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/wait.h>
30 #include <linux/blockgroup_lock.h>
31 #include <linux/percpu_counter.h>
32 #include <crypto/hash.h>
34 #include <linux/compat.h>
38 * The fourth extended filesystem constants/structures
42 * Define EXT4FS_DEBUG to produce debug messages
50 #define ext4_debug(f, a...) \
52 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
53 __FILE__, __LINE__, __func__); \
54 printk(KERN_DEBUG f, ## a); \
57 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
60 #define EXT4_ERROR_INODE(inode, fmt, a...) \
61 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
63 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
64 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
66 #define EXT4_ERROR_FILE(file, block, fmt, a...) \
67 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
69 /* data type for block offset of block group */
70 typedef int ext4_grpblk_t
;
72 /* data type for filesystem-wide blocks number */
73 typedef unsigned long long ext4_fsblk_t
;
75 /* data type for file logical block number */
76 typedef __u32 ext4_lblk_t
;
78 /* data type for block group number */
79 typedef unsigned int ext4_group_t
;
82 * Flags used in mballoc's allocation_context flags field.
84 * Also used to show what's going on for debugging purposes when the
85 * flag field is exported via the traceport interface
88 /* prefer goal again. length */
89 #define EXT4_MB_HINT_MERGE 0x0001
90 /* blocks already reserved */
91 #define EXT4_MB_HINT_RESERVED 0x0002
92 /* metadata is being allocated */
93 #define EXT4_MB_HINT_METADATA 0x0004
94 /* first blocks in the file */
95 #define EXT4_MB_HINT_FIRST 0x0008
96 /* search for the best chunk */
97 #define EXT4_MB_HINT_BEST 0x0010
98 /* data is being allocated */
99 #define EXT4_MB_HINT_DATA 0x0020
100 /* don't preallocate (for tails) */
101 #define EXT4_MB_HINT_NOPREALLOC 0x0040
102 /* allocate for locality group */
103 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080
104 /* allocate goal blocks or none */
105 #define EXT4_MB_HINT_GOAL_ONLY 0x0100
106 /* goal is meaningful */
107 #define EXT4_MB_HINT_TRY_GOAL 0x0200
108 /* blocks already pre-reserved by delayed allocation */
109 #define EXT4_MB_DELALLOC_RESERVED 0x0400
110 /* We are doing stream allocation */
111 #define EXT4_MB_STREAM_ALLOC 0x0800
112 /* Use reserved root blocks if needed */
113 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000
115 struct ext4_allocation_request
{
116 /* target inode for block we're allocating */
118 /* how many blocks we want to allocate */
120 /* logical block in target inode */
122 /* the closest logical allocated block to the left */
124 /* the closest logical allocated block to the right */
126 /* phys. target (a hint) */
128 /* phys. block for the closest logical allocated block to the left */
130 /* phys. block for the closest logical allocated block to the right */
132 /* flags. see above EXT4_MB_HINT_* */
137 * Logical to physical block mapping, used by ext4_map_blocks()
139 * This structure is used to pass requests into ext4_map_blocks() as
140 * well as to store the information returned by ext4_map_blocks(). It
141 * takes less room on the stack than a struct buffer_head.
143 #define EXT4_MAP_NEW (1 << BH_New)
144 #define EXT4_MAP_MAPPED (1 << BH_Mapped)
145 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten)
146 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary)
147 #define EXT4_MAP_UNINIT (1 << BH_Uninit)
148 /* Sometimes (in the bigalloc case, from ext4_da_get_block_prep) the caller of
149 * ext4_map_blocks wants to know whether or not the underlying cluster has
150 * already been accounted for. EXT4_MAP_FROM_CLUSTER conveys to the caller that
151 * the requested mapping was from previously mapped (or delayed allocated)
152 * cluster. We use BH_AllocFromCluster only for this flag. BH_AllocFromCluster
153 * should never appear on buffer_head's state flags.
155 #define EXT4_MAP_FROM_CLUSTER (1 << BH_AllocFromCluster)
156 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
157 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\
158 EXT4_MAP_UNINIT | EXT4_MAP_FROM_CLUSTER)
160 struct ext4_map_blocks
{
164 unsigned int m_flags
;
168 * For delayed allocation tracking
170 struct mpage_da_data
{
172 sector_t b_blocknr
; /* start block number of extent */
173 size_t b_size
; /* size of extent */
174 unsigned long b_state
; /* state of the extent */
175 unsigned long first_page
, next_page
; /* extent of pages */
176 struct writeback_control
*wbc
;
183 * Flags for ext4_io_end->flags
185 #define EXT4_IO_END_UNWRITTEN 0x0001
186 #define EXT4_IO_END_ERROR 0x0002
187 #define EXT4_IO_END_QUEUED 0x0004
188 #define EXT4_IO_END_DIRECT 0x0008
189 #define EXT4_IO_END_IN_FSYNC 0x0010
191 struct ext4_io_page
{
196 #define MAX_IO_PAGES 128
199 * For converting uninitialized extents on a work queue.
201 * 'page' is only used from the writepage() path; 'pages' is only used for
202 * buffered writes; they are used to keep page references until conversion
203 * takes place. For AIO/DIO, neither field is filled in.
205 typedef struct ext4_io_end
{
206 struct list_head list
; /* per-file finished IO list */
207 struct inode
*inode
; /* file being written to */
208 unsigned int flag
; /* unwritten or not */
209 struct page
*page
; /* for writepage() path */
210 loff_t offset
; /* offset in the file */
211 ssize_t size
; /* size of the extent */
212 struct work_struct work
; /* data work queue */
213 struct kiocb
*iocb
; /* iocb struct for AIO */
214 int result
; /* error value for AIO */
215 int num_io_pages
; /* for writepages() */
216 struct ext4_io_page
*pages
[MAX_IO_PAGES
]; /* for writepages() */
219 struct ext4_io_submit
{
222 ext4_io_end_t
*io_end
;
223 struct ext4_io_page
*io_page
;
224 sector_t io_next_block
;
228 * Special inodes numbers
230 #define EXT4_BAD_INO 1 /* Bad blocks inode */
231 #define EXT4_ROOT_INO 2 /* Root inode */
232 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */
233 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
234 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
235 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
236 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
237 #define EXT4_JOURNAL_INO 8 /* Journal inode */
239 /* First non-reserved inode for old ext4 filesystems */
240 #define EXT4_GOOD_OLD_FIRST_INO 11
243 * Maximal count of links to a file
245 #define EXT4_LINK_MAX 65000
248 * Macro-instructions used to manage several block sizes
250 #define EXT4_MIN_BLOCK_SIZE 1024
251 #define EXT4_MAX_BLOCK_SIZE 65536
252 #define EXT4_MIN_BLOCK_LOG_SIZE 10
253 #define EXT4_MAX_BLOCK_LOG_SIZE 16
255 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
257 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
259 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
260 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
261 EXT4_SB(s)->s_cluster_bits)
263 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
264 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
266 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
269 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
270 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
271 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
273 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
274 EXT4_GOOD_OLD_INODE_SIZE : \
276 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
277 EXT4_GOOD_OLD_FIRST_INO : \
280 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
282 /* Translate a block number to a cluster number */
283 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
284 /* Translate a cluster number to a block number */
285 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
286 /* Translate # of blks to # of clusters */
287 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
288 (sbi)->s_cluster_bits)
291 * Structure of a blocks group descriptor
293 struct ext4_group_desc
295 __le32 bg_block_bitmap_lo
; /* Blocks bitmap block */
296 __le32 bg_inode_bitmap_lo
; /* Inodes bitmap block */
297 __le32 bg_inode_table_lo
; /* Inodes table block */
298 __le16 bg_free_blocks_count_lo
;/* Free blocks count */
299 __le16 bg_free_inodes_count_lo
;/* Free inodes count */
300 __le16 bg_used_dirs_count_lo
; /* Directories count */
301 __le16 bg_flags
; /* EXT4_BG_flags (INODE_UNINIT, etc) */
302 __le32 bg_exclude_bitmap_lo
; /* Exclude bitmap for snapshots */
303 __le16 bg_block_bitmap_csum_lo
;/* crc32c(s_uuid+grp_num+bbitmap) LE */
304 __le16 bg_inode_bitmap_csum_lo
;/* crc32c(s_uuid+grp_num+ibitmap) LE */
305 __le16 bg_itable_unused_lo
; /* Unused inodes count */
306 __le16 bg_checksum
; /* crc16(sb_uuid+group+desc) */
307 __le32 bg_block_bitmap_hi
; /* Blocks bitmap block MSB */
308 __le32 bg_inode_bitmap_hi
; /* Inodes bitmap block MSB */
309 __le32 bg_inode_table_hi
; /* Inodes table block MSB */
310 __le16 bg_free_blocks_count_hi
;/* Free blocks count MSB */
311 __le16 bg_free_inodes_count_hi
;/* Free inodes count MSB */
312 __le16 bg_used_dirs_count_hi
; /* Directories count MSB */
313 __le16 bg_itable_unused_hi
; /* Unused inodes count MSB */
314 __le32 bg_exclude_bitmap_hi
; /* Exclude bitmap block MSB */
315 __le16 bg_block_bitmap_csum_hi
;/* crc32c(s_uuid+grp_num+bbitmap) BE */
316 __le16 bg_inode_bitmap_csum_hi
;/* crc32c(s_uuid+grp_num+ibitmap) BE */
320 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
321 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
323 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
324 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
328 * Structure of a flex block group info
332 atomic_t free_inodes
;
333 atomic_t free_clusters
;
337 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
338 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
339 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
342 * Macro-instructions used to manage group descriptors
344 #define EXT4_MIN_DESC_SIZE 32
345 #define EXT4_MIN_DESC_SIZE_64BIT 64
346 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
347 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
349 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
350 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
351 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
352 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
353 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
355 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
356 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
357 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
361 * Constants relative to the data blocks
363 #define EXT4_NDIR_BLOCKS 12
364 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
365 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
366 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
367 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
372 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
373 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
374 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
375 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
376 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
377 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
378 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
379 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
380 /* Reserved for compression usage... */
381 #define EXT4_DIRTY_FL 0x00000100
382 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
383 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
384 #define EXT4_ECOMPR_FL 0x00000800 /* Compression error */
385 /* End compression flags --- maybe not all used */
386 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
387 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
388 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
389 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
390 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
391 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
392 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
393 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
394 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
395 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
396 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
398 #define EXT4_FL_USER_VISIBLE 0x004BDFFF /* User visible flags */
399 #define EXT4_FL_USER_MODIFIABLE 0x004B80FF /* User modifiable flags */
401 /* Flags that should be inherited by new inodes from their parent. */
402 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
403 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
404 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
405 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
407 /* Flags that are appropriate for regular files (all but dir-specific ones). */
408 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
410 /* Flags that are appropriate for non-directories/regular files. */
411 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
413 /* Mask out flags that are inappropriate for the given type of inode. */
414 static inline __u32
ext4_mask_flags(umode_t mode
, __u32 flags
)
418 else if (S_ISREG(mode
))
419 return flags
& EXT4_REG_FLMASK
;
421 return flags
& EXT4_OTHER_FLMASK
;
425 * Inode flags used for atomic set/get
428 EXT4_INODE_SECRM
= 0, /* Secure deletion */
429 EXT4_INODE_UNRM
= 1, /* Undelete */
430 EXT4_INODE_COMPR
= 2, /* Compress file */
431 EXT4_INODE_SYNC
= 3, /* Synchronous updates */
432 EXT4_INODE_IMMUTABLE
= 4, /* Immutable file */
433 EXT4_INODE_APPEND
= 5, /* writes to file may only append */
434 EXT4_INODE_NODUMP
= 6, /* do not dump file */
435 EXT4_INODE_NOATIME
= 7, /* do not update atime */
436 /* Reserved for compression usage... */
437 EXT4_INODE_DIRTY
= 8,
438 EXT4_INODE_COMPRBLK
= 9, /* One or more compressed clusters */
439 EXT4_INODE_NOCOMPR
= 10, /* Don't compress */
440 EXT4_INODE_ECOMPR
= 11, /* Compression error */
441 /* End compression flags --- maybe not all used */
442 EXT4_INODE_INDEX
= 12, /* hash-indexed directory */
443 EXT4_INODE_IMAGIC
= 13, /* AFS directory */
444 EXT4_INODE_JOURNAL_DATA
= 14, /* file data should be journaled */
445 EXT4_INODE_NOTAIL
= 15, /* file tail should not be merged */
446 EXT4_INODE_DIRSYNC
= 16, /* dirsync behaviour (directories only) */
447 EXT4_INODE_TOPDIR
= 17, /* Top of directory hierarchies*/
448 EXT4_INODE_HUGE_FILE
= 18, /* Set to each huge file */
449 EXT4_INODE_EXTENTS
= 19, /* Inode uses extents */
450 EXT4_INODE_EA_INODE
= 21, /* Inode used for large EA */
451 EXT4_INODE_EOFBLOCKS
= 22, /* Blocks allocated beyond EOF */
452 EXT4_INODE_RESERVED
= 31, /* reserved for ext4 lib */
455 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
456 #define CHECK_FLAG_VALUE(FLAG) if (!TEST_FLAG_VALUE(FLAG)) { \
457 printk(KERN_EMERG "EXT4 flag fail: " #FLAG ": %d %d\n", \
458 EXT4_##FLAG##_FL, EXT4_INODE_##FLAG); BUG_ON(1); }
461 * Since it's pretty easy to mix up bit numbers and hex values, and we
462 * can't do a compile-time test for ENUM values, we use a run-time
463 * test to make sure that EXT4_XXX_FL is consistent with respect to
464 * EXT4_INODE_XXX. If all is well the printk and BUG_ON will all drop
465 * out so it won't cost any extra space in the compiled kernel image.
