ext4: remove EXT4_STATE_ORDERED_MODE
[sfrench/cifs-2.6.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/version.h>
30 #include <linux/wait.h>
31 #include <linux/blockgroup_lock.h>
32 #include <linux/percpu_counter.h>
33 #include <linux/ratelimit.h>
34 #include <crypto/hash.h>
35 #include <linux/falloc.h>
36 #ifdef __KERNEL__
37 #include <linux/compat.h>
38 #endif
39
40 /*
41  * The fourth extended filesystem constants/structures
42  */
43
44 /*
45  * with AGGRESSIVE_CHECK allocator runs consistency checks over
46  * structures. these checks slow things down a lot
47  */
48 #define AGGRESSIVE_CHECK__
49
50 /*
51  * with DOUBLE_CHECK defined mballoc creates persistent in-core
52  * bitmaps, maintains and uses them to check for double allocations
53  */
54 #define DOUBLE_CHECK__
55
56 /*
57  * Define EXT4FS_DEBUG to produce debug messages
58  */
59 #undef EXT4FS_DEBUG
60
61 /*
62  * Debug code
63  */
64 #ifdef EXT4FS_DEBUG
65 #define ext4_debug(f, a...)                                             \
66         do {                                                            \
67                 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:",        \
68                         __FILE__, __LINE__, __func__);                  \
69                 printk(KERN_DEBUG f, ## a);                             \
70         } while (0)
71 #else
72 #define ext4_debug(fmt, ...)    no_printk(fmt, ##__VA_ARGS__)
73 #endif
74
75 /*
76  * Turn on EXT_DEBUG to get lots of info about extents operations.
77  */
78 #define EXT_DEBUG__
79 #ifdef EXT_DEBUG
80 #define ext_debug(fmt, ...)     printk(fmt, ##__VA_ARGS__)
81 #else
82 #define ext_debug(fmt, ...)     no_printk(fmt, ##__VA_ARGS__)
83 #endif
84
85 /* data type for block offset of block group */
86 typedef int ext4_grpblk_t;
87
88 /* data type for filesystem-wide blocks number */
89 typedef unsigned long long ext4_fsblk_t;
90
91 /* data type for file logical block number */
92 typedef __u32 ext4_lblk_t;
93
94 /* data type for block group number */
95 typedef unsigned int ext4_group_t;
96
97 enum SHIFT_DIRECTION {
98         SHIFT_LEFT = 0,
99         SHIFT_RIGHT,
100 };
101
102 /*
103  * Flags used in mballoc's allocation_context flags field.
104  *
105  * Also used to show what's going on for debugging purposes when the
106  * flag field is exported via the traceport interface
107  */
108
109 /* prefer goal again. length */
110 #define EXT4_MB_HINT_MERGE              0x0001
111 /* blocks already reserved */
112 #define EXT4_MB_HINT_RESERVED           0x0002
113 /* metadata is being allocated */
114 #define EXT4_MB_HINT_METADATA           0x0004
115 /* first blocks in the file */
116 #define EXT4_MB_HINT_FIRST              0x0008
117 /* search for the best chunk */
118 #define EXT4_MB_HINT_BEST               0x0010
119 /* data is being allocated */
120 #define EXT4_MB_HINT_DATA               0x0020
121 /* don't preallocate (for tails) */
122 #define EXT4_MB_HINT_NOPREALLOC         0x0040
123 /* allocate for locality group */
124 #define EXT4_MB_HINT_GROUP_ALLOC        0x0080
125 /* allocate goal blocks or none */
126 #define EXT4_MB_HINT_GOAL_ONLY          0x0100
127 /* goal is meaningful */
128 #define EXT4_MB_HINT_TRY_GOAL           0x0200
129 /* blocks already pre-reserved by delayed allocation */
130 #define EXT4_MB_DELALLOC_RESERVED       0x0400
131 /* We are doing stream allocation */
132 #define EXT4_MB_STREAM_ALLOC            0x0800
133 /* Use reserved root blocks if needed */
134 #define EXT4_MB_USE_ROOT_BLOCKS         0x1000
135 /* Use blocks from reserved pool */
136 #define EXT4_MB_USE_RESERVED            0x2000
137
138 struct ext4_allocation_request {
139         /* target inode for block we're allocating */
140         struct inode *inode;
141         /* how many blocks we want to allocate */
142         unsigned int len;
143         /* logical block in target inode */
144         ext4_lblk_t logical;
145         /* the closest logical allocated block to the left */
146         ext4_lblk_t lleft;
147         /* the closest logical allocated block to the right */
148         ext4_lblk_t lright;
149         /* phys. target (a hint) */
150         ext4_fsblk_t goal;
151         /* phys. block for the closest logical allocated block to the left */
152         ext4_fsblk_t pleft;
153         /* phys. block for the closest logical allocated block to the right */
154         ext4_fsblk_t pright;
155         /* flags. see above EXT4_MB_HINT_* */
156         unsigned int flags;
157 };
158
159 /*
160  * Logical to physical block mapping, used by ext4_map_blocks()
161  *
162  * This structure is used to pass requests into ext4_map_blocks() as
163  * well as to store the information returned by ext4_map_blocks().  It
164  * takes less room on the stack than a struct buffer_head.
