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/version.h>
30 #include <linux/wait.h>
31 #include <linux/blockgroup_lock.h>
32 #include <linux/percpu_counter.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(f, a...) do {} while (0)
60 #define EXT4_ERROR_INODE(inode, fmt, a...) \
61 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
63 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
64 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
66 #define EXT4_ERROR_FILE(file, block, fmt, a...) \
67 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
69 /* data type for block offset of block group */
70 typedef int ext4_grpblk_t;
72 /* data type for filesystem-wide blocks number */
73 typedef unsigned long long ext4_fsblk_t;
75 /* data type for file logical block number */
76 typedef __u32 ext4_lblk_t;
78 /* data type for block group number */
79 typedef unsigned int ext4_group_t;
82 * Flags used in mballoc's allocation_context flags field.
84 * Also used to show what's going on for debugging purposes when the
85 * flag field is exported via the traceport interface
88 /* prefer goal again. length */
89 #define EXT4_MB_HINT_MERGE 0x0001
90 /* blocks already reserved */
91 #define EXT4_MB_HINT_RESERVED 0x0002
92 /* metadata is being allocated */
93 #define EXT4_MB_HINT_METADATA 0x0004
94 /* first blocks in the file */
95 #define EXT4_MB_HINT_FIRST 0x0008
96 /* search for the best chunk */
97 #define EXT4_MB_HINT_BEST 0x0010
98 /* data is being allocated */
99 #define EXT4_MB_HINT_DATA 0x0020
100 /* don't preallocate (for tails) */
101 #define EXT4_MB_HINT_NOPREALLOC 0x0040
102 /* allocate for locality group */
103 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080
104 /* allocate goal blocks or none */
105 #define EXT4_MB_HINT_GOAL_ONLY 0x0100
106 /* goal is meaningful */
107 #define EXT4_MB_HINT_TRY_GOAL 0x0200
108 /* blocks already pre-reserved by delayed allocation */
109 #define EXT4_MB_DELALLOC_RESERVED 0x0400
110 /* We are doing stream allocation */
111 #define EXT4_MB_STREAM_ALLOC 0x0800
112 /* Use reserved root blocks if needed */
113 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000
115 struct ext4_allocation_request {
116 /* target inode for block we're allocating */
118 /* how many blocks we want to allocate */
120 /* logical block in target inode */
122 /* the closest logical allocated block to the left */
124 /* the closest logical allocated block to the right */
126 /* phys. target (a hint) */
128 /* phys. block for the closest logical allocated block to the left */
130 /* phys. block for the closest logical allocated block to the right */
132 /* flags. see above EXT4_MB_HINT_* */
137 * Logical to physical block mapping, used by ext4_map_blocks()
139 * This structure is used to pass requests into ext4_map_blocks() as
140 * well as to store the information returned by ext4_map_blocks(). It
141 * takes less room on the stack than a struct buffer_head.
143 #define EXT4_MAP_NEW (1 << BH_New)
144 #define EXT4_MAP_MAPPED (1 << BH_Mapped)
145 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten)
146 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary)
147 #define EXT4_MAP_UNINIT (1 << BH_Uninit)
148 /* Sometimes (in the bigalloc case, from ext4_da_get_block_prep) the caller of
149 * ext4_map_blocks wants to know whether or not the underlying cluster has
150 * already been accounted for. EXT4_MAP_FROM_CLUSTER conveys to the caller that
151 * the requested mapping was from previously mapped (or delayed allocated)
152 * cluster. We use BH_AllocFromCluster only for this flag. BH_AllocFromCluster
153 * should never appear on buffer_head's state flags.
155 #define EXT4_MAP_FROM_CLUSTER (1 << BH_AllocFromCluster)
156 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
157 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\
158 EXT4_MAP_UNINIT | EXT4_MAP_FROM_CLUSTER)
160 struct ext4_map_blocks {
164 unsigned int m_flags;
168 * For delayed allocation tracking
170 struct mpage_da_data {
172 sector_t b_blocknr; /* start block number of extent */
173 size_t b_size; /* size of extent */
174 unsigned long b_state; /* state of the extent */
175 unsigned long first_page, next_page; /* extent of pages */
176 struct writeback_control *wbc;
183 * Flags for ext4_io_end->flags
185 #define EXT4_IO_END_UNWRITTEN 0x0001
186 #define EXT4_IO_END_ERROR 0x0002
187 #define EXT4_IO_END_QUEUED 0x0004
188 #define EXT4_IO_END_DIRECT 0x0008
189 #define EXT4_IO_END_IN_FSYNC 0x0010
191 struct ext4_io_page {
196 #define MAX_IO_PAGES 128
198 typedef struct ext4_io_end {
199 struct list_head list; /* per-file finished IO list */
200 struct inode *inode; /* file being written to */
201 unsigned int flag; /* unwritten or not */
202 struct page *page; /* page struct for buffer write */
203 loff_t offset; /* offset in the file */
204 ssize_t size; /* size of the extent */
205 struct work_struct work; /* data work queue */
206 struct kiocb *iocb; /* iocb struct for AIO */
207 int result; /* error value for AIO */
209 struct ext4_io_page *pages[MAX_IO_PAGES];
212 struct ext4_io_submit {
215 ext4_io_end_t *io_end;
216 struct ext4_io_page *io_page;
217 sector_t io_next_block;
221 * Special inodes numbers
223 #define EXT4_BAD_INO 1 /* Bad blocks inode */
224 #define EXT4_ROOT_INO 2 /* Root inode */
225 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */
226 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
227 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
228 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
229 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
230 #define EXT4_JOURNAL_INO 8 /* Journal inode */
232 /* First non-reserved inode for old ext4 filesystems */
233 #define EXT4_GOOD_OLD_FIRST_INO 11
236 * Maximal count of links to a file
238 #define EXT4_LINK_MAX 65000
241 * Macro-instructions used to manage several block sizes
243 #define EXT4_MIN_BLOCK_SIZE 1024
244 #define EXT4_MAX_BLOCK_SIZE 65536
245 #define EXT4_MIN_BLOCK_LOG_SIZE 10
246 #define EXT4_MAX_BLOCK_LOG_SIZE 16
248 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
250 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
252 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
253 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
254 EXT4_SB(s)->s_cluster_bits)
256 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
257 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
259 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
262 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
263 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
264 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
266 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
267 EXT4_GOOD_OLD_INODE_SIZE : \
269 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
270 EXT4_GOOD_OLD_FIRST_INO : \
273 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
275 /* Translate a block number to a cluster number */
276 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
277 /* Translate a cluster number to a block number */
278 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
279 /* Translate # of blks to # of clusters */
280 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
281 (sbi)->s_cluster_bits)
282 /* Mask out the low bits to get the starting block of the cluster */
283 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
284 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
285 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
286 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
287 /* Get the cluster offset */
288 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
289 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
290 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
291 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
294 * Structure of a blocks group descriptor
296 struct ext4_group_desc
298 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
299 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
300 __le32 bg_inode_table_lo; /* Inodes table block */
301 __le16 bg_free_blocks_count_lo;/* Free blocks count */
302 __le16 bg_free_inodes_count_lo;/* Free inodes count */
303 __le16 bg_used_dirs_count_lo; /* Directories count */
304 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
305 __u32 bg_reserved[2]; /* Likely block/inode bitmap checksum */
306 __le16 bg_itable_unused_lo; /* Unused inodes count */
307 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
308 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
309 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
310 __le32 bg_inode_table_hi; /* Inodes table block MSB */
311 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
312 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
313 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
314 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
315 __u32 bg_reserved2[3];
319 * Structure of a flex block group info
323 atomic64_t free_clusters;
324 atomic_t free_inodes;
328 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
329 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
330 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
333 * Macro-instructions used to manage group descriptors
335 #define EXT4_MIN_DESC_SIZE 32
336 #define EXT4_MIN_DESC_SIZE_64BIT 64
337 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
338 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
340 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
341 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
342 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
343 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
344 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
346 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
347 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
348 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
352 * Constants relative to the data blocks
354 #define EXT4_NDIR_BLOCKS 12
355 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
356 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
357 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
358 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
363 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
364 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
365 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
366 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
367 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
368 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
369 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
370 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
371 /* Reserved for compression usage... */
372 #define EXT4_DIRTY_FL 0x00000100
373 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
374 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
375 #define EXT4_ECOMPR_FL 0x00000800 /* Compression error */
376 /* End compression flags --- maybe not all used */
377 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
378 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
379 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
380 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
381 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
382 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
