Linux Kernel  3.7.1
 All Data Structures Namespaces Files Functions Variables Typedefs Enumerations Enumerator Macros Groups Pages
ext4.h
Go to the documentation of this file.
1 /*
2  * ext4.h
3  *
4  * Copyright (C) 1992, 1993, 1994, 1995
5  * Remy Card ([email protected])
6  * Laboratoire MASI - Institut Blaise Pascal
7  * Universite Pierre et Marie Curie (Paris VI)
8  *
9  * from
10  *
11  * linux/include/linux/minix_fs.h
12  *
13  * Copyright (C) 1991, 1992 Linus Torvalds
14  */
15 
16 #ifndef _EXT4_H
17 #define _EXT4_H
18 
19 #include <linux/types.h>
20 #include <linux/blkdev.h>
21 #include <linux/magic.h>
22 #include <linux/jbd2.h>
23 #include <linux/quota.h>
24 #include <linux/rwsem.h>
25 #include <linux/rbtree.h>
26 #include <linux/seqlock.h>
27 #include <linux/mutex.h>
28 #include <linux/timer.h>
29 #include <linux/wait.h>
30 #include <linux/blockgroup_lock.h>
31 #include <linux/percpu_counter.h>
32 #include <crypto/hash.h>
33 #ifdef __KERNEL__
34 #include <linux/compat.h>
35 #endif
36 
37 /*
38  * The fourth extended filesystem constants/structures
39  */
40 
41 /*
42  * Define EXT4FS_DEBUG to produce debug messages
43  */
44 #undef EXT4FS_DEBUG
45 
46 /*
47  * Debug code
48  */
49 #ifdef EXT4FS_DEBUG
50 #define ext4_debug(f, a...) \
51  do { \
52  printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
53  __FILE__, __LINE__, __func__); \
54  printk(KERN_DEBUG f, ## a); \
55  } while (0)
56 #else
57 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
58 #endif
59 
60 #define EXT4_ERROR_INODE(inode, fmt, a...) \
61  ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
62 
63 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
64  ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
65 
66 #define EXT4_ERROR_FILE(file, block, fmt, a...) \
67  ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
68 
69 /* data type for block offset of block group */
70 typedef int ext4_grpblk_t;
71 
72 /* data type for filesystem-wide blocks number */
73 typedef unsigned long long ext4_fsblk_t;
74 
75 /* data type for file logical block number */
77 
78 /* data type for block group number */
79 typedef unsigned int ext4_group_t;
80 
81 /*
82  * Flags used in mballoc's allocation_context flags field.
83  *
84  * Also used to show what's going on for debugging purposes when the
85  * flag field is exported via the traceport interface
86  */
87 
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
114 
116  /* target inode for block we're allocating */
117  struct inode *inode;
118  /* how many blocks we want to allocate */
119  unsigned int len;
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_* */
133  unsigned int flags;
134 };
135 
136 /*
137  * Logical to physical block mapping, used by ext4_map_blocks()
138  *
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.
142  */
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.
154  */
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)
159 
163  unsigned int m_len;
164  unsigned int m_flags;
165 };
166 
167 /*
168  * For delayed allocation tracking
169  */
171  struct inode *inode;
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 */
177  int io_done;
179  int retval;
180 };
181 
182 /*
183  * Flags for ext4_io_end->flags
184  */
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 
190 struct ext4_io_page {
191  struct page *p_page;
193 };
194 
195 #define MAX_IO_PAGES 128
196 
197 /*
198  * For converting uninitialized extents on a work queue.
199  *
200  * 'page' is only used from the writepage() path; 'pages' is only used for
201  * buffered writes; they are used to keep page references until conversion
202  * takes place. For AIO/DIO, neither field is filled in.
203  */
204 typedef struct ext4_io_end {
205  struct list_head list; /* per-file finished IO list */
206  struct inode *inode; /* file being written to */
207  unsigned int flag; /* unwritten or not */
208  struct page *page; /* for writepage() path */
209  loff_t offset; /* offset in the file */
210  ssize_t size; /* size of the extent */
211  struct work_struct work; /* data work queue */
212  struct kiocb *iocb; /* iocb struct for AIO */
213  int result; /* error value for AIO */
214  int num_io_pages; /* for writepages() */
215  struct ext4_io_page *pages[MAX_IO_PAGES]; /* for writepages() */
216 } ext4_io_end_t;
217 
219  int io_op;
220  struct bio *io_bio;
224 };
225 
226 /*
227  * Special inodes numbers
228  */
229 #define EXT4_BAD_INO 1 /* Bad blocks inode */
230 #define EXT4_ROOT_INO 2 /* Root inode */
231 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */
232 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
233 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
234 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
235 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
236 #define EXT4_JOURNAL_INO 8 /* Journal inode */
237 
238 /* First non-reserved inode for old ext4 filesystems */
239 #define EXT4_GOOD_OLD_FIRST_INO 11
240 
241 /*
242  * Maximal count of links to a file
243  */
244 #define EXT4_LINK_MAX 65000
245 
246 /*
247  * Macro-instructions used to manage several block sizes
248  */
249 #define EXT4_MIN_BLOCK_SIZE 1024
250 #define EXT4_MAX_BLOCK_SIZE 65536
251 #define EXT4_MIN_BLOCK_LOG_SIZE 10
252 #define EXT4_MAX_BLOCK_LOG_SIZE 16
253 #ifdef __KERNEL__
254 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
255 #else
256 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
257 #endif
258 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
259 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
260  EXT4_SB(s)->s_cluster_bits)
261 #ifdef __KERNEL__
262 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
263 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
264 #else
265 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
266 #endif
267 #ifdef __KERNEL__
268 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
269 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
270 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
271 #else
272 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
273  EXT4_GOOD_OLD_INODE_SIZE : \
274  (s)->s_inode_size)
275 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
276  EXT4_GOOD_OLD_FIRST_INO : \
277  (s)->s_first_ino)
278 #endif
279 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
280 
281 /* Translate a block number to a cluster number */
282 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
283 /* Translate a cluster number to a block number */
284 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
285 /* Translate # of blks to # of clusters */
286 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
287  (sbi)->s_cluster_bits)
288 
289 /*
290  * Structure of a blocks group descriptor
291  */
293 {
294  __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
295  __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
296  __le32 bg_inode_table_lo; /* Inodes table block */
297  __le16 bg_free_blocks_count_lo;/* Free blocks count */
298  __le16 bg_free_inodes_count_lo;/* Free inodes count */
299  __le16 bg_used_dirs_count_lo; /* Directories count */
300  __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
301  __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
302  __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
303  __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
304  __le16 bg_itable_unused_lo; /* Unused inodes count */
305  __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
306  __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
307  __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
308  __le32 bg_inode_table_hi; /* Inodes table block MSB */
309  __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
310  __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
311  __le16 bg_used_dirs_count_hi; /* Directories count MSB */
312  __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
313  __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
314  __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
315  __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
317 };
318 
319 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
320  (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
321  sizeof(__le16))
322 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
323  (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
324  sizeof(__le16))
325 
326 /*
327  * Structure of a flex block group info
328  */
329 
330 struct flex_groups {
334 };
335 
336 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
337 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
338 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
339 
340 /*
341  * Macro-instructions used to manage group descriptors
342  */
343 #define EXT4_MIN_DESC_SIZE 32
344 #define EXT4_MIN_DESC_SIZE_64BIT 64
345 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
346 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
347 #ifdef __KERNEL__
348 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
349 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
350 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
351 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
352 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
353 #else
354 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
355 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
356 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
357 #endif
358 
359 /*
360  * Constants relative to the data blocks
361  */
362 #define EXT4_NDIR_BLOCKS 12
363 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
364 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
365 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
366 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
367 
368 /*
369  * Inode flags
370  */
371 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
372 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
373 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
374 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
375 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
376 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
377 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
378 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
379 /* Reserved for compression usage... */
380 #define EXT4_DIRTY_FL 0x00000100
381 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
382 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
383 #define EXT4_ECOMPR_FL 0x00000800 /* Compression error */
384 /* End compression flags --- maybe not all used */
385 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
386 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
387 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
388 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
389 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
390 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
391 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
392 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
393 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
394 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
395 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
396 
397 #define EXT4_FL_USER_VISIBLE 0x004BDFFF /* User visible flags */
398 #define EXT4_FL_USER_MODIFIABLE 0x004B80FF /* User modifiable flags */
399 
400 /* Flags that should be inherited by new inodes from their parent. */
401 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
402  EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
403  EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
404  EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
405 
406 /* Flags that are appropriate for regular files (all but dir-specific ones). */
407 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
408 
409 /* Flags that are appropriate for non-directories/regular files. */
410 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
411 
412 /* Mask out flags that are inappropriate for the given type of inode. */
413 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
414 {
415  if (S_ISDIR(mode))
416  return flags;
417  else if (S_ISREG(mode))
418  return flags & EXT4_REG_FLMASK;
419  else
420  return flags & EXT4_OTHER_FLMASK;
421 }
422 
423 /*
424  * Inode flags used for atomic set/get
425  */
426 enum {
427  EXT4_INODE_SECRM = 0, /* Secure deletion */
428  EXT4_INODE_UNRM = 1, /* Undelete */
429  EXT4_INODE_COMPR = 2, /* Compress file */
430  EXT4_INODE_SYNC = 3, /* Synchronous updates */
431  EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
432  EXT4_INODE_APPEND = 5, /* writes to file may only append */
433  EXT4_INODE_NODUMP = 6, /* do not dump file */
434  EXT4_INODE_NOATIME = 7, /* do not update atime */
435 /* Reserved for compression usage... */
437  EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
438  EXT4_INODE_NOCOMPR = 10, /* Don't compress */
439  EXT4_INODE_ECOMPR = 11, /* Compression error */
440 /* End compression flags --- maybe not all used */
441  EXT4_INODE_INDEX = 12, /* hash-indexed directory */
442  EXT4_INODE_IMAGIC = 13, /* AFS directory */
443  EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
444  EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
445  EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
446  EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
447  EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
448  EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
449  EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
450  EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
451  EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
452 };
453 
454 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
455 #define CHECK_FLAG_VALUE(FLAG) if (!TEST_FLAG_VALUE(FLAG)) { \
456  printk(KERN_EMERG "EXT4 flag fail: " #FLAG ": %d %d\n", \
457  EXT4_##FLAG##_FL, EXT4_INODE_##FLAG); BUG_ON(1); }
458 
459 /*
460  * Since it's pretty easy to mix up bit numbers and hex values, and we
461  * can't do a compile-time test for ENUM values, we use a run-time
462  * test to make sure that EXT4_XXX_FL is consistent with respect to
463  * EXT4_INODE_XXX. If all is well the printk and BUG_ON will all drop
464  * out so it won't cost any extra space in the compiled kernel image.
