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>
32 #include <crypto/hash.h>
34 #include <linux/compat.h>
38 * The fourth extended filesystem constants/structures
42 * Define EXT4FS_DEBUG to produce debug messages
50 #define ext4_debug(f, a...) \
52 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
53 __FILE__, __LINE__, __func__); \
54 printk(KERN_DEBUG f, ## a); \
57 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
61 * Turn on EXT_DEBUG to get lots of info about extents operations.
65 #define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__)
67 #define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
70 #define EXT4_ERROR_INODE(inode, fmt, a...) \
71 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
73 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
74 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
76 #define EXT4_ERROR_FILE(file, block, fmt, a...) \
77 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
79 /* data type for block offset of block group */
80 typedef int ext4_grpblk_t;
82 /* data type for filesystem-wide blocks number */
83 typedef unsigned long long ext4_fsblk_t;
85 /* data type for file logical block number */
86 typedef __u32 ext4_lblk_t;
88 /* data type for block group number */
89 typedef unsigned int ext4_group_t;
92 * Flags used in mballoc's allocation_context flags field.
94 * Also used to show what's going on for debugging purposes when the
95 * flag field is exported via the traceport interface
98 /* prefer goal again. length */
99 #define EXT4_MB_HINT_MERGE 0x0001
100 /* blocks already reserved */
101 #define EXT4_MB_HINT_RESERVED 0x0002
102 /* metadata is being allocated */
103 #define EXT4_MB_HINT_METADATA 0x0004
104 /* first blocks in the file */
105 #define EXT4_MB_HINT_FIRST 0x0008
106 /* search for the best chunk */
107 #define EXT4_MB_HINT_BEST 0x0010
108 /* data is being allocated */
109 #define EXT4_MB_HINT_DATA 0x0020
110 /* don't preallocate (for tails) */
111 #define EXT4_MB_HINT_NOPREALLOC 0x0040
112 /* allocate for locality group */
113 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080
114 /* allocate goal blocks or none */
115 #define EXT4_MB_HINT_GOAL_ONLY 0x0100
116 /* goal is meaningful */
117 #define EXT4_MB_HINT_TRY_GOAL 0x0200
118 /* blocks already pre-reserved by delayed allocation */
119 #define EXT4_MB_DELALLOC_RESERVED 0x0400
120 /* We are doing stream allocation */
121 #define EXT4_MB_STREAM_ALLOC 0x0800
122 /* Use reserved root blocks if needed */
123 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000
125 struct ext4_allocation_request {
126 /* target inode for block we're allocating */
128 /* how many blocks we want to allocate */
130 /* logical block in target inode */
132 /* the closest logical allocated block to the left */
134 /* the closest logical allocated block to the right */
136 /* phys. target (a hint) */
138 /* phys. block for the closest logical allocated block to the left */
140 /* phys. block for the closest logical allocated block to the right */
142 /* flags. see above EXT4_MB_HINT_* */
147 * Logical to physical block mapping, used by ext4_map_blocks()
149 * This structure is used to pass requests into ext4_map_blocks() as
150 * well as to store the information returned by ext4_map_blocks(). It
151 * takes less room on the stack than a struct buffer_head.
153 #define EXT4_MAP_NEW (1 << BH_New)
154 #define EXT4_MAP_MAPPED (1 << BH_Mapped)
155 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten)
156 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary)
157 #define EXT4_MAP_UNINIT (1 << BH_Uninit)
158 /* Sometimes (in the bigalloc case, from ext4_da_get_block_prep) the caller of
159 * ext4_map_blocks wants to know whether or not the underlying cluster has
160 * already been accounted for. EXT4_MAP_FROM_CLUSTER conveys to the caller that
161 * the requested mapping was from previously mapped (or delayed allocated)
162 * cluster. We use BH_AllocFromCluster only for this flag. BH_AllocFromCluster
163 * should never appear on buffer_head's state flags.
165 #define EXT4_MAP_FROM_CLUSTER (1 << BH_AllocFromCluster)
166 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
167 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\
168 EXT4_MAP_UNINIT | EXT4_MAP_FROM_CLUSTER)
170 struct ext4_map_blocks {
174 unsigned int m_flags;
178 * For delayed allocation tracking
180 struct mpage_da_data {
182 sector_t b_blocknr; /* start block number of extent */
183 size_t b_size; /* size of extent */
184 unsigned long b_state; /* state of the extent */
185 unsigned long first_page, next_page; /* extent of pages */
186 struct writeback_control *wbc;
193 * Flags for ext4_io_end->flags
195 #define EXT4_IO_END_UNWRITTEN 0x0001
196 #define EXT4_IO_END_ERROR 0x0002
197 #define EXT4_IO_END_DIRECT 0x0004
199 struct ext4_io_page {
204 #define MAX_IO_PAGES 128
207 * For converting uninitialized extents on a work queue.
209 * 'page' is only used from the writepage() path; 'pages' is only used for
210 * buffered writes; they are used to keep page references until conversion
211 * takes place. For AIO/DIO, neither field is filled in.
213 typedef struct ext4_io_end {
214 struct list_head list; /* per-file finished IO list */
215 struct inode *inode; /* file being written to */
216 unsigned int flag; /* unwritten or not */
217 loff_t offset; /* offset in the file */
218 ssize_t size; /* size of the extent */
219 struct kiocb *iocb; /* iocb struct for AIO */
220 int result; /* error value for AIO */
221 int num_io_pages; /* for writepages() */
222 struct ext4_io_page *pages[MAX_IO_PAGES]; /* for writepages() */
225 struct ext4_io_submit {
228 ext4_io_end_t *io_end;
229 struct ext4_io_page *io_page;
230 sector_t io_next_block;
234 * Special inodes numbers
236 #define EXT4_BAD_INO 1 /* Bad blocks inode */
237 #define EXT4_ROOT_INO 2 /* Root inode */
238 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */
239 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
240 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
241 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
242 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
243 #define EXT4_JOURNAL_INO 8 /* Journal inode */
245 /* First non-reserved inode for old ext4 filesystems */
246 #define EXT4_GOOD_OLD_FIRST_INO 11
249 * Maximal count of links to a file
251 #define EXT4_LINK_MAX 65000
254 * Macro-instructions used to manage several block sizes
256 #define EXT4_MIN_BLOCK_SIZE 1024
257 #define EXT4_MAX_BLOCK_SIZE 65536
258 #define EXT4_MIN_BLOCK_LOG_SIZE 10
259 #define EXT4_MAX_BLOCK_LOG_SIZE 16
261 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
263 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
265 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
266 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
267 EXT4_SB(s)->s_cluster_bits)
269 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
270 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
272 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
275 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
276 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
277 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
279 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
280 EXT4_GOOD_OLD_INODE_SIZE : \
282 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
283 EXT4_GOOD_OLD_FIRST_INO : \
286 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
288 /* Translate a block number to a cluster number */
289 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
290 /* Translate a cluster number to a block number */
291 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
292 /* Translate # of blks to # of clusters */
293 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
294 (sbi)->s_cluster_bits)
297 * Structure of a blocks group descriptor
299 struct ext4_group_desc
301 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
302 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
303 __le32 bg_inode_table_lo; /* Inodes table block */
304 __le16 bg_free_blocks_count_lo;/* Free blocks count */
305 __le16 bg_free_inodes_count_lo;/* Free inodes count */
306 __le16 bg_used_dirs_count_lo; /* Directories count */
307 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
308 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
309 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
310 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
311 __le16 bg_itable_unused_lo; /* Unused inodes count */
312 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
313 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
314 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
315 __le32 bg_inode_table_hi; /* Inodes table block MSB */
316 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
317 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
318 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
319 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
320 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
321 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
322 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
326 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
327 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
329 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
330 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
334 * Structure of a flex block group info
338 atomic_t free_inodes;
339 atomic_t free_clusters;
343 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
344 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
345 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
348 * Macro-instructions used to manage group descriptors
350 #define EXT4_MIN_DESC_SIZE 32
351 #define EXT4_MIN_DESC_SIZE_64BIT 64
352 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
353 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
355 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
356 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
357 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
358 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
359 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
361 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
362 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
363 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
367 * Constants relative to the data blocks
369 #define EXT4_NDIR_BLOCKS 12
370 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
371 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
372 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
373 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
378 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
379 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
380 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
381 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
382 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
383 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
384 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
385 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
386 /* Reserved for compression usage... */
387 #define EXT4_DIRTY_FL 0x00000100
388 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
389 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
390 #define EXT4_ECOMPR_FL 0x00000800 /* Compression error */
391 /* End compression flags --- maybe not all used */
392 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
393 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
394 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
395 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
396 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
397 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
398 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
399 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
400 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
401 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
402 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
403 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
405 #define EXT4_FL_USER_VISIBLE 0x004BDFFF /* User visible flags */
406 #define EXT4_FL_USER_MODIFIABLE 0x004B80FF /* User modifiable flags */
408 /* Flags that should be inherited by new inodes from their parent. */
409 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
410 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
411 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
412 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
414 /* Flags that are appropriate for regular files (all but dir-specific ones). */
415 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
417 /* Flags that are appropriate for non-directories/regular files. */
418 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
420 /* Mask out flags that are inappropriate for the given type of inode. */
421 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
425 else if (S_ISREG(mode))
426 return flags & EXT4_REG_FLMASK;
428 return flags & EXT4_OTHER_FLMASK;
432 * Inode flags used for atomic set/get
435 EXT4_INODE_SECRM = 0, /* Secure deletion */
436 EXT4_INODE_UNRM = 1, /* Undelete */
437 EXT4_INODE_COMPR = 2, /* Compress file */
438 EXT4_INODE_SYNC = 3, /* Synchronous updates */
439 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
440 EXT4_INODE_APPEND = 5, /* writes to file may only append */
441 EXT4_INODE_NODUMP = 6, /* do not dump file */
442 EXT4_INODE_NOATIME = 7, /* do not update atime */
443 /* Reserved for compression usage... */
444 EXT4_INODE_DIRTY = 8,
445 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
446 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
447 EXT4_INODE_ECOMPR = 11, /* Compression error */
448 /* End compression flags --- maybe not all used */
449 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
450 EXT4_INODE_IMAGIC = 13, /* AFS directory */
451 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
452 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
453 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
454 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
455 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
456 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
457 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
458 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
459 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
460 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
464 * Since it's pretty easy to mix up bit numbers and hex values, we use a
465 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
466 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
467 * any extra space in the compiled kernel image, otherwise, the build will fail.
