]> git.kernelconcepts.de Git - karo-tx-linux.git/blob - fs/ext4/ext4.h
Merge remote-tracking branch 'ext4/dev'
[karo-tx-linux.git] / fs / ext4 / ext4.h
1 /*
2  *  ext4.h
3  *
4  * Copyright (C) 1992, 1993, 1994, 1995
5  * Remy Card (card@masi.ibp.fr)
6  * Laboratoire MASI - Institut Blaise Pascal
7  * Universite Pierre et Marie Curie (Paris VI)
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 <linux/ratelimit.h>
33 #include <crypto/hash.h>
34 #include <linux/falloc.h>
35 #ifdef __KERNEL__
36 #include <linux/compat.h>
37 #endif
38
39 /*
40  * The fourth extended filesystem constants/structures
41  */
42
43 /*
44  * Define EXT4FS_DEBUG to produce debug messages
45  */
46 #undef EXT4FS_DEBUG
47
48 /*
49  * Debug code
50  */
51 #ifdef EXT4FS_DEBUG
52 #define ext4_debug(f, a...)                                             \
53         do {                                                            \
54                 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:",        \
55                         __FILE__, __LINE__, __func__);                  \
56                 printk(KERN_DEBUG f, ## a);                             \
57         } while (0)
58 #else
59 #define ext4_debug(fmt, ...)    no_printk(fmt, ##__VA_ARGS__)
60 #endif
61
62 /*
63  * Turn on EXT_DEBUG to get lots of info about extents operations.
64  */
65 #define EXT_DEBUG__
66 #ifdef EXT_DEBUG
67 #define ext_debug(fmt, ...)     printk(fmt, ##__VA_ARGS__)
68 #else
69 #define ext_debug(fmt, ...)     no_printk(fmt, ##__VA_ARGS__)
70 #endif
71
72 /* data type for block offset of block group */
73 typedef int ext4_grpblk_t;
74
75 /* data type for filesystem-wide blocks number */
76 typedef unsigned long long ext4_fsblk_t;
77
78 /* data type for file logical block number */
79 typedef __u32 ext4_lblk_t;
80
81 /* data type for block group number */
82 typedef unsigned int ext4_group_t;
83
84 enum SHIFT_DIRECTION {
85         SHIFT_LEFT = 0,
86         SHIFT_RIGHT,
87 };
88
89 /*
90  * Flags used in mballoc's allocation_context flags field.
91  *
92  * Also used to show what's going on for debugging purposes when the
93  * flag field is exported via the traceport interface
94  */
95
96 /* prefer goal again. length */
97 #define EXT4_MB_HINT_MERGE              0x0001
98 /* blocks already reserved */
99 #define EXT4_MB_HINT_RESERVED           0x0002
100 /* metadata is being allocated */
101 #define EXT4_MB_HINT_METADATA           0x0004
102 /* first blocks in the file */
103 #define EXT4_MB_HINT_FIRST              0x0008
104 /* search for the best chunk */
105 #define EXT4_MB_HINT_BEST               0x0010
106 /* data is being allocated */
107 #define EXT4_MB_HINT_DATA               0x0020
108 /* don't preallocate (for tails) */
109 #define EXT4_MB_HINT_NOPREALLOC         0x0040
110 /* allocate for locality group */
111 #define EXT4_MB_HINT_GROUP_ALLOC        0x0080
112 /* allocate goal blocks or none */
113 #define EXT4_MB_HINT_GOAL_ONLY          0x0100
114 /* goal is meaningful */
115 #define EXT4_MB_HINT_TRY_GOAL           0x0200
116 /* blocks already pre-reserved by delayed allocation */
117 #define EXT4_MB_DELALLOC_RESERVED       0x0400
118 /* We are doing stream allocation */
119 #define EXT4_MB_STREAM_ALLOC            0x0800
120 /* Use reserved root blocks if needed */
121 #define EXT4_MB_USE_ROOT_BLOCKS         0x1000
122 /* Use blocks from reserved pool */
123 #define EXT4_MB_USE_RESERVED            0x2000
124
125 struct ext4_allocation_request {
126         /* target inode for block we're allocating */
127         struct inode *inode;
128         /* how many blocks we want to allocate */
129         unsigned int len;
130         /* logical block in target inode */
131         ext4_lblk_t logical;
132         /* the closest logical allocated block to the left */
133         ext4_lblk_t lleft;
134         /* the closest logical allocated block to the right */
135         ext4_lblk_t lright;
136         /* phys. target (a hint) */
137         ext4_fsblk_t goal;
138         /* phys. block for the closest logical allocated block to the left */
139         ext4_fsblk_t pleft;
140         /* phys. block for the closest logical allocated block to the right */
141         ext4_fsblk_t pright;
142         /* flags. see above EXT4_MB_HINT_* */
143         unsigned int flags;
144 };
145
146 /*
147  * Logical to physical block mapping, used by ext4_map_blocks()
148  *
149  * This structure is used to pass requests into ext4_map_blocks() as
150  * well as to store the information returned by ext4_map_blocks().  It
151  * takes less room on the stack than a struct buffer_head.
152  */
153 #define EXT4_MAP_NEW            (1 << BH_New)
154 #define EXT4_MAP_MAPPED         (1 << BH_Mapped)
155 #define EXT4_MAP_UNWRITTEN      (1 << BH_Unwritten)
156 #define EXT4_MAP_BOUNDARY       (1 << BH_Boundary)
157 #define EXT4_MAP_FLAGS          (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
158                                  EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
159
160 struct ext4_map_blocks {
161         ext4_fsblk_t m_pblk;
162         ext4_lblk_t m_lblk;
163         unsigned int m_len;
164         unsigned int m_flags;
165 };
166
167 /*
168  * Flags for ext4_io_end->flags
169  */
170 #define EXT4_IO_END_UNWRITTEN   0x0001
171
172 /*
173  * For converting unwritten extents on a work queue. 'handle' is used for
174  * buffered writeback.
175  */
176 typedef struct ext4_io_end {
177         struct list_head        list;           /* per-file finished IO list */
178         handle_t                *handle;        /* handle reserved for extent
179                                                  * conversion */
180         struct inode            *inode;         /* file being written to */
181         struct bio              *bio;           /* Linked list of completed
182                                                  * bios covering the extent */
183         unsigned int            flag;           /* unwritten or not */
184         loff_t                  offset;         /* offset in the file */
185         ssize_t                 size;           /* size of the extent */
186         atomic_t                count;          /* reference counter */
187 } ext4_io_end_t;
188
189 struct ext4_io_submit {
190         struct writeback_control *io_wbc;
191         struct bio              *io_bio;
192         ext4_io_end_t           *io_end;
193         sector_t                io_next_block;
194 };
195
196 /*
197  * Special inodes numbers
198  */
199 #define EXT4_BAD_INO             1      /* Bad blocks inode */
200 #define EXT4_ROOT_INO            2      /* Root inode */
201 #define EXT4_USR_QUOTA_INO       3      /* User quota inode */
202 #define EXT4_GRP_QUOTA_INO       4      /* Group quota inode */
203 #define EXT4_BOOT_LOADER_INO     5      /* Boot loader inode */
204 #define EXT4_UNDEL_DIR_INO       6      /* Undelete directory inode */
205 #define EXT4_RESIZE_INO          7      /* Reserved group descriptors inode */
206 #define EXT4_JOURNAL_INO         8      /* Journal inode */
207
208 /* First non-reserved inode for old ext4 filesystems */
209 #define EXT4_GOOD_OLD_FIRST_INO 11
210
211 /*
212  * Maximal count of links to a file
213  */
214 #define EXT4_LINK_MAX           65000
215
216 /*
217  * Macro-instructions used to manage several block sizes
218  */
219 #define EXT4_MIN_BLOCK_SIZE             1024
220 #define EXT4_MAX_BLOCK_SIZE             65536
221 #define EXT4_MIN_BLOCK_LOG_SIZE         10
222 #define EXT4_MAX_BLOCK_LOG_SIZE         16
223 #ifdef __KERNEL__
224 # define EXT4_BLOCK_SIZE(s)             ((s)->s_blocksize)
225 #else
226 # define EXT4_BLOCK_SIZE(s)             (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
227 #endif
228 #define EXT4_ADDR_PER_BLOCK(s)          (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
229 #define EXT4_CLUSTER_SIZE(s)            (EXT4_BLOCK_SIZE(s) << \
230                                          EXT4_SB(s)->s_cluster_bits)
231 #ifdef __KERNEL__
232 # define EXT4_BLOCK_SIZE_BITS(s)        ((s)->s_blocksize_bits)
233 # define EXT4_CLUSTER_BITS(s)           (EXT4_SB(s)->s_cluster_bits)
234 #else
235 # define EXT4_BLOCK_SIZE_BITS(s)        ((s)->s_log_block_size + 10)
236 #endif
237 #ifdef __KERNEL__
238 #define EXT4_ADDR_PER_BLOCK_BITS(s)     (EXT4_SB(s)->s_addr_per_block_bits)
239 #define EXT4_INODE_SIZE(s)              (EXT4_SB(s)->s_inode_size)
240 #define EXT4_FIRST_INO(s)               (EXT4_SB(s)->s_first_ino)
241 #else
242 #define EXT4_INODE_SIZE(s)      (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
243                                  EXT4_GOOD_OLD_INODE_SIZE : \
244                                  (s)->s_inode_size)
245 #define EXT4_FIRST_INO(s)       (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
246                                  EXT4_GOOD_OLD_FIRST_INO : \
247                                  (s)->s_first_ino)
248 #endif
249 #define EXT4_BLOCK_ALIGN(size, blkbits)         ALIGN((size), (1 << (blkbits)))
250
251 /* Translate a block number to a cluster number */
252 #define EXT4_B2C(sbi, blk)      ((blk) >> (sbi)->s_cluster_bits)
253 /* Translate a cluster number to a block number */
254 #define EXT4_C2B(sbi, cluster)  ((cluster) << (sbi)->s_cluster_bits)
255 /* Translate # of blks to # of clusters */
256 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
257                                  (sbi)->s_cluster_bits)
258 /* Mask out the low bits to get the starting block of the cluster */
259 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) &                              \
260                                   ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
261 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) &                              \
262                                   ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
263 /* Get the cluster offset */
264 #define EXT4_PBLK_COFF(s, pblk) ((pblk) &                               \
265                                  ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
266 #define EXT4_LBLK_COFF(s, lblk) ((lblk) &                               \
267                                  ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
268
269 /*
270  * Structure of a blocks group descriptor
271  */
272 struct ext4_group_desc
273 {
274         __le32  bg_block_bitmap_lo;     /* Blocks bitmap block */
275         __le32  bg_inode_bitmap_lo;     /* Inodes bitmap block */
276         __le32  bg_inode_table_lo;      /* Inodes table block */
277         __le16  bg_free_blocks_count_lo;/* Free blocks count */
278         __le16  bg_free_inodes_count_lo;/* Free inodes count */
279         __le16  bg_used_dirs_count_lo;  /* Directories count */
280         __le16  bg_flags;               /* EXT4_BG_flags (INODE_UNINIT, etc) */
281         __le32  bg_exclude_bitmap_lo;   /* Exclude bitmap for snapshots */
282         __le16  bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
283         __le16  bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
284         __le16  bg_itable_unused_lo;    /* Unused inodes count */
285         __le16  bg_checksum;            /* crc16(sb_uuid+group+desc) */
286         __le32  bg_block_bitmap_hi;     /* Blocks bitmap block MSB */
287         __le32  bg_inode_bitmap_hi;     /* Inodes bitmap block MSB */
288         __le32  bg_inode_table_hi;      /* Inodes table block MSB */
289         __le16  bg_free_blocks_count_hi;/* Free blocks count MSB */
290         __le16  bg_free_inodes_count_hi;/* Free inodes count MSB */
291         __le16  bg_used_dirs_count_hi;  /* Directories count MSB */
292         __le16  bg_itable_unused_hi;    /* Unused inodes count MSB */
293         __le32  bg_exclude_bitmap_hi;   /* Exclude bitmap block MSB */
294         __le16  bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
295         __le16  bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
296         __u32   bg_reserved;
297 };
298
299 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END        \
300         (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
301          sizeof(__le16))
302 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END        \
303         (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
304          sizeof(__le16))
305
306 /*
307  * Structure of a flex block group info
308  */
309
310 struct flex_groups {
311         atomic64_t      free_clusters;
312         atomic_t        free_inodes;
313         atomic_t        used_dirs;
314 };
315
316 #define EXT4_BG_INODE_UNINIT    0x0001 /* Inode table/bitmap not in use */
317 #define EXT4_BG_BLOCK_UNINIT    0x0002 /* Block bitmap not in use */
318 #define EXT4_BG_INODE_ZEROED    0x0004 /* On-disk itable initialized to zero */
319
320 /*
321  * Macro-instructions used to manage group descriptors
322  */
323 #define EXT4_MIN_DESC_SIZE              32
324 #define EXT4_MIN_DESC_SIZE_64BIT        64
325 #define EXT4_MAX_DESC_SIZE              EXT4_MIN_BLOCK_SIZE
326 #define EXT4_DESC_SIZE(s)               (EXT4_SB(s)->s_desc_size)
327 #ifdef __KERNEL__
328 # define EXT4_BLOCKS_PER_GROUP(s)       (EXT4_SB(s)->s_blocks_per_group)
329 # define EXT4_CLUSTERS_PER_GROUP(s)     (EXT4_SB(s)->s_clusters_per_group)
330 # define EXT4_DESC_PER_BLOCK(s)         (EXT4_SB(s)->s_desc_per_block)
331 # define EXT4_INODES_PER_GROUP(s)       (EXT4_SB(s)->s_inodes_per_group)
332 # define EXT4_DESC_PER_BLOCK_BITS(s)    (EXT4_SB(s)->s_desc_per_block_bits)
333 #else
334 # define EXT4_BLOCKS_PER_GROUP(s)       ((s)->s_blocks_per_group)
335 # define EXT4_DESC_PER_BLOCK(s)         (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
336 # define EXT4_INODES_PER_GROUP(s)       ((s)->s_inodes_per_group)
337 #endif
338
339 /*
340  * Constants relative to the data blocks
341  */
342 #define EXT4_NDIR_BLOCKS                12
343 #define EXT4_IND_BLOCK                  EXT4_NDIR_BLOCKS
344 #define EXT4_DIND_BLOCK                 (EXT4_IND_BLOCK + 1)
345 #define EXT4_TIND_BLOCK                 (EXT4_DIND_BLOCK + 1)
346 #define EXT4_N_BLOCKS                   (EXT4_TIND_BLOCK + 1)
347
348 /*
349  * Inode flags
350  */
351 #define EXT4_SECRM_FL                   0x00000001 /* Secure deletion */
352 #define EXT4_UNRM_FL                    0x00000002 /* Undelete */
353 #define EXT4_COMPR_FL                   0x00000004 /* Compress file */
354 #define EXT4_SYNC_FL                    0x00000008 /* Synchronous updates */
355 #define EXT4_IMMUTABLE_FL               0x00000010 /* Immutable file */
356 #define EXT4_APPEND_FL                  0x00000020 /* writes to file may only append */
357 #define EXT4_NODUMP_FL                  0x00000040 /* do not dump file */
358 #define EXT4_NOATIME_FL                 0x00000080 /* do not update atime */
359 /* Reserved for compression usage... */
360 #define EXT4_DIRTY_FL                   0x00000100
361 #define EXT4_COMPRBLK_FL                0x00000200 /* One or more compressed clusters */
362 #define EXT4_NOCOMPR_FL                 0x00000400 /* Don't compress */
363         /* nb: was previously EXT2_ECOMPR_FL */
364 #define EXT4_ENCRYPT_FL                 0x00000800 /* encrypted file */
365 /* End compression flags --- maybe not all used */
366 #define EXT4_INDEX_FL                   0x00001000 /* hash-indexed directory */
367 #define EXT4_IMAGIC_FL                  0x00002000 /* AFS directory */
368 #define EXT4_JOURNAL_DATA_FL            0x00004000 /* file data should be journaled */
369 #define EXT4_NOTAIL_FL                  0x00008000 /* file tail should not be merged */
370 #define EXT4_DIRSYNC_FL                 0x00010000 /* dirsync behaviour (directories only) */
371 #define EXT4_TOPDIR_FL                  0x00020000 /* Top of directory hierarchies*/
372 #define EXT4_HUGE_FILE_FL               0x00040000 /* Set to each huge file */
373 #define EXT4_EXTENTS_FL                 0x00080000 /* Inode uses extents */
374 #define EXT4_EA_INODE_FL                0x00200000 /* Inode used for large EA */
375 #define EXT4_EOFBLOCKS_FL               0x00400000 /* Blocks allocated beyond EOF */
376 #define EXT4_INLINE_DATA_FL             0x10000000 /* Inode has inline data. */
377 #define EXT4_PROJINHERIT_FL             0x20000000 /* Create with parents projid */
378 #define EXT4_RESERVED_FL                0x80000000 /* reserved for ext4 lib */
379
380 #define EXT4_FL_USER_VISIBLE            0x004BDFFF /* User visible flags */
381 #define EXT4_FL_USER_MODIFIABLE         0x004380FF /* User modifiable flags */
382
383 /* Flags that should be inherited by new inodes from their parent. */
384 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
385                            EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
386                            EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
387                            EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
388
389 /* Flags that are appropriate for regular files (all but dir-specific ones). */
390 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
391
392 /* Flags that are appropriate for non-directories/regular files. */
393 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
394
395 /* Mask out flags that are inappropriate for the given type of inode. */
396 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
397 {
398         if (S_ISDIR(mode))
399                 return flags;
400         else if (S_ISREG(mode))
401                 return flags & EXT4_REG_FLMASK;
402         else
403                 return flags & EXT4_OTHER_FLMASK;
404 }
405
406 /*
407  * Inode flags used for atomic set/get
408  */
409 enum {
410         EXT4_INODE_SECRM        = 0,    /* Secure deletion */
411         EXT4_INODE_UNRM         = 1,    /* Undelete */
412         EXT4_INODE_COMPR        = 2,    /* Compress file */
413         EXT4_INODE_SYNC         = 3,    /* Synchronous updates */
414         EXT4_INODE_IMMUTABLE    = 4,    /* Immutable file */
415         EXT4_INODE_APPEND       = 5,    /* writes to file may only append */
416         EXT4_INODE_NODUMP       = 6,    /* do not dump file */
417         EXT4_INODE_NOATIME      = 7,    /* do not update atime */
418 /* Reserved for compression usage... */
419         EXT4_INODE_DIRTY        = 8,
420         EXT4_INODE_COMPRBLK     = 9,    /* One or more compressed clusters */
421         EXT4_INODE_NOCOMPR      = 10,   /* Don't compress */
422         EXT4_INODE_ENCRYPT      = 11,   /* Encrypted file */
423 /* End compression flags --- maybe not all used */
424         EXT4_INODE_INDEX        = 12,   /* hash-indexed directory */
425         EXT4_INODE_IMAGIC       = 13,   /* AFS directory */
426         EXT4_INODE_JOURNAL_DATA = 14,   /* file data should be journaled */
427         EXT4_INODE_NOTAIL       = 15,   /* file tail should not be merged */
428         EXT4_INODE_DIRSYNC      = 16,   /* dirsync behaviour (directories only) */
429         EXT4_INODE_TOPDIR       = 17,   /* Top of directory hierarchies*/
430         EXT4_INODE_HUGE_FILE    = 18,   /* Set to each huge file */
431         EXT4_INODE_EXTENTS      = 19,   /* Inode uses extents */
432         EXT4_INODE_EA_INODE     = 21,   /* Inode used for large EA */
433         EXT4_INODE_EOFBLOCKS    = 22,   /* Blocks allocated beyond EOF */
434         EXT4_INODE_INLINE_DATA  = 28,   /* Data in inode. */
435         EXT4_INODE_PROJINHERIT  = 29,   /* Create with parents projid */
436         EXT4_INODE_RESERVED     = 31,   /* reserved for ext4 lib */
437 };
438
439 /*
440  * Since it's pretty easy to mix up bit numbers and hex values, we use a
441  * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
442  * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
443  * any extra space in the compiled kernel image, otherwise, the build will fail.
