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>
34 * The fourth extended filesystem constants/structures
38 * Define EXT4FS_DEBUG to produce debug messages
46 #define ext4_debug(f, a...) \
48 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
49 __FILE__, __LINE__, __func__); \
50 printk(KERN_DEBUG f, ## a); \
53 #define ext4_debug(f, a...) do {} while (0)
56 /* data type for block offset of block group */
57 typedef int ext4_grpblk_t
;
59 /* data type for filesystem-wide blocks number */
60 typedef unsigned long long ext4_fsblk_t
;
62 /* data type for file logical block number */
63 typedef __u32 ext4_lblk_t
;
65 /* data type for block group number */
66 typedef unsigned int ext4_group_t
;
69 /* prefer goal again. length */
70 #define EXT4_MB_HINT_MERGE 1
71 /* blocks already reserved */
72 #define EXT4_MB_HINT_RESERVED 2
73 /* metadata is being allocated */
74 #define EXT4_MB_HINT_METADATA 4
75 /* first blocks in the file */
76 #define EXT4_MB_HINT_FIRST 8
77 /* search for the best chunk */
78 #define EXT4_MB_HINT_BEST 16
79 /* data is being allocated */
80 #define EXT4_MB_HINT_DATA 32
81 /* don't preallocate (for tails) */
82 #define EXT4_MB_HINT_NOPREALLOC 64
83 /* allocate for locality group */
84 #define EXT4_MB_HINT_GROUP_ALLOC 128
85 /* allocate goal blocks or none */
86 #define EXT4_MB_HINT_GOAL_ONLY 256
87 /* goal is meaningful */
88 #define EXT4_MB_HINT_TRY_GOAL 512
89 /* blocks already pre-reserved by delayed allocation */
90 #define EXT4_MB_DELALLOC_RESERVED 1024
91 /* We are doing stream allocation */
92 #define EXT4_MB_STREAM_ALLOC 2048
95 struct ext4_allocation_request
{
96 /* target inode for block we're allocating */
98 /* how many blocks we want to allocate */
100 /* logical block in target inode */
102 /* the closest logical allocated block to the left */
104 /* the closest logical allocated block to the right */
106 /* phys. target (a hint) */
108 /* phys. block for the closest logical allocated block to the left */
110 /* phys. block for the closest logical allocated block to the right */
112 /* flags. see above EXT4_MB_HINT_* */
116 #define DIO_AIO_UNWRITTEN 0x1
117 typedef struct ext4_io_end
{
118 struct list_head list
; /* per-file finished AIO list */
119 struct inode
*inode
; /* file being written to */
120 unsigned int flag
; /* sync IO or AIO */
121 int error
; /* I/O error code */
122 ext4_lblk_t offset
; /* offset in the file */
123 size_t size
; /* size of the extent */
124 struct work_struct work
; /* data work queue */
128 * Delayed allocation stuff
131 struct mpage_da_data
{
133 sector_t b_blocknr
; /* start block number of extent */
134 size_t b_size
; /* size of extent */
135 unsigned long b_state
; /* state of the extent */
136 unsigned long first_page
, next_page
; /* extent of pages */
137 struct writeback_control
*wbc
;
144 * Special inodes numbers
146 #define EXT4_BAD_INO 1 /* Bad blocks inode */
147 #define EXT4_ROOT_INO 2 /* Root inode */
148 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
149 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
150 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
151 #define EXT4_JOURNAL_INO 8 /* Journal inode */
153 /* First non-reserved inode for old ext4 filesystems */
154 #define EXT4_GOOD_OLD_FIRST_INO 11
157 * Maximal count of links to a file
159 #define EXT4_LINK_MAX 65000
162 * Macro-instructions used to manage several block sizes
164 #define EXT4_MIN_BLOCK_SIZE 1024
165 #define EXT4_MAX_BLOCK_SIZE 65536
166 #define EXT4_MIN_BLOCK_LOG_SIZE 10
168 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
170 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
172 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
174 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
176 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
179 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
180 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
181 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
183 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
184 EXT4_GOOD_OLD_INODE_SIZE : \
186 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
187 EXT4_GOOD_OLD_FIRST_INO : \
190 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
193 * Structure of a blocks group descriptor
195 struct ext4_group_desc
197 __le32 bg_block_bitmap_lo
; /* Blocks bitmap block */
198 __le32 bg_inode_bitmap_lo
; /* Inodes bitmap block */
199 __le32 bg_inode_table_lo
; /* Inodes table block */
200 __le16 bg_free_blocks_count_lo
;/* Free blocks count */
201 __le16 bg_free_inodes_count_lo
;/* Free inodes count */
202 __le16 bg_used_dirs_count_lo
; /* Directories count */
203 __le16 bg_flags
; /* EXT4_BG_flags (INODE_UNINIT, etc) */
204 __u32 bg_reserved
[2]; /* Likely block/inode bitmap checksum */
205 __le16 bg_itable_unused_lo
; /* Unused inodes count */
206 __le16 bg_checksum
; /* crc16(sb_uuid+group+desc) */
207 __le32 bg_block_bitmap_hi
; /* Blocks bitmap block MSB */
208 __le32 bg_inode_bitmap_hi
; /* Inodes bitmap block MSB */
209 __le32 bg_inode_table_hi
; /* Inodes table block MSB */
210 __le16 bg_free_blocks_count_hi
;/* Free blocks count MSB */
211 __le16 bg_free_inodes_count_hi
;/* Free inodes count MSB */
212 __le16 bg_used_dirs_count_hi
; /* Directories count MSB */
213 __le16 bg_itable_unused_hi
; /* Unused inodes count MSB */
214 __u32 bg_reserved2
[3];
218 * Structure of a flex block group info
222 atomic_t free_inodes
;
223 atomic_t free_blocks
;
227 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
228 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
229 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
232 * Macro-instructions used to manage group descriptors
234 #define EXT4_MIN_DESC_SIZE 32
235 #define EXT4_MIN_DESC_SIZE_64BIT 64
236 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
237 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
239 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
240 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
241 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
242 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
244 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
245 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
246 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
250 * Constants relative to the data blocks
252 #define EXT4_NDIR_BLOCKS 12
253 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
254 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
255 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
256 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
261 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
262 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
263 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
264 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
265 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
266 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
267 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
268 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
269 /* Reserved for compression usage... */
270 #define EXT4_DIRTY_FL 0x00000100
271 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
272 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
273 #define EXT4_ECOMPR_FL 0x00000800 /* Compression error */
274 /* End compression flags --- maybe not all used */
275 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
276 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
277 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
278 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
279 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
280 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
281 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
282 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
283 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
285 #define EXT4_FL_USER_VISIBLE 0x000BDFFF /* User visible flags */
286 #define EXT4_FL_USER_MODIFIABLE 0x000B80FF /* User modifiable flags */
288 /* Flags that should be inherited by new inodes from their parent. */
289 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
290 EXT4_SYNC_FL | EXT4_IMMUTABLE_FL | EXT4_APPEND_FL |\