466 * But it's important that these values are the same, since we are
467 * using EXT4_INODE_XXX to test for the flag values, but EXT4_XX_FL
468 * must be consistent with the values of FS_XXX_FL defined in
469 * include/linux/fs.h and the on-disk values found in ext2, ext3, and
470 * ext4 filesystems, and of course the values defined in e2fsprogs.
472 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
474 static inline void ext4_check_flag_values(void)
476 CHECK_FLAG_VALUE(SECRM
);
477 CHECK_FLAG_VALUE(UNRM
);
478 CHECK_FLAG_VALUE(COMPR
);
479 CHECK_FLAG_VALUE(SYNC
);
480 CHECK_FLAG_VALUE(IMMUTABLE
);
481 CHECK_FLAG_VALUE(APPEND
);
482 CHECK_FLAG_VALUE(NODUMP
);
483 CHECK_FLAG_VALUE(NOATIME
);
484 CHECK_FLAG_VALUE(DIRTY
);
485 CHECK_FLAG_VALUE(COMPRBLK
);
486 CHECK_FLAG_VALUE(NOCOMPR
);
487 CHECK_FLAG_VALUE(ECOMPR
);
488 CHECK_FLAG_VALUE(INDEX
);
489 CHECK_FLAG_VALUE(IMAGIC
);
490 CHECK_FLAG_VALUE(JOURNAL_DATA
);
491 CHECK_FLAG_VALUE(NOTAIL
);
492 CHECK_FLAG_VALUE(DIRSYNC
);
493 CHECK_FLAG_VALUE(TOPDIR
);
494 CHECK_FLAG_VALUE(HUGE_FILE
);
495 CHECK_FLAG_VALUE(EXTENTS
);
496 CHECK_FLAG_VALUE(EA_INODE
);
497 CHECK_FLAG_VALUE(EOFBLOCKS
);
498 CHECK_FLAG_VALUE(RESERVED
);
501 /* Used to pass group descriptor data when online resize is done */
502 struct ext4_new_group_input
{
503 __u32 group
; /* Group number for this data */
504 __u64 block_bitmap
; /* Absolute block number of block bitmap */
505 __u64 inode_bitmap
; /* Absolute block number of inode bitmap */
506 __u64 inode_table
; /* Absolute block number of inode table start */
507 __u32 blocks_count
; /* Total number of blocks in this group */
508 __u16 reserved_blocks
; /* Number of reserved blocks in this group */
512 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
513 struct compat_ext4_new_group_input
{
515 compat_u64 block_bitmap
;
516 compat_u64 inode_bitmap
;
517 compat_u64 inode_table
;
524 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
525 struct ext4_new_group_data
{
531 __u16 reserved_blocks
;
533 __u32 free_blocks_count
;
536 /* Indexes used to index group tables in ext4_new_group_data */
538 BLOCK_BITMAP
= 0, /* block bitmap */
539 INODE_BITMAP
, /* inode bitmap */
540 INODE_TABLE
, /* inode tables */
545 * Flags used by ext4_map_blocks()
547 /* Allocate any needed blocks and/or convert an unitialized
548 extent to be an initialized ext4 */
549 #define EXT4_GET_BLOCKS_CREATE 0x0001
550 /* Request the creation of an unitialized extent */
551 #define EXT4_GET_BLOCKS_UNINIT_EXT 0x0002
552 #define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT (EXT4_GET_BLOCKS_UNINIT_EXT|\
553 EXT4_GET_BLOCKS_CREATE)
554 /* Caller is from the delayed allocation writeout path,
555 so set the magic i_delalloc_reserve_flag after taking the
556 inode allocation semaphore for */
557 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
558 /* caller is from the direct IO path, request to creation of an
559 unitialized extents if not allocated, split the uninitialized
560 extent if blocks has been preallocated already*/
561 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
562 #define EXT4_GET_BLOCKS_CONVERT 0x0010
563 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
564 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
565 /* Convert extent to initialized after IO complete */
566 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
567 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
568 /* Punch out blocks of an extent */
569 #define EXT4_GET_BLOCKS_PUNCH_OUT_EXT 0x0020
570 /* Don't normalize allocation size (used for fallocate) */
571 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
572 /* Request will not result in inode size update (user for fallocate) */
573 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080
574 /* Do not take i_data_sem locking in ext4_map_blocks */
575 #define EXT4_GET_BLOCKS_NO_LOCK 0x0100
578 * Flags used by ext4_free_blocks
580 #define EXT4_FREE_BLOCKS_METADATA 0x0001
581 #define EXT4_FREE_BLOCKS_FORGET 0x0002
582 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
583 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
584 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
585 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
588 * Flags used by ext4_discard_partial_page_buffers
590 #define EXT4_DISCARD_PARTIAL_PG_ZERO_UNMAPPED 0x0001
595 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
596 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
597 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
598 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
599 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
600 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
601 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
602 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
603 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
604 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
605 #define EXT4_IOC_MIGRATE _IO('f', 9)
606 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
607 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
608 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
609 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
610 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
612 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
614 * ioctl commands in 32 bit emulation
616 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
617 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
618 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
619 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
620 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
621 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
622 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
623 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
624 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
625 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
628 /* Max physical block we can address w/o extents */
629 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
632 * Structure of an inode on the disk
635 __le16 i_mode
; /* File mode */
636 __le16 i_uid
; /* Low 16 bits of Owner Uid */
637 __le32 i_size_lo
; /* Size in bytes */
638 __le32 i_atime
; /* Access time */
639 __le32 i_ctime
; /* Inode Change time */
640 __le32 i_mtime
; /* Modification time */
641 __le32 i_dtime
; /* Deletion Time */
642 __le16 i_gid
; /* Low 16 bits of Group Id */
643 __le16 i_links_count
; /* Links count */
644 __le32 i_blocks_lo
; /* Blocks count */
645 __le32 i_flags
; /* File flags */
651 __u32 h_i_translator
;
656 } osd1
; /* OS dependent 1 */
657 __le32 i_block
[EXT4_N_BLOCKS
];/* Pointers to blocks */
658 __le32 i_generation
; /* File version (for NFS) */
659 __le32 i_file_acl_lo
; /* File ACL */
661 __le32 i_obso_faddr
; /* Obsoleted fragment address */
664 __le16 l_i_blocks_high
; /* were l_i_reserved1 */
665 __le16 l_i_file_acl_high
;
666 __le16 l_i_uid_high
; /* these 2 fields */
667 __le16 l_i_gid_high
; /* were reserved2[0] */
668 __le16 l_i_checksum_lo
;/* crc32c(uuid+inum+inode) LE */
672 __le16 h_i_reserved1
; /* Obsoleted fragment number/size which are removed in ext4 */
679 __le16 h_i_reserved1
; /* Obsoleted fragment number/size which are removed in ext4 */
680 __le16 m_i_file_acl_high
;
681 __u32 m_i_reserved2
[2];
683 } osd2
; /* OS dependent 2 */
684 __le16 i_extra_isize
;
685 __le16 i_checksum_hi
; /* crc32c(uuid+inum+inode) BE */
686 __le32 i_ctime_extra
; /* extra Change time (nsec << 2 | epoch) */
687 __le32 i_mtime_extra
; /* extra Modification time(nsec << 2 | epoch) */
688 __le32 i_atime_extra
; /* extra Access time (nsec << 2 | epoch) */
689 __le32 i_crtime
; /* File Creation time */
690 __le32 i_crtime_extra
; /* extra FileCreationtime (nsec << 2 | epoch) */
691 __le32 i_version_hi
; /* high 32 bits for 64-bit version */
695 __u32 reserved
; /* should be zero */
696 __u32 donor_fd
; /* donor file descriptor */
697 __u64 orig_start
; /* logical start offset in block for orig */
698 __u64 donor_start
; /* logical start offset in block for donor */
699 __u64 len
; /* block length to be moved */
700 __u64 moved_len
; /* moved block length */
703 #define EXT4_EPOCH_BITS 2
704 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
705 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
708 * Extended fields will fit into an inode if the filesystem was formatted
709 * with large inodes (-I 256 or larger) and there are not currently any EAs
710 * consuming all of the available space. For new inodes we always reserve
711 * enough space for the kernel's known extended fields, but for inodes
712 * created with an old kernel this might not have been the case. None of
713 * the extended inode fields is critical for correct filesystem operation.