165  */
166 #define EXT4_MAP_NEW            (1 << BH_New)
167 #define EXT4_MAP_MAPPED         (1 << BH_Mapped)
168 #define EXT4_MAP_UNWRITTEN      (1 << BH_Unwritten)
169 #define EXT4_MAP_BOUNDARY       (1 << BH_Boundary)
170 #define EXT4_MAP_FLAGS          (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
171                                  EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
172
173 struct ext4_map_blocks {
174         ext4_fsblk_t m_pblk;
175         ext4_lblk_t m_lblk;
176         unsigned int m_len;
177         unsigned int m_flags;
178 };
179
180 /*
181  * Flags for ext4_io_end->flags
182  */
183 #define EXT4_IO_END_UNWRITTEN   0x0001
184
185 /*
186  * For converting unwritten extents on a work queue. 'handle' is used for
187  * buffered writeback.
188  */
189 typedef struct ext4_io_end {
190         struct list_head        list;           /* per-file finished IO list */
191         handle_t                *handle;        /* handle reserved for extent
192                                                  * conversion */
193         struct inode            *inode;         /* file being written to */
194         struct bio              *bio;           /* Linked list of completed
195                                                  * bios covering the extent */
196         unsigned int            flag;           /* unwritten or not */
197         atomic_t                count;          /* reference counter */
198         loff_t                  offset;         /* offset in the file */
199         ssize_t                 size;           /* size of the extent */
200 } ext4_io_end_t;
201
202 struct ext4_io_submit {
203         struct writeback_control *io_wbc;
204         struct bio              *io_bio;
205         ext4_io_end_t           *io_end;
206         sector_t                io_next_block;
207 };
208
209 /*
210  * Special inodes numbers
211  */
212 #define EXT4_BAD_INO             1      /* Bad blocks inode */
213 #define EXT4_ROOT_INO            2      /* Root inode */
214 #define EXT4_USR_QUOTA_INO       3      /* User quota inode */
215 #define EXT4_GRP_QUOTA_INO       4      /* Group quota inode */
216 #define EXT4_BOOT_LOADER_INO     5      /* Boot loader inode */
217 #define EXT4_UNDEL_DIR_INO       6      /* Undelete directory inode */
218 #define EXT4_RESIZE_INO          7      /* Reserved group descriptors inode */
219 #define EXT4_JOURNAL_INO         8      /* Journal inode */
220
221 /* First non-reserved inode for old ext4 filesystems */
222 #define EXT4_GOOD_OLD_FIRST_INO 11
223
224 /*
225  * Maximal count of links to a file
226  */
227 #define EXT4_LINK_MAX           65000
228
229 /*
230  * Macro-instructions used to manage several block sizes
231  */
232 #define EXT4_MIN_BLOCK_SIZE             1024
233 #define EXT4_MAX_BLOCK_SIZE             65536
234 #define EXT4_MIN_BLOCK_LOG_SIZE         10
235 #define EXT4_MAX_BLOCK_LOG_SIZE         16
236 #ifdef __KERNEL__
237 # define EXT4_BLOCK_SIZE(s)             ((s)->s_blocksize)
238 #else
239 # define EXT4_BLOCK_SIZE(s)             (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
240 #endif
241 #define EXT4_ADDR_PER_BLOCK(s)          (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
242 #define EXT4_CLUSTER_SIZE(s)            (EXT4_BLOCK_SIZE(s) << \
243                                          EXT4_SB(s)->s_cluster_bits)
244 #ifdef __KERNEL__
245 # define EXT4_BLOCK_SIZE_BITS(s)        ((s)->s_blocksize_bits)
246 # define EXT4_CLUSTER_BITS(s)           (EXT4_SB(s)->s_cluster_bits)
247 #else
248 # define EXT4_BLOCK_SIZE_BITS(s)        ((s)->s_log_block_size + 10)
249 #endif
250 #ifdef __KERNEL__
251 #define EXT4_ADDR_PER_BLOCK_BITS(s)     (EXT4_SB(s)->s_addr_per_block_bits)
252 #define EXT4_INODE_SIZE(s)              (EXT4_SB(s)->s_inode_size)
253 #define EXT4_FIRST_INO(s)               (EXT4_SB(s)->s_first_ino)
254 #else
255 #define EXT4_INODE_SIZE(s)      (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
256                                  EXT4_GOOD_OLD_INODE_SIZE : \
257                                  (s)->s_inode_size)
258 #define EXT4_FIRST_INO(s)       (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
259                                  EXT4_GOOD_OLD_FIRST_INO : \
260                                  (s)->s_first_ino)
261 #endif