468 * It's important that these values are the same, since we are using
469 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
470 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
471 * values found in ext2, ext3 and ext4 filesystems, and of course the values
472 * defined in e2fsprogs.
474 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
476 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
477 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
479 static inline void ext4_check_flag_values(void)
481 CHECK_FLAG_VALUE(SECRM);
482 CHECK_FLAG_VALUE(UNRM);
483 CHECK_FLAG_VALUE(COMPR);
484 CHECK_FLAG_VALUE(SYNC);
485 CHECK_FLAG_VALUE(IMMUTABLE);
486 CHECK_FLAG_VALUE(APPEND);
487 CHECK_FLAG_VALUE(NODUMP);
488 CHECK_FLAG_VALUE(NOATIME);
489 CHECK_FLAG_VALUE(DIRTY);
490 CHECK_FLAG_VALUE(COMPRBLK);
491 CHECK_FLAG_VALUE(NOCOMPR);
492 CHECK_FLAG_VALUE(ECOMPR);
493 CHECK_FLAG_VALUE(INDEX);
494 CHECK_FLAG_VALUE(IMAGIC);
495 CHECK_FLAG_VALUE(JOURNAL_DATA);
496 CHECK_FLAG_VALUE(NOTAIL);
497 CHECK_FLAG_VALUE(DIRSYNC);
498 CHECK_FLAG_VALUE(TOPDIR);
499 CHECK_FLAG_VALUE(HUGE_FILE);
500 CHECK_FLAG_VALUE(EXTENTS);
501 CHECK_FLAG_VALUE(EA_INODE);
502 CHECK_FLAG_VALUE(EOFBLOCKS);
503 CHECK_FLAG_VALUE(INLINE_DATA);
504 CHECK_FLAG_VALUE(RESERVED);
507 /* Used to pass group descriptor data when online resize is done */
508 struct ext4_new_group_input {
509 __u32 group; /* Group number for this data */
510 __u64 block_bitmap; /* Absolute block number of block bitmap */
511 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
512 __u64 inode_table; /* Absolute block number of inode table start */
513 __u32 blocks_count; /* Total number of blocks in this group */
514 __u16 reserved_blocks; /* Number of reserved blocks in this group */
518 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
519 struct compat_ext4_new_group_input {
521 compat_u64 block_bitmap;
522 compat_u64 inode_bitmap;
523 compat_u64 inode_table;
530 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
531 struct ext4_new_group_data {
537 __u16 reserved_blocks;
539 __u32 free_blocks_count;
542 /* Indexes used to index group tables in ext4_new_group_data */
544 BLOCK_BITMAP = 0, /* block bitmap */
545 INODE_BITMAP, /* inode bitmap */
546 INODE_TABLE, /* inode tables */
551 * Flags used by ext4_map_blocks()
553 /* Allocate any needed blocks and/or convert an unitialized
554 extent to be an initialized ext4 */
555 #define EXT4_GET_BLOCKS_CREATE 0x0001
556 /* Request the creation of an unitialized extent */
557 #define EXT4_GET_BLOCKS_UNINIT_EXT 0x0002
558 #define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT (EXT4_GET_BLOCKS_UNINIT_EXT|\
559 EXT4_GET_BLOCKS_CREATE)
560 /* Caller is from the delayed allocation writeout path,
561 so set the magic i_delalloc_reserve_flag after taking the
562 inode allocation semaphore for */
563 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
564 /* caller is from the direct IO path, request to creation of an
565 unitialized extents if not allocated, split the uninitialized
566 extent if blocks has been preallocated already*/
567 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
568 #define EXT4_GET_BLOCKS_CONVERT 0x0010
569 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
570 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
571 /* Convert extent to initialized after IO complete */
572 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
573 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
574 /* Punch out blocks of an extent */
575 #define EXT4_GET_BLOCKS_PUNCH_OUT_EXT 0x0020
576 /* Don't normalize allocation size (used for fallocate) */
577 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
578 /* Request will not result in inode size update (user for fallocate) */
579 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080
580 /* Do not take i_data_sem locking in ext4_map_blocks */
581 #define EXT4_GET_BLOCKS_NO_LOCK 0x0100
582 /* Do not put hole in extent cache */
583 #define EXT4_GET_BLOCKS_NO_PUT_HOLE 0x0200
586 * Flags used by ext4_free_blocks
588 #define EXT4_FREE_BLOCKS_METADATA 0x0001
589 #define EXT4_FREE_BLOCKS_FORGET 0x0002
590 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
591 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
592 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
593 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
596 * Flags used by ext4_discard_partial_page_buffers
598 #define EXT4_DISCARD_PARTIAL_PG_ZERO_UNMAPPED 0x0001
603 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
604 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
605 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
606 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
607 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
608 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
609 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
610 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
611 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
612 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
613 #define EXT4_IOC_MIGRATE _IO('f', 9)
614 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
615 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
616 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
617 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
618 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
620 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
622 * ioctl commands in 32 bit emulation
624 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
625 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
626 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
627 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
628 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
629 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
630 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
631 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
632 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
633 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
636 /* Max physical block we can address w/o extents */
637 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
640 * Structure of an inode on the disk
643 __le16 i_mode; /* File mode */
644 __le16 i_uid; /* Low 16 bits of Owner Uid */
645 __le32 i_size_lo; /* Size in bytes */
646 __le32 i_atime; /* Access time */
647 __le32 i_ctime; /* Inode Change time */
648 __le32 i_mtime; /* Modification time */
649 __le32 i_dtime; /* Deletion Time */
650 __le16 i_gid; /* Low 16 bits of Group Id */
651 __le16 i_links_count; /* Links count */
652 __le32 i_blocks_lo; /* Blocks count */
653 __le32 i_flags; /* File flags */
659 __u32 h_i_translator;
664 } osd1; /* OS dependent 1 */
665 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
666 __le32 i_generation; /* File version (for NFS) */
667 __le32 i_file_acl_lo; /* File ACL */
669 __le32 i_obso_faddr; /* Obsoleted fragment address */
672 __le16 l_i_blocks_high; /* were l_i_reserved1 */
673 __le16 l_i_file_acl_high;
674 __le16 l_i_uid_high; /* these 2 fields */
675 __le16 l_i_gid_high; /* were reserved2[0] */
676 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
680 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
687 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
688 __le16 m_i_file_acl_high;
689 __u32 m_i_reserved2[2];
691 } osd2; /* OS dependent 2 */
692 __le16 i_extra_isize;
693 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
694 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
695 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
696 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
697 __le32 i_crtime; /* File Creation time */
698 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
699 __le32 i_version_hi; /* high 32 bits for 64-bit version */
703 __u32 reserved; /* should be zero */
704 __u32 donor_fd; /* donor file descriptor */
705 __u64 orig_start; /* logical start offset in block for orig */
706 __u64 donor_start; /* logical start offset in block for donor */
707 __u64 len; /* block length to be moved */
708 __u64 moved_len; /* moved block length */
711 #define EXT4_EPOCH_BITS 2
712 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
713 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
716 * Extended fields will fit into an inode if the filesystem was formatted
717 * with large inodes (-I 256 or larger) and there are not currently any EAs
718 * consuming all of the available space. For new inodes we always reserve
719 * enough space for the kernel's known extended fields, but for inodes
720 * created with an old kernel this might not have been the case. None of
721 * the extended inode fields is critical for correct filesystem operation.
722 * This macro checks if a certain field fits in the inode. Note that
723 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
725 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
726 ((offsetof(typeof(*ext4_inode), field) + \
727 sizeof((ext4_inode)->field)) \
728 <= (EXT4_GOOD_OLD_INODE_SIZE + \
729 (einode)->i_extra_isize)) \
731 static inline __le32 ext4_encode_extra_time(struct timespec *time)
733 return cpu_to_le32((sizeof(time->tv_sec) > 4 ?