383 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
384 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
385 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
386 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
387 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
389 #define EXT4_FL_USER_VISIBLE 0x004BDFFF /* User visible flags */
390 #define EXT4_FL_USER_MODIFIABLE 0x004B80FF /* User modifiable flags */
392 /* Flags that should be inherited by new inodes from their parent. */
393 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
394 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
395 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
396 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
398 /* Flags that are appropriate for regular files (all but dir-specific ones). */
399 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
401 /* Flags that are appropriate for non-directories/regular files. */
402 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
404 /* Mask out flags that are inappropriate for the given type of inode. */
405 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
409 else if (S_ISREG(mode))
410 return flags & EXT4_REG_FLMASK;
412 return flags & EXT4_OTHER_FLMASK;
416 * Inode flags used for atomic set/get
419 EXT4_INODE_SECRM = 0, /* Secure deletion */
420 EXT4_INODE_UNRM = 1, /* Undelete */
421 EXT4_INODE_COMPR = 2, /* Compress file */
422 EXT4_INODE_SYNC = 3, /* Synchronous updates */
423 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
424 EXT4_INODE_APPEND = 5, /* writes to file may only append */
425 EXT4_INODE_NODUMP = 6, /* do not dump file */
426 EXT4_INODE_NOATIME = 7, /* do not update atime */
427 /* Reserved for compression usage... */
428 EXT4_INODE_DIRTY = 8,
429 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
430 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
431 EXT4_INODE_ECOMPR = 11, /* Compression error */
432 /* End compression flags --- maybe not all used */
433 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
434 EXT4_INODE_IMAGIC = 13, /* AFS directory */
435 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
436 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
437 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
438 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
439 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
440 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
441 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
442 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
443 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
446 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
447 #define CHECK_FLAG_VALUE(FLAG) if (!TEST_FLAG_VALUE(FLAG)) { \
448 printk(KERN_EMERG "EXT4 flag fail: " #FLAG ": %d %d\n", \
449 EXT4_##FLAG##_FL, EXT4_INODE_##FLAG); BUG_ON(1); }
452 * Since it's pretty easy to mix up bit numbers and hex values, and we
453 * can't do a compile-time test for ENUM values, we use a run-time
454 * test to make sure that EXT4_XXX_FL is consistent with respect to
455 * EXT4_INODE_XXX. If all is well the printk and BUG_ON will all drop
456 * out so it won't cost any extra space in the compiled kernel image.
457 * But it's important that these values are the same, since we are
458 * using EXT4_INODE_XXX to test for the flag values, but EXT4_XX_FL
459 * must be consistent with the values of FS_XXX_FL defined in
460 * include/linux/fs.h and the on-disk values found in ext2, ext3, and
461 * ext4 filesystems, and of course the values defined in e2fsprogs.
463 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
465 static inline void ext4_check_flag_values(void)
467 CHECK_FLAG_VALUE(SECRM);
468 CHECK_FLAG_VALUE(UNRM);
469 CHECK_FLAG_VALUE(COMPR);
470 CHECK_FLAG_VALUE(SYNC);
471 CHECK_FLAG_VALUE(IMMUTABLE);
472 CHECK_FLAG_VALUE(APPEND);
473 CHECK_FLAG_VALUE(NODUMP);
474 CHECK_FLAG_VALUE(NOATIME);
475 CHECK_FLAG_VALUE(DIRTY);
476 CHECK_FLAG_VALUE(COMPRBLK);
477 CHECK_FLAG_VALUE(NOCOMPR);
478 CHECK_FLAG_VALUE(ECOMPR);
479 CHECK_FLAG_VALUE(INDEX);
480 CHECK_FLAG_VALUE(IMAGIC);
481 CHECK_FLAG_VALUE(JOURNAL_DATA);
482 CHECK_FLAG_VALUE(NOTAIL);
483 CHECK_FLAG_VALUE(DIRSYNC);
484 CHECK_FLAG_VALUE(TOPDIR);
485 CHECK_FLAG_VALUE(HUGE_FILE);
486 CHECK_FLAG_VALUE(EXTENTS);
487 CHECK_FLAG_VALUE(EA_INODE);
488 CHECK_FLAG_VALUE(EOFBLOCKS);
489 CHECK_FLAG_VALUE(RESERVED);
492 /* Used to pass group descriptor data when online resize is done */
493 struct ext4_new_group_input {
494 __u32 group; /* Group number for this data */
495 __u64 block_bitmap; /* Absolute block number of block bitmap */
496 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
497 __u64 inode_table; /* Absolute block number of inode table start */
498 __u32 blocks_count; /* Total number of blocks in this group */
499 __u16 reserved_blocks; /* Number of reserved blocks in this group */
503 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
504 struct compat_ext4_new_group_input {
506 compat_u64 block_bitmap;
507 compat_u64 inode_bitmap;
508 compat_u64 inode_table;
515 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
516 struct ext4_new_group_data {
522 __u16 reserved_blocks;
524 __u32 free_blocks_count;
528 * Flags used by ext4_map_blocks()
530 /* Allocate any needed blocks and/or convert an unitialized
531 extent to be an initialized ext4 */
532 #define EXT4_GET_BLOCKS_CREATE 0x0001
533 /* Request the creation of an unitialized extent */
534 #define EXT4_GET_BLOCKS_UNINIT_EXT 0x0002
535 #define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT (EXT4_GET_BLOCKS_UNINIT_EXT|\
536 EXT4_GET_BLOCKS_CREATE)
537 /* Caller is from the delayed allocation writeout path,
538 so set the magic i_delalloc_reserve_flag after taking the
539 inode allocation semaphore for */
540 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
541 /* caller is from the direct IO path, request to creation of an
542 unitialized extents if not allocated, split the uninitialized
543 extent if blocks has been preallocated already*/
544 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
545 #define EXT4_GET_BLOCKS_CONVERT 0x0010
546 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
547 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
548 /* Convert extent to initialized after IO complete */
549 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
550 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
551 /* Punch out blocks of an extent */
552 #define EXT4_GET_BLOCKS_PUNCH_OUT_EXT 0x0020
553 /* Don't normalize allocation size (used for fallocate) */
554 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
555 /* Request will not result in inode size update (user for fallocate) */
556 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080
559 * Flags used by ext4_free_blocks
561 #define EXT4_FREE_BLOCKS_METADATA 0x0001
562 #define EXT4_FREE_BLOCKS_FORGET 0x0002
563 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
564 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
565 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
566 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
569 * Flags used by ext4_discard_partial_page_buffers
571 #define EXT4_DISCARD_PARTIAL_PG_ZERO_UNMAPPED 0x0001
576 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
577 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
578 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
579 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
580 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
581 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
582 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
583 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
584 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
585 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
586 #define EXT4_IOC_MIGRATE _IO('f', 9)
587 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
588 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
589 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
590 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
592 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
594 * ioctl commands in 32 bit emulation
596 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
597 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
598 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
599 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
600 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
601 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
602 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
603 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
604 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
605 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
608 /* Max physical block we can address w/o extents */
609 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
612 * Structure of an inode on the disk
615 __le16 i_mode; /* File mode */
616 __le16 i_uid; /* Low 16 bits of Owner Uid */
617 __le32 i_size_lo; /* Size in bytes */
618 __le32 i_atime; /* Access time */
619 __le32 i_ctime; /* Inode Change time */
620 __le32 i_mtime; /* Modification time */
621 __le32 i_dtime; /* Deletion Time */
622 __le16 i_gid; /* Low 16 bits of Group Id */
623 __le16 i_links_count; /* Links count */
624 __le32 i_blocks_lo; /* Blocks count */
625 __le32 i_flags; /* File flags */
631 __u32 h_i_translator;
636 } osd1; /* OS dependent 1 */
637 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
638 __le32 i_generation; /* File version (for NFS) */
639 __le32 i_file_acl_lo; /* File ACL */
641 __le32 i_obso_faddr; /* Obsoleted fragment address */
644 __le16 l_i_blocks_high; /* were l_i_reserved1 */
645 __le16 l_i_file_acl_high;
646 __le16 l_i_uid_high; /* these 2 fields */
647 __le16 l_i_gid_high; /* were reserved2[0] */
651 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
658 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
659 __le16 m_i_file_acl_high;
660 __u32 m_i_reserved2[2];
662 } osd2; /* OS dependent 2 */
663 __le16 i_extra_isize;
665 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
666 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
667 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
668 __le32 i_crtime; /* File Creation time */
669 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
670 __le32 i_version_hi; /* high 32 bits for 64-bit version */
674 __u32 reserved; /* should be zero */
675 __u32 donor_fd; /* donor file descriptor */
676 __u64 orig_start; /* logical start offset in block for orig */
677 __u64 donor_start; /* logical start offset in block for donor */
678 __u64 len; /* block length to be moved */
679 __u64 moved_len; /* moved block length */
682 #define EXT4_EPOCH_BITS 2
683 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
684 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
687 * Extended fields will fit into an inode if the filesystem was formatted
688 * with large inodes (-I 256 or larger) and there are not currently any EAs
689 * consuming all of the available space. For new inodes we always reserve
690 * enough space for the kernel's known extended fields, but for inodes
691 * created with an old kernel this might not have been the case. None of
692 * the extended inode fields is critical for correct filesystem operation.