262 #define EXT4_BLOCK_ALIGN(size, blkbits)         ALIGN((size), (1 << (blkbits)))
263
264 /* Translate a block number to a cluster number */
265 #define EXT4_B2C(sbi, blk)      ((blk) >> (sbi)->s_cluster_bits)
266 /* Translate a cluster number to a block number */
267 #define EXT4_C2B(sbi, cluster)  ((cluster) << (sbi)->s_cluster_bits)
268 /* Translate # of blks to # of clusters */
269 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
270                                  (sbi)->s_cluster_bits)
271 /* Mask out the low bits to get the starting block of the cluster */
272 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) &                              \
273                                   ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
274 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) &                              \
275                                   ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
276 /* Get the cluster offset */
277 #define EXT4_PBLK_COFF(s, pblk) ((pblk) &                               \
278                                  ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
279 #define EXT4_LBLK_COFF(s, lblk) ((lblk) &                               \
280                                  ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
281
282 /*
283  * Structure of a blocks group descriptor
284  */
285 struct ext4_group_desc
286 {
287         __le32  bg_block_bitmap_lo;     /* Blocks bitmap block */
288         __le32  bg_inode_bitmap_lo;     /* Inodes bitmap block */
289         __le32  bg_inode_table_lo;      /* Inodes table block */
290         __le16  bg_free_blocks_count_lo;/* Free blocks count */
291         __le16  bg_free_inodes_count_lo;/* Free inodes count */
292         __le16  bg_used_dirs_count_lo;  /* Directories count */
293         __le16  bg_flags;               /* EXT4_BG_flags (INODE_UNINIT, etc) */
294         __le32  bg_exclude_bitmap_lo;   /* Exclude bitmap for snapshots */
295         __le16  bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
296         __le16  bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
297         __le16  bg_itable_unused_lo;    /* Unused inodes count */
298         __le16  bg_checksum;            /* crc16(sb_uuid+group+desc) */
299         __le32  bg_block_bitmap_hi;     /* Blocks bitmap block MSB */
300         __le32  bg_inode_bitmap_hi;     /* Inodes bitmap block MSB */
301         __le32  bg_inode_table_hi;      /* Inodes table block MSB */
302         __le16  bg_free_blocks_count_hi;/* Free blocks count MSB */
303         __le16  bg_free_inodes_count_hi;/* Free inodes count MSB */
304         __le16  bg_used_dirs_count_hi;  /* Directories count MSB */
305         __le16  bg_itable_unused_hi;    /* Unused inodes count MSB */
306         __le32  bg_exclude_bitmap_hi;   /* Exclude bitmap block MSB */
307         __le16  bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
308         __le16  bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
309         __u32   bg_reserved;
310 };
311
312 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END        \
313         (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
314          sizeof(__le16))
315 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END        \
316         (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
317          sizeof(__le16))
318
319 /*
320  * Structure of a flex block group info
321  */
322
323 struct flex_groups {
324         atomic64_t      free_clusters;
325         atomic_t        free_inodes;
326         atomic_t        used_dirs;
327 };
328
329 #define EXT4_BG_INODE_UNINIT    0x0001 /* Inode table/bitmap not in use */
330 #define EXT4_BG_BLOCK_UNINIT    0x0002 /* Block bitmap not in use */
331 #define EXT4_BG_INODE_ZEROED    0x0004 /* On-disk itable initialized to zero */
332
333 /*
334  * Macro-instructions used to manage group descriptors
335  */
336 #define EXT4_MIN_DESC_SIZE              32
337 #define EXT4_MIN_DESC_SIZE_64BIT        64
338 #define EXT4_MAX_DESC_SIZE              EXT4_MIN_BLOCK_SIZE
339 #define EXT4_DESC_SIZE(s)               (EXT4_SB(s)->s_desc_size)
340 #ifdef __KERNEL__
341 # define EXT4_BLOCKS_PER_GROUP(s)       (EXT4_SB(s)->s_blocks_per_group)
342 # define EXT4_CLUSTERS_PER_GROUP(s)     (EXT4_SB(s)->s_clusters_per_group)