734 (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) |
735 ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK));
738 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
740 if (sizeof(time->tv_sec) > 4)
741 time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK)
743 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
746 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
748 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
749 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
750 (raw_inode)->xtime ## _extra = \
751 ext4_encode_extra_time(&(inode)->xtime); \
754 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
756 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
757 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
758 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
759 (raw_inode)->xtime ## _extra = \
760 ext4_encode_extra_time(&(einode)->xtime); \
763 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
765 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
766 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
767 ext4_decode_extra_time(&(inode)->xtime, \
768 raw_inode->xtime ## _extra); \
770 (inode)->xtime.tv_nsec = 0; \
773 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
775 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
776 (einode)->xtime.tv_sec = \
777 (signed)le32_to_cpu((raw_inode)->xtime); \
778 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
779 ext4_decode_extra_time(&(einode)->xtime, \
780 raw_inode->xtime ## _extra); \
782 (einode)->xtime.tv_nsec = 0; \
785 #define i_disk_version osd1.linux1.l_i_version
787 #if defined(__KERNEL__) || defined(__linux__)
788 #define i_reserved1 osd1.linux1.l_i_reserved1
789 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
790 #define i_blocks_high osd2.linux2.l_i_blocks_high
791 #define i_uid_low i_uid
792 #define i_gid_low i_gid
793 #define i_uid_high osd2.linux2.l_i_uid_high
794 #define i_gid_high osd2.linux2.l_i_gid_high
795 #define i_checksum_lo osd2.linux2.l_i_checksum_lo
797 #elif defined(__GNU__)
799 #define i_translator osd1.hurd1.h_i_translator
800 #define i_uid_high osd2.hurd2.h_i_uid_high
801 #define i_gid_high osd2.hurd2.h_i_gid_high
802 #define i_author osd2.hurd2.h_i_author
804 #elif defined(__masix__)
806 #define i_reserved1 osd1.masix1.m_i_reserved1
807 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
808 #define i_reserved2 osd2.masix2.m_i_reserved2
810 #endif /* defined(__KERNEL__) || defined(__linux__) */
812 #include "extents_status.h"
815 * fourth extended file system inode data in memory
817 struct ext4_inode_info {
818 __le32 i_data[15]; /* unconverted */
820 ext4_fsblk_t i_file_acl;
823 * i_block_group is the number of the block group which contains
824 * this file's inode. Constant across the lifetime of the inode,
825 * it is ued for making block allocation decisions - we try to
826 * place a file's data blocks near its inode block, and new inodes
827 * near to their parent directory's inode.
829 ext4_group_t i_block_group;
830 ext4_lblk_t i_dir_start_lookup;
831 #if (BITS_PER_LONG < 64)
832 unsigned long i_state_flags; /* Dynamic state flags */
834 unsigned long i_flags;
837 * Extended attributes can be read independently of the main file
838 * data. Taking i_mutex even when reading would cause contention
839 * between readers of EAs and writers of regular file data, so
840 * instead we synchronize on xattr_sem when reading or changing
843 struct rw_semaphore xattr_sem;
845 struct list_head i_orphan; /* unlinked but open inodes */
848 * i_disksize keeps track of what the inode size is ON DISK, not
849 * in memory. During truncate, i_size is set to the new size by
850 * the VFS prior to calling ext4_truncate(), but the filesystem won't
851 * set i_disksize to 0 until the truncate is actually under way.
853 * The intent is that i_disksize always represents the blocks which
854 * are used by this file. This allows recovery to restart truncate
855 * on orphans if we crash during truncate. We actually write i_disksize
856 * into the on-disk inode when writing inodes out, instead of i_size.
858 * The only time when i_disksize and i_size may be different is when
859 * a truncate is in progress. The only things which change i_disksize
860 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
865 * i_data_sem is for serialising ext4_truncate() against
866 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
867 * data tree are chopped off during truncate. We can't do that in
868 * ext4 because whenever we perform intermediate commits during
869 * truncate, the inode and all the metadata blocks *must* be in a
870 * consistent state which allows truncation of the orphans to restart
871 * during recovery. Hence we must fix the get_block-vs-truncate race
872 * by other means, so we have i_data_sem.
874 struct rw_semaphore i_data_sem;
875 struct inode vfs_inode;
876 struct jbd2_inode *jinode;
879 * File creation time. Its function is same as that of
880 * struct timespec i_{a,c,m}time in the generic inode.
882 struct timespec i_crtime;
885 struct list_head i_prealloc_list;
886 spinlock_t i_prealloc_lock;
888 /* extents status tree */
889 struct ext4_es_tree i_es_tree;
893 ext4_group_t i_last_alloc_group;
895 /* allocation reservation info for delalloc */
896 /* In case of bigalloc, these refer to clusters rather than blocks */
897 unsigned int i_reserved_data_blocks;
898 unsigned int i_reserved_meta_blocks;
899 unsigned int i_allocated_meta_blocks;
900 ext4_lblk_t i_da_metadata_calc_last_lblock;
901 int i_da_metadata_calc_len;
903 /* on-disk additional length */
906 /* Indicate the inline data space. */
911 /* quota space reservation, managed internally by quota code */
912 qsize_t i_reserved_quota;
915 /* completed IOs that might need unwritten extents handling */
916 struct list_head i_completed_io_list;
917 spinlock_t i_completed_io_lock;
918 atomic_t i_ioend_count; /* Number of outstanding io_end structs */
919 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
920 struct work_struct i_unwritten_work; /* deferred extent conversion */
922 spinlock_t i_block_reservation_lock;
925 * Transactions that contain inode's metadata needed to complete
926 * fsync and fdatasync, respectively.
929 tid_t i_datasync_tid;
931 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
938 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
939 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
940 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
943 * Misc. filesystem flags
945 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
946 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
947 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
952 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
953 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
954 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
955 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
956 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
957 #define EXT4_MOUNT_ERRORS_MASK 0x00070
958 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
959 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
960 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
961 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
962 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
963 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
964 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
965 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
966 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
967 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
968 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
969 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
970 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
971 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
972 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
973 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
974 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
975 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
976 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
977 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
978 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
979 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
980 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
982 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
983 specified delalloc */
985 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
987 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
989 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
992 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
994 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
996 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
999 #define ext4_test_and_set_bit __test_and_set_bit_le
1000 #define ext4_set_bit __set_bit_le
1001 #define ext4_set_bit_atomic ext2_set_bit_atomic
1002 #define ext4_test_and_clear_bit __test_and_clear_bit_le
1003 #define ext4_clear_bit __clear_bit_le
1004 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
1005 #define ext4_test_bit test_bit_le
1006 #define ext4_find_next_zero_bit find_next_zero_bit_le
1007 #define ext4_find_next_bit find_next_bit_le
1009 extern void ext4_set_bits(void *bm, int cur, int len);
1012 * Maximal mount counts between two filesystem checks
1014 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1015 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1018 * Behaviour when detecting errors
1020 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1021 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1022 #define EXT4_ERRORS_PANIC 3 /* Panic */
1023 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1025 /* Metadata checksum algorithm codes */
1026 #define EXT4_CRC32C_CHKSUM 1
1029 * Structure of the super block
1031 struct ext4_super_block {
1032 /*00*/ __le32 s_inodes_count; /* Inodes count */
1033 __le32 s_blocks_count_lo; /* Blocks count */
1034 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1035 __le32 s_free_blocks_count_lo; /* Free blocks count */
1036 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1037 __le32 s_first_data_block; /* First Data Block */
1038 __le32 s_log_block_size; /* Block size */
1039 __le32 s_log_cluster_size; /* Allocation cluster size */
1040 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1041 __le32 s_clusters_per_group; /* # Clusters per group */
1042 __le32 s_inodes_per_group; /* # Inodes per group */
1043 __le32 s_mtime; /* Mount time */
1044 /*30*/ __le32 s_wtime; /* Write time */
1045 __le16 s_mnt_count; /* Mount count */
1046 __le16 s_max_mnt_count; /* Maximal mount count */
1047 __le16 s_magic; /* Magic signature */
1048 __le16 s_state; /* File system state */
1049 __le16 s_errors; /* Behaviour when detecting errors */
1050 __le16 s_minor_rev_level; /* minor revision level */
1051 /*40*/ __le32 s_lastcheck; /* time of last check */
1052 __le32 s_checkinterval; /* max. time between checks */
1053 __le32 s_creator_os; /* OS */
1054 __le32 s_rev_level; /* Revision level */
1055 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1056 __le16 s_def_resgid; /* Default gid for reserved blocks */
1058 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1060 * Note: the difference between the compatible feature set and
1061 * the incompatible feature set is that if there is a bit set
1062 * in the incompatible feature set that the kernel doesn't
1063 * know about, it should refuse to mount the filesystem.
1065 * e2fsck's requirements are more strict; if it doesn't know
1066 * about a feature in either the compatible or incompatible
1067 * feature set, it must abort and not try to meddle with
1068 * things it doesn't understand...
