4 * Copyright (C) 1992, 1993, 1994, 1995
5 * Remy Card (card@masi.ibp.fr)
6 * Laboratoire MASI - Institut Blaise Pascal
7 * Universite Pierre et Marie Curie (Paris VI)
11 * linux/include/linux/minix_fs.h
13 * Copyright (C) 1991, 1992 Linus Torvalds
19 #include <linux/types.h>
20 #include <linux/blkdev.h>
21 #include <linux/magic.h>
22 #include <linux/jbd2.h>
23 #include <linux/quota.h>
24 #include <linux/rwsem.h>
25 #include <linux/rbtree.h>
26 #include <linux/seqlock.h>
27 #include <linux/mutex.h>
28 #include <linux/timer.h>
29 #include <linux/wait.h>
30 #include <linux/blockgroup_lock.h>
31 #include <linux/percpu_counter.h>
33 #include <linux/compat.h>
37 * The fourth extended filesystem constants/structures
41 * Define EXT4FS_DEBUG to produce debug messages
49 #define ext4_debug(f, a...) \
51 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
52 __FILE__, __LINE__, __func__); \
53 printk(KERN_DEBUG f, ## a); \
56 #define ext4_debug(f, a...) do {} while (0)
59 #define EXT4_ERROR_INODE(inode, fmt, a...) \
60 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
62 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
63 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
65 #define EXT4_ERROR_FILE(file, block, fmt, a...) \
66 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
68 /* data type for block offset of block group */
69 typedef int ext4_grpblk_t
;
71 /* data type for filesystem-wide blocks number */
72 typedef unsigned long long ext4_fsblk_t
;
74 /* data type for file logical block number */
75 typedef __u32 ext4_lblk_t
;
77 /* data type for block group number */
78 typedef unsigned int ext4_group_t
;
81 * Flags used in mballoc's allocation_context flags field.
83 * Also used to show what's going on for debugging purposes when the
84 * flag field is exported via the traceport interface
87 /* prefer goal again. length */
88 #define EXT4_MB_HINT_MERGE 0x0001
89 /* blocks already reserved */
90 #define EXT4_MB_HINT_RESERVED 0x0002
91 /* metadata is being allocated */
92 #define EXT4_MB_HINT_METADATA 0x0004
93 /* first blocks in the file */
94 #define EXT4_MB_HINT_FIRST 0x0008
95 /* search for the best chunk */
96 #define EXT4_MB_HINT_BEST 0x0010
97 /* data is being allocated */
98 #define EXT4_MB_HINT_DATA 0x0020
99 /* don't preallocate (for tails) */
100 #define EXT4_MB_HINT_NOPREALLOC 0x0040
101 /* allocate for locality group */
102 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080
103 /* allocate goal blocks or none */
104 #define EXT4_MB_HINT_GOAL_ONLY 0x0100
105 /* goal is meaningful */
106 #define EXT4_MB_HINT_TRY_GOAL 0x0200
107 /* blocks already pre-reserved by delayed allocation */
108 #define EXT4_MB_DELALLOC_RESERVED 0x0400
109 /* We are doing stream allocation */
110 #define EXT4_MB_STREAM_ALLOC 0x0800
111 /* Use reserved root blocks if needed */
112 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000
114 struct ext4_allocation_request
{
115 /* target inode for block we're allocating */
117 /* how many blocks we want to allocate */
119 /* logical block in target inode */
121 /* the closest logical allocated block to the left */
123 /* the closest logical allocated block to the right */
125 /* phys. target (a hint) */
127 /* phys. block for the closest logical allocated block to the left */
129 /* phys. block for the closest logical allocated block to the right */
131 /* flags. see above EXT4_MB_HINT_* */
136 * Logical to physical block mapping, used by ext4_map_blocks()
138 * This structure is used to pass requests into ext4_map_blocks() as
139 * well as to store the information returned by ext4_map_blocks(). It
140 * takes less room on the stack than a struct buffer_head.
142 #define EXT4_MAP_NEW (1 << BH_New)
143 #define EXT4_MAP_MAPPED (1 << BH_Mapped)
144 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten)
145 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary)
146 #define EXT4_MAP_UNINIT (1 << BH_Uninit)
147 /* Sometimes (in the bigalloc case, from ext4_da_get_block_prep) the caller of
148 * ext4_map_blocks wants to know whether or not the underlying cluster has
149 * already been accounted for. EXT4_MAP_FROM_CLUSTER conveys to the caller that
150 * the requested mapping was from previously mapped (or delayed allocated)
151 * cluster. We use BH_AllocFromCluster only for this flag. BH_AllocFromCluster
152 * should never appear on buffer_head's state flags.
154 #define EXT4_MAP_FROM_CLUSTER (1 << BH_AllocFromCluster)
155 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
156 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\
157 EXT4_MAP_UNINIT | EXT4_MAP_FROM_CLUSTER)
159 struct ext4_map_blocks
{
163 unsigned int m_flags
;
167 * For delayed allocation tracking
169 struct mpage_da_data
{
171 sector_t b_blocknr
; /* start block number of extent */
172 size_t b_size
; /* size of extent */
173 unsigned long b_state
; /* state of the extent */
174 unsigned long first_page
, next_page
; /* extent of pages */
175 struct writeback_control
*wbc
;
182 * Flags for ext4_io_end->flags
184 #define EXT4_IO_END_UNWRITTEN 0x0001
185 #define EXT4_IO_END_ERROR 0x0002
186 #define EXT4_IO_END_QUEUED 0x0004
188 struct ext4_io_page
{
193 #define MAX_IO_PAGES 128
195 typedef struct ext4_io_end
{
196 struct list_head list
; /* per-file finished IO list */
197 struct inode
*inode
; /* file being written to */
198 unsigned int flag
; /* unwritten or not */
199 struct page
*page
; /* page struct for buffer write */
200 loff_t offset
; /* offset in the file */
201 ssize_t size
; /* size of the extent */
202 struct work_struct work
; /* data work queue */
203 struct kiocb
*iocb
; /* iocb struct for AIO */
204 int result
; /* error value for AIO */
206 struct ext4_io_page
*pages
[MAX_IO_PAGES
];
209 struct ext4_io_submit
{
212 ext4_io_end_t
*io_end
;
213 struct ext4_io_page
*io_page
;
214 sector_t io_next_block
;
218 * Special inodes numbers
220 #define EXT4_BAD_INO 1 /* Bad blocks inode */
221 #define EXT4_ROOT_INO 2 /* Root inode */
222 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */
223 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
224 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
225 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
226 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
227 #define EXT4_JOURNAL_INO 8 /* Journal inode */
229 /* First non-reserved inode for old ext4 filesystems */
230 #define EXT4_GOOD_OLD_FIRST_INO 11
233 * Maximal count of links to a file
235 #define EXT4_LINK_MAX 65000
238 * Macro-instructions used to manage several block sizes
240 #define EXT4_MIN_BLOCK_SIZE 1024
241 #define EXT4_MAX_BLOCK_SIZE 65536
242 #define EXT4_MIN_BLOCK_LOG_SIZE 10
243 #define EXT4_MAX_BLOCK_LOG_SIZE 16
245 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
247 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
249 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
250 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
251 EXT4_SB(s)->s_cluster_bits)
253 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
254 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
256 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
259 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
260 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
261 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
263 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
264 EXT4_GOOD_OLD_INODE_SIZE : \
266 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
267 EXT4_GOOD_OLD_FIRST_INO : \
270 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
272 /* Translate a block number to a cluster number */
273 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
274 /* Translate a cluster number to a block number */
275 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
276 /* Translate # of blks to # of clusters */
277 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
278 (sbi)->s_cluster_bits)
281 * Structure of a blocks group descriptor
283 struct ext4_group_desc
285 __le32 bg_block_bitmap_lo
; /* Blocks bitmap block */
286 __le32 bg_inode_bitmap_lo
; /* Inodes bitmap block */
287 __le32 bg_inode_table_lo
; /* Inodes table block */
288 __le16 bg_free_blocks_count_lo
;/* Free blocks count */
289 __le16 bg_free_inodes_count_lo
;/* Free inodes count */
290 __le16 bg_used_dirs_count_lo
; /* Directories count */
291 __le16 bg_flags
; /* EXT4_BG_flags (INODE_UNINIT, etc) */
292 __u32 bg_reserved
[2]; /* Likely block/inode bitmap checksum */
293 __le16 bg_itable_unused_lo
; /* Unused inodes count */
294 __le16 bg_checksum
; /* crc16(sb_uuid+group+desc) */
295 __le32 bg_block_bitmap_hi
; /* Blocks bitmap block MSB */
296 __le32 bg_inode_bitmap_hi
; /* Inodes bitmap block MSB */
297 __le32 bg_inode_table_hi
; /* Inodes table block MSB */
298 __le16 bg_free_blocks_count_hi
;/* Free blocks count MSB */
299 __le16 bg_free_inodes_count_hi
;/* Free inodes count MSB */
300 __le16 bg_used_dirs_count_hi
; /* Directories count MSB */
301 __le16 bg_itable_unused_hi
; /* Unused inodes count MSB */
302 __u32 bg_reserved2
[3];
306 * Structure of a flex block group info
310 atomic_t free_inodes
;
311 atomic_t free_clusters
;
315 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
316 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
317 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
320 * Macro-instructions used to manage group descriptors
322 #define EXT4_MIN_DESC_SIZE 32
323 #define EXT4_MIN_DESC_SIZE_64BIT 64
324 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
325 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
327 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
328 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
329 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
330 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
331 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
333 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
334 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
335 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
339 * Constants relative to the data blocks
341 #define EXT4_NDIR_BLOCKS 12
342 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
343 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
344 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
345 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
350 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
351 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
352 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
353 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
354 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
355 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
356 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
357 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
358 /* Reserved for compression usage... */
359 #define EXT4_DIRTY_FL 0x00000100
360 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
361 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
362 #define EXT4_ECOMPR_FL 0x00000800 /* Compression error */
363 /* End compression flags --- maybe not all used */
364 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
365 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
366 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
367 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
368 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
369 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
370 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
371 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
372 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
373 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
374 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
376 #define EXT4_FL_USER_VISIBLE 0x004BDFFF /* User visible flags */
377 #define EXT4_FL_USER_MODIFIABLE 0x004B80FF /* User modifiable flags */
379 /* Flags that should be inherited by new inodes from their parent. */
380 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
381 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
382 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
383 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
385 /* Flags that are appropriate for regular files (all but dir-specific ones). */
386 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
388 /* Flags that are appropriate for non-directories/regular files. */
389 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
391 /* Mask out flags that are inappropriate for the given type of inode. */
392 static inline __u32
ext4_mask_flags(umode_t mode
, __u32 flags
)
396 else if (S_ISREG(mode
))
397 return flags
& EXT4_REG_FLMASK
;
399 return flags
& EXT4_OTHER_FLMASK
;
403 * Inode flags used for atomic set/get
406 EXT4_INODE_SECRM
= 0, /* Secure deletion */
407 EXT4_INODE_UNRM
= 1, /* Undelete */
408 EXT4_INODE_COMPR
= 2, /* Compress file */
409 EXT4_INODE_SYNC
= 3, /* Synchronous updates */
410 EXT4_INODE_IMMUTABLE
= 4, /* Immutable file */
411 EXT4_INODE_APPEND
= 5, /* writes to file may only append */
412 EXT4_INODE_NODUMP
= 6, /* do not dump file */
413 EXT4_INODE_NOATIME
= 7, /* do not update atime */
414 /* Reserved for compression usage... */
415 EXT4_INODE_DIRTY
= 8,
416 EXT4_INODE_COMPRBLK
= 9, /* One or more compressed clusters */
417 EXT4_INODE_NOCOMPR
= 10, /* Don't compress */
418 EXT4_INODE_ECOMPR
= 11, /* Compression error */
419 /* End compression flags --- maybe not all used */
420 EXT4_INODE_INDEX
= 12, /* hash-indexed directory */
421 EXT4_INODE_IMAGIC
= 13, /* AFS directory */
422 EXT4_INODE_JOURNAL_DATA
= 14, /* file data should be journaled */
423 EXT4_INODE_NOTAIL
= 15, /* file tail should not be merged */
424 EXT4_INODE_DIRSYNC
= 16, /* dirsync behaviour (directories only) */
425 EXT4_INODE_TOPDIR
= 17, /* Top of directory hierarchies*/
426 EXT4_INODE_HUGE_FILE
= 18, /* Set to each huge file */
427 EXT4_INODE_EXTENTS
= 19, /* Inode uses extents */
428 EXT4_INODE_EA_INODE
= 21, /* Inode used for large EA */
429 EXT4_INODE_EOFBLOCKS
= 22, /* Blocks allocated beyond EOF */
430 EXT4_INODE_RESERVED
= 31, /* reserved for ext4 lib */
433 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
434 #define CHECK_FLAG_VALUE(FLAG) if (!TEST_FLAG_VALUE(FLAG)) { \
435 printk(KERN_EMERG "EXT4 flag fail: " #FLAG ": %d %d\n", \
436 EXT4_##FLAG##_FL, EXT4_INODE_##FLAG); BUG_ON(1); }
439 * Since it's pretty easy to mix up bit numbers and hex values, and we
440 * can't do a compile-time test for ENUM values, we use a run-time
441 * test to make sure that EXT4_XXX_FL is consistent with respect to
442 * EXT4_INODE_XXX. If all is well the printk and BUG_ON will all drop
443 * out so it won't cost any extra space in the compiled kernel image.