444  * It's important that these values are the same, since we are using
445  * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
446  * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
447  * values found in ext2, ext3 and ext4 filesystems, and of course the values
448  * defined in e2fsprogs.
449  *
450  * It's not paranoia if the Murphy's Law really *is* out to get you.  :-)
451  */
452 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
453 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
454
455 static inline void ext4_check_flag_values(void)
456 {
457         CHECK_FLAG_VALUE(SECRM);
458         CHECK_FLAG_VALUE(UNRM);
459         CHECK_FLAG_VALUE(COMPR);
460         CHECK_FLAG_VALUE(SYNC);
461         CHECK_FLAG_VALUE(IMMUTABLE);
462         CHECK_FLAG_VALUE(APPEND);
463         CHECK_FLAG_VALUE(NODUMP);
464         CHECK_FLAG_VALUE(NOATIME);
465         CHECK_FLAG_VALUE(DIRTY);
466         CHECK_FLAG_VALUE(COMPRBLK);
467         CHECK_FLAG_VALUE(NOCOMPR);
468         CHECK_FLAG_VALUE(ENCRYPT);
469         CHECK_FLAG_VALUE(INDEX);
470         CHECK_FLAG_VALUE(IMAGIC);
471         CHECK_FLAG_VALUE(JOURNAL_DATA);
472         CHECK_FLAG_VALUE(NOTAIL);
473         CHECK_FLAG_VALUE(DIRSYNC);
474         CHECK_FLAG_VALUE(TOPDIR);
475         CHECK_FLAG_VALUE(HUGE_FILE);
476         CHECK_FLAG_VALUE(EXTENTS);
477         CHECK_FLAG_VALUE(EA_INODE);
478         CHECK_FLAG_VALUE(EOFBLOCKS);
479         CHECK_FLAG_VALUE(INLINE_DATA);
480         CHECK_FLAG_VALUE(PROJINHERIT);
481         CHECK_FLAG_VALUE(RESERVED);
482 }
483
484 /* Used to pass group descriptor data when online resize is done */
485 struct ext4_new_group_input {
486         __u32 group;            /* Group number for this data */
487         __u64 block_bitmap;     /* Absolute block number of block bitmap */
488         __u64 inode_bitmap;     /* Absolute block number of inode bitmap */
489         __u64 inode_table;      /* Absolute block number of inode table start */
490         __u32 blocks_count;     /* Total number of blocks in this group */
491         __u16 reserved_blocks;  /* Number of reserved blocks in this group */
492         __u16 unused;
493 };
494
495 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
496 struct compat_ext4_new_group_input {
497         u32 group;
498         compat_u64 block_bitmap;
499         compat_u64 inode_bitmap;
500         compat_u64 inode_table;
501         u32 blocks_count;
502         u16 reserved_blocks;
503         u16 unused;
504 };
505 #endif
506
507 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
508 struct ext4_new_group_data {
509         __u32 group;
510         __u64 block_bitmap;
511         __u64 inode_bitmap;
512         __u64 inode_table;
513         __u32 blocks_count;
514         __u16 reserved_blocks;
515         __u16 unused;
516         __u32 free_blocks_count;
517 };
518
519 /* Indexes used to index group tables in ext4_new_group_data */
520 enum {
521         BLOCK_BITMAP = 0,       /* block bitmap */
522         INODE_BITMAP,           /* inode bitmap */
523         INODE_TABLE,            /* inode tables */
524         GROUP_TABLE_COUNT,
525 };
526
527 /*
528  * Flags used by ext4_map_blocks()
529  */
530         /* Allocate any needed blocks and/or convert an unwritten
531            extent to be an initialized ext4 */
532 #define EXT4_GET_BLOCKS_CREATE                  0x0001
533         /* Request the creation of an unwritten extent */
534 #define EXT4_GET_BLOCKS_UNWRIT_EXT              0x0002
535 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT       (EXT4_GET_BLOCKS_UNWRIT_EXT|\
536                                                  EXT4_GET_BLOCKS_CREATE)
537         /* Caller is from the delayed allocation writeout path
538          * finally doing the actual allocation of delayed blocks */
539 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE        0x0004
540         /* caller is from the direct IO path, request to creation of an
541         unwritten extents if not allocated, split the unwritten
542         extent if blocks has been preallocated already*/
543 #define EXT4_GET_BLOCKS_PRE_IO                  0x0008
544 #define EXT4_GET_BLOCKS_CONVERT                 0x0010
545 #define EXT4_GET_BLOCKS_IO_CREATE_EXT           (EXT4_GET_BLOCKS_PRE_IO|\
546                                          EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
547         /* Convert extent to initialized after IO complete */
548 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT          (EXT4_GET_BLOCKS_CONVERT|\
549                                          EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
550         /* Eventual metadata allocation (due to growing extent tree)
551          * should not fail, so try to use reserved blocks for that.*/
552 #define EXT4_GET_BLOCKS_METADATA_NOFAIL         0x0020
553         /* Don't normalize allocation size (used for fallocate) */
554 #define EXT4_GET_BLOCKS_NO_NORMALIZE            0x0040
555         /* Request will not result in inode size update (user for fallocate) */
556 #define EXT4_GET_BLOCKS_KEEP_SIZE               0x0080
557         /* Do not take i_data_sem locking in ext4_map_blocks */
558 #define EXT4_GET_BLOCKS_NO_LOCK                 0x0100
559         /* Convert written extents to unwritten */
560 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN       0x0200
561
562 /*
563  * The bit position of these flags must not overlap with any of the
564  * EXT4_GET_BLOCKS_*.  They are used by ext4_find_extent(),
565  * read_extent_tree_block(), ext4_split_extent_at(),
566  * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
567  * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
568  * caching the extents when reading from the extent tree while a
569  * truncate or punch hole operation is in progress.
570  */
571 #define EXT4_EX_NOCACHE                         0x40000000
572 #define EXT4_EX_FORCE_CACHE                     0x20000000
573
574 /*
575  * Flags used by ext4_free_blocks
576  */
577 #define EXT4_FREE_BLOCKS_METADATA       0x0001
578 #define EXT4_FREE_BLOCKS_FORGET         0x0002
579 #define EXT4_FREE_BLOCKS_VALIDATED      0x0004
580 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
581 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER   0x0010
582 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER    0x0020
583
584 /* Encryption algorithms */
585 #define EXT4_ENCRYPTION_MODE_INVALID            0
586 #define EXT4_ENCRYPTION_MODE_AES_256_XTS        1
587 #define EXT4_ENCRYPTION_MODE_AES_256_GCM        2
588 #define EXT4_ENCRYPTION_MODE_AES_256_CBC        3
589 #define EXT4_ENCRYPTION_MODE_AES_256_CTS        4
590
591 #include "ext4_crypto.h"
592
593 /*
594  * ioctl commands
595  */
596 #define EXT4_IOC_GETFLAGS               FS_IOC_GETFLAGS
597 #define EXT4_IOC_SETFLAGS               FS_IOC_SETFLAGS
598 #define EXT4_IOC_GETVERSION             _IOR('f', 3, long)
599 #define EXT4_IOC_SETVERSION             _IOW('f', 4, long)
600 #define EXT4_IOC_GETVERSION_OLD         FS_IOC_GETVERSION
601 #define EXT4_IOC_SETVERSION_OLD         FS_IOC_SETVERSION
602 #define EXT4_IOC_GETRSVSZ               _IOR('f', 5, long)
603 #define EXT4_IOC_SETRSVSZ               _IOW('f', 6, long)
604 #define EXT4_IOC_GROUP_EXTEND           _IOW('f', 7, unsigned long)
605 #define EXT4_IOC_GROUP_ADD              _IOW('f', 8, struct ext4_new_group_input)
606 #define EXT4_IOC_MIGRATE                _IO('f', 9)
607  /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
608  /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
609 #define EXT4_IOC_ALLOC_DA_BLKS          _IO('f', 12)
610 #define EXT4_IOC_MOVE_EXT               _IOWR('f', 15, struct move_extent)
611 #define EXT4_IOC_RESIZE_FS              _IOW('f', 16, __u64)
612 #define EXT4_IOC_SWAP_BOOT              _IO('f', 17)
613 #define EXT4_IOC_PRECACHE_EXTENTS       _IO('f', 18)
614 #define EXT4_IOC_SET_ENCRYPTION_POLICY  _IOR('f', 19, struct ext4_encryption_policy)
615 #define EXT4_IOC_GET_ENCRYPTION_PWSALT  _IOW('f', 20, __u8[16])
616 #define EXT4_IOC_GET_ENCRYPTION_POLICY  _IOW('f', 21, struct ext4_encryption_policy)
617
618 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
619 /*
620  * ioctl commands in 32 bit emulation
621  */
622 #define EXT4_IOC32_GETFLAGS             FS_IOC32_GETFLAGS
623 #define EXT4_IOC32_SETFLAGS             FS_IOC32_SETFLAGS
624 #define EXT4_IOC32_GETVERSION           _IOR('f', 3, int)
625 #define EXT4_IOC32_SETVERSION           _IOW('f', 4, int)
626 #define EXT4_IOC32_GETRSVSZ             _IOR('f', 5, int)
627 #define EXT4_IOC32_SETRSVSZ             _IOW('f', 6, int)
628 #define EXT4_IOC32_GROUP_EXTEND         _IOW('f', 7, unsigned int)
629 #define EXT4_IOC32_GROUP_ADD            _IOW('f', 8, struct compat_ext4_new_group_input)
630 #define EXT4_IOC32_GETVERSION_OLD       FS_IOC32_GETVERSION
631 #define EXT4_IOC32_SETVERSION_OLD       FS_IOC32_SETVERSION
632 #endif
633
634 /* Max physical block we can address w/o extents */
635 #define EXT4_MAX_BLOCK_FILE_PHYS        0xFFFFFFFF
636
637 /*
638  * Structure of an inode on the disk
639  */
640 struct ext4_inode {
641         __le16  i_mode;         /* File mode */
642         __le16  i_uid;          /* Low 16 bits of Owner Uid */
643         __le32  i_size_lo;      /* Size in bytes */
644         __le32  i_atime;        /* Access time */
645         __le32  i_ctime;        /* Inode Change time */
646         __le32  i_mtime;        /* Modification time */
647         __le32  i_dtime;        /* Deletion Time */
648         __le16  i_gid;          /* Low 16 bits of Group Id */
649         __le16  i_links_count;  /* Links count */
650         __le32  i_blocks_lo;    /* Blocks count */
651         __le32  i_flags;        /* File flags */
652         union {
653                 struct {
654                         __le32  l_i_version;
655                 } linux1;
656                 struct {
657                         __u32  h_i_translator;
658                 } hurd1;
659                 struct {
660                         __u32  m_i_reserved1;
661                 } masix1;
662         } osd1;                         /* OS dependent 1 */
663         __le32  i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
664         __le32  i_generation;   /* File version (for NFS) */
665         __le32  i_file_acl_lo;  /* File ACL */
666         __le32  i_size_high;
667         __le32  i_obso_faddr;   /* Obsoleted fragment address */
668         union {
669                 struct {
670                         __le16  l_i_blocks_high; /* were l_i_reserved1 */
671                         __le16  l_i_file_acl_high;
672                         __le16  l_i_uid_high;   /* these 2 fields */
673                         __le16  l_i_gid_high;   /* were reserved2[0] */
674                         __le16  l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
675                         __le16  l_i_reserved;
676                 } linux2;
677                 struct {
678                         __le16  h_i_reserved1;  /* Obsoleted fragment number/size which are removed in ext4 */
679                         __u16   h_i_mode_high;
680                         __u16   h_i_uid_high;
681                         __u16   h_i_gid_high;
682                         __u32   h_i_author;
683                 } hurd2;
684                 struct {
685                         __le16  h_i_reserved1;  /* Obsoleted fragment number/size which are removed in ext4 */
686                         __le16  m_i_file_acl_high;
687                         __u32   m_i_reserved2[2];
688                 } masix2;
689         } osd2;                         /* OS dependent 2 */
690         __le16  i_extra_isize;
691         __le16  i_checksum_hi;  /* crc32c(uuid+inum+inode) BE */
692         __le32  i_ctime_extra;  /* extra Change time      (nsec << 2 | epoch) */
693         __le32  i_mtime_extra;  /* extra Modification time(nsec << 2 | epoch) */
694         __le32  i_atime_extra;  /* extra Access time      (nsec << 2 | epoch) */
695         __le32  i_crtime;       /* File Creation time */
696         __le32  i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
697         __le32  i_version_hi;   /* high 32 bits for 64-bit version */
698         __le32  i_projid;       /* Project ID */
699 };
700
701 struct move_extent {
702         __u32 reserved;         /* should be zero */
703         __u32 donor_fd;         /* donor file descriptor */
704         __u64 orig_start;       /* logical start offset in block for orig */
705         __u64 donor_start;      /* logical start offset in block for donor */
706         __u64 len;              /* block length to be moved */
707         __u64 moved_len;        /* moved block length */
708 };
709
710 #define EXT4_EPOCH_BITS 2
711 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
712 #define EXT4_NSEC_MASK  (~0UL << EXT4_EPOCH_BITS)
713
714 /*
715  * Extended fields will fit into an inode if the filesystem was formatted
716  * with large inodes (-I 256 or larger) and there are not currently any EAs
717  * consuming all of the available space. For new inodes we always reserve
718  * enough space for the kernel's known extended fields, but for inodes
719  * created with an old kernel this might not have been the case. None of
720  * the extended inode fields is critical for correct filesystem operation.