343 # define EXT4_DESC_PER_BLOCK(s)         (EXT4_SB(s)->s_desc_per_block)
344 # define EXT4_INODES_PER_GROUP(s)       (EXT4_SB(s)->s_inodes_per_group)
345 # define EXT4_DESC_PER_BLOCK_BITS(s)    (EXT4_SB(s)->s_desc_per_block_bits)
346 #else
347 # define EXT4_BLOCKS_PER_GROUP(s)       ((s)->s_blocks_per_group)
348 # define EXT4_DESC_PER_BLOCK(s)         (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
349 # define EXT4_INODES_PER_GROUP(s)       ((s)->s_inodes_per_group)
350 #endif
351
352 /*
353  * Constants relative to the data blocks
354  */
355 #define EXT4_NDIR_BLOCKS                12
356 #define EXT4_IND_BLOCK                  EXT4_NDIR_BLOCKS
357 #define EXT4_DIND_BLOCK                 (EXT4_IND_BLOCK + 1)
358 #define EXT4_TIND_BLOCK                 (EXT4_DIND_BLOCK + 1)
359 #define EXT4_N_BLOCKS                   (EXT4_TIND_BLOCK + 1)
360
361 /*
362  * Inode flags
363  */
364 #define EXT4_SECRM_FL                   0x00000001 /* Secure deletion */
365 #define EXT4_UNRM_FL                    0x00000002 /* Undelete */
366 #define EXT4_COMPR_FL                   0x00000004 /* Compress file */
367 #define EXT4_SYNC_FL                    0x00000008 /* Synchronous updates */
368 #define EXT4_IMMUTABLE_FL               0x00000010 /* Immutable file */
369 #define EXT4_APPEND_FL                  0x00000020 /* writes to file may only append */
370 #define EXT4_NODUMP_FL                  0x00000040 /* do not dump file */
371 #define EXT4_NOATIME_FL                 0x00000080 /* do not update atime */
372 /* Reserved for compression usage... */
373 #define EXT4_DIRTY_FL                   0x00000100
374 #define EXT4_COMPRBLK_FL                0x00000200 /* One or more compressed clusters */
375 #define EXT4_NOCOMPR_FL                 0x00000400 /* Don't compress */
376         /* nb: was previously EXT2_ECOMPR_FL */
377 #define EXT4_ENCRYPT_FL                 0x00000800 /* encrypted file */
378 /* End compression flags --- maybe not all used */
379 #define EXT4_INDEX_FL                   0x00001000 /* hash-indexed directory */
380 #define EXT4_IMAGIC_FL                  0x00002000 /* AFS directory */
381 #define EXT4_JOURNAL_DATA_FL            0x00004000 /* file data should be journaled */
382 #define EXT4_NOTAIL_FL                  0x00008000 /* file tail should not be merged */
383 #define EXT4_DIRSYNC_FL                 0x00010000 /* dirsync behaviour (directories only) */
384 #define EXT4_TOPDIR_FL                  0x00020000 /* Top of directory hierarchies*/
385 #define EXT4_HUGE_FILE_FL               0x00040000 /* Set to each huge file */
386 #define EXT4_EXTENTS_FL                 0x00080000 /* Inode uses extents */
387 #define EXT4_EA_INODE_FL                0x00200000 /* Inode used for large EA */
388 #define EXT4_EOFBLOCKS_FL               0x00400000 /* Blocks allocated beyond EOF */
389 #define EXT4_INLINE_DATA_FL             0x10000000 /* Inode has inline data. */
390 #define EXT4_PROJINHERIT_FL             0x20000000 /* Create with parents projid */
391 #define EXT4_RESERVED_FL                0x80000000 /* reserved for ext4 lib */
392
393 #define EXT4_FL_USER_VISIBLE            0x304BDFFF /* User visible flags */
394 #define EXT4_FL_USER_MODIFIABLE         0x204380FF /* User modifiable flags */
395
396 #define EXT4_FL_XFLAG_VISIBLE           (EXT4_SYNC_FL | \
397                                          EXT4_IMMUTABLE_FL | \
398                                          EXT4_APPEND_FL | \
399                                          EXT4_NODUMP_FL | \
400                                          EXT4_NOATIME_FL | \
401                                          EXT4_PROJINHERIT_FL)
402
403 /* Flags that should be inherited by new inodes from their parent. */
404 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
405                            EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
406                            EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
407                            EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
408                            EXT4_PROJINHERIT_FL)
409
410 /* Flags that are appropriate for regular files (all but dir-specific ones). */
411 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
412
413 /* Flags that are appropriate for non-directories/regular files. */
