1 // SPDX-License-Identifier: GPL-2.0
5 * Copyright (C) 1992, 1993, 1994, 1995
7 * Laboratoire MASI - Institut Blaise Pascal
8 * Universite Pierre et Marie Curie (Paris VI)
12 * linux/include/linux/minix_fs.h
14 * Copyright (C) 1991, 1992 Linus Torvalds
20 #include <linux/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>
45 #include <linux/compiler.h>
48 * The fourth extended filesystem constants/structures
52 * with AGGRESSIVE_CHECK allocator runs consistency checks over
53 * structures. these checks slow things down a lot
55 #define AGGRESSIVE_CHECK__
58 * with DOUBLE_CHECK defined mballoc creates persistent in-core
59 * bitmaps, maintains and uses them to check for double allocations
61 #define DOUBLE_CHECK__
64 * Define EXT4FS_DEBUG to produce debug messages
72 #define ext4_debug(f, a...) \
74 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
75 __FILE__, __LINE__, __func__); \
76 printk(KERN_DEBUG f, ## a); \
79 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
83 * Turn on EXT_DEBUG to get lots of info about extents operations.
87 #define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__)
89 #define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
92 /* data type for block offset of block group */
93 typedef int ext4_grpblk_t;
95 /* data type for filesystem-wide blocks number */
96 typedef unsigned long long ext4_fsblk_t;
98 /* data type for file logical block number */
99 typedef __u32 ext4_lblk_t;
101 /* data type for block group number */
102 typedef unsigned int ext4_group_t;
104 enum SHIFT_DIRECTION {
110 * Flags used in mballoc's allocation_context flags field.
112 * Also used to show what's going on for debugging purposes when the
113 * flag field is exported via the traceport interface
116 /* prefer goal again. length */
117 #define EXT4_MB_HINT_MERGE 0x0001
118 /* blocks already reserved */
119 #define EXT4_MB_HINT_RESERVED 0x0002
120 /* metadata is being allocated */
121 #define EXT4_MB_HINT_METADATA 0x0004
122 /* first blocks in the file */
123 #define EXT4_MB_HINT_FIRST 0x0008
124 /* search for the best chunk */
125 #define EXT4_MB_HINT_BEST 0x0010
126 /* data is being allocated */
127 #define EXT4_MB_HINT_DATA 0x0020
128 /* don't preallocate (for tails) */
129 #define EXT4_MB_HINT_NOPREALLOC 0x0040
130 /* allocate for locality group */
131 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080
132 /* allocate goal blocks or none */
133 #define EXT4_MB_HINT_GOAL_ONLY 0x0100
134 /* goal is meaningful */
135 #define EXT4_MB_HINT_TRY_GOAL 0x0200
136 /* blocks already pre-reserved by delayed allocation */
137 #define EXT4_MB_DELALLOC_RESERVED 0x0400
138 /* We are doing stream allocation */
139 #define EXT4_MB_STREAM_ALLOC 0x0800
140 /* Use reserved root blocks if needed */
141 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000
142 /* Use blocks from reserved pool */
143 #define EXT4_MB_USE_RESERVED 0x2000
145 struct ext4_allocation_request {
146 /* target inode for block we're allocating */
148 /* how many blocks we want to allocate */
150 /* logical block in target inode */
152 /* the closest logical allocated block to the left */
154 /* the closest logical allocated block to the right */
156 /* phys. target (a hint) */
158 /* phys. block for the closest logical allocated block to the left */
160 /* phys. block for the closest logical allocated block to the right */
162 /* flags. see above EXT4_MB_HINT_* */
167 * Logical to physical block mapping, used by ext4_map_blocks()
169 * This structure is used to pass requests into ext4_map_blocks() as
170 * well as to store the information returned by ext4_map_blocks(). It
171 * takes less room on the stack than a struct buffer_head.
173 #define EXT4_MAP_NEW (1 << BH_New)
174 #define EXT4_MAP_MAPPED (1 << BH_Mapped)
175 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten)
176 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary)
177 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
178 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
180 struct ext4_map_blocks {
184 unsigned int m_flags;
188 * Flags for ext4_io_end->flags
190 #define EXT4_IO_END_UNWRITTEN 0x0001
193 * For converting unwritten extents on a work queue. 'handle' is used for
194 * buffered writeback.
196 typedef struct ext4_io_end {
197 struct list_head list; /* per-file finished IO list */
198 handle_t *handle; /* handle reserved for extent
200 struct inode *inode; /* file being written to */
201 struct bio *bio; /* Linked list of completed
202 * bios covering the extent */
203 unsigned int flag; /* unwritten or not */
204 atomic_t count; /* reference counter */
205 loff_t offset; /* offset in the file */
206 ssize_t size; /* size of the extent */
209 struct ext4_io_submit {
210 struct writeback_control *io_wbc;
212 ext4_io_end_t *io_end;
213 sector_t io_next_block;
217 * Special inodes numbers
219 #define EXT4_BAD_INO 1 /* Bad blocks inode */
220 #define EXT4_ROOT_INO 2 /* Root inode */
221 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */
222 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
223 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
224 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
225 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
226 #define EXT4_JOURNAL_INO 8 /* Journal inode */
228 /* First non-reserved inode for old ext4 filesystems */
229 #define EXT4_GOOD_OLD_FIRST_INO 11
232 * Maximal count of links to a file
234 #define EXT4_LINK_MAX 65000
237 * Macro-instructions used to manage several block sizes
239 #define EXT4_MIN_BLOCK_SIZE 1024
240 #define EXT4_MAX_BLOCK_SIZE 65536
241 #define EXT4_MIN_BLOCK_LOG_SIZE 10
242 #define EXT4_MAX_BLOCK_LOG_SIZE 16
243 #define EXT4_MAX_CLUSTER_LOG_SIZE 30
245 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
247 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
249 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
250 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
251 EXT4_SB(s)->s_cluster_bits)
253 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
254 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
256 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
259 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
260 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
261 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
263 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
264 EXT4_GOOD_OLD_INODE_SIZE : \
266 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
267 EXT4_GOOD_OLD_FIRST_INO : \
270 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
271 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \
272 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \
275 /* Translate a block number to a cluster number */
276 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
277 /* Translate a cluster number to a block number */
278 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
279 /* Translate # of blks to # of clusters */
280 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
281 (sbi)->s_cluster_bits)
282 /* Mask out the low bits to get the starting block of the cluster */
283 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
284 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
285 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
286 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
287 /* Get the cluster offset */
288 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
289 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
290 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
291 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
294 * Structure of a blocks group descriptor
296 struct ext4_group_desc
298 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
299 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
300 __le32 bg_inode_table_lo; /* Inodes table block */
301 __le16 bg_free_blocks_count_lo;/* Free blocks count */
302 __le16 bg_free_inodes_count_lo;/* Free inodes count */
303 __le16 bg_used_dirs_count_lo; /* Directories count */
304 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
305 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
306 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
307 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
308 __le16 bg_itable_unused_lo; /* Unused inodes count */
309 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
310 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
311 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
312 __le32 bg_inode_table_hi; /* Inodes table block MSB */
313 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
314 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
315 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
316 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
317 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
318 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
319 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
323 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
324 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
326 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
327 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
331 * Structure of a flex block group info
335 atomic64_t free_clusters;
336 atomic_t free_inodes;
340 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
341 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
342 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
345 * Macro-instructions used to manage group descriptors
347 #define EXT4_MIN_DESC_SIZE 32
348 #define EXT4_MIN_DESC_SIZE_64BIT 64
349 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
350 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
352 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
353 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
354 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
355 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
356 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
358 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
359 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
360 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
364 * Constants relative to the data blocks
366 #define EXT4_NDIR_BLOCKS 12
367 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
368 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
369 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
370 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
375 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
376 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
377 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
378 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
379 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
380 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
381 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
382 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
383 /* Reserved for compression usage... */
384 #define EXT4_DIRTY_FL 0x00000100
385 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
386 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
387 /* nb: was previously EXT2_ECOMPR_FL */
388 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */
389 /* End compression flags --- maybe not all used */
390 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
391 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
392 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
393 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
394 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
395 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
396 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
397 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
398 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
399 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
400 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
401 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
402 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
404 #define EXT4_FL_USER_VISIBLE 0x304BDFFF /* User visible flags */
405 #define EXT4_FL_USER_MODIFIABLE 0x204BC0FF /* User modifiable flags */
407 /* Flags we can manipulate with through EXT4_IOC_FSSETXATTR */
408 #define EXT4_FL_XFLAG_VISIBLE (EXT4_SYNC_FL | \
409 EXT4_IMMUTABLE_FL | \
415 /* Flags that should be inherited by new inodes from their parent. */
416 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
417 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
418 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
419 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
422 /* Flags that are appropriate for regular files (all but dir-specific ones). */
423 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
425 /* Flags that are appropriate for non-directories/regular files. */
426 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
428 /* Mask out flags that are inappropriate for the given type of inode. */
429 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
433 else if (S_ISREG(mode))
434 return flags & EXT4_REG_FLMASK;
436 return flags & EXT4_OTHER_FLMASK;
440 * Inode flags used for atomic set/get
443 EXT4_INODE_SECRM = 0, /* Secure deletion */
444 EXT4_INODE_UNRM = 1, /* Undelete */
445 EXT4_INODE_COMPR = 2, /* Compress file */
446 EXT4_INODE_SYNC = 3, /* Synchronous updates */
447 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
448 EXT4_INODE_APPEND = 5, /* writes to file may only append */
449 EXT4_INODE_NODUMP = 6, /* do not dump file */
450 EXT4_INODE_NOATIME = 7, /* do not update atime */
451 /* Reserved for compression usage... */
452 EXT4_INODE_DIRTY = 8,
453 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
454 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
455 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */
456 /* End compression flags --- maybe not all used */
457 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
458 EXT4_INODE_IMAGIC = 13, /* AFS directory */
459 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
460 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
461 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
462 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
463 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
464 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
465 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
466 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
467 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
468 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */
469 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
473 * Since it's pretty easy to mix up bit numbers and hex values, we use a
474 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
475 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
476 * any extra space in the compiled kernel image, otherwise, the build will fail.
477 * It's important that these values are the same, since we are using
478 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
479 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
480 * values found in ext2, ext3 and ext4 filesystems, and of course the values
481 * defined in e2fsprogs.