721  * This macro checks if a certain field fits in the inode. Note that
722  * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
723  */
724 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field)   \
725         ((offsetof(typeof(*ext4_inode), field) +        \
726           sizeof((ext4_inode)->field))                  \
727         <= (EXT4_GOOD_OLD_INODE_SIZE +                  \
728             (einode)->i_extra_isize))                   \
729
730 static inline __le32 ext4_encode_extra_time(struct timespec *time)
731 {
732        return cpu_to_le32((sizeof(time->tv_sec) > 4 ?
733                            (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) |
734                           ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK));
735 }
736
737 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
738 {
739        if (sizeof(time->tv_sec) > 4)
740                time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK)
741                                << 32;
742        time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
743 }
744
745 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode)                          \
746 do {                                                                           \
747         (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec);               \
748         if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     \
749                 (raw_inode)->xtime ## _extra =                                 \
750                                 ext4_encode_extra_time(&(inode)->xtime);       \
751 } while (0)
752
753 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode)                        \
754 do {                                                                           \
755         if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))                      \
756                 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec);      \
757         if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))            \
758                 (raw_inode)->xtime ## _extra =                                 \
759                                 ext4_encode_extra_time(&(einode)->xtime);      \
760 } while (0)
761
762 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode)                          \
763 do {                                                                           \
764         (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime);       \
765         if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     \
766                 ext4_decode_extra_time(&(inode)->xtime,                        \
767                                        raw_inode->xtime ## _extra);            \
768         else                                                                   \
769                 (inode)->xtime.tv_nsec = 0;                                    \
770 } while (0)
771
772 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode)                        \
773 do {                                                                           \
774         if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))                      \
775                 (einode)->xtime.tv_sec =                                       \
776                         (signed)le32_to_cpu((raw_inode)->xtime);               \
777         else                                                                   \
778                 (einode)->xtime.tv_sec = 0;                                    \
779         if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))            \
780                 ext4_decode_extra_time(&(einode)->xtime,                       \
781                                        raw_inode->xtime ## _extra);            \
782         else                                                                   \
783                 (einode)->xtime.tv_nsec = 0;                                   \
784 } while (0)
785
786 #define i_disk_version osd1.linux1.l_i_version
787
788 #if defined(__KERNEL__) || defined(__linux__)
789 #define i_reserved1     osd1.linux1.l_i_reserved1
790 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
791 #define i_blocks_high   osd2.linux2.l_i_blocks_high
792 #define i_uid_low       i_uid
793 #define i_gid_low       i_gid
794 #define i_uid_high      osd2.linux2.l_i_uid_high
795 #define i_gid_high      osd2.linux2.l_i_gid_high
796 #define i_checksum_lo   osd2.linux2.l_i_checksum_lo
797
798 #elif defined(__GNU__)
799
800 #define i_translator    osd1.hurd1.h_i_translator
801 #define i_uid_high      osd2.hurd2.h_i_uid_high
802 #define i_gid_high      osd2.hurd2.h_i_gid_high
803 #define i_author        osd2.hurd2.h_i_author
804
805 #elif defined(__masix__)
806
807 #define i_reserved1     osd1.masix1.m_i_reserved1
808 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
809 #define i_reserved2     osd2.masix2.m_i_reserved2
810
811 #endif /* defined(__KERNEL__) || defined(__linux__) */
812
813 #include "extents_status.h"
814
815 /*
816  * fourth extended file system inode data in memory
817  */
818 struct ext4_inode_info {
819         __le32  i_data[15];     /* unconverted */
820         __u32   i_dtime;
821         ext4_fsblk_t    i_file_acl;
822
823         /*
824          * i_block_group is the number of the block group which contains
825          * this file's inode.  Constant across the lifetime of the inode,
826          * it is ued for making block allocation decisions - we try to
827          * place a file's data blocks near its inode block, and new inodes
828          * near to their parent directory's inode.
829          */
830         ext4_group_t    i_block_group;
831         ext4_lblk_t     i_dir_start_lookup;
832 #if (BITS_PER_LONG < 64)
833         unsigned long   i_state_flags;          /* Dynamic state flags */
834 #endif
835         unsigned long   i_flags;
836
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
846         struct list_head i_orphan;      /* unlinked but open inodes */
847
848         /*
849          * i_disksize keeps track of what the inode size is ON DISK, not
850          * in memory.  During truncate, i_size is set to the new size by
851          * the VFS prior to calling ext4_truncate(), but the filesystem won't
852          * set i_disksize to 0 until the truncate is actually under way.
853          *
854          * The intent is that i_disksize always represents the blocks which
855          * are used by this file.  This allows recovery to restart truncate
856          * on orphans if we crash during truncate.  We actually write i_disksize
857          * into the on-disk inode when writing inodes out, instead of i_size.
858          *
859          * The only time when i_disksize and i_size may be different is when
860          * a truncate is in progress.  The only things which change i_disksize
861          * are ext4_get_block (growth) and ext4_truncate (shrinkth).
862          */
863         loff_t  i_disksize;
864
865         /*
866          * i_data_sem is for serialising ext4_truncate() against
867          * ext4_getblock().  In the 2.4 ext2 design, great chunks of inode's
868          * data tree are chopped off during truncate. We can't do that in
869          * ext4 because whenever we perform intermediate commits during
870          * truncate, the inode and all the metadata blocks *must* be in a
871          * consistent state which allows truncation of the orphans to restart
872          * during recovery.  Hence we must fix the get_block-vs-truncate race
873          * by other means, so we have i_data_sem.
874          */
875         struct rw_semaphore i_data_sem;
876         struct inode vfs_inode;
877         struct jbd2_inode *jinode;
878
879         spinlock_t i_raw_lock;  /* protects updates to the raw inode */
880
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          */
885         struct timespec i_crtime;
886
887         /* mballoc */
888         struct list_head i_prealloc_list;
889         spinlock_t i_prealloc_lock;
890
891         /* extents status tree */
892         struct ext4_es_tree i_es_tree;
893         rwlock_t i_es_lock;
894         struct list_head i_es_list;
895         unsigned int i_es_all_nr;       /* protected by i_es_lock */
896         unsigned int i_es_shk_nr;       /* protected by i_es_lock */
897         ext4_lblk_t i_es_shrink_lblk;   /* Offset where we start searching for
898                                            extents to shrink. Protected by
899                                            i_es_lock  */
900
901         /* ialloc */
902         ext4_group_t    i_last_alloc_group;
903
904         /* allocation reservation info for delalloc */
905         /* In case of bigalloc, these refer to clusters rather than blocks */
906         unsigned int i_reserved_data_blocks;
907         unsigned int i_reserved_meta_blocks;
908         unsigned int i_allocated_meta_blocks;
909         ext4_lblk_t i_da_metadata_calc_last_lblock;
910         int i_da_metadata_calc_len;
911
912         /* on-disk additional length */
913         __u16 i_extra_isize;
914
915         /* Indicate the inline data space. */
916         u16 i_inline_off;
917         u16 i_inline_size;
918
919 #ifdef CONFIG_QUOTA
920         /* quota space reservation, managed internally by quota code */
921         qsize_t i_reserved_quota;
922 #endif
923
924         /* Lock protecting lists below */
925         spinlock_t i_completed_io_lock;
926         /*
927          * Completed IOs that need unwritten extents handling and have
928          * transaction reserved
929          */
930         struct list_head i_rsv_conversion_list;
931         /*
932          * Completed IOs that need unwritten extents handling and don't have
933          * transaction reserved
934          */
935         atomic_t i_ioend_count; /* Number of outstanding io_end structs */
936         atomic_t i_unwritten; /* Nr. of inflight conversions pending */
937         struct work_struct i_rsv_conversion_work;
938
939         spinlock_t i_block_reservation_lock;
940
941         /*
942          * Transactions that contain inode's metadata needed to complete
943          * fsync and fdatasync, respectively.
944          */
945         tid_t i_sync_tid;
946         tid_t i_datasync_tid;
947
948 #ifdef CONFIG_QUOTA
949         struct dquot *i_dquot[MAXQUOTAS];
950 #endif
951
952         /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
953         __u32 i_csum_seed;
954
955 #ifdef CONFIG_EXT4_FS_ENCRYPTION
956         /* Encryption params */
957         struct ext4_crypt_info *i_crypt_info;
958 #endif
959 };
960
961 /*
962  * File system states
963  */
964 #define EXT4_VALID_FS                   0x0001  /* Unmounted cleanly */
965 #define EXT4_ERROR_FS                   0x0002  /* Errors detected */
966 #define EXT4_ORPHAN_FS                  0x0004  /* Orphans being recovered */
967
968 /*
969  * Misc. filesystem flags
970  */
971 #define EXT2_FLAGS_SIGNED_HASH          0x0001  /* Signed dirhash in use */
972 #define EXT2_FLAGS_UNSIGNED_HASH        0x0002  /* Unsigned dirhash in use */
973 #define EXT2_FLAGS_TEST_FILESYS         0x0004  /* to test development code */
974
975 /*
976  * Mount flags set via mount options or defaults
977  */
978 #define EXT4_MOUNT_GRPID                0x00004 /* Create files with directory's group */
979 #define EXT4_MOUNT_DEBUG                0x00008 /* Some debugging messages */
980 #define EXT4_MOUNT_ERRORS_CONT          0x00010 /* Continue on errors */
981 #define EXT4_MOUNT_ERRORS_RO            0x00020 /* Remount fs ro on errors */
982 #define EXT4_MOUNT_ERRORS_PANIC         0x00040 /* Panic on errors */
983 #define EXT4_MOUNT_ERRORS_MASK          0x00070
984 #define EXT4_MOUNT_MINIX_DF             0x00080 /* Mimics the Minix statfs */
985 #define EXT4_MOUNT_NOLOAD               0x00100 /* Don't use existing journal*/
986 #ifdef CONFIG_FS_DAX
987 #define EXT4_MOUNT_DAX                  0x00200 /* Direct Access */
988 #else
989 #define EXT4_MOUNT_DAX                  0
990 #endif
991 #define EXT4_MOUNT_DATA_FLAGS           0x00C00 /* Mode for data writes: */
992 #define EXT4_MOUNT_JOURNAL_DATA         0x00400 /* Write data to journal */
993 #define EXT4_MOUNT_ORDERED_DATA         0x00800 /* Flush data before commit */
994 #define EXT4_MOUNT_WRITEBACK_DATA       0x00C00 /* No data ordering */
995 #define EXT4_MOUNT_UPDATE_JOURNAL       0x01000 /* Update the journal format */
996 #define EXT4_MOUNT_NO_UID32             0x02000  /* Disable 32-bit UIDs */
997 #define EXT4_MOUNT_XATTR_USER           0x04000 /* Extended user attributes */
998 #define EXT4_MOUNT_POSIX_ACL            0x08000 /* POSIX Access Control Lists */
999 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC     0x10000 /* No auto delalloc mapping */
1000 #define EXT4_MOUNT_BARRIER              0x20000 /* Use block barriers */
1001 #define EXT4_MOUNT_QUOTA                0x80000 /* Some quota option set */
1002 #define EXT4_MOUNT_USRQUOTA             0x100000 /* "old" user quota */
1003 #define EXT4_MOUNT_GRPQUOTA             0x200000 /* "old" group quota */
1004 #define EXT4_MOUNT_DIOREAD_NOLOCK       0x400000 /* Enable support for dio read nolocking */
1005 #define EXT4_MOUNT_JOURNAL_CHECKSUM     0x800000 /* Journal checksums */
1006 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1007 #define EXT4_MOUNT_DELALLOC             0x8000000 /* Delalloc support */
1008 #define EXT4_MOUNT_DATA_ERR_ABORT       0x10000000 /* Abort on file data write */
1009 #define EXT4_MOUNT_BLOCK_VALIDITY       0x20000000 /* Block validity checking */
1010 #define EXT4_MOUNT_DISCARD              0x40000000 /* Issue DISCARD requests */
1011 #define EXT4_MOUNT_INIT_INODE_TABLE     0x80000000 /* Initialize uninitialized itables */
1012
1013 /*
1014  * Mount flags set either automatically (could not be set by mount option)
1015  * based on per file system feature or property or in special cases such as
1016  * distinguishing between explicit mount option definition and default.
1017  */
1018 #define EXT4_MOUNT2_EXPLICIT_DELALLOC   0x00000001 /* User explicitly
1019                                                       specified delalloc */
1020 #define EXT4_MOUNT2_STD_GROUP_SIZE      0x00000002 /* We have standard group
1021                                                       size of blocksize * 8
1022                                                       blocks */
1023 #define EXT4_MOUNT2_HURD_COMPAT         0x00000004 /* Support HURD-castrated
1024                                                       file systems */
1025
1026 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM   0x00000008 /* User explicitly
1027                                                 specified journal checksum */
1028
1029 #define clear_opt(sb, opt)              EXT4_SB(sb)->s_mount_opt &= \
1030                                                 ~EXT4_MOUNT_##opt
1031 #define set_opt(sb, opt)                EXT4_SB(sb)->s_mount_opt |= \
1032                                                 EXT4_MOUNT_##opt
1033 #define test_opt(sb, opt)               (EXT4_SB(sb)->s_mount_opt & \
1034                                          EXT4_MOUNT_##opt)
1035
1036 #define clear_opt2(sb, opt)             EXT4_SB(sb)->s_mount_opt2 &= \
1037                                                 ~EXT4_MOUNT2_##opt
1038 #define set_opt2(sb, opt)               EXT4_SB(sb)->s_mount_opt2 |= \
1039                                                 EXT4_MOUNT2_##opt
1040 #define test_opt2(sb, opt)              (EXT4_SB(sb)->s_mount_opt2 & \
1041                                          EXT4_MOUNT2_##opt)
1042
1043 #define ext4_test_and_set_bit           __test_and_set_bit_le
1044 #define ext4_set_bit                    __set_bit_le
1045 #define ext4_set_bit_atomic             ext2_set_bit_atomic
1046 #define ext4_test_and_clear_bit         __test_and_clear_bit_le
1047 #define ext4_clear_bit                  __clear_bit_le
1048 #define ext4_clear_bit_atomic           ext2_clear_bit_atomic
1049 #define ext4_test_bit                   test_bit_le
1050 #define ext4_find_next_zero_bit         find_next_zero_bit_le
1051 #define ext4_find_next_bit              find_next_bit_le
1052
1053 extern void ext4_set_bits(void *bm, int cur, int len);
1054
1055 /*
1056  * Maximal mount counts between two filesystem checks
1057  */
1058 #define EXT4_DFL_MAX_MNT_COUNT          20      /* Allow 20 mounts */
1059 #define EXT4_DFL_CHECKINTERVAL          0       /* Don't use interval check */
1060
1061 /*
1062  * Behaviour when detecting errors
1063  */
1064 #define EXT4_ERRORS_CONTINUE            1       /* Continue execution */
1065 #define EXT4_ERRORS_RO                  2       /* Remount fs read-only */
1066 #define EXT4_ERRORS_PANIC               3       /* Panic */
1067 #define EXT4_ERRORS_DEFAULT             EXT4_ERRORS_CONTINUE
1068
1069 /* Metadata checksum algorithm codes */
1070 #define EXT4_CRC32C_CHKSUM              1
1071
1072 /*
1073  * Structure of the super block
1074  */
1075 struct ext4_super_block {
1076 /*00*/  __le32  s_inodes_count;         /* Inodes count */
1077         __le32  s_blocks_count_lo;      /* Blocks count */
1078         __le32  s_r_blocks_count_lo;    /* Reserved blocks count */
1079         __le32  s_free_blocks_count_lo; /* Free blocks count */
1080 /*10*/  __le32  s_free_inodes_count;    /* Free inodes count */
1081         __le32  s_first_data_block;     /* First Data Block */
1082         __le32  s_log_block_size;       /* Block size */
1083         __le32  s_log_cluster_size;     /* Allocation cluster size */
1084 /*20*/  __le32  s_blocks_per_group;     /* # Blocks per group */
1085         __le32  s_clusters_per_group;   /* # Clusters per group */
1086         __le32  s_inodes_per_group;     /* # Inodes per group */
1087         __le32  s_mtime;                /* Mount time */
1088 /*30*/  __le32  s_wtime;                /* Write time */
1089         __le16  s_mnt_count;            /* Mount count */
1090         __le16  s_max_mnt_count;        /* Maximal mount count */
1091         __le16  s_magic;                /* Magic signature */
1092         __le16  s_state;                /* File system state */
1093         __le16  s_errors;               /* Behaviour when detecting errors */
1094         __le16  s_minor_rev_level;      /* minor revision level */
1095 /*40*/  __le32  s_lastcheck;            /* time of last check */
1096         __le32  s_checkinterval;        /* max. time between checks */
1097         __le32  s_creator_os;           /* OS */
1098         __le32  s_rev_level;            /* Revision level */
1099 /*50*/  __le16  s_def_resuid;           /* Default uid for reserved blocks */
1100         __le16  s_def_resgid;           /* Default gid for reserved blocks */
1101         /*
1102          * These fields are for EXT4_DYNAMIC_REV superblocks only.
