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
111 /* Use reserved root blocks if needed */
112 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000
114 struct ext4_allocation_request
{
115 /* target inode for block we're allocating */
117 /* how many blocks we want to allocate */
119 /* logical block in target inode */
121 /* the closest logical allocated block to the left */
123 /* the closest logical allocated block to the right */
125 /* phys. target (a hint) */
127 /* phys. block for the closest logical allocated block to the left */
129 /* phys. block for the closest logical allocated block to the right */
131 /* flags. see above EXT4_MB_HINT_* */
136 * Logical to physical block mapping, used by ext4_map_blocks()
138 * This structure is used to pass requests into ext4_map_blocks() as
139 * well as to store the information returned by ext4_map_blocks(). It
140 * takes less room on the stack than a struct buffer_head.
142 #define EXT4_MAP_NEW (1 << BH_New)
143 #define EXT4_MAP_MAPPED (1 << BH_Mapped)
144 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten)
145 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary)
146 #define EXT4_MAP_UNINIT (1 << BH_Uninit)
147 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
148 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\
151 struct ext4_map_blocks
{
155 unsigned int m_flags
;
159 * For delayed allocation tracking
161 struct mpage_da_data
{
163 sector_t b_blocknr
; /* start block number of extent */
164 size_t b_size
; /* size of extent */
165 unsigned long b_state
; /* state of the extent */
166 unsigned long first_page
, next_page
; /* extent of pages */
167 struct writeback_control
*wbc
;
174 * Flags for ext4_io_end->flags
176 #define EXT4_IO_END_UNWRITTEN 0x0001
177 #define EXT4_IO_END_ERROR 0x0002
178 #define EXT4_IO_END_QUEUED 0x0004
180 struct ext4_io_page
{
185 #define MAX_IO_PAGES 128
187 typedef struct ext4_io_end
{
188 struct list_head list
; /* per-file finished IO list */
189 struct inode
*inode
; /* file being written to */
190 unsigned int flag
; /* unwritten or not */
191 struct page
*page
; /* page struct for buffer write */
192 loff_t offset
; /* offset in the file */
193 ssize_t size
; /* size of the extent */
194 struct work_struct work
; /* data work queue */
195 struct kiocb
*iocb
; /* iocb struct for AIO */
196 int result
; /* error value for AIO */
198 struct ext4_io_page
*pages
[MAX_IO_PAGES
];
201 struct ext4_io_submit
{
204 ext4_io_end_t
*io_end
;
205 struct ext4_io_page
*io_page
;
206 sector_t io_next_block
;
210 * Special inodes numbers
212 #define EXT4_BAD_INO 1 /* Bad blocks inode */
213 #define EXT4_ROOT_INO 2 /* Root inode */
214 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */
215 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
216 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
217 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
218 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
219 #define EXT4_JOURNAL_INO 8 /* Journal inode */
221 /* First non-reserved inode for old ext4 filesystems */
222 #define EXT4_GOOD_OLD_FIRST_INO 11
225 * Maximal count of links to a file
227 #define EXT4_LINK_MAX 65000
230 * Macro-instructions used to manage several block sizes
232 #define EXT4_MIN_BLOCK_SIZE 1024
233 #define EXT4_MAX_BLOCK_SIZE 65536
234 #define EXT4_MIN_BLOCK_LOG_SIZE 10
235 #define EXT4_MAX_BLOCK_LOG_SIZE 16
237 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
239 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
241 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
243 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
245 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
248 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
249 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
250 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
252 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
253 EXT4_GOOD_OLD_INODE_SIZE : \
255 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
256 EXT4_GOOD_OLD_FIRST_INO : \
259 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
262 * Structure of a blocks group descriptor
264 struct ext4_group_desc
266 __le32 bg_block_bitmap_lo
; /* Blocks bitmap block */
267 __le32 bg_inode_bitmap_lo
; /* Inodes bitmap block */
268 __le32 bg_inode_table_lo
; /* Inodes table block */
269 __le16 bg_free_blocks_count_lo
;/* Free blocks count */
270 __le16 bg_free_inodes_count_lo
;/* Free inodes count */
271 __le16 bg_used_dirs_count_lo
; /* Directories count */
272 __le16 bg_flags
; /* EXT4_BG_flags (INODE_UNINIT, etc) */
273 __u32 bg_reserved
[2]; /* Likely block/inode bitmap checksum */
274 __le16 bg_itable_unused_lo
; /* Unused inodes count */
275 __le16 bg_checksum
; /* crc16(sb_uuid+group+desc) */
276 __le32 bg_block_bitmap_hi
; /* Blocks bitmap block MSB */
277 __le32 bg_inode_bitmap_hi
; /* Inodes bitmap block MSB */
278 __le32 bg_inode_table_hi
; /* Inodes table block MSB */
279 __le16 bg_free_blocks_count_hi
;/* Free blocks count MSB */
280 __le16 bg_free_inodes_count_hi
;/* Free inodes count MSB */
281 __le16 bg_used_dirs_count_hi
; /* Directories count MSB */
282 __le16 bg_itable_unused_hi
; /* Unused inodes count MSB */
283 __u32 bg_reserved2
[3];
287 * Structure of a flex block group info
291 atomic_t free_inodes
;
292 atomic_t free_blocks
;
296 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
297 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
298 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
301 * Macro-instructions used to manage group descriptors
303 #define EXT4_MIN_DESC_SIZE 32
304 #define EXT4_MIN_DESC_SIZE_64BIT 64
305 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
306 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
308 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
309 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
310 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
311 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
313 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
314 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
315 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
319 * Constants relative to the data blocks
321 #define EXT4_NDIR_BLOCKS 12
322 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
323 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
324 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
325 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
330 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
331 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
332 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
333 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
334 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
335 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
336 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
337 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
338 /* Reserved for compression usage... */
339 #define EXT4_DIRTY_FL 0x00000100
340 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
341 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
342 #define EXT4_ECOMPR_FL 0x00000800 /* Compression error */
343 /* End compression flags --- maybe not all used */
344 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
345 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
346 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
347 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
348 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
349 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
350 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
351 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
352 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
353 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
354 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
356 #define EXT4_FL_USER_VISIBLE 0x004BDFFF /* User visible flags */
357 #define EXT4_FL_USER_MODIFIABLE 0x004B80FF /* User modifiable flags */
359 /* Flags that should be inherited by new inodes from their parent. */
360 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
361 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
362 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
363 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
365 /* Flags that are appropriate for regular files (all but dir-specific ones). */
366 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
368 /* Flags that are appropriate for non-directories/regular files. */
369 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
371 /* Mask out flags that are inappropriate for the given type of inode. */
372 static inline __u32
ext4_mask_flags(umode_t mode
, __u32 flags
)
376 else if (S_ISREG(mode
))
377 return flags
& EXT4_REG_FLMASK
;
379 return flags
& EXT4_OTHER_FLMASK
;
383 * Inode flags used for atomic set/get
386 EXT4_INODE_SECRM
= 0, /* Secure deletion */
387 EXT4_INODE_UNRM
= 1, /* Undelete */
388 EXT4_INODE_COMPR
= 2, /* Compress file */
389 EXT4_INODE_SYNC
= 3, /* Synchronous updates */
390 EXT4_INODE_IMMUTABLE
= 4, /* Immutable file */
391 EXT4_INODE_APPEND
= 5, /* writes to file may only append */
392 EXT4_INODE_NODUMP
= 6, /* do not dump file */
393 EXT4_INODE_NOATIME
= 7, /* do not update atime */
394 /* Reserved for compression usage... */
395 EXT4_INODE_DIRTY
= 8,
396 EXT4_INODE_COMPRBLK
= 9, /* One or more compressed clusters */
397 EXT4_INODE_NOCOMPR
= 10, /* Don't compress */
398 EXT4_INODE_ECOMPR
= 11, /* Compression error */
399 /* End compression flags --- maybe not all used */
400 EXT4_INODE_INDEX
= 12, /* hash-indexed directory */
401 EXT4_INODE_IMAGIC
= 13, /* AFS directory */
402 EXT4_INODE_JOURNAL_DATA
= 14, /* file data should be journaled */
403 EXT4_INODE_NOTAIL
= 15, /* file tail should not be merged */
404 EXT4_INODE_DIRSYNC
= 16, /* dirsync behaviour (directories only) */
405 EXT4_INODE_TOPDIR
= 17, /* Top of directory hierarchies*/
406 EXT4_INODE_HUGE_FILE
= 18, /* Set to each huge file */
407 EXT4_INODE_EXTENTS
= 19, /* Inode uses extents */
408 EXT4_INODE_EA_INODE
= 21, /* Inode used for large EA */
409 EXT4_INODE_EOFBLOCKS
= 22, /* Blocks allocated beyond EOF */
410 EXT4_INODE_RESERVED
= 31, /* reserved for ext4 lib */
413 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
414 #define CHECK_FLAG_VALUE(FLAG) if (!TEST_FLAG_VALUE(FLAG)) { \
415 printk(KERN_EMERG "EXT4 flag fail: " #FLAG ": %d %d\n", \
416 EXT4_##FLAG##_FL, EXT4_INODE_##FLAG); BUG_ON(1); }
419 * Since it's pretty easy to mix up bit numbers and hex values, and we
420 * can't do a compile-time test for ENUM values, we use a run-time
421 * test to make sure that EXT4_XXX_FL is consistent with respect to
422 * EXT4_INODE_XXX. If all is well the printk and BUG_ON will all drop
423 * out so it won't cost any extra space in the compiled kernel image.
424 * But it's important that these values are the same, since we are
425 * using EXT4_INODE_XXX to test for the flag values, but EXT4_XX_FL
426 * must be consistent with the values of FS_XXX_FL defined in
427 * include/linux/fs.h and the on-disk values found in ext2, ext3, and
428 * ext4 filesystems, and of course the values defined in e2fsprogs.
