4 * Copyright (C) 1992, 1993, 1994, 1995
5 * Remy Card (card@masi.ibp.fr)
6 * Laboratoire MASI - Institut Blaise Pascal
7 * Universite Pierre et Marie Curie (Paris VI)
11 * linux/include/linux/minix_fs.h
13 * Copyright (C) 1991, 1992 Linus Torvalds
19 #include <linux/types.h>
20 #include <linux/blkdev.h>
21 #include <linux/magic.h>
22 #include <linux/jbd2.h>
23 #include <linux/quota.h>
24 #include <linux/rwsem.h>
25 #include <linux/rbtree.h>
26 #include <linux/seqlock.h>
27 #include <linux/mutex.h>
28 #include <linux/timer.h>
29 #include <linux/version.h>
30 #include <linux/wait.h>
31 #include <linux/blockgroup_lock.h>
32 #include <linux/percpu_counter.h>
33 #include <linux/ratelimit.h>
34 #include <crypto/hash.h>
35 #include <linux/falloc.h>
37 #include <linux/compat.h>
41 * The fourth extended filesystem constants/structures
45 * with AGGRESSIVE_CHECK allocator runs consistency checks over
46 * structures. these checks slow things down a lot
48 #define AGGRESSIVE_CHECK__
51 * with DOUBLE_CHECK defined mballoc creates persistent in-core
52 * bitmaps, maintains and uses them to check for double allocations
54 #define DOUBLE_CHECK__
57 * Define EXT4FS_DEBUG to produce debug messages
65 #define ext4_debug(f, a...) \
67 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
68 __FILE__, __LINE__, __func__); \
69 printk(KERN_DEBUG f, ## a); \
72 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
76 * Turn on EXT_DEBUG to get lots of info about extents operations.
80 #define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__)
82 #define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
85 /* data type for block offset of block group */
86 typedef int ext4_grpblk_t;
88 /* data type for filesystem-wide blocks number */
89 typedef unsigned long long ext4_fsblk_t;
91 /* data type for file logical block number */
92 typedef __u32 ext4_lblk_t;
94 /* data type for block group number */
95 typedef unsigned int ext4_group_t;
97 enum SHIFT_DIRECTION {
103 * Flags used in mballoc's allocation_context flags field.
105 * Also used to show what's going on for debugging purposes when the
106 * flag field is exported via the traceport interface
109 /* prefer goal again. length */
110 #define EXT4_MB_HINT_MERGE 0x0001
111 /* blocks already reserved */
112 #define EXT4_MB_HINT_RESERVED 0x0002
113 /* metadata is being allocated */
114 #define EXT4_MB_HINT_METADATA 0x0004
115 /* first blocks in the file */
116 #define EXT4_MB_HINT_FIRST 0x0008
117 /* search for the best chunk */
118 #define EXT4_MB_HINT_BEST 0x0010
119 /* data is being allocated */
120 #define EXT4_MB_HINT_DATA 0x0020
121 /* don't preallocate (for tails) */
122 #define EXT4_MB_HINT_NOPREALLOC 0x0040
123 /* allocate for locality group */
124 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080
125 /* allocate goal blocks or none */
126 #define EXT4_MB_HINT_GOAL_ONLY 0x0100
127 /* goal is meaningful */
128 #define EXT4_MB_HINT_TRY_GOAL 0x0200
129 /* blocks already pre-reserved by delayed allocation */
130 #define EXT4_MB_DELALLOC_RESERVED 0x0400
131 /* We are doing stream allocation */
132 #define EXT4_MB_STREAM_ALLOC 0x0800
133 /* Use reserved root blocks if needed */
134 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000
135 /* Use blocks from reserved pool */
136 #define EXT4_MB_USE_RESERVED 0x2000
138 struct ext4_allocation_request {
139 /* target inode for block we're allocating */
141 /* how many blocks we want to allocate */
143 /* logical block in target inode */
145 /* the closest logical allocated block to the left */
147 /* the closest logical allocated block to the right */
149 /* phys. target (a hint) */
151 /* phys. block for the closest logical allocated block to the left */
153 /* phys. block for the closest logical allocated block to the right */
155 /* flags. see above EXT4_MB_HINT_* */
160 * Logical to physical block mapping, used by ext4_map_blocks()
162 * This structure is used to pass requests into ext4_map_blocks() as
163 * well as to store the information returned by ext4_map_blocks(). It
164 * takes less room on the stack than a struct buffer_head.
166 #define EXT4_MAP_NEW (1 << BH_New)
167 #define EXT4_MAP_MAPPED (1 << BH_Mapped)
168 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten)
169 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary)
170 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
171 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
173 struct ext4_map_blocks {
177 unsigned int m_flags;
181 * Flags for ext4_io_end->flags
183 #define EXT4_IO_END_UNWRITTEN 0x0001
186 * For converting unwritten extents on a work queue. 'handle' is used for
187 * buffered writeback.
189 typedef struct ext4_io_end {
190 struct list_head list; /* per-file finished IO list */
191 handle_t *handle; /* handle reserved for extent
193 struct inode *inode; /* file being written to */
194 struct bio *bio; /* Linked list of completed
195 * bios covering the extent */
196 unsigned int flag; /* unwritten or not */
197 atomic_t count; /* reference counter */
198 loff_t offset; /* offset in the file */
199 ssize_t size; /* size of the extent */
202 struct ext4_io_submit {
203 struct writeback_control *io_wbc;
205 ext4_io_end_t *io_end;
206 sector_t io_next_block;
210 * Special inodes numbers
212 #define EXT4_BAD_INO 1 /* Bad blocks inode */
213 #define EXT4_ROOT_INO 2 /* Root inode */
214 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */
215 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
216 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
217 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
218 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
219 #define EXT4_JOURNAL_INO 8 /* Journal inode */
221 /* First non-reserved inode for old ext4 filesystems */
222 #define EXT4_GOOD_OLD_FIRST_INO 11
225 * Maximal count of links to a file
227 #define EXT4_LINK_MAX 65000
230 * Macro-instructions used to manage several block sizes
232 #define EXT4_MIN_BLOCK_SIZE 1024
233 #define EXT4_MAX_BLOCK_SIZE 65536
234 #define EXT4_MIN_BLOCK_LOG_SIZE 10
235 #define EXT4_MAX_BLOCK_LOG_SIZE 16
237 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
239 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
241 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
242 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
243 EXT4_SB(s)->s_cluster_bits)
245 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
246 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
248 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
251 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
252 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
253 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
255 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
256 EXT4_GOOD_OLD_INODE_SIZE : \
258 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
259 EXT4_GOOD_OLD_FIRST_INO : \
262 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
264 /* Translate a block number to a cluster number */
265 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
266 /* Translate a cluster number to a block number */
267 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
268 /* Translate # of blks to # of clusters */
269 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
270 (sbi)->s_cluster_bits)
271 /* Mask out the low bits to get the starting block of the cluster */
272 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
273 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
274 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
275 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
276 /* Get the cluster offset */
277 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
278 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
279 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
280 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
283 * Structure of a blocks group descriptor
285 struct ext4_group_desc
287 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
288 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
289 __le32 bg_inode_table_lo; /* Inodes table block */
290 __le16 bg_free_blocks_count_lo;/* Free blocks count */
291 __le16 bg_free_inodes_count_lo;/* Free inodes count */
292 __le16 bg_used_dirs_count_lo; /* Directories count */
293 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
294 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
295 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
296 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
297 __le16 bg_itable_unused_lo; /* Unused inodes count */
298 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
299 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
300 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
301 __le32 bg_inode_table_hi; /* Inodes table block MSB */
302 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
303 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
304 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
305 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
306 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
307 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
308 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
312 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
313 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
315 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
316 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
320 * Structure of a flex block group info
324 atomic64_t free_clusters;
325 atomic_t free_inodes;
329 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
330 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
331 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
334 * Macro-instructions used to manage group descriptors
336 #define EXT4_MIN_DESC_SIZE 32
337 #define EXT4_MIN_DESC_SIZE_64BIT 64
338 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
339 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
341 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
342 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
343 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
344 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
345 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
347 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
348 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
349 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
353 * Constants relative to the data blocks
355 #define EXT4_NDIR_BLOCKS 12
356 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
357 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
358 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
359 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
364 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
365 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
366 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
367 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
368 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
369 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
370 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
371 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
372 /* Reserved for compression usage... */
373 #define EXT4_DIRTY_FL 0x00000100
374 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
375 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
376 /* nb: was previously EXT2_ECOMPR_FL */
377 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */
378 /* End compression flags --- maybe not all used */
379 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
380 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
381 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
382 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
383 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
384 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
385 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
386 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
387 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
388 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
389 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
390 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
391 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
393 #define EXT4_FL_USER_VISIBLE 0x304BDFFF /* User visible flags */
394 #define EXT4_FL_USER_MODIFIABLE 0x204380FF /* User modifiable flags */
396 #define EXT4_FL_XFLAG_VISIBLE (EXT4_SYNC_FL | \
397 EXT4_IMMUTABLE_FL | \
403 /* Flags that should be inherited by new inodes from their parent. */
404 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
405 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
406 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
407 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
410 /* Flags that are appropriate for regular files (all but dir-specific ones). */
411 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
413 /* Flags that are appropriate for non-directories/regular files. */
414 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
416 /* Mask out flags that are inappropriate for the given type of inode. */
417 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
421 else if (S_ISREG(mode))
422 return flags & EXT4_REG_FLMASK;
424 return flags & EXT4_OTHER_FLMASK;
428 * Inode flags used for atomic set/get
431 EXT4_INODE_SECRM = 0, /* Secure deletion */
432 EXT4_INODE_UNRM = 1, /* Undelete */
433 EXT4_INODE_COMPR = 2, /* Compress file */
434 EXT4_INODE_SYNC = 3, /* Synchronous updates */
435 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
436 EXT4_INODE_APPEND = 5, /* writes to file may only append */
437 EXT4_INODE_NODUMP = 6, /* do not dump file */
438 EXT4_INODE_NOATIME = 7, /* do not update atime */
439 /* Reserved for compression usage... */
440 EXT4_INODE_DIRTY = 8,
441 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
442 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
443 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */
444 /* End compression flags --- maybe not all used */
445 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
446 EXT4_INODE_IMAGIC = 13, /* AFS directory */
447 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
448 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
449 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
450 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
451 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
452 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
453 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
454 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
455 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
456 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */
457 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
461 * Since it's pretty easy to mix up bit numbers and hex values, we use a
462 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
463 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
464 * any extra space in the compiled kernel image, otherwise, the build will fail.
465 * It's important that these values are the same, since we are using
466 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
467 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
468 * values found in ext2, ext3 and ext4 filesystems, and of course the values
469 * defined in e2fsprogs.