465  * But it's important that these values are the same, since we are
466  * using EXT4_INODE_XXX to test for the flag values, but EXT4_XX_FL
467  * must be consistent with the values of FS_XXX_FL defined in
468  * include/linux/fs.h and the on-disk values found in ext2, ext3, and
469  * ext4 filesystems, and of course the values defined in e2fsprogs.
470  *
471  * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
472  */
473 static inline void ext4_check_flag_values(void)
474 {
475  CHECK_FLAG_VALUE(SECRM);
476  CHECK_FLAG_VALUE(UNRM);
477  CHECK_FLAG_VALUE(COMPR);
479  CHECK_FLAG_VALUE(IMMUTABLE);
480  CHECK_FLAG_VALUE(APPEND);
481  CHECK_FLAG_VALUE(NODUMP);
482  CHECK_FLAG_VALUE(NOATIME);
484  CHECK_FLAG_VALUE(COMPRBLK);
485  CHECK_FLAG_VALUE(NOCOMPR);
486  CHECK_FLAG_VALUE(ECOMPR);
488  CHECK_FLAG_VALUE(IMAGIC);
489  CHECK_FLAG_VALUE(JOURNAL_DATA);
490  CHECK_FLAG_VALUE(NOTAIL);
491  CHECK_FLAG_VALUE(DIRSYNC);
492  CHECK_FLAG_VALUE(TOPDIR);
493  CHECK_FLAG_VALUE(HUGE_FILE);
494  CHECK_FLAG_VALUE(EXTENTS);
495  CHECK_FLAG_VALUE(EA_INODE);
496  CHECK_FLAG_VALUE(EOFBLOCKS);
498 }
499 
500 /* Used to pass group descriptor data when online resize is done */
502  __u32 group; /* Group number for this data */
503  __u64 block_bitmap; /* Absolute block number of block bitmap */
504  __u64 inode_bitmap; /* Absolute block number of inode bitmap */
505  __u64 inode_table; /* Absolute block number of inode table start */
506  __u32 blocks_count; /* Total number of blocks in this group */
507  __u16 reserved_blocks; /* Number of reserved blocks in this group */
509 };
510 
511 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
512 struct compat_ext4_new_group_input {
513  u32 group;
514  compat_u64 block_bitmap;
515  compat_u64 inode_bitmap;
516  compat_u64 inode_table;
517  u32 blocks_count;
518  u16 reserved_blocks;
519  u16 unused;
520 };
521 #endif
522 
523 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
533 };
534 
535 /* Indexes used to index group tables in ext4_new_group_data */
536 enum {
537  BLOCK_BITMAP = 0, /* block bitmap */
538  INODE_BITMAP, /* inode bitmap */
539  INODE_TABLE, /* inode tables */
541 };
542 
543 /*
544  * Flags used by ext4_map_blocks()
545  */
546  /* Allocate any needed blocks and/or convert an unitialized
547  extent to be an initialized ext4 */
548 #define EXT4_GET_BLOCKS_CREATE 0x0001
549  /* Request the creation of an unitialized extent */
550 #define EXT4_GET_BLOCKS_UNINIT_EXT 0x0002
551 #define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT (EXT4_GET_BLOCKS_UNINIT_EXT|\
552  EXT4_GET_BLOCKS_CREATE)
553  /* Caller is from the delayed allocation writeout path,
554  so set the magic i_delalloc_reserve_flag after taking the
555  inode allocation semaphore for */
556 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
557  /* caller is from the direct IO path, request to creation of an
558  unitialized extents if not allocated, split the uninitialized
559  extent if blocks has been preallocated already*/
560 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
561 #define EXT4_GET_BLOCKS_CONVERT 0x0010
562 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
563  EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
564  /* Convert extent to initialized after IO complete */
565 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
566  EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
567  /* Punch out blocks of an extent */
568 #define EXT4_GET_BLOCKS_PUNCH_OUT_EXT 0x0020
569  /* Don't normalize allocation size (used for fallocate) */
570 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
571  /* Request will not result in inode size update (user for fallocate) */
572 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080
573  /* Do not take i_data_sem locking in ext4_map_blocks */
574 #define EXT4_GET_BLOCKS_NO_LOCK 0x0100
575 
576 /*
577  * Flags used by ext4_free_blocks
578  */
579 #define EXT4_FREE_BLOCKS_METADATA 0x0001
580 #define EXT4_FREE_BLOCKS_FORGET 0x0002
581 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
582 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
583 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
584 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
585 
586 /*
587  * Flags used by ext4_discard_partial_page_buffers
588  */
589 #define EXT4_DISCARD_PARTIAL_PG_ZERO_UNMAPPED 0x0001
590 
591 /*
592  * ioctl commands
593  */
594 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
595 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
596 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
597 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
598 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
599 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
600 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
601 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
602 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
603 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
604 #define EXT4_IOC_MIGRATE _IO('f', 9)
605  /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
606  /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
607 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
608 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
609 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
610 
611 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
612 /*
613  * ioctl commands in 32 bit emulation
614  */
615 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
616 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
617 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
618 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
619 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
620 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
621 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
622 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
623 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
624 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
625 #endif
626 
627 /* Max physical block we can address w/o extents */
628 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
629 
630 /*
631  * Structure of an inode on the disk
632  */
633 struct ext4_inode {
634  __le16 i_mode; /* File mode */
635  __le16 i_uid; /* Low 16 bits of Owner Uid */
636  __le32 i_size_lo; /* Size in bytes */
637  __le32 i_atime; /* Access time */
638  __le32 i_ctime; /* Inode Change time */
639  __le32 i_mtime; /* Modification time */
640  __le32 i_dtime; /* Deletion Time */
641  __le16 i_gid; /* Low 16 bits of Group Id */
642  __le16 i_links_count; /* Links count */
643  __le32 i_blocks_lo; /* Blocks count */
644  __le32 i_flags; /* File flags */
645  union {
646  struct {
648  } linux1;
649  struct {
651  } hurd1;
652  struct {
654  } masix1;
655  } osd1; /* OS dependent 1 */
656  __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
657  __le32 i_generation; /* File version (for NFS) */
658  __le32 i_file_acl_lo; /* File ACL */
660  __le32 i_obso_faddr; /* Obsoleted fragment address */
661  union {
662  struct {
663  __le16 l_i_blocks_high; /* were l_i_reserved1 */
665  __le16 l_i_uid_high; /* these 2 fields */
666  __le16 l_i_gid_high; /* were reserved2[0] */
667  __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
669  } linux2;
670  struct {
671  __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
676  } hurd2;
677  struct {
678  __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
681  } masix2;
682  } osd2; /* OS dependent 2 */
684  __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
685  __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
686  __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
687  __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
688  __le32 i_crtime; /* File Creation time */
689  __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
690  __le32 i_version_hi; /* high 32 bits for 64-bit version */
691 };
692 
693 struct move_extent {
694  __u32 reserved; /* should be zero */
695  __u32 donor_fd; /* donor file descriptor */
696  __u64 orig_start; /* logical start offset in block for orig */
697  __u64 donor_start; /* logical start offset in block for donor */
698  __u64 len; /* block length to be moved */
699  __u64 moved_len; /* moved block length */
700 };
701 
702 #define EXT4_EPOCH_BITS 2
703 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
704 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
705 
706 /*
707  * Extended fields will fit into an inode if the filesystem was formatted
708  * with large inodes (-I 256 or larger) and there are not currently any EAs
709  * consuming all of the available space. For new inodes we always reserve
710  * enough space for the kernel's known extended fields, but for inodes
711  * created with an old kernel this might not have been the case. None of
712  * the extended inode fields is critical for correct filesystem operation.
713  * This macro checks if a certain field fits in the inode. Note that
714  * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
715  */
716 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
717  ((offsetof(typeof(*ext4_inode), field) + \
718  sizeof((ext4_inode)->field)) \
719  <= (EXT4_GOOD_OLD_INODE_SIZE + \
720  (einode)->i_extra_isize)) \
721 
722 static inline __le32 ext4_encode_extra_time(struct timespec *time)
723 {
724  return cpu_to_le32((sizeof(time->tv_sec) > 4 ?