430 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
432 static inline void ext4_check_flag_values(void)
434 CHECK_FLAG_VALUE(SECRM
);
435 CHECK_FLAG_VALUE(UNRM
);
436 CHECK_FLAG_VALUE(COMPR
);
437 CHECK_FLAG_VALUE(SYNC
);
438 CHECK_FLAG_VALUE(IMMUTABLE
);
439 CHECK_FLAG_VALUE(APPEND
);
440 CHECK_FLAG_VALUE(NODUMP
);
441 CHECK_FLAG_VALUE(NOATIME
);
442 CHECK_FLAG_VALUE(DIRTY
);
443 CHECK_FLAG_VALUE(COMPRBLK
);
444 CHECK_FLAG_VALUE(NOCOMPR
);
445 CHECK_FLAG_VALUE(ECOMPR
);
446 CHECK_FLAG_VALUE(INDEX
);
447 CHECK_FLAG_VALUE(IMAGIC
);
448 CHECK_FLAG_VALUE(JOURNAL_DATA
);
449 CHECK_FLAG_VALUE(NOTAIL
);
450 CHECK_FLAG_VALUE(DIRSYNC
);
451 CHECK_FLAG_VALUE(TOPDIR
);
452 CHECK_FLAG_VALUE(HUGE_FILE
);
453 CHECK_FLAG_VALUE(EXTENTS
);
454 CHECK_FLAG_VALUE(EA_INODE
);
455 CHECK_FLAG_VALUE(EOFBLOCKS
);
456 CHECK_FLAG_VALUE(RESERVED
);
459 /* Used to pass group descriptor data when online resize is done */
460 struct ext4_new_group_input
{
461 __u32 group
; /* Group number for this data */
462 __u64 block_bitmap
; /* Absolute block number of block bitmap */
463 __u64 inode_bitmap
; /* Absolute block number of inode bitmap */
464 __u64 inode_table
; /* Absolute block number of inode table start */
465 __u32 blocks_count
; /* Total number of blocks in this group */
466 __u16 reserved_blocks
; /* Number of reserved blocks in this group */
470 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
471 struct compat_ext4_new_group_input
{
473 compat_u64 block_bitmap
;
474 compat_u64 inode_bitmap
;
475 compat_u64 inode_table
;
482 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
483 struct ext4_new_group_data
{
489 __u16 reserved_blocks
;
491 __u32 free_blocks_count
;
495 * Flags used by ext4_map_blocks()
497 /* Allocate any needed blocks and/or convert an unitialized
498 extent to be an initialized ext4 */
499 #define EXT4_GET_BLOCKS_CREATE 0x0001
500 /* Request the creation of an unitialized extent */
501 #define EXT4_GET_BLOCKS_UNINIT_EXT 0x0002
502 #define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT (EXT4_GET_BLOCKS_UNINIT_EXT|\
503 EXT4_GET_BLOCKS_CREATE)
504 /* Caller is from the delayed allocation writeout path,
505 so set the magic i_delalloc_reserve_flag after taking the
506 inode allocation semaphore for */
507 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
508 /* caller is from the direct IO path, request to creation of an
509 unitialized extents if not allocated, split the uninitialized
510 extent if blocks has been preallocated already*/
511 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
512 #define EXT4_GET_BLOCKS_CONVERT 0x0010
513 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
514 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
515 /* Convert extent to initialized after IO complete */
516 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
517 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
518 /* Punch out blocks of an extent */
519 #define EXT4_GET_BLOCKS_PUNCH_OUT_EXT 0x0020
520 /* Don't normalize allocation size (used for fallocate) */
521 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
524 * Flags used by ext4_free_blocks
526 #define EXT4_FREE_BLOCKS_METADATA 0x0001
527 #define EXT4_FREE_BLOCKS_FORGET 0x0002
528 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
529 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
532 * Flags used by ext4_discard_partial_page_buffers
534 #define EXT4_DISCARD_PARTIAL_PG_ZERO_UNMAPPED 0x0001
539 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
540 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
541 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
542 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
543 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
544 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
545 #ifdef CONFIG_JBD2_DEBUG
546 #define EXT4_IOC_WAIT_FOR_READONLY _IOR('f', 99, long)
548 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
549 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
550 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
551 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
552 #define EXT4_IOC_MIGRATE _IO('f', 9)
553 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
554 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
555 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
556 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
558 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
560 * ioctl commands in 32 bit emulation
562 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
563 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
564 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
565 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
566 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
567 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
568 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
569 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
570 #ifdef CONFIG_JBD2_DEBUG
571 #define EXT4_IOC32_WAIT_FOR_READONLY _IOR('f', 99, int)
573 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
574 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
577 /* Max physical block we can address w/o extents */
578 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
581 * Structure of an inode on the disk
584 __le16 i_mode
; /* File mode */
585 __le16 i_uid
; /* Low 16 bits of Owner Uid */
586 __le32 i_size_lo
; /* Size in bytes */
587 __le32 i_atime
; /* Access time */
588 __le32 i_ctime
; /* Inode Change time */
589 __le32 i_mtime
; /* Modification time */
590 __le32 i_dtime
; /* Deletion Time */
591 __le16 i_gid
; /* Low 16 bits of Group Id */
592 __le16 i_links_count
; /* Links count */
593 __le32 i_blocks_lo
; /* Blocks count */
594 __le32 i_flags
; /* File flags */
600 __u32 h_i_translator
;
605 } osd1
; /* OS dependent 1 */
606 __le32 i_block
[EXT4_N_BLOCKS
];/* Pointers to blocks */
607 __le32 i_generation
; /* File version (for NFS) */
608 __le32 i_file_acl_lo
; /* File ACL */
610 __le32 i_obso_faddr
; /* Obsoleted fragment address */
613 __le16 l_i_blocks_high
; /* were l_i_reserved1 */
614 __le16 l_i_file_acl_high
;
615 __le16 l_i_uid_high
; /* these 2 fields */
616 __le16 l_i_gid_high
; /* were reserved2[0] */
620 __le16 h_i_reserved1
; /* Obsoleted fragment number/size which are removed in ext4 */
627 __le16 h_i_reserved1
; /* Obsoleted fragment number/size which are removed in ext4 */
628 __le16 m_i_file_acl_high
;
629 __u32 m_i_reserved2
[2];
631 } osd2
; /* OS dependent 2 */
632 __le16 i_extra_isize
;
634 __le32 i_ctime_extra
; /* extra Change time (nsec << 2 | epoch) */
635 __le32 i_mtime_extra
; /* extra Modification time(nsec << 2 | epoch) */
636 __le32 i_atime_extra
; /* extra Access time (nsec << 2 | epoch) */
637 __le32 i_crtime
; /* File Creation time */
638 __le32 i_crtime_extra
; /* extra FileCreationtime (nsec << 2 | epoch) */
639 __le32 i_version_hi
; /* high 32 bits for 64-bit version */
643 __u32 reserved
; /* should be zero */
644 __u32 donor_fd
; /* donor file descriptor */
645 __u64 orig_start
; /* logical start offset in block for orig */
646 __u64 donor_start
; /* logical start offset in block for donor */
647 __u64 len
; /* block length to be moved */
648 __u64 moved_len
; /* moved block length */
651 #define EXT4_EPOCH_BITS 2
652 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
653 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
656 * Extended fields will fit into an inode if the filesystem was formatted
657 * with large inodes (-I 256 or larger) and there are not currently any EAs
658 * consuming all of the available space. For new inodes we always reserve
659 * enough space for the kernel's known extended fields, but for inodes
660 * created with an old kernel this might not have been the case. None of
661 * the extended inode fields is critical for correct filesystem operation.
662 * This macro checks if a certain field fits in the inode. Note that
663 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
665 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
666 ((offsetof(typeof(*ext4_inode), field) + \
667 sizeof((ext4_inode)->field)) \
668 <= (EXT4_GOOD_OLD_INODE_SIZE + \
669 (einode)->i_extra_isize)) \
671 static inline __le32 ext4_encode_extra_time(struct timespec *time)
673 return cpu_to_le32((sizeof(time
->tv_sec
) > 4 ?
674 (time
->tv_sec
>> 32) & EXT4_EPOCH_MASK
: 0) |
675 ((time
->tv_nsec
<< EXT4_EPOCH_BITS
) & EXT4_NSEC_MASK
));
678 static inline void ext4_decode_extra_time(struct timespec
*time
, __le32 extra
)
680 if (sizeof(time
->tv_sec
) > 4)
681 time
->tv_sec
|= (__u64
)(le32_to_cpu(extra
) & EXT4_EPOCH_MASK
)
683 time
->tv_nsec
= (le32_to_cpu(extra
) & EXT4_NSEC_MASK
) >> EXT4_EPOCH_BITS
;
686 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
688 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
689 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
690 (raw_inode)->xtime ## _extra = \
691 ext4_encode_extra_time(&(inode)->xtime); \
694 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
696 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
697 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
698 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
699 (raw_inode)->xtime ## _extra = \
700 ext4_encode_extra_time(&(einode)->xtime); \
703 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
705 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
706 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
707 ext4_decode_extra_time(&(inode)->xtime, \
708 raw_inode->xtime ## _extra); \
710 (inode)->xtime.tv_nsec = 0; \
713 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
715 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
716 (einode)->xtime.tv_sec = \
717 (signed)le32_to_cpu((raw_inode)->xtime); \
718 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
719 ext4_decode_extra_time(&(einode)->xtime, \
720 raw_inode->xtime ## _extra); \
722 (einode)->xtime.tv_nsec = 0; \
725 #define i_disk_version osd1.linux1.l_i_version
727 #if defined(__KERNEL__) || defined(__linux__)
728 #define i_reserved1 osd1.linux1.l_i_reserved1
729 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
730 #define i_blocks_high osd2.linux2.l_i_blocks_high
731 #define i_uid_low i_uid
732 #define i_gid_low i_gid
733 #define i_uid_high osd2.linux2.l_i_uid_high
734 #define i_gid_high osd2.linux2.l_i_gid_high
735 #define i_reserved2 osd2.linux2.l_i_reserved2
737 #elif defined(__GNU__)
739 #define i_translator osd1.hurd1.h_i_translator
740 #define i_uid_high osd2.hurd2.h_i_uid_high
741 #define i_gid_high osd2.hurd2.h_i_gid_high
742 #define i_author osd2.hurd2.h_i_author
744 #elif defined(__masix__)
746 #define i_reserved1 osd1.masix1.m_i_reserved1
747 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
748 #define i_reserved2 osd2.masix2.m_i_reserved2
750 #endif /* defined(__KERNEL__) || defined(__linux__) */
753 * storage for cached extent
754 * If ec_len == 0, then the cache is invalid.
