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 * Flags used in mballoc's allocation_context flags field.
71 * Also used to show what's going on for debugging purposes when the
72 * flag field is exported via the traceport interface
75 /* prefer goal again. length */
76 #define EXT4_MB_HINT_MERGE 0x0001
77 /* blocks already reserved */
78 #define EXT4_MB_HINT_RESERVED 0x0002
79 /* metadata is being allocated */
80 #define EXT4_MB_HINT_METADATA 0x0004
81 /* first blocks in the file */
82 #define EXT4_MB_HINT_FIRST 0x0008
83 /* search for the best chunk */
84 #define EXT4_MB_HINT_BEST 0x0010
85 /* data is being allocated */
86 #define EXT4_MB_HINT_DATA 0x0020
87 /* don't preallocate (for tails) */
88 #define EXT4_MB_HINT_NOPREALLOC 0x0040
89 /* allocate for locality group */
90 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080
91 /* allocate goal blocks or none */
92 #define EXT4_MB_HINT_GOAL_ONLY 0x0100
93 /* goal is meaningful */
94 #define EXT4_MB_HINT_TRY_GOAL 0x0200
95 /* blocks already pre-reserved by delayed allocation */
96 #define EXT4_MB_DELALLOC_RESERVED 0x0400
97 /* We are doing stream allocation */
98 #define EXT4_MB_STREAM_ALLOC 0x0800
101 struct ext4_allocation_request
{
102 /* target inode for block we're allocating */
104 /* how many blocks we want to allocate */
106 /* logical block in target inode */
108 /* the closest logical allocated block to the left */
110 /* the closest logical allocated block to the right */
112 /* phys. target (a hint) */
114 /* phys. block for the closest logical allocated block to the left */
116 /* phys. block for the closest logical allocated block to the right */
118 /* flags. see above EXT4_MB_HINT_* */
123 * For delayed allocation tracking
125 struct mpage_da_data
{
127 sector_t b_blocknr
; /* start block number of extent */
128 size_t b_size
; /* size of extent */
129 unsigned long b_state
; /* state of the extent */
130 unsigned long first_page
, next_page
; /* extent of pages */
131 struct writeback_control
*wbc
;
136 #define DIO_AIO_UNWRITTEN 0x1
137 typedef struct ext4_io_end
{
138 struct list_head list
; /* per-file finished AIO list */
139 struct inode
*inode
; /* file being written to */
140 unsigned int flag
; /* unwritten or not */
141 int error
; /* I/O error code */
142 loff_t offset
; /* offset in the file */
143 ssize_t size
; /* size of the extent */
144 struct work_struct work
; /* data work queue */
148 * Special inodes numbers
150 #define EXT4_BAD_INO 1 /* Bad blocks inode */
151 #define EXT4_ROOT_INO 2 /* Root inode */
152 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
153 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
154 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
155 #define EXT4_JOURNAL_INO 8 /* Journal inode */
157 /* First non-reserved inode for old ext4 filesystems */
158 #define EXT4_GOOD_OLD_FIRST_INO 11
161 * Maximal count of links to a file
163 #define EXT4_LINK_MAX 65000
166 * Macro-instructions used to manage several block sizes
168 #define EXT4_MIN_BLOCK_SIZE 1024
169 #define EXT4_MAX_BLOCK_SIZE 65536
170 #define EXT4_MIN_BLOCK_LOG_SIZE 10
172 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
174 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
176 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
178 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
180 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
183 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
184 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
185 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
187 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
188 EXT4_GOOD_OLD_INODE_SIZE : \
190 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
191 EXT4_GOOD_OLD_FIRST_INO : \
194 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
197 * Structure of a blocks group descriptor
199 struct ext4_group_desc
201 __le32 bg_block_bitmap_lo
; /* Blocks bitmap block */
202 __le32 bg_inode_bitmap_lo
; /* Inodes bitmap block */
203 __le32 bg_inode_table_lo
; /* Inodes table block */
204 __le16 bg_free_blocks_count_lo
;/* Free blocks count */
205 __le16 bg_free_inodes_count_lo
;/* Free inodes count */
206 __le16 bg_used_dirs_count_lo
; /* Directories count */
207 __le16 bg_flags
; /* EXT4_BG_flags (INODE_UNINIT, etc) */
208 __u32 bg_reserved
[2]; /* Likely block/inode bitmap checksum */
209 __le16 bg_itable_unused_lo
; /* Unused inodes count */
210 __le16 bg_checksum
; /* crc16(sb_uuid+group+desc) */
211 __le32 bg_block_bitmap_hi
; /* Blocks bitmap block MSB */
212 __le32 bg_inode_bitmap_hi
; /* Inodes bitmap block MSB */
213 __le32 bg_inode_table_hi
; /* Inodes table block MSB */
214 __le16 bg_free_blocks_count_hi
;/* Free blocks count MSB */
215 __le16 bg_free_inodes_count_hi
;/* Free inodes count MSB */
216 __le16 bg_used_dirs_count_hi
; /* Directories count MSB */
217 __le16 bg_itable_unused_hi
; /* Unused inodes count MSB */
218 __u32 bg_reserved2
[3];
222 * Structure of a flex block group info
226 atomic_t free_inodes
;
227 atomic_t free_blocks
;
231 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
232 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
233 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
236 * Macro-instructions used to manage group descriptors
238 #define EXT4_MIN_DESC_SIZE 32
239 #define EXT4_MIN_DESC_SIZE_64BIT 64
240 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
241 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
243 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
244 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
245 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
246 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
248 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
249 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
250 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
254 * Constants relative to the data blocks
256 #define EXT4_NDIR_BLOCKS 12
257 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
258 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
259 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
260 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
265 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
266 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
267 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
268 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
269 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
270 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
271 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
272 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
273 /* Reserved for compression usage... */
274 #define EXT4_DIRTY_FL 0x00000100
275 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
276 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
277 #define EXT4_ECOMPR_FL 0x00000800 /* Compression error */
278 /* End compression flags --- maybe not all used */
279 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
280 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
281 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
282 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
283 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
284 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
285 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
286 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
287 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
289 #define EXT4_FL_USER_VISIBLE 0x000BDFFF /* User visible flags */
290 #define EXT4_FL_USER_MODIFIABLE 0x000B80FF /* User modifiable flags */
292 /* Flags that should be inherited by new inodes from their parent. */
293 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
294 EXT4_SYNC_FL | EXT4_IMMUTABLE_FL | EXT4_APPEND_FL |\
295 EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
296 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
297 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