414 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
415
416 /* Mask out flags that are inappropriate for the given type of inode. */
417 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
418 {
419         if (S_ISDIR(mode))
420                 return flags;
421         else if (S_ISREG(mode))
422                 return flags & EXT4_REG_FLMASK;
423         else
424                 return flags & EXT4_OTHER_FLMASK;
425 }
426
427 /*
428  * Inode flags used for atomic set/get
429  */
430 enum {
431         EXT4_INODE_SECRM        = 0,    /* Secure deletion */
432         EXT4_INODE_UNRM         = 1,    /* Undelete */
433         EXT4_INODE_COMPR        = 2,    /* Compress file */
434         EXT4_INODE_SYNC         = 3,    /* Synchronous updates */
435         EXT4_INODE_IMMUTABLE    = 4,    /* Immutable file */
436         EXT4_INODE_APPEND       = 5,    /* writes to file may only append */
437         EXT4_INODE_NODUMP       = 6,    /* do not dump file */
438         EXT4_INODE_NOATIME      = 7,    /* do not update atime */
439 /* Reserved for compression usage... */
440         EXT4_INODE_DIRTY        = 8,
441         EXT4_INODE_COMPRBLK     = 9,    /* One or more compressed clusters */
442         EXT4_INODE_NOCOMPR      = 10,   /* Don't compress */
443         EXT4_INODE_ENCRYPT      = 11,   /* Encrypted file */
444 /* End compression flags --- maybe not all used */
445         EXT4_INODE_INDEX        = 12,   /* hash-indexed directory */
446         EXT4_INODE_IMAGIC       = 13,   /* AFS directory */
447         EXT4_INODE_JOURNAL_DATA = 14,   /* file data should be journaled */
448         EXT4_INODE_NOTAIL       = 15,   /* file tail should not be merged */
449         EXT4_INODE_DIRSYNC      = 16,   /* dirsync behaviour (directories only) */
450         EXT4_INODE_TOPDIR       = 17,   /* Top of directory hierarchies*/
451         EXT4_INODE_HUGE_FILE    = 18,   /* Set to each huge file */
452         EXT4_INODE_EXTENTS      = 19,   /* Inode uses extents */
453         EXT4_INODE_EA_INODE     = 21,   /* Inode used for large EA */
454         EXT4_INODE_EOFBLOCKS    = 22,   /* Blocks allocated beyond EOF */
455         EXT4_INODE_INLINE_DATA  = 28,   /* Data in inode. */
456         EXT4_INODE_PROJINHERIT  = 29,   /* Create with parents projid */
457         EXT4_INODE_RESERVED     = 31,   /* reserved for ext4 lib */
458 };
459
460 /*
461  * Since it's pretty easy to mix up bit numbers and hex values, we use a
462  * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
463  * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
464  * any extra space in the compiled kernel image, otherwise, the build will fail.
465  * It's important that these values are the same, since we are using
466  * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
467  * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
468  * values found in ext2, ext3 and ext4 filesystems, and of course the values
469  * defined in e2fsprogs.
470  *
471  * It's not paranoia if the Murphy's Law really *is* out to get you.  :-)
472  */
473 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
474 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
475
476 static inline void ext4_check_flag_values(void)
477 {
478         CHECK_FLAG_VALUE(SECRM);
479         CHECK_FLAG_VALUE(UNRM);
480         CHECK_FLAG_VALUE(COMPR);
481         CHECK_FLAG_VALUE(SYNC);
482         CHECK_FLAG_VALUE(IMMUTABLE);
483         CHECK_FLAG_VALUE(APPEND);
484         CHECK_FLAG_VALUE(NODUMP);
485         CHECK_FLAG_VALUE(NOATIME);
486         CHECK_FLAG_VALUE(DIRTY);
487         CHECK_FLAG_VALUE(COMPRBLK);
488         CHECK_FLAG_VALUE(NOCOMPR);
489         CHECK_FLAG_VALUE(ENCRYPT);
490         CHECK_FLAG_VALUE(INDEX);
491         CHECK_FLAG_VALUE(IMAGIC);
492         CHECK_FLAG_VALUE(JOURNAL_DATA);
493         CHECK_FLAG_VALUE(NOTAIL);
494         CHECK_FLAG_VALUE(DIRSYNC);
495         CHECK_FLAG_VALUE(TOPDIR);
496         CHECK_FLAG_VALUE(HUGE_FILE);
497         CHECK_FLAG_VALUE(EXTENTS);
498         CHECK_FLAG_VALUE(EA_INODE);
499         CHECK_FLAG_VALUE(EOFBLOCKS);
500         CHECK_FLAG_VALUE(INLINE_DATA);
501         CHECK_FLAG_VALUE(PROJINHERIT);
502         CHECK_FLAG_VALUE(RESERVED);
503 }
504
505 /* Used to pass group descriptor data when online resize is done */
