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_QUEUED 0x0004
198 #define EXT4_IO_END_DIRECT 0x0008
200 struct ext4_io_page {
205 #define MAX_IO_PAGES 128
208 * For converting uninitialized extents on a work queue.
210 * 'page' is only used from the writepage() path; 'pages' is only used for
211 * buffered writes; they are used to keep page references until conversion
212 * takes place. For AIO/DIO, neither field is filled in.
214 typedef struct ext4_io_end {
215 struct list_head list; /* per-file finished IO list */
216 struct inode *inode; /* file being written to */
217 unsigned int flag; /* unwritten or not */
218 struct page *page; /* for writepage() path */
219 loff_t offset; /* offset in the file */
220 ssize_t size; /* size of the extent */
221 struct work_struct work; /* data work queue */
222 struct kiocb *iocb; /* iocb struct for AIO */
223 int result; /* error value for AIO */
224 int num_io_pages; /* for writepages() */
225 struct ext4_io_page *pages[MAX_IO_PAGES]; /* for writepages() */
228 struct ext4_io_submit {
231 ext4_io_end_t *io_end;
232 struct ext4_io_page *io_page;
233 sector_t io_next_block;
237 * Special inodes numbers
239 #define EXT4_BAD_INO 1 /* Bad blocks inode */
240 #define EXT4_ROOT_INO 2 /* Root inode */
241 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */
242 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
243 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
244 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
245 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
246 #define EXT4_JOURNAL_INO 8 /* Journal inode */
248 /* First non-reserved inode for old ext4 filesystems */
249 #define EXT4_GOOD_OLD_FIRST_INO 11
252 * Maximal count of links to a file
254 #define EXT4_LINK_MAX 65000
257 * Macro-instructions used to manage several block sizes
259 #define EXT4_MIN_BLOCK_SIZE 1024
260 #define EXT4_MAX_BLOCK_SIZE 65536
261 #define EXT4_MIN_BLOCK_LOG_SIZE 10
262 #define EXT4_MAX_BLOCK_LOG_SIZE 16
264 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
266 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
268 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
269 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
270 EXT4_SB(s)->s_cluster_bits)
272 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
273 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
275 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
278 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
279 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
280 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
282 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
283 EXT4_GOOD_OLD_INODE_SIZE : \
285 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
286 EXT4_GOOD_OLD_FIRST_INO : \
289 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
291 /* Translate a block number to a cluster number */
292 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
293 /* Translate a cluster number to a block number */
294 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
295 /* Translate # of blks to # of clusters */
296 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
297 (sbi)->s_cluster_bits)
300 * Structure of a blocks group descriptor
302 struct ext4_group_desc
304 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
305 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
306 __le32 bg_inode_table_lo; /* Inodes table block */
307 __le16 bg_free_blocks_count_lo;/* Free blocks count */
308 __le16 bg_free_inodes_count_lo;/* Free inodes count */
309 __le16 bg_used_dirs_count_lo; /* Directories count */
310 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
311 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
312 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
313 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
314 __le16 bg_itable_unused_lo; /* Unused inodes count */
315 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
316 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
317 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
318 __le32 bg_inode_table_hi; /* Inodes table block MSB */
319 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
320 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
321 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
322 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
323 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
324 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
325 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
329 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
330 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
332 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
333 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
337 * Structure of a flex block group info
341 atomic_t free_inodes;
342 atomic_t free_clusters;
346 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
347 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
348 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
351 * Macro-instructions used to manage group descriptors
353 #define EXT4_MIN_DESC_SIZE 32
354 #define EXT4_MIN_DESC_SIZE_64BIT 64
355 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
356 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
358 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
359 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
360 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
361 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
362 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
364 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
365 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
366 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
370 * Constants relative to the data blocks
372 #define EXT4_NDIR_BLOCKS 12
373 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
374 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
375 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
376 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
381 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
382 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
383 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
384 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
385 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
386 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
387 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
388 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
389 /* Reserved for compression usage... */
390 #define EXT4_DIRTY_FL 0x00000100
391 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
392 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
393 #define EXT4_ECOMPR_FL 0x00000800 /* Compression error */
394 /* End compression flags --- maybe not all used */
395 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
396 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
397 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
398 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
399 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
400 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
401 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
402 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
403 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
404 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
405 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
406 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
408 #define EXT4_FL_USER_VISIBLE 0x004BDFFF /* User visible flags */
409 #define EXT4_FL_USER_MODIFIABLE 0x004B80FF /* User modifiable flags */
411 /* Flags that should be inherited by new inodes from their parent. */
412 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
413 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
414 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
415 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
417 /* Flags that are appropriate for regular files (all but dir-specific ones). */
418 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
420 /* Flags that are appropriate for non-directories/regular files. */
421 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
423 /* Mask out flags that are inappropriate for the given type of inode. */
424 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
428 else if (S_ISREG(mode))
429 return flags & EXT4_REG_FLMASK;
431 return flags & EXT4_OTHER_FLMASK;
435 * Inode flags used for atomic set/get
438 EXT4_INODE_SECRM = 0, /* Secure deletion */
439 EXT4_INODE_UNRM = 1, /* Undelete */
440 EXT4_INODE_COMPR = 2, /* Compress file */
441 EXT4_INODE_SYNC = 3, /* Synchronous updates */
442 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
443 EXT4_INODE_APPEND = 5, /* writes to file may only append */
444 EXT4_INODE_NODUMP = 6, /* do not dump file */
445 EXT4_INODE_NOATIME = 7, /* do not update atime */
446 /* Reserved for compression usage... */
447 EXT4_INODE_DIRTY = 8,
448 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
449 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
450 EXT4_INODE_ECOMPR = 11, /* Compression error */
451 /* End compression flags --- maybe not all used */
452 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
453 EXT4_INODE_IMAGIC = 13, /* AFS directory */
454 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
455 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
456 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
457 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
458 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
459 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
460 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
461 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
462 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
463 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
466 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
467 #define CHECK_FLAG_VALUE(FLAG) if (!TEST_FLAG_VALUE(FLAG)) { \
468 printk(KERN_EMERG "EXT4 flag fail: " #FLAG ": %d %d\n", \
469 EXT4_##FLAG##_FL, EXT4_INODE_##FLAG); BUG_ON(1); }
472 * Since it's pretty easy to mix up bit numbers and hex values, and we
473 * can't do a compile-time test for ENUM values, we use a run-time
474 * test to make sure that EXT4_XXX_FL is consistent with respect to
475 * EXT4_INODE_XXX. If all is well the printk and BUG_ON will all drop
476 * out so it won't cost any extra space in the compiled kernel image.
477 * But it's important that these values are the same, since we are
478 * using EXT4_INODE_XXX to test for the flag values, but EXT4_XX_FL
479 * must be consistent with the values of FS_XXX_FL defined in
480 * include/linux/fs.h and the on-disk values found in ext2, ext3, and
481 * ext4 filesystems, and of course the values defined in e2fsprogs.