1070 __le32 s_first_ino; /* First non-reserved inode */
1071 __le16 s_inode_size; /* size of inode structure */
1072 __le16 s_block_group_nr; /* block group # of this superblock */
1073 __le32 s_feature_compat; /* compatible feature set */
1074 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1075 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1076 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1077 /*78*/ char s_volume_name[16]; /* volume name */
1078 /*88*/ char s_last_mounted[64]; /* directory where last mounted */
1079 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1081 * Performance hints. Directory preallocation should only
1082 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1084 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1085 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1086 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1088 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1090 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1091 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1092 __le32 s_journal_dev; /* device number of journal file */
1093 __le32 s_last_orphan; /* start of list of inodes to delete */
1094 __le32 s_hash_seed[4]; /* HTREE hash seed */
1095 __u8 s_def_hash_version; /* Default hash version to use */
1096 __u8 s_jnl_backup_type;
1097 __le16 s_desc_size; /* size of group descriptor */
1098 /*100*/ __le32 s_default_mount_opts;
1099 __le32 s_first_meta_bg; /* First metablock block group */
1100 __le32 s_mkfs_time; /* When the filesystem was created */
1101 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1102 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1103 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1104 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1105 __le32 s_free_blocks_count_hi; /* Free blocks count */
1106 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1107 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1108 __le32 s_flags; /* Miscellaneous flags */
1109 __le16 s_raid_stride; /* RAID stride */
1110 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1111 __le64 s_mmp_block; /* Block for multi-mount protection */
1112 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1113 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1114 __u8 s_checksum_type; /* metadata checksum algorithm used */
1115 __le16 s_reserved_pad;
1116 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1117 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1118 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1119 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1120 snapshot's future use */
1121 __le32 s_snapshot_list; /* inode number of the head of the
1122 on-disk snapshot list */
1123 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1124 __le32 s_error_count; /* number of fs errors */
1125 __le32 s_first_error_time; /* first time an error happened */
1126 __le32 s_first_error_ino; /* inode involved in first error */
1127 __le64 s_first_error_block; /* block involved of first error */
1128 __u8 s_first_error_func[32]; /* function where the error happened */
1129 __le32 s_first_error_line; /* line number where error happened */
1130 __le32 s_last_error_time; /* most recent time of an error */
1131 __le32 s_last_error_ino; /* inode involved in last error */
1132 __le32 s_last_error_line; /* line number where error happened */
1133 __le64 s_last_error_block; /* block involved of last error */
1134 __u8 s_last_error_func[32]; /* function where the error happened */
1135 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1136 __u8 s_mount_opts[64];
1137 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1138 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1139 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1140 __le32 s_reserved[108]; /* Padding to the end of the block */
1141 __le32 s_checksum; /* crc32c(superblock) */
1144 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1149 * run-time mount flags
1151 #define EXT4_MF_MNTDIR_SAMPLED 0x0001
1152 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1155 * fourth extended-fs super-block data in memory
1157 struct ext4_sb_info {
1158 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1159 unsigned long s_inodes_per_block;/* Number of inodes per block */
1160 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1161 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1162 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1163 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1164 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1165 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1166 ext4_group_t s_groups_count; /* Number of groups in the fs */
1167 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1168 unsigned long s_overhead; /* # of fs overhead clusters */
1169 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1170 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1171 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1172 struct buffer_head * s_sbh; /* Buffer containing the super block */
1173 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1174 struct buffer_head **s_group_desc;
1175 unsigned int s_mount_opt;
1176 unsigned int s_mount_opt2;
1177 unsigned int s_mount_flags;
1178 unsigned int s_def_mount_opt;
1179 ext4_fsblk_t s_sb_block;
1182 unsigned short s_mount_state;
1183 unsigned short s_pad;
1184 int s_addr_per_block_bits;
1185 int s_desc_per_block_bits;
1188 unsigned int s_inode_readahead_blks;
1189 unsigned int s_inode_goal;
1190 spinlock_t s_next_gen_lock;
1191 u32 s_next_generation;
1193 int s_def_hash_version;
1194 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1195 struct percpu_counter s_freeclusters_counter;
1196 struct percpu_counter s_freeinodes_counter;
1197 struct percpu_counter s_dirs_counter;
1198 struct percpu_counter s_dirtyclusters_counter;
1199 struct blockgroup_lock *s_blockgroup_lock;
1200 struct proc_dir_entry *s_proc;
1201 struct kobject s_kobj;
1202 struct completion s_kobj_unregister;
1203 struct super_block *s_sb;
1206 struct journal_s *s_journal;
1207 struct list_head s_orphan;
1208 struct mutex s_orphan_lock;
1209 unsigned long s_resize_flags; /* Flags indicating if there
1211 unsigned long s_commit_interval;
1212 u32 s_max_batch_time;
1213 u32 s_min_batch_time;
1214 struct block_device *journal_bdev;
1216 char *s_qf_names[MAXQUOTAS]; /* Names of quota files with journalled quota */
1217 int s_jquota_fmt; /* Format of quota to use */
1219 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1220 struct rb_root system_blks;
1222 #ifdef EXTENTS_STATS
1223 /* ext4 extents stats */
1224 unsigned long s_ext_min;
1225 unsigned long s_ext_max;
1226 unsigned long s_depth_max;
1227 spinlock_t s_ext_stats_lock;
1228 unsigned long s_ext_blocks;
1229 unsigned long s_ext_extents;
1232 /* for buddy allocator */
1233 struct ext4_group_info ***s_group_info;
1234 struct inode *s_buddy_cache;
1235 spinlock_t s_md_lock;
1236 unsigned short *s_mb_offsets;
1237 unsigned int *s_mb_maxs;
1238 unsigned int s_group_info_size;
1241 unsigned long s_stripe;
1242 unsigned int s_mb_stream_request;
1243 unsigned int s_mb_max_to_scan;
1244 unsigned int s_mb_min_to_scan;
1245 unsigned int s_mb_stats;
1246 unsigned int s_mb_order2_reqs;
1247 unsigned int s_mb_group_prealloc;
1248 unsigned int s_max_writeback_mb_bump;
1249 unsigned int s_max_dir_size_kb;
1250 /* where last allocation was done - for stream allocation */
1251 unsigned long s_mb_last_group;
1252 unsigned long s_mb_last_start;
1254 /* stats for buddy allocator */
1255 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1256 atomic_t s_bal_success; /* we found long enough chunks */
1257 atomic_t s_bal_allocated; /* in blocks */
1258 atomic_t s_bal_ex_scanned; /* total extents scanned */
1259 atomic_t s_bal_goals; /* goal hits */
1260 atomic_t s_bal_breaks; /* too long searches */
1261 atomic_t s_bal_2orders; /* 2^order hits */
1262 spinlock_t s_bal_lock;
1263 unsigned long s_mb_buddies_generated;
1264 unsigned long long s_mb_generation_time;
1265 atomic_t s_mb_lost_chunks;
1266 atomic_t s_mb_preallocated;
1267 atomic_t s_mb_discarded;
1268 atomic_t s_lock_busy;
1270 /* locality groups */
1271 struct ext4_locality_group __percpu *s_locality_groups;
1273 /* for write statistics */
1274 unsigned long s_sectors_written_start;
1275 u64 s_kbytes_written;
1277 /* the size of zero-out chunk */
1278 unsigned int s_extent_max_zeroout_kb;
1280 unsigned int s_log_groups_per_flex;
1281 struct flex_groups *s_flex_groups;
1282 ext4_group_t s_flex_groups_allocated;
1284 /* workqueue for dio unwritten */
1285 struct workqueue_struct *dio_unwritten_wq;
1287 /* timer for periodic error stats printing */
1288 struct timer_list s_err_report;
1290 /* Lazy inode table initialization info */
1291 struct ext4_li_request *s_li_request;
1292 /* Wait multiplier for lazy initialization thread */
1293 unsigned int s_li_wait_mult;
1295 /* Kernel thread for multiple mount protection */
1296 struct task_struct *s_mmp_tsk;
1298 /* record the last minlen when FITRIM is called. */
1299 atomic_t s_last_trim_minblks;
1301 /* Reference to checksum algorithm driver via cryptoapi */
1302 struct crypto_shash *s_chksum_driver;
1304 /* Precomputed FS UUID checksum for seeding other checksums */
1308 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1310 return sb->s_fs_info;
1312 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1314 return container_of(inode, struct ext4_inode_info, vfs_inode);
1317 static inline struct timespec ext4_current_time(struct inode *inode)
1319 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1320 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1323 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1325 return ino == EXT4_ROOT_INO ||
1326 ino == EXT4_USR_QUOTA_INO ||
1327 ino == EXT4_GRP_QUOTA_INO ||
1328 ino == EXT4_JOURNAL_INO ||
1329 ino == EXT4_RESIZE_INO ||
1330 (ino >= EXT4_FIRST_INO(sb) &&
1331 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1334 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
1335 struct ext4_io_end *io_end)
1337 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
1338 io_end->flag |= EXT4_IO_END_UNWRITTEN;
1339 atomic_inc(&EXT4_I(inode)->i_unwritten);
1343 static inline ext4_io_end_t *ext4_inode_aio(struct inode *inode)
1345 return inode->i_private;
1348 static inline void ext4_inode_aio_set(struct inode *inode, ext4_io_end_t *io)
1350 inode->i_private = io;
1354 * Inode dynamic state flags
1357 EXT4_STATE_JDATA, /* journaled data exists */
1358 EXT4_STATE_NEW, /* inode is newly created */
1359 EXT4_STATE_XATTR, /* has in-inode xattrs */
1360 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1361 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1362 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1363 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
1364 EXT4_STATE_NEWENTRY, /* File just added to dir */
1365 EXT4_STATE_DELALLOC_RESERVED, /* blks already reserved for delalloc */
1366 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read
1368 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1371 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1372 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1374 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1376 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1378 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1380 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1382 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1385 EXT4_INODE_BIT_FNS(flag, flags, 0)
1386 #if (BITS_PER_LONG < 64)
1387 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1389 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1391 (ei)->i_state_flags = 0;
1394 EXT4_INODE_BIT_FNS(state, flags, 32)
1396 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1398 /* We depend on the fact that callers will set i_flags */
1402 /* Assume that user mode programs are passing in an ext4fs superblock, not
1403 * a kernel struct super_block. This will allow us to call the feature-test
1404 * macros from user land. */
1405 #define EXT4_SB(sb) (sb)
1408 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1411 * Codes for operating systems
1413 #define EXT4_OS_LINUX 0
1414 #define EXT4_OS_HURD 1
1415 #define EXT4_OS_MASIX 2
1416 #define EXT4_OS_FREEBSD 3
1417 #define EXT4_OS_LITES 4
1422 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1423 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1425 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1426 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1428 #define EXT4_GOOD_OLD_INODE_SIZE 128
1431 * Feature set definitions
1434 #define EXT4_HAS_COMPAT_FEATURE(sb,mask) \
1435 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1436 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \
1437 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1438 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \
1439 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1440 #define EXT4_SET_COMPAT_FEATURE(sb,mask) \
1441 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1442 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \
1443 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1444 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \
1445 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1446 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \
1447 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1448 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
1449 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1450 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \
1451 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1453 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1454 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1455 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1456 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1457 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1458 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1460 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1461 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1462 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1463 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1464 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1465 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1466 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1467 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1468 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1470 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1471 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1472 * all other data structures' checksums. However, the METADATA_CSUM and
1473 * GDT_CSUM bits are mutually exclusive.