714 * This macro checks if a certain field fits in the inode. Note that
715 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
717 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
718 ((offsetof(typeof(*ext4_inode), field) + \
719 sizeof((ext4_inode)->field)) \
720 <= (EXT4_GOOD_OLD_INODE_SIZE + \
721 (einode)->i_extra_isize)) \
723 static inline __le32 ext4_encode_extra_time(struct timespec *time)
725 return cpu_to_le32((sizeof(time
->tv_sec
) > 4 ?
726 (time
->tv_sec
>> 32) & EXT4_EPOCH_MASK
: 0) |
727 ((time
->tv_nsec
<< EXT4_EPOCH_BITS
) & EXT4_NSEC_MASK
));
730 static inline void ext4_decode_extra_time(struct timespec
*time
, __le32 extra
)
732 if (sizeof(time
->tv_sec
) > 4)
733 time
->tv_sec
|= (__u64
)(le32_to_cpu(extra
) & EXT4_EPOCH_MASK
)
735 time
->tv_nsec
= (le32_to_cpu(extra
) & EXT4_NSEC_MASK
) >> EXT4_EPOCH_BITS
;
738 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
740 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
741 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
742 (raw_inode)->xtime ## _extra = \
743 ext4_encode_extra_time(&(inode)->xtime); \
746 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
748 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
749 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
750 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
751 (raw_inode)->xtime ## _extra = \
752 ext4_encode_extra_time(&(einode)->xtime); \
755 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
757 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
758 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
759 ext4_decode_extra_time(&(inode)->xtime, \
760 raw_inode->xtime ## _extra); \
762 (inode)->xtime.tv_nsec = 0; \
765 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
767 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
768 (einode)->xtime.tv_sec = \
769 (signed)le32_to_cpu((raw_inode)->xtime); \
770 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
771 ext4_decode_extra_time(&(einode)->xtime, \
772 raw_inode->xtime ## _extra); \
774 (einode)->xtime.tv_nsec = 0; \
777 #define i_disk_version osd1.linux1.l_i_version
779 #if defined(__KERNEL__) || defined(__linux__)
780 #define i_reserved1 osd1.linux1.l_i_reserved1
781 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
782 #define i_blocks_high osd2.linux2.l_i_blocks_high
783 #define i_uid_low i_uid
784 #define i_gid_low i_gid
785 #define i_uid_high osd2.linux2.l_i_uid_high
786 #define i_gid_high osd2.linux2.l_i_gid_high
787 #define i_checksum_lo osd2.linux2.l_i_checksum_lo
789 #elif defined(__GNU__)
791 #define i_translator osd1.hurd1.h_i_translator
792 #define i_uid_high osd2.hurd2.h_i_uid_high
793 #define i_gid_high osd2.hurd2.h_i_gid_high
794 #define i_author osd2.hurd2.h_i_author
796 #elif defined(__masix__)
798 #define i_reserved1 osd1.masix1.m_i_reserved1
799 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
800 #define i_reserved2 osd2.masix2.m_i_reserved2
802 #endif /* defined(__KERNEL__) || defined(__linux__) */
805 * storage for cached extent
806 * If ec_len == 0, then the cache is invalid.
807 * If ec_start == 0, then the cache represents a gap (null mapping)
809 struct ext4_ext_cache
{
810 ext4_fsblk_t ec_start
;
811 ext4_lblk_t ec_block
;
812 __u32 ec_len
; /* must be 32bit to return holes */
816 * fourth extended file system inode data in memory
818 struct ext4_inode_info
{
819 __le32 i_data
[15]; /* unconverted */
821 ext4_fsblk_t i_file_acl
;
824 * i_block_group is the number of the block group which contains
825 * this file's inode. Constant across the lifetime of the inode,
826 * it is ued for making block allocation decisions - we try to
827 * place a file's data blocks near its inode block, and new inodes
828 * near to their parent directory's inode.
830 ext4_group_t i_block_group
;
831 ext4_lblk_t i_dir_start_lookup
;
832 #if (BITS_PER_LONG < 64)
833 unsigned long i_state_flags
; /* Dynamic state flags */
835 unsigned long i_flags
;
837 #ifdef CONFIG_EXT4_FS_XATTR
839 * Extended attributes can be read independently of the main file
840 * data. Taking i_mutex even when reading would cause contention
841 * between readers of EAs and writers of regular file data, so
842 * instead we synchronize on xattr_sem when reading or changing
845 struct rw_semaphore xattr_sem
;
848 struct list_head i_orphan
; /* unlinked but open inodes */
851 * i_disksize keeps track of what the inode size is ON DISK, not
852 * in memory. During truncate, i_size is set to the new size by
853 * the VFS prior to calling ext4_truncate(), but the filesystem won't
854 * set i_disksize to 0 until the truncate is actually under way.
856 * The intent is that i_disksize always represents the blocks which
857 * are used by this file. This allows recovery to restart truncate
858 * on orphans if we crash during truncate. We actually write i_disksize
859 * into the on-disk inode when writing inodes out, instead of i_size.
861 * The only time when i_disksize and i_size may be different is when
862 * a truncate is in progress. The only things which change i_disksize
863 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
868 * i_data_sem is for serialising ext4_truncate() against
869 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
870 * data tree are chopped off during truncate. We can't do that in
871 * ext4 because whenever we perform intermediate commits during
872 * truncate, the inode and all the metadata blocks *must* be in a
873 * consistent state which allows truncation of the orphans to restart
874 * during recovery. Hence we must fix the get_block-vs-truncate race
875 * by other means, so we have i_data_sem.
877 struct rw_semaphore i_data_sem
;
878 struct inode vfs_inode
;
879 struct jbd2_inode
*jinode
;
881 struct ext4_ext_cache i_cached_extent
;
883 * File creation time. Its function is same as that of
884 * struct timespec i_{a,c,m}time in the generic inode.
886 struct timespec i_crtime
;
889 struct list_head i_prealloc_list
;
890 spinlock_t i_prealloc_lock
;
893 ext4_group_t i_last_alloc_group
;
895 /* allocation reservation info for delalloc */
896 /* In case of bigalloc, these refer to clusters rather than blocks */
897 unsigned int i_reserved_data_blocks
;
898 unsigned int i_reserved_meta_blocks
;
899 unsigned int i_allocated_meta_blocks
;
900 ext4_lblk_t i_da_metadata_calc_last_lblock
;
901 int i_da_metadata_calc_len
;
903 /* on-disk additional length */
907 /* quota space reservation, managed internally by quota code */
908 qsize_t i_reserved_quota
;
911 /* completed IOs that might need unwritten extents handling */
912 struct list_head i_completed_io_list
;
913 spinlock_t i_completed_io_lock
;
914 atomic_t i_ioend_count
; /* Number of outstanding io_end structs */
915 /* current io_end structure for async DIO write*/
916 ext4_io_end_t
*cur_aio_dio
;
917 atomic_t i_aiodio_unwritten
; /* Nr. of inflight conversions pending */
919 spinlock_t i_block_reservation_lock
;
922 * Transactions that contain inode's metadata needed to complete
923 * fsync and fdatasync, respectively.
926 tid_t i_datasync_tid
;
928 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
935 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
936 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
937 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
940 * Misc. filesystem flags
942 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
943 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
944 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
949 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
950 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
951 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
952 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
953 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
954 #define EXT4_MOUNT_ERRORS_MASK 0x00070
955 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
956 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
957 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
958 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
959 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
960 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
961 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
962 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
963 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
964 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
965 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
966 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
967 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
968 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
969 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
970 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
971 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
972 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
973 #define EXT4_MOUNT_MBLK_IO_SUBMIT 0x4000000 /* multi-block io submits */
974 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
975 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
976 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
977 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
978 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
980 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
981 specified delalloc */
983 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
985 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
987 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
990 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
992 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
994 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
997 #define ext4_test_and_set_bit __test_and_set_bit_le
998 #define ext4_set_bit __set_bit_le
999 #define ext4_set_bit_atomic ext2_set_bit_atomic
1000 #define ext4_test_and_clear_bit __test_and_clear_bit_le
1001 #define ext4_clear_bit __clear_bit_le
1002 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
1003 #define ext4_test_bit test_bit_le
1004 #define ext4_find_next_zero_bit find_next_zero_bit_le
1005 #define ext4_find_next_bit find_next_bit_le
1007 extern void ext4_set_bits(void *bm
, int cur
, int len
);
1010 * Maximal mount counts between two filesystem checks
1012 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1013 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1016 * Behaviour when detecting errors
1018 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1019 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1020 #define EXT4_ERRORS_PANIC 3 /* Panic */
1021 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1023 /* Metadata checksum algorithm codes */
1024 #define EXT4_CRC32C_CHKSUM 1
1027 * Structure of the super block
1029 struct ext4_super_block
{
1030 /*00*/ __le32 s_inodes_count
; /* Inodes count */
1031 __le32 s_blocks_count_lo
; /* Blocks count */
1032 __le32 s_r_blocks_count_lo
; /* Reserved blocks count */
1033 __le32 s_free_blocks_count_lo
; /* Free blocks count */
1034 /*10*/ __le32 s_free_inodes_count
; /* Free inodes count */
1035 __le32 s_first_data_block
; /* First Data Block */
1036 __le32 s_log_block_size
; /* Block size */
1037 __le32 s_log_cluster_size
; /* Allocation cluster size */
1038 /*20*/ __le32 s_blocks_per_group
; /* # Blocks per group */
1039 __le32 s_clusters_per_group
; /* # Clusters per group */
1040 __le32 s_inodes_per_group
; /* # Inodes per group */
1041 __le32 s_mtime
; /* Mount time */
1042 /*30*/ __le32 s_wtime
; /* Write time */
1043 __le16 s_mnt_count
; /* Mount count */
1044 __le16 s_max_mnt_count
; /* Maximal mount count */
1045 __le16 s_magic
; /* Magic signature */
1046 __le16 s_state
; /* File system state */
1047 __le16 s_errors
; /* Behaviour when detecting errors */
1048 __le16 s_minor_rev_level
; /* minor revision level */
1049 /*40*/ __le32 s_lastcheck
; /* time of last check */
1050 __le32 s_checkinterval
; /* max. time between checks */
1051 __le32 s_creator_os
; /* OS */
1052 __le32 s_rev_level
; /* Revision level */
1053 /*50*/ __le16 s_def_resuid
; /* Default uid for reserved blocks */
1054 __le16 s_def_resgid
; /* Default gid for reserved blocks */
1056 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1058 * Note: the difference between the compatible feature set and
1059 * the incompatible feature set is that if there is a bit set
1060 * in the incompatible feature set that the kernel doesn't
1061 * know about, it should refuse to mount the filesystem.
1063 * e2fsck's requirements are more strict; if it doesn't know
1064 * about a feature in either the compatible or incompatible
1065 * feature set, it must abort and not try to meddle with
1066 * things it doesn't understand...