444 * But it's important that these values are the same, since we are
445 * using EXT4_INODE_XXX to test for the flag values, but EXT4_XX_FL
446 * must be consistent with the values of FS_XXX_FL defined in
447 * include/linux/fs.h and the on-disk values found in ext2, ext3, and
448 * ext4 filesystems, and of course the values defined in e2fsprogs.
450 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
452 static inline void ext4_check_flag_values(void)
454 CHECK_FLAG_VALUE(SECRM
);
455 CHECK_FLAG_VALUE(UNRM
);
456 CHECK_FLAG_VALUE(COMPR
);
457 CHECK_FLAG_VALUE(SYNC
);
458 CHECK_FLAG_VALUE(IMMUTABLE
);
459 CHECK_FLAG_VALUE(APPEND
);
460 CHECK_FLAG_VALUE(NODUMP
);
461 CHECK_FLAG_VALUE(NOATIME
);
462 CHECK_FLAG_VALUE(DIRTY
);
463 CHECK_FLAG_VALUE(COMPRBLK
);
464 CHECK_FLAG_VALUE(NOCOMPR
);
465 CHECK_FLAG_VALUE(ECOMPR
);
466 CHECK_FLAG_VALUE(INDEX
);
467 CHECK_FLAG_VALUE(IMAGIC
);
468 CHECK_FLAG_VALUE(JOURNAL_DATA
);
469 CHECK_FLAG_VALUE(NOTAIL
);
470 CHECK_FLAG_VALUE(DIRSYNC
);
471 CHECK_FLAG_VALUE(TOPDIR
);
472 CHECK_FLAG_VALUE(HUGE_FILE
);
473 CHECK_FLAG_VALUE(EXTENTS
);
474 CHECK_FLAG_VALUE(EA_INODE
);
475 CHECK_FLAG_VALUE(EOFBLOCKS
);
476 CHECK_FLAG_VALUE(RESERVED
);
479 /* Used to pass group descriptor data when online resize is done */
480 struct ext4_new_group_input
{
481 __u32 group
; /* Group number for this data */
482 __u64 block_bitmap
; /* Absolute block number of block bitmap */
483 __u64 inode_bitmap
; /* Absolute block number of inode bitmap */
484 __u64 inode_table
; /* Absolute block number of inode table start */
485 __u32 blocks_count
; /* Total number of blocks in this group */
486 __u16 reserved_blocks
; /* Number of reserved blocks in this group */
490 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
491 struct compat_ext4_new_group_input
{
493 compat_u64 block_bitmap
;
494 compat_u64 inode_bitmap
;
495 compat_u64 inode_table
;
502 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
503 struct ext4_new_group_data
{
509 __u16 reserved_blocks
;
511 __u32 free_blocks_count
;
515 * Flags used by ext4_map_blocks()
517 /* Allocate any needed blocks and/or convert an unitialized
518 extent to be an initialized ext4 */
519 #define EXT4_GET_BLOCKS_CREATE 0x0001
520 /* Request the creation of an unitialized extent */
521 #define EXT4_GET_BLOCKS_UNINIT_EXT 0x0002
522 #define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT (EXT4_GET_BLOCKS_UNINIT_EXT|\
523 EXT4_GET_BLOCKS_CREATE)
524 /* Caller is from the delayed allocation writeout path,
525 so set the magic i_delalloc_reserve_flag after taking the
526 inode allocation semaphore for */
527 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
528 /* caller is from the direct IO path, request to creation of an
529 unitialized extents if not allocated, split the uninitialized
530 extent if blocks has been preallocated already*/
531 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
532 #define EXT4_GET_BLOCKS_CONVERT 0x0010
533 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
534 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
535 /* Convert extent to initialized after IO complete */
536 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
537 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
538 /* Punch out blocks of an extent */
539 #define EXT4_GET_BLOCKS_PUNCH_OUT_EXT 0x0020
540 /* Don't normalize allocation size (used for fallocate) */
541 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
544 * Flags used by ext4_free_blocks
546 #define EXT4_FREE_BLOCKS_METADATA 0x0001
547 #define EXT4_FREE_BLOCKS_FORGET 0x0002
548 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
549 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
550 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
551 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
554 * Flags used by ext4_discard_partial_page_buffers
556 #define EXT4_DISCARD_PARTIAL_PG_ZERO_UNMAPPED 0x0001
561 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
562 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
563 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
564 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
565 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
566 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
567 #ifdef CONFIG_JBD2_DEBUG
568 #define EXT4_IOC_WAIT_FOR_READONLY _IOR('f', 99, long)
570 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
571 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
572 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
573 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
574 #define EXT4_IOC_MIGRATE _IO('f', 9)
575 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
576 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
577 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
578 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
580 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
582 * ioctl commands in 32 bit emulation
584 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
585 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
586 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
587 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
588 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
589 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
590 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
591 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
592 #ifdef CONFIG_JBD2_DEBUG
593 #define EXT4_IOC32_WAIT_FOR_READONLY _IOR('f', 99, int)
595 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
596 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
599 /* Max physical block we can address w/o extents */
600 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
603 * Structure of an inode on the disk
606 __le16 i_mode
; /* File mode */
607 __le16 i_uid
; /* Low 16 bits of Owner Uid */
608 __le32 i_size_lo
; /* Size in bytes */
609 __le32 i_atime
; /* Access time */
610 __le32 i_ctime
; /* Inode Change time */
611 __le32 i_mtime
; /* Modification time */
612 __le32 i_dtime
; /* Deletion Time */
613 __le16 i_gid
; /* Low 16 bits of Group Id */
614 __le16 i_links_count
; /* Links count */
615 __le32 i_blocks_lo
; /* Blocks count */
616 __le32 i_flags
; /* File flags */
622 __u32 h_i_translator
;
627 } osd1
; /* OS dependent 1 */
628 __le32 i_block
[EXT4_N_BLOCKS
];/* Pointers to blocks */
629 __le32 i_generation
; /* File version (for NFS) */
630 __le32 i_file_acl_lo
; /* File ACL */
632 __le32 i_obso_faddr
; /* Obsoleted fragment address */
635 __le16 l_i_blocks_high
; /* were l_i_reserved1 */
636 __le16 l_i_file_acl_high
;
637 __le16 l_i_uid_high
; /* these 2 fields */
638 __le16 l_i_gid_high
; /* were reserved2[0] */
642 __le16 h_i_reserved1
; /* Obsoleted fragment number/size which are removed in ext4 */
649 __le16 h_i_reserved1
; /* Obsoleted fragment number/size which are removed in ext4 */
650 __le16 m_i_file_acl_high
;
651 __u32 m_i_reserved2
[2];
653 } osd2
; /* OS dependent 2 */
654 __le16 i_extra_isize
;
656 __le32 i_ctime_extra
; /* extra Change time (nsec << 2 | epoch) */
657 __le32 i_mtime_extra
; /* extra Modification time(nsec << 2 | epoch) */
658 __le32 i_atime_extra
; /* extra Access time (nsec << 2 | epoch) */
659 __le32 i_crtime
; /* File Creation time */
660 __le32 i_crtime_extra
; /* extra FileCreationtime (nsec << 2 | epoch) */
661 __le32 i_version_hi
; /* high 32 bits for 64-bit version */
665 __u32 reserved
; /* should be zero */
666 __u32 donor_fd
; /* donor file descriptor */
667 __u64 orig_start
; /* logical start offset in block for orig */
668 __u64 donor_start
; /* logical start offset in block for donor */
669 __u64 len
; /* block length to be moved */
670 __u64 moved_len
; /* moved block length */
673 #define EXT4_EPOCH_BITS 2
674 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
675 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
678 * Extended fields will fit into an inode if the filesystem was formatted
679 * with large inodes (-I 256 or larger) and there are not currently any EAs
680 * consuming all of the available space. For new inodes we always reserve
681 * enough space for the kernel's known extended fields, but for inodes
682 * created with an old kernel this might not have been the case. None of
683 * the extended inode fields is critical for correct filesystem operation.