483 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
485 static inline void ext4_check_flag_values(void)
487 CHECK_FLAG_VALUE(SECRM);
488 CHECK_FLAG_VALUE(UNRM);
489 CHECK_FLAG_VALUE(COMPR);
490 CHECK_FLAG_VALUE(SYNC);
491 CHECK_FLAG_VALUE(IMMUTABLE);
492 CHECK_FLAG_VALUE(APPEND);
493 CHECK_FLAG_VALUE(NODUMP);
494 CHECK_FLAG_VALUE(NOATIME);
495 CHECK_FLAG_VALUE(DIRTY);
496 CHECK_FLAG_VALUE(COMPRBLK);
497 CHECK_FLAG_VALUE(NOCOMPR);
498 CHECK_FLAG_VALUE(ECOMPR);
499 CHECK_FLAG_VALUE(INDEX);
500 CHECK_FLAG_VALUE(IMAGIC);
501 CHECK_FLAG_VALUE(JOURNAL_DATA);
502 CHECK_FLAG_VALUE(NOTAIL);
503 CHECK_FLAG_VALUE(DIRSYNC);
504 CHECK_FLAG_VALUE(TOPDIR);
505 CHECK_FLAG_VALUE(HUGE_FILE);
506 CHECK_FLAG_VALUE(EXTENTS);
507 CHECK_FLAG_VALUE(EA_INODE);
508 CHECK_FLAG_VALUE(EOFBLOCKS);
509 CHECK_FLAG_VALUE(INLINE_DATA);
510 CHECK_FLAG_VALUE(RESERVED);
513 /* Used to pass group descriptor data when online resize is done */
514 struct ext4_new_group_input {
515 __u32 group; /* Group number for this data */
516 __u64 block_bitmap; /* Absolute block number of block bitmap */
517 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
518 __u64 inode_table; /* Absolute block number of inode table start */
519 __u32 blocks_count; /* Total number of blocks in this group */
520 __u16 reserved_blocks; /* Number of reserved blocks in this group */
524 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
525 struct compat_ext4_new_group_input {
527 compat_u64 block_bitmap;
528 compat_u64 inode_bitmap;
529 compat_u64 inode_table;
536 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
537 struct ext4_new_group_data {
543 __u16 reserved_blocks;
545 __u32 free_blocks_count;
548 /* Indexes used to index group tables in ext4_new_group_data */
550 BLOCK_BITMAP = 0, /* block bitmap */
551 INODE_BITMAP, /* inode bitmap */
552 INODE_TABLE, /* inode tables */
557 * Flags used by ext4_map_blocks()
559 /* Allocate any needed blocks and/or convert an unitialized
560 extent to be an initialized ext4 */
561 #define EXT4_GET_BLOCKS_CREATE 0x0001
562 /* Request the creation of an unitialized extent */
563 #define EXT4_GET_BLOCKS_UNINIT_EXT 0x0002
564 #define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT (EXT4_GET_BLOCKS_UNINIT_EXT|\
565 EXT4_GET_BLOCKS_CREATE)
566 /* Caller is from the delayed allocation writeout path,
567 so set the magic i_delalloc_reserve_flag after taking the
568 inode allocation semaphore for */
569 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
570 /* caller is from the direct IO path, request to creation of an
571 unitialized extents if not allocated, split the uninitialized
572 extent if blocks has been preallocated already*/
573 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
574 #define EXT4_GET_BLOCKS_CONVERT 0x0010
575 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
576 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
577 /* Convert extent to initialized after IO complete */
578 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
579 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
580 /* Punch out blocks of an extent */
581 #define EXT4_GET_BLOCKS_PUNCH_OUT_EXT 0x0020
582 /* Don't normalize allocation size (used for fallocate) */
583 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
584 /* Request will not result in inode size update (user for fallocate) */
585 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080
586 /* Do not take i_data_sem locking in ext4_map_blocks */
587 #define EXT4_GET_BLOCKS_NO_LOCK 0x0100
590 * Flags used by ext4_free_blocks
592 #define EXT4_FREE_BLOCKS_METADATA 0x0001
593 #define EXT4_FREE_BLOCKS_FORGET 0x0002
594 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
595 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
596 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
597 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
600 * Flags used by ext4_discard_partial_page_buffers
602 #define EXT4_DISCARD_PARTIAL_PG_ZERO_UNMAPPED 0x0001
607 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
608 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
609 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
610 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
611 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
612 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
613 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
614 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
615 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
616 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
617 #define EXT4_IOC_MIGRATE _IO('f', 9)
618 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
619 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
620 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
621 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
622 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
624 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
626 * ioctl commands in 32 bit emulation
628 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
629 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
630 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
631 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
632 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
633 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
634 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
635 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
636 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
637 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
640 /* Max physical block we can address w/o extents */
641 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
644 * Structure of an inode on the disk
647 __le16 i_mode; /* File mode */
648 __le16 i_uid; /* Low 16 bits of Owner Uid */
649 __le32 i_size_lo; /* Size in bytes */
650 __le32 i_atime; /* Access time */
651 __le32 i_ctime; /* Inode Change time */
652 __le32 i_mtime; /* Modification time */
653 __le32 i_dtime; /* Deletion Time */
654 __le16 i_gid; /* Low 16 bits of Group Id */
655 __le16 i_links_count; /* Links count */
656 __le32 i_blocks_lo; /* Blocks count */
657 __le32 i_flags; /* File flags */
663 __u32 h_i_translator;
668 } osd1; /* OS dependent 1 */
669 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
670 __le32 i_generation; /* File version (for NFS) */
671 __le32 i_file_acl_lo; /* File ACL */
673 __le32 i_obso_faddr; /* Obsoleted fragment address */
676 __le16 l_i_blocks_high; /* were l_i_reserved1 */
677 __le16 l_i_file_acl_high;
678 __le16 l_i_uid_high; /* these 2 fields */
679 __le16 l_i_gid_high; /* were reserved2[0] */
680 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
684 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
691 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
692 __le16 m_i_file_acl_high;
693 __u32 m_i_reserved2[2];
695 } osd2; /* OS dependent 2 */
696 __le16 i_extra_isize;
697 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
698 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
699 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
700 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
701 __le32 i_crtime; /* File Creation time */
702 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
703 __le32 i_version_hi; /* high 32 bits for 64-bit version */
707 __u32 reserved; /* should be zero */
708 __u32 donor_fd; /* donor file descriptor */
709 __u64 orig_start; /* logical start offset in block for orig */
710 __u64 donor_start; /* logical start offset in block for donor */
711 __u64 len; /* block length to be moved */
712 __u64 moved_len; /* moved block length */
715 #define EXT4_EPOCH_BITS 2
716 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
717 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
720 * Extended fields will fit into an inode if the filesystem was formatted
721 * with large inodes (-I 256 or larger) and there are not currently any EAs
722 * consuming all of the available space. For new inodes we always reserve
723 * enough space for the kernel's known extended fields, but for inodes
724 * created with an old kernel this might not have been the case. None of
725 * the extended inode fields is critical for correct filesystem operation.
726 * This macro checks if a certain field fits in the inode. Note that
727 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
729 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
730 ((offsetof(typeof(*ext4_inode), field) + \
731 sizeof((ext4_inode)->field)) \
732 <= (EXT4_GOOD_OLD_INODE_SIZE + \
733 (einode)->i_extra_isize)) \
735 static inline __le32 ext4_encode_extra_time(struct timespec *time)
737 return cpu_to_le32((sizeof(time->tv_sec) > 4 ?
738 (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) |
739 ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK));
742 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
744 if (sizeof(time->tv_sec) > 4)
745 time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK)
747 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
750 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
752 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
753 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
754 (raw_inode)->xtime ## _extra = \
755 ext4_encode_extra_time(&(inode)->xtime); \
758 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
760 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
761 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
762 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
763 (raw_inode)->xtime ## _extra = \
764 ext4_encode_extra_time(&(einode)->xtime); \
767 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
769 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
770 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
771 ext4_decode_extra_time(&(inode)->xtime, \
772 raw_inode->xtime ## _extra); \
774 (inode)->xtime.tv_nsec = 0; \
777 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
779 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
780 (einode)->xtime.tv_sec = \
781 (signed)le32_to_cpu((raw_inode)->xtime); \
782 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
783 ext4_decode_extra_time(&(einode)->xtime, \
784 raw_inode->xtime ## _extra); \
786 (einode)->xtime.tv_nsec = 0; \
789 #define i_disk_version osd1.linux1.l_i_version
791 #if defined(__KERNEL__) || defined(__linux__)
792 #define i_reserved1 osd1.linux1.l_i_reserved1
793 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
794 #define i_blocks_high osd2.linux2.l_i_blocks_high
795 #define i_uid_low i_uid
796 #define i_gid_low i_gid
797 #define i_uid_high osd2.linux2.l_i_uid_high
798 #define i_gid_high osd2.linux2.l_i_gid_high
799 #define i_checksum_lo osd2.linux2.l_i_checksum_lo
801 #elif defined(__GNU__)
803 #define i_translator osd1.hurd1.h_i_translator
804 #define i_uid_high osd2.hurd2.h_i_uid_high
805 #define i_gid_high osd2.hurd2.h_i_gid_high
806 #define i_author osd2.hurd2.h_i_author
808 #elif defined(__masix__)
810 #define i_reserved1 osd1.masix1.m_i_reserved1
811 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
812 #define i_reserved2 osd2.masix2.m_i_reserved2
814 #endif /* defined(__KERNEL__) || defined(__linux__) */
817 * storage for cached extent
818 * If ec_len == 0, then the cache is invalid.