1068 __le32 s_first_ino
; /* First non-reserved inode */
1069 __le16 s_inode_size
; /* size of inode structure */
1070 __le16 s_block_group_nr
; /* block group # of this superblock */
1071 __le32 s_feature_compat
; /* compatible feature set */
1072 /*60*/ __le32 s_feature_incompat
; /* incompatible feature set */
1073 __le32 s_feature_ro_compat
; /* readonly-compatible feature set */
1074 /*68*/ __u8 s_uuid
[16]; /* 128-bit uuid for volume */
1075 /*78*/ char s_volume_name
[16]; /* volume name */
1076 /*88*/ char s_last_mounted
[64]; /* directory where last mounted */
1077 /*C8*/ __le32 s_algorithm_usage_bitmap
; /* For compression */
1079 * Performance hints. Directory preallocation should only
1080 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1082 __u8 s_prealloc_blocks
; /* Nr of blocks to try to preallocate*/
1083 __u8 s_prealloc_dir_blocks
; /* Nr to preallocate for dirs */
1084 __le16 s_reserved_gdt_blocks
; /* Per group desc for online growth */
1086 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1088 /*D0*/ __u8 s_journal_uuid
[16]; /* uuid of journal superblock */
1089 /*E0*/ __le32 s_journal_inum
; /* inode number of journal file */
1090 __le32 s_journal_dev
; /* device number of journal file */
1091 __le32 s_last_orphan
; /* start of list of inodes to delete */
1092 __le32 s_hash_seed
[4]; /* HTREE hash seed */
1093 __u8 s_def_hash_version
; /* Default hash version to use */
1094 __u8 s_jnl_backup_type
;
1095 __le16 s_desc_size
; /* size of group descriptor */
1096 /*100*/ __le32 s_default_mount_opts
;
1097 __le32 s_first_meta_bg
; /* First metablock block group */
1098 __le32 s_mkfs_time
; /* When the filesystem was created */
1099 __le32 s_jnl_blocks
[17]; /* Backup of the journal inode */
1100 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1101 /*150*/ __le32 s_blocks_count_hi
; /* Blocks count */
1102 __le32 s_r_blocks_count_hi
; /* Reserved blocks count */
1103 __le32 s_free_blocks_count_hi
; /* Free blocks count */
1104 __le16 s_min_extra_isize
; /* All inodes have at least # bytes */
1105 __le16 s_want_extra_isize
; /* New inodes should reserve # bytes */
1106 __le32 s_flags
; /* Miscellaneous flags */
1107 __le16 s_raid_stride
; /* RAID stride */
1108 __le16 s_mmp_update_interval
; /* # seconds to wait in MMP checking */
1109 __le64 s_mmp_block
; /* Block for multi-mount protection */
1110 __le32 s_raid_stripe_width
; /* blocks on all data disks (N*stride)*/
1111 __u8 s_log_groups_per_flex
; /* FLEX_BG group size */
1112 __u8 s_checksum_type
; /* metadata checksum algorithm used */
1113 __le16 s_reserved_pad
;
1114 __le64 s_kbytes_written
; /* nr of lifetime kilobytes written */
1115 __le32 s_snapshot_inum
; /* Inode number of active snapshot */
1116 __le32 s_snapshot_id
; /* sequential ID of active snapshot */
1117 __le64 s_snapshot_r_blocks_count
; /* reserved blocks for active
1118 snapshot's future use */
1119 __le32 s_snapshot_list
; /* inode number of the head of the
1120 on-disk snapshot list */
1121 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1122 __le32 s_error_count
; /* number of fs errors */
1123 __le32 s_first_error_time
; /* first time an error happened */
1124 __le32 s_first_error_ino
; /* inode involved in first error */
1125 __le64 s_first_error_block
; /* block involved of first error */
1126 __u8 s_first_error_func
[32]; /* function where the error happened */
1127 __le32 s_first_error_line
; /* line number where error happened */
1128 __le32 s_last_error_time
; /* most recent time of an error */
1129 __le32 s_last_error_ino
; /* inode involved in last error */
1130 __le32 s_last_error_line
; /* line number where error happened */
1131 __le64 s_last_error_block
; /* block involved of last error */
1132 __u8 s_last_error_func
[32]; /* function where the error happened */
1133 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1134 __u8 s_mount_opts
[64];
1135 __le32 s_usr_quota_inum
; /* inode for tracking user quota */
1136 __le32 s_grp_quota_inum
; /* inode for tracking group quota */
1137 __le32 s_overhead_clusters
; /* overhead blocks/clusters in fs */
1138 __le32 s_reserved
[108]; /* Padding to the end of the block */
1139 __le32 s_checksum
; /* crc32c(superblock) */
1142 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1147 * run-time mount flags
1149 #define EXT4_MF_MNTDIR_SAMPLED 0x0001
1150 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1153 * fourth extended-fs super-block data in memory
1155 struct ext4_sb_info
{
1156 unsigned long s_desc_size
; /* Size of a group descriptor in bytes */
1157 unsigned long s_inodes_per_block
;/* Number of inodes per block */
1158 unsigned long s_blocks_per_group
;/* Number of blocks in a group */
1159 unsigned long s_clusters_per_group
; /* Number of clusters in a group */
1160 unsigned long s_inodes_per_group
;/* Number of inodes in a group */
1161 unsigned long s_itb_per_group
; /* Number of inode table blocks per group */
1162 unsigned long s_gdb_count
; /* Number of group descriptor blocks */
1163 unsigned long s_desc_per_block
; /* Number of group descriptors per block */
1164 ext4_group_t s_groups_count
; /* Number of groups in the fs */
1165 ext4_group_t s_blockfile_groups
;/* Groups acceptable for non-extent files */
1166 unsigned long s_overhead
; /* # of fs overhead clusters */
1167 unsigned int s_cluster_ratio
; /* Number of blocks per cluster */
1168 unsigned int s_cluster_bits
; /* log2 of s_cluster_ratio */
1169 loff_t s_bitmap_maxbytes
; /* max bytes for bitmap files */
1170 struct buffer_head
* s_sbh
; /* Buffer containing the super block */
1171 struct ext4_super_block
*s_es
; /* Pointer to the super block in the buffer */
1172 struct buffer_head
**s_group_desc
;
1173 unsigned int s_mount_opt
;
1174 unsigned int s_mount_opt2
;
1175 unsigned int s_mount_flags
;
1176 unsigned int s_def_mount_opt
;
1177 ext4_fsblk_t s_sb_block
;
1180 unsigned short s_mount_state
;
1181 unsigned short s_pad
;
1182 int s_addr_per_block_bits
;
1183 int s_desc_per_block_bits
;
1186 unsigned int s_inode_readahead_blks
;
1187 unsigned int s_inode_goal
;
1188 spinlock_t s_next_gen_lock
;
1189 u32 s_next_generation
;
1191 int s_def_hash_version
;
1192 int s_hash_unsigned
; /* 3 if hash should be signed, 0 if not */
1193 struct percpu_counter s_freeclusters_counter
;
1194 struct percpu_counter s_freeinodes_counter
;
1195 struct percpu_counter s_dirs_counter
;
1196 struct percpu_counter s_dirtyclusters_counter
;
1197 struct blockgroup_lock
*s_blockgroup_lock
;
1198 struct proc_dir_entry
*s_proc
;
1199 struct kobject s_kobj
;
1200 struct completion s_kobj_unregister
;
1201 struct super_block
*s_sb
;
1204 struct journal_s
*s_journal
;
1205 struct list_head s_orphan
;
1206 struct mutex s_orphan_lock
;
1207 unsigned long s_resize_flags
; /* Flags indicating if there
1209 unsigned long s_commit_interval
;
1210 u32 s_max_batch_time
;
1211 u32 s_min_batch_time
;
1212 struct block_device
*journal_bdev
;
1214 char *s_qf_names
[MAXQUOTAS
]; /* Names of quota files with journalled quota */
1215 int s_jquota_fmt
; /* Format of quota to use */
1217 unsigned int s_want_extra_isize
; /* New inodes should reserve # bytes */
1218 struct rb_root system_blks
;
1220 #ifdef EXTENTS_STATS
1221 /* ext4 extents stats */
1222 unsigned long s_ext_min
;
1223 unsigned long s_ext_max
;
1224 unsigned long s_depth_max
;
1225 spinlock_t s_ext_stats_lock
;
1226 unsigned long s_ext_blocks
;
1227 unsigned long s_ext_extents
;
1230 /* for buddy allocator */
1231 struct ext4_group_info
***s_group_info
;
1232 struct inode
*s_buddy_cache
;
1233 spinlock_t s_md_lock
;
1234 unsigned short *s_mb_offsets
;
1235 unsigned int *s_mb_maxs
;
1238 unsigned long s_stripe
;
1239 unsigned int s_mb_stream_request
;
1240 unsigned int s_mb_max_to_scan
;
1241 unsigned int s_mb_min_to_scan
;
1242 unsigned int s_mb_stats
;
1243 unsigned int s_mb_order2_reqs
;
1244 unsigned int s_mb_group_prealloc
;
1245 unsigned int s_max_writeback_mb_bump
;
1246 /* where last allocation was done - for stream allocation */
1247 unsigned long s_mb_last_group
;
1248 unsigned long s_mb_last_start
;
1250 /* stats for buddy allocator */
1251 atomic_t s_bal_reqs
; /* number of reqs with len > 1 */
1252 atomic_t s_bal_success
; /* we found long enough chunks */
1253 atomic_t s_bal_allocated
; /* in blocks */
1254 atomic_t s_bal_ex_scanned
; /* total extents scanned */
1255 atomic_t s_bal_goals
; /* goal hits */
1256 atomic_t s_bal_breaks
; /* too long searches */
1257 atomic_t s_bal_2orders
; /* 2^order hits */
1258 spinlock_t s_bal_lock
;
1259 unsigned long s_mb_buddies_generated
;
1260 unsigned long long s_mb_generation_time
;
1261 atomic_t s_mb_lost_chunks
;
1262 atomic_t s_mb_preallocated
;
1263 atomic_t s_mb_discarded
;
1264 atomic_t s_lock_busy
;
1266 /* locality groups */
1267 struct ext4_locality_group __percpu
*s_locality_groups
;
1269 /* for write statistics */
1270 unsigned long s_sectors_written_start
;
1271 u64 s_kbytes_written
;
1273 unsigned int s_log_groups_per_flex
;
1274 struct flex_groups
*s_flex_groups
;
1276 /* workqueue for dio unwritten */
1277 struct workqueue_struct
*dio_unwritten_wq
;
1279 /* timer for periodic error stats printing */
1280 struct timer_list s_err_report
;
1282 /* Lazy inode table initialization info */
1283 struct ext4_li_request
*s_li_request
;
1284 /* Wait multiplier for lazy initialization thread */
1285 unsigned int s_li_wait_mult
;
1287 /* Kernel thread for multiple mount protection */
1288 struct task_struct
*s_mmp_tsk
;
1290 /* record the last minlen when FITRIM is called. */
1291 atomic_t s_last_trim_minblks
;
1293 /* Reference to checksum algorithm driver via cryptoapi */
1294 struct crypto_shash
*s_chksum_driver
;
1296 /* Precomputed FS UUID checksum for seeding other checksums */
1300 static inline struct ext4_sb_info
*EXT4_SB(struct super_block
*sb
)
1302 return sb
->s_fs_info
;
1304 static inline struct ext4_inode_info
*EXT4_I(struct inode
*inode
)
1306 return container_of(inode
, struct ext4_inode_info
, vfs_inode
);
1309 static inline struct timespec
ext4_current_time(struct inode
*inode
)
1311 return (inode
->i_sb
->s_time_gran
< NSEC_PER_SEC
) ?