471 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
473 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
474 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
476 static inline void ext4_check_flag_values(void)
478 CHECK_FLAG_VALUE(SECRM);
479 CHECK_FLAG_VALUE(UNRM);
480 CHECK_FLAG_VALUE(COMPR);
481 CHECK_FLAG_VALUE(SYNC);
482 CHECK_FLAG_VALUE(IMMUTABLE);
483 CHECK_FLAG_VALUE(APPEND);
484 CHECK_FLAG_VALUE(NODUMP);
485 CHECK_FLAG_VALUE(NOATIME);
486 CHECK_FLAG_VALUE(DIRTY);
487 CHECK_FLAG_VALUE(COMPRBLK);
488 CHECK_FLAG_VALUE(NOCOMPR);
489 CHECK_FLAG_VALUE(ENCRYPT);
490 CHECK_FLAG_VALUE(INDEX);
491 CHECK_FLAG_VALUE(IMAGIC);
492 CHECK_FLAG_VALUE(JOURNAL_DATA);
493 CHECK_FLAG_VALUE(NOTAIL);
494 CHECK_FLAG_VALUE(DIRSYNC);
495 CHECK_FLAG_VALUE(TOPDIR);
496 CHECK_FLAG_VALUE(HUGE_FILE);
497 CHECK_FLAG_VALUE(EXTENTS);
498 CHECK_FLAG_VALUE(EA_INODE);
499 CHECK_FLAG_VALUE(EOFBLOCKS);
500 CHECK_FLAG_VALUE(INLINE_DATA);
501 CHECK_FLAG_VALUE(PROJINHERIT);
502 CHECK_FLAG_VALUE(RESERVED);
505 /* Used to pass group descriptor data when online resize is done */
506 struct ext4_new_group_input {
507 __u32 group; /* Group number for this data */
508 __u64 block_bitmap; /* Absolute block number of block bitmap */
509 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
510 __u64 inode_table; /* Absolute block number of inode table start */
511 __u32 blocks_count; /* Total number of blocks in this group */
512 __u16 reserved_blocks; /* Number of reserved blocks in this group */
516 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
517 struct compat_ext4_new_group_input {
519 compat_u64 block_bitmap;
520 compat_u64 inode_bitmap;
521 compat_u64 inode_table;
528 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
529 struct ext4_new_group_data {
535 __u16 reserved_blocks;
537 __u32 free_blocks_count;
540 /* Indexes used to index group tables in ext4_new_group_data */
542 BLOCK_BITMAP = 0, /* block bitmap */
543 INODE_BITMAP, /* inode bitmap */
544 INODE_TABLE, /* inode tables */
549 * Flags used by ext4_map_blocks()
551 /* Allocate any needed blocks and/or convert an unwritten
552 extent to be an initialized ext4 */
553 #define EXT4_GET_BLOCKS_CREATE 0x0001
554 /* Request the creation of an unwritten extent */
555 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002
556 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\
557 EXT4_GET_BLOCKS_CREATE)
558 /* Caller is from the delayed allocation writeout path
559 * finally doing the actual allocation of delayed blocks */
560 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
561 /* caller is from the direct IO path, request to creation of an
562 unwritten extents if not allocated, split the unwritten
563 extent if blocks has been preallocated already*/
564 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
565 #define EXT4_GET_BLOCKS_CONVERT 0x0010
566 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
567 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
568 /* Convert extent to initialized after IO complete */
569 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
570 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
571 /* Eventual metadata allocation (due to growing extent tree)
572 * should not fail, so try to use reserved blocks for that.*/
573 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
574 /* Don't normalize allocation size (used for fallocate) */
575 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
576 /* Request will not result in inode size update (user for fallocate) */
577 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080
578 /* Convert written extents to unwritten */
579 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100
580 /* Write zeros to newly created written extents */
581 #define EXT4_GET_BLOCKS_ZERO 0x0200
582 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\
583 EXT4_GET_BLOCKS_ZERO)
584 /* Caller will submit data before dropping transaction handle. This
585 * allows jbd2 to avoid submitting data before commit. */
586 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400
589 * The bit position of these flags must not overlap with any of the
590 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(),
591 * read_extent_tree_block(), ext4_split_extent_at(),
592 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
593 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
594 * caching the extents when reading from the extent tree while a
595 * truncate or punch hole operation is in progress.
597 #define EXT4_EX_NOCACHE 0x40000000
598 #define EXT4_EX_FORCE_CACHE 0x20000000
601 * Flags used by ext4_free_blocks
603 #define EXT4_FREE_BLOCKS_METADATA 0x0001
604 #define EXT4_FREE_BLOCKS_FORGET 0x0002
605 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
606 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
607 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
608 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
610 /* Encryption algorithms */
611 #define EXT4_ENCRYPTION_MODE_INVALID 0
612 #define EXT4_ENCRYPTION_MODE_AES_256_XTS 1
613 #define EXT4_ENCRYPTION_MODE_AES_256_GCM 2
614 #define EXT4_ENCRYPTION_MODE_AES_256_CBC 3
615 #define EXT4_ENCRYPTION_MODE_AES_256_CTS 4
617 #include "ext4_crypto.h"
622 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
623 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
624 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
625 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
626 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
627 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
628 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
629 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
630 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
631 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
632 #define EXT4_IOC_MIGRATE _IO('f', 9)
633 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
634 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
635 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
636 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
637 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
638 #define EXT4_IOC_SWAP_BOOT _IO('f', 17)
639 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18)
640 #define EXT4_IOC_SET_ENCRYPTION_POLICY _IOR('f', 19, struct ext4_encryption_policy)
641 #define EXT4_IOC_GET_ENCRYPTION_PWSALT _IOW('f', 20, __u8[16])
642 #define EXT4_IOC_GET_ENCRYPTION_POLICY _IOW('f', 21, struct ext4_encryption_policy)
644 #ifndef FS_IOC_FSGETXATTR
645 /* Until the uapi changes get merged for project quota... */
647 #define FS_IOC_FSGETXATTR _IOR('X', 31, struct fsxattr)
648 #define FS_IOC_FSSETXATTR _IOW('X', 32, struct fsxattr)
651 * Structure for FS_IOC_FSGETXATTR and FS_IOC_FSSETXATTR.
654 __u32 fsx_xflags; /* xflags field value (get/set) */
655 __u32 fsx_extsize; /* extsize field value (get/set)*/
656 __u32 fsx_nextents; /* nextents field value (get) */
657 __u32 fsx_projid; /* project identifier (get/set) */
658 unsigned char fsx_pad[12];
662 * Flags for the fsx_xflags field
664 #define FS_XFLAG_REALTIME 0x00000001 /* data in realtime volume */
665 #define FS_XFLAG_PREALLOC 0x00000002 /* preallocated file extents */
666 #define FS_XFLAG_IMMUTABLE 0x00000008 /* file cannot be modified */
667 #define FS_XFLAG_APPEND 0x00000010 /* all writes append */
668 #define FS_XFLAG_SYNC 0x00000020 /* all writes synchronous */
669 #define FS_XFLAG_NOATIME 0x00000040 /* do not update access time */
670 #define FS_XFLAG_NODUMP 0x00000080 /* do not include in backups */
671 #define FS_XFLAG_RTINHERIT 0x00000100 /* create with rt bit set */
672 #define FS_XFLAG_PROJINHERIT 0x00000200 /* create with parents projid */
673 #define FS_XFLAG_NOSYMLINKS 0x00000400 /* disallow symlink creation */
674 #define FS_XFLAG_EXTSIZE 0x00000800 /* extent size allocator hint */
675 #define FS_XFLAG_EXTSZINHERIT 0x00001000 /* inherit inode extent size */
676 #define FS_XFLAG_NODEFRAG 0x00002000 /* do not defragment */
677 #define FS_XFLAG_FILESTREAM 0x00004000 /* use filestream allocator */
678 #define FS_XFLAG_HASATTR 0x80000000 /* no DIFLAG for this */
679 #endif /* !defined(FS_IOC_FSGETXATTR) */
681 #define EXT4_IOC_FSGETXATTR FS_IOC_FSGETXATTR
682 #define EXT4_IOC_FSSETXATTR FS_IOC_FSSETXATTR
684 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
686 * ioctl commands in 32 bit emulation
688 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
689 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
690 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
691 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
692 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
693 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
694 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
695 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
696 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
697 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
700 /* Max physical block we can address w/o extents */
701 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
704 * Structure of an inode on the disk
707 __le16 i_mode; /* File mode */
708 __le16 i_uid; /* Low 16 bits of Owner Uid */
709 __le32 i_size_lo; /* Size in bytes */
710 __le32 i_atime; /* Access time */
711 __le32 i_ctime; /* Inode Change time */
712 __le32 i_mtime; /* Modification time */
713 __le32 i_dtime; /* Deletion Time */
714 __le16 i_gid; /* Low 16 bits of Group Id */
715 __le16 i_links_count; /* Links count */
716 __le32 i_blocks_lo; /* Blocks count */
717 __le32 i_flags; /* File flags */
723 __u32 h_i_translator;
728 } osd1; /* OS dependent 1 */
729 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
730 __le32 i_generation; /* File version (for NFS) */
731 __le32 i_file_acl_lo; /* File ACL */
733 __le32 i_obso_faddr; /* Obsoleted fragment address */
736 __le16 l_i_blocks_high; /* were l_i_reserved1 */
737 __le16 l_i_file_acl_high;
738 __le16 l_i_uid_high; /* these 2 fields */
739 __le16 l_i_gid_high; /* were reserved2[0] */
740 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
744 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
751 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
752 __le16 m_i_file_acl_high;
753 __u32 m_i_reserved2[2];
755 } osd2; /* OS dependent 2 */
756 __le16 i_extra_isize;
757 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
758 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
759 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
760 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
761 __le32 i_crtime; /* File Creation time */
762 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
763 __le32 i_version_hi; /* high 32 bits for 64-bit version */
764 __le32 i_projid; /* Project ID */
768 __u32 reserved; /* should be zero */
769 __u32 donor_fd; /* donor file descriptor */
770 __u64 orig_start; /* logical start offset in block for orig */
771 __u64 donor_start; /* logical start offset in block for donor */
772 __u64 len; /* block length to be moved */
773 __u64 moved_len; /* moved block length */
776 #define EXT4_EPOCH_BITS 2
777 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
778 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
781 * Extended fields will fit into an inode if the filesystem was formatted
782 * with large inodes (-I 256 or larger) and there are not currently any EAs
783 * consuming all of the available space. For new inodes we always reserve
784 * enough space for the kernel's known extended fields, but for inodes
785 * created with an old kernel this might not have been the case. None of
786 * the extended inode fields is critical for correct filesystem operation.
787 * This macro checks if a certain field fits in the inode. Note that
788 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
790 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
791 ((offsetof(typeof(*ext4_inode), field) + \
792 sizeof((ext4_inode)->field)) \
793 <= (EXT4_GOOD_OLD_INODE_SIZE + \
794 (einode)->i_extra_isize)) \
797 * We use an encoding that preserves the times for extra epoch "00":
799 * extra msb of adjust for signed
800 * epoch 32-bit 32-bit tv_sec to
801 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
802 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
803 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
804 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
805 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
806 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
807 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
808 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
809 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
811 * Note that previous versions of the kernel on 64-bit systems would
812 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
813 * 1970. e2fsck will correct this, assuming that it is run on the
814 * affected filesystem before 2242.