291 EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
292 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
293 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
295 /* Flags that are appropriate for regular files (all but dir-specific ones). */
296 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
298 /* Flags that are appropriate for non-directories/regular files. */
299 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
301 /* Mask out flags that are inappropriate for the given type of inode. */
302 static inline __u32
ext4_mask_flags(umode_t mode
, __u32 flags
)
306 else if (S_ISREG(mode
))
307 return flags
& EXT4_REG_FLMASK
;
309 return flags
& EXT4_OTHER_FLMASK
;
313 * Inode dynamic state flags
315 #define EXT4_STATE_JDATA 0x00000001 /* journaled data exists */
316 #define EXT4_STATE_NEW 0x00000002 /* inode is newly created */
317 #define EXT4_STATE_XATTR 0x00000004 /* has in-inode xattrs */
318 #define EXT4_STATE_NO_EXPAND 0x00000008 /* No space for expansion */
319 #define EXT4_STATE_DA_ALLOC_CLOSE 0x00000010 /* Alloc DA blks on close */
320 #define EXT4_STATE_EXT_MIGRATE 0x00000020 /* Inode is migrating */
321 #define EXT4_STATE_DIO_UNWRITTEN 0x00000040 /* need convert on dio done*/
323 /* Used to pass group descriptor data when online resize is done */
324 struct ext4_new_group_input
{
325 __u32 group
; /* Group number for this data */
326 __u64 block_bitmap
; /* Absolute block number of block bitmap */
327 __u64 inode_bitmap
; /* Absolute block number of inode bitmap */
328 __u64 inode_table
; /* Absolute block number of inode table start */
329 __u32 blocks_count
; /* Total number of blocks in this group */
330 __u16 reserved_blocks
; /* Number of reserved blocks in this group */
334 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
335 struct ext4_new_group_data
{
341 __u16 reserved_blocks
;
343 __u32 free_blocks_count
;
347 * Flags used by ext4_get_blocks()
349 /* Allocate any needed blocks and/or convert an unitialized
350 extent to be an initialized ext4 */
351 #define EXT4_GET_BLOCKS_CREATE 0x0001
352 /* Request the creation of an unitialized extent */
353 #define EXT4_GET_BLOCKS_UNINIT_EXT 0x0002
354 #define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT (EXT4_GET_BLOCKS_UNINIT_EXT|\
355 EXT4_GET_BLOCKS_CREATE)
356 /* Caller is from the delayed allocation writeout path,
357 so set the magic i_delalloc_reserve_flag after taking the
358 inode allocation semaphore for */
359 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
360 /* Call ext4_da_update_reserve_space() after successfully
361 allocating the blocks */
362 #define EXT4_GET_BLOCKS_UPDATE_RESERVE_SPACE 0x0008
363 /* caller is from the direct IO path, request to creation of an
364 unitialized extents if not allocated, split the uninitialized
365 extent if blocks has been preallocated already*/
366 #define EXT4_GET_BLOCKS_DIO 0x0010
367 #define EXT4_GET_BLOCKS_CONVERT 0x0020
368 #define EXT4_GET_BLOCKS_DIO_CREATE_EXT (EXT4_GET_BLOCKS_DIO|\
369 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
370 /* Convert extent to initialized after direct IO complete */
371 #define EXT4_GET_BLOCKS_DIO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
372 EXT4_GET_BLOCKS_DIO_CREATE_EXT)
377 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
378 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
379 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
380 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
381 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
382 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
383 #ifdef CONFIG_JBD2_DEBUG
384 #define EXT4_IOC_WAIT_FOR_READONLY _IOR('f', 99, long)
386 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
387 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
388 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
389 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
390 #define EXT4_IOC_MIGRATE _IO('f', 9)
391 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
392 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
393 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
394 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
397 * ioctl commands in 32 bit emulation
399 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
400 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
401 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
402 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
403 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
404 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
405 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
406 #ifdef CONFIG_JBD2_DEBUG
407 #define EXT4_IOC32_WAIT_FOR_READONLY _IOR('f', 99, int)
409 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
410 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
416 struct ext4_mount_options
{
417 unsigned long s_mount_opt
;
420 unsigned long s_commit_interval
;
421 u32 s_min_batch_time
, s_max_batch_time
;
424 char *s_qf_names
[MAXQUOTAS
];
428 /* Max physical block we can addres w/o extents */
429 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
432 * Structure of an inode on the disk
435 __le16 i_mode
; /* File mode */
436 __le16 i_uid
; /* Low 16 bits of Owner Uid */
437 __le32 i_size_lo
; /* Size in bytes */
438 __le32 i_atime
; /* Access time */
439 __le32 i_ctime
; /* Inode Change time */
440 __le32 i_mtime
; /* Modification time */
441 __le32 i_dtime
; /* Deletion Time */
442 __le16 i_gid
; /* Low 16 bits of Group Id */
443 __le16 i_links_count
; /* Links count */
444 __le32 i_blocks_lo
; /* Blocks count */
445 __le32 i_flags
; /* File flags */
451 __u32 h_i_translator
;
456 } osd1
; /* OS dependent 1 */
457 __le32 i_block
[EXT4_N_BLOCKS
];/* Pointers to blocks */
458 __le32 i_generation
; /* File version (for NFS) */
459 __le32 i_file_acl_lo
; /* File ACL */
461 __le32 i_obso_faddr
; /* Obsoleted fragment address */
464 __le16 l_i_blocks_high
; /* were l_i_reserved1 */
465 __le16 l_i_file_acl_high
;
466 __le16 l_i_uid_high
; /* these 2 fields */
467 __le16 l_i_gid_high
; /* were reserved2[0] */
471 __le16 h_i_reserved1
; /* Obsoleted fragment number/size which are removed in ext4 */
478 __le16 h_i_reserved1
; /* Obsoleted fragment number/size which are removed in ext4 */
479 __le16 m_i_file_acl_high
;
480 __u32 m_i_reserved2
[2];
482 } osd2
; /* OS dependent 2 */
483 __le16 i_extra_isize
;
485 __le32 i_ctime_extra
; /* extra Change time (nsec << 2 | epoch) */
486 __le32 i_mtime_extra
; /* extra Modification time(nsec << 2 | epoch) */
487 __le32 i_atime_extra
; /* extra Access time (nsec << 2 | epoch) */
488 __le32 i_crtime
; /* File Creation time */
489 __le32 i_crtime_extra
; /* extra FileCreationtime (nsec << 2 | epoch) */
490 __le32 i_version_hi
; /* high 32 bits for 64-bit version */
494 __u32 reserved
; /* should be zero */
495 __u32 donor_fd
; /* donor file descriptor */
496 __u64 orig_start
; /* logical start offset in block for orig */
497 __u64 donor_start
; /* logical start offset in block for donor */
498 __u64 len
; /* block length to be moved */
499 __u64 moved_len
; /* moved block length */
501 #define MAX_DEFRAG_SIZE ((1UL<<31) - 1)
503 #define EXT4_EPOCH_BITS 2
504 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
505 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
508 * Extended fields will fit into an inode if the filesystem was formatted
509 * with large inodes (-I 256 or larger) and there are not currently any EAs
510 * consuming all of the available space. For new inodes we always reserve
511 * enough space for the kernel's known extended fields, but for inodes
512 * created with an old kernel this might not have been the case. None of
513 * the extended inode fields is critical for correct filesystem operation.
514 * This macro checks if a certain field fits in the inode. Note that
515 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
517 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
518 ((offsetof(typeof(*ext4_inode), field) + \
519 sizeof((ext4_inode)->field)) \
520 <= (EXT4_GOOD_OLD_INODE_SIZE + \
521 (einode)->i_extra_isize)) \
523 static inline __le32 ext4_encode_extra_time(struct timespec *time)
525 return cpu_to_le32((sizeof(time
->tv_sec
) > 4 ?
