4 * Copyright (C) 1992, 1993, 1994, 1995
5 * Remy Card (card@masi.ibp.fr)
6 * Laboratoire MASI - Institut Blaise Pascal
7 * Universite Pierre et Marie Curie (Paris VI)
11 * linux/include/linux/minix_fs.h
13 * Copyright (C) 1991, 1992 Linus Torvalds
19 #include <linux/types.h>
20 #include <linux/blkdev.h>
21 #include <linux/magic.h>
22 #include <linux/jbd2.h>
23 #include <linux/quota.h>
24 #include <linux/rwsem.h>
25 #include <linux/rbtree.h>
26 #include <linux/seqlock.h>
27 #include <linux/mutex.h>
28 #include <linux/timer.h>
29 #include <linux/wait.h>
30 #include <linux/blockgroup_lock.h>
31 #include <linux/percpu_counter.h>
33 #include <linux/compat.h>
37 * The fourth extended filesystem constants/structures
41 * Define EXT4FS_DEBUG to produce debug messages
49 #define ext4_debug(f, a...) \
51 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
52 __FILE__, __LINE__, __func__); \
53 printk(KERN_DEBUG f, ## a); \
56 #define ext4_debug(f, a...) do {} while (0)
59 #define EXT4_ERROR_INODE(inode, fmt, a...) \
60 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
62 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
63 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
65 #define EXT4_ERROR_FILE(file, block, fmt, a...) \
66 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
68 /* data type for block offset of block group */
69 typedef int ext4_grpblk_t
;
71 /* data type for filesystem-wide blocks number */
72 typedef unsigned long long ext4_fsblk_t
;
74 /* data type for file logical block number */
75 typedef __u32 ext4_lblk_t
;
77 /* data type for block group number */
78 typedef unsigned int ext4_group_t
;
81 * Flags used in mballoc's allocation_context flags field.
83 * Also used to show what's going on for debugging purposes when the
84 * flag field is exported via the traceport interface
87 /* prefer goal again. length */
88 #define EXT4_MB_HINT_MERGE 0x0001
89 /* blocks already reserved */
90 #define EXT4_MB_HINT_RESERVED 0x0002
91 /* metadata is being allocated */
92 #define EXT4_MB_HINT_METADATA 0x0004
93 /* first blocks in the file */
94 #define EXT4_MB_HINT_FIRST 0x0008
95 /* search for the best chunk */
96 #define EXT4_MB_HINT_BEST 0x0010
97 /* data is being allocated */
98 #define EXT4_MB_HINT_DATA 0x0020
99 /* don't preallocate (for tails) */
100 #define EXT4_MB_HINT_NOPREALLOC 0x0040
101 /* allocate for locality group */
102 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080
103 /* allocate goal blocks or none */
104 #define EXT4_MB_HINT_GOAL_ONLY 0x0100
105 /* goal is meaningful */
106 #define EXT4_MB_HINT_TRY_GOAL 0x0200
107 /* blocks already pre-reserved by delayed allocation */
108 #define EXT4_MB_DELALLOC_RESERVED 0x0400
109 /* We are doing stream allocation */
110 #define EXT4_MB_STREAM_ALLOC 0x0800
111 /* Use reserved root blocks if needed */
112 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000
114 struct ext4_allocation_request
{
115 /* target inode for block we're allocating */
117 /* how many blocks we want to allocate */
119 /* logical block in target inode */
121 /* the closest logical allocated block to the left */
123 /* the closest logical allocated block to the right */
125 /* phys. target (a hint) */
127 /* phys. block for the closest logical allocated block to the left */
129 /* phys. block for the closest logical allocated block to the right */
131 /* flags. see above EXT4_MB_HINT_* */
136 * Logical to physical block mapping, used by ext4_map_blocks()
138 * This structure is used to pass requests into ext4_map_blocks() as
139 * well as to store the information returned by ext4_map_blocks(). It
140 * takes less room on the stack than a struct buffer_head.
142 #define EXT4_MAP_NEW (1 << BH_New)
143 #define EXT4_MAP_MAPPED (1 << BH_Mapped)
144 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten)
145 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary)
146 #define EXT4_MAP_UNINIT (1 << BH_Uninit)
147 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
148 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\
151 struct ext4_map_blocks
{
155 unsigned int m_flags
;
159 * For delayed allocation tracking
161 struct mpage_da_data
{
163 sector_t b_blocknr
; /* start block number of extent */
164 size_t b_size
; /* size of extent */
165 unsigned long b_state
; /* state of the extent */
166 unsigned long first_page
, next_page
; /* extent of pages */
167 struct writeback_control
*wbc
;
174 * Flags for ext4_io_end->flags
176 #define EXT4_IO_END_UNWRITTEN 0x0001
177 #define EXT4_IO_END_ERROR 0x0002
178 #define EXT4_IO_END_QUEUED 0x0004
180 struct ext4_io_page
{
185 #define MAX_IO_PAGES 128
187 typedef struct ext4_io_end
{
188 struct list_head list
; /* per-file finished IO list */
189 struct inode
*inode
; /* file being written to */
190 unsigned int flag
; /* unwritten or not */
191 struct page
*page
; /* page struct for buffer write */
192 loff_t offset
; /* offset in the file */
193 ssize_t size
; /* size of the extent */
194 struct work_struct work
; /* data work queue */
195 struct kiocb
*iocb
; /* iocb struct for AIO */
196 int result
; /* error value for AIO */
198 struct ext4_io_page
*pages
[MAX_IO_PAGES
];
201 struct ext4_io_submit
{
204 ext4_io_end_t
*io_end
;
205 struct ext4_io_page
*io_page
;
206 sector_t io_next_block
;
210 * Special inodes numbers
212 #define EXT4_BAD_INO 1 /* Bad blocks inode */
213 #define EXT4_ROOT_INO 2 /* Root inode */
214 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */
215 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
216 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
217 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
218 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
219 #define EXT4_JOURNAL_INO 8 /* Journal inode */
221 /* First non-reserved inode for old ext4 filesystems */
222 #define EXT4_GOOD_OLD_FIRST_INO 11
225 * Maximal count of links to a file
227 #define EXT4_LINK_MAX 65000
230 * Macro-instructions used to manage several block sizes
232 #define EXT4_MIN_BLOCK_SIZE 1024
233 #define EXT4_MAX_BLOCK_SIZE 65536
234 #define EXT4_MIN_BLOCK_LOG_SIZE 10
235 #define EXT4_MAX_BLOCK_LOG_SIZE 16
237 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
239 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
241 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
242 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
243 EXT4_SB(s)->s_cluster_bits)
245 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
246 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
248 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
251 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
252 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
253 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
255 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
256 EXT4_GOOD_OLD_INODE_SIZE : \
258 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
259 EXT4_GOOD_OLD_FIRST_INO : \
262 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
264 /* Translate a block number to a cluster number */
265 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
266 /* Translate a cluster number to a block number */
267 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
268 /* Translate # of blks to # of clusters */
269 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
270 (sbi)->s_cluster_bits)
273 * Structure of a blocks group descriptor
275 struct ext4_group_desc
277 __le32 bg_block_bitmap_lo
; /* Blocks bitmap block */
278 __le32 bg_inode_bitmap_lo
; /* Inodes bitmap block */
279 __le32 bg_inode_table_lo
; /* Inodes table block */
280 __le16 bg_free_blocks_count_lo
;/* Free blocks count */
281 __le16 bg_free_inodes_count_lo
;/* Free inodes count */
282 __le16 bg_used_dirs_count_lo
; /* Directories count */
283 __le16 bg_flags
; /* EXT4_BG_flags (INODE_UNINIT, etc) */
284 __u32 bg_reserved
[2]; /* Likely block/inode bitmap checksum */
285 __le16 bg_itable_unused_lo
; /* Unused inodes count */
286 __le16 bg_checksum
; /* crc16(sb_uuid+group+desc) */
287 __le32 bg_block_bitmap_hi
; /* Blocks bitmap block MSB */
288 __le32 bg_inode_bitmap_hi
; /* Inodes bitmap block MSB */
289 __le32 bg_inode_table_hi
; /* Inodes table block MSB */
290 __le16 bg_free_blocks_count_hi
;/* Free blocks count MSB */
291 __le16 bg_free_inodes_count_hi
;/* Free inodes count MSB */
292 __le16 bg_used_dirs_count_hi
; /* Directories count MSB */
293 __le16 bg_itable_unused_hi
; /* Unused inodes count MSB */
294 __u32 bg_reserved2
[3];
298 * Structure of a flex block group info
302 atomic_t free_inodes
;
303 atomic_t free_clusters
;
307 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
308 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
309 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
312 * Macro-instructions used to manage group descriptors
314 #define EXT4_MIN_DESC_SIZE 32
315 #define EXT4_MIN_DESC_SIZE_64BIT 64
316 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
317 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
319 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
320 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
321 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
322 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
323 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
325 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
326 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
327 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
331 * Constants relative to the data blocks
333 #define EXT4_NDIR_BLOCKS 12
334 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
335 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
336 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
337 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
342 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
343 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
344 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
345 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
346 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
347 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
348 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
349 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
350 /* Reserved for compression usage... */
351 #define EXT4_DIRTY_FL 0x00000100
352 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
353 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
354 #define EXT4_ECOMPR_FL 0x00000800 /* Compression error */
355 /* End compression flags --- maybe not all used */
356 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
357 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
358 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
359 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
360 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
361 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
362 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
363 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
364 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
365 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
366 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
368 #define EXT4_FL_USER_VISIBLE 0x004BDFFF /* User visible flags */
369 #define EXT4_FL_USER_MODIFIABLE 0x004B80FF /* User modifiable flags */
371 /* Flags that should be inherited by new inodes from their parent. */
372 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
373 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
374 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
375 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
377 /* Flags that are appropriate for regular files (all but dir-specific ones). */
378 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
380 /* Flags that are appropriate for non-directories/regular files. */
381 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
383 /* Mask out flags that are inappropriate for the given type of inode. */
384 static inline __u32
ext4_mask_flags(umode_t mode
, __u32 flags
)
388 else if (S_ISREG(mode
))
389 return flags
& EXT4_REG_FLMASK
;
391 return flags
& EXT4_OTHER_FLMASK
;
395 * Inode flags used for atomic set/get
398 EXT4_INODE_SECRM
= 0, /* Secure deletion */
399 EXT4_INODE_UNRM
= 1, /* Undelete */
400 EXT4_INODE_COMPR
= 2, /* Compress file */
401 EXT4_INODE_SYNC
= 3, /* Synchronous updates */
402 EXT4_INODE_IMMUTABLE
= 4, /* Immutable file */
403 EXT4_INODE_APPEND
= 5, /* writes to file may only append */
404 EXT4_INODE_NODUMP
= 6, /* do not dump file */
405 EXT4_INODE_NOATIME
= 7, /* do not update atime */
406 /* Reserved for compression usage... */
407 EXT4_INODE_DIRTY
= 8,
408 EXT4_INODE_COMPRBLK
= 9, /* One or more compressed clusters */
409 EXT4_INODE_NOCOMPR
= 10, /* Don't compress */
410 EXT4_INODE_ECOMPR
= 11, /* Compression error */
411 /* End compression flags --- maybe not all used */
412 EXT4_INODE_INDEX
= 12, /* hash-indexed directory */
413 EXT4_INODE_IMAGIC
= 13, /* AFS directory */
414 EXT4_INODE_JOURNAL_DATA
= 14, /* file data should be journaled */
415 EXT4_INODE_NOTAIL
= 15, /* file tail should not be merged */
416 EXT4_INODE_DIRSYNC
= 16, /* dirsync behaviour (directories only) */
417 EXT4_INODE_TOPDIR
= 17, /* Top of directory hierarchies*/
418 EXT4_INODE_HUGE_FILE
= 18, /* Set to each huge file */
419 EXT4_INODE_EXTENTS
= 19, /* Inode uses extents */
420 EXT4_INODE_EA_INODE
= 21, /* Inode used for large EA */
421 EXT4_INODE_EOFBLOCKS
= 22, /* Blocks allocated beyond EOF */
422 EXT4_INODE_RESERVED
= 31, /* reserved for ext4 lib */
425 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
426 #define CHECK_FLAG_VALUE(FLAG) if (!TEST_FLAG_VALUE(FLAG)) { \
427 printk(KERN_EMERG "EXT4 flag fail: " #FLAG ": %d %d\n", \
428 EXT4_##FLAG##_FL, EXT4_INODE_##FLAG); BUG_ON(1); }
431 * Since it's pretty easy to mix up bit numbers and hex values, and we
432 * can't do a compile-time test for ENUM values, we use a run-time
433 * test to make sure that EXT4_XXX_FL is consistent with respect to
434 * EXT4_INODE_XXX. If all is well the printk and BUG_ON will all drop
435 * out so it won't cost any extra space in the compiled kernel image.