693 * This macro checks if a certain field fits in the inode. Note that
694 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
696 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
697 ((offsetof(typeof(*ext4_inode), field) + \
698 sizeof((ext4_inode)->field)) \
699 <= (EXT4_GOOD_OLD_INODE_SIZE + \
700 (einode)->i_extra_isize)) \
703 * We use an encoding that preserves the times for extra epoch "00":
705 * extra msb of adjust for signed
706 * epoch 32-bit 32-bit tv_sec to
707 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
708 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
709 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
710 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
711 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
712 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
713 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
714 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
715 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
717 * Note that previous versions of the kernel on 64-bit systems would
718 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
719 * 1970. e2fsck will correct this, assuming that it is run on the
720 * affected filesystem before 2242.
723 static inline __le32 ext4_encode_extra_time(struct timespec *time)
725 u32 extra = sizeof(time->tv_sec) > 4 ?
726 ((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK : 0;
727 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
730 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
732 if (unlikely(sizeof(time->tv_sec) > 4 &&
733 (extra & cpu_to_le32(EXT4_EPOCH_MASK)))) {
734 #if LINUX_VERSION_CODE < KERNEL_VERSION(4,20,0)
735 /* Handle legacy encoding of pre-1970 dates with epoch
736 * bits 1,1. We assume that by kernel version 4.20,
737 * everyone will have run fsck over the affected
738 * filesystems to correct the problem. (This
739 * backwards compatibility may be removed before this
740 * time, at the discretion of the ext4 developers.)
742 u64 extra_bits = le32_to_cpu(extra) & EXT4_EPOCH_MASK;
743 if (extra_bits == 3 && ((time->tv_sec) & 0x80000000) != 0)
745 time->tv_sec += extra_bits << 32;
747 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
750 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
753 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
755 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
756 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
757 (raw_inode)->xtime ## _extra = \
758 ext4_encode_extra_time(&(inode)->xtime); \
761 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
763 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
764 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
765 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
766 (raw_inode)->xtime ## _extra = \
767 ext4_encode_extra_time(&(einode)->xtime); \
770 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
772 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
773 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
774 ext4_decode_extra_time(&(inode)->xtime, \
775 raw_inode->xtime ## _extra); \
777 (inode)->xtime.tv_nsec = 0; \
780 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
782 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
783 (einode)->xtime.tv_sec = \
784 (signed)le32_to_cpu((raw_inode)->xtime); \
786 (einode)->xtime.tv_sec = 0; \
787 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
788 ext4_decode_extra_time(&(einode)->xtime, \
789 raw_inode->xtime ## _extra); \
791 (einode)->xtime.tv_nsec = 0; \
794 #define i_disk_version osd1.linux1.l_i_version
796 #if defined(__KERNEL__) || defined(__linux__)
797 #define i_reserved1 osd1.linux1.l_i_reserved1
798 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
799 #define i_blocks_high osd2.linux2.l_i_blocks_high
800 #define i_uid_low i_uid
801 #define i_gid_low i_gid
802 #define i_uid_high osd2.linux2.l_i_uid_high
803 #define i_gid_high osd2.linux2.l_i_gid_high
804 #define i_reserved2 osd2.linux2.l_i_reserved2
806 #elif defined(__GNU__)
808 #define i_translator osd1.hurd1.h_i_translator
809 #define i_uid_high osd2.hurd2.h_i_uid_high
810 #define i_gid_high osd2.hurd2.h_i_gid_high
811 #define i_author osd2.hurd2.h_i_author
813 #elif defined(__masix__)
815 #define i_reserved1 osd1.masix1.m_i_reserved1
816 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
817 #define i_reserved2 osd2.masix2.m_i_reserved2
819 #endif /* defined(__KERNEL__) || defined(__linux__) */
822 * storage for cached extent
823 * If ec_len == 0, then the cache is invalid.
824 * If ec_start == 0, then the cache represents a gap (null mapping)
826 struct ext4_ext_cache {
827 ext4_fsblk_t ec_start;
828 ext4_lblk_t ec_block;
829 __u32 ec_len; /* must be 32bit to return holes */
833 * fourth extended file system inode data in memory
835 struct ext4_inode_info {
836 __le32 i_data[15]; /* unconverted */
838 ext4_fsblk_t i_file_acl;
841 * i_block_group is the number of the block group which contains
842 * this file's inode. Constant across the lifetime of the inode,
843 * it is ued for making block allocation decisions - we try to
844 * place a file's data blocks near its inode block, and new inodes
845 * near to their parent directory's inode.
847 ext4_group_t i_block_group;
848 ext4_lblk_t i_dir_start_lookup;
849 #if (BITS_PER_LONG < 64)
850 unsigned long i_state_flags; /* Dynamic state flags */
852 unsigned long i_flags;
854 #ifdef CONFIG_EXT4_FS_XATTR
856 * Extended attributes can be read independently of the main file
857 * data. Taking i_mutex even when reading would cause contention
858 * between readers of EAs and writers of regular file data, so
859 * instead we synchronize on xattr_sem when reading or changing
862 struct rw_semaphore xattr_sem;
865 struct list_head i_orphan; /* unlinked but open inodes */
868 * i_disksize keeps track of what the inode size is ON DISK, not
869 * in memory. During truncate, i_size is set to the new size by
870 * the VFS prior to calling ext4_truncate(), but the filesystem won't
871 * set i_disksize to 0 until the truncate is actually under way.
873 * The intent is that i_disksize always represents the blocks which
874 * are used by this file. This allows recovery to restart truncate
875 * on orphans if we crash during truncate. We actually write i_disksize
876 * into the on-disk inode when writing inodes out, instead of i_size.
878 * The only time when i_disksize and i_size may be different is when
879 * a truncate is in progress. The only things which change i_disksize
880 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
885 * i_data_sem is for serialising ext4_truncate() against
886 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
887 * data tree are chopped off during truncate. We can't do that in
888 * ext4 because whenever we perform intermediate commits during
889 * truncate, the inode and all the metadata blocks *must* be in a
890 * consistent state which allows truncation of the orphans to restart
891 * during recovery. Hence we must fix the get_block-vs-truncate race
892 * by other means, so we have i_data_sem.
894 struct rw_semaphore i_data_sem;
895 struct inode vfs_inode;
896 struct jbd2_inode *jinode;
898 struct ext4_ext_cache i_cached_extent;
900 * File creation time. Its function is same as that of
901 * struct timespec i_{a,c,m}time in the generic inode.
903 struct timespec i_crtime;
906 struct list_head i_prealloc_list;
907 spinlock_t i_prealloc_lock;
910 ext4_group_t i_last_alloc_group;
912 /* allocation reservation info for delalloc */
913 /* In case of bigalloc, these refer to clusters rather than blocks */
914 unsigned int i_reserved_data_blocks;
915 unsigned int i_reserved_meta_blocks;
916 unsigned int i_allocated_meta_blocks;
917 ext4_lblk_t i_da_metadata_calc_last_lblock;
918 int i_da_metadata_calc_len;
920 /* on-disk additional length */
924 /* quota space reservation, managed internally by quota code */
925 qsize_t i_reserved_quota;
928 /* completed IOs that might need unwritten extents handling */
929 struct list_head i_completed_io_list;
930 spinlock_t i_completed_io_lock;
931 atomic_t i_ioend_count; /* Number of outstanding io_end structs */
932 /* current io_end structure for async DIO write*/
933 ext4_io_end_t *cur_aio_dio;
934 atomic_t i_aiodio_unwritten; /* Nr. of inflight conversions pending */
936 spinlock_t i_block_reservation_lock;
939 * Transactions that contain inode's metadata needed to complete
940 * fsync and fdatasync, respectively.