526 (time
->tv_sec
>> 32) & EXT4_EPOCH_MASK
: 0) |
527 ((time
->tv_nsec
<< EXT4_EPOCH_BITS
) & EXT4_NSEC_MASK
));
530 static inline void ext4_decode_extra_time(struct timespec
*time
, __le32 extra
)
532 if (sizeof(time
->tv_sec
) > 4)
533 time
->tv_sec
|= (__u64
)(le32_to_cpu(extra
) & EXT4_EPOCH_MASK
)
535 time
->tv_nsec
= (le32_to_cpu(extra
) & EXT4_NSEC_MASK
) >> EXT4_EPOCH_BITS
;
538 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
540 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
541 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
542 (raw_inode)->xtime ## _extra = \
543 ext4_encode_extra_time(&(inode)->xtime); \
546 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
548 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
549 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
550 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
551 (raw_inode)->xtime ## _extra = \
552 ext4_encode_extra_time(&(einode)->xtime); \
555 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
557 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
558 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
559 ext4_decode_extra_time(&(inode)->xtime, \
560 raw_inode->xtime ## _extra); \
563 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
565 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
566 (einode)->xtime.tv_sec = \
567 (signed)le32_to_cpu((raw_inode)->xtime); \
568 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
569 ext4_decode_extra_time(&(einode)->xtime, \
570 raw_inode->xtime ## _extra); \
573 #define i_disk_version osd1.linux1.l_i_version
575 #if defined(__KERNEL__) || defined(__linux__)
576 #define i_reserved1 osd1.linux1.l_i_reserved1
577 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
578 #define i_blocks_high osd2.linux2.l_i_blocks_high
579 #define i_uid_low i_uid
580 #define i_gid_low i_gid
581 #define i_uid_high osd2.linux2.l_i_uid_high
582 #define i_gid_high osd2.linux2.l_i_gid_high
583 #define i_reserved2 osd2.linux2.l_i_reserved2
585 #elif defined(__GNU__)
587 #define i_translator osd1.hurd1.h_i_translator
588 #define i_uid_high osd2.hurd2.h_i_uid_high
589 #define i_gid_high osd2.hurd2.h_i_gid_high
590 #define i_author osd2.hurd2.h_i_author
592 #elif defined(__masix__)
594 #define i_reserved1 osd1.masix1.m_i_reserved1
595 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
596 #define i_reserved2 osd2.masix2.m_i_reserved2
598 #endif /* defined(__KERNEL__) || defined(__linux__) */
601 * storage for cached extent
603 struct ext4_ext_cache
{
604 ext4_fsblk_t ec_start
;
605 ext4_lblk_t ec_block
;
606 __u32 ec_len
; /* must be 32bit to return holes */
611 * fourth extended file system inode data in memory
613 struct ext4_inode_info
{
614 __le32 i_data
[15]; /* unconverted */
616 ext4_fsblk_t i_file_acl
;
620 * i_block_group is the number of the block group which contains
621 * this file's inode. Constant across the lifetime of the inode,
622 * it is ued for making block allocation decisions - we try to
623 * place a file's data blocks near its inode block, and new inodes
624 * near to their parent directory's inode.
626 ext4_group_t i_block_group
;
627 __u32 i_state
; /* Dynamic state flags for ext4 */
629 ext4_lblk_t i_dir_start_lookup
;
630 #ifdef CONFIG_EXT4_FS_XATTR
632 * Extended attributes can be read independently of the main file
633 * data. Taking i_mutex even when reading would cause contention
634 * between readers of EAs and writers of regular file data, so
635 * instead we synchronize on xattr_sem when reading or changing
638 struct rw_semaphore xattr_sem
;
641 struct list_head i_orphan
; /* unlinked but open inodes */
644 * i_disksize keeps track of what the inode size is ON DISK, not
645 * in memory. During truncate, i_size is set to the new size by
646 * the VFS prior to calling ext4_truncate(), but the filesystem won't
647 * set i_disksize to 0 until the truncate is actually under way.
649 * The intent is that i_disksize always represents the blocks which
650 * are used by this file. This allows recovery to restart truncate
651 * on orphans if we crash during truncate. We actually write i_disksize
652 * into the on-disk inode when writing inodes out, instead of i_size.
654 * The only time when i_disksize and i_size may be different is when
655 * a truncate is in progress. The only things which change i_disksize
656 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
661 * i_data_sem is for serialising ext4_truncate() against
662 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
663 * data tree are chopped off during truncate. We can't do that in
664 * ext4 because whenever we perform intermediate commits during
665 * truncate, the inode and all the metadata blocks *must* be in a
666 * consistent state which allows truncation of the orphans to restart
667 * during recovery. Hence we must fix the get_block-vs-truncate race
668 * by other means, so we have i_data_sem.
670 struct rw_semaphore i_data_sem
;
671 struct inode vfs_inode
;
672 struct jbd2_inode jinode
;
674 struct ext4_ext_cache i_cached_extent
;
676 * File creation time. Its function is same as that of
677 * struct timespec i_{a,c,m}time in the generic inode.
679 struct timespec i_crtime
;
682 struct list_head i_prealloc_list
;
683 spinlock_t i_prealloc_lock
;
686 ext4_group_t i_last_alloc_group
;
688 /* allocation reservation info for delalloc */
689 unsigned int i_reserved_data_blocks
;
690 unsigned int i_reserved_meta_blocks
;
691 unsigned int i_allocated_meta_blocks
;
692 unsigned short i_delalloc_reserved_flag
;
694 /* on-disk additional length */
697 spinlock_t i_block_reservation_lock
;
699 /* completed async DIOs that might need unwritten extents handling */
700 struct list_head i_aio_dio_complete_list
;
701 /* current io_end structure for async DIO write*/
702 ext4_io_end_t
*cur_aio_dio
;
705 * Transactions that contain inode's metadata needed to complete
706 * fsync and fdatasync, respectively.
709 tid_t i_datasync_tid
;
715 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
716 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
717 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
720 * Misc. filesystem flags
722 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
723 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
724 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
729 #define EXT4_MOUNT_OLDALLOC 0x00002 /* Don't use the new Orlov allocator */
730 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
731 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
732 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
733 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
734 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
735 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
736 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
737 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
738 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
739 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
740 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
741 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
742 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
743 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
744 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
745 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
746 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
747 #define EXT4_MOUNT_NOBH 0x40000 /* No bufferheads */
748 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
749 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
750 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
751 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
752 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
753 #define EXT4_MOUNT_I_VERSION 0x2000000 /* i_version support */
754 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
755 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
756 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
757 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
759 #define clear_opt(o, opt) o &= ~EXT4_MOUNT_##opt
760 #define set_opt(o, opt) o |= EXT4_MOUNT_##opt
761 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
764 #define ext4_set_bit ext2_set_bit
765 #define ext4_set_bit_atomic ext2_set_bit_atomic
766 #define ext4_clear_bit ext2_clear_bit
767 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
768 #define ext4_test_bit ext2_test_bit
769 #define ext4_find_first_zero_bit ext2_find_first_zero_bit
770 #define ext4_find_next_zero_bit ext2_find_next_zero_bit
771 #define ext4_find_next_bit ext2_find_next_bit
774 * Maximal mount counts between two filesystem checks
776 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
777 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
780 * Behaviour when detecting errors
782 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
783 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
784 #define EXT4_ERRORS_PANIC 3 /* Panic */
785 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
788 * Structure of the super block
790 struct ext4_super_block
{
791 /*00*/ __le32 s_inodes_count
; /* Inodes count */
792 __le32 s_blocks_count_lo
; /* Blocks count */
793 __le32 s_r_blocks_count_lo
; /* Reserved blocks count */
794 __le32 s_free_blocks_count_lo
; /* Free blocks count */
795 /*10*/ __le32 s_free_inodes_count
; /* Free inodes count */
796 __le32 s_first_data_block
; /* First Data Block */
797 __le32 s_log_block_size
; /* Block size */
798 __le32 s_obso_log_frag_size
; /* Obsoleted fragment size */
799 /*20*/ __le32 s_blocks_per_group
; /* # Blocks per group */
800 __le32 s_obso_frags_per_group
; /* Obsoleted fragments per group */
801 __le32 s_inodes_per_group
; /* # Inodes per group */
802 __le32 s_mtime
; /* Mount time */
803 /*30*/ __le32 s_wtime
; /* Write time */
804 __le16 s_mnt_count
; /* Mount count */
805 __le16 s_max_mnt_count
; /* Maximal mount count */
806 __le16 s_magic
; /* Magic signature */
807 __le16 s_state
; /* File system state */
808 __le16 s_errors
; /* Behaviour when detecting errors */
809 __le16 s_minor_rev_level
; /* minor revision level */
810 /*40*/ __le32 s_lastcheck
; /* time of last check */
811 __le32 s_checkinterval
; /* max. time between checks */
812 __le32 s_creator_os
; /* OS */
813 __le32 s_rev_level
; /* Revision level */
814 /*50*/ __le16 s_def_resuid
; /* Default uid for reserved blocks */
815 __le16 s_def_resgid
; /* Default gid for reserved blocks */
817 * These fields are for EXT4_DYNAMIC_REV superblocks only.