436 * But it's important that these values are the same, since we are
437 * using EXT4_INODE_XXX to test for the flag values, but EXT4_XX_FL
438 * must be consistent with the values of FS_XXX_FL defined in
439 * include/linux/fs.h and the on-disk values found in ext2, ext3, and
440 * ext4 filesystems, and of course the values defined in e2fsprogs.
442 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
444 static inline void ext4_check_flag_values(void)
446 CHECK_FLAG_VALUE(SECRM
);
447 CHECK_FLAG_VALUE(UNRM
);
448 CHECK_FLAG_VALUE(COMPR
);
449 CHECK_FLAG_VALUE(SYNC
);
450 CHECK_FLAG_VALUE(IMMUTABLE
);
451 CHECK_FLAG_VALUE(APPEND
);
452 CHECK_FLAG_VALUE(NODUMP
);
453 CHECK_FLAG_VALUE(NOATIME
);
454 CHECK_FLAG_VALUE(DIRTY
);
455 CHECK_FLAG_VALUE(COMPRBLK
);
456 CHECK_FLAG_VALUE(NOCOMPR
);
457 CHECK_FLAG_VALUE(ECOMPR
);
458 CHECK_FLAG_VALUE(INDEX
);
459 CHECK_FLAG_VALUE(IMAGIC
);
460 CHECK_FLAG_VALUE(JOURNAL_DATA
);
461 CHECK_FLAG_VALUE(NOTAIL
);
462 CHECK_FLAG_VALUE(DIRSYNC
);
463 CHECK_FLAG_VALUE(TOPDIR
);
464 CHECK_FLAG_VALUE(HUGE_FILE
);
465 CHECK_FLAG_VALUE(EXTENTS
);
466 CHECK_FLAG_VALUE(EA_INODE
);
467 CHECK_FLAG_VALUE(EOFBLOCKS
);
468 CHECK_FLAG_VALUE(RESERVED
);
471 /* Used to pass group descriptor data when online resize is done */
472 struct ext4_new_group_input
{
473 __u32 group
; /* Group number for this data */
474 __u64 block_bitmap
; /* Absolute block number of block bitmap */
475 __u64 inode_bitmap
; /* Absolute block number of inode bitmap */
476 __u64 inode_table
; /* Absolute block number of inode table start */
477 __u32 blocks_count
; /* Total number of blocks in this group */
478 __u16 reserved_blocks
; /* Number of reserved blocks in this group */
482 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
483 struct compat_ext4_new_group_input
{
485 compat_u64 block_bitmap
;
486 compat_u64 inode_bitmap
;
487 compat_u64 inode_table
;
494 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
495 struct ext4_new_group_data
{
501 __u16 reserved_blocks
;
503 __u32 free_blocks_count
;
507 * Flags used by ext4_map_blocks()
509 /* Allocate any needed blocks and/or convert an unitialized
510 extent to be an initialized ext4 */
511 #define EXT4_GET_BLOCKS_CREATE 0x0001
512 /* Request the creation of an unitialized extent */
513 #define EXT4_GET_BLOCKS_UNINIT_EXT 0x0002
514 #define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT (EXT4_GET_BLOCKS_UNINIT_EXT|\
515 EXT4_GET_BLOCKS_CREATE)
516 /* Caller is from the delayed allocation writeout path,
517 so set the magic i_delalloc_reserve_flag after taking the
518 inode allocation semaphore for */
519 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
520 /* caller is from the direct IO path, request to creation of an
521 unitialized extents if not allocated, split the uninitialized
522 extent if blocks has been preallocated already*/
523 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
524 #define EXT4_GET_BLOCKS_CONVERT 0x0010
525 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
526 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
527 /* Convert extent to initialized after IO complete */
528 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
529 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
530 /* Punch out blocks of an extent */
531 #define EXT4_GET_BLOCKS_PUNCH_OUT_EXT 0x0020
532 /* Don't normalize allocation size (used for fallocate) */
533 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
536 * Flags used by ext4_free_blocks
538 #define EXT4_FREE_BLOCKS_METADATA 0x0001
539 #define EXT4_FREE_BLOCKS_FORGET 0x0002
540 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
541 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
542 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
543 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
546 * Flags used by ext4_discard_partial_page_buffers
548 #define EXT4_DISCARD_PARTIAL_PG_ZERO_UNMAPPED 0x0001
553 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
554 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
555 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
556 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
557 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
558 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
559 #ifdef CONFIG_JBD2_DEBUG
560 #define EXT4_IOC_WAIT_FOR_READONLY _IOR('f', 99, long)
562 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
563 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
564 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
565 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
566 #define EXT4_IOC_MIGRATE _IO('f', 9)
567 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
568 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
569 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
570 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
572 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
574 * ioctl commands in 32 bit emulation
576 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
577 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
578 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
579 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
580 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
581 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
582 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
583 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
584 #ifdef CONFIG_JBD2_DEBUG
585 #define EXT4_IOC32_WAIT_FOR_READONLY _IOR('f', 99, int)
587 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
588 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
591 /* Max physical block we can address w/o extents */
592 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
595 * Structure of an inode on the disk
598 __le16 i_mode
; /* File mode */
599 __le16 i_uid
; /* Low 16 bits of Owner Uid */
600 __le32 i_size_lo
; /* Size in bytes */
601 __le32 i_atime
; /* Access time */
602 __le32 i_ctime
; /* Inode Change time */
603 __le32 i_mtime
; /* Modification time */
604 __le32 i_dtime
; /* Deletion Time */
605 __le16 i_gid
; /* Low 16 bits of Group Id */
606 __le16 i_links_count
; /* Links count */
607 __le32 i_blocks_lo
; /* Blocks count */
608 __le32 i_flags
; /* File flags */
614 __u32 h_i_translator
;
619 } osd1
; /* OS dependent 1 */
620 __le32 i_block
[EXT4_N_BLOCKS
];/* Pointers to blocks */
621 __le32 i_generation
; /* File version (for NFS) */
622 __le32 i_file_acl_lo
; /* File ACL */
624 __le32 i_obso_faddr
; /* Obsoleted fragment address */
627 __le16 l_i_blocks_high
; /* were l_i_reserved1 */
628 __le16 l_i_file_acl_high
;
629 __le16 l_i_uid_high
; /* these 2 fields */
630 __le16 l_i_gid_high
; /* were reserved2[0] */
634 __le16 h_i_reserved1
; /* Obsoleted fragment number/size which are removed in ext4 */
641 __le16 h_i_reserved1
; /* Obsoleted fragment number/size which are removed in ext4 */
642 __le16 m_i_file_acl_high
;
643 __u32 m_i_reserved2
[2];
645 } osd2
; /* OS dependent 2 */
646 __le16 i_extra_isize
;
648 __le32 i_ctime_extra
; /* extra Change time (nsec << 2 | epoch) */
649 __le32 i_mtime_extra
; /* extra Modification time(nsec << 2 | epoch) */
650 __le32 i_atime_extra
; /* extra Access time (nsec << 2 | epoch) */
651 __le32 i_crtime
; /* File Creation time */
652 __le32 i_crtime_extra
; /* extra FileCreationtime (nsec << 2 | epoch) */
653 __le32 i_version_hi
; /* high 32 bits for 64-bit version */
657 __u32 reserved
; /* should be zero */
658 __u32 donor_fd
; /* donor file descriptor */
659 __u64 orig_start
; /* logical start offset in block for orig */
660 __u64 donor_start
; /* logical start offset in block for donor */
661 __u64 len
; /* block length to be moved */
662 __u64 moved_len
; /* moved block length */
665 #define EXT4_EPOCH_BITS 2
666 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
667 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
670 * Extended fields will fit into an inode if the filesystem was formatted
671 * with large inodes (-I 256 or larger) and there are not currently any EAs
672 * consuming all of the available space. For new inodes we always reserve
673 * enough space for the kernel's known extended fields, but for inodes
674 * created with an old kernel this might not have been the case. None of
675 * the extended inode fields is critical for correct filesystem operation.
