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 #define EXT4_ERROR_INODE(inode, fmt, a...) \
60 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
62 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
63 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
65 #define EXT4_ERROR_FILE(file, block, fmt, a...) \
66 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
68 /* data type for block offset of block group */
69 typedef int ext4_grpblk_t
;
71 /* data type for filesystem-wide blocks number */
72 typedef unsigned long long ext4_fsblk_t
;
74 /* data type for file logical block number */
75 typedef __u32 ext4_lblk_t
;
77 /* data type for block group number */
78 typedef unsigned int ext4_group_t
;
81 * Flags used in mballoc's allocation_context flags field.
83 * Also used to show what's going on for debugging purposes when the
84 * flag field is exported via the traceport interface
87 /* prefer goal again. length */
88 #define EXT4_MB_HINT_MERGE 0x0001
89 /* blocks already reserved */
90 #define EXT4_MB_HINT_RESERVED 0x0002
91 /* metadata is being allocated */
92 #define EXT4_MB_HINT_METADATA 0x0004
93 /* first blocks in the file */
94 #define EXT4_MB_HINT_FIRST 0x0008
95 /* search for the best chunk */
96 #define EXT4_MB_HINT_BEST 0x0010
97 /* data is being allocated */
98 #define EXT4_MB_HINT_DATA 0x0020
99 /* don't preallocate (for tails) */
100 #define EXT4_MB_HINT_NOPREALLOC 0x0040
101 /* allocate for locality group */
102 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080
103 /* allocate goal blocks or none */
104 #define EXT4_MB_HINT_GOAL_ONLY 0x0100
105 /* goal is meaningful */
106 #define EXT4_MB_HINT_TRY_GOAL 0x0200
107 /* blocks already pre-reserved by delayed allocation */
108 #define EXT4_MB_DELALLOC_RESERVED 0x0400
109 /* We are doing stream allocation */
110 #define EXT4_MB_STREAM_ALLOC 0x0800
113 struct ext4_allocation_request
{
114 /* target inode for block we're allocating */
116 /* how many blocks we want to allocate */
118 /* logical block in target inode */
120 /* the closest logical allocated block to the left */
122 /* the closest logical allocated block to the right */
124 /* phys. target (a hint) */
126 /* phys. block for the closest logical allocated block to the left */
128 /* phys. block for the closest logical allocated block to the right */
130 /* flags. see above EXT4_MB_HINT_* */
135 * Logical to physical block mapping, used by ext4_map_blocks()
137 * This structure is used to pass requests into ext4_map_blocks() as
138 * well as to store the information returned by ext4_map_blocks(). It
139 * takes less room on the stack than a struct buffer_head.
141 #define EXT4_MAP_NEW (1 << BH_New)
142 #define EXT4_MAP_MAPPED (1 << BH_Mapped)
143 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten)
144 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary)
145 #define EXT4_MAP_UNINIT (1 << BH_Uninit)
146 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
147 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\
150 struct ext4_map_blocks
{
154 unsigned int m_flags
;
158 * For delayed allocation tracking
160 struct mpage_da_data
{
162 sector_t b_blocknr
; /* start block number of extent */
163 size_t b_size
; /* size of extent */
164 unsigned long b_state
; /* state of the extent */
165 unsigned long first_page
, next_page
; /* extent of pages */
166 struct writeback_control
*wbc
;
173 * Flags for ext4_io_end->flags
175 #define EXT4_IO_END_UNWRITTEN 0x0001
176 #define EXT4_IO_END_ERROR 0x0002
178 struct ext4_io_page
{
183 #define MAX_IO_PAGES 128
185 typedef struct ext4_io_end
{
186 struct list_head list
; /* per-file finished IO list */
187 struct inode
*inode
; /* file being written to */
188 unsigned int flag
; /* unwritten or not */
189 struct page
*page
; /* page struct for buffer write */
190 loff_t offset
; /* offset in the file */
191 ssize_t size
; /* size of the extent */
192 struct work_struct work
; /* data work queue */
193 struct kiocb
*iocb
; /* iocb struct for AIO */
194 int result
; /* error value for AIO */
196 struct ext4_io_page
*pages
[MAX_IO_PAGES
];
199 struct ext4_io_submit
{
202 ext4_io_end_t
*io_end
;
203 struct ext4_io_page
*io_page
;
204 sector_t io_next_block
;
208 * Special inodes numbers
210 #define EXT4_BAD_INO 1 /* Bad blocks inode */
211 #define EXT4_ROOT_INO 2 /* Root inode */
212 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
213 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
214 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
215 #define EXT4_JOURNAL_INO 8 /* Journal inode */
217 /* First non-reserved inode for old ext4 filesystems */
218 #define EXT4_GOOD_OLD_FIRST_INO 11
221 * Maximal count of links to a file
223 #define EXT4_LINK_MAX 65000
226 * Macro-instructions used to manage several block sizes
228 #define EXT4_MIN_BLOCK_SIZE 1024
229 #define EXT4_MAX_BLOCK_SIZE 65536
230 #define EXT4_MIN_BLOCK_LOG_SIZE 10
231 #define EXT4_MAX_BLOCK_LOG_SIZE 16
233 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
235 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
237 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
239 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
241 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
244 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
245 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
246 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
248 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
249 EXT4_GOOD_OLD_INODE_SIZE : \
251 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
252 EXT4_GOOD_OLD_FIRST_INO : \
255 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
258 * Structure of a blocks group descriptor
260 struct ext4_group_desc
262 __le32 bg_block_bitmap_lo
; /* Blocks bitmap block */
263 __le32 bg_inode_bitmap_lo
; /* Inodes bitmap block */
264 __le32 bg_inode_table_lo
; /* Inodes table block */
265 __le16 bg_free_blocks_count_lo
;/* Free blocks count */
266 __le16 bg_free_inodes_count_lo
;/* Free inodes count */
267 __le16 bg_used_dirs_count_lo
; /* Directories count */
268 __le16 bg_flags
; /* EXT4_BG_flags (INODE_UNINIT, etc) */
269 __u32 bg_reserved
[2]; /* Likely block/inode bitmap checksum */
270 __le16 bg_itable_unused_lo
; /* Unused inodes count */
271 __le16 bg_checksum
; /* crc16(sb_uuid+group+desc) */
272 __le32 bg_block_bitmap_hi
; /* Blocks bitmap block MSB */
273 __le32 bg_inode_bitmap_hi
; /* Inodes bitmap block MSB */
274 __le32 bg_inode_table_hi
; /* Inodes table block MSB */
275 __le16 bg_free_blocks_count_hi
;/* Free blocks count MSB */
276 __le16 bg_free_inodes_count_hi
;/* Free inodes count MSB */
277 __le16 bg_used_dirs_count_hi
; /* Directories count MSB */
278 __le16 bg_itable_unused_hi
; /* Unused inodes count MSB */
279 __u32 bg_reserved2
[3];
283 * Structure of a flex block group info
287 atomic_t free_inodes
;
288 atomic_t free_blocks
;
292 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
293 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
294 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
297 * Macro-instructions used to manage group descriptors
299 #define EXT4_MIN_DESC_SIZE 32
300 #define EXT4_MIN_DESC_SIZE_64BIT 64
301 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
302 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
304 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
305 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
306 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
307 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
309 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
310 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
311 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
315 * Constants relative to the data blocks
317 #define EXT4_NDIR_BLOCKS 12
318 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
319 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
320 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
321 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
326 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
327 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
328 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
329 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
330 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
331 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
332 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
333 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
334 /* Reserved for compression usage... */
335 #define EXT4_DIRTY_FL 0x00000100
336 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
337 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
338 #define EXT4_ECOMPR_FL 0x00000800 /* Compression error */
339 /* End compression flags --- maybe not all used */
340 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
341 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
342 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
343 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
344 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
345 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
346 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
347 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
348 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
349 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
350 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
352 #define EXT4_FL_USER_VISIBLE 0x004BDFFF /* User visible flags */
353 #define EXT4_FL_USER_MODIFIABLE 0x004B80FF /* User modifiable flags */
355 /* Flags that should be inherited by new inodes from their parent. */
356 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
357 EXT4_SYNC_FL | EXT4_IMMUTABLE_FL | EXT4_APPEND_FL |\
358 EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
359 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
360 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
362 /* Flags that are appropriate for regular files (all but dir-specific ones). */
363 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
365 /* Flags that are appropriate for non-directories/regular files. */
366 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
368 /* Mask out flags that are inappropriate for the given type of inode. */
369 static inline __u32
ext4_mask_flags(umode_t mode
, __u32 flags
)
373 else if (S_ISREG(mode
))
374 return flags
& EXT4_REG_FLMASK
;
376 return flags
& EXT4_OTHER_FLMASK
;
380 * Inode flags used for atomic set/get
383 EXT4_INODE_SECRM
= 0, /* Secure deletion */
384 EXT4_INODE_UNRM
= 1, /* Undelete */
385 EXT4_INODE_COMPR
= 2, /* Compress file */
386 EXT4_INODE_SYNC
= 3, /* Synchronous updates */
387 EXT4_INODE_IMMUTABLE
= 4, /* Immutable file */
388 EXT4_INODE_APPEND
= 5, /* writes to file may only append */
389 EXT4_INODE_NODUMP
= 6, /* do not dump file */
390 EXT4_INODE_NOATIME
= 7, /* do not update atime */
391 /* Reserved for compression usage... */
392 EXT4_INODE_DIRTY
= 8,
393 EXT4_INODE_COMPRBLK
= 9, /* One or more compressed clusters */
394 EXT4_INODE_NOCOMPR
= 10, /* Don't compress */
395 EXT4_INODE_ECOMPR
= 11, /* Compression error */
396 /* End compression flags --- maybe not all used */
397 EXT4_INODE_INDEX
= 12, /* hash-indexed directory */
398 EXT4_INODE_IMAGIC
= 13, /* AFS directory */
399 EXT4_INODE_JOURNAL_DATA
= 14, /* file data should be journaled */
400 EXT4_INODE_NOTAIL
= 15, /* file tail should not be merged */
401 EXT4_INODE_DIRSYNC
= 16, /* dirsync behaviour (directories only) */
402 EXT4_INODE_TOPDIR
= 17, /* Top of directory hierarchies*/
403 EXT4_INODE_HUGE_FILE
= 18, /* Set to each huge file */
404 EXT4_INODE_EXTENTS
= 19, /* Inode uses extents */
405 EXT4_INODE_EA_INODE
= 21, /* Inode used for large EA */
406 EXT4_INODE_EOFBLOCKS
= 22, /* Blocks allocated beyond EOF */
407 EXT4_INODE_RESERVED
= 31, /* reserved for ext4 lib */
410 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
411 #define CHECK_FLAG_VALUE(FLAG) if (!TEST_FLAG_VALUE(FLAG)) { \
412 printk(KERN_EMERG "EXT4 flag fail: " #FLAG ": %d %d\n", \
413 EXT4_##FLAG##_FL, EXT4_INODE_##FLAG); BUG_ON(1); }
416 * Since it's pretty easy to mix up bit numbers and hex values, and we
417 * can't do a compile-time test for ENUM values, we use a run-time
418 * test to make sure that EXT4_XXX_FL is consistent with respect to
419 * EXT4_INODE_XXX. If all is well the printk and BUG_ON will all drop
420 * out so it won't cost any extra space in the compiled kernel image.
