GUI: Fix Tomato RAF theme for all builds. Compilation typo.
[tomato.git] / release / src-rt-6.x.4708 / linux / linux-2.6.36 / fs / ext4 / ext4.h
blob7ca85e513be85a48b37b26f535d10c6ccb821748
1 /*
2 * ext4.h
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)
9 * from
11 * linux/include/linux/minix_fs.h
13 * Copyright (C) 1991, 1992 Linus Torvalds
16 #ifndef _EXT4_H
17 #define _EXT4_H
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 #ifdef __KERNEL__
33 #include <linux/compat.h>
34 #endif
37 * The fourth extended filesystem constants/structures
41 * Define EXT4FS_DEBUG to produce debug messages
43 #undef EXT4FS_DEBUG
46 * Debug code
48 #ifdef EXT4FS_DEBUG
49 #define ext4_debug(f, a...) \
50 do { \
51 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
52 __FILE__, __LINE__, __func__); \
53 printk(KERN_DEBUG f, ## a); \
54 } while (0)
55 #else
56 #define ext4_debug(f, a...) do {} while (0)
57 #endif
59 #define EXT4_ERROR_INODE(inode, fmt, a...) \
60 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
62 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
63 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
65 #define EXT4_ERROR_FILE(file, fmt, a...) \
66 ext4_error_file(__func__, __LINE__, (file), (fmt), ## a)
68 /* data type for block offset of block group */
69 typedef int ext4_grpblk_t;
71 /* data type for filesystem-wide blocks number */
72 typedef unsigned long long ext4_fsblk_t;
74 /* data type for file logical block number */
75 typedef __u32 ext4_lblk_t;
77 /* data type for block group number */
78 typedef unsigned int ext4_group_t;
81 * Flags used in mballoc's allocation_context flags field.
83 * Also used to show what's going on for debugging purposes when the
84 * flag field is exported via the traceport interface
87 /* prefer goal again. length */
88 #define EXT4_MB_HINT_MERGE 0x0001
89 /* blocks already reserved */
90 #define EXT4_MB_HINT_RESERVED 0x0002
91 /* metadata is being allocated */
92 #define EXT4_MB_HINT_METADATA 0x0004
93 /* first blocks in the file */
94 #define EXT4_MB_HINT_FIRST 0x0008
95 /* search for the best chunk */
96 #define EXT4_MB_HINT_BEST 0x0010
97 /* data is being allocated */
98 #define EXT4_MB_HINT_DATA 0x0020
99 /* don't preallocate (for tails) */
100 #define EXT4_MB_HINT_NOPREALLOC 0x0040
101 /* allocate for locality group */
102 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080
103 /* allocate goal blocks or none */
104 #define EXT4_MB_HINT_GOAL_ONLY 0x0100
105 /* goal is meaningful */
106 #define EXT4_MB_HINT_TRY_GOAL 0x0200
107 /* blocks already pre-reserved by delayed allocation */
108 #define EXT4_MB_DELALLOC_RESERVED 0x0400
109 /* We are doing stream allocation */
110 #define EXT4_MB_STREAM_ALLOC 0x0800
113 struct ext4_allocation_request {
114 /* target inode for block we're allocating */
115 struct inode *inode;
116 /* how many blocks we want to allocate */
117 unsigned int len;
118 /* logical block in target inode */
119 ext4_lblk_t logical;
120 /* the closest logical allocated block to the left */
121 ext4_lblk_t lleft;
122 /* the closest logical allocated block to the right */
123 ext4_lblk_t lright;
124 /* phys. target (a hint) */
125 ext4_fsblk_t goal;
126 /* phys. block for the closest logical allocated block to the left */
127 ext4_fsblk_t pleft;
128 /* phys. block for the closest logical allocated block to the right */
129 ext4_fsblk_t pright;
130 /* flags. see above EXT4_MB_HINT_* */
131 unsigned int flags;
135 * Logical to physical block mapping, used by ext4_map_blocks()
137 * This structure is used to pass requests into ext4_map_blocks() as
138 * well as to store the information returned by ext4_map_blocks(). It
139 * takes less room on the stack than a struct buffer_head.
141 #define EXT4_MAP_NEW (1 << BH_New)
142 #define EXT4_MAP_MAPPED (1 << BH_Mapped)
143 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten)
144 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary)
145 #define EXT4_MAP_UNINIT (1 << BH_Uninit)
146 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
147 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\
148 EXT4_MAP_UNINIT)
150 struct ext4_map_blocks {
151 ext4_fsblk_t m_pblk;
152 ext4_lblk_t m_lblk;
153 unsigned int m_len;
154 unsigned int m_flags;
158 * For delayed allocation tracking
160 struct mpage_da_data {
161 struct inode *inode;
162 sector_t b_blocknr; /* start block number of extent */
163 size_t b_size; /* size of extent */
164 unsigned long b_state; /* state of the extent */
165 unsigned long first_page, next_page; /* extent of pages */
166 struct writeback_control *wbc;
167 int io_done;
168 int pages_written;
169 int retval;
171 #define EXT4_IO_UNWRITTEN 0x1
172 typedef struct ext4_io_end {
173 struct list_head list; /* per-file finished IO list */
174 struct inode *inode; /* file being written to */
175 unsigned int flag; /* unwritten or not */
176 struct page *page; /* page struct for buffer write */
177 loff_t offset; /* offset in the file */
178 ssize_t size; /* size of the extent */
179 struct work_struct work; /* data work queue */
180 struct kiocb *iocb; /* iocb struct for AIO */
181 int result; /* error value for AIO */
182 } ext4_io_end_t;
185 * Special inodes numbers
187 #define EXT4_BAD_INO 1 /* Bad blocks inode */
188 #define EXT4_ROOT_INO 2 /* Root inode */
189 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
190 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
191 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
192 #define EXT4_JOURNAL_INO 8 /* Journal inode */
194 /* First non-reserved inode for old ext4 filesystems */
195 #define EXT4_GOOD_OLD_FIRST_INO 11
198 * Maximal count of links to a file
200 #define EXT4_LINK_MAX 65000
203 * Macro-instructions used to manage several block sizes
205 #define EXT4_MIN_BLOCK_SIZE 1024
206 #define EXT4_MAX_BLOCK_SIZE 65536
207 #define EXT4_MIN_BLOCK_LOG_SIZE 10
208 #ifdef __KERNEL__
209 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
210 #else
211 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
212 #endif
213 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
214 #ifdef __KERNEL__
215 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
216 #else
217 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
218 #endif
219 #ifdef __KERNEL__
220 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
221 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
222 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
223 #else
224 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
225 EXT4_GOOD_OLD_INODE_SIZE : \
226 (s)->s_inode_size)
227 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
228 EXT4_GOOD_OLD_FIRST_INO : \
229 (s)->s_first_ino)
230 #endif
231 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
234 * Structure of a blocks group descriptor
236 struct ext4_group_desc
238 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
239 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
240 __le32 bg_inode_table_lo; /* Inodes table block */
241 __le16 bg_free_blocks_count_lo;/* Free blocks count */
242 __le16 bg_free_inodes_count_lo;/* Free inodes count */
243 __le16 bg_used_dirs_count_lo; /* Directories count */
244 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
245 __u32 bg_reserved[2]; /* Likely block/inode bitmap checksum */
246 __le16 bg_itable_unused_lo; /* Unused inodes count */
247 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
248 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
249 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
250 __le32 bg_inode_table_hi; /* Inodes table block MSB */
251 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
252 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
253 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
254 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
255 __u32 bg_reserved2[3];
259 * Structure of a flex block group info
262 struct flex_groups {
263 atomic_t free_inodes;
264 atomic_t free_blocks;
265 atomic_t used_dirs;
268 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
269 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
270 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
273 * Macro-instructions used to manage group descriptors
275 #define EXT4_MIN_DESC_SIZE 32
276 #define EXT4_MIN_DESC_SIZE_64BIT 64
277 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
278 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
279 #ifdef __KERNEL__
280 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
281 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
282 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
283 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
284 #else
285 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
286 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
287 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
288 #endif
291 * Constants relative to the data blocks
293 #define EXT4_NDIR_BLOCKS 12
294 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
295 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
296 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
297 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
300 * Inode flags
302 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
303 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