725  (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) |
726  ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK));
727 }
728 
729 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
730 {
731  if (sizeof(time->tv_sec) > 4)
732  time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK)
733  << 32;
734  time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
735 }
736 
737 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
738 do { \
739  (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
740  if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
741  (raw_inode)->xtime ## _extra = \
742  ext4_encode_extra_time(&(inode)->xtime); \
743 } while (0)
744 
745 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
746 do { \
747  if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
748  (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
749  if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
750  (raw_inode)->xtime ## _extra = \
751  ext4_encode_extra_time(&(einode)->xtime); \
752 } while (0)
753 
754 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
755 do { \
756  (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
757  if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
758  ext4_decode_extra_time(&(inode)->xtime, \
759  raw_inode->xtime ## _extra); \
760  else \
761  (inode)->xtime.tv_nsec = 0; \
762 } while (0)
763 
764 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
765 do { \
766  if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
767  (einode)->xtime.tv_sec = \
768  (signed)le32_to_cpu((raw_inode)->xtime); \
769  if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
770  ext4_decode_extra_time(&(einode)->xtime, \
771  raw_inode->xtime ## _extra); \
772  else \
773  (einode)->xtime.tv_nsec = 0; \
774 } while (0)
775 
776 #define i_disk_version osd1.linux1.l_i_version
777 
778 #if defined(__KERNEL__) || defined(__linux__)
779 #define i_reserved1 osd1.linux1.l_i_reserved1
780 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
781 #define i_blocks_high osd2.linux2.l_i_blocks_high
782 #define i_uid_low i_uid
783 #define i_gid_low i_gid
784 #define i_uid_high osd2.linux2.l_i_uid_high
785 #define i_gid_high osd2.linux2.l_i_gid_high
786 #define i_checksum_lo osd2.linux2.l_i_checksum_lo
787 
788 #elif defined(__GNU__)
789 
790 #define i_translator osd1.hurd1.h_i_translator
791 #define i_uid_high osd2.hurd2.h_i_uid_high
792 #define i_gid_high osd2.hurd2.h_i_gid_high
793 #define i_author osd2.hurd2.h_i_author
794 
795 #elif defined(__masix__)
796 
797 #define i_reserved1 osd1.masix1.m_i_reserved1
798 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
799 #define i_reserved2 osd2.masix2.m_i_reserved2
800 
801 #endif /* defined(__KERNEL__) || defined(__linux__) */
802 
803 /*
804  * storage for cached extent
805  * If ec_len == 0, then the cache is invalid.
806  * If ec_start == 0, then the cache represents a gap (null mapping)
807  */
811  __u32 ec_len; /* must be 32bit to return holes */
812 };
813 
814 /*
815  * fourth extended file system inode data in memory
816  */
818  __le32 i_data[15]; /* unconverted */
821 
822  /*
823  * i_block_group is the number of the block group which contains
824  * this file's inode. Constant across the lifetime of the inode,
825  * it is ued for making block allocation decisions - we try to
826  * place a file's data blocks near its inode block, and new inodes
827  * near to their parent directory's inode.
828  */
831 #if (BITS_PER_LONG < 64)
832  unsigned long i_state_flags; /* Dynamic state flags */
833 #endif
834  unsigned long i_flags;
835 
836 #ifdef CONFIG_EXT4_FS_XATTR
837  /*
838  * Extended attributes can be read independently of the main file
839  * data. Taking i_mutex even when reading would cause contention
840  * between readers of EAs and writers of regular file data, so
841  * instead we synchronize on xattr_sem when reading or changing
842  * EAs.
843  */
844  struct rw_semaphore xattr_sem;
845 #endif
846 
847  struct list_head i_orphan; /* unlinked but open inodes */
848 
849  /*
850  * i_disksize keeps track of what the inode size is ON DISK, not
851  * in memory. During truncate, i_size is set to the new size by
852  * the VFS prior to calling ext4_truncate(), but the filesystem won't
853  * set i_disksize to 0 until the truncate is actually under way.
854  *
855  * The intent is that i_disksize always represents the blocks which
856  * are used by this file. This allows recovery to restart truncate
857  * on orphans if we crash during truncate. We actually write i_disksize
858  * into the on-disk inode when writing inodes out, instead of i_size.
859  *
860  * The only time when i_disksize and i_size may be different is when
861  * a truncate is in progress. The only things which change i_disksize
862  * are ext4_get_block (growth) and ext4_truncate (shrinkth).
863  */
864  loff_t i_disksize;
865 
866  /*
867  * i_data_sem is for serialising ext4_truncate() against
868  * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
869  * data tree are chopped off during truncate. We can't do that in
870  * ext4 because whenever we perform intermediate commits during
871  * truncate, the inode and all the metadata blocks *must* be in a
872  * consistent state which allows truncation of the orphans to restart
873  * during recovery. Hence we must fix the get_block-vs-truncate race
874  * by other means, so we have i_data_sem.
875  */
877  struct inode vfs_inode;
878  struct jbd2_inode *jinode;
879 
881  /*
882  * File creation time. Its function is same as that of
883  * struct timespec i_{a,c,m}time in the generic inode.
884  */
886 
887  /* mballoc */
890 
891  /* ialloc */
893 
894  /* allocation reservation info for delalloc */
895  /* In case of bigalloc, these refer to clusters rather than blocks */
901 
902  /* on-disk additional length */
904 
905 #ifdef CONFIG_QUOTA
906  /* quota space reservation, managed internally by quota code */
907  qsize_t i_reserved_quota;
908 #endif
909 
910  /* completed IOs that might need unwritten extents handling */
913  atomic_t i_ioend_count; /* Number of outstanding io_end structs */
914  atomic_t i_unwritten; /* Nr. of inflight conversions pending */
915 
917 
918  /*
919  * Transactions that contain inode's metadata needed to complete
920  * fsync and fdatasync, respectively.
921  */
924 
925  /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
927 };
928 
929 /*
930  * File system states
931  */
932 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
933 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
934 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
935 
936 /*
937  * Misc. filesystem flags
938  */
939 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
940 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
941 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
942 
943 /*
944  * Mount flags
945  */
946 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
947 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
948 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
949 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
950 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
951 #define EXT4_MOUNT_ERRORS_MASK 0x00070
952 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
953 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
954 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
955 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
956 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
957 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
958 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
959 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
960 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
961 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
962 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
963 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
964 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
965 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
966 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
967 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
968 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
969 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
970 #define EXT4_MOUNT_MBLK_IO_SUBMIT 0x4000000 /* multi-block io submits */
971 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
972 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
973 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
974 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
975 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
976 
977 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
978  specified delalloc */
980 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
981  ~EXT4_MOUNT_##opt
982 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
983  EXT4_MOUNT_##opt
984 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
985  EXT4_MOUNT_##opt)
987 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
988  ~EXT4_MOUNT2_##opt
989 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
990  EXT4_MOUNT2_##opt
991 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
992  EXT4_MOUNT2_##opt)
994 #define ext4_test_and_set_bit __test_and_set_bit_le
995 #define ext4_set_bit __set_bit_le
996 #define ext4_set_bit_atomic ext2_set_bit_atomic
997 #define ext4_test_and_clear_bit __test_and_clear_bit_le
998 #define ext4_clear_bit __clear_bit_le
999 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
1000 #define ext4_test_bit test_bit_le
1001 #define ext4_find_next_zero_bit find_next_zero_bit_le
1002 #define ext4_find_next_bit find_next_bit_le
1003 
1004 extern void ext4_set_bits(void *bm, int cur, int len);
1005 
1006 /*
1007  * Maximal mount counts between two filesystem checks
1008  */
1009 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1010 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1011 
1012 /*
1013  * Behaviour when detecting errors
1014  */
1015 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1016 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1017 #define EXT4_ERRORS_PANIC 3 /* Panic */
1018 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1019 
1020 /* Metadata checksum algorithm codes */
1021 #define EXT4_CRC32C_CHKSUM 1
1022 
1023 /*
1024  * Structure of the super block
1025  */
1027 /*00*/ __le32 s_inodes_count; /* Inodes count */
1028  __le32 s_blocks_count_lo; /* Blocks count */
1029  __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1030  __le32 s_free_blocks_count_lo; /* Free blocks count */
1031 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1032  __le32 s_first_data_block; /* First Data Block */
1033  __le32 s_log_block_size; /* Block size */
1034  __le32 s_log_cluster_size; /* Allocation cluster size */
1035 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1036  __le32 s_clusters_per_group; /* # Clusters per group */
1037  __le32 s_inodes_per_group; /* # Inodes per group */
1038  __le32 s_mtime; /* Mount time */
1039 /*30*/ __le32 s_wtime; /* Write time */
1040  __le16 s_mnt_count; /* Mount count */
1041  __le16 s_max_mnt_count; /* Maximal mount count */
1042  __le16 s_magic; /* Magic signature */
1043  __le16 s_state; /* File system state */
1044  __le16 s_errors; /* Behaviour when detecting errors */
1045  __le16 s_minor_rev_level; /* minor revision level */
1046 /*40*/ __le32 s_lastcheck; /* time of last check */
1047  __le32 s_checkinterval; /* max. time between checks */
1049  __le32 s_rev_level; /* Revision level */
1050 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1051  __le16 s_def_resgid; /* Default gid for reserved blocks */
1052  /*
1053  * These fields are for EXT4_DYNAMIC_REV superblocks only.
1054  *
1055  * Note: the difference between the compatible feature set and
1056  * the incompatible feature set is that if there is a bit set
1057  * in the incompatible feature set that the kernel doesn't
1058  * know about, it should refuse to mount the filesystem.
1059  *
1060  * e2fsck's requirements are more strict; if it doesn't know
1061  * about a feature in either the compatible or incompatible
1062  * feature set, it must abort and not try to meddle with
1063  * things it doesn't understand...