483 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
485 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
486 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
488 static inline void ext4_check_flag_values(void)
490 CHECK_FLAG_VALUE(SECRM);
491 CHECK_FLAG_VALUE(UNRM);
492 CHECK_FLAG_VALUE(COMPR);
493 CHECK_FLAG_VALUE(SYNC);
494 CHECK_FLAG_VALUE(IMMUTABLE);
495 CHECK_FLAG_VALUE(APPEND);
496 CHECK_FLAG_VALUE(NODUMP);
497 CHECK_FLAG_VALUE(NOATIME);
498 CHECK_FLAG_VALUE(DIRTY);
499 CHECK_FLAG_VALUE(COMPRBLK);
500 CHECK_FLAG_VALUE(NOCOMPR);
501 CHECK_FLAG_VALUE(ENCRYPT);
502 CHECK_FLAG_VALUE(INDEX);
503 CHECK_FLAG_VALUE(IMAGIC);
504 CHECK_FLAG_VALUE(JOURNAL_DATA);
505 CHECK_FLAG_VALUE(NOTAIL);
506 CHECK_FLAG_VALUE(DIRSYNC);
507 CHECK_FLAG_VALUE(TOPDIR);
508 CHECK_FLAG_VALUE(HUGE_FILE);
509 CHECK_FLAG_VALUE(EXTENTS);
510 CHECK_FLAG_VALUE(EA_INODE);
511 CHECK_FLAG_VALUE(EOFBLOCKS);
512 CHECK_FLAG_VALUE(INLINE_DATA);
513 CHECK_FLAG_VALUE(PROJINHERIT);
514 CHECK_FLAG_VALUE(RESERVED);
517 /* Used to pass group descriptor data when online resize is done */
518 struct ext4_new_group_input {
519 __u32 group; /* Group number for this data */
520 __u64 block_bitmap; /* Absolute block number of block bitmap */
521 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
522 __u64 inode_table; /* Absolute block number of inode table start */
523 __u32 blocks_count; /* Total number of blocks in this group */
524 __u16 reserved_blocks; /* Number of reserved blocks in this group */
528 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
529 struct compat_ext4_new_group_input {
531 compat_u64 block_bitmap;
532 compat_u64 inode_bitmap;
533 compat_u64 inode_table;
540 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
541 struct ext4_new_group_data {
547 __u16 reserved_blocks;
549 __u32 free_clusters_count;
552 /* Indexes used to index group tables in ext4_new_group_data */
554 BLOCK_BITMAP = 0, /* block bitmap */
555 INODE_BITMAP, /* inode bitmap */
556 INODE_TABLE, /* inode tables */
561 * Flags used by ext4_map_blocks()
563 /* Allocate any needed blocks and/or convert an unwritten
564 extent to be an initialized ext4 */
565 #define EXT4_GET_BLOCKS_CREATE 0x0001
566 /* Request the creation of an unwritten extent */
567 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002
568 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\
569 EXT4_GET_BLOCKS_CREATE)
570 /* Caller is from the delayed allocation writeout path
571 * finally doing the actual allocation of delayed blocks */
572 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
573 /* caller is from the direct IO path, request to creation of an
574 unwritten extents if not allocated, split the unwritten
575 extent if blocks has been preallocated already*/
576 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
577 #define EXT4_GET_BLOCKS_CONVERT 0x0010
578 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
579 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
580 /* Convert extent to initialized after IO complete */
581 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
582 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
583 /* Eventual metadata allocation (due to growing extent tree)
584 * should not fail, so try to use reserved blocks for that.*/
585 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
586 /* Don't normalize allocation size (used for fallocate) */
587 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
588 /* Request will not result in inode size update (user for fallocate) */
589 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080
590 /* Convert written extents to unwritten */
591 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100
592 /* Write zeros to newly created written extents */
593 #define EXT4_GET_BLOCKS_ZERO 0x0200
594 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\
595 EXT4_GET_BLOCKS_ZERO)
596 /* Caller will submit data before dropping transaction handle. This
597 * allows jbd2 to avoid submitting data before commit. */
598 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400
601 * The bit position of these flags must not overlap with any of the
602 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(),
603 * read_extent_tree_block(), ext4_split_extent_at(),
604 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
605 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
606 * caching the extents when reading from the extent tree while a
607 * truncate or punch hole operation is in progress.
609 #define EXT4_EX_NOCACHE 0x40000000
610 #define EXT4_EX_FORCE_CACHE 0x20000000
613 * Flags used by ext4_free_blocks
615 #define EXT4_FREE_BLOCKS_METADATA 0x0001
616 #define EXT4_FREE_BLOCKS_FORGET 0x0002
617 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
618 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
619 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
620 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
621 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040
626 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
627 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
628 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
629 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
630 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
631 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
632 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
633 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
634 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
635 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
636 #define EXT4_IOC_MIGRATE _IO('f', 9)
637 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
638 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
639 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
640 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
641 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
642 #define EXT4_IOC_SWAP_BOOT _IO('f', 17)
643 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18)
644 #define EXT4_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY
645 #define EXT4_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT
646 #define EXT4_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY
648 #define EXT4_IOC_FSGETXATTR FS_IOC_FSGETXATTR
649 #define EXT4_IOC_FSSETXATTR FS_IOC_FSSETXATTR
651 #define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32)
654 * Flags for going down operation
656 #define EXT4_GOING_FLAGS_DEFAULT 0x0 /* going down */
657 #define EXT4_GOING_FLAGS_LOGFLUSH 0x1 /* flush log but not data */
658 #define EXT4_GOING_FLAGS_NOLOGFLUSH 0x2 /* don't flush log nor data */
661 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
663 * ioctl commands in 32 bit emulation
665 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
666 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
667 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
668 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
669 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
670 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
671 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
672 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
673 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
674 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
677 /* Max physical block we can address w/o extents */
678 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
680 /* Max logical block we can support */
681 #define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFF
684 * Structure of an inode on the disk
687 __le16 i_mode; /* File mode */
688 __le16 i_uid; /* Low 16 bits of Owner Uid */
689 __le32 i_size_lo; /* Size in bytes */
690 __le32 i_atime; /* Access time */
691 __le32 i_ctime; /* Inode Change time */
692 __le32 i_mtime; /* Modification time */
693 __le32 i_dtime; /* Deletion Time */
694 __le16 i_gid; /* Low 16 bits of Group Id */
695 __le16 i_links_count; /* Links count */
696 __le32 i_blocks_lo; /* Blocks count */
697 __le32 i_flags; /* File flags */
703 __u32 h_i_translator;
708 } osd1; /* OS dependent 1 */
709 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
710 __le32 i_generation; /* File version (for NFS) */
711 __le32 i_file_acl_lo; /* File ACL */
713 __le32 i_obso_faddr; /* Obsoleted fragment address */
716 __le16 l_i_blocks_high; /* were l_i_reserved1 */
717 __le16 l_i_file_acl_high;
718 __le16 l_i_uid_high; /* these 2 fields */
719 __le16 l_i_gid_high; /* were reserved2[0] */
720 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
724 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
731 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
732 __le16 m_i_file_acl_high;
733 __u32 m_i_reserved2[2];
735 } osd2; /* OS dependent 2 */
736 __le16 i_extra_isize;
737 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
738 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
739 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
740 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
741 __le32 i_crtime; /* File Creation time */
742 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
743 __le32 i_version_hi; /* high 32 bits for 64-bit version */
744 __le32 i_projid; /* Project ID */
748 __u32 reserved; /* should be zero */
749 __u32 donor_fd; /* donor file descriptor */
750 __u64 orig_start; /* logical start offset in block for orig */
751 __u64 donor_start; /* logical start offset in block for donor */
752 __u64 len; /* block length to be moved */
753 __u64 moved_len; /* moved block length */
756 #define EXT4_EPOCH_BITS 2
757 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
758 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
761 * Extended fields will fit into an inode if the filesystem was formatted
762 * with large inodes (-I 256 or larger) and there are not currently any EAs
763 * consuming all of the available space. For new inodes we always reserve
764 * enough space for the kernel's known extended fields, but for inodes
765 * created with an old kernel this might not have been the case. None of
766 * the extended inode fields is critical for correct filesystem operation.
767 * This macro checks if a certain field fits in the inode. Note that
768 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
770 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
771 ((offsetof(typeof(*ext4_inode), field) + \
772 sizeof((ext4_inode)->field)) \
773 <= (EXT4_GOOD_OLD_INODE_SIZE + \
774 (einode)->i_extra_isize)) \
777 * We use an encoding that preserves the times for extra epoch "00":
779 * extra msb of adjust for signed
780 * epoch 32-bit 32-bit tv_sec to
781 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
782 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
783 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
784 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
785 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
786 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
787 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
788 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
789 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
791 * Note that previous versions of the kernel on 64-bit systems would
792 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
793 * 1970. e2fsck will correct this, assuming that it is run on the
794 * affected filesystem before 2242.
797 static inline __le32 ext4_encode_extra_time(struct timespec64 *time)
799 u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK;
800 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
803 static inline void ext4_decode_extra_time(struct timespec64 *time,
806 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK))) {
809 /* Handle legacy encoding of pre-1970 dates with epoch
810 * bits 1,1. (This backwards compatibility may be removed
811 * at the discretion of the ext4 developers.)
813 u64 extra_bits = le32_to_cpu(extra) & EXT4_EPOCH_MASK;
814 if (extra_bits == 3 && ((time->tv_sec) & 0x80000000) != 0)
816 time->tv_sec += extra_bits << 32;
818 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
821 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
824 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
826 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
827 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {\
828 (raw_inode)->xtime ## _extra = \
829 ext4_encode_extra_time(&(inode)->xtime); \
833 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
835 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
836 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
837 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
838 (raw_inode)->xtime ## _extra = \
839 ext4_encode_extra_time(&(einode)->xtime); \
842 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
844 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
845 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \
846 ext4_decode_extra_time(&(inode)->xtime, \
847 raw_inode->xtime ## _extra); \
850 (inode)->xtime.tv_nsec = 0; \
854 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
856 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
857 (einode)->xtime.tv_sec = \
858 (signed)le32_to_cpu((raw_inode)->xtime); \
860 (einode)->xtime.tv_sec = 0; \
861 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
862 ext4_decode_extra_time(&(einode)->xtime, \
863 raw_inode->xtime ## _extra); \
865 (einode)->xtime.tv_nsec = 0; \
868 #define i_disk_version osd1.linux1.l_i_version
870 #if defined(__KERNEL__) || defined(__linux__)
871 #define i_reserved1 osd1.linux1.l_i_reserved1
872 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
873 #define i_blocks_high osd2.linux2.l_i_blocks_high
874 #define i_uid_low i_uid
875 #define i_gid_low i_gid
876 #define i_uid_high osd2.linux2.l_i_uid_high
877 #define i_gid_high osd2.linux2.l_i_gid_high
878 #define i_checksum_lo osd2.linux2.l_i_checksum_lo
880 #elif defined(__GNU__)
882 #define i_translator osd1.hurd1.h_i_translator
883 #define i_uid_high osd2.hurd2.h_i_uid_high
884 #define i_gid_high osd2.hurd2.h_i_gid_high
885 #define i_author osd2.hurd2.h_i_author
887 #elif defined(__masix__)
889 #define i_reserved1 osd1.masix1.m_i_reserved1
890 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
891 #define i_reserved2 osd2.masix2.m_i_reserved2
893 #endif /* defined(__KERNEL__) || defined(__linux__) */
895 #include "extents_status.h"
898 * Lock subclasses for i_data_sem in the ext4_inode_info structure.