819 * Note: the difference between the compatible feature set and
820 * the incompatible feature set is that if there is a bit set
821 * in the incompatible feature set that the kernel doesn't
822 * know about, it should refuse to mount the filesystem.
824 * e2fsck's requirements are more strict; if it doesn't know
825 * about a feature in either the compatible or incompatible
826 * feature set, it must abort and not try to meddle with
827 * things it doesn't understand...
829 __le32 s_first_ino
; /* First non-reserved inode */
830 __le16 s_inode_size
; /* size of inode structure */
831 __le16 s_block_group_nr
; /* block group # of this superblock */
832 __le32 s_feature_compat
; /* compatible feature set */
833 /*60*/ __le32 s_feature_incompat
; /* incompatible feature set */
834 __le32 s_feature_ro_compat
; /* readonly-compatible feature set */
835 /*68*/ __u8 s_uuid
[16]; /* 128-bit uuid for volume */
836 /*78*/ char s_volume_name
[16]; /* volume name */
837 /*88*/ char s_last_mounted
[64]; /* directory where last mounted */
838 /*C8*/ __le32 s_algorithm_usage_bitmap
; /* For compression */
840 * Performance hints. Directory preallocation should only
841 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
843 __u8 s_prealloc_blocks
; /* Nr of blocks to try to preallocate*/
844 __u8 s_prealloc_dir_blocks
; /* Nr to preallocate for dirs */
845 __le16 s_reserved_gdt_blocks
; /* Per group desc for online growth */
847 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
849 /*D0*/ __u8 s_journal_uuid
[16]; /* uuid of journal superblock */
850 /*E0*/ __le32 s_journal_inum
; /* inode number of journal file */
851 __le32 s_journal_dev
; /* device number of journal file */
852 __le32 s_last_orphan
; /* start of list of inodes to delete */
853 __le32 s_hash_seed
[4]; /* HTREE hash seed */
854 __u8 s_def_hash_version
; /* Default hash version to use */
855 __u8 s_reserved_char_pad
;
856 __le16 s_desc_size
; /* size of group descriptor */
857 /*100*/ __le32 s_default_mount_opts
;
858 __le32 s_first_meta_bg
; /* First metablock block group */
859 __le32 s_mkfs_time
; /* When the filesystem was created */
860 __le32 s_jnl_blocks
[17]; /* Backup of the journal inode */
861 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
862 /*150*/ __le32 s_blocks_count_hi
; /* Blocks count */
863 __le32 s_r_blocks_count_hi
; /* Reserved blocks count */
864 __le32 s_free_blocks_count_hi
; /* Free blocks count */
865 __le16 s_min_extra_isize
; /* All inodes have at least # bytes */
866 __le16 s_want_extra_isize
; /* New inodes should reserve # bytes */
867 __le32 s_flags
; /* Miscellaneous flags */
868 __le16 s_raid_stride
; /* RAID stride */
869 __le16 s_mmp_interval
; /* # seconds to wait in MMP checking */
870 __le64 s_mmp_block
; /* Block for multi-mount protection */
871 __le32 s_raid_stripe_width
; /* blocks on all data disks (N*stride)*/
872 __u8 s_log_groups_per_flex
; /* FLEX_BG group size */
873 __u8 s_reserved_char_pad2
;
874 __le16 s_reserved_pad
;
875 __le64 s_kbytes_written
; /* nr of lifetime kilobytes written */
876 __u32 s_reserved
[160]; /* Padding to the end of the block */
882 * run-time mount flags
884 #define EXT4_MF_MNTDIR_SAMPLED 0x0001
885 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
888 * fourth extended-fs super-block data in memory
890 struct ext4_sb_info
{
891 unsigned long s_desc_size
; /* Size of a group descriptor in bytes */
892 unsigned long s_inodes_per_block
;/* Number of inodes per block */
893 unsigned long s_blocks_per_group
;/* Number of blocks in a group */
894 unsigned long s_inodes_per_group
;/* Number of inodes in a group */
895 unsigned long s_itb_per_group
; /* Number of inode table blocks per group */
896 unsigned long s_gdb_count
; /* Number of group descriptor blocks */
897 unsigned long s_desc_per_block
; /* Number of group descriptors per block */
898 ext4_group_t s_groups_count
; /* Number of groups in the fs */
899 ext4_group_t s_blockfile_groups
;/* Groups acceptable for non-extent files */
900 unsigned long s_overhead_last
; /* Last calculated overhead */
901 unsigned long s_blocks_last
; /* Last seen block count */
902 loff_t s_bitmap_maxbytes
; /* max bytes for bitmap files */
903 struct buffer_head
* s_sbh
; /* Buffer containing the super block */
904 struct ext4_super_block
*s_es
; /* Pointer to the super block in the buffer */
905 struct buffer_head
**s_group_desc
;
906 unsigned int s_mount_opt
;
907 unsigned int s_mount_flags
;
908 ext4_fsblk_t s_sb_block
;
911 unsigned short s_mount_state
;
912 unsigned short s_pad
;
913 int s_addr_per_block_bits
;
914 int s_desc_per_block_bits
;
917 unsigned int s_inode_readahead_blks
;
918 unsigned int s_inode_goal
;
919 spinlock_t s_next_gen_lock
;
920 u32 s_next_generation
;
922 int s_def_hash_version
;
923 int s_hash_unsigned
; /* 3 if hash should be signed, 0 if not */
924 struct percpu_counter s_freeblocks_counter
;
925 struct percpu_counter s_freeinodes_counter
;
926 struct percpu_counter s_dirs_counter
;
927 struct percpu_counter s_dirtyblocks_counter
;
928 struct blockgroup_lock
*s_blockgroup_lock
;
929 struct proc_dir_entry
*s_proc
;
930 struct kobject s_kobj
;
931 struct completion s_kobj_unregister
;
934 struct inode
*s_journal_inode
;
935 struct journal_s
*s_journal
;
936 struct list_head s_orphan
;
937 struct mutex s_orphan_lock
;
938 struct mutex s_resize_lock
;
939 unsigned long s_commit_interval
;
940 u32 s_max_batch_time
;
941 u32 s_min_batch_time
;
942 struct block_device
*journal_bdev
;
943 #ifdef CONFIG_JBD2_DEBUG
944 struct timer_list turn_ro_timer
; /* For turning read-only (crash simulation) */
945 wait_queue_head_t ro_wait_queue
; /* For people waiting for the fs to go read-only */
948 char *s_qf_names
[MAXQUOTAS
]; /* Names of quota files with journalled quota */
949 int s_jquota_fmt
; /* Format of quota to use */
951 unsigned int s_want_extra_isize
; /* New inodes should reserve # bytes */
952 struct rb_root system_blks
;
955 /* ext4 extents stats */
956 unsigned long s_ext_min
;
957 unsigned long s_ext_max
;
958 unsigned long s_depth_max
;
959 spinlock_t s_ext_stats_lock
;
960 unsigned long s_ext_blocks
;
961 unsigned long s_ext_extents
;
964 /* for buddy allocator */
965 struct ext4_group_info
***s_group_info
;
966 struct inode
*s_buddy_cache
;
967 long s_blocks_reserved
;
968 spinlock_t s_reserve_lock
;
969 spinlock_t s_md_lock
;
970 tid_t s_last_transaction
;
971 unsigned short *s_mb_offsets
;
972 unsigned int *s_mb_maxs
;
975 unsigned long s_stripe
;
976 unsigned int s_mb_stream_request
;
977 unsigned int s_mb_max_to_scan
;
978 unsigned int s_mb_min_to_scan
;
979 unsigned int s_mb_stats
;
980 unsigned int s_mb_order2_reqs
;
981 unsigned int s_mb_group_prealloc
;
982 unsigned int s_max_writeback_mb_bump
;
983 /* where last allocation was done - for stream allocation */
984 unsigned long s_mb_last_group
;
985 unsigned long s_mb_last_start
;
987 /* history to debug policy */
988 struct ext4_mb_history
*s_mb_history
;
989 int s_mb_history_cur
;
990 int s_mb_history_max
;
991 int s_mb_history_num
;
992 spinlock_t s_mb_history_lock
;
993 int s_mb_history_filter
;
995 /* stats for buddy allocator */
996 spinlock_t s_mb_pa_lock
;
997 atomic_t s_bal_reqs
; /* number of reqs with len > 1 */
998 atomic_t s_bal_success
; /* we found long enough chunks */
999 atomic_t s_bal_allocated
; /* in blocks */
1000 atomic_t s_bal_ex_scanned
; /* total extents scanned */
1001 atomic_t s_bal_goals
; /* goal hits */
1002 atomic_t s_bal_breaks
; /* too long searches */
1003 atomic_t s_bal_2orders
; /* 2^order hits */
1004 spinlock_t s_bal_lock
;
1005 unsigned long s_mb_buddies_generated
;
1006 unsigned long long s_mb_generation_time
;
1007 atomic_t s_mb_lost_chunks
;
1008 atomic_t s_mb_preallocated
;
1009 atomic_t s_mb_discarded
;
1010 atomic_t s_lock_busy
;
1012 /* locality groups */
1013 struct ext4_locality_group
*s_locality_groups
;
1015 /* for write statistics */
1016 unsigned long s_sectors_written_start
;
1017 u64 s_kbytes_written
;
1019 unsigned int s_log_groups_per_flex
;
1020 struct flex_groups
*s_flex_groups
;
1022 /* workqueue for dio unwritten */