817 static inline __le32 ext4_encode_extra_time(struct timespec *time)
819 u32 extra = sizeof(time->tv_sec) > 4 ?
820 ((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK : 0;
821 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
824 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
826 if (unlikely(sizeof(time->tv_sec) > 4 &&
827 (extra & cpu_to_le32(EXT4_EPOCH_MASK)))) {
828 #if LINUX_VERSION_CODE < KERNEL_VERSION(4,20,0)
829 /* Handle legacy encoding of pre-1970 dates with epoch
830 * bits 1,1. We assume that by kernel version 4.20,
831 * everyone will have run fsck over the affected
832 * filesystems to correct the problem. (This
833 * backwards compatibility may be removed before this
834 * time, at the discretion of the ext4 developers.)
836 u64 extra_bits = le32_to_cpu(extra) & EXT4_EPOCH_MASK;
837 if (extra_bits == 3 && ((time->tv_sec) & 0x80000000) != 0)
839 time->tv_sec += extra_bits << 32;
841 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
844 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
847 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
849 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
850 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
851 (raw_inode)->xtime ## _extra = \
852 ext4_encode_extra_time(&(inode)->xtime); \
855 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
857 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
858 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
859 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
860 (raw_inode)->xtime ## _extra = \
861 ext4_encode_extra_time(&(einode)->xtime); \
864 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
866 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
867 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
868 ext4_decode_extra_time(&(inode)->xtime, \
869 raw_inode->xtime ## _extra); \
871 (inode)->xtime.tv_nsec = 0; \
874 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
876 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
877 (einode)->xtime.tv_sec = \
878 (signed)le32_to_cpu((raw_inode)->xtime); \
880 (einode)->xtime.tv_sec = 0; \
881 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
882 ext4_decode_extra_time(&(einode)->xtime, \
883 raw_inode->xtime ## _extra); \
885 (einode)->xtime.tv_nsec = 0; \
888 #define i_disk_version osd1.linux1.l_i_version
890 #if defined(__KERNEL__) || defined(__linux__)
891 #define i_reserved1 osd1.linux1.l_i_reserved1
892 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
893 #define i_blocks_high osd2.linux2.l_i_blocks_high
894 #define i_uid_low i_uid
895 #define i_gid_low i_gid
896 #define i_uid_high osd2.linux2.l_i_uid_high
897 #define i_gid_high osd2.linux2.l_i_gid_high
898 #define i_checksum_lo osd2.linux2.l_i_checksum_lo
900 #elif defined(__GNU__)
902 #define i_translator osd1.hurd1.h_i_translator
903 #define i_uid_high osd2.hurd2.h_i_uid_high
904 #define i_gid_high osd2.hurd2.h_i_gid_high
905 #define i_author osd2.hurd2.h_i_author
907 #elif defined(__masix__)
909 #define i_reserved1 osd1.masix1.m_i_reserved1
910 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
911 #define i_reserved2 osd2.masix2.m_i_reserved2
913 #endif /* defined(__KERNEL__) || defined(__linux__) */
915 #include "extents_status.h"
918 * Lock subclasses for i_data_sem in the ext4_inode_info structure.
920 * These are needed to avoid lockdep false positives when we need to
921 * allocate blocks to the quota inode during ext4_map_blocks(), while
922 * holding i_data_sem for a normal (non-quota) inode. Since we don't
923 * do quota tracking for the quota inode, this avoids deadlock (as
924 * well as infinite recursion, since it isn't turtles all the way
927 * I_DATA_SEM_NORMAL - Used for most inodes
928 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode
929 * where the second inode has larger inode number
931 * I_DATA_SEM_QUOTA - Used for quota inodes only
934 I_DATA_SEM_NORMAL = 0,
941 * fourth extended file system inode data in memory
943 struct ext4_inode_info {
944 __le32 i_data[15]; /* unconverted */
946 ext4_fsblk_t i_file_acl;
949 * i_block_group is the number of the block group which contains
950 * this file's inode. Constant across the lifetime of the inode,
951 * it is ued for making block allocation decisions - we try to
952 * place a file's data blocks near its inode block, and new inodes
953 * near to their parent directory's inode.
955 ext4_group_t i_block_group;
956 ext4_lblk_t i_dir_start_lookup;
957 #if (BITS_PER_LONG < 64)
958 unsigned long i_state_flags; /* Dynamic state flags */
960 unsigned long i_flags;
963 * Extended attributes can be read independently of the main file
964 * data. Taking i_mutex even when reading would cause contention
965 * between readers of EAs and writers of regular file data, so
966 * instead we synchronize on xattr_sem when reading or changing
969 struct rw_semaphore xattr_sem;
971 struct list_head i_orphan; /* unlinked but open inodes */
974 * i_disksize keeps track of what the inode size is ON DISK, not
975 * in memory. During truncate, i_size is set to the new size by
976 * the VFS prior to calling ext4_truncate(), but the filesystem won't
977 * set i_disksize to 0 until the truncate is actually under way.
979 * The intent is that i_disksize always represents the blocks which
980 * are used by this file. This allows recovery to restart truncate
981 * on orphans if we crash during truncate. We actually write i_disksize
982 * into the on-disk inode when writing inodes out, instead of i_size.
984 * The only time when i_disksize and i_size may be different is when
985 * a truncate is in progress. The only things which change i_disksize
986 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
991 * i_data_sem is for serialising ext4_truncate() against
992 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
993 * data tree are chopped off during truncate. We can't do that in
994 * ext4 because whenever we perform intermediate commits during
995 * truncate, the inode and all the metadata blocks *must* be in a
996 * consistent state which allows truncation of the orphans to restart
997 * during recovery. Hence we must fix the get_block-vs-truncate race
998 * by other means, so we have i_data_sem.
1000 struct rw_semaphore i_data_sem;
1002 * i_mmap_sem is for serializing page faults with truncate / punch hole
1003 * operations. We have to make sure that new page cannot be faulted in
1004 * a section of the inode that is being punched. We cannot easily use
1005 * i_data_sem for this since we need protection for the whole punch
1006 * operation and i_data_sem ranks below transaction start so we have
1007 * to occasionally drop it.
1009 struct rw_semaphore i_mmap_sem;
1010 struct inode vfs_inode;
1011 struct jbd2_inode *jinode;
1013 spinlock_t i_raw_lock; /* protects updates to the raw inode */
1016 * File creation time. Its function is same as that of
1017 * struct timespec i_{a,c,m}time in the generic inode.
1019 struct timespec i_crtime;
1022 struct list_head i_prealloc_list;
1023 spinlock_t i_prealloc_lock;
1025 /* extents status tree */
1026 struct ext4_es_tree i_es_tree;
1028 struct list_head i_es_list;
1029 unsigned int i_es_all_nr; /* protected by i_es_lock */
1030 unsigned int i_es_shk_nr; /* protected by i_es_lock */
1031 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for
1032 extents to shrink. Protected by
1036 ext4_group_t i_last_alloc_group;
1038 /* allocation reservation info for delalloc */
1039 /* In case of bigalloc, these refer to clusters rather than blocks */
1040 unsigned int i_reserved_data_blocks;
1041 unsigned int i_reserved_meta_blocks;
1042 unsigned int i_allocated_meta_blocks;
1043 ext4_lblk_t i_da_metadata_calc_last_lblock;
1044 int i_da_metadata_calc_len;
1046 /* on-disk additional length */
1047 __u16 i_extra_isize;
1049 /* Indicate the inline data space. */
1054 /* quota space reservation, managed internally by quota code */
1055 qsize_t i_reserved_quota;
1058 /* Lock protecting lists below */
1059 spinlock_t i_completed_io_lock;
1061 * Completed IOs that need unwritten extents handling and have
1062 * transaction reserved
1064 struct list_head i_rsv_conversion_list;
1065 struct work_struct i_rsv_conversion_work;
1066 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1068 spinlock_t i_block_reservation_lock;
1071 * Transactions that contain inode's metadata needed to complete
1072 * fsync and fdatasync, respectively.
1075 tid_t i_datasync_tid;
1078 struct dquot *i_dquot[MAXQUOTAS];
1081 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1084 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1085 /* Encryption params */
1086 struct ext4_crypt_info *i_crypt_info;
1092 * File system states
1094 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
1095 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
1096 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
1099 * Misc. filesystem flags
1101 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
1102 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
1103 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
1106 * Mount flags set via mount options or defaults
1108 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
1109 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
1110 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
1111 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
1112 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
1113 #define EXT4_MOUNT_ERRORS_MASK 0x00070
1114 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
1115 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
1116 #ifdef CONFIG_FS_DAX
1117 #define EXT4_MOUNT_DAX 0x00200 /* Direct Access */
1119 #define EXT4_MOUNT_DAX 0
1121 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
1122 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
1123 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
1124 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
1125 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
1126 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
1127 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
1128 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
1129 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
1130 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
1131 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
1132 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
1133 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
1134 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
1135 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
1136 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1137 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
1138 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
1139 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
1140 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
1141 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
1144 * Mount flags set either automatically (could not be set by mount option)
1145 * based on per file system feature or property or in special cases such as
1146 * distinguishing between explicit mount option definition and default.
1148 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1149 specified delalloc */
1150 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1151 size of blocksize * 8
1153 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1156 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly
1157 specified journal checksum */
1159 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1161 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1163 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1166 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1168 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1170 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1173 #define ext4_test_and_set_bit __test_and_set_bit_le
1174 #define ext4_set_bit __set_bit_le
1175 #define ext4_set_bit_atomic ext2_set_bit_atomic
1176 #define ext4_test_and_clear_bit __test_and_clear_bit_le
1177 #define ext4_clear_bit __clear_bit_le
1178 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
1179 #define ext4_test_bit test_bit_le
1180 #define ext4_find_next_zero_bit find_next_zero_bit_le
1181 #define ext4_find_next_bit find_next_bit_le
1183 extern void ext4_set_bits(void *bm, int cur, int len);
1186 * Maximal mount counts between two filesystem checks
1188 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1189 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1192 * Behaviour when detecting errors
1194 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1195 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1196 #define EXT4_ERRORS_PANIC 3 /* Panic */
1197 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1199 /* Metadata checksum algorithm codes */
1200 #define EXT4_CRC32C_CHKSUM 1
1203 * Structure of the super block
1205 struct ext4_super_block {
1206 /*00*/ __le32 s_inodes_count; /* Inodes count */
1207 __le32 s_blocks_count_lo; /* Blocks count */
1208 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1209 __le32 s_free_blocks_count_lo; /* Free blocks count */
1210 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1211 __le32 s_first_data_block; /* First Data Block */
1212 __le32 s_log_block_size; /* Block size */
1213 __le32 s_log_cluster_size; /* Allocation cluster size */
1214 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1215 __le32 s_clusters_per_group; /* # Clusters per group */
1216 __le32 s_inodes_per_group; /* # Inodes per group */
1217 __le32 s_mtime; /* Mount time */
1218 /*30*/ __le32 s_wtime; /* Write time */
1219 __le16 s_mnt_count; /* Mount count */
1220 __le16 s_max_mnt_count; /* Maximal mount count */
1221 __le16 s_magic; /* Magic signature */
1222 __le16 s_state; /* File system state */
1223 __le16 s_errors; /* Behaviour when detecting errors */
1224 __le16 s_minor_rev_level; /* minor revision level */
1225 /*40*/ __le32 s_lastcheck; /* time of last check */
1226 __le32 s_checkinterval; /* max. time between checks */
1227 __le32 s_creator_os; /* OS */
1228 __le32 s_rev_level; /* Revision level */
1229 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1230 __le16 s_def_resgid; /* Default gid for reserved blocks */
1232 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1234 * Note: the difference between the compatible feature set and
1235 * the incompatible feature set is that if there is a bit set
1236 * in the incompatible feature set that the kernel doesn't
1237 * know about, it should refuse to mount the filesystem.