676 * This macro checks if a certain field fits in the inode. Note that
677 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
679 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
680 ((offsetof(typeof(*ext4_inode), field) + \
681 sizeof((ext4_inode)->field)) \
682 <= (EXT4_GOOD_OLD_INODE_SIZE + \
683 (einode)->i_extra_isize)) \
685 static inline __le32 ext4_encode_extra_time(struct timespec *time)
687 return cpu_to_le32((sizeof(time
->tv_sec
) > 4 ?
688 (time
->tv_sec
>> 32) & EXT4_EPOCH_MASK
: 0) |
689 ((time
->tv_nsec
<< EXT4_EPOCH_BITS
) & EXT4_NSEC_MASK
));
692 static inline void ext4_decode_extra_time(struct timespec
*time
, __le32 extra
)
694 if (sizeof(time
->tv_sec
) > 4)
695 time
->tv_sec
|= (__u64
)(le32_to_cpu(extra
) & EXT4_EPOCH_MASK
)
697 time
->tv_nsec
= (le32_to_cpu(extra
) & EXT4_NSEC_MASK
) >> EXT4_EPOCH_BITS
;
700 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
702 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
703 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
704 (raw_inode)->xtime ## _extra = \
705 ext4_encode_extra_time(&(inode)->xtime); \
708 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
710 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
711 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
712 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
713 (raw_inode)->xtime ## _extra = \
714 ext4_encode_extra_time(&(einode)->xtime); \
717 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
719 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
720 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
721 ext4_decode_extra_time(&(inode)->xtime, \
722 raw_inode->xtime ## _extra); \
724 (inode)->xtime.tv_nsec = 0; \
727 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
729 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
730 (einode)->xtime.tv_sec = \
731 (signed)le32_to_cpu((raw_inode)->xtime); \
732 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
733 ext4_decode_extra_time(&(einode)->xtime, \
734 raw_inode->xtime ## _extra); \
736 (einode)->xtime.tv_nsec = 0; \
739 #define i_disk_version osd1.linux1.l_i_version
741 #if defined(__KERNEL__) || defined(__linux__)
742 #define i_reserved1 osd1.linux1.l_i_reserved1
743 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
744 #define i_blocks_high osd2.linux2.l_i_blocks_high
745 #define i_uid_low i_uid
746 #define i_gid_low i_gid
747 #define i_uid_high osd2.linux2.l_i_uid_high
748 #define i_gid_high osd2.linux2.l_i_gid_high
749 #define i_reserved2 osd2.linux2.l_i_reserved2
751 #elif defined(__GNU__)
753 #define i_translator osd1.hurd1.h_i_translator
754 #define i_uid_high osd2.hurd2.h_i_uid_high
755 #define i_gid_high osd2.hurd2.h_i_gid_high
756 #define i_author osd2.hurd2.h_i_author
758 #elif defined(__masix__)
760 #define i_reserved1 osd1.masix1.m_i_reserved1
761 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
762 #define i_reserved2 osd2.masix2.m_i_reserved2
764 #endif /* defined(__KERNEL__) || defined(__linux__) */
767 * storage for cached extent
768 * If ec_len == 0, then the cache is invalid.
769 * If ec_start == 0, then the cache represents a gap (null mapping)
771 struct ext4_ext_cache
{
772 ext4_fsblk_t ec_start
;
773 ext4_lblk_t ec_block
;
774 __u32 ec_len
; /* must be 32bit to return holes */
778 * fourth extended file system inode data in memory
780 struct ext4_inode_info
{
781 __le32 i_data
[15]; /* unconverted */
783 ext4_fsblk_t i_file_acl
;
786 * i_block_group is the number of the block group which contains
787 * this file's inode. Constant across the lifetime of the inode,
788 * it is ued for making block allocation decisions - we try to
789 * place a file's data blocks near its inode block, and new inodes
790 * near to their parent directory's inode.
792 ext4_group_t i_block_group
;
793 ext4_lblk_t i_dir_start_lookup
;
794 #if (BITS_PER_LONG < 64)
795 unsigned long i_state_flags
; /* Dynamic state flags */
797 unsigned long i_flags
;
799 #ifdef CONFIG_EXT4_FS_XATTR
801 * Extended attributes can be read independently of the main file
802 * data. Taking i_mutex even when reading would cause contention
803 * between readers of EAs and writers of regular file data, so
804 * instead we synchronize on xattr_sem when reading or changing
807 struct rw_semaphore xattr_sem
;
810 struct list_head i_orphan
; /* unlinked but open inodes */
813 * i_disksize keeps track of what the inode size is ON DISK, not
814 * in memory. During truncate, i_size is set to the new size by
815 * the VFS prior to calling ext4_truncate(), but the filesystem won't
816 * set i_disksize to 0 until the truncate is actually under way.
818 * The intent is that i_disksize always represents the blocks which
819 * are used by this file. This allows recovery to restart truncate
820 * on orphans if we crash during truncate. We actually write i_disksize
821 * into the on-disk inode when writing inodes out, instead of i_size.
823 * The only time when i_disksize and i_size may be different is when
824 * a truncate is in progress. The only things which change i_disksize
825 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
830 * i_data_sem is for serialising ext4_truncate() against
831 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
832 * data tree are chopped off during truncate. We can't do that in
833 * ext4 because whenever we perform intermediate commits during
834 * truncate, the inode and all the metadata blocks *must* be in a
835 * consistent state which allows truncation of the orphans to restart
836 * during recovery. Hence we must fix the get_block-vs-truncate race
837 * by other means, so we have i_data_sem.
839 struct rw_semaphore i_data_sem
;
840 struct inode vfs_inode
;
841 struct jbd2_inode
*jinode
;
843 struct ext4_ext_cache i_cached_extent
;
845 * File creation time. Its function is same as that of
846 * struct timespec i_{a,c,m}time in the generic inode.
848 struct timespec i_crtime
;
851 struct list_head i_prealloc_list
;
852 spinlock_t i_prealloc_lock
;
855 ext4_group_t i_last_alloc_group
;
857 /* allocation reservation info for delalloc */
858 /* In case of bigalloc, these refer to clusters rather than blocks */
859 unsigned int i_reserved_data_blocks
;
860 unsigned int i_reserved_meta_blocks
;
861 unsigned int i_allocated_meta_blocks
;
862 ext4_lblk_t i_da_metadata_calc_last_lblock
;
863 int i_da_metadata_calc_len
;
865 /* on-disk additional length */
869 /* quota space reservation, managed internally by quota code */
870 qsize_t i_reserved_quota
;
873 /* completed IOs that might need unwritten extents handling */
874 struct list_head i_completed_io_list
;
875 spinlock_t i_completed_io_lock
;
876 atomic_t i_ioend_count
; /* Number of outstanding io_end structs */
877 /* current io_end structure for async DIO write*/
878 ext4_io_end_t
*cur_aio_dio
;
879 atomic_t i_aiodio_unwritten
; /* Nr. of inflight conversions pending */
881 spinlock_t i_block_reservation_lock
;
884 * Transactions that contain inode's metadata needed to complete
885 * fsync and fdatasync, respectively.
888 tid_t i_datasync_tid
;
894 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
895 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
896 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
899 * Misc. filesystem flags
901 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
902 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
903 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
908 #define EXT4_MOUNT_OLDALLOC 0x00002 /* Don't use the new Orlov allocator */
909 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
910 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
911 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
912 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
913 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
914 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
915 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
916 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
917 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
918 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
919 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
920 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
921 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
922 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
923 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
924 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
925 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
926 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
927 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
928 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
929 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
930 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
931 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
932 #define EXT4_MOUNT_I_VERSION 0x2000000 /* i_version support */
933 #define EXT4_MOUNT_MBLK_IO_SUBMIT 0x4000000 /* multi-block io submits */
934 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
935 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
936 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
937 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
938 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
940 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
941 specified delalloc */
943 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
945 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
947 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
950 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
952 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
954 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
957 #define ext4_set_bit __test_and_set_bit_le
958 #define ext4_set_bit_atomic ext2_set_bit_atomic
959 #define ext4_clear_bit __test_and_clear_bit_le
960 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
961 #define ext4_test_bit test_bit_le
962 #define ext4_find_first_zero_bit find_first_zero_bit_le
963 #define ext4_find_next_zero_bit find_next_zero_bit_le
964 #define ext4_find_next_bit find_next_bit_le
966 extern void ext4_set_bits(void *bm
, int cur
, int len
);
969 * Maximal mount counts between two filesystem checks
971 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
972 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
975 * Behaviour when detecting errors
977 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
978 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
979 #define EXT4_ERRORS_PANIC 3 /* Panic */
980 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
983 * Structure of the super block
985 struct ext4_super_block
{
986 /*00*/ __le32 s_inodes_count
; /* Inodes count */
987 __le32 s_blocks_count_lo
; /* Blocks count */
988 __le32 s_r_blocks_count_lo
; /* Reserved blocks count */
989 __le32 s_free_blocks_count_lo
; /* Free blocks count */
990 /*10*/ __le32 s_free_inodes_count
; /* Free inodes count */
991 __le32 s_first_data_block
; /* First Data Block */
992 __le32 s_log_block_size
; /* Block size */
993 __le32 s_log_cluster_size
; /* Allocation cluster size */
994 /*20*/ __le32 s_blocks_per_group
; /* # Blocks per group */
995 __le32 s_clusters_per_group
; /* # Clusters per group */
996 __le32 s_inodes_per_group
; /* # Inodes per group */
997 __le32 s_mtime
; /* Mount time */
998 /*30*/ __le32 s_wtime
; /* Write time */
999 __le16 s_mnt_count
; /* Mount count */
1000 __le16 s_max_mnt_count
; /* Maximal mount count */
1001 __le16 s_magic
; /* Magic signature */
1002 __le16 s_state
; /* File system state */
1003 __le16 s_errors
; /* Behaviour when detecting errors */
1004 __le16 s_minor_rev_level
; /* minor revision level */
1005 /*40*/ __le32 s_lastcheck
; /* time of last check */
1006 __le32 s_checkinterval
; /* max. time between checks */
1007 __le32 s_creator_os
; /* OS */
1008 __le32 s_rev_level
; /* Revision level */
1009 /*50*/ __le16 s_def_resuid
; /* Default uid for reserved blocks */
1010 __le16 s_def_resgid
; /* Default gid for reserved blocks */
1012 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1014 * Note: the difference between the compatible feature set and
1015 * the incompatible feature set is that if there is a bit set
1016 * in the incompatible feature set that the kernel doesn't
1017 * know about, it should refuse to mount the filesystem.
