void ext4_exit_system_zone(void);
3321 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
3322 ext4_fsblk_t start_blk,
3323 unsigned int count);
3324 extern int ext4_check_blockref(const char *, unsigned int,
3325 struct inode *, __le32 *, unsigned int);
3328 struct ext4_ext_path;
3332 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3335 #define EXT_MAX_BLOCKS 0xffffffff
3337 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
3338 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3339 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3340 struct ext4_map_blocks *map, int flags);
3341 extern int ext4_ext_truncate(handle_t *, struct inode *);
3342 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3344 extern void ext4_ext_init(struct super_block *);
3345 extern void ext4_ext_release(struct super_block *);
3346 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3348 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3349 loff_t offset, ssize_t len);
3350 extern int ext4_convert_unwritten_io_end_vec(handle_t *handle,
3351 ext4_io_end_t *io_end);
3352 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3353 struct ext4_map_blocks *map, int flags);
3354 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3356 struct ext4_ext_path *path);
3357 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3358 struct ext4_ext_path **,
3359 struct ext4_extent *, int);
3360 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3361 struct ext4_ext_path **,
3363 extern void ext4_ext_drop_refs(struct ext4_ext_path *);
3364 extern int ext4_ext_check_inode(struct inode *inode);
3365 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3366 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3367 __u64 start, __u64 len);
3368 extern int ext4_get_es_cache(struct inode *inode,
3369 struct fiemap_extent_info *fieinfo,
3370 __u64 start, __u64 len);
3371 extern int ext4_ext_precache(struct inode *inode);
3372 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3373 struct inode *inode2, ext4_lblk_t lblk1,
3374 ext4_lblk_t lblk2, ext4_lblk_t count,
3375 int mark_unwritten,int *err);
3376 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu);
3377 extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode,
3378 int check_cred, int restart_cred,
3383 extern void ext4_double_down_write_data_sem(struct inode *first,
3384 struct inode *second);
3385 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3386 struct inode *donor_inode);
3387 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3388 __u64 start_orig, __u64 start_donor,
3389 __u64 len, __u64 *moved_len);
3392 extern int __init ext4_init_pageio(void);
3393 extern void ext4_exit_pageio(void);
3394 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3395 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3396 extern int ext4_put_io_end(ext4_io_end_t *io_end);
3397 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3398 extern void ext4_io_submit_init(struct ext4_io_submit *io,
3399 struct writeback_control *wbc);
3400 extern void ext4_end_io_rsv_work(struct work_struct *work);
3401 extern void ext4_io_submit(struct ext4_io_submit *io);
3402 extern int ext4_bio_write_page(struct ext4_io_submit *io,
3405 struct writeback_control *wbc,
3407 extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end);
3408 extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end);
3411 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3414 extern const struct fsverity_operations ext4_verityops;
3417 * Add new method to test whether block and inode bitmaps are properly
3418 * initialized. With uninit_bg reading the block from disk is not enough
3419 * to mark the bitmap uptodate. We need to also zero-out the bitmap
3421 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3423 static inline int bitmap_uptodate(struct buffer_head *bh)
3425 return (buffer_uptodate(bh) &&
3426 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3428 static inline void set_bitmap_uptodate(struct buffer_head *bh)
3430 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3433 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
3435 /* For ioend & aio unwritten conversion wait queues */
3436 #define EXT4_WQ_HASH_SZ 37
3437 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
3439 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3441 extern int ext4_resize_begin(struct super_block *sb);
3442 extern void ext4_resize_end(struct super_block *sb);
3444 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
3445 struct ext4_io_end *io_end)
3447 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
3448 io_end->flag |= EXT4_IO_END_UNWRITTEN;
3449 atomic_inc(&EXT4_I(inode)->i_unwritten);
3453 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3455 struct inode *inode = io_end->inode;
3457 if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
3458 io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3459 /* Wake up anyone waiting on unwritten extent conversion */
3460 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
3461 wake_up_all(ext4_ioend_wq(inode));
3465 extern const struct iomap_ops ext4_iomap_ops;
3466 extern const struct iomap_ops ext4_iomap_overwrite_ops;
3467 extern const struct iomap_ops ext4_iomap_report_ops;
3469 static inline int ext4_buffer_uptodate(struct buffer_head *bh)
3472 * If the buffer has the write error flag, we have failed
3473 * to write out data in the block. In this case, we don't
3474 * have to read the block because we may read the old data
3477 if (!buffer_uptodate(bh) && buffer_write_io_error(bh))
3478 set_buffer_uptodate(bh);
3479 return buffer_uptodate(bh);
3482 #endif /* __KERNEL__ */
3484 #define EFSBADCRC EBADMSG /* Bad CRC detected */
3485 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
3487 #endif /* _EXT4_H */