943 tid_t i_datasync_tid;
949 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
950 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
951 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
954 * Misc. filesystem flags
956 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
957 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
958 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
963 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
964 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
965 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
966 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
967 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
968 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
969 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
970 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
971 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
972 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
973 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
974 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
975 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
976 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
977 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
978 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
979 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
980 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
981 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
982 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
983 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
984 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
985 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
986 #define EXT4_MOUNT_I_VERSION 0x2000000 /* i_version support */
987 #define EXT4_MOUNT_MBLK_IO_SUBMIT 0x4000000 /* multi-block io submits */
988 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
989 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
990 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
991 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
992 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
994 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
995 specified delalloc */
997 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
999 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1001 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1004 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1006 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1008 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1011 #define ext4_set_bit __test_and_set_bit_le
1012 #define ext4_set_bit_atomic ext2_set_bit_atomic
1013 #define ext4_clear_bit __test_and_clear_bit_le
1014 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
1015 #define ext4_test_bit test_bit_le
1016 #define ext4_find_first_zero_bit find_first_zero_bit_le
1017 #define ext4_find_next_zero_bit find_next_zero_bit_le
1018 #define ext4_find_next_bit find_next_bit_le
1020 extern void ext4_set_bits(void *bm, int cur, int len);
1023 * Maximal mount counts between two filesystem checks
1025 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1026 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1029 * Behaviour when detecting errors
1031 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1032 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1033 #define EXT4_ERRORS_PANIC 3 /* Panic */
1034 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1037 * Structure of the super block
1039 struct ext4_super_block {
1040 /*00*/ __le32 s_inodes_count; /* Inodes count */
1041 __le32 s_blocks_count_lo; /* Blocks count */
1042 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1043 __le32 s_free_blocks_count_lo; /* Free blocks count */
1044 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1045 __le32 s_first_data_block; /* First Data Block */
1046 __le32 s_log_block_size; /* Block size */
1047 __le32 s_log_cluster_size; /* Allocation cluster size */
1048 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1049 __le32 s_clusters_per_group; /* # Clusters per group */
1050 __le32 s_inodes_per_group; /* # Inodes per group */
1051 __le32 s_mtime; /* Mount time */
1052 /*30*/ __le32 s_wtime; /* Write time */
1053 __le16 s_mnt_count; /* Mount count */
1054 __le16 s_max_mnt_count; /* Maximal mount count */
1055 __le16 s_magic; /* Magic signature */
1056 __le16 s_state; /* File system state */
1057 __le16 s_errors; /* Behaviour when detecting errors */
1058 __le16 s_minor_rev_level; /* minor revision level */
1059 /*40*/ __le32 s_lastcheck; /* time of last check */
1060 __le32 s_checkinterval; /* max. time between checks */
1061 __le32 s_creator_os; /* OS */
1062 __le32 s_rev_level; /* Revision level */
1063 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1064 __le16 s_def_resgid; /* Default gid for reserved blocks */
1066 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1068 * Note: the difference between the compatible feature set and
1069 * the incompatible feature set is that if there is a bit set
1070 * in the incompatible feature set that the kernel doesn't
1071 * know about, it should refuse to mount the filesystem.
1073 * e2fsck's requirements are more strict; if it doesn't know
1074 * about a feature in either the compatible or incompatible
1075 * feature set, it must abort and not try to meddle with
1076 * things it doesn't understand...
1078 __le32 s_first_ino; /* First non-reserved inode */
1079 __le16 s_inode_size; /* size of inode structure */
1080 __le16 s_block_group_nr; /* block group # of this superblock */
1081 __le32 s_feature_compat; /* compatible feature set */
1082 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1083 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1084 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1085 /*78*/ char s_volume_name[16]; /* volume name */
1086 /*88*/ char s_last_mounted[64]; /* directory where last mounted */
1087 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1089 * Performance hints. Directory preallocation should only
1090 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1092 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1093 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1094 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1096 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1098 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1099 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1100 __le32 s_journal_dev; /* device number of journal file */
1101 __le32 s_last_orphan; /* start of list of inodes to delete */
1102 __le32 s_hash_seed[4]; /* HTREE hash seed */
1103 __u8 s_def_hash_version; /* Default hash version to use */
1104 __u8 s_jnl_backup_type;
1105 __le16 s_desc_size; /* size of group descriptor */
1106 /*100*/ __le32 s_default_mount_opts;
1107 __le32 s_first_meta_bg; /* First metablock block group */
1108 __le32 s_mkfs_time; /* When the filesystem was created */
1109 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1110 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1111 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1112 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1113 __le32 s_free_blocks_count_hi; /* Free blocks count */
1114 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1115 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1116 __le32 s_flags; /* Miscellaneous flags */
1117 __le16 s_raid_stride; /* RAID stride */
1118 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1119 __le64 s_mmp_block; /* Block for multi-mount protection */
1120 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1121 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1122 __u8 s_reserved_char_pad;
1123 __le16 s_reserved_pad;
1124 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1125 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1126 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1127 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1128 snapshot's future use */
1129 __le32 s_snapshot_list; /* inode number of the head of the
1130 on-disk snapshot list */
1131 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1132 __le32 s_error_count; /* number of fs errors */
1133 __le32 s_first_error_time; /* first time an error happened */
1134 __le32 s_first_error_ino; /* inode involved in first error */
1135 __le64 s_first_error_block; /* block involved of first error */
1136 __u8 s_first_error_func[32]; /* function where the error happened */
1137 __le32 s_first_error_line; /* line number where error happened */
1138 __le32 s_last_error_time; /* most recent time of an error */
1139 __le32 s_last_error_ino; /* inode involved in last error */
1140 __le32 s_last_error_line; /* line number where error happened */
1141 __le64 s_last_error_block; /* block involved of last error */
1142 __u8 s_last_error_func[32]; /* function where the error happened */
1143 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1144 __u8 s_mount_opts[64];
1145 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1146 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1147 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1148 __le32 s_reserved[109]; /* Padding to the end of the block */
1151 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1156 * run-time mount flags
1158 #define EXT4_MF_MNTDIR_SAMPLED 0x0001
1159 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1162 * fourth extended-fs super-block data in memory
1164 struct ext4_sb_info {
1165 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1166 unsigned long s_inodes_per_block;/* Number of inodes per block */
1167 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1168 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1169 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1170 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1171 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1172 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1173 ext4_group_t s_groups_count; /* Number of groups in the fs */
1174 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1175 unsigned long s_overhead; /* # of fs overhead clusters */
1176 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1177 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1178 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1179 struct buffer_head * s_sbh; /* Buffer containing the super block */
1180 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1181 struct buffer_head **s_group_desc;
1182 unsigned int s_mount_opt;
1183 unsigned int s_mount_opt2;
1184 unsigned int s_mount_flags;
1185 ext4_fsblk_t s_sb_block;
1188 unsigned short s_mount_state;
1189 unsigned short s_pad;
1190 int s_addr_per_block_bits;
1191 int s_desc_per_block_bits;
1194 unsigned int s_inode_readahead_blks;
1195 unsigned int s_inode_goal;
1196 spinlock_t s_next_gen_lock;
1197 u32 s_next_generation;
1199 int s_def_hash_version;
1200 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1201 struct percpu_counter s_freeclusters_counter;
1202 struct percpu_counter s_freeinodes_counter;
1203 struct percpu_counter s_dirs_counter;
1204 struct percpu_counter s_dirtyclusters_counter;
1205 struct blockgroup_lock *s_blockgroup_lock;
1206 struct proc_dir_entry *s_proc;
1207 struct kobject s_kobj;
1208 struct completion s_kobj_unregister;
1211 struct journal_s *s_journal;
1212 struct list_head s_orphan;
1213 struct mutex s_orphan_lock;
1214 unsigned long s_resize_flags; /* Flags indicating if there
1216 unsigned long s_commit_interval;