1475 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
1477 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1478 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1479 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1480 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1481 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1482 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1483 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1484 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1485 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1486 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1487 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1488 #define EXT4_FEATURE_INCOMPAT_BG_USE_META_CSUM 0x2000 /* use crc32c for bg */
1489 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
1490 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
1492 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1493 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1494 EXT4_FEATURE_INCOMPAT_META_BG)
1495 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1496 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1497 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1499 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1500 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1501 EXT4_FEATURE_INCOMPAT_RECOVER| \
1502 EXT4_FEATURE_INCOMPAT_META_BG)
1503 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1504 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1505 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1507 #define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR
1508 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1509 EXT4_FEATURE_INCOMPAT_RECOVER| \
1510 EXT4_FEATURE_INCOMPAT_META_BG| \
1511 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1512 EXT4_FEATURE_INCOMPAT_64BIT| \
1513 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1514 EXT4_FEATURE_INCOMPAT_MMP | \
1515 EXT4_FEATURE_INCOMPAT_INLINE_DATA)
1516 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1517 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1518 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1519 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1520 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1521 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1522 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1523 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1524 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1525 EXT4_FEATURE_RO_COMPAT_QUOTA)
1528 * Default values for user and/or group using reserved blocks
1530 #define EXT4_DEF_RESUID 0
1531 #define EXT4_DEF_RESGID 0
1533 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
1536 * Default mount options
1538 #define EXT4_DEFM_DEBUG 0x0001
1539 #define EXT4_DEFM_BSDGROUPS 0x0002
1540 #define EXT4_DEFM_XATTR_USER 0x0004
1541 #define EXT4_DEFM_ACL 0x0008
1542 #define EXT4_DEFM_UID16 0x0010
1543 #define EXT4_DEFM_JMODE 0x0060
1544 #define EXT4_DEFM_JMODE_DATA 0x0020
1545 #define EXT4_DEFM_JMODE_ORDERED 0x0040
1546 #define EXT4_DEFM_JMODE_WBACK 0x0060
1547 #define EXT4_DEFM_NOBARRIER 0x0100
1548 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1549 #define EXT4_DEFM_DISCARD 0x0400
1550 #define EXT4_DEFM_NODELALLOC 0x0800
1553 * Default journal batch times
1555 #define EXT4_DEF_MIN_BATCH_TIME 0
1556 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1559 * Minimum number of groups in a flexgroup before we separate out
1560 * directories into the first block group of a flexgroup
1562 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1565 * Structure of a directory entry
1567 #define EXT4_NAME_LEN 255
1569 struct ext4_dir_entry {
1570 __le32 inode; /* Inode number */
1571 __le16 rec_len; /* Directory entry length */
1572 __le16 name_len; /* Name length */
1573 char name[EXT4_NAME_LEN]; /* File name */
1577 * The new version of the directory entry. Since EXT4 structures are
1578 * stored in intel byte order, and the name_len field could never be
1579 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1582 struct ext4_dir_entry_2 {
1583 __le32 inode; /* Inode number */
1584 __le16 rec_len; /* Directory entry length */
1585 __u8 name_len; /* Name length */
1587 char name[EXT4_NAME_LEN]; /* File name */
1591 * This is a bogus directory entry at the end of each leaf block that
1592 * records checksums.
1594 struct ext4_dir_entry_tail {
1595 __le32 det_reserved_zero1; /* Pretend to be unused */
1596 __le16 det_rec_len; /* 12 */
1597 __u8 det_reserved_zero2; /* Zero name length */
1598 __u8 det_reserved_ft; /* 0xDE, fake file type */
1599 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
1602 #define EXT4_DIRENT_TAIL(block, blocksize) \
1603 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1605 sizeof(struct ext4_dir_entry_tail))))
1608 * Ext4 directory file types. Only the low 3 bits are used. The
1609 * other bits are reserved for now.
1611 #define EXT4_FT_UNKNOWN 0
1612 #define EXT4_FT_REG_FILE 1
1613 #define EXT4_FT_DIR 2
1614 #define EXT4_FT_CHRDEV 3
1615 #define EXT4_FT_BLKDEV 4
1616 #define EXT4_FT_FIFO 5
1617 #define EXT4_FT_SOCK 6
1618 #define EXT4_FT_SYMLINK 7
1620 #define EXT4_FT_MAX 8
1622 #define EXT4_FT_DIR_CSUM 0xDE
1625 * EXT4_DIR_PAD defines the directory entries boundaries
1627 * NOTE: It must be a multiple of 4
1629 #define EXT4_DIR_PAD 4
1630 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1631 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1633 #define EXT4_MAX_REC_LEN ((1<<16)-1)
1636 * If we ever get support for fs block sizes > page_size, we'll need
1637 * to remove the #if statements in the next two functions...
1639 static inline unsigned int
1640 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1642 unsigned len = le16_to_cpu(dlen);
1644 #if (PAGE_CACHE_SIZE >= 65536)
1645 if (len == EXT4_MAX_REC_LEN || len == 0)
1647 return (len & 65532) | ((len & 3) << 16);
1653 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1655 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1657 #if (PAGE_CACHE_SIZE >= 65536)
1659 return cpu_to_le16(len);
1660 if (len == blocksize) {
1661 if (blocksize == 65536)
1662 return cpu_to_le16(EXT4_MAX_REC_LEN);
1664 return cpu_to_le16(0);
1666 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1668 return cpu_to_le16(len);
1673 * Hash Tree Directory indexing
1674 * (c) Daniel Phillips, 2001
1677 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1678 EXT4_FEATURE_COMPAT_DIR_INDEX) && \
1679 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
1680 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1681 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1683 /* Legal values for the dx_root hash_version field: */
1685 #define DX_HASH_LEGACY 0
1686 #define DX_HASH_HALF_MD4 1
1687 #define DX_HASH_TEA 2
1688 #define DX_HASH_LEGACY_UNSIGNED 3
1689 #define DX_HASH_HALF_MD4_UNSIGNED 4
1690 #define DX_HASH_TEA_UNSIGNED 5
1692 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
1693 const void *address, unsigned int length)
1696 struct shash_desc shash;
1701 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
1703 desc.shash.tfm = sbi->s_chksum_driver;
1704 desc.shash.flags = 0;
1705 *(u32 *)desc.ctx = crc;
1707 err = crypto_shash_update(&desc.shash, address, length);
1710 return *(u32 *)desc.ctx;
1715 /* hash info structure used by the directory hash */
1725 /* 32 and 64 bit signed EOF for dx directories */
1726 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
1727 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
1731 * Control parameters used by ext4_htree_next_block
1733 #define HASH_NB_ALWAYS 1
1737 * Describe an inode's exact location on disk and in memory
1741 struct buffer_head *bh;
1742 unsigned long offset;
1743 ext4_group_t block_group;
1746 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
1748 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
1752 * This structure is stuffed into the struct file's private_data field
1753 * for directories. It is where we put information so that we can do
1754 * readdir operations in hash tree order.
1756 struct dir_private_info {
1757 struct rb_root root;
1758 struct rb_node *curr_node;
1759 struct fname *extra_fname;
1762 __u32 curr_minor_hash;
1766 /* calculate the first block number of the group */
1767 static inline ext4_fsblk_t
1768 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
1770 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
1771 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
1775 * Special error return code only used by dx_probe() and its callers.