1239 * e2fsck's requirements are more strict; if it doesn't know
1240 * about a feature in either the compatible or incompatible
1241 * feature set, it must abort and not try to meddle with
1242 * things it doesn't understand...
1244 __le32 s_first_ino; /* First non-reserved inode */
1245 __le16 s_inode_size; /* size of inode structure */
1246 __le16 s_block_group_nr; /* block group # of this superblock */
1247 __le32 s_feature_compat; /* compatible feature set */
1248 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1249 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1250 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1251 /*78*/ char s_volume_name[16]; /* volume name */
1252 /*88*/ char s_last_mounted[64]; /* directory where last mounted */
1253 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1255 * Performance hints. Directory preallocation should only
1256 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1258 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1259 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1260 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1262 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1264 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1265 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1266 __le32 s_journal_dev; /* device number of journal file */
1267 __le32 s_last_orphan; /* start of list of inodes to delete */
1268 __le32 s_hash_seed[4]; /* HTREE hash seed */
1269 __u8 s_def_hash_version; /* Default hash version to use */
1270 __u8 s_jnl_backup_type;
1271 __le16 s_desc_size; /* size of group descriptor */
1272 /*100*/ __le32 s_default_mount_opts;
1273 __le32 s_first_meta_bg; /* First metablock block group */
1274 __le32 s_mkfs_time; /* When the filesystem was created */
1275 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1276 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1277 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1278 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1279 __le32 s_free_blocks_count_hi; /* Free blocks count */
1280 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1281 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1282 __le32 s_flags; /* Miscellaneous flags */
1283 __le16 s_raid_stride; /* RAID stride */
1284 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1285 __le64 s_mmp_block; /* Block for multi-mount protection */
1286 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1287 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1288 __u8 s_checksum_type; /* metadata checksum algorithm used */
1289 __u8 s_encryption_level; /* versioning level for encryption */
1290 __u8 s_reserved_pad; /* Padding to next 32bits */
1291 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1292 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1293 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1294 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1295 snapshot's future use */
1296 __le32 s_snapshot_list; /* inode number of the head of the
1297 on-disk snapshot list */
1298 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1299 __le32 s_error_count; /* number of fs errors */
1300 __le32 s_first_error_time; /* first time an error happened */
1301 __le32 s_first_error_ino; /* inode involved in first error */
1302 __le64 s_first_error_block; /* block involved of first error */
1303 __u8 s_first_error_func[32]; /* function where the error happened */
1304 __le32 s_first_error_line; /* line number where error happened */
1305 __le32 s_last_error_time; /* most recent time of an error */
1306 __le32 s_last_error_ino; /* inode involved in last error */
1307 __le32 s_last_error_line; /* line number where error happened */
1308 __le64 s_last_error_block; /* block involved of last error */
1309 __u8 s_last_error_func[32]; /* function where the error happened */
1310 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1311 __u8 s_mount_opts[64];
1312 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1313 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1314 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1315 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
1316 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */
1317 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */
1318 __le32 s_lpf_ino; /* Location of the lost+found inode */
1319 __le32 s_prj_quota_inum; /* inode for tracking project quota */
1320 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */
1321 __le32 s_reserved[98]; /* Padding to the end of the block */
1322 __le32 s_checksum; /* crc32c(superblock) */
1325 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1330 * run-time mount flags
1332 #define EXT4_MF_MNTDIR_SAMPLED 0x0001
1333 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1334 #define EXT4_MF_TEST_DUMMY_ENCRYPTION 0x0004
1336 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1337 #define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \
1338 EXT4_MF_TEST_DUMMY_ENCRYPTION))
1340 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1343 /* Number of quota types we support */
1344 #define EXT4_MAXQUOTAS 3
1347 * fourth extended-fs super-block data in memory
1349 struct ext4_sb_info {
1350 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1351 unsigned long s_inodes_per_block;/* Number of inodes per block */
1352 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1353 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1354 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1355 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1356 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1357 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1358 ext4_group_t s_groups_count; /* Number of groups in the fs */
1359 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1360 unsigned long s_overhead; /* # of fs overhead clusters */
1361 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1362 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1363 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1364 struct buffer_head * s_sbh; /* Buffer containing the super block */
1365 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1366 struct buffer_head **s_group_desc;
1367 unsigned int s_mount_opt;
1368 unsigned int s_mount_opt2;
1369 unsigned int s_mount_flags;
1370 unsigned int s_def_mount_opt;
1371 ext4_fsblk_t s_sb_block;
1372 atomic64_t s_resv_clusters;
1375 unsigned short s_mount_state;
1376 unsigned short s_pad;
1377 int s_addr_per_block_bits;
1378 int s_desc_per_block_bits;
1381 unsigned int s_inode_readahead_blks;
1382 unsigned int s_inode_goal;
1383 spinlock_t s_next_gen_lock;
1384 u32 s_next_generation;
1386 int s_def_hash_version;
1387 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1388 struct percpu_counter s_freeclusters_counter;
1389 struct percpu_counter s_freeinodes_counter;
1390 struct percpu_counter s_dirs_counter;
1391 struct percpu_counter s_dirtyclusters_counter;
1392 struct blockgroup_lock *s_blockgroup_lock;
1393 struct proc_dir_entry *s_proc;
1394 struct kobject s_kobj;
1395 struct completion s_kobj_unregister;
1396 struct super_block *s_sb;
1399 struct journal_s *s_journal;
1400 struct list_head s_orphan;
1401 struct mutex s_orphan_lock;
1402 unsigned long s_resize_flags; /* Flags indicating if there
1404 unsigned long s_commit_interval;
1405 u32 s_max_batch_time;
1406 u32 s_min_batch_time;
1407 struct block_device *journal_bdev;
1409 char *s_qf_names[EXT4_MAXQUOTAS]; /* Names of quota files with journalled quota */
1410 int s_jquota_fmt; /* Format of quota to use */
1412 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1413 struct rb_root system_blks;
1415 #ifdef EXTENTS_STATS
1416 /* ext4 extents stats */
1417 unsigned long s_ext_min;
1418 unsigned long s_ext_max;
1419 unsigned long s_depth_max;
1420 spinlock_t s_ext_stats_lock;
1421 unsigned long s_ext_blocks;
1422 unsigned long s_ext_extents;
1425 /* for buddy allocator */
1426 struct ext4_group_info ***s_group_info;
1427 struct inode *s_buddy_cache;
1428 spinlock_t s_md_lock;
1429 unsigned short *s_mb_offsets;
1430 unsigned int *s_mb_maxs;
1431 unsigned int s_group_info_size;
1434 unsigned long s_stripe;
1435 unsigned int s_mb_stream_request;
1436 unsigned int s_mb_max_to_scan;
1437 unsigned int s_mb_min_to_scan;
1438 unsigned int s_mb_stats;
1439 unsigned int s_mb_order2_reqs;
1440 unsigned int s_mb_group_prealloc;
1441 unsigned int s_max_dir_size_kb;
1442 /* where last allocation was done - for stream allocation */
1443 unsigned long s_mb_last_group;
1444 unsigned long s_mb_last_start;
1446 /* stats for buddy allocator */
1447 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1448 atomic_t s_bal_success; /* we found long enough chunks */
1449 atomic_t s_bal_allocated; /* in blocks */
1450 atomic_t s_bal_ex_scanned; /* total extents scanned */
1451 atomic_t s_bal_goals; /* goal hits */
1452 atomic_t s_bal_breaks; /* too long searches */
1453 atomic_t s_bal_2orders; /* 2^order hits */
1454 spinlock_t s_bal_lock;
1455 unsigned long s_mb_buddies_generated;
1456 unsigned long long s_mb_generation_time;
1457 atomic_t s_mb_lost_chunks;
1458 atomic_t s_mb_preallocated;
1459 atomic_t s_mb_discarded;
1460 atomic_t s_lock_busy;
1462 /* locality groups */
1463 struct ext4_locality_group __percpu *s_locality_groups;
1465 /* for write statistics */
1466 unsigned long s_sectors_written_start;
1467 u64 s_kbytes_written;
1469 /* the size of zero-out chunk */
1470 unsigned int s_extent_max_zeroout_kb;
1472 unsigned int s_log_groups_per_flex;
1473 struct flex_groups *s_flex_groups;
1474 ext4_group_t s_flex_groups_allocated;
1476 /* workqueue for reserved extent conversions (buffered io) */
1477 struct workqueue_struct *rsv_conversion_wq;
1479 /* timer for periodic error stats printing */
1480 struct timer_list s_err_report;
1482 /* Lazy inode table initialization info */
1483 struct ext4_li_request *s_li_request;
1484 /* Wait multiplier for lazy initialization thread */
1485 unsigned int s_li_wait_mult;
1487 /* Kernel thread for multiple mount protection */
1488 struct task_struct *s_mmp_tsk;
1490 /* record the last minlen when FITRIM is called. */
1491 atomic_t s_last_trim_minblks;
1493 /* Reference to checksum algorithm driver via cryptoapi */
1494 struct crypto_shash *s_chksum_driver;
1496 /* Precomputed FS UUID checksum for seeding other checksums */
1499 /* Reclaim extents from extent status tree */
1500 struct shrinker s_es_shrinker;
1501 struct list_head s_es_list; /* List of inodes with reclaimable extents */
1503 struct ext4_es_stats s_es_stats;
1504 struct mb_cache *s_mb_cache;
1505 spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1507 /* Ratelimit ext4 messages. */
1508 struct ratelimit_state s_err_ratelimit_state;
1509 struct ratelimit_state s_warning_ratelimit_state;
1510 struct ratelimit_state s_msg_ratelimit_state;
1513 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1515 return sb->s_fs_info;
1517 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1519 return container_of(inode, struct ext4_inode_info, vfs_inode);
1522 static inline struct timespec ext4_current_time(struct inode *inode)
1524 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1525 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1528 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1530 return ino == EXT4_ROOT_INO ||
1531 ino == EXT4_USR_QUOTA_INO ||
1532 ino == EXT4_GRP_QUOTA_INO ||
1533 ino == EXT4_BOOT_LOADER_INO ||
1534 ino == EXT4_JOURNAL_INO ||
1535 ino == EXT4_RESIZE_INO ||
1536 (ino >= EXT4_FIRST_INO(sb) &&
1537 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1541 * Inode dynamic state flags
1544 EXT4_STATE_JDATA, /* journaled data exists */
1545 EXT4_STATE_NEW, /* inode is newly created */
1546 EXT4_STATE_XATTR, /* has in-inode xattrs */
1547 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1548 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1549 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1550 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
1551 EXT4_STATE_NEWENTRY, /* File just added to dir */
1552 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read
1554 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1555 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
1558 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1559 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1561 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1563 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1565 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1567 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1569 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1572 /* Add these declarations here only so that these functions can be
1573 * found by name. Otherwise, they are very hard to locate. */
1574 static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1575 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1576 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1577 EXT4_INODE_BIT_FNS(flag, flags, 0)
1579 /* Add these declarations here only so that these functions can be
1580 * found by name. Otherwise, they are very hard to locate. */
1581 static inline int ext4_test_inode_state(struct inode *inode, int bit);
1582 static inline void ext4_set_inode_state(struct inode *inode, int bit);
1583 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1584 #if (BITS_PER_LONG < 64)
1585 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1587 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1589 (ei)->i_state_flags = 0;
1592 EXT4_INODE_BIT_FNS(state, flags, 32)
1594 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1596 /* We depend on the fact that callers will set i_flags */
1600 /* Assume that user mode programs are passing in an ext4fs superblock, not
1601 * a kernel struct super_block. This will allow us to call the feature-test
1602 * macros from user land. */
1603 #define EXT4_SB(sb) (sb)
1607 * Returns true if the inode is inode is encrypted