1019 * e2fsck's requirements are more strict; if it doesn't know
1020 * about a feature in either the compatible or incompatible
1021 * feature set, it must abort and not try to meddle with
1022 * things it doesn't understand...
1024 __le32 s_first_ino
; /* First non-reserved inode */
1025 __le16 s_inode_size
; /* size of inode structure */
1026 __le16 s_block_group_nr
; /* block group # of this superblock */
1027 __le32 s_feature_compat
; /* compatible feature set */
1028 /*60*/ __le32 s_feature_incompat
; /* incompatible feature set */
1029 __le32 s_feature_ro_compat
; /* readonly-compatible feature set */
1030 /*68*/ __u8 s_uuid
[16]; /* 128-bit uuid for volume */
1031 /*78*/ char s_volume_name
[16]; /* volume name */
1032 /*88*/ char s_last_mounted
[64]; /* directory where last mounted */
1033 /*C8*/ __le32 s_algorithm_usage_bitmap
; /* For compression */
1035 * Performance hints. Directory preallocation should only
1036 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1038 __u8 s_prealloc_blocks
; /* Nr of blocks to try to preallocate*/
1039 __u8 s_prealloc_dir_blocks
; /* Nr to preallocate for dirs */
1040 __le16 s_reserved_gdt_blocks
; /* Per group desc for online growth */
1042 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1044 /*D0*/ __u8 s_journal_uuid
[16]; /* uuid of journal superblock */
1045 /*E0*/ __le32 s_journal_inum
; /* inode number of journal file */
1046 __le32 s_journal_dev
; /* device number of journal file */
1047 __le32 s_last_orphan
; /* start of list of inodes to delete */
1048 __le32 s_hash_seed
[4]; /* HTREE hash seed */
1049 __u8 s_def_hash_version
; /* Default hash version to use */
1050 __u8 s_jnl_backup_type
;
1051 __le16 s_desc_size
; /* size of group descriptor */
1052 /*100*/ __le32 s_default_mount_opts
;
1053 __le32 s_first_meta_bg
; /* First metablock block group */
1054 __le32 s_mkfs_time
; /* When the filesystem was created */
1055 __le32 s_jnl_blocks
[17]; /* Backup of the journal inode */
1056 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1057 /*150*/ __le32 s_blocks_count_hi
; /* Blocks count */
1058 __le32 s_r_blocks_count_hi
; /* Reserved blocks count */
1059 __le32 s_free_blocks_count_hi
; /* Free blocks count */
1060 __le16 s_min_extra_isize
; /* All inodes have at least # bytes */
1061 __le16 s_want_extra_isize
; /* New inodes should reserve # bytes */
1062 __le32 s_flags
; /* Miscellaneous flags */
1063 __le16 s_raid_stride
; /* RAID stride */
1064 __le16 s_mmp_update_interval
; /* # seconds to wait in MMP checking */
1065 __le64 s_mmp_block
; /* Block for multi-mount protection */
1066 __le32 s_raid_stripe_width
; /* blocks on all data disks (N*stride)*/
1067 __u8 s_log_groups_per_flex
; /* FLEX_BG group size */
1068 __u8 s_reserved_char_pad
;
1069 __le16 s_reserved_pad
;
1070 __le64 s_kbytes_written
; /* nr of lifetime kilobytes written */
1071 __le32 s_snapshot_inum
; /* Inode number of active snapshot */
1072 __le32 s_snapshot_id
; /* sequential ID of active snapshot */
1073 __le64 s_snapshot_r_blocks_count
; /* reserved blocks for active
1074 snapshot's future use */
1075 __le32 s_snapshot_list
; /* inode number of the head of the
1076 on-disk snapshot list */
1077 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1078 __le32 s_error_count
; /* number of fs errors */
1079 __le32 s_first_error_time
; /* first time an error happened */
1080 __le32 s_first_error_ino
; /* inode involved in first error */
1081 __le64 s_first_error_block
; /* block involved of first error */
1082 __u8 s_first_error_func
[32]; /* function where the error happened */
1083 __le32 s_first_error_line
; /* line number where error happened */
1084 __le32 s_last_error_time
; /* most recent time of an error */
1085 __le32 s_last_error_ino
; /* inode involved in last error */
1086 __le32 s_last_error_line
; /* line number where error happened */
1087 __le64 s_last_error_block
; /* block involved of last error */
1088 __u8 s_last_error_func
[32]; /* function where the error happened */
1089 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1090 __u8 s_mount_opts
[64];
1091 __le32 s_usr_quota_inum
; /* inode for tracking user quota */
1092 __le32 s_grp_quota_inum
; /* inode for tracking group quota */
1093 __le32 s_overhead_clusters
; /* overhead blocks/clusters in fs */
1094 __le32 s_reserved
[109]; /* Padding to the end of the block */
1097 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1102 * run-time mount flags
1104 #define EXT4_MF_MNTDIR_SAMPLED 0x0001
1105 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1108 * fourth extended-fs super-block data in memory
1110 struct ext4_sb_info
{
1111 unsigned long s_desc_size
; /* Size of a group descriptor in bytes */
1112 unsigned long s_inodes_per_block
;/* Number of inodes per block */
1113 unsigned long s_blocks_per_group
;/* Number of blocks in a group */
1114 unsigned long s_clusters_per_group
; /* Number of clusters in a group */
1115 unsigned long s_inodes_per_group
;/* Number of inodes in a group */
1116 unsigned long s_itb_per_group
; /* Number of inode table blocks per group */
1117 unsigned long s_gdb_count
; /* Number of group descriptor blocks */
1118 unsigned long s_desc_per_block
; /* Number of group descriptors per block */
1119 ext4_group_t s_groups_count
; /* Number of groups in the fs */
1120 ext4_group_t s_blockfile_groups
;/* Groups acceptable for non-extent files */
1121 unsigned long s_overhead_last
; /* Last calculated overhead */
1122 unsigned long s_blocks_last
; /* Last seen block count */
1123 unsigned int s_cluster_ratio
; /* Number of blocks per cluster */
1124 unsigned int s_cluster_bits
; /* log2 of s_cluster_ratio */
1125 loff_t s_bitmap_maxbytes
; /* max bytes for bitmap files */
1126 struct buffer_head
* s_sbh
; /* Buffer containing the super block */
1127 struct ext4_super_block
*s_es
; /* Pointer to the super block in the buffer */
1128 struct buffer_head
**s_group_desc
;
1129 unsigned int s_mount_opt
;
1130 unsigned int s_mount_opt2
;
1131 unsigned int s_mount_flags
;
1132 ext4_fsblk_t s_sb_block
;
1135 unsigned short s_mount_state
;
1136 unsigned short s_pad
;
1137 int s_addr_per_block_bits
;
1138 int s_desc_per_block_bits
;
1141 unsigned int s_inode_readahead_blks
;
1142 unsigned int s_inode_goal
;
1143 spinlock_t s_next_gen_lock
;
1144 u32 s_next_generation
;
1146 int s_def_hash_version
;
1147 int s_hash_unsigned
; /* 3 if hash should be signed, 0 if not */
1148 struct percpu_counter s_freeclusters_counter
;
1149 struct percpu_counter s_freeinodes_counter
;
1150 struct percpu_counter s_dirs_counter
;
1151 struct percpu_counter s_dirtyclusters_counter
;
1152 struct blockgroup_lock
*s_blockgroup_lock
;
1153 struct proc_dir_entry
*s_proc
;
1154 struct kobject s_kobj
;
1155 struct completion s_kobj_unregister
;
1158 struct journal_s
*s_journal
;
1159 struct list_head s_orphan
;
1160 struct mutex s_orphan_lock
;
1161 unsigned long s_resize_flags
; /* Flags indicating if there
1163 unsigned long s_commit_interval
;
1164 u32 s_max_batch_time
;
1165 u32 s_min_batch_time
;
1166 struct block_device
*journal_bdev
;
1167 #ifdef CONFIG_JBD2_DEBUG
1168 struct timer_list turn_ro_timer
; /* For turning read-only (crash simulation) */
1169 wait_queue_head_t ro_wait_queue
; /* For people waiting for the fs to go read-only */
1172 char *s_qf_names
[MAXQUOTAS
]; /* Names of quota files with journalled quota */
1173 int s_jquota_fmt
; /* Format of quota to use */
1175 unsigned int s_want_extra_isize
; /* New inodes should reserve # bytes */
1176 struct rb_root system_blks
;
1178 #ifdef EXTENTS_STATS
1179 /* ext4 extents stats */
1180 unsigned long s_ext_min
;
1181 unsigned long s_ext_max
;
1182 unsigned long s_depth_max
;
1183 spinlock_t s_ext_stats_lock
;
1184 unsigned long s_ext_blocks
;
1185 unsigned long s_ext_extents
;
1187 /* ext4 extent cache stats */
1188 unsigned long extent_cache_hits
;
1189 unsigned long extent_cache_misses
;
1191 /* for buddy allocator */
1192 struct ext4_group_info
***s_group_info
;
1193 struct inode
*s_buddy_cache
;
1194 spinlock_t s_md_lock
;
1195 unsigned short *s_mb_offsets
;
1196 unsigned int *s_mb_maxs
;
1199 unsigned long s_stripe
;
1200 unsigned int s_mb_stream_request
;
1201 unsigned int s_mb_max_to_scan
;
1202 unsigned int s_mb_min_to_scan
;
1203 unsigned int s_mb_stats
;
1204 unsigned int s_mb_order2_reqs
;
1205 unsigned int s_mb_group_prealloc
;
1206 unsigned int s_max_writeback_mb_bump
;
1207 /* where last allocation was done - for stream allocation */
1208 unsigned long s_mb_last_group
;
1209 unsigned long s_mb_last_start
;
1211 /* stats for buddy allocator */
1212 atomic_t s_bal_reqs
; /* number of reqs with len > 1 */
1213 atomic_t s_bal_success
; /* we found long enough chunks */
1214 atomic_t s_bal_allocated
; /* in blocks */
1215 atomic_t s_bal_ex_scanned
; /* total extents scanned */
1216 atomic_t s_bal_goals
; /* goal hits */
1217 atomic_t s_bal_breaks
; /* too long searches */
1218 atomic_t s_bal_2orders
; /* 2^order hits */
1219 spinlock_t s_bal_lock
;
1220 unsigned long s_mb_buddies_generated
;
1221 unsigned long long s_mb_generation_time
;
1222 atomic_t s_mb_lost_chunks
;
1223 atomic_t s_mb_preallocated
;
1224 atomic_t s_mb_discarded
;
1225 atomic_t s_lock_busy
;
1227 /* locality groups */
1228 struct ext4_locality_group __percpu
*s_locality_groups
;
1230 /* for write statistics */
1231 unsigned long s_sectors_written_start
;
1232 u64 s_kbytes_written
;
1234 unsigned int s_log_groups_per_flex
;
1235 struct flex_groups
*s_flex_groups
;
1237 /* workqueue for dio unwritten */
1238 struct workqueue_struct
*dio_unwritten_wq
;
1240 /* timer for periodic error stats printing */
1241 struct timer_list s_err_report
;
1243 /* Lazy inode table initialization info */
1244 struct ext4_li_request
*s_li_request
;
1245 /* Wait multiplier for lazy initialization thread */
1246 unsigned int s_li_wait_mult
;
1248 /* Kernel thread for multiple mount protection */
1249 struct task_struct
*s_mmp_tsk
;
1251 /* record the last minlen when FITRIM is called. */
1252 atomic_t s_last_trim_minblks
;
1255 static inline struct ext4_sb_info
*EXT4_SB(struct super_block
*sb
)
1257 return sb
->s_fs_info
;
1259 static inline struct ext4_inode_info
*EXT4_I(struct inode
*inode
)
1261 return container_of(inode
, struct ext4_inode_info
, vfs_inode
);
1264 static inline struct timespec
ext4_current_time(struct inode
*inode
)
1266 return (inode
->i_sb
->s_time_gran
< NSEC_PER_SEC
) ?