819 * If ec_start == 0, then the cache represents a gap (null mapping)
821 struct ext4_ext_cache {
822 ext4_fsblk_t ec_start;
823 ext4_lblk_t ec_block;
824 __u32 ec_len; /* must be 32bit to return holes */
827 #include "extents_status.h"
830 * fourth extended file system inode data in memory
832 struct ext4_inode_info {
833 __le32 i_data[15]; /* unconverted */
835 ext4_fsblk_t i_file_acl;
838 * i_block_group is the number of the block group which contains
839 * this file's inode. Constant across the lifetime of the inode,
840 * it is ued for making block allocation decisions - we try to
841 * place a file's data blocks near its inode block, and new inodes
842 * near to their parent directory's inode.
844 ext4_group_t i_block_group;
845 ext4_lblk_t i_dir_start_lookup;
846 #if (BITS_PER_LONG < 64)
847 unsigned long i_state_flags; /* Dynamic state flags */
849 unsigned long i_flags;
851 #ifdef CONFIG_EXT4_FS_XATTR
853 * Extended attributes can be read independently of the main file
854 * data. Taking i_mutex even when reading would cause contention
855 * between readers of EAs and writers of regular file data, so
856 * instead we synchronize on xattr_sem when reading or changing
859 struct rw_semaphore xattr_sem;
862 struct list_head i_orphan; /* unlinked but open inodes */
865 * i_disksize keeps track of what the inode size is ON DISK, not
866 * in memory. During truncate, i_size is set to the new size by
867 * the VFS prior to calling ext4_truncate(), but the filesystem won't
868 * set i_disksize to 0 until the truncate is actually under way.
870 * The intent is that i_disksize always represents the blocks which
871 * are used by this file. This allows recovery to restart truncate
872 * on orphans if we crash during truncate. We actually write i_disksize
873 * into the on-disk inode when writing inodes out, instead of i_size.
875 * The only time when i_disksize and i_size may be different is when
876 * a truncate is in progress. The only things which change i_disksize
877 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
882 * i_data_sem is for serialising ext4_truncate() against
883 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
884 * data tree are chopped off during truncate. We can't do that in
885 * ext4 because whenever we perform intermediate commits during
886 * truncate, the inode and all the metadata blocks *must* be in a
887 * consistent state which allows truncation of the orphans to restart
888 * during recovery. Hence we must fix the get_block-vs-truncate race
889 * by other means, so we have i_data_sem.
891 struct rw_semaphore i_data_sem;
892 struct inode vfs_inode;
893 struct jbd2_inode *jinode;
895 struct ext4_ext_cache i_cached_extent;
897 * File creation time. Its function is same as that of
898 * struct timespec i_{a,c,m}time in the generic inode.
900 struct timespec i_crtime;
903 struct list_head i_prealloc_list;
904 spinlock_t i_prealloc_lock;
906 /* extents status tree */
907 struct ext4_es_tree i_es_tree;
911 ext4_group_t i_last_alloc_group;
913 /* allocation reservation info for delalloc */
914 /* In case of bigalloc, these refer to clusters rather than blocks */
915 unsigned int i_reserved_data_blocks;
916 unsigned int i_reserved_meta_blocks;
917 unsigned int i_allocated_meta_blocks;
918 ext4_lblk_t i_da_metadata_calc_last_lblock;
919 int i_da_metadata_calc_len;
921 /* on-disk additional length */
924 /* Indicate the inline data space. */
929 /* quota space reservation, managed internally by quota code */
930 qsize_t i_reserved_quota;
933 /* completed IOs that might need unwritten extents handling */
934 struct list_head i_completed_io_list;
935 spinlock_t i_completed_io_lock;
936 atomic_t i_ioend_count; /* Number of outstanding io_end structs */
937 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
939 spinlock_t i_block_reservation_lock;
942 * Transactions that contain inode's metadata needed to complete
943 * fsync and fdatasync, respectively.
946 tid_t i_datasync_tid;
948 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
955 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
956 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
957 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
960 * Misc. filesystem flags
962 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
963 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
964 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
969 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
970 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
971 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
972 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
973 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
974 #define EXT4_MOUNT_ERRORS_MASK 0x00070
975 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
976 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
977 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
978 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
979 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
980 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
981 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
982 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
983 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
984 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
985 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
986 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
987 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
988 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
989 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
990 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
991 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
992 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
993 #define EXT4_MOUNT_MBLK_IO_SUBMIT 0x4000000 /* multi-block io submits */
994 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
995 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
996 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
997 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
998 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
1000 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1001 specified delalloc */
1003 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1005 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1007 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1010 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1012 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1014 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1017 #define ext4_test_and_set_bit __test_and_set_bit_le
1018 #define ext4_set_bit __set_bit_le
1019 #define ext4_set_bit_atomic ext2_set_bit_atomic
1020 #define ext4_test_and_clear_bit __test_and_clear_bit_le
1021 #define ext4_clear_bit __clear_bit_le
1022 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
1023 #define ext4_test_bit test_bit_le
1024 #define ext4_find_next_zero_bit find_next_zero_bit_le
1025 #define ext4_find_next_bit find_next_bit_le
1027 extern void ext4_set_bits(void *bm, int cur, int len);
1030 * Maximal mount counts between two filesystem checks
1032 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1033 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1036 * Behaviour when detecting errors
1038 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1039 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1040 #define EXT4_ERRORS_PANIC 3 /* Panic */
1041 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1043 /* Metadata checksum algorithm codes */
1044 #define EXT4_CRC32C_CHKSUM 1
1047 * Structure of the super block
1049 struct ext4_super_block {
1050 /*00*/ __le32 s_inodes_count; /* Inodes count */
1051 __le32 s_blocks_count_lo; /* Blocks count */
1052 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1053 __le32 s_free_blocks_count_lo; /* Free blocks count */
1054 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1055 __le32 s_first_data_block; /* First Data Block */
1056 __le32 s_log_block_size; /* Block size */
1057 __le32 s_log_cluster_size; /* Allocation cluster size */
1058 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1059 __le32 s_clusters_per_group; /* # Clusters per group */
1060 __le32 s_inodes_per_group; /* # Inodes per group */
1061 __le32 s_mtime; /* Mount time */
1062 /*30*/ __le32 s_wtime; /* Write time */
1063 __le16 s_mnt_count; /* Mount count */
1064 __le16 s_max_mnt_count; /* Maximal mount count */
1065 __le16 s_magic; /* Magic signature */
1066 __le16 s_state; /* File system state */
1067 __le16 s_errors; /* Behaviour when detecting errors */
1068 __le16 s_minor_rev_level; /* minor revision level */
1069 /*40*/ __le32 s_lastcheck; /* time of last check */
1070 __le32 s_checkinterval; /* max. time between checks */
1071 __le32 s_creator_os; /* OS */
1072 __le32 s_rev_level; /* Revision level */
1073 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1074 __le16 s_def_resgid; /* Default gid for reserved blocks */
1076 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1078 * Note: the difference between the compatible feature set and
1079 * the incompatible feature set is that if there is a bit set
1080 * in the incompatible feature set that the kernel doesn't
1081 * know about, it should refuse to mount the filesystem.
1083 * e2fsck's requirements are more strict; if it doesn't know
1084 * about a feature in either the compatible or incompatible
1085 * feature set, it must abort and not try to meddle with
1086 * things it doesn't understand...