304 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
305 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
306 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
307 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
308 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
309 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
310 /* Reserved for compression usage... */
311 #define EXT4_DIRTY_FL 0x00000100
312 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
313 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
314 #define EXT4_ECOMPR_FL 0x00000800 /* Compression error */
315 /* End compression flags --- maybe not all used */
316 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
317 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
318 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
319 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
320 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
321 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
322 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
323 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
324 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
325 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
326 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
328 #define EXT4_FL_USER_VISIBLE 0x004BDFFF /* User visible flags */
329 #define EXT4_FL_USER_MODIFIABLE 0x004B80FF /* User modifiable flags */
331 /* Flags that should be inherited by new inodes from their parent. */
332 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
333 EXT4_SYNC_FL | EXT4_IMMUTABLE_FL | EXT4_APPEND_FL |\
334 EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
335 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
336 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
338 /* Flags that are appropriate for regular files (all but dir-specific ones). */
339 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
341 /* Flags that are appropriate for non-directories/regular files. */
342 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
344 /* Mask out flags that are inappropriate for the given type of inode. */
345 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
347 if (S_ISDIR(mode))
348 return flags;
349 else if (S_ISREG(mode))
350 return flags & EXT4_REG_FLMASK;
351 else
352 return flags & EXT4_OTHER_FLMASK;
356 * Inode flags used for atomic set/get
358 enum {
359 EXT4_INODE_SECRM = 0, /* Secure deletion */
360 EXT4_INODE_UNRM = 1, /* Undelete */
361 EXT4_INODE_COMPR = 2, /* Compress file */
362 EXT4_INODE_SYNC = 3, /* Synchronous updates */
363 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
364 EXT4_INODE_APPEND = 5, /* writes to file may only append */
365 EXT4_INODE_NODUMP = 6, /* do not dump file */
366 EXT4_INODE_NOATIME = 7, /* do not update atime */
367 /* Reserved for compression usage... */
368 EXT4_INODE_DIRTY = 8,
369 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
370 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
371 EXT4_INODE_ECOMPR = 11, /* Compression error */
372 /* End compression flags --- maybe not all used */
373 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
374 EXT4_INODE_IMAGIC = 13, /* AFS directory */
375 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
376 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
377 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
378 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
379 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
380 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
381 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
382 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
383 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
386 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
387 #define CHECK_FLAG_VALUE(FLAG) if (!TEST_FLAG_VALUE(FLAG)) { \
388 printk(KERN_EMERG "EXT4 flag fail: " #FLAG ": %d %d\n", \
389 EXT4_##FLAG##_FL, EXT4_INODE_##FLAG); BUG_ON(1); }
392 * Since it's pretty easy to mix up bit numbers and hex values, and we
393 * can't do a compile-time test for ENUM values, we use a run-time
394 * test to make sure that EXT4_XXX_FL is consistent with respect to
395 * EXT4_INODE_XXX. If all is well the printk and BUG_ON will all drop
396 * out so it won't cost any extra space in the compiled kernel image.
397 * But it's important that these values are the same, since we are
398 * using EXT4_INODE_XXX to test for the flag values, but EXT4_XX_FL
399 * must be consistent with the values of FS_XXX_FL defined in
400 * include/linux/fs.h and the on-disk values found in ext2, ext3, and
401 * ext4 filesystems, and of course the values defined in e2fsprogs.
403 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
405 static inline void ext4_check_flag_values(void)
407 CHECK_FLAG_VALUE(SECRM);
408 CHECK_FLAG_VALUE(UNRM);
409 CHECK_FLAG_VALUE(COMPR);
410 CHECK_FLAG_VALUE(SYNC);
411 CHECK_FLAG_VALUE(IMMUTABLE);
412 CHECK_FLAG_VALUE(APPEND);
413 CHECK_FLAG_VALUE(NODUMP);
414 CHECK_FLAG_VALUE(NOATIME);
415 CHECK_FLAG_VALUE(DIRTY);
416 CHECK_FLAG_VALUE(COMPRBLK);
417 CHECK_FLAG_VALUE(NOCOMPR);
418 CHECK_FLAG_VALUE(ECOMPR);
419 CHECK_FLAG_VALUE(INDEX);
420 CHECK_FLAG_VALUE(IMAGIC);
421 CHECK_FLAG_VALUE(JOURNAL_DATA);
422 CHECK_FLAG_VALUE(NOTAIL);
423 CHECK_FLAG_VALUE(DIRSYNC);
424 CHECK_FLAG_VALUE(TOPDIR);
425 CHECK_FLAG_VALUE(HUGE_FILE);
426 CHECK_FLAG_VALUE(EXTENTS);
427 CHECK_FLAG_VALUE(EA_INODE);
428 CHECK_FLAG_VALUE(EOFBLOCKS);
429 CHECK_FLAG_VALUE(RESERVED);
432 /* Used to pass group descriptor data when online resize is done */
433 struct ext4_new_group_input {
434 __u32 group; /* Group number for this data */
435 __u64 block_bitmap; /* Absolute block number of block bitmap */
436 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
437 __u64 inode_table; /* Absolute block number of inode table start */
438 __u32 blocks_count; /* Total number of blocks in this group */
439 __u16 reserved_blocks; /* Number of reserved blocks in this group */
440 __u16 unused;
443 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
444 struct compat_ext4_new_group_input {
445 u32 group;
446 compat_u64 block_bitmap;
447 compat_u64 inode_bitmap;
448 compat_u64 inode_table;
449 u32 blocks_count;
450 u16 reserved_blocks;
451 u16 unused;
453 #endif
455 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
456 struct ext4_new_group_data {
457 __u32 group;
458 __u64 block_bitmap;
459 __u64 inode_bitmap;
460 __u64 inode_table;
461 __u32 blocks_count;
462 __u16 reserved_blocks;
463 __u16 unused;
464 __u32 free_blocks_count;
468 * Flags used by ext4_map_blocks()
470 /* Allocate any needed blocks and/or convert an unitialized
471 extent to be an initialized ext4 */
472 #define EXT4_GET_BLOCKS_CREATE 0x0001
473 /* Request the creation of an unitialized extent */
474 #define EXT4_GET_BLOCKS_UNINIT_EXT 0x0002
475 #define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT (EXT4_GET_BLOCKS_UNINIT_EXT|\
476 EXT4_GET_BLOCKS_CREATE)
477 /* Caller is from the delayed allocation writeout path,
478 so set the magic i_delalloc_reserve_flag after taking the
479 inode allocation semaphore for */
480 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
481 /* caller is from the direct IO path, request to creation of an
482 unitialized extents if not allocated, split the uninitialized
483 extent if blocks has been preallocated already*/
484 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
485 #define EXT4_GET_BLOCKS_CONVERT 0x0010
486 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
487 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
488 /* Convert extent to initialized after IO complete */
489 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
490 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
493 * Flags used by ext4_free_blocks
495 #define EXT4_FREE_BLOCKS_METADATA 0x0001
496 #define EXT4_FREE_BLOCKS_FORGET 0x0002
497 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
500 * ioctl commands
502 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
503 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
504 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
505 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
506 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
507 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
508 #ifdef CONFIG_JBD2_DEBUG
509 #define EXT4_IOC_WAIT_FOR_READONLY _IOR('f', 99, long)
510 #endif
511 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
512 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
513 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
514 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
515 #define EXT4_IOC_MIGRATE _IO('f', 9)
516 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
517 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
518 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
519 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
521 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
523 * ioctl commands in 32 bit emulation
525 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
526 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
527 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
528 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