1312 current_fs_time(inode
->i_sb
) : CURRENT_TIME_SEC
;
1315 static inline int ext4_valid_inum(struct super_block
*sb
, unsigned long ino
)
1317 return ino
== EXT4_ROOT_INO
||
1318 ino
== EXT4_USR_QUOTA_INO
||
1319 ino
== EXT4_GRP_QUOTA_INO
||
1320 ino
== EXT4_JOURNAL_INO
||
1321 ino
== EXT4_RESIZE_INO
||
1322 (ino
>= EXT4_FIRST_INO(sb
) &&
1323 ino
<= le32_to_cpu(EXT4_SB(sb
)->s_es
->s_inodes_count
));
1326 static inline void ext4_set_io_unwritten_flag(struct inode
*inode
,
1327 struct ext4_io_end
*io_end
)
1329 if (!(io_end
->flag
& EXT4_IO_END_UNWRITTEN
)) {
1330 io_end
->flag
|= EXT4_IO_END_UNWRITTEN
;
1331 atomic_inc(&EXT4_I(inode
)->i_aiodio_unwritten
);
1336 * Inode dynamic state flags
1339 EXT4_STATE_JDATA
, /* journaled data exists */
1340 EXT4_STATE_NEW
, /* inode is newly created */
1341 EXT4_STATE_XATTR
, /* has in-inode xattrs */
1342 EXT4_STATE_NO_EXPAND
, /* No space for expansion */
1343 EXT4_STATE_DA_ALLOC_CLOSE
, /* Alloc DA blks on close */
1344 EXT4_STATE_EXT_MIGRATE
, /* Inode is migrating */
1345 EXT4_STATE_DIO_UNWRITTEN
, /* need convert on dio done*/
1346 EXT4_STATE_NEWENTRY
, /* File just added to dir */
1347 EXT4_STATE_DELALLOC_RESERVED
, /* blks already reserved for delalloc */
1350 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1351 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1353 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1355 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1357 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1359 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1361 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1364 EXT4_INODE_BIT_FNS(flag
, flags
, 0)
1365 #if (BITS_PER_LONG < 64)
1366 EXT4_INODE_BIT_FNS(state
, state_flags
, 0)
1368 static inline void ext4_clear_state_flags(struct ext4_inode_info
*ei
)
1370 (ei
)->i_state_flags
= 0;
1373 EXT4_INODE_BIT_FNS(state
, flags
, 32)
1375 static inline void ext4_clear_state_flags(struct ext4_inode_info
*ei
)
1377 /* We depend on the fact that callers will set i_flags */
1381 /* Assume that user mode programs are passing in an ext4fs superblock, not
1382 * a kernel struct super_block. This will allow us to call the feature-test
1383 * macros from user land. */
1384 #define EXT4_SB(sb) (sb)
1387 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1390 * Codes for operating systems
1392 #define EXT4_OS_LINUX 0
1393 #define EXT4_OS_HURD 1
1394 #define EXT4_OS_MASIX 2
1395 #define EXT4_OS_FREEBSD 3
1396 #define EXT4_OS_LITES 4
1401 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1402 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1404 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1405 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1407 #define EXT4_GOOD_OLD_INODE_SIZE 128
1410 * Feature set definitions
1413 #define EXT4_HAS_COMPAT_FEATURE(sb,mask) \
1414 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1415 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \
1416 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1417 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \
1418 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1419 #define EXT4_SET_COMPAT_FEATURE(sb,mask) \
1420 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1421 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \
1422 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1423 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \
1424 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1425 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \
1426 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1427 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
1428 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1429 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \
1430 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1432 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1433 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1434 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1435 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1436 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1437 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1439 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1440 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1441 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1442 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1443 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1444 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1445 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1446 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1447 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1449 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1450 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1451 * all other data structures' checksums. However, the METADATA_CSUM and
1452 * GDT_CSUM bits are mutually exclusive.
1454 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
1456 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1457 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1458 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1459 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1460 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1461 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1462 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1463 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1464 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1465 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1466 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1467 #define EXT4_FEATURE_INCOMPAT_BG_USE_META_CSUM 0x2000 /* use crc32c for bg */
1468 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
1469 #define EXT4_FEATURE_INCOMPAT_INLINEDATA 0x8000 /* data in inode */
1471 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1472 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1473 EXT4_FEATURE_INCOMPAT_META_BG)
1474 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1475 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1476 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1478 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1479 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1480 EXT4_FEATURE_INCOMPAT_RECOVER| \
1481 EXT4_FEATURE_INCOMPAT_META_BG)
1482 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1483 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1484 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1486 #define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR
1487 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1488 EXT4_FEATURE_INCOMPAT_RECOVER| \
1489 EXT4_FEATURE_INCOMPAT_META_BG| \
1490 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1491 EXT4_FEATURE_INCOMPAT_64BIT| \
1492 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1493 EXT4_FEATURE_INCOMPAT_MMP)
1494 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1495 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1496 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1497 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1498 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1499 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1500 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1501 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1502 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1503 EXT4_FEATURE_RO_COMPAT_QUOTA)
1506 * Default values for user and/or group using reserved blocks
1508 #define EXT4_DEF_RESUID 0
1509 #define EXT4_DEF_RESGID 0
1511 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
1514 * Default mount options
1516 #define EXT4_DEFM_DEBUG 0x0001
1517 #define EXT4_DEFM_BSDGROUPS 0x0002
1518 #define EXT4_DEFM_XATTR_USER 0x0004
1519 #define EXT4_DEFM_ACL 0x0008
1520 #define EXT4_DEFM_UID16 0x0010
1521 #define EXT4_DEFM_JMODE 0x0060
1522 #define EXT4_DEFM_JMODE_DATA 0x0020
1523 #define EXT4_DEFM_JMODE_ORDERED 0x0040
1524 #define EXT4_DEFM_JMODE_WBACK 0x0060
1525 #define EXT4_DEFM_NOBARRIER 0x0100
1526 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1527 #define EXT4_DEFM_DISCARD 0x0400
1528 #define EXT4_DEFM_NODELALLOC 0x0800
1531 * Default journal batch times
1533 #define EXT4_DEF_MIN_BATCH_TIME 0
1534 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1537 * Minimum number of groups in a flexgroup before we separate out
1538 * directories into the first block group of a flexgroup
1540 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1543 * Structure of a directory entry
1545 #define EXT4_NAME_LEN 255
1547 struct ext4_dir_entry
{
1548 __le32 inode
; /* Inode number */
1549 __le16 rec_len
; /* Directory entry length */
1550 __le16 name_len
; /* Name length */
1551 char name
[EXT4_NAME_LEN
]; /* File name */
1555 * The new version of the directory entry. Since EXT4 structures are
1556 * stored in intel byte order, and the name_len field could never be
1557 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1560 struct ext4_dir_entry_2
{
1561 __le32 inode
; /* Inode number */
1562 __le16 rec_len
; /* Directory entry length */
1563 __u8 name_len
; /* Name length */
1565 char name
[EXT4_NAME_LEN
]; /* File name */
1569 * This is a bogus directory entry at the end of each leaf block that
1570 * records checksums.
1572 struct ext4_dir_entry_tail
{
1573 __le32 det_reserved_zero1
; /* Pretend to be unused */
1574 __le16 det_rec_len
; /* 12 */
1575 __u8 det_reserved_zero2
; /* Zero name length */
1576 __u8 det_reserved_ft
; /* 0xDE, fake file type */
1577 __le32 det_checksum
; /* crc32c(uuid+inum+dirblock) */
1581 * Ext4 directory file types. Only the low 3 bits are used. The
1582 * other bits are reserved for now.
1584 #define EXT4_FT_UNKNOWN 0
1585 #define EXT4_FT_REG_FILE 1
1586 #define EXT4_FT_DIR 2
1587 #define EXT4_FT_CHRDEV 3
1588 #define EXT4_FT_BLKDEV 4
1589 #define EXT4_FT_FIFO 5
1590 #define EXT4_FT_SOCK 6
1591 #define EXT4_FT_SYMLINK 7
1593 #define EXT4_FT_MAX 8
1595 #define EXT4_FT_DIR_CSUM 0xDE
1598 * EXT4_DIR_PAD defines the directory entries boundaries
1600 * NOTE: It must be a multiple of 4
1602 #define EXT4_DIR_PAD 4
1603 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1604 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1606 #define EXT4_MAX_REC_LEN ((1<<16)-1)
1609 * If we ever get support for fs block sizes > page_size, we'll need
1610 * to remove the #if statements in the next two functions...
1612 static inline unsigned int
1613 ext4_rec_len_from_disk(__le16 dlen
, unsigned blocksize
)
1615 unsigned len
= le16_to_cpu(dlen
);
1617 #if (PAGE_CACHE_SIZE >= 65536)
1618 if (len
== EXT4_MAX_REC_LEN
|| len
== 0)
1620 return (len
& 65532) | ((len
& 3) << 16);
1626 static inline __le16
ext4_rec_len_to_disk(unsigned len
, unsigned blocksize
)
1628 if ((len
> blocksize
) || (blocksize
> (1 << 18)) || (len
& 3))
1630 #if (PAGE_CACHE_SIZE >= 65536)
1632 return cpu_to_le16(len
);
1633 if (len
== blocksize
) {
1634 if (blocksize
== 65536)
1635 return cpu_to_le16(EXT4_MAX_REC_LEN
);
1637 return cpu_to_le16(0);
1639 return cpu_to_le16((len
& 65532) | ((len
>> 16) & 3));
1641 return cpu_to_le16(len
);
1646 * Hash Tree Directory indexing
1647 * (c) Daniel Phillips, 2001
1650 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1651 EXT4_FEATURE_COMPAT_DIR_INDEX) && \
1652 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
1653 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1654 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1656 /* Legal values for the dx_root hash_version field: */
1658 #define DX_HASH_LEGACY 0
1659 #define DX_HASH_HALF_MD4 1
1660 #define DX_HASH_TEA 2
1661 #define DX_HASH_LEGACY_UNSIGNED 3
1662 #define DX_HASH_HALF_MD4_UNSIGNED 4
1663 #define DX_HASH_TEA_UNSIGNED 5
1665 static inline u32
ext4_chksum(struct ext4_sb_info
*sbi
, u32 crc
,
1666 const void *address
, unsigned int length
)
1669 struct shash_desc shash
;
1674 BUG_ON(crypto_shash_descsize(sbi
->s_chksum_driver
)!=sizeof(desc
.ctx
));
1676 desc
.shash
.tfm
= sbi
->s_chksum_driver
;
1677 desc
.shash
.flags
= 0;
1678 *(u32
*)desc
.ctx
= crc
;
1680 err
= crypto_shash_update(&desc
.shash
, address
, length
);
1683 return *(u32
*)desc
.ctx
;
1688 /* hash info structure used by the directory hash */
1698 /* 32 and 64 bit signed EOF for dx directories */
1699 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
1700 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
1704 * Control parameters used by ext4_htree_next_block
1706 #define HASH_NB_ALWAYS 1
1710 * Describe an inode's exact location on disk and in memory
1714 struct buffer_head
*bh
;
1715 unsigned long offset
;
1716 ext4_group_t block_group
;
1719 static inline struct ext4_inode
*ext4_raw_inode(struct ext4_iloc
*iloc
)
1721 return (struct ext4_inode
*) (iloc
->bh
->b_data
+ iloc
->offset
);
1725 * This structure is stuffed into the struct file's private_data field
1726 * for directories. It is where we put information so that we can do
1727 * readdir operations in hash tree order.