1088 __le32 s_first_ino; /* First non-reserved inode */
1089 __le16 s_inode_size; /* size of inode structure */
1090 __le16 s_block_group_nr; /* block group # of this superblock */
1091 __le32 s_feature_compat; /* compatible feature set */
1092 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1093 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1094 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1095 /*78*/ char s_volume_name[16]; /* volume name */
1096 /*88*/ char s_last_mounted[64]; /* directory where last mounted */
1097 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1099 * Performance hints. Directory preallocation should only
1100 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1102 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1103 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1104 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1106 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1108 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1109 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1110 __le32 s_journal_dev; /* device number of journal file */
1111 __le32 s_last_orphan; /* start of list of inodes to delete */
1112 __le32 s_hash_seed[4]; /* HTREE hash seed */
1113 __u8 s_def_hash_version; /* Default hash version to use */
1114 __u8 s_jnl_backup_type;
1115 __le16 s_desc_size; /* size of group descriptor */
1116 /*100*/ __le32 s_default_mount_opts;
1117 __le32 s_first_meta_bg; /* First metablock block group */
1118 __le32 s_mkfs_time; /* When the filesystem was created */
1119 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1120 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1121 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1122 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1123 __le32 s_free_blocks_count_hi; /* Free blocks count */
1124 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1125 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1126 __le32 s_flags; /* Miscellaneous flags */
1127 __le16 s_raid_stride; /* RAID stride */
1128 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1129 __le64 s_mmp_block; /* Block for multi-mount protection */
1130 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1131 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1132 __u8 s_checksum_type; /* metadata checksum algorithm used */
1133 __le16 s_reserved_pad;
1134 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1135 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1136 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1137 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1138 snapshot's future use */
1139 __le32 s_snapshot_list; /* inode number of the head of the
1140 on-disk snapshot list */
1141 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1142 __le32 s_error_count; /* number of fs errors */
1143 __le32 s_first_error_time; /* first time an error happened */
1144 __le32 s_first_error_ino; /* inode involved in first error */
1145 __le64 s_first_error_block; /* block involved of first error */
1146 __u8 s_first_error_func[32]; /* function where the error happened */
1147 __le32 s_first_error_line; /* line number where error happened */
1148 __le32 s_last_error_time; /* most recent time of an error */
1149 __le32 s_last_error_ino; /* inode involved in last error */
1150 __le32 s_last_error_line; /* line number where error happened */
1151 __le64 s_last_error_block; /* block involved of last error */
1152 __u8 s_last_error_func[32]; /* function where the error happened */
1153 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1154 __u8 s_mount_opts[64];
1155 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1156 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1157 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1158 __le32 s_reserved[108]; /* Padding to the end of the block */
1159 __le32 s_checksum; /* crc32c(superblock) */
1162 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1167 * run-time mount flags
1169 #define EXT4_MF_MNTDIR_SAMPLED 0x0001
1170 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1173 * fourth extended-fs super-block data in memory
1175 struct ext4_sb_info {
1176 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1177 unsigned long s_inodes_per_block;/* Number of inodes per block */
1178 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1179 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1180 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1181 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1182 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1183 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1184 ext4_group_t s_groups_count; /* Number of groups in the fs */
1185 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1186 unsigned long s_overhead; /* # of fs overhead clusters */
1187 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1188 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1189 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1190 struct buffer_head * s_sbh; /* Buffer containing the super block */
1191 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1192 struct buffer_head **s_group_desc;
1193 unsigned int s_mount_opt;
1194 unsigned int s_mount_opt2;
1195 unsigned int s_mount_flags;
1196 unsigned int s_def_mount_opt;
1197 ext4_fsblk_t s_sb_block;
1200 unsigned short s_mount_state;
1201 unsigned short s_pad;
1202 int s_addr_per_block_bits;
1203 int s_desc_per_block_bits;
1206 unsigned int s_inode_readahead_blks;
1207 unsigned int s_inode_goal;
1208 spinlock_t s_next_gen_lock;
1209 u32 s_next_generation;
1211 int s_def_hash_version;
1212 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1213 struct percpu_counter s_freeclusters_counter;
1214 struct percpu_counter s_freeinodes_counter;
1215 struct percpu_counter s_dirs_counter;
1216 struct percpu_counter s_dirtyclusters_counter;
1217 struct blockgroup_lock *s_blockgroup_lock;
1218 struct proc_dir_entry *s_proc;
1219 struct kobject s_kobj;
1220 struct completion s_kobj_unregister;
1221 struct super_block *s_sb;
1224 struct journal_s *s_journal;
1225 struct list_head s_orphan;
1226 struct mutex s_orphan_lock;
1227 unsigned long s_resize_flags; /* Flags indicating if there
1229 unsigned long s_commit_interval;
1230 u32 s_max_batch_time;
1231 u32 s_min_batch_time;
1232 struct block_device *journal_bdev;
1234 char *s_qf_names[MAXQUOTAS]; /* Names of quota files with journalled quota */
1235 int s_jquota_fmt; /* Format of quota to use */
1237 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1238 struct rb_root system_blks;
1240 #ifdef EXTENTS_STATS
1241 /* ext4 extents stats */
1242 unsigned long s_ext_min;
1243 unsigned long s_ext_max;
1244 unsigned long s_depth_max;
1245 spinlock_t s_ext_stats_lock;
1246 unsigned long s_ext_blocks;
1247 unsigned long s_ext_extents;
1250 /* for buddy allocator */
1251 struct ext4_group_info ***s_group_info;
1252 struct inode *s_buddy_cache;
1253 spinlock_t s_md_lock;
1254 unsigned short *s_mb_offsets;
1255 unsigned int *s_mb_maxs;
1256 unsigned int s_group_info_size;
1259 unsigned long s_stripe;
1260 unsigned int s_mb_stream_request;
1261 unsigned int s_mb_max_to_scan;
1262 unsigned int s_mb_min_to_scan;
1263 unsigned int s_mb_stats;
1264 unsigned int s_mb_order2_reqs;
1265 unsigned int s_mb_group_prealloc;
1266 unsigned int s_max_writeback_mb_bump;
1267 unsigned int s_max_dir_size_kb;
1268 /* where last allocation was done - for stream allocation */
1269 unsigned long s_mb_last_group;
1270 unsigned long s_mb_last_start;
1272 /* stats for buddy allocator */
1273 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1274 atomic_t s_bal_success; /* we found long enough chunks */
1275 atomic_t s_bal_allocated; /* in blocks */
1276 atomic_t s_bal_ex_scanned; /* total extents scanned */
1277 atomic_t s_bal_goals; /* goal hits */
1278 atomic_t s_bal_breaks; /* too long searches */
1279 atomic_t s_bal_2orders; /* 2^order hits */
1280 spinlock_t s_bal_lock;
1281 unsigned long s_mb_buddies_generated;
1282 unsigned long long s_mb_generation_time;
1283 atomic_t s_mb_lost_chunks;
1284 atomic_t s_mb_preallocated;
1285 atomic_t s_mb_discarded;
1286 atomic_t s_lock_busy;
1288 /* locality groups */
1289 struct ext4_locality_group __percpu *s_locality_groups;
1291 /* for write statistics */
1292 unsigned long s_sectors_written_start;
1293 u64 s_kbytes_written;
1295 /* the size of zero-out chunk */
1296 unsigned int s_extent_max_zeroout_kb;
1298 unsigned int s_log_groups_per_flex;
1299 struct flex_groups *s_flex_groups;
1300 ext4_group_t s_flex_groups_allocated;
1302 /* workqueue for dio unwritten */
1303 struct workqueue_struct *dio_unwritten_wq;
1305 /* timer for periodic error stats printing */
1306 struct timer_list s_err_report;
1308 /* Lazy inode table initialization info */
1309 struct ext4_li_request *s_li_request;
1310 /* Wait multiplier for lazy initialization thread */
1311 unsigned int s_li_wait_mult;
1313 /* Kernel thread for multiple mount protection */
1314 struct task_struct *s_mmp_tsk;
1316 /* record the last minlen when FITRIM is called. */
1317 atomic_t s_last_trim_minblks;
1319 /* Reference to checksum algorithm driver via cryptoapi */
1320 struct crypto_shash *s_chksum_driver;
1322 /* Precomputed FS UUID checksum for seeding other checksums */
1326 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1328 return sb->s_fs_info;
1330 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1332 return container_of(inode, struct ext4_inode_info, vfs_inode);
1335 static inline struct timespec ext4_current_time(struct inode *inode)
1337 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1338 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1341 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1343 return ino == EXT4_ROOT_INO ||
1344 ino == EXT4_USR_QUOTA_INO ||
1345 ino == EXT4_GRP_QUOTA_INO ||
1346 ino == EXT4_JOURNAL_INO ||
1347 ino == EXT4_RESIZE_INO ||
1348 (ino >= EXT4_FIRST_INO(sb) &&
1349 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1352 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
1353 struct ext4_io_end *io_end)
1355 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
1356 io_end->flag |= EXT4_IO_END_UNWRITTEN;
1357 atomic_inc(&EXT4_I(inode)->i_unwritten);
1361 static inline ext4_io_end_t *ext4_inode_aio(struct inode *inode)
1363 return inode->i_private;
1366 static inline void ext4_inode_aio_set(struct inode *inode, ext4_io_end_t *io)
1368 inode->i_private = io;
1372 * Inode dynamic state flags
1375 EXT4_STATE_JDATA, /* journaled data exists */
1376 EXT4_STATE_NEW, /* inode is newly created */
1377 EXT4_STATE_XATTR, /* has in-inode xattrs */
1378 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1379 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1380 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1381 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
1382 EXT4_STATE_NEWENTRY, /* File just added to dir */
1383 EXT4_STATE_DELALLOC_RESERVED, /* blks already reserved for delalloc */
1384 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read
1386 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1389 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1390 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1392 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1394 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1396 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1398 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1400 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1403 EXT4_INODE_BIT_FNS(flag, flags, 0)
1404 #if (BITS_PER_LONG < 64)
1405 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1407 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1409 (ei)->i_state_flags = 0;
1412 EXT4_INODE_BIT_FNS(state, flags, 32)
1414 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1416 /* We depend on the fact that callers will set i_flags */
1420 /* Assume that user mode programs are passing in an ext4fs superblock, not
1421 * a kernel struct super_block. This will allow us to call the feature-test
1422 * macros from user land. */
1423 #define EXT4_SB(sb) (sb)
1426 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1429 * Codes for operating systems
1431 #define EXT4_OS_LINUX 0
1432 #define EXT4_OS_HURD 1
1433 #define EXT4_OS_MASIX 2
1434 #define EXT4_OS_FREEBSD 3
1435 #define EXT4_OS_LITES 4
1440 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1441 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1443 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1444 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1446 #define EXT4_GOOD_OLD_INODE_SIZE 128
1449 * Feature set definitions
1452 #define EXT4_HAS_COMPAT_FEATURE(sb,mask) \
1453 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1454 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \
1455 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1456 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \
1457 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1458 #define EXT4_SET_COMPAT_FEATURE(sb,mask) \
1459 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1460 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \
1461 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1462 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \
1463 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1464 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \
1465 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1466 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
1467 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1468 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \
1469 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1471 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1472 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1473 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1474 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1475 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1476 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1478 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1479 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1480 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1481 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1482 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1483 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1484 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1485 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1486 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1488 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1489 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1490 * all other data structures' checksums. However, the METADATA_CSUM and
1491 * GDT_CSUM bits are mutually exclusive.