1064  */
1065  __le32 s_first_ino; /* First non-reserved inode */
1066  __le16 s_inode_size; /* size of inode structure */
1067  __le16 s_block_group_nr; /* block group # of this superblock */
1068  __le32 s_feature_compat; /* compatible feature set */
1069 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1070  __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1071 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1072 /*78*/ char s_volume_name[16]; /* volume name */
1073 /*88*/ char s_last_mounted[64]; /* directory where last mounted */
1074 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1075  /*
1076  * Performance hints. Directory preallocation should only
1077  * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1078  */
1079  __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1080  __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1081  __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1082  /*
1083  * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1084  */
1085 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1086 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1087  __le32 s_journal_dev; /* device number of journal file */
1088  __le32 s_last_orphan; /* start of list of inodes to delete */
1089  __le32 s_hash_seed[4]; /* HTREE hash seed */
1090  __u8 s_def_hash_version; /* Default hash version to use */
1092  __le16 s_desc_size; /* size of group descriptor */
1094  __le32 s_first_meta_bg; /* First metablock block group */
1095  __le32 s_mkfs_time; /* When the filesystem was created */
1096  __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1097  /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1098 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1099  __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1100  __le32 s_free_blocks_count_hi; /* Free blocks count */
1101  __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1102  __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1103  __le32 s_flags; /* Miscellaneous flags */
1104  __le16 s_raid_stride; /* RAID stride */
1105  __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1106  __le64 s_mmp_block; /* Block for multi-mount protection */
1107  __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1108  __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1109  __u8 s_checksum_type; /* metadata checksum algorithm used */
1111  __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1112  __le32 s_snapshot_inum; /* Inode number of active snapshot */
1113  __le32 s_snapshot_id; /* sequential ID of active snapshot */
1114  __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1115  snapshot's future use */
1116  __le32 s_snapshot_list; /* inode number of the head of the
1117  on-disk snapshot list */
1118 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1119  __le32 s_error_count; /* number of fs errors */
1120  __le32 s_first_error_time; /* first time an error happened */
1121  __le32 s_first_error_ino; /* inode involved in first error */
1122  __le64 s_first_error_block; /* block involved of first error */
1123  __u8 s_first_error_func[32]; /* function where the error happened */
1124  __le32 s_first_error_line; /* line number where error happened */
1125  __le32 s_last_error_time; /* most recent time of an error */
1126  __le32 s_last_error_ino; /* inode involved in last error */
1127  __le32 s_last_error_line; /* line number where error happened */
1128  __le64 s_last_error_block; /* block involved of last error */
1129  __u8 s_last_error_func[32]; /* function where the error happened */
1130 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1132  __le32 s_usr_quota_inum; /* inode for tracking user quota */
1133  __le32 s_grp_quota_inum; /* inode for tracking group quota */
1134  __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1135  __le32 s_reserved[108]; /* Padding to the end of the block */
1136  __le32 s_checksum; /* crc32c(superblock) */
1137 };
1139 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1140 
1141 #ifdef __KERNEL__
1142 
1143 /*
1144  * run-time mount flags
1145  */
1146 #define EXT4_MF_MNTDIR_SAMPLED 0x0001
1147 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1148 
1149 /*
1150  * fourth extended-fs super-block data in memory
1151  */
1152 struct ext4_sb_info {
1153  unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1154  unsigned long s_inodes_per_block;/* Number of inodes per block */
1155  unsigned long s_blocks_per_group;/* Number of blocks in a group */
1156  unsigned long s_clusters_per_group; /* Number of clusters in a group */
1157  unsigned long s_inodes_per_group;/* Number of inodes in a group */
1158  unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1159  unsigned long s_gdb_count; /* Number of group descriptor blocks */
1160  unsigned long s_desc_per_block; /* Number of group descriptors per block */
1161  ext4_group_t s_groups_count; /* Number of groups in the fs */
1162  ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1163  unsigned long s_overhead; /* # of fs overhead clusters */
1164  unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1165  unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1166  loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1167  struct buffer_head * s_sbh; /* Buffer containing the super block */
1168  struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1169  struct buffer_head **s_group_desc;
1170  unsigned int s_mount_opt;
1171  unsigned int s_mount_opt2;
1172  unsigned int s_mount_flags;
1173  unsigned int s_def_mount_opt;
1174  ext4_fsblk_t s_sb_block;
1175  kuid_t s_resuid;
1176  kgid_t s_resgid;
1177  unsigned short s_mount_state;
1178  unsigned short s_pad;
1179  int s_addr_per_block_bits;
1180  int s_desc_per_block_bits;
1181  int s_inode_size;
1182  int s_first_ino;
1183  unsigned int s_inode_readahead_blks;
1184  unsigned int s_inode_goal;
1185  spinlock_t s_next_gen_lock;
1186  u32 s_next_generation;
1187  u32 s_hash_seed[4];
1188  int s_def_hash_version;
1189  int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1190  struct percpu_counter s_freeclusters_counter;
1191  struct percpu_counter s_freeinodes_counter;
1192  struct percpu_counter s_dirs_counter;
1193  struct percpu_counter s_dirtyclusters_counter;
1194  struct blockgroup_lock *s_blockgroup_lock;
1195  struct proc_dir_entry *s_proc;
1196  struct kobject s_kobj;
1197  struct completion s_kobj_unregister;
1198  struct super_block *s_sb;
1199 
1200  /* Journaling */
1201  struct journal_s *s_journal;
1202  struct list_head s_orphan;
1203  struct mutex s_orphan_lock;
1204  unsigned long s_resize_flags; /* Flags indicating if there
1205  is a resizer */
1206  unsigned long s_commit_interval;
1207  u32 s_max_batch_time;
1208  u32 s_min_batch_time;
1209  struct block_device *journal_bdev;
1210 #ifdef CONFIG_QUOTA
1211  char *s_qf_names[MAXQUOTAS]; /* Names of quota files with journalled quota */
1212  int s_jquota_fmt; /* Format of quota to use */
1213 #endif
1214  unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1215  struct rb_root system_blks;
1216 
1217 #ifdef EXTENTS_STATS
1218  /* ext4 extents stats */
1219  unsigned long s_ext_min;
1220  unsigned long s_ext_max;
1221  unsigned long s_depth_max;
1222  spinlock_t s_ext_stats_lock;
1223  unsigned long s_ext_blocks;
1224  unsigned long s_ext_extents;
1225 #endif
1226 
1227  /* for buddy allocator */
1228  struct ext4_group_info ***s_group_info;
1229  struct inode *s_buddy_cache;
1230  spinlock_t s_md_lock;
1231  unsigned short *s_mb_offsets;
1232  unsigned int *s_mb_maxs;
1233  unsigned int s_group_info_size;
1234 
1235  /* tunables */
1236  unsigned long s_stripe;
1237  unsigned int s_mb_stream_request;
1238  unsigned int s_mb_max_to_scan;
1239  unsigned int s_mb_min_to_scan;
1240  unsigned int s_mb_stats;
1241  unsigned int s_mb_order2_reqs;
1242  unsigned int s_mb_group_prealloc;
1243  unsigned int s_max_writeback_mb_bump;
1244  unsigned int s_max_dir_size_kb;
1245  /* where last allocation was done - for stream allocation */
1246  unsigned long s_mb_last_group;
1247  unsigned long s_mb_last_start;
1248 
1249  /* stats for buddy allocator */
1250  atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1251  atomic_t s_bal_success; /* we found long enough chunks */
1252  atomic_t s_bal_allocated; /* in blocks */
1253  atomic_t s_bal_ex_scanned; /* total extents scanned */
1254  atomic_t s_bal_goals; /* goal hits */
1255  atomic_t s_bal_breaks; /* too long searches */
1256  atomic_t s_bal_2orders; /* 2^order hits */
1257  spinlock_t s_bal_lock;
1258  unsigned long s_mb_buddies_generated;
1259  unsigned long long s_mb_generation_time;
1260  atomic_t s_mb_lost_chunks;
1261  atomic_t s_mb_preallocated;
1262  atomic_t s_mb_discarded;
1263  atomic_t s_lock_busy;
1264 
1265  /* locality groups */
1266  struct ext4_locality_group __percpu *s_locality_groups;
1267 
1268  /* for write statistics */
1269  unsigned long s_sectors_written_start;
1270  u64 s_kbytes_written;
1271 
1272  /* the size of zero-out chunk */
1273  unsigned int s_extent_max_zeroout_kb;
1274 
1275  unsigned int s_log_groups_per_flex;
1276  struct flex_groups *s_flex_groups;
1277  ext4_group_t s_flex_groups_allocated;
1278 
1279  /* workqueue for dio unwritten */
1280  struct workqueue_struct *dio_unwritten_wq;
1281 
1282  /* timer for periodic error stats printing */
1283  struct timer_list s_err_report;
1284 
1285  /* Lazy inode table initialization info */
1286  struct ext4_li_request *s_li_request;
1287  /* Wait multiplier for lazy initialization thread */
1288  unsigned int s_li_wait_mult;
1289 
1290  /* Kernel thread for multiple mount protection */
1291  struct task_struct *s_mmp_tsk;
1292 
1293  /* record the last minlen when FITRIM is called. */
1294  atomic_t s_last_trim_minblks;
1295 
1296  /* Reference to checksum algorithm driver via cryptoapi */
1297  struct crypto_shash *s_chksum_driver;
1298 
1299  /* Precomputed FS UUID checksum for seeding other checksums */
1300  __u32 s_csum_seed;
1301 };
1302 
1303 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1304 {
1305  return sb->s_fs_info;
1306 }
1307 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1308 {
1309  return container_of(inode, struct ext4_inode_info, vfs_inode);
1310 }
1311 
1312 static inline struct timespec ext4_current_time(struct inode *inode)
1313 {
1314  return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1316 }
1317 
1318 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1319 {
1320  return ino == EXT4_ROOT_INO ||
1321  ino == EXT4_USR_QUOTA_INO ||
1322  ino == EXT4_GRP_QUOTA_INO ||
1323  ino == EXT4_JOURNAL_INO ||
1324  ino == EXT4_RESIZE_INO ||
1325  (ino >= EXT4_FIRST_INO(sb) &&
1326  ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1327 }
1328 
1329 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
1330  struct ext4_io_end *io_end)