1777 #define ERR_BAD_DX_DIR -75000
1779 void ext4_get_group_no_and_offset(struct super_block *sb, ext4_fsblk_t blocknr,
1780 ext4_group_t *blockgrpp, ext4_grpblk_t *offsetp);
1783 * Timeout and state flag for lazy initialization inode thread.
1785 #define EXT4_DEF_LI_WAIT_MULT 10
1786 #define EXT4_DEF_LI_MAX_START_DELAY 5
1787 #define EXT4_LAZYINIT_QUIT 0x0001
1788 #define EXT4_LAZYINIT_RUNNING 0x0002
1791 * Lazy inode table initialization info
1793 struct ext4_lazy_init {
1794 unsigned long li_state;
1795 struct list_head li_request_list;
1796 struct mutex li_list_mtx;
1799 struct ext4_li_request {
1800 struct super_block *lr_super;
1801 struct ext4_sb_info *lr_sbi;
1802 ext4_group_t lr_next_group;
1803 struct list_head lr_request;
1804 unsigned long lr_next_sched;
1805 unsigned long lr_timeout;
1808 struct ext4_features {
1809 struct kobject f_kobj;
1810 struct completion f_kobj_unregister;
1814 * This structure will be used for multiple mount protection. It will be
1815 * written into the block number saved in the s_mmp_block field in the
1816 * superblock. Programs that check MMP should assume that if
1817 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
1818 * to use the filesystem, regardless of how old the timestamp is.
1820 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
1821 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
1822 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
1823 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
1826 __le32 mmp_magic; /* Magic number for MMP */
1827 __le32 mmp_seq; /* Sequence no. updated periodically */
1830 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
1831 * purposes and do not affect the correctness of the algorithm
1833 __le64 mmp_time; /* Time last updated */
1834 char mmp_nodename[64]; /* Node which last updated MMP block */
1835 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
1838 * mmp_check_interval is used to verify if the MMP block has been
1839 * updated on the block device. The value is updated based on the
1840 * maximum time to write the MMP block during an update cycle.
1842 __le16 mmp_check_interval;
1845 __le32 mmp_pad2[226];
1846 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
1849 /* arguments passed to the mmp thread */
1851 struct buffer_head *bh; /* bh from initial read_mmp_block() */
1852 struct super_block *sb; /* super block of the fs */
1856 * Check interval multiplier
1857 * The MMP block is written every update interval and initially checked every
1858 * update interval x the multiplier (the value is then adapted based on the
1859 * write latency). The reason is that writes can be delayed under load and we
1860 * don't want readers to incorrectly assume that the filesystem is no longer
1863 #define EXT4_MMP_CHECK_MULT 2UL
1866 * Minimum interval for MMP checking in seconds.
1868 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
1871 * Maximum interval for MMP checking in seconds.
1873 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
1876 * Function prototypes
1880 * Ok, these declarations are also in <linux/kernel.h> but none of the
1881 * ext4 source programs needs to include it so they are duplicated here.
1883 # define NORET_TYPE /**/
1884 # define ATTRIB_NORET __attribute__((noreturn))
1885 # define NORET_AND noreturn,
1888 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
1889 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
1890 struct ext4_group_desc *gdp,
1891 struct buffer_head *bh, int sz);
1892 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
1893 struct ext4_group_desc *gdp,
1894 struct buffer_head *bh, int sz);
1895 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
1896 struct ext4_group_desc *gdp,
1897 struct buffer_head *bh);
1898 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
1899 struct ext4_group_desc *gdp,
1900 struct buffer_head *bh);
1903 extern void ext4_validate_block_bitmap(struct super_block *sb,
1904 struct ext4_group_desc *desc,
1905 unsigned int block_group,
1906 struct buffer_head *bh);
1907 extern unsigned int ext4_block_group(struct super_block *sb,
1908 ext4_fsblk_t blocknr);
1909 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
1910 ext4_fsblk_t blocknr);
1911 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
1912 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
1913 ext4_group_t group);
1914 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
1917 unsigned long *count,
1919 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
1920 s64 nclusters, unsigned int flags);
1921 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
1922 extern void ext4_check_blocks_bitmap(struct super_block *);
1923 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
1924 ext4_group_t block_group,
1925 struct buffer_head ** bh);
1926 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
1928 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
1929 ext4_group_t block_group);
1930 extern int ext4_wait_block_bitmap(struct super_block *sb,
1931 ext4_group_t block_group,
1932 struct buffer_head *bh);
1933 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
1934 ext4_group_t block_group);
1935 extern void ext4_init_block_bitmap(struct super_block *sb,
1936 struct buffer_head *bh,
1938 struct ext4_group_desc *desc);
1939 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
1940 ext4_group_t block_group,
1941 struct ext4_group_desc *gdp);
1942 extern unsigned ext4_num_overhead_clusters(struct super_block *sb,
1943 ext4_group_t block_group,
1944 struct ext4_group_desc *gdp);
1945 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
1948 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
1950 struct ext4_dir_entry_2 *,
1951 struct buffer_head *, char *, int,
1953 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
1954 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
1955 (de), (bh), (buf), (size), (offset)))
1956 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
1958 struct ext4_dir_entry_2 *dirent);
1959 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
1960 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
1961 struct buffer_head *bh,
1962 void *buf, int buf_size,
1963 const char *name, int namelen,
1964 struct ext4_dir_entry_2 **dest_de);
1965 void ext4_insert_dentry(struct inode *inode,
1966 struct ext4_dir_entry_2 *de,
1968 const char *name, int namelen);
1969 static inline void ext4_update_dx_flag(struct inode *inode)
1971 if (!EXT4_HAS_COMPAT_FEATURE(inode->i_sb,
1972 EXT4_FEATURE_COMPAT_DIR_INDEX))
1973 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
1975 static unsigned char ext4_filetype_table[] = {
1976 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
1979 static inline unsigned char get_dtype(struct super_block *sb, int filetype)
1981 if (!EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE) ||
1982 (filetype >= EXT4_FT_MAX))
1985 return ext4_filetype_table[filetype];
1989 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
1990 extern int ext4_flush_unwritten_io(struct inode *);
1993 extern int ext4fs_dirhash(const char *name, int len, struct
1994 dx_hash_info *hinfo);
1997 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
1998 const struct qstr *qstr, __u32 goal,
1999 uid_t *owner, int handle_type,
2000 unsigned int line_no, int nblocks);
2002 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner) \
2003 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2005 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2007 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2008 (type), __LINE__, (nblocks))
2011 extern void ext4_free_inode(handle_t *, struct inode *);
2012 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2013 extern unsigned long ext4_count_free_inodes(struct super_block *);
2014 extern unsigned long ext4_count_dirs(struct super_block *);
2015 extern void ext4_check_inodes_bitmap(struct super_block *);
2016 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2017 extern int ext4_init_inode_table(struct super_block *sb,
2018 ext4_group_t group, int barrier);
2019 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2022 extern long ext4_mb_stats;
2023 extern long ext4_mb_max_to_scan;
2024 extern int ext4_mb_init(struct super_block *);
2025 extern int ext4_mb_release(struct super_block *);
2026 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2027 struct ext4_allocation_request *, int *);
2028 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2029 extern void ext4_discard_preallocations(struct inode *);
2030 extern int __init ext4_init_mballoc(void);
2031 extern void ext4_exit_mballoc(void);
2032 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2033 struct buffer_head *bh, ext4_fsblk_t block,
2034 unsigned long count, int flags);
2035 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2036 ext4_group_t ngroups);
2037 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2038 ext4_group_t i, struct ext4_group_desc *desc);
2039 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2040 ext4_fsblk_t block, unsigned long count);
2041 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2044 struct buffer_head *ext4_getblk(handle_t *, struct inode *,
2045 ext4_lblk_t, int, int *);
2046 struct buffer_head *ext4_bread(handle_t *, struct inode *,
2047 ext4_lblk_t, int, int *);
2048 int ext4_get_block_write(struct inode *inode, sector_t iblock,
2049 struct buffer_head *bh_result, int create);
2050 int ext4_get_block(struct inode *inode, sector_t iblock,
2051 struct buffer_head *bh_result, int create);
2052 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2053 struct buffer_head *bh, int create);
2054 int ext4_walk_page_buffers(handle_t *handle,
2055 struct buffer_head *head,
2059 int (*fn)(handle_t *handle,
2060 struct buffer_head *bh));
2061 int do_journal_get_write_access(handle_t *handle,
2062 struct buffer_head *bh);
2063 #define FALL_BACK_TO_NONDELALLOC 1
2064 #define CONVERT_INLINE_DATA 2
2066 extern struct inode *ext4_iget(struct super_block *, unsigned long);
2067 extern int ext4_write_inode(struct inode *, struct writeback_control *);
2068 extern int ext4_setattr(struct dentry *, struct iattr *);
2069 extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
2070 struct kstat *stat);
2071 extern void ext4_evict_inode(struct inode *);
2072 extern void ext4_clear_inode(struct inode *);
2073 extern int ext4_sync_inode(handle_t *, struct inode *);
2074 extern void ext4_dirty_inode(struct inode *, int);
2075 extern int ext4_change_inode_journal_flag(struct inode *, int);
2076 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2077 extern int ext4_can_truncate(struct inode *inode);
2078 extern void ext4_truncate(struct inode *);
2079 extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length);
2080 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
2081 extern void ext4_set_inode_flags(struct inode *);
2082 extern void ext4_get_inode_flags(struct ext4_inode_info *);