1493 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
1495 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1496 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1497 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1498 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1499 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1500 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1501 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1502 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1503 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1504 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1505 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1506 #define EXT4_FEATURE_INCOMPAT_BG_USE_META_CSUM 0x2000 /* use crc32c for bg */
1507 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
1508 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
1510 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1511 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1512 EXT4_FEATURE_INCOMPAT_META_BG)
1513 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1514 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1515 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1517 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1518 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1519 EXT4_FEATURE_INCOMPAT_RECOVER| \
1520 EXT4_FEATURE_INCOMPAT_META_BG)
1521 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1522 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1523 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1525 #define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR
1526 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1527 EXT4_FEATURE_INCOMPAT_RECOVER| \
1528 EXT4_FEATURE_INCOMPAT_META_BG| \
1529 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1530 EXT4_FEATURE_INCOMPAT_64BIT| \
1531 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1532 EXT4_FEATURE_INCOMPAT_MMP)
1533 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1534 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1535 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1536 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1537 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1538 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1539 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1540 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1541 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1542 EXT4_FEATURE_RO_COMPAT_QUOTA)
1545 * Default values for user and/or group using reserved blocks
1547 #define EXT4_DEF_RESUID 0
1548 #define EXT4_DEF_RESGID 0
1550 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
1553 * Default mount options
1555 #define EXT4_DEFM_DEBUG 0x0001
1556 #define EXT4_DEFM_BSDGROUPS 0x0002
1557 #define EXT4_DEFM_XATTR_USER 0x0004
1558 #define EXT4_DEFM_ACL 0x0008
1559 #define EXT4_DEFM_UID16 0x0010
1560 #define EXT4_DEFM_JMODE 0x0060
1561 #define EXT4_DEFM_JMODE_DATA 0x0020
1562 #define EXT4_DEFM_JMODE_ORDERED 0x0040
1563 #define EXT4_DEFM_JMODE_WBACK 0x0060
1564 #define EXT4_DEFM_NOBARRIER 0x0100
1565 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1566 #define EXT4_DEFM_DISCARD 0x0400
1567 #define EXT4_DEFM_NODELALLOC 0x0800
1570 * Default journal batch times
1572 #define EXT4_DEF_MIN_BATCH_TIME 0
1573 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1576 * Minimum number of groups in a flexgroup before we separate out
1577 * directories into the first block group of a flexgroup
1579 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1582 * Structure of a directory entry
1584 #define EXT4_NAME_LEN 255
1586 struct ext4_dir_entry {
1587 __le32 inode; /* Inode number */
1588 __le16 rec_len; /* Directory entry length */
1589 __le16 name_len; /* Name length */
1590 char name[EXT4_NAME_LEN]; /* File name */
1594 * The new version of the directory entry. Since EXT4 structures are
1595 * stored in intel byte order, and the name_len field could never be
1596 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1599 struct ext4_dir_entry_2 {
1600 __le32 inode; /* Inode number */
1601 __le16 rec_len; /* Directory entry length */
1602 __u8 name_len; /* Name length */
1604 char name[EXT4_NAME_LEN]; /* File name */
1608 * This is a bogus directory entry at the end of each leaf block that
1609 * records checksums.
1611 struct ext4_dir_entry_tail {
1612 __le32 det_reserved_zero1; /* Pretend to be unused */
1613 __le16 det_rec_len; /* 12 */
1614 __u8 det_reserved_zero2; /* Zero name length */
1615 __u8 det_reserved_ft; /* 0xDE, fake file type */
1616 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
1619 #define EXT4_DIRENT_TAIL(block, blocksize) \
1620 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1622 sizeof(struct ext4_dir_entry_tail))))
1625 * Ext4 directory file types. Only the low 3 bits are used. The
1626 * other bits are reserved for now.
1628 #define EXT4_FT_UNKNOWN 0
1629 #define EXT4_FT_REG_FILE 1
1630 #define EXT4_FT_DIR 2
1631 #define EXT4_FT_CHRDEV 3
1632 #define EXT4_FT_BLKDEV 4
1633 #define EXT4_FT_FIFO 5
1634 #define EXT4_FT_SOCK 6
1635 #define EXT4_FT_SYMLINK 7
1637 #define EXT4_FT_MAX 8
1639 #define EXT4_FT_DIR_CSUM 0xDE
1642 * EXT4_DIR_PAD defines the directory entries boundaries
1644 * NOTE: It must be a multiple of 4
1646 #define EXT4_DIR_PAD 4
1647 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1648 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1650 #define EXT4_MAX_REC_LEN ((1<<16)-1)
1653 * If we ever get support for fs block sizes > page_size, we'll need
1654 * to remove the #if statements in the next two functions...
1656 static inline unsigned int
1657 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1659 unsigned len = le16_to_cpu(dlen);
1661 #if (PAGE_CACHE_SIZE >= 65536)
1662 if (len == EXT4_MAX_REC_LEN || len == 0)
1664 return (len & 65532) | ((len & 3) << 16);
1670 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1672 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1674 #if (PAGE_CACHE_SIZE >= 65536)
1676 return cpu_to_le16(len);
1677 if (len == blocksize) {
1678 if (blocksize == 65536)
1679 return cpu_to_le16(EXT4_MAX_REC_LEN);
1681 return cpu_to_le16(0);
1683 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1685 return cpu_to_le16(len);
1690 * Hash Tree Directory indexing
1691 * (c) Daniel Phillips, 2001
1694 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1695 EXT4_FEATURE_COMPAT_DIR_INDEX) && \
1696 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
1697 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1698 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1700 /* Legal values for the dx_root hash_version field: */
1702 #define DX_HASH_LEGACY 0
1703 #define DX_HASH_HALF_MD4 1
1704 #define DX_HASH_TEA 2
1705 #define DX_HASH_LEGACY_UNSIGNED 3
1706 #define DX_HASH_HALF_MD4_UNSIGNED 4
1707 #define DX_HASH_TEA_UNSIGNED 5
1709 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
1710 const void *address, unsigned int length)
1713 struct shash_desc shash;
1718 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
1720 desc.shash.tfm = sbi->s_chksum_driver;
1721 desc.shash.flags = 0;
1722 *(u32 *)desc.ctx = crc;
1724 err = crypto_shash_update(&desc.shash, address, length);
1727 return *(u32 *)desc.ctx;
1732 /* hash info structure used by the directory hash */
1742 /* 32 and 64 bit signed EOF for dx directories */
1743 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
1744 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
1748 * Control parameters used by ext4_htree_next_block
1750 #define HASH_NB_ALWAYS 1
1754 * Describe an inode's exact location on disk and in memory
1758 struct buffer_head *bh;
1759 unsigned long offset;
1760 ext4_group_t block_group;
1763 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
1765 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
1769 * This structure is stuffed into the struct file's private_data field
1770 * for directories. It is where we put information so that we can do
1771 * readdir operations in hash tree order.