529 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
530 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
531 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
532 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
533 #ifdef CONFIG_JBD2_DEBUG
534 #define EXT4_IOC32_WAIT_FOR_READONLY _IOR('f', 99, int)
535 #endif
536 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
537 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
538 #endif
542 * Mount options
544 struct ext4_mount_options {
545 unsigned long s_mount_opt;
546 uid_t s_resuid;
547 gid_t s_resgid;
548 unsigned long s_commit_interval;
549 u32 s_min_batch_time, s_max_batch_time;
550 #ifdef CONFIG_QUOTA
551 int s_jquota_fmt;
552 char *s_qf_names[MAXQUOTAS];
553 #endif
556 /* Max physical block we can addres w/o extents */
557 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
560 * Structure of an inode on the disk
562 struct ext4_inode {
563 __le16 i_mode; /* File mode */
564 __le16 i_uid; /* Low 16 bits of Owner Uid */
565 __le32 i_size_lo; /* Size in bytes */
566 __le32 i_atime; /* Access time */
567 __le32 i_ctime; /* Inode Change time */
568 __le32 i_mtime; /* Modification time */
569 __le32 i_dtime; /* Deletion Time */
570 __le16 i_gid; /* Low 16 bits of Group Id */
571 __le16 i_links_count; /* Links count */
572 __le32 i_blocks_lo; /* Blocks count */
573 __le32 i_flags; /* File flags */
574 union {
575 struct {
576 __le32 l_i_version;
577 } linux1;
578 struct {
579 __u32 h_i_translator;
580 } hurd1;
581 struct {
582 __u32 m_i_reserved1;
583 } masix1;
584 } osd1; /* OS dependent 1 */
585 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
586 __le32 i_generation; /* File version (for NFS) */
587 __le32 i_file_acl_lo; /* File ACL */
588 __le32 i_size_high;
589 __le32 i_obso_faddr; /* Obsoleted fragment address */
590 union {
591 struct {
592 __le16 l_i_blocks_high; /* were l_i_reserved1 */
593 __le16 l_i_file_acl_high;
594 __le16 l_i_uid_high; /* these 2 fields */
595 __le16 l_i_gid_high; /* were reserved2[0] */
596 __u32 l_i_reserved2;
597 } linux2;
598 struct {
599 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
600 __u16 h_i_mode_high;
601 __u16 h_i_uid_high;
602 __u16 h_i_gid_high;
603 __u32 h_i_author;
604 } hurd2;
605 struct {
606 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
607 __le16 m_i_file_acl_high;
608 __u32 m_i_reserved2[2];
609 } masix2;
610 } osd2; /* OS dependent 2 */
611 __le16 i_extra_isize;
612 __le16 i_pad1;
613 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
614 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
615 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
616 __le32 i_crtime; /* File Creation time */
617 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
618 __le32 i_version_hi; /* high 32 bits for 64-bit version */
621 struct move_extent {
622 __u32 reserved; /* should be zero */
623 __u32 donor_fd; /* donor file descriptor */
624 __u64 orig_start; /* logical start offset in block for orig */
625 __u64 donor_start; /* logical start offset in block for donor */
626 __u64 len; /* block length to be moved */
627 __u64 moved_len; /* moved block length */
630 #define EXT4_EPOCH_BITS 2
631 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
632 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
635 * Extended fields will fit into an inode if the filesystem was formatted
636 * with large inodes (-I 256 or larger) and there are not currently any EAs
637 * consuming all of the available space. For new inodes we always reserve
638 * enough space for the kernel's known extended fields, but for inodes
639 * created with an old kernel this might not have been the case. None of
640 * the extended inode fields is critical for correct filesystem operation.
641 * This macro checks if a certain field fits in the inode. Note that
642 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
644 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
645 ((offsetof(typeof(*ext4_inode), field) + \
646 sizeof((ext4_inode)->field)) \
647 <= (EXT4_GOOD_OLD_INODE_SIZE + \
648 (einode)->i_extra_isize)) \
650 static inline __le32 ext4_encode_extra_time(struct timespec *time)
652 return cpu_to_le32((sizeof(time->tv_sec) > 4 ?
653 (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) |
654 ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK));
657 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
659 if (sizeof(time->tv_sec) > 4)
660 time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK)
661 << 32;
662 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
665 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
666 do { \
667 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
668 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
669 (raw_inode)->xtime ## _extra = \
670 ext4_encode_extra_time(&(inode)->xtime); \
671 } while (0)
673 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
674 do { \
675 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
676 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
677 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
678 (raw_inode)->xtime ## _extra = \
679 ext4_encode_extra_time(&(einode)->xtime); \
680 } while (0)
682 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
683 do { \
684 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
685 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
686 ext4_decode_extra_time(&(inode)->xtime, \
687 raw_inode->xtime ## _extra); \
688 } while (0)
690 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
691 do { \
692 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
693 (einode)->xtime.tv_sec = \
694 (signed)le32_to_cpu((raw_inode)->xtime); \
695 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
696 ext4_decode_extra_time(&(einode)->xtime, \
697 raw_inode->xtime ## _extra); \
698 } while (0)
700 #define i_disk_version osd1.linux1.l_i_version
702 #if defined(__KERNEL__) || defined(__linux__)
703 #define i_reserved1 osd1.linux1.l_i_reserved1
704 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
705 #define i_blocks_high osd2.linux2.l_i_blocks_high
706 #define i_uid_low i_uid
707 #define i_gid_low i_gid
708 #define i_uid_high osd2.linux2.l_i_uid_high
709 #define i_gid_high osd2.linux2.l_i_gid_high
710 #define i_reserved2 osd2.linux2.l_i_reserved2
712 #elif defined(__GNU__)
714 #define i_translator osd1.hurd1.h_i_translator
715 #define i_uid_high osd2.hurd2.h_i_uid_high
716 #define i_gid_high osd2.hurd2.h_i_gid_high
717 #define i_author osd2.hurd2.h_i_author
719 #elif defined(__masix__)
721 #define i_reserved1 osd1.masix1.m_i_reserved1
722 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
723 #define i_reserved2 osd2.masix2.m_i_reserved2
725 #endif /* defined(__KERNEL__) || defined(__linux__) */
728 * storage for cached extent
730 struct ext4_ext_cache {
731 ext4_fsblk_t ec_start;
732 ext4_lblk_t ec_block;
733 __u32 ec_len; /* must be 32bit to return holes */
734 __u32 ec_type;
738 * fourth extended file system inode data in memory
740 struct ext4_inode_info {
741 __le32 i_data[15]; /* unconverted */
742 __u32 i_dtime;
743 ext4_fsblk_t i_file_acl;
746 * i_block_group is the number of the block group which contains
747 * this file's inode. Constant across the lifetime of the inode,
748 * it is ued for making block allocation decisions - we try to
749 * place a file's data blocks near its inode block, and new inodes
750 * near to their parent directory's inode.
752 ext4_group_t i_block_group;
753 unsigned long i_state_flags; /* Dynamic state flags */
754 unsigned long i_flags;
756 ext4_lblk_t i_dir_start_lookup;
757 #ifdef CONFIG_EXT4_FS_XATTR
759 * Extended attributes can be read independently of the main file
760 * data. Taking i_mutex even when reading would cause contention
761 * between readers of EAs and writers of regular file data, so
762 * instead we synchronize on xattr_sem when reading or changing
763 * EAs.
765 struct rw_semaphore xattr_sem;
766 #endif
768 struct list_head i_orphan; /* unlinked but open inodes */
771 * i_disksize keeps track of what the inode size is ON DISK, not
772 * in memory. During truncate, i_size is set to the new size by
773 * the VFS prior to calling ext4_truncate(), but the filesystem won't
774 * set i_disksize to 0 until the truncate is actually under way.
776 * The intent is that i_disksize always represents the blocks which
777 * are used by this file. This allows recovery to restart truncate
778 * on orphans if we crash during truncate. We actually write i_disksize
779 * into the on-disk inode when writing inodes out, instead of i_size.
781 * The only time when i_disksize and i_size may be different is when
782 * a truncate is in progress. The only things which change i_disksize
783 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
785 loff_t i_disksize;
788 * i_data_sem is for serialising ext4_truncate() against
789 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
790 * data tree are chopped off during truncate. We can't do that in
791 * ext4 because whenever we perform intermediate commits during
792 * truncate, the inode and all the metadata blocks *must* be in a
793 * consistent state which allows truncation of the orphans to restart
794 * during recovery. Hence we must fix the get_block-vs-truncate race
795 * by other means, so we have i_data_sem.