684 * This macro checks if a certain field fits in the inode. Note that
685 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
687 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
688 ((offsetof(typeof(*ext4_inode), field) + \
689 sizeof((ext4_inode)->field)) \
690 <= (EXT4_GOOD_OLD_INODE_SIZE + \
691 (einode)->i_extra_isize)) \
693 static inline __le32 ext4_encode_extra_time(struct timespec *time)
695 return cpu_to_le32((sizeof(time
->tv_sec
) > 4 ?
696 (time
->tv_sec
>> 32) & EXT4_EPOCH_MASK
: 0) |
697 ((time
->tv_nsec
<< EXT4_EPOCH_BITS
) & EXT4_NSEC_MASK
));
700 static inline void ext4_decode_extra_time(struct timespec
*time
, __le32 extra
)
702 if (sizeof(time
->tv_sec
) > 4)
703 time
->tv_sec
|= (__u64
)(le32_to_cpu(extra
) & EXT4_EPOCH_MASK
)
705 time
->tv_nsec
= (le32_to_cpu(extra
) & EXT4_NSEC_MASK
) >> EXT4_EPOCH_BITS
;
708 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
710 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
711 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
712 (raw_inode)->xtime ## _extra = \
713 ext4_encode_extra_time(&(inode)->xtime); \
716 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
718 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
719 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
720 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
721 (raw_inode)->xtime ## _extra = \
722 ext4_encode_extra_time(&(einode)->xtime); \
725 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
727 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
728 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
729 ext4_decode_extra_time(&(inode)->xtime, \
730 raw_inode->xtime ## _extra); \
732 (inode)->xtime.tv_nsec = 0; \
735 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
737 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
738 (einode)->xtime.tv_sec = \
739 (signed)le32_to_cpu((raw_inode)->xtime); \
740 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
741 ext4_decode_extra_time(&(einode)->xtime, \
742 raw_inode->xtime ## _extra); \
744 (einode)->xtime.tv_nsec = 0; \
747 #define i_disk_version osd1.linux1.l_i_version
749 #if defined(__KERNEL__) || defined(__linux__)
750 #define i_reserved1 osd1.linux1.l_i_reserved1
751 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
752 #define i_blocks_high osd2.linux2.l_i_blocks_high
753 #define i_uid_low i_uid
754 #define i_gid_low i_gid
755 #define i_uid_high osd2.linux2.l_i_uid_high
756 #define i_gid_high osd2.linux2.l_i_gid_high
757 #define i_reserved2 osd2.linux2.l_i_reserved2
759 #elif defined(__GNU__)
761 #define i_translator osd1.hurd1.h_i_translator
762 #define i_uid_high osd2.hurd2.h_i_uid_high
763 #define i_gid_high osd2.hurd2.h_i_gid_high
764 #define i_author osd2.hurd2.h_i_author
766 #elif defined(__masix__)
768 #define i_reserved1 osd1.masix1.m_i_reserved1
769 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
770 #define i_reserved2 osd2.masix2.m_i_reserved2
772 #endif /* defined(__KERNEL__) || defined(__linux__) */
775 * storage for cached extent
776 * If ec_len == 0, then the cache is invalid.
777 * If ec_start == 0, then the cache represents a gap (null mapping)
779 struct ext4_ext_cache
{
780 ext4_fsblk_t ec_start
;
781 ext4_lblk_t ec_block
;
782 __u32 ec_len
; /* must be 32bit to return holes */
786 * fourth extended file system inode data in memory
788 struct ext4_inode_info
{
789 __le32 i_data
[15]; /* unconverted */
791 ext4_fsblk_t i_file_acl
;
794 * i_block_group is the number of the block group which contains
795 * this file's inode. Constant across the lifetime of the inode,
796 * it is ued for making block allocation decisions - we try to
797 * place a file's data blocks near its inode block, and new inodes
798 * near to their parent directory's inode.
800 ext4_group_t i_block_group
;
801 ext4_lblk_t i_dir_start_lookup
;
802 #if (BITS_PER_LONG < 64)
803 unsigned long i_state_flags
; /* Dynamic state flags */
805 unsigned long i_flags
;
807 #ifdef CONFIG_EXT4_FS_XATTR
809 * Extended attributes can be read independently of the main file
810 * data. Taking i_mutex even when reading would cause contention
811 * between readers of EAs and writers of regular file data, so
812 * instead we synchronize on xattr_sem when reading or changing
815 struct rw_semaphore xattr_sem
;
818 struct list_head i_orphan
; /* unlinked but open inodes */
821 * i_disksize keeps track of what the inode size is ON DISK, not
822 * in memory. During truncate, i_size is set to the new size by
823 * the VFS prior to calling ext4_truncate(), but the filesystem won't
824 * set i_disksize to 0 until the truncate is actually under way.
826 * The intent is that i_disksize always represents the blocks which
827 * are used by this file. This allows recovery to restart truncate
828 * on orphans if we crash during truncate. We actually write i_disksize
829 * into the on-disk inode when writing inodes out, instead of i_size.
831 * The only time when i_disksize and i_size may be different is when
832 * a truncate is in progress. The only things which change i_disksize
833 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
838 * i_data_sem is for serialising ext4_truncate() against
839 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
840 * data tree are chopped off during truncate. We can't do that in
841 * ext4 because whenever we perform intermediate commits during
842 * truncate, the inode and all the metadata blocks *must* be in a
843 * consistent state which allows truncation of the orphans to restart
844 * during recovery. Hence we must fix the get_block-vs-truncate race
845 * by other means, so we have i_data_sem.
847 struct rw_semaphore i_data_sem
;
848 struct inode vfs_inode
;
849 struct jbd2_inode
*jinode
;
851 struct ext4_ext_cache i_cached_extent
;
853 * File creation time. Its function is same as that of
854 * struct timespec i_{a,c,m}time in the generic inode.
856 struct timespec i_crtime
;
859 struct list_head i_prealloc_list
;
860 spinlock_t i_prealloc_lock
;
863 ext4_group_t i_last_alloc_group
;
865 /* allocation reservation info for delalloc */
866 /* In case of bigalloc, these refer to clusters rather than blocks */
867 unsigned int i_reserved_data_blocks
;
868 unsigned int i_reserved_meta_blocks
;
869 unsigned int i_allocated_meta_blocks
;
870 ext4_lblk_t i_da_metadata_calc_last_lblock
;
871 int i_da_metadata_calc_len
;
873 /* on-disk additional length */
877 /* quota space reservation, managed internally by quota code */
878 qsize_t i_reserved_quota
;
881 /* completed IOs that might need unwritten extents handling */
882 struct list_head i_completed_io_list
;
883 spinlock_t i_completed_io_lock
;
884 atomic_t i_ioend_count
; /* Number of outstanding io_end structs */
885 /* current io_end structure for async DIO write*/
886 ext4_io_end_t
*cur_aio_dio
;
887 atomic_t i_aiodio_unwritten
; /* Nr. of inflight conversions pending */
889 spinlock_t i_block_reservation_lock
;
892 * Transactions that contain inode's metadata needed to complete
893 * fsync and fdatasync, respectively.
896 tid_t i_datasync_tid
;
902 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
903 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
904 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
907 * Misc. filesystem flags
909 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
910 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
911 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
916 #define EXT4_MOUNT_OLDALLOC 0x00002 /* Don't use the new Orlov allocator */
917 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
918 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
919 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
920 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
921 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
922 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
923 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
924 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
925 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
926 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
927 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
928 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
929 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
930 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
931 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
932 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
933 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
934 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
935 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
936 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
937 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
938 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
939 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
940 #define EXT4_MOUNT_I_VERSION 0x2000000 /* i_version support */
941 #define EXT4_MOUNT_MBLK_IO_SUBMIT 0x4000000 /* multi-block io submits */
942 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
943 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
944 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
945 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
946 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
948 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
949 specified delalloc */
951 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
953 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
955 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
958 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
960 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
962 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
965 #define ext4_set_bit __test_and_set_bit_le
966 #define ext4_set_bit_atomic ext2_set_bit_atomic
967 #define ext4_clear_bit __test_and_clear_bit_le
968 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
969 #define ext4_test_bit test_bit_le
970 #define ext4_find_first_zero_bit find_first_zero_bit_le
971 #define ext4_find_next_zero_bit find_next_zero_bit_le
972 #define ext4_find_next_bit find_next_bit_le
974 extern void ext4_set_bits(void *bm
, int cur
, int len
);
977 * Maximal mount counts between two filesystem checks
979 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
980 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
983 * Behaviour when detecting errors
985 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
986 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
987 #define EXT4_ERRORS_PANIC 3 /* Panic */
988 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
991 * Structure of the super block
993 struct ext4_super_block
{
994 /*00*/ __le32 s_inodes_count
; /* Inodes count */
995 __le32 s_blocks_count_lo
; /* Blocks count */
996 __le32 s_r_blocks_count_lo
; /* Reserved blocks count */
997 __le32 s_free_blocks_count_lo
; /* Free blocks count */
998 /*10*/ __le32 s_free_inodes_count
; /* Free inodes count */
999 __le32 s_first_data_block
; /* First Data Block */
1000 __le32 s_log_block_size
; /* Block size */
1001 __le32 s_log_cluster_size
; /* Allocation cluster size */
1002 /*20*/ __le32 s_blocks_per_group
; /* # Blocks per group */
1003 __le32 s_clusters_per_group
; /* # Clusters per group */
1004 __le32 s_inodes_per_group
; /* # Inodes per group */
1005 __le32 s_mtime
; /* Mount time */
1006 /*30*/ __le32 s_wtime
; /* Write time */
1007 __le16 s_mnt_count
; /* Mount count */
1008 __le16 s_max_mnt_count
; /* Maximal mount count */
1009 __le16 s_magic
; /* Magic signature */
1010 __le16 s_state
; /* File system state */
1011 __le16 s_errors
; /* Behaviour when detecting errors */
1012 __le16 s_minor_rev_level
; /* minor revision level */
1013 /*40*/ __le32 s_lastcheck
; /* time of last check */
1014 __le32 s_checkinterval
; /* max. time between checks */
1015 __le32 s_creator_os
; /* OS */
1016 __le32 s_rev_level
; /* Revision level */
1017 /*50*/ __le16 s_def_resuid
; /* Default uid for reserved blocks */
1018 __le16 s_def_resgid
; /* Default gid for reserved blocks */
1020 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1022 * Note: the difference between the compatible feature set and
1023 * the incompatible feature set is that if there is a bit set
1024 * in the incompatible feature set that the kernel doesn't
1025 * know about, it should refuse to mount the filesystem.