1773 struct dir_private_info {
1774 struct rb_root root;
1775 struct rb_node *curr_node;
1776 struct fname *extra_fname;
1779 __u32 curr_minor_hash;
1783 /* calculate the first block number of the group */
1784 static inline ext4_fsblk_t
1785 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
1787 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
1788 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
1792 * Special error return code only used by dx_probe() and its callers.
1794 #define ERR_BAD_DX_DIR -75000
1796 void ext4_get_group_no_and_offset(struct super_block *sb, ext4_fsblk_t blocknr,
1797 ext4_group_t *blockgrpp, ext4_grpblk_t *offsetp);
1800 * Timeout and state flag for lazy initialization inode thread.
1802 #define EXT4_DEF_LI_WAIT_MULT 10
1803 #define EXT4_DEF_LI_MAX_START_DELAY 5
1804 #define EXT4_LAZYINIT_QUIT 0x0001
1805 #define EXT4_LAZYINIT_RUNNING 0x0002
1808 * Lazy inode table initialization info
1810 struct ext4_lazy_init {
1811 unsigned long li_state;
1812 struct list_head li_request_list;
1813 struct mutex li_list_mtx;
1816 struct ext4_li_request {
1817 struct super_block *lr_super;
1818 struct ext4_sb_info *lr_sbi;
1819 ext4_group_t lr_next_group;
1820 struct list_head lr_request;
1821 unsigned long lr_next_sched;
1822 unsigned long lr_timeout;
1825 struct ext4_features {
1826 struct kobject f_kobj;
1827 struct completion f_kobj_unregister;
1831 * This structure will be used for multiple mount protection. It will be
1832 * written into the block number saved in the s_mmp_block field in the
1833 * superblock. Programs that check MMP should assume that if
1834 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
1835 * to use the filesystem, regardless of how old the timestamp is.
1837 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
1838 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
1839 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
1840 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
1843 __le32 mmp_magic; /* Magic number for MMP */
1844 __le32 mmp_seq; /* Sequence no. updated periodically */
1847 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
1848 * purposes and do not affect the correctness of the algorithm
1850 __le64 mmp_time; /* Time last updated */
1851 char mmp_nodename[64]; /* Node which last updated MMP block */
1852 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
1855 * mmp_check_interval is used to verify if the MMP block has been
1856 * updated on the block device. The value is updated based on the
1857 * maximum time to write the MMP block during an update cycle.
1859 __le16 mmp_check_interval;
1862 __le32 mmp_pad2[226];
1863 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
1866 /* arguments passed to the mmp thread */
1868 struct buffer_head *bh; /* bh from initial read_mmp_block() */
1869 struct super_block *sb; /* super block of the fs */
1873 * Check interval multiplier
1874 * The MMP block is written every update interval and initially checked every
1875 * update interval x the multiplier (the value is then adapted based on the
1876 * write latency). The reason is that writes can be delayed under load and we
1877 * don't want readers to incorrectly assume that the filesystem is no longer
1880 #define EXT4_MMP_CHECK_MULT 2UL
1883 * Minimum interval for MMP checking in seconds.
1885 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
1888 * Maximum interval for MMP checking in seconds.
1890 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
1893 * Function prototypes
1897 * Ok, these declarations are also in <linux/kernel.h> but none of the
1898 * ext4 source programs needs to include it so they are duplicated here.
1900 # define NORET_TYPE /**/
1901 # define ATTRIB_NORET __attribute__((noreturn))
1902 # define NORET_AND noreturn,
1905 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
1906 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
1907 struct ext4_group_desc *gdp,
1908 struct buffer_head *bh, int sz);
1909 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
1910 struct ext4_group_desc *gdp,
1911 struct buffer_head *bh, int sz);
1912 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
1913 struct ext4_group_desc *gdp,
1914 struct buffer_head *bh);
1915 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
1916 struct ext4_group_desc *gdp,
1917 struct buffer_head *bh);
1920 extern void ext4_validate_block_bitmap(struct super_block *sb,
1921 struct ext4_group_desc *desc,
1922 unsigned int block_group,
1923 struct buffer_head *bh);
1924 extern unsigned int ext4_block_group(struct super_block *sb,
1925 ext4_fsblk_t blocknr);
1926 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
1927 ext4_fsblk_t blocknr);
1928 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
1929 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
1930 ext4_group_t group);
1931 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
1934 unsigned long *count,
1936 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
1937 s64 nclusters, unsigned int flags);
1938 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
1939 extern void ext4_check_blocks_bitmap(struct super_block *);
1940 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
1941 ext4_group_t block_group,
1942 struct buffer_head ** bh);
1943 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
1945 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
1946 ext4_group_t block_group);
1947 extern int ext4_wait_block_bitmap(struct super_block *sb,
1948 ext4_group_t block_group,
1949 struct buffer_head *bh);
1950 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
1951 ext4_group_t block_group);
1952 extern void ext4_init_block_bitmap(struct super_block *sb,
1953 struct buffer_head *bh,
1955 struct ext4_group_desc *desc);
1956 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
1957 ext4_group_t block_group,
1958 struct ext4_group_desc *gdp);
1959 extern unsigned ext4_num_overhead_clusters(struct super_block *sb,
1960 ext4_group_t block_group,
1961 struct ext4_group_desc *gdp);
1962 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
1965 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
1967 struct ext4_dir_entry_2 *,
1968 struct buffer_head *, char *, int,
1970 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
1971 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
1972 (de), (bh), (buf), (size), (offset)))
1973 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
1975 struct ext4_dir_entry_2 *dirent);
1976 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
1977 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
1978 struct buffer_head *bh,
1979 void *buf, int buf_size,
1980 const char *name, int namelen,
1981 struct ext4_dir_entry_2 **dest_de);
1982 void ext4_insert_dentry(struct inode *inode,
1983 struct ext4_dir_entry_2 *de,
1985 const char *name, int namelen);
1986 static inline void ext4_update_dx_flag(struct inode *inode)
1988 if (!EXT4_HAS_COMPAT_FEATURE(inode->i_sb,
1989 EXT4_FEATURE_COMPAT_DIR_INDEX))
1990 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
1992 static unsigned char ext4_filetype_table[] = {
1993 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
1996 static inline unsigned char get_dtype(struct super_block *sb, int filetype)
1998 if (!EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE) ||
1999 (filetype >= EXT4_FT_MAX))
2002 return ext4_filetype_table[filetype];
2006 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2007 extern int ext4_flush_unwritten_io(struct inode *);
2010 extern int ext4fs_dirhash(const char *name, int len, struct
2011 dx_hash_info *hinfo);
2014 extern struct inode *ext4_new_inode(handle_t *, struct inode *, umode_t,
2015 const struct qstr *qstr, __u32 goal,
2017 extern void ext4_free_inode(handle_t *, struct inode *);
2018 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2019 extern unsigned long ext4_count_free_inodes(struct super_block *);
2020 extern unsigned long ext4_count_dirs(struct super_block *);
2021 extern void ext4_check_inodes_bitmap(struct super_block *);
2022 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2023 extern int ext4_init_inode_table(struct super_block *sb,
2024 ext4_group_t group, int barrier);
2025 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2028 extern long ext4_mb_stats;
2029 extern long ext4_mb_max_to_scan;
2030 extern int ext4_mb_init(struct super_block *);
2031 extern int ext4_mb_release(struct super_block *);
2032 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2033 struct ext4_allocation_request *, int *);
2034 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2035 extern void ext4_discard_preallocations(struct inode *);
2036 extern int __init ext4_init_mballoc(void);
2037 extern void ext4_exit_mballoc(void);
2038 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2039 struct buffer_head *bh, ext4_fsblk_t block,
2040 unsigned long count, int flags);
2041 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2042 ext4_group_t ngroups);
2043 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2044 ext4_group_t i, struct ext4_group_desc *desc);
2045 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2046 ext4_fsblk_t block, unsigned long count);
2047 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2050 struct buffer_head *ext4_getblk(handle_t *, struct inode *,
2051 ext4_lblk_t, int, int *);
2052 struct buffer_head *ext4_bread(handle_t *, struct inode *,
2053 ext4_lblk_t, int, int *);
2054 int ext4_get_block_write(struct inode *inode, sector_t iblock,
2055 struct buffer_head *bh_result, int create);
2056 int ext4_get_block(struct inode *inode, sector_t iblock,
2057 struct buffer_head *bh_result, int create);
2058 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2059 struct buffer_head *bh, int create);
2060 int ext4_walk_page_buffers(handle_t *handle,
2061 struct buffer_head *head,
2065 int (*fn)(handle_t *handle,
2066 struct buffer_head *bh));
2067 int do_journal_get_write_access(handle_t *handle,
2068 struct buffer_head *bh);
2069 #define FALL_BACK_TO_NONDELALLOC 1
2070 #define CONVERT_INLINE_DATA 2
2072 extern struct inode *ext4_iget(struct super_block *, unsigned long);
2073 extern int ext4_write_inode(struct inode *, struct writeback_control *);
2074 extern int ext4_setattr(struct dentry *, struct iattr *);