755 * If ec_start == 0, then the cache represents a gap (null mapping)
757 struct ext4_ext_cache
{
758 ext4_fsblk_t ec_start
;
759 ext4_lblk_t ec_block
;
760 __u32 ec_len
; /* must be 32bit to return holes */
764 * fourth extended file system inode data in memory
766 struct ext4_inode_info
{
767 __le32 i_data
[15]; /* unconverted */
769 ext4_fsblk_t i_file_acl
;
772 * i_block_group is the number of the block group which contains
773 * this file's inode. Constant across the lifetime of the inode,
774 * it is ued for making block allocation decisions - we try to
775 * place a file's data blocks near its inode block, and new inodes
776 * near to their parent directory's inode.
778 ext4_group_t i_block_group
;
779 ext4_lblk_t i_dir_start_lookup
;
780 #if (BITS_PER_LONG < 64)
781 unsigned long i_state_flags
; /* Dynamic state flags */
783 unsigned long i_flags
;
785 #ifdef CONFIG_EXT4_FS_XATTR
787 * Extended attributes can be read independently of the main file
788 * data. Taking i_mutex even when reading would cause contention
789 * between readers of EAs and writers of regular file data, so
790 * instead we synchronize on xattr_sem when reading or changing
793 struct rw_semaphore xattr_sem
;
796 struct list_head i_orphan
; /* unlinked but open inodes */
799 * i_disksize keeps track of what the inode size is ON DISK, not
800 * in memory. During truncate, i_size is set to the new size by
801 * the VFS prior to calling ext4_truncate(), but the filesystem won't
802 * set i_disksize to 0 until the truncate is actually under way.
804 * The intent is that i_disksize always represents the blocks which
805 * are used by this file. This allows recovery to restart truncate
806 * on orphans if we crash during truncate. We actually write i_disksize
807 * into the on-disk inode when writing inodes out, instead of i_size.
809 * The only time when i_disksize and i_size may be different is when
810 * a truncate is in progress. The only things which change i_disksize
811 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
816 * i_data_sem is for serialising ext4_truncate() against
817 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
818 * data tree are chopped off during truncate. We can't do that in
819 * ext4 because whenever we perform intermediate commits during
820 * truncate, the inode and all the metadata blocks *must* be in a
821 * consistent state which allows truncation of the orphans to restart
822 * during recovery. Hence we must fix the get_block-vs-truncate race
823 * by other means, so we have i_data_sem.
825 struct rw_semaphore i_data_sem
;
826 struct inode vfs_inode
;
827 struct jbd2_inode
*jinode
;
829 struct ext4_ext_cache i_cached_extent
;
831 * File creation time. Its function is same as that of
832 * struct timespec i_{a,c,m}time in the generic inode.
834 struct timespec i_crtime
;
837 struct list_head i_prealloc_list
;
838 spinlock_t i_prealloc_lock
;
841 ext4_group_t i_last_alloc_group
;
843 /* allocation reservation info for delalloc */
844 unsigned int i_reserved_data_blocks
;
845 unsigned int i_reserved_meta_blocks
;
846 unsigned int i_allocated_meta_blocks
;
847 ext4_lblk_t i_da_metadata_calc_last_lblock
;
848 int i_da_metadata_calc_len
;
850 /* on-disk additional length */
854 /* quota space reservation, managed internally by quota code */
855 qsize_t i_reserved_quota
;
858 /* completed IOs that might need unwritten extents handling */
859 struct list_head i_completed_io_list
;
860 spinlock_t i_completed_io_lock
;
861 atomic_t i_ioend_count
; /* Number of outstanding io_end structs */
862 /* current io_end structure for async DIO write*/
863 ext4_io_end_t
*cur_aio_dio
;
864 atomic_t i_aiodio_unwritten
; /* Nr. of inflight conversions pending */
866 spinlock_t i_block_reservation_lock
;
869 * Transactions that contain inode's metadata needed to complete
870 * fsync and fdatasync, respectively.
873 tid_t i_datasync_tid
;
879 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
880 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
881 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
884 * Misc. filesystem flags
886 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
887 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
888 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
893 #define EXT4_MOUNT_OLDALLOC 0x00002 /* Don't use the new Orlov allocator */
894 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
895 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
896 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
897 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
898 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
899 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
900 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
901 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
902 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
903 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
904 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
905 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
906 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
907 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
908 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
909 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
910 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
911 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
912 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
913 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
914 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
915 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
916 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
917 #define EXT4_MOUNT_I_VERSION 0x2000000 /* i_version support */
918 #define EXT4_MOUNT_MBLK_IO_SUBMIT 0x4000000 /* multi-block io submits */
919 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
920 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
921 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
922 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
923 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
925 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
926 specified delalloc */
928 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
930 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
932 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
935 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
937 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
939 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
942 #define ext4_set_bit __test_and_set_bit_le
943 #define ext4_set_bit_atomic ext2_set_bit_atomic
944 #define ext4_clear_bit __test_and_clear_bit_le
945 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
946 #define ext4_test_bit test_bit_le
947 #define ext4_find_first_zero_bit find_first_zero_bit_le
948 #define ext4_find_next_zero_bit find_next_zero_bit_le
949 #define ext4_find_next_bit find_next_bit_le
951 extern void ext4_set_bits(void *bm
, int cur
, int len
);
954 * Maximal mount counts between two filesystem checks
956 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
957 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
960 * Behaviour when detecting errors
962 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
963 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
964 #define EXT4_ERRORS_PANIC 3 /* Panic */
965 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
968 * Structure of the super block
970 struct ext4_super_block
{
971 /*00*/ __le32 s_inodes_count
; /* Inodes count */
972 __le32 s_blocks_count_lo
; /* Blocks count */
973 __le32 s_r_blocks_count_lo
; /* Reserved blocks count */
974 __le32 s_free_blocks_count_lo
; /* Free blocks count */
975 /*10*/ __le32 s_free_inodes_count
; /* Free inodes count */
976 __le32 s_first_data_block
; /* First Data Block */
977 __le32 s_log_block_size
; /* Block size */
978 __le32 s_obso_log_frag_size
; /* Obsoleted fragment size */
979 /*20*/ __le32 s_blocks_per_group
; /* # Blocks per group */
980 __le32 s_obso_frags_per_group
; /* Obsoleted fragments per group */
981 __le32 s_inodes_per_group
; /* # Inodes per group */
982 __le32 s_mtime
; /* Mount time */
983 /*30*/ __le32 s_wtime
; /* Write time */
984 __le16 s_mnt_count
; /* Mount count */
985 __le16 s_max_mnt_count
; /* Maximal mount count */
986 __le16 s_magic
; /* Magic signature */
987 __le16 s_state
; /* File system state */
988 __le16 s_errors
; /* Behaviour when detecting errors */
989 __le16 s_minor_rev_level
; /* minor revision level */
990 /*40*/ __le32 s_lastcheck
; /* time of last check */
991 __le32 s_checkinterval
; /* max. time between checks */
992 __le32 s_creator_os
; /* OS */
993 __le32 s_rev_level
; /* Revision level */
994 /*50*/ __le16 s_def_resuid
; /* Default uid for reserved blocks */
995 __le16 s_def_resgid
; /* Default gid for reserved blocks */
997 * These fields are for EXT4_DYNAMIC_REV superblocks only.
999 * Note: the difference between the compatible feature set and
1000 * the incompatible feature set is that if there is a bit set
1001 * in the incompatible feature set that the kernel doesn't
1002 * know about, it should refuse to mount the filesystem.
1004 * e2fsck's requirements are more strict; if it doesn't know
1005 * about a feature in either the compatible or incompatible
1006 * feature set, it must abort and not try to meddle with
1007 * things it doesn't understand...