1027 * e2fsck's requirements are more strict; if it doesn't know
1028 * about a feature in either the compatible or incompatible
1029 * feature set, it must abort and not try to meddle with
1030 * things it doesn't understand...
1032 __le32 s_first_ino
; /* First non-reserved inode */
1033 __le16 s_inode_size
; /* size of inode structure */
1034 __le16 s_block_group_nr
; /* block group # of this superblock */
1035 __le32 s_feature_compat
; /* compatible feature set */
1036 /*60*/ __le32 s_feature_incompat
; /* incompatible feature set */
1037 __le32 s_feature_ro_compat
; /* readonly-compatible feature set */
1038 /*68*/ __u8 s_uuid
[16]; /* 128-bit uuid for volume */
1039 /*78*/ char s_volume_name
[16]; /* volume name */
1040 /*88*/ char s_last_mounted
[64]; /* directory where last mounted */
1041 /*C8*/ __le32 s_algorithm_usage_bitmap
; /* For compression */
1043 * Performance hints. Directory preallocation should only
1044 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1046 __u8 s_prealloc_blocks
; /* Nr of blocks to try to preallocate*/
1047 __u8 s_prealloc_dir_blocks
; /* Nr to preallocate for dirs */
1048 __le16 s_reserved_gdt_blocks
; /* Per group desc for online growth */
1050 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1052 /*D0*/ __u8 s_journal_uuid
[16]; /* uuid of journal superblock */
1053 /*E0*/ __le32 s_journal_inum
; /* inode number of journal file */
1054 __le32 s_journal_dev
; /* device number of journal file */
1055 __le32 s_last_orphan
; /* start of list of inodes to delete */
1056 __le32 s_hash_seed
[4]; /* HTREE hash seed */
1057 __u8 s_def_hash_version
; /* Default hash version to use */
1058 __u8 s_jnl_backup_type
;
1059 __le16 s_desc_size
; /* size of group descriptor */
1060 /*100*/ __le32 s_default_mount_opts
;
1061 __le32 s_first_meta_bg
; /* First metablock block group */
1062 __le32 s_mkfs_time
; /* When the filesystem was created */
1063 __le32 s_jnl_blocks
[17]; /* Backup of the journal inode */
1064 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1065 /*150*/ __le32 s_blocks_count_hi
; /* Blocks count */
1066 __le32 s_r_blocks_count_hi
; /* Reserved blocks count */
1067 __le32 s_free_blocks_count_hi
; /* Free blocks count */
1068 __le16 s_min_extra_isize
; /* All inodes have at least # bytes */
1069 __le16 s_want_extra_isize
; /* New inodes should reserve # bytes */
1070 __le32 s_flags
; /* Miscellaneous flags */
1071 __le16 s_raid_stride
; /* RAID stride */
1072 __le16 s_mmp_update_interval
; /* # seconds to wait in MMP checking */
1073 __le64 s_mmp_block
; /* Block for multi-mount protection */
1074 __le32 s_raid_stripe_width
; /* blocks on all data disks (N*stride)*/
1075 __u8 s_log_groups_per_flex
; /* FLEX_BG group size */
1076 __u8 s_reserved_char_pad
;
1077 __le16 s_reserved_pad
;
1078 __le64 s_kbytes_written
; /* nr of lifetime kilobytes written */
1079 __le32 s_snapshot_inum
; /* Inode number of active snapshot */
1080 __le32 s_snapshot_id
; /* sequential ID of active snapshot */
1081 __le64 s_snapshot_r_blocks_count
; /* reserved blocks for active
1082 snapshot's future use */
1083 __le32 s_snapshot_list
; /* inode number of the head of the
1084 on-disk snapshot list */
1085 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1086 __le32 s_error_count
; /* number of fs errors */
1087 __le32 s_first_error_time
; /* first time an error happened */
1088 __le32 s_first_error_ino
; /* inode involved in first error */
1089 __le64 s_first_error_block
; /* block involved of first error */
1090 __u8 s_first_error_func
[32]; /* function where the error happened */
1091 __le32 s_first_error_line
; /* line number where error happened */
1092 __le32 s_last_error_time
; /* most recent time of an error */
1093 __le32 s_last_error_ino
; /* inode involved in last error */
1094 __le32 s_last_error_line
; /* line number where error happened */
1095 __le64 s_last_error_block
; /* block involved of last error */
1096 __u8 s_last_error_func
[32]; /* function where the error happened */
1097 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1098 __u8 s_mount_opts
[64];
1099 __le32 s_usr_quota_inum
; /* inode for tracking user quota */
1100 __le32 s_grp_quota_inum
; /* inode for tracking group quota */
1101 __le32 s_overhead_clusters
; /* overhead blocks/clusters in fs */
1102 __le32 s_reserved
[109]; /* Padding to the end of the block */
1105 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1110 * run-time mount flags
1112 #define EXT4_MF_MNTDIR_SAMPLED 0x0001
1113 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1116 * fourth extended-fs super-block data in memory
1118 struct ext4_sb_info
{
1119 unsigned long s_desc_size
; /* Size of a group descriptor in bytes */
1120 unsigned long s_inodes_per_block
;/* Number of inodes per block */
1121 unsigned long s_blocks_per_group
;/* Number of blocks in a group */
1122 unsigned long s_clusters_per_group
; /* Number of clusters in a group */
1123 unsigned long s_inodes_per_group
;/* Number of inodes in a group */
1124 unsigned long s_itb_per_group
; /* Number of inode table blocks per group */
1125 unsigned long s_gdb_count
; /* Number of group descriptor blocks */
1126 unsigned long s_desc_per_block
; /* Number of group descriptors per block */
1127 ext4_group_t s_groups_count
; /* Number of groups in the fs */
1128 ext4_group_t s_blockfile_groups
;/* Groups acceptable for non-extent files */
1129 unsigned long s_overhead_last
; /* Last calculated overhead */
1130 unsigned long s_blocks_last
; /* Last seen block count */
1131 unsigned int s_cluster_ratio
; /* Number of blocks per cluster */
1132 unsigned int s_cluster_bits
; /* log2 of s_cluster_ratio */
1133 loff_t s_bitmap_maxbytes
; /* max bytes for bitmap files */
1134 struct buffer_head
* s_sbh
; /* Buffer containing the super block */
1135 struct ext4_super_block
*s_es
; /* Pointer to the super block in the buffer */
1136 struct buffer_head
**s_group_desc
;
1137 unsigned int s_mount_opt
;
1138 unsigned int s_mount_opt2
;
1139 unsigned int s_mount_flags
;
1140 ext4_fsblk_t s_sb_block
;
1143 unsigned short s_mount_state
;
1144 unsigned short s_pad
;
1145 int s_addr_per_block_bits
;
1146 int s_desc_per_block_bits
;
1149 unsigned int s_inode_readahead_blks
;
1150 unsigned int s_inode_goal
;
1151 spinlock_t s_next_gen_lock
;
1152 u32 s_next_generation
;
1154 int s_def_hash_version
;
1155 int s_hash_unsigned
; /* 3 if hash should be signed, 0 if not */
1156 struct percpu_counter s_freeclusters_counter
;
1157 struct percpu_counter s_freeinodes_counter
;
1158 struct percpu_counter s_dirs_counter
;
1159 struct percpu_counter s_dirtyclusters_counter
;
1160 struct blockgroup_lock
*s_blockgroup_lock
;
1161 struct proc_dir_entry
*s_proc
;
1162 struct kobject s_kobj
;
1163 struct completion s_kobj_unregister
;
1166 struct journal_s
*s_journal
;
1167 struct list_head s_orphan
;
1168 struct mutex s_orphan_lock
;
1169 unsigned long s_resize_flags
; /* Flags indicating if there
1171 unsigned long s_commit_interval
;
1172 u32 s_max_batch_time
;
1173 u32 s_min_batch_time
;
1174 struct block_device
*journal_bdev
;
1175 #ifdef CONFIG_JBD2_DEBUG
1176 struct timer_list turn_ro_timer
; /* For turning read-only (crash simulation) */
1177 wait_queue_head_t ro_wait_queue
; /* For people waiting for the fs to go read-only */
1180 char *s_qf_names
[MAXQUOTAS
]; /* Names of quota files with journalled quota */
1181 int s_jquota_fmt
; /* Format of quota to use */
1183 unsigned int s_want_extra_isize
; /* New inodes should reserve # bytes */
1184 struct rb_root system_blks
;
1186 #ifdef EXTENTS_STATS
1187 /* ext4 extents stats */
1188 unsigned long s_ext_min
;
1189 unsigned long s_ext_max
;
1190 unsigned long s_depth_max
;
1191 spinlock_t s_ext_stats_lock
;
1192 unsigned long s_ext_blocks
;
1193 unsigned long s_ext_extents
;
1195 /* ext4 extent cache stats */
1196 unsigned long extent_cache_hits
;
1197 unsigned long extent_cache_misses
;
1199 /* for buddy allocator */
1200 struct ext4_group_info
***s_group_info
;
1201 struct inode
*s_buddy_cache
;
1202 spinlock_t s_md_lock
;
1203 unsigned short *s_mb_offsets
;
1204 unsigned int *s_mb_maxs
;
1207 unsigned long s_stripe
;
1208 unsigned int s_mb_stream_request
;
1209 unsigned int s_mb_max_to_scan
;
1210 unsigned int s_mb_min_to_scan
;
1211 unsigned int s_mb_stats
;
1212 unsigned int s_mb_order2_reqs
;
1213 unsigned int s_mb_group_prealloc
;
1214 unsigned int s_max_writeback_mb_bump
;
1215 /* where last allocation was done - for stream allocation */
1216 unsigned long s_mb_last_group
;
1217 unsigned long s_mb_last_start
;
1219 /* stats for buddy allocator */
1220 atomic_t s_bal_reqs
; /* number of reqs with len > 1 */
1221 atomic_t s_bal_success
; /* we found long enough chunks */
1222 atomic_t s_bal_allocated
; /* in blocks */
1223 atomic_t s_bal_ex_scanned
; /* total extents scanned */
1224 atomic_t s_bal_goals
; /* goal hits */
1225 atomic_t s_bal_breaks
; /* too long searches */
1226 atomic_t s_bal_2orders
; /* 2^order hits */
1227 spinlock_t s_bal_lock
;
1228 unsigned long s_mb_buddies_generated
;
1229 unsigned long long s_mb_generation_time
;
1230 atomic_t s_mb_lost_chunks
;
1231 atomic_t s_mb_preallocated
;
1232 atomic_t s_mb_discarded
;
1233 atomic_t s_lock_busy
;
1235 /* locality groups */
1236 struct ext4_locality_group __percpu
*s_locality_groups
;
1238 /* for write statistics */
1239 unsigned long s_sectors_written_start
;
1240 u64 s_kbytes_written
;
1242 unsigned int s_log_groups_per_flex
;
1243 struct flex_groups
*s_flex_groups
;
1245 /* workqueue for dio unwritten */
1246 struct workqueue_struct
*dio_unwritten_wq
;
1248 /* timer for periodic error stats printing */
1249 struct timer_list s_err_report
;
1251 /* Lazy inode table initialization info */
1252 struct ext4_li_request
*s_li_request
;
1253 /* Wait multiplier for lazy initialization thread */
1254 unsigned int s_li_wait_mult
;
1256 /* Kernel thread for multiple mount protection */
1257 struct task_struct
*s_mmp_tsk
;
1259 /* record the last minlen when FITRIM is called. */
1260 atomic_t s_last_trim_minblks
;
1263 static inline struct ext4_sb_info
*EXT4_SB(struct super_block
*sb
)
1265 return sb
->s_fs_info
;
1267 static inline struct ext4_inode_info
*EXT4_I(struct inode
*inode
)
1269 return container_of(inode
, struct ext4_inode_info
, vfs_inode
);
1272 static inline struct timespec
ext4_current_time(struct inode
*inode
)
1274 return (inode
->i_sb
->s_time_gran
< NSEC_PER_SEC
) ?