1103          *
1104          * Note: the difference between the compatible feature set and
1105          * the incompatible feature set is that if there is a bit set
1106          * in the incompatible feature set that the kernel doesn't
1107          * know about, it should refuse to mount the filesystem.
1108          *
1109          * e2fsck's requirements are more strict; if it doesn't know
1110          * about a feature in either the compatible or incompatible
1111          * feature set, it must abort and not try to meddle with
1112          * things it doesn't understand...
1113          */
1114         __le32  s_first_ino;            /* First non-reserved inode */
1115         __le16  s_inode_size;           /* size of inode structure */
1116         __le16  s_block_group_nr;       /* block group # of this superblock */
1117         __le32  s_feature_compat;       /* compatible feature set */
1118 /*60*/  __le32  s_feature_incompat;     /* incompatible feature set */
1119         __le32  s_feature_ro_compat;    /* readonly-compatible feature set */
1120 /*68*/  __u8    s_uuid[16];             /* 128-bit uuid for volume */
1121 /*78*/  char    s_volume_name[16];      /* volume name */
1122 /*88*/  char    s_last_mounted[64];     /* directory where last mounted */
1123 /*C8*/  __le32  s_algorithm_usage_bitmap; /* For compression */
1124         /*
1125          * Performance hints.  Directory preallocation should only
1126          * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1127          */
1128         __u8    s_prealloc_blocks;      /* Nr of blocks to try to preallocate*/
1129         __u8    s_prealloc_dir_blocks;  /* Nr to preallocate for dirs */
1130         __le16  s_reserved_gdt_blocks;  /* Per group desc for online growth */
1131         /*
1132          * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1133          */
1134 /*D0*/  __u8    s_journal_uuid[16];     /* uuid of journal superblock */
1135 /*E0*/  __le32  s_journal_inum;         /* inode number of journal file */
1136         __le32  s_journal_dev;          /* device number of journal file */
1137         __le32  s_last_orphan;          /* start of list of inodes to delete */
1138         __le32  s_hash_seed[4];         /* HTREE hash seed */
1139         __u8    s_def_hash_version;     /* Default hash version to use */
1140         __u8    s_jnl_backup_type;
1141         __le16  s_desc_size;            /* size of group descriptor */
1142 /*100*/ __le32  s_default_mount_opts;
1143         __le32  s_first_meta_bg;        /* First metablock block group */
1144         __le32  s_mkfs_time;            /* When the filesystem was created */
1145         __le32  s_jnl_blocks[17];       /* Backup of the journal inode */
1146         /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1147 /*150*/ __le32  s_blocks_count_hi;      /* Blocks count */
1148         __le32  s_r_blocks_count_hi;    /* Reserved blocks count */
1149         __le32  s_free_blocks_count_hi; /* Free blocks count */
1150         __le16  s_min_extra_isize;      /* All inodes have at least # bytes */
1151         __le16  s_want_extra_isize;     /* New inodes should reserve # bytes */
1152         __le32  s_flags;                /* Miscellaneous flags */
1153         __le16  s_raid_stride;          /* RAID stride */
1154         __le16  s_mmp_update_interval;  /* # seconds to wait in MMP checking */
1155         __le64  s_mmp_block;            /* Block for multi-mount protection */
1156         __le32  s_raid_stripe_width;    /* blocks on all data disks (N*stride)*/
1157         __u8    s_log_groups_per_flex;  /* FLEX_BG group size */
1158         __u8    s_checksum_type;        /* metadata checksum algorithm used */
1159         __u8    s_encryption_level;     /* versioning level for encryption */
1160         __u8    s_reserved_pad;         /* Padding to next 32bits */
1161         __le64  s_kbytes_written;       /* nr of lifetime kilobytes written */
1162         __le32  s_snapshot_inum;        /* Inode number of active snapshot */
1163         __le32  s_snapshot_id;          /* sequential ID of active snapshot */
1164         __le64  s_snapshot_r_blocks_count; /* reserved blocks for active
1165                                               snapshot's future use */
1166         __le32  s_snapshot_list;        /* inode number of the head of the
1167                                            on-disk snapshot list */
1168 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1169         __le32  s_error_count;          /* number of fs errors */
1170         __le32  s_first_error_time;     /* first time an error happened */
1171         __le32  s_first_error_ino;      /* inode involved in first error */
1172         __le64  s_first_error_block;    /* block involved of first error */
1173         __u8    s_first_error_func[32]; /* function where the error happened */
1174         __le32  s_first_error_line;     /* line number where error happened */
1175         __le32  s_last_error_time;      /* most recent time of an error */
1176         __le32  s_last_error_ino;       /* inode involved in last error */
1177         __le32  s_last_error_line;      /* line number where error happened */
1178         __le64  s_last_error_block;     /* block involved of last error */
1179         __u8    s_last_error_func[32];  /* function where the error happened */
1180 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1181         __u8    s_mount_opts[64];
1182         __le32  s_usr_quota_inum;       /* inode for tracking user quota */
1183         __le32  s_grp_quota_inum;       /* inode for tracking group quota */
1184         __le32  s_overhead_clusters;    /* overhead blocks/clusters in fs */
1185         __le32  s_backup_bgs[2];        /* groups with sparse_super2 SBs */
1186         __u8    s_encrypt_algos[4];     /* Encryption algorithms in use  */
1187         __u8    s_encrypt_pw_salt[16];  /* Salt used for string2key algorithm */
1188         __le32  s_lpf_ino;              /* Location of the lost+found inode */
1189         __le32  s_prj_quota_inum;       /* inode for tracking project quota */
1190         __le32  s_checksum_seed;        /* crc32c(uuid) if csum_seed set */
1191         __le32  s_reserved[98];         /* Padding to the end of the block */
1192         __le32  s_checksum;             /* crc32c(superblock) */
1193 };
1194
1195 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1196
1197 #ifdef __KERNEL__
1198
1199 /*
1200  * run-time mount flags
1201  */
1202 #define EXT4_MF_MNTDIR_SAMPLED          0x0001
1203 #define EXT4_MF_FS_ABORTED              0x0002  /* Fatal error detected */
1204 #define EXT4_MF_TEST_DUMMY_ENCRYPTION   0x0004
1205
1206 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1207 #define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \
1208                                                 EXT4_MF_TEST_DUMMY_ENCRYPTION))
1209 #else
1210 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1211 #endif
1212
1213 /* Number of quota types we support */
1214 #define EXT4_MAXQUOTAS 2
1215
1216 /*
1217  * fourth extended-fs super-block data in memory
1218  */
1219 struct ext4_sb_info {
1220         unsigned long s_desc_size;      /* Size of a group descriptor in bytes */
1221         unsigned long s_inodes_per_block;/* Number of inodes per block */
1222         unsigned long s_blocks_per_group;/* Number of blocks in a group */
1223         unsigned long s_clusters_per_group; /* Number of clusters in a group */
1224         unsigned long s_inodes_per_group;/* Number of inodes in a group */
1225         unsigned long s_itb_per_group;  /* Number of inode table blocks per group */
1226         unsigned long s_gdb_count;      /* Number of group descriptor blocks */
1227         unsigned long s_desc_per_block; /* Number of group descriptors per block */
1228         ext4_group_t s_groups_count;    /* Number of groups in the fs */
1229         ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1230         unsigned long s_overhead;  /* # of fs overhead clusters */
1231         unsigned int s_cluster_ratio;   /* Number of blocks per cluster */
1232         unsigned int s_cluster_bits;    /* log2 of s_cluster_ratio */
1233         loff_t s_bitmap_maxbytes;       /* max bytes for bitmap files */
1234         struct buffer_head * s_sbh;     /* Buffer containing the super block */
1235         struct ext4_super_block *s_es;  /* Pointer to the super block in the buffer */
1236         struct buffer_head **s_group_desc;
1237         unsigned int s_mount_opt;
1238         unsigned int s_mount_opt2;
1239         unsigned int s_mount_flags;
1240         unsigned int s_def_mount_opt;
1241         ext4_fsblk_t s_sb_block;
1242         atomic64_t s_resv_clusters;
1243         kuid_t s_resuid;
1244         kgid_t s_resgid;
1245         unsigned short s_mount_state;
1246         unsigned short s_pad;
1247         int s_addr_per_block_bits;
1248         int s_desc_per_block_bits;
1249         int s_inode_size;
1250         int s_first_ino;
1251         unsigned int s_inode_readahead_blks;
1252         unsigned int s_inode_goal;
1253         spinlock_t s_next_gen_lock;
1254         u32 s_next_generation;
1255         u32 s_hash_seed[4];
1256         int s_def_hash_version;
1257         int s_hash_unsigned;    /* 3 if hash should be signed, 0 if not */
1258         struct percpu_counter s_freeclusters_counter;
1259         struct percpu_counter s_freeinodes_counter;
1260         struct percpu_counter s_dirs_counter;
1261         struct percpu_counter s_dirtyclusters_counter;
1262         struct blockgroup_lock *s_blockgroup_lock;
1263         struct proc_dir_entry *s_proc;
1264         struct kobject s_kobj;
1265         struct completion s_kobj_unregister;
1266         struct super_block *s_sb;
1267
1268         /* Journaling */
1269         struct journal_s *s_journal;
1270         struct list_head s_orphan;
1271         struct mutex s_orphan_lock;
1272         unsigned long s_resize_flags;           /* Flags indicating if there
1273                                                    is a resizer */
1274         unsigned long s_commit_interval;
1275         u32 s_max_batch_time;
1276         u32 s_min_batch_time;
1277         struct block_device *journal_bdev;
1278 #ifdef CONFIG_QUOTA
1279         char *s_qf_names[EXT4_MAXQUOTAS];       /* Names of quota files with journalled quota */
1280         int s_jquota_fmt;                       /* Format of quota to use */
1281 #endif
1282         unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1283         struct rb_root system_blks;
1284
1285 #ifdef EXTENTS_STATS
1286         /* ext4 extents stats */
1287         unsigned long s_ext_min;
1288         unsigned long s_ext_max;
1289         unsigned long s_depth_max;
1290         spinlock_t s_ext_stats_lock;
1291         unsigned long s_ext_blocks;
1292         unsigned long s_ext_extents;
1293 #endif
1294
1295         /* for buddy allocator */
1296         struct ext4_group_info ***s_group_info;
1297         struct inode *s_buddy_cache;
1298         spinlock_t s_md_lock;
1299         unsigned short *s_mb_offsets;
1300         unsigned int *s_mb_maxs;
1301         unsigned int s_group_info_size;
1302
1303         /* tunables */
1304         unsigned long s_stripe;
1305         unsigned int s_mb_stream_request;
1306         unsigned int s_mb_max_to_scan;
1307         unsigned int s_mb_min_to_scan;
1308         unsigned int s_mb_stats;
1309         unsigned int s_mb_order2_reqs;
1310         unsigned int s_mb_group_prealloc;
1311         unsigned int s_max_dir_size_kb;
1312         /* where last allocation was done - for stream allocation */
1313         unsigned long s_mb_last_group;
1314         unsigned long s_mb_last_start;
1315
1316         /* stats for buddy allocator */
1317         atomic_t s_bal_reqs;    /* number of reqs with len > 1 */
1318         atomic_t s_bal_success; /* we found long enough chunks */
1319         atomic_t s_bal_allocated;       /* in blocks */
1320         atomic_t s_bal_ex_scanned;      /* total extents scanned */
1321         atomic_t s_bal_goals;   /* goal hits */
1322         atomic_t s_bal_breaks;  /* too long searches */
1323         atomic_t s_bal_2orders; /* 2^order hits */
1324         spinlock_t s_bal_lock;
1325         unsigned long s_mb_buddies_generated;
1326         unsigned long long s_mb_generation_time;
1327         atomic_t s_mb_lost_chunks;
1328         atomic_t s_mb_preallocated;
1329         atomic_t s_mb_discarded;
1330         atomic_t s_lock_busy;
1331
1332         /* locality groups */
1333         struct ext4_locality_group __percpu *s_locality_groups;
1334
1335         /* for write statistics */
1336         unsigned long s_sectors_written_start;
1337         u64 s_kbytes_written;
1338
1339         /* the size of zero-out chunk */
1340         unsigned int s_extent_max_zeroout_kb;
1341
1342         unsigned int s_log_groups_per_flex;
1343         struct flex_groups *s_flex_groups;
1344         ext4_group_t s_flex_groups_allocated;
1345
1346         /* workqueue for reserved extent conversions (buffered io) */
1347         struct workqueue_struct *rsv_conversion_wq;
1348
1349         /* timer for periodic error stats printing */
1350         struct timer_list s_err_report;
1351
1352         /* Lazy inode table initialization info */
1353         struct ext4_li_request *s_li_request;
1354         /* Wait multiplier for lazy initialization thread */
1355         unsigned int s_li_wait_mult;
1356
1357         /* Kernel thread for multiple mount protection */
1358         struct task_struct *s_mmp_tsk;
1359
1360         /* record the last minlen when FITRIM is called. */
1361         atomic_t s_last_trim_minblks;
1362
1363         /* Reference to checksum algorithm driver via cryptoapi */
1364         struct crypto_shash *s_chksum_driver;
1365
1366         /* Precomputed FS UUID checksum for seeding other checksums */
1367         __u32 s_csum_seed;
1368
1369         /* Reclaim extents from extent status tree */
1370         struct shrinker s_es_shrinker;
1371         struct list_head s_es_list;     /* List of inodes with reclaimable extents */
1372         long s_es_nr_inode;
1373         struct ext4_es_stats s_es_stats;
1374         struct mb_cache *s_mb_cache;
1375         spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1376
1377         /* Ratelimit ext4 messages. */
1378         struct ratelimit_state s_err_ratelimit_state;
1379         struct ratelimit_state s_warning_ratelimit_state;
1380         struct ratelimit_state s_msg_ratelimit_state;
1381 };
1382
1383 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1384 {
1385         return sb->s_fs_info;
1386 }
1387 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1388 {
1389         return container_of(inode, struct ext4_inode_info, vfs_inode);
1390 }
1391
1392 static inline struct timespec ext4_current_time(struct inode *inode)
1393 {
1394         return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1395                 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1396 }
1397
1398 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1399 {
1400         return ino == EXT4_ROOT_INO ||
1401                 ino == EXT4_USR_QUOTA_INO ||
1402                 ino == EXT4_GRP_QUOTA_INO ||
1403                 ino == EXT4_BOOT_LOADER_INO ||
1404                 ino == EXT4_JOURNAL_INO ||
1405                 ino == EXT4_RESIZE_INO ||
1406                 (ino >= EXT4_FIRST_INO(sb) &&
1407                  ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1408 }
1409
1410 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
1411                                               struct ext4_io_end *io_end)
1412 {
1413         if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
1414                 io_end->flag |= EXT4_IO_END_UNWRITTEN;
1415                 atomic_inc(&EXT4_I(inode)->i_unwritten);
1416         }
1417 }
1418
1419 static inline ext4_io_end_t *ext4_inode_aio(struct inode *inode)
1420 {
1421         return inode->i_private;
1422 }
1423
1424 static inline void ext4_inode_aio_set(struct inode *inode, ext4_io_end_t *io)
1425 {
1426         inode->i_private = io;
1427 }
1428
1429 /*
1430  * Inode dynamic state flags
1431  */
1432 enum {
1433         EXT4_STATE_JDATA,               /* journaled data exists */
1434         EXT4_STATE_NEW,                 /* inode is newly created */
1435         EXT4_STATE_XATTR,               /* has in-inode xattrs */
1436         EXT4_STATE_NO_EXPAND,           /* No space for expansion */
1437         EXT4_STATE_DA_ALLOC_CLOSE,      /* Alloc DA blks on close */
1438         EXT4_STATE_EXT_MIGRATE,         /* Inode is migrating */
1439         EXT4_STATE_DIO_UNWRITTEN,       /* need convert on dio done*/
1440         EXT4_STATE_NEWENTRY,            /* File just added to dir */
1441         EXT4_STATE_DIOREAD_LOCK,        /* Disable support for dio read
1442                                            nolocking */
1443         EXT4_STATE_MAY_INLINE_DATA,     /* may have in-inode data */
1444         EXT4_STATE_ORDERED_MODE,        /* data=ordered mode */
1445         EXT4_STATE_EXT_PRECACHED,       /* extents have been precached */
1446 };
1447
1448 #define EXT4_INODE_BIT_FNS(name, field, offset)                         \
1449 static inline int ext4_test_inode_##name(struct inode *inode, int bit)  \
1450 {                                                                       \
1451         return test_bit(bit + (offset), &EXT4_I(inode)->i_##field);     \
1452 }                                                                       \
1453 static inline void ext4_set_inode_##name(struct inode *inode, int bit)  \
1454 {                                                                       \
1455         set_bit(bit + (offset), &EXT4_I(inode)->i_##field);             \
1456 }                                                                       \
1457 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1458 {                                                                       \
1459         clear_bit(bit + (offset), &EXT4_I(inode)->i_##field);           \
1460 }
1461
1462 /* Add these declarations here only so that these functions can be
1463  * found by name.  Otherwise, they are very hard to locate. */
1464 static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1465 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1466 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1467 EXT4_INODE_BIT_FNS(flag, flags, 0)