421 * But it's important that these values are the same, since we are
422 * using EXT4_INODE_XXX to test for the flag values, but EXT4_XX_FL
423 * must be consistent with the values of FS_XXX_FL defined in
424 * include/linux/fs.h and the on-disk values found in ext2, ext3, and
425 * ext4 filesystems, and of course the values defined in e2fsprogs.
427 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
429 static inline void ext4_check_flag_values(void)
431 CHECK_FLAG_VALUE(SECRM
);
432 CHECK_FLAG_VALUE(UNRM
);
433 CHECK_FLAG_VALUE(COMPR
);
434 CHECK_FLAG_VALUE(SYNC
);
435 CHECK_FLAG_VALUE(IMMUTABLE
);
436 CHECK_FLAG_VALUE(APPEND
);
437 CHECK_FLAG_VALUE(NODUMP
);
438 CHECK_FLAG_VALUE(NOATIME
);
439 CHECK_FLAG_VALUE(DIRTY
);
440 CHECK_FLAG_VALUE(COMPRBLK
);
441 CHECK_FLAG_VALUE(NOCOMPR
);
442 CHECK_FLAG_VALUE(ECOMPR
);
443 CHECK_FLAG_VALUE(INDEX
);
444 CHECK_FLAG_VALUE(IMAGIC
);
445 CHECK_FLAG_VALUE(JOURNAL_DATA
);
446 CHECK_FLAG_VALUE(NOTAIL
);
447 CHECK_FLAG_VALUE(DIRSYNC
);
448 CHECK_FLAG_VALUE(TOPDIR
);
449 CHECK_FLAG_VALUE(HUGE_FILE
);
450 CHECK_FLAG_VALUE(EXTENTS
);
451 CHECK_FLAG_VALUE(EA_INODE
);
452 CHECK_FLAG_VALUE(EOFBLOCKS
);
453 CHECK_FLAG_VALUE(RESERVED
);
456 /* Used to pass group descriptor data when online resize is done */
457 struct ext4_new_group_input
{
458 __u32 group
; /* Group number for this data */
459 __u64 block_bitmap
; /* Absolute block number of block bitmap */
460 __u64 inode_bitmap
; /* Absolute block number of inode bitmap */
461 __u64 inode_table
; /* Absolute block number of inode table start */
462 __u32 blocks_count
; /* Total number of blocks in this group */
463 __u16 reserved_blocks
; /* Number of reserved blocks in this group */
467 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
468 struct compat_ext4_new_group_input
{
470 compat_u64 block_bitmap
;
471 compat_u64 inode_bitmap
;
472 compat_u64 inode_table
;
479 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
480 struct ext4_new_group_data
{
486 __u16 reserved_blocks
;
488 __u32 free_blocks_count
;
492 * Flags used by ext4_map_blocks()
494 /* Allocate any needed blocks and/or convert an unitialized
495 extent to be an initialized ext4 */
496 #define EXT4_GET_BLOCKS_CREATE 0x0001
497 /* Request the creation of an unitialized extent */
498 #define EXT4_GET_BLOCKS_UNINIT_EXT 0x0002
499 #define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT (EXT4_GET_BLOCKS_UNINIT_EXT|\
500 EXT4_GET_BLOCKS_CREATE)
501 /* Caller is from the delayed allocation writeout path,
502 so set the magic i_delalloc_reserve_flag after taking the
503 inode allocation semaphore for */
504 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
505 /* caller is from the direct IO path, request to creation of an
506 unitialized extents if not allocated, split the uninitialized
507 extent if blocks has been preallocated already*/
508 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
509 #define EXT4_GET_BLOCKS_CONVERT 0x0010
510 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
511 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
512 /* Convert extent to initialized after IO complete */
513 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
514 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
517 * Flags used by ext4_free_blocks
519 #define EXT4_FREE_BLOCKS_METADATA 0x0001
520 #define EXT4_FREE_BLOCKS_FORGET 0x0002
521 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
526 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
527 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
528 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
529 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
530 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
531 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
532 #ifdef CONFIG_JBD2_DEBUG
533 #define EXT4_IOC_WAIT_FOR_READONLY _IOR('f', 99, long)
535 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
536 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
537 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
538 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
539 #define EXT4_IOC_MIGRATE _IO('f', 9)
540 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
541 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
542 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
543 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
545 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
547 * ioctl commands in 32 bit emulation
549 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
550 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
551 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
552 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
553 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
554 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
555 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
556 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
557 #ifdef CONFIG_JBD2_DEBUG
558 #define EXT4_IOC32_WAIT_FOR_READONLY _IOR('f', 99, int)
560 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
561 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
564 /* Max physical block we can addres w/o extents */
565 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
568 * Structure of an inode on the disk
571 __le16 i_mode
; /* File mode */
572 __le16 i_uid
; /* Low 16 bits of Owner Uid */
573 __le32 i_size_lo
; /* Size in bytes */
574 __le32 i_atime
; /* Access time */
575 __le32 i_ctime
; /* Inode Change time */
576 __le32 i_mtime
; /* Modification time */
577 __le32 i_dtime
; /* Deletion Time */
578 __le16 i_gid
; /* Low 16 bits of Group Id */
579 __le16 i_links_count
; /* Links count */
580 __le32 i_blocks_lo
; /* Blocks count */
581 __le32 i_flags
; /* File flags */
587 __u32 h_i_translator
;
592 } osd1
; /* OS dependent 1 */
593 __le32 i_block
[EXT4_N_BLOCKS
];/* Pointers to blocks */
594 __le32 i_generation
; /* File version (for NFS) */
595 __le32 i_file_acl_lo
; /* File ACL */
597 __le32 i_obso_faddr
; /* Obsoleted fragment address */
600 __le16 l_i_blocks_high
; /* were l_i_reserved1 */
601 __le16 l_i_file_acl_high
;
602 __le16 l_i_uid_high
; /* these 2 fields */
603 __le16 l_i_gid_high
; /* were reserved2[0] */
607 __le16 h_i_reserved1
; /* Obsoleted fragment number/size which are removed in ext4 */
614 __le16 h_i_reserved1
; /* Obsoleted fragment number/size which are removed in ext4 */
615 __le16 m_i_file_acl_high
;
616 __u32 m_i_reserved2
[2];
618 } osd2
; /* OS dependent 2 */
619 __le16 i_extra_isize
;
621 __le32 i_ctime_extra
; /* extra Change time (nsec << 2 | epoch) */
622 __le32 i_mtime_extra
; /* extra Modification time(nsec << 2 | epoch) */
623 __le32 i_atime_extra
; /* extra Access time (nsec << 2 | epoch) */
624 __le32 i_crtime
; /* File Creation time */
625 __le32 i_crtime_extra
; /* extra FileCreationtime (nsec << 2 | epoch) */
626 __le32 i_version_hi
; /* high 32 bits for 64-bit version */
630 __u32 reserved
; /* should be zero */
631 __u32 donor_fd
; /* donor file descriptor */
632 __u64 orig_start
; /* logical start offset in block for orig */
633 __u64 donor_start
; /* logical start offset in block for donor */
634 __u64 len
; /* block length to be moved */
635 __u64 moved_len
; /* moved block length */
638 #define EXT4_EPOCH_BITS 2
639 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
640 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
643 * Extended fields will fit into an inode if the filesystem was formatted
644 * with large inodes (-I 256 or larger) and there are not currently any EAs
645 * consuming all of the available space. For new inodes we always reserve
646 * enough space for the kernel's known extended fields, but for inodes
647 * created with an old kernel this might not have been the case. None of
648 * the extended inode fields is critical for correct filesystem operation.
