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>
33 #include <linux/compat.h>
37 * The fourth extended filesystem constants/structures
41 * Define EXT4FS_DEBUG to produce debug messages
49 #define ext4_debug(f, a...) \
51 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
52 __FILE__, __LINE__, __func__); \
53 printk(KERN_DEBUG f, ## a); \
56 #define ext4_debug(f, a...) do {} while (0)
59 /* data type for block offset of block group */
60 typedef int ext4_grpblk_t
;
62 /* data type for filesystem-wide blocks number */
63 typedef unsigned long long ext4_fsblk_t
;
65 /* data type for file logical block number */
66 typedef __u32 ext4_lblk_t
;
68 /* data type for block group number */
69 typedef unsigned int ext4_group_t
;
72 * Flags used in mballoc's allocation_context flags field.
74 * Also used to show what's going on for debugging purposes when the
75 * flag field is exported via the traceport interface
78 /* prefer goal again. length */
79 #define EXT4_MB_HINT_MERGE 0x0001
80 /* blocks already reserved */
81 #define EXT4_MB_HINT_RESERVED 0x0002
82 /* metadata is being allocated */
83 #define EXT4_MB_HINT_METADATA 0x0004
84 /* first blocks in the file */
85 #define EXT4_MB_HINT_FIRST 0x0008
86 /* search for the best chunk */
87 #define EXT4_MB_HINT_BEST 0x0010
88 /* data is being allocated */
89 #define EXT4_MB_HINT_DATA 0x0020
90 /* don't preallocate (for tails) */
91 #define EXT4_MB_HINT_NOPREALLOC 0x0040
92 /* allocate for locality group */
93 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080
94 /* allocate goal blocks or none */
95 #define EXT4_MB_HINT_GOAL_ONLY 0x0100
96 /* goal is meaningful */
97 #define EXT4_MB_HINT_TRY_GOAL 0x0200
98 /* blocks already pre-reserved by delayed allocation */
99 #define EXT4_MB_DELALLOC_RESERVED 0x0400
100 /* We are doing stream allocation */
101 #define EXT4_MB_STREAM_ALLOC 0x0800
104 struct ext4_allocation_request
{
105 /* target inode for block we're allocating */
107 /* how many blocks we want to allocate */
109 /* logical block in target inode */
111 /* the closest logical allocated block to the left */
113 /* the closest logical allocated block to the right */
115 /* phys. target (a hint) */
117 /* phys. block for the closest logical allocated block to the left */
119 /* phys. block for the closest logical allocated block to the right */
121 /* flags. see above EXT4_MB_HINT_* */
126 * For delayed allocation tracking
128 struct mpage_da_data
{
130 sector_t b_blocknr
; /* start block number of extent */
131 size_t b_size
; /* size of extent */
132 unsigned long b_state
; /* state of the extent */
133 unsigned long first_page
, next_page
; /* extent of pages */
134 struct writeback_control
*wbc
;
139 #define DIO_AIO_UNWRITTEN 0x1
140 typedef struct ext4_io_end
{
141 struct list_head list
; /* per-file finished AIO list */
142 struct inode
*inode
; /* file being written to */
143 unsigned int flag
; /* unwritten or not */
144 int error
; /* I/O error code */
145 loff_t offset
; /* offset in the file */
146 ssize_t size
; /* size of the extent */
147 struct work_struct work
; /* data work queue */
151 * Special inodes numbers
153 #define EXT4_BAD_INO 1 /* Bad blocks inode */
154 #define EXT4_ROOT_INO 2 /* Root inode */
155 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
156 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
157 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
158 #define EXT4_JOURNAL_INO 8 /* Journal inode */
160 /* First non-reserved inode for old ext4 filesystems */
161 #define EXT4_GOOD_OLD_FIRST_INO 11
164 * Maximal count of links to a file
166 #define EXT4_LINK_MAX 65000
169 * Macro-instructions used to manage several block sizes
171 #define EXT4_MIN_BLOCK_SIZE 1024
172 #define EXT4_MAX_BLOCK_SIZE 65536
173 #define EXT4_MIN_BLOCK_LOG_SIZE 10
175 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
177 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
179 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
181 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
183 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
186 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
187 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
188 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
190 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
191 EXT4_GOOD_OLD_INODE_SIZE : \
193 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
194 EXT4_GOOD_OLD_FIRST_INO : \
197 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
200 * Structure of a blocks group descriptor
202 struct ext4_group_desc
204 __le32 bg_block_bitmap_lo
; /* Blocks bitmap block */
205 __le32 bg_inode_bitmap_lo
; /* Inodes bitmap block */
206 __le32 bg_inode_table_lo
; /* Inodes table block */
207 __le16 bg_free_blocks_count_lo
;/* Free blocks count */
208 __le16 bg_free_inodes_count_lo
;/* Free inodes count */
209 __le16 bg_used_dirs_count_lo
; /* Directories count */
210 __le16 bg_flags
; /* EXT4_BG_flags (INODE_UNINIT, etc) */
211 __u32 bg_reserved
[2]; /* Likely block/inode bitmap checksum */
212 __le16 bg_itable_unused_lo
; /* Unused inodes count */
213 __le16 bg_checksum
; /* crc16(sb_uuid+group+desc) */
214 __le32 bg_block_bitmap_hi
; /* Blocks bitmap block MSB */
215 __le32 bg_inode_bitmap_hi
; /* Inodes bitmap block MSB */
216 __le32 bg_inode_table_hi
; /* Inodes table block MSB */
217 __le16 bg_free_blocks_count_hi
;/* Free blocks count MSB */
218 __le16 bg_free_inodes_count_hi
;/* Free inodes count MSB */
219 __le16 bg_used_dirs_count_hi
; /* Directories count MSB */
220 __le16 bg_itable_unused_hi
; /* Unused inodes count MSB */
221 __u32 bg_reserved2
[3];
225 * Structure of a flex block group info
229 atomic_t free_inodes
;
230 atomic_t free_blocks
;
234 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
235 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
236 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
239 * Macro-instructions used to manage group descriptors
241 #define EXT4_MIN_DESC_SIZE 32
242 #define EXT4_MIN_DESC_SIZE_64BIT 64
243 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
244 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
246 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
247 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
248 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
249 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
251 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
252 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
253 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
257 * Constants relative to the data blocks
259 #define EXT4_NDIR_BLOCKS 12
260 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
261 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
262 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
263 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
268 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
269 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
270 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
271 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
272 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
273 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
274 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
275 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
276 /* Reserved for compression usage... */
277 #define EXT4_DIRTY_FL 0x00000100
278 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
279 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
280 #define EXT4_ECOMPR_FL 0x00000800 /* Compression error */
281 /* End compression flags --- maybe not all used */
282 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
283 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
284 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
285 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
286 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
287 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
288 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
289 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
290 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
291 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
292 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
294 #define EXT4_FL_USER_VISIBLE 0x004BDFFF /* User visible flags */
295 #define EXT4_FL_USER_MODIFIABLE 0x004B80FF /* User modifiable flags */
297 /* Flags that should be inherited by new inodes from their parent. */
298 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
299 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
300 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
301 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
303 /* Flags that are appropriate for regular files (all but dir-specific ones). */
304 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
306 /* Flags that are appropriate for non-directories/regular files. */
307 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
309 /* Mask out flags that are inappropriate for the given type of inode. */
310 static inline __u32
ext4_mask_flags(umode_t mode
, __u32 flags
)
314 else if (S_ISREG(mode
))
315 return flags
& EXT4_REG_FLMASK
;
317 return flags
& EXT4_OTHER_FLMASK
;
321 * Inode flags used for atomic set/get
324 EXT4_INODE_SECRM
= 0, /* Secure deletion */
325 EXT4_INODE_UNRM
= 1, /* Undelete */
326 EXT4_INODE_COMPR
= 2, /* Compress file */
327 EXT4_INODE_SYNC
= 3, /* Synchronous updates */
328 EXT4_INODE_IMMUTABLE
= 4, /* Immutable file */
329 EXT4_INODE_APPEND
= 5, /* writes to file may only append */
330 EXT4_INODE_NODUMP
= 6, /* do not dump file */
331 EXT4_INODE_NOATIME
= 7, /* do not update atime */
332 /* Reserved for compression usage... */
333 EXT4_INODE_DIRTY
= 8,
334 EXT4_INODE_COMPRBLK
= 9, /* One or more compressed clusters */
335 EXT4_INODE_NOCOMPR
= 10, /* Don't compress */
336 EXT4_INODE_ECOMPR
= 11, /* Compression error */
337 /* End compression flags --- maybe not all used */
338 EXT4_INODE_INDEX
= 12, /* hash-indexed directory */
339 EXT4_INODE_IMAGIC
= 13, /* AFS directory */
340 EXT4_INODE_JOURNAL_DATA
= 14, /* file data should be journaled */
341 EXT4_INODE_NOTAIL
= 15, /* file tail should not be merged */
342 EXT4_INODE_DIRSYNC
= 16, /* dirsync behaviour (directories only) */
343 EXT4_INODE_TOPDIR
= 17, /* Top of directory hierarchies*/
344 EXT4_INODE_HUGE_FILE
= 18, /* Set to each huge file */
345 EXT4_INODE_EXTENTS
= 19, /* Inode uses extents */
346 EXT4_INODE_EA_INODE
= 21, /* Inode used for large EA */
347 EXT4_INODE_EOFBLOCKS
= 22, /* Blocks allocated beyond EOF */
348 EXT4_INODE_RESERVED
= 31, /* reserved for ext4 lib */
351 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
352 #define CHECK_FLAG_VALUE(FLAG) if (!TEST_FLAG_VALUE(FLAG)) { \
353 printk(KERN_EMERG "EXT4 flag fail: " #FLAG ": %d %d\n", \
354 EXT4_##FLAG##_FL, EXT4_INODE_##FLAG); BUG_ON(1); }
357 * Since it's pretty easy to mix up bit numbers and hex values, and we
358 * can't do a compile-time test for ENUM values, we use a run-time
359 * test to make sure that EXT4_XXX_FL is consistent with respect to
360 * EXT4_INODE_XXX. If all is well the printk and BUG_ON will all drop
361 * out so it won't cost any extra space in the compiled kernel image.