900 * These are needed to avoid lockdep false positives when we need to
901 * allocate blocks to the quota inode during ext4_map_blocks(), while
902 * holding i_data_sem for a normal (non-quota) inode. Since we don't
903 * do quota tracking for the quota inode, this avoids deadlock (as
904 * well as infinite recursion, since it isn't turtles all the way
907 * I_DATA_SEM_NORMAL - Used for most inodes
908 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode
909 * where the second inode has larger inode number
911 * I_DATA_SEM_QUOTA - Used for quota inodes only
914 I_DATA_SEM_NORMAL = 0,
921 * fourth extended file system inode data in memory
923 struct ext4_inode_info {
924 __le32 i_data[15]; /* unconverted */
926 ext4_fsblk_t i_file_acl;
929 * i_block_group is the number of the block group which contains
930 * this file's inode. Constant across the lifetime of the inode,
931 * it is used for making block allocation decisions - we try to
932 * place a file's data blocks near its inode block, and new inodes
933 * near to their parent directory's inode.
935 ext4_group_t i_block_group;
936 ext4_lblk_t i_dir_start_lookup;
937 #if (BITS_PER_LONG < 64)
938 unsigned long i_state_flags; /* Dynamic state flags */
940 unsigned long i_flags;
943 * Extended attributes can be read independently of the main file
944 * data. Taking i_mutex even when reading would cause contention
945 * between readers of EAs and writers of regular file data, so
946 * instead we synchronize on xattr_sem when reading or changing
949 struct rw_semaphore xattr_sem;
951 struct list_head i_orphan; /* unlinked but open inodes */
954 * i_disksize keeps track of what the inode size is ON DISK, not
955 * in memory. During truncate, i_size is set to the new size by
956 * the VFS prior to calling ext4_truncate(), but the filesystem won't
957 * set i_disksize to 0 until the truncate is actually under way.
959 * The intent is that i_disksize always represents the blocks which
960 * are used by this file. This allows recovery to restart truncate
961 * on orphans if we crash during truncate. We actually write i_disksize
962 * into the on-disk inode when writing inodes out, instead of i_size.
964 * The only time when i_disksize and i_size may be different is when
965 * a truncate is in progress. The only things which change i_disksize
966 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
971 * i_data_sem is for serialising ext4_truncate() against
972 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
973 * data tree are chopped off during truncate. We can't do that in
974 * ext4 because whenever we perform intermediate commits during
975 * truncate, the inode and all the metadata blocks *must* be in a
976 * consistent state which allows truncation of the orphans to restart
977 * during recovery. Hence we must fix the get_block-vs-truncate race
978 * by other means, so we have i_data_sem.
980 struct rw_semaphore i_data_sem;
982 * i_mmap_sem is for serializing page faults with truncate / punch hole
983 * operations. We have to make sure that new page cannot be faulted in
984 * a section of the inode that is being punched. We cannot easily use
985 * i_data_sem for this since we need protection for the whole punch
986 * operation and i_data_sem ranks below transaction start so we have
987 * to occasionally drop it.
989 struct rw_semaphore i_mmap_sem;
990 struct inode vfs_inode;
991 struct jbd2_inode *jinode;
993 spinlock_t i_raw_lock; /* protects updates to the raw inode */
996 * File creation time. Its function is same as that of
997 * struct timespec64 i_{a,c,m}time in the generic inode.
999 struct timespec64 i_crtime;
1002 struct list_head i_prealloc_list;
1003 spinlock_t i_prealloc_lock;
1005 /* extents status tree */
1006 struct ext4_es_tree i_es_tree;
1008 struct list_head i_es_list;
1009 unsigned int i_es_all_nr; /* protected by i_es_lock */
1010 unsigned int i_es_shk_nr; /* protected by i_es_lock */
1011 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for
1012 extents to shrink. Protected by
1016 ext4_group_t i_last_alloc_group;
1018 /* allocation reservation info for delalloc */
1019 /* In case of bigalloc, this refer to clusters rather than blocks */
1020 unsigned int i_reserved_data_blocks;
1021 ext4_lblk_t i_da_metadata_calc_last_lblock;
1022 int i_da_metadata_calc_len;
1024 /* pending cluster reservations for bigalloc file systems */
1025 struct ext4_pending_tree i_pending_tree;
1027 /* on-disk additional length */
1028 __u16 i_extra_isize;
1030 /* Indicate the inline data space. */
1035 /* quota space reservation, managed internally by quota code */
1036 qsize_t i_reserved_quota;
1039 /* Lock protecting lists below */
1040 spinlock_t i_completed_io_lock;
1042 * Completed IOs that need unwritten extents handling and have
1043 * transaction reserved
1045 struct list_head i_rsv_conversion_list;
1046 struct work_struct i_rsv_conversion_work;
1047 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1049 spinlock_t i_block_reservation_lock;
1052 * Transactions that contain inode's metadata needed to complete
1053 * fsync and fdatasync, respectively.
1056 tid_t i_datasync_tid;
1059 struct dquot *i_dquot[MAXQUOTAS];
1062 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1069 * File system states
1071 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
1072 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
1073 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
1076 * Misc. filesystem flags
1078 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
1079 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
1080 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
1083 * Mount flags set via mount options or defaults
1085 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */
1086 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
1087 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
1088 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
1089 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
1090 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
1091 #define EXT4_MOUNT_ERRORS_MASK 0x00070
1092 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
1093 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
1094 #ifdef CONFIG_FS_DAX
1095 #define EXT4_MOUNT_DAX 0x00200 /* Direct Access */
1097 #define EXT4_MOUNT_DAX 0
1099 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
1100 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
1101 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
1102 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
1103 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
1104 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
1105 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
1106 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
1107 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
1108 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
1109 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */
1110 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota,
1111 * enable enforcement for hidden
1113 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable
1114 * enforcement for hidden quota
1116 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota
1118 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
1119 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
1120 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1121 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */
1122 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
1123 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
1124 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
1125 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
1126 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
1129 * Mount flags set either automatically (could not be set by mount option)
1130 * based on per file system feature or property or in special cases such as
1131 * distinguishing between explicit mount option definition and default.
1133 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1134 specified delalloc */
1135 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1136 size of blocksize * 8
1138 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1141 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly
1142 specified journal checksum */
1144 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1146 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1148 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1151 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1153 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1155 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1158 #define ext4_test_and_set_bit __test_and_set_bit_le
1159 #define ext4_set_bit __set_bit_le
1160 #define ext4_set_bit_atomic ext2_set_bit_atomic
1161 #define ext4_test_and_clear_bit __test_and_clear_bit_le
1162 #define ext4_clear_bit __clear_bit_le
1163 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
1164 #define ext4_test_bit test_bit_le
1165 #define ext4_find_next_zero_bit find_next_zero_bit_le
1166 #define ext4_find_next_bit find_next_bit_le
1168 extern void ext4_set_bits(void *bm, int cur, int len);
1171 * Maximal mount counts between two filesystem checks
1173 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1174 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1177 * Behaviour when detecting errors
1179 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1180 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1181 #define EXT4_ERRORS_PANIC 3 /* Panic */
1182 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1184 /* Metadata checksum algorithm codes */
1185 #define EXT4_CRC32C_CHKSUM 1
1188 * Structure of the super block
1190 struct ext4_super_block {
1191 /*00*/ __le32 s_inodes_count; /* Inodes count */
1192 __le32 s_blocks_count_lo; /* Blocks count */
1193 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1194 __le32 s_free_blocks_count_lo; /* Free blocks count */
1195 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1196 __le32 s_first_data_block; /* First Data Block */
1197 __le32 s_log_block_size; /* Block size */
1198 __le32 s_log_cluster_size; /* Allocation cluster size */
1199 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1200 __le32 s_clusters_per_group; /* # Clusters per group */
1201 __le32 s_inodes_per_group; /* # Inodes per group */
1202 __le32 s_mtime; /* Mount time */
1203 /*30*/ __le32 s_wtime; /* Write time */
1204 __le16 s_mnt_count; /* Mount count */
1205 __le16 s_max_mnt_count; /* Maximal mount count */
1206 __le16 s_magic; /* Magic signature */
1207 __le16 s_state; /* File system state */
1208 __le16 s_errors; /* Behaviour when detecting errors */
1209 __le16 s_minor_rev_level; /* minor revision level */
1210 /*40*/ __le32 s_lastcheck; /* time of last check */
1211 __le32 s_checkinterval; /* max. time between checks */
1212 __le32 s_creator_os; /* OS */
1213 __le32 s_rev_level; /* Revision level */
1214 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1215 __le16 s_def_resgid; /* Default gid for reserved blocks */
1217 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1219 * Note: the difference between the compatible feature set and
1220 * the incompatible feature set is that if there is a bit set
1221 * in the incompatible feature set that the kernel doesn't
1222 * know about, it should refuse to mount the filesystem.
1224 * e2fsck's requirements are more strict; if it doesn't know
1225 * about a feature in either the compatible or incompatible
1226 * feature set, it must abort and not try to meddle with
1227 * things it doesn't understand...