299 /* Flags that are appropriate for regular files (all but dir-specific ones). */
300 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
302 /* Flags that are appropriate for non-directories/regular files. */
303 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
305 /* Mask out flags that are inappropriate for the given type of inode. */
306 static inline __u32
ext4_mask_flags(umode_t mode
, __u32 flags
)
310 else if (S_ISREG(mode
))
311 return flags
& EXT4_REG_FLMASK
;
313 return flags
& EXT4_OTHER_FLMASK
;
317 * Inode dynamic state flags
319 #define EXT4_STATE_JDATA 0x00000001 /* journaled data exists */
320 #define EXT4_STATE_NEW 0x00000002 /* inode is newly created */
321 #define EXT4_STATE_XATTR 0x00000004 /* has in-inode xattrs */
322 #define EXT4_STATE_NO_EXPAND 0x00000008 /* No space for expansion */
323 #define EXT4_STATE_DA_ALLOC_CLOSE 0x00000010 /* Alloc DA blks on close */
324 #define EXT4_STATE_EXT_MIGRATE 0x00000020 /* Inode is migrating */
325 #define EXT4_STATE_DIO_UNWRITTEN 0x00000040 /* need convert on dio done*/
327 /* Used to pass group descriptor data when online resize is done */
328 struct ext4_new_group_input
{
329 __u32 group
; /* Group number for this data */
330 __u64 block_bitmap
; /* Absolute block number of block bitmap */
331 __u64 inode_bitmap
; /* Absolute block number of inode bitmap */
332 __u64 inode_table
; /* Absolute block number of inode table start */
333 __u32 blocks_count
; /* Total number of blocks in this group */
334 __u16 reserved_blocks
; /* Number of reserved blocks in this group */
338 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
339 struct ext4_new_group_data
{
345 __u16 reserved_blocks
;
347 __u32 free_blocks_count
;
351 * Flags used by ext4_get_blocks()
353 /* Allocate any needed blocks and/or convert an unitialized
354 extent to be an initialized ext4 */
355 #define EXT4_GET_BLOCKS_CREATE 0x0001
356 /* Request the creation of an unitialized extent */
357 #define EXT4_GET_BLOCKS_UNINIT_EXT 0x0002
358 #define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT (EXT4_GET_BLOCKS_UNINIT_EXT|\
359 EXT4_GET_BLOCKS_CREATE)
360 /* Caller is from the delayed allocation writeout path,
361 so set the magic i_delalloc_reserve_flag after taking the
362 inode allocation semaphore for */
363 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
364 /* Call ext4_da_update_reserve_space() after successfully
365 allocating the blocks */
366 #define EXT4_GET_BLOCKS_UPDATE_RESERVE_SPACE 0x0008
367 /* caller is from the direct IO path, request to creation of an
368 unitialized extents if not allocated, split the uninitialized
369 extent if blocks has been preallocated already*/
370 #define EXT4_GET_BLOCKS_DIO 0x0010
371 #define EXT4_GET_BLOCKS_CONVERT 0x0020
372 #define EXT4_GET_BLOCKS_DIO_CREATE_EXT (EXT4_GET_BLOCKS_DIO|\
373 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
374 /* Convert extent to initialized after direct IO complete */
375 #define EXT4_GET_BLOCKS_DIO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
376 EXT4_GET_BLOCKS_DIO_CREATE_EXT)
381 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
382 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
383 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
384 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
385 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
386 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
387 #ifdef CONFIG_JBD2_DEBUG
388 #define EXT4_IOC_WAIT_FOR_READONLY _IOR('f', 99, long)
390 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
391 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
392 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
393 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
394 #define EXT4_IOC_MIGRATE _IO('f', 9)
395 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
396 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
397 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
398 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
401 * ioctl commands in 32 bit emulation
403 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
404 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
405 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
406 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
407 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
408 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
409 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
410 #ifdef CONFIG_JBD2_DEBUG
411 #define EXT4_IOC32_WAIT_FOR_READONLY _IOR('f', 99, int)
413 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
414 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
420 struct ext4_mount_options
{
421 unsigned long s_mount_opt
;
424 unsigned long s_commit_interval
;
425 u32 s_min_batch_time
, s_max_batch_time
;
428 char *s_qf_names
[MAXQUOTAS
];
432 /* Max physical block we can addres w/o extents */
433 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
436 * Structure of an inode on the disk
439 __le16 i_mode
; /* File mode */
440 __le16 i_uid
; /* Low 16 bits of Owner Uid */
441 __le32 i_size_lo
; /* Size in bytes */
442 __le32 i_atime
; /* Access time */
443 __le32 i_ctime
; /* Inode Change time */
444 __le32 i_mtime
; /* Modification time */
445 __le32 i_dtime
; /* Deletion Time */
446 __le16 i_gid
; /* Low 16 bits of Group Id */
447 __le16 i_links_count
; /* Links count */
448 __le32 i_blocks_lo
; /* Blocks count */
449 __le32 i_flags
; /* File flags */
455 __u32 h_i_translator
;
460 } osd1
; /* OS dependent 1 */
461 __le32 i_block
[EXT4_N_BLOCKS
];/* Pointers to blocks */
462 __le32 i_generation
; /* File version (for NFS) */
463 __le32 i_file_acl_lo
; /* File ACL */
465 __le32 i_obso_faddr
; /* Obsoleted fragment address */
468 __le16 l_i_blocks_high
; /* were l_i_reserved1 */
469 __le16 l_i_file_acl_high
;
470 __le16 l_i_uid_high
; /* these 2 fields */
471 __le16 l_i_gid_high
; /* were reserved2[0] */
475 __le16 h_i_reserved1
; /* Obsoleted fragment number/size which are removed in ext4 */
482 __le16 h_i_reserved1
; /* Obsoleted fragment number/size which are removed in ext4 */
483 __le16 m_i_file_acl_high
;
484 __u32 m_i_reserved2
[2];
486 } osd2
; /* OS dependent 2 */
487 __le16 i_extra_isize
;
489 __le32 i_ctime_extra
; /* extra Change time (nsec << 2 | epoch) */
490 __le32 i_mtime_extra
; /* extra Modification time(nsec << 2 | epoch) */
491 __le32 i_atime_extra
; /* extra Access time (nsec << 2 | epoch) */
492 __le32 i_crtime
; /* File Creation time */
493 __le32 i_crtime_extra
; /* extra FileCreationtime (nsec << 2 | epoch) */
494 __le32 i_version_hi
; /* high 32 bits for 64-bit version */
498 __u32 reserved
; /* should be zero */
499 __u32 donor_fd
; /* donor file descriptor */
500 __u64 orig_start
; /* logical start offset in block for orig */
501 __u64 donor_start
; /* logical start offset in block for donor */
502 __u64 len
; /* block length to be moved */
503 __u64 moved_len
; /* moved block length */
506 #define EXT4_EPOCH_BITS 2
507 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
508 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
511 * Extended fields will fit into an inode if the filesystem was formatted
512 * with large inodes (-I 256 or larger) and there are not currently any EAs
513 * consuming all of the available space. For new inodes we always reserve
514 * enough space for the kernel's known extended fields, but for inodes
515 * created with an old kernel this might not have been the case. None of
516 * the extended inode fields is critical for correct filesystem operation.
517 * This macro checks if a certain field fits in the inode. Note that
518 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
520 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
521 ((offsetof(typeof(*ext4_inode), field) + \
522 sizeof((ext4_inode)->field)) \
523 <= (EXT4_GOOD_OLD_INODE_SIZE + \
524 (einode)->i_extra_isize)) \
526 static inline __le32 ext4_encode_extra_time(struct timespec *time)
528 return cpu_to_le32((sizeof(time
->tv_sec
) > 4 ?