1729 struct dir_private_info
{
1730 struct rb_root root
;
1731 struct rb_node
*curr_node
;
1732 struct fname
*extra_fname
;
1735 __u32 curr_minor_hash
;
1739 /* calculate the first block number of the group */
1740 static inline ext4_fsblk_t
1741 ext4_group_first_block_no(struct super_block
*sb
, ext4_group_t group_no
)
1743 return group_no
* (ext4_fsblk_t
)EXT4_BLOCKS_PER_GROUP(sb
) +
1744 le32_to_cpu(EXT4_SB(sb
)->s_es
->s_first_data_block
);
1748 * Special error return code only used by dx_probe() and its callers.
1750 #define ERR_BAD_DX_DIR -75000
1752 void ext4_get_group_no_and_offset(struct super_block
*sb
, ext4_fsblk_t blocknr
,
1753 ext4_group_t
*blockgrpp
, ext4_grpblk_t
*offsetp
);
1756 * Timeout and state flag for lazy initialization inode thread.
1758 #define EXT4_DEF_LI_WAIT_MULT 10
1759 #define EXT4_DEF_LI_MAX_START_DELAY 5
1760 #define EXT4_LAZYINIT_QUIT 0x0001
1761 #define EXT4_LAZYINIT_RUNNING 0x0002
1764 * Lazy inode table initialization info
1766 struct ext4_lazy_init
{
1767 unsigned long li_state
;
1768 struct list_head li_request_list
;
1769 struct mutex li_list_mtx
;
1772 struct ext4_li_request
{
1773 struct super_block
*lr_super
;
1774 struct ext4_sb_info
*lr_sbi
;
1775 ext4_group_t lr_next_group
;
1776 struct list_head lr_request
;
1777 unsigned long lr_next_sched
;
1778 unsigned long lr_timeout
;
1781 struct ext4_features
{
1782 struct kobject f_kobj
;
1783 struct completion f_kobj_unregister
;
1787 * This structure will be used for multiple mount protection. It will be
1788 * written into the block number saved in the s_mmp_block field in the
1789 * superblock. Programs that check MMP should assume that if
1790 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
1791 * to use the filesystem, regardless of how old the timestamp is.
1793 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
1794 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
1795 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
1796 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
1799 __le32 mmp_magic
; /* Magic number for MMP */
1800 __le32 mmp_seq
; /* Sequence no. updated periodically */
1803 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
1804 * purposes and do not affect the correctness of the algorithm
1806 __le64 mmp_time
; /* Time last updated */
1807 char mmp_nodename
[64]; /* Node which last updated MMP block */
1808 char mmp_bdevname
[32]; /* Bdev which last updated MMP block */
1811 * mmp_check_interval is used to verify if the MMP block has been
1812 * updated on the block device. The value is updated based on the
1813 * maximum time to write the MMP block during an update cycle.
1815 __le16 mmp_check_interval
;
1818 __le32 mmp_pad2
[226];
1819 __le32 mmp_checksum
; /* crc32c(uuid+mmp_block) */
1822 /* arguments passed to the mmp thread */
1824 struct buffer_head
*bh
; /* bh from initial read_mmp_block() */
1825 struct super_block
*sb
; /* super block of the fs */
1829 * Check interval multiplier
1830 * The MMP block is written every update interval and initially checked every
1831 * update interval x the multiplier (the value is then adapted based on the
1832 * write latency). The reason is that writes can be delayed under load and we
1833 * don't want readers to incorrectly assume that the filesystem is no longer
1836 #define EXT4_MMP_CHECK_MULT 2UL
1839 * Minimum interval for MMP checking in seconds.
1841 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
1844 * Maximum interval for MMP checking in seconds.
1846 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
1849 * Function prototypes
1853 * Ok, these declarations are also in <linux/kernel.h> but none of the
1854 * ext4 source programs needs to include it so they are duplicated here.
1856 # define NORET_TYPE /**/
1857 # define ATTRIB_NORET __attribute__((noreturn))
1858 # define NORET_AND noreturn,
1861 extern unsigned int ext4_count_free(char *bitmap
, unsigned numchars
);
1862 void ext4_inode_bitmap_csum_set(struct super_block
*sb
, ext4_group_t group
,
1863 struct ext4_group_desc
*gdp
,
1864 struct buffer_head
*bh
, int sz
);
1865 int ext4_inode_bitmap_csum_verify(struct super_block
*sb
, ext4_group_t group
,
1866 struct ext4_group_desc
*gdp
,
1867 struct buffer_head
*bh
, int sz
);
1868 void ext4_block_bitmap_csum_set(struct super_block
*sb
, ext4_group_t group
,
1869 struct ext4_group_desc
*gdp
,
1870 struct buffer_head
*bh
);
1871 int ext4_block_bitmap_csum_verify(struct super_block
*sb
, ext4_group_t group
,
1872 struct ext4_group_desc
*gdp
,
1873 struct buffer_head
*bh
);
1876 extern void ext4_validate_block_bitmap(struct super_block
*sb
,
1877 struct ext4_group_desc
*desc
,
1878 unsigned int block_group
,
1879 struct buffer_head
*bh
);
1880 extern unsigned int ext4_block_group(struct super_block
*sb
,
1881 ext4_fsblk_t blocknr
);
1882 extern ext4_grpblk_t
ext4_block_group_offset(struct super_block
*sb
,
1883 ext4_fsblk_t blocknr
);
1884 extern int ext4_bg_has_super(struct super_block
*sb
, ext4_group_t group
);
1885 extern unsigned long ext4_bg_num_gdb(struct super_block
*sb
,
1886 ext4_group_t group
);
1887 extern ext4_fsblk_t
ext4_new_meta_blocks(handle_t
*handle
, struct inode
*inode
,
1890 unsigned long *count
,
1892 extern int ext4_claim_free_clusters(struct ext4_sb_info
*sbi
,
1893 s64 nclusters
, unsigned int flags
);
1894 extern ext4_fsblk_t
ext4_count_free_clusters(struct super_block
*);
1895 extern void ext4_check_blocks_bitmap(struct super_block
*);
1896 extern struct ext4_group_desc
* ext4_get_group_desc(struct super_block
* sb
,
1897 ext4_group_t block_group
,
1898 struct buffer_head
** bh
);
1899 extern int ext4_should_retry_alloc(struct super_block
*sb
, int *retries
);
1901 extern struct buffer_head
*ext4_read_block_bitmap_nowait(struct super_block
*sb
,
1902 ext4_group_t block_group
);
1903 extern int ext4_wait_block_bitmap(struct super_block
*sb
,
1904 ext4_group_t block_group
,
1905 struct buffer_head
*bh
);
1906 extern struct buffer_head
*ext4_read_block_bitmap(struct super_block
*sb
,
1907 ext4_group_t block_group
);
1908 extern void ext4_init_block_bitmap(struct super_block
*sb
,
1909 struct buffer_head
*bh
,
1911 struct ext4_group_desc
*desc
);
1912 extern unsigned ext4_free_clusters_after_init(struct super_block
*sb
,
1913 ext4_group_t block_group
,
1914 struct ext4_group_desc
*gdp
);
1915 extern unsigned ext4_num_overhead_clusters(struct super_block
*sb
,
1916 ext4_group_t block_group
,
1917 struct ext4_group_desc
*gdp
);
1918 ext4_fsblk_t
ext4_inode_to_goal_block(struct inode
*);
1921 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode
*,
1923 struct ext4_dir_entry_2
*,
1924 struct buffer_head
*, unsigned int);
1925 #define ext4_check_dir_entry(dir, filp, de, bh, offset) \
1926 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
1927 (de), (bh), (offset)))
1928 extern int ext4_htree_store_dirent(struct file
*dir_file
, __u32 hash
,
1930 struct ext4_dir_entry_2
*dirent
);
1931 extern void ext4_htree_free_dir_info(struct dir_private_info
*p
);
1934 extern int ext4_sync_file(struct file
*, loff_t
, loff_t
, int);
1935 extern int ext4_flush_completed_IO(struct inode
*);
1938 extern int ext4fs_dirhash(const char *name
, int len
, struct
1939 dx_hash_info
*hinfo
);
1942 extern struct inode
*ext4_new_inode(handle_t
*, struct inode
*, umode_t
,
1943 const struct qstr
*qstr
, __u32 goal
,
1945 extern void ext4_free_inode(handle_t
*, struct inode
*);
1946 extern struct inode
* ext4_orphan_get(struct super_block
*, unsigned long);
1947 extern unsigned long ext4_count_free_inodes(struct super_block
*);
1948 extern unsigned long ext4_count_dirs(struct super_block
*);
1949 extern void ext4_check_inodes_bitmap(struct super_block
*);
1950 extern void ext4_mark_bitmap_end(int start_bit
, int end_bit
, char *bitmap
);
1951 extern int ext4_init_inode_table(struct super_block
*sb
,
1952 ext4_group_t group
, int barrier
);
1953 extern void ext4_end_bitmap_read(struct buffer_head
*bh
, int uptodate
);
1956 extern long ext4_mb_stats
;
1957 extern long ext4_mb_max_to_scan
;
1958 extern int ext4_mb_init(struct super_block
*);
1959 extern int ext4_mb_release(struct super_block
*);
1960 extern ext4_fsblk_t
ext4_mb_new_blocks(handle_t
*,
1961 struct ext4_allocation_request
*, int *);
1962 extern int ext4_mb_reserve_blocks(struct super_block
*, int);
1963 extern void ext4_discard_preallocations(struct inode
*);
1964 extern int __init
ext4_init_mballoc(void);
1965 extern void ext4_exit_mballoc(void);
1966 extern void ext4_free_blocks(handle_t
*handle
, struct inode
*inode
,
1967 struct buffer_head
*bh
, ext4_fsblk_t block
,
1968 unsigned long count
, int flags
);
1969 extern int ext4_mb_add_groupinfo(struct super_block
*sb
,
1970 ext4_group_t i
, struct ext4_group_desc
*desc
);
1971 extern int ext4_group_add_blocks(handle_t
*handle
, struct super_block
*sb
,
1972 ext4_fsblk_t block
, unsigned long count
);
1973 extern int ext4_trim_fs(struct super_block
*, struct fstrim_range
*);
1976 struct buffer_head
*ext4_getblk(handle_t
*, struct inode
*,
1977 ext4_lblk_t
, int, int *);
1978 struct buffer_head
*ext4_bread(handle_t
*, struct inode
*,
1979 ext4_lblk_t
, int, int *);
1980 int ext4_get_block(struct inode
*inode
, sector_t iblock
,
1981 struct buffer_head
*bh_result
, int create
);
1983 extern struct inode
*ext4_iget(struct super_block
*, unsigned long);
1984 extern int ext4_write_inode(struct inode
*, struct writeback_control
*);
1985 extern int ext4_setattr(struct dentry
*, struct iattr
*);
1986 extern int ext4_getattr(struct vfsmount
*mnt
, struct dentry
*dentry
,
1987 struct kstat
*stat
);
1988 extern void ext4_evict_inode(struct inode
*);
1989 extern void ext4_clear_inode(struct inode
*);
1990 extern int ext4_sync_inode(handle_t
*, struct inode
*);
1991 extern void ext4_dirty_inode(struct inode
*, int);
1992 extern int ext4_change_inode_journal_flag(struct inode
*, int);
1993 extern int ext4_get_inode_loc(struct inode
*, struct ext4_iloc
*);
1994 extern int ext4_can_truncate(struct inode
*inode
);
1995 extern void ext4_truncate(struct inode
*);
1996 extern int ext4_punch_hole(struct file
*file
, loff_t offset
, loff_t length
);
1997 extern int ext4_truncate_restart_trans(handle_t
*, struct inode
*, int nblocks
);