362 * But it's important that these values are the same, since we are
363 * using EXT4_INODE_XXX to test for the flag values, but EXT4_XX_FL
364 * must be consistent with the values of FS_XXX_FL defined in
365 * include/linux/fs.h and the on-disk values found in ext2, ext3, and
366 * ext4 filesystems, and of course the values defined in e2fsprogs.
368 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
370 static inline void ext4_check_flag_values(void)
372 CHECK_FLAG_VALUE(SECRM
);
373 CHECK_FLAG_VALUE(UNRM
);
374 CHECK_FLAG_VALUE(COMPR
);
375 CHECK_FLAG_VALUE(SYNC
);
376 CHECK_FLAG_VALUE(IMMUTABLE
);
377 CHECK_FLAG_VALUE(APPEND
);
378 CHECK_FLAG_VALUE(NODUMP
);
379 CHECK_FLAG_VALUE(NOATIME
);
380 CHECK_FLAG_VALUE(DIRTY
);
381 CHECK_FLAG_VALUE(COMPRBLK
);
382 CHECK_FLAG_VALUE(NOCOMPR
);
383 CHECK_FLAG_VALUE(ECOMPR
);
384 CHECK_FLAG_VALUE(INDEX
);
385 CHECK_FLAG_VALUE(IMAGIC
);
386 CHECK_FLAG_VALUE(JOURNAL_DATA
);
387 CHECK_FLAG_VALUE(NOTAIL
);
388 CHECK_FLAG_VALUE(DIRSYNC
);
389 CHECK_FLAG_VALUE(TOPDIR
);
390 CHECK_FLAG_VALUE(HUGE_FILE
);
391 CHECK_FLAG_VALUE(EXTENTS
);
392 CHECK_FLAG_VALUE(EA_INODE
);
393 CHECK_FLAG_VALUE(EOFBLOCKS
);
394 CHECK_FLAG_VALUE(RESERVED
);
397 /* Used to pass group descriptor data when online resize is done */
398 struct ext4_new_group_input
{
399 __u32 group
; /* Group number for this data */
400 __u64 block_bitmap
; /* Absolute block number of block bitmap */
401 __u64 inode_bitmap
; /* Absolute block number of inode bitmap */
402 __u64 inode_table
; /* Absolute block number of inode table start */
403 __u32 blocks_count
; /* Total number of blocks in this group */
404 __u16 reserved_blocks
; /* Number of reserved blocks in this group */
408 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
409 struct compat_ext4_new_group_input
{
411 compat_u64 block_bitmap
;
412 compat_u64 inode_bitmap
;
413 compat_u64 inode_table
;
420 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
421 struct ext4_new_group_data
{
427 __u16 reserved_blocks
;
429 __u32 free_blocks_count
;
433 * Flags used by ext4_get_blocks()
435 /* Allocate any needed blocks and/or convert an unitialized
436 extent to be an initialized ext4 */
437 #define EXT4_GET_BLOCKS_CREATE 0x0001
438 /* Request the creation of an unitialized extent */
439 #define EXT4_GET_BLOCKS_UNINIT_EXT 0x0002
440 #define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT (EXT4_GET_BLOCKS_UNINIT_EXT|\
441 EXT4_GET_BLOCKS_CREATE)
442 /* Caller is from the delayed allocation writeout path,
443 so set the magic i_delalloc_reserve_flag after taking the
444 inode allocation semaphore for */
445 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
446 /* caller is from the direct IO path, request to creation of an
447 unitialized extents if not allocated, split the uninitialized
448 extent if blocks has been preallocated already*/
449 #define EXT4_GET_BLOCKS_DIO 0x0008
450 #define EXT4_GET_BLOCKS_CONVERT 0x0010
451 #define EXT4_GET_BLOCKS_DIO_CREATE_EXT (EXT4_GET_BLOCKS_DIO|\
452 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
453 /* Convert extent to initialized after direct IO complete */
454 #define EXT4_GET_BLOCKS_DIO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
455 EXT4_GET_BLOCKS_DIO_CREATE_EXT)
460 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
461 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
462 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
463 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
464 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
465 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
466 #ifdef CONFIG_JBD2_DEBUG
467 #define EXT4_IOC_WAIT_FOR_READONLY _IOR('f', 99, long)
469 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
470 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
471 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
472 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
473 #define EXT4_IOC_MIGRATE _IO('f', 9)
474 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
475 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
476 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
477 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
479 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
481 * ioctl commands in 32 bit emulation
483 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
484 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
485 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
486 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
487 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
488 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
489 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
490 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
491 #ifdef CONFIG_JBD2_DEBUG
492 #define EXT4_IOC32_WAIT_FOR_READONLY _IOR('f', 99, int)
494 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
495 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
502 struct ext4_mount_options
{
503 unsigned long s_mount_opt
;
506 unsigned long s_commit_interval
;
507 u32 s_min_batch_time
, s_max_batch_time
;
510 char *s_qf_names
[MAXQUOTAS
];
514 /* Max physical block we can addres w/o extents */
515 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
518 * Structure of an inode on the disk
521 __le16 i_mode
; /* File mode */
522 __le16 i_uid
; /* Low 16 bits of Owner Uid */
523 __le32 i_size_lo
; /* Size in bytes */
524 __le32 i_atime
; /* Access time */
525 __le32 i_ctime
; /* Inode Change time */
526 __le32 i_mtime
; /* Modification time */
527 __le32 i_dtime
; /* Deletion Time */
528 __le16 i_gid
; /* Low 16 bits of Group Id */
529 __le16 i_links_count
; /* Links count */
530 __le32 i_blocks_lo
; /* Blocks count */
531 __le32 i_flags
; /* File flags */
537 __u32 h_i_translator
;
542 } osd1
; /* OS dependent 1 */
543 __le32 i_block
[EXT4_N_BLOCKS
];/* Pointers to blocks */
544 __le32 i_generation
; /* File version (for NFS) */
545 __le32 i_file_acl_lo
; /* File ACL */
547 __le32 i_obso_faddr
; /* Obsoleted fragment address */
550 __le16 l_i_blocks_high
; /* were l_i_reserved1 */
551 __le16 l_i_file_acl_high
;
552 __le16 l_i_uid_high
; /* these 2 fields */
553 __le16 l_i_gid_high
; /* were reserved2[0] */
557 __le16 h_i_reserved1
; /* Obsoleted fragment number/size which are removed in ext4 */
564 __le16 h_i_reserved1
; /* Obsoleted fragment number/size which are removed in ext4 */
565 __le16 m_i_file_acl_high
;
566 __u32 m_i_reserved2
[2];
568 } osd2
; /* OS dependent 2 */
569 __le16 i_extra_isize
;
571 __le32 i_ctime_extra
; /* extra Change time (nsec << 2 | epoch) */
572 __le32 i_mtime_extra
; /* extra Modification time(nsec << 2 | epoch) */
573 __le32 i_atime_extra
; /* extra Access time (nsec << 2 | epoch) */
574 __le32 i_crtime
; /* File Creation time */
575 __le32 i_crtime_extra
; /* extra FileCreationtime (nsec << 2 | epoch) */
576 __le32 i_version_hi
; /* high 32 bits for 64-bit version */
580 __u32 reserved
; /* should be zero */
581 __u32 donor_fd
; /* donor file descriptor */
582 __u64 orig_start
; /* logical start offset in block for orig */
583 __u64 donor_start
; /* logical start offset in block for donor */
584 __u64 len
; /* block length to be moved */
585 __u64 moved_len
; /* moved block length */
588 #define EXT4_EPOCH_BITS 2
589 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
590 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
593 * Extended fields will fit into an inode if the filesystem was formatted
594 * with large inodes (-I 256 or larger) and there are not currently any EAs
595 * consuming all of the available space. For new inodes we always reserve
596 * enough space for the kernel's known extended fields, but for inodes
597 * created with an old kernel this might not have been the case. None of
598 * the extended inode fields is critical for correct filesystem operation.