1023 struct workqueue_struct
*dio_unwritten_wq
;
1026 static inline struct ext4_sb_info
*EXT4_SB(struct super_block
*sb
)
1028 return sb
->s_fs_info
;
1030 static inline struct ext4_inode_info
*EXT4_I(struct inode
*inode
)
1032 return container_of(inode
, struct ext4_inode_info
, vfs_inode
);
1035 static inline struct timespec
ext4_current_time(struct inode
*inode
)
1037 return (inode
->i_sb
->s_time_gran
< NSEC_PER_SEC
) ?
1038 current_fs_time(inode
->i_sb
) : CURRENT_TIME_SEC
;
1041 static inline int ext4_valid_inum(struct super_block
*sb
, unsigned long ino
)
1043 return ino
== EXT4_ROOT_INO
||
1044 ino
== EXT4_JOURNAL_INO
||
1045 ino
== EXT4_RESIZE_INO
||
1046 (ino
>= EXT4_FIRST_INO(sb
) &&
1047 ino
<= le32_to_cpu(EXT4_SB(sb
)->s_es
->s_inodes_count
));
1050 /* Assume that user mode programs are passing in an ext4fs superblock, not
1051 * a kernel struct super_block. This will allow us to call the feature-test
1052 * macros from user land. */
1053 #define EXT4_SB(sb) (sb)
1056 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1059 * Codes for operating systems
1061 #define EXT4_OS_LINUX 0
1062 #define EXT4_OS_HURD 1
1063 #define EXT4_OS_MASIX 2
1064 #define EXT4_OS_FREEBSD 3
1065 #define EXT4_OS_LITES 4
1070 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1071 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1073 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1074 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1076 #define EXT4_GOOD_OLD_INODE_SIZE 128
1079 * Feature set definitions
1082 #define EXT4_HAS_COMPAT_FEATURE(sb,mask) \
1083 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1084 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \
1085 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1086 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \
1087 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1088 #define EXT4_SET_COMPAT_FEATURE(sb,mask) \
1089 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1090 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \
1091 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1092 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \
1093 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1094 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \
1095 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1096 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
1097 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1098 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \
1099 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1101 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1102 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1103 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1104 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1105 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1106 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1108 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1109 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1110 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1111 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1112 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1113 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1114 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1116 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1117 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1118 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1119 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1120 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1121 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1122 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1123 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1124 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1126 #define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR
1127 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1128 EXT4_FEATURE_INCOMPAT_RECOVER| \
1129 EXT4_FEATURE_INCOMPAT_META_BG| \
1130 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1131 EXT4_FEATURE_INCOMPAT_64BIT| \
1132 EXT4_FEATURE_INCOMPAT_FLEX_BG)
1133 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1134 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1135 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1136 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1137 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1138 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1139 EXT4_FEATURE_RO_COMPAT_HUGE_FILE)
1142 * Default values for user and/or group using reserved blocks
1144 #define EXT4_DEF_RESUID 0
1145 #define EXT4_DEF_RESGID 0
1147 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
1150 * Default mount options
1152 #define EXT4_DEFM_DEBUG 0x0001
1153 #define EXT4_DEFM_BSDGROUPS 0x0002
1154 #define EXT4_DEFM_XATTR_USER 0x0004
1155 #define EXT4_DEFM_ACL 0x0008
1156 #define EXT4_DEFM_UID16 0x0010
1157 #define EXT4_DEFM_JMODE 0x0060
1158 #define EXT4_DEFM_JMODE_DATA 0x0020
1159 #define EXT4_DEFM_JMODE_ORDERED 0x0040
1160 #define EXT4_DEFM_JMODE_WBACK 0x0060
1163 * Default journal batch times
1165 #define EXT4_DEF_MIN_BATCH_TIME 0
1166 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1169 * Minimum number of groups in a flexgroup before we separate out
1170 * directories into the first block group of a flexgroup
1172 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1175 * Structure of a directory entry
1177 #define EXT4_NAME_LEN 255
1179 struct ext4_dir_entry
{
1180 __le32 inode
; /* Inode number */
1181 __le16 rec_len
; /* Directory entry length */
1182 __le16 name_len
; /* Name length */
1183 char name
[EXT4_NAME_LEN
]; /* File name */
1187 * The new version of the directory entry. Since EXT4 structures are
1188 * stored in intel byte order, and the name_len field could never be
1189 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1192 struct ext4_dir_entry_2
{
1193 __le32 inode
; /* Inode number */
1194 __le16 rec_len
; /* Directory entry length */
1195 __u8 name_len
; /* Name length */
1197 char name
[EXT4_NAME_LEN
]; /* File name */
1201 * Ext4 directory file types. Only the low 3 bits are used. The
1202 * other bits are reserved for now.
1204 #define EXT4_FT_UNKNOWN 0
1205 #define EXT4_FT_REG_FILE 1
1206 #define EXT4_FT_DIR 2
1207 #define EXT4_FT_CHRDEV 3
1208 #define EXT4_FT_BLKDEV 4
1209 #define EXT4_FT_FIFO 5
1210 #define EXT4_FT_SOCK 6
1211 #define EXT4_FT_SYMLINK 7
1213 #define EXT4_FT_MAX 8
1216 * EXT4_DIR_PAD defines the directory entries boundaries
1218 * NOTE: It must be a multiple of 4
1220 #define EXT4_DIR_PAD 4
1221 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1222 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1224 #define EXT4_MAX_REC_LEN ((1<<16)-1)
1227 * Hash Tree Directory indexing
1228 * (c) Daniel Phillips, 2001
1231 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1232 EXT4_FEATURE_COMPAT_DIR_INDEX) && \
1233 (EXT4_I(dir)->i_flags & EXT4_INDEX_FL))
1234 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1235 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1237 /* Legal values for the dx_root hash_version field: */
1239 #define DX_HASH_LEGACY 0
1240 #define DX_HASH_HALF_MD4 1
1241 #define DX_HASH_TEA 2
1242 #define DX_HASH_LEGACY_UNSIGNED 3
1243 #define DX_HASH_HALF_MD4_UNSIGNED 4
1244 #define DX_HASH_TEA_UNSIGNED 5
1248 /* hash info structure used by the directory hash */
1257 #define EXT4_HTREE_EOF 0x7fffffff
1260 * Control parameters used by ext4_htree_next_block
1262 #define HASH_NB_ALWAYS 1
1266 * Describe an inode's exact location on disk and in memory
1270 struct buffer_head
*bh
;
1271 unsigned long offset
;
1272 ext4_group_t block_group
;
1275 static inline struct ext4_inode
*ext4_raw_inode(struct ext4_iloc
*iloc
)
1277 return (struct ext4_inode
*) (iloc
->bh
->b_data
+ iloc
->offset
);
1281 * This structure is stuffed into the struct file's private_data field
1282 * for directories. It is where we put information so that we can do
1283 * readdir operations in hash tree order.