797 struct rw_semaphore i_data_sem;
798 struct inode vfs_inode;
799 struct jbd2_inode jinode;
801 struct ext4_ext_cache i_cached_extent;
803 * File creation time. Its function is same as that of
804 * struct timespec i_{a,c,m}time in the generic inode.
806 struct timespec i_crtime;
808 /* mballoc */
809 struct list_head i_prealloc_list;
810 spinlock_t i_prealloc_lock;
812 /* ialloc */
813 ext4_group_t i_last_alloc_group;
815 /* allocation reservation info for delalloc */
816 unsigned int i_reserved_data_blocks;
817 unsigned int i_reserved_meta_blocks;
818 unsigned int i_allocated_meta_blocks;
819 unsigned short i_delalloc_reserved_flag;
820 sector_t i_da_metadata_calc_last_lblock;
821 int i_da_metadata_calc_len;
823 /* on-disk additional length */
824 __u16 i_extra_isize;
826 spinlock_t i_block_reservation_lock;
827 #ifdef CONFIG_QUOTA
828 /* quota space reservation, managed internally by quota code */
829 qsize_t i_reserved_quota;
830 #endif
832 /* completed IOs that might need unwritten extents handling */
833 struct list_head i_completed_io_list;
834 spinlock_t i_completed_io_lock;
835 /* current io_end structure for async DIO write*/
836 ext4_io_end_t *cur_aio_dio;
839 * Transactions that contain inode's metadata needed to complete
840 * fsync and fdatasync, respectively.
842 tid_t i_sync_tid;
843 tid_t i_datasync_tid;
847 * File system states
849 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
850 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
851 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
854 * Misc. filesystem flags
856 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
857 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
858 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
861 * Mount flags
863 #define EXT4_MOUNT_OLDALLOC 0x00002 /* Don't use the new Orlov allocator */
864 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
865 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
866 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
867 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
868 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
869 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
870 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
871 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
872 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
873 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
874 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
875 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
876 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
877 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
878 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
879 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
880 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
881 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
882 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
883 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
884 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
885 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
886 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
887 #define EXT4_MOUNT_I_VERSION 0x2000000 /* i_version support */
888 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
889 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
890 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
891 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
893 #define clear_opt(o, opt) o &= ~EXT4_MOUNT_##opt
894 #define set_opt(o, opt) o |= EXT4_MOUNT_##opt
895 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
896 EXT4_MOUNT_##opt)
898 #define ext4_set_bit ext2_set_bit
899 #define ext4_set_bit_atomic ext2_set_bit_atomic
900 #define ext4_clear_bit ext2_clear_bit
901 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
902 #define ext4_test_bit ext2_test_bit
903 #define ext4_find_first_zero_bit ext2_find_first_zero_bit
904 #define ext4_find_next_zero_bit ext2_find_next_zero_bit
905 #define ext4_find_next_bit ext2_find_next_bit
908 * Maximal mount counts between two filesystem checks
910 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
911 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
914 * Behaviour when detecting errors
916 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
917 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
918 #define EXT4_ERRORS_PANIC 3 /* Panic */
919 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
922 * Structure of the super block
924 struct ext4_super_block {
925 /*00*/ __le32 s_inodes_count; /* Inodes count */
926 __le32 s_blocks_count_lo; /* Blocks count */
927 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
928 __le32 s_free_blocks_count_lo; /* Free blocks count */
929 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */
930 __le32 s_first_data_block; /* First Data Block */
931 __le32 s_log_block_size; /* Block size */
932 __le32 s_obso_log_frag_size; /* Obsoleted fragment size */
933 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
934 __le32 s_obso_frags_per_group; /* Obsoleted fragments per group */
935 __le32 s_inodes_per_group; /* # Inodes per group */
936 __le32 s_mtime; /* Mount time */
937 /*30*/ __le32 s_wtime; /* Write time */
938 __le16 s_mnt_count; /* Mount count */
939 __le16 s_max_mnt_count; /* Maximal mount count */
940 __le16 s_magic; /* Magic signature */
941 __le16 s_state; /* File system state */
942 __le16 s_errors; /* Behaviour when detecting errors */
943 __le16 s_minor_rev_level; /* minor revision level */
944 /*40*/ __le32 s_lastcheck; /* time of last check */
945 __le32 s_checkinterval; /* max. time between checks */
946 __le32 s_creator_os; /* OS */
947 __le32 s_rev_level; /* Revision level */
948 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
949 __le16 s_def_resgid; /* Default gid for reserved blocks */
951 * These fields are for EXT4_DYNAMIC_REV superblocks only.
953 * Note: the difference between the compatible feature set and
954 * the incompatible feature set is that if there is a bit set
955 * in the incompatible feature set that the kernel doesn't
956 * know about, it should refuse to mount the filesystem.
958 * e2fsck's requirements are more strict; if it doesn't know
959 * about a feature in either the compatible or incompatible
960 * feature set, it must abort and not try to meddle with
961 * things it doesn't understand...
963 __le32 s_first_ino; /* First non-reserved inode */
964 __le16 s_inode_size; /* size of inode structure */
965 __le16 s_block_group_nr; /* block group # of this superblock */
966 __le32 s_feature_compat; /* compatible feature set */
967 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */
968 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
969 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
970 /*78*/ char s_volume_name[16]; /* volume name */
971 /*88*/ char s_last_mounted[64]; /* directory where last mounted */
972 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
974 * Performance hints. Directory preallocation should only
975 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
977 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
978 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
979 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
981 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
983 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
984 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */
985 __le32 s_journal_dev; /* device number of journal file */
986 __le32 s_last_orphan; /* start of list of inodes to delete */
987 __le32 s_hash_seed[4]; /* HTREE hash seed */
988 __u8 s_def_hash_version; /* Default hash version to use */
989 __u8 s_jnl_backup_type;
990 __le16 s_desc_size; /* size of group descriptor */
991 /*100*/ __le32 s_default_mount_opts;
992 __le32 s_first_meta_bg; /* First metablock block group */
993 __le32 s_mkfs_time; /* When the filesystem was created */
994 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
995 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
996 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */
997 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
998 __le32 s_free_blocks_count_hi; /* Free blocks count */
999 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1000 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1001 __le32 s_flags; /* Miscellaneous flags */
1002 __le16 s_raid_stride; /* RAID stride */
1003 __le16 s_mmp_interval; /* # seconds to wait in MMP checking */
1004 __le64 s_mmp_block; /* Block for multi-mount protection */
1005 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1006 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1007 __u8 s_reserved_char_pad;
1008 __le16 s_reserved_pad;
1009 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1010 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1011 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1012 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1013 snapshot's future use */
1014 __le32 s_snapshot_list; /* inode number of the head of the
1015 on-disk snapshot list */
1016 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1017 __le32 s_error_count; /* number of fs errors */
1018 __le32 s_first_error_time; /* first time an error happened */
1019 __le32 s_first_error_ino; /* inode involved in first error */
1020 __le64 s_first_error_block; /* block involved of first error */
1021 __u8 s_first_error_func[32]; /* function where the error happened */
1022 __le32 s_first_error_line; /* line number where error happened */
1023 __le32 s_last_error_time; /* most recent time of an error */
1024 __le32 s_last_error_ino; /* inode involved in last error */
1025 __le32 s_last_error_line; /* line number where error happened */
1026 __le64 s_last_error_block; /* block involved of last error */
1027 __u8 s_last_error_func[32]; /* function where the error happened */
1028 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1029 __u8 s_mount_opts[64];
1030 __le32 s_reserved[112]; /* Padding to the end of the block */
1033 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1035 #ifdef __KERNEL__
1038 * run-time mount flags
1040 #define EXT4_MF_MNTDIR_SAMPLED 0x0001
1041 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1044 * fourth extended-fs super-block data in memory
1046 struct ext4_sb_info {
1047 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1048 unsigned long s_inodes_per_block;/* Number of inodes per block */
1049 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1050 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1051 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1052 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1053 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1054 ext4_group_t s_groups_count; /* Number of groups in the fs */
1055 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1056 unsigned long s_overhead_last; /* Last calculated overhead */
1057 unsigned long s_blocks_last; /* Last seen block count */