599 * This macro checks if a certain field fits in the inode. Note that
600 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
602 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
603 ((offsetof(typeof(*ext4_inode), field) + \
604 sizeof((ext4_inode)->field)) \
605 <= (EXT4_GOOD_OLD_INODE_SIZE + \
606 (einode)->i_extra_isize)) \
608 static inline __le32 ext4_encode_extra_time(struct timespec *time)
610 return cpu_to_le32((sizeof(time
->tv_sec
) > 4 ?
611 (time
->tv_sec
>> 32) & EXT4_EPOCH_MASK
: 0) |
612 ((time
->tv_nsec
<< EXT4_EPOCH_BITS
) & EXT4_NSEC_MASK
));
615 static inline void ext4_decode_extra_time(struct timespec
*time
, __le32 extra
)
617 if (sizeof(time
->tv_sec
) > 4)
618 time
->tv_sec
|= (__u64
)(le32_to_cpu(extra
) & EXT4_EPOCH_MASK
)
620 time
->tv_nsec
= (le32_to_cpu(extra
) & EXT4_NSEC_MASK
) >> EXT4_EPOCH_BITS
;
623 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
625 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
626 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
627 (raw_inode)->xtime ## _extra = \
628 ext4_encode_extra_time(&(inode)->xtime); \
631 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
633 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
634 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
635 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
636 (raw_inode)->xtime ## _extra = \
637 ext4_encode_extra_time(&(einode)->xtime); \
640 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
642 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
643 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
644 ext4_decode_extra_time(&(inode)->xtime, \
645 raw_inode->xtime ## _extra); \
648 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
650 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
651 (einode)->xtime.tv_sec = \
652 (signed)le32_to_cpu((raw_inode)->xtime); \
653 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
654 ext4_decode_extra_time(&(einode)->xtime, \
655 raw_inode->xtime ## _extra); \
658 #define i_disk_version osd1.linux1.l_i_version
660 #if defined(__KERNEL__) || defined(__linux__)
661 #define i_reserved1 osd1.linux1.l_i_reserved1
662 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
663 #define i_blocks_high osd2.linux2.l_i_blocks_high
664 #define i_uid_low i_uid
665 #define i_gid_low i_gid
666 #define i_uid_high osd2.linux2.l_i_uid_high
667 #define i_gid_high osd2.linux2.l_i_gid_high
668 #define i_reserved2 osd2.linux2.l_i_reserved2
670 #elif defined(__GNU__)
672 #define i_translator osd1.hurd1.h_i_translator
673 #define i_uid_high osd2.hurd2.h_i_uid_high
674 #define i_gid_high osd2.hurd2.h_i_gid_high
675 #define i_author osd2.hurd2.h_i_author
677 #elif defined(__masix__)
679 #define i_reserved1 osd1.masix1.m_i_reserved1
680 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
681 #define i_reserved2 osd2.masix2.m_i_reserved2
683 #endif /* defined(__KERNEL__) || defined(__linux__) */
686 * storage for cached extent
688 struct ext4_ext_cache
{
689 ext4_fsblk_t ec_start
;
690 ext4_lblk_t ec_block
;
691 __u32 ec_len
; /* must be 32bit to return holes */
696 * fourth extended file system inode data in memory
698 struct ext4_inode_info
{
699 __le32 i_data
[15]; /* unconverted */
701 ext4_fsblk_t i_file_acl
;
704 * i_block_group is the number of the block group which contains
705 * this file's inode. Constant across the lifetime of the inode,
706 * it is ued for making block allocation decisions - we try to
707 * place a file's data blocks near its inode block, and new inodes
708 * near to their parent directory's inode.
710 ext4_group_t i_block_group
;
711 unsigned long i_state_flags
; /* Dynamic state flags */
712 unsigned long i_flags
;
714 ext4_lblk_t i_dir_start_lookup
;
715 #ifdef CONFIG_EXT4_FS_XATTR
717 * Extended attributes can be read independently of the main file
718 * data. Taking i_mutex even when reading would cause contention
719 * between readers of EAs and writers of regular file data, so
720 * instead we synchronize on xattr_sem when reading or changing
723 struct rw_semaphore xattr_sem
;
726 struct list_head i_orphan
; /* unlinked but open inodes */
729 * i_disksize keeps track of what the inode size is ON DISK, not
730 * in memory. During truncate, i_size is set to the new size by
731 * the VFS prior to calling ext4_truncate(), but the filesystem won't
732 * set i_disksize to 0 until the truncate is actually under way.
734 * The intent is that i_disksize always represents the blocks which
735 * are used by this file. This allows recovery to restart truncate
736 * on orphans if we crash during truncate. We actually write i_disksize
737 * into the on-disk inode when writing inodes out, instead of i_size.
739 * The only time when i_disksize and i_size may be different is when
740 * a truncate is in progress. The only things which change i_disksize
741 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
746 * i_data_sem is for serialising ext4_truncate() against
747 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
748 * data tree are chopped off during truncate. We can't do that in
749 * ext4 because whenever we perform intermediate commits during
750 * truncate, the inode and all the metadata blocks *must* be in a
751 * consistent state which allows truncation of the orphans to restart
752 * during recovery. Hence we must fix the get_block-vs-truncate race
753 * by other means, so we have i_data_sem.
755 struct rw_semaphore i_data_sem
;
756 struct inode vfs_inode
;
757 struct jbd2_inode jinode
;
759 struct ext4_ext_cache i_cached_extent
;
761 * File creation time. Its function is same as that of
762 * struct timespec i_{a,c,m}time in the generic inode.
764 struct timespec i_crtime
;
767 struct list_head i_prealloc_list
;
768 spinlock_t i_prealloc_lock
;
771 ext4_group_t i_last_alloc_group
;
773 /* allocation reservation info for delalloc */
774 unsigned int i_reserved_data_blocks
;
775 unsigned int i_reserved_meta_blocks
;
776 unsigned int i_allocated_meta_blocks
;
777 unsigned short i_delalloc_reserved_flag
;
778 sector_t i_da_metadata_calc_last_lblock
;
779 int i_da_metadata_calc_len
;
781 /* on-disk additional length */
784 spinlock_t i_block_reservation_lock
;
786 /* quota space reservation, managed internally by quota code */
787 qsize_t i_reserved_quota
;
790 /* completed async DIOs that might need unwritten extents handling */
791 struct list_head i_aio_dio_complete_list
;
792 /* current io_end structure for async DIO write*/
793 ext4_io_end_t
*cur_aio_dio
;
796 * Transactions that contain inode's metadata needed to complete
797 * fsync and fdatasync, respectively.