1285 struct dir_private_info
{
1286 struct rb_root root
;
1287 struct rb_node
*curr_node
;
1288 struct fname
*extra_fname
;
1291 __u32 curr_minor_hash
;
1295 /* calculate the first block number of the group */
1296 static inline ext4_fsblk_t
1297 ext4_group_first_block_no(struct super_block
*sb
, ext4_group_t group_no
)
1299 return group_no
* (ext4_fsblk_t
)EXT4_BLOCKS_PER_GROUP(sb
) +
1300 le32_to_cpu(EXT4_SB(sb
)->s_es
->s_first_data_block
);
1304 * Special error return code only used by dx_probe() and its callers.
1306 #define ERR_BAD_DX_DIR -75000
1308 void ext4_get_group_no_and_offset(struct super_block
*sb
, ext4_fsblk_t blocknr
,
1309 ext4_group_t
*blockgrpp
, ext4_grpblk_t
*offsetp
);
1311 extern struct proc_dir_entry
*ext4_proc_root
;
1314 * Function prototypes
1318 * Ok, these declarations are also in <linux/kernel.h> but none of the
1319 * ext4 source programs needs to include it so they are duplicated here.
1321 # define NORET_TYPE /**/
1322 # define ATTRIB_NORET __attribute__((noreturn))
1323 # define NORET_AND noreturn,
1326 extern unsigned int ext4_count_free(struct buffer_head
*, unsigned);
1329 extern unsigned int ext4_block_group(struct super_block
*sb
,
1330 ext4_fsblk_t blocknr
);
1331 extern ext4_grpblk_t
ext4_block_group_offset(struct super_block
*sb
,
1332 ext4_fsblk_t blocknr
);
1333 extern int ext4_bg_has_super(struct super_block
*sb
, ext4_group_t group
);
1334 extern unsigned long ext4_bg_num_gdb(struct super_block
*sb
,
1335 ext4_group_t group
);
1336 extern ext4_fsblk_t
ext4_new_meta_blocks(handle_t
*handle
, struct inode
*inode
,
1337 ext4_fsblk_t goal
, unsigned long *count
, int *errp
);
1338 extern int ext4_claim_free_blocks(struct ext4_sb_info
*sbi
, s64 nblocks
);
1339 extern int ext4_has_free_blocks(struct ext4_sb_info
*sbi
, s64 nblocks
);
1340 extern void ext4_free_blocks(handle_t
*handle
, struct inode
*inode
,
1341 ext4_fsblk_t block
, unsigned long count
, int metadata
);
1342 extern void ext4_add_groupblocks(handle_t
*handle
, struct super_block
*sb
,
1343 ext4_fsblk_t block
, unsigned long count
);
1344 extern ext4_fsblk_t
ext4_count_free_blocks(struct super_block
*);
1345 extern void ext4_check_blocks_bitmap(struct super_block
*);
1346 extern struct ext4_group_desc
* ext4_get_group_desc(struct super_block
* sb
,
1347 ext4_group_t block_group
,
1348 struct buffer_head
** bh
);
1349 extern int ext4_should_retry_alloc(struct super_block
*sb
, int *retries
);
1350 struct buffer_head
*ext4_read_block_bitmap(struct super_block
*sb
,
1351 ext4_group_t block_group
);
1352 extern unsigned ext4_init_block_bitmap(struct super_block
*sb
,
1353 struct buffer_head
*bh
,
1355 struct ext4_group_desc
*desc
);
1356 #define ext4_free_blocks_after_init(sb, group, desc) \
1357 ext4_init_block_bitmap(sb, NULL, group, desc)
1360 extern int ext4_check_dir_entry(const char *, struct inode
*,
1361 struct ext4_dir_entry_2
*,
1362 struct buffer_head
*, unsigned int);
1363 extern int ext4_htree_store_dirent(struct file
*dir_file
, __u32 hash
,
1365 struct ext4_dir_entry_2
*dirent
);
1366 extern void ext4_htree_free_dir_info(struct dir_private_info
*p
);
1369 extern int ext4_sync_file(struct file
*, struct dentry
*, int);
1372 extern int ext4fs_dirhash(const char *name
, int len
, struct
1373 dx_hash_info
*hinfo
);
1376 extern struct inode
*ext4_new_inode(handle_t
*, struct inode
*, int,
1377 const struct qstr
*qstr
, __u32 goal
);
1378 extern void ext4_free_inode(handle_t
*, struct inode
*);
1379 extern struct inode
* ext4_orphan_get(struct super_block
*, unsigned long);
1380 extern unsigned long ext4_count_free_inodes(struct super_block
*);
1381 extern unsigned long ext4_count_dirs(struct super_block
*);
1382 extern void ext4_check_inodes_bitmap(struct super_block
*);
1383 extern unsigned ext4_init_inode_bitmap(struct super_block
*sb
,
1384 struct buffer_head
*bh
,
1386 struct ext4_group_desc
*desc
);
1387 extern void mark_bitmap_end(int start_bit
, int end_bit
, char *bitmap
);
1390 extern long ext4_mb_stats
;
1391 extern long ext4_mb_max_to_scan
;
1392 extern int ext4_mb_init(struct super_block
*, int);
1393 extern int ext4_mb_release(struct super_block
*);
1394 extern ext4_fsblk_t
ext4_mb_new_blocks(handle_t
*,
1395 struct ext4_allocation_request
*, int *);
1396 extern int ext4_mb_reserve_blocks(struct super_block
*, int);
1397 extern void ext4_discard_preallocations(struct inode
*);
1398 extern int __init
init_ext4_mballoc(void);
1399 extern void exit_ext4_mballoc(void);
1400 extern void ext4_mb_free_blocks(handle_t
*, struct inode
*,
1401 ext4_fsblk_t
, unsigned long, int, unsigned long *);
1402 extern int ext4_mb_add_groupinfo(struct super_block
*sb
,
1403 ext4_group_t i
, struct ext4_group_desc
*desc
);
1404 extern void ext4_mb_update_group_info(struct ext4_group_info
*grp
,
1406 extern int ext4_mb_get_buddy_cache_lock(struct super_block
*, ext4_group_t
);
1407 extern void ext4_mb_put_buddy_cache_lock(struct super_block
*,
1410 int ext4_forget(handle_t
*handle
, int is_metadata
, struct inode
*inode
,
1411 struct buffer_head
*bh
, ext4_fsblk_t blocknr
);
1412 struct buffer_head
*ext4_getblk(handle_t
*, struct inode
*,
1413 ext4_lblk_t
, int, int *);
1414 struct buffer_head
*ext4_bread(handle_t
*, struct inode
*,
1415 ext4_lblk_t
, int, int *);
1416 int ext4_get_block(struct inode
*inode
, sector_t iblock
,
1417 struct buffer_head
*bh_result
, int create
);
1419 extern struct inode
*ext4_iget(struct super_block
*, unsigned long);
1420 extern int ext4_write_inode(struct inode
*, int);
1421 extern int ext4_setattr(struct dentry
*, struct iattr
*);
1422 extern int ext4_getattr(struct vfsmount
*mnt
, struct dentry
*dentry
,
1423 struct kstat
*stat
);
1424 extern void ext4_delete_inode(struct inode
*);
1425 extern int ext4_sync_inode(handle_t
*, struct inode
*);
1426 extern void ext4_dirty_inode(struct inode
*);
1427 extern int ext4_change_inode_journal_flag(struct inode
*, int);
1428 extern int ext4_get_inode_loc(struct inode
*, struct ext4_iloc
*);
1429 extern int ext4_can_truncate(struct inode
*inode
);
1430 extern void ext4_truncate(struct inode
*);
1431 extern int ext4_truncate_restart_trans(handle_t