1609 static inline int ext4_encrypted_inode(struct inode *inode)
1611 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1612 return ext4_test_inode_flag(inode, EXT4_INODE_ENCRYPT);
1618 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1621 * Codes for operating systems
1623 #define EXT4_OS_LINUX 0
1624 #define EXT4_OS_HURD 1
1625 #define EXT4_OS_MASIX 2
1626 #define EXT4_OS_FREEBSD 3
1627 #define EXT4_OS_LITES 4
1632 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1633 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1635 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1636 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1638 #define EXT4_GOOD_OLD_INODE_SIZE 128
1641 * Feature set definitions
1644 /* Use the ext4_{has,set,clear}_feature_* helpers; these will be removed */
1645 #define EXT4_HAS_COMPAT_FEATURE(sb,mask) \
1646 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1647 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \
1648 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1649 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \
1650 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1651 #define EXT4_SET_COMPAT_FEATURE(sb,mask) \
1652 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1653 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \
1654 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1655 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \
1656 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1657 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \
1658 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1659 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
1660 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1661 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \
1662 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1664 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1665 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1666 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1667 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1668 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1669 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1670 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
1672 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1673 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1674 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1675 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1676 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1677 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1678 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1679 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1680 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1682 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1683 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1684 * all other data structures' checksums. However, the METADATA_CSUM and
1685 * GDT_CSUM bits are mutually exclusive.
1687 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
1688 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000
1689 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000
1691 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1692 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1693 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1694 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1695 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1696 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1697 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1698 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1699 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1700 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1701 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1702 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000
1703 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
1704 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
1705 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000
1707 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
1708 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1710 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1711 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
1713 static inline void ext4_set_feature_##name(struct super_block *sb) \
1715 EXT4_SB(sb)->s_es->s_feature_compat |= \
1716 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1718 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1720 EXT4_SB(sb)->s_es->s_feature_compat &= \
1721 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1724 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
1725 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1727 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1728 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
1730 static inline void ext4_set_feature_##name(struct super_block *sb) \
1732 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
1733 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1735 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1737 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
1738 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1741 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
1742 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1744 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1745 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
1747 static inline void ext4_set_feature_##name(struct super_block *sb) \
1749 EXT4_SB(sb)->s_es->s_feature_incompat |= \
1750 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1752 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1754 EXT4_SB(sb)->s_es->s_feature_incompat &= \
1755 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1758 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC)
1759 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES)
1760 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL)
1761 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR)
1762 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE)
1763 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX)
1764 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2)
1766 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER)
1767 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE)
1768 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR)
1769 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE)
1770 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM)
1771 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK)
1772 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE)
1773 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA)
1774 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC)
1775 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM)
1776 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY)
1777 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT)
1779 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION)
1780 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE)
1781 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER)
1782 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV)
1783 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG)
1784 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS)
1785 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT)
1786 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP)
1787 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG)
1788 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE)
1789 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA)
1790 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED)
1791 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR)
1792 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA)
1793 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT)
1795 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1796 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1797 EXT4_FEATURE_INCOMPAT_META_BG)
1798 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1799 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1800 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1802 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1803 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1804 EXT4_FEATURE_INCOMPAT_RECOVER| \
1805 EXT4_FEATURE_INCOMPAT_META_BG)
1806 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1807 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1808 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1810 #define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1811 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1812 EXT4_FEATURE_INCOMPAT_RECOVER| \
1813 EXT4_FEATURE_INCOMPAT_META_BG| \
1814 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1815 EXT4_FEATURE_INCOMPAT_64BIT| \
1816 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1817 EXT4_FEATURE_INCOMPAT_MMP | \
1818 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
1819 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
1820 EXT4_FEATURE_INCOMPAT_CSUM_SEED)
1821 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1822 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1823 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1824 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1825 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1826 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1827 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1828 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1829 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1830 EXT4_FEATURE_RO_COMPAT_QUOTA |\
1831 EXT4_FEATURE_RO_COMPAT_PROJECT)
1833 #define EXTN_FEATURE_FUNCS(ver) \
1834 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
1836 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1837 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
1839 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
1841 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1842 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
1844 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
1846 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1847 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
1850 EXTN_FEATURE_FUNCS(2)
1851 EXTN_FEATURE_FUNCS(3)
1852 EXTN_FEATURE_FUNCS(4)
1854 static inline bool ext4_has_compat_features(struct super_block *sb)
1856 return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
1858 static inline bool ext4_has_ro_compat_features(struct super_block *sb)
1860 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
1862 static inline bool ext4_has_incompat_features(struct super_block *sb)
1864 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
1868 * Default values for user and/or group using reserved blocks
1870 #define EXT4_DEF_RESUID 0
1871 #define EXT4_DEF_RESGID 0
1874 * Default project ID
1876 #define EXT4_DEF_PROJID 0
1878 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
1881 * Default mount options
1883 #define EXT4_DEFM_DEBUG 0x0001
1884 #define EXT4_DEFM_BSDGROUPS 0x0002
1885 #define EXT4_DEFM_XATTR_USER 0x0004
1886 #define EXT4_DEFM_ACL 0x0008
1887 #define EXT4_DEFM_UID16 0x0010
1888 #define EXT4_DEFM_JMODE 0x0060
1889 #define EXT4_DEFM_JMODE_DATA 0x0020
1890 #define EXT4_DEFM_JMODE_ORDERED 0x0040
1891 #define EXT4_DEFM_JMODE_WBACK 0x0060
1892 #define EXT4_DEFM_NOBARRIER 0x0100
1893 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1894 #define EXT4_DEFM_DISCARD 0x0400
1895 #define EXT4_DEFM_NODELALLOC 0x0800
1898 * Default journal batch times
1900 #define EXT4_DEF_MIN_BATCH_TIME 0
1901 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1904 * Minimum number of groups in a flexgroup before we separate out
1905 * directories into the first block group of a flexgroup
1907 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1910 * Structure of a directory entry
1912 #define EXT4_NAME_LEN 255
1914 struct ext4_dir_entry {
1915 __le32 inode; /* Inode number */
1916 __le16 rec_len; /* Directory entry length */
1917 __le16 name_len; /* Name length */
1918 char name[EXT4_NAME_LEN]; /* File name */
1922 * The new version of the directory entry. Since EXT4 structures are
1923 * stored in intel byte order, and the name_len field could never be
1924 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1927 struct ext4_dir_entry_2 {
1928 __le32 inode; /* Inode number */
1929 __le16 rec_len; /* Directory entry length */
1930 __u8 name_len; /* Name length */
1932 char name[EXT4_NAME_LEN]; /* File name */
1936 * This is a bogus directory entry at the end of each leaf block that
1937 * records checksums.
1939 struct ext4_dir_entry_tail {
1940 __le32 det_reserved_zero1; /* Pretend to be unused */
1941 __le16 det_rec_len; /* 12 */
1942 __u8 det_reserved_zero2; /* Zero name length */
1943 __u8 det_reserved_ft; /* 0xDE, fake file type */
1944 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
1947 #define EXT4_DIRENT_TAIL(block, blocksize) \
1948 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1950 sizeof(struct ext4_dir_entry_tail))))
1953 * Ext4 directory file types. Only the low 3 bits are used. The
1954 * other bits are reserved for now.
1956 #define EXT4_FT_UNKNOWN 0
1957 #define EXT4_FT_REG_FILE 1
1958 #define EXT4_FT_DIR 2
1959 #define EXT4_FT_CHRDEV 3
1960 #define EXT4_FT_BLKDEV 4
1961 #define EXT4_FT_FIFO 5
1962 #define EXT4_FT_SOCK 6
1963 #define EXT4_FT_SYMLINK 7
1965 #define EXT4_FT_MAX 8
1967 #define EXT4_FT_DIR_CSUM 0xDE
1970 * EXT4_DIR_PAD defines the directory entries boundaries
1972 * NOTE: It must be a multiple of 4
1974 #define EXT4_DIR_PAD 4
1975 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1976 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1978 #define EXT4_MAX_REC_LEN ((1<<16)-1)
1981 * If we ever get support for fs block sizes > page_size, we'll need
1982 * to remove the #if statements in the next two functions...