649 * This macro checks if a certain field fits in the inode. Note that
650 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
652 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
653 ((offsetof(typeof(*ext4_inode), field) + \
654 sizeof((ext4_inode)->field)) \
655 <= (EXT4_GOOD_OLD_INODE_SIZE + \
656 (einode)->i_extra_isize)) \
658 static inline __le32 ext4_encode_extra_time(struct timespec *time)
660 return cpu_to_le32((sizeof(time
->tv_sec
) > 4 ?
661 (time
->tv_sec
>> 32) & EXT4_EPOCH_MASK
: 0) |
662 ((time
->tv_nsec
<< EXT4_EPOCH_BITS
) & EXT4_NSEC_MASK
));
665 static inline void ext4_decode_extra_time(struct timespec
*time
, __le32 extra
)
667 if (sizeof(time
->tv_sec
) > 4)
668 time
->tv_sec
|= (__u64
)(le32_to_cpu(extra
) & EXT4_EPOCH_MASK
)
670 time
->tv_nsec
= (le32_to_cpu(extra
) & EXT4_NSEC_MASK
) >> EXT4_EPOCH_BITS
;
673 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
675 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
676 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
677 (raw_inode)->xtime ## _extra = \
678 ext4_encode_extra_time(&(inode)->xtime); \
681 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
683 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
684 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
685 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
686 (raw_inode)->xtime ## _extra = \
687 ext4_encode_extra_time(&(einode)->xtime); \
690 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
692 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
693 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
694 ext4_decode_extra_time(&(inode)->xtime, \
695 raw_inode->xtime ## _extra); \
697 (inode)->xtime.tv_nsec = 0; \
700 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
702 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
703 (einode)->xtime.tv_sec = \
704 (signed)le32_to_cpu((raw_inode)->xtime); \
705 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
706 ext4_decode_extra_time(&(einode)->xtime, \
707 raw_inode->xtime ## _extra); \
709 (einode)->xtime.tv_nsec = 0; \
712 #define i_disk_version osd1.linux1.l_i_version
714 #if defined(__KERNEL__) || defined(__linux__)
715 #define i_reserved1 osd1.linux1.l_i_reserved1
716 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
717 #define i_blocks_high osd2.linux2.l_i_blocks_high
718 #define i_uid_low i_uid
719 #define i_gid_low i_gid
720 #define i_uid_high osd2.linux2.l_i_uid_high
721 #define i_gid_high osd2.linux2.l_i_gid_high
722 #define i_reserved2 osd2.linux2.l_i_reserved2
724 #elif defined(__GNU__)
726 #define i_translator osd1.hurd1.h_i_translator
727 #define i_uid_high osd2.hurd2.h_i_uid_high
728 #define i_gid_high osd2.hurd2.h_i_gid_high
729 #define i_author osd2.hurd2.h_i_author
731 #elif defined(__masix__)
733 #define i_reserved1 osd1.masix1.m_i_reserved1
734 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
735 #define i_reserved2 osd2.masix2.m_i_reserved2
737 #endif /* defined(__KERNEL__) || defined(__linux__) */
740 * storage for cached extent
741 * If ec_len == 0, then the cache is invalid.
742 * If ec_start == 0, then the cache represents a gap (null mapping)
744 struct ext4_ext_cache
{
745 ext4_fsblk_t ec_start
;
746 ext4_lblk_t ec_block
;
747 __u32 ec_len
; /* must be 32bit to return holes */
751 * fourth extended file system inode data in memory
753 struct ext4_inode_info
{
754 __le32 i_data
[15]; /* unconverted */
756 ext4_fsblk_t i_file_acl
;
759 * i_block_group is the number of the block group which contains
760 * this file's inode. Constant across the lifetime of the inode,
761 * it is ued for making block allocation decisions - we try to
762 * place a file's data blocks near its inode block, and new inodes
763 * near to their parent directory's inode.
765 ext4_group_t i_block_group
;
766 ext4_lblk_t i_dir_start_lookup
;
767 #if (BITS_PER_LONG < 64)
768 unsigned long i_state_flags
; /* Dynamic state flags */
770 unsigned long i_flags
;
772 #ifdef CONFIG_EXT4_FS_XATTR
774 * Extended attributes can be read independently of the main file
775 * data. Taking i_mutex even when reading would cause contention
776 * between readers of EAs and writers of regular file data, so
777 * instead we synchronize on xattr_sem when reading or changing
780 struct rw_semaphore xattr_sem
;
783 struct list_head i_orphan
; /* unlinked but open inodes */
786 * i_disksize keeps track of what the inode size is ON DISK, not
787 * in memory. During truncate, i_size is set to the new size by
788 * the VFS prior to calling ext4_truncate(), but the filesystem won't
789 * set i_disksize to 0 until the truncate is actually under way.
791 * The intent is that i_disksize always represents the blocks which
792 * are used by this file. This allows recovery to restart truncate
793 * on orphans if we crash during truncate. We actually write i_disksize
794 * into the on-disk inode when writing inodes out, instead of i_size.
796 * The only time when i_disksize and i_size may be different is when
797 * a truncate is in progress. The only things which change i_disksize
798 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
803 * i_data_sem is for serialising ext4_truncate() against
804 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
805 * data tree are chopped off during truncate. We can't do that in
806 * ext4 because whenever we perform intermediate commits during
807 * truncate, the inode and all the metadata blocks *must* be in a
808 * consistent state which allows truncation of the orphans to restart
809 * during recovery. Hence we must fix the get_block-vs-truncate race
810 * by other means, so we have i_data_sem.
812 struct rw_semaphore i_data_sem
;
813 struct inode vfs_inode
;
814 struct jbd2_inode
*jinode
;
816 struct ext4_ext_cache i_cached_extent
;
818 * File creation time. Its function is same as that of
819 * struct timespec i_{a,c,m}time in the generic inode.
821 struct timespec i_crtime
;
824 struct list_head i_prealloc_list
;
825 spinlock_t i_prealloc_lock
;
828 ext4_group_t i_last_alloc_group
;
830 /* allocation reservation info for delalloc */
831 unsigned int i_reserved_data_blocks
;
832 unsigned int i_reserved_meta_blocks
;
833 unsigned int i_allocated_meta_blocks
;
834 ext4_lblk_t i_da_metadata_calc_last_lblock
;
835 int i_da_metadata_calc_len
;
837 /* on-disk additional length */
841 /* quota space reservation, managed internally by quota code */
842 qsize_t i_reserved_quota
;
845 /* completed IOs that might need unwritten extents handling */
846 struct list_head i_completed_io_list
;
847 spinlock_t i_completed_io_lock
;
848 atomic_t i_ioend_count
; /* Number of outstanding io_end structs */
849 /* current io_end structure for async DIO write*/
850 ext4_io_end_t
*cur_aio_dio
;
852 spinlock_t i_block_reservation_lock
;
855 * Transactions that contain inode's metadata needed to complete
856 * fsync and fdatasync, respectively.
859 tid_t i_datasync_tid
;
865 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
866 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
867 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
870 * Misc. filesystem flags
872 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
873 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
874 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
879 #define EXT4_MOUNT_OLDALLOC 0x00002 /* Don't use the new Orlov allocator */
880 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
881 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
882 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
883 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
884 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
885 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
886 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
887 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
888 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
889 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
890 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
891 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
892 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
893 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
894 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
895 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
896 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
897 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
898 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
899 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
900 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
901 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
902 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
903 #define EXT4_MOUNT_I_VERSION 0x2000000 /* i_version support */
904 #define EXT4_MOUNT_MBLK_IO_SUBMIT 0x4000000 /* multi-block io submits */
905 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
906 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
907 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
908 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
909 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
911 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
913 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
915 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
918 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
920 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
922 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
925 #define ext4_set_bit ext2_set_bit
926 #define ext4_set_bit_atomic ext2_set_bit_atomic
927 #define ext4_clear_bit ext2_clear_bit
928 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
929 #define ext4_test_bit ext2_test_bit
930 #define ext4_find_first_zero_bit ext2_find_first_zero_bit
931 #define ext4_find_next_zero_bit ext2_find_next_zero_bit
932 #define ext4_find_next_bit ext2_find_next_bit
935 * Maximal mount counts between two filesystem checks
937 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
938 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
941 * Behaviour when detecting errors
943 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
944 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
945 #define EXT4_ERRORS_PANIC 3 /* Panic */
946 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
949 * Structure of the super block
951 struct ext4_super_block
{
952 /*00*/ __le32 s_inodes_count
; /* Inodes count */
953 __le32 s_blocks_count_lo
; /* Blocks count */
954 __le32 s_r_blocks_count_lo
; /* Reserved blocks count */
955 __le32 s_free_blocks_count_lo
; /* Free blocks count */
956 /*10*/ __le32 s_free_inodes_count
; /* Free inodes count */
957 __le32 s_first_data_block
; /* First Data Block */
958 __le32 s_log_block_size
; /* Block size */
959 __le32 s_obso_log_frag_size
; /* Obsoleted fragment size */
960 /*20*/ __le32 s_blocks_per_group
; /* # Blocks per group */
961 __le32 s_obso_frags_per_group
; /* Obsoleted fragments per group */
962 __le32 s_inodes_per_group
; /* # Inodes per group */
963 __le32 s_mtime
; /* Mount time */
964 /*30*/ __le32 s_wtime
; /* Write time */
965 __le16 s_mnt_count
; /* Mount count */
966 __le16 s_max_mnt_count
; /* Maximal mount count */
967 __le16 s_magic
; /* Magic signature */
968 __le16 s_state
; /* File system state */
969 __le16 s_errors
; /* Behaviour when detecting errors */
970 __le16 s_minor_rev_level
; /* minor revision level */
971 /*40*/ __le32 s_lastcheck
; /* time of last check */
972 __le32 s_checkinterval
; /* max. time between checks */
973 __le32 s_creator_os
; /* OS */
974 __le32 s_rev_level
; /* Revision level */
975 /*50*/ __le16 s_def_resuid
; /* Default uid for reserved blocks */
976 __le16 s_def_resgid
; /* Default gid for reserved blocks */
978 * These fields are for EXT4_DYNAMIC_REV superblocks only.