1229 __le32 s_first_ino; /* First non-reserved inode */
1230 __le16 s_inode_size; /* size of inode structure */
1231 __le16 s_block_group_nr; /* block group # of this superblock */
1232 __le32 s_feature_compat; /* compatible feature set */
1233 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1234 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1235 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1236 /*78*/ char s_volume_name[16]; /* volume name */
1237 /*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */
1238 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1240 * Performance hints. Directory preallocation should only
1241 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1243 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1244 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1245 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1247 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1249 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1250 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1251 __le32 s_journal_dev; /* device number of journal file */
1252 __le32 s_last_orphan; /* start of list of inodes to delete */
1253 __le32 s_hash_seed[4]; /* HTREE hash seed */
1254 __u8 s_def_hash_version; /* Default hash version to use */
1255 __u8 s_jnl_backup_type;
1256 __le16 s_desc_size; /* size of group descriptor */
1257 /*100*/ __le32 s_default_mount_opts;
1258 __le32 s_first_meta_bg; /* First metablock block group */
1259 __le32 s_mkfs_time; /* When the filesystem was created */
1260 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1261 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1262 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1263 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1264 __le32 s_free_blocks_count_hi; /* Free blocks count */
1265 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1266 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1267 __le32 s_flags; /* Miscellaneous flags */
1268 __le16 s_raid_stride; /* RAID stride */
1269 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1270 __le64 s_mmp_block; /* Block for multi-mount protection */
1271 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1272 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1273 __u8 s_checksum_type; /* metadata checksum algorithm used */
1274 __u8 s_encryption_level; /* versioning level for encryption */
1275 __u8 s_reserved_pad; /* Padding to next 32bits */
1276 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1277 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1278 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1279 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1280 snapshot's future use */
1281 __le32 s_snapshot_list; /* inode number of the head of the
1282 on-disk snapshot list */
1283 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1284 __le32 s_error_count; /* number of fs errors */
1285 __le32 s_first_error_time; /* first time an error happened */
1286 __le32 s_first_error_ino; /* inode involved in first error */
1287 __le64 s_first_error_block; /* block involved of first error */
1288 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */
1289 __le32 s_first_error_line; /* line number where error happened */
1290 __le32 s_last_error_time; /* most recent time of an error */
1291 __le32 s_last_error_ino; /* inode involved in last error */
1292 __le32 s_last_error_line; /* line number where error happened */
1293 __le64 s_last_error_block; /* block involved of last error */
1294 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */
1295 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1296 __u8 s_mount_opts[64];
1297 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1298 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1299 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1300 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
1301 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */
1302 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */
1303 __le32 s_lpf_ino; /* Location of the lost+found inode */
1304 __le32 s_prj_quota_inum; /* inode for tracking project quota */
1305 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */
1308 __u8 s_mkfs_time_hi;
1309 __u8 s_lastcheck_hi;
1310 __u8 s_first_error_time_hi;
1311 __u8 s_last_error_time_hi;
1313 __le32 s_reserved[96]; /* Padding to the end of the block */
1314 __le32 s_checksum; /* crc32c(superblock) */
1317 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1322 * run-time mount flags
1324 #define EXT4_MF_MNTDIR_SAMPLED 0x0001
1325 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1326 #define EXT4_MF_TEST_DUMMY_ENCRYPTION 0x0004
1328 #ifdef CONFIG_FS_ENCRYPTION
1329 #define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \
1330 EXT4_MF_TEST_DUMMY_ENCRYPTION))
1332 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1335 /* Number of quota types we support */
1336 #define EXT4_MAXQUOTAS 3
1339 * fourth extended-fs super-block data in memory
1341 struct ext4_sb_info {
1342 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1343 unsigned long s_inodes_per_block;/* Number of inodes per block */
1344 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1345 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1346 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1347 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1348 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1349 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1350 ext4_group_t s_groups_count; /* Number of groups in the fs */
1351 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1352 unsigned long s_overhead; /* # of fs overhead clusters */
1353 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1354 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1355 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1356 struct buffer_head * s_sbh; /* Buffer containing the super block */
1357 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1358 struct buffer_head **s_group_desc;
1359 unsigned int s_mount_opt;
1360 unsigned int s_mount_opt2;
1361 unsigned int s_mount_flags;
1362 unsigned int s_def_mount_opt;
1363 ext4_fsblk_t s_sb_block;
1364 atomic64_t s_resv_clusters;
1367 unsigned short s_mount_state;
1368 unsigned short s_pad;
1369 int s_addr_per_block_bits;
1370 int s_desc_per_block_bits;
1373 unsigned int s_inode_readahead_blks;
1374 unsigned int s_inode_goal;
1376 int s_def_hash_version;
1377 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1378 struct percpu_counter s_freeclusters_counter;
1379 struct percpu_counter s_freeinodes_counter;
1380 struct percpu_counter s_dirs_counter;
1381 struct percpu_counter s_dirtyclusters_counter;
1382 struct blockgroup_lock *s_blockgroup_lock;
1383 struct proc_dir_entry *s_proc;
1384 struct kobject s_kobj;
1385 struct completion s_kobj_unregister;
1386 struct super_block *s_sb;
1389 struct journal_s *s_journal;
1390 struct list_head s_orphan;
1391 struct mutex s_orphan_lock;
1392 unsigned long s_ext4_flags; /* Ext4 superblock flags */
1393 unsigned long s_commit_interval;
1394 u32 s_max_batch_time;
1395 u32 s_min_batch_time;
1396 struct block_device *journal_bdev;
1398 /* Names of quota files with journalled quota */
1399 char __rcu *s_qf_names[EXT4_MAXQUOTAS];
1400 int s_jquota_fmt; /* Format of quota to use */
1402 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1403 struct rb_root system_blks;
1405 #ifdef EXTENTS_STATS
1406 /* ext4 extents stats */
1407 unsigned long s_ext_min;
1408 unsigned long s_ext_max;
1409 unsigned long s_depth_max;
1410 spinlock_t s_ext_stats_lock;
1411 unsigned long s_ext_blocks;
1412 unsigned long s_ext_extents;
1415 /* for buddy allocator */
1416 struct ext4_group_info ***s_group_info;
1417 struct inode *s_buddy_cache;
1418 spinlock_t s_md_lock;
1419 unsigned short *s_mb_offsets;
1420 unsigned int *s_mb_maxs;
1421 unsigned int s_group_info_size;
1422 unsigned int s_mb_free_pending;
1423 struct list_head s_freed_data_list; /* List of blocks to be freed
1424 after commit completed */
1427 unsigned long s_stripe;
1428 unsigned int s_mb_stream_request;
1429 unsigned int s_mb_max_to_scan;
1430 unsigned int s_mb_min_to_scan;
1431 unsigned int s_mb_stats;
1432 unsigned int s_mb_order2_reqs;
1433 unsigned int s_mb_group_prealloc;
1434 unsigned int s_max_dir_size_kb;
1435 /* where last allocation was done - for stream allocation */
1436 unsigned long s_mb_last_group;
1437 unsigned long s_mb_last_start;
1439 /* stats for buddy allocator */
1440 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1441 atomic_t s_bal_success; /* we found long enough chunks */
1442 atomic_t s_bal_allocated; /* in blocks */
1443 atomic_t s_bal_ex_scanned; /* total extents scanned */
1444 atomic_t s_bal_goals; /* goal hits */
1445 atomic_t s_bal_breaks; /* too long searches */
1446 atomic_t s_bal_2orders; /* 2^order hits */
1447 spinlock_t s_bal_lock;
1448 unsigned long s_mb_buddies_generated;
1449 unsigned long long s_mb_generation_time;
1450 atomic_t s_mb_lost_chunks;
1451 atomic_t s_mb_preallocated;
1452 atomic_t s_mb_discarded;
1453 atomic_t s_lock_busy;
1455 /* locality groups */
1456 struct ext4_locality_group __percpu *s_locality_groups;
1458 /* for write statistics */
1459 unsigned long s_sectors_written_start;
1460 u64 s_kbytes_written;
1462 /* the size of zero-out chunk */
1463 unsigned int s_extent_max_zeroout_kb;
1465 unsigned int s_log_groups_per_flex;
1466 struct flex_groups *s_flex_groups;
1467 ext4_group_t s_flex_groups_allocated;
1469 /* workqueue for reserved extent conversions (buffered io) */
1470 struct workqueue_struct *rsv_conversion_wq;
1472 /* timer for periodic error stats printing */
1473 struct timer_list s_err_report;
1475 /* Lazy inode table initialization info */
1476 struct ext4_li_request *s_li_request;
1477 /* Wait multiplier for lazy initialization thread */
1478 unsigned int s_li_wait_mult;
1480 /* Kernel thread for multiple mount protection */
1481 struct task_struct *s_mmp_tsk;
1483 /* record the last minlen when FITRIM is called. */
1484 atomic_t s_last_trim_minblks;
1486 /* Reference to checksum algorithm driver via cryptoapi */
1487 struct crypto_shash *s_chksum_driver;
1489 /* Precomputed FS UUID checksum for seeding other checksums */
1492 /* Reclaim extents from extent status tree */
1493 struct shrinker s_es_shrinker;
1494 struct list_head s_es_list; /* List of inodes with reclaimable extents */
1496 struct ext4_es_stats s_es_stats;
1497 struct mb_cache *s_ea_block_cache;
1498 struct mb_cache *s_ea_inode_cache;
1499 spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1501 /* Ratelimit ext4 messages. */
1502 struct ratelimit_state s_err_ratelimit_state;
1503 struct ratelimit_state s_warning_ratelimit_state;
1504 struct ratelimit_state s_msg_ratelimit_state;
1506 /* Barrier between changing inodes' journal flags and writepages ops. */
1507 struct percpu_rw_semaphore s_journal_flag_rwsem;
1508 struct dax_device *s_daxdev;
1511 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1513 return sb->s_fs_info;
1515 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1517 return container_of(inode, struct ext4_inode_info, vfs_inode);
1520 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1522 return ino == EXT4_ROOT_INO ||
1523 (ino >= EXT4_FIRST_INO(sb) &&
1524 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1528 * Inode dynamic state flags
1531 EXT4_STATE_JDATA, /* journaled data exists */
1532 EXT4_STATE_NEW, /* inode is newly created */
1533 EXT4_STATE_XATTR, /* has in-inode xattrs */
1534 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1535 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1536 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1537 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
1538 EXT4_STATE_NEWENTRY, /* File just added to dir */
1539 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1540 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
1541 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */
1544 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1545 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1547 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1549 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1551 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1553 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1555 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1558 /* Add these declarations here only so that these functions can be
1559 * found by name. Otherwise, they are very hard to locate. */
1560 static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1561 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1562 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1563 EXT4_INODE_BIT_FNS(flag, flags, 0)
1565 /* Add these declarations here only so that these functions can be
1566 * found by name. Otherwise, they are very hard to locate. */
1567 static inline int ext4_test_inode_state(struct inode *inode, int bit);
1568 static inline void ext4_set_inode_state(struct inode *inode, int bit);
1569 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1570 #if (BITS_PER_LONG < 64)
1571 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1573 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1575 (ei)->i_state_flags = 0;
1578 EXT4_INODE_BIT_FNS(state, flags, 32)
1580 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1582 /* We depend on the fact that callers will set i_flags */
1586 /* Assume that user mode programs are passing in an ext4fs superblock, not
1587 * a kernel struct super_block. This will allow us to call the feature-test
1588 * macros from user land. */
1589 #define EXT4_SB(sb) (sb)
1593 * Returns true if the inode is inode is encrypted
1595 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1598 * Codes for operating systems
1600 #define EXT4_OS_LINUX 0
1601 #define EXT4_OS_HURD 1
1602 #define EXT4_OS_MASIX 2
1603 #define EXT4_OS_FREEBSD 3
1604 #define EXT4_OS_LITES 4
1609 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1610 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1612 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1613 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1615 #define EXT4_GOOD_OLD_INODE_SIZE 128
1618 * Feature set definitions
1621 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1622 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1623 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1624 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1625 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1626 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1627 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
1629 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1630 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1631 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1632 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1633 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1634 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1635 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1636 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1637 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1639 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1640 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1641 * all other data structures' checksums. However, the METADATA_CSUM and
1642 * GDT_CSUM bits are mutually exclusive.