1984 static inline unsigned int
1985 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1987 unsigned len = le16_to_cpu(dlen);
1989 #if (PAGE_SIZE >= 65536)
1990 if (len == EXT4_MAX_REC_LEN || len == 0)
1992 return (len & 65532) | ((len & 3) << 16);
1998 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
2000 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
2002 #if (PAGE_SIZE >= 65536)
2004 return cpu_to_le16(len);
2005 if (len == blocksize) {
2006 if (blocksize == 65536)
2007 return cpu_to_le16(EXT4_MAX_REC_LEN);
2009 return cpu_to_le16(0);
2011 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
2013 return cpu_to_le16(len);
2018 * Hash Tree Directory indexing
2019 * (c) Daniel Phillips, 2001
2022 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
2023 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2024 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
2025 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2027 /* Legal values for the dx_root hash_version field: */
2029 #define DX_HASH_LEGACY 0
2030 #define DX_HASH_HALF_MD4 1
2031 #define DX_HASH_TEA 2
2032 #define DX_HASH_LEGACY_UNSIGNED 3
2033 #define DX_HASH_HALF_MD4_UNSIGNED 4
2034 #define DX_HASH_TEA_UNSIGNED 5
2036 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2037 const void *address, unsigned int length)
2040 struct shash_desc shash;
2045 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2047 desc.shash.tfm = sbi->s_chksum_driver;
2048 desc.shash.flags = 0;
2049 *(u32 *)desc.ctx = crc;
2051 err = crypto_shash_update(&desc.shash, address, length);
2054 return *(u32 *)desc.ctx;
2059 /* hash info structure used by the directory hash */
2069 /* 32 and 64 bit signed EOF for dx directories */
2070 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
2071 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
2075 * Control parameters used by ext4_htree_next_block
2077 #define HASH_NB_ALWAYS 1
2079 struct ext4_filename {
2080 const struct qstr *usr_fname;
2081 struct ext4_str disk_name;
2082 struct dx_hash_info hinfo;
2083 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2084 struct ext4_str crypto_buf;
2088 #define fname_name(p) ((p)->disk_name.name)
2089 #define fname_len(p) ((p)->disk_name.len)
2092 * Describe an inode's exact location on disk and in memory
2096 struct buffer_head *bh;
2097 unsigned long offset;
2098 ext4_group_t block_group;
2101 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2103 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2107 * This structure is stuffed into the struct file's private_data field
2108 * for directories. It is where we put information so that we can do
2109 * readdir operations in hash tree order.
2111 struct dir_private_info {
2112 struct rb_root root;
2113 struct rb_node *curr_node;
2114 struct fname *extra_fname;
2117 __u32 curr_minor_hash;
2121 /* calculate the first block number of the group */
2122 static inline ext4_fsblk_t
2123 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2125 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2126 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2130 * Special error return code only used by dx_probe() and its callers.
2132 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
2135 * Timeout and state flag for lazy initialization inode thread.
2137 #define EXT4_DEF_LI_WAIT_MULT 10
2138 #define EXT4_DEF_LI_MAX_START_DELAY 5
2139 #define EXT4_LAZYINIT_QUIT 0x0001
2140 #define EXT4_LAZYINIT_RUNNING 0x0002
2143 * Lazy inode table initialization info
2145 struct ext4_lazy_init {
2146 unsigned long li_state;
2147 struct list_head li_request_list;
2148 struct mutex li_list_mtx;
2151 struct ext4_li_request {
2152 struct super_block *lr_super;
2153 struct ext4_sb_info *lr_sbi;
2154 ext4_group_t lr_next_group;
2155 struct list_head lr_request;
2156 unsigned long lr_next_sched;
2157 unsigned long lr_timeout;
2160 struct ext4_features {
2161 struct kobject f_kobj;
2162 struct completion f_kobj_unregister;
2166 * This structure will be used for multiple mount protection. It will be
2167 * written into the block number saved in the s_mmp_block field in the
2168 * superblock. Programs that check MMP should assume that if
2169 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2170 * to use the filesystem, regardless of how old the timestamp is.
2172 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
2173 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2174 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
2175 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
2178 __le32 mmp_magic; /* Magic number for MMP */
2179 __le32 mmp_seq; /* Sequence no. updated periodically */
2182 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2183 * purposes and do not affect the correctness of the algorithm
2185 __le64 mmp_time; /* Time last updated */
2186 char mmp_nodename[64]; /* Node which last updated MMP block */
2187 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
2190 * mmp_check_interval is used to verify if the MMP block has been
2191 * updated on the block device. The value is updated based on the
2192 * maximum time to write the MMP block during an update cycle.
2194 __le16 mmp_check_interval;
2197 __le32 mmp_pad2[226];
2198 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
2201 /* arguments passed to the mmp thread */
2203 struct buffer_head *bh; /* bh from initial read_mmp_block() */
2204 struct super_block *sb; /* super block of the fs */
2208 * Check interval multiplier
2209 * The MMP block is written every update interval and initially checked every
2210 * update interval x the multiplier (the value is then adapted based on the
2211 * write latency). The reason is that writes can be delayed under load and we
2212 * don't want readers to incorrectly assume that the filesystem is no longer
2215 #define EXT4_MMP_CHECK_MULT 2UL
2218 * Minimum interval for MMP checking in seconds.
2220 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
2223 * Maximum interval for MMP checking in seconds.
2225 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
2228 * Function prototypes
2232 * Ok, these declarations are also in <linux/kernel.h> but none of the
2233 * ext4 source programs needs to include it so they are duplicated here.
2235 # define NORET_TYPE /**/
2236 # define ATTRIB_NORET __attribute__((noreturn))
2237 # define NORET_AND noreturn,
2240 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2241 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2242 struct ext4_group_desc *gdp,
2243 struct buffer_head *bh, int sz);
2244 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2245 struct ext4_group_desc *gdp,
2246 struct buffer_head *bh, int sz);
2247 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2248 struct ext4_group_desc *gdp,
2249 struct buffer_head *bh);
2250 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2251 struct ext4_group_desc *gdp,
2252 struct buffer_head *bh);
2255 extern void ext4_get_group_no_and_offset(struct super_block *sb,
2256 ext4_fsblk_t blocknr,
2257 ext4_group_t *blockgrpp,
2258 ext4_grpblk_t *offsetp);
2259 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2260 ext4_fsblk_t block);
2262 extern unsigned int ext4_block_group(struct super_block *sb,
2263 ext4_fsblk_t blocknr);
2264 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2265 ext4_fsblk_t blocknr);
2266 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2267 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2268 ext4_group_t group);
2269 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2272 unsigned long *count,
2274 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2275 s64 nclusters, unsigned int flags);
2276 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2277 extern void ext4_check_blocks_bitmap(struct super_block *);
2278 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2279 ext4_group_t block_group,
2280 struct buffer_head ** bh);
2281 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2283 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2284 ext4_group_t block_group);
2285 extern int ext4_wait_block_bitmap(struct super_block *sb,
2286 ext4_group_t block_group,
2287 struct buffer_head *bh);
2288 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2289 ext4_group_t block_group);
2290 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2291 ext4_group_t block_group,
2292 struct ext4_group_desc *gdp);
2293 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2295 /* crypto_policy.c */
2296 int ext4_is_child_context_consistent_with_parent(struct inode *parent,
2297 struct inode *child);
2298 int ext4_inherit_context(struct inode *parent, struct inode *child);
2299 void ext4_to_hex(char *dst, char *src, size_t src_size);
2300 int ext4_process_policy(const struct ext4_encryption_policy *policy,
2301 struct inode *inode);
2302 int ext4_get_policy(struct inode *inode,
2303 struct ext4_encryption_policy *policy);
2306 extern struct kmem_cache *ext4_crypt_info_cachep;
2307 bool ext4_valid_contents_enc_mode(uint32_t mode);
2308 uint32_t ext4_validate_encryption_key_size(uint32_t mode, uint32_t size);
2309 extern struct workqueue_struct *ext4_read_workqueue;
2310 struct ext4_crypto_ctx *ext4_get_crypto_ctx(struct inode *inode,
2312 void ext4_release_crypto_ctx(struct ext4_crypto_ctx *ctx);
2313 void ext4_restore_control_page(struct page *data_page);
2314 struct page *ext4_encrypt(struct inode *inode,
2315 struct page *plaintext_page,
2317 int ext4_decrypt(struct page *page);
2318 int ext4_encrypted_zeroout(struct inode *inode, ext4_lblk_t lblk,
2319 ext4_fsblk_t pblk, ext4_lblk_t len);
2320 extern const struct dentry_operations ext4_encrypted_d_ops;
2322 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2323 int ext4_init_crypto(void);
2324 void ext4_exit_crypto(void);
2325 static inline int ext4_sb_has_crypto(struct super_block *sb)
2327 return ext4_has_feature_encrypt(sb);
2330 static inline int ext4_init_crypto(void) { return 0; }
2331 static inline void ext4_exit_crypto(void) { }
2332 static inline int ext4_sb_has_crypto(struct super_block *sb)
2338 /* crypto_fname.c */
2339 bool ext4_valid_filenames_enc_mode(uint32_t mode);
2340 u32 ext4_fname_crypto_round_up(u32 size, u32 blksize);
2341 unsigned ext4_fname_encrypted_size(struct inode *inode, u32 ilen);
2342 int ext4_fname_crypto_alloc_buffer(struct inode *inode,
2343 u32 ilen, struct ext4_str *crypto_str);
2344 int _ext4_fname_disk_to_usr(struct inode *inode,
2345 struct dx_hash_info *hinfo,
2346 const struct ext4_str *iname,
2347 struct ext4_str *oname);
2348 int ext4_fname_disk_to_usr(struct inode *inode,
2349 struct dx_hash_info *hinfo,
2350 const struct ext4_dir_entry_2 *de,
2351 struct ext4_str *oname);
2352 int ext4_fname_usr_to_disk(struct inode *inode,
2353 const struct qstr *iname,
2354 struct ext4_str *oname);
2355 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2356 void ext4_fname_crypto_free_buffer(struct ext4_str *crypto_str);
2357 int ext4_fname_setup_filename(struct inode *dir, const struct qstr *iname,
2358 int lookup, struct ext4_filename *fname);
2359 void ext4_fname_free_filename(struct ext4_filename *fname);
2362 int ext4_setup_fname_crypto(struct inode *inode)
2366 static inline void ext4_fname_crypto_free_buffer(struct ext4_str *p) { }
2367 static inline int ext4_fname_setup_filename(struct inode *dir,
2368 const struct qstr *iname,
2369 int lookup, struct ext4_filename *fname)
2371 fname->usr_fname = iname;
2372 fname->disk_name.name = (unsigned char *) iname->name;
2373 fname->disk_name.len = iname->len;
2376 static inline void ext4_fname_free_filename(struct ext4_filename *fname) { }
2381 void ext4_free_crypt_info(struct ext4_crypt_info *ci);
2382 void ext4_free_encryption_info(struct inode *inode, struct ext4_crypt_info *ci);
2383 int _ext4_get_encryption_info(struct inode *inode);
2385 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2386 int ext4_has_encryption_key(struct inode *inode);
2388 static inline int ext4_get_encryption_info(struct inode *inode)
2390 struct ext4_crypt_info *ci = EXT4_I(inode)->i_crypt_info;
2393 (ci->ci_keyring_key &&
2394 (ci->ci_keyring_key->flags & ((1 << KEY_FLAG_INVALIDATED) |
2395 (1 << KEY_FLAG_REVOKED) |
2396 (1 << KEY_FLAG_DEAD)))))
2397 return _ext4_get_encryption_info(inode);
2401 static inline struct ext4_crypt_info *ext4_encryption_info(struct inode *inode)
2403 return EXT4_I(inode)->i_crypt_info;
2407 static inline int ext4_has_encryption_key(struct inode *inode)
2411 static inline int ext4_get_encryption_info(struct inode *inode)
2415 static inline struct ext4_crypt_info *ext4_encryption_info(struct inode *inode)
2423 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2425 struct ext4_dir_entry_2 *,
2426 struct buffer_head *, char *, int,
2428 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2429 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2430 (de), (bh), (buf), (size), (offset)))
2431 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2433 struct ext4_dir_entry_2 *dirent,