1267 current_fs_time(inode
->i_sb
) : CURRENT_TIME_SEC
;
1270 static inline int ext4_valid_inum(struct super_block
*sb
, unsigned long ino
)
1272 return ino
== EXT4_ROOT_INO
||
1273 ino
== EXT4_JOURNAL_INO
||
1274 ino
== EXT4_RESIZE_INO
||
1275 (ino
>= EXT4_FIRST_INO(sb
) &&
1276 ino
<= le32_to_cpu(EXT4_SB(sb
)->s_es
->s_inodes_count
));
1280 * Inode dynamic state flags
1283 EXT4_STATE_JDATA
, /* journaled data exists */
1284 EXT4_STATE_NEW
, /* inode is newly created */
1285 EXT4_STATE_XATTR
, /* has in-inode xattrs */
1286 EXT4_STATE_NO_EXPAND
, /* No space for expansion */
1287 EXT4_STATE_DA_ALLOC_CLOSE
, /* Alloc DA blks on close */
1288 EXT4_STATE_EXT_MIGRATE
, /* Inode is migrating */
1289 EXT4_STATE_DIO_UNWRITTEN
, /* need convert on dio done*/
1290 EXT4_STATE_NEWENTRY
, /* File just added to dir */
1291 EXT4_STATE_DELALLOC_RESERVED
, /* blks already reserved for delalloc */
1294 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1295 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1297 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1299 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1301 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1303 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1305 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1308 EXT4_INODE_BIT_FNS(flag
, flags
, 0)
1309 #if (BITS_PER_LONG < 64)
1310 EXT4_INODE_BIT_FNS(state
, state_flags
, 0)
1312 static inline void ext4_clear_state_flags(struct ext4_inode_info
*ei
)
1314 (ei
)->i_state_flags
= 0;
1317 EXT4_INODE_BIT_FNS(state
, flags
, 32)
1319 static inline void ext4_clear_state_flags(struct ext4_inode_info
*ei
)
1321 /* We depend on the fact that callers will set i_flags */
1325 /* Assume that user mode programs are passing in an ext4fs superblock, not
1326 * a kernel struct super_block. This will allow us to call the feature-test
1327 * macros from user land. */
1328 #define EXT4_SB(sb) (sb)
1331 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1334 * Codes for operating systems
1336 #define EXT4_OS_LINUX 0
1337 #define EXT4_OS_HURD 1
1338 #define EXT4_OS_MASIX 2
1339 #define EXT4_OS_FREEBSD 3
1340 #define EXT4_OS_LITES 4
1345 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1346 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1348 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1349 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1351 #define EXT4_GOOD_OLD_INODE_SIZE 128
1354 * Feature set definitions
1357 #define EXT4_HAS_COMPAT_FEATURE(sb,mask) \
1358 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1359 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \
1360 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1361 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \
1362 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1363 #define EXT4_SET_COMPAT_FEATURE(sb,mask) \
1364 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1365 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \
1366 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1367 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \
1368 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1369 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \
1370 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1371 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
1372 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1373 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \
1374 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1376 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1377 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1378 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1379 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1380 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1381 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1383 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1384 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1385 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1386 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1387 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1388 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1389 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1390 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1391 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1393 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1394 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1395 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1396 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1397 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1398 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1399 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1400 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1401 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1402 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1403 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1405 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1406 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1407 EXT4_FEATURE_INCOMPAT_META_BG)
1408 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1409 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1410 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1412 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1413 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1414 EXT4_FEATURE_INCOMPAT_RECOVER| \
1415 EXT4_FEATURE_INCOMPAT_META_BG)
1416 #define EXT3_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 EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR
1421 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1422 EXT4_FEATURE_INCOMPAT_RECOVER| \
1423 EXT4_FEATURE_INCOMPAT_META_BG| \
1424 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1425 EXT4_FEATURE_INCOMPAT_64BIT| \
1426 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1427 EXT4_FEATURE_INCOMPAT_MMP)
1428 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1429 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1430 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1431 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1432 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1433 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1434 EXT4_FEATURE_RO_COMPAT_HUGE_FILE)
1437 * Default values for user and/or group using reserved blocks
1439 #define EXT4_DEF_RESUID 0
1440 #define EXT4_DEF_RESGID 0
1442 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
1445 * Default mount options
1447 #define EXT4_DEFM_DEBUG 0x0001
1448 #define EXT4_DEFM_BSDGROUPS 0x0002
1449 #define EXT4_DEFM_XATTR_USER 0x0004
1450 #define EXT4_DEFM_ACL 0x0008
1451 #define EXT4_DEFM_UID16 0x0010
1452 #define EXT4_DEFM_JMODE 0x0060
1453 #define EXT4_DEFM_JMODE_DATA 0x0020
1454 #define EXT4_DEFM_JMODE_ORDERED 0x0040
1455 #define EXT4_DEFM_JMODE_WBACK 0x0060
1456 #define EXT4_DEFM_NOBARRIER 0x0100
1457 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1458 #define EXT4_DEFM_DISCARD 0x0400
1459 #define EXT4_DEFM_NODELALLOC 0x0800
1462 * Default journal batch times
1464 #define EXT4_DEF_MIN_BATCH_TIME 0
1465 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1468 * Minimum number of groups in a flexgroup before we separate out
1469 * directories into the first block group of a flexgroup
1471 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1474 * Structure of a directory entry
1476 #define EXT4_NAME_LEN 255
1478 struct ext4_dir_entry
{
1479 __le32 inode
; /* Inode number */
1480 __le16 rec_len
; /* Directory entry length */
1481 __le16 name_len
; /* Name length */
1482 char name
[EXT4_NAME_LEN
]; /* File name */
1486 * The new version of the directory entry. Since EXT4 structures are
1487 * stored in intel byte order, and the name_len field could never be
1488 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1491 struct ext4_dir_entry_2
{
1492 __le32 inode
; /* Inode number */
1493 __le16 rec_len
; /* Directory entry length */
1494 __u8 name_len
; /* Name length */
1496 char name
[EXT4_NAME_LEN
]; /* File name */
1500 * Ext4 directory file types. Only the low 3 bits are used. The
1501 * other bits are reserved for now.
1503 #define EXT4_FT_UNKNOWN 0
1504 #define EXT4_FT_REG_FILE 1
1505 #define EXT4_FT_DIR 2
1506 #define EXT4_FT_CHRDEV 3
1507 #define EXT4_FT_BLKDEV 4
1508 #define EXT4_FT_FIFO 5
1509 #define EXT4_FT_SOCK 6
1510 #define EXT4_FT_SYMLINK 7
1512 #define EXT4_FT_MAX 8
1515 * EXT4_DIR_PAD defines the directory entries boundaries
1517 * NOTE: It must be a multiple of 4
1519 #define EXT4_DIR_PAD 4
1520 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1521 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1523 #define EXT4_MAX_REC_LEN ((1<<16)-1)
1526 * If we ever get support for fs block sizes > page_size, we'll need
1527 * to remove the #if statements in the next two functions...
1529 static inline unsigned int
1530 ext4_rec_len_from_disk(__le16 dlen
, unsigned blocksize
)
1532 unsigned len
= le16_to_cpu(dlen
);
1534 #if (PAGE_CACHE_SIZE >= 65536)
1535 if (len
== EXT4_MAX_REC_LEN
|| len
== 0)
1537 return (len
& 65532) | ((len
& 3) << 16);
1543 static inline __le16
ext4_rec_len_to_disk(unsigned len
, unsigned blocksize
)
1545 if ((len
> blocksize
) || (blocksize
> (1 << 18)) || (len
& 3))
1547 #if (PAGE_CACHE_SIZE >= 65536)
1549 return cpu_to_le16(len
);
1550 if (len
== blocksize
) {
1551 if (blocksize
== 65536)
1552 return cpu_to_le16(EXT4_MAX_REC_LEN
);
1554 return cpu_to_le16(0);
1556 return cpu_to_le16((len
& 65532) | ((len
>> 16) & 3));
1558 return cpu_to_le16(len
);
1563 * Hash Tree Directory indexing
1564 * (c) Daniel Phillips, 2001
1567 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1568 EXT4_FEATURE_COMPAT_DIR_INDEX) && \
1569 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
1570 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1571 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1573 /* Legal values for the dx_root hash_version field: */
1575 #define DX_HASH_LEGACY 0
1576 #define DX_HASH_HALF_MD4 1
1577 #define DX_HASH_TEA 2
1578 #define DX_HASH_LEGACY_UNSIGNED 3
1579 #define DX_HASH_HALF_MD4_UNSIGNED 4
1580 #define DX_HASH_TEA_UNSIGNED 5
1584 /* hash info structure used by the directory hash */
1593 #define EXT4_HTREE_EOF 0x7fffffff
1596 * Control parameters used by ext4_htree_next_block
1598 #define HASH_NB_ALWAYS 1
1602 * Describe an inode's exact location on disk and in memory
1606 struct buffer_head
*bh
;
1607 unsigned long offset
;
1608 ext4_group_t block_group
;
1611 static inline struct ext4_inode
*ext4_raw_inode(struct ext4_iloc
*iloc
)
1613 return (struct ext4_inode
*) (iloc
->bh
->b_data
+ iloc
->offset
);
1617 * This structure is stuffed into the struct file's private_data field
1618 * for directories. It is where we put information so that we can do
1619 * readdir operations in hash tree order.