1217 u32 s_max_batch_time;
1218 u32 s_min_batch_time;
1219 struct block_device *journal_bdev;
1221 char *s_qf_names[MAXQUOTAS]; /* Names of quota files with journalled quota */
1222 int s_jquota_fmt; /* Format of quota to use */
1224 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1225 struct rb_root system_blks;
1227 #ifdef EXTENTS_STATS
1228 /* ext4 extents stats */
1229 unsigned long s_ext_min;
1230 unsigned long s_ext_max;
1231 unsigned long s_depth_max;
1232 spinlock_t s_ext_stats_lock;
1233 unsigned long s_ext_blocks;
1234 unsigned long s_ext_extents;
1237 /* for buddy allocator */
1238 struct ext4_group_info ***s_group_info;
1239 struct inode *s_buddy_cache;
1240 spinlock_t s_md_lock;
1241 unsigned short *s_mb_offsets;
1242 unsigned int *s_mb_maxs;
1245 unsigned long s_stripe;
1246 unsigned int s_mb_stream_request;
1247 unsigned int s_mb_max_to_scan;
1248 unsigned int s_mb_min_to_scan;
1249 unsigned int s_mb_stats;
1250 unsigned int s_mb_order2_reqs;
1251 unsigned int s_mb_group_prealloc;
1252 unsigned int s_max_writeback_mb_bump;
1253 /* where last allocation was done - for stream allocation */
1254 unsigned long s_mb_last_group;
1255 unsigned long s_mb_last_start;
1257 /* stats for buddy allocator */
1258 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1259 atomic_t s_bal_success; /* we found long enough chunks */
1260 atomic_t s_bal_allocated; /* in blocks */
1261 atomic_t s_bal_ex_scanned; /* total extents scanned */
1262 atomic_t s_bal_goals; /* goal hits */
1263 atomic_t s_bal_breaks; /* too long searches */
1264 atomic_t s_bal_2orders; /* 2^order hits */
1265 spinlock_t s_bal_lock;
1266 unsigned long s_mb_buddies_generated;
1267 unsigned long long s_mb_generation_time;
1268 atomic_t s_mb_lost_chunks;
1269 atomic_t s_mb_preallocated;
1270 atomic_t s_mb_discarded;
1271 atomic_t s_lock_busy;
1273 /* locality groups */
1274 struct ext4_locality_group __percpu *s_locality_groups;
1276 /* for write statistics */
1277 unsigned long s_sectors_written_start;
1278 u64 s_kbytes_written;
1280 unsigned int s_log_groups_per_flex;
1281 struct flex_groups *s_flex_groups;
1283 /* workqueue for dio unwritten */
1284 struct workqueue_struct *dio_unwritten_wq;
1286 /* timer for periodic error stats printing */
1287 struct timer_list s_err_report;
1289 /* Lazy inode table initialization info */
1290 struct ext4_li_request *s_li_request;
1291 /* Wait multiplier for lazy initialization thread */
1292 unsigned int s_li_wait_mult;
1294 /* Kernel thread for multiple mount protection */
1295 struct task_struct *s_mmp_tsk;
1297 /* record the last minlen when FITRIM is called. */
1298 atomic_t s_last_trim_minblks;
1301 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1303 return sb->s_fs_info;
1305 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1307 return container_of(inode, struct ext4_inode_info, vfs_inode);
1310 static inline struct timespec ext4_current_time(struct inode *inode)
1312 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1313 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1316 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1318 return ino == EXT4_ROOT_INO ||
1319 ino == EXT4_JOURNAL_INO ||
1320 ino == EXT4_RESIZE_INO ||
1321 (ino >= EXT4_FIRST_INO(sb) &&
1322 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1325 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
1326 struct ext4_io_end *io_end)
1328 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
1329 io_end->flag |= EXT4_IO_END_UNWRITTEN;
1330 atomic_inc(&EXT4_I(inode)->i_aiodio_unwritten);
1335 * Inode dynamic state flags
1338 EXT4_STATE_JDATA, /* journaled data exists */
1339 EXT4_STATE_NEW, /* inode is newly created */
1340 EXT4_STATE_XATTR, /* has in-inode xattrs */
1341 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1342 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1343 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1344 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
1345 EXT4_STATE_NEWENTRY, /* File just added to dir */
1346 EXT4_STATE_DELALLOC_RESERVED, /* blks already reserved for delalloc */
1349 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1350 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1352 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1354 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1356 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1358 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1360 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1363 EXT4_INODE_BIT_FNS(flag, flags, 0)
1364 #if (BITS_PER_LONG < 64)
1365 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1367 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1369 (ei)->i_state_flags = 0;
1372 EXT4_INODE_BIT_FNS(state, flags, 32)
1374 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1376 /* We depend on the fact that callers will set i_flags */
1380 /* Assume that user mode programs are passing in an ext4fs superblock, not
1381 * a kernel struct super_block. This will allow us to call the feature-test
1382 * macros from user land. */
1383 #define EXT4_SB(sb) (sb)
1386 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1389 * Codes for operating systems
1391 #define EXT4_OS_LINUX 0
1392 #define EXT4_OS_HURD 1
1393 #define EXT4_OS_MASIX 2
1394 #define EXT4_OS_FREEBSD 3
1395 #define EXT4_OS_LITES 4
1400 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1401 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1403 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1404 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1406 #define EXT4_GOOD_OLD_INODE_SIZE 128
1409 * Feature set definitions
1412 #define EXT4_HAS_COMPAT_FEATURE(sb,mask) \
1413 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1414 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \
1415 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1416 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \
1417 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1418 #define EXT4_SET_COMPAT_FEATURE(sb,mask) \
1419 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1420 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \
1421 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1422 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \
1423 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1424 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \
1425 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1426 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
1427 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1428 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \
1429 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1431 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1432 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1433 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1434 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1435 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1436 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1438 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1439 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1440 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1441 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1442 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1443 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1444 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1445 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1446 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1448 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1449 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1450 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1451 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1452 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1453 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1454 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1455 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1456 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1457 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1458 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1460 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1461 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1462 EXT4_FEATURE_INCOMPAT_META_BG)
1463 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1464 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1465 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1467 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1468 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1469 EXT4_FEATURE_INCOMPAT_RECOVER| \
1470 EXT4_FEATURE_INCOMPAT_META_BG)
1471 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1472 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1473 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1475 #define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR
1476 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1477 EXT4_FEATURE_INCOMPAT_RECOVER| \
1478 EXT4_FEATURE_INCOMPAT_META_BG| \
1479 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1480 EXT4_FEATURE_INCOMPAT_64BIT| \
1481 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1482 EXT4_FEATURE_INCOMPAT_MMP)
1483 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1484 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1485 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1486 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1487 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1488 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1489 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1490 EXT4_FEATURE_RO_COMPAT_BIGALLOC)
1493 * Default values for user and/or group using reserved blocks
1495 #define EXT4_DEF_RESUID 0
1496 #define EXT4_DEF_RESGID 0
1498 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
1501 * Default mount options
1503 #define EXT4_DEFM_DEBUG 0x0001
1504 #define EXT4_DEFM_BSDGROUPS 0x0002
1505 #define EXT4_DEFM_XATTR_USER 0x0004
1506 #define EXT4_DEFM_ACL 0x0008
1507 #define EXT4_DEFM_UID16 0x0010
1508 #define EXT4_DEFM_JMODE 0x0060
1509 #define EXT4_DEFM_JMODE_DATA 0x0020
1510 #define EXT4_DEFM_JMODE_ORDERED 0x0040
1511 #define EXT4_DEFM_JMODE_WBACK 0x0060
1512 #define EXT4_DEFM_NOBARRIER 0x0100
1513 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1514 #define EXT4_DEFM_DISCARD 0x0400
1515 #define EXT4_DEFM_NODELALLOC 0x0800
1518 * Default journal batch times
1520 #define EXT4_DEF_MIN_BATCH_TIME 0
1521 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1524 * Minimum number of groups in a flexgroup before we separate out
1525 * directories into the first block group of a flexgroup
1527 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1530 * Structure of a directory entry
1532 #define EXT4_NAME_LEN 255
1534 struct ext4_dir_entry {
1535 __le32 inode; /* Inode number */
1536 __le16 rec_len; /* Directory entry length */
1537 __le16 name_len; /* Name length */
1538 char name[EXT4_NAME_LEN]; /* File name */
1542 * The new version of the directory entry. Since EXT4 structures are
1543 * stored in intel byte order, and the name_len field could never be
1544 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1547 struct ext4_dir_entry_2 {
1548 __le32 inode; /* Inode number */
1549 __le16 rec_len; /* Directory entry length */
1550 __u8 name_len; /* Name length */
1552 char name[EXT4_NAME_LEN]; /* File name */
1556 * Ext4 directory file types. Only the low 3 bits are used. The
1557 * other bits are reserved for now.