1468
1469 /* Add these declarations here only so that these functions can be
1470  * found by name.  Otherwise, they are very hard to locate. */
1471 static inline int ext4_test_inode_state(struct inode *inode, int bit);
1472 static inline void ext4_set_inode_state(struct inode *inode, int bit);
1473 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1474 #if (BITS_PER_LONG < 64)
1475 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1476
1477 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1478 {
1479         (ei)->i_state_flags = 0;
1480 }
1481 #else
1482 EXT4_INODE_BIT_FNS(state, flags, 32)
1483
1484 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1485 {
1486         /* We depend on the fact that callers will set i_flags */
1487 }
1488 #endif
1489 #else
1490 /* Assume that user mode programs are passing in an ext4fs superblock, not
1491  * a kernel struct super_block.  This will allow us to call the feature-test
1492  * macros from user land. */
1493 #define EXT4_SB(sb)     (sb)
1494 #endif
1495
1496 /*
1497  * Returns true if the inode is inode is encrypted
1498  */
1499 static inline int ext4_encrypted_inode(struct inode *inode)
1500 {
1501 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1502         return ext4_test_inode_flag(inode, EXT4_INODE_ENCRYPT);
1503 #else
1504         return 0;
1505 #endif
1506 }
1507
1508 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1509
1510 /*
1511  * Codes for operating systems
1512  */
1513 #define EXT4_OS_LINUX           0
1514 #define EXT4_OS_HURD            1
1515 #define EXT4_OS_MASIX           2
1516 #define EXT4_OS_FREEBSD         3
1517 #define EXT4_OS_LITES           4
1518
1519 /*
1520  * Revision levels
1521  */
1522 #define EXT4_GOOD_OLD_REV       0       /* The good old (original) format */
1523 #define EXT4_DYNAMIC_REV        1       /* V2 format w/ dynamic inode sizes */
1524
1525 #define EXT4_CURRENT_REV        EXT4_GOOD_OLD_REV
1526 #define EXT4_MAX_SUPP_REV       EXT4_DYNAMIC_REV
1527
1528 #define EXT4_GOOD_OLD_INODE_SIZE 128
1529
1530 /*
1531  * Feature set definitions
1532  */
1533
1534 /* Use the ext4_{has,set,clear}_feature_* helpers; these will be removed */
1535 #define EXT4_HAS_COMPAT_FEATURE(sb,mask)                        \
1536         ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1537 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask)                     \
1538         ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1539 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask)                      \
1540         ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1541 #define EXT4_SET_COMPAT_FEATURE(sb,mask)                        \
1542         EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1543 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask)                     \
1544         EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1545 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask)                      \
1546         EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1547 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask)                      \
1548         EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1549 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask)                   \
1550         EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1551 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask)                    \
1552         EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1553
1554 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC        0x0001
1555 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES       0x0002
1556 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL         0x0004
1557 #define EXT4_FEATURE_COMPAT_EXT_ATTR            0x0008
1558 #define EXT4_FEATURE_COMPAT_RESIZE_INODE        0x0010
1559 #define EXT4_FEATURE_COMPAT_DIR_INDEX           0x0020
1560 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2       0x0200
1561
1562 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER     0x0001
1563 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE       0x0002
1564 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR        0x0004
1565 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE        0x0008
1566 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM         0x0010
1567 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK        0x0020
1568 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE      0x0040
1569 #define EXT4_FEATURE_RO_COMPAT_QUOTA            0x0100
1570 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC         0x0200
1571 /*
1572  * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM).  When
1573  * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1574  * all other data structures' checksums.  However, the METADATA_CSUM and
1575  * GDT_CSUM bits are mutually exclusive.
1576  */
1577 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM    0x0400
1578 #define EXT4_FEATURE_RO_COMPAT_READONLY         0x1000
1579 #define EXT4_FEATURE_RO_COMPAT_PROJECT          0x2000
1580
1581 #define EXT4_FEATURE_INCOMPAT_COMPRESSION       0x0001
1582 #define EXT4_FEATURE_INCOMPAT_FILETYPE          0x0002
1583 #define EXT4_FEATURE_INCOMPAT_RECOVER           0x0004 /* Needs recovery */
1584 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV       0x0008 /* Journal device */
1585 #define EXT4_FEATURE_INCOMPAT_META_BG           0x0010
1586 #define EXT4_FEATURE_INCOMPAT_EXTENTS           0x0040 /* extents support */
1587 #define EXT4_FEATURE_INCOMPAT_64BIT             0x0080
1588 #define EXT4_FEATURE_INCOMPAT_MMP               0x0100
1589 #define EXT4_FEATURE_INCOMPAT_FLEX_BG           0x0200
1590 #define EXT4_FEATURE_INCOMPAT_EA_INODE          0x0400 /* EA in inode */
1591 #define EXT4_FEATURE_INCOMPAT_DIRDATA           0x1000 /* data in dirent */
1592 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED         0x2000
1593 #define EXT4_FEATURE_INCOMPAT_LARGEDIR          0x4000 /* >2GB or 3-lvl htree */
1594 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA       0x8000 /* data in inode */
1595 #define EXT4_FEATURE_INCOMPAT_ENCRYPT           0x10000
1596
1597 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
1598 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1599 { \
1600         return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1601                 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
1602 } \
1603 static inline void ext4_set_feature_##name(struct super_block *sb) \
1604 { \
1605         EXT4_SB(sb)->s_es->s_feature_compat |= \
1606                 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1607 } \
1608 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1609 { \
1610         EXT4_SB(sb)->s_es->s_feature_compat &= \
1611                 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1612 }
1613
1614 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
1615 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1616 { \
1617         return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1618                 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
1619 } \
1620 static inline void ext4_set_feature_##name(struct super_block *sb) \
1621 { \
1622         EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
1623                 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1624 } \
1625 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1626 { \
1627         EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
1628                 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1629 }
1630
1631 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
1632 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1633 { \
1634         return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1635                 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
1636 } \
1637 static inline void ext4_set_feature_##name(struct super_block *sb) \
1638 { \
1639         EXT4_SB(sb)->s_es->s_feature_incompat |= \
1640                 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1641 } \
1642 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1643 { \
1644         EXT4_SB(sb)->s_es->s_feature_incompat &= \
1645                 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1646 }
1647
1648 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc,         DIR_PREALLOC)
1649 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes,        IMAGIC_INODES)
1650 EXT4_FEATURE_COMPAT_FUNCS(journal,              HAS_JOURNAL)
1651 EXT4_FEATURE_COMPAT_FUNCS(xattr,                EXT_ATTR)
1652 EXT4_FEATURE_COMPAT_FUNCS(resize_inode,         RESIZE_INODE)
1653 EXT4_FEATURE_COMPAT_FUNCS(dir_index,            DIR_INDEX)
1654 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2,        SPARSE_SUPER2)
1655
1656 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super,      SPARSE_SUPER)
1657 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file,        LARGE_FILE)
1658 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir,         BTREE_DIR)
1659 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file,         HUGE_FILE)
1660 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum,          GDT_CSUM)
1661 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink,         DIR_NLINK)
1662 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize,       EXTRA_ISIZE)
1663 EXT4_FEATURE_RO_COMPAT_FUNCS(quota,             QUOTA)
1664 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc,          BIGALLOC)
1665 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum,     METADATA_CSUM)
1666 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly,          READONLY)
1667 EXT4_FEATURE_RO_COMPAT_FUNCS(project,           PROJECT)
1668
1669 EXT4_FEATURE_INCOMPAT_FUNCS(compression,        COMPRESSION)
1670 EXT4_FEATURE_INCOMPAT_FUNCS(filetype,           FILETYPE)
1671 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery,     RECOVER)
1672 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev,        JOURNAL_DEV)
1673 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg,            META_BG)
1674 EXT4_FEATURE_INCOMPAT_FUNCS(extents,            EXTENTS)
1675 EXT4_FEATURE_INCOMPAT_FUNCS(64bit,              64BIT)
1676 EXT4_FEATURE_INCOMPAT_FUNCS(mmp,                MMP)
1677 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg,            FLEX_BG)
1678 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode,           EA_INODE)
1679 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata,            DIRDATA)
1680 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed,          CSUM_SEED)
1681 EXT4_FEATURE_INCOMPAT_FUNCS(largedir,           LARGEDIR)
1682 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data,        INLINE_DATA)
1683 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt,            ENCRYPT)
1684
1685 #define EXT2_FEATURE_COMPAT_SUPP        EXT4_FEATURE_COMPAT_EXT_ATTR
1686 #define EXT2_FEATURE_INCOMPAT_SUPP      (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1687                                          EXT4_FEATURE_INCOMPAT_META_BG)
1688 #define EXT2_FEATURE_RO_COMPAT_SUPP     (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1689                                          EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1690                                          EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1691
1692 #define EXT3_FEATURE_COMPAT_SUPP        EXT4_FEATURE_COMPAT_EXT_ATTR
1693 #define EXT3_FEATURE_INCOMPAT_SUPP      (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1694                                          EXT4_FEATURE_INCOMPAT_RECOVER| \
1695                                          EXT4_FEATURE_INCOMPAT_META_BG)
1696 #define EXT3_FEATURE_RO_COMPAT_SUPP     (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1697                                          EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1698                                          EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1699
1700 #define EXT4_FEATURE_COMPAT_SUPP        EXT4_FEATURE_COMPAT_EXT_ATTR
1701 #define EXT4_FEATURE_INCOMPAT_SUPP      (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1702                                          EXT4_FEATURE_INCOMPAT_RECOVER| \
1703                                          EXT4_FEATURE_INCOMPAT_META_BG| \
1704                                          EXT4_FEATURE_INCOMPAT_EXTENTS| \
1705                                          EXT4_FEATURE_INCOMPAT_64BIT| \
1706                                          EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1707                                          EXT4_FEATURE_INCOMPAT_MMP | \
1708                                          EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
1709                                          EXT4_FEATURE_INCOMPAT_ENCRYPT | \
1710                                          EXT4_FEATURE_INCOMPAT_CSUM_SEED)
1711 #define EXT4_FEATURE_RO_COMPAT_SUPP     (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1712                                          EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1713                                          EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1714                                          EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1715                                          EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1716                                          EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1717                                          EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1718                                          EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1719                                          EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1720                                          EXT4_FEATURE_RO_COMPAT_QUOTA)
1721
1722 #define EXTN_FEATURE_FUNCS(ver) \
1723 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
1724 { \
1725         return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1726                 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
1727 } \
1728 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
1729 { \
1730         return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1731                 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
1732 } \
1733 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
1734 { \
1735         return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1736                 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
1737 }
1738
1739 EXTN_FEATURE_FUNCS(2)
1740 EXTN_FEATURE_FUNCS(3)
1741 EXTN_FEATURE_FUNCS(4)
1742
1743 static inline bool ext4_has_compat_features(struct super_block *sb)
1744 {
1745         return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
1746 }
1747 static inline bool ext4_has_ro_compat_features(struct super_block *sb)
1748 {
1749         return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
1750 }
1751 static inline bool ext4_has_incompat_features(struct super_block *sb)
1752 {
1753         return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
1754 }
1755
1756 /*
1757  * Default values for user and/or group using reserved blocks
1758  */
1759 #define EXT4_DEF_RESUID         0
1760 #define EXT4_DEF_RESGID         0
1761
1762 #define EXT4_DEF_INODE_READAHEAD_BLKS   32
1763
1764 /*
1765  * Default mount options
1766  */
1767 #define EXT4_DEFM_DEBUG         0x0001
1768 #define EXT4_DEFM_BSDGROUPS     0x0002
1769 #define EXT4_DEFM_XATTR_USER    0x0004
1770 #define EXT4_DEFM_ACL           0x0008
1771 #define EXT4_DEFM_UID16         0x0010
1772 #define EXT4_DEFM_JMODE         0x0060
1773 #define EXT4_DEFM_JMODE_DATA    0x0020
1774 #define EXT4_DEFM_JMODE_ORDERED 0x0040
1775 #define EXT4_DEFM_JMODE_WBACK   0x0060
1776 #define EXT4_DEFM_NOBARRIER     0x0100
1777 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1778 #define EXT4_DEFM_DISCARD       0x0400
1779 #define EXT4_DEFM_NODELALLOC    0x0800
1780
1781 /*
1782  * Default journal batch times
1783  */
1784 #define EXT4_DEF_MIN_BATCH_TIME 0
1785 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1786
1787 /*
1788  * Minimum number of groups in a flexgroup before we separate out
1789  * directories into the first block group of a flexgroup
1790  */
1791 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1792
1793 /*
1794  * Structure of a directory entry
1795  */
1796 #define EXT4_NAME_LEN 255
1797
1798 struct ext4_dir_entry {
1799         __le32  inode;                  /* Inode number */
1800         __le16  rec_len;                /* Directory entry length */
1801         __le16  name_len;               /* Name length */
1802         char    name[EXT4_NAME_LEN];    /* File name */
1803 };
1804
1805 /*
1806  * The new version of the directory entry.  Since EXT4 structures are
1807  * stored in intel byte order, and the name_len field could never be
1808  * bigger than 255 chars, it's safe to reclaim the extra byte for the
1809  * file_type field.
1810  */
1811 struct ext4_dir_entry_2 {
1812         __le32  inode;                  /* Inode number */
1813         __le16  rec_len;                /* Directory entry length */
1814         __u8    name_len;               /* Name length */
1815         __u8    file_type;
1816         char    name[EXT4_NAME_LEN];    /* File name */
1817 };
1818
1819 /*
1820  * This is a bogus directory entry at the end of each leaf block that
1821  * records checksums.
1822  */
1823 struct ext4_dir_entry_tail {
1824         __le32  det_reserved_zero1;     /* Pretend to be unused */
1825         __le16  det_rec_len;            /* 12 */
1826         __u8    det_reserved_zero2;     /* Zero name length */
1827         __u8    det_reserved_ft;        /* 0xDE, fake file type */
1828         __le32  det_checksum;           /* crc32c(uuid+inum+dirblock) */
1829 };
1830
1831 #define EXT4_DIRENT_TAIL(block, blocksize) \
1832         ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1833                                         ((blocksize) - \
1834                                          sizeof(struct ext4_dir_entry_tail))))
1835
1836 /*
1837  * Ext4 directory file types.  Only the low 3 bits are used.  The
1838  * other bits are reserved for now.