1331 {
1332  if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
1333  io_end->flag |= EXT4_IO_END_UNWRITTEN;
1334  atomic_inc(&EXT4_I(inode)->i_unwritten);
1335  }
1336 }
1337 
1338 static inline ext4_io_end_t *ext4_inode_aio(struct inode *inode)
1339 {
1340  return inode->i_private;
1341 }
1342 
1343 static inline void ext4_inode_aio_set(struct inode *inode, ext4_io_end_t *io)
1344 {
1345  inode->i_private = io;
1346 }
1347 
1348 /*
1349  * Inode dynamic state flags
1350  */
1351 enum {
1352  EXT4_STATE_JDATA, /* journaled data exists */
1353  EXT4_STATE_NEW, /* inode is newly created */
1354  EXT4_STATE_XATTR, /* has in-inode xattrs */
1355  EXT4_STATE_NO_EXPAND, /* No space for expansion */
1356  EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1357  EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1358  EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
1359  EXT4_STATE_NEWENTRY, /* File just added to dir */
1360  EXT4_STATE_DELALLOC_RESERVED, /* blks already reserved for delalloc */
1361  EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read
1362  nolocking */
1363 };
1364 
1365 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1366 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1367 { \
1368  return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1369 } \
1370 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1371 { \
1372  set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1373 } \
1374 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1375 { \
1376  clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1377 }
1378 
1379 EXT4_INODE_BIT_FNS(flag, flags, 0)
1380 #if (BITS_PER_LONG < 64)
1381 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1382 
1383 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1384 {
1385  (ei)->i_state_flags = 0;
1386 }
1387 #else
1388 EXT4_INODE_BIT_FNS(state, flags, 32)
1389 
1390 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1391 {
1392  /* We depend on the fact that callers will set i_flags */
1393 }
1394 #endif
1395 #else
1396 /* Assume that user mode programs are passing in an ext4fs superblock, not
1397  * a kernel struct super_block. This will allow us to call the feature-test
1398  * macros from user land. */
1399 #define EXT4_SB(sb) (sb)
1400 #endif
1402 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1403 
1404 /*
1405  * Codes for operating systems
1406  */
1407 #define EXT4_OS_LINUX 0
1408 #define EXT4_OS_HURD 1
1409 #define EXT4_OS_MASIX 2
1410 #define EXT4_OS_FREEBSD 3
1411 #define EXT4_OS_LITES 4
1412 
1413 /*
1414  * Revision levels
1415  */
1416 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1417 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1419 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1420 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1422 #define EXT4_GOOD_OLD_INODE_SIZE 128
1423 
1424 /*
1425  * Feature set definitions
1426  */
1428 #define EXT4_HAS_COMPAT_FEATURE(sb,mask) \
1429  ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1430 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \
1431  ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1432 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \
1433  ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1434 #define EXT4_SET_COMPAT_FEATURE(sb,mask) \
1435  EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1436 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \
1437  EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1438 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \
1439  EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1440 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \
1441  EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1442 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
1443  EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1444 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \
1445  EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1447 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1448 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1449 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1450 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1451 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1452 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1454 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1455 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1456 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1457 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1458 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1459 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1460 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1461 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1462 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1463 /*
1464  * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1465  * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1466  * all other data structures' checksums. However, the METADATA_CSUM and
1467  * GDT_CSUM bits are mutually exclusive.
1468  */
1469 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
1471 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1472 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1473 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1474 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1475 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1476 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1477 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1478 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1479 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1480 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1481 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1482 #define EXT4_FEATURE_INCOMPAT_BG_USE_META_CSUM 0x2000 /* use crc32c for bg */
1483 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
1484 #define EXT4_FEATURE_INCOMPAT_INLINEDATA 0x8000 /* data in inode */
1486 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1487 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1488  EXT4_FEATURE_INCOMPAT_META_BG)
1489 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1490  EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1491  EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1493 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1494 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1495  EXT4_FEATURE_INCOMPAT_RECOVER| \
1496  EXT4_FEATURE_INCOMPAT_META_BG)
1497 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1498  EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1499  EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1501 #define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR
1502 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1503  EXT4_FEATURE_INCOMPAT_RECOVER| \
1504  EXT4_FEATURE_INCOMPAT_META_BG| \
1505  EXT4_FEATURE_INCOMPAT_EXTENTS| \
1506  EXT4_FEATURE_INCOMPAT_64BIT| \
1507  EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1508  EXT4_FEATURE_INCOMPAT_MMP)
1509 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1510  EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1511  EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1512  EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1513  EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1514  EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1515  EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1516  EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1517  EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1518  EXT4_FEATURE_RO_COMPAT_QUOTA)
1519 
1520 /*
1521  * Default values for user and/or group using reserved blocks
1522  */
1523 #define EXT4_DEF_RESUID 0
1524 #define EXT4_DEF_RESGID 0
1526 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
1527 
1528 /*
1529  * Default mount options
1530  */
1531 #define EXT4_DEFM_DEBUG 0x0001
1532 #define EXT4_DEFM_BSDGROUPS 0x0002
1533 #define EXT4_DEFM_XATTR_USER 0x0004
1534 #define EXT4_DEFM_ACL 0x0008
1535 #define EXT4_DEFM_UID16 0x0010
1536 #define EXT4_DEFM_JMODE 0x0060
1537 #define EXT4_DEFM_JMODE_DATA 0x0020
1538 #define EXT4_DEFM_JMODE_ORDERED 0x0040
1539 #define EXT4_DEFM_JMODE_WBACK 0x0060
1540 #define EXT4_DEFM_NOBARRIER 0x0100
1541 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1542 #define EXT4_DEFM_DISCARD 0x0400
1543 #define EXT4_DEFM_NODELALLOC 0x0800
1544 
1545 /*
1546  * Default journal batch times
1547  */
1548 #define EXT4_DEF_MIN_BATCH_TIME 0
1549 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1550 
1551 /*
1552  * Minimum number of groups in a flexgroup before we separate out
1553  * directories into the first block group of a flexgroup
1554  */
1555 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1556 
1557 /*
1558  * Structure of a directory entry
1559  */
1560 #define EXT4_NAME_LEN 255
1563  __le32 inode; /* Inode number */
1564  __le16 rec_len; /* Directory entry length */
1565  __le16 name_len; /* Name length */
1566  char name[EXT4_NAME_LEN]; /* File name */
1567 };
1568 
1569 /*
1570  * The new version of the directory entry. Since EXT4 structures are
1571  * stored in intel byte order, and the name_len field could never be
1572  * bigger than 255 chars, it's safe to reclaim the extra byte for the
1573  * file_type field.
1574  */
1576  __le32 inode; /* Inode number */
1577  __le16 rec_len; /* Directory entry length */
1578  __u8 name_len; /* Name length */
1580  char name[EXT4_NAME_LEN]; /* File name */
1581 };
1582 
1583 /*
1584  * This is a bogus directory entry at the end of each leaf block that
1585  * records checksums.
1586  */
1588  __le32 det_reserved_zero1; /* Pretend to be unused */
1590  __u8 det_reserved_zero2; /* Zero name length */
1591  __u8 det_reserved_ft; /* 0xDE, fake file type */
1592  __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
1593 };
1594 
1595 /*
1596  * Ext4 directory file types. Only the low 3 bits are used. The
1597  * other bits are reserved for now.
1598  */
1599 #define EXT4_FT_UNKNOWN 0
1600 #define EXT4_FT_REG_FILE 1
1601 #define EXT4_FT_DIR 2
1602 #define EXT4_FT_CHRDEV 3
1603 #define EXT4_FT_BLKDEV 4
1604 #define EXT4_FT_FIFO 5
1605 #define EXT4_FT_SOCK 6
1606 #define EXT4_FT_SYMLINK 7
1608 #define EXT4_FT_MAX 8
1610 #define EXT4_FT_DIR_CSUM 0xDE
1611 
1612 /*
1613  * EXT4_DIR_PAD defines the directory entries boundaries
1614  *
1615  * NOTE: It must be a multiple of 4
1616  */
1617 #define EXT4_DIR_PAD 4
1618 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1619 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1620  ~EXT4_DIR_ROUND)
1621 #define EXT4_MAX_REC_LEN ((1<<16)-1)
1622 
1623 /*
1624  * If we ever get support for fs block sizes > page_size, we'll need
1625  * to remove the #if statements in the next two functions...