529 (time
->tv_sec
>> 32) & EXT4_EPOCH_MASK
: 0) |
530 ((time
->tv_nsec
<< EXT4_EPOCH_BITS
) & EXT4_NSEC_MASK
));
533 static inline void ext4_decode_extra_time(struct timespec
*time
, __le32 extra
)
535 if (sizeof(time
->tv_sec
) > 4)
536 time
->tv_sec
|= (__u64
)(le32_to_cpu(extra
) & EXT4_EPOCH_MASK
)
538 time
->tv_nsec
= (le32_to_cpu(extra
) & EXT4_NSEC_MASK
) >> EXT4_EPOCH_BITS
;
541 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
543 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
544 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
545 (raw_inode)->xtime ## _extra = \
546 ext4_encode_extra_time(&(inode)->xtime); \
549 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
551 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
552 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
553 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
554 (raw_inode)->xtime ## _extra = \
555 ext4_encode_extra_time(&(einode)->xtime); \
558 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
560 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
561 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
562 ext4_decode_extra_time(&(inode)->xtime, \
563 raw_inode->xtime ## _extra); \
566 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
568 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
569 (einode)->xtime.tv_sec = \
570 (signed)le32_to_cpu((raw_inode)->xtime); \
571 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
572 ext4_decode_extra_time(&(einode)->xtime, \
573 raw_inode->xtime ## _extra); \
576 #define i_disk_version osd1.linux1.l_i_version
578 #if defined(__KERNEL__) || defined(__linux__)
579 #define i_reserved1 osd1.linux1.l_i_reserved1
580 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
581 #define i_blocks_high osd2.linux2.l_i_blocks_high
582 #define i_uid_low i_uid
583 #define i_gid_low i_gid
584 #define i_uid_high osd2.linux2.l_i_uid_high
585 #define i_gid_high osd2.linux2.l_i_gid_high
586 #define i_reserved2 osd2.linux2.l_i_reserved2
588 #elif defined(__GNU__)
590 #define i_translator osd1.hurd1.h_i_translator
591 #define i_uid_high osd2.hurd2.h_i_uid_high
592 #define i_gid_high osd2.hurd2.h_i_gid_high
593 #define i_author osd2.hurd2.h_i_author
595 #elif defined(__masix__)
597 #define i_reserved1 osd1.masix1.m_i_reserved1
598 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
599 #define i_reserved2 osd2.masix2.m_i_reserved2
601 #endif /* defined(__KERNEL__) || defined(__linux__) */
604 * storage for cached extent
606 struct ext4_ext_cache
{
607 ext4_fsblk_t ec_start
;
608 ext4_lblk_t ec_block
;
609 __u32 ec_len
; /* must be 32bit to return holes */
614 * fourth extended file system inode data in memory
616 struct ext4_inode_info
{
617 __le32 i_data
[15]; /* unconverted */
619 ext4_fsblk_t i_file_acl
;
623 * i_block_group is the number of the block group which contains
624 * this file's inode. Constant across the lifetime of the inode,
625 * it is ued for making block allocation decisions - we try to
626 * place a file's data blocks near its inode block, and new inodes
627 * near to their parent directory's inode.
629 ext4_group_t i_block_group
;
630 __u32 i_state
; /* Dynamic state flags for ext4 */
632 ext4_lblk_t i_dir_start_lookup
;
633 #ifdef CONFIG_EXT4_FS_XATTR
635 * Extended attributes can be read independently of the main file
636 * data. Taking i_mutex even when reading would cause contention
637 * between readers of EAs and writers of regular file data, so
638 * instead we synchronize on xattr_sem when reading or changing
641 struct rw_semaphore xattr_sem
;
644 struct list_head i_orphan
; /* unlinked but open inodes */
647 * i_disksize keeps track of what the inode size is ON DISK, not
648 * in memory. During truncate, i_size is set to the new size by
649 * the VFS prior to calling ext4_truncate(), but the filesystem won't
650 * set i_disksize to 0 until the truncate is actually under way.
652 * The intent is that i_disksize always represents the blocks which
653 * are used by this file. This allows recovery to restart truncate
654 * on orphans if we crash during truncate. We actually write i_disksize
655 * into the on-disk inode when writing inodes out, instead of i_size.
657 * The only time when i_disksize and i_size may be different is when
658 * a truncate is in progress. The only things which change i_disksize
659 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
664 * i_data_sem is for serialising ext4_truncate() against
665 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
666 * data tree are chopped off during truncate. We can't do that in
667 * ext4 because whenever we perform intermediate commits during
668 * truncate, the inode and all the metadata blocks *must* be in a
669 * consistent state which allows truncation of the orphans to restart
670 * during recovery. Hence we must fix the get_block-vs-truncate race
671 * by other means, so we have i_data_sem.
673 struct rw_semaphore i_data_sem
;
674 struct inode vfs_inode
;
675 struct jbd2_inode jinode
;
677 struct ext4_ext_cache i_cached_extent
;
679 * File creation time. Its function is same as that of
680 * struct timespec i_{a,c,m}time in the generic inode.
682 struct timespec i_crtime
;
685 struct list_head i_prealloc_list
;
686 spinlock_t i_prealloc_lock
;
689 ext4_group_t i_last_alloc_group
;
691 /* allocation reservation info for delalloc */
692 unsigned int i_reserved_data_blocks
;
693 unsigned int i_reserved_meta_blocks
;
694 unsigned int i_allocated_meta_blocks
;
695 unsigned short i_delalloc_reserved_flag
;
697 /* on-disk additional length */
700 spinlock_t i_block_reservation_lock
;
702 /* quota space reservation, managed internally by quota code */
703 qsize_t i_reserved_quota
;
706 /* completed async DIOs that might need unwritten extents handling */
707 struct list_head i_aio_dio_complete_list
;
708 /* current io_end structure for async DIO write*/
709 ext4_io_end_t
*cur_aio_dio
;
712 * Transactions that contain inode's metadata needed to complete
713 * fsync and fdatasync, respectively.