1839  */
1840 #define EXT4_FT_UNKNOWN         0
1841 #define EXT4_FT_REG_FILE        1
1842 #define EXT4_FT_DIR             2
1843 #define EXT4_FT_CHRDEV          3
1844 #define EXT4_FT_BLKDEV          4
1845 #define EXT4_FT_FIFO            5
1846 #define EXT4_FT_SOCK            6
1847 #define EXT4_FT_SYMLINK         7
1848
1849 #define EXT4_FT_MAX             8
1850
1851 #define EXT4_FT_DIR_CSUM        0xDE
1852
1853 /*
1854  * EXT4_DIR_PAD defines the directory entries boundaries
1855  *
1856  * NOTE: It must be a multiple of 4
1857  */
1858 #define EXT4_DIR_PAD                    4
1859 #define EXT4_DIR_ROUND                  (EXT4_DIR_PAD - 1)
1860 #define EXT4_DIR_REC_LEN(name_len)      (((name_len) + 8 + EXT4_DIR_ROUND) & \
1861                                          ~EXT4_DIR_ROUND)
1862 #define EXT4_MAX_REC_LEN                ((1<<16)-1)
1863
1864 /*
1865  * If we ever get support for fs block sizes > page_size, we'll need
1866  * to remove the #if statements in the next two functions...
1867  */
1868 static inline unsigned int
1869 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1870 {
1871         unsigned len = le16_to_cpu(dlen);
1872
1873 #if (PAGE_CACHE_SIZE >= 65536)
1874         if (len == EXT4_MAX_REC_LEN || len == 0)
1875                 return blocksize;
1876         return (len & 65532) | ((len & 3) << 16);
1877 #else
1878         return len;
1879 #endif
1880 }
1881
1882 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1883 {
1884         if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1885                 BUG();
1886 #if (PAGE_CACHE_SIZE >= 65536)
1887         if (len < 65536)
1888                 return cpu_to_le16(len);
1889         if (len == blocksize) {
1890                 if (blocksize == 65536)
1891                         return cpu_to_le16(EXT4_MAX_REC_LEN);
1892                 else
1893                         return cpu_to_le16(0);
1894         }
1895         return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1896 #else
1897         return cpu_to_le16(len);
1898 #endif
1899 }
1900
1901 /*
1902  * Hash Tree Directory indexing
1903  * (c) Daniel Phillips, 2001
1904  */
1905
1906 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
1907                     ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
1908 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1909 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1910
1911 /* Legal values for the dx_root hash_version field: */
1912
1913 #define DX_HASH_LEGACY          0
1914 #define DX_HASH_HALF_MD4        1
1915 #define DX_HASH_TEA             2
1916 #define DX_HASH_LEGACY_UNSIGNED 3
1917 #define DX_HASH_HALF_MD4_UNSIGNED       4
1918 #define DX_HASH_TEA_UNSIGNED            5
1919
1920 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
1921                               const void *address, unsigned int length)
1922 {
1923         struct {
1924                 struct shash_desc shash;
1925                 char ctx[4];
1926         } desc;
1927         int err;
1928
1929         BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
1930
1931         desc.shash.tfm = sbi->s_chksum_driver;
1932         desc.shash.flags = 0;
1933         *(u32 *)desc.ctx = crc;
1934
1935         err = crypto_shash_update(&desc.shash, address, length);
1936         BUG_ON(err);
1937
1938         return *(u32 *)desc.ctx;
1939 }
1940
1941 #ifdef __KERNEL__
1942
1943 /* hash info structure used by the directory hash */
1944 struct dx_hash_info
1945 {
1946         u32             hash;
1947         u32             minor_hash;
1948         int             hash_version;
1949         u32             *seed;
1950 };
1951
1952
1953 /* 32 and 64 bit signed EOF for dx directories */
1954 #define EXT4_HTREE_EOF_32BIT   ((1UL  << (32 - 1)) - 1)
1955 #define EXT4_HTREE_EOF_64BIT   ((1ULL << (64 - 1)) - 1)
1956
1957
1958 /*
1959  * Control parameters used by ext4_htree_next_block
1960  */
1961 #define HASH_NB_ALWAYS          1
1962
1963 struct ext4_filename {
1964         const struct qstr *usr_fname;
1965         struct ext4_str disk_name;
1966         struct dx_hash_info hinfo;
1967 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1968         struct ext4_str crypto_buf;
1969 #endif
1970 };
1971
1972 #define fname_name(p) ((p)->disk_name.name)
1973 #define fname_len(p)  ((p)->disk_name.len)
1974
1975 /*
1976  * Describe an inode's exact location on disk and in memory
1977  */
1978 struct ext4_iloc
1979 {
1980         struct buffer_head *bh;
1981         unsigned long offset;
1982         ext4_group_t block_group;
1983 };
1984
1985 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
1986 {
1987         return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
1988 }
1989
1990 /*
1991  * This structure is stuffed into the struct file's private_data field
1992  * for directories.  It is where we put information so that we can do
1993  * readdir operations in hash tree order.
1994  */
1995 struct dir_private_info {
1996         struct rb_root  root;
1997         struct rb_node  *curr_node;
1998         struct fname    *extra_fname;
1999         loff_t          last_pos;
2000         __u32           curr_hash;
2001         __u32           curr_minor_hash;
2002         __u32           next_hash;
2003 };
2004
2005 /* calculate the first block number of the group */
2006 static inline ext4_fsblk_t
2007 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2008 {
2009         return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2010                 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2011 }
2012
2013 /*
2014  * Special error return code only used by dx_probe() and its callers.
2015  */
2016 #define ERR_BAD_DX_DIR  (-(MAX_ERRNO - 1))
2017
2018 /*
2019  * Timeout and state flag for lazy initialization inode thread.
2020  */
2021 #define EXT4_DEF_LI_WAIT_MULT                   10
2022 #define EXT4_DEF_LI_MAX_START_DELAY             5
2023 #define EXT4_LAZYINIT_QUIT                      0x0001
2024 #define EXT4_LAZYINIT_RUNNING                   0x0002
2025
2026 /*
2027  * Lazy inode table initialization info
2028  */
2029 struct ext4_lazy_init {
2030         unsigned long           li_state;
2031         struct list_head        li_request_list;
2032         struct mutex            li_list_mtx;
2033 };
2034
2035 struct ext4_li_request {
2036         struct super_block      *lr_super;
2037         struct ext4_sb_info     *lr_sbi;
2038         ext4_group_t            lr_next_group;
2039         struct list_head        lr_request;
2040         unsigned long           lr_next_sched;
2041         unsigned long           lr_timeout;
2042 };
2043
2044 struct ext4_features {
2045         struct kobject f_kobj;
2046         struct completion f_kobj_unregister;
2047 };
2048
2049 /*
2050  * This structure will be used for multiple mount protection. It will be
2051  * written into the block number saved in the s_mmp_block field in the
2052  * superblock. Programs that check MMP should assume that if
2053  * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2054  * to use the filesystem, regardless of how old the timestamp is.
2055  */
2056 #define EXT4_MMP_MAGIC     0x004D4D50U /* ASCII for MMP */
2057 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2058 #define EXT4_MMP_SEQ_FSCK  0xE24D4D50U /* mmp_seq value when being fscked */
2059 #define EXT4_MMP_SEQ_MAX   0xE24D4D4FU /* maximum valid mmp_seq value */
2060
2061 struct mmp_struct {
2062         __le32  mmp_magic;              /* Magic number for MMP */
2063         __le32  mmp_seq;                /* Sequence no. updated periodically */
2064
2065         /*
2066          * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2067          * purposes and do not affect the correctness of the algorithm
2068          */
2069         __le64  mmp_time;               /* Time last updated */
2070         char    mmp_nodename[64];       /* Node which last updated MMP block */
2071         char    mmp_bdevname[32];       /* Bdev which last updated MMP block */
2072
2073         /*
2074          * mmp_check_interval is used to verify if the MMP block has been
2075          * updated on the block device. The value is updated based on the
2076          * maximum time to write the MMP block during an update cycle.
2077          */
2078         __le16  mmp_check_interval;
2079
2080         __le16  mmp_pad1;
2081         __le32  mmp_pad2[226];
2082         __le32  mmp_checksum;           /* crc32c(uuid+mmp_block) */
2083 };
2084
2085 /* arguments passed to the mmp thread */
2086 struct mmpd_data {
2087         struct buffer_head *bh; /* bh from initial read_mmp_block() */
2088         struct super_block *sb;  /* super block of the fs */
2089 };
2090
2091 /*
2092  * Check interval multiplier
2093  * The MMP block is written every update interval and initially checked every
2094  * update interval x the multiplier (the value is then adapted based on the
2095  * write latency). The reason is that writes can be delayed under load and we
2096  * don't want readers to incorrectly assume that the filesystem is no longer
2097  * in use.
2098  */
2099 #define EXT4_MMP_CHECK_MULT             2UL
2100
2101 /*
2102  * Minimum interval for MMP checking in seconds.
2103  */
2104 #define EXT4_MMP_MIN_CHECK_INTERVAL     5UL
2105
2106 /*
2107  * Maximum interval for MMP checking in seconds.
2108  */
2109 #define EXT4_MMP_MAX_CHECK_INTERVAL     300UL
2110
2111 /*
2112  * Function prototypes
2113  */
2114
2115 /*
2116  * Ok, these declarations are also in <linux/kernel.h> but none of the
2117  * ext4 source programs needs to include it so they are duplicated here.
2118  */
2119 # define NORET_TYPE     /**/
2120 # define ATTRIB_NORET   __attribute__((noreturn))
2121 # define NORET_AND      noreturn,
2122
2123 /* bitmap.c */
2124 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2125 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2126                                 struct ext4_group_desc *gdp,
2127                                 struct buffer_head *bh, int sz);
2128 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2129                                   struct ext4_group_desc *gdp,
2130                                   struct buffer_head *bh, int sz);
2131 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2132                                 struct ext4_group_desc *gdp,
2133                                 struct buffer_head *bh);
2134 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2135                                   struct ext4_group_desc *gdp,
2136                                   struct buffer_head *bh);
2137
2138 /* balloc.c */
2139 extern void ext4_get_group_no_and_offset(struct super_block *sb,
2140                                          ext4_fsblk_t blocknr,
2141                                          ext4_group_t *blockgrpp,
2142                                          ext4_grpblk_t *offsetp);
2143 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2144                                           ext4_fsblk_t block);
2145
2146 extern unsigned int ext4_block_group(struct super_block *sb,
2147                         ext4_fsblk_t blocknr);
2148 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2149                         ext4_fsblk_t blocknr);
2150 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2151 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2152                         ext4_group_t group);
2153 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2154                                          ext4_fsblk_t goal,
2155                                          unsigned int flags,
2156                                          unsigned long *count,
2157                                          int *errp);
2158 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2159                                     s64 nclusters, unsigned int flags);
2160 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2161 extern void ext4_check_blocks_bitmap(struct super_block *);
2162 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2163                                                     ext4_group_t block_group,
2164                                                     struct buffer_head ** bh);
2165 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2166
2167 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2168                                                 ext4_group_t block_group);
2169 extern int ext4_wait_block_bitmap(struct super_block *sb,
2170                                   ext4_group_t block_group,
2171                                   struct buffer_head *bh);
2172 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2173                                                   ext4_group_t block_group);
2174 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2175                                               ext4_group_t block_group,
2176                                               struct ext4_group_desc *gdp);
2177 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2178
2179 /* crypto_policy.c */
2180 int ext4_is_child_context_consistent_with_parent(struct inode *parent,
2181                                                  struct inode *child);
2182 int ext4_inherit_context(struct inode *parent, struct inode *child);
2183 void ext4_to_hex(char *dst, char *src, size_t src_size);
2184 int ext4_process_policy(const struct ext4_encryption_policy *policy,
2185                         struct inode *inode);
2186 int ext4_get_policy(struct inode *inode,
2187                     struct ext4_encryption_policy *policy);
2188
2189 /* crypto.c */
2190 extern struct kmem_cache *ext4_crypt_info_cachep;
2191 bool ext4_valid_contents_enc_mode(uint32_t mode);
2192 uint32_t ext4_validate_encryption_key_size(uint32_t mode, uint32_t size);
2193 extern struct workqueue_struct *ext4_read_workqueue;
2194 struct ext4_crypto_ctx *ext4_get_crypto_ctx(struct inode *inode);
2195 void ext4_release_crypto_ctx(struct ext4_crypto_ctx *ctx);
2196 void ext4_restore_control_page(struct page *data_page);
2197 struct page *ext4_encrypt(struct inode *inode,
2198                           struct page *plaintext_page);
2199 int ext4_decrypt(struct page *page);
2200 int ext4_encrypted_zeroout(struct inode *inode, struct ext4_extent *ex);
2201
2202 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2203 int ext4_init_crypto(void);
2204 void ext4_exit_crypto(void);
2205 static inline int ext4_sb_has_crypto(struct super_block *sb)
2206 {
2207         return ext4_has_feature_encrypt(sb);
2208 }
2209 #else
2210 static inline int ext4_init_crypto(void) { return 0; }
2211 static inline void ext4_exit_crypto(void) { }
2212 static inline int ext4_sb_has_crypto(struct super_block *sb)
2213 {
2214         return 0;
2215 }
2216 #endif
2217
2218 /* crypto_fname.c */
2219 bool ext4_valid_filenames_enc_mode(uint32_t mode);
2220 u32 ext4_fname_crypto_round_up(u32 size, u32 blksize);
2221 unsigned ext4_fname_encrypted_size(struct inode *inode, u32 ilen);
2222 int ext4_fname_crypto_alloc_buffer(struct inode *inode,
2223                                    u32 ilen, struct ext4_str *crypto_str);
2224 int _ext4_fname_disk_to_usr(struct inode *inode,
2225                             struct dx_hash_info *hinfo,
2226                             const struct ext4_str *iname,
2227                             struct ext4_str *oname);
2228 int ext4_fname_disk_to_usr(struct inode *inode,
2229                            struct dx_hash_info *hinfo,
2230                            const struct ext4_dir_entry_2 *de,
2231                            struct ext4_str *oname);
2232 int ext4_fname_usr_to_disk(struct inode *inode,
2233                            const struct qstr *iname,
2234                            struct ext4_str *oname);
2235 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2236 void ext4_fname_crypto_free_buffer(struct ext4_str *crypto_str);
2237 int ext4_fname_setup_filename(struct inode *dir, const struct qstr *iname,
2238                               int lookup, struct ext4_filename *fname);
2239 void ext4_fname_free_filename(struct ext4_filename *fname);
2240 #else
2241 static inline
2242 int ext4_setup_fname_crypto(struct inode *inode)
2243 {
2244         return 0;
2245 }
2246 static inline void ext4_fname_crypto_free_buffer(struct ext4_str *p) { }
2247 static inline int ext4_fname_setup_filename(struct inode *dir,
2248                                      const struct qstr *iname,
2249                                      int lookup, struct ext4_filename *fname)
2250 {
2251         fname->usr_fname = iname;
2252         fname->disk_name.name = (unsigned char *) iname->name;
2253         fname->disk_name.len = iname->len;
2254         return 0;
2255 }
2256 static inline void ext4_fname_free_filename(struct ext4_filename *fname) { }
2257 #endif
2258
2259
2260 /* crypto_key.c */
2261 void ext4_free_crypt_info(struct ext4_crypt_info *ci);
2262 void ext4_free_encryption_info(struct inode *inode, struct ext4_crypt_info *ci);
2263 int _ext4_get_encryption_info(struct inode *inode);
2264
2265 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2266 int ext4_has_encryption_key(struct inode *inode);
2267
2268 static inline int ext4_get_encryption_info(struct inode *inode)
2269 {
2270         struct ext4_crypt_info *ci = EXT4_I(inode)->i_crypt_info;
2271
2272         if (!ci ||
2273             (ci->ci_keyring_key &&
2274              (ci->ci_keyring_key->flags & ((1 << KEY_FLAG_INVALIDATED) |
2275                                            (1 << KEY_FLAG_REVOKED) |
2276                                            (1 << KEY_FLAG_DEAD)))))
2277                 return _ext4_get_encryption_info(inode);
2278         return 0;
2279 }
2280
2281 static inline struct ext4_crypt_info *ext4_encryption_info(struct inode *inode)
2282 {
2283         return EXT4_I(inode)->i_crypt_info;
2284 }
2285
2286 #else
2287 static inline int ext4_has_encryption_key(struct inode *inode)
2288 {
2289         return 0;
2290 }
2291 static inline int ext4_get_encryption_info(struct inode *inode)
2292 {
2293         return 0;
2294 }
2295 static inline struct ext4_crypt_info *ext4_encryption_info(struct inode *inode)
2296 {
2297         return NULL;
2298 }
2299 #endif
2300
2301
2302 /* dir.c */
2303 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2304                                   struct file *,
2305                                   struct ext4_dir_entry_2 *,
2306                                   struct buffer_head *, char *, int,
2307                                   unsigned int);
2308 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset)      \
2309         unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2310                                         (de), (bh), (buf), (size), (offset)))
2311 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2312                                 __u32 minor_hash,
2313                                 struct ext4_dir_entry_2 *dirent,
2314                                 struct ext4_str *ent_name);
2315 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2316 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2317                              struct buffer_head *bh,
2318                              void *buf, int buf_size,
2319                              struct ext4_filename *fname,
2320                              struct ext4_dir_entry_2 **dest_de);
2321 int ext4_insert_dentry(struct inode *dir,
2322                        struct inode *inode,
2323                        struct ext4_dir_entry_2 *de,
2324                        int buf_size,
2325                        struct ext4_filename *fname);
2326 static inline void ext4_update_dx_flag(struct inode *inode)
2327 {
2328         if (!ext4_has_feature_dir_index(inode->i_sb))
2329                 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2330 }
2331 static unsigned char ext4_filetype_table[] = {
2332         DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2333 };
2334
2335 static inline  unsigned char get_dtype(struct super_block *sb, int filetype)
2336 {
2337         if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2338                 return DT_UNKNOWN;
2339
2340         return ext4_filetype_table[filetype];
2341 }
2342 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2343                              void *buf, int buf_size);
2344
2345 /* fsync.c */
2346 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2347
2348 /* hash.c */
2349 extern int ext4fs_dirhash(const char *name, int len, struct
2350                           dx_hash_info *hinfo);
2351
2352 /* ialloc.c */
2353 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2354                                       const struct qstr *qstr, __u32 goal,
2355                                       uid_t *owner, int handle_type,
2356                                       unsigned int line_no, int nblocks);
2357
2358 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner) \
2359         __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2360                          0, 0, 0)
2361 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2362                                     type, nblocks)                  \
2363         __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2364                          (type), __LINE__, (nblocks))
2365
2366
2367 extern void ext4_free_inode(handle_t *, struct inode *);