506 struct ext4_new_group_input {
507         __u32 group;            /* Group number for this data */
508         __u64 block_bitmap;     /* Absolute block number of block bitmap */
509         __u64 inode_bitmap;     /* Absolute block number of inode bitmap */
510         __u64 inode_table;      /* Absolute block number of inode table start */
511         __u32 blocks_count;     /* Total number of blocks in this group */
512         __u16 reserved_blocks;  /* Number of reserved blocks in this group */
513         __u16 unused;
514 };
515
516 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
517 struct compat_ext4_new_group_input {
518         u32 group;
519         compat_u64 block_bitmap;
520         compat_u64 inode_bitmap;
521         compat_u64 inode_table;
522         u32 blocks_count;
523         u16 reserved_blocks;
524         u16 unused;
525 };
526 #endif
527
528 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
529 struct ext4_new_group_data {
530         __u32 group;
531         __u64 block_bitmap;
532         __u64 inode_bitmap;
533         __u64 inode_table;
534         __u32 blocks_count;
535         __u16 reserved_blocks;
536         __u16 unused;
537         __u32 free_blocks_count;
538 };
539
540 /* Indexes used to index group tables in ext4_new_group_data */
541 enum {
542         BLOCK_BITMAP = 0,       /* block bitmap */
543         INODE_BITMAP,           /* inode bitmap */
544         INODE_TABLE,            /* inode tables */
545         GROUP_TABLE_COUNT,
546 };
547
548 /*
549  * Flags used by ext4_map_blocks()
550  */
551         /* Allocate any needed blocks and/or convert an unwritten
552            extent to be an initialized ext4 */
553 #define EXT4_GET_BLOCKS_CREATE                  0x0001
554         /* Request the creation of an unwritten extent */
555 #define EXT4_GET_BLOCKS_UNWRIT_EXT              0x0002
556 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT       (EXT4_GET_BLOCKS_UNWRIT_EXT|\
557                                                  EXT4_GET_BLOCKS_CREATE)
558         /* Caller is from the delayed allocation writeout path
559          * finally doing the actual allocation of delayed blocks */
560 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE        0x0004
561         /* caller is from the direct IO path, request to creation of an
562         unwritten extents if not allocated, split the unwritten
563         extent if blocks has been preallocated already*/
564 #define EXT4_GET_BLOCKS_PRE_IO                  0x0008
565 #define EXT4_GET_BLOCKS_CONVERT                 0x0010
566 #define EXT4_GET_BLOCKS_IO_CREATE_EXT           (EXT4_GET_BLOCKS_PRE_IO|\
567                                          EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
568         /* Convert extent to initialized after IO complete */
569 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT          (EXT4_GET_BLOCKS_CONVERT|\
570                                          EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
571         /* Eventual metadata allocation (due to growing extent tree)
572          * should not fail, so try to use reserved blocks for that.*/
573 #define EXT4_GET_BLOCKS_METADATA_NOFAIL         0x0020
574         /* Don't normalize allocation size (used for fallocate) */
575 #define EXT4_GET_BLOCKS_NO_NORMALIZE            0x0040
576         /* Request will not result in inode size update (user for fallocate) */
577 #define EXT4_GET_BLOCKS_KEEP_SIZE               0x0080
578         /* Convert written extents to unwritten */
579 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN       0x0100
580         /* Write zeros to newly created written extents */
581 #define EXT4_GET_BLOCKS_ZERO                    0x0200
582 #define EXT4_GET_BLOCKS_CREATE_ZERO             (EXT4_GET_BLOCKS_CREATE |\
583                                         EXT4_GET_BLOCKS_ZERO)
584
585 /*
586  * The bit position of these flags must not overlap with any of the
587  * EXT4_GET_BLOCKS_*.  They are used by ext4_find_extent(),
588  * read_extent_tree_block(), ext4_split_extent_at(),
589  * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
590  * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
591  * caching the extents when reading from the extent tree while a
592  * truncate or punch hole operation is in progress.