1009 __le32 s_first_ino
; /* First non-reserved inode */
1010 __le16 s_inode_size
; /* size of inode structure */
1011 __le16 s_block_group_nr
; /* block group # of this superblock */
1012 __le32 s_feature_compat
; /* compatible feature set */
1013 /*60*/ __le32 s_feature_incompat
; /* incompatible feature set */
1014 __le32 s_feature_ro_compat
; /* readonly-compatible feature set */
1015 /*68*/ __u8 s_uuid
[16]; /* 128-bit uuid for volume */
1016 /*78*/ char s_volume_name
[16]; /* volume name */
1017 /*88*/ char s_last_mounted
[64]; /* directory where last mounted */
1018 /*C8*/ __le32 s_algorithm_usage_bitmap
; /* For compression */
1020 * Performance hints. Directory preallocation should only
1021 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1023 __u8 s_prealloc_blocks
; /* Nr of blocks to try to preallocate*/
1024 __u8 s_prealloc_dir_blocks
; /* Nr to preallocate for dirs */
1025 __le16 s_reserved_gdt_blocks
; /* Per group desc for online growth */
1027 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1029 /*D0*/ __u8 s_journal_uuid
[16]; /* uuid of journal superblock */
1030 /*E0*/ __le32 s_journal_inum
; /* inode number of journal file */
1031 __le32 s_journal_dev
; /* device number of journal file */
1032 __le32 s_last_orphan
; /* start of list of inodes to delete */
1033 __le32 s_hash_seed
[4]; /* HTREE hash seed */
1034 __u8 s_def_hash_version
; /* Default hash version to use */
1035 __u8 s_jnl_backup_type
;
1036 __le16 s_desc_size
; /* size of group descriptor */
1037 /*100*/ __le32 s_default_mount_opts
;
1038 __le32 s_first_meta_bg
; /* First metablock block group */
1039 __le32 s_mkfs_time
; /* When the filesystem was created */
1040 __le32 s_jnl_blocks
[17]; /* Backup of the journal inode */
1041 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1042 /*150*/ __le32 s_blocks_count_hi
; /* Blocks count */
1043 __le32 s_r_blocks_count_hi
; /* Reserved blocks count */
1044 __le32 s_free_blocks_count_hi
; /* Free blocks count */
1045 __le16 s_min_extra_isize
; /* All inodes have at least # bytes */
1046 __le16 s_want_extra_isize
; /* New inodes should reserve # bytes */
1047 __le32 s_flags
; /* Miscellaneous flags */
1048 __le16 s_raid_stride
; /* RAID stride */
1049 __le16 s_mmp_update_interval
; /* # seconds to wait in MMP checking */
1050 __le64 s_mmp_block
; /* Block for multi-mount protection */
1051 __le32 s_raid_stripe_width
; /* blocks on all data disks (N*stride)*/
1052 __u8 s_log_groups_per_flex
; /* FLEX_BG group size */
1053 __u8 s_reserved_char_pad
;
1054 __le16 s_reserved_pad
;
1055 __le64 s_kbytes_written
; /* nr of lifetime kilobytes written */
1056 __le32 s_snapshot_inum
; /* Inode number of active snapshot */
1057 __le32 s_snapshot_id
; /* sequential ID of active snapshot */
1058 __le64 s_snapshot_r_blocks_count
; /* reserved blocks for active
1059 snapshot's future use */
1060 __le32 s_snapshot_list
; /* inode number of the head of the
1061 on-disk snapshot list */
1062 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1063 __le32 s_error_count
; /* number of fs errors */
1064 __le32 s_first_error_time
; /* first time an error happened */
1065 __le32 s_first_error_ino
; /* inode involved in first error */
1066 __le64 s_first_error_block
; /* block involved of first error */
1067 __u8 s_first_error_func
[32]; /* function where the error happened */
1068 __le32 s_first_error_line
; /* line number where error happened */
1069 __le32 s_last_error_time
; /* most recent time of an error */
1070 __le32 s_last_error_ino
; /* inode involved in last error */
1071 __le32 s_last_error_line
; /* line number where error happened */
1072 __le64 s_last_error_block
; /* block involved of last error */
1073 __u8 s_last_error_func
[32]; /* function where the error happened */
1074 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1075 __u8 s_mount_opts
[64];
1076 __le32 s_reserved
[112]; /* Padding to the end of the block */
1079 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1084 * run-time mount flags
1086 #define EXT4_MF_MNTDIR_SAMPLED 0x0001
1087 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1090 * fourth extended-fs super-block data in memory
1092 struct ext4_sb_info
{
1093 unsigned long s_desc_size
; /* Size of a group descriptor in bytes */
1094 unsigned long s_inodes_per_block
;/* Number of inodes per block */
1095 unsigned long s_blocks_per_group
;/* Number of blocks in a group */
1096 unsigned long s_inodes_per_group
;/* Number of inodes in a group */
1097 unsigned long s_itb_per_group
; /* Number of inode table blocks per group */
1098 unsigned long s_gdb_count
; /* Number of group descriptor blocks */
1099 unsigned long s_desc_per_block
; /* Number of group descriptors per block */
1100 ext4_group_t s_groups_count
; /* Number of groups in the fs */
1101 ext4_group_t s_blockfile_groups
;/* Groups acceptable for non-extent files */
1102 unsigned long s_overhead_last
; /* Last calculated overhead */
1103 unsigned long s_blocks_last
; /* Last seen block count */
1104 loff_t s_bitmap_maxbytes
; /* max bytes for bitmap files */
1105 struct buffer_head
* s_sbh
; /* Buffer containing the super block */
1106 struct ext4_super_block
*s_es
; /* Pointer to the super block in the buffer */
1107 struct buffer_head
**s_group_desc
;
1108 unsigned int s_mount_opt
;
1109 unsigned int s_mount_opt2
;
1110 unsigned int s_mount_flags
;
1111 ext4_fsblk_t s_sb_block
;
1114 unsigned short s_mount_state
;
1115 unsigned short s_pad
;
1116 int s_addr_per_block_bits
;
1117 int s_desc_per_block_bits
;
1120 unsigned int s_inode_readahead_blks
;
1121 unsigned int s_inode_goal
;
1122 spinlock_t s_next_gen_lock
;
1123 u32 s_next_generation
;
1125 int s_def_hash_version
;
1126 int s_hash_unsigned
; /* 3 if hash should be signed, 0 if not */
1127 struct percpu_counter s_freeblocks_counter
;
1128 struct percpu_counter s_freeinodes_counter
;
1129 struct percpu_counter s_dirs_counter
;
1130 struct percpu_counter s_dirtyblocks_counter
;
1131 struct blockgroup_lock
*s_blockgroup_lock
;
1132 struct proc_dir_entry
*s_proc
;
1133 struct kobject s_kobj
;
1134 struct completion s_kobj_unregister
;
1137 struct journal_s
*s_journal
;
1138 struct list_head s_orphan
;
1139 struct mutex s_orphan_lock
;
1140 unsigned long s_resize_flags
; /* Flags indicating if there
1142 unsigned long s_commit_interval
;
1143 u32 s_max_batch_time
;
1144 u32 s_min_batch_time
;
1145 struct block_device
*journal_bdev
;
1146 #ifdef CONFIG_JBD2_DEBUG
1147 struct timer_list turn_ro_timer
; /* For turning read-only (crash simulation) */
1148 wait_queue_head_t ro_wait_queue
; /* For people waiting for the fs to go read-only */
1151 char *s_qf_names
[MAXQUOTAS
]; /* Names of quota files with journalled quota */
1152 int s_jquota_fmt
; /* Format of quota to use */
1154 unsigned int s_want_extra_isize
; /* New inodes should reserve # bytes */
1155 struct rb_root system_blks
;
1157 #ifdef EXTENTS_STATS
1158 /* ext4 extents stats */
1159 unsigned long s_ext_min
;
1160 unsigned long s_ext_max
;
1161 unsigned long s_depth_max
;
1162 spinlock_t s_ext_stats_lock
;
1163 unsigned long s_ext_blocks
;
1164 unsigned long s_ext_extents
;
1166 /* ext4 extent cache stats */
1167 unsigned long extent_cache_hits
;
1168 unsigned long extent_cache_misses
;
1170 /* for buddy allocator */
1171 struct ext4_group_info
***s_group_info
;
1172 struct inode
*s_buddy_cache
;
1173 spinlock_t s_md_lock
;
1174 unsigned short *s_mb_offsets
;
1175 unsigned int *s_mb_maxs
;
1178 unsigned long s_stripe
;
1179 unsigned int s_mb_stream_request
;
1180 unsigned int s_mb_max_to_scan
;
1181 unsigned int s_mb_min_to_scan
;
1182 unsigned int s_mb_stats
;
1183 unsigned int s_mb_order2_reqs
;
1184 unsigned int s_mb_group_prealloc
;
1185 unsigned int s_max_writeback_mb_bump
;
1186 /* where last allocation was done - for stream allocation */
1187 unsigned long s_mb_last_group
;
1188 unsigned long s_mb_last_start
;
1190 /* stats for buddy allocator */
1191 atomic_t s_bal_reqs
; /* number of reqs with len > 1 */
1192 atomic_t s_bal_success
; /* we found long enough chunks */
1193 atomic_t s_bal_allocated
; /* in blocks */
1194 atomic_t s_bal_ex_scanned
; /* total extents scanned */
1195 atomic_t s_bal_goals
; /* goal hits */
1196 atomic_t s_bal_breaks
; /* too long searches */
1197 atomic_t s_bal_2orders
; /* 2^order hits */
1198 spinlock_t s_bal_lock
;
1199 unsigned long s_mb_buddies_generated
;
1200 unsigned long long s_mb_generation_time
;
1201 atomic_t s_mb_lost_chunks
;
1202 atomic_t s_mb_preallocated
;
1203 atomic_t s_mb_discarded
;
1204 atomic_t s_lock_busy
;
1206 /* locality groups */
1207 struct ext4_locality_group __percpu
*s_locality_groups
;
1209 /* for write statistics */
1210 unsigned long s_sectors_written_start
;
1211 u64 s_kbytes_written
;
1213 unsigned int s_log_groups_per_flex
;
1214 struct flex_groups
*s_flex_groups
;
1216 /* workqueue for dio unwritten */
1217 struct workqueue_struct
*dio_unwritten_wq
;
1219 /* timer for periodic error stats printing */
1220 struct timer_list s_err_report
;
1222 /* Lazy inode table initialization info */
1223 struct ext4_li_request
*s_li_request
;
1224 /* Wait multiplier for lazy initialization thread */
1225 unsigned int s_li_wait_mult
;
1227 /* Kernel thread for multiple mount protection */
1228 struct task_struct
*s_mmp_tsk
;
1230 /* record the last minlen when FITRIM is called. */
1231 atomic_t s_last_trim_minblks
;
1234 static inline struct ext4_sb_info
*EXT4_SB(struct super_block
*sb
)
1236 return sb
->s_fs_info
;
1238 static inline struct ext4_inode_info
*EXT4_I(struct inode
*inode
)
1240 return container_of(inode
, struct ext4_inode_info
, vfs_inode
);
1243 static inline struct timespec
ext4_current_time(struct inode
*inode
)
1245 return (inode
->i_sb
->s_time_gran
< NSEC_PER_SEC
) ?