1621 struct dir_private_info
{
1622 struct rb_root root
;
1623 struct rb_node
*curr_node
;
1624 struct fname
*extra_fname
;
1627 __u32 curr_minor_hash
;
1631 /* calculate the first block number of the group */
1632 static inline ext4_fsblk_t
1633 ext4_group_first_block_no(struct super_block
*sb
, ext4_group_t group_no
)
1635 return group_no
* (ext4_fsblk_t
)EXT4_BLOCKS_PER_GROUP(sb
) +
1636 le32_to_cpu(EXT4_SB(sb
)->s_es
->s_first_data_block
);
1640 * Special error return code only used by dx_probe() and its callers.
1642 #define ERR_BAD_DX_DIR -75000
1644 void ext4_get_group_no_and_offset(struct super_block
*sb
, ext4_fsblk_t blocknr
,
1645 ext4_group_t
*blockgrpp
, ext4_grpblk_t
*offsetp
);
1648 * Timeout and state flag for lazy initialization inode thread.
1650 #define EXT4_DEF_LI_WAIT_MULT 10
1651 #define EXT4_DEF_LI_MAX_START_DELAY 5
1652 #define EXT4_LAZYINIT_QUIT 0x0001
1653 #define EXT4_LAZYINIT_RUNNING 0x0002
1656 * Lazy inode table initialization info
1658 struct ext4_lazy_init
{
1659 unsigned long li_state
;
1660 struct list_head li_request_list
;
1661 struct mutex li_list_mtx
;
1664 struct ext4_li_request
{
1665 struct super_block
*lr_super
;
1666 struct ext4_sb_info
*lr_sbi
;
1667 ext4_group_t lr_next_group
;
1668 struct list_head lr_request
;
1669 unsigned long lr_next_sched
;
1670 unsigned long lr_timeout
;
1673 struct ext4_features
{
1674 struct kobject f_kobj
;
1675 struct completion f_kobj_unregister
;
1679 * This structure will be used for multiple mount protection. It will be
1680 * written into the block number saved in the s_mmp_block field in the
1681 * superblock. Programs that check MMP should assume that if
1682 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
1683 * to use the filesystem, regardless of how old the timestamp is.
1685 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
1686 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
1687 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
1688 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
1691 __le32 mmp_magic
; /* Magic number for MMP */
1692 __le32 mmp_seq
; /* Sequence no. updated periodically */
1695 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
1696 * purposes and do not affect the correctness of the algorithm
1698 __le64 mmp_time
; /* Time last updated */
1699 char mmp_nodename
[64]; /* Node which last updated MMP block */
1700 char mmp_bdevname
[32]; /* Bdev which last updated MMP block */
1703 * mmp_check_interval is used to verify if the MMP block has been
1704 * updated on the block device. The value is updated based on the
1705 * maximum time to write the MMP block during an update cycle.
1707 __le16 mmp_check_interval
;
1710 __le32 mmp_pad2
[227];
1713 /* arguments passed to the mmp thread */
1715 struct buffer_head
*bh
; /* bh from initial read_mmp_block() */
1716 struct super_block
*sb
; /* super block of the fs */
1720 * Check interval multiplier
1721 * The MMP block is written every update interval and initially checked every
1722 * update interval x the multiplier (the value is then adapted based on the
1723 * write latency). The reason is that writes can be delayed under load and we
1724 * don't want readers to incorrectly assume that the filesystem is no longer
1727 #define EXT4_MMP_CHECK_MULT 2UL
1730 * Minimum interval for MMP checking in seconds.
1732 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
1735 * Maximum interval for MMP checking in seconds.
1737 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
1740 * Function prototypes
1744 * Ok, these declarations are also in <linux/kernel.h> but none of the
1745 * ext4 source programs needs to include it so they are duplicated here.
1747 # define NORET_TYPE /**/
1748 # define ATTRIB_NORET __attribute__((noreturn))
1749 # define NORET_AND noreturn,
1752 extern unsigned int ext4_count_free(struct buffer_head
*, unsigned);
1755 extern unsigned int ext4_block_group(struct super_block
*sb
,
1756 ext4_fsblk_t blocknr
);
1757 extern ext4_grpblk_t
ext4_block_group_offset(struct super_block
*sb
,
1758 ext4_fsblk_t blocknr
);
1759 extern int ext4_bg_has_super(struct super_block
*sb
, ext4_group_t group
);
1760 extern unsigned long ext4_bg_num_gdb(struct super_block
*sb
,
1761 ext4_group_t group
);
1762 extern ext4_fsblk_t
ext4_new_meta_blocks(handle_t
*handle
, struct inode
*inode
,
1765 unsigned long *count
,
1767 extern int ext4_claim_free_blocks(struct ext4_sb_info
*sbi
,
1768 s64 nblocks
, unsigned int flags
);
1769 extern ext4_fsblk_t
ext4_count_free_blocks(struct super_block
*);
1770 extern void ext4_check_blocks_bitmap(struct super_block
*);
1771 extern struct ext4_group_desc
* ext4_get_group_desc(struct super_block
* sb
,
1772 ext4_group_t block_group
,
1773 struct buffer_head
** bh
);
1774 extern int ext4_should_retry_alloc(struct super_block
*sb
, int *retries
);
1775 struct buffer_head
*ext4_read_block_bitmap(struct super_block
*sb
,
1776 ext4_group_t block_group
);
1777 extern void ext4_init_block_bitmap(struct super_block
*sb
,
1778 struct buffer_head
*bh
,
1780 struct ext4_group_desc
*desc
);
1781 extern unsigned ext4_free_blocks_after_init(struct super_block
*sb
,
1782 ext4_group_t block_group
,
1783 struct ext4_group_desc
*gdp
);
1784 extern unsigned ext4_num_base_meta_clusters(struct super_block
*sb
,
1785 ext4_group_t block_group
);
1786 extern unsigned ext4_num_overhead_clusters(struct super_block
*sb
,
1787 ext4_group_t block_group
,
1788 struct ext4_group_desc
*gdp
);
1789 ext4_fsblk_t
ext4_inode_to_goal_block(struct inode
*);
1792 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode
*,
1794 struct ext4_dir_entry_2
*,
1795 struct buffer_head
*, unsigned int);
1796 #define ext4_check_dir_entry(dir, filp, de, bh, offset) \
1797 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
1798 (de), (bh), (offset)))
1799 extern int ext4_htree_store_dirent(struct file
*dir_file
, __u32 hash
,
1801 struct ext4_dir_entry_2
*dirent
);
1802 extern void ext4_htree_free_dir_info(struct dir_private_info
*p
);
1805 extern int ext4_sync_file(struct file
*, loff_t
, loff_t
, int);
1806 extern int ext4_flush_completed_IO(struct inode
*);
1809 extern int ext4fs_dirhash(const char *name
, int len
, struct
1810 dx_hash_info
*hinfo
);
1813 extern struct inode
*ext4_new_inode(handle_t
*, struct inode
*, int,
1814 const struct qstr
*qstr
, __u32 goal
);
1815 extern void ext4_free_inode(handle_t
*, struct inode
*);
1816 extern struct inode
* ext4_orphan_get(struct super_block
*, unsigned long);
1817 extern unsigned long ext4_count_free_inodes(struct super_block
*);
1818 extern unsigned long ext4_count_dirs(struct super_block
*);
1819 extern void ext4_check_inodes_bitmap(struct super_block
*);
1820 extern void ext4_mark_bitmap_end(int start_bit
, int end_bit
, char *bitmap
);
1821 extern int ext4_init_inode_table(struct super_block
*sb
,
1822 ext4_group_t group
, int barrier
);
1825 extern long ext4_mb_stats
;
1826 extern long ext4_mb_max_to_scan
;
1827 extern int ext4_mb_init(struct super_block
*, int);
1828 extern int ext4_mb_release(struct super_block
*);
1829 extern ext4_fsblk_t
ext4_mb_new_blocks(handle_t
*,
1830 struct ext4_allocation_request
*, int *);
1831 extern int ext4_mb_reserve_blocks(struct super_block
*, int);
1832 extern void ext4_discard_preallocations(struct inode
*);
1833 extern int __init
ext4_init_mballoc(void);
1834 extern void ext4_exit_mballoc(void);
1835 extern void ext4_free_blocks(handle_t
*handle
, struct inode
*inode
,
1836 struct buffer_head
*bh
, ext4_fsblk_t block
,
1837 unsigned long count
, int flags
);
1838 extern int ext4_mb_add_groupinfo(struct super_block
*sb
,
1839 ext4_group_t i
, struct ext4_group_desc
*desc
);
1840 extern int ext4_group_add_blocks(handle_t
*handle
, struct super_block
*sb
,
1841 ext4_fsblk_t block
, unsigned long count
);
1842 extern int ext4_trim_fs(struct super_block
*, struct fstrim_range
*);
1845 struct buffer_head
*ext4_getblk(handle_t
*, struct inode
*,
1846 ext4_lblk_t
, int, int *);
1847 struct buffer_head
*ext4_bread(handle_t
*, struct inode
*,
1848 ext4_lblk_t
, int, int *);
1849 int ext4_get_block(struct inode
*inode
, sector_t iblock
,
1850 struct buffer_head
*bh_result
, int create
);
1852 extern struct inode
*ext4_iget(struct super_block
*, unsigned long);
1853 extern int ext4_write_inode(struct inode
*, struct writeback_control
*);
1854 extern int ext4_setattr(struct dentry
*, struct iattr
*);
1855 extern int ext4_getattr(struct vfsmount
*mnt
, struct dentry
*dentry
,
1856 struct kstat
*stat
);
1857 extern void ext4_evict_inode(struct inode
*);
1858 extern void ext4_clear_inode(struct inode
*);
1859 extern int ext4_sync_inode(handle_t
*, struct inode
*);
1860 extern void ext4_dirty_inode(struct inode
*, int);
1861 extern int ext4_change_inode_journal_flag(struct inode
*, int);
1862 extern int ext4_get_inode_loc(struct inode
*, struct ext4_iloc
*);
1863 extern int ext4_can_truncate(struct inode
*inode
);
1864 extern void ext4_truncate(struct inode
*);
1865 extern int ext4_punch_hole(struct file
*file
, loff_t offset
, loff_t length
);
1866 extern int ext4_truncate_restart_trans(handle_t
*, struct inode
*, int nblocks
);
1867 extern void ext4_set_inode_flags(struct inode
*);
1868 extern void ext4_get_inode_flags(struct ext4_inode_info
*);
1869 extern int ext4_alloc_da_blocks(struct inode
*inode
);
1870 extern void ext4_set_aops(struct inode
*inode
);
1871 extern int ext4_writepage_trans_blocks(struct inode
*);
1872 extern int ext4_chunk_trans_blocks(struct inode
*, int nrblocks
);
1873 extern int ext4_block_truncate_page(handle_t
*handle
,
1874 struct address_space
*mapping
, loff_t from
);
1875 extern int ext4_block_zero_page_range(handle_t
*handle
,
1876 struct address_space