1998 extern void ext4_set_inode_flags(struct inode
*);
1999 extern void ext4_get_inode_flags(struct ext4_inode_info
*);
2000 extern int ext4_alloc_da_blocks(struct inode
*inode
);
2001 extern void ext4_set_aops(struct inode
*inode
);
2002 extern int ext4_writepage_trans_blocks(struct inode
*);
2003 extern int ext4_chunk_trans_blocks(struct inode
*, int nrblocks
);
2004 extern int ext4_discard_partial_page_buffers(handle_t
*handle
,
2005 struct address_space
*mapping
, loff_t from
,
2006 loff_t length
, int flags
);
2007 extern int ext4_page_mkwrite(struct vm_area_struct
*vma
, struct vm_fault
*vmf
);
2008 extern qsize_t
*ext4_get_reserved_space(struct inode
*inode
);
2009 extern void ext4_da_update_reserve_space(struct inode
*inode
,
2010 int used
, int quota_claim
);
2013 extern int ext4_ind_map_blocks(handle_t
*handle
, struct inode
*inode
,
2014 struct ext4_map_blocks
*map
, int flags
);
2015 extern ssize_t
ext4_ind_direct_IO(int rw
, struct kiocb
*iocb
,
2016 const struct iovec
*iov
, loff_t offset
,
2017 unsigned long nr_segs
);
2018 extern int ext4_ind_calc_metadata_amount(struct inode
*inode
, sector_t lblock
);
2019 extern int ext4_ind_trans_blocks(struct inode
*inode
, int nrblocks
, int chunk
);
2020 extern void ext4_ind_truncate(struct inode
*inode
);
2023 extern long ext4_ioctl(struct file
*, unsigned int, unsigned long);
2024 extern long ext4_compat_ioctl(struct file
*, unsigned int, unsigned long);
2027 extern int ext4_ext_migrate(struct inode
*);
2030 extern int ext4_dirent_csum_verify(struct inode
*inode
,
2031 struct ext4_dir_entry
*dirent
);
2032 extern int ext4_orphan_add(handle_t
*, struct inode
*);
2033 extern int ext4_orphan_del(handle_t
*, struct inode
*);
2034 extern int ext4_htree_fill_tree(struct file
*dir_file
, __u32 start_hash
,
2035 __u32 start_minor_hash
, __u32
*next_hash
);
2038 extern int ext4_group_add(struct super_block
*sb
,
2039 struct ext4_new_group_data
*input
);
2040 extern int ext4_group_extend(struct super_block
*sb
,
2041 struct ext4_super_block
*es
,
2042 ext4_fsblk_t n_blocks_count
);
2043 extern int ext4_resize_fs(struct super_block
*sb
, ext4_fsblk_t n_blocks_count
);
2046 extern int ext4_calculate_overhead(struct super_block
*sb
);
2047 extern int ext4_superblock_csum_verify(struct super_block
*sb
,
2048 struct ext4_super_block
*es
);
2049 extern void ext4_superblock_csum_set(struct super_block
*sb
);
2050 extern void *ext4_kvmalloc(size_t size
, gfp_t flags
);
2051 extern void *ext4_kvzalloc(size_t size
, gfp_t flags
);
2052 extern void ext4_kvfree(void *ptr
);
2053 extern __printf(4, 5)
2054 void __ext4_error(struct super_block
*, const char *, unsigned int,
2056 #define ext4_error(sb, message...) __ext4_error(sb, __func__, \
2057 __LINE__, ## message)
2058 extern __printf(5, 6)
2059 void ext4_error_inode(struct inode
*, const char *, unsigned int, ext4_fsblk_t
,
2061 extern __printf(5, 6)
2062 void ext4_error_file(struct file
*, const char *, unsigned int, ext4_fsblk_t
,
2064 extern void __ext4_std_error(struct super_block
*, const char *,
2066 extern __printf(4, 5)
2067 void __ext4_abort(struct super_block
*, const char *, unsigned int,
2069 #define ext4_abort(sb, message...) __ext4_abort(sb, __func__, \
2070 __LINE__, ## message)
2071 extern __printf(4, 5)
2072 void __ext4_warning(struct super_block
*, const char *, unsigned int,
2074 #define ext4_warning(sb, message...) __ext4_warning(sb, __func__, \
2075 __LINE__, ## message)
2076 extern __printf(3, 4)
2077 void ext4_msg(struct super_block
*, const char *, const char *, ...);
2078 extern void __dump_mmp_msg(struct super_block
*, struct mmp_struct
*mmp
,
2079 const char *, unsigned int, const char *);
2080 #define dump_mmp_msg(sb, mmp, msg) __dump_mmp_msg(sb, mmp, __func__, \
2082 extern __printf(7, 8)
2083 void __ext4_grp_locked_error(const char *, unsigned int,
2084 struct super_block
*, ext4_group_t
,
2085 unsigned long, ext4_fsblk_t
,
2087 #define ext4_grp_locked_error(sb, grp, message...) \
2088 __ext4_grp_locked_error(__func__, __LINE__, (sb), (grp), ## message)
2089 extern void ext4_update_dynamic_rev(struct super_block
*sb
);
2090 extern int ext4_update_compat_feature(handle_t
*handle
, struct super_block
*sb
,
2092 extern int ext4_update_rocompat_feature(handle_t
*handle
,
2093 struct super_block
*sb
, __u32 rocompat
);
2094 extern int ext4_update_incompat_feature(handle_t
*handle
,
2095 struct super_block
*sb
, __u32 incompat
);
2096 extern ext4_fsblk_t
ext4_block_bitmap(struct super_block
*sb
,
2097 struct ext4_group_desc
*bg
);
2098 extern ext4_fsblk_t
ext4_inode_bitmap(struct super_block
*sb
,
2099 struct ext4_group_desc
*bg
);
2100 extern ext4_fsblk_t
ext4_inode_table(struct super_block
*sb
,
2101 struct ext4_group_desc
*bg
);
2102 extern __u32
ext4_free_group_clusters(struct super_block
*sb
,
2103 struct ext4_group_desc
*bg
);
2104 extern __u32
ext4_free_inodes_count(struct super_block
*sb
,
2105 struct ext4_group_desc
*bg
);
2106 extern __u32
ext4_used_dirs_count(struct super_block
*sb
,
2107 struct ext4_group_desc
*bg
);
2108 extern __u32
ext4_itable_unused_count(struct super_block
*sb
,
2109 struct ext4_group_desc
*bg
);
2110 extern void ext4_block_bitmap_set(struct super_block
*sb
,
2111 struct ext4_group_desc
*bg
, ext4_fsblk_t blk
);
2112 extern void ext4_inode_bitmap_set(struct super_block
*sb
,
2113 struct ext4_group_desc
*bg
, ext4_fsblk_t blk
);
2114 extern void ext4_inode_table_set(struct super_block
*sb
,
2115 struct ext4_group_desc
*bg
, ext4_fsblk_t blk
);
2116 extern void ext4_free_group_clusters_set(struct super_block
*sb
,
2117 struct ext4_group_desc
*bg
,
2119 extern void ext4_free_inodes_set(struct super_block
*sb
,
2120 struct ext4_group_desc
*bg
, __u32 count
);
2121 extern void ext4_used_dirs_set(struct super_block
*sb
,
2122 struct ext4_group_desc
*bg
, __u32 count
);
2123 extern void ext4_itable_unused_set(struct super_block
*sb
,
2124 struct ext4_group_desc
*bg
, __u32 count
);
2125 extern int ext4_group_desc_csum_verify(struct super_block
*sb
, __u32 group
,
2126 struct ext4_group_desc
*gdp
);
2127 extern void ext4_group_desc_csum_set(struct super_block
*sb
, __u32 group
,
2128 struct ext4_group_desc
*gdp
);
2130 static inline int ext4_has_group_desc_csum(struct super_block
*sb
)
2132 return EXT4_HAS_RO_COMPAT_FEATURE(sb
,
2133 EXT4_FEATURE_RO_COMPAT_GDT_CSUM
|
2134 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM
);
2137 static inline ext4_fsblk_t
ext4_blocks_count(struct ext4_super_block
*es
)
2139 return ((ext4_fsblk_t
)le32_to_cpu(es
->s_blocks_count_hi
) << 32) |
2140 le32_to_cpu(es
->s_blocks_count_lo
);
2143 static inline ext4_fsblk_t
ext4_r_blocks_count(struct ext4_super_block
*es
)
2145 return ((ext4_fsblk_t
)le32_to_cpu(es
->s_r_blocks_count_hi
) << 32) |
2146 le32_to_cpu(es
->s_r_blocks_count_lo
);
2149 static inline ext4_fsblk_t
ext4_free_blocks_count(struct ext4_super_block
*es
)
2151 return ((ext4_fsblk_t
)le32_to_cpu(es
->s_free_blocks_count_hi
) << 32) |
2152 le32_to_cpu(es
->s_free_blocks_count_lo
);
2155 static inline void ext4_blocks_count_set(struct ext4_super_block
*es
,
2158 es
->s_blocks_count_lo
= cpu_to_le32((u32
)blk
);
2159 es
->s_blocks_count_hi
= cpu_to_le32(blk
>> 32);
2162 static inline void ext4_free_blocks_count_set(struct ext4_super_block
*es
,
2165 es
->s_free_blocks_count_lo
= cpu_to_le32((u32
)blk
);
2166 es
->s_free_blocks_count_hi
= cpu_to_le32(blk
>> 32);
2169 static inline void ext4_r_blocks_count_set(struct ext4_super_block
*es
,
2172 es
->s_r_blocks_count_lo
= cpu_to_le32((u32
)blk
);
2173 es
->s_r_blocks_count_hi
= cpu_to_le32(blk
>> 32);
2176 static inline loff_t
ext4_isize(struct ext4_inode
*raw_inode
)
2178 if (S_ISREG(le16_to_cpu(raw_inode
->i_mode
)))
2179 return ((loff_t
)le32_to_cpu(raw_inode
->i_size_high
) << 32) |
2180 le32_to_cpu(raw_inode
->i_size_lo
);
2182 return (loff_t
) le32_to_cpu(raw_inode
->i_size_lo
);
2185 static inline void ext4_isize_set(struct ext4_inode
*raw_inode
, loff_t i_size
)
2187 raw_inode
->i_size_lo
= cpu_to_le32(i_size
);
2188 raw_inode
->i_size_high
= cpu_to_le32(i_size
>> 32);
2192 struct ext4_group_info
*ext4_get_group_info(struct super_block
*sb
,
2195 struct ext4_group_info
***grp_info
;
2196 long indexv
, indexh
;
2197 grp_info
= EXT4_SB(sb
)->s_group_info
;
2198 indexv
= group
>> (EXT4_DESC_PER_BLOCK_BITS(sb
));
2199 indexh
= group
& ((EXT4_DESC_PER_BLOCK(sb
)) - 1);
2200 return grp_info
[indexv
][indexh
];
2204 * Reading s_groups_count requires using smp_rmb() afterwards. See
2205 * the locking protocol documented in the comments of ext4_group_add()
2208 static inline ext4_group_t
ext4_get_groups_count(struct super_block
*sb
)
2210 ext4_group_t ngroups
= EXT4_SB(sb
)->s_groups_count
;
2216 static inline ext4_group_t
ext4_flex_group(struct ext4_sb_info
*sbi
,
2217 ext4_group_t block_group
)
2219 return block_group
>> sbi
->s_log_groups_per_flex
;
2222 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info
*sbi
)
2224 return 1 << sbi
->s_log_groups_per_flex
;
2227 #define ext4_std_error(sb, errno) \
2230 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
2234 /* Each CPU can accumulate percpu_counter_batch clusters in their local
2235 * counters. So we need to make sure we have free clusters more
2236 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
2238 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2240 #define EXT4_FREECLUSTERS_WATERMARK 0
2243 static inline void ext4_update_i_disksize(struct inode
*inode
, loff_t newsize
)
2246 * XXX: replace with spinlock if seen contended -bzzz
2248 down_write(&EXT4_I(inode
)->i_data_sem
);
2249 if (newsize
> EXT4_I(inode
)->i_disksize
)
2250 EXT4_I(inode
)->i_disksize
= newsize
;
2251 up_write(&EXT4_I(inode
)->i_data_sem
);
2255 struct ext4_group_info
{
2256 unsigned long bb_state
;
2257 struct rb_root bb_free_root
;
2258 ext4_grpblk_t bb_first_free
; /* first free block */
2259 ext4_grpblk_t bb_free
; /* total free blocks */
2260 ext4_grpblk_t bb_fragments
; /* nr of freespace fragments */
2261 ext4_grpblk_t bb_largest_free_order
;/* order of largest frag in BG */
2262 struct list_head bb_prealloc_list
;
2266 struct rw_semaphore alloc_sem
;
2267 ext4_grpblk_t bb_counters
[]; /* Nr of free power-of-two-block
2268 * regions, index is order.