1626  */
1627 static inline unsigned int
1628 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1629 {
1630  unsigned len = le16_to_cpu(dlen);
1631 
1632 #if (PAGE_CACHE_SIZE >= 65536)
1633  if (len == EXT4_MAX_REC_LEN || len == 0)
1634  return blocksize;
1635  return (len & 65532) | ((len & 3) << 16);
1636 #else
1637  return len;
1638 #endif
1639 }
1640 
1641 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1642 {
1643  if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1644  BUG();
1645 #if (PAGE_CACHE_SIZE >= 65536)
1646  if (len < 65536)
1647  return cpu_to_le16(len);
1648  if (len == blocksize) {
1649  if (blocksize == 65536)
1650  return cpu_to_le16(EXT4_MAX_REC_LEN);
1651  else
1652  return cpu_to_le16(0);
1653  }
1654  return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1655 #else
1656  return cpu_to_le16(len);
1657 #endif
1658 }
1659 
1660 /*
1661  * Hash Tree Directory indexing
1662  * (c) Daniel Phillips, 2001
1663  */
1665 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1666  EXT4_FEATURE_COMPAT_DIR_INDEX) && \
1667  ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
1668 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1669 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1670 
1671 /* Legal values for the dx_root hash_version field: */
1673 #define DX_HASH_LEGACY 0
1674 #define DX_HASH_HALF_MD4 1
1675 #define DX_HASH_TEA 2
1676 #define DX_HASH_LEGACY_UNSIGNED 3
1677 #define DX_HASH_HALF_MD4_UNSIGNED 4
1678 #define DX_HASH_TEA_UNSIGNED 5
1679 
1680 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
1681  const void *address, unsigned int length)
1682 {
1683  struct {
1684  struct shash_desc shash;
1685  char ctx[4];
1686  } desc;
1687  int err;
1688 
1689  BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
1690 
1691  desc.shash.tfm = sbi->s_chksum_driver;
1692  desc.shash.flags = 0;
1693  *(u32 *)desc.ctx = crc;
1694 
1695  err = crypto_shash_update(&desc.shash, address, length);
1696  BUG_ON(err);
1697 
1698  return *(u32 *)desc.ctx;
1699 }
1700 
1701 #ifdef __KERNEL__
1702 
1703 /* hash info structure used by the directory hash */
1704 struct dx_hash_info
1705 {
1706  u32 hash;
1707  u32 minor_hash;
1708  int hash_version;
1709  u32 *seed;
1710 };
1711 
1712 
1713 /* 32 and 64 bit signed EOF for dx directories */
1714 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
1715 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
1716 
1717 
1718 /*
1719  * Control parameters used by ext4_htree_next_block
1720  */
1721 #define HASH_NB_ALWAYS 1
1722 
1723 
1724 /*
1725  * Describe an inode's exact location on disk and in memory
1726  */
1727 struct ext4_iloc
1728 {
1729  struct buffer_head *bh;
1730  unsigned long offset;
1732 };
1733 
1734 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
1735 {
1736  return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
1737 }
1738 
1739 /*
1740  * This structure is stuffed into the struct file's private_data field
1741  * for directories. It is where we put information so that we can do
1742  * readdir operations in hash tree order.
1743  */
1744 struct dir_private_info {
1745  struct rb_root root;
1746  struct rb_node *curr_node;
1747  struct fname *extra_fname;
1748  loff_t last_pos;
1749  __u32 curr_hash;
1751  __u32 next_hash;
1752 };
1753 
1754 /* calculate the first block number of the group */
1755 static inline ext4_fsblk_t
1756 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
1757 {
1758  return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
1759  le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
1760 }
1761 
1762 /*
1763  * Special error return code only used by dx_probe() and its callers.
1764  */
1765 #define ERR_BAD_DX_DIR -75000
1766 
1768  ext4_group_t *blockgrpp, ext4_grpblk_t *offsetp);
1769 
1770 /*
1771  * Timeout and state flag for lazy initialization inode thread.
1772  */
1773 #define EXT4_DEF_LI_WAIT_MULT 10
1774 #define EXT4_DEF_LI_MAX_START_DELAY 5
1775 #define EXT4_LAZYINIT_QUIT 0x0001
1776 #define EXT4_LAZYINIT_RUNNING 0x0002
1777 
1778 /*
1779  * Lazy inode table initialization info
1780  */
1781 struct ext4_lazy_init {
1782  unsigned long li_state;
1783  struct list_head li_request_list;
1784  struct mutex li_list_mtx;
1785 };
1786 
1787 struct ext4_li_request {
1788  struct super_block *lr_super;
1789  struct ext4_sb_info *lr_sbi;
1790  ext4_group_t lr_next_group;
1791  struct list_head lr_request;
1792  unsigned long lr_next_sched;
1793  unsigned long lr_timeout;
1794 };
1795 
1796 struct ext4_features {
1797  struct kobject f_kobj;
1798  struct completion f_kobj_unregister;
1799 };
1800 
1801 /*
1802  * This structure will be used for multiple mount protection. It will be
1803  * written into the block number saved in the s_mmp_block field in the
1804  * superblock. Programs that check MMP should assume that if
1805  * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
1806  * to use the filesystem, regardless of how old the timestamp is.
1807  */
1808 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
1809 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
1810 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
1811 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
1812 
1813 struct mmp_struct {
1814  __le32 mmp_magic; /* Magic number for MMP */
1815  __le32 mmp_seq; /* Sequence no. updated periodically */
1816 
1817  /*
1818  * mmp_time, mmp_nodename & mmp_bdevname are only used for information
1819  * purposes and do not affect the correctness of the algorithm
1820  */
1821  __le64 mmp_time; /* Time last updated */
1822  char mmp_nodename[64]; /* Node which last updated MMP block */
1823  char mmp_bdevname[32]; /* Bdev which last updated MMP block */
1824 
1825  /*
1826  * mmp_check_interval is used to verify if the MMP block has been
1827  * updated on the block device. The value is updated based on the
1828  * maximum time to write the MMP block during an update cycle.
1829  */
1830  __le16 mmp_check_interval;
1831 
1832  __le16 mmp_pad1;
1833  __le32 mmp_pad2[226];
1834  __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
1835 };
1836 
1837 /* arguments passed to the mmp thread */
1838 struct mmpd_data {
1839  struct buffer_head *bh; /* bh from initial read_mmp_block() */
1840  struct super_block *sb; /* super block of the fs */
1841 };
1842 
1843 /*
1844  * Check interval multiplier
1845  * The MMP block is written every update interval and initially checked every
1846  * update interval x the multiplier (the value is then adapted based on the
1847  * write latency). The reason is that writes can be delayed under load and we
1848  * don't want readers to incorrectly assume that the filesystem is no longer
1849  * in use.
1850  */
1851 #define EXT4_MMP_CHECK_MULT 2UL
1852 
1853 /*
1854  * Minimum interval for MMP checking in seconds.
1855  */
1856 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
1857 
1858 /*
1859  * Maximum interval for MMP checking in seconds.
1860  */
1861 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
1862 
1863 /*
1864  * Function prototypes
1865  */
1866 
1867 /*
1868  * Ok, these declarations are also in <linux/kernel.h> but none of the
1869  * ext4 source programs needs to include it so they are duplicated here.