593  */
594 #define EXT4_EX_NOCACHE                         0x40000000
595 #define EXT4_EX_FORCE_CACHE                     0x20000000
596
597 /*
598  * Flags used by ext4_free_blocks
599  */
600 #define EXT4_FREE_BLOCKS_METADATA       0x0001
601 #define EXT4_FREE_BLOCKS_FORGET         0x0002
602 #define EXT4_FREE_BLOCKS_VALIDATED      0x0004
603 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
604 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER   0x0010
605 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER    0x0020
606
607 /* Encryption algorithms */
608 #define EXT4_ENCRYPTION_MODE_INVALID            0
609 #define EXT4_ENCRYPTION_MODE_AES_256_XTS        1
610 #define EXT4_ENCRYPTION_MODE_AES_256_GCM        2
611 #define EXT4_ENCRYPTION_MODE_AES_256_CBC        3
612 #define EXT4_ENCRYPTION_MODE_AES_256_CTS        4
613
614 #include "ext4_crypto.h"
615
616 /*
617  * ioctl commands
618  */
619 #define EXT4_IOC_GETFLAGS               FS_IOC_GETFLAGS
620 #define EXT4_IOC_SETFLAGS               FS_IOC_SETFLAGS
621 #define EXT4_IOC_GETVERSION             _IOR('f', 3, long)
622 #define EXT4_IOC_SETVERSION             _IOW('f', 4, long)
623 #define EXT4_IOC_GETVERSION_OLD         FS_IOC_GETVERSION
624 #define EXT4_IOC_SETVERSION_OLD         FS_IOC_SETVERSION
625 #define EXT4_IOC_GETRSVSZ               _IOR('f', 5, long)
626 #define EXT4_IOC_SETRSVSZ               _IOW('f', 6, long)
627 #define EXT4_IOC_GROUP_EXTEND           _IOW('f', 7, unsigned long)
628 #define EXT4_IOC_GROUP_ADD              _IOW('f', 8, struct ext4_new_group_input)
629 #define EXT4_IOC_MIGRATE                _IO('f', 9)
630  /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
631  /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
632 #define EXT4_IOC_ALLOC_DA_BLKS          _IO('f', 12)
633 #define EXT4_IOC_MOVE_EXT               _IOWR('f', 15, struct move_extent)
634 #define EXT4_IOC_RESIZE_FS              _IOW('f', 16, __u64)
635 #define EXT4_IOC_SWAP_BOOT              _IO('f', 17)
636 #define EXT4_IOC_PRECACHE_EXTENTS       _IO('f', 18)
637 #define EXT4_IOC_SET_ENCRYPTION_POLICY  _IOR('f', 19, struct ext4_encryption_policy)
638 #define EXT4_IOC_GET_ENCRYPTION_PWSALT  _IOW('f', 20, __u8[16])
639 #define EXT4_IOC_GET_ENCRYPTION_POLICY  _IOW('f', 21, struct ext4_encryption_policy)
640
641 #ifndef FS_IOC_FSGETXATTR
642 /* Until the uapi changes get merged for project quota... */
643
644 #define FS_IOC_FSGETXATTR               _IOR('X', 31, struct fsxattr)
645 #define FS_IOC_FSSETXATTR               _IOW('X', 32, struct fsxattr)
646
647 /*
648  * Structure for FS_IOC_FSGETXATTR and FS_IOC_FSSETXATTR.
649  */
650 struct fsxattr {
651         __u32           fsx_xflags;     /* xflags field value (get/set) */
652         __u32           fsx_extsize;    /* extsize field value (get/set)*/
653         __u32           fsx_nextents;   /* nextents field value (get)   */
654         __u32           fsx_projid;     /* project identifier (get/set) */
655         unsigned char   fsx_pad[12];
656 };
657
658 /*
659  * Flags for the fsx_xflags field
660  */
661 #define FS_XFLAG_REALTIME       0x00000001      /* data in realtime volume */
662 #define FS_XFLAG_PREALLOC       0x00000002      /* preallocated file extents */
663 #define FS_XFLAG_IMMUTABLE      0x00000008      /* file cannot be modified */
664 #define FS_XFLAG_APPEND         0x00000010      /* all writes append */
665 #define FS_XFLAG_SYNC           0x00000020      /* all writes synchronous */
666 #define FS_XFLAG_NOATIME        0x00000040      /* do not update access time */
667 #define FS_XFLAG_NODUMP         0x00000080      /* do not include in backups */
668 #define FS_XFLAG_RTINHERIT      0x00000100      /* create with rt bit set */
669 #define FS_XFLAG_PROJINHERIT    0x00000200      /* create with parents projid */
670 #define FS_XFLAG_NOSYMLINKS     0x00000400      /* disallow symlink creation */
671 #define FS_XFLAG_EXTSIZE        0x00000800      /* extent size allocator hint */
672 #define FS_XFLAG_EXTSZINHERIT   0x00001000      /* inherit inode extent size */
673 #define FS_XFLAG_NODEFRAG       0x00002000      /* do not defragment */
674 #define FS_XFLAG_FILESTREAM     0x00004000      /* use filestream allocator */
675 #define FS_XFLAG_HASATTR        0x80000000      /* no DIFLAG for this */
676 #endif /* !defined(FS_IOC_FSGETXATTR) */
677
678 #define EXT4_IOC_FSGETXATTR             FS_IOC_FSGETXATTR
679 #define EXT4_IOC_FSSETXATTR             FS_IOC_FSSETXATTR
680
681 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
682 /*
683  * ioctl commands in 32 bit emulation
684  */
685 #define EXT4_IOC32_GETFLAGS             FS_IOC32_GETFLAGS
686 #define EXT4_IOC32_SETFLAGS             FS_IOC32_SETFLAGS