2434 struct ext4_str *ent_name);
2435 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2436 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2437 struct buffer_head *bh,
2438 void *buf, int buf_size,
2439 struct ext4_filename *fname,
2440 struct ext4_dir_entry_2 **dest_de);
2441 int ext4_insert_dentry(struct inode *dir,
2442 struct inode *inode,
2443 struct ext4_dir_entry_2 *de,
2445 struct ext4_filename *fname);
2446 static inline void ext4_update_dx_flag(struct inode *inode)
2448 if (!ext4_has_feature_dir_index(inode->i_sb))
2449 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2451 static unsigned char ext4_filetype_table[] = {
2452 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2455 static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2457 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2460 return ext4_filetype_table[filetype];
2462 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2463 void *buf, int buf_size);
2466 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2469 extern int ext4fs_dirhash(const char *name, int len, struct
2470 dx_hash_info *hinfo);
2473 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2474 const struct qstr *qstr, __u32 goal,
2475 uid_t *owner, int handle_type,
2476 unsigned int line_no, int nblocks);
2478 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner) \
2479 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2481 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2483 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2484 (type), __LINE__, (nblocks))
2487 extern void ext4_free_inode(handle_t *, struct inode *);
2488 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2489 extern unsigned long ext4_count_free_inodes(struct super_block *);
2490 extern unsigned long ext4_count_dirs(struct super_block *);
2491 extern void ext4_check_inodes_bitmap(struct super_block *);
2492 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2493 extern int ext4_init_inode_table(struct super_block *sb,
2494 ext4_group_t group, int barrier);
2495 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2498 extern const struct file_operations ext4_seq_mb_groups_fops;
2499 extern long ext4_mb_stats;
2500 extern long ext4_mb_max_to_scan;
2501 extern int ext4_mb_init(struct super_block *);
2502 extern int ext4_mb_release(struct super_block *);
2503 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2504 struct ext4_allocation_request *, int *);
2505 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2506 extern void ext4_discard_preallocations(struct inode *);
2507 extern int __init ext4_init_mballoc(void);
2508 extern void ext4_exit_mballoc(void);
2509 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2510 struct buffer_head *bh, ext4_fsblk_t block,
2511 unsigned long count, int flags);
2512 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2513 ext4_group_t ngroups);
2514 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2515 ext4_group_t i, struct ext4_group_desc *desc);
2516 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2517 ext4_fsblk_t block, unsigned long count);
2518 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2521 int ext4_inode_is_fast_symlink(struct inode *inode);
2522 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2523 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2524 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2525 struct buffer_head *bh_result, int create);
2526 int ext4_dax_mmap_get_block(struct inode *inode, sector_t iblock,
2527 struct buffer_head *bh_result, int create);
2528 int ext4_get_block(struct inode *inode, sector_t iblock,
2529 struct buffer_head *bh_result, int create);
2530 int ext4_dio_get_block(struct inode *inode, sector_t iblock,
2531 struct buffer_head *bh_result, int create);
2532 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2533 struct buffer_head *bh, int create);
2534 int ext4_walk_page_buffers(handle_t *handle,
2535 struct buffer_head *head,
2539 int (*fn)(handle_t *handle,
2540 struct buffer_head *bh));
2541 int do_journal_get_write_access(handle_t *handle,
2542 struct buffer_head *bh);
2543 #define FALL_BACK_TO_NONDELALLOC 1
2544 #define CONVERT_INLINE_DATA 2
2546 extern struct inode *ext4_iget(struct super_block *, unsigned long);
2547 extern struct inode *ext4_iget_normal(struct super_block *, unsigned long);
2548 extern int ext4_write_inode(struct inode *, struct writeback_control *);
2549 extern int ext4_setattr(struct dentry *, struct iattr *);
2550 extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
2551 struct kstat *stat);
2552 extern void ext4_evict_inode(struct inode *);
2553 extern void ext4_clear_inode(struct inode *);
2554 extern int ext4_sync_inode(handle_t *, struct inode *);
2555 extern void ext4_dirty_inode(struct inode *, int);
2556 extern int ext4_change_inode_journal_flag(struct inode *, int);
2557 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2558 extern int ext4_inode_attach_jinode(struct inode *inode);
2559 extern int ext4_can_truncate(struct inode *inode);
2560 extern void ext4_truncate(struct inode *);
2561 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2562 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
2563 extern void ext4_set_inode_flags(struct inode *);
2564 extern void ext4_get_inode_flags(struct ext4_inode_info *);
2565 extern int ext4_alloc_da_blocks(struct inode *inode);
2566 extern void ext4_set_aops(struct inode *inode);
2567 extern int ext4_writepage_trans_blocks(struct inode *);
2568 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2569 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2570 loff_t lstart, loff_t lend);
2571 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
2572 extern int ext4_filemap_fault(struct vm_area_struct *vma, struct vm_fault *vmf);
2573 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2574 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
2575 extern void ext4_da_update_reserve_space(struct inode *inode,
2576 int used, int quota_claim);
2577 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
2578 ext4_fsblk_t pblk, ext4_lblk_t len);
2579 extern int ext4_get_next_extent(struct inode *inode, ext4_lblk_t lblk,
2580 unsigned int map_len,
2581 struct extent_status *result);
2584 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2585 struct ext4_map_blocks *map, int flags);
2586 extern ssize_t ext4_ind_direct_IO(struct kiocb *iocb, struct iov_iter *iter,
2588 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2589 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2590 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2591 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2592 ext4_lblk_t start, ext4_lblk_t end);
2595 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2596 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2599 extern int ext4_ext_migrate(struct inode *);
2600 extern int ext4_ind_migrate(struct inode *inode);
2603 extern int ext4_dirent_csum_verify(struct inode *inode,
2604 struct ext4_dir_entry *dirent);
2605 extern int ext4_orphan_add(handle_t *, struct inode *);
2606 extern int ext4_orphan_del(handle_t *, struct inode *);
2607 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2608 __u32 start_minor_hash, __u32 *next_hash);
2609 extern int ext4_search_dir(struct buffer_head *bh,
2613 struct ext4_filename *fname,
2614 const struct qstr *d_name,
2615 unsigned int offset,
2616 struct ext4_dir_entry_2 **res_dir);
2617 extern int ext4_generic_delete_entry(handle_t *handle,
2619 struct ext4_dir_entry_2 *de_del,
2620 struct buffer_head *bh,
2624 extern int ext4_empty_dir(struct inode *inode);
2627 extern int ext4_group_add(struct super_block *sb,
2628 struct ext4_new_group_data *input);
2629 extern int ext4_group_extend(struct super_block *sb,
2630 struct ext4_super_block *es,
2631 ext4_fsblk_t n_blocks_count);
2632 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2635 extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
2636 extern int ext4_calculate_overhead(struct super_block *sb);
2637 extern void ext4_superblock_csum_set(struct super_block *sb);
2638 extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2639 extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2640 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2641 ext4_group_t ngroup);
2642 extern const char *ext4_decode_error(struct super_block *sb, int errno,
2645 extern __printf(4, 5)
2646 void __ext4_error(struct super_block *, const char *, unsigned int,
2648 extern __printf(5, 6)
2649 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2651 extern __printf(5, 6)
2652 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2654 extern void __ext4_std_error(struct super_block *, const char *,
2656 extern __printf(4, 5)
2657 void __ext4_abort(struct super_block *, const char *, unsigned int,
2659 extern __printf(4, 5)
2660 void __ext4_warning(struct super_block *, const char *, unsigned int,
2662 extern __printf(4, 5)
2663 void __ext4_warning_inode(const struct inode *inode, const char *function,
2664 unsigned int line, const char *fmt, ...);
2665 extern __printf(3, 4)
2666 void __ext4_msg(struct super_block *, const char *, const char *, ...);
2667 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2668 const char *, unsigned int, const char *);
2669 extern __printf(7, 8)
2670 void __ext4_grp_locked_error(const char *, unsigned int,
2671 struct super_block *, ext4_group_t,
2672 unsigned long, ext4_fsblk_t,
2675 #define EXT4_ERROR_INODE(inode, fmt, a...) \
2676 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
2678 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
2679 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
2681 #define EXT4_ERROR_FILE(file, block, fmt, a...) \
2682 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
2684 #ifdef CONFIG_PRINTK
2686 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2687 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__)
2688 #define ext4_error_file(file, func, line, block, fmt, ...) \
2689 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2690 #define ext4_error(sb, fmt, ...) \
2691 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2692 #define ext4_abort(sb, fmt, ...) \
2693 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2694 #define ext4_warning(sb, fmt, ...) \
2695 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2696 #define ext4_warning_inode(inode, fmt, ...) \
2697 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
2698 #define ext4_msg(sb, level, fmt, ...) \
2699 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2700 #define dump_mmp_msg(sb, mmp, msg) \
2701 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2702 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2703 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2708 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2710 no_printk(fmt, ##__VA_ARGS__); \
2711 __ext4_error_inode(inode, "", 0, block, " "); \
2713 #define ext4_error_file(file, func, line, block, fmt, ...) \
2715 no_printk(fmt, ##__VA_ARGS__); \
2716 __ext4_error_file(file, "", 0, block, " "); \
2718 #define ext4_error(sb, fmt, ...) \
2720 no_printk(fmt, ##__VA_ARGS__); \
2721 __ext4_error(sb, "", 0, " "); \
2723 #define ext4_abort(sb, fmt, ...) \
2725 no_printk(fmt, ##__VA_ARGS__); \
2726 __ext4_abort(sb, "", 0, " "); \
2728 #define ext4_warning(sb, fmt, ...) \
2730 no_printk(fmt, ##__VA_ARGS__); \
2731 __ext4_warning(sb, "", 0, " "); \
2733 #define ext4_warning_inode(inode, fmt, ...) \
2735 no_printk(fmt, ##__VA_ARGS__); \
2736 __ext4_warning_inode(inode, "", 0, " "); \
2738 #define ext4_msg(sb, level, fmt, ...) \
2740 no_printk(fmt, ##__VA_ARGS__); \
2741 __ext4_msg(sb, "", " "); \
2743 #define dump_mmp_msg(sb, mmp, msg) \
2744 __dump_mmp_msg(sb, mmp, "", 0, "")
2745 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2747 no_printk(fmt, ##__VA_ARGS__); \
2748 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
2753 extern void ext4_update_dynamic_rev(struct super_block *sb);
2754 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2756 extern int ext4_update_rocompat_feature(handle_t *handle,
2757 struct super_block *sb, __u32 rocompat);
2758 extern int ext4_update_incompat_feature(handle_t *handle,
2759 struct super_block *sb, __u32 incompat);
2760 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2761 struct ext4_group_desc *bg);
2762 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2763 struct ext4_group_desc *bg);
2764 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2765 struct ext4_group_desc *bg);
2766 extern __u32 ext4_free_group_clusters(struct super_block *sb,
2767 struct ext4_group_desc *bg);
2768 extern __u32 ext4_free_inodes_count(struct super_block *sb,
2769 struct ext4_group_desc *bg);
2770 extern __u32 ext4_used_dirs_count(struct super_block *sb,
2771 struct ext4_group_desc *bg);
2772 extern __u32 ext4_itable_unused_count(struct super_block *sb,
2773 struct ext4_group_desc *bg);