2083 extern int ext4_alloc_da_blocks(struct inode *inode);
2084 extern void ext4_set_aops(struct inode *inode);
2085 extern int ext4_writepage_trans_blocks(struct inode *);
2086 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2087 extern int ext4_discard_partial_page_buffers(handle_t *handle,
2088 struct address_space *mapping, loff_t from,
2089 loff_t length, int flags);
2090 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
2091 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2092 extern void ext4_da_update_reserve_space(struct inode *inode,
2093 int used, int quota_claim);
2096 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2097 struct ext4_map_blocks *map, int flags);
2098 extern ssize_t ext4_ind_direct_IO(int rw, struct kiocb *iocb,
2099 const struct iovec *iov, loff_t offset,
2100 unsigned long nr_segs);
2101 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2102 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks, int chunk);
2103 extern void ext4_ind_truncate(struct inode *inode);
2104 extern int ext4_ind_punch_hole(struct file *file, loff_t offset, loff_t length);
2107 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2108 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2111 extern int ext4_ext_migrate(struct inode *);
2114 extern int ext4_dirent_csum_verify(struct inode *inode,
2115 struct ext4_dir_entry *dirent);
2116 extern int ext4_orphan_add(handle_t *, struct inode *);
2117 extern int ext4_orphan_del(handle_t *, struct inode *);
2118 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2119 __u32 start_minor_hash, __u32 *next_hash);
2120 extern int search_dir(struct buffer_head *bh,
2124 const struct qstr *d_name,
2125 unsigned int offset,
2126 struct ext4_dir_entry_2 **res_dir);
2127 extern int ext4_generic_delete_entry(handle_t *handle,
2129 struct ext4_dir_entry_2 *de_del,
2130 struct buffer_head *bh,
2136 extern int ext4_group_add(struct super_block *sb,
2137 struct ext4_new_group_data *input);
2138 extern int ext4_group_extend(struct super_block *sb,
2139 struct ext4_super_block *es,
2140 ext4_fsblk_t n_blocks_count);
2141 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2144 extern int ext4_calculate_overhead(struct super_block *sb);
2145 extern int ext4_superblock_csum_verify(struct super_block *sb,
2146 struct ext4_super_block *es);
2147 extern void ext4_superblock_csum_set(struct super_block *sb);
2148 extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2149 extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2150 extern void ext4_kvfree(void *ptr);
2151 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2152 ext4_group_t ngroup);
2153 extern const char *ext4_decode_error(struct super_block *sb, int errno,
2155 extern __printf(4, 5)
2156 void __ext4_error(struct super_block *, const char *, unsigned int,
2158 #define ext4_error(sb, message...) __ext4_error(sb, __func__, \
2159 __LINE__, ## message)
2160 extern __printf(5, 6)
2161 void ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2163 extern __printf(5, 6)
2164 void ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2166 extern void __ext4_std_error(struct super_block *, const char *,
2168 extern __printf(4, 5)
2169 void __ext4_abort(struct super_block *, const char *, unsigned int,
2171 #define ext4_abort(sb, message...) __ext4_abort(sb, __func__, \
2172 __LINE__, ## message)
2173 extern __printf(4, 5)
2174 void __ext4_warning(struct super_block *, const char *, unsigned int,
2176 #define ext4_warning(sb, message...) __ext4_warning(sb, __func__, \
2177 __LINE__, ## message)
2178 extern __printf(3, 4)
2179 void ext4_msg(struct super_block *, const char *, const char *, ...);
2180 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2181 const char *, unsigned int, const char *);
2182 #define dump_mmp_msg(sb, mmp, msg) __dump_mmp_msg(sb, mmp, __func__, \
2184 extern __printf(7, 8)
2185 void __ext4_grp_locked_error(const char *, unsigned int,
2186 struct super_block *, ext4_group_t,
2187 unsigned long, ext4_fsblk_t,
2189 #define ext4_grp_locked_error(sb, grp, message...) \
2190 __ext4_grp_locked_error(__func__, __LINE__, (sb), (grp), ## message)
2191 extern void ext4_update_dynamic_rev(struct super_block *sb);
2192 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2194 extern int ext4_update_rocompat_feature(handle_t *handle,
2195 struct super_block *sb, __u32 rocompat);
2196 extern int ext4_update_incompat_feature(handle_t *handle,
2197 struct super_block *sb, __u32 incompat);
2198 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2199 struct ext4_group_desc *bg);
2200 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2201 struct ext4_group_desc *bg);
2202 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2203 struct ext4_group_desc *bg);
2204 extern __u32 ext4_free_group_clusters(struct super_block *sb,
2205 struct ext4_group_desc *bg);
2206 extern __u32 ext4_free_inodes_count(struct super_block *sb,
2207 struct ext4_group_desc *bg);
2208 extern __u32 ext4_used_dirs_count(struct super_block *sb,
2209 struct ext4_group_desc *bg);
2210 extern __u32 ext4_itable_unused_count(struct super_block *sb,
2211 struct ext4_group_desc *bg);
2212 extern void ext4_block_bitmap_set(struct super_block *sb,
2213 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2214 extern void ext4_inode_bitmap_set(struct super_block *sb,
2215 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2216 extern void ext4_inode_table_set(struct super_block *sb,
2217 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2218 extern void ext4_free_group_clusters_set(struct super_block *sb,
2219 struct ext4_group_desc *bg,
2221 extern void ext4_free_inodes_set(struct super_block *sb,
2222 struct ext4_group_desc *bg, __u32 count);
2223 extern void ext4_used_dirs_set(struct super_block *sb,
2224 struct ext4_group_desc *bg, __u32 count);
2225 extern void ext4_itable_unused_set(struct super_block *sb,
2226 struct ext4_group_desc *bg, __u32 count);
2227 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2228 struct ext4_group_desc *gdp);
2229 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2230 struct ext4_group_desc *gdp);
2231 extern int ext4_register_li_request(struct super_block *sb,
2232 ext4_group_t first_not_zeroed);
2234 static inline int ext4_has_group_desc_csum(struct super_block *sb)
2236 return EXT4_HAS_RO_COMPAT_FEATURE(sb,
2237 EXT4_FEATURE_RO_COMPAT_GDT_CSUM |
2238 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM);
2241 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2243 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2244 le32_to_cpu(es->s_blocks_count_lo);
2247 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2249 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2250 le32_to_cpu(es->s_r_blocks_count_lo);
2253 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2255 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2256 le32_to_cpu(es->s_free_blocks_count_lo);
2259 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2262 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2263 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2266 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2269 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2270 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2273 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2276 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2277 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2280 static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
2282 if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2283 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2284 le32_to_cpu(raw_inode->i_size_lo);
2286 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2289 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2291 raw_inode->i_size_lo = cpu_to_le32(i_size);
2292 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2296 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2299 struct ext4_group_info ***grp_info;
2300 long indexv, indexh;
2301 grp_info = EXT4_SB(sb)->s_group_info;
2302 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2303 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2304 return grp_info[indexv][indexh];
2308 * Reading s_groups_count requires using smp_rmb() afterwards. See
2309 * the locking protocol documented in the comments of ext4_group_add()
2312 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2314 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
2320 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2321 ext4_group_t block_group)
2323 return block_group >> sbi->s_log_groups_per_flex;
2326 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2328 return 1 << sbi->s_log_groups_per_flex;
2331 #define ext4_std_error(sb, errno) \
2334 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
2338 /* Each CPU can accumulate percpu_counter_batch clusters in their local
2339 * counters. So we need to make sure we have free clusters more
2340 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
2342 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2344 #define EXT4_FREECLUSTERS_WATERMARK 0
2347 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2350 * XXX: replace with spinlock if seen contended -bzzz
2352 down_write(&EXT4_I(inode)->i_data_sem);
2353 if (newsize > EXT4_I(inode)->i_disksize)
2354 EXT4_I(inode)->i_disksize = newsize;
2355 up_write(&EXT4_I(inode)->i_data_sem);
2359 struct ext4_group_info {
2360 unsigned long bb_state;
2361 struct rb_root bb_free_root;
2362 ext4_grpblk_t bb_first_free; /* first free block */
2363 ext4_grpblk_t bb_free; /* total free blocks */
2364 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
2365 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
2366 struct list_head bb_prealloc_list;
2370 struct rw_semaphore alloc_sem;
2371 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
2372 * regions, index is order.
2373 * bb_counters[3] = 5 means
2374 * 5 free 8-block regions. */
2377 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
2378 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
2380 #define EXT4_MB_GRP_NEED_INIT(grp) \
2381 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2383 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
2384 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2385 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
2386 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2387 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
2388 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2390 #define EXT4_MAX_CONTENTION 8
2391 #define EXT4_CONTENTION_THRESHOLD 2
2393 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2396 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2400 * Returns true if the filesystem is busy enough that attempts to
2401 * access the block group locks has run into contention.