716 tid_t i_datasync_tid
;
722 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
723 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
724 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
727 * Misc. filesystem flags
729 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
730 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
731 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
736 #define EXT4_MOUNT_OLDALLOC 0x00002 /* Don't use the new Orlov allocator */
737 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
738 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
739 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
740 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
741 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
742 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
743 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
744 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
745 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
746 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
747 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
748 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
749 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
750 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
751 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
752 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
753 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
754 #define EXT4_MOUNT_NOBH 0x40000 /* No bufferheads */
755 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
756 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
757 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
758 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
759 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
760 #define EXT4_MOUNT_I_VERSION 0x2000000 /* i_version support */
761 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
762 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
763 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
764 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
766 #define clear_opt(o, opt) o &= ~EXT4_MOUNT_##opt
767 #define set_opt(o, opt) o |= EXT4_MOUNT_##opt
768 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
771 #define ext4_set_bit ext2_set_bit
772 #define ext4_set_bit_atomic ext2_set_bit_atomic
773 #define ext4_clear_bit ext2_clear_bit
774 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
775 #define ext4_test_bit ext2_test_bit
776 #define ext4_find_first_zero_bit ext2_find_first_zero_bit
777 #define ext4_find_next_zero_bit ext2_find_next_zero_bit
778 #define ext4_find_next_bit ext2_find_next_bit
781 * Maximal mount counts between two filesystem checks
783 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
784 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
787 * Behaviour when detecting errors
789 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
790 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
791 #define EXT4_ERRORS_PANIC 3 /* Panic */
792 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
795 * Structure of the super block
797 struct ext4_super_block
{
798 /*00*/ __le32 s_inodes_count
; /* Inodes count */
799 __le32 s_blocks_count_lo
; /* Blocks count */
800 __le32 s_r_blocks_count_lo
; /* Reserved blocks count */
801 __le32 s_free_blocks_count_lo
; /* Free blocks count */
802 /*10*/ __le32 s_free_inodes_count
; /* Free inodes count */
803 __le32 s_first_data_block
; /* First Data Block */
804 __le32 s_log_block_size
; /* Block size */
805 __le32 s_obso_log_frag_size
; /* Obsoleted fragment size */
806 /*20*/ __le32 s_blocks_per_group
; /* # Blocks per group */
807 __le32 s_obso_frags_per_group
; /* Obsoleted fragments per group */
808 __le32 s_inodes_per_group
; /* # Inodes per group */
809 __le32 s_mtime
; /* Mount time */
810 /*30*/ __le32 s_wtime
; /* Write time */
811 __le16 s_mnt_count
; /* Mount count */
812 __le16 s_max_mnt_count
; /* Maximal mount count */
813 __le16 s_magic
; /* Magic signature */
814 __le16 s_state
; /* File system state */
815 __le16 s_errors
; /* Behaviour when detecting errors */
816 __le16 s_minor_rev_level
; /* minor revision level */
817 /*40*/ __le32 s_lastcheck
; /* time of last check */
818 __le32 s_checkinterval
; /* max. time between checks */
819 __le32 s_creator_os
; /* OS */
820 __le32 s_rev_level
; /* Revision level */
821 /*50*/ __le16 s_def_resuid
; /* Default uid for reserved blocks */
822 __le16 s_def_resgid
; /* Default gid for reserved blocks */
824 * These fields are for EXT4_DYNAMIC_REV superblocks only.
826 * Note: the difference between the compatible feature set and
827 * the incompatible feature set is that if there is a bit set
828 * in the incompatible feature set that the kernel doesn't
829 * know about, it should refuse to mount the filesystem.
831 * e2fsck's requirements are more strict; if it doesn't know
832 * about a feature in either the compatible or incompatible
833 * feature set, it must abort and not try to meddle with
834 * things it doesn't understand...
836 __le32 s_first_ino
; /* First non-reserved inode */
837 __le16 s_inode_size
; /* size of inode structure */
838 __le16 s_block_group_nr
; /* block group # of this superblock */
839 __le32 s_feature_compat
; /* compatible feature set */
840 /*60*/ __le32 s_feature_incompat
; /* incompatible feature set */
841 __le32 s_feature_ro_compat
; /* readonly-compatible feature set */
842 /*68*/ __u8 s_uuid
[16]; /* 128-bit uuid for volume */
843 /*78*/ char s_volume_name
[16]; /* volume name */
844 /*88*/ char s_last_mounted
[64]; /* directory where last mounted */
845 /*C8*/ __le32 s_algorithm_usage_bitmap
; /* For compression */
847 * Performance hints. Directory preallocation should only
848 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
850 __u8 s_prealloc_blocks
; /* Nr of blocks to try to preallocate*/
851 __u8 s_prealloc_dir_blocks
; /* Nr to preallocate for dirs */
852 __le16 s_reserved_gdt_blocks
; /* Per group desc for online growth */
854 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
856 /*D0*/ __u8 s_journal_uuid
[16]; /* uuid of journal superblock */
857 /*E0*/ __le32 s_journal_inum
; /* inode number of journal file */
858 __le32 s_journal_dev
; /* device number of journal file */
859 __le32 s_last_orphan
; /* start of list of inodes to delete */
860 __le32 s_hash_seed
[4]; /* HTREE hash seed */
861 __u8 s_def_hash_version
; /* Default hash version to use */
862 __u8 s_reserved_char_pad
;
863 __le16 s_desc_size
; /* size of group descriptor */
864 /*100*/ __le32 s_default_mount_opts
;
865 __le32 s_first_meta_bg
; /* First metablock block group */
866 __le32 s_mkfs_time
; /* When the filesystem was created */
867 __le32 s_jnl_blocks
[17]; /* Backup of the journal inode */
868 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
869 /*150*/ __le32 s_blocks_count_hi
; /* Blocks count */
870 __le32 s_r_blocks_count_hi
; /* Reserved blocks count */
871 __le32 s_free_blocks_count_hi
; /* Free blocks count */
872 __le16 s_min_extra_isize
; /* All inodes have at least # bytes */
873 __le16 s_want_extra_isize
; /* New inodes should reserve # bytes */
874 __le32 s_flags
; /* Miscellaneous flags */
875 __le16 s_raid_stride
; /* RAID stride */
876 __le16 s_mmp_interval
; /* # seconds to wait in MMP checking */
877 __le64 s_mmp_block
; /* Block for multi-mount protection */
878 __le32 s_raid_stripe_width
; /* blocks on all data disks (N*stride)*/
879 __u8 s_log_groups_per_flex
; /* FLEX_BG group size */
880 __u8 s_reserved_char_pad2
;
881 __le16 s_reserved_pad
;
882 __le64 s_kbytes_written
; /* nr of lifetime kilobytes written */
883 __u32 s_reserved
[160]; /* Padding to the end of the block */
889 * run-time mount flags
891 #define EXT4_MF_MNTDIR_SAMPLED 0x0001
892 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
895 * fourth extended-fs super-block data in memory
897 struct ext4_sb_info
{
898 unsigned long s_desc_size
; /* Size of a group descriptor in bytes */
899 unsigned long s_inodes_per_block
;/* Number of inodes per block */
900 unsigned long s_blocks_per_group
;/* Number of blocks in a group */
901 unsigned long s_inodes_per_group
;/* Number of inodes in a group */
902 unsigned long s_itb_per_group
; /* Number of inode table blocks per group */
903 unsigned long s_gdb_count
; /* Number of group descriptor blocks */
904 unsigned long s_desc_per_block
; /* Number of group descriptors per block */
905 ext4_group_t s_groups_count
; /* Number of groups in the fs */
906 ext4_group_t s_blockfile_groups
;/* Groups acceptable for non-extent files */
907 unsigned long s_overhead_last
; /* Last calculated overhead */
908 unsigned long s_blocks_last
; /* Last seen block count */
909 loff_t s_bitmap_maxbytes
; /* max bytes for bitmap files */
910 struct buffer_head
* s_sbh
; /* Buffer containing the super block */
911 struct ext4_super_block
*s_es
; /* Pointer to the super block in the buffer */
912 struct buffer_head
**s_group_desc
;
913 unsigned int s_mount_opt
;
914 unsigned int s_mount_flags
;
915 ext4_fsblk_t s_sb_block
;
918 unsigned short s_mount_state
;
919 unsigned short s_pad
;
920 int s_addr_per_block_bits
;
921 int s_desc_per_block_bits
;
924 unsigned int s_inode_readahead_blks
;
925 unsigned int s_inode_goal
;
926 spinlock_t s_next_gen_lock
;
927 u32 s_next_generation
;