1275 current_fs_time(inode
->i_sb
) : CURRENT_TIME_SEC
;
1278 static inline int ext4_valid_inum(struct super_block
*sb
, unsigned long ino
)
1280 return ino
== EXT4_ROOT_INO
||
1281 ino
== EXT4_JOURNAL_INO
||
1282 ino
== EXT4_RESIZE_INO
||
1283 (ino
>= EXT4_FIRST_INO(sb
) &&
1284 ino
<= le32_to_cpu(EXT4_SB(sb
)->s_es
->s_inodes_count
));
1288 * Inode dynamic state flags
1291 EXT4_STATE_JDATA
, /* journaled data exists */
1292 EXT4_STATE_NEW
, /* inode is newly created */
1293 EXT4_STATE_XATTR
, /* has in-inode xattrs */
1294 EXT4_STATE_NO_EXPAND
, /* No space for expansion */
1295 EXT4_STATE_DA_ALLOC_CLOSE
, /* Alloc DA blks on close */
1296 EXT4_STATE_EXT_MIGRATE
, /* Inode is migrating */
1297 EXT4_STATE_DIO_UNWRITTEN
, /* need convert on dio done*/
1298 EXT4_STATE_NEWENTRY
, /* File just added to dir */
1299 EXT4_STATE_DELALLOC_RESERVED
, /* blks already reserved for delalloc */
1302 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1303 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1305 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1307 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1309 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1311 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1313 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1316 EXT4_INODE_BIT_FNS(flag
, flags
, 0)
1317 #if (BITS_PER_LONG < 64)
1318 EXT4_INODE_BIT_FNS(state
, state_flags
, 0)
1320 static inline void ext4_clear_state_flags(struct ext4_inode_info
*ei
)
1322 (ei
)->i_state_flags
= 0;
1325 EXT4_INODE_BIT_FNS(state
, flags
, 32)
1327 static inline void ext4_clear_state_flags(struct ext4_inode_info
*ei
)
1329 /* We depend on the fact that callers will set i_flags */
1333 /* Assume that user mode programs are passing in an ext4fs superblock, not
1334 * a kernel struct super_block. This will allow us to call the feature-test
1335 * macros from user land. */
1336 #define EXT4_SB(sb) (sb)
1339 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1342 * Codes for operating systems
1344 #define EXT4_OS_LINUX 0
1345 #define EXT4_OS_HURD 1
1346 #define EXT4_OS_MASIX 2
1347 #define EXT4_OS_FREEBSD 3
1348 #define EXT4_OS_LITES 4
1353 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1354 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1356 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1357 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1359 #define EXT4_GOOD_OLD_INODE_SIZE 128
1362 * Feature set definitions
1365 #define EXT4_HAS_COMPAT_FEATURE(sb,mask) \
1366 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1367 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \
1368 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1369 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \
1370 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1371 #define EXT4_SET_COMPAT_FEATURE(sb,mask) \
1372 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1373 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \
1374 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1375 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \
1376 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1377 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \
1378 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1379 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
1380 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1381 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \
1382 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1384 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1385 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1386 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1387 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1388 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1389 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1391 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1392 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1393 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1394 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1395 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1396 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1397 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1398 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1399 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1401 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1402 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1403 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1404 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1405 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1406 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1407 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1408 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1409 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1410 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1411 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1413 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1414 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1415 EXT4_FEATURE_INCOMPAT_META_BG)
1416 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1417 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1418 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1420 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1421 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1422 EXT4_FEATURE_INCOMPAT_RECOVER| \
1423 EXT4_FEATURE_INCOMPAT_META_BG)
1424 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1425 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1426 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1428 #define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR
1429 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1430 EXT4_FEATURE_INCOMPAT_RECOVER| \
1431 EXT4_FEATURE_INCOMPAT_META_BG| \
1432 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1433 EXT4_FEATURE_INCOMPAT_64BIT| \
1434 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1435 EXT4_FEATURE_INCOMPAT_MMP)
1436 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1437 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1438 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1439 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1440 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1441 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1442 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1443 EXT4_FEATURE_RO_COMPAT_BIGALLOC)
1446 * Default values for user and/or group using reserved blocks
1448 #define EXT4_DEF_RESUID 0
1449 #define EXT4_DEF_RESGID 0
1451 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
1454 * Default mount options
1456 #define EXT4_DEFM_DEBUG 0x0001
1457 #define EXT4_DEFM_BSDGROUPS 0x0002
1458 #define EXT4_DEFM_XATTR_USER 0x0004
1459 #define EXT4_DEFM_ACL 0x0008
1460 #define EXT4_DEFM_UID16 0x0010
1461 #define EXT4_DEFM_JMODE 0x0060
1462 #define EXT4_DEFM_JMODE_DATA 0x0020
1463 #define EXT4_DEFM_JMODE_ORDERED 0x0040
1464 #define EXT4_DEFM_JMODE_WBACK 0x0060
1465 #define EXT4_DEFM_NOBARRIER 0x0100
1466 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1467 #define EXT4_DEFM_DISCARD 0x0400
1468 #define EXT4_DEFM_NODELALLOC 0x0800
1471 * Default journal batch times
1473 #define EXT4_DEF_MIN_BATCH_TIME 0
1474 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1477 * Minimum number of groups in a flexgroup before we separate out
1478 * directories into the first block group of a flexgroup
1480 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1483 * Structure of a directory entry
1485 #define EXT4_NAME_LEN 255
1487 struct ext4_dir_entry
{
1488 __le32 inode
; /* Inode number */
1489 __le16 rec_len
; /* Directory entry length */
1490 __le16 name_len
; /* Name length */
1491 char name
[EXT4_NAME_LEN
]; /* File name */
1495 * The new version of the directory entry. Since EXT4 structures are
1496 * stored in intel byte order, and the name_len field could never be
1497 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1500 struct ext4_dir_entry_2
{
1501 __le32 inode
; /* Inode number */
1502 __le16 rec_len
; /* Directory entry length */
1503 __u8 name_len
; /* Name length */
1505 char name
[EXT4_NAME_LEN
]; /* File name */
1509 * Ext4 directory file types. Only the low 3 bits are used. The
1510 * other bits are reserved for now.
1512 #define EXT4_FT_UNKNOWN 0
1513 #define EXT4_FT_REG_FILE 1
1514 #define EXT4_FT_DIR 2
1515 #define EXT4_FT_CHRDEV 3
1516 #define EXT4_FT_BLKDEV 4
1517 #define EXT4_FT_FIFO 5
1518 #define EXT4_FT_SOCK 6
1519 #define EXT4_FT_SYMLINK 7
1521 #define EXT4_FT_MAX 8
1524 * EXT4_DIR_PAD defines the directory entries boundaries
1526 * NOTE: It must be a multiple of 4
1528 #define EXT4_DIR_PAD 4
1529 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1530 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1532 #define EXT4_MAX_REC_LEN ((1<<16)-1)
1535 * If we ever get support for fs block sizes > page_size, we'll need
1536 * to remove the #if statements in the next two functions...
1538 static inline unsigned int
1539 ext4_rec_len_from_disk(__le16 dlen
, unsigned blocksize
)
1541 unsigned len
= le16_to_cpu(dlen
);
1543 #if (PAGE_CACHE_SIZE >= 65536)
1544 if (len
== EXT4_MAX_REC_LEN
|| len
== 0)
1546 return (len
& 65532) | ((len
& 3) << 16);
1552 static inline __le16
ext4_rec_len_to_disk(unsigned len
, unsigned blocksize
)
1554 if ((len
> blocksize
) || (blocksize
> (1 << 18)) || (len
& 3))
1556 #if (PAGE_CACHE_SIZE >= 65536)
1558 return cpu_to_le16(len
);
1559 if (len
== blocksize
) {
1560 if (blocksize
== 65536)
1561 return cpu_to_le16(EXT4_MAX_REC_LEN
);
1563 return cpu_to_le16(0);
1565 return cpu_to_le16((len
& 65532) | ((len
>> 16) & 3));
1567 return cpu_to_le16(len
);
1572 * Hash Tree Directory indexing
1573 * (c) Daniel Phillips, 2001
1576 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1577 EXT4_FEATURE_COMPAT_DIR_INDEX) && \
1578 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
1579 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1580 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1582 /* Legal values for the dx_root hash_version field: */
1584 #define DX_HASH_LEGACY 0
1585 #define DX_HASH_HALF_MD4 1
1586 #define DX_HASH_TEA 2
1587 #define DX_HASH_LEGACY_UNSIGNED 3
1588 #define DX_HASH_HALF_MD4_UNSIGNED 4
1589 #define DX_HASH_TEA_UNSIGNED 5
1593 /* hash info structure used by the directory hash */
1602 #define EXT4_HTREE_EOF 0x7fffffff
1605 * Control parameters used by ext4_htree_next_block
1607 #define HASH_NB_ALWAYS 1
1611 * Describe an inode's exact location on disk and in memory
1615 struct buffer_head
*bh
;
1616 unsigned long offset
;
1617 ext4_group_t block_group
;
1620 static inline struct ext4_inode
*ext4_raw_inode(struct ext4_iloc
*iloc
)
1622 return (struct ext4_inode
*) (iloc
->bh
->b_data
+ iloc
->offset
);
1626 * This structure is stuffed into the struct file's private_data field
1627 * for directories. It is where we put information so that we can do
1628 * readdir operations in hash tree order.