1246 current_fs_time(inode
->i_sb
) : CURRENT_TIME_SEC
;
1249 static inline int ext4_valid_inum(struct super_block
*sb
, unsigned long ino
)
1251 return ino
== EXT4_ROOT_INO
||
1252 ino
== EXT4_JOURNAL_INO
||
1253 ino
== EXT4_RESIZE_INO
||
1254 (ino
>= EXT4_FIRST_INO(sb
) &&
1255 ino
<= le32_to_cpu(EXT4_SB(sb
)->s_es
->s_inodes_count
));
1259 * Inode dynamic state flags
1262 EXT4_STATE_JDATA
, /* journaled data exists */
1263 EXT4_STATE_NEW
, /* inode is newly created */
1264 EXT4_STATE_XATTR
, /* has in-inode xattrs */
1265 EXT4_STATE_NO_EXPAND
, /* No space for expansion */
1266 EXT4_STATE_DA_ALLOC_CLOSE
, /* Alloc DA blks on close */
1267 EXT4_STATE_EXT_MIGRATE
, /* Inode is migrating */
1268 EXT4_STATE_DIO_UNWRITTEN
, /* need convert on dio done*/
1269 EXT4_STATE_NEWENTRY
, /* File just added to dir */
1270 EXT4_STATE_DELALLOC_RESERVED
, /* blks already reserved for delalloc */
1273 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1274 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1276 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1278 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1280 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1282 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1284 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1287 EXT4_INODE_BIT_FNS(flag
, flags
, 0)
1288 #if (BITS_PER_LONG < 64)
1289 EXT4_INODE_BIT_FNS(state
, state_flags
, 0)
1291 static inline void ext4_clear_state_flags(struct ext4_inode_info
*ei
)
1293 (ei
)->i_state_flags
= 0;
1296 EXT4_INODE_BIT_FNS(state
, flags
, 32)
1298 static inline void ext4_clear_state_flags(struct ext4_inode_info
*ei
)
1300 /* We depend on the fact that callers will set i_flags */
1304 /* Assume that user mode programs are passing in an ext4fs superblock, not
1305 * a kernel struct super_block. This will allow us to call the feature-test
1306 * macros from user land. */
1307 #define EXT4_SB(sb) (sb)
1310 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1313 * Codes for operating systems
1315 #define EXT4_OS_LINUX 0
1316 #define EXT4_OS_HURD 1
1317 #define EXT4_OS_MASIX 2
1318 #define EXT4_OS_FREEBSD 3
1319 #define EXT4_OS_LITES 4
1324 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1325 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1327 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1328 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1330 #define EXT4_GOOD_OLD_INODE_SIZE 128
1333 * Feature set definitions
1336 #define EXT4_HAS_COMPAT_FEATURE(sb,mask) \
1337 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1338 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \
1339 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1340 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \
1341 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1342 #define EXT4_SET_COMPAT_FEATURE(sb,mask) \
1343 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1344 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \
1345 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1346 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \
1347 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1348 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \
1349 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1350 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
1351 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1352 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \
1353 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1355 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1356 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1357 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1358 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1359 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1360 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1362 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1363 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1364 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1365 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1366 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1367 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1368 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1369 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1371 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1372 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1373 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1374 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1375 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1376 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1377 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1378 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1379 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1380 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1381 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1383 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1384 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1385 EXT4_FEATURE_INCOMPAT_META_BG)
1386 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1387 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1388 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1390 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1391 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1392 EXT4_FEATURE_INCOMPAT_RECOVER| \
1393 EXT4_FEATURE_INCOMPAT_META_BG)
1394 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1395 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1396 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1398 #define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR
1399 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1400 EXT4_FEATURE_INCOMPAT_RECOVER| \
1401 EXT4_FEATURE_INCOMPAT_META_BG| \
1402 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1403 EXT4_FEATURE_INCOMPAT_64BIT| \
1404 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1405 EXT4_FEATURE_INCOMPAT_MMP)
1406 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1407 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1408 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1409 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1410 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1411 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1412 EXT4_FEATURE_RO_COMPAT_HUGE_FILE)
1415 * Default values for user and/or group using reserved blocks
1417 #define EXT4_DEF_RESUID 0
1418 #define EXT4_DEF_RESGID 0
1420 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
1423 * Default mount options
1425 #define EXT4_DEFM_DEBUG 0x0001
1426 #define EXT4_DEFM_BSDGROUPS 0x0002
1427 #define EXT4_DEFM_XATTR_USER 0x0004
1428 #define EXT4_DEFM_ACL 0x0008
1429 #define EXT4_DEFM_UID16 0x0010
1430 #define EXT4_DEFM_JMODE 0x0060
1431 #define EXT4_DEFM_JMODE_DATA 0x0020
1432 #define EXT4_DEFM_JMODE_ORDERED 0x0040
1433 #define EXT4_DEFM_JMODE_WBACK 0x0060
1434 #define EXT4_DEFM_NOBARRIER 0x0100
1435 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1436 #define EXT4_DEFM_DISCARD 0x0400
1437 #define EXT4_DEFM_NODELALLOC 0x0800
1440 * Default journal batch times
1442 #define EXT4_DEF_MIN_BATCH_TIME 0
1443 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1446 * Minimum number of groups in a flexgroup before we separate out
1447 * directories into the first block group of a flexgroup
1449 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1452 * Structure of a directory entry
1454 #define EXT4_NAME_LEN 255
1456 struct ext4_dir_entry
{
1457 __le32 inode
; /* Inode number */
1458 __le16 rec_len
; /* Directory entry length */
1459 __le16 name_len
; /* Name length */
1460 char name
[EXT4_NAME_LEN
]; /* File name */
1464 * The new version of the directory entry. Since EXT4 structures are
1465 * stored in intel byte order, and the name_len field could never be
1466 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1469 struct ext4_dir_entry_2
{
1470 __le32 inode
; /* Inode number */
1471 __le16 rec_len
; /* Directory entry length */
1472 __u8 name_len
; /* Name length */
1474 char name
[EXT4_NAME_LEN
]; /* File name */
1478 * Ext4 directory file types. Only the low 3 bits are used. The
1479 * other bits are reserved for now.
1481 #define EXT4_FT_UNKNOWN 0
1482 #define EXT4_FT_REG_FILE 1
1483 #define EXT4_FT_DIR 2
1484 #define EXT4_FT_CHRDEV 3
1485 #define EXT4_FT_BLKDEV 4
1486 #define EXT4_FT_FIFO 5
1487 #define EXT4_FT_SOCK 6
1488 #define EXT4_FT_SYMLINK 7
1490 #define EXT4_FT_MAX 8
1493 * EXT4_DIR_PAD defines the directory entries boundaries
1495 * NOTE: It must be a multiple of 4
1497 #define EXT4_DIR_PAD 4
1498 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1499 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1501 #define EXT4_MAX_REC_LEN ((1<<16)-1)
1504 * If we ever get support for fs block sizes > page_size, we'll need
1505 * to remove the #if statements in the next two functions...
1507 static inline unsigned int
1508 ext4_rec_len_from_disk(__le16 dlen
, unsigned blocksize
)
1510 unsigned len
= le16_to_cpu(dlen
);
1512 #if (PAGE_CACHE_SIZE >= 65536)
1513 if (len
== EXT4_MAX_REC_LEN
|| len
== 0)
1515 return (len
& 65532) | ((len
& 3) << 16);
1521 static inline __le16
ext4_rec_len_to_disk(unsigned len
, unsigned blocksize
)
1523 if ((len
> blocksize
) || (blocksize
> (1 << 18)) || (len
& 3))
1525 #if (PAGE_CACHE_SIZE >= 65536)
1527 return cpu_to_le16(len
);
1528 if (len
== blocksize
) {
1529 if (blocksize
== 65536)
1530 return cpu_to_le16(EXT4_MAX_REC_LEN
);
1532 return cpu_to_le16(0);
1534 return cpu_to_le16((len
& 65532) | ((len
>> 16) & 3));
1536 return cpu_to_le16(len
);
1541 * Hash Tree Directory indexing
1542 * (c) Daniel Phillips, 2001
1545 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1546 EXT4_FEATURE_COMPAT_DIR_INDEX) && \
1547 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
1548 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1549 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1551 /* Legal values for the dx_root hash_version field: */
1553 #define DX_HASH_LEGACY 0
1554 #define DX_HASH_HALF_MD4 1
1555 #define DX_HASH_TEA 2
1556 #define DX_HASH_LEGACY_UNSIGNED 3
1557 #define DX_HASH_HALF_MD4_UNSIGNED 4
1558 #define DX_HASH_TEA_UNSIGNED 5
1562 /* hash info structure used by the directory hash */
1571 #define EXT4_HTREE_EOF 0x7fffffff
1574 * Control parameters used by ext4_htree_next_block
1576 #define HASH_NB_ALWAYS 1
1580 * Describe an inode's exact location on disk and in memory
1584 struct buffer_head
*bh
;
1585 unsigned long offset
;
1586 ext4_group_t block_group
;
1589 static inline struct ext4_inode
*ext4_raw_inode(struct ext4_iloc
*iloc
)
1591 return (struct ext4_inode
*) (iloc
->bh
->b_data
+ iloc
->offset
);
1595 * This structure is stuffed into the struct file's private_data field
1596 * for directories. It is where we put information so that we can do
1597 * readdir operations in hash tree order.
1599 struct dir_private_info
{
1600 struct rb_root root
;
1601 struct rb_node
*curr_node
;
1602 struct fname
*extra_fname
;
1605 __u32 curr_minor_hash
;
1609 /* calculate the first block number of the group */
1610 static inline ext4_fsblk_t
1611 ext4_group_first_block_no(struct super_block
*sb
, ext4_group_t group_no
)
1613 return group_no
* (ext4_fsblk_t
)EXT4_BLOCKS_PER_GROUP(sb
) +
1614 le32_to_cpu(EXT4_SB(sb
)->s_es
->s_first_data_block
);
1618 * Special error return code only used by dx_probe() and its callers.