929 int s_def_hash_version
;
930 int s_hash_unsigned
; /* 3 if hash should be signed, 0 if not */
931 struct percpu_counter s_freeblocks_counter
;
932 struct percpu_counter s_freeinodes_counter
;
933 struct percpu_counter s_dirs_counter
;
934 struct percpu_counter s_dirtyblocks_counter
;
935 struct blockgroup_lock
*s_blockgroup_lock
;
936 struct proc_dir_entry
*s_proc
;
937 struct kobject s_kobj
;
938 struct completion s_kobj_unregister
;
941 struct inode
*s_journal_inode
;
942 struct journal_s
*s_journal
;
943 struct list_head s_orphan
;
944 struct mutex s_orphan_lock
;
945 struct mutex s_resize_lock
;
946 unsigned long s_commit_interval
;
947 u32 s_max_batch_time
;
948 u32 s_min_batch_time
;
949 struct block_device
*journal_bdev
;
950 #ifdef CONFIG_JBD2_DEBUG
951 struct timer_list turn_ro_timer
; /* For turning read-only (crash simulation) */
952 wait_queue_head_t ro_wait_queue
; /* For people waiting for the fs to go read-only */
955 char *s_qf_names
[MAXQUOTAS
]; /* Names of quota files with journalled quota */
956 int s_jquota_fmt
; /* Format of quota to use */
958 unsigned int s_want_extra_isize
; /* New inodes should reserve # bytes */
959 struct rb_root system_blks
;
962 /* ext4 extents stats */
963 unsigned long s_ext_min
;
964 unsigned long s_ext_max
;
965 unsigned long s_depth_max
;
966 spinlock_t s_ext_stats_lock
;
967 unsigned long s_ext_blocks
;
968 unsigned long s_ext_extents
;
971 /* for buddy allocator */
972 struct ext4_group_info
***s_group_info
;
973 struct inode
*s_buddy_cache
;
974 long s_blocks_reserved
;
975 spinlock_t s_reserve_lock
;
976 spinlock_t s_md_lock
;
977 tid_t s_last_transaction
;
978 unsigned short *s_mb_offsets
;
979 unsigned int *s_mb_maxs
;
982 unsigned long s_stripe
;
983 unsigned int s_mb_stream_request
;
984 unsigned int s_mb_max_to_scan
;
985 unsigned int s_mb_min_to_scan
;
986 unsigned int s_mb_stats
;
987 unsigned int s_mb_order2_reqs
;
988 unsigned int s_mb_group_prealloc
;
989 unsigned int s_max_writeback_mb_bump
;
990 /* where last allocation was done - for stream allocation */
991 unsigned long s_mb_last_group
;
992 unsigned long s_mb_last_start
;
994 /* stats for buddy allocator */
995 spinlock_t s_mb_pa_lock
;
996 atomic_t s_bal_reqs
; /* number of reqs with len > 1 */
997 atomic_t s_bal_success
; /* we found long enough chunks */
998 atomic_t s_bal_allocated
; /* in blocks */
999 atomic_t s_bal_ex_scanned
; /* total extents scanned */
1000 atomic_t s_bal_goals
; /* goal hits */
1001 atomic_t s_bal_breaks
; /* too long searches */
1002 atomic_t s_bal_2orders
; /* 2^order hits */
1003 spinlock_t s_bal_lock
;
1004 unsigned long s_mb_buddies_generated
;
1005 unsigned long long s_mb_generation_time
;
1006 atomic_t s_mb_lost_chunks
;
1007 atomic_t s_mb_preallocated
;
1008 atomic_t s_mb_discarded
;
1009 atomic_t s_lock_busy
;
1011 /* locality groups */
1012 struct ext4_locality_group
*s_locality_groups
;
1014 /* for write statistics */
1015 unsigned long s_sectors_written_start
;
1016 u64 s_kbytes_written
;
1018 unsigned int s_log_groups_per_flex
;
1019 struct flex_groups
*s_flex_groups
;
1021 /* workqueue for dio unwritten */
1022 struct workqueue_struct
*dio_unwritten_wq
;
1025 static inline struct ext4_sb_info
*EXT4_SB(struct super_block
*sb
)
1027 return sb
->s_fs_info
;
1029 static inline struct ext4_inode_info
*EXT4_I(struct inode
*inode
)
1031 return container_of(inode
, struct ext4_inode_info
, vfs_inode
);
1034 static inline struct timespec
ext4_current_time(struct inode
*inode
)
1036 return (inode
->i_sb
->s_time_gran
< NSEC_PER_SEC
) ?
1037 current_fs_time(inode
->i_sb
) : CURRENT_TIME_SEC
;
1040 static inline int ext4_valid_inum(struct super_block
*sb
, unsigned long ino
)
1042 return ino
== EXT4_ROOT_INO
||
1043 ino
== EXT4_JOURNAL_INO
||
1044 ino
== EXT4_RESIZE_INO
||
1045 (ino
>= EXT4_FIRST_INO(sb
) &&
1046 ino
<= le32_to_cpu(EXT4_SB(sb
)->s_es
->s_inodes_count
));
1049 /* Assume that user mode programs are passing in an ext4fs superblock, not
1050 * a kernel struct super_block. This will allow us to call the feature-test
1051 * macros from user land. */
1052 #define EXT4_SB(sb) (sb)
1055 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1058 * Codes for operating systems
1060 #define EXT4_OS_LINUX 0
1061 #define EXT4_OS_HURD 1
1062 #define EXT4_OS_MASIX 2
1063 #define EXT4_OS_FREEBSD 3
1064 #define EXT4_OS_LITES 4
1069 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1070 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1072 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1073 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1075 #define EXT4_GOOD_OLD_INODE_SIZE 128
1078 * Feature set definitions
1081 #define EXT4_HAS_COMPAT_FEATURE(sb,mask) \
1082 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1083 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \
1084 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1085 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \
1086 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1087 #define EXT4_SET_COMPAT_FEATURE(sb,mask) \
1088 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1089 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \
1090 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1091 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \
1092 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1093 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \
1094 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1095 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
1096 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1097 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \
1098 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1100 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1101 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1102 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1103 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1104 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1105 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1107 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1108 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1109 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1110 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1111 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1112 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1113 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1115 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1116 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1117 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1118 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1119 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1120 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1121 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1122 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1123 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1125 #define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR
1126 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1127 EXT4_FEATURE_INCOMPAT_RECOVER| \
1128 EXT4_FEATURE_INCOMPAT_META_BG| \
1129 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1130 EXT4_FEATURE_INCOMPAT_64BIT| \
1131 EXT4_FEATURE_INCOMPAT_FLEX_BG)
1132 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1133 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1134 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1135 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1136 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1137 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1138 EXT4_FEATURE_RO_COMPAT_HUGE_FILE)
1141 * Default values for user and/or group using reserved blocks
1143 #define EXT4_DEF_RESUID 0
1144 #define EXT4_DEF_RESGID 0
1146 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
1149 * Default mount options
1151 #define EXT4_DEFM_DEBUG 0x0001
1152 #define EXT4_DEFM_BSDGROUPS 0x0002
1153 #define EXT4_DEFM_XATTR_USER 0x0004
1154 #define EXT4_DEFM_ACL 0x0008
1155 #define EXT4_DEFM_UID16 0x0010
1156 #define EXT4_DEFM_JMODE 0x0060
1157 #define EXT4_DEFM_JMODE_DATA 0x0020
1158 #define EXT4_DEFM_JMODE_ORDERED 0x0040
1159 #define EXT4_DEFM_JMODE_WBACK 0x0060
1162 * Default journal batch times
1164 #define EXT4_DEF_MIN_BATCH_TIME 0
1165 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1168 * Minimum number of groups in a flexgroup before we separate out
1169 * directories into the first block group of a flexgroup
1171 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1174 * Structure of a directory entry
1176 #define EXT4_NAME_LEN 255
1178 struct ext4_dir_entry
{
1179 __le32 inode
; /* Inode number */
1180 __le16 rec_len
; /* Directory entry length */
1181 __le16 name_len
; /* Name length */
1182 char name
[EXT4_NAME_LEN
]; /* File name */
1186 * The new version of the directory entry. Since EXT4 structures are
1187 * stored in intel byte order, and the name_len field could never be
1188 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1191 struct ext4_dir_entry_2
{
1192 __le32 inode
; /* Inode number */
1193 __le16 rec_len
; /* Directory entry length */
1194 __u8 name_len
; /* Name length */
1196 char name
[EXT4_NAME_LEN
]; /* File name */
1200 * Ext4 directory file types. Only the low 3 bits are used. The
1201 * other bits are reserved for now.