980 * Note: the difference between the compatible feature set and
981 * the incompatible feature set is that if there is a bit set
982 * in the incompatible feature set that the kernel doesn't
983 * know about, it should refuse to mount the filesystem.
985 * e2fsck's requirements are more strict; if it doesn't know
986 * about a feature in either the compatible or incompatible
987 * feature set, it must abort and not try to meddle with
988 * things it doesn't understand...
990 __le32 s_first_ino
; /* First non-reserved inode */
991 __le16 s_inode_size
; /* size of inode structure */
992 __le16 s_block_group_nr
; /* block group # of this superblock */
993 __le32 s_feature_compat
; /* compatible feature set */
994 /*60*/ __le32 s_feature_incompat
; /* incompatible feature set */
995 __le32 s_feature_ro_compat
; /* readonly-compatible feature set */
996 /*68*/ __u8 s_uuid
[16]; /* 128-bit uuid for volume */
997 /*78*/ char s_volume_name
[16]; /* volume name */
998 /*88*/ char s_last_mounted
[64]; /* directory where last mounted */
999 /*C8*/ __le32 s_algorithm_usage_bitmap
; /* For compression */
1001 * Performance hints. Directory preallocation should only
1002 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1004 __u8 s_prealloc_blocks
; /* Nr of blocks to try to preallocate*/
1005 __u8 s_prealloc_dir_blocks
; /* Nr to preallocate for dirs */
1006 __le16 s_reserved_gdt_blocks
; /* Per group desc for online growth */
1008 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1010 /*D0*/ __u8 s_journal_uuid
[16]; /* uuid of journal superblock */
1011 /*E0*/ __le32 s_journal_inum
; /* inode number of journal file */
1012 __le32 s_journal_dev
; /* device number of journal file */
1013 __le32 s_last_orphan
; /* start of list of inodes to delete */
1014 __le32 s_hash_seed
[4]; /* HTREE hash seed */
1015 __u8 s_def_hash_version
; /* Default hash version to use */
1016 __u8 s_jnl_backup_type
;
1017 __le16 s_desc_size
; /* size of group descriptor */
1018 /*100*/ __le32 s_default_mount_opts
;
1019 __le32 s_first_meta_bg
; /* First metablock block group */
1020 __le32 s_mkfs_time
; /* When the filesystem was created */
1021 __le32 s_jnl_blocks
[17]; /* Backup of the journal inode */
1022 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1023 /*150*/ __le32 s_blocks_count_hi
; /* Blocks count */
1024 __le32 s_r_blocks_count_hi
; /* Reserved blocks count */
1025 __le32 s_free_blocks_count_hi
; /* Free blocks count */
1026 __le16 s_min_extra_isize
; /* All inodes have at least # bytes */
1027 __le16 s_want_extra_isize
; /* New inodes should reserve # bytes */
1028 __le32 s_flags
; /* Miscellaneous flags */
1029 __le16 s_raid_stride
; /* RAID stride */
1030 __le16 s_mmp_interval
; /* # seconds to wait in MMP checking */
1031 __le64 s_mmp_block
; /* Block for multi-mount protection */
1032 __le32 s_raid_stripe_width
; /* blocks on all data disks (N*stride)*/
1033 __u8 s_log_groups_per_flex
; /* FLEX_BG group size */
1034 __u8 s_reserved_char_pad
;
1035 __le16 s_reserved_pad
;
1036 __le64 s_kbytes_written
; /* nr of lifetime kilobytes written */
1037 __le32 s_snapshot_inum
; /* Inode number of active snapshot */
1038 __le32 s_snapshot_id
; /* sequential ID of active snapshot */
1039 __le64 s_snapshot_r_blocks_count
; /* reserved blocks for active
1040 snapshot's future use */
1041 __le32 s_snapshot_list
; /* inode number of the head of the
1042 on-disk snapshot list */
1043 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1044 __le32 s_error_count
; /* number of fs errors */
1045 __le32 s_first_error_time
; /* first time an error happened */
1046 __le32 s_first_error_ino
; /* inode involved in first error */
1047 __le64 s_first_error_block
; /* block involved of first error */
1048 __u8 s_first_error_func
[32]; /* function where the error happened */
1049 __le32 s_first_error_line
; /* line number where error happened */
1050 __le32 s_last_error_time
; /* most recent time of an error */
1051 __le32 s_last_error_ino
; /* inode involved in last error */
1052 __le32 s_last_error_line
; /* line number where error happened */
1053 __le64 s_last_error_block
; /* block involved of last error */
1054 __u8 s_last_error_func
[32]; /* function where the error happened */
1055 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1056 __u8 s_mount_opts
[64];
1057 __le32 s_reserved
[112]; /* Padding to the end of the block */
1060 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1065 * run-time mount flags
1067 #define EXT4_MF_MNTDIR_SAMPLED 0x0001
1068 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1071 * fourth extended-fs super-block data in memory
1073 struct ext4_sb_info
{
1074 unsigned long s_desc_size
; /* Size of a group descriptor in bytes */
1075 unsigned long s_inodes_per_block
;/* Number of inodes per block */
1076 unsigned long s_blocks_per_group
;/* Number of blocks in a group */
1077 unsigned long s_inodes_per_group
;/* Number of inodes in a group */
1078 unsigned long s_itb_per_group
; /* Number of inode table blocks per group */
1079 unsigned long s_gdb_count
; /* Number of group descriptor blocks */
1080 unsigned long s_desc_per_block
; /* Number of group descriptors per block */
1081 ext4_group_t s_groups_count
; /* Number of groups in the fs */
1082 ext4_group_t s_blockfile_groups
;/* Groups acceptable for non-extent files */
1083 unsigned long s_overhead_last
; /* Last calculated overhead */
1084 unsigned long s_blocks_last
; /* Last seen block count */
1085 loff_t s_bitmap_maxbytes
; /* max bytes for bitmap files */
1086 struct buffer_head
* s_sbh
; /* Buffer containing the super block */
1087 struct ext4_super_block
*s_es
; /* Pointer to the super block in the buffer */
1088 struct buffer_head
**s_group_desc
;
1089 unsigned int s_mount_opt
;
1090 unsigned int s_mount_opt2
;
1091 unsigned int s_mount_flags
;
1092 ext4_fsblk_t s_sb_block
;
1095 unsigned short s_mount_state
;
1096 unsigned short s_pad
;
1097 int s_addr_per_block_bits
;
1098 int s_desc_per_block_bits
;
1101 unsigned int s_inode_readahead_blks
;
1102 unsigned int s_inode_goal
;
1103 spinlock_t s_next_gen_lock
;
1104 u32 s_next_generation
;
1106 int s_def_hash_version
;
1107 int s_hash_unsigned
; /* 3 if hash should be signed, 0 if not */
1108 struct percpu_counter s_freeblocks_counter
;
1109 struct percpu_counter s_freeinodes_counter
;
1110 struct percpu_counter s_dirs_counter
;
1111 struct percpu_counter s_dirtyblocks_counter
;
1112 struct blockgroup_lock
*s_blockgroup_lock
;
1113 struct proc_dir_entry
*s_proc
;
1114 struct kobject s_kobj
;
1115 struct completion s_kobj_unregister
;
1118 struct journal_s
*s_journal
;
1119 struct list_head s_orphan
;
1120 struct mutex s_orphan_lock
;
1121 struct mutex s_resize_lock
;
1122 unsigned long s_commit_interval
;
1123 u32 s_max_batch_time
;
1124 u32 s_min_batch_time
;
1125 struct block_device
*journal_bdev
;
1126 #ifdef CONFIG_JBD2_DEBUG
1127 struct timer_list turn_ro_timer
; /* For turning read-only (crash simulation) */
1128 wait_queue_head_t ro_wait_queue
; /* For people waiting for the fs to go read-only */
1131 char *s_qf_names
[MAXQUOTAS
]; /* Names of quota files with journalled quota */
1132 int s_jquota_fmt
; /* Format of quota to use */
1134 unsigned int s_want_extra_isize
; /* New inodes should reserve # bytes */
1135 struct rb_root system_blks
;
1137 #ifdef EXTENTS_STATS
1138 /* ext4 extents stats */
1139 unsigned long s_ext_min
;
1140 unsigned long s_ext_max
;
1141 unsigned long s_depth_max
;
1142 spinlock_t s_ext_stats_lock
;
1143 unsigned long s_ext_blocks
;
1144 unsigned long s_ext_extents
;
1147 /* for buddy allocator */
1148 struct ext4_group_info
***s_group_info
;
1149 struct inode
*s_buddy_cache
;
1150 spinlock_t s_md_lock
;
1151 unsigned short *s_mb_offsets
;
1152 unsigned int *s_mb_maxs
;
1155 unsigned long s_stripe
;
1156 unsigned int s_mb_stream_request
;
1157 unsigned int s_mb_max_to_scan
;
1158 unsigned int s_mb_min_to_scan
;
1159 unsigned int s_mb_stats
;
1160 unsigned int s_mb_order2_reqs
;
1161 unsigned int s_mb_group_prealloc
;
1162 unsigned int s_max_writeback_mb_bump
;
1163 /* where last allocation was done - for stream allocation */
1164 unsigned long s_mb_last_group
;
1165 unsigned long s_mb_last_start
;
1167 /* stats for buddy allocator */
1168 atomic_t s_bal_reqs
; /* number of reqs with len > 1 */
1169 atomic_t s_bal_success
; /* we found long enough chunks */
1170 atomic_t s_bal_allocated
; /* in blocks */
1171 atomic_t s_bal_ex_scanned
; /* total extents scanned */
1172 atomic_t s_bal_goals
; /* goal hits */
1173 atomic_t s_bal_breaks
; /* too long searches */
1174 atomic_t s_bal_2orders
; /* 2^order hits */
1175 spinlock_t s_bal_lock
;
1176 unsigned long s_mb_buddies_generated
;
1177 unsigned long long s_mb_generation_time
;
1178 atomic_t s_mb_lost_chunks
;
1179 atomic_t s_mb_preallocated
;
1180 atomic_t s_mb_discarded
;
1181 atomic_t s_lock_busy
;
1183 /* locality groups */
1184 struct ext4_locality_group __percpu
*s_locality_groups
;
1186 /* for write statistics */
1187 unsigned long s_sectors_written_start