1620 #define ERR_BAD_DX_DIR -75000
1622 void ext4_get_group_no_and_offset(struct super_block
*sb
, ext4_fsblk_t blocknr
,
1623 ext4_group_t
*blockgrpp
, ext4_grpblk_t
*offsetp
);
1626 * Timeout and state flag for lazy initialization inode thread.
1628 #define EXT4_DEF_LI_WAIT_MULT 10
1629 #define EXT4_DEF_LI_MAX_START_DELAY 5
1630 #define EXT4_LAZYINIT_QUIT 0x0001
1631 #define EXT4_LAZYINIT_RUNNING 0x0002
1634 * Lazy inode table initialization info
1636 struct ext4_lazy_init
{
1637 unsigned long li_state
;
1638 struct list_head li_request_list
;
1639 struct mutex li_list_mtx
;
1642 struct ext4_li_request
{
1643 struct super_block
*lr_super
;
1644 struct ext4_sb_info
*lr_sbi
;
1645 ext4_group_t lr_next_group
;
1646 struct list_head lr_request
;
1647 unsigned long lr_next_sched
;
1648 unsigned long lr_timeout
;
1651 struct ext4_features
{
1652 struct kobject f_kobj
;
1653 struct completion f_kobj_unregister
;
1657 * This structure will be used for multiple mount protection. It will be
1658 * written into the block number saved in the s_mmp_block field in the
1659 * superblock. Programs that check MMP should assume that if
1660 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
1661 * to use the filesystem, regardless of how old the timestamp is.
1663 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
1664 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
1665 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
1666 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
1669 __le32 mmp_magic
; /* Magic number for MMP */
1670 __le32 mmp_seq
; /* Sequence no. updated periodically */
1673 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
1674 * purposes and do not affect the correctness of the algorithm
1676 __le64 mmp_time
; /* Time last updated */
1677 char mmp_nodename
[64]; /* Node which last updated MMP block */
1678 char mmp_bdevname
[32]; /* Bdev which last updated MMP block */
1681 * mmp_check_interval is used to verify if the MMP block has been
1682 * updated on the block device. The value is updated based on the
1683 * maximum time to write the MMP block during an update cycle.
1685 __le16 mmp_check_interval
;
1688 __le32 mmp_pad2
[227];
1691 /* arguments passed to the mmp thread */
1693 struct buffer_head
*bh
; /* bh from initial read_mmp_block() */
1694 struct super_block
*sb
; /* super block of the fs */
1698 * Check interval multiplier
1699 * The MMP block is written every update interval and initially checked every
1700 * update interval x the multiplier (the value is then adapted based on the
1701 * write latency). The reason is that writes can be delayed under load and we
1702 * don't want readers to incorrectly assume that the filesystem is no longer
1705 #define EXT4_MMP_CHECK_MULT 2UL
1708 * Minimum interval for MMP checking in seconds.
1710 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
1713 * Maximum interval for MMP checking in seconds.
1715 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
1718 * Function prototypes
1722 * Ok, these declarations are also in <linux/kernel.h> but none of the
1723 * ext4 source programs needs to include it so they are duplicated here.
1725 # define NORET_TYPE /**/
1726 # define ATTRIB_NORET __attribute__((noreturn))
1727 # define NORET_AND noreturn,
1730 extern unsigned int ext4_count_free(struct buffer_head
*, unsigned);
1733 extern unsigned int ext4_block_group(struct super_block
*sb
,
1734 ext4_fsblk_t blocknr
);
1735 extern ext4_grpblk_t
ext4_block_group_offset(struct super_block
*sb
,
1736 ext4_fsblk_t blocknr
);
1737 extern int ext4_bg_has_super(struct super_block
*sb
, ext4_group_t group
);
1738 extern unsigned long ext4_bg_num_gdb(struct super_block
*sb
,
1739 ext4_group_t group
);
1740 extern ext4_fsblk_t
ext4_new_meta_blocks(handle_t
*handle
, struct inode
*inode
,
1743 unsigned long *count
,
1745 extern int ext4_claim_free_blocks(struct ext4_sb_info
*sbi
,
1746 s64 nblocks
, unsigned int flags
);
1747 extern ext4_fsblk_t
ext4_count_free_blocks(struct super_block
*);
1748 extern void ext4_check_blocks_bitmap(struct super_block
*);
1749 extern struct ext4_group_desc
* ext4_get_group_desc(struct super_block
* sb
,
1750 ext4_group_t block_group
,
1751 struct buffer_head
** bh
);
1752 extern int ext4_should_retry_alloc(struct super_block
*sb
, int *retries
);
1753 struct buffer_head
*ext4_read_block_bitmap(struct super_block
*sb
,
1754 ext4_group_t block_group
);
1755 extern unsigned ext4_init_block_bitmap(struct super_block
*sb
,
1756 struct buffer_head
*bh
,
1758 struct ext4_group_desc
*desc
);
1759 #define ext4_free_blocks_after_init(sb, group, desc) \
1760 ext4_init_block_bitmap(sb, NULL, group, desc)
1761 ext4_fsblk_t
ext4_inode_to_goal_block(struct inode
*);
1764 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode
*,
1766 struct ext4_dir_entry_2
*,
1767 struct buffer_head
*, unsigned int);
1768 #define ext4_check_dir_entry(dir, filp, de, bh, offset) \
1769 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
1770 (de), (bh), (offset)))
1771 extern int ext4_htree_store_dirent(struct file
*dir_file
, __u32 hash
,
1773 struct ext4_dir_entry_2
*dirent
);
1774 extern void ext4_htree_free_dir_info(struct dir_private_info
*p
);
1777 extern int ext4_sync_file(struct file
*, loff_t
, loff_t
, int);
1778 extern int ext4_flush_completed_IO(struct inode
*);
1781 extern int ext4fs_dirhash(const char *name
, int len
, struct
1782 dx_hash_info
*hinfo
);
1785 extern struct inode
*ext4_new_inode(handle_t
*, struct inode
*, int,
1786 const struct qstr
*qstr
, __u32 goal
);
1787 extern void ext4_free_inode(handle_t
*, struct inode
*);
1788 extern struct inode
* ext4_orphan_get(struct super_block
*, unsigned long);
1789 extern unsigned long ext4_count_free_inodes(struct super_block
*);
1790 extern unsigned long ext4_count_dirs(struct super_block
*);
1791 extern void ext4_check_inodes_bitmap(struct super_block
*);
1792 extern void ext4_mark_bitmap_end(int start_bit
, int end_bit
, char *bitmap
);
1793 extern int ext4_init_inode_table(struct super_block
*sb
,
1794 ext4_group_t group
, int barrier
);
1797 extern long ext4_mb_stats
;
1798 extern long ext4_mb_max_to_scan
;
1799 extern int ext4_mb_init(struct super_block
*, int);
1800 extern int ext4_mb_release(struct super_block
*);
1801 extern ext4_fsblk_t
ext4_mb_new_blocks(handle_t
*,
1802 struct ext4_allocation_request
*, int *);
1803 extern int ext4_mb_reserve_blocks(struct super_block
*, int);
1804 extern void ext4_discard_preallocations(struct inode
*);
1805 extern int __init
ext4_init_mballoc(void);
1806 extern void ext4_exit_mballoc(void);
1807 extern void ext4_free_blocks(handle_t
*handle
, struct inode
*inode
,
1808 struct buffer_head
*bh
, ext4_fsblk_t block
,
1809 unsigned long count
, int flags
);
1810 extern int ext4_mb_add_groupinfo(struct super_block
*sb
,
1811 ext4_group_t i
, struct ext4_group_desc
*desc
);
1812 extern int ext4_group_add_blocks(handle_t
*handle
, struct super_block
*sb
,
1813 ext4_fsblk_t block
, unsigned long count
);
1814 extern int ext4_trim_fs(struct super_block
*, struct fstrim_range
*);
1817 struct buffer_head
*ext4_getblk(handle_t
*, struct inode
*,
1818 ext4_lblk_t
, int, int *);
1819 struct buffer_head
*ext4_bread(handle_t
*, struct inode
*,
1820 ext4_lblk_t
, int, int *);
1821 int ext4_get_block(struct inode
*inode
, sector_t iblock
,
1822 struct buffer_head
*bh_result
, int create
);
1824 extern struct inode
*ext4_iget(struct super_block
*, unsigned long);
1825 extern int ext4_write_inode(struct inode
*, struct writeback_control
*);
1826 extern int ext4_setattr(struct dentry
*, struct iattr
*);
1827 extern int ext4_getattr(struct vfsmount
*mnt
, struct dentry
*dentry
,
1828 struct kstat
*stat
);
1829 extern void ext4_evict_inode(struct inode
*);
1830 extern void ext4_clear_inode(struct inode
*);
1831 extern int ext4_sync_inode(handle_t
*, struct inode
*);
1832 extern void ext4_dirty_inode(struct inode
*, int);
1833 extern int ext4_change_inode_journal_flag(struct inode
*, int);
1834 extern int ext4_get_inode_loc(struct inode
*, struct ext4_iloc
*);
1835 extern int ext4_can_truncate(struct inode
*inode
);
1836 extern void ext4_truncate(struct inode
*);
1837 extern int ext4_punch_hole(struct file
*file
, loff_t offset
, loff_t length
);
1838 extern int ext4_truncate_restart_trans(handle_t
*, struct inode
*, int nblocks
);
1839 extern void ext4_set_inode_flags(struct inode
*);
1840 extern void ext4_get_inode_flags(struct ext4_inode_info
*);
1841 extern int ext4_alloc_da_blocks(struct inode
*inode
);
1842 extern void ext4_set_aops(struct inode
*inode
);
1843 extern int ext4_writepage_trans_blocks(struct inode
*);
1844 extern int ext4_chunk_trans_blocks(struct inode
*, int nrblocks
);
1845 extern int ext4_block_truncate_page(handle_t
*handle
,
1846 struct address_space
*mapping
, loff_t from
);
1847 extern int ext4_block_zero_page_range(handle_t
*handle
,
1848 struct address_space
*mapping
, loff_t from
, loff_t length
);
1849 extern int ext4_discard_partial_page_buffers(handle_t
*handle
,
1850 struct address_space
*mapping
, loff_t from
,
1851 loff_t length
, int flags
);
1852 extern int ext4_discard_partial_page_buffers_no_lock(handle_t
*handle
,
1853 struct inode
*inode
, struct page
*page
, loff_t from
,
1854 loff_t length
, int flags
);