2075 extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
2076 struct kstat *stat);
2077 extern void ext4_evict_inode(struct inode *);
2078 extern void ext4_clear_inode(struct inode *);
2079 extern int ext4_sync_inode(handle_t *, struct inode *);
2080 extern void ext4_dirty_inode(struct inode *, int);
2081 extern int ext4_change_inode_journal_flag(struct inode *, int);
2082 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2083 extern int ext4_can_truncate(struct inode *inode);
2084 extern void ext4_truncate(struct inode *);
2085 extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length);
2086 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
2087 extern void ext4_set_inode_flags(struct inode *);
2088 extern void ext4_get_inode_flags(struct ext4_inode_info *);
2089 extern int ext4_alloc_da_blocks(struct inode *inode);
2090 extern void ext4_set_aops(struct inode *inode);
2091 extern int ext4_writepage_trans_blocks(struct inode *);
2092 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2093 extern int ext4_discard_partial_page_buffers(handle_t *handle,
2094 struct address_space *mapping, loff_t from,
2095 loff_t length, int flags);
2096 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
2097 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2098 extern void ext4_da_update_reserve_space(struct inode *inode,
2099 int used, int quota_claim);
2102 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2103 struct ext4_map_blocks *map, int flags);
2104 extern ssize_t ext4_ind_direct_IO(int rw, struct kiocb *iocb,
2105 const struct iovec *iov, loff_t offset,
2106 unsigned long nr_segs);
2107 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2108 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks, int chunk);
2109 extern void ext4_ind_truncate(struct inode *inode);
2112 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2113 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2116 extern int ext4_ext_migrate(struct inode *);
2119 extern int ext4_dirent_csum_verify(struct inode *inode,
2120 struct ext4_dir_entry *dirent);
2121 extern int ext4_orphan_add(handle_t *, struct inode *);
2122 extern int ext4_orphan_del(handle_t *, struct inode *);
2123 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2124 __u32 start_minor_hash, __u32 *next_hash);
2125 extern int search_dir(struct buffer_head *bh,
2129 const struct qstr *d_name,
2130 unsigned int offset,
2131 struct ext4_dir_entry_2 **res_dir);
2134 extern int ext4_group_add(struct super_block *sb,
2135 struct ext4_new_group_data *input);
2136 extern int ext4_group_extend(struct super_block *sb,
2137 struct ext4_super_block *es,
2138 ext4_fsblk_t n_blocks_count);
2139 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2142 extern int ext4_calculate_overhead(struct super_block *sb);
2143 extern int ext4_superblock_csum_verify(struct super_block *sb,
2144 struct ext4_super_block *es);
2145 extern void ext4_superblock_csum_set(struct super_block *sb);
2146 extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2147 extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2148 extern void ext4_kvfree(void *ptr);
2149 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2150 ext4_group_t ngroup);
2151 extern __printf(4, 5)
2152 void __ext4_error(struct super_block *, const char *, unsigned int,
2154 #define ext4_error(sb, message...) __ext4_error(sb, __func__, \
2155 __LINE__, ## message)
2156 extern __printf(5, 6)
2157 void ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2159 extern __printf(5, 6)
2160 void ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2162 extern void __ext4_std_error(struct super_block *, const char *,
2164 extern __printf(4, 5)
2165 void __ext4_abort(struct super_block *, const char *, unsigned int,
2167 #define ext4_abort(sb, message...) __ext4_abort(sb, __func__, \
2168 __LINE__, ## message)
2169 extern __printf(4, 5)
2170 void __ext4_warning(struct super_block *, const char *, unsigned int,
2172 #define ext4_warning(sb, message...) __ext4_warning(sb, __func__, \
2173 __LINE__, ## message)
2174 extern __printf(3, 4)
2175 void ext4_msg(struct super_block *, const char *, const char *, ...);
2176 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2177 const char *, unsigned int, const char *);
2178 #define dump_mmp_msg(sb, mmp, msg) __dump_mmp_msg(sb, mmp, __func__, \
2180 extern __printf(7, 8)
2181 void __ext4_grp_locked_error(const char *, unsigned int,
2182 struct super_block *, ext4_group_t,
2183 unsigned long, ext4_fsblk_t,
2185 #define ext4_grp_locked_error(sb, grp, message...) \
2186 __ext4_grp_locked_error(__func__, __LINE__, (sb), (grp), ## message)
2187 extern void ext4_update_dynamic_rev(struct super_block *sb);
2188 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2190 extern int ext4_update_rocompat_feature(handle_t *handle,
2191 struct super_block *sb, __u32 rocompat);
2192 extern int ext4_update_incompat_feature(handle_t *handle,
2193 struct super_block *sb, __u32 incompat);
2194 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2195 struct ext4_group_desc *bg);
2196 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2197 struct ext4_group_desc *bg);
2198 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2199 struct ext4_group_desc *bg);
2200 extern __u32 ext4_free_group_clusters(struct super_block *sb,
2201 struct ext4_group_desc *bg);
2202 extern __u32 ext4_free_inodes_count(struct super_block *sb,
2203 struct ext4_group_desc *bg);
2204 extern __u32 ext4_used_dirs_count(struct super_block *sb,
2205 struct ext4_group_desc *bg);
2206 extern __u32 ext4_itable_unused_count(struct super_block *sb,
2207 struct ext4_group_desc *bg);
2208 extern void ext4_block_bitmap_set(struct super_block *sb,
2209 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2210 extern void ext4_inode_bitmap_set(struct super_block *sb,
2211 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2212 extern void ext4_inode_table_set(struct super_block *sb,
2213 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2214 extern void ext4_free_group_clusters_set(struct super_block *sb,
2215 struct ext4_group_desc *bg,
2217 extern void ext4_free_inodes_set(struct super_block *sb,
2218 struct ext4_group_desc *bg, __u32 count);
2219 extern void ext4_used_dirs_set(struct super_block *sb,
2220 struct ext4_group_desc *bg, __u32 count);
2221 extern void ext4_itable_unused_set(struct super_block *sb,
2222 struct ext4_group_desc *bg, __u32 count);
2223 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2224 struct ext4_group_desc *gdp);
2225 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2226 struct ext4_group_desc *gdp);
2228 static inline int ext4_has_group_desc_csum(struct super_block *sb)
2230 return EXT4_HAS_RO_COMPAT_FEATURE(sb,
2231 EXT4_FEATURE_RO_COMPAT_GDT_CSUM |
2232 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM);
2235 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2237 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2238 le32_to_cpu(es->s_blocks_count_lo);
2241 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2243 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2244 le32_to_cpu(es->s_r_blocks_count_lo);
2247 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2249 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2250 le32_to_cpu(es->s_free_blocks_count_lo);
2253 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2256 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2257 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2260 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2263 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2264 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2267 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2270 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2271 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2274 static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
2276 if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2277 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2278 le32_to_cpu(raw_inode->i_size_lo);
2280 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2283 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2285 raw_inode->i_size_lo = cpu_to_le32(i_size);
2286 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2290 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2293 struct ext4_group_info ***grp_info;
2294 long indexv, indexh;
2295 grp_info = EXT4_SB(sb)->s_group_info;
2296 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2297 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2298 return grp_info[indexv][indexh];
2302 * Reading s_groups_count requires using smp_rmb() afterwards. See
2303 * the locking protocol documented in the comments of ext4_group_add()
2306 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2308 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
2314 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2315 ext4_group_t block_group)
2317 return block_group >> sbi->s_log_groups_per_flex;
2320 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2322 return 1 << sbi->s_log_groups_per_flex;
2325 #define ext4_std_error(sb, errno) \
2328 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
2332 /* Each CPU can accumulate percpu_counter_batch clusters in their local
2333 * counters. So we need to make sure we have free clusters more
2334 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
2336 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2338 #define EXT4_FREECLUSTERS_WATERMARK 0
2341 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2344 * XXX: replace with spinlock if seen contended -bzzz
2346 down_write(&EXT4_I(inode)->i_data_sem);
2347 if (newsize > EXT4_I(inode)->i_disksize)
2348 EXT4_I(inode)->i_disksize = newsize;
2349 up_write(&EXT4_I(inode)->i_data_sem);
2353 struct ext4_group_info {
2354 unsigned long bb_state;
2355 struct rb_root bb_free_root;
2356 ext4_grpblk_t bb_first_free; /* first free block */
2357 ext4_grpblk_t bb_free; /* total free blocks */
2358 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
2359 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
2360 struct list_head bb_prealloc_list;
2364 struct rw_semaphore alloc_sem;
2365 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
2366 * regions, index is order.