1630 struct dir_private_info
{
1631 struct rb_root root
;
1632 struct rb_node
*curr_node
;
1633 struct fname
*extra_fname
;
1636 __u32 curr_minor_hash
;
1640 /* calculate the first block number of the group */
1641 static inline ext4_fsblk_t
1642 ext4_group_first_block_no(struct super_block
*sb
, ext4_group_t group_no
)
1644 return group_no
* (ext4_fsblk_t
)EXT4_BLOCKS_PER_GROUP(sb
) +
1645 le32_to_cpu(EXT4_SB(sb
)->s_es
->s_first_data_block
);
1649 * Special error return code only used by dx_probe() and its callers.
1651 #define ERR_BAD_DX_DIR -75000
1653 void ext4_get_group_no_and_offset(struct super_block
*sb
, ext4_fsblk_t blocknr
,
1654 ext4_group_t
*blockgrpp
, ext4_grpblk_t
*offsetp
);
1657 * Timeout and state flag for lazy initialization inode thread.
1659 #define EXT4_DEF_LI_WAIT_MULT 10
1660 #define EXT4_DEF_LI_MAX_START_DELAY 5
1661 #define EXT4_LAZYINIT_QUIT 0x0001
1662 #define EXT4_LAZYINIT_RUNNING 0x0002
1665 * Lazy inode table initialization info
1667 struct ext4_lazy_init
{
1668 unsigned long li_state
;
1669 struct list_head li_request_list
;
1670 struct mutex li_list_mtx
;
1673 struct ext4_li_request
{
1674 struct super_block
*lr_super
;
1675 struct ext4_sb_info
*lr_sbi
;
1676 ext4_group_t lr_next_group
;
1677 struct list_head lr_request
;
1678 unsigned long lr_next_sched
;
1679 unsigned long lr_timeout
;
1682 struct ext4_features
{
1683 struct kobject f_kobj
;
1684 struct completion f_kobj_unregister
;
1688 * This structure will be used for multiple mount protection. It will be
1689 * written into the block number saved in the s_mmp_block field in the
1690 * superblock. Programs that check MMP should assume that if
1691 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
1692 * to use the filesystem, regardless of how old the timestamp is.
1694 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
1695 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
1696 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
1697 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
1700 __le32 mmp_magic
; /* Magic number for MMP */
1701 __le32 mmp_seq
; /* Sequence no. updated periodically */
1704 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
1705 * purposes and do not affect the correctness of the algorithm
1707 __le64 mmp_time
; /* Time last updated */
1708 char mmp_nodename
[64]; /* Node which last updated MMP block */
1709 char mmp_bdevname
[32]; /* Bdev which last updated MMP block */
1712 * mmp_check_interval is used to verify if the MMP block has been
1713 * updated on the block device. The value is updated based on the
1714 * maximum time to write the MMP block during an update cycle.
1716 __le16 mmp_check_interval
;
1719 __le32 mmp_pad2
[227];
1722 /* arguments passed to the mmp thread */
1724 struct buffer_head
*bh
; /* bh from initial read_mmp_block() */
1725 struct super_block
*sb
; /* super block of the fs */
1729 * Check interval multiplier
1730 * The MMP block is written every update interval and initially checked every
1731 * update interval x the multiplier (the value is then adapted based on the
1732 * write latency). The reason is that writes can be delayed under load and we
1733 * don't want readers to incorrectly assume that the filesystem is no longer
1736 #define EXT4_MMP_CHECK_MULT 2UL
1739 * Minimum interval for MMP checking in seconds.
1741 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
1744 * Maximum interval for MMP checking in seconds.
1746 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
1749 * Function prototypes
1753 * Ok, these declarations are also in <linux/kernel.h> but none of the
1754 * ext4 source programs needs to include it so they are duplicated here.
1756 # define NORET_TYPE /**/
1757 # define ATTRIB_NORET __attribute__((noreturn))
1758 # define NORET_AND noreturn,
1761 extern unsigned int ext4_count_free(struct buffer_head
*, unsigned);
1764 extern unsigned int ext4_block_group(struct super_block
*sb
,
1765 ext4_fsblk_t blocknr
);
1766 extern ext4_grpblk_t
ext4_block_group_offset(struct super_block
*sb
,
1767 ext4_fsblk_t blocknr
);
1768 extern int ext4_bg_has_super(struct super_block
*sb
, ext4_group_t group
);
1769 extern unsigned long ext4_bg_num_gdb(struct super_block
*sb
,
1770 ext4_group_t group
);
1771 extern ext4_fsblk_t
ext4_new_meta_blocks(handle_t
*handle
, struct inode
*inode
,
1774 unsigned long *count
,
1776 extern int ext4_claim_free_blocks(struct ext4_sb_info
*sbi
,
1777 s64 nblocks
, unsigned int flags
);
1778 extern ext4_fsblk_t
ext4_count_free_blocks(struct super_block
*);
1779 extern void ext4_check_blocks_bitmap(struct super_block
*);
1780 extern struct ext4_group_desc
* ext4_get_group_desc(struct super_block
* sb
,
1781 ext4_group_t block_group
,
1782 struct buffer_head
** bh
);
1783 extern int ext4_should_retry_alloc(struct super_block
*sb
, int *retries
);
1784 struct buffer_head
*ext4_read_block_bitmap(struct super_block
*sb
,
1785 ext4_group_t block_group
);
1786 extern void ext4_init_block_bitmap(struct super_block
*sb
,
1787 struct buffer_head
*bh
,
1789 struct ext4_group_desc
*desc
);
1790 extern unsigned ext4_free_blocks_after_init(struct super_block
*sb
,
1791 ext4_group_t block_group
,
1792 struct ext4_group_desc
*gdp
);
1793 extern unsigned ext4_num_base_meta_clusters(struct super_block
*sb
,
1794 ext4_group_t block_group
);
1795 extern unsigned ext4_num_overhead_clusters(struct super_block
*sb
,
1796 ext4_group_t block_group
,
1797 struct ext4_group_desc
*gdp
);
1798 ext4_fsblk_t
ext4_inode_to_goal_block(struct inode
*);
1801 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode
*,
1803 struct ext4_dir_entry_2
*,
1804 struct buffer_head
*, unsigned int);
1805 #define ext4_check_dir_entry(dir, filp, de, bh, offset) \
1806 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
1807 (de), (bh), (offset)))
1808 extern int ext4_htree_store_dirent(struct file
*dir_file
, __u32 hash
,
1810 struct ext4_dir_entry_2
*dirent
);
1811 extern void ext4_htree_free_dir_info(struct dir_private_info
*p
);
1814 extern int ext4_sync_file(struct file
*, loff_t
, loff_t
, int);
1815 extern int ext4_flush_completed_IO(struct inode
*);
1818 extern int ext4fs_dirhash(const char *name
, int len
, struct
1819 dx_hash_info
*hinfo
);
1822 extern struct inode
*ext4_new_inode(handle_t
*, struct inode
*, int,
1823 const struct qstr
*qstr
, __u32 goal
);
1824 extern void ext4_free_inode(handle_t
*, struct inode
*);
1825 extern struct inode
* ext4_orphan_get(struct super_block
*, unsigned long);
1826 extern unsigned long ext4_count_free_inodes(struct super_block
*);
1827 extern unsigned long ext4_count_dirs(struct super_block
*);
1828 extern void ext4_check_inodes_bitmap(struct super_block
*);
1829 extern void ext4_mark_bitmap_end(int start_bit
, int end_bit
, char *bitmap
);
1830 extern int ext4_init_inode_table(struct super_block
*sb
,
1831 ext4_group_t group
, int barrier
);
1834 extern long ext4_mb_stats
;
1835 extern long ext4_mb_max_to_scan
;
1836 extern int ext4_mb_init(struct super_block
*, int);
1837 extern int ext4_mb_release(struct super_block
*);
1838 extern ext4_fsblk_t
ext4_mb_new_blocks(handle_t
*,
1839 struct ext4_allocation_request
*, int *);
1840 extern int ext4_mb_reserve_blocks(struct super_block
*, int);
1841 extern void ext4_discard_preallocations(struct inode
*);
1842 extern int __init
ext4_init_mballoc(void);
1843 extern void ext4_exit_mballoc(void);
1844 extern void ext4_free_blocks(handle_t
*handle
, struct inode
*inode
,
1845 struct buffer_head
*bh
, ext4_fsblk_t block
,
1846 unsigned long count
, int flags
);
1847 extern int ext4_mb_add_groupinfo(struct super_block
*sb
,
1848 ext4_group_t i
, struct ext4_group_desc
*desc
);
1849 extern int ext4_group_add_blocks(handle_t
*handle
, struct super_block
*sb
,
1850 ext4_fsblk_t block
, unsigned long count
);
1851 extern int ext4_trim_fs(struct super_block
*, struct fstrim_range
*);
1854 struct buffer_head
*ext4_getblk(handle_t
*, struct inode
*,
1855 ext4_lblk_t
, int, int *);
1856 struct buffer_head
*ext4_bread(handle_t
*, struct inode
*,
1857 ext4_lblk_t
, int, int *);
1858 int ext4_get_block(struct inode
*inode
, sector_t iblock
,
1859 struct buffer_head
*bh_result
, int create
);
1861 extern struct inode
*ext4_iget(struct super_block
*, unsigned long);
1862 extern int ext4_write_inode(struct inode
*, struct writeback_control
*);
1863 extern int ext4_setattr(struct dentry
*, struct iattr
*);
1864 extern int ext4_getattr(struct vfsmount
*mnt
, struct dentry
*dentry
,
1865 struct kstat
*stat
);
1866 extern void ext4_evict_inode(struct inode
*);
1867 extern void ext4_clear_inode(struct inode
*);
1868 extern int ext4_sync_inode(handle_t
*, struct inode
*);
1869 extern void ext4_dirty_inode(struct inode
*, int);
1870 extern int ext4_change_inode_journal_flag(struct inode
*, int);
1871 extern int ext4_get_inode_loc(struct inode
*, struct ext4_iloc
*);
1872 extern int ext4_can_truncate(struct inode
*inode
);
1873 extern void ext4_truncate(struct inode
*);
1874 extern int ext4_punch_hole(struct file
*file
, loff_t offset
, loff_t length
);
1875 extern int ext4_truncate_restart_trans(handle_t
*, struct inode
*, int nblocks
);
1876 extern void ext4_set_inode_flags(struct inode
*);
1877 extern void ext4_get_inode_flags(struct ext4_inode_info
*);
1878 extern int ext4_alloc_da_blocks(struct inode
*inode
);
1879 extern void ext4_set_aops(struct inode
*inode
);
1880 extern int ext4_writepage_trans_blocks(struct inode
*);
1881 extern int ext4_chunk_trans_blocks(struct inode
*, int nrblocks
);
1882 extern int ext4_block_truncate_page(handle_t
*handle
,
1883 struct address_space
*mapping
, loff_t from
);
1884 extern int ext4_block_zero_page_range(handle_t
*handle
,
1885 struct address_space
*mapping
, loff_t from
, loff_t length