800 tid_t i_datasync_tid
;
806 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
807 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
808 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
811 * Misc. filesystem flags
813 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
814 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
815 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
820 #define EXT4_MOUNT_OLDALLOC 0x00002 /* Don't use the new Orlov allocator */
821 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
822 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
823 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
824 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
825 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
826 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
827 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
828 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
829 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
830 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
831 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
832 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
833 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
834 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
835 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
836 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
837 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
838 #define EXT4_MOUNT_NOBH 0x40000 /* No bufferheads */
839 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
840 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
841 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
842 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
843 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
844 #define EXT4_MOUNT_I_VERSION 0x2000000 /* i_version support */
845 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
846 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
847 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
848 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
850 #define clear_opt(o, opt) o &= ~EXT4_MOUNT_##opt
851 #define set_opt(o, opt) o |= EXT4_MOUNT_##opt
852 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
855 #define ext4_set_bit ext2_set_bit
856 #define ext4_set_bit_atomic ext2_set_bit_atomic
857 #define ext4_clear_bit ext2_clear_bit
858 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
859 #define ext4_test_bit ext2_test_bit
860 #define ext4_find_first_zero_bit ext2_find_first_zero_bit
861 #define ext4_find_next_zero_bit ext2_find_next_zero_bit
862 #define ext4_find_next_bit ext2_find_next_bit
865 * Maximal mount counts between two filesystem checks
867 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
868 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
871 * Behaviour when detecting errors
873 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
874 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
875 #define EXT4_ERRORS_PANIC 3 /* Panic */
876 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
879 * Structure of the super block
881 struct ext4_super_block
{
882 /*00*/ __le32 s_inodes_count
; /* Inodes count */
883 __le32 s_blocks_count_lo
; /* Blocks count */
884 __le32 s_r_blocks_count_lo
; /* Reserved blocks count */
885 __le32 s_free_blocks_count_lo
; /* Free blocks count */
886 /*10*/ __le32 s_free_inodes_count
; /* Free inodes count */
887 __le32 s_first_data_block
; /* First Data Block */
888 __le32 s_log_block_size
; /* Block size */
889 __le32 s_obso_log_frag_size
; /* Obsoleted fragment size */
890 /*20*/ __le32 s_blocks_per_group
; /* # Blocks per group */
891 __le32 s_obso_frags_per_group
; /* Obsoleted fragments per group */
892 __le32 s_inodes_per_group
; /* # Inodes per group */
893 __le32 s_mtime
; /* Mount time */
894 /*30*/ __le32 s_wtime
; /* Write time */
895 __le16 s_mnt_count
; /* Mount count */
896 __le16 s_max_mnt_count
; /* Maximal mount count */
897 __le16 s_magic
; /* Magic signature */
898 __le16 s_state
; /* File system state */
899 __le16 s_errors
; /* Behaviour when detecting errors */
900 __le16 s_minor_rev_level
; /* minor revision level */
901 /*40*/ __le32 s_lastcheck
; /* time of last check */
902 __le32 s_checkinterval
; /* max. time between checks */
903 __le32 s_creator_os
; /* OS */
904 __le32 s_rev_level
; /* Revision level */
905 /*50*/ __le16 s_def_resuid
; /* Default uid for reserved blocks */
906 __le16 s_def_resgid
; /* Default gid for reserved blocks */
908 * These fields are for EXT4_DYNAMIC_REV superblocks only.
910 * Note: the difference between the compatible feature set and
911 * the incompatible feature set is that if there is a bit set
912 * in the incompatible feature set that the kernel doesn't
913 * know about, it should refuse to mount the filesystem.
915 * e2fsck's requirements are more strict; if it doesn't know
916 * about a feature in either the compatible or incompatible
917 * feature set, it must abort and not try to meddle with
918 * things it doesn't understand...
920 __le32 s_first_ino
; /* First non-reserved inode */
921 __le16 s_inode_size
; /* size of inode structure */
922 __le16 s_block_group_nr
; /* block group # of this superblock */
923 __le32 s_feature_compat
; /* compatible feature set */
924 /*60*/ __le32 s_feature_incompat
; /* incompatible feature set */
925 __le32 s_feature_ro_compat
; /* readonly-compatible feature set */
926 /*68*/ __u8 s_uuid
[16]; /* 128-bit uuid for volume */
927 /*78*/ char s_volume_name
[16]; /* volume name */
928 /*88*/ char s_last_mounted
[64]; /* directory where last mounted */
929 /*C8*/ __le32 s_algorithm_usage_bitmap
; /* For compression */
931 * Performance hints. Directory preallocation should only
932 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
934 __u8 s_prealloc_blocks
; /* Nr of blocks to try to preallocate*/
935 __u8 s_prealloc_dir_blocks
; /* Nr to preallocate for dirs */
936 __le16 s_reserved_gdt_blocks
; /* Per group desc for online growth */
938 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
940 /*D0*/ __u8 s_journal_uuid
[16]; /* uuid of journal superblock */
941 /*E0*/ __le32 s_journal_inum
; /* inode number of journal file */
942 __le32 s_journal_dev
; /* device number of journal file */
943 __le32 s_last_orphan
; /* start of list of inodes to delete */
944 __le32 s_hash_seed
[4]; /* HTREE hash seed */
945 __u8 s_def_hash_version
; /* Default hash version to use */
946 __u8 s_reserved_char_pad
;
947 __le16 s_desc_size
; /* size of group descriptor */
948 /*100*/ __le32 s_default_mount_opts
;
949 __le32 s_first_meta_bg
; /* First metablock block group */
950 __le32 s_mkfs_time
; /* When the filesystem was created */
951 __le32 s_jnl_blocks
[17]; /* Backup of the journal inode */
952 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
953 /*150*/ __le32 s_blocks_count_hi
; /* Blocks count */
954 __le32 s_r_blocks_count_hi
; /* Reserved blocks count */
955 __le32 s_free_blocks_count_hi
; /* Free blocks count */
956 __le16 s_min_extra_isize
; /* All inodes have at least # bytes */
957 __le16 s_want_extra_isize
; /* New inodes should reserve # bytes */
958 __le32 s_flags
; /* Miscellaneous flags */
959 __le16 s_raid_stride
; /* RAID stride */
960 __le16 s_mmp_interval
; /* # seconds to wait in MMP checking */
961 __le64 s_mmp_block
; /* Block for multi-mount protection */
962 __le32 s_raid_stripe_width
; /* blocks on all data disks (N*stride)*/
963 __u8 s_log_groups_per_flex
; /* FLEX_BG group size */
964 __u8 s_reserved_char_pad2
;
965 __le16 s_reserved_pad
;
966 __le64 s_kbytes_written
; /* nr of lifetime kilobytes written */
967 __u32 s_reserved
[160]; /* Padding to the end of the block */
973 * run-time mount flags
975 #define EXT4_MF_MNTDIR_SAMPLED 0x0001
976 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
979 * fourth extended-fs super-block data in memory
981 struct ext4_sb_info
{
982 unsigned long s_desc_size
; /* Size of a group descriptor in bytes */
983 unsigned long s_inodes_per_block
;/* Number of inodes per block */
984 unsigned long s_blocks_per_group
;/* Number of blocks in a group */
985 unsigned long s_inodes_per_group
;/* Number of inodes in a group */
986 unsigned long s_itb_per_group
; /* Number of inode table blocks per group */
987 unsigned long s_gdb_count
; /* Number of group descriptor blocks */
988 unsigned long s_desc_per_block
; /* Number of group descriptors per block */
989 ext4_group_t s_groups_count
; /* Number of groups in the fs */
990 ext4_group_t s_blockfile_groups
;/* Groups acceptable for non-extent files */
991 unsigned long s_overhead_last
; /* Last calculated overhead */
992 unsigned long s_blocks_last
; /* Last seen block count */
993 loff_t s_bitmap_maxbytes
; /* max bytes for bitmap files */
994 struct buffer_head
* s_sbh
; /* Buffer containing the super block */
995 struct ext4_super_block
*s_es
; /* Pointer to the super block in the buffer */
996 struct buffer_head
**s_group_desc
;
997 unsigned int s_mount_opt
;
998 unsigned int s_mount_flags
;
999 ext4_fsblk_t s_sb_block
;
1002 unsigned short s_mount_state
;
1003 unsigned short s_pad
;
1004 int s_addr_per_block_bits
;
1005 int s_desc_per_block_bits
;
1008 unsigned int s_inode_readahead_blks
;
1009 unsigned int s_inode_goal
;
1010 spinlock_t s_next_gen_lock
;
1011 u32 s_next_generation
;
1013 int s_def_hash_version
;
1014 int s_hash_unsigned
; /* 3 if hash should be signed, 0 if not */
1015 struct percpu_counter s_freeblocks_counter
;
1016 struct percpu_counter s_freeinodes_counter
;
1017 struct percpu_counter s_dirs_counter
;
1018 struct percpu_counter s_dirtyblocks_counter
;
1019 struct blockgroup_lock
*s_blockgroup_lock
;
1020 struct proc_dir_entry
*s_proc
;
1021 struct kobject s_kobj
;
1022 struct completion s_kobj_unregister
;
1025 struct inode
*s_journal_inode
;
1026 struct journal_s
*s_journal