2403 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2405 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2408 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2410 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2411 if (spin_trylock(lock))
2413 * We're able to grab the lock right away, so drop the
2414 * lock contention counter.
2416 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2419 * The lock is busy, so bump the contention counter,
2420 * and then wait on the spin lock.
2422 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2423 EXT4_MAX_CONTENTION);
2428 static inline void ext4_unlock_group(struct super_block *sb,
2431 spin_unlock(ext4_group_lock_ptr(sb, group));
2435 * Block validity checking
2437 #define ext4_check_indirect_blockref(inode, bh) \
2438 ext4_check_blockref(__func__, __LINE__, inode, \
2439 (__le32 *)(bh)->b_data, \
2440 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2442 #define ext4_ind_check_inode(inode) \
2443 ext4_check_blockref(__func__, __LINE__, inode, \
2444 EXT4_I(inode)->i_data, \
2448 * Inodes and files operations
2452 extern const struct file_operations ext4_dir_operations;
2455 extern const struct inode_operations ext4_file_inode_operations;
2456 extern const struct file_operations ext4_file_operations;
2457 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
2458 extern void ext4_unwritten_wait(struct inode *inode);
2461 extern int ext4_has_inline_data(struct inode *inode);
2462 extern int ext4_get_inline_size(struct inode *inode);
2463 extern int ext4_get_max_inline_size(struct inode *inode);
2464 extern int ext4_find_inline_data_nolock(struct inode *inode);
2465 extern void ext4_write_inline_data(struct inode *inode,
2466 struct ext4_iloc *iloc,
2467 void *buffer, loff_t pos,
2469 extern int ext4_prepare_inline_data(handle_t *handle, struct inode *inode,
2471 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
2473 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
2475 extern int ext4_readpage_inline(struct inode *inode, struct page *page);
2476 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
2477 struct inode *inode,
2478 loff_t pos, unsigned len,
2480 struct page **pagep);
2481 extern int ext4_write_inline_data_end(struct inode *inode,
2482 loff_t pos, unsigned len,
2485 extern struct buffer_head *
2486 ext4_journalled_write_inline_data(struct inode *inode,
2489 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
2490 struct inode *inode,
2491 loff_t pos, unsigned len,
2493 struct page **pagep,
2495 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
2496 unsigned len, unsigned copied,
2498 extern int ext4_try_add_inline_entry(handle_t *handle, struct dentry *dentry,
2499 struct inode *inode);
2500 extern int ext4_try_create_inline_dir(handle_t *handle,
2501 struct inode *parent,
2502 struct inode *inode);
2503 extern int ext4_read_inline_dir(struct file *filp,
2504 void *dirent, filldir_t filldir,
2505 int *has_inline_data);
2506 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
2507 const struct qstr *d_name,
2508 struct ext4_dir_entry_2 **res_dir,
2509 int *has_inline_data);
2510 extern int ext4_delete_inline_entry(handle_t *handle,
2512 struct ext4_dir_entry_2 *de_del,
2513 struct buffer_head *bh,
2514 int *has_inline_data);
2515 extern int empty_inline_dir(struct inode *dir, int *has_inline_data);
2516 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
2517 struct ext4_dir_entry_2 **parent_de,
2519 extern int ext4_inline_data_fiemap(struct inode *inode,
2520 struct fiemap_extent_info *fieinfo,
2522 extern int ext4_try_to_evict_inline_data(handle_t *handle,
2523 struct inode *inode,
2525 extern void ext4_inline_data_truncate(struct inode *inode, int *has_inline);
2527 extern int ext4_convert_inline_data(struct inode *inode);
2530 extern const struct inode_operations ext4_dir_inode_operations;
2531 extern const struct inode_operations ext4_special_inode_operations;
2532 extern struct dentry *ext4_get_parent(struct dentry *child);
2533 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
2534 struct ext4_dir_entry_2 *de,
2535 int blocksize, int csum_size,
2536 unsigned int parent_ino, int dotdot_real_len);
2537 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t,
2538 unsigned int blocksize);
2539 extern int ext4_handle_dirty_dirent_node(handle_t *handle,
2540 struct inode *inode,
2541 struct buffer_head *bh);
2544 extern const struct inode_operations ext4_symlink_inode_operations;
2545 extern const struct inode_operations ext4_fast_symlink_inode_operations;
2547 /* block_validity */
2548 extern void ext4_release_system_zone(struct super_block *sb);
2549 extern int ext4_setup_system_zone(struct super_block *sb);
2550 extern int __init ext4_init_system_zone(void);
2551 extern void ext4_exit_system_zone(void);
2552 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
2553 ext4_fsblk_t start_blk,
2554 unsigned int count);
2555 extern int ext4_check_blockref(const char *, unsigned int,
2556 struct inode *, __le32 *, unsigned int);
2559 struct ext4_ext_path;
2562 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
2563 extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
2564 extern int ext4_ext_index_trans_blocks(struct inode *inode, int nrblocks,
2566 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
2567 struct ext4_map_blocks *map, int flags);
2568 extern void ext4_ext_truncate(struct inode *);
2569 extern int ext4_ext_punch_hole(struct file *file, loff_t offset,
2571 extern void ext4_ext_init(struct super_block *);
2572 extern void ext4_ext_release(struct super_block *);
2573 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
2575 extern int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset,
2577 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
2578 struct ext4_map_blocks *map, int flags);
2579 extern int ext4_ext_calc_metadata_amount(struct inode *inode,
2580 ext4_lblk_t lblocks);
2581 extern int ext4_extent_tree_init(handle_t *, struct inode *);
2582 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
2584 struct ext4_ext_path *path);
2585 extern int ext4_can_extents_be_merged(struct inode *inode,
2586 struct ext4_extent *ex1,
2587 struct ext4_extent *ex2);
2588 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
2589 struct ext4_ext_path *,
2590 struct ext4_extent *, int);
2591 extern struct ext4_ext_path *ext4_ext_find_extent(struct inode *, ext4_lblk_t,
2592 struct ext4_ext_path *);
2593 extern void ext4_ext_drop_refs(struct ext4_ext_path *);
2594 extern int ext4_ext_check_inode(struct inode *inode);
2595 extern int ext4_find_delalloc_range(struct inode *inode,
2596 ext4_lblk_t lblk_start,
2597 ext4_lblk_t lblk_end);
2598 extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk);
2599 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
2600 __u64 start, __u64 len);
2604 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
2605 __u64 start_orig, __u64 start_donor,
2606 __u64 len, __u64 *moved_len);
2609 extern int __init ext4_init_pageio(void);
2610 extern void ext4_add_complete_io(ext4_io_end_t *io_end);
2611 extern void ext4_exit_pageio(void);
2612 extern void ext4_ioend_wait(struct inode *);
2613 extern void ext4_free_io_end(ext4_io_end_t *io);
2614 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
2615 extern void ext4_end_io_work(struct work_struct *work);
2616 extern void ext4_io_submit(struct ext4_io_submit *io);
2617 extern int ext4_bio_write_page(struct ext4_io_submit *io,
2620 struct writeback_control *wbc);
2623 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
2624 extern void ext4_mmp_csum_set(struct super_block *sb, struct mmp_struct *mmp);
2625 extern int ext4_mmp_csum_verify(struct super_block *sb,
2626 struct mmp_struct *mmp);
2628 /* BH_Uninit flag: blocks are allocated but uninitialized on disk */
2629 enum ext4_state_bits {
2630 BH_Uninit /* blocks are allocated but uninitialized on disk */
2631 = BH_JBDPrivateStart,
2632 BH_AllocFromCluster, /* allocated blocks were part of already
2633 * allocated cluster. Note that this flag will
2634 * never, ever appear in a buffer_head's state
2635 * flag. See EXT4_MAP_FROM_CLUSTER to see where
2639 BUFFER_FNS(Uninit, uninit)
2640 TAS_BUFFER_FNS(Uninit, uninit)
2643 * Add new method to test whether block and inode bitmaps are properly
2644 * initialized. With uninit_bg reading the block from disk is not enough
2645 * to mark the bitmap uptodate. We need to also zero-out the bitmap
2647 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
2649 static inline int bitmap_uptodate(struct buffer_head *bh)
2651 return (buffer_uptodate(bh) &&
2652 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
2654 static inline void set_bitmap_uptodate(struct buffer_head *bh)
2656 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
2660 * Disable DIO read nolock optimization, so new dioreaders will be forced
2663 static inline void ext4_inode_block_unlocked_dio(struct inode *inode)
2665 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
2668 static inline void ext4_inode_resume_unlocked_dio(struct inode *inode)
2671 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
2674 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
2676 /* For ioend & aio unwritten conversion wait queues */
2677 #define EXT4_WQ_HASH_SZ 37
2678 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
2680 #define ext4_aio_mutex(v) (&ext4__aio_mutex[((unsigned long)(v)) %\
2682 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
2683 extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ];
2685 #define EXT4_RESIZING 0
2686 extern int ext4_resize_begin(struct super_block *sb);
2687 extern void ext4_resize_end(struct super_block *sb);
2689 #endif /* __KERNEL__ */
2691 #endif /* _EXT4_H */