1058 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1059 struct buffer_head * s_sbh; /* Buffer containing the super block */
1060 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1061 struct buffer_head **s_group_desc;
1062 unsigned int s_mount_opt;
1063 unsigned int s_mount_flags;
1064 ext4_fsblk_t s_sb_block;
1065 uid_t s_resuid;
1066 gid_t s_resgid;
1067 unsigned short s_mount_state;
1068 unsigned short s_pad;
1069 int s_addr_per_block_bits;
1070 int s_desc_per_block_bits;
1071 int s_inode_size;
1072 int s_first_ino;
1073 unsigned int s_inode_readahead_blks;
1074 unsigned int s_inode_goal;
1075 spinlock_t s_next_gen_lock;
1076 u32 s_next_generation;
1077 u32 s_hash_seed[4];
1078 int s_def_hash_version;
1079 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1080 struct percpu_counter s_freeblocks_counter;
1081 struct percpu_counter s_freeinodes_counter;
1082 struct percpu_counter s_dirs_counter;
1083 struct percpu_counter s_dirtyblocks_counter;
1084 struct blockgroup_lock *s_blockgroup_lock;
1085 struct proc_dir_entry *s_proc;
1086 struct kobject s_kobj;
1087 struct completion s_kobj_unregister;
1089 /* Journaling */
1090 struct inode *s_journal_inode;
1091 struct journal_s *s_journal;
1092 struct list_head s_orphan;
1093 struct mutex s_orphan_lock;
1094 struct mutex s_resize_lock;
1095 unsigned long s_commit_interval;
1096 u32 s_max_batch_time;
1097 u32 s_min_batch_time;
1098 struct block_device *journal_bdev;
1099 #ifdef CONFIG_JBD2_DEBUG
1100 struct timer_list turn_ro_timer; /* For turning read-only (crash simulation) */
1101 wait_queue_head_t ro_wait_queue; /* For people waiting for the fs to go read-only */
1102 #endif
1103 #ifdef CONFIG_QUOTA
1104 char *s_qf_names[MAXQUOTAS]; /* Names of quota files with journalled quota */
1105 int s_jquota_fmt; /* Format of quota to use */
1106 #endif
1107 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1108 struct rb_root system_blks;
1110 #ifdef EXTENTS_STATS
1111 /* ext4 extents stats */
1112 unsigned long s_ext_min;
1113 unsigned long s_ext_max;
1114 unsigned long s_depth_max;
1115 spinlock_t s_ext_stats_lock;
1116 unsigned long s_ext_blocks;
1117 unsigned long s_ext_extents;
1118 #endif
1120 /* for buddy allocator */
1121 struct ext4_group_info ***s_group_info;
1122 struct inode *s_buddy_cache;
1123 long s_blocks_reserved;
1124 spinlock_t s_reserve_lock;
1125 spinlock_t s_md_lock;
1126 tid_t s_last_transaction;
1127 unsigned short *s_mb_offsets;
1128 unsigned int *s_mb_maxs;
1130 /* tunables */
1131 unsigned long s_stripe;
1132 unsigned int s_mb_stream_request;
1133 unsigned int s_mb_max_to_scan;
1134 unsigned int s_mb_min_to_scan;
1135 unsigned int s_mb_stats;
1136 unsigned int s_mb_order2_reqs;
1137 unsigned int s_mb_group_prealloc;
1138 unsigned int s_max_writeback_mb_bump;
1139 /* where last allocation was done - for stream allocation */
1140 unsigned long s_mb_last_group;
1141 unsigned long s_mb_last_start;
1143 /* stats for buddy allocator */
1144 spinlock_t s_mb_pa_lock;
1145 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1146 atomic_t s_bal_success; /* we found long enough chunks */
1147 atomic_t s_bal_allocated; /* in blocks */
1148 atomic_t s_bal_ex_scanned; /* total extents scanned */
1149 atomic_t s_bal_goals; /* goal hits */
1150 atomic_t s_bal_breaks; /* too long searches */
1151 atomic_t s_bal_2orders; /* 2^order hits */
1152 spinlock_t s_bal_lock;
1153 unsigned long s_mb_buddies_generated;
1154 unsigned long long s_mb_generation_time;
1155 atomic_t s_mb_lost_chunks;
1156 atomic_t s_mb_preallocated;
1157 atomic_t s_mb_discarded;
1158 atomic_t s_lock_busy;
1160 /* locality groups */
1161 struct ext4_locality_group __percpu *s_locality_groups;
1163 /* for write statistics */
1164 unsigned long s_sectors_written_start;
1165 u64 s_kbytes_written;
1167 unsigned int s_log_groups_per_flex;
1168 struct flex_groups *s_flex_groups;
1170 /* workqueue for dio unwritten */
1171 struct workqueue_struct *dio_unwritten_wq;
1173 /* timer for periodic error stats printing */
1174 struct timer_list s_err_report;
1177 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1179 return sb->s_fs_info;
1181 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1183 return container_of(inode, struct ext4_inode_info, vfs_inode);
1186 static inline struct timespec ext4_current_time(struct inode *inode)
1188 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1189 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1192 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1194 return ino == EXT4_ROOT_INO ||
1195 ino == EXT4_JOURNAL_INO ||
1196 ino == EXT4_RESIZE_INO ||
1197 (ino >= EXT4_FIRST_INO(sb) &&
1198 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1202 * Inode dynamic state flags
1204 enum {
1205 EXT4_STATE_JDATA, /* journaled data exists */
1206 EXT4_STATE_NEW, /* inode is newly created */
1207 EXT4_STATE_XATTR, /* has in-inode xattrs */
1208 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1209 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1210 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1211 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
1212 EXT4_STATE_NEWENTRY, /* File just added to dir */
1215 #define EXT4_INODE_BIT_FNS(name, field) \
1216 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1218 return test_bit(bit, &EXT4_I(inode)->i_##field); \
1220 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1222 set_bit(bit, &EXT4_I(inode)->i_##field); \
1224 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1226 clear_bit(bit, &EXT4_I(inode)->i_##field); \
1229 EXT4_INODE_BIT_FNS(flag, flags)
1230 EXT4_INODE_BIT_FNS(state, state_flags)
1231 #else
1232 /* Assume that user mode programs are passing in an ext4fs superblock, not
1233 * a kernel struct super_block. This will allow us to call the feature-test
1234 * macros from user land. */
1235 #define EXT4_SB(sb) (sb)
1236 #endif
1238 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1241 * Codes for operating systems
1243 #define EXT4_OS_LINUX 0
1244 #define EXT4_OS_HURD 1
1245 #define EXT4_OS_MASIX 2
1246 #define EXT4_OS_FREEBSD 3
1247 #define EXT4_OS_LITES 4
1250 * Revision levels
1252 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1253 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1255 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1256 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1258 #define EXT4_GOOD_OLD_INODE_SIZE 128
1261 * Feature set definitions
1264 #define EXT4_HAS_COMPAT_FEATURE(sb,mask) \
1265 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1266 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \
1267 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1268 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \
1269 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1270 #define EXT4_SET_COMPAT_FEATURE(sb,mask) \
1271 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1272 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \
1273 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1274 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \
1275 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1276 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \
1277 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1278 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
1279 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1280 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \
1281 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1283 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1284 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1285 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1286 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1287 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1288 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1290 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1291 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1292 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1293 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1294 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1295 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1296 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1298 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1299 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1300 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1301 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1302 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1303 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1304 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1305 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1306 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1307 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1308 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1310 #define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR
1311 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1312 EXT4_FEATURE_INCOMPAT_RECOVER| \
1313 EXT4_FEATURE_INCOMPAT_META_BG| \
1314 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1315 EXT4_FEATURE_INCOMPAT_64BIT| \
1316 EXT4_FEATURE_INCOMPAT_FLEX_BG)
1317 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1318 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1319 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1320 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1321 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1322 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1323 EXT4_FEATURE_RO_COMPAT_HUGE_FILE)
1326 * Default values for user and/or group using reserved blocks
1328 #define EXT4_DEF_RESUID 0
1329 #define EXT4_DEF_RESGID 0
1331 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
1334 * Default mount options
1336 #define EXT4_DEFM_DEBUG 0x0001
1337 #define EXT4_DEFM_BSDGROUPS 0x0002
1338 #define EXT4_DEFM_XATTR_USER 0x0004
1339 #define EXT4_DEFM_ACL 0x0008
1340 #define EXT4_DEFM_UID16 0x0010
1341 #define EXT4_DEFM_JMODE 0x0060
1342 #define EXT4_DEFM_JMODE_DATA 0x0020
1343 #define EXT4_DEFM_JMODE_ORDERED 0x0040
1344 #define EXT4_DEFM_JMODE_WBACK 0x0060
1345 #define EXT4_DEFM_NOBARRIER 0x0100
1346 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1347 #define EXT4_DEFM_DISCARD 0x0400
1348 #define EXT4_DEFM_NODELALLOC 0x0800
1351 * Default journal batch times
1353 #define EXT4_DEF_MIN_BATCH_TIME 0
1354 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1357 * Minimum number of groups in a flexgroup before we separate out
1358 * directories into the first block group of a flexgroup
1360 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1363 * Structure of a directory entry
1365 #define EXT4_NAME_LEN 255
1367 struct ext4_dir_entry {
1368 __le32 inode; /* Inode number */
1369 __le16 rec_len; /* Directory entry length */
1370 __le16 name_len; /* Name length */
1371 char name[EXT4_NAME_LEN]; /* File name */
1375 * The new version of the directory entry. Since EXT4 structures are
1376 * stored in intel byte order, and the name_len field could never be
1377 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1378 * file_type field.