);
1886 extern int ext4_discard_partial_page_buffers(handle_t
*handle
,
1887 struct address_space
*mapping
, loff_t from
,
1888 loff_t length
, int flags
);
1889 extern int ext4_discard_partial_page_buffers_no_lock(handle_t
*handle
,
1890 struct inode
*inode
, struct page
*page
, loff_t from
,
1891 loff_t length
, int flags
);
1892 extern int ext4_page_mkwrite(struct vm_area_struct
*vma
, struct vm_fault
*vmf
);
1893 extern qsize_t
*ext4_get_reserved_space(struct inode
*inode
);
1894 extern void ext4_da_update_reserve_space(struct inode
*inode
,
1895 int used
, int quota_claim
);
1896 extern int ext4_da_reserve_space(struct inode
*inode
, ext4_lblk_t lblock
);
1899 extern int ext4_ind_map_blocks(handle_t
*handle
, struct inode
*inode
,
1900 struct ext4_map_blocks
*map
, int flags
);
1901 extern ssize_t
ext4_ind_direct_IO(int rw
, struct kiocb
*iocb
,
1902 const struct iovec
*iov
, loff_t offset
,
1903 unsigned long nr_segs
);
1904 extern int ext4_ind_calc_metadata_amount(struct inode
*inode
, sector_t lblock
);
1905 extern int ext4_ind_trans_blocks(struct inode
*inode
, int nrblocks
, int chunk
);
1906 extern void ext4_ind_truncate(struct inode
*inode
);
1909 extern long ext4_ioctl(struct file
*, unsigned int, unsigned long);
1910 extern long ext4_compat_ioctl(struct file
*, unsigned int, unsigned long);
1913 extern int ext4_ext_migrate(struct inode
*);
1916 extern int ext4_orphan_add(handle_t
*, struct inode
*);
1917 extern int ext4_orphan_del(handle_t
*, struct inode
*);
1918 extern int ext4_htree_fill_tree(struct file
*dir_file
, __u32 start_hash
,
1919 __u32 start_minor_hash
, __u32
*next_hash
);
1922 extern int ext4_group_add(struct super_block
*sb
,
1923 struct ext4_new_group_data
*input
);
1924 extern int ext4_group_extend(struct super_block
*sb
,
1925 struct ext4_super_block
*es
,
1926 ext4_fsblk_t n_blocks_count
);
1929 extern void *ext4_kvmalloc(size_t size
, gfp_t flags
);
1930 extern void *ext4_kvzalloc(size_t size
, gfp_t flags
);
1931 extern void ext4_kvfree(void *ptr
);
1932 extern void __ext4_error(struct super_block
*, const char *, unsigned int,
1934 __attribute__ ((format (printf
, 4, 5)));
1935 #define ext4_error(sb, message...) __ext4_error(sb, __func__, \
1936 __LINE__, ## message)
1937 extern void ext4_error_inode(struct inode
*, const char *, unsigned int,
1938 ext4_fsblk_t
, const char *, ...)
1939 __attribute__ ((format (printf
, 5, 6)));
1940 extern void ext4_error_file(struct file
*, const char *, unsigned int,
1941 ext4_fsblk_t
, const char *, ...)
1942 __attribute__ ((format (printf
, 5, 6)));
1943 extern void __ext4_std_error(struct super_block
*, const char *,
1945 extern void __ext4_abort(struct super_block
*, const char *, unsigned int,
1947 __attribute__ ((format (printf
, 4, 5)));
1948 #define ext4_abort(sb, message...) __ext4_abort(sb, __func__, \
1949 __LINE__, ## message)
1950 extern void __ext4_warning(struct super_block
*, const char *, unsigned int,
1952 __attribute__ ((format (printf
, 4, 5)));
1953 #define ext4_warning(sb, message...) __ext4_warning(sb, __func__, \
1954 __LINE__, ## message)
1955 extern void ext4_msg(struct super_block
*, const char *, const char *, ...)
1956 __attribute__ ((format (printf
, 3, 4)));
1957 extern void __dump_mmp_msg(struct super_block
*, struct mmp_struct
*mmp
,
1958 const char *, unsigned int, const char *);
1959 #define dump_mmp_msg(sb, mmp, msg) __dump_mmp_msg(sb, mmp, __func__, \
1961 extern void __ext4_grp_locked_error(const char *, unsigned int, \
1962 struct super_block
*, ext4_group_t
, \
1963 unsigned long, ext4_fsblk_t
, \
1965 __attribute__ ((format (printf
, 7, 8)));
1966 #define ext4_grp_locked_error(sb, grp, message...) \
1967 __ext4_grp_locked_error(__func__, __LINE__, (sb), (grp), ## message)
1968 extern void ext4_update_dynamic_rev(struct super_block
*sb
);
1969 extern int ext4_update_compat_feature(handle_t
*handle
, struct super_block
*sb
,
1971 extern int ext4_update_rocompat_feature(handle_t
*handle
,
1972 struct super_block
*sb
, __u32 rocompat
);
1973 extern int ext4_update_incompat_feature(handle_t
*handle
,
1974 struct super_block
*sb
, __u32 incompat
);
1975 extern ext4_fsblk_t
ext4_block_bitmap(struct super_block
*sb
,
1976 struct ext4_group_desc
*bg
);
1977 extern ext4_fsblk_t
ext4_inode_bitmap(struct super_block
*sb
,
1978 struct ext4_group_desc
*bg
);
1979 extern ext4_fsblk_t
ext4_inode_table(struct super_block
*sb
,
1980 struct ext4_group_desc
*bg
);
1981 extern __u32
ext4_free_group_clusters(struct super_block
*sb
,
1982 struct ext4_group_desc
*bg
);
1983 extern __u32
ext4_free_inodes_count(struct super_block
*sb
,
1984 struct ext4_group_desc
*bg
);
1985 extern __u32
ext4_used_dirs_count(struct super_block
*sb
,
1986 struct ext4_group_desc
*bg
);
1987 extern __u32
ext4_itable_unused_count(struct super_block
*sb
,
1988 struct ext4_group_desc
*bg
);
1989 extern void ext4_block_bitmap_set(struct super_block
*sb
,
1990 struct ext4_group_desc
*bg
, ext4_fsblk_t blk
);
1991 extern void ext4_inode_bitmap_set(struct super_block
*sb
,
1992 struct ext4_group_desc
*bg
, ext4_fsblk_t blk
);
1993 extern void ext4_inode_table_set(struct super_block
*sb
,
1994 struct ext4_group_desc
*bg
, ext4_fsblk_t blk
);
1995 extern void ext4_free_group_clusters_set(struct super_block
*sb
,
1996 struct ext4_group_desc
*bg
,
1998 extern void ext4_free_inodes_set(struct super_block
*sb
,
1999 struct ext4_group_desc
*bg
, __u32 count
);
2000 extern void ext4_used_dirs_set(struct super_block
*sb
,
2001 struct ext4_group_desc
*bg
, __u32 count
);
2002 extern void ext4_itable_unused_set(struct super_block
*sb
,
2003 struct ext4_group_desc
*bg
, __u32 count
);
2004 extern __le16
ext4_group_desc_csum(struct ext4_sb_info
*sbi
, __u32 group
,
2005 struct ext4_group_desc
*gdp
);
2006 extern int ext4_group_desc_csum_verify(struct ext4_sb_info
*sbi
, __u32 group
,
2007 struct ext4_group_desc
*gdp
);
2009 static inline ext4_fsblk_t
ext4_blocks_count(struct ext4_super_block
*es
)
2011 return ((ext4_fsblk_t
)le32_to_cpu(es
->s_blocks_count_hi
) << 32) |
2012 le32_to_cpu(es
->s_blocks_count_lo
);
2015 static inline ext4_fsblk_t
ext4_r_blocks_count(struct ext4_super_block
*es
)
2017 return ((ext4_fsblk_t
)le32_to_cpu(es
->s_r_blocks_count_hi
) << 32) |
2018 le32_to_cpu(es
->s_r_blocks_count_lo
);
2021 static inline ext4_fsblk_t
ext4_free_blocks_count(struct ext4_super_block
*es
)
2023 return ((ext4_fsblk_t
)le32_to_cpu(es
->s_free_blocks_count_hi
) << 32) |
2024 le32_to_cpu(es
->s_free_blocks_count_lo
);
2027 static inline void ext4_blocks_count_set(struct ext4_super_block
*es
,
2030 es
->s_blocks_count_lo
= cpu_to_le32((u32
)blk
);
2031 es
->s_blocks_count_hi
= cpu_to_le32(blk
>> 32);
2034 static inline void ext4_free_blocks_count_set(struct ext4_super_block
*es
,
2037 es
->s_free_blocks_count_lo
= cpu_to_le32((u32
)blk
);
2038 es
->s_free_blocks_count_hi
= cpu_to_le32(blk
>> 32);
2041 static inline void ext4_r_blocks_count_set(struct ext4_super_block
*es
,
2044 es
->s_r_blocks_count_lo
= cpu_to_le32((u32
)blk
);
2045 es
->s_r_blocks_count_hi
= cpu_to_le32(blk
>> 32);
2048 static inline loff_t
ext4_isize(struct ext4_inode
*raw_inode
)
2050 if (S_ISREG(le16_to_cpu(raw_inode
->i_mode
)))
2051 return ((loff_t
)le32_to_cpu(raw_inode
->i_size_high
) << 32) |
2052 le32_to_cpu(raw_inode
->i_size_lo
);
2054 return (loff_t
) le32_to_cpu(raw_inode
->i_size_lo
);
2057 static inline void ext4_isize_set(struct ext4_inode
*raw_inode
, loff_t i_size
)
2059 raw_inode
->i_size_lo
= cpu_to_le32(i_size
);
2060 raw_inode
->i_size_high
= cpu_to_le32(i_size
>> 32);
2064 struct ext4_group_info
*ext4_get_group_info(struct super_block
*sb
,
2067 struct ext4_group_info
***grp_info
;
2068 long indexv
, indexh
;
2069 grp_info
= EXT4_SB(sb
)->s_group_info
;
2070 indexv
= group
>> (EXT4_DESC_PER_BLOCK_BITS(sb
));
2071 indexh
= group
& ((EXT4_DESC_PER_BLOCK(sb
)) - 1);
2072 return grp_info
[indexv
][indexh
];
2076 * Reading s_groups_count requires using smp_rmb() afterwards. See
2077 * the locking protocol documented in the comments of ext4_group_add()
2080 static inline ext4_group_t
ext4_get_groups_count(struct super_block
*sb
)
2082 ext4_group_t ngroups
= EXT4_SB(sb
)->s_groups_count
;
2088 static inline ext4_group_t
ext4_flex_group(struct ext4_sb_info
*sbi
,
2089 ext4_group_t block_group
)
2091 return block_group
>> sbi
->s_log_groups_per_flex
;
2094 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info
*sbi
)
2096 return 1 << sbi
->s_log_groups_per_flex
;
2099 #define ext4_std_error(sb, errno) \
2102 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
2106 /* Each CPU can accumulate percpu_counter_batch blocks in their local
2107 * counters. So we need to make sure we have free blocks more
2108 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
2110 #define EXT4_FREEBLOCKS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2112 #define EXT4_FREEBLOCKS_WATERMARK 0
2115 static inline void ext4_update_i_disksize(struct inode
*inode
, loff_t newsize
)
2118 * XXX: replace with spinlock if seen contended -bzzz
2120 down_write(&EXT4_I(inode
)->i_data_sem
);
2121 if (newsize
> EXT4_I(inode
)->i_disksize
)
2122 EXT4_I(inode
)->i_disksize
= newsize
;
2123 up_write(&EXT4_I(inode
)->i_data_sem
);
2127 struct ext4_group_info
{
2128 unsigned long bb_state
;
2129 struct rb_root bb_free_root
;
2130 ext4_grpblk_t bb_first_free
; /* first free block */
2131 ext4_grpblk_t bb_free
; /* total free blocks */
2132 ext4_grpblk_t bb_fragments
; /* nr of freespace fragments */
2133 ext4_grpblk_t bb_largest_free_order
;/* order of largest frag in BG */
2134 struct list_head bb_prealloc_list
;
2138 struct rw_semaphore alloc_sem
;
2139 ext4_grpblk_t bb_counters
[]; /* Nr of free power-of-two-block
2140 * regions, index is order.