;
1188 u64 s_kbytes_written
;
1190 unsigned int s_log_groups_per_flex
;
1191 struct flex_groups
*s_flex_groups
;
1193 /* workqueue for dio unwritten */
1194 struct workqueue_struct
*dio_unwritten_wq
;
1196 /* timer for periodic error stats printing */
1197 struct timer_list s_err_report
;
1199 /* Lazy inode table initialization info */
1200 struct ext4_li_request
*s_li_request
;
1201 /* Wait multiplier for lazy initialization thread */
1202 unsigned int s_li_wait_mult
;
1205 static inline struct ext4_sb_info
*EXT4_SB(struct super_block
*sb
)
1207 return sb
->s_fs_info
;
1209 static inline struct ext4_inode_info
*EXT4_I(struct inode
*inode
)
1211 return container_of(inode
, struct ext4_inode_info
, vfs_inode
);
1214 static inline struct timespec
ext4_current_time(struct inode
*inode
)
1216 return (inode
->i_sb
->s_time_gran
< NSEC_PER_SEC
) ?
1217 current_fs_time(inode
->i_sb
) : CURRENT_TIME_SEC
;
1220 static inline int ext4_valid_inum(struct super_block
*sb
, unsigned long ino
)
1222 return ino
== EXT4_ROOT_INO
||
1223 ino
== EXT4_JOURNAL_INO
||
1224 ino
== EXT4_RESIZE_INO
||
1225 (ino
>= EXT4_FIRST_INO(sb
) &&
1226 ino
<= le32_to_cpu(EXT4_SB(sb
)->s_es
->s_inodes_count
));
1230 * Inode dynamic state flags
1233 EXT4_STATE_JDATA
, /* journaled data exists */
1234 EXT4_STATE_NEW
, /* inode is newly created */
1235 EXT4_STATE_XATTR
, /* has in-inode xattrs */
1236 EXT4_STATE_NO_EXPAND
, /* No space for expansion */
1237 EXT4_STATE_DA_ALLOC_CLOSE
, /* Alloc DA blks on close */
1238 EXT4_STATE_EXT_MIGRATE
, /* Inode is migrating */
1239 EXT4_STATE_DIO_UNWRITTEN
, /* need convert on dio done*/
1240 EXT4_STATE_NEWENTRY
, /* File just added to dir */
1241 EXT4_STATE_DELALLOC_RESERVED
, /* blks already reserved for delalloc */
1244 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1245 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1247 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1249 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1251 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1253 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1255 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1258 EXT4_INODE_BIT_FNS(flag
, flags
, 0)
1259 #if (BITS_PER_LONG < 64)
1260 EXT4_INODE_BIT_FNS(state
, state_flags
, 0)
1262 static inline void ext4_clear_state_flags(struct ext4_inode_info
*ei
)
1264 (ei
)->i_state_flags
= 0;
1267 EXT4_INODE_BIT_FNS(state
, flags
, 32)
1269 static inline void ext4_clear_state_flags(struct ext4_inode_info
*ei
)
1271 /* We depend on the fact that callers will set i_flags */
1275 /* Assume that user mode programs are passing in an ext4fs superblock, not
1276 * a kernel struct super_block. This will allow us to call the feature-test
1277 * macros from user land. */
1278 #define EXT4_SB(sb) (sb)
1281 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1284 * Codes for operating systems
1286 #define EXT4_OS_LINUX 0
1287 #define EXT4_OS_HURD 1
1288 #define EXT4_OS_MASIX 2
1289 #define EXT4_OS_FREEBSD 3
1290 #define EXT4_OS_LITES 4
1295 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1296 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1298 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1299 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1301 #define EXT4_GOOD_OLD_INODE_SIZE 128
1304 * Feature set definitions
1307 #define EXT4_HAS_COMPAT_FEATURE(sb,mask) \
1308 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1309 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \
1310 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1311 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \
1312 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1313 #define EXT4_SET_COMPAT_FEATURE(sb,mask) \
1314 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1315 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \
1316 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1317 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \
1318 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1319 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \
1320 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1321 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
1322 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1323 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \
1324 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1326 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1327 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1328 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1329 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1330 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1331 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1333 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1334 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1335 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1336 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1337 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1338 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1339 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1341 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1342 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1343 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1344 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1345 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1346 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1347 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1348 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1349 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1350 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1351 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1353 #define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR
1354 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1355 EXT4_FEATURE_INCOMPAT_RECOVER| \
1356 EXT4_FEATURE_INCOMPAT_META_BG| \
1357 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1358 EXT4_FEATURE_INCOMPAT_64BIT| \
1359 EXT4_FEATURE_INCOMPAT_FLEX_BG)
1360 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1361 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1362 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1363 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1364 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1365 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1366 EXT4_FEATURE_RO_COMPAT_HUGE_FILE)
1369 * Default values for user and/or group using reserved blocks
1371 #define EXT4_DEF_RESUID 0
1372 #define EXT4_DEF_RESGID 0
1374 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
1377 * Default mount options
1379 #define EXT4_DEFM_DEBUG 0x0001
1380 #define EXT4_DEFM_BSDGROUPS 0x0002
1381 #define EXT4_DEFM_XATTR_USER 0x0004
1382 #define EXT4_DEFM_ACL 0x0008
1383 #define EXT4_DEFM_UID16 0x0010
1384 #define EXT4_DEFM_JMODE 0x0060
1385 #define EXT4_DEFM_JMODE_DATA 0x0020
1386 #define EXT4_DEFM_JMODE_ORDERED 0x0040
1387 #define EXT4_DEFM_JMODE_WBACK 0x0060
1388 #define EXT4_DEFM_NOBARRIER 0x0100
1389 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1390 #define EXT4_DEFM_DISCARD 0x0400
1391 #define EXT4_DEFM_NODELALLOC 0x0800
1394 * Default journal batch times
1396 #define EXT4_DEF_MIN_BATCH_TIME 0
1397 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1400 * Minimum number of groups in a flexgroup before we separate out
1401 * directories into the first block group of a flexgroup
1403 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1406 * Structure of a directory entry
1408 #define EXT4_NAME_LEN 255
1410 struct ext4_dir_entry
{
1411 __le32 inode
; /* Inode number */
1412 __le16 rec_len
; /* Directory entry length */
1413 __le16 name_len
; /* Name length */
1414 char name
[EXT4_NAME_LEN
]; /* File name */
1418 * The new version of the directory entry. Since EXT4 structures are
1419 * stored in intel byte order, and the name_len field could never be
1420 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1423 struct ext4_dir_entry_2
{
1424 __le32 inode
; /* Inode number */
1425 __le16 rec_len
; /* Directory entry length */
1426 __u8 name_len
; /* Name length */
1428 char name
[EXT4_NAME_LEN
]; /* File name */
1432 * Ext4 directory file types. Only the low 3 bits are used. The
1433 * other bits are reserved for now.
1435 #define EXT4_FT_UNKNOWN 0
1436 #define EXT4_FT_REG_FILE 1
1437 #define EXT4_FT_DIR 2
1438 #define EXT4_FT_CHRDEV 3
1439 #define EXT4_FT_BLKDEV 4
1440 #define EXT4_FT_FIFO 5
1441 #define EXT4_FT_SOCK 6
1442 #define EXT4_FT_SYMLINK 7
1444 #define EXT4_FT_MAX 8
1447 * EXT4_DIR_PAD defines the directory entries boundaries
1449 * NOTE: It must be a multiple of 4
1451 #define EXT4_DIR_PAD 4
1452 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1453 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1455 #define EXT4_MAX_REC_LEN ((1<<16)-1)
1458 * If we ever get support for fs block sizes > page_size, we'll need
1459 * to remove the #if statements in the next two functions...