1870  */
1871 # define NORET_TYPE
1872 # define ATTRIB_NORET __attribute__((noreturn))
1873 # define NORET_AND noreturn,
1874 
1875 /* bitmap.c */
1876 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
1878  struct ext4_group_desc *gdp,
1879  struct buffer_head *bh, int sz);
1881  struct ext4_group_desc *gdp,
1882  struct buffer_head *bh, int sz);
1884  struct ext4_group_desc *gdp,
1885  struct buffer_head *bh);
1887  struct ext4_group_desc *gdp,
1888  struct buffer_head *bh);
1889 
1890 /* balloc.c */
1891 extern void ext4_validate_block_bitmap(struct super_block *sb,
1892  struct ext4_group_desc *desc,
1893  unsigned int block_group,
1894  struct buffer_head *bh);
1895 extern unsigned int ext4_block_group(struct super_block *sb,
1897 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
1899 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
1900 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
1902 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
1903  ext4_fsblk_t goal,
1904  unsigned int flags,
1905  unsigned long *count,
1906  int *errp);
1907 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
1908  s64 nclusters, unsigned int flags);
1910 extern void ext4_check_blocks_bitmap(struct super_block *);
1911 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
1913  struct buffer_head ** bh);
1914 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
1915 
1916 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
1918 extern int ext4_wait_block_bitmap(struct super_block *sb,
1920  struct buffer_head *bh);
1921 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
1923 extern void ext4_init_block_bitmap(struct super_block *sb,
1924  struct buffer_head *bh,
1926  struct ext4_group_desc *desc);
1927 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
1929  struct ext4_group_desc *gdp);
1930 extern unsigned ext4_num_overhead_clusters(struct super_block *sb,
1932  struct ext4_group_desc *gdp);
1933 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
1934 
1935 /* dir.c */
1936 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
1937  struct file *,
1938  struct ext4_dir_entry_2 *,
1939  struct buffer_head *, unsigned int);
1940 #define ext4_check_dir_entry(dir, filp, de, bh, offset) \
1941  unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
1942  (de), (bh), (offset)))
1943 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
1944  __u32 minor_hash,
1945  struct ext4_dir_entry_2 *dirent);
1946 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
1947 
1948 /* fsync.c */
1949 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
1950 extern int ext4_flush_unwritten_io(struct inode *);
1951 
1952 /* hash.c */
1953 extern int ext4fs_dirhash(const char *name, int len, struct
1954  dx_hash_info *hinfo);
1955 
1956 /* ialloc.c */
1957 extern struct inode *ext4_new_inode(handle_t *, struct inode *, umode_t,
1958  const struct qstr *qstr, __u32 goal,
1959  uid_t *owner);
1960 extern void ext4_free_inode(handle_t *, struct inode *);
1961 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
1962 extern unsigned long ext4_count_free_inodes(struct super_block *);
1963 extern unsigned long ext4_count_dirs(struct super_block *);
1964 extern void ext4_check_inodes_bitmap(struct super_block *);
1965 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
1966 extern int ext4_init_inode_table(struct super_block *sb,
1967  ext4_group_t group, int barrier);
1968 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
1969 
1970 /* mballoc.c */
1971 extern long ext4_mb_stats;
1972 extern long ext4_mb_max_to_scan;
1973 extern int ext4_mb_init(struct super_block *);
1974 extern int ext4_mb_release(struct super_block *);
1975 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
1976  struct ext4_allocation_request *, int *);
1977 extern int ext4_mb_reserve_blocks(struct super_block *, int);
1978 extern void ext4_discard_preallocations(struct inode *);
1979 extern int __init ext4_init_mballoc(void);
1980 extern void ext4_exit_mballoc(void);
1981 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
1982  struct buffer_head *bh, ext4_fsblk_t block,
1983  unsigned long count, int flags);
1984 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
1985  ext4_group_t ngroups);
1986 extern int ext4_mb_add_groupinfo(struct super_block *sb,
1987  ext4_group_t i, struct ext4_group_desc *desc);
1988 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
1989  ext4_fsblk_t block, unsigned long count);
1990 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
1991 
1992 /* inode.c */
1993 struct buffer_head *ext4_getblk(handle_t *, struct inode *,
1994  ext4_lblk_t, int, int *);
1995 struct buffer_head *ext4_bread(handle_t *, struct inode *,
1996  ext4_lblk_t, int, int *);
1997 int ext4_get_block(struct inode *inode, sector_t iblock,
1998  struct buffer_head *bh_result, int create);
1999 
2000 extern struct inode *ext4_iget(struct super_block *, unsigned long);
2001 extern int ext4_write_inode(struct inode *, struct writeback_control *);
2002 extern int ext4_setattr(struct dentry *, struct iattr *);
2003 extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
2004  struct kstat *stat);
2005 extern void ext4_evict_inode(struct inode *);
2006 extern void ext4_clear_inode(struct inode *);
2007 extern int ext4_sync_inode(handle_t *, struct inode *);
2008 extern void ext4_dirty_inode(struct inode *, int);
2009 extern int ext4_change_inode_journal_flag(struct inode *, int);
2010 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2011 extern int ext4_can_truncate(struct inode *inode);
2012 extern void ext4_truncate(struct inode *);
2013 extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length);
2014 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
2015 extern void ext4_set_inode_flags(struct inode *);
2016 extern void ext4_get_inode_flags(struct ext4_inode_info *);
2017 extern int ext4_alloc_da_blocks(struct inode *inode);
2018 extern void ext4_set_aops(struct inode *inode);
2019 extern int ext4_writepage_trans_blocks(struct inode *);
2020 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2021 extern int ext4_discard_partial_page_buffers(handle_t *handle,
2022  struct address_space *mapping, loff_t from,
2023  loff_t length, int flags);
2024 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
2025 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2026 extern void ext4_da_update_reserve_space(struct inode *inode,
2027  int used, int quota_claim);
2028 
2029 /* indirect.c */
2030 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2031  struct ext4_map_blocks *map, int flags);
2032 extern ssize_t ext4_ind_direct_IO(int rw, struct kiocb *iocb,
2033  const struct iovec *iov, loff_t offset,
2034  unsigned long nr_segs);
2035 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2036 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks, int chunk);
2037 extern void ext4_ind_truncate(struct inode *inode);
2038 
2039 /* ioctl.c */
2040 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2041 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2042 
2043 /* migrate.c */
2044 extern int ext4_ext_migrate(struct inode *);
2045 
2046 /* namei.c */
2047 extern int ext4_dirent_csum_verify(struct inode *inode,
2048  struct ext4_dir_entry *dirent);
2049 extern int ext4_orphan_add(handle_t *, struct inode *);
2050 extern int ext4_orphan_del(handle_t *, struct inode *);
2051 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2052  __u32 start_minor_hash, __u32 *next_hash);
2053 
2054 /* resize.c */
2055 extern int ext4_group_add(struct super_block *sb,
2056  struct ext4_new_group_data *input);
2057 extern int ext4_group_extend(struct super_block *sb,
2058  struct ext4_super_block *es,
2059  ext4_fsblk_t n_blocks_count);
2060 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2061 
2062 /* super.c */
2063 extern int ext4_calculate_overhead(struct super_block *sb);
2064 extern int ext4_superblock_csum_verify(struct super_block *sb,
2065  struct ext4_super_block *es);
2066 extern void ext4_superblock_csum_set(struct super_block *sb);
2067 extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2068 extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2069 extern void ext4_kvfree(void *ptr);
2070 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2071  ext4_group_t ngroup);
2072 extern __printf(4, 5)
2073 void __ext4_error(struct super_block *, const char *, unsigned int,
2074  const char *, ...);
2075 #define ext4_error(sb, message...) __ext4_error(sb, __func__, \
2076  __LINE__, ## message)
2077 extern __printf(5, 6)
2078 void ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2079  const char *, ...);
2080 extern __printf(5, 6)
2081 void ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2082  const char *, ...);
2083 extern void __ext4_std_error(struct super_block *, const char *,
2084  unsigned int, int);
2085 extern __printf(4, 5)
2086 void __ext4_abort(struct super_block *, const char *, unsigned int,
2087  const char *, ...);
2088 #define ext4_abort(sb, message...) __ext4_abort(sb, __func__, \
2089  __LINE__, ## message)
2090 extern __printf(4, 5)
2091 void __ext4_warning(struct super_block *, const char *, unsigned int,
2092  const char *, ...);
2093 #define ext4_warning(sb, message...) __ext4_warning(sb, __func__, \
2094  __LINE__, ## message)
2095 extern __printf(3, 4)
2096 void ext4_msg(struct super_block *, const char *, const char *, ...);
2097 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2098  const char *, unsigned int, const char *);
2099 #define dump_mmp_msg(sb, mmp, msg) __dump_mmp_msg(sb, mmp, __func__, \
2100  __LINE__, msg)
2101 extern __printf(7, 8)
2102 void __ext4_grp_locked_error(const char *, unsigned int,
2103  struct super_block *, ext4_group_t,
2104  unsigned long, ext4_fsblk_t,
2105  const char *, ...);
2106 #define ext4_grp_locked_error(sb, grp, message...) \
2107  __ext4_grp_locked_error(__func__, __LINE__, (sb), (grp), ## message)
2108 extern void ext4_update_dynamic_rev(struct super_block *sb);
2109 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2110  __u32 compat);
2111 extern int ext4_update_rocompat_feature(handle_t *handle,
2112  struct super_block *sb, __u32 rocompat);
2113 extern int ext4_update_incompat_feature(handle_t *handle,
2114  struct super_block *sb, __u32 incompat);
2115 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2116  struct ext4_group_desc *bg);
2117 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2118  struct ext4_group_desc *bg);
2119 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2120  struct ext4_group_desc *bg);
2121 extern __u32 ext4_free_group_clusters(struct super_block *sb,
2122  struct ext4_group_desc *bg);
2123 extern __u32 ext4_free_inodes_count(struct super_block *sb,
2124  struct ext4_group_desc *bg);
2125 extern __u32 ext4_used_dirs_count(struct super_block *sb,
2126  struct ext4_group_desc *bg);
2127 extern __u32 ext4_itable_unused_count(struct super_block *sb,
2128  struct ext4_group_desc *bg);
2129 extern void ext4_block_bitmap_set(struct super_block *sb,
2130  struct ext4_group_desc *bg, ext4_fsblk_t blk);
2131 extern void ext4_inode_bitmap_set(struct super_block *sb,
2132  struct ext4_group_desc *bg, ext4_fsblk_t blk);
2133 extern void ext4_inode_table_set(struct super_block *sb,
2134  struct ext4_group_desc *bg, ext4_fsblk_t blk);
2135 extern void ext4_free_group_clusters_set(struct super_block *sb,
2136  struct ext4_group_desc *bg,
2137  __u32 count);
2138 extern void ext4_free_inodes_set(struct super_block *sb,
2139  struct ext4_group_desc *bg, __u32 count);
2140 extern void ext4_used_dirs_set(struct super_block *sb,
2141  struct ext4_group_desc *bg, __u32 count);
2142 extern void ext4_itable_unused_set(struct super_block *sb,
2143  struct ext4_group_desc *bg, __u32 count);
2144 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2145  struct ext4_group_desc *gdp);
2146 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2147  struct ext4_group_desc *gdp);
2148 
2149 static inline int ext4_has_group_desc_csum(struct super_block *sb)
2150 {
2151  return EXT4_HAS_RO_COMPAT_FEATURE(sb,
2154 }
2155 
2156 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2157 {
2158  return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2160 }
2161 
2162 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2163 {
2164  return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2166 }
2167 
2168 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2169 {
2170  return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2172 }
2173 
2174 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2175  ext4_fsblk_t blk)
2176 {
2177  es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2178  es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2179 }
2180 
2181 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2182  ext4_fsblk_t blk)
2183 {
2185  es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2186 }
2187 
2188 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2189  ext4_fsblk_t blk)
2190 {
2191  es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2192  es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2193 }
2194 
2195 static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
2196 {
2197  if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2198  return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2199  le32_to_cpu(raw_inode->i_size_lo);
2200  else
2201  return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2202 }
2203 
2204 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2205 {
2206  raw_inode->i_size_lo = cpu_to_le32(i_size);
2207  raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2208 }
2209 
2210 static inline
2211 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2212  ext4_group_t group)
2213 {
2214  struct ext4_group_info ***grp_info;
2215  long indexv, indexh;
2216  grp_info = EXT4_SB(sb)->s_group_info;
2217  indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2218  indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2219  return grp_info[indexv][indexh];
2220 }
2221 
2222 /*
2223  * Reading s_groups_count requires using smp_rmb() afterwards. See
2224  * the locking protocol documented in the comments of ext4_group_add()
2225  * in resize.c
2226  */
2227 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2228 {
2229  ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
2230 
2231  smp_rmb();
2232  return ngroups;
2233 }
2234 
2235 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2236  ext4_group_t block_group)
2237 {
2238  return block_group >> sbi->s_log_groups_per_flex;
2239 }
2240 
2241 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2242 {
2243  return 1 << sbi->s_log_groups_per_flex;
2244 }
2245 
2246 #define ext4_std_error(sb, errno) \
2247 do { \
2248  if ((errno)) \
2249  __ext4_std_error((sb), __func__, __LINE__, (errno)); \
2250 } while (0)
2251 
2252 #ifdef CONFIG_SMP
2253 /* Each CPU can accumulate percpu_counter_batch clusters in their local
2254  * counters. So we need to make sure we have free clusters more
2255  * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
2256  */
2257 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2258 #else
2259 #define EXT4_FREECLUSTERS_WATERMARK 0
2260 #endif
2261 
2262 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2263 {
2264  /*
2265  * XXX: replace with spinlock if seen contended -bzzz
2266  */
2267  down_write(&EXT4_I(inode)->i_data_sem);
2268  if (newsize > EXT4_I(inode)->i_disksize)
2269  EXT4_I(inode)->i_disksize = newsize;
2270  up_write(&EXT4_I(inode)->i_data_sem);
2271  return ;
2272 }
2273 
2274 struct ext4_group_info {
2275  unsigned long bb_state;
2276  struct rb_root bb_free_root;
2277  ext4_grpblk_t bb_first_free; /* first free block */
2278  ext4_grpblk_t bb_free; /* total free blocks */
2279  ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
2280  ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
2281  struct list_head bb_prealloc_list;
2282 #ifdef DOUBLE_CHECK
2283  void *bb_bitmap;
2284 #endif
2285  struct rw_semaphore alloc_sem;
2286  ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
2287  * regions, index is order.