1203 #define EXT4_FT_UNKNOWN 0
1204 #define EXT4_FT_REG_FILE 1
1205 #define EXT4_FT_DIR 2
1206 #define EXT4_FT_CHRDEV 3
1207 #define EXT4_FT_BLKDEV 4
1208 #define EXT4_FT_FIFO 5
1209 #define EXT4_FT_SOCK 6
1210 #define EXT4_FT_SYMLINK 7
1212 #define EXT4_FT_MAX 8
1215 * EXT4_DIR_PAD defines the directory entries boundaries
1217 * NOTE: It must be a multiple of 4
1219 #define EXT4_DIR_PAD 4
1220 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1221 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1223 #define EXT4_MAX_REC_LEN ((1<<16)-1)
1226 * Hash Tree Directory indexing
1227 * (c) Daniel Phillips, 2001
1230 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1231 EXT4_FEATURE_COMPAT_DIR_INDEX) && \
1232 (EXT4_I(dir)->i_flags & EXT4_INDEX_FL))
1233 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1234 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1236 /* Legal values for the dx_root hash_version field: */
1238 #define DX_HASH_LEGACY 0
1239 #define DX_HASH_HALF_MD4 1
1240 #define DX_HASH_TEA 2
1241 #define DX_HASH_LEGACY_UNSIGNED 3
1242 #define DX_HASH_HALF_MD4_UNSIGNED 4
1243 #define DX_HASH_TEA_UNSIGNED 5
1247 /* hash info structure used by the directory hash */
1256 #define EXT4_HTREE_EOF 0x7fffffff
1259 * Control parameters used by ext4_htree_next_block
1261 #define HASH_NB_ALWAYS 1
1265 * Describe an inode's exact location on disk and in memory
1269 struct buffer_head
*bh
;
1270 unsigned long offset
;
1271 ext4_group_t block_group
;
1274 static inline struct ext4_inode
*ext4_raw_inode(struct ext4_iloc
*iloc
)
1276 return (struct ext4_inode
*) (iloc
->bh
->b_data
+ iloc
->offset
);
1280 * This structure is stuffed into the struct file's private_data field
1281 * for directories. It is where we put information so that we can do
1282 * readdir operations in hash tree order.
1284 struct dir_private_info
{
1285 struct rb_root root
;
1286 struct rb_node
*curr_node
;
1287 struct fname
*extra_fname
;
1290 __u32 curr_minor_hash
;
1294 /* calculate the first block number of the group */
1295 static inline ext4_fsblk_t
1296 ext4_group_first_block_no(struct super_block
*sb
, ext4_group_t group_no
)
1298 return group_no
* (ext4_fsblk_t
)EXT4_BLOCKS_PER_GROUP(sb
) +
1299 le32_to_cpu(EXT4_SB(sb
)->s_es
->s_first_data_block
);
1303 * Special error return code only used by dx_probe() and its callers.
1305 #define ERR_BAD_DX_DIR -75000
1307 void ext4_get_group_no_and_offset(struct super_block
*sb
, ext4_fsblk_t blocknr
,
1308 ext4_group_t
*blockgrpp
, ext4_grpblk_t
*offsetp
);
1310 extern struct proc_dir_entry
*ext4_proc_root
;
1313 * Function prototypes
1317 * Ok, these declarations are also in <linux/kernel.h> but none of the
1318 * ext4 source programs needs to include it so they are duplicated here.
1320 # define NORET_TYPE /**/
1321 # define ATTRIB_NORET __attribute__((noreturn))
1322 # define NORET_AND noreturn,
1325 extern unsigned int ext4_count_free(struct buffer_head
*, unsigned);
1328 extern unsigned int ext4_block_group(struct super_block
*sb
,
1329 ext4_fsblk_t blocknr
);
1330 extern ext4_grpblk_t
ext4_block_group_offset(struct super_block
*sb
,
1331 ext4_fsblk_t blocknr
);
1332 extern int ext4_bg_has_super(struct super_block
*sb
, ext4_group_t group
);
1333 extern unsigned long ext4_bg_num_gdb(struct super_block
*sb
,
1334 ext4_group_t group
);
1335 extern ext4_fsblk_t
ext4_new_meta_blocks(handle_t
*handle
, struct inode
*inode
,
1336 ext4_fsblk_t goal
, unsigned long *count
, int *errp
);
1337 extern int ext4_claim_free_blocks(struct ext4_sb_info
*sbi
, s64 nblocks
);
1338 extern int ext4_has_free_blocks(struct ext4_sb_info
*sbi
, s64 nblocks
);
1339 extern void ext4_free_blocks(handle_t
*handle
, struct inode
*inode
,
1340 ext4_fsblk_t block
, unsigned long count
, int metadata
);
1341 extern void ext4_add_groupblocks(handle_t
*handle
, struct super_block
*sb
,
1342 ext4_fsblk_t block
, unsigned long count
);
1343 extern ext4_fsblk_t
ext4_count_free_blocks(struct super_block
*);
1344 extern void ext4_check_blocks_bitmap(struct super_block
*);
1345 extern struct ext4_group_desc
* ext4_get_group_desc(struct super_block
* sb
,
1346 ext4_group_t block_group
,
1347 struct buffer_head
** bh
);
1348 extern int ext4_should_retry_alloc(struct super_block
*sb
, int *retries
);
1349 struct buffer_head
*ext4_read_block_bitmap(struct super_block
*sb
,
1350 ext4_group_t block_group
);
1351 extern unsigned ext4_init_block_bitmap(struct super_block
*sb
,
1352 struct buffer_head
*bh
,
1354 struct ext4_group_desc
*desc
);
1355 #define ext4_free_blocks_after_init(sb, group, desc) \
1356 ext4_init_block_bitmap(sb, NULL, group, desc)
1359 extern int ext4_check_dir_entry(const char *, struct inode
*,
1360 struct ext4_dir_entry_2
*,
1361 struct buffer_head
*, unsigned int);
1362 extern int ext4_htree_store_dirent(struct file
*dir_file
, __u32 hash
,
1364 struct ext4_dir_entry_2
*dirent
);
1365 extern void ext4_htree_free_dir_info(struct dir_private_info
*p
);
1368 extern int ext4_sync_file(struct file
*, struct dentry
*, int);
1371 extern int ext4fs_dirhash(const char *name
, int len
, struct
1372 dx_hash_info
*hinfo
);
1375 extern struct inode