;
1027 struct list_head s_orphan
;
1028 struct mutex s_orphan_lock
;
1029 struct mutex s_resize_lock
;
1030 unsigned long s_commit_interval
;
1031 u32 s_max_batch_time
;
1032 u32 s_min_batch_time
;
1033 struct block_device
*journal_bdev
;
1034 #ifdef CONFIG_JBD2_DEBUG
1035 struct timer_list turn_ro_timer
; /* For turning read-only (crash simulation) */
1036 wait_queue_head_t ro_wait_queue
; /* For people waiting for the fs to go read-only */
1039 char *s_qf_names
[MAXQUOTAS
]; /* Names of quota files with journalled quota */
1040 int s_jquota_fmt
; /* Format of quota to use */
1042 unsigned int s_want_extra_isize
; /* New inodes should reserve # bytes */
1043 struct rb_root system_blks
;
1045 #ifdef EXTENTS_STATS
1046 /* ext4 extents stats */
1047 unsigned long s_ext_min
;
1048 unsigned long s_ext_max
;
1049 unsigned long s_depth_max
;
1050 spinlock_t s_ext_stats_lock
;
1051 unsigned long s_ext_blocks
;
1052 unsigned long s_ext_extents
;
1055 /* for buddy allocator */
1056 struct ext4_group_info
***s_group_info
;
1057 struct inode
*s_buddy_cache
;
1058 long s_blocks_reserved
;
1059 spinlock_t s_reserve_lock
;
1060 spinlock_t s_md_lock
;
1061 tid_t s_last_transaction
;
1062 unsigned short *s_mb_offsets
;
1063 unsigned int *s_mb_maxs
;
1066 unsigned long s_stripe
;
1067 unsigned int s_mb_stream_request
;
1068 unsigned int s_mb_max_to_scan
;
1069 unsigned int s_mb_min_to_scan
;
1070 unsigned int s_mb_stats
;
1071 unsigned int s_mb_order2_reqs
;
1072 unsigned int s_mb_group_prealloc
;
1073 unsigned int s_max_writeback_mb_bump
;
1074 /* where last allocation was done - for stream allocation */
1075 unsigned long s_mb_last_group
;
1076 unsigned long s_mb_last_start
;
1078 /* stats for buddy allocator */
1079 spinlock_t s_mb_pa_lock
;
1080 atomic_t s_bal_reqs
; /* number of reqs with len > 1 */
1081 atomic_t s_bal_success
; /* we found long enough chunks */
1082 atomic_t s_bal_allocated
; /* in blocks */
1083 atomic_t s_bal_ex_scanned
; /* total extents scanned */
1084 atomic_t s_bal_goals
; /* goal hits */
1085 atomic_t s_bal_breaks
; /* too long searches */
1086 atomic_t s_bal_2orders
; /* 2^order hits */
1087 spinlock_t s_bal_lock
;
1088 unsigned long s_mb_buddies_generated
;
1089 unsigned long long s_mb_generation_time
;
1090 atomic_t s_mb_lost_chunks
;
1091 atomic_t s_mb_preallocated
;
1092 atomic_t s_mb_discarded
;
1093 atomic_t s_lock_busy
;
1095 /* locality groups */
1096 struct ext4_locality_group
*s_locality_groups
;
1098 /* for write statistics */
1099 unsigned long s_sectors_written_start
;
1100 u64 s_kbytes_written
;
1102 unsigned int s_log_groups_per_flex
;
1103 struct flex_groups
*s_flex_groups
;
1105 /* workqueue for dio unwritten */
1106 struct workqueue_struct
*dio_unwritten_wq
;
1109 static inline struct ext4_sb_info
*EXT4_SB(struct super_block
*sb
)
1111 return sb
->s_fs_info
;
1113 static inline struct ext4_inode_info
*EXT4_I(struct inode
*inode
)
1115 return container_of(inode
, struct ext4_inode_info
, vfs_inode
);
1118 static inline struct timespec
ext4_current_time(struct inode
*inode
)
1120 return (inode
->i_sb
->s_time_gran
< NSEC_PER_SEC
) ?
1121 current_fs_time(inode
->i_sb
) : CURRENT_TIME_SEC
;
1124 static inline int ext4_valid_inum(struct super_block
*sb
, unsigned long ino
)
1126 return ino
== EXT4_ROOT_INO
||
1127 ino
== EXT4_JOURNAL_INO
||
1128 ino
== EXT4_RESIZE_INO
||
1129 (ino
>= EXT4_FIRST_INO(sb
) &&
1130 ino
<= le32_to_cpu(EXT4_SB(sb
)->s_es
->s_inodes_count
));
1134 * Inode dynamic state flags
1137 EXT4_STATE_JDATA
, /* journaled data exists */
1138 EXT4_STATE_NEW
, /* inode is newly created */
1139 EXT4_STATE_XATTR
, /* has in-inode xattrs */
1140 EXT4_STATE_NO_EXPAND
, /* No space for expansion */
1141 EXT4_STATE_DA_ALLOC_CLOSE
, /* Alloc DA blks on close */
1142 EXT4_STATE_EXT_MIGRATE
, /* Inode is migrating */
1143 EXT4_STATE_DIO_UNWRITTEN
, /* need convert on dio done*/
1144 EXT4_STATE_NEWENTRY
, /* File just added to dir */
1147 #define EXT4_INODE_BIT_FNS(name, field) \
1148 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1150 return test_bit(bit, &EXT4_I(inode)->i_##field); \
1152 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1154 set_bit(bit, &EXT4_I(inode)->i_##field); \
1156 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1158 clear_bit(bit, &EXT4_I(inode)->i_##field); \
1161 EXT4_INODE_BIT_FNS(flag
, flags
)
1162 EXT4_INODE_BIT_FNS(state
, state_flags
)
1164 /* Assume that user mode programs are passing in an ext4fs superblock, not
1165 * a kernel struct super_block. This will allow us to call the feature-test
1166 * macros from user land. */
1167 #define EXT4_SB(sb) (sb)
1170 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1173 * Codes for operating systems
1175 #define EXT4_OS_LINUX 0
1176 #define EXT4_OS_HURD 1
1177 #define EXT4_OS_MASIX 2
1178 #define EXT4_OS_FREEBSD 3
1179 #define EXT4_OS_LITES 4
1184 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1185 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1187 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1188 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1190 #define EXT4_GOOD_OLD_INODE_SIZE 128
1193 * Feature set definitions
1196 #define EXT4_HAS_COMPAT_FEATURE(sb,mask) \
1197 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1198 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \
1199 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1200 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \
1201 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1202 #define EXT4_SET_COMPAT_FEATURE(sb,mask) \
1203 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1204 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \
1205 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1206 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \
1207 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1208 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \
1209 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1210 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
1211 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1212 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \
1213 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1215 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1216 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1217 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1218 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1219 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1220 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1222 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1223 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1224 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1225 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1226 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1227 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1228 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1230 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1231 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1232 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1233 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1234 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1235 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1236 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1237 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1238 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1240 #define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR
1241 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1242 EXT4_FEATURE_INCOMPAT_RECOVER| \
1243 EXT4_FEATURE_INCOMPAT_META_BG| \
1244 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1245 EXT4_FEATURE_INCOMPAT_64BIT| \
1246 EXT4_FEATURE_INCOMPAT_FLEX_BG)
1247 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1248 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1249 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1250 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1251 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1252 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1253 EXT4_FEATURE_RO_COMPAT_HUGE_FILE)
1256 * Default values for user and/or group using reserved blocks
1258 #define EXT4_DEF_RESUID 0
1259 #define EXT4_DEF_RESGID 0
1261 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
1264 * Default mount options
1266 #define EXT4_DEFM_DEBUG 0x0001
1267 #define EXT4_DEFM_BSDGROUPS 0x0002
1268 #define EXT4_DEFM_XATTR_USER 0x0004
1269 #define EXT4_DEFM_ACL 0x0008
1270 #define EXT4_DEFM_UID16 0x0010
1271 #define EXT4_DEFM_JMODE 0x0060
1272 #define EXT4_DEFM_JMODE_DATA 0x0020
1273 #define EXT4_DEFM_JMODE_ORDERED 0x0040
1274 #define EXT4_DEFM_JMODE_WBACK 0x0060
1277 * Default journal batch times
1279 #define EXT4_DEF_MIN_BATCH_TIME 0
1280 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1283 * Minimum number of groups in a flexgroup before we separate out
1284 * directories into the first block group of a flexgroup
1286 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1289 * Structure of a directory entry
1291 #define EXT4_NAME_LEN 255
1293 struct ext4_dir_entry
{
1294 __le32 inode
; /* Inode number */
1295 __le16 rec_len
; /* Directory entry length */
1296 __le16 name_len
; /* Name length */
1297 char name
[EXT4_NAME_LEN
]; /* File name */
1301 * The new version of the directory entry. Since EXT4 structures are
1302 * stored in intel byte order, and the name_len field could never be
1303 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1306 struct ext4_dir_entry_2
{
1307 __le32 inode
; /* Inode number */
1308 __le16 rec_len
; /* Directory entry length */
1309 __u8 name_len
; /* Name length */
1311 char name
[EXT4_NAME_LEN
]; /* File name */
1315 * Ext4 directory file types. Only the low 3 bits are used. The
1316 * other bits are reserved for now.