1855 extern int ext4_page_mkwrite(struct vm_area_struct
*vma
, struct vm_fault
*vmf
);
1856 extern qsize_t
*ext4_get_reserved_space(struct inode
*inode
);
1857 extern void ext4_da_update_reserve_space(struct inode
*inode
,
1858 int used
, int quota_claim
);
1861 extern int ext4_ind_map_blocks(handle_t
*handle
, struct inode
*inode
,
1862 struct ext4_map_blocks
*map
, int flags
);
1863 extern ssize_t
ext4_ind_direct_IO(int rw
, struct kiocb
*iocb
,
1864 const struct iovec
*iov
, loff_t offset
,
1865 unsigned long nr_segs
);
1866 extern int ext4_ind_calc_metadata_amount(struct inode
*inode
, sector_t lblock
);
1867 extern int ext4_ind_trans_blocks(struct inode
*inode
, int nrblocks
, int chunk
);
1868 extern void ext4_ind_truncate(struct inode
*inode
);
1871 extern long ext4_ioctl(struct file
*, unsigned int, unsigned long);
1872 extern long ext4_compat_ioctl(struct file
*, unsigned int, unsigned long);
1875 extern int ext4_ext_migrate(struct inode
*);
1878 extern int ext4_orphan_add(handle_t
*, struct inode
*);
1879 extern int ext4_orphan_del(handle_t
*, struct inode
*);
1880 extern int ext4_htree_fill_tree(struct file
*dir_file
, __u32 start_hash
,
1881 __u32 start_minor_hash
, __u32
*next_hash
);
1884 extern int ext4_group_add(struct super_block
*sb
,
1885 struct ext4_new_group_data
*input
);
1886 extern int ext4_group_extend(struct super_block
*sb
,
1887 struct ext4_super_block
*es
,
1888 ext4_fsblk_t n_blocks_count
);
1891 extern void *ext4_kvmalloc(size_t size
, gfp_t flags
);
1892 extern void *ext4_kvzalloc(size_t size
, gfp_t flags
);
1893 extern void ext4_kvfree(void *ptr
);
1894 extern void __ext4_error(struct super_block
*, const char *, unsigned int,
1896 __attribute__ ((format (printf
, 4, 5)));
1897 #define ext4_error(sb, message...) __ext4_error(sb, __func__, \
1898 __LINE__, ## message)
1899 extern void ext4_error_inode(struct inode
*, const char *, unsigned int,
1900 ext4_fsblk_t
, const char *, ...)
1901 __attribute__ ((format (printf
, 5, 6)));
1902 extern void ext4_error_file(struct file
*, const char *, unsigned int,
1903 ext4_fsblk_t
, const char *, ...)
1904 __attribute__ ((format (printf
, 5, 6)));
1905 extern void __ext4_std_error(struct super_block
*, const char *,
1907 extern void __ext4_abort(struct super_block
*, const char *, unsigned int,
1909 __attribute__ ((format (printf
, 4, 5)));
1910 #define ext4_abort(sb, message...) __ext4_abort(sb, __func__, \
1911 __LINE__, ## message)
1912 extern void __ext4_warning(struct super_block
*, const char *, unsigned int,
1914 __attribute__ ((format (printf
, 4, 5)));
1915 #define ext4_warning(sb, message...) __ext4_warning(sb, __func__, \
1916 __LINE__, ## message)
1917 extern void ext4_msg(struct super_block
*, const char *, const char *, ...)
1918 __attribute__ ((format (printf
, 3, 4)));
1919 extern void __dump_mmp_msg(struct super_block
*, struct mmp_struct
*mmp
,
1920 const char *, unsigned int, const char *);
1921 #define dump_mmp_msg(sb, mmp, msg) __dump_mmp_msg(sb, mmp, __func__, \
1923 extern void __ext4_grp_locked_error(const char *, unsigned int, \
1924 struct super_block
*, ext4_group_t
, \
1925 unsigned long, ext4_fsblk_t
, \
1927 __attribute__ ((format (printf
, 7, 8)));
1928 #define ext4_grp_locked_error(sb, grp, message...) \
1929 __ext4_grp_locked_error(__func__, __LINE__, (sb), (grp), ## message)
1930 extern void ext4_update_dynamic_rev(struct super_block
*sb
);
1931 extern int ext4_update_compat_feature(handle_t
*handle
, struct super_block
*sb
,
1933 extern int ext4_update_rocompat_feature(handle_t
*handle
,
1934 struct super_block
*sb
, __u32 rocompat
);
1935 extern int ext4_update_incompat_feature(handle_t
*handle
,
1936 struct super_block
*sb
, __u32 incompat
);
1937 extern ext4_fsblk_t
ext4_block_bitmap(struct super_block
*sb
,
1938 struct ext4_group_desc
*bg
);
1939 extern ext4_fsblk_t
ext4_inode_bitmap(struct super_block
*sb
,
1940 struct ext4_group_desc
*bg
);
1941 extern ext4_fsblk_t
ext4_inode_table(struct super_block
*sb
,
1942 struct ext4_group_desc
*bg
);
1943 extern __u32
ext4_free_blks_count(struct super_block
*sb
,
1944 struct ext4_group_desc
*bg
);
1945 extern __u32
ext4_free_inodes_count(struct super_block
*sb
,
1946 struct ext4_group_desc
*bg
);
1947 extern __u32
ext4_used_dirs_count(struct super_block
*sb
,
1948 struct ext4_group_desc
*bg
);
1949 extern __u32
ext4_itable_unused_count(struct super_block
*sb
,
1950 struct ext4_group_desc
*bg
);
1951 extern void ext4_block_bitmap_set(struct super_block
*sb
,
1952 struct ext4_group_desc
*bg
, ext4_fsblk_t blk
);
1953 extern void ext4_inode_bitmap_set(struct super_block
*sb
,
1954 struct ext4_group_desc
*bg
, ext4_fsblk_t blk
);
1955 extern void ext4_inode_table_set(struct super_block
*sb
,
1956 struct ext4_group_desc
*bg
, ext4_fsblk_t blk
);
1957 extern void ext4_free_blks_set(struct super_block
*sb
,
1958 struct ext4_group_desc
*bg
, __u32 count
);
1959 extern void ext4_free_inodes_set(struct super_block
*sb
,
1960 struct ext4_group_desc
*bg
, __u32 count
);
1961 extern void ext4_used_dirs_set(struct super_block
*sb
,
1962 struct ext4_group_desc
*bg
, __u32 count
);
1963 extern void ext4_itable_unused_set(struct super_block
*sb
,
1964 struct ext4_group_desc
*bg
, __u32 count
);
1965 extern __le16
ext4_group_desc_csum(struct ext4_sb_info
*sbi
, __u32 group
,
1966 struct ext4_group_desc
*gdp
);
1967 extern int ext4_group_desc_csum_verify(struct ext4_sb_info
*sbi
, __u32 group
,
1968 struct ext4_group_desc
*gdp
);
1970 static inline ext4_fsblk_t
ext4_blocks_count(struct ext4_super_block
*es
)
1972 return ((ext4_fsblk_t
)le32_to_cpu(es
->s_blocks_count_hi
) << 32) |
1973 le32_to_cpu(es
->s_blocks_count_lo
);
1976 static inline ext4_fsblk_t
ext4_r_blocks_count(struct ext4_super_block
*es
)
1978 return ((ext4_fsblk_t
)le32_to_cpu(es
->s_r_blocks_count_hi
) << 32) |
1979 le32_to_cpu(es
->s_r_blocks_count_lo
);
1982 static inline ext4_fsblk_t
ext4_free_blocks_count(struct ext4_super_block
*es
)
1984 return ((ext4_fsblk_t
)le32_to_cpu(es
->s_free_blocks_count_hi
) << 32) |
1985 le32_to_cpu(es
->s_free_blocks_count_lo
);
1988 static inline void ext4_blocks_count_set(struct ext4_super_block
*es
,
1991 es
->s_blocks_count_lo
= cpu_to_le32((u32
)blk
);
1992 es
->s_blocks_count_hi
= cpu_to_le32(blk
>> 32);
1995 static inline void ext4_free_blocks_count_set(struct ext4_super_block
*es
,
1998 es
->s_free_blocks_count_lo
= cpu_to_le32((u32
)blk
);
1999 es
->s_free_blocks_count_hi
= cpu_to_le32(blk
>> 32);
2002 static inline void ext4_r_blocks_count_set(struct ext4_super_block
*es
,
2005 es
->s_r_blocks_count_lo
= cpu_to_le32((u32
)blk
);
2006 es
->s_r_blocks_count_hi
= cpu_to_le32(blk
>> 32);
2009 static inline loff_t
ext4_isize(struct ext4_inode
*raw_inode
)
2011 if (S_ISREG(le16_to_cpu(raw_inode
->i_mode
)))
2012 return ((loff_t
)le32_to_cpu(raw_inode
->i_size_high
) << 32) |
2013 le32_to_cpu(raw_inode
->i_size_lo
);
2015 return (loff_t
) le32_to_cpu(raw_inode
->i_size_lo
);
2018 static inline void ext4_isize_set(struct ext4_inode
*raw_inode
, loff_t i_size
)
2020 raw_inode
->i_size_lo
= cpu_to_le32(i_size
);
2021 raw_inode
->i_size_high
= cpu_to_le32(i_size
>> 32);
2025 struct ext4_group_info
*ext4_get_group_info(struct super_block
*sb
,
2028 struct ext4_group_info
***grp_info
;
2029 long indexv
, indexh
;
2030 grp_info
= EXT4_SB(sb
)->s_group_info
;
2031 indexv
= group
>> (EXT4_DESC_PER_BLOCK_BITS(sb
));
2032 indexh
= group
& ((EXT4_DESC_PER_BLOCK(sb
)) - 1);
2033 return grp_info
[indexv
][indexh
];
2037 * Reading s_groups_count requires using smp_rmb() afterwards. See
2038 * the locking protocol documented in the comments of ext4_group_add()
2041 static inline ext4_group_t
ext4_get_groups_count(struct super_block
*sb
)
2043 ext4_group_t ngroups
= EXT4_SB(sb
)->s_groups_count
;
2049 static inline ext4_group_t
ext4_flex_group(struct ext4_sb_info
*sbi
,
2050 ext4_group_t block_group
)
2052 return block_group
>> sbi
->s_log_groups_per_flex
;
2055 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info
*sbi
)
2057 return 1 << sbi
->s_log_groups_per_flex
;
2060 #define ext4_std_error(sb, errno) \
2063 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
2067 /* Each CPU can accumulate percpu_counter_batch blocks in their local
2068 * counters. So we need to make sure we have free blocks more
2069 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
2071 #define EXT4_FREEBLOCKS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2073 #define EXT4_FREEBLOCKS_WATERMARK 0
2076 static inline void ext4_update_i_disksize(struct inode
*inode
, loff_t newsize
)
2079 * XXX: replace with spinlock if seen contended -bzzz
2081 down_write(&EXT4_I(inode
)->i_data_sem
);
2082 if (newsize
> EXT4_I(inode
)->i_disksize
)
2083 EXT4_I(inode
)->i_disksize
= newsize
;
2084 up_write(&EXT4_I(inode
)->i_data_sem
);
2088 struct ext4_group_info
{
2089 unsigned long bb_state
;
2090 struct rb_root bb_free_root
;
2091 ext4_grpblk_t bb_first_free
; /* first free block */
2092 ext4_grpblk_t bb_free
; /* total free blocks */
2093 ext4_grpblk_t bb_fragments
; /* nr of freespace fragments */
2094 ext4_grpblk_t bb_largest_free_order
;/* order of largest frag in BG */
2095 struct list_head bb_prealloc_list
;
2099 struct rw_semaphore alloc_sem
;
2100 ext4_grpblk_t bb_counters
[]; /* Nr of free power-of-two-block
2101 * regions, index is order.