2141 * bb_counters[3] = 5 means
2142 * 5 free 8-block regions. */
2145 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
2146 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
2148 #define EXT4_MB_GRP_NEED_INIT(grp) \
2149 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2151 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
2152 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2153 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
2154 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2155 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
2156 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2158 #define EXT4_MAX_CONTENTION 8
2159 #define EXT4_CONTENTION_THRESHOLD 2
2161 static inline spinlock_t
*ext4_group_lock_ptr(struct super_block
*sb
,
2164 return bgl_lock_ptr(EXT4_SB(sb
)->s_blockgroup_lock
, group
);
2168 * Returns true if the filesystem is busy enough that attempts to
2169 * access the block group locks has run into contention.
2171 static inline int ext4_fs_is_busy(struct ext4_sb_info
*sbi
)
2173 return (atomic_read(&sbi
->s_lock_busy
) > EXT4_CONTENTION_THRESHOLD
);
2176 static inline void ext4_lock_group(struct super_block
*sb
, ext4_group_t group
)
2178 spinlock_t
*lock
= ext4_group_lock_ptr(sb
, group
);
2179 if (spin_trylock(lock
))
2181 * We're able to grab the lock right away, so drop the
2182 * lock contention counter.
2184 atomic_add_unless(&EXT4_SB(sb
)->s_lock_busy
, -1, 0);
2187 * The lock is busy, so bump the contention counter,
2188 * and then wait on the spin lock.
2190 atomic_add_unless(&EXT4_SB(sb
)->s_lock_busy
, 1,
2191 EXT4_MAX_CONTENTION
);
2196 static inline void ext4_unlock_group(struct super_block
*sb
,
2199 spin_unlock(ext4_group_lock_ptr(sb
, group
));
2202 static inline void ext4_mark_super_dirty(struct super_block
*sb
)
2204 if (EXT4_SB(sb
)->s_journal
== NULL
)
2209 * Block validity checking
2211 #define ext4_check_indirect_blockref(inode, bh) \
2212 ext4_check_blockref(__func__, __LINE__, inode, \
2213 (__le32 *)(bh)->b_data, \
2214 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2216 #define ext4_ind_check_inode(inode) \
2217 ext4_check_blockref(__func__, __LINE__, inode, \
2218 EXT4_I(inode)->i_data, \
2222 * Inodes and files operations
2226 extern const struct file_operations ext4_dir_operations
;
2229 extern const struct inode_operations ext4_file_inode_operations
;
2230 extern const struct file_operations ext4_file_operations
;
2231 extern loff_t
ext4_llseek(struct file
*file
, loff_t offset
, int origin
);
2234 extern const struct inode_operations ext4_dir_inode_operations
;
2235 extern const struct inode_operations ext4_special_inode_operations
;
2236 extern struct dentry
*ext4_get_parent(struct dentry
*child
);
2239 extern const struct inode_operations ext4_symlink_inode_operations
;
2240 extern const struct inode_operations ext4_fast_symlink_inode_operations
;
2242 /* block_validity */
2243 extern void ext4_release_system_zone(struct super_block
*sb
);
2244 extern int ext4_setup_system_zone(struct super_block
*sb
);
2245 extern int __init
ext4_init_system_zone(void);
2246 extern void ext4_exit_system_zone(void);
2247 extern int ext4_data_block_valid(struct ext4_sb_info
*sbi
,
2248 ext4_fsblk_t start_blk
,
2249 unsigned int count
);
2250 extern int ext4_check_blockref(const char *, unsigned int,
2251 struct inode
*, __le32
*, unsigned int);
2254 extern int ext4_ext_tree_init(handle_t
*handle
, struct inode
*);
2255 extern int ext4_ext_writepage_trans_blocks(struct inode
*, int);
2256 extern int ext4_ext_index_trans_blocks(struct inode
*inode
, int nrblocks
,
2258 extern int ext4_ext_map_blocks(handle_t
*handle
, struct inode
*inode
,
2259 struct ext4_map_blocks
*map
, int flags
);
2260 extern void ext4_ext_truncate(struct inode
*);
2261 extern int ext4_ext_punch_hole(struct file
*file
, loff_t offset
,
2263 extern void ext4_ext_init(struct super_block
*);
2264 extern void ext4_ext_release(struct super_block
*);
2265 extern long ext4_fallocate(struct file
*file
, int mode
, loff_t offset
,
2267 extern int ext4_convert_unwritten_extents(struct inode
*inode
, loff_t offset
,
2269 extern int ext4_map_blocks(handle_t
*handle
, struct inode
*inode
,
2270 struct ext4_map_blocks
*map
, int flags
);
2271 extern int ext4_fiemap(struct inode
*inode
, struct fiemap_extent_info
*fieinfo
,
2272 __u64 start
, __u64 len
);
2274 extern int ext4_move_extents(struct file
*o_filp
, struct file
*d_filp
,
2275 __u64 start_orig
, __u64 start_donor
,
2276 __u64 len
, __u64
*moved_len
);
2279 extern int __init
ext4_init_pageio(void);
2280 extern void ext4_exit_pageio(void);
2281 extern void ext4_ioend_wait(struct inode
*);
2282 extern void ext4_free_io_end(ext4_io_end_t
*io
);
2283 extern ext4_io_end_t
*ext4_init_io_end(struct inode
*inode
, gfp_t flags
);
2284 extern int ext4_end_io_nolock(ext4_io_end_t
*io
);
2285 extern void ext4_io_submit(struct ext4_io_submit
*io
);
2286 extern int ext4_bio_write_page(struct ext4_io_submit
*io
,
2289 struct writeback_control
*wbc
);
2292 extern int ext4_multi_mount_protect(struct super_block
*, ext4_fsblk_t
);
2294 /* BH_Uninit flag: blocks are allocated but uninitialized on disk */
2295 enum ext4_state_bits
{
2296 BH_Uninit
/* blocks are allocated but uninitialized on disk */
2297 = BH_JBDPrivateStart
,
2298 BH_AllocFromCluster
, /* allocated blocks were part of already
2299 * allocated cluster. Note that this flag will
2300 * never, ever appear in a buffer_head's state
2301 * flag. See EXT4_MAP_FROM_CLUSTER to see where
2305 BUFFER_FNS(Uninit
, uninit
)
2306 TAS_BUFFER_FNS(Uninit
, uninit
)
2309 * Add new method to test wether block and inode bitmaps are properly
2310 * initialized. With uninit_bg reading the block from disk is not enough
2311 * to mark the bitmap uptodate. We need to also zero-out the bitmap
2313 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
2315 static inline int bitmap_uptodate(struct buffer_head
*bh
)
2317 return (buffer_uptodate(bh
) &&
2318 test_bit(BH_BITMAP_UPTODATE
, &(bh
)->b_state
));
2320 static inline void set_bitmap_uptodate(struct buffer_head
*bh
)
2322 set_bit(BH_BITMAP_UPTODATE
, &(bh
)->b_state
);
2325 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
2327 /* For ioend & aio unwritten conversion wait queues */
2328 #define EXT4_WQ_HASH_SZ 37
2329 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
2331 #define ext4_aio_mutex(v) (&ext4__aio_mutex[((unsigned long)(v)) %\
2333 extern wait_queue_head_t ext4__ioend_wq
[EXT4_WQ_HASH_SZ
];
2334 extern struct mutex ext4__aio_mutex
[EXT4_WQ_HASH_SZ
];
2336 #define EXT4_RESIZING 0
2337 extern int ext4_resize_begin(struct super_block
*sb
);
2338 extern void ext4_resize_end(struct super_block
*sb
);
2340 #endif /* __KERNEL__ */
2342 #endif /* _EXT4_H */