2368 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2369 extern unsigned long ext4_count_free_inodes(struct super_block *);
2370 extern unsigned long ext4_count_dirs(struct super_block *);
2371 extern void ext4_check_inodes_bitmap(struct super_block *);
2372 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2373 extern int ext4_init_inode_table(struct super_block *sb,
2374                                  ext4_group_t group, int barrier);
2375 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2376
2377 /* mballoc.c */
2378 extern const struct file_operations ext4_seq_mb_groups_fops;
2379 extern long ext4_mb_stats;
2380 extern long ext4_mb_max_to_scan;
2381 extern int ext4_mb_init(struct super_block *);
2382 extern int ext4_mb_release(struct super_block *);
2383 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2384                                 struct ext4_allocation_request *, int *);
2385 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2386 extern void ext4_discard_preallocations(struct inode *);
2387 extern int __init ext4_init_mballoc(void);
2388 extern void ext4_exit_mballoc(void);
2389 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2390                              struct buffer_head *bh, ext4_fsblk_t block,
2391                              unsigned long count, int flags);
2392 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2393                                    ext4_group_t ngroups);
2394 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2395                 ext4_group_t i, struct ext4_group_desc *desc);
2396 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2397                                 ext4_fsblk_t block, unsigned long count);
2398 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2399
2400 /* inode.c */
2401 int ext4_inode_is_fast_symlink(struct inode *inode);
2402 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2403 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2404 int ext4_get_block_write(struct inode *inode, sector_t iblock,
2405                          struct buffer_head *bh_result, int create);
2406 int ext4_get_block_dax(struct inode *inode, sector_t iblock,
2407                          struct buffer_head *bh_result, int create);
2408 int ext4_get_block(struct inode *inode, sector_t iblock,
2409                                 struct buffer_head *bh_result, int create);
2410 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2411                            struct buffer_head *bh, int create);
2412 int ext4_walk_page_buffers(handle_t *handle,
2413                            struct buffer_head *head,
2414                            unsigned from,
2415                            unsigned to,
2416                            int *partial,
2417                            int (*fn)(handle_t *handle,
2418                                      struct buffer_head *bh));
2419 int do_journal_get_write_access(handle_t *handle,
2420                                 struct buffer_head *bh);
2421 #define FALL_BACK_TO_NONDELALLOC 1
2422 #define CONVERT_INLINE_DATA      2
2423
2424 extern struct inode *ext4_iget(struct super_block *, unsigned long);
2425 extern struct inode *ext4_iget_normal(struct super_block *, unsigned long);
2426 extern int  ext4_write_inode(struct inode *, struct writeback_control *);
2427 extern int  ext4_setattr(struct dentry *, struct iattr *);
2428 extern int  ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
2429                                 struct kstat *stat);
2430 extern void ext4_evict_inode(struct inode *);
2431 extern void ext4_clear_inode(struct inode *);
2432 extern int  ext4_sync_inode(handle_t *, struct inode *);
2433 extern void ext4_dirty_inode(struct inode *, int);
2434 extern int ext4_change_inode_journal_flag(struct inode *, int);
2435 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2436 extern int ext4_inode_attach_jinode(struct inode *inode);
2437 extern int ext4_can_truncate(struct inode *inode);
2438 extern void ext4_truncate(struct inode *);
2439 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2440 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
2441 extern void ext4_set_inode_flags(struct inode *);
2442 extern void ext4_get_inode_flags(struct ext4_inode_info *);
2443 extern int ext4_alloc_da_blocks(struct inode *inode);
2444 extern void ext4_set_aops(struct inode *inode);
2445 extern int ext4_writepage_trans_blocks(struct inode *);
2446 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2447 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2448                              loff_t lstart, loff_t lend);
2449 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
2450 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2451 extern void ext4_da_update_reserve_space(struct inode *inode,
2452                                         int used, int quota_claim);
2453
2454 /* indirect.c */
2455 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2456                                 struct ext4_map_blocks *map, int flags);
2457 extern ssize_t ext4_ind_direct_IO(struct kiocb *iocb, struct iov_iter *iter,
2458                                   loff_t offset);
2459 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2460 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2461 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2462 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2463                                  ext4_lblk_t start, ext4_lblk_t end);
2464
2465 /* ioctl.c */
2466 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2467 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2468
2469 /* migrate.c */
2470 extern int ext4_ext_migrate(struct inode *);
2471 extern int ext4_ind_migrate(struct inode *inode);
2472
2473 /* namei.c */
2474 extern int ext4_dirent_csum_verify(struct inode *inode,
2475                                    struct ext4_dir_entry *dirent);
2476 extern int ext4_orphan_add(handle_t *, struct inode *);
2477 extern int ext4_orphan_del(handle_t *, struct inode *);
2478 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2479                                 __u32 start_minor_hash, __u32 *next_hash);
2480 extern int ext4_search_dir(struct buffer_head *bh,
2481                            char *search_buf,
2482                            int buf_size,
2483                            struct inode *dir,
2484                            struct ext4_filename *fname,
2485                            const struct qstr *d_name,
2486                            unsigned int offset,
2487                            struct ext4_dir_entry_2 **res_dir);
2488 extern int ext4_generic_delete_entry(handle_t *handle,
2489                                      struct inode *dir,
2490                                      struct ext4_dir_entry_2 *de_del,
2491                                      struct buffer_head *bh,
2492                                      void *entry_buf,
2493                                      int buf_size,
2494                                      int csum_size);
2495 extern int ext4_empty_dir(struct inode *inode);
2496
2497 /* resize.c */
2498 extern int ext4_group_add(struct super_block *sb,
2499                                 struct ext4_new_group_data *input);
2500 extern int ext4_group_extend(struct super_block *sb,
2501                                 struct ext4_super_block *es,
2502                                 ext4_fsblk_t n_blocks_count);
2503 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2504
2505 /* super.c */
2506 extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
2507 extern int ext4_calculate_overhead(struct super_block *sb);
2508 extern void ext4_superblock_csum_set(struct super_block *sb);
2509 extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2510 extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2511 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2512                                     ext4_group_t ngroup);
2513 extern const char *ext4_decode_error(struct super_block *sb, int errno,
2514                                      char nbuf[16]);
2515
2516 extern __printf(4, 5)
2517 void __ext4_error(struct super_block *, const char *, unsigned int,
2518                   const char *, ...);
2519 extern __printf(5, 6)
2520 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2521                       const char *, ...);
2522 extern __printf(5, 6)
2523 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2524                      const char *, ...);
2525 extern void __ext4_std_error(struct super_block *, const char *,
2526                              unsigned int, int);
2527 extern __printf(4, 5)
2528 void __ext4_abort(struct super_block *, const char *, unsigned int,
2529                   const char *, ...);
2530 extern __printf(4, 5)
2531 void __ext4_warning(struct super_block *, const char *, unsigned int,
2532                     const char *, ...);
2533 extern __printf(4, 5)
2534 void __ext4_warning_inode(const struct inode *inode, const char *function,
2535                           unsigned int line, const char *fmt, ...);
2536 extern __printf(3, 4)
2537 void __ext4_msg(struct super_block *, const char *, const char *, ...);
2538 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2539                            const char *, unsigned int, const char *);
2540 extern __printf(7, 8)
2541 void __ext4_grp_locked_error(const char *, unsigned int,
2542                              struct super_block *, ext4_group_t,
2543                              unsigned long, ext4_fsblk_t,
2544                              const char *, ...);
2545
2546 #define EXT4_ERROR_INODE(inode, fmt, a...) \
2547         ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
2548
2549 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...)                 \
2550         ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
2551
2552 #define EXT4_ERROR_FILE(file, block, fmt, a...)                         \
2553         ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
2554
2555 #ifdef CONFIG_PRINTK
2556
2557 #define ext4_error_inode(inode, func, line, block, fmt, ...)            \
2558         __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__)
2559 #define ext4_error_file(file, func, line, block, fmt, ...)              \
2560         __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2561 #define ext4_error(sb, fmt, ...)                                        \
2562         __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2563 #define ext4_abort(sb, fmt, ...)                                        \
2564         __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2565 #define ext4_warning(sb, fmt, ...)                                      \
2566         __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2567 #define ext4_warning_inode(inode, fmt, ...)                             \
2568         __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
2569 #define ext4_msg(sb, level, fmt, ...)                           \
2570         __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2571 #define dump_mmp_msg(sb, mmp, msg)                                      \
2572         __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2573 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...)            \
2574         __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2575                                 fmt, ##__VA_ARGS__)
2576
2577 #else
2578
2579 #define ext4_error_inode(inode, func, line, block, fmt, ...)            \
2580 do {                                                                    \
2581         no_printk(fmt, ##__VA_ARGS__);                                  \
2582         __ext4_error_inode(inode, "", 0, block, " ");                   \
2583 } while (0)
2584 #define ext4_error_file(file, func, line, block, fmt, ...)              \
2585 do {                                                                    \
2586         no_printk(fmt, ##__VA_ARGS__);                                  \
2587         __ext4_error_file(file, "", 0, block, " ");                     \
2588 } while (0)
2589 #define ext4_error(sb, fmt, ...)                                        \
2590 do {                                                                    \
2591         no_printk(fmt, ##__VA_ARGS__);                                  \
2592         __ext4_error(sb, "", 0, " ");                                   \
2593 } while (0)
2594 #define ext4_abort(sb, fmt, ...)                                        \
2595 do {                                                                    \
2596         no_printk(fmt, ##__VA_ARGS__);                                  \
2597         __ext4_abort(sb, "", 0, " ");                                   \
2598 } while (0)
2599 #define ext4_warning(sb, fmt, ...)                                      \
2600 do {                                                                    \
2601         no_printk(fmt, ##__VA_ARGS__);                                  \
2602         __ext4_warning(sb, "", 0, " ");                                 \
2603 } while (0)
2604 #define ext4_warning_inode(inode, fmt, ...)                             \
2605 do {                                                                    \
2606         no_printk(fmt, ##__VA_ARGS__);                                  \
2607         __ext4_warning_inode(inode, "", 0, " ");                        \
2608 } while (0)
2609 #define ext4_msg(sb, level, fmt, ...)                                   \
2610 do {                                                                    \
2611         no_printk(fmt, ##__VA_ARGS__);                                  \
2612         __ext4_msg(sb, "", " ");                                        \
2613 } while (0)
2614 #define dump_mmp_msg(sb, mmp, msg)                                      \
2615         __dump_mmp_msg(sb, mmp, "", 0, "")
2616 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...)            \
2617 do {                                                                    \
2618         no_printk(fmt, ##__VA_ARGS__);                          \
2619         __ext4_grp_locked_error("", 0, sb, grp, ino, block, " ");       \
2620 } while (0)
2621
2622 #endif
2623
2624 extern void ext4_update_dynamic_rev(struct super_block *sb);
2625 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2626                                         __u32 compat);
2627 extern int ext4_update_rocompat_feature(handle_t *handle,
2628                                         struct super_block *sb, __u32 rocompat);
2629 extern int ext4_update_incompat_feature(handle_t *handle,
2630                                         struct super_block *sb, __u32 incompat);
2631 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2632                                       struct ext4_group_desc *bg);
2633 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2634                                       struct ext4_group_desc *bg);
2635 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2636                                      struct ext4_group_desc *bg);
2637 extern __u32 ext4_free_group_clusters(struct super_block *sb,
2638                                       struct ext4_group_desc *bg);
2639 extern __u32 ext4_free_inodes_count(struct super_block *sb,
2640                                  struct ext4_group_desc *bg);
2641 extern __u32 ext4_used_dirs_count(struct super_block *sb,
2642                                 struct ext4_group_desc *bg);
2643 extern __u32 ext4_itable_unused_count(struct super_block *sb,
2644                                    struct ext4_group_desc *bg);
2645 extern void ext4_block_bitmap_set(struct super_block *sb,
2646                                   struct ext4_group_desc *bg, ext4_fsblk_t blk);
2647 extern void ext4_inode_bitmap_set(struct super_block *sb,
2648                                   struct ext4_group_desc *bg, ext4_fsblk_t blk);
2649 extern void ext4_inode_table_set(struct super_block *sb,
2650                                  struct ext4_group_desc *bg, ext4_fsblk_t blk);
2651 extern void ext4_free_group_clusters_set(struct super_block *sb,
2652                                          struct ext4_group_desc *bg,
2653                                          __u32 count);
2654 extern void ext4_free_inodes_set(struct super_block *sb,
2655                                 struct ext4_group_desc *bg, __u32 count);
2656 extern void ext4_used_dirs_set(struct super_block *sb,
2657                                 struct ext4_group_desc *bg, __u32 count);
2658 extern void ext4_itable_unused_set(struct super_block *sb,
2659                                    struct ext4_group_desc *bg, __u32 count);
2660 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2661                                        struct ext4_group_desc *gdp);
2662 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2663                                      struct ext4_group_desc *gdp);
2664 extern int ext4_register_li_request(struct super_block *sb,
2665                                     ext4_group_t first_not_zeroed);
2666
2667 static inline int ext4_has_group_desc_csum(struct super_block *sb)
2668 {
2669         return ext4_has_feature_gdt_csum(sb) ||
2670                EXT4_SB(sb)->s_chksum_driver != NULL;
2671 }
2672
2673 static inline int ext4_has_metadata_csum(struct super_block *sb)
2674 {
2675         WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
2676                      !EXT4_SB(sb)->s_chksum_driver);
2677
2678         return (EXT4_SB(sb)->s_chksum_driver != NULL);
2679 }
2680 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2681 {
2682         return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2683                 le32_to_cpu(es->s_blocks_count_lo);
2684 }
2685
2686 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2687 {
2688         return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2689                 le32_to_cpu(es->s_r_blocks_count_lo);
2690 }
2691
2692 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2693 {
2694         return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2695                 le32_to_cpu(es->s_free_blocks_count_lo);
2696 }
2697
2698 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2699                                          ext4_fsblk_t blk)
2700 {
2701         es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2702         es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2703 }
2704
2705 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2706                                               ext4_fsblk_t blk)
2707 {
2708         es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2709         es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2710 }
2711
2712 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2713                                            ext4_fsblk_t blk)
2714 {
2715         es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2716         es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2717 }
2718
2719 static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
2720 {
2721         if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2722                 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2723                         le32_to_cpu(raw_inode->i_size_lo);
2724         else
2725                 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2726 }
2727
2728 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2729 {
2730         raw_inode->i_size_lo = cpu_to_le32(i_size);
2731         raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2732 }
2733
2734 static inline
2735 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2736                                             ext4_group_t group)
2737 {
2738          struct ext4_group_info ***grp_info;
2739          long indexv, indexh;
2740          BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
2741          grp_info = EXT4_SB(sb)->s_group_info;
2742          indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2743          indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2744          return grp_info[indexv][indexh];
2745 }
2746
2747 /*
2748  * Reading s_groups_count requires using smp_rmb() afterwards.  See
2749  * the locking protocol documented in the comments of ext4_group_add()
2750  * in resize.c
2751  */
2752 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2753 {
2754         ext4_group_t    ngroups = EXT4_SB(sb)->s_groups_count;
2755
2756         smp_rmb();
2757         return ngroups;
2758 }
2759
2760 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2761                                              ext4_group_t block_group)
2762 {
2763         return block_group >> sbi->s_log_groups_per_flex;
2764 }
2765
2766 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2767 {
2768         return 1 << sbi->s_log_groups_per_flex;
2769 }
2770
2771 #define ext4_std_error(sb, errno)                               \
2772 do {                                                            \
2773         if ((errno))                                            \
2774                 __ext4_std_error((sb), __func__, __LINE__, (errno));    \
2775 } while (0)
2776
2777 #ifdef CONFIG_SMP
2778 /* Each CPU can accumulate percpu_counter_batch clusters in their local
2779  * counters. So we need to make sure we have free clusters more
2780  * than percpu_counter_batch  * nr_cpu_ids. Also add a window of 4 times.