1644 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
1645 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000
1646 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000
1648 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1649 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1650 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1651 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1652 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1653 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1654 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1655 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1656 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1657 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1658 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1659 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000
1660 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
1661 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
1662 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000
1664 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
1665 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1667 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1668 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
1670 static inline void ext4_set_feature_##name(struct super_block *sb) \
1672 EXT4_SB(sb)->s_es->s_feature_compat |= \
1673 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1675 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1677 EXT4_SB(sb)->s_es->s_feature_compat &= \
1678 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1681 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
1682 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1684 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1685 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
1687 static inline void ext4_set_feature_##name(struct super_block *sb) \
1689 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
1690 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1692 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1694 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
1695 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1698 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
1699 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1701 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1702 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
1704 static inline void ext4_set_feature_##name(struct super_block *sb) \
1706 EXT4_SB(sb)->s_es->s_feature_incompat |= \
1707 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1709 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1711 EXT4_SB(sb)->s_es->s_feature_incompat &= \
1712 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1715 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC)
1716 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES)
1717 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL)
1718 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR)
1719 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE)
1720 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX)
1721 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2)
1723 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER)
1724 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE)
1725 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR)
1726 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE)
1727 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM)
1728 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK)
1729 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE)
1730 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA)
1731 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC)
1732 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM)
1733 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY)
1734 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT)
1736 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION)
1737 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE)
1738 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER)
1739 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV)
1740 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG)
1741 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS)
1742 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT)
1743 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP)
1744 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG)
1745 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE)
1746 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA)
1747 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED)
1748 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR)
1749 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA)
1750 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT)
1752 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1753 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1754 EXT4_FEATURE_INCOMPAT_META_BG)
1755 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1756 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1757 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1759 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1760 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1761 EXT4_FEATURE_INCOMPAT_RECOVER| \
1762 EXT4_FEATURE_INCOMPAT_META_BG)
1763 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1764 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1765 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1767 #define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1768 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1769 EXT4_FEATURE_INCOMPAT_RECOVER| \
1770 EXT4_FEATURE_INCOMPAT_META_BG| \
1771 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1772 EXT4_FEATURE_INCOMPAT_64BIT| \
1773 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1774 EXT4_FEATURE_INCOMPAT_EA_INODE| \
1775 EXT4_FEATURE_INCOMPAT_MMP | \
1776 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
1777 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
1778 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \
1779 EXT4_FEATURE_INCOMPAT_LARGEDIR)
1780 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1781 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1782 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1783 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1784 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1785 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1786 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1787 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1788 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1789 EXT4_FEATURE_RO_COMPAT_QUOTA |\
1790 EXT4_FEATURE_RO_COMPAT_PROJECT)
1792 #define EXTN_FEATURE_FUNCS(ver) \
1793 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
1795 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1796 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
1798 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
1800 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1801 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
1803 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
1805 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1806 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
1809 EXTN_FEATURE_FUNCS(2)
1810 EXTN_FEATURE_FUNCS(3)
1811 EXTN_FEATURE_FUNCS(4)
1813 static inline bool ext4_has_compat_features(struct super_block *sb)
1815 return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
1817 static inline bool ext4_has_ro_compat_features(struct super_block *sb)
1819 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
1821 static inline bool ext4_has_incompat_features(struct super_block *sb)
1823 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
1829 #define EXT4_FLAGS_RESIZING 0
1830 #define EXT4_FLAGS_SHUTDOWN 1
1832 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi)
1834 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags);
1839 * Default values for user and/or group using reserved blocks
1841 #define EXT4_DEF_RESUID 0
1842 #define EXT4_DEF_RESGID 0
1845 * Default project ID
1847 #define EXT4_DEF_PROJID 0
1849 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
1852 * Default mount options
1854 #define EXT4_DEFM_DEBUG 0x0001
1855 #define EXT4_DEFM_BSDGROUPS 0x0002
1856 #define EXT4_DEFM_XATTR_USER 0x0004
1857 #define EXT4_DEFM_ACL 0x0008
1858 #define EXT4_DEFM_UID16 0x0010
1859 #define EXT4_DEFM_JMODE 0x0060
1860 #define EXT4_DEFM_JMODE_DATA 0x0020
1861 #define EXT4_DEFM_JMODE_ORDERED 0x0040
1862 #define EXT4_DEFM_JMODE_WBACK 0x0060
1863 #define EXT4_DEFM_NOBARRIER 0x0100
1864 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1865 #define EXT4_DEFM_DISCARD 0x0400
1866 #define EXT4_DEFM_NODELALLOC 0x0800
1869 * Default journal batch times
1871 #define EXT4_DEF_MIN_BATCH_TIME 0
1872 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1875 * Minimum number of groups in a flexgroup before we separate out
1876 * directories into the first block group of a flexgroup
1878 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1881 * Structure of a directory entry
1883 #define EXT4_NAME_LEN 255
1885 struct ext4_dir_entry {
1886 __le32 inode; /* Inode number */
1887 __le16 rec_len; /* Directory entry length */
1888 __le16 name_len; /* Name length */
1889 char name[EXT4_NAME_LEN]; /* File name */
1893 * The new version of the directory entry. Since EXT4 structures are
1894 * stored in intel byte order, and the name_len field could never be
1895 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1898 struct ext4_dir_entry_2 {
1899 __le32 inode; /* Inode number */
1900 __le16 rec_len; /* Directory entry length */
1901 __u8 name_len; /* Name length */
1903 char name[EXT4_NAME_LEN]; /* File name */
1907 * This is a bogus directory entry at the end of each leaf block that
1908 * records checksums.
1910 struct ext4_dir_entry_tail {
1911 __le32 det_reserved_zero1; /* Pretend to be unused */
1912 __le16 det_rec_len; /* 12 */
1913 __u8 det_reserved_zero2; /* Zero name length */
1914 __u8 det_reserved_ft; /* 0xDE, fake file type */
1915 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
1918 #define EXT4_DIRENT_TAIL(block, blocksize) \
1919 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1921 sizeof(struct ext4_dir_entry_tail))))
1924 * Ext4 directory file types. Only the low 3 bits are used. The
1925 * other bits are reserved for now.
1927 #define EXT4_FT_UNKNOWN 0
1928 #define EXT4_FT_REG_FILE 1
1929 #define EXT4_FT_DIR 2
1930 #define EXT4_FT_CHRDEV 3
1931 #define EXT4_FT_BLKDEV 4
1932 #define EXT4_FT_FIFO 5
1933 #define EXT4_FT_SOCK 6
1934 #define EXT4_FT_SYMLINK 7
1936 #define EXT4_FT_MAX 8
1938 #define EXT4_FT_DIR_CSUM 0xDE
1941 * EXT4_DIR_PAD defines the directory entries boundaries
1943 * NOTE: It must be a multiple of 4
1945 #define EXT4_DIR_PAD 4
1946 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1947 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1949 #define EXT4_MAX_REC_LEN ((1<<16)-1)
1952 * If we ever get support for fs block sizes > page_size, we'll need
1953 * to remove the #if statements in the next two functions...
1955 static inline unsigned int
1956 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1958 unsigned len = le16_to_cpu(dlen);
1960 #if (PAGE_SIZE >= 65536)
1961 if (len == EXT4_MAX_REC_LEN || len == 0)
1963 return (len & 65532) | ((len & 3) << 16);
1969 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1971 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1973 #if (PAGE_SIZE >= 65536)
1975 return cpu_to_le16(len);
1976 if (len == blocksize) {
1977 if (blocksize == 65536)
1978 return cpu_to_le16(EXT4_MAX_REC_LEN);
1980 return cpu_to_le16(0);
1982 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1984 return cpu_to_le16(len);
1989 * Hash Tree Directory indexing
1990 * (c) Daniel Phillips, 2001
1993 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
1994 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
1995 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \
1996 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir)))
1997 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1999 /* Legal values for the dx_root hash_version field: */
2001 #define DX_HASH_LEGACY 0
2002 #define DX_HASH_HALF_MD4 1
2003 #define DX_HASH_TEA 2
2004 #define DX_HASH_LEGACY_UNSIGNED 3
2005 #define DX_HASH_HALF_MD4_UNSIGNED 4
2006 #define DX_HASH_TEA_UNSIGNED 5
2008 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2009 const void *address, unsigned int length)
2012 struct shash_desc shash;
2016 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2018 desc.shash.tfm = sbi->s_chksum_driver;
2019 desc.shash.flags = 0;
2020 *(u32 *)desc.ctx = crc;
2022 BUG_ON(crypto_shash_update(&desc.shash, address, length));
2024 return *(u32 *)desc.ctx;
2029 /* hash info structure used by the directory hash */
2039 /* 32 and 64 bit signed EOF for dx directories */
2040 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
2041 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
2045 * Control parameters used by ext4_htree_next_block
2047 #define HASH_NB_ALWAYS 1
2049 struct ext4_filename {
2050 const struct qstr *usr_fname;
2051 struct fscrypt_str disk_name;
2052 struct dx_hash_info hinfo;
2053 #ifdef CONFIG_FS_ENCRYPTION
2054 struct fscrypt_str crypto_buf;
2058 #define fname_name(p) ((p)->disk_name.name)
2059 #define fname_len(p) ((p)->disk_name.len)
2062 * Describe an inode's exact location on disk and in memory
2066 struct buffer_head *bh;
2067 unsigned long offset;
2068 ext4_group_t block_group;
2071 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2073 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2076 static inline bool ext4_is_quota_file(struct inode *inode)
2078 return IS_NOQUOTA(inode) &&
2079 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL);
2083 * This structure is stuffed into the struct file's private_data field
2084 * for directories. It is where we put information so that we can do
2085 * readdir operations in hash tree order.