*mapping
, loff_t from
, loff_t length
);
1877 extern int ext4_discard_partial_page_buffers(handle_t
*handle
,
1878 struct address_space
*mapping
, loff_t from
,
1879 loff_t length
, int flags
);
1880 extern int ext4_discard_partial_page_buffers_no_lock(handle_t
*handle
,
1881 struct inode
*inode
, struct page
*page
, loff_t from
,
1882 loff_t length
, int flags
);
1883 extern int ext4_page_mkwrite(struct vm_area_struct
*vma
, struct vm_fault
*vmf
);
1884 extern qsize_t
*ext4_get_reserved_space(struct inode
*inode
);
1885 extern void ext4_da_update_reserve_space(struct inode
*inode
,
1886 int used
, int quota_claim
);
1889 extern int ext4_ind_map_blocks(handle_t
*handle
, struct inode
*inode
,
1890 struct ext4_map_blocks
*map
, int flags
);
1891 extern ssize_t
ext4_ind_direct_IO(int rw
, struct kiocb
*iocb
,
1892 const struct iovec
*iov
, loff_t offset
,
1893 unsigned long nr_segs
);
1894 extern int ext4_ind_calc_metadata_amount(struct inode
*inode
, sector_t lblock
);
1895 extern int ext4_ind_trans_blocks(struct inode
*inode
, int nrblocks
, int chunk
);
1896 extern void ext4_ind_truncate(struct inode
*inode
);
1899 extern long ext4_ioctl(struct file
*, unsigned int, unsigned long);
1900 extern long ext4_compat_ioctl(struct file
*, unsigned int, unsigned long);
1903 extern int ext4_ext_migrate(struct inode
*);
1906 extern int ext4_orphan_add(handle_t
*, struct inode
*);
1907 extern int ext4_orphan_del(handle_t
*, struct inode
*);
1908 extern int ext4_htree_fill_tree(struct file
*dir_file
, __u32 start_hash
,
1909 __u32 start_minor_hash
, __u32
*next_hash
);
1912 extern int ext4_group_add(struct super_block
*sb
,
1913 struct ext4_new_group_data
*input
);
1914 extern int ext4_group_extend(struct super_block
*sb
,
1915 struct ext4_super_block
*es
,
1916 ext4_fsblk_t n_blocks_count
);
1919 extern void *ext4_kvmalloc(size_t size
, gfp_t flags
);
1920 extern void *ext4_kvzalloc(size_t size
, gfp_t flags
);
1921 extern void ext4_kvfree(void *ptr
);
1922 extern void __ext4_error(struct super_block
*, const char *, unsigned int,
1924 __attribute__ ((format (printf
, 4, 5)));
1925 #define ext4_error(sb, message...) __ext4_error(sb, __func__, \
1926 __LINE__, ## message)
1927 extern void ext4_error_inode(struct inode
*, const char *, unsigned int,
1928 ext4_fsblk_t
, const char *, ...)
1929 __attribute__ ((format (printf
, 5, 6)));
1930 extern void ext4_error_file(struct file
*, const char *, unsigned int,
1931 ext4_fsblk_t
, const char *, ...)
1932 __attribute__ ((format (printf
, 5, 6)));
1933 extern void __ext4_std_error(struct super_block
*, const char *,
1935 extern void __ext4_abort(struct super_block
*, const char *, unsigned int,
1937 __attribute__ ((format (printf
, 4, 5)));
1938 #define ext4_abort(sb, message...) __ext4_abort(sb, __func__, \
1939 __LINE__, ## message)
1940 extern void __ext4_warning(struct super_block
*, const char *, unsigned int,
1942 __attribute__ ((format (printf
, 4, 5)));
1943 #define ext4_warning(sb, message...) __ext4_warning(sb, __func__, \
1944 __LINE__, ## message)
1945 extern void ext4_msg(struct super_block
*, const char *, const char *, ...)
1946 __attribute__ ((format (printf
, 3, 4)));
1947 extern void __dump_mmp_msg(struct super_block
*, struct mmp_struct
*mmp
,
1948 const char *, unsigned int, const char *);
1949 #define dump_mmp_msg(sb, mmp, msg) __dump_mmp_msg(sb, mmp, __func__, \
1951 extern void __ext4_grp_locked_error(const char *, unsigned int, \
1952 struct super_block
*, ext4_group_t
, \
1953 unsigned long, ext4_fsblk_t
, \
1955 __attribute__ ((format (printf
, 7, 8)));
1956 #define ext4_grp_locked_error(sb, grp, message...) \
1957 __ext4_grp_locked_error(__func__, __LINE__, (sb), (grp), ## message)
1958 extern void ext4_update_dynamic_rev(struct super_block
*sb
);
1959 extern int ext4_update_compat_feature(handle_t
*handle
, struct super_block
*sb
,
1961 extern int ext4_update_rocompat_feature(handle_t
*handle
,
1962 struct super_block
*sb
, __u32 rocompat
);
1963 extern int ext4_update_incompat_feature(handle_t
*handle
,
1964 struct super_block
*sb
, __u32 incompat
);
1965 extern ext4_fsblk_t
ext4_block_bitmap(struct super_block
*sb
,
1966 struct ext4_group_desc
*bg
);
1967 extern ext4_fsblk_t
ext4_inode_bitmap(struct super_block
*sb
,
1968 struct ext4_group_desc
*bg
);
1969 extern ext4_fsblk_t
ext4_inode_table(struct super_block
*sb
,
1970 struct ext4_group_desc
*bg
);
1971 extern __u32
ext4_free_blks_count(struct super_block
*sb
,
1972 struct ext4_group_desc
*bg
);
1973 extern __u32
ext4_free_inodes_count(struct super_block
*sb
,
1974 struct ext4_group_desc
*bg
);
1975 extern __u32
ext4_used_dirs_count(struct super_block
*sb
,
1976 struct ext4_group_desc
*bg
);
1977 extern __u32
ext4_itable_unused_count(struct super_block
*sb
,
1978 struct ext4_group_desc
*bg
);
1979 extern void ext4_block_bitmap_set(struct super_block
*sb
,
1980 struct ext4_group_desc
*bg
, ext4_fsblk_t blk
);
1981 extern void ext4_inode_bitmap_set(struct super_block
*sb
,
1982 struct ext4_group_desc
*bg
, ext4_fsblk_t blk
);
1983 extern void ext4_inode_table_set(struct super_block
*sb
,
1984 struct ext4_group_desc
*bg
, ext4_fsblk_t blk
);
1985 extern void ext4_free_blks_set(struct super_block
*sb
,
1986 struct ext4_group_desc
*bg
, __u32 count
);
1987 extern void ext4_free_inodes_set(struct super_block
*sb
,
1988 struct ext4_group_desc
*bg
, __u32 count
);
1989 extern void ext4_used_dirs_set(struct super_block
*sb
,
1990 struct ext4_group_desc
*bg
, __u32 count
);
1991 extern void ext4_itable_unused_set(struct super_block
*sb
,
1992 struct ext4_group_desc
*bg
, __u32 count
);
1993 extern __le16
ext4_group_desc_csum(struct ext4_sb_info
*sbi
, __u32 group
,
1994 struct ext4_group_desc
*gdp
);
1995 extern int ext4_group_desc_csum_verify(struct ext4_sb_info
*sbi
, __u32 group
,
1996 struct ext4_group_desc
*gdp
);
1998 static inline ext4_fsblk_t
ext4_blocks_count(struct ext4_super_block
*es
)
2000 return ((ext4_fsblk_t
)le32_to_cpu(es
->s_blocks_count_hi
) << 32) |
2001 le32_to_cpu(es
->s_blocks_count_lo
);
2004 static inline ext4_fsblk_t
ext4_r_blocks_count(struct ext4_super_block
*es
)
2006 return ((ext4_fsblk_t
)le32_to_cpu(es
->s_r_blocks_count_hi
) << 32) |
2007 le32_to_cpu(es
->s_r_blocks_count_lo
);
2010 static inline ext4_fsblk_t
ext4_free_blocks_count(struct ext4_super_block
*es
)
2012 return ((ext4_fsblk_t
)le32_to_cpu(es
->s_free_blocks_count_hi
) << 32) |
2013 le32_to_cpu(es
->s_free_blocks_count_lo
);
2016 static inline void ext4_blocks_count_set(struct ext4_super_block
*es
,
2019 es
->s_blocks_count_lo
= cpu_to_le32((u32
)blk
);
2020 es
->s_blocks_count_hi
= cpu_to_le32(blk
>> 32);
2023 static inline void ext4_free_blocks_count_set(struct ext4_super_block
*es
,
2026 es
->s_free_blocks_count_lo
= cpu_to_le32((u32
)blk
);
2027 es
->s_free_blocks_count_hi
= cpu_to_le32(blk
>> 32);
2030 static inline void ext4_r_blocks_count_set(struct ext4_super_block
*es
,
2033 es
->s_r_blocks_count_lo
= cpu_to_le32((u32
)blk
);
2034 es
->s_r_blocks_count_hi
= cpu_to_le32(blk
>> 32);
2037 static inline loff_t
ext4_isize(struct ext4_inode
*raw_inode
)
2039 if (S_ISREG(le16_to_cpu(raw_inode
->i_mode
)))
2040 return ((loff_t
)le32_to_cpu(raw_inode
->i_size_high
) << 32) |
2041 le32_to_cpu(raw_inode
->i_size_lo
);
2043 return (loff_t
) le32_to_cpu(raw_inode
->i_size_lo
);
2046 static inline void ext4_isize_set(struct ext4_inode
*raw_inode
, loff_t i_size
)
2048 raw_inode
->i_size_lo
= cpu_to_le32(i_size
);
2049 raw_inode
->i_size_high
= cpu_to_le32(i_size
>> 32);
2053 struct ext4_group_info
*ext4_get_group_info(struct super_block
*sb
,
2056 struct ext4_group_info
***grp_info
;
2057 long indexv
, indexh
;
2058 grp_info
= EXT4_SB(sb
)->s_group_info
;
2059 indexv
= group
>> (EXT4_DESC_PER_BLOCK_BITS(sb
));
2060 indexh
= group
& ((EXT4_DESC_PER_BLOCK(sb
)) - 1);
2061 return grp_info
[indexv
][indexh
];
2065 * Reading s_groups_count requires using smp_rmb() afterwards. See
2066 * the locking protocol documented in the comments of ext4_group_add()
2069 static inline ext4_group_t
ext4_get_groups_count(struct super_block
*sb
)
2071 ext4_group_t ngroups
= EXT4_SB(sb
)->s_groups_count
;
2077 static inline ext4_group_t
ext4_flex_group(struct ext4_sb_info
*sbi
,
2078 ext4_group_t block_group
)
2080 return block_group
>> sbi
->s_log_groups_per_flex
;
2083 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info
*sbi
)
2085 return 1 << sbi
->s_log_groups_per_flex
;
2088 #define ext4_std_error(sb, errno) \
2091 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
2095 /* Each CPU can accumulate percpu_counter_batch blocks in their local
2096 * counters. So we need to make sure we have free blocks more
2097 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
2099 #define EXT4_FREEBLOCKS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2101 #define EXT4_FREEBLOCKS_WATERMARK 0
2104 static inline void ext4_update_i_disksize(struct inode
*inode
, loff_t newsize
)
2107 * XXX: replace with spinlock if seen contended -bzzz
2109 down_write(&EXT4_I(inode
)->i_data_sem
);
2110 if (newsize
> EXT4_I(inode
)->i_disksize
)
2111 EXT4_I(inode
)->i_disksize
= newsize
;
2112 up_write(&EXT4_I(inode
)->i_data_sem
);
2116 struct ext4_group_info
{
2117 unsigned long bb_state
;
2118 struct rb_root bb_free_root
;
2119 ext4_grpblk_t bb_first_free
; /* first free block */
2120 ext4_grpblk_t bb_free
; /* total free blocks */
2121 ext4_grpblk_t bb_fragments
; /* nr of freespace fragments */
2122 ext4_grpblk_t bb_largest_free_order
;/* order of largest frag in BG */
2123 struct list_head bb_prealloc_list
;
2127 struct rw_semaphore alloc_sem
;
2128 ext4_grpblk_t bb_counters
[]; /* Nr of free power-of-two-block
2129 * regions, index is order.