1380 struct ext4_dir_entry_2 {
1381 __le32 inode; /* Inode number */
1382 __le16 rec_len; /* Directory entry length */
1383 __u8 name_len; /* Name length */
1384 __u8 file_type;
1385 char name[EXT4_NAME_LEN]; /* File name */
1389 * Ext4 directory file types. Only the low 3 bits are used. The
1390 * other bits are reserved for now.
1392 #define EXT4_FT_UNKNOWN 0
1393 #define EXT4_FT_REG_FILE 1
1394 #define EXT4_FT_DIR 2
1395 #define EXT4_FT_CHRDEV 3
1396 #define EXT4_FT_BLKDEV 4
1397 #define EXT4_FT_FIFO 5
1398 #define EXT4_FT_SOCK 6
1399 #define EXT4_FT_SYMLINK 7
1401 #define EXT4_FT_MAX 8
1404 * EXT4_DIR_PAD defines the directory entries boundaries
1406 * NOTE: It must be a multiple of 4
1408 #define EXT4_DIR_PAD 4
1409 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1410 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1411 ~EXT4_DIR_ROUND)
1412 #define EXT4_MAX_REC_LEN ((1<<16)-1)
1415 * If we ever get support for fs block sizes > page_size, we'll need
1416 * to remove the #if statements in the next two functions...
1418 static inline unsigned int
1419 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1421 unsigned len = le16_to_cpu(dlen);
1423 #if (PAGE_CACHE_SIZE >= 65536)
1424 if (len == EXT4_MAX_REC_LEN || len == 0)
1425 return blocksize;
1426 return (len & 65532) | ((len & 3) << 16);
1427 #else
1428 return len;
1429 #endif
1432 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1434 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1435 BUG();
1436 #if (PAGE_CACHE_SIZE >= 65536)
1437 if (len < 65536)
1438 return cpu_to_le16(len);
1439 if (len == blocksize) {
1440 if (blocksize == 65536)
1441 return cpu_to_le16(EXT4_MAX_REC_LEN);
1442 else
1443 return cpu_to_le16(0);
1445 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1446 #else
1447 return cpu_to_le16(len);
1448 #endif
1452 * Hash Tree Directory indexing
1453 * (c) Daniel Phillips, 2001
1456 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1457 EXT4_FEATURE_COMPAT_DIR_INDEX) && \
1458 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
1459 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1460 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1462 /* Legal values for the dx_root hash_version field: */
1464 #define DX_HASH_LEGACY 0
1465 #define DX_HASH_HALF_MD4 1
1466 #define DX_HASH_TEA 2
1467 #define DX_HASH_LEGACY_UNSIGNED 3
1468 #define DX_HASH_HALF_MD4_UNSIGNED 4
1469 #define DX_HASH_TEA_UNSIGNED 5
1471 #ifdef __KERNEL__
1473 /* hash info structure used by the directory hash */
1474 struct dx_hash_info
1476 u32 hash;
1477 u32 minor_hash;
1478 int hash_version;
1479 u32 *seed;
1482 #define EXT4_HTREE_EOF 0x7fffffff
1485 * Control parameters used by ext4_htree_next_block
1487 #define HASH_NB_ALWAYS 1
1491 * Describe an inode's exact location on disk and in memory
1493 struct ext4_iloc
1495 struct buffer_head *bh;
1496 unsigned long offset;
1497 ext4_group_t block_group;
1500 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
1502 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
1506 * This structure is stuffed into the struct file's private_data field
1507 * for directories. It is where we put information so that we can do
1508 * readdir operations in hash tree order.
1510 struct dir_private_info {
1511 struct rb_root root;
1512 struct rb_node *curr_node;
1513 struct fname *extra_fname;
1514 loff_t last_pos;
1515 __u32 curr_hash;
1516 __u32 curr_minor_hash;
1517 __u32 next_hash;
1520 /* calculate the first block number of the group */
1521 static inline ext4_fsblk_t
1522 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
1524 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
1525 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
1529 * Special error return code only used by dx_probe() and its callers.
1531 #define ERR_BAD_DX_DIR -75000
1533 void ext4_get_group_no_and_offset(struct super_block *sb, ext4_fsblk_t blocknr,
1534 ext4_group_t *blockgrpp, ext4_grpblk_t *offsetp);
1536 extern struct proc_dir_entry *ext4_proc_root;
1539 * Function prototypes
1543 * Ok, these declarations are also in <linux/kernel.h> but none of the
1544 * ext4 source programs needs to include it so they are duplicated here.