2774 extern void ext4_block_bitmap_set(struct super_block *sb,
2775 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2776 extern void ext4_inode_bitmap_set(struct super_block *sb,
2777 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2778 extern void ext4_inode_table_set(struct super_block *sb,
2779 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2780 extern void ext4_free_group_clusters_set(struct super_block *sb,
2781 struct ext4_group_desc *bg,
2783 extern void ext4_free_inodes_set(struct super_block *sb,
2784 struct ext4_group_desc *bg, __u32 count);
2785 extern void ext4_used_dirs_set(struct super_block *sb,
2786 struct ext4_group_desc *bg, __u32 count);
2787 extern void ext4_itable_unused_set(struct super_block *sb,
2788 struct ext4_group_desc *bg, __u32 count);
2789 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2790 struct ext4_group_desc *gdp);
2791 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2792 struct ext4_group_desc *gdp);
2793 extern int ext4_register_li_request(struct super_block *sb,
2794 ext4_group_t first_not_zeroed);
2796 static inline int ext4_has_group_desc_csum(struct super_block *sb)
2798 return ext4_has_feature_gdt_csum(sb) ||
2799 EXT4_SB(sb)->s_chksum_driver != NULL;
2802 static inline int ext4_has_metadata_csum(struct super_block *sb)
2804 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
2805 !EXT4_SB(sb)->s_chksum_driver);
2807 return (EXT4_SB(sb)->s_chksum_driver != NULL);
2809 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2811 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2812 le32_to_cpu(es->s_blocks_count_lo);
2815 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2817 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2818 le32_to_cpu(es->s_r_blocks_count_lo);
2821 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2823 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2824 le32_to_cpu(es->s_free_blocks_count_lo);
2827 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2830 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2831 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2834 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2837 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2838 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2841 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2844 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2845 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2848 static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
2850 if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2851 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2852 le32_to_cpu(raw_inode->i_size_lo);
2854 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2857 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2859 raw_inode->i_size_lo = cpu_to_le32(i_size);
2860 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2864 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2867 struct ext4_group_info ***grp_info;
2868 long indexv, indexh;
2869 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
2870 grp_info = EXT4_SB(sb)->s_group_info;
2871 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2872 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2873 return grp_info[indexv][indexh];
2877 * Reading s_groups_count requires using smp_rmb() afterwards. See
2878 * the locking protocol documented in the comments of ext4_group_add()
2881 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2883 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
2889 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2890 ext4_group_t block_group)
2892 return block_group >> sbi->s_log_groups_per_flex;
2895 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2897 return 1 << sbi->s_log_groups_per_flex;
2900 #define ext4_std_error(sb, errno) \
2903 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
2907 /* Each CPU can accumulate percpu_counter_batch clusters in their local
2908 * counters. So we need to make sure we have free clusters more
2909 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
2911 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2913 #define EXT4_FREECLUSTERS_WATERMARK 0
2916 /* Update i_disksize. Requires i_mutex to avoid races with truncate */
2917 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2919 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
2920 !inode_is_locked(inode));
2921 down_write(&EXT4_I(inode)->i_data_sem);
2922 if (newsize > EXT4_I(inode)->i_disksize)
2923 EXT4_I(inode)->i_disksize = newsize;
2924 up_write(&EXT4_I(inode)->i_data_sem);
2927 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
2928 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
2932 if (newsize > inode->i_size) {
2933 i_size_write(inode, newsize);
2936 if (newsize > EXT4_I(inode)->i_disksize) {
2937 ext4_update_i_disksize(inode, newsize);
2943 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
2946 struct ext4_group_info {
2947 unsigned long bb_state;
2948 struct rb_root bb_free_root;
2949 ext4_grpblk_t bb_first_free; /* first free block */
2950 ext4_grpblk_t bb_free; /* total free blocks */
2951 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
2952 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
2953 struct list_head bb_prealloc_list;
2957 struct rw_semaphore alloc_sem;
2958 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
2959 * regions, index is order.
2960 * bb_counters[3] = 5 means
2961 * 5 free 8-block regions. */
2964 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
2965 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
2966 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
2967 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
2969 #define EXT4_MB_GRP_NEED_INIT(grp) \
2970 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2971 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
2972 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2973 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
2974 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2976 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
2977 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2978 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
2979 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2980 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
2981 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2983 #define EXT4_MAX_CONTENTION 8
2984 #define EXT4_CONTENTION_THRESHOLD 2
2986 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2989 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2993 * Returns true if the filesystem is busy enough that attempts to
2994 * access the block group locks has run into contention.
2996 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2998 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
3001 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
3003 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
3004 if (spin_trylock(lock))
3006 * We're able to grab the lock right away, so drop the
3007 * lock contention counter.
3009 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
3012 * The lock is busy, so bump the contention counter,
3013 * and then wait on the spin lock.
3015 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
3016 EXT4_MAX_CONTENTION);
3021 static inline void ext4_unlock_group(struct super_block *sb,
3024 spin_unlock(ext4_group_lock_ptr(sb, group));
3028 * Block validity checking
3030 #define ext4_check_indirect_blockref(inode, bh) \
3031 ext4_check_blockref(__func__, __LINE__, inode, \
3032 (__le32 *)(bh)->b_data, \
3033 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
3035 #define ext4_ind_check_inode(inode) \
3036 ext4_check_blockref(__func__, __LINE__, inode, \
3037 EXT4_I(inode)->i_data, \
3041 * Inodes and files operations
3045 extern const struct file_operations ext4_dir_operations;
3048 extern const struct inode_operations ext4_file_inode_operations;
3049 extern const struct file_operations ext4_file_operations;
3050 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
3053 extern int ext4_get_max_inline_size(struct inode *inode);
3054 extern int ext4_find_inline_data_nolock(struct inode *inode);
3055 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
3057 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
3059 extern int ext4_readpage_inline(struct inode *inode, struct page *page);
3060 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3061 struct inode *inode,
3062 loff_t pos, unsigned len,
3064 struct page **pagep);
3065 extern int ext4_write_inline_data_end(struct inode *inode,
3066 loff_t pos, unsigned len,
3069 extern struct buffer_head *
3070 ext4_journalled_write_inline_data(struct inode *inode,
3073 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
3074 struct inode *inode,
3075 loff_t pos, unsigned len,
3077 struct page **pagep,
3079 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
3080 unsigned len, unsigned copied,
3082 extern int ext4_try_add_inline_entry(handle_t *handle,
3083 struct ext4_filename *fname,
3084 struct inode *dir, struct inode *inode);
3085 extern int ext4_try_create_inline_dir(handle_t *handle,
3086 struct inode *parent,
3087 struct inode *inode);
3088 extern int ext4_read_inline_dir(struct file *filp,
3089 struct dir_context *ctx,
3090 int *has_inline_data);
3091 extern int htree_inlinedir_to_tree(struct file *dir_file,
3092 struct inode *dir, ext4_lblk_t block,
3093 struct dx_hash_info *hinfo,
3094 __u32 start_hash, __u32 start_minor_hash,
3095 int *has_inline_data);
3096 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3097 struct ext4_filename *fname,
3098 const struct qstr *d_name,
3099 struct ext4_dir_entry_2 **res_dir,
3100 int *has_inline_data);
3101 extern int ext4_delete_inline_entry(handle_t *handle,
3103 struct ext4_dir_entry_2 *de_del,
3104 struct buffer_head *bh,
3105 int *has_inline_data);
3106 extern int empty_inline_dir(struct inode *dir, int *has_inline_data);
3107 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3108 struct ext4_dir_entry_2 **parent_de,
3110 extern int ext4_inline_data_fiemap(struct inode *inode,
3111 struct fiemap_extent_info *fieinfo,
3112 int *has_inline, __u64 start, __u64 len);
3113 extern int ext4_try_to_evict_inline_data(handle_t *handle,
3114 struct inode *inode,
3116 extern void ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3118 extern int ext4_convert_inline_data(struct inode *inode);
3120 static inline int ext4_has_inline_data(struct inode *inode)
3122 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3123 EXT4_I(inode)->i_inline_off;
3127 extern const struct inode_operations ext4_dir_inode_operations;
3128 extern const struct inode_operations ext4_special_inode_operations;
3129 extern struct dentry *ext4_get_parent(struct dentry *child);
3130 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3131 struct ext4_dir_entry_2 *de,
3132 int blocksize, int csum_size,
3133 unsigned int parent_ino, int dotdot_real_len);
3134 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t,
3135 unsigned int blocksize);
3136 extern int ext4_handle_dirty_dirent_node(handle_t *handle,
3137 struct inode *inode,
3138 struct buffer_head *bh);
3140 static unsigned char ext4_type_by_mode[S_IFMT >> S_SHIFT] = {
3141 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
3142 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
3143 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
3144 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
3145 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
3146 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
3147 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
3150 static inline void ext4_set_de_type(struct super_block *sb,
3151 struct ext4_dir_entry_2 *de,
3153 if (ext4_has_feature_filetype(sb))
3154 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3158 extern int ext4_mpage_readpages(struct address_space *mapping,
3159 struct list_head *pages, struct page *page,
3163 extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3164 extern const struct inode_operations ext4_symlink_inode_operations;
3165 extern const struct inode_operations ext4_fast_symlink_inode_operations;
3168 extern int ext4_register_sysfs(struct super_block *sb);
3169 extern void ext4_unregister_sysfs(struct super_block *sb);
3170 extern int __init ext4_init_sysfs(void);
3171 extern void ext4_exit_sysfs(void);
3173 /* block_validity */
3174 extern void ext4_release_system_zone(struct super_block *sb);
3175 extern int ext4_setup_system_zone(struct super_block *sb);
3176 extern int __init ext4_init_system_zone(void);
3177 extern void ext4_exit_system_zone(void);
3178 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
3179 ext4_fsblk_t start_blk,
3180 unsigned int count);
3181 extern int ext4_check_blockref(const char *, unsigned int,
3182 struct inode *, __le32 *, unsigned int);
3185 struct ext4_ext_path;
3189 * Maximum number of logical blocks in a file; ext4_extent's ee_block is