2102 * bb_counters[3] = 5 means
2103 * 5 free 8-block regions. */
2106 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
2107 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
2109 #define EXT4_MB_GRP_NEED_INIT(grp) \
2110 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2112 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
2113 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2114 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
2115 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2116 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
2117 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2119 #define EXT4_MAX_CONTENTION 8
2120 #define EXT4_CONTENTION_THRESHOLD 2
2122 static inline spinlock_t
*ext4_group_lock_ptr(struct super_block
*sb
,
2125 return bgl_lock_ptr(EXT4_SB(sb
)->s_blockgroup_lock
, group
);
2129 * Returns true if the filesystem is busy enough that attempts to
2130 * access the block group locks has run into contention.
2132 static inline int ext4_fs_is_busy(struct ext4_sb_info
*sbi
)
2134 return (atomic_read(&sbi
->s_lock_busy
) > EXT4_CONTENTION_THRESHOLD
);
2137 static inline void ext4_lock_group(struct super_block
*sb
, ext4_group_t group
)
2139 spinlock_t
*lock
= ext4_group_lock_ptr(sb
, group
);
2140 if (spin_trylock(lock
))
2142 * We're able to grab the lock right away, so drop the
2143 * lock contention counter.
2145 atomic_add_unless(&EXT4_SB(sb
)->s_lock_busy
, -1, 0);
2148 * The lock is busy, so bump the contention counter,
2149 * and then wait on the spin lock.
2151 atomic_add_unless(&EXT4_SB(sb
)->s_lock_busy
, 1,
2152 EXT4_MAX_CONTENTION
);
2157 static inline void ext4_unlock_group(struct super_block
*sb
,
2160 spin_unlock(ext4_group_lock_ptr(sb
, group
));
2163 static inline void ext4_mark_super_dirty(struct super_block
*sb
)
2165 if (EXT4_SB(sb
)->s_journal
== NULL
)
2170 * Block validity checking
2172 #define ext4_check_indirect_blockref(inode, bh) \
2173 ext4_check_blockref(__func__, __LINE__, inode, \
2174 (__le32 *)(bh)->b_data, \
2175 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2177 #define ext4_ind_check_inode(inode) \
2178 ext4_check_blockref(__func__, __LINE__, inode, \
2179 EXT4_I(inode)->i_data, \
2183 * Inodes and files operations
2187 extern const struct file_operations ext4_dir_operations
;
2190 extern const struct inode_operations ext4_file_inode_operations
;
2191 extern const struct file_operations ext4_file_operations
;
2192 extern loff_t
ext4_llseek(struct file
*file
, loff_t offset
, int origin
);
2195 extern const struct inode_operations ext4_dir_inode_operations
;
2196 extern const struct inode_operations ext4_special_inode_operations
;
2197 extern struct dentry
*ext4_get_parent(struct dentry
*child
);
2200 extern const struct inode_operations ext4_symlink_inode_operations
;
2201 extern const struct inode_operations ext4_fast_symlink_inode_operations
;
2203 /* block_validity */
2204 extern void ext4_release_system_zone(struct super_block
*sb
);
2205 extern int ext4_setup_system_zone(struct super_block
*sb
);
2206 extern int __init
ext4_init_system_zone(void);
2207 extern void ext4_exit_system_zone(void);
2208 extern int ext4_data_block_valid(struct ext4_sb_info
*sbi
,
2209 ext4_fsblk_t start_blk
,
2210 unsigned int count
);
2211 extern int ext4_check_blockref(const char *, unsigned int,
2212 struct inode
*, __le32
*, unsigned int);
2215 extern int ext4_ext_tree_init(handle_t
*handle
, struct inode
*);
2216 extern int ext4_ext_writepage_trans_blocks(struct inode
*, int);
2217 extern int ext4_ext_index_trans_blocks(struct inode
*inode
, int nrblocks
,
2219 extern int ext4_ext_map_blocks(handle_t
*handle
, struct inode
*inode
,
2220 struct ext4_map_blocks
*map
, int flags
);
2221 extern void ext4_ext_truncate(struct inode
*);
2222 extern int ext4_ext_punch_hole(struct file
*file
, loff_t offset
,
2224 extern void ext4_ext_init(struct super_block
*);
2225 extern void ext4_ext_release(struct super_block
*);
2226 extern long ext4_fallocate(struct file
*file
, int mode
, loff_t offset
,
2228 extern int ext4_convert_unwritten_extents(struct inode
*inode
, loff_t offset
,
2230 extern int ext4_map_blocks(handle_t
*handle
, struct inode
*inode
,
2231 struct ext4_map_blocks
*map
, int flags
);
2232 extern int ext4_fiemap(struct inode
*inode
, struct fiemap_extent_info
*fieinfo
,
2233 __u64 start
, __u64 len
);
2235 extern int ext4_move_extents(struct file
*o_filp
, struct file
*d_filp
,
2236 __u64 start_orig
, __u64 start_donor
,
2237 __u64 len
, __u64
*moved_len
);
2240 extern int __init
ext4_init_pageio(void);
2241 extern void ext4_exit_pageio(void);
2242 extern void ext4_ioend_wait(struct inode
*);
2243 extern void ext4_free_io_end(ext4_io_end_t
*io
);
2244 extern ext4_io_end_t
*ext4_init_io_end(struct inode
*inode
, gfp_t flags
);
2245 extern int ext4_end_io_nolock(ext4_io_end_t
*io
);
2246 extern void ext4_io_submit(struct ext4_io_submit
*io
);
2247 extern int ext4_bio_write_page(struct ext4_io_submit
*io
,
2250 struct writeback_control
*wbc
);
2253 extern int ext4_multi_mount_protect(struct super_block
*, ext4_fsblk_t
);
2255 /* BH_Uninit flag: blocks are allocated but uninitialized on disk */
2256 enum ext4_state_bits
{
2257 BH_Uninit
/* blocks are allocated but uninitialized on disk */
2258 = BH_JBDPrivateStart
,
2261 BUFFER_FNS(Uninit
, uninit
)
2262 TAS_BUFFER_FNS(Uninit
, uninit
)
2265 * Add new method to test wether block and inode bitmaps are properly
2266 * initialized. With uninit_bg reading the block from disk is not enough
2267 * to mark the bitmap uptodate. We need to also zero-out the bitmap
2269 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
2271 static inline int bitmap_uptodate(struct buffer_head
*bh
)
2273 return (buffer_uptodate(bh
) &&
2274 test_bit(BH_BITMAP_UPTODATE
, &(bh
)->b_state
));
2276 static inline void set_bitmap_uptodate(struct buffer_head
*bh
)
2278 set_bit(BH_BITMAP_UPTODATE
, &(bh
)->b_state
);
2281 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
2283 /* For ioend & aio unwritten conversion wait queues */
2284 #define EXT4_WQ_HASH_SZ 37
2285 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
2287 #define ext4_aio_mutex(v) (&ext4__aio_mutex[((unsigned long)(v)) %\
2289 extern wait_queue_head_t ext4__ioend_wq
[EXT4_WQ_HASH_SZ
];
2290 extern struct mutex ext4__aio_mutex
[EXT4_WQ_HASH_SZ
];
2292 #define EXT4_RESIZING 0
2293 extern int ext4_resize_begin(struct super_block
*sb
);
2294 extern void ext4_resize_end(struct super_block
*sb
);
2296 #endif /* __KERNEL__ */
2298 #endif /* _EXT4_H */