1546 # define NORET_TYPE /**/
1547 # define ATTRIB_NORET __attribute__((noreturn))
1548 # define NORET_AND noreturn,
1550 /* bitmap.c */
1551 extern unsigned int ext4_count_free(struct buffer_head *, unsigned);
1553 /* balloc.c */
1554 extern unsigned int ext4_block_group(struct super_block *sb,
1555 ext4_fsblk_t blocknr);
1556 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
1557 ext4_fsblk_t blocknr);
1558 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
1559 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
1560 ext4_group_t group);
1561 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
1562 ext4_fsblk_t goal, unsigned long *count, int *errp);
1563 extern int ext4_claim_free_blocks(struct ext4_sb_info *sbi, s64 nblocks);
1564 extern int ext4_has_free_blocks(struct ext4_sb_info *sbi, s64 nblocks);
1565 extern void ext4_add_groupblocks(handle_t *handle, struct super_block *sb,
1566 ext4_fsblk_t block, unsigned long count);
1567 extern ext4_fsblk_t ext4_count_free_blocks(struct super_block *);
1568 extern void ext4_check_blocks_bitmap(struct super_block *);
1569 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
1570 ext4_group_t block_group,
1571 struct buffer_head ** bh);
1572 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
1573 struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
1574 ext4_group_t block_group);
1575 extern unsigned ext4_init_block_bitmap(struct super_block *sb,
1576 struct buffer_head *bh,
1577 ext4_group_t group,
1578 struct ext4_group_desc *desc);
1579 #define ext4_free_blocks_after_init(sb, group, desc) \
1580 ext4_init_block_bitmap(sb, NULL, group, desc)
1582 /* dir.c */
1583 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
1584 struct ext4_dir_entry_2 *,
1585 struct buffer_head *, unsigned int);
1586 #define ext4_check_dir_entry(dir, de, bh, offset) \
1587 __ext4_check_dir_entry(__func__, __LINE__, (dir), (de), (bh), (offset))
1588 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
1589 __u32 minor_hash,
1590 struct ext4_dir_entry_2 *dirent);
1591 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
1593 /* fsync.c */
1594 extern int ext4_sync_file(struct file *, int);
1596 /* hash.c */
1597 extern int ext4fs_dirhash(const char *name, int len, struct
1598 dx_hash_info *hinfo);
1600 /* ialloc.c */
1601 extern struct inode *ext4_new_inode(handle_t *, struct inode *, int,
1602 const struct qstr *qstr, __u32 goal);
1603 extern void ext4_free_inode(handle_t *, struct inode *);
1604 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
1605 extern unsigned long ext4_count_free_inodes(struct super_block *);
1606 extern unsigned long ext4_count_dirs(struct super_block *);
1607 extern void ext4_check_inodes_bitmap(struct super_block *);
1608 extern unsigned ext4_init_inode_bitmap(struct super_block *sb,
1609 struct buffer_head *bh,
1610 ext4_group_t group,
1611 struct ext4_group_desc *desc);
1612 extern void mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
1614 /* mballoc.c */
1615 extern long ext4_mb_stats;
1616 extern long ext4_mb_max_to_scan;
1617 extern int ext4_mb_init(struct super_block *, int);
1618 extern int ext4_mb_release(struct super_block *);
1619 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
1620 struct ext4_allocation_request *, int *);
1621 extern int ext4_mb_reserve_blocks(struct super_block *, int);
1622 extern void ext4_discard_preallocations(struct inode *);
1623 extern int __init init_ext4_mballoc(void);
1624 extern void exit_ext4_mballoc(void);
1625 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
1626 struct buffer_head *bh, ext4_fsblk_t block,
1627 unsigned long count, int flags);
1628 extern int ext4_mb_add_groupinfo(struct super_block *sb,
1629 ext4_group_t i, struct ext4_group_desc *desc);
1630 extern int ext4_mb_get_buddy_cache_lock(struct super_block *, ext4_group_t);
1631 extern void ext4_mb_put_buddy_cache_lock(struct super_block *,
1632 ext4_group_t, int);
1633 /* inode.c */
1634 struct buffer_head *ext4_getblk(handle_t *, struct inode *,
1635 ext4_lblk_t, int, int *);
1636 struct buffer_head *ext4_bread(handle_t *, struct inode *,
1637 ext4_lblk_t, int, int *);
1638 int ext4_get_block(struct inode *inode, sector_t iblock,
1639 struct buffer_head *bh_result, int create);
1641 extern struct inode *ext4_iget(struct super_block *, unsigned long);
1642 extern int ext4_write_inode(struct inode *, struct writeback_control *);
1643 extern int ext4_setattr(struct dentry *, struct iattr *);
1644 extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
1645 struct kstat *stat);
1646 extern void ext4_evict_inode(struct inode *);
1647 extern void ext4_clear_inode(struct inode *);
1648 extern int ext4_sync_inode(handle_t *, struct inode *);
1649 extern void ext4_dirty_inode(struct inode *);
1650 extern int ext4_change_inode_journal_flag(struct inode *, int);
1651 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
1652 extern int ext4_can_truncate(struct inode *inode);
1653 extern void ext4_truncate(struct inode *);
1654 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
1655 extern void ext4_set_inode_flags(struct inode *);
1656 extern void ext4_get_inode_flags(struct ext4_inode_info *);
1657 extern int ext4_alloc_da_blocks(struct inode *inode);
1658 extern void ext4_set_aops(struct inode *inode);
1659 extern int ext4_writepage_trans_blocks(struct inode *);
1660 extern int ext4_meta_trans_blocks(struct inode *, int nrblocks, int idxblocks);
1661 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
1662 extern int ext4_block_truncate_page(handle_t *handle,
1663 struct address_space *mapping, loff_t from);
1664 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
1665 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
1666 extern int flush_completed_IO(struct inode *inode);
1667 extern void ext4_da_update_reserve_space(struct inode *inode,
1668 int used, int quota_claim);
1669 /* ioctl.c */
1670 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
1671 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
1673 /* migrate.c */
1674 extern int ext4_ext_migrate(struct inode *);
1676 /* namei.c */
1677 extern int ext4_orphan_add(handle_t *, struct inode *);
1678 extern int ext4_orphan_del(handle_t *, struct inode *);
1679 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
1680 __u32 start_minor_hash, __u32 *next_hash);
1682 /* resize.c */
1683 extern int ext4_group_add(struct super_block *sb,
1684 struct ext4_new_group_data *input);
1685 extern int ext4_group_extend(struct super_block *sb,
1686 struct ext4_super_block *es,
1687 ext4_fsblk_t n_blocks_count);
1689 /* super.c */
1690 extern void __ext4_error(struct super_block *, const char *, unsigned int,
1691 const char *, ...)
1692 __attribute__ ((format (printf, 4, 5)));
1693 #define ext4_error(sb, message...) __ext4_error(sb, __func__, \
1694 __LINE__, ## message)
1695 extern void ext4_error_inode(struct inode *, const char *, unsigned int,
1696 ext4_fsblk_t, const char *, ...)
1697 __attribute__ ((format (printf, 5, 6)));
1698 extern void ext4_error_file(struct file *, const char *, unsigned int,
1699 const char *, ...)
1700 __attribute__ ((format (printf, 4, 5)));
1701 extern void __ext4_std_error(struct super_block *, const char *,
1702 unsigned int, int);
1703 extern void __ext4_abort(struct super_block *, const char *, unsigned int,
1704 const char *, ...)
1705 __attribute__ ((format (printf, 4, 5)));
1706 #define ext4_abort(sb, message...) __ext4_abort(sb, __func__, \
1707 __LINE__, ## message)
1708 extern void __ext4_warning(struct super_block *, const char *, unsigned int,
1709 const char *, ...)
1710 __attribute__ ((format (printf, 4, 5)));
1711 #define ext4_warning(sb, message...) __ext4_warning(sb, __func__, \
1712 __LINE__, ## message)
1713 extern void ext4_msg(struct super_block *, const char *, const char *, ...)
1714 __attribute__ ((format (printf, 3, 4)));
1715 extern void __ext4_grp_locked_error(const char *, unsigned int, \
1716 struct super_block *, ext4_group_t, \
1717 unsigned long, ext4_fsblk_t, \
1718 const char *, ...)