2269 * bb_counters[3] = 5 means
2270 * 5 free 8-block regions. */
2273 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
2274 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
2276 #define EXT4_MB_GRP_NEED_INIT(grp) \
2277 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2279 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
2280 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2281 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
2282 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2283 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
2284 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2286 #define EXT4_MAX_CONTENTION 8
2287 #define EXT4_CONTENTION_THRESHOLD 2
2289 static inline spinlock_t
*ext4_group_lock_ptr(struct super_block
*sb
,
2292 return bgl_lock_ptr(EXT4_SB(sb
)->s_blockgroup_lock
, group
);
2296 * Returns true if the filesystem is busy enough that attempts to
2297 * access the block group locks has run into contention.
2299 static inline int ext4_fs_is_busy(struct ext4_sb_info
*sbi
)
2301 return (atomic_read(&sbi
->s_lock_busy
) > EXT4_CONTENTION_THRESHOLD
);
2304 static inline void ext4_lock_group(struct super_block
*sb
, ext4_group_t group
)
2306 spinlock_t
*lock
= ext4_group_lock_ptr(sb
, group
);
2307 if (spin_trylock(lock
))
2309 * We're able to grab the lock right away, so drop the
2310 * lock contention counter.
2312 atomic_add_unless(&EXT4_SB(sb
)->s_lock_busy
, -1, 0);
2315 * The lock is busy, so bump the contention counter,
2316 * and then wait on the spin lock.
2318 atomic_add_unless(&EXT4_SB(sb
)->s_lock_busy
, 1,
2319 EXT4_MAX_CONTENTION
);
2324 static inline void ext4_unlock_group(struct super_block
*sb
,
2327 spin_unlock(ext4_group_lock_ptr(sb
, group
));
2331 * Block validity checking
2333 #define ext4_check_indirect_blockref(inode, bh) \
2334 ext4_check_blockref(__func__, __LINE__, inode, \
2335 (__le32 *)(bh)->b_data, \
2336 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2338 #define ext4_ind_check_inode(inode) \
2339 ext4_check_blockref(__func__, __LINE__, inode, \
2340 EXT4_I(inode)->i_data, \
2344 * Inodes and files operations
2348 extern const struct file_operations ext4_dir_operations
;
2351 extern const struct inode_operations ext4_file_inode_operations
;
2352 extern const struct file_operations ext4_file_operations
;
2353 extern loff_t
ext4_llseek(struct file
*file
, loff_t offset
, int origin
);
2356 extern const struct inode_operations ext4_dir_inode_operations
;
2357 extern const struct inode_operations ext4_special_inode_operations
;
2358 extern struct dentry
*ext4_get_parent(struct dentry
*child
);
2361 extern const struct inode_operations ext4_symlink_inode_operations
;
2362 extern const struct inode_operations ext4_fast_symlink_inode_operations
;
2364 /* block_validity */
2365 extern void ext4_release_system_zone(struct super_block
*sb
);
2366 extern int ext4_setup_system_zone(struct super_block
*sb
);
2367 extern int __init
ext4_init_system_zone(void);
2368 extern void ext4_exit_system_zone(void);
2369 extern int ext4_data_block_valid(struct ext4_sb_info
*sbi
,
2370 ext4_fsblk_t start_blk
,
2371 unsigned int count
);
2372 extern int ext4_check_blockref(const char *, unsigned int,
2373 struct inode
*, __le32
*, unsigned int);
2376 extern int ext4_ext_tree_init(handle_t
*handle
, struct inode
*);
2377 extern int ext4_ext_writepage_trans_blocks(struct inode
*, int);
2378 extern int ext4_ext_index_trans_blocks(struct inode
*inode
, int nrblocks
,
2380 extern int ext4_ext_map_blocks(handle_t
*handle
, struct inode
*inode
,
2381 struct ext4_map_blocks
*map
, int flags
);
2382 extern void ext4_ext_truncate(struct inode
*);
2383 extern int ext4_ext_punch_hole(struct file
*file
, loff_t offset
,
2385 extern void ext4_ext_init(struct super_block
*);
2386 extern void ext4_ext_release(struct super_block
*);
2387 extern long ext4_fallocate(struct file
*file
, int mode
, loff_t offset
,
2389 extern int ext4_convert_unwritten_extents(struct inode
*inode
, loff_t offset
,
2391 extern int ext4_map_blocks(handle_t
*handle
, struct inode
*inode
,
2392 struct ext4_map_blocks
*map
, int flags
);
2393 extern int ext4_fiemap(struct inode
*inode
, struct fiemap_extent_info
*fieinfo
,
2394 __u64 start
, __u64 len
);
2396 extern int ext4_move_extents(struct file
*o_filp
, struct file
*d_filp
,
2397 __u64 start_orig
, __u64 start_donor
,
2398 __u64 len
, __u64
*moved_len
);
2401 extern int __init
ext4_init_pageio(void);
2402 extern void ext4_exit_pageio(void);
2403 extern void ext4_ioend_wait(struct inode
*);
2404 extern void ext4_free_io_end(ext4_io_end_t
*io
);
2405 extern ext4_io_end_t
*ext4_init_io_end(struct inode
*inode
, gfp_t flags
);
2406 extern int ext4_end_io_nolock(ext4_io_end_t
*io
);
2407 extern void ext4_io_submit(struct ext4_io_submit
*io
);
2408 extern int ext4_bio_write_page(struct ext4_io_submit
*io
,
2411 struct writeback_control
*wbc
);
2414 extern int ext4_multi_mount_protect(struct super_block
*, ext4_fsblk_t
);
2415 extern void ext4_mmp_csum_set(struct super_block
*sb
, struct mmp_struct
*mmp
);
2416 extern int ext4_mmp_csum_verify(struct super_block
*sb
,
2417 struct mmp_struct
*mmp
);
2419 /* BH_Uninit flag: blocks are allocated but uninitialized on disk */
2420 enum ext4_state_bits
{
2421 BH_Uninit
/* blocks are allocated but uninitialized on disk */
2422 = BH_JBDPrivateStart
,
2423 BH_AllocFromCluster
, /* allocated blocks were part of already
2424 * allocated cluster. Note that this flag will
2425 * never, ever appear in a buffer_head's state
2426 * flag. See EXT4_MAP_FROM_CLUSTER to see where
2428 BH_Da_Mapped
, /* Delayed allocated block that now has a mapping. This
2429 * flag is set when ext4_map_blocks is called on a
2430 * delayed allocated block to get its real mapping. */
2433 BUFFER_FNS(Uninit
, uninit
)
2434 TAS_BUFFER_FNS(Uninit
, uninit
)
2435 BUFFER_FNS(Da_Mapped
, da_mapped
)
2438 * Add new method to test wether block and inode bitmaps are properly
2439 * initialized. With uninit_bg reading the block from disk is not enough
2440 * to mark the bitmap uptodate. We need to also zero-out the bitmap
2442 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
2444 static inline int bitmap_uptodate(struct buffer_head
*bh
)
2446 return (buffer_uptodate(bh
) &&
2447 test_bit(BH_BITMAP_UPTODATE
, &(bh
)->b_state
));
2449 static inline void set_bitmap_uptodate(struct buffer_head
*bh
)
2451 set_bit(BH_BITMAP_UPTODATE
, &(bh
)->b_state
);
2454 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
2456 /* For ioend & aio unwritten conversion wait queues */
2457 #define EXT4_WQ_HASH_SZ 37
2458 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
2460 #define ext4_aio_mutex(v) (&ext4__aio_mutex[((unsigned long)(v)) %\
2462 extern wait_queue_head_t ext4__ioend_wq
[EXT4_WQ_HASH_SZ
];
2463 extern struct mutex ext4__aio_mutex
[EXT4_WQ_HASH_SZ
];
2465 #define EXT4_RESIZING 0
2466 extern int ext4_resize_begin(struct super_block
*sb
);
2467 extern void ext4_resize_end(struct super_block
*sb
);
2469 #endif /* __KERNEL__ */
2471 #include "ext4_extents.h"
2473 #endif /* _EXT4_H */