2087 struct dir_private_info {
2088 struct rb_root root;
2089 struct rb_node *curr_node;
2090 struct fname *extra_fname;
2093 __u32 curr_minor_hash;
2097 /* calculate the first block number of the group */
2098 static inline ext4_fsblk_t
2099 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2101 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2102 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2106 * Special error return code only used by dx_probe() and its callers.
2108 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
2110 /* htree levels for ext4 */
2111 #define EXT4_HTREE_LEVEL_COMPAT 2
2112 #define EXT4_HTREE_LEVEL 3
2114 static inline int ext4_dir_htree_level(struct super_block *sb)
2116 return ext4_has_feature_largedir(sb) ?
2117 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT;
2121 * Timeout and state flag for lazy initialization inode thread.
2123 #define EXT4_DEF_LI_WAIT_MULT 10
2124 #define EXT4_DEF_LI_MAX_START_DELAY 5
2125 #define EXT4_LAZYINIT_QUIT 0x0001
2126 #define EXT4_LAZYINIT_RUNNING 0x0002
2129 * Lazy inode table initialization info
2131 struct ext4_lazy_init {
2132 unsigned long li_state;
2133 struct list_head li_request_list;
2134 struct mutex li_list_mtx;
2137 struct ext4_li_request {
2138 struct super_block *lr_super;
2139 struct ext4_sb_info *lr_sbi;
2140 ext4_group_t lr_next_group;
2141 struct list_head lr_request;
2142 unsigned long lr_next_sched;
2143 unsigned long lr_timeout;
2146 struct ext4_features {
2147 struct kobject f_kobj;
2148 struct completion f_kobj_unregister;
2152 * This structure will be used for multiple mount protection. It will be
2153 * written into the block number saved in the s_mmp_block field in the
2154 * superblock. Programs that check MMP should assume that if
2155 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2156 * to use the filesystem, regardless of how old the timestamp is.
2158 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
2159 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2160 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
2161 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
2164 __le32 mmp_magic; /* Magic number for MMP */
2165 __le32 mmp_seq; /* Sequence no. updated periodically */
2168 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2169 * purposes and do not affect the correctness of the algorithm
2171 __le64 mmp_time; /* Time last updated */
2172 char mmp_nodename[64]; /* Node which last updated MMP block */
2173 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
2176 * mmp_check_interval is used to verify if the MMP block has been
2177 * updated on the block device. The value is updated based on the
2178 * maximum time to write the MMP block during an update cycle.
2180 __le16 mmp_check_interval;
2183 __le32 mmp_pad2[226];
2184 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
2187 /* arguments passed to the mmp thread */
2189 struct buffer_head *bh; /* bh from initial read_mmp_block() */
2190 struct super_block *sb; /* super block of the fs */
2194 * Check interval multiplier
2195 * The MMP block is written every update interval and initially checked every
2196 * update interval x the multiplier (the value is then adapted based on the
2197 * write latency). The reason is that writes can be delayed under load and we
2198 * don't want readers to incorrectly assume that the filesystem is no longer
2201 #define EXT4_MMP_CHECK_MULT 2UL
2204 * Minimum interval for MMP checking in seconds.
2206 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
2209 * Maximum interval for MMP checking in seconds.
2211 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
2214 * Function prototypes
2218 * Ok, these declarations are also in <linux/kernel.h> but none of the
2219 * ext4 source programs needs to include it so they are duplicated here.
2221 # define NORET_TYPE /**/
2222 # define ATTRIB_NORET __attribute__((noreturn))
2223 # define NORET_AND noreturn,
2226 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2227 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2228 struct ext4_group_desc *gdp,
2229 struct buffer_head *bh, int sz);
2230 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2231 struct ext4_group_desc *gdp,
2232 struct buffer_head *bh, int sz);
2233 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2234 struct ext4_group_desc *gdp,
2235 struct buffer_head *bh);
2236 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2237 struct ext4_group_desc *gdp,
2238 struct buffer_head *bh);
2241 extern void ext4_get_group_no_and_offset(struct super_block *sb,
2242 ext4_fsblk_t blocknr,
2243 ext4_group_t *blockgrpp,
2244 ext4_grpblk_t *offsetp);
2245 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2246 ext4_fsblk_t block);
2248 extern unsigned int ext4_block_group(struct super_block *sb,
2249 ext4_fsblk_t blocknr);
2250 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2251 ext4_fsblk_t blocknr);
2252 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2253 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2254 ext4_group_t group);
2255 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2258 unsigned long *count,
2260 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2261 s64 nclusters, unsigned int flags);
2262 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2263 extern void ext4_check_blocks_bitmap(struct super_block *);
2264 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2265 ext4_group_t block_group,
2266 struct buffer_head ** bh);
2267 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2269 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2270 ext4_group_t block_group);
2271 extern int ext4_wait_block_bitmap(struct super_block *sb,
2272 ext4_group_t block_group,
2273 struct buffer_head *bh);
2274 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2275 ext4_group_t block_group);
2276 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2277 ext4_group_t block_group,
2278 struct ext4_group_desc *gdp);
2279 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2281 #ifdef CONFIG_FS_ENCRYPTION
2282 static inline int ext4_fname_setup_filename(struct inode *dir,
2283 const struct qstr *iname,
2284 int lookup, struct ext4_filename *fname)
2286 struct fscrypt_name name;
2289 memset(fname, 0, sizeof(struct ext4_filename));
2291 err = fscrypt_setup_filename(dir, iname, lookup, &name);
2293 fname->usr_fname = name.usr_fname;
2294 fname->disk_name = name.disk_name;
2295 fname->hinfo.hash = name.hash;
2296 fname->hinfo.minor_hash = name.minor_hash;
2297 fname->crypto_buf = name.crypto_buf;
2301 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2303 struct fscrypt_name name;
2305 name.crypto_buf = fname->crypto_buf;
2306 fscrypt_free_filename(&name);
2308 fname->crypto_buf.name = NULL;
2309 fname->usr_fname = NULL;
2310 fname->disk_name.name = NULL;
2313 static inline int ext4_fname_setup_filename(struct inode *dir,
2314 const struct qstr *iname,
2315 int lookup, struct ext4_filename *fname)
2317 fname->usr_fname = iname;
2318 fname->disk_name.name = (unsigned char *) iname->name;
2319 fname->disk_name.len = iname->len;
2322 static inline void ext4_fname_free_filename(struct ext4_filename *fname) { }
2327 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2329 struct ext4_dir_entry_2 *,
2330 struct buffer_head *, char *, int,
2332 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2333 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2334 (de), (bh), (buf), (size), (offset)))
2335 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2337 struct ext4_dir_entry_2 *dirent,
2338 struct fscrypt_str *ent_name);
2339 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2340 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2341 struct buffer_head *bh,
2342 void *buf, int buf_size,
2343 struct ext4_filename *fname,
2344 struct ext4_dir_entry_2 **dest_de);
2345 void ext4_insert_dentry(struct inode *inode,
2346 struct ext4_dir_entry_2 *de,
2348 struct ext4_filename *fname);
2349 static inline void ext4_update_dx_flag(struct inode *inode)
2351 if (!ext4_has_feature_dir_index(inode->i_sb))
2352 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2354 static const unsigned char ext4_filetype_table[] = {
2355 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2358 static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2360 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2363 return ext4_filetype_table[filetype];
2365 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2366 void *buf, int buf_size);
2369 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2372 extern int ext4fs_dirhash(const char *name, int len, struct
2373 dx_hash_info *hinfo);
2376 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2377 const struct qstr *qstr, __u32 goal,
2378 uid_t *owner, __u32 i_flags,
2379 int handle_type, unsigned int line_no,
2382 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \
2383 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2385 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2387 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2388 0, (type), __LINE__, (nblocks))
2391 extern void ext4_free_inode(handle_t *, struct inode *);
2392 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2393 extern unsigned long ext4_count_free_inodes(struct super_block *);
2394 extern unsigned long ext4_count_dirs(struct super_block *);
2395 extern void ext4_check_inodes_bitmap(struct super_block *);
2396 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2397 extern int ext4_init_inode_table(struct super_block *sb,
2398 ext4_group_t group, int barrier);
2399 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2402 extern const struct seq_operations ext4_mb_seq_groups_ops;
2403 extern long ext4_mb_stats;
2404 extern long ext4_mb_max_to_scan;
2405 extern int ext4_mb_init(struct super_block *);
2406 extern int ext4_mb_release(struct super_block *);
2407 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2408 struct ext4_allocation_request *, int *);
2409 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2410 extern void ext4_discard_preallocations(struct inode *);
2411 extern int __init ext4_init_mballoc(void);
2412 extern void ext4_exit_mballoc(void);
2413 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2414 struct buffer_head *bh, ext4_fsblk_t block,
2415 unsigned long count, int flags);
2416 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2417 ext4_group_t ngroups);
2418 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2419 ext4_group_t i, struct ext4_group_desc *desc);
2420 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2421 ext4_fsblk_t block, unsigned long count);
2422 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2423 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid);
2426 int ext4_inode_is_fast_symlink(struct inode *inode);
2427 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2428 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2429 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count,
2430 bool wait, struct buffer_head **bhs);
2431 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2432 struct buffer_head *bh_result, int create);
2433 int ext4_get_block(struct inode *inode, sector_t iblock,
2434 struct buffer_head *bh_result, int create);
2435 int ext4_dio_get_block(struct inode *inode, sector_t iblock,
2436 struct buffer_head *bh_result, int create);
2437 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2438 struct buffer_head *bh, int create);
2439 int ext4_walk_page_buffers(handle_t *handle,
2440 struct buffer_head *head,
2444 int (*fn)(handle_t *handle,
2445 struct buffer_head *bh));
2446 int do_journal_get_write_access(handle_t *handle,
2447 struct buffer_head *bh);
2448 #define FALL_BACK_TO_NONDELALLOC 1
2449 #define CONVERT_INLINE_DATA 2
2452 EXT4_IGET_NORMAL = 0,
2453 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */
2454 EXT4_IGET_HANDLE = 0x0002 /* Inode # is from a handle */
2457 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
2458 ext4_iget_flags flags, const char *function,
2461 #define ext4_iget(sb, ino, flags) \
2462 __ext4_iget((sb), (ino), (flags), __func__, __LINE__)
2464 extern int ext4_write_inode(struct inode *, struct writeback_control *);
2465 extern int ext4_setattr(struct dentry *, struct iattr *);
2466 extern int ext4_getattr(const struct path *, struct kstat *, u32, unsigned int);
2467 extern void ext4_evict_inode(struct inode *);
2468 extern void ext4_clear_inode(struct inode *);
2469 extern int ext4_file_getattr(const struct path *, struct kstat *, u32, unsigned int);
2470 extern int ext4_sync_inode(handle_t *, struct inode *);
2471 extern void ext4_dirty_inode(struct inode *, int);
2472 extern int ext4_change_inode_journal_flag(struct inode *, int);
2473 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2474 extern int ext4_inode_attach_jinode(struct inode *inode);
2475 extern int ext4_can_truncate(struct inode *inode);
2476 extern int ext4_truncate(struct inode *);
2477 extern int ext4_break_layouts(struct inode *);
2478 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2479 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