2781  */
2782 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2783 #else
2784 #define EXT4_FREECLUSTERS_WATERMARK 0
2785 #endif
2786
2787 /* Update i_disksize. Requires i_mutex to avoid races with truncate */
2788 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2789 {
2790         WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
2791                      !mutex_is_locked(&inode->i_mutex));
2792         down_write(&EXT4_I(inode)->i_data_sem);
2793         if (newsize > EXT4_I(inode)->i_disksize)
2794                 EXT4_I(inode)->i_disksize = newsize;
2795         up_write(&EXT4_I(inode)->i_data_sem);
2796 }
2797
2798 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
2799 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
2800 {
2801         int changed = 0;
2802
2803         if (newsize > inode->i_size) {
2804                 i_size_write(inode, newsize);
2805                 changed = 1;
2806         }
2807         if (newsize > EXT4_I(inode)->i_disksize) {
2808                 ext4_update_i_disksize(inode, newsize);
2809                 changed |= 2;
2810         }
2811         return changed;
2812 }
2813
2814 struct ext4_group_info {
2815         unsigned long   bb_state;
2816         struct rb_root  bb_free_root;
2817         ext4_grpblk_t   bb_first_free;  /* first free block */
2818         ext4_grpblk_t   bb_free;        /* total free blocks */
2819         ext4_grpblk_t   bb_fragments;   /* nr of freespace fragments */
2820         ext4_grpblk_t   bb_largest_free_order;/* order of largest frag in BG */
2821         struct          list_head bb_prealloc_list;
2822 #ifdef DOUBLE_CHECK
2823         void            *bb_bitmap;
2824 #endif
2825         struct rw_semaphore alloc_sem;
2826         ext4_grpblk_t   bb_counters[];  /* Nr of free power-of-two-block
2827                                          * regions, index is order.
2828                                          * bb_counters[3] = 5 means
2829                                          * 5 free 8-block regions. */
2830 };
2831
2832 #define EXT4_GROUP_INFO_NEED_INIT_BIT           0
2833 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT         1
2834 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT     2
2835 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT     3
2836
2837 #define EXT4_MB_GRP_NEED_INIT(grp)      \
2838         (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2839 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp)        \
2840         (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2841 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp)        \
2842         (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2843
2844 #define EXT4_MB_GRP_WAS_TRIMMED(grp)    \
2845         (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2846 #define EXT4_MB_GRP_SET_TRIMMED(grp)    \
2847         (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2848 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp)  \
2849         (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2850
2851 #define EXT4_MAX_CONTENTION             8
2852 #define EXT4_CONTENTION_THRESHOLD       2
2853
2854 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2855                                               ext4_group_t group)
2856 {
2857         return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2858 }
2859
2860 /*
2861  * Returns true if the filesystem is busy enough that attempts to
2862  * access the block group locks has run into contention.
2863  */
2864 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2865 {
2866         return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2867 }
2868
2869 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2870 {
2871         spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2872         if (spin_trylock(lock))
2873                 /*
2874                  * We're able to grab the lock right away, so drop the
2875                  * lock contention counter.
2876                  */
2877                 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2878         else {
2879                 /*
2880                  * The lock is busy, so bump the contention counter,
2881                  * and then wait on the spin lock.
2882                  */
2883                 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2884                                   EXT4_MAX_CONTENTION);
2885                 spin_lock(lock);
2886         }
2887 }
2888
2889 static inline void ext4_unlock_group(struct super_block *sb,
2890                                         ext4_group_t group)
2891 {
2892         spin_unlock(ext4_group_lock_ptr(sb, group));
2893 }
2894
2895 /*
2896  * Block validity checking
2897  */
2898 #define ext4_check_indirect_blockref(inode, bh)                         \
2899         ext4_check_blockref(__func__, __LINE__, inode,                  \
2900                             (__le32 *)(bh)->b_data,                     \
2901                             EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2902
2903 #define ext4_ind_check_inode(inode)                                     \
2904         ext4_check_blockref(__func__, __LINE__, inode,                  \
2905                             EXT4_I(inode)->i_data,                      \
2906                             EXT4_NDIR_BLOCKS)
2907
2908 /*
2909  * Inodes and files operations
2910  */
2911
2912 /* dir.c */
2913 extern const struct file_operations ext4_dir_operations;
2914
2915 /* file.c */
2916 extern const struct inode_operations ext4_file_inode_operations;
2917 extern const struct file_operations ext4_file_operations;
2918 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
2919
2920 /* inline.c */
2921 extern int ext4_get_max_inline_size(struct inode *inode);
2922 extern int ext4_find_inline_data_nolock(struct inode *inode);
2923 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
2924                                  unsigned int len);
2925 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
2926
2927 extern int ext4_readpage_inline(struct inode *inode, struct page *page);
2928 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
2929                                          struct inode *inode,
2930                                          loff_t pos, unsigned len,
2931                                          unsigned flags,
2932                                          struct page **pagep);
2933 extern int ext4_write_inline_data_end(struct inode *inode,
2934                                       loff_t pos, unsigned len,
2935                                       unsigned copied,
2936                                       struct page *page);
2937 extern struct buffer_head *
2938 ext4_journalled_write_inline_data(struct inode *inode,
2939                                   unsigned len,
2940                                   struct page *page);
2941 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
2942                                            struct inode *inode,
2943                                            loff_t pos, unsigned len,
2944                                            unsigned flags,
2945                                            struct page **pagep,
2946                                            void **fsdata);
2947 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
2948                                          unsigned len, unsigned copied,
2949                                          struct page *page);
2950 extern int ext4_try_add_inline_entry(handle_t *handle,
2951                                      struct ext4_filename *fname,
2952                                      struct dentry *dentry,
2953                                      struct inode *inode);
2954 extern int ext4_try_create_inline_dir(handle_t *handle,
2955                                       struct inode *parent,
2956                                       struct inode *inode);
2957 extern int ext4_read_inline_dir(struct file *filp,
2958                                 struct dir_context *ctx,
2959                                 int *has_inline_data);
2960 extern int htree_inlinedir_to_tree(struct file *dir_file,
2961                                    struct inode *dir, ext4_lblk_t block,
2962                                    struct dx_hash_info *hinfo,
2963                                    __u32 start_hash, __u32 start_minor_hash,
2964                                    int *has_inline_data);
2965 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
2966                                         struct ext4_filename *fname,
2967                                         const struct qstr *d_name,
2968                                         struct ext4_dir_entry_2 **res_dir,
2969                                         int *has_inline_data);
2970 extern int ext4_delete_inline_entry(handle_t *handle,
2971                                     struct inode *dir,
2972                                     struct ext4_dir_entry_2 *de_del,
2973                                     struct buffer_head *bh,
2974                                     int *has_inline_data);
2975 extern int empty_inline_dir(struct inode *dir, int *has_inline_data);
2976 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
2977                                         struct ext4_dir_entry_2 **parent_de,
2978                                         int *retval);
2979 extern int ext4_inline_data_fiemap(struct inode *inode,
2980                                    struct fiemap_extent_info *fieinfo,
2981                                    int *has_inline, __u64 start, __u64 len);
2982 extern int ext4_try_to_evict_inline_data(handle_t *handle,
2983                                          struct inode *inode,
2984                                          int needed);
2985 extern void ext4_inline_data_truncate(struct inode *inode, int *has_inline);
2986
2987 extern int ext4_convert_inline_data(struct inode *inode);
2988
2989 static inline int ext4_has_inline_data(struct inode *inode)
2990 {
2991         return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
2992                EXT4_I(inode)->i_inline_off;
2993 }
2994
2995 /* namei.c */
2996 extern const struct inode_operations ext4_dir_inode_operations;
2997 extern const struct inode_operations ext4_special_inode_operations;
2998 extern struct dentry *ext4_get_parent(struct dentry *child);
2999 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3000                                  struct ext4_dir_entry_2 *de,
3001                                  int blocksize, int csum_size,
3002                                  unsigned int parent_ino, int dotdot_real_len);
3003 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t,
3004                                    unsigned int blocksize);
3005 extern int ext4_handle_dirty_dirent_node(handle_t *handle,
3006                                          struct inode *inode,
3007                                          struct buffer_head *bh);
3008 #define S_SHIFT 12
3009 static unsigned char ext4_type_by_mode[S_IFMT >> S_SHIFT] = {
3010         [S_IFREG >> S_SHIFT]    = EXT4_FT_REG_FILE,
3011         [S_IFDIR >> S_SHIFT]    = EXT4_FT_DIR,
3012         [S_IFCHR >> S_SHIFT]    = EXT4_FT_CHRDEV,
3013         [S_IFBLK >> S_SHIFT]    = EXT4_FT_BLKDEV,
3014         [S_IFIFO >> S_SHIFT]    = EXT4_FT_FIFO,
3015         [S_IFSOCK >> S_SHIFT]   = EXT4_FT_SOCK,
3016         [S_IFLNK >> S_SHIFT]    = EXT4_FT_SYMLINK,
3017 };
3018
3019 static inline void ext4_set_de_type(struct super_block *sb,
3020                                 struct ext4_dir_entry_2 *de,
3021                                 umode_t mode) {
3022         if (ext4_has_feature_filetype(sb))
3023                 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3024 }
3025
3026 /* readpages.c */
3027 extern int ext4_mpage_readpages(struct address_space *mapping,
3028                                 struct list_head *pages, struct page *page,
3029                                 unsigned nr_pages);
3030
3031 /* symlink.c */
3032 extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3033 extern const struct inode_operations ext4_symlink_inode_operations;
3034 extern const struct inode_operations ext4_fast_symlink_inode_operations;
3035
3036 /* sysfs.c */
3037 extern int ext4_register_sysfs(struct super_block *sb);
3038 extern void ext4_unregister_sysfs(struct super_block *sb);
3039 extern int __init ext4_init_sysfs(void);
3040 extern void ext4_exit_sysfs(void);
3041
3042 /* block_validity */
3043 extern void ext4_release_system_zone(struct super_block *sb);
3044 extern int ext4_setup_system_zone(struct super_block *sb);
3045 extern int __init ext4_init_system_zone(void);
3046 extern void ext4_exit_system_zone(void);
3047 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
3048                                  ext4_fsblk_t start_blk,
3049                                  unsigned int count);
3050 extern int ext4_check_blockref(const char *, unsigned int,
3051                                struct inode *, __le32 *, unsigned int);
3052
3053 /* extents.c */
3054 struct ext4_ext_path;
3055 struct ext4_extent;
3056
3057 /*
3058  * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3059  * __le32.
3060  */
3061 #define EXT_MAX_BLOCKS  0xffffffff
3062
3063 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
3064 extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
3065 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3066 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3067                                struct ext4_map_blocks *map, int flags);
3068 extern void ext4_ext_truncate(handle_t *, struct inode *);
3069 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3070                                  ext4_lblk_t end);
3071 extern void ext4_ext_init(struct super_block *);
3072 extern void ext4_ext_release(struct super_block *);
3073 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3074                           loff_t len);
3075 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3076                                           loff_t offset, ssize_t len);
3077 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3078                            struct ext4_map_blocks *map, int flags);
3079 extern int ext4_ext_calc_metadata_amount(struct inode *inode,
3080                                          ext4_lblk_t lblocks);
3081 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3082                                                    int num,
3083                                                    struct ext4_ext_path *path);
3084 extern int ext4_can_extents_be_merged(struct inode *inode,
3085                                       struct ext4_extent *ex1,
3086                                       struct ext4_extent *ex2);
3087 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3088                                   struct ext4_ext_path **,
3089                                   struct ext4_extent *, int);
3090 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3091                                               struct ext4_ext_path **,
3092                                               int flags);
3093 extern void ext4_ext_drop_refs(struct ext4_ext_path *);
3094 extern int ext4_ext_check_inode(struct inode *inode);
3095 extern int ext4_find_delalloc_range(struct inode *inode,
3096                                     ext4_lblk_t lblk_start,
3097                                     ext4_lblk_t lblk_end);
3098 extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk);
3099 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3100 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3101                         __u64 start, __u64 len);
3102 extern int ext4_ext_precache(struct inode *inode);
3103 extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len);
3104 extern int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len);
3105 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3106                                 struct inode *inode2, ext4_lblk_t lblk1,
3107                              ext4_lblk_t lblk2,  ext4_lblk_t count,
3108                              int mark_unwritten,int *err);
3109
3110 /* move_extent.c */
3111 extern void ext4_double_down_write_data_sem(struct inode *first,
3112                                             struct inode *second);
3113 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3114                                           struct inode *donor_inode);
3115 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3116                              __u64 start_orig, __u64 start_donor,
3117                              __u64 len, __u64 *moved_len);
3118
3119 /* page-io.c */
3120 extern int __init ext4_init_pageio(void);
3121 extern void ext4_exit_pageio(void);
3122 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3123 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3124 extern int ext4_put_io_end(ext4_io_end_t *io_end);
3125 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3126 extern void ext4_io_submit_init(struct ext4_io_submit *io,
3127                                 struct writeback_control *wbc);
3128 extern void ext4_end_io_rsv_work(struct work_struct *work);
3129 extern void ext4_io_submit(struct ext4_io_submit *io);
3130 extern int ext4_bio_write_page(struct ext4_io_submit *io,
3131                                struct page *page,
3132                                int len,
3133                                struct writeback_control *wbc,
3134                                bool keep_towrite);
3135
3136 /* mmp.c */
3137 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3138
3139 /*
3140  * Add new method to test whether block and inode bitmaps are properly
3141  * initialized. With uninit_bg reading the block from disk is not enough
3142  * to mark the bitmap uptodate. We need to also zero-out the bitmap
3143  */
3144 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3145
3146 static inline int bitmap_uptodate(struct buffer_head *bh)
3147 {
3148         return (buffer_uptodate(bh) &&
3149                         test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3150 }
3151 static inline void set_bitmap_uptodate(struct buffer_head *bh)
3152 {
3153         set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3154 }
3155
3156 /*
3157  * Disable DIO read nolock optimization, so new dioreaders will be forced
3158  * to grab i_mutex
3159  */
3160 static inline void ext4_inode_block_unlocked_dio(struct inode *inode)
3161 {
3162         ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
3163         smp_mb();
3164 }
3165 static inline void ext4_inode_resume_unlocked_dio(struct inode *inode)
3166 {
3167         smp_mb();
3168         ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
3169 }
3170
3171 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
3172
3173 /* For ioend & aio unwritten conversion wait queues */
3174 #define EXT4_WQ_HASH_SZ         37
3175 #define ext4_ioend_wq(v)   (&ext4__ioend_wq[((unsigned long)(v)) %\
3176                                             EXT4_WQ_HASH_SZ])
3177 #define ext4_aio_mutex(v)  (&ext4__aio_mutex[((unsigned long)(v)) %\
3178                                              EXT4_WQ_HASH_SZ])
3179 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3180 extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ];
3181
3182 #define EXT4_RESIZING   0
3183 extern int ext4_resize_begin(struct super_block *sb);
3184 extern void ext4_resize_end(struct super_block *sb);
3185
3186 #endif  /* __KERNEL__ */
3187
3188 #define EFSBADCRC       EBADMSG         /* Bad CRC detected */
3189 #define EFSCORRUPTED    EUCLEAN         /* Filesystem is corrupted */
3190
3191 #endif  /* _EXT4_H */