1559 #define EXT4_FT_UNKNOWN 0
1560 #define EXT4_FT_REG_FILE 1
1561 #define EXT4_FT_DIR 2
1562 #define EXT4_FT_CHRDEV 3
1563 #define EXT4_FT_BLKDEV 4
1564 #define EXT4_FT_FIFO 5
1565 #define EXT4_FT_SOCK 6
1566 #define EXT4_FT_SYMLINK 7
1568 #define EXT4_FT_MAX 8
1571 * EXT4_DIR_PAD defines the directory entries boundaries
1573 * NOTE: It must be a multiple of 4
1575 #define EXT4_DIR_PAD 4
1576 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1577 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1579 #define EXT4_MAX_REC_LEN ((1<<16)-1)
1582 * If we ever get support for fs block sizes > page_size, we'll need
1583 * to remove the #if statements in the next two functions...
1585 static inline unsigned int
1586 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1588 unsigned len = le16_to_cpu(dlen);
1590 #if (PAGE_CACHE_SIZE >= 65536)
1591 if (len == EXT4_MAX_REC_LEN || len == 0)
1593 return (len & 65532) | ((len & 3) << 16);
1599 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1601 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1603 #if (PAGE_CACHE_SIZE >= 65536)
1605 return cpu_to_le16(len);
1606 if (len == blocksize) {
1607 if (blocksize == 65536)
1608 return cpu_to_le16(EXT4_MAX_REC_LEN);
1610 return cpu_to_le16(0);
1612 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1614 return cpu_to_le16(len);
1619 * Hash Tree Directory indexing
1620 * (c) Daniel Phillips, 2001
1623 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1624 EXT4_FEATURE_COMPAT_DIR_INDEX) && \
1625 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
1626 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1627 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1629 /* Legal values for the dx_root hash_version field: */
1631 #define DX_HASH_LEGACY 0
1632 #define DX_HASH_HALF_MD4 1
1633 #define DX_HASH_TEA 2
1634 #define DX_HASH_LEGACY_UNSIGNED 3
1635 #define DX_HASH_HALF_MD4_UNSIGNED 4
1636 #define DX_HASH_TEA_UNSIGNED 5
1640 /* hash info structure used by the directory hash */
1650 /* 32 and 64 bit signed EOF for dx directories */
1651 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
1652 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
1656 * Control parameters used by ext4_htree_next_block
1658 #define HASH_NB_ALWAYS 1
1662 * Describe an inode's exact location on disk and in memory
1666 struct buffer_head *bh;
1667 unsigned long offset;
1668 ext4_group_t block_group;
1671 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
1673 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
1677 * This structure is stuffed into the struct file's private_data field
1678 * for directories. It is where we put information so that we can do
1679 * readdir operations in hash tree order.
1681 struct dir_private_info {
1682 struct rb_root root;
1683 struct rb_node *curr_node;
1684 struct fname *extra_fname;
1687 __u32 curr_minor_hash;
1691 /* calculate the first block number of the group */
1692 static inline ext4_fsblk_t
1693 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
1695 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
1696 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
1700 * Special error return code only used by dx_probe() and its callers.
1702 #define ERR_BAD_DX_DIR -75000
1704 void ext4_get_group_no_and_offset(struct super_block *sb, ext4_fsblk_t blocknr,
1705 ext4_group_t *blockgrpp, ext4_grpblk_t *offsetp);
1708 * Timeout and state flag for lazy initialization inode thread.
1710 #define EXT4_DEF_LI_WAIT_MULT 10
1711 #define EXT4_DEF_LI_MAX_START_DELAY 5
1712 #define EXT4_LAZYINIT_QUIT 0x0001
1713 #define EXT4_LAZYINIT_RUNNING 0x0002
1716 * Lazy inode table initialization info
1718 struct ext4_lazy_init {
1719 unsigned long li_state;
1720 struct list_head li_request_list;
1721 struct mutex li_list_mtx;
1724 struct ext4_li_request {
1725 struct super_block *lr_super;
1726 struct ext4_sb_info *lr_sbi;
1727 ext4_group_t lr_next_group;
1728 struct list_head lr_request;
1729 unsigned long lr_next_sched;
1730 unsigned long lr_timeout;
1733 struct ext4_features {
1734 struct kobject f_kobj;
1735 struct completion f_kobj_unregister;
1739 * This structure will be used for multiple mount protection. It will be
1740 * written into the block number saved in the s_mmp_block field in the
1741 * superblock. Programs that check MMP should assume that if
1742 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
1743 * to use the filesystem, regardless of how old the timestamp is.
1745 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
1746 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
1747 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
1748 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
1751 __le32 mmp_magic; /* Magic number for MMP */
1752 __le32 mmp_seq; /* Sequence no. updated periodically */
1755 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
1756 * purposes and do not affect the correctness of the algorithm
1758 __le64 mmp_time; /* Time last updated */
1759 char mmp_nodename[64]; /* Node which last updated MMP block */
1760 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
1763 * mmp_check_interval is used to verify if the MMP block has been
1764 * updated on the block device. The value is updated based on the
1765 * maximum time to write the MMP block during an update cycle.
1767 __le16 mmp_check_interval;
1770 __le32 mmp_pad2[227];
1773 /* arguments passed to the mmp thread */
1775 struct buffer_head *bh; /* bh from initial read_mmp_block() */
1776 struct super_block *sb; /* super block of the fs */
1780 * Check interval multiplier
1781 * The MMP block is written every update interval and initially checked every
1782 * update interval x the multiplier (the value is then adapted based on the
1783 * write latency). The reason is that writes can be delayed under load and we
1784 * don't want readers to incorrectly assume that the filesystem is no longer
1787 #define EXT4_MMP_CHECK_MULT 2UL
1790 * Minimum interval for MMP checking in seconds.
1792 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
1795 * Maximum interval for MMP checking in seconds.
1797 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
1800 * Function prototypes
1804 * Ok, these declarations are also in <linux/kernel.h> but none of the
1805 * ext4 source programs needs to include it so they are duplicated here.