*ext4_new_inode(handle_t
*, struct inode
*, int,
1376 const struct qstr
*qstr
, __u32 goal
);
1377 extern void ext4_free_inode(handle_t
*, struct inode
*);
1378 extern struct inode
* ext4_orphan_get(struct super_block
*, unsigned long);
1379 extern unsigned long ext4_count_free_inodes(struct super_block
*);
1380 extern unsigned long ext4_count_dirs(struct super_block
*);
1381 extern void ext4_check_inodes_bitmap(struct super_block
*);
1382 extern unsigned ext4_init_inode_bitmap(struct super_block
*sb
,
1383 struct buffer_head
*bh
,
1385 struct ext4_group_desc
*desc
);
1386 extern void mark_bitmap_end(int start_bit
, int end_bit
, char *bitmap
);
1389 extern long ext4_mb_stats
;
1390 extern long ext4_mb_max_to_scan
;
1391 extern int ext4_mb_init(struct super_block
*, int);
1392 extern int ext4_mb_release(struct super_block
*);
1393 extern ext4_fsblk_t
ext4_mb_new_blocks(handle_t
*,
1394 struct ext4_allocation_request
*, int *);
1395 extern int ext4_mb_reserve_blocks(struct super_block
*, int);
1396 extern void ext4_discard_preallocations(struct inode
*);
1397 extern int __init
init_ext4_mballoc(void);
1398 extern void exit_ext4_mballoc(void);
1399 extern void ext4_mb_free_blocks(handle_t
*, struct inode
*,
1400 ext4_fsblk_t
, unsigned long, int, unsigned long *);
1401 extern int ext4_mb_add_groupinfo(struct super_block
*sb
,
1402 ext4_group_t i
, struct ext4_group_desc
*desc
);
1403 extern int ext4_mb_get_buddy_cache_lock(struct super_block
*, ext4_group_t
);
1404 extern void ext4_mb_put_buddy_cache_lock(struct super_block
*,
1407 int ext4_forget(handle_t
*handle
, int is_metadata
, struct inode
*inode
,
1408 struct buffer_head
*bh
, ext4_fsblk_t blocknr
);
1409 struct buffer_head
*ext4_getblk(handle_t
*, struct inode
*,
1410 ext4_lblk_t
, int, int *);
1411 struct buffer_head
*ext4_bread(handle_t
*, struct inode
*,
1412 ext4_lblk_t
, int, int *);
1413 int ext4_get_block(struct inode
*inode
, sector_t iblock
,
1414 struct buffer_head
*bh_result
, int create
);
1416 extern struct inode
*ext4_iget(struct super_block
*, unsigned long);
1417 extern int ext4_write_inode(struct inode
*, int);
1418 extern int ext4_setattr(struct dentry
*, struct iattr
*);
1419 extern int ext4_getattr(struct vfsmount
*mnt
, struct dentry
*dentry
,
1420 struct kstat
*stat
);
1421 extern void ext4_delete_inode(struct inode
*);
1422 extern int ext4_sync_inode(handle_t
*, struct inode
*);
1423 extern void ext4_dirty_inode(struct inode
*);
1424 extern int ext4_change_inode_journal_flag(struct inode
*, int);
1425 extern int ext4_get_inode_loc(struct inode
*, struct ext4_iloc
*);
1426 extern int ext4_can_truncate(struct inode
*inode
);
1427 extern void ext4_truncate(struct inode
*);
1428 extern int ext4_truncate_restart_trans(handle_t
*, struct inode
*, int nblocks
);
1429 extern void ext4_set_inode_flags(struct inode
*);
1430 extern void ext4_get_inode_flags(struct ext4_inode_info
*);
1431 extern int ext4_alloc_da_blocks(struct inode
*inode
);
1432 extern void ext4_set_aops(struct inode
*inode
);
1433 extern int ext4_writepage_trans_blocks(struct inode
*);
1434 extern int ext4_meta_trans_blocks(struct inode
*, int nrblocks
, int idxblocks
);
1435 extern int ext4_chunk_trans_blocks(struct inode
*, int nrblocks
);
1436 extern int ext4_block_truncate_page(handle_t
*handle
,
1437 struct address_space
*mapping
, loff_t from
);
1438 extern int ext4_page_mkwrite(struct vm_area_struct
*vma
, struct vm_fault
*vmf
);
1439 extern qsize_t
*ext4_get_reserved_space(struct inode
*inode
);
1440 extern int flush_aio_dio_completed_IO(struct inode
*inode
);
1442 extern long ext4_ioctl(struct file
*, unsigned int, unsigned long);
1443 extern long ext4_compat_ioctl(struct file
*, unsigned int, unsigned long);
1446 extern int ext4_ext_migrate(struct inode
*);
1449 extern unsigned int ext4_rec_len_from_disk(__le16 dlen
, unsigned blocksize
);
1450 extern __le16
ext4_rec_len_to_disk(unsigned len
, unsigned blocksize
);
1451 extern int ext4_orphan_add(handle_t
*, struct inode
*);
1452 extern int ext4_orphan_del(handle_t
*, struct inode
*);
1453 extern int ext4_htree_fill_tree(struct file
*dir_file
, __u32 start_hash
,
1454 __u32 start_minor_hash
, __u32
*next_hash
);
1457 extern int ext4_group_add(struct super_block
*sb
,
1458 struct ext4_new_group_data
*input
);
1459 extern int ext4_group_extend(struct super_block
*sb
,
1460 struct ext4_super_block
*es
,
1461 ext4_fsblk_t n_blocks_count
);
1464 extern void ext4_error(struct super_block
*, const char *, const char *, ...)
1465 __attribute__ ((format (printf
, 3, 4)));
1466 extern void __ext4_std_error(struct super_block
*, const char *, int);
1467 extern void ext4_abort(struct super_block
*, const char *, const char *, ...)
1468 __attribute__ ((format (printf
, 3, 4)));
1469 extern void ext4_warning(struct super_block
*, const char *, const char *, ...)
1470 __attribute__ ((format (printf
, 3, 4)));
1471 extern void ext4_msg(struct super_block
*, const char *, const char *, ...)
1472 __attribute__ ((format (printf
, 3, 4)));
1473 extern void ext4_grp_locked_error(struct super_block
*, ext4_group_t
,
1474 const char *, const char *, ...)