1461 static inline unsigned int
1462 ext4_rec_len_from_disk(__le16 dlen
, unsigned blocksize
)
1464 unsigned len
= le16_to_cpu(dlen
);
1466 #if (PAGE_CACHE_SIZE >= 65536)
1467 if (len
== EXT4_MAX_REC_LEN
|| len
== 0)
1469 return (len
& 65532) | ((len
& 3) << 16);
1475 static inline __le16
ext4_rec_len_to_disk(unsigned len
, unsigned blocksize
)
1477 if ((len
> blocksize
) || (blocksize
> (1 << 18)) || (len
& 3))
1479 #if (PAGE_CACHE_SIZE >= 65536)
1481 return cpu_to_le16(len
);
1482 if (len
== blocksize
) {
1483 if (blocksize
== 65536)
1484 return cpu_to_le16(EXT4_MAX_REC_LEN
);
1486 return cpu_to_le16(0);
1488 return cpu_to_le16((len
& 65532) | ((len
>> 16) & 3));
1490 return cpu_to_le16(len
);
1495 * Hash Tree Directory indexing
1496 * (c) Daniel Phillips, 2001
1499 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1500 EXT4_FEATURE_COMPAT_DIR_INDEX) && \
1501 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
1502 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1503 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1505 /* Legal values for the dx_root hash_version field: */
1507 #define DX_HASH_LEGACY 0
1508 #define DX_HASH_HALF_MD4 1
1509 #define DX_HASH_TEA 2
1510 #define DX_HASH_LEGACY_UNSIGNED 3
1511 #define DX_HASH_HALF_MD4_UNSIGNED 4
1512 #define DX_HASH_TEA_UNSIGNED 5
1516 /* hash info structure used by the directory hash */
1525 #define EXT4_HTREE_EOF 0x7fffffff
1528 * Control parameters used by ext4_htree_next_block
1530 #define HASH_NB_ALWAYS 1
1534 * Describe an inode's exact location on disk and in memory
1538 struct buffer_head
*bh
;
1539 unsigned long offset
;
1540 ext4_group_t block_group
;
1543 static inline struct ext4_inode
*ext4_raw_inode(struct ext4_iloc
*iloc
)
1545 return (struct ext4_inode
*) (iloc
->bh
->b_data
+ iloc
->offset
);
1549 * This structure is stuffed into the struct file's private_data field
1550 * for directories. It is where we put information so that we can do
1551 * readdir operations in hash tree order.
1553 struct dir_private_info
{
1554 struct rb_root root
;
1555 struct rb_node
*curr_node
;
1556 struct fname
*extra_fname
;
1559 __u32 curr_minor_hash
;
1563 /* calculate the first block number of the group */
1564 static inline ext4_fsblk_t
1565 ext4_group_first_block_no(struct super_block
*sb
, ext4_group_t group_no
)
1567 return group_no
* (ext4_fsblk_t
)EXT4_BLOCKS_PER_GROUP(sb
) +
1568 le32_to_cpu(EXT4_SB(sb
)->s_es
->s_first_data_block
);
1572 * Special error return code only used by dx_probe() and its callers.
1574 #define ERR_BAD_DX_DIR -75000
1576 void ext4_get_group_no_and_offset(struct super_block
*sb
, ext4_fsblk_t blocknr
,
1577 ext4_group_t
*blockgrpp
, ext4_grpblk_t
*offsetp
);
1580 * Timeout and state flag for lazy initialization inode thread.
1582 #define EXT4_DEF_LI_WAIT_MULT 10
1583 #define EXT4_DEF_LI_MAX_START_DELAY 5
1584 #define EXT4_LAZYINIT_QUIT 0x0001
1585 #define EXT4_LAZYINIT_RUNNING 0x0002
1588 * Lazy inode table initialization info
1590 struct ext4_lazy_init
{
1591 unsigned long li_state
;
1593 wait_queue_head_t li_wait_daemon
;
1594 wait_queue_head_t li_wait_task
;
1595 struct timer_list li_timer
;
1596 struct task_struct
*li_task
;
1598 struct list_head li_request_list
;
1599 struct mutex li_list_mtx
;
1602 struct ext4_li_request
{
1603 struct super_block
*lr_super
;
1604 struct ext4_sb_info
*lr_sbi
;
1605 ext4_group_t lr_next_group
;
1606 struct list_head lr_request
;
1607 unsigned long lr_next_sched
;
1608 unsigned long lr_timeout
;
1611 struct ext4_features
{
1612 struct kobject f_kobj
;
1613 struct completion f_kobj_unregister
;
1617 * Function prototypes
1621 * Ok, these declarations are also in <linux/kernel.h> but none of the
1622 * ext4 source programs needs to include it so they are duplicated here.
1624 # define NORET_TYPE /**/
1625 # define ATTRIB_NORET __attribute__((noreturn))
1626 # define NORET_AND noreturn,
1629 extern unsigned int ext4_count_free(struct buffer_head
*, unsigned);
1632 extern unsigned int ext4_block_group(struct super_block
*sb
,
1633 ext4_fsblk_t blocknr
);
1634 extern ext4_grpblk_t
ext4_block_group_offset(struct super_block
*sb
,
1635 ext4_fsblk_t blocknr
);
1636 extern int ext4_bg_has_super(struct super_block
*sb
, ext4_group_t group
);
1637 extern unsigned long ext4_bg_num_gdb(struct super_block
*sb
,
1638 ext4_group_t group
);
1639 extern ext4_fsblk_t
ext4_new_meta_blocks(handle_t
*handle
, struct inode
*inode
,
1640 ext4_fsblk_t goal
, unsigned long *count
, int *errp
);
1641 extern int ext4_claim_free_blocks(struct ext4_sb_info
*sbi
, s64 nblocks
);
1642 extern void ext4_add_groupblocks(handle_t
*handle
, struct super_block
*sb
,
1643 ext4_fsblk_t block
, unsigned long count
);
1644 extern ext4_fsblk_t
ext4_count_free_blocks(struct super_block
*);
1645 extern void ext4_check_blocks_bitmap(struct super_block
*);
1646 extern struct ext4_group_desc
* ext4_get_group_desc(struct super_block
* sb
,
1647 ext4_group_t block_group
,
1648 struct buffer_head
** bh
);
1649 extern int ext4_should_retry_alloc(struct super_block
*sb
, int *retries
);
1650 struct buffer_head
*ext4_read_block_bitmap(struct super_block
*sb
,
1651 ext4_group_t block_group
);
1652 extern unsigned ext4_init_block_bitmap(struct super_block
*sb
,
1653 struct buffer_head
*bh
,
1655 struct ext4_group_desc
*desc
);
1656 #define ext4_free_blocks_after_init(sb, group, desc) \
1657 ext4_init_block_bitmap(sb, NULL, group, desc)
1660 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode
*,
1662 struct ext4_dir_entry_2
*,
1663 struct buffer_head
*, unsigned int);
1664 #define ext4_check_dir_entry(dir, filp, de, bh, offset) \
1665 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
1666 (de), (bh), (offset)))
1667 extern int ext4_htree_store_dirent(struct file
*dir_file
, __u32 hash
,
1669 struct ext4_dir_entry_2
*dirent
);
1670 extern void ext4_htree_free_dir_info(struct dir_private_info
*p
);
1673 extern int ext4_sync_file(struct file
*, int);
1674 extern int ext4_flush_completed_IO(struct inode
*);
1677 extern int ext4fs_dirhash(const char *name
, int len
, struct
1678 dx_hash_info
*hinfo
);
1681 extern struct inode
*ext4_new_inode(handle_t
*, struct inode
*, int,
1682 const struct qstr
*qstr
, __u32 goal
);
1683 extern void ext4_free_inode(handle_t
*, struct inode
*);
1684 extern struct inode
* ext4_orphan_get(struct super_block
*, unsigned long);
1685 extern unsigned long ext4_count_free_inodes(struct super_block
*);
1686 extern unsigned long ext4_count_dirs(struct super_block
*);
1687 extern void ext4_check_inodes_bitmap(struct super_block
*);
1688 extern void ext4_mark_bitmap_end(int start_bit
, int end_bit
, char *bitmap
);
1689 extern int ext4_init_inode_table(struct super_block
*sb
,
1690 ext4_group_t group
, int barrier
);
1693 extern long ext4_mb_stats
;
1694 extern long ext4_mb_max_to_scan
;
1695 extern int ext4_mb_init(struct super_block
*, int);
1696 extern int ext4_mb_release(struct super_block
*);
1697 extern ext4_fsblk_t
ext4_mb_new_blocks(handle_t
*,
1698 struct ext4_allocation_request
*, int *);
1699 extern int ext4_mb_reserve_blocks(struct super_block
*, int);
1700 extern void ext4_discard_preallocations(struct inode
*);
1701 extern int __init
ext4_init_mballoc(void);
1702 extern void ext4_exit_mballoc(void);
1703 extern void ext4_free_blocks(handle_t
*handle
, struct inode
*inode
,
1704 struct buffer_head
*bh
, ext4_fsblk_t block
,
1705 unsigned long count
, int flags
);
1706 extern int ext4_mb_add_groupinfo(struct super_block
*sb
,
1707 ext4_group_t i
, struct ext4_group_desc
*desc
);
1708 extern int ext4_trim_fs(struct super_block