1318 #define EXT4_FT_UNKNOWN 0
1319 #define EXT4_FT_REG_FILE 1
1320 #define EXT4_FT_DIR 2
1321 #define EXT4_FT_CHRDEV 3
1322 #define EXT4_FT_BLKDEV 4
1323 #define EXT4_FT_FIFO 5
1324 #define EXT4_FT_SOCK 6
1325 #define EXT4_FT_SYMLINK 7
1327 #define EXT4_FT_MAX 8
1330 * EXT4_DIR_PAD defines the directory entries boundaries
1332 * NOTE: It must be a multiple of 4
1334 #define EXT4_DIR_PAD 4
1335 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1336 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1338 #define EXT4_MAX_REC_LEN ((1<<16)-1)
1341 * Hash Tree Directory indexing
1342 * (c) Daniel Phillips, 2001
1345 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1346 EXT4_FEATURE_COMPAT_DIR_INDEX) && \
1347 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
1348 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1349 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1351 /* Legal values for the dx_root hash_version field: */
1353 #define DX_HASH_LEGACY 0
1354 #define DX_HASH_HALF_MD4 1
1355 #define DX_HASH_TEA 2
1356 #define DX_HASH_LEGACY_UNSIGNED 3
1357 #define DX_HASH_HALF_MD4_UNSIGNED 4
1358 #define DX_HASH_TEA_UNSIGNED 5
1362 /* hash info structure used by the directory hash */
1371 #define EXT4_HTREE_EOF 0x7fffffff
1374 * Control parameters used by ext4_htree_next_block
1376 #define HASH_NB_ALWAYS 1
1380 * Describe an inode's exact location on disk and in memory
1384 struct buffer_head
*bh
;
1385 unsigned long offset
;
1386 ext4_group_t block_group
;
1389 static inline struct ext4_inode
*ext4_raw_inode(struct ext4_iloc
*iloc
)
1391 return (struct ext4_inode
*) (iloc
->bh
->b_data
+ iloc
->offset
);
1395 * This structure is stuffed into the struct file's private_data field
1396 * for directories. It is where we put information so that we can do
1397 * readdir operations in hash tree order.
1399 struct dir_private_info
{
1400 struct rb_root root
;
1401 struct rb_node
*curr_node
;
1402 struct fname
*extra_fname
;
1405 __u32 curr_minor_hash
;
1409 /* calculate the first block number of the group */
1410 static inline ext4_fsblk_t
1411 ext4_group_first_block_no(struct super_block
*sb
, ext4_group_t group_no
)
1413 return group_no
* (ext4_fsblk_t
)EXT4_BLOCKS_PER_GROUP(sb
) +
1414 le32_to_cpu(EXT4_SB(sb
)->s_es
->s_first_data_block
);
1418 * Special error return code only used by dx_probe() and its callers.
1420 #define ERR_BAD_DX_DIR -75000
1422 void ext4_get_group_no_and_offset(struct super_block
*sb
, ext4_fsblk_t blocknr
,
1423 ext4_group_t
*blockgrpp
, ext4_grpblk_t
*offsetp
);
1425 extern struct proc_dir_entry
*ext4_proc_root
;
1428 * Function prototypes
1432 * Ok, these declarations are also in <linux/kernel.h> but none of the
1433 * ext4 source programs needs to include it so they are duplicated here.
1435 # define NORET_TYPE /**/
1436 # define ATTRIB_NORET __attribute__((noreturn))
1437 # define NORET_AND noreturn,
1440 extern unsigned int ext4_count_free(struct buffer_head
*, unsigned);
1443 extern unsigned int ext4_block_group(struct super_block
*sb
,
1444 ext4_fsblk_t blocknr
);
1445 extern ext4_grpblk_t
ext4_block_group_offset(struct super_block
*sb
,
1446 ext4_fsblk_t blocknr
);
1447 extern int ext4_bg_has_super(struct super_block
*sb
, ext4_group_t group
);
1448 extern unsigned long ext4_bg_num_gdb(struct super_block
*sb
,
1449 ext4_group_t group
);
1450 extern ext4_fsblk_t
ext4_new_meta_blocks(handle_t
*handle
, struct inode
*inode
,
1451 ext4_fsblk_t goal
, unsigned long *count
, int *errp
);
1452 extern int ext4_claim_free_blocks(struct ext4_sb_info
*sbi
, s64 nblocks
);
1453 extern int ext4_has_free_blocks(struct ext4_sb_info
*sbi
, s64 nblocks
);
1454 extern void ext4_free_blocks(handle_t
*handle
, struct inode
*inode
,
1455 ext4_fsblk_t block
, unsigned long count
, int metadata
);
1456 extern void ext4_add_groupblocks(handle_t
*handle
, struct super_block
*sb
,
1457 ext4_fsblk_t block
, unsigned long count
);
1458 extern ext4_fsblk_t
ext4_count_free_blocks(struct super_block
*);
1459 extern void ext4_check_blocks_bitmap(struct super_block
*);
1460 extern struct ext4_group_desc
* ext4_get_group_desc(struct super_block
* sb
,
1461 ext4_group_t block_group
,
1462 struct buffer_head
** bh
);
1463 extern int ext4_should_retry_alloc(struct super_block
*sb
, int *retries
);
1464 struct buffer_head
*ext4_read_block_bitmap(struct super_block
*sb
,
1465 ext4_group_t block_group
);
1466 extern unsigned ext4_init_block_bitmap(struct super_block
*sb
,
1467 struct buffer_head
*bh
,
1469 struct ext4_group_desc
*desc
);
1470 #define ext4_free_blocks_after_init(sb, group, desc) \
1471 ext4_init_block_bitmap(sb, NULL, group, desc)
1474 extern int ext4_check_dir_entry(const char *, struct inode
*,
1475 struct ext4_dir_entry_2
*,
1476 struct buffer_head
*, unsigned int);
1477 extern int ext4_htree_store_dirent(struct file
*dir_file
, __u32 hash
,
1479 struct ext4_dir_entry_2
*dirent
);
1480 extern void ext4_htree_free_dir_info(struct dir_private_info
*p
);
1483 extern int ext4_sync_file(struct file
*, struct dentry
*, int);
1486 extern int ext4fs_dirhash(const char *name
, int len
, struct
1487 dx_hash_info
*hinfo
);
1490 extern struct inode
*ext4_new_inode(handle_t
*, struct inode
*, int,
1491 const struct qstr
*qstr
, __u32 goal
);
1492 extern void ext4_free_inode(handle_t
*, struct inode
*);
1493 extern struct inode
* ext4_orphan_get(struct super_block
*, unsigned long);
1494 extern unsigned long ext4_count_free_inodes(struct super_block
*);
1495 extern unsigned long ext4_count_dirs(struct super_block
*);
1496 extern void ext4_check_inodes_bitmap(struct super_block
*);
1497 extern unsigned ext4_init_inode_bitmap(struct super_block
*sb
,
1498 struct buffer_head
*bh
,
1500 struct ext4_group_desc
*desc
);
1501 extern void mark_bitmap_end(int start_bit
, int end_bit
, char *bitmap
);
1504 extern long ext4_mb_stats
;
1505 extern long ext4_mb_max_to_scan
;
1506 extern int ext4_mb_init(struct super_block
*, int);
1507 extern int ext4_mb_release(struct super_block
*);
1508 extern ext4_fsblk_t
ext4_mb_new_blocks(handle_t
*,
1509 struct ext4_allocation_request
*, int *);
1510 extern int ext4_mb_reserve_blocks(struct super_block
*, int);
1511 extern void ext4_discard_preallocations(struct inode
*);
1512 extern int __init
init_ext4_mballoc(void);
1513 extern void exit_ext4_mballoc(void);
1514 extern void ext4_mb_free_blocks(handle_t
*, struct inode
*,
1515 ext4_fsblk_t
, unsigned long, int, unsigned long *);
1516 extern int ext4_mb_add_groupinfo(struct super_block
*sb
,
1517 ext4_group_t i
, struct ext4_group_desc
*desc
);
1518 extern int ext4_mb_get_buddy_cache_lock(struct super_block
*, ext4_group_t
);
1519 extern void ext4_mb_put_buddy_cache_lock(struct super_block
*,
1522 int ext4_forget(handle_t
*handle
, int is_metadata
, struct inode
*inode
,
1523 struct buffer_head
*bh
, ext4_fsblk_t blocknr
);
1524 struct buffer_head