2367 * bb_counters[3] = 5 means
2368 * 5 free 8-block regions. */
2371 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
2372 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
2374 #define EXT4_MB_GRP_NEED_INIT(grp) \
2375 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2377 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
2378 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2379 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
2380 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2381 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
2382 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2384 #define EXT4_MAX_CONTENTION 8
2385 #define EXT4_CONTENTION_THRESHOLD 2
2387 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2390 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2394 * Returns true if the filesystem is busy enough that attempts to
2395 * access the block group locks has run into contention.
2397 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2399 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2402 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2404 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2405 if (spin_trylock(lock))
2407 * We're able to grab the lock right away, so drop the
2408 * lock contention counter.
2410 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2413 * The lock is busy, so bump the contention counter,
2414 * and then wait on the spin lock.
2416 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2417 EXT4_MAX_CONTENTION);
2422 static inline void ext4_unlock_group(struct super_block *sb,
2425 spin_unlock(ext4_group_lock_ptr(sb, group));
2429 * Block validity checking
2431 #define ext4_check_indirect_blockref(inode, bh) \
2432 ext4_check_blockref(__func__, __LINE__, inode, \
2433 (__le32 *)(bh)->b_data, \
2434 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2436 #define ext4_ind_check_inode(inode) \
2437 ext4_check_blockref(__func__, __LINE__, inode, \
2438 EXT4_I(inode)->i_data, \
2442 * Inodes and files operations
2446 extern const struct file_operations ext4_dir_operations;
2449 extern const struct inode_operations ext4_file_inode_operations;
2450 extern const struct file_operations ext4_file_operations;
2451 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
2452 extern void ext4_unwritten_wait(struct inode *inode);
2455 extern const struct inode_operations ext4_dir_inode_operations;
2456 extern const struct inode_operations ext4_special_inode_operations;
2457 extern struct dentry *ext4_get_parent(struct dentry *child);
2458 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
2459 struct ext4_dir_entry_2 *de,
2460 int blocksize, int csum_size,
2461 unsigned int parent_ino, int dotdot_real_len);
2462 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t,
2463 unsigned int blocksize);
2464 extern int ext4_handle_dirty_dirent_node(handle_t *handle,
2465 struct inode *inode,
2466 struct buffer_head *bh);
2469 extern const struct inode_operations ext4_symlink_inode_operations;
2470 extern const struct inode_operations ext4_fast_symlink_inode_operations;
2472 /* block_validity */
2473 extern void ext4_release_system_zone(struct super_block *sb);
2474 extern int ext4_setup_system_zone(struct super_block *sb);
2475 extern int __init ext4_init_system_zone(void);
2476 extern void ext4_exit_system_zone(void);
2477 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
2478 ext4_fsblk_t start_blk,
2479 unsigned int count);
2480 extern int ext4_check_blockref(const char *, unsigned int,
2481 struct inode *, __le32 *, unsigned int);
2484 struct ext4_ext_path;
2487 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
2488 extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
2489 extern int ext4_ext_index_trans_blocks(struct inode *inode, int nrblocks,
2491 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
2492 struct ext4_map_blocks *map, int flags);
2493 extern void ext4_ext_truncate(struct inode *);
2494 extern int ext4_ext_punch_hole(struct file *file, loff_t offset,
2496 extern void ext4_ext_init(struct super_block *);
2497 extern void ext4_ext_release(struct super_block *);
2498 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
2500 extern int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset,
2502 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
2503 struct ext4_map_blocks *map, int flags);
2504 extern int ext4_ext_calc_metadata_amount(struct inode *inode,
2505 ext4_lblk_t lblocks);
2506 extern int ext4_extent_tree_init(handle_t *, struct inode *);
2507 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
2509 struct ext4_ext_path *path);
2510 extern int ext4_can_extents_be_merged(struct inode *inode,
2511 struct ext4_extent *ex1,
2512 struct ext4_extent *ex2);
2513 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
2514 struct ext4_ext_path *,
2515 struct ext4_extent *, int);
2516 extern struct ext4_ext_path *ext4_ext_find_extent(struct inode *, ext4_lblk_t,
2517 struct ext4_ext_path *);
2518 extern void ext4_ext_drop_refs(struct ext4_ext_path *);
2519 extern int ext4_ext_check_inode(struct inode *inode);
2520 extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk);
2521 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
2522 __u64 start, __u64 len);
2526 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
2527 __u64 start_orig, __u64 start_donor,
2528 __u64 len, __u64 *moved_len);
2531 extern int __init ext4_init_pageio(void);
2532 extern void ext4_add_complete_io(ext4_io_end_t *io_end);
2533 extern void ext4_exit_pageio(void);
2534 extern void ext4_ioend_wait(struct inode *);
2535 extern void ext4_free_io_end(ext4_io_end_t *io);
2536 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
2537 extern void ext4_io_submit(struct ext4_io_submit *io);
2538 extern int ext4_bio_write_page(struct ext4_io_submit *io,
2541 struct writeback_control *wbc);
2544 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
2545 extern void ext4_mmp_csum_set(struct super_block *sb, struct mmp_struct *mmp);
2546 extern int ext4_mmp_csum_verify(struct super_block *sb,
2547 struct mmp_struct *mmp);
2549 /* BH_Uninit flag: blocks are allocated but uninitialized on disk */
2550 enum ext4_state_bits {
2551 BH_Uninit /* blocks are allocated but uninitialized on disk */
2552 = BH_JBDPrivateStart,
2553 BH_AllocFromCluster, /* allocated blocks were part of already
2554 * allocated cluster. Note that this flag will
2555 * never, ever appear in a buffer_head's state
2556 * flag. See EXT4_MAP_FROM_CLUSTER to see where
2560 BUFFER_FNS(Uninit, uninit)
2561 TAS_BUFFER_FNS(Uninit, uninit)
2564 * Add new method to test wether block and inode bitmaps are properly
2565 * initialized. With uninit_bg reading the block from disk is not enough
2566 * to mark the bitmap uptodate. We need to also zero-out the bitmap
2568 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
2570 static inline int bitmap_uptodate(struct buffer_head *bh)
2572 return (buffer_uptodate(bh) &&
2573 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
2575 static inline void set_bitmap_uptodate(struct buffer_head *bh)
2577 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
2581 * Disable DIO read nolock optimization, so new dioreaders will be forced
2584 static inline void ext4_inode_block_unlocked_dio(struct inode *inode)
2586 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
2589 static inline void ext4_inode_resume_unlocked_dio(struct inode *inode)
2592 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
2595 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
2597 /* For ioend & aio unwritten conversion wait queues */
2598 #define EXT4_WQ_HASH_SZ 37
2599 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
2601 #define ext4_aio_mutex(v) (&ext4__aio_mutex[((unsigned long)(v)) %\
2603 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
2604 extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ];
2606 #define EXT4_RESIZING 0
2607 extern int ext4_resize_begin(struct super_block *sb);
2608 extern void ext4_resize_end(struct super_block *sb);
2610 #endif /* __KERNEL__ */
2612 #endif /* _EXT4_H */