*, struct fstrim_range
*);
1711 struct buffer_head
*ext4_getblk(handle_t
*, struct inode
*,
1712 ext4_lblk_t
, int, int *);
1713 struct buffer_head
*ext4_bread(handle_t
*, struct inode
*,
1714 ext4_lblk_t
, int, int *);
1715 int ext4_get_block(struct inode
*inode
, sector_t iblock
,
1716 struct buffer_head
*bh_result
, int create
);
1718 extern struct inode
*ext4_iget(struct super_block
*, unsigned long);
1719 extern int ext4_write_inode(struct inode
*, struct writeback_control
*);
1720 extern int ext4_setattr(struct dentry
*, struct iattr
*);
1721 extern int ext4_getattr(struct vfsmount
*mnt
, struct dentry
*dentry
,
1722 struct kstat
*stat
);
1723 extern void ext4_evict_inode(struct inode
*);
1724 extern void ext4_clear_inode(struct inode
*);
1725 extern int ext4_sync_inode(handle_t
*, struct inode
*);
1726 extern void ext4_dirty_inode(struct inode
*);
1727 extern int ext4_change_inode_journal_flag(struct inode
*, int);
1728 extern int ext4_get_inode_loc(struct inode
*, struct ext4_iloc
*);
1729 extern int ext4_can_truncate(struct inode
*inode
);
1730 extern void ext4_truncate(struct inode
*);
1731 extern int ext4_truncate_restart_trans(handle_t
*, struct inode
*, int nblocks
);
1732 extern void ext4_set_inode_flags(struct inode
*);
1733 extern void ext4_get_inode_flags(struct ext4_inode_info
*);
1734 extern int ext4_alloc_da_blocks(struct inode
*inode
);
1735 extern void ext4_set_aops(struct inode
*inode
);
1736 extern int ext4_writepage_trans_blocks(struct inode
*);
1737 extern int ext4_chunk_trans_blocks(struct inode
*, int nrblocks
);
1738 extern int ext4_block_truncate_page(handle_t
*handle
,
1739 struct address_space
*mapping
, loff_t from
);
1740 extern int ext4_page_mkwrite(struct vm_area_struct
*vma
, struct vm_fault
*vmf
);
1741 extern qsize_t
*ext4_get_reserved_space(struct inode
*inode
);
1742 extern void ext4_da_update_reserve_space(struct inode
*inode
,
1743 int used
, int quota_claim
);
1745 extern long ext4_ioctl(struct file
*, unsigned int, unsigned long);
1746 extern long ext4_compat_ioctl(struct file
*, unsigned int, unsigned long);
1749 extern int ext4_ext_migrate(struct inode
*);
1752 extern int ext4_orphan_add(handle_t
*, struct inode
*);
1753 extern int ext4_orphan_del(handle_t
*, struct inode
*);
1754 extern int ext4_htree_fill_tree(struct file
*dir_file
, __u32 start_hash
,
1755 __u32 start_minor_hash
, __u32
*next_hash
);
1758 extern int ext4_group_add(struct super_block
*sb
,
1759 struct ext4_new_group_data
*input
);
1760 extern int ext4_group_extend(struct super_block
*sb
,
1761 struct ext4_super_block
*es
,
1762 ext4_fsblk_t n_blocks_count
);
1765 extern void __ext4_error(struct super_block
*, const char *, unsigned int,
1767 __attribute__ ((format (printf
, 4, 5)));
1768 #define ext4_error(sb, message...) __ext4_error(sb, __func__, \
1769 __LINE__, ## message)
1770 extern void ext4_error_inode(struct inode
*, const char *, unsigned int,
1771 ext4_fsblk_t
, const char *, ...)
1772 __attribute__ ((format (printf
, 5, 6)));
1773 extern void ext4_error_file(struct file
*, const char *, unsigned int,
1774 ext4_fsblk_t
, const char *, ...)
1775 __attribute__ ((format (printf
, 5, 6)));
1776 extern void __ext4_std_error(struct super_block
*, const char *,
1778 extern void __ext4_abort(struct super_block
*, const char *, unsigned int,
1780 __attribute__ ((format (printf
, 4, 5)));
1781 #define ext4_abort(sb, message...) __ext4_abort(sb, __func__, \
1782 __LINE__, ## message)
1783 extern void __ext4_warning(struct super_block
*, const char *, unsigned int,
1785 __attribute__ ((format (printf
, 4, 5)));
1786 #define ext4_warning(sb, message...) __ext4_warning(sb, __func__, \
1787 __LINE__, ## message)
1788 extern void ext4_msg(struct super_block
*, const char *, const char *, ...)
1789 __attribute__ ((format (printf
, 3, 4)));
1790 extern void __ext4_grp_locked_error(const char *, unsigned int, \
1791 struct super_block
*, ext4_group_t
, \
1792 unsigned long, ext4_fsblk_t
, \
1794 __attribute__ ((format (printf
, 7, 8)));
1795 #define ext4_grp_locked_error(sb, grp, message...) \
1796 __ext4_grp_locked_error(__func__, __LINE__, (sb), (grp), ## message)
1797 extern void ext4_update_dynamic_rev(struct super_block
*sb
);
1798 extern int ext4_update_compat_feature(handle_t
*handle
, struct super_block
*sb
,
1800 extern int ext4_update_rocompat_feature(handle_t
*handle
,
1801 struct super_block
*sb
, __u32 rocompat
);
1802 extern int ext4_update_incompat_feature(handle_t
*handle
,
1803 struct super_block
*sb
, __u32 incompat
);
1804 extern ext4_fsblk_t
ext4_block_bitmap(struct super_block
*sb
,
1805 struct ext4_group_desc
*bg
);
1806 extern ext4_fsblk_t
ext4_inode_bitmap(struct super_block
*sb
,
1807 struct ext4_group_desc
*bg
);
1808 extern ext4_fsblk_t
ext4_inode_table(struct super_block
*sb
,
1809 struct ext4_group_desc
*bg
);
1810 extern __u32
ext4_free_blks_count(struct super_block
*sb
,
1811 struct ext4_group_desc
*bg
);
1812 extern __u32
ext4_free_inodes_count(struct super_block
*sb
,
1813 struct ext4_group_desc
*bg
);
1814 extern __u32
ext4_used_dirs_count(struct super_block
*sb
,
1815 struct ext4_group_desc
*bg
);
1816 extern __u32
ext4_itable_unused_count(struct super_block
*sb
,
1817 struct ext4_group_desc
*bg
);
1818 extern void ext4_block_bitmap_set(struct super_block
*sb
,
1819 struct ext4_group_desc
*bg
, ext4_fsblk_t blk
);
1820 extern void ext4_inode_bitmap_set(struct super_block
*sb
,
1821 struct ext4_group_desc
*bg
, ext4_fsblk_t blk
);
1822 extern void ext4_inode_table_set(struct super_block
*sb
,
1823 struct ext4_group_desc
*bg
, ext4_fsblk_t blk
);
1824 extern void ext4_free_blks_set(struct super_block
*sb
,
1825 struct ext4_group_desc
*bg
, __u32 count
);
1826 extern void ext4_free_inodes_set(struct super_block
*sb
,
1827 struct ext4_group_desc
*bg
, __u32 count
);
1828 extern void ext4_used_dirs_set(struct super_block
*sb
,
1829 struct ext4_group_desc
*bg
, __u32 count
);
1830 extern void ext4_itable_unused_set(struct super_block
*sb
,
1831 struct ext4_group_desc
*bg
, __u32 count
);
1832 extern __le16
ext4_group_desc_csum(struct ext4_sb_info
*sbi
, __u32 group
,
1833 struct ext4_group_desc
*gdp
);
1834 extern int ext4_group_desc_csum_verify(struct ext4_sb_info
*sbi
, __u32 group
,
1835 struct ext4_group_desc
*gdp
);
1837 static inline ext4_fsblk_t
ext4_blocks_count(struct ext4_super_block
*es
)
1839 return ((ext4_fsblk_t
)le32_to_cpu(es
->s_blocks_count_hi
) << 32) |
1840 le32_to_cpu(es
->s_blocks_count_lo
);
1843 static inline ext4_fsblk_t
ext4_r_blocks_count(struct ext4_super_block
*es
)
1845 return ((ext4_fsblk_t
)le32_to_cpu(es
->s_r_blocks_count_hi
) << 32) |
1846 le32_to_cpu(es
->s_r_blocks_count_lo
);
1849 static inline ext4_fsblk_t
ext4_free_blocks_count(struct ext4_super_block
*es
)
1851 return ((ext4_fsblk_t
)le32_to_cpu(es
->s_free_blocks_count_hi
) << 32) |
1852 le32_to_cpu(es
->s_free_blocks_count_lo
);
1855 static inline void ext4_blocks_count_set(struct ext4_super_block
*es
,
1858 es
->s_blocks_count_lo
= cpu_to_le32((u32
)blk
);
1859 es
->s_blocks_count_hi
= cpu_to_le32(blk
>> 32);
1862 static inline void ext4_free_blocks_count_set(struct ext4_super_block
*es
,
1865 es
->s_free_blocks_count_lo
= cpu_to_le32((u32
)blk
);
1866 es
->s_free_blocks_count_hi
= cpu_to_le32(blk
>> 32);
1869 static inline void ext4_r_blocks_count_set(struct ext4_super_block
*es
,
1872 es
->s_r_blocks_count_lo
= cpu_to_le32((u32
)blk
);
1873 es
->s_r_blocks_count_hi
= cpu_to_le32(blk
>> 32);
1876 static inline loff_t
ext4_isize(struct ext4_inode
*raw_inode
)
1878 if (S_ISREG(le16_to_cpu(raw_inode
->i_mode