1475 __attribute__ ((format (printf
, 4, 5)));
1476 extern void ext4_update_dynamic_rev(struct super_block
*sb
);
1477 extern int ext4_update_compat_feature(handle_t
*handle
, struct super_block
*sb
,
1479 extern int ext4_update_rocompat_feature(handle_t
*handle
,
1480 struct super_block
*sb
, __u32 rocompat
);
1481 extern int ext4_update_incompat_feature(handle_t
*handle
,
1482 struct super_block
*sb
, __u32 incompat
);
1483 extern ext4_fsblk_t
ext4_block_bitmap(struct super_block
*sb
,
1484 struct ext4_group_desc
*bg
);
1485 extern ext4_fsblk_t
ext4_inode_bitmap(struct super_block
*sb
,
1486 struct ext4_group_desc
*bg
);
1487 extern ext4_fsblk_t
ext4_inode_table(struct super_block
*sb
,
1488 struct ext4_group_desc
*bg
);
1489 extern __u32
ext4_free_blks_count(struct super_block
*sb
,
1490 struct ext4_group_desc
*bg
);
1491 extern __u32
ext4_free_inodes_count(struct super_block
*sb
,
1492 struct ext4_group_desc
*bg
);
1493 extern __u32
ext4_used_dirs_count(struct super_block
*sb
,
1494 struct ext4_group_desc
*bg
);
1495 extern __u32
ext4_itable_unused_count(struct super_block
*sb
,
1496 struct ext4_group_desc
*bg
);
1497 extern void ext4_block_bitmap_set(struct super_block
*sb
,
1498 struct ext4_group_desc
*bg
, ext4_fsblk_t blk
);
1499 extern void ext4_inode_bitmap_set(struct super_block
*sb
,
1500 struct ext4_group_desc
*bg
, ext4_fsblk_t blk
);
1501 extern void ext4_inode_table_set(struct super_block
*sb
,
1502 struct ext4_group_desc
*bg
, ext4_fsblk_t blk
);
1503 extern void ext4_free_blks_set(struct super_block
*sb
,
1504 struct ext4_group_desc
*bg
, __u32 count
);
1505 extern void ext4_free_inodes_set(struct super_block
*sb
,
1506 struct ext4_group_desc
*bg
, __u32 count
);
1507 extern void ext4_used_dirs_set(struct super_block
*sb
,
1508 struct ext4_group_desc
*bg
, __u32 count
);
1509 extern void ext4_itable_unused_set(struct super_block
*sb
,
1510 struct ext4_group_desc
*bg
, __u32 count
);
1511 extern __le16
ext4_group_desc_csum(struct ext4_sb_info
*sbi
, __u32 group
,
1512 struct ext4_group_desc
*gdp
);
1513 extern int ext4_group_desc_csum_verify(struct ext4_sb_info
*sbi
, __u32 group
,
1514 struct ext4_group_desc
*gdp
);
1516 static inline ext4_fsblk_t
ext4_blocks_count(struct ext4_super_block
*es
)
1518 return ((ext4_fsblk_t
)le32_to_cpu(es
->s_blocks_count_hi
) << 32) |
1519 le32_to_cpu(es
->s_blocks_count_lo
);
1522 static inline ext4_fsblk_t
ext4_r_blocks_count(struct ext4_super_block
*es
)
1524 return ((ext4_fsblk_t
)le32_to_cpu(es
->s_r_blocks_count_hi
) << 32) |
1525 le32_to_cpu(es
->s_r_blocks_count_lo
);
1528 static inline ext4_fsblk_t
ext4_free_blocks_count(struct ext4_super_block
*es
)
1530 return ((ext4_fsblk_t
)le32_to_cpu(es
->s_free_blocks_count_hi
) << 32) |
1531 le32_to_cpu(es
->s_free_blocks_count_lo
);
1534 static inline void ext4_blocks_count_set(struct ext4_super_block
*es
,
1537 es
->s_blocks_count_lo
= cpu_to_le32((u32
)blk
);
1538 es
->s_blocks_count_hi
= cpu_to_le32(blk
>> 32);
1541 static inline void ext4_free_blocks_count_set(struct ext4_super_block
*es
,
1544 es
->s_free_blocks_count_lo
= cpu_to_le32((u32
)blk
);
1545 es
->s_free_blocks_count_hi
= cpu_to_le32(blk
>> 32);
1548 static inline void ext4_r_blocks_count_set(struct ext4_super_block
*es
,
1551 es
->s_r_blocks_count_lo
= cpu_to_le32((u32
)blk
);
1552 es
->s_r_blocks_count_hi
= cpu_to_le32(blk
>> 32);
1555 static inline loff_t
ext4_isize(struct ext4_inode
*raw_inode
)
1557 if (S_ISREG(le16_to_cpu(raw_inode
->i_mode
)))
1558 return ((loff_t
)le32_to_cpu(raw_inode
->i_size_high
) << 32) |
1559 le32_to_cpu(raw_inode
->i_size_lo
);
1561 return (loff_t
) le32_to_cpu(raw_inode
->i_size_lo
);
1564 static inline void ext4_isize_set(struct ext4_inode
*raw_inode
, loff_t i_size
)
1566 raw_inode
->i_size_lo
= cpu_to_le32(i_size
);
1567 raw_inode
->i_size_high
= cpu_to_le32(i_size
>> 32);
1571 struct ext4_group_info
*ext4_get_group_info(struct super_block
*sb
,
1574 struct ext4_group_info
***grp_info
;
1575 long indexv
, indexh
;
1576 grp_info
= EXT4_SB(sb
)->s_group_info
;
1577 indexv
= group
>> (EXT4_DESC_PER_BLOCK_BITS(sb
));
1578 indexh
= group
& ((EXT4_DESC_PER_BLOCK(sb
)) - 1);
1579 return grp_info
[indexv
][indexh
];
1583 * Reading s_groups_count requires using smp_rmb() afterwards. See
1584 * the locking protocol documented in the comments of ext4_group_add()
1587 static inline ext4_group_t
ext4_get_groups_count(struct super_block
*sb
)
1589 ext4_group_t ngroups
= EXT4_SB(sb
)->s_groups_count
;
1595 static inline ext4_group_t
ext4_flex_group(struct ext4_sb_info
*sbi
,
1596 ext4_group_t block_group
)
1598 return block_group
>> sbi
->s_log_groups_per_flex
;
1601 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info
*sbi
)
1603 return 1 << sbi
->s_log_groups_per_flex
;
1606 #define ext4_std_error(sb, errno) \
1609 __ext4_std_error((sb), __func__, (errno)); \
1613 /* Each CPU can accumulate percpu_counter_batch blocks in their local
1614 * counters. So we need to make sure we have free blocks more
1615 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
1617 #define EXT4_FREEBLOCKS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
1619 #define EXT4_FREEBLOCKS_WATERMARK 0
1622 static inline void ext4_update_i_disksize(struct inode
*inode
, loff_t newsize
)
1625 * XXX: replace with spinlock if seen contended -bzzz
1627 down_write(&EXT4_I(inode
)->i_data_sem
);
1628 if (newsize
> EXT4_I(inode
)->i_disksize
)
1629 EXT4_I(inode
)->i_disksize
= newsize
;
1630 up_write(&EXT4_I(inode
)->i_data_sem
);
1634 struct ext4_group_info
{
1635 unsigned long bb_state
;
1636 struct rb_root bb_free_root
;
1637 ext4_grpblk_t bb_first_free
; /* first free block */
1638 ext4_grpblk_t bb_free
; /* total free blocks */
1639 ext4_grpblk_t bb_fragments
; /* nr of freespace fragments */
1640 struct list_head bb_prealloc_list
;
1644 struct rw_semaphore alloc_sem
;
1645 ext4_grpblk_t bb_counters
[]; /* Nr of free power-of-two-block
1646 * regions, index is order.
1647 * bb_counters[3] = 5 means
1648 * 5 free 8-block regions. */
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 */