1807 # define NORET_TYPE /**/
1808 # define ATTRIB_NORET __attribute__((noreturn))
1809 # define NORET_AND noreturn,
1812 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
1815 extern unsigned int ext4_block_group(struct super_block *sb,
1816 ext4_fsblk_t blocknr);
1817 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
1818 ext4_fsblk_t blocknr);
1819 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
1820 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
1821 ext4_group_t group);
1822 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
1825 unsigned long *count,
1827 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
1828 s64 nclusters, unsigned int flags);
1829 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
1830 extern void ext4_check_blocks_bitmap(struct super_block *);
1831 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
1832 ext4_group_t block_group,
1833 struct buffer_head ** bh);
1834 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
1835 struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
1836 ext4_group_t block_group);
1837 extern void ext4_init_block_bitmap(struct super_block *sb,
1838 struct buffer_head *bh,
1840 struct ext4_group_desc *desc);
1841 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
1842 ext4_group_t block_group,
1843 struct ext4_group_desc *gdp);
1844 extern unsigned ext4_num_base_meta_clusters(struct super_block *sb,
1845 ext4_group_t block_group);
1846 extern unsigned ext4_num_overhead_clusters(struct super_block *sb,
1847 ext4_group_t block_group,
1848 struct ext4_group_desc *gdp);
1849 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
1852 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
1854 struct ext4_dir_entry_2 *,
1855 struct buffer_head *, unsigned int);
1856 #define ext4_check_dir_entry(dir, filp, de, bh, offset) \
1857 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
1858 (de), (bh), (offset)))
1859 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
1861 struct ext4_dir_entry_2 *dirent);
1862 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
1865 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
1866 extern int ext4_flush_completed_IO(struct inode *);
1869 extern int ext4fs_dirhash(const char *name, int len, struct
1870 dx_hash_info *hinfo);
1873 extern struct inode *ext4_new_inode(handle_t *, struct inode *, int,
1874 const struct qstr *qstr, __u32 goal,
1876 extern void ext4_free_inode(handle_t *, struct inode *);
1877 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
1878 extern unsigned long ext4_count_free_inodes(struct super_block *);
1879 extern unsigned long ext4_count_dirs(struct super_block *);
1880 extern void ext4_check_inodes_bitmap(struct super_block *);
1881 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
1882 extern int ext4_init_inode_table(struct super_block *sb,
1883 ext4_group_t group, int barrier);
1886 extern long ext4_mb_stats;
1887 extern long ext4_mb_max_to_scan;
1888 extern int ext4_mb_init(struct super_block *, int);
1889 extern int ext4_mb_release(struct super_block *);
1890 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
1891 struct ext4_allocation_request *, int *);
1892 extern int ext4_mb_reserve_blocks(struct super_block *, int);
1893 extern void ext4_discard_preallocations(struct inode *);
1894 extern int __init ext4_init_mballoc(void);
1895 extern void ext4_exit_mballoc(void);
1896 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
1897 struct buffer_head *bh, ext4_fsblk_t block,
1898 unsigned long count, int flags);
1899 extern int ext4_mb_add_groupinfo(struct super_block *sb,
1900 ext4_group_t i, struct ext4_group_desc *desc);
1901 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
1902 ext4_fsblk_t block, unsigned long count);
1903 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
1906 struct buffer_head *ext4_getblk(handle_t *, struct inode *,
1907 ext4_lblk_t, int, int *);
1908 struct buffer_head *ext4_bread(handle_t *, struct inode *,
1909 ext4_lblk_t, int, int *);
1910 int ext4_get_block(struct inode *inode, sector_t iblock,
1911 struct buffer_head *bh_result, int create);
1913 extern struct inode *ext4_iget(struct super_block *, unsigned long);
1914 extern struct inode *ext4_iget_normal(struct super_block *, unsigned long);
1915 extern int ext4_write_inode(struct inode *, struct writeback_control *);
1916 extern int ext4_setattr(struct dentry *, struct iattr *);
1917 extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
1918 struct kstat *stat);
1919 extern void ext4_evict_inode(struct inode *);
1920 extern void ext4_clear_inode(struct inode *);
1921 extern int ext4_sync_inode(handle_t *, struct inode *);
1922 extern void ext4_dirty_inode(struct inode *, int);
1923 extern int ext4_change_inode_journal_flag(struct inode *, int);
1924 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
1925 extern int ext4_can_truncate(struct inode *inode);
1926 extern void ext4_truncate(struct inode *);
1927 extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length);
1928 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
1929 extern void ext4_set_inode_flags(struct inode *);
1930 extern void ext4_get_inode_flags(struct ext4_inode_info *);
1931 extern int ext4_alloc_da_blocks(struct inode *inode);
1932 extern void ext4_set_aops(struct inode *inode);
1933 extern int ext4_writepage_trans_blocks(struct inode *);
1934 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
1935 extern int ext4_block_truncate_page(handle_t *handle,
1936 struct address_space *mapping, loff_t from);
1937 extern int ext4_block_zero_page_range(handle_t *handle,
1938 struct address_space *mapping, loff_t from, loff_t length);
1939 extern int ext4_discard_partial_page_buffers(handle_t *handle,
1940 struct address_space *mapping, loff_t from,
1941 loff_t length, int flags);
1942 extern int ext4_discard_partial_page_buffers_no_lock(handle_t *handle,
1943 struct inode *inode, struct page *page, loff_t from,
1944 loff_t length, int flags);
1945 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
1946 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
1947 extern void ext4_da_update_reserve_space(struct inode *inode,
1948 int used, int quota_claim);
1951 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
1952 struct ext4_map_blocks *map, int flags);
1953 extern ssize_t ext4_ind_direct_IO(int rw, struct kiocb *iocb,
1954 const struct iovec *iov, loff_t offset,
1955 unsigned long nr_segs);
1956 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
1957 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks, int chunk);
1958 extern void ext4_ind_truncate(struct inode *inode);
1961 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
1962 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
1965 extern int ext4_ext_migrate(struct inode *);
1968 extern int ext4_orphan_add(handle_t *, struct inode *);
1969 extern int ext4_orphan_del(handle_t *, struct inode *);
1970 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
1971 __u32 start_minor_hash, __u32 *next_hash);
1974 extern int ext4_group_add(struct super_block *sb,
1975 struct ext4_new_group_data *input);
1976 extern int ext4_group_extend(struct super_block *sb,
1977 struct ext4_super_block *es,
1978 ext4_fsblk_t n_blocks_count);
1981 extern int ext4_calculate_overhead(struct super_block *sb);
1982 extern void *ext4_kvmalloc(size_t size, gfp_t flags);
1983 extern void *ext4_kvzalloc(size_t size, gfp_t flags);
1984 extern void ext4_kvfree(void *ptr);
1985 extern __printf(4, 5)
1986 void __ext4_error(struct super_block *, const char *, unsigned int,
1988 #define ext4_error(sb, message...) __ext4_error(sb, __func__, \
1989 __LINE__, ## message)
1990 extern __printf(5, 6)
1991 void ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
1993 extern __printf(5, 6)
1994 void ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
1996 extern void __ext4_std_error(struct super_block *, const char *,
1998 extern __printf(4, 5)
1999 void __ext4_abort(struct super_block *, const char *, unsigned int,
2001 #define ext4_abort(sb, message...) __ext4_abort(sb, __func__, \
2002 __LINE__, ## message)
2003 extern __printf(4, 5)
2004 void __ext4_warning(struct super_block *, const char *, unsigned int,
2006 #define ext4_warning(sb, message...) __ext4_warning(sb, __func__, \
2007 __LINE__, ## message)
2008 extern __printf(3, 4)
2009 void ext4_msg(struct super_block *, const char *, const char *, ...);
2010 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2011 const char *, unsigned int, const char *);
2012 #define dump_mmp_msg(sb, mmp, msg) __dump_mmp_msg(sb, mmp, __func__, \
2014 extern __printf(7, 8)
2015 void __ext4_grp_locked_error(const char *, unsigned int,
2016 struct super_block *, ext4_group_t,
2017 unsigned long, ext4_fsblk_t,
2019 #define ext4_grp_locked_error(sb, grp, message...) \
2020 __ext4_grp_locked_error(__func__, __LINE__, (sb), (grp), ## message)
2021 extern void ext4_update_dynamic_rev(struct super_block *sb);
2022 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2024 extern int ext4_update_rocompat_feature(handle_t *handle,
2025 struct super_block *sb, __u32 rocompat);
2026 extern int ext4_update_incompat_feature(handle_t *handle,
2027 struct super_block *sb, __u32 incompat);
2028 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2029 struct ext4_group_desc *bg);
2030 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2031 struct ext4_group_desc *bg);
2032 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2033 struct ext4_group_desc *bg);
2034 extern __u32 ext4_free_group_clusters(struct super_block *sb,
2035 struct ext4_group_desc *bg);
2036 extern __u32 ext4_free_inodes_count(struct super_block *sb,
2037 struct ext4_group_desc *bg);
2038 extern __u32 ext4_used_dirs_count(struct super_block *sb,
2039 struct ext4_group_desc *bg);
2040 extern __u32 ext4_itable_unused_count(struct super_block *sb,
2041 struct ext4_group_desc *bg);
2042 extern void ext4_block_bitmap_set(struct super_block *sb,
2043 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2044 extern void ext4_inode_bitmap_set(struct super_block *sb,
2045 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2046 extern void ext4_inode_table_set(struct super_block *sb,
2047 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2048 extern void ext4_free_group_clusters_set(struct super_block *sb,
2049 struct ext4_group_desc *bg,
2051 extern void ext4_free_inodes_set(struct super_block *sb,
2052 struct ext4_group_desc *bg, __u32 count);
2053 extern void ext4_used_dirs_set(struct super_block *sb,