2480 extern void ext4_set_inode_flags(struct inode *);
2481 extern int ext4_alloc_da_blocks(struct inode *inode);
2482 extern void ext4_set_aops(struct inode *inode);
2483 extern int ext4_writepage_trans_blocks(struct inode *);
2484 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2485 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2486 loff_t lstart, loff_t lend);
2487 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf);
2488 extern vm_fault_t ext4_filemap_fault(struct vm_fault *vmf);
2489 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2490 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
2491 extern void ext4_da_release_space(struct inode *inode, int to_free);
2492 extern void ext4_da_update_reserve_space(struct inode *inode,
2493 int used, int quota_claim);
2494 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
2495 ext4_fsblk_t pblk, ext4_lblk_t len);
2498 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2499 struct ext4_map_blocks *map, int flags);
2500 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2501 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2502 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2503 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2504 ext4_lblk_t start, ext4_lblk_t end);
2507 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2508 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2511 extern int ext4_ext_migrate(struct inode *);
2512 extern int ext4_ind_migrate(struct inode *inode);
2515 extern int ext4_dirent_csum_verify(struct inode *inode,
2516 struct ext4_dir_entry *dirent);
2517 extern int ext4_orphan_add(handle_t *, struct inode *);
2518 extern int ext4_orphan_del(handle_t *, struct inode *);
2519 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2520 __u32 start_minor_hash, __u32 *next_hash);
2521 extern int ext4_search_dir(struct buffer_head *bh,
2525 struct ext4_filename *fname,
2526 unsigned int offset,
2527 struct ext4_dir_entry_2 **res_dir);
2528 extern int ext4_generic_delete_entry(handle_t *handle,
2530 struct ext4_dir_entry_2 *de_del,
2531 struct buffer_head *bh,
2535 extern bool ext4_empty_dir(struct inode *inode);
2538 extern int ext4_group_add(struct super_block *sb,
2539 struct ext4_new_group_data *input);
2540 extern int ext4_group_extend(struct super_block *sb,
2541 struct ext4_super_block *es,
2542 ext4_fsblk_t n_blocks_count);
2543 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2546 extern struct buffer_head *ext4_sb_bread(struct super_block *sb,
2547 sector_t block, int op_flags);
2548 extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
2549 extern int ext4_calculate_overhead(struct super_block *sb);
2550 extern void ext4_superblock_csum_set(struct super_block *sb);
2551 extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2552 extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2553 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2554 ext4_group_t ngroup);
2555 extern const char *ext4_decode_error(struct super_block *sb, int errno,
2557 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb,
2558 ext4_group_t block_group,
2559 unsigned int flags);
2561 extern __printf(4, 5)
2562 void __ext4_error(struct super_block *, const char *, unsigned int,
2564 extern __printf(5, 6)
2565 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2567 extern __printf(5, 6)
2568 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2570 extern void __ext4_std_error(struct super_block *, const char *,
2572 extern __printf(4, 5)
2573 void __ext4_abort(struct super_block *, const char *, unsigned int,
2575 extern __printf(4, 5)
2576 void __ext4_warning(struct super_block *, const char *, unsigned int,
2578 extern __printf(4, 5)
2579 void __ext4_warning_inode(const struct inode *inode, const char *function,
2580 unsigned int line, const char *fmt, ...);
2581 extern __printf(3, 4)
2582 void __ext4_msg(struct super_block *, const char *, const char *, ...);
2583 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2584 const char *, unsigned int, const char *);
2585 extern __printf(7, 8)
2586 void __ext4_grp_locked_error(const char *, unsigned int,
2587 struct super_block *, ext4_group_t,
2588 unsigned long, ext4_fsblk_t,
2591 #define EXT4_ERROR_INODE(inode, fmt, a...) \
2592 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
2594 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
2595 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
2597 #define EXT4_ERROR_FILE(file, block, fmt, a...) \
2598 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
2600 #ifdef CONFIG_PRINTK
2602 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2603 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__)
2604 #define ext4_error_file(file, func, line, block, fmt, ...) \
2605 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2606 #define ext4_error(sb, fmt, ...) \
2607 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2608 #define ext4_abort(sb, fmt, ...) \
2609 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2610 #define ext4_warning(sb, fmt, ...) \
2611 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2612 #define ext4_warning_inode(inode, fmt, ...) \
2613 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
2614 #define ext4_msg(sb, level, fmt, ...) \
2615 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2616 #define dump_mmp_msg(sb, mmp, msg) \
2617 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2618 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2619 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2624 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2626 no_printk(fmt, ##__VA_ARGS__); \
2627 __ext4_error_inode(inode, "", 0, block, " "); \
2629 #define ext4_error_file(file, func, line, block, fmt, ...) \
2631 no_printk(fmt, ##__VA_ARGS__); \
2632 __ext4_error_file(file, "", 0, block, " "); \
2634 #define ext4_error(sb, fmt, ...) \
2636 no_printk(fmt, ##__VA_ARGS__); \
2637 __ext4_error(sb, "", 0, " "); \
2639 #define ext4_abort(sb, fmt, ...) \
2641 no_printk(fmt, ##__VA_ARGS__); \
2642 __ext4_abort(sb, "", 0, " "); \
2644 #define ext4_warning(sb, fmt, ...) \
2646 no_printk(fmt, ##__VA_ARGS__); \
2647 __ext4_warning(sb, "", 0, " "); \
2649 #define ext4_warning_inode(inode, fmt, ...) \
2651 no_printk(fmt, ##__VA_ARGS__); \
2652 __ext4_warning_inode(inode, "", 0, " "); \
2654 #define ext4_msg(sb, level, fmt, ...) \
2656 no_printk(fmt, ##__VA_ARGS__); \
2657 __ext4_msg(sb, "", " "); \
2659 #define dump_mmp_msg(sb, mmp, msg) \
2660 __dump_mmp_msg(sb, mmp, "", 0, "")
2661 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2663 no_printk(fmt, ##__VA_ARGS__); \
2664 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
2669 extern void ext4_update_dynamic_rev(struct super_block *sb);
2670 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2672 extern int ext4_update_rocompat_feature(handle_t *handle,
2673 struct super_block *sb, __u32 rocompat);
2674 extern int ext4_update_incompat_feature(handle_t *handle,
2675 struct super_block *sb, __u32 incompat);
2676 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2677 struct ext4_group_desc *bg);
2678 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2679 struct ext4_group_desc *bg);
2680 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2681 struct ext4_group_desc *bg);
2682 extern __u32 ext4_free_group_clusters(struct super_block *sb,
2683 struct ext4_group_desc *bg);
2684 extern __u32 ext4_free_inodes_count(struct super_block *sb,
2685 struct ext4_group_desc *bg);
2686 extern __u32 ext4_used_dirs_count(struct super_block *sb,
2687 struct ext4_group_desc *bg);
2688 extern __u32 ext4_itable_unused_count(struct super_block *sb,
2689 struct ext4_group_desc *bg);
2690 extern void ext4_block_bitmap_set(struct super_block *sb,
2691 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2692 extern void ext4_inode_bitmap_set(struct super_block *sb,
2693 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2694 extern void ext4_inode_table_set(struct super_block *sb,
2695 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2696 extern void ext4_free_group_clusters_set(struct super_block *sb,
2697 struct ext4_group_desc *bg,
2699 extern void ext4_free_inodes_set(struct super_block *sb,
2700 struct ext4_group_desc *bg, __u32 count);
2701 extern void ext4_used_dirs_set(struct super_block *sb,
2702 struct ext4_group_desc *bg, __u32 count);
2703 extern void ext4_itable_unused_set(struct super_block *sb,
2704 struct ext4_group_desc *bg, __u32 count);
2705 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2706 struct ext4_group_desc *gdp);
2707 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2708 struct ext4_group_desc *gdp);
2709 extern int ext4_register_li_request(struct super_block *sb,
2710 ext4_group_t first_not_zeroed);
2712 static inline int ext4_has_metadata_csum(struct super_block *sb)
2714 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
2715 !EXT4_SB(sb)->s_chksum_driver);
2717 return ext4_has_feature_metadata_csum(sb) &&
2718 (EXT4_SB(sb)->s_chksum_driver != NULL);
2721 static inline int ext4_has_group_desc_csum(struct super_block *sb)
2723 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb);
2726 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2728 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2729 le32_to_cpu(es->s_blocks_count_lo);
2732 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2734 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2735 le32_to_cpu(es->s_r_blocks_count_lo);
2738 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2740 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2741 le32_to_cpu(es->s_free_blocks_count_lo);
2744 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2747 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2748 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2751 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2754 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2755 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2758 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2761 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2762 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2765 static inline loff_t ext4_isize(struct super_block *sb,
2766 struct ext4_inode *raw_inode)
2768 if (ext4_has_feature_largedir(sb) ||
2769 S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2770 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2771 le32_to_cpu(raw_inode->i_size_lo);
2773 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2776 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2778 raw_inode->i_size_lo = cpu_to_le32(i_size);
2779 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2783 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2786 struct ext4_group_info ***grp_info;
2787 long indexv, indexh;
2788 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
2789 grp_info = EXT4_SB(sb)->s_group_info;
2790 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2791 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2792 return grp_info[indexv][indexh];
2796 * Reading s_groups_count requires using smp_rmb() afterwards. See
2797 * the locking protocol documented in the comments of ext4_group_add()
2800 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2802 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
2808 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2809 ext4_group_t block_group)
2811 return block_group >> sbi->s_log_groups_per_flex;
2814 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2816 return 1 << sbi->s_log_groups_per_flex;
2819 #define ext4_std_error(sb, errno) \
2822 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
2826 /* Each CPU can accumulate percpu_counter_batch clusters in their local
2827 * counters. So we need to make sure we have free clusters more
2828 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
2830 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2832 #define EXT4_FREECLUSTERS_WATERMARK 0
2835 /* Update i_disksize. Requires i_mutex to avoid races with truncate */
2836 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2838 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
2839 !inode_is_locked(inode));
2840 down_write(&EXT4_I(inode)->i_data_sem);
2841 if (newsize > EXT4_I(inode)->i_disksize)
2842 EXT4_I(inode)->i_disksize = newsize;
2843 up_write(&EXT4_I(inode)->i_data_sem);
2846 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
2847 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
2851 if (newsize > inode->i_size) {
2852 i_size_write(inode, newsize);
2855 if (newsize > EXT4_I(inode)->i_disksize) {
2856 ext4_update_i_disksize(inode, newsize);
2862 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
2865 struct ext4_group_info {
2866 unsigned long bb_state;
2867 struct rb_root bb_free_root;
2868 ext4_grpblk_t bb_first_free; /* first free block */
2869 ext4_grpblk_t bb_free; /* total free blocks */
2870 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
2871 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
2872 struct list_head bb_prealloc_list;
2876 struct rw_semaphore alloc_sem;
2877 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
2878 * regions, index is order.