)))
1879 return ((loff_t
)le32_to_cpu(raw_inode
->i_size_high
) << 32) |
1880 le32_to_cpu(raw_inode
->i_size_lo
);
1882 return (loff_t
) le32_to_cpu(raw_inode
->i_size_lo
);
1885 static inline void ext4_isize_set(struct ext4_inode
*raw_inode
, loff_t i_size
)
1887 raw_inode
->i_size_lo
= cpu_to_le32(i_size
);
1888 raw_inode
->i_size_high
= cpu_to_le32(i_size
>> 32);
1892 struct ext4_group_info
*ext4_get_group_info(struct super_block
*sb
,
1895 struct ext4_group_info
***grp_info
;
1896 long indexv
, indexh
;
1897 grp_info
= EXT4_SB(sb
)->s_group_info
;
1898 indexv
= group
>> (EXT4_DESC_PER_BLOCK_BITS(sb
));
1899 indexh
= group
& ((EXT4_DESC_PER_BLOCK(sb
)) - 1);
1900 return grp_info
[indexv
][indexh
];
1904 * Reading s_groups_count requires using smp_rmb() afterwards. See
1905 * the locking protocol documented in the comments of ext4_group_add()
1908 static inline ext4_group_t
ext4_get_groups_count(struct super_block
*sb
)
1910 ext4_group_t ngroups
= EXT4_SB(sb
)->s_groups_count
;
1916 static inline ext4_group_t
ext4_flex_group(struct ext4_sb_info
*sbi
,
1917 ext4_group_t block_group
)
1919 return block_group
>> sbi
->s_log_groups_per_flex
;
1922 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info
*sbi
)
1924 return 1 << sbi
->s_log_groups_per_flex
;
1927 #define ext4_std_error(sb, errno) \
1930 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
1934 /* Each CPU can accumulate percpu_counter_batch blocks in their local
1935 * counters. So we need to make sure we have free blocks more
1936 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
1938 #define EXT4_FREEBLOCKS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
1940 #define EXT4_FREEBLOCKS_WATERMARK 0
1943 static inline void ext4_update_i_disksize(struct inode
*inode
, loff_t newsize
)
1946 * XXX: replace with spinlock if seen contended -bzzz
1948 down_write(&EXT4_I(inode
)->i_data_sem
);
1949 if (newsize
> EXT4_I(inode
)->i_disksize
)
1950 EXT4_I(inode
)->i_disksize
= newsize
;
1951 up_write(&EXT4_I(inode
)->i_data_sem
);
1955 struct ext4_group_info
{
1956 unsigned long bb_state
;
1957 struct rb_root bb_free_root
;
1958 ext4_grpblk_t bb_first_free
; /* first free block */
1959 ext4_grpblk_t bb_free
; /* total free blocks */
1960 ext4_grpblk_t bb_fragments
; /* nr of freespace fragments */
1961 ext4_grpblk_t bb_largest_free_order
;/* order of largest frag in BG */
1962 struct list_head bb_prealloc_list
;
1966 struct rw_semaphore alloc_sem
;
1967 ext4_grpblk_t bb_counters
[]; /* Nr of free power-of-two-block
1968 * regions, index is order.
1969 * bb_counters[3] = 5 means
1970 * 5 free 8-block regions. */
1973 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
1975 #define EXT4_MB_GRP_NEED_INIT(grp) \
1976 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
1978 #define EXT4_MAX_CONTENTION 8
1979 #define EXT4_CONTENTION_THRESHOLD 2
1981 static inline spinlock_t
*ext4_group_lock_ptr(struct super_block
*sb
,
1984 return bgl_lock_ptr(EXT4_SB(sb
)->s_blockgroup_lock
, group
);
1988 * Returns true if the filesystem is busy enough that attempts to
1989 * access the block group locks has run into contention.
1991 static inline int ext4_fs_is_busy(struct ext4_sb_info
*sbi
)
1993 return (atomic_read(&sbi
->s_lock_busy
) > EXT4_CONTENTION_THRESHOLD
);
1996 static inline void ext4_lock_group(struct super_block
*sb
, ext4_group_t group
)
1998 spinlock_t
*lock
= ext4_group_lock_ptr(sb
, group
);
1999 if (spin_trylock(lock
))
2001 * We're able to grab the lock right away, so drop the
2002 * lock contention counter.
2004 atomic_add_unless(&EXT4_SB(sb
)->s_lock_busy
, -1, 0);
2007 * The lock is busy, so bump the contention counter,
2008 * and then wait on the spin lock.
2010 atomic_add_unless(&EXT4_SB(sb
)->s_lock_busy
, 1,
2011 EXT4_MAX_CONTENTION
);
2016 static inline void ext4_unlock_group(struct super_block
*sb
,
2019 spin_unlock(ext4_group_lock_ptr(sb
, group
));
2022 static inline void ext4_mark_super_dirty(struct super_block
*sb
)
2024 if (EXT4_SB(sb
)->s_journal
== NULL
)
2029 * Inodes and files operations
2033 extern const struct file_operations ext4_dir_operations
;
2036 extern const struct inode_operations ext4_file_inode_operations
;
2037 extern const struct file_operations ext4_file_operations
;
2038 extern loff_t
ext4_llseek(struct file
*file
, loff_t offset
, int origin
);
2041 extern const struct inode_operations ext4_dir_inode_operations
;
2042 extern const struct inode_operations ext4_special_inode_operations
;
2043 extern struct dentry
*ext4_get_parent(struct dentry
*child
);
2046 extern const struct inode_operations ext4_symlink_inode_operations
;
2047 extern const struct inode_operations ext4_fast_symlink_inode_operations
;
2049 /* block_validity */
2050 extern void ext4_release_system_zone(struct super_block
*sb
);
2051 extern int ext4_setup_system_zone(struct super_block
*sb
);
2052 extern int __init
ext4_init_system_zone(void);
2053 extern void ext4_exit_system_zone(void);
2054 extern int ext4_data_block_valid(struct ext4_sb_info
*sbi
,
2055 ext4_fsblk_t start_blk
,
2056 unsigned int count
);
2059 extern int ext4_ext_tree_init(handle_t
*handle
, struct inode
*);
2060 extern int ext4_ext_writepage_trans_blocks(struct inode
*, int);
2061 extern int ext4_ext_index_trans_blocks(struct inode
*inode
, int nrblocks
,
2063 extern int ext4_ext_map_blocks(handle_t
*handle
, struct inode
*inode
,
2064 struct ext4_map_blocks
*map
, int flags
);
2065 extern void ext4_ext_truncate(struct inode
*);
2066 extern void ext4_ext_init(struct super_block
*);
2067 extern void ext4_ext_release(struct super_block
*);
2068 extern long ext4_fallocate(struct inode
*inode
, int mode
, loff_t offset
,
2070 extern int ext4_convert_unwritten_extents(struct inode
*inode
, loff_t offset
,
2072 extern int ext4_map_blocks(handle_t
*handle
, struct inode
*inode
,
2073 struct ext4_map_blocks
*map
, int flags
);
2074 extern int ext4_fiemap(struct inode
*inode
, struct fiemap_extent_info
*fieinfo
,
2075 __u64 start
, __u64 len
);
2077 extern int ext4_move_extents(struct file
*o_filp
, struct file
*d_filp
,
2078 __u64 start_orig
, __u64 start_donor
,
2079 __u64 len
, __u64
*moved_len
);
2082 extern int __init
ext4_init_pageio(void);
2083 extern void ext4_exit_pageio(void);
2084 extern void ext4_ioend_wait(struct inode
*);
2085 extern void ext4_free_io_end(ext4_io_end_t
*io
);
2086 extern ext4_io_end_t
*ext4_init_io_end(struct inode
*inode
, gfp_t flags
);
2087 extern int ext4_end_io_nolock(ext4_io_end_t
*io
);
2088 extern void ext4_io_submit(struct ext4_io_submit
*io
);
2089 extern int ext4_bio_write_page(struct ext4_io_submit
*io
,
2092 struct writeback_control
*wbc
);
2094 /* BH_Uninit flag: blocks are allocated but uninitialized on disk */
2095 enum ext4_state_bits
{
2096 BH_Uninit
/* blocks are allocated but uninitialized on disk */
2097 = BH_JBDPrivateStart
,
2100 BUFFER_FNS(Uninit
, uninit
)
2101 TAS_BUFFER_FNS(Uninit
, uninit
)
2104 * Add new method to test wether block and inode bitmaps are properly
2105 * initialized. With uninit_bg reading the block from disk is not enough
2106 * to mark the bitmap uptodate. We need to also zero-out the bitmap
2108 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
2110 static inline int bitmap_uptodate(struct buffer_head
*bh
)
2112 return (buffer_uptodate(bh
) &&
2113 test_bit(BH_BITMAP_UPTODATE
, &(bh
)->b_state
));
2115 static inline void set_bitmap_uptodate(struct buffer_head
*bh
)
2117 set_bit(BH_BITMAP_UPTODATE
, &(bh
)->b_state
);
2120 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
2122 #endif /* __KERNEL__ */
2124 #endif /* _EXT4_H */