*, struct inode
*, int nblocks
);
1432 extern void ext4_set_inode_flags(struct inode
*);
1433 extern void ext4_get_inode_flags(struct ext4_inode_info
*);
1434 extern int ext4_alloc_da_blocks(struct inode
*inode
);
1435 extern void ext4_set_aops(struct inode
*inode
);
1436 extern int ext4_writepage_trans_blocks(struct inode
*);
1437 extern int ext4_meta_trans_blocks(struct inode
*, int nrblocks
, int idxblocks
);
1438 extern int ext4_chunk_trans_blocks(struct inode
*, int nrblocks
);
1439 extern int ext4_block_truncate_page(handle_t
*handle
,
1440 struct address_space
*mapping
, loff_t from
);
1441 extern int ext4_page_mkwrite(struct vm_area_struct
*vma
, struct vm_fault
*vmf
);
1442 extern qsize_t
ext4_get_reserved_space(struct inode
*inode
);
1443 extern int flush_aio_dio_completed_IO(struct inode
*inode
);
1445 extern long ext4_ioctl(struct file
*, unsigned int, unsigned long);
1446 extern long ext4_compat_ioctl(struct file
*, unsigned int, unsigned long);
1449 extern int ext4_ext_migrate(struct inode
*);
1452 extern unsigned int ext4_rec_len_from_disk(__le16 dlen
, unsigned blocksize
);
1453 extern __le16
ext4_rec_len_to_disk(unsigned len
, unsigned blocksize
);
1454 extern int ext4_orphan_add(handle_t
*, struct inode
*);
1455 extern int ext4_orphan_del(handle_t
*, struct inode
*);
1456 extern int ext4_htree_fill_tree(struct file
*dir_file
, __u32 start_hash
,
1457 __u32 start_minor_hash
, __u32
*next_hash
);
1460 extern int ext4_group_add(struct super_block
*sb
,
1461 struct ext4_new_group_data
*input
);
1462 extern int ext4_group_extend(struct super_block
*sb
,
1463 struct ext4_super_block
*es
,
1464 ext4_fsblk_t n_blocks_count
);
1467 extern void ext4_error(struct super_block
*, const char *, const char *, ...)
1468 __attribute__ ((format (printf
, 3, 4)));
1469 extern void __ext4_std_error(struct super_block
*, const char *, int);
1470 extern void ext4_abort(struct super_block
*, const char *, const char *, ...)
1471 __attribute__ ((format (printf
, 3, 4)));
1472 extern void ext4_warning(struct super_block
*, const char *, const char *, ...)
1473 __attribute__ ((format (printf
, 3, 4)));
1474 extern void ext4_msg(struct super_block
*, const char *, const char *, ...)
1475 __attribute__ ((format (printf
, 3, 4)));
1476 extern void ext4_grp_locked_error(struct super_block
*, ext4_group_t
,
1477 const char *, const char *, ...)
1478 __attribute__ ((format (printf
, 4, 5)));
1479 extern void ext4_update_dynamic_rev(struct super_block
*sb
);
1480 extern int ext4_update_compat_feature(handle_t
*handle
, struct super_block
*sb
,
1482 extern int ext4_update_rocompat_feature(handle_t
*handle
,
1483 struct super_block
*sb
, __u32 rocompat
);
1484 extern int ext4_update_incompat_feature(handle_t
*handle
,
1485 struct super_block
*sb
, __u32 incompat
);
1486 extern ext4_fsblk_t
ext4_block_bitmap(struct super_block
*sb
,
1487 struct ext4_group_desc
*bg
);
1488 extern ext4_fsblk_t
ext4_inode_bitmap(struct super_block
*sb
,
1489 struct ext4_group_desc
*bg
);
1490 extern ext4_fsblk_t
ext4_inode_table(struct super_block
*sb
,
1491 struct ext4_group_desc
*bg
);
1492 extern __u32
ext4_free_blks_count(struct super_block
*sb
,
1493 struct ext4_group_desc
*bg
);
1494 extern __u32
ext4_free_inodes_count(struct super_block
*sb
,
1495 struct ext4_group_desc
*bg
);
1496 extern __u32
ext4_used_dirs_count(struct super_block
*sb
,
1497 struct ext4_group_desc
*bg
);
1498 extern __u32
ext4_itable_unused_count(struct super_block
*sb
,
1499 struct ext4_group_desc
*bg
);
1500 extern void ext4_block_bitmap_set(struct super_block
*sb
,
1501 struct ext4_group_desc
*bg
, ext4_fsblk_t blk
);
1502 extern void ext4_inode_bitmap_set(struct super_block
*sb
,
1503 struct ext4_group_desc
*bg
, ext4_fsblk_t blk
);
1504 extern void ext4_inode_table_set(struct super_block
*sb
,
1505 struct ext4_group_desc
*bg
, ext4_fsblk_t blk
);
1506 extern void ext4_free_blks_set(struct super_block
*sb
,
1507 struct ext4_group_desc
*bg
, __u32 count
);
1508 extern void ext4_free_inodes_set(struct super_block
*sb
,
1509 struct ext4_group_desc
*bg
, __u32 count
);
1510 extern void ext4_used_dirs_set(struct super_block
*sb
,
1511 struct ext4_group_desc
*bg
, __u32 count
);
1512 extern void ext4_itable_unused_set(struct super_block
*sb
,
1513 struct ext4_group_desc
*bg
, __u32 count
);
1514 extern __le16
ext4_group_desc_csum(struct ext4_sb_info
*sbi
, __u32 group
,
1515 struct ext4_group_desc
*gdp
);
1516 extern int ext4_group_desc_csum_verify(struct ext4_sb_info
*sbi
, __u32 group
,
1517 struct ext4_group_desc
*gdp
);
1519 static inline ext4_fsblk_t
ext4_blocks_count(struct ext4_super_block
*es
)
1521 return ((ext4_fsblk_t
)le32_to_cpu(es
->s_blocks_count_hi
) << 32) |
1522 le32_to_cpu(es
->s_blocks_count_lo
);
1525 static inline ext4_fsblk_t
ext4_r_blocks_count(struct ext4_super_block
*es
)
1527 return ((ext4_fsblk_t
)le32_to_cpu(es
->s_r_blocks_count_hi
) << 32) |
1528 le32_to_cpu(es
->s_r_blocks_count_lo
);
1531 static inline ext4_fsblk_t
ext4_free_blocks_count(struct ext4_super_block
*es
)
1533 return ((ext4_fsblk_t
)le32_to_cpu(es
->s_free_blocks_count_hi
) << 32) |
1534 le32_to_cpu(es
->s_free_blocks_count_lo
);
1537 static inline void ext4_blocks_count_set(struct ext4_super_block
*es
,
1540 es
->s_blocks_count_lo
= cpu_to_le32((u32
)blk
);
1541 es
->s_blocks_count_hi
= cpu_to_le32(blk
>> 32);
1544 static inline void ext4_free_blocks_count_set(struct ext4_super_block
*es
,
1547 es
->s_free_blocks_count_lo
= cpu_to_le32((u32
)blk
);
1548 es
->s_free_blocks_count_hi
= cpu_to_le32(blk
>> 32);
1551 static inline void ext4_r_blocks_count_set(struct ext4_super_block
*es
,
1554 es
->s_r_blocks_count_lo
= cpu_to_le32((u32
)blk
);
1555 es
->s_r_blocks_count_hi
= cpu_to_le32(blk
>> 32);
1558 static inline loff_t
ext4_isize(struct ext4_inode
*raw_inode
)
1560 if (S_ISREG(le16_to_cpu(raw_inode
->i_mode
)))
1561 return ((loff_t
)le32_to_cpu(raw_inode
->i_size_high
) << 32) |
1562 le32_to_cpu(raw_inode
->i_size_lo
);
1564 return (loff_t
) le32_to_cpu(raw_inode
->i_size_lo
);
1567 static inline void ext4_isize_set(struct ext4_inode
*raw_inode