1719 __attribute__ ((format (printf, 7, 8)));
1720 #define ext4_grp_locked_error(sb, grp, message...) \
1721 __ext4_grp_locked_error(__func__, __LINE__, (sb), (grp), ## message)
1722 extern void ext4_update_dynamic_rev(struct super_block *sb);
1723 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
1724 __u32 compat);
1725 extern int ext4_update_rocompat_feature(handle_t *handle,
1726 struct super_block *sb, __u32 rocompat);
1727 extern int ext4_update_incompat_feature(handle_t *handle,
1728 struct super_block *sb, __u32 incompat);
1729 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
1730 struct ext4_group_desc *bg);
1731 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
1732 struct ext4_group_desc *bg);
1733 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
1734 struct ext4_group_desc *bg);
1735 extern __u32 ext4_free_blks_count(struct super_block *sb,
1736 struct ext4_group_desc *bg);
1737 extern __u32 ext4_free_inodes_count(struct super_block *sb,
1738 struct ext4_group_desc *bg);
1739 extern __u32 ext4_used_dirs_count(struct super_block *sb,
1740 struct ext4_group_desc *bg);
1741 extern __u32 ext4_itable_unused_count(struct super_block *sb,
1742 struct ext4_group_desc *bg);
1743 extern void ext4_block_bitmap_set(struct super_block *sb,
1744 struct ext4_group_desc *bg, ext4_fsblk_t blk);
1745 extern void ext4_inode_bitmap_set(struct super_block *sb,
1746 struct ext4_group_desc *bg, ext4_fsblk_t blk);
1747 extern void ext4_inode_table_set(struct super_block *sb,
1748 struct ext4_group_desc *bg, ext4_fsblk_t blk);
1749 extern void ext4_free_blks_set(struct super_block *sb,
1750 struct ext4_group_desc *bg, __u32 count);
1751 extern void ext4_free_inodes_set(struct super_block *sb,
1752 struct ext4_group_desc *bg, __u32 count);
1753 extern void ext4_used_dirs_set(struct super_block *sb,
1754 struct ext4_group_desc *bg, __u32 count);
1755 extern void ext4_itable_unused_set(struct super_block *sb,
1756 struct ext4_group_desc *bg, __u32 count);
1757 extern __le16 ext4_group_desc_csum(struct ext4_sb_info *sbi, __u32 group,
1758 struct ext4_group_desc *gdp);
1759 extern int ext4_group_desc_csum_verify(struct ext4_sb_info *sbi, __u32 group,
1760 struct ext4_group_desc *gdp);
1762 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
1764 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
1765 le32_to_cpu(es->s_blocks_count_lo);
1768 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
1770 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
1771 le32_to_cpu(es->s_r_blocks_count_lo);
1774 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
1776 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
1777 le32_to_cpu(es->s_free_blocks_count_lo);
1780 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
1781 ext4_fsblk_t blk)
1783 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
1784 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
1787 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
1788 ext4_fsblk_t blk)
1790 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
1791 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
1794 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
1795 ext4_fsblk_t blk)
1797 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
1798 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
1801 static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
1803 if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
1804 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
1805 le32_to_cpu(raw_inode->i_size_lo);
1806 else
1807 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
1810 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
1812 raw_inode->i_size_lo = cpu_to_le32(i_size);
1813 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
1816 static inline
1817 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
1818 ext4_group_t group)
1820 struct ext4_group_info ***grp_info;
1821 long indexv, indexh;
1822 grp_info = EXT4_SB(sb)->s_group_info;
1823 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
1824 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
1825 return grp_info[indexv][indexh];
1829 * Reading s_groups_count requires using smp_rmb() afterwards. See
1830 * the locking protocol documented in the comments of ext4_group_add()
1831 * in resize.c
1833 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
1835 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
1837 smp_rmb();
1838 return ngroups;
1841 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
1842 ext4_group_t block_group)
1844 return block_group >> sbi->s_log_groups_per_flex;
1847 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
1849 return 1 << sbi->s_log_groups_per_flex;
1852 #define ext4_std_error(sb, errno) \
1853 do { \
1854 if ((errno)) \
1855 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
1856 } while (0)
1858 #ifdef CONFIG_SMP
1859 /* Each CPU can accumulate percpu_counter_batch blocks in their local
1860 * counters. So we need to make sure we have free blocks more
1861 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
1863 #define EXT4_FREEBLOCKS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
1864 #else
1865 #define EXT4_FREEBLOCKS_WATERMARK 0
1866 #endif
1868 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
1870 down_write(&EXT4_I(inode)->i_data_sem);
1871 if (newsize > EXT4_I(inode)->i_disksize)
1872 EXT4_I(inode)->i_disksize = newsize;
1873 up_write(&EXT4_I(inode)->i_data_sem);
1874 return ;
1877 struct ext4_group_info {
1878 unsigned long bb_state;
1879 struct rb_root bb_free_root;
1880 ext4_grpblk_t bb_first_free; /* first free block */
1881 ext4_grpblk_t bb_free; /* total free blocks */
1882 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
1883 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
1884 struct list_head bb_prealloc_list;
1885 #ifdef DOUBLE_CHECK
1886 void *bb_bitmap;
1887 #endif
1888 struct rw_semaphore alloc_sem;
1889 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
1890 * regions, index is order.
1891 * bb_counters[3] = 5 means
1892 * 5 free 8-block regions. */
1895 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
1897 #define EXT4_MB_GRP_NEED_INIT(grp) \
1898 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
1900 #define EXT4_MAX_CONTENTION 8
1901 #define EXT4_CONTENTION_THRESHOLD 2
1903 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
1904 ext4_group_t group)
1906 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
1910 * Returns true if the filesystem is busy enough that attempts to
1911 * access the block group locks has run into contention.
1913 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
1915 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
1918 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
1920 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
1921 if (spin_trylock(lock))
1923 * We're able to grab the lock right away, so drop the
1924 * lock contention counter.
1926 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
1927 else {
1929 * The lock is busy, so bump the contention counter,
1930 * and then wait on the spin lock.
1932 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
1933 EXT4_MAX_CONTENTION);
1934 spin_lock(lock);
1938 static inline void ext4_unlock_group(struct super_block *sb,
1939 ext4_group_t group)
1941 spin_unlock(ext4_group_lock_ptr(sb, group));
1944 static inline void ext4_mark_super_dirty(struct super_block *sb)
1946 if (EXT4_SB(sb)->s_journal == NULL)
1947 sb->s_dirt =1;
1951 * Inodes and files operations
1954 /* dir.c */
1955 extern const struct file_operations ext4_dir_operations;
1957 /* file.c */
1958 extern const struct inode_operations ext4_file_inode_operations;
1959 extern const struct file_operations ext4_file_operations;
1961 /* namei.c */
1962 extern const struct inode_operations ext4_dir_inode_operations;
1963 extern const struct inode_operations ext4_special_inode_operations;
1964 extern struct dentry *ext4_get_parent(struct dentry *child);
1966 /* symlink.c */
1967 extern const struct inode_operations ext4_symlink_inode_operations;
1968 extern const struct inode_operations ext4_fast_symlink_inode_operations;
1970 /* block_validity */
1971 extern void ext4_release_system_zone(struct super_block *sb);
1972 extern int ext4_setup_system_zone(struct super_block *sb);
1973 extern int __init init_ext4_system_zone(void);
1974 extern void exit_ext4_system_zone(void);
1975 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
1976 ext4_fsblk_t start_blk,
1977 unsigned int count);
1979 /* extents.c */
1980 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
1981 extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
1982 extern int ext4_ext_index_trans_blocks(struct inode *inode, int nrblocks,
1983 int chunk);
1984 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
1985 struct ext4_map_blocks *map, int flags);
1986 extern void ext4_ext_truncate(struct inode *);
1987 extern void ext4_ext_init(struct super_block *);
1988 extern void ext4_ext_release(struct super_block *);
1989 extern long ext4_fallocate(struct inode *inode, int mode, loff_t offset,
1990 loff_t len);
1991 extern int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset,
1992 ssize_t len);
1993 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
1994 struct ext4_map_blocks *map, int flags);
1995 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
1996 __u64 start, __u64 len);
1997 /* move_extent.c */
1998 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
1999 __u64 start_orig, __u64 start_donor,
2000 __u64 len, __u64 *moved_len);
2003 /* BH_Uninit flag: blocks are allocated but uninitialized on disk */
2004 enum ext4_state_bits {
2005 BH_Uninit /* blocks are allocated but uninitialized on disk */
2006 = BH_JBDPrivateStart,
2009 BUFFER_FNS(Uninit, uninit)
2010 TAS_BUFFER_FNS(Uninit, uninit)
2013 * Add new method to test wether block and inode bitmaps are properly
2014 * initialized. With uninit_bg reading the block from disk is not enough
2015 * to mark the bitmap uptodate. We need to also zero-out the bitmap
2017 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
2019 static inline int bitmap_uptodate(struct buffer_head *bh)
2021 return (buffer_uptodate(bh) &&
2022 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
2024 static inline void set_bitmap_uptodate(struct buffer_head *bh)
2026 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
2029 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
2031 #endif /* __KERNEL__ */
2033 #endif /* _EXT4_H */