2879 * bb_counters[3] = 5 means
2880 * 5 free 8-block regions. */
2883 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
2884 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
2885 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
2886 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
2887 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \
2888 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT)
2889 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \
2890 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT)
2892 #define EXT4_MB_GRP_NEED_INIT(grp) \
2893 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2894 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
2895 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2896 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
2897 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2899 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
2900 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2901 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
2902 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2903 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
2904 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2906 #define EXT4_MAX_CONTENTION 8
2907 #define EXT4_CONTENTION_THRESHOLD 2
2909 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2912 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2916 * Returns true if the filesystem is busy enough that attempts to
2917 * access the block group locks has run into contention.
2919 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2921 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2924 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2926 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2927 if (spin_trylock(lock))
2929 * We're able to grab the lock right away, so drop the
2930 * lock contention counter.
2932 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2935 * The lock is busy, so bump the contention counter,
2936 * and then wait on the spin lock.
2938 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2939 EXT4_MAX_CONTENTION);
2944 static inline void ext4_unlock_group(struct super_block *sb,
2947 spin_unlock(ext4_group_lock_ptr(sb, group));
2951 * Block validity checking
2953 #define ext4_check_indirect_blockref(inode, bh) \
2954 ext4_check_blockref(__func__, __LINE__, inode, \
2955 (__le32 *)(bh)->b_data, \
2956 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2958 #define ext4_ind_check_inode(inode) \
2959 ext4_check_blockref(__func__, __LINE__, inode, \
2960 EXT4_I(inode)->i_data, \
2964 * Inodes and files operations
2968 extern const struct file_operations ext4_dir_operations;
2971 extern const struct inode_operations ext4_file_inode_operations;
2972 extern const struct file_operations ext4_file_operations;
2973 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
2976 extern int ext4_get_max_inline_size(struct inode *inode);
2977 extern int ext4_find_inline_data_nolock(struct inode *inode);
2978 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
2980 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
2982 extern int ext4_readpage_inline(struct inode *inode, struct page *page);
2983 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
2984 struct inode *inode,
2985 loff_t pos, unsigned len,
2987 struct page **pagep);
2988 extern int ext4_write_inline_data_end(struct inode *inode,
2989 loff_t pos, unsigned len,
2992 extern struct buffer_head *
2993 ext4_journalled_write_inline_data(struct inode *inode,
2996 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
2997 struct inode *inode,
2998 loff_t pos, unsigned len,
3000 struct page **pagep,
3002 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
3003 unsigned len, unsigned copied,
3005 extern int ext4_try_add_inline_entry(handle_t *handle,
3006 struct ext4_filename *fname,
3007 struct inode *dir, struct inode *inode);
3008 extern int ext4_try_create_inline_dir(handle_t *handle,
3009 struct inode *parent,
3010 struct inode *inode);
3011 extern int ext4_read_inline_dir(struct file *filp,
3012 struct dir_context *ctx,
3013 int *has_inline_data);
3014 extern int htree_inlinedir_to_tree(struct file *dir_file,
3015 struct inode *dir, ext4_lblk_t block,
3016 struct dx_hash_info *hinfo,
3017 __u32 start_hash, __u32 start_minor_hash,
3018 int *has_inline_data);
3019 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3020 struct ext4_filename *fname,
3021 struct ext4_dir_entry_2 **res_dir,
3022 int *has_inline_data);
3023 extern int ext4_delete_inline_entry(handle_t *handle,
3025 struct ext4_dir_entry_2 *de_del,
3026 struct buffer_head *bh,
3027 int *has_inline_data);
3028 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3029 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3030 struct ext4_dir_entry_2 **parent_de,
3032 extern int ext4_inline_data_fiemap(struct inode *inode,
3033 struct fiemap_extent_info *fieinfo,
3034 int *has_inline, __u64 start, __u64 len);
3037 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap);
3039 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3041 extern int ext4_convert_inline_data(struct inode *inode);
3043 static inline int ext4_has_inline_data(struct inode *inode)
3045 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3046 EXT4_I(inode)->i_inline_off;
3050 extern const struct inode_operations ext4_dir_inode_operations;
3051 extern const struct inode_operations ext4_special_inode_operations;
3052 extern struct dentry *ext4_get_parent(struct dentry *child);
3053 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3054 struct ext4_dir_entry_2 *de,
3055 int blocksize, int csum_size,
3056 unsigned int parent_ino, int dotdot_real_len);
3057 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t,
3058 unsigned int blocksize);
3059 extern int ext4_handle_dirty_dirent_node(handle_t *handle,
3060 struct inode *inode,
3061 struct buffer_head *bh);
3063 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = {
3064 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
3065 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
3066 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
3067 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
3068 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
3069 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
3070 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
3073 static inline void ext4_set_de_type(struct super_block *sb,
3074 struct ext4_dir_entry_2 *de,
3076 if (ext4_has_feature_filetype(sb))
3077 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3081 extern int ext4_mpage_readpages(struct address_space *mapping,
3082 struct list_head *pages, struct page *page,
3083 unsigned nr_pages, bool is_readahead);
3086 extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3087 extern const struct inode_operations ext4_symlink_inode_operations;
3088 extern const struct inode_operations ext4_fast_symlink_inode_operations;
3091 extern int ext4_register_sysfs(struct super_block *sb);
3092 extern void ext4_unregister_sysfs(struct super_block *sb);
3093 extern int __init ext4_init_sysfs(void);
3094 extern void ext4_exit_sysfs(void);
3096 /* block_validity */
3097 extern void ext4_release_system_zone(struct super_block *sb);
3098 extern int ext4_setup_system_zone(struct super_block *sb);
3099 extern int __init ext4_init_system_zone(void);
3100 extern void ext4_exit_system_zone(void);
3101 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
3102 ext4_fsblk_t start_blk,
3103 unsigned int count);
3104 extern int ext4_check_blockref(const char *, unsigned int,
3105 struct inode *, __le32 *, unsigned int);
3108 struct ext4_ext_path;
3112 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3115 #define EXT_MAX_BLOCKS 0xffffffff
3117 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
3118 extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
3119 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3120 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3121 struct ext4_map_blocks *map, int flags);
3122 extern int ext4_ext_truncate(handle_t *, struct inode *);
3123 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3125 extern void ext4_ext_init(struct super_block *);
3126 extern void ext4_ext_release(struct super_block *);
3127 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3129 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3130 loff_t offset, ssize_t len);
3131 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3132 struct ext4_map_blocks *map, int flags);
3133 extern int ext4_ext_calc_metadata_amount(struct inode *inode,
3134 ext4_lblk_t lblocks);
3135 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3137 struct ext4_ext_path *path);
3138 extern int ext4_can_extents_be_merged(struct inode *inode,
3139 struct ext4_extent *ex1,
3140 struct ext4_extent *ex2);
3141 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3142 struct ext4_ext_path **,
3143 struct ext4_extent *, int);
3144 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3145 struct ext4_ext_path **,
3147 extern void ext4_ext_drop_refs(struct ext4_ext_path *);
3148 extern int ext4_ext_check_inode(struct inode *inode);
3149 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3150 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3151 __u64 start, __u64 len);
3152 extern int ext4_ext_precache(struct inode *inode);
3153 extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len);
3154 extern int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len);
3155 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3156 struct inode *inode2, ext4_lblk_t lblk1,
3157 ext4_lblk_t lblk2, ext4_lblk_t count,
3158 int mark_unwritten,int *err);
3159 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu);
3162 extern void ext4_double_down_write_data_sem(struct inode *first,
3163 struct inode *second);
3164 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3165 struct inode *donor_inode);
3166 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3167 __u64 start_orig, __u64 start_donor,
3168 __u64 len, __u64 *moved_len);
3171 extern int __init ext4_init_pageio(void);
3172 extern void ext4_exit_pageio(void);
3173 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3174 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3175 extern int ext4_put_io_end(ext4_io_end_t *io_end);
3176 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3177 extern void ext4_io_submit_init(struct ext4_io_submit *io,
3178 struct writeback_control *wbc);
3179 extern void ext4_end_io_rsv_work(struct work_struct *work);
3180 extern void ext4_io_submit(struct ext4_io_submit *io);
3181 extern int ext4_bio_write_page(struct ext4_io_submit *io,
3184 struct writeback_control *wbc,
3188 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3191 * Add new method to test whether block and inode bitmaps are properly
3192 * initialized. With uninit_bg reading the block from disk is not enough
3193 * to mark the bitmap uptodate. We need to also zero-out the bitmap
3195 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3197 static inline int bitmap_uptodate(struct buffer_head *bh)
3199 return (buffer_uptodate(bh) &&
3200 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3202 static inline void set_bitmap_uptodate(struct buffer_head *bh)
3204 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3207 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
3209 /* For ioend & aio unwritten conversion wait queues */
3210 #define EXT4_WQ_HASH_SZ 37
3211 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
3213 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3215 extern int ext4_resize_begin(struct super_block *sb);
3216 extern void ext4_resize_end(struct super_block *sb);
3218 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
3219 struct ext4_io_end *io_end)
3221 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
3222 io_end->flag |= EXT4_IO_END_UNWRITTEN;
3223 atomic_inc(&EXT4_I(inode)->i_unwritten);
3227 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3229 struct inode *inode = io_end->inode;
3231 if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
3232 io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3233 /* Wake up anyone waiting on unwritten extent conversion */
3234 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
3235 wake_up_all(ext4_ioend_wq(inode));
3239 extern const struct iomap_ops ext4_iomap_ops;
3241 #endif /* __KERNEL__ */
3243 #define EFSBADCRC EBADMSG /* Bad CRC detected */
3244 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
3246 #endif /* _EXT4_H */