*ext4_getblk(handle_t
*, struct inode
*,
1525 ext4_lblk_t
, int, int *);
1526 struct buffer_head
*ext4_bread(handle_t
*, struct inode
*,
1527 ext4_lblk_t
, int, int *);
1528 int ext4_get_block(struct inode
*inode
, sector_t iblock
,
1529 struct buffer_head
*bh_result
, int create
);
1531 extern struct inode
*ext4_iget(struct super_block
*, unsigned long);
1532 extern int ext4_write_inode(struct inode
*, int);
1533 extern int ext4_setattr(struct dentry
*, struct iattr
*);
1534 extern int ext4_getattr(struct vfsmount
*mnt
, struct dentry
*dentry
,
1535 struct kstat
*stat
);
1536 extern void ext4_delete_inode(struct inode
*);
1537 extern int ext4_sync_inode(handle_t
*, struct inode
*);
1538 extern void ext4_dirty_inode(struct inode
*);
1539 extern int ext4_change_inode_journal_flag(struct inode
*, int);
1540 extern int ext4_get_inode_loc(struct inode
*, struct ext4_iloc
*);
1541 extern int ext4_can_truncate(struct inode
*inode
);
1542 extern void ext4_truncate(struct inode
*);
1543 extern int ext4_truncate_restart_trans(handle_t
*, struct inode
*, int nblocks
);
1544 extern void ext4_set_inode_flags(struct inode
*);
1545 extern void ext4_get_inode_flags(struct ext4_inode_info
*);
1546 extern int ext4_alloc_da_blocks(struct inode
*inode
);
1547 extern void ext4_set_aops(struct inode
*inode
);
1548 extern int ext4_writepage_trans_blocks(struct inode
*);
1549 extern int ext4_meta_trans_blocks(struct inode
*, int nrblocks
, int idxblocks
);
1550 extern int ext4_chunk_trans_blocks(struct inode
*, int nrblocks
);
1551 extern int ext4_block_truncate_page(handle_t
*handle
,
1552 struct address_space
*mapping
, loff_t from
);
1553 extern int ext4_page_mkwrite(struct vm_area_struct
*vma
, struct vm_fault
*vmf
);
1554 extern qsize_t
*ext4_get_reserved_space(struct inode
*inode
);
1555 extern int flush_aio_dio_completed_IO(struct inode
*inode
);
1556 extern void ext4_da_update_reserve_space(struct inode
*inode
,
1557 int used
, int quota_claim
);
1559 extern long ext4_ioctl(struct file
*, unsigned int, unsigned long);
1560 extern long ext4_compat_ioctl(struct file
*, unsigned int, unsigned long);
1563 extern int ext4_ext_migrate(struct inode
*);
1566 extern unsigned int ext4_rec_len_from_disk(__le16 dlen
, unsigned blocksize
);
1567 extern __le16
ext4_rec_len_to_disk(unsigned len
, unsigned blocksize
);
1568 extern int ext4_orphan_add(handle_t
*, struct inode
*);
1569 extern int ext4_orphan_del(handle_t
*, struct inode
*);
1570 extern int ext4_htree_fill_tree(struct file
*dir_file
, __u32 start_hash
,
1571 __u32 start_minor_hash
, __u32
*next_hash
);
1574 extern int ext4_group_add(struct super_block
*sb
,
1575 struct ext4_new_group_data
*input
);
1576 extern int ext4_group_extend(struct super_block
*sb
,
1577 struct ext4_super_block
*es
,
1578 ext4_fsblk_t n_blocks_count
);
1581 extern void ext4_error(struct super_block
*, const char *, const char *, ...)
1582 __attribute__ ((format (printf
, 3, 4)));
1583 extern void __ext4_std_error(struct super_block
*, const char *, int);
1584 extern void ext4_abort(struct super_block
*, const char *, const char *, ...)
1585 __attribute__ ((format (printf
, 3, 4)));
1586 extern void ext4_warning(struct super_block
*, const char *, const char *, ...)
1587 __attribute__ ((format (printf
, 3, 4)));
1588 extern void ext4_msg(struct super_block
*, const char *, const char *, ...)
1589 __attribute__ ((format (printf
, 3, 4)));
1590 extern void ext4_grp_locked_error(struct super_block
*, ext4_group_t
,
1591 const char *, const char *, ...)
1592 __attribute__ ((format (printf
, 4, 5)));
1593 extern void ext4_update_dynamic_rev(struct super_block
*sb
);
1594 extern int ext4_update_compat_feature(handle_t
*handle
, struct super_block
*sb
,
1596 extern int ext4_update_rocompat_feature(handle_t
*handle
,
1597 struct super_block
*sb
, __u32 rocompat
);
1598 extern int ext4_update_incompat_feature(handle_t
*handle
,
1599 struct super_block
*sb
, __u32 incompat
);
1600 extern ext4_fsblk_t
ext4_block_bitmap(struct super_block
*sb
,
1601 struct ext4_group_desc
*bg
);
1602 extern ext4_fsblk_t
ext4_inode_bitmap(struct super_block
*sb
,
1603 struct ext4_group_desc
*bg
);
1604 extern ext4_fsblk_t
ext4_inode_table(struct super_block
*sb
,
1605 struct ext4_group_desc
*bg
);
1606 extern __u32
ext4_free_blks_count(struct super_block
*sb
,
1607 struct ext4_group_desc
*bg
);
1608 extern __u32
ext4_free_inodes_count(struct super_block
*sb
,
1609 struct ext4_group_desc
*bg
);
1610 extern __u32
ext4_used_dirs_count(struct super_block
*sb
,
1611 struct ext4_group_desc
*bg
);
1612 extern __u32
ext4_itable_unused_count(struct super_block
*sb
,
1613 struct ext4_group_desc
*bg
);
1614 extern void ext4_block_bitmap_set(struct super_block
*sb
,
1615 struct ext4_group_desc
*bg
, ext4_fsblk_t blk
);
1616 extern void ext4_inode_bitmap_set(struct super_block
*sb
,
1617 struct ext4_group_desc
*bg
, ext4_fsblk_t blk
);
1618 extern void ext4_inode_table_set(struct super_block
*sb
,
1619 struct ext4_group_desc
*bg
, ext4_fsblk_t blk
);
1620 extern void ext4_free_blks_set(struct super_block
*sb
,
1621 struct ext4_group_desc
*bg
, __u32 count
);
1622 extern void ext4_free_inodes_set(struct super_block
*sb
,
1623 struct ext4_group_desc
*bg
, __u32 count
);
1624 extern void ext4_used_dirs_set(struct super_block
*sb
,
1625 struct ext4_group_desc
*bg
, __u32 count
);
1626 extern void ext4_itable_unused_set(struct super_block
*sb
,
1627 struct ext4_group_desc
*bg
, __u32 count
);
1628 extern __le16
ext4_group_desc_csum(struct ext4_sb_info
*sbi
, __u32 group
,
1629 struct ext4_group_desc
*gdp
);
1630 extern int ext4_group_desc_csum_verify(struct ext4_sb_info
*sbi
, __u32 group
,
1631 struct ext4_group_desc
*gdp
);
1633 static inline ext4_fsblk_t
ext4_blocks_count(struct ext4_super_block
*es
)
1635 return ((ext4_fsblk_t
)le32_to_cpu(es
->s_blocks_count_hi
) << 32) |
1636 le32_to_cpu(es
->s_blocks_count_lo
);
1639 static inline ext4_fsblk_t
ext4_r_blocks_count(struct ext4_super_block
*es
)
1641 return ((ext4_fsblk_t
)le32_to_cpu(es
->s_r_blocks_count_hi
) << 32) |
1642 le32_to_cpu(es
->s_r_blocks_count_lo
);
1645 static inline ext4_fsblk_t
ext4_free_blocks_count(struct ext4_super_block
*es
)
1647 return ((ext4_fsblk_t
)le32_to_cpu(es
->s_free_blocks_count_hi
) << 32) |
1648 le32_to_cpu(es
->s_free_blocks_count_lo
);
1651 static inline void ext4_blocks_count_set(struct ext4_super_block
*es
,
1654 es
->s_blocks_count_lo
= cpu_to_le32((u32
)blk
);
1655 es
->s_blocks_count_hi
= cpu_to_le32(blk
>> 32);
1658 static inline void ext4_free_blocks_count_set(struct ext4_super_block
*es
,
1661 es
->s_free_blocks_count_lo
= cpu_to_le32((u32
)blk
);
1662 es
->s_free_blocks_count_hi
= cpu_to_le32(blk
>> 32);
1665 static inline void ext4_r_blocks_count_set(struct ext4_super_block
*es
,
1668 es
->s_r_blocks_count_lo
= cpu_to_le32((u32
)blk
);
1669 es
->s_r_blocks_count_hi