2054 struct ext4_group_desc *bg, __u32 count);
2055 extern void ext4_itable_unused_set(struct super_block *sb,
2056 struct ext4_group_desc *bg, __u32 count);
2057 extern __le16 ext4_group_desc_csum(struct ext4_sb_info *sbi, __u32 group,
2058 struct ext4_group_desc *gdp);
2059 extern int ext4_group_desc_csum_verify(struct ext4_sb_info *sbi, __u32 group,
2060 struct ext4_group_desc *gdp);
2062 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2064 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2065 le32_to_cpu(es->s_blocks_count_lo);
2068 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2070 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2071 le32_to_cpu(es->s_r_blocks_count_lo);
2074 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2076 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2077 le32_to_cpu(es->s_free_blocks_count_lo);
2080 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2083 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2084 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2087 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2090 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2091 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2094 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2097 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2098 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2101 static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
2103 if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2104 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2105 le32_to_cpu(raw_inode->i_size_lo);
2107 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2110 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2112 raw_inode->i_size_lo = cpu_to_le32(i_size);
2113 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2117 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2120 struct ext4_group_info ***grp_info;
2121 long indexv, indexh;
2122 grp_info = EXT4_SB(sb)->s_group_info;
2123 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2124 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2125 return grp_info[indexv][indexh];
2129 * Reading s_groups_count requires using smp_rmb() afterwards. See
2130 * the locking protocol documented in the comments of ext4_group_add()
2133 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2135 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
2141 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2142 ext4_group_t block_group)
2144 return block_group >> sbi->s_log_groups_per_flex;
2147 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2149 return 1 << sbi->s_log_groups_per_flex;
2152 #define ext4_std_error(sb, errno) \
2155 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
2159 /* Each CPU can accumulate percpu_counter_batch clusters in their local
2160 * counters. So we need to make sure we have free clusters more
2161 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
2163 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2165 #define EXT4_FREECLUSTERS_WATERMARK 0
2168 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2171 * XXX: replace with spinlock if seen contended -bzzz
2173 down_write(&EXT4_I(inode)->i_data_sem);
2174 if (newsize > EXT4_I(inode)->i_disksize)
2175 EXT4_I(inode)->i_disksize = newsize;
2176 up_write(&EXT4_I(inode)->i_data_sem);
2180 struct ext4_group_info {
2181 unsigned long bb_state;
2182 struct rb_root bb_free_root;
2183 ext4_grpblk_t bb_first_free; /* first free block */
2184 ext4_grpblk_t bb_free; /* total free blocks */
2185 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
2186 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
2187 struct list_head bb_prealloc_list;
2191 struct rw_semaphore alloc_sem;
2192 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
2193 * regions, index is order.
2194 * bb_counters[3] = 5 means
2195 * 5 free 8-block regions. */
2198 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
2199 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
2201 #define EXT4_MB_GRP_NEED_INIT(grp) \
2202 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2204 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
2205 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2206 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
2207 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2208 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
2209 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2211 #define EXT4_MAX_CONTENTION 8
2212 #define EXT4_CONTENTION_THRESHOLD 2
2214 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2217 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2221 * Returns true if the filesystem is busy enough that attempts to
2222 * access the block group locks has run into contention.
2224 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2226 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2229 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2231 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2232 if (spin_trylock(lock))
2234 * We're able to grab the lock right away, so drop the
2235 * lock contention counter.
2237 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2240 * The lock is busy, so bump the contention counter,
2241 * and then wait on the spin lock.
2243 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2244 EXT4_MAX_CONTENTION);
2249 static inline void ext4_unlock_group(struct super_block *sb,
2252 spin_unlock(ext4_group_lock_ptr(sb, group));
2255 static inline void ext4_mark_super_dirty(struct super_block *sb)
2257 if (EXT4_SB(sb)->s_journal == NULL)
2262 * Block validity checking
2264 #define ext4_check_indirect_blockref(inode, bh) \
2265 ext4_check_blockref(__func__, __LINE__, inode, \
2266 (__le32 *)(bh)->b_data, \
2267 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2269 #define ext4_ind_check_inode(inode) \
2270 ext4_check_blockref(__func__, __LINE__, inode, \
2271 EXT4_I(inode)->i_data, \
2275 * Inodes and files operations
2279 extern const struct file_operations ext4_dir_operations;
2282 extern const struct inode_operations ext4_file_inode_operations;
2283 extern const struct file_operations ext4_file_operations;
2284 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
2287 extern const struct inode_operations ext4_dir_inode_operations;
2288 extern const struct inode_operations ext4_special_inode_operations;
2289 extern struct dentry *ext4_get_parent(struct dentry *child);
2292 extern const struct inode_operations ext4_symlink_inode_operations;
2293 extern const struct inode_operations ext4_fast_symlink_inode_operations;
2295 /* block_validity */
2296 extern void ext4_release_system_zone(struct super_block *sb);
2297 extern int ext4_setup_system_zone(struct super_block *sb);
2298 extern int __init ext4_init_system_zone(void);
2299 extern void ext4_exit_system_zone(void);
2300 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
2301 ext4_fsblk_t start_blk,
2302 unsigned int count);
2303 extern int ext4_check_blockref(const char *, unsigned int,
2304 struct inode *, __le32 *, unsigned int);
2307 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
2308 extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
2309 extern int ext4_ext_index_trans_blocks(struct inode *inode, int nrblocks,
2311 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
2312 struct ext4_map_blocks *map, int flags);
2313 extern void ext4_ext_truncate(struct inode *);
2314 extern int ext4_ext_punch_hole(struct file *file, loff_t offset,
2316 extern void ext4_ext_init(struct super_block *);
2317 extern void ext4_ext_release(struct super_block *);
2318 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
2320 extern int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset,
2322 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
2323 struct ext4_map_blocks *map, int flags);
2324 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
2325 __u64 start, __u64 len);
2327 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
2328 __u64 start_orig, __u64 start_donor,
2329 __u64 len, __u64 *moved_len);
2332 extern int __init ext4_init_pageio(void);
2333 extern void ext4_exit_pageio(void);
2334 extern void ext4_ioend_wait(struct inode *);
2335 extern void ext4_free_io_end(ext4_io_end_t *io);
2336 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
2337 extern int ext4_end_io_nolock(ext4_io_end_t *io);
2338 extern void ext4_io_submit(struct ext4_io_submit *io);
2339 extern int ext4_bio_write_page(struct ext4_io_submit *io,
2342 struct writeback_control *wbc);
2345 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
2347 /* BH_Uninit flag: blocks are allocated but uninitialized on disk */
2348 enum ext4_state_bits {
2349 BH_Uninit /* blocks are allocated but uninitialized on disk */
2350 = BH_JBDPrivateStart,
2351 BH_AllocFromCluster, /* allocated blocks were part of already
2352 * allocated cluster. Note that this flag will
2353 * never, ever appear in a buffer_head's state
2354 * flag. See EXT4_MAP_FROM_CLUSTER to see where
2356 BH_Da_Mapped, /* Delayed allocated block that now has a mapping. This
2357 * flag is set when ext4_map_blocks is called on a
2358 * delayed allocated block to get its real mapping. */
2361 BUFFER_FNS(Uninit, uninit)
2362 TAS_BUFFER_FNS(Uninit, uninit)
2363 BUFFER_FNS(Da_Mapped, da_mapped)
2366 * Add new method to test wether block and inode bitmaps are properly
2367 * initialized. With uninit_bg reading the block from disk is not enough
2368 * to mark the bitmap uptodate. We need to also zero-out the bitmap
2370 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
2372 static inline int bitmap_uptodate(struct buffer_head *bh)
2374 return (buffer_uptodate(bh) &&
2375 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
2377 static inline void set_bitmap_uptodate(struct buffer_head *bh)
2379 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
2382 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
2384 /* For ioend & aio unwritten conversion wait queues */
2385 #define EXT4_WQ_HASH_SZ 37
2386 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
2388 #define ext4_aio_mutex(v) (&ext4__aio_mutex[((unsigned long)(v)) %\
2390 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
2391 extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ];
2393 #define EXT4_RESIZING 0
2394 extern int ext4_resize_begin(struct super_block *sb);
2395 extern void ext4_resize_end(struct super_block *sb);
2397 #endif /* __KERNEL__ */
2399 #include "ext4_extents.h"
2401 #endif /* _EXT4_H */