, loff_t i_size
)
1569 raw_inode
->i_size_lo
= cpu_to_le32(i_size
);
1570 raw_inode
->i_size_high
= cpu_to_le32(i_size
>> 32);
1574 struct ext4_group_info
*ext4_get_group_info(struct super_block
*sb
,
1577 struct ext4_group_info
***grp_info
;
1578 long indexv
, indexh
;
1579 grp_info
= EXT4_SB(sb
)->s_group_info
;
1580 indexv
= group
>> (EXT4_DESC_PER_BLOCK_BITS(sb
));
1581 indexh
= group
& ((EXT4_DESC_PER_BLOCK(sb
)) - 1);
1582 return grp_info
[indexv
][indexh
];
1586 * Reading s_groups_count requires using smp_rmb() afterwards. See
1587 * the locking protocol documented in the comments of ext4_group_add()
1590 static inline ext4_group_t
ext4_get_groups_count(struct super_block
*sb
)
1592 ext4_group_t ngroups
= EXT4_SB(sb
)->s_groups_count
;
1598 static inline ext4_group_t
ext4_flex_group(struct ext4_sb_info
*sbi
,
1599 ext4_group_t block_group
)
1601 return block_group
>> sbi
->s_log_groups_per_flex
;
1604 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info
*sbi
)
1606 return 1 << sbi
->s_log_groups_per_flex
;
1609 #define ext4_std_error(sb, errno) \
1612 __ext4_std_error((sb), __func__, (errno)); \
1616 /* Each CPU can accumulate percpu_counter_batch blocks in their local
1617 * counters. So we need to make sure we have free blocks more
1618 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
1620 #define EXT4_FREEBLOCKS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
1622 #define EXT4_FREEBLOCKS_WATERMARK 0
1625 static inline void ext4_update_i_disksize(struct inode
*inode
, loff_t newsize
)
1628 * XXX: replace with spinlock if seen contended -bzzz
1630 down_write(&EXT4_I(inode
)->i_data_sem
);
1631 if (newsize
> EXT4_I(inode
)->i_disksize
)
1632 EXT4_I(inode
)->i_disksize
= newsize
;
1633 up_write(&EXT4_I(inode
)->i_data_sem
);
1637 struct ext4_group_info
{
1638 unsigned long bb_state
;
1639 struct rb_root bb_free_root
;
1640 unsigned short bb_first_free
;
1641 unsigned short bb_free
;
1642 unsigned short bb_fragments
;
1643 struct list_head bb_prealloc_list
;
1647 struct rw_semaphore alloc_sem
;
1648 unsigned short bb_counters
[];
1651 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
1653 #define EXT4_MB_GRP_NEED_INIT(grp) \
1654 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
1656 #define EXT4_MAX_CONTENTION 8
1657 #define EXT4_CONTENTION_THRESHOLD 2
1659 static inline spinlock_t
*ext4_group_lock_ptr(struct super_block
*sb
,
1662 return bgl_lock_ptr(EXT4_SB(sb
)->s_blockgroup_lock
, group
);
1666 * Returns true if the filesystem is busy enough that attempts to
1667 * access the block group locks has run into contention.
1669 static inline int ext4_fs_is_busy(struct ext4_sb_info
*sbi
)
1671 return (atomic_read(&sbi
->s_lock_busy
) > EXT4_CONTENTION_THRESHOLD
);
1674 static inline void ext4_lock_group(struct super_block
*sb
, ext4_group_t group
)
1676 spinlock_t
*lock
= ext4_group_lock_ptr(sb
, group
);
1677 if (spin_trylock(lock
))
1679 * We're able to grab the lock right away, so drop the
1680 * lock contention counter.
1682 atomic_add_unless(&EXT4_SB(sb
)->s_lock_busy
, -1, 0);
1685 * The lock is busy, so bump the contention counter,
1686 * and then wait on the spin lock.
1688 atomic_add_unless(&EXT4_SB(sb
)->s_lock_busy
, 1,
1689 EXT4_MAX_CONTENTION
);
1694 static inline void ext4_unlock_group(struct super_block
*sb
,
1697 spin_unlock(ext4_group_lock_ptr(sb
, group
));
1701 * Inodes and files operations
1705 extern const struct file_operations ext4_dir_operations
;
1708 extern const struct inode_operations ext4_file_inode_operations
;
1709 extern const struct file_operations ext4_file_operations
;
1712 extern const struct inode_operations ext4_dir_inode_operations
;
1713 extern const struct inode_operations ext4_special_inode_operations
;
1714 extern struct dentry
*ext4_get_parent(struct dentry
*child
);
1717 extern const struct inode_operations ext4_symlink_inode_operations
;
1718 extern const struct inode_operations ext4_fast_symlink_inode_operations
;
1720 /* block_validity */
1721 extern void ext4_release_system_zone(struct super_block
*sb
);
1722 extern int ext4_setup_system_zone(struct super_block
*sb
);
1723 extern int __init
init_ext4_system_zone(void);
1724 extern void exit_ext4_system_zone(void);
1725 extern int ext4_data_block_valid(struct ext4_sb_info
*sbi
,
1726 ext4_fsblk_t start_blk
,
1727 unsigned int count
);
1730 extern int ext4_ext_tree_init(handle_t
*handle
, struct inode
*);
1731 extern int ext4_ext_writepage_trans_blocks(struct inode
*, int);
1732 extern int ext4_ext_index_trans_blocks(struct inode
*inode
, int nrblocks
,
1734 extern int ext4_ext_get_blocks(handle_t
*handle
, struct inode
*inode
,
1735 ext4_lblk_t iblock
, unsigned int max_blocks
,
1736 struct buffer_head
*bh_result
, int flags
);
1737 extern void ext4_ext_truncate(struct inode
*);
1738 extern void ext4_ext_init(struct super_block
*);
1739 extern void ext4_ext_release(struct super_block
*);
1740 extern long ext4_fallocate(struct inode
*inode
, int mode
, loff_t offset
,
1742 extern int ext4_convert_unwritten_extents(struct inode
*inode
, loff_t offset
,
1744 extern int ext4_get_blocks(handle_t
*handle
, struct inode
*inode
,
1745 sector_t block
, unsigned int max_blocks
,
1746 struct buffer_head
*bh
, int flags
);
1747 extern int ext4_fiemap(struct inode
*inode
, struct fiemap_extent_info
*fieinfo
,
1748 __u64 start
, __u64 len
);
1750 extern int ext4_move_extents(struct file
*o_filp
, struct file
*d_filp
,
1751 __u64 start_orig
, __u64 start_donor
,
1752 __u64 len
, __u64
*moved_len
);
1756 * Add new method to test wether block and inode bitmaps are properly
1757 * initialized. With uninit_bg reading the block from disk is not enough
1758 * to mark the bitmap uptodate. We need to also zero-out the bitmap
1760 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
1762 static inline int bitmap_uptodate(struct buffer_head
*bh
)
1764 return (buffer_uptodate(bh
) &&
1765 test_bit(BH_BITMAP_UPTODATE
, &(bh
)->b_state
));
1767 static inline void set_bitmap_uptodate(struct buffer_head
*bh
)
1769 set_bit(BH_BITMAP_UPTODATE
, &(bh
)->b_state
);
1772 #endif /* __KERNEL__ */
1774 #endif /* _EXT4_H */