2130 * bb_counters[3] = 5 means
2131 * 5 free 8-block regions. */
2134 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
2135 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
2137 #define EXT4_MB_GRP_NEED_INIT(grp) \
2138 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2140 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
2141 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2142 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
2143 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2144 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
2145 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2147 #define EXT4_MAX_CONTENTION 8
2148 #define EXT4_CONTENTION_THRESHOLD 2
2150 static inline spinlock_t
*ext4_group_lock_ptr(struct super_block
*sb
,
2153 return bgl_lock_ptr(EXT4_SB(sb
)->s_blockgroup_lock
, group
);
2157 * Returns true if the filesystem is busy enough that attempts to
2158 * access the block group locks has run into contention.
2160 static inline int ext4_fs_is_busy(struct ext4_sb_info
*sbi
)
2162 return (atomic_read(&sbi
->s_lock_busy
) > EXT4_CONTENTION_THRESHOLD
);
2165 static inline void ext4_lock_group(struct super_block
*sb
, ext4_group_t group
)
2167 spinlock_t
*lock
= ext4_group_lock_ptr(sb
, group
);
2168 if (spin_trylock(lock
))
2170 * We're able to grab the lock right away, so drop the
2171 * lock contention counter.
2173 atomic_add_unless(&EXT4_SB(sb
)->s_lock_busy
, -1, 0);
2176 * The lock is busy, so bump the contention counter,
2177 * and then wait on the spin lock.
2179 atomic_add_unless(&EXT4_SB(sb
)->s_lock_busy
, 1,
2180 EXT4_MAX_CONTENTION
);
2185 static inline void ext4_unlock_group(struct super_block
*sb
,
2188 spin_unlock(ext4_group_lock_ptr(sb
, group
));
2191 static inline void ext4_mark_super_dirty(struct super_block
*sb
)
2193 if (EXT4_SB(sb
)->s_journal
== NULL
)
2198 * Block validity checking
2200 #define ext4_check_indirect_blockref(inode, bh) \
2201 ext4_check_blockref(__func__, __LINE__, inode, \
2202 (__le32 *)(bh)->b_data, \
2203 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2205 #define ext4_ind_check_inode(inode) \
2206 ext4_check_blockref(__func__, __LINE__, inode, \
2207 EXT4_I(inode)->i_data, \
2211 * Inodes and files operations
2215 extern const struct file_operations ext4_dir_operations
;
2218 extern const struct inode_operations ext4_file_inode_operations
;
2219 extern const struct file_operations ext4_file_operations
;
2220 extern loff_t
ext4_llseek(struct file
*file
, loff_t offset
, int origin
);
2223 extern const struct inode_operations ext4_dir_inode_operations
;
2224 extern const struct inode_operations ext4_special_inode_operations
;
2225 extern struct dentry
*ext4_get_parent(struct dentry
*child
);
2228 extern const struct inode_operations ext4_symlink_inode_operations
;
2229 extern const struct inode_operations ext4_fast_symlink_inode_operations
;
2231 /* block_validity */
2232 extern void ext4_release_system_zone(struct super_block
*sb
);
2233 extern int ext4_setup_system_zone(struct super_block
*sb
);
2234 extern int __init
ext4_init_system_zone(void);
2235 extern void ext4_exit_system_zone(void);
2236 extern int ext4_data_block_valid(struct ext4_sb_info
*sbi
,
2237 ext4_fsblk_t start_blk
,
2238 unsigned int count
);
2239 extern int ext4_check_blockref(const char *, unsigned int,
2240 struct inode
*, __le32
*, unsigned int);
2243 extern int ext4_ext_tree_init(handle_t
*handle
, struct inode
*);
2244 extern int ext4_ext_writepage_trans_blocks(struct inode
*, int);
2245 extern int ext4_ext_index_trans_blocks(struct inode
*inode
, int nrblocks
,
2247 extern int ext4_ext_map_blocks(handle_t
*handle
, struct inode
*inode
,
2248 struct ext4_map_blocks
*map
, int flags
);
2249 extern void ext4_ext_truncate(struct inode
*);
2250 extern int ext4_ext_punch_hole(struct file
*file
, loff_t offset
,
2252 extern void ext4_ext_init(struct super_block
*);
2253 extern void ext4_ext_release(struct super_block
*);
2254 extern long ext4_fallocate(struct file
*file
, int mode
, loff_t offset
,
2256 extern int ext4_convert_unwritten_extents(struct inode
*inode
, loff_t offset
,
2258 extern int ext4_map_blocks(handle_t
*handle
, struct inode
*inode
,
2259 struct ext4_map_blocks
*map
, int flags
);
2260 extern int ext4_fiemap(struct inode
*inode
, struct fiemap_extent_info
*fieinfo
,
2261 __u64 start
, __u64 len
);
2263 extern int ext4_move_extents(struct file
*o_filp
, struct file
*d_filp
,
2264 __u64 start_orig
, __u64 start_donor
,
2265 __u64 len
, __u64
*moved_len
);
2268 extern int __init
ext4_init_pageio(void);
2269 extern void ext4_exit_pageio(void);
2270 extern void ext4_ioend_wait(struct inode
*);
2271 extern void ext4_free_io_end(ext4_io_end_t
*io
);
2272 extern ext4_io_end_t
*ext4_init_io_end(struct inode
*inode
, gfp_t flags
);
2273 extern int ext4_end_io_nolock(ext4_io_end_t
*io
);
2274 extern void ext4_io_submit(struct ext4_io_submit
*io
);
2275 extern int ext4_bio_write_page(struct ext4_io_submit
*io
,
2278 struct writeback_control
*wbc
);
2281 extern int ext4_multi_mount_protect(struct super_block
*, ext4_fsblk_t
);
2283 /* BH_Uninit flag: blocks are allocated but uninitialized on disk */
2284 enum ext4_state_bits
{
2285 BH_Uninit
/* blocks are allocated but uninitialized on disk */
2286 = BH_JBDPrivateStart
,
2289 BUFFER_FNS(Uninit
, uninit
)
2290 TAS_BUFFER_FNS(Uninit
, uninit
)
2293 * Add new method to test wether block and inode bitmaps are properly
2294 * initialized. With uninit_bg reading the block from disk is not enough
2295 * to mark the bitmap uptodate. We need to also zero-out the bitmap
2297 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
2299 static inline int bitmap_uptodate(struct buffer_head
*bh
)
2301 return (buffer_uptodate(bh
) &&
2302 test_bit(BH_BITMAP_UPTODATE
, &(bh
)->b_state
));
2304 static inline void set_bitmap_uptodate(struct buffer_head
*bh
)
2306 set_bit(BH_BITMAP_UPTODATE
, &(bh
)->b_state
);
2309 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
2311 /* For ioend & aio unwritten conversion wait queues */
2312 #define EXT4_WQ_HASH_SZ 37
2313 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
2315 #define ext4_aio_mutex(v) (&ext4__aio_mutex[((unsigned long)(v)) %\
2317 extern wait_queue_head_t ext4__ioend_wq
[EXT4_WQ_HASH_SZ
];
2318 extern struct mutex ext4__aio_mutex
[EXT4_WQ_HASH_SZ
];
2320 #define EXT4_RESIZING 0
2321 extern int ext4_resize_begin(struct super_block
*sb
);
2322 extern void ext4_resize_end(struct super_block
*sb
);
2324 #endif /* __KERNEL__ */
2326 #endif /* _EXT4_H */