= cpu_to_le32(blk
>> 32);
1672 static inline loff_t
ext4_isize(struct ext4_inode
*raw_inode
)
1674 if (S_ISREG(le16_to_cpu(raw_inode
->i_mode
)))
1675 return ((loff_t
)le32_to_cpu(raw_inode
->i_size_high
) << 32) |
1676 le32_to_cpu(raw_inode
->i_size_lo
);
1678 return (loff_t
) le32_to_cpu(raw_inode
->i_size_lo
);
1681 static inline void ext4_isize_set(struct ext4_inode
*raw_inode
, loff_t i_size
)
1683 raw_inode
->i_size_lo
= cpu_to_le32(i_size
);
1684 raw_inode
->i_size_high
= cpu_to_le32(i_size
>> 32);
1688 struct ext4_group_info
*ext4_get_group_info(struct super_block
*sb
,
1691 struct ext4_group_info
***grp_info
;
1692 long indexv
, indexh
;
1693 grp_info
= EXT4_SB(sb
)->s_group_info
;
1694 indexv
= group
>> (EXT4_DESC_PER_BLOCK_BITS(sb
));
1695 indexh
= group
& ((EXT4_DESC_PER_BLOCK(sb
)) - 1);
1696 return grp_info
[indexv
][indexh
];
1700 * Reading s_groups_count requires using smp_rmb() afterwards. See
1701 * the locking protocol documented in the comments of ext4_group_add()
1704 static inline ext4_group_t
ext4_get_groups_count(struct super_block
*sb
)
1706 ext4_group_t ngroups
= EXT4_SB(sb
)->s_groups_count
;
1712 static inline ext4_group_t
ext4_flex_group(struct ext4_sb_info
*sbi
,
1713 ext4_group_t block_group
)
1715 return block_group
>> sbi
->s_log_groups_per_flex
;
1718 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info
*sbi
)
1720 return 1 << sbi
->s_log_groups_per_flex
;
1723 #define ext4_std_error(sb, errno) \
1726 __ext4_std_error((sb), __func__, (errno)); \
1730 /* Each CPU can accumulate percpu_counter_batch blocks in their local
1731 * counters. So we need to make sure we have free blocks more
1732 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
1734 #define EXT4_FREEBLOCKS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
1736 #define EXT4_FREEBLOCKS_WATERMARK 0
1739 static inline void ext4_update_i_disksize(struct inode
*inode
, loff_t newsize
)
1742 * XXX: replace with spinlock if seen contended -bzzz
1744 down_write(&EXT4_I(inode
)->i_data_sem
);
1745 if (newsize
> EXT4_I(inode
)->i_disksize
)
1746 EXT4_I(inode
)->i_disksize
= newsize
;
1747 up_write(&EXT4_I(inode
)->i_data_sem
);
1751 struct ext4_group_info
{
1752 unsigned long bb_state
;
1753 struct rb_root bb_free_root
;
1754 ext4_grpblk_t bb_first_free
; /* first free block */
1755 ext4_grpblk_t bb_free
; /* total free blocks */
1756 ext4_grpblk_t bb_fragments
; /* nr of freespace fragments */
1757 ext4_grpblk_t bb_largest_free_order
;/* order of largest frag in BG */
1758 struct list_head bb_prealloc_list
;
1762 struct rw_semaphore alloc_sem
;
1763 ext4_grpblk_t bb_counters
[]; /* Nr of free power-of-two-block
1764 * regions, index is order.
1765 * bb_counters[3] = 5 means
1766 * 5 free 8-block regions. */
1769 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
1771 #define EXT4_MB_GRP_NEED_INIT(grp) \
1772 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
1774 #define EXT4_MAX_CONTENTION 8
1775 #define EXT4_CONTENTION_THRESHOLD 2
1777 static inline spinlock_t
*ext4_group_lock_ptr(struct super_block
*sb
,
1780 return bgl_lock_ptr(EXT4_SB(sb
)->s_blockgroup_lock
, group
);
1784 * Returns true if the filesystem is busy enough that attempts to
1785 * access the block group locks has run into contention.
1787 static inline int ext4_fs_is_busy(struct ext4_sb_info
*sbi
)
1789 return (atomic_read(&sbi
->s_lock_busy
) > EXT4_CONTENTION_THRESHOLD
);
1792 static inline void ext4_lock_group(struct super_block
*sb
, ext4_group_t group
)
1794 spinlock_t
*lock
= ext4_group_lock_ptr(sb
, group
);
1795 if (spin_trylock(lock
))
1797 * We're able to grab the lock right away, so drop the
1798 * lock contention counter.
1800 atomic_add_unless(&EXT4_SB(sb
)->s_lock_busy
, -1, 0);
1803 * The lock is busy, so bump the contention counter,
1804 * and then wait on the spin lock.
1806 atomic_add_unless(&EXT4_SB(sb
)->s_lock_busy
, 1,
1807 EXT4_MAX_CONTENTION
);
1812 static inline void ext4_unlock_group(struct super_block
*sb
,
1815 spin_unlock(ext4_group_lock_ptr(sb
, group
));
1819 * Inodes and files operations
1823 extern const struct file_operations ext4_dir_operations
;
1826 extern const struct inode_operations ext4_file_inode_operations
;
1827 extern const struct file_operations ext4_file_operations
;
1830 extern const struct inode_operations ext4_dir_inode_operations
;
1831 extern const struct inode_operations ext4_special_inode_operations
;
1832 extern struct dentry
*ext4_get_parent(struct dentry
*child
);
1835 extern const struct inode_operations ext4_symlink_inode_operations
;
1836 extern const struct inode_operations ext4_fast_symlink_inode_operations
;
1838 /* block_validity */
1839 extern void ext4_release_system_zone(struct super_block
*sb
);
1840 extern int ext4_setup_system_zone(struct super_block
*sb
);
1841 extern int __init
init_ext4_system_zone(void);
1842 extern void exit_ext4_system_zone(void);
1843 extern int ext4_data_block_valid(struct ext4_sb_info
*sbi
,
1844 ext4_fsblk_t start_blk
,
1845 unsigned int count
);
1848 extern int ext4_ext_tree_init(handle_t
*handle
, struct inode
*);
1849 extern int ext4_ext_writepage_trans_blocks(struct inode
*, int);
1850 extern int ext4_ext_index_trans_blocks(struct inode
*inode
, int nrblocks
,
1852 extern int ext4_ext_get_blocks(handle_t
*handle
, struct inode
*inode
,
1853 ext4_lblk_t iblock
, unsigned int max_blocks
,
1854 struct buffer_head
*bh_result
, int flags
);
1855 extern void ext4_ext_truncate(struct inode
*);
1856 extern void ext4_ext_init(struct super_block
*);
1857 extern void ext4_ext_release(struct super_block
*);
1858 extern long ext4_fallocate(struct inode
*inode
, int mode
, loff_t offset
,
1860 extern int ext4_convert_unwritten_extents(struct inode
*inode
, loff_t offset
,
1862 extern int ext4_get_blocks(handle_t
*handle
, struct inode
*inode
,
1863 sector_t block
, unsigned int max_blocks
,
1864 struct buffer_head
*bh
, int flags
);
1865 extern int ext4_fiemap(struct inode
*inode
, struct fiemap_extent_info
*fieinfo
,
1866 __u64 start
, __u64 len
);
1868 extern int ext4_move_extents(struct file
*o_filp
, struct file
*d_filp
,
1869 __u64 start_orig
, __u64 start_donor
,
1870 __u64 len
, __u64
*moved_len
);
1874 * Add new method to test wether block and inode bitmaps are properly
1875 * initialized. With uninit_bg reading the block from disk is not enough
1876 * to mark the bitmap uptodate. We need to also zero-out the bitmap
1878 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
1880 static inline int bitmap_uptodate(struct buffer_head
*bh
)
1882 return (buffer_uptodate(bh
) &&
1883 test_bit(BH_BITMAP_UPTODATE
, &(bh
)->b_state
));
1885 static inline void set_bitmap_uptodate(struct buffer_head
*bh
)
1887 set_bit(BH_BITMAP_UPTODATE
, &(bh
)->b_state
);
1890 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
1892 #endif /* __KERNEL__ */
1894 #endif /* _EXT4_H */