2288  * bb_counters[3] = 5 means
2289  * 5 free 8-block regions. */
2290 };
2291 
2292 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
2293 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
2294 
2295 #define EXT4_MB_GRP_NEED_INIT(grp) \
2296  (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2297 
2298 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
2299  (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2300 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
2301  (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2302 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
2303  (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2304 
2305 #define EXT4_MAX_CONTENTION 8
2306 #define EXT4_CONTENTION_THRESHOLD 2
2307 
2308 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2309  ext4_group_t group)
2310 {
2311  return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2312 }
2313 
2314 /*
2315  * Returns true if the filesystem is busy enough that attempts to
2316  * access the block group locks has run into contention.
2317  */
2318 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2319 {
2320  return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2321 }
2322 
2323 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2324 {
2325  spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2326  if (spin_trylock(lock))
2327  /*
2328  * We're able to grab the lock right away, so drop the
2329  * lock contention counter.
2330  */
2331  atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2332  else {
2333  /*
2334  * The lock is busy, so bump the contention counter,
2335  * and then wait on the spin lock.
2336  */
2337  atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2338  EXT4_MAX_CONTENTION);
2339  spin_lock(lock);
2340  }
2341 }
2342 
2343 static inline void ext4_unlock_group(struct super_block *sb,
2344  ext4_group_t group)
2345 {
2346  spin_unlock(ext4_group_lock_ptr(sb, group));
2347 }
2348 
2349 /*
2350  * Block validity checking
2351  */
2352 #define ext4_check_indirect_blockref(inode, bh) \
2353  ext4_check_blockref(__func__, __LINE__, inode, \
2354  (__le32 *)(bh)->b_data, \
2355  EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2356 
2357 #define ext4_ind_check_inode(inode) \
2358  ext4_check_blockref(__func__, __LINE__, inode, \
2359  EXT4_I(inode)->i_data, \
2360  EXT4_NDIR_BLOCKS)
2361 
2362 /*
2363  * Inodes and files operations
2364  */
2365 
2366 /* dir.c */
2367 extern const struct file_operations ext4_dir_operations;
2368 
2369 /* file.c */
2370 extern const struct inode_operations ext4_file_inode_operations;
2371 extern const struct file_operations ext4_file_operations;
2372 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
2373 extern void ext4_unwritten_wait(struct inode *inode);
2374 
2375 /* namei.c */
2376 extern const struct inode_operations ext4_dir_inode_operations;
2378 extern struct dentry *ext4_get_parent(struct dentry *child);
2379 
2380 /* symlink.c */
2383 
2384 /* block_validity */
2385 extern void ext4_release_system_zone(struct super_block *sb);
2386 extern int ext4_setup_system_zone(struct super_block *sb);
2387 extern int __init ext4_init_system_zone(void);
2388 extern void ext4_exit_system_zone(void);
2389 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
2390  ext4_fsblk_t start_blk,
2391  unsigned int count);
2392 extern int ext4_check_blockref(const char *, unsigned int,
2393  struct inode *, __le32 *, unsigned int);
2394 
2395 /* extents.c */
2396 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
2397 extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
2398 extern int ext4_ext_index_trans_blocks(struct inode *inode, int nrblocks,
2399  int chunk);
2400 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
2401  struct ext4_map_blocks *map, int flags);
2402 extern void ext4_ext_truncate(struct inode *);
2403 extern int ext4_ext_punch_hole(struct file *file, loff_t offset,
2404  loff_t length);
2405 extern void ext4_ext_init(struct super_block *);
2406 extern void ext4_ext_release(struct super_block *);
2407 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
2408  loff_t len);
2409 extern int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset,
2410  ssize_t len);
2411 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
2412  struct ext4_map_blocks *map, int flags);
2413 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
2414  __u64 start, __u64 len);
2415 /* move_extent.c */
2416 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
2417  __u64 start_orig, __u64 start_donor,
2418  __u64 len, __u64 *moved_len);
2419 
2420 /* page-io.c */
2421 extern int __init ext4_init_pageio(void);
2422 extern void ext4_add_complete_io(ext4_io_end_t *io_end);
2423 extern void ext4_exit_pageio(void);
2424 extern void ext4_ioend_wait(struct inode *);
2425 extern void ext4_free_io_end(ext4_io_end_t *io);
2426 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
2427 extern void ext4_io_submit(struct ext4_io_submit *io);
2428 extern int ext4_bio_write_page(struct ext4_io_submit *io,
2429  struct page *page,
2430  int len,
2431  struct writeback_control *wbc);
2432 
2433 /* mmp.c */
2434 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
2435 extern void ext4_mmp_csum_set(struct super_block *sb, struct mmp_struct *mmp);
2436 extern int ext4_mmp_csum_verify(struct super_block *sb,
2437  struct mmp_struct *mmp);
2438 
2439 /* BH_Uninit flag: blocks are allocated but uninitialized on disk */
2440 enum ext4_state_bits {
2441  BH_Uninit /* blocks are allocated but uninitialized on disk */
2443  BH_AllocFromCluster, /* allocated blocks were part of already
2444  * allocated cluster. Note that this flag will
2445  * never, ever appear in a buffer_head's state
2446  * flag. See EXT4_MAP_FROM_CLUSTER to see where
2447  * this is used. */
2448  BH_Da_Mapped, /* Delayed allocated block that now has a mapping. This
2449  * flag is set when ext4_map_blocks is called on a
2450  * delayed allocated block to get its real mapping. */
2451 };
2452 
2453 BUFFER_FNS(Uninit, uninit)
2454 TAS_BUFFER_FNS(Uninit, uninit)
2455 BUFFER_FNS(Da_Mapped, da_mapped)
2456 
2457 /*
2458  * Add new method to test wether block and inode bitmaps are properly
2459  * initialized. With uninit_bg reading the block from disk is not enough
2460  * to mark the bitmap uptodate. We need to also zero-out the bitmap
2461  */
2462 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
2463 
2464 static inline int bitmap_uptodate(struct buffer_head *bh)
2465 {
2466  return (buffer_uptodate(bh) &&
2467  test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
2468 }
2469 static inline void set_bitmap_uptodate(struct buffer_head *bh)
2470 {
2471  set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
2472 }
2473 
2474 /*
2475  * Disable DIO read nolock optimization, so new dioreaders will be forced
2476  * to grab i_mutex
2477  */
2478 static inline void ext4_inode_block_unlocked_dio(struct inode *inode)
2479 {
2480  ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
2481  smp_mb();
2482 }
2483 static inline void ext4_inode_resume_unlocked_dio(struct inode *inode)
2484 {
2485  smp_mb();
2486  ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
2487 }
2488 
2489 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
2490 
2491 /* For ioend & aio unwritten conversion wait queues */
2492 #define EXT4_WQ_HASH_SZ 37
2493 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
2494  EXT4_WQ_HASH_SZ])
2495 #define ext4_aio_mutex(v) (&ext4__aio_mutex[((unsigned long)(v)) %\
2496  EXT4_WQ_HASH_SZ])
2497 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
2498 extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ];
2499 
2500 #define EXT4_RESIZING 0
2501 extern int ext4_resize_begin(struct super_block *sb);
2502 extern void ext4_resize_end(struct super_block *sb);
2503 
2504 #endif /* __KERNEL__ */
2505 
2506 #include "ext4_extents.h"
2507 
2508 #endif /* _EXT4_H */