687 #define EXT4_IOC32_GETVERSION           _IOR('f', 3, int)
688 #define EXT4_IOC32_SETVERSION           _IOW('f', 4, int)
689 #define EXT4_IOC32_GETRSVSZ             _IOR('f', 5, int)
690 #define EXT4_IOC32_SETRSVSZ             _IOW('f', 6, int)
691 #define EXT4_IOC32_GROUP_EXTEND         _IOW('f', 7, unsigned int)
692 #define EXT4_IOC32_GROUP_ADD            _IOW('f', 8, struct compat_ext4_new_group_input)
693 #define EXT4_IOC32_GETVERSION_OLD       FS_IOC32_GETVERSION
694 #define EXT4_IOC32_SETVERSION_OLD       FS_IOC32_SETVERSION
695 #endif
696
697 /* Max physical block we can address w/o extents */
698 #define EXT4_MAX_BLOCK_FILE_PHYS        0xFFFFFFFF
699
700 /*
701  * Structure of an inode on the disk
702  */
703 struct ext4_inode {
704         __le16  i_mode;         /* File mode */
705         __le16  i_uid;          /* Low 16 bits of Owner Uid */
706         __le32  i_size_lo;      /* Size in bytes */
707         __le32  i_atime;        /* Access time */
708         __le32  i_ctime;        /* Inode Change time */
709         __le32  i_mtime;        /* Modification time */
710         __le32  i_dtime;        /* Deletion Time */
711         __le16  i_gid;          /* Low 16 bits of Group Id */
712         __le16  i_links_count;  /* Links count */
713         __le32  i_blocks_lo;    /* Blocks count */
714         __le32  i_flags;        /* File flags */
715         union {
716                 struct {
717                         __le32  l_i_version;
718                 } linux1;
719                 struct {
720                         __u32  h_i_translator;
721                 } hurd1;
722                 struct {
723                         __u32  m_i_reserved1;
724                 } masix1;
725         } osd1;                         /* OS dependent 1 */
726         __le32  i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
727         __le32  i_generation;   /* File version (for NFS) */
728         __le32  i_file_acl_lo;  /* File ACL */
729         __le32  i_size_high;
730         __le32  i_obso_faddr;   /* Obsoleted fragment address */
731         union {
732                 struct {
733                         __le16  l_i_blocks_high; /* were l_i_reserved1 */
734                         __le16  l_i_file_acl_high;
735                         __le16  l_i_uid_high;   /* these 2 fields */
736                         __le16  l_i_gid_high;   /* were reserved2[0] */
737                         __le16  l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
738                         __le16  l_i_reserved;
739                 } linux2;
740                 struct {
741                         __le16  h_i_reserved1;  /* Obsoleted fragment number/size which are removed in ext4 */
742                         __u16   h_i_mode_high;
743                         __u16   h_i_uid_high;
744                         __u16   h_i_gid_high;
745                         __u32   h_i_author;
746                 } hurd2;
747                 struct {
748                         __le16  h_i_reserved1;  /* Obsoleted fragment number/size which are removed in ext4 */
749                         __le16  m_i_file_acl_high;
750                         __u32   m_i_reserved2[2];
751                 } masix2;
752         } osd2;                         /* OS dependent 2 */
753         __le16  i_extra_isize;
754         __le16  i_checksum_hi;  /* crc32c(uuid+inum+inode) BE */
755         __le32  i_ctime_extra;  /* extra Change time      (nsec << 2 | epoch) */
756         __le32  i_mtime_extra;  /* extra Modification time(nsec << 2 | epoch) */
757         __le32  i_atime_extra;  /* extra Access time      (nsec << 2 | epoch) */
758         __le32  i_crtime;       /* File Creation time */
759         __le32  i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
760         __le32  i_version_hi;   /* high 32 bits for 64-bit version */
761         __le32  i_projid;       /* Project ID */
762 };
763
764 struct move_extent {
765         __u32 reserved;         /* should be zero */
766         __u32 donor_fd;         /* donor file descriptor */
767         __u64 orig_start;       /* logical start offset in block for orig */
768         __u64 donor_start;      /* logical start offset in block for donor */
769         __u64 len;              /* block length to be moved */
770         __u64 moved_len;        /* moved block length */
771 };
772
773 #define EXT4_EPOCH_BITS 2
774 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
775 #define EXT4_NSEC_MASK  (~0UL << EXT4_EPOCH_BITS)
776
777 /*
778  * Extended fields will fit into an inode if the filesystem was formatted
779  * with large inodes (-I 256 or larger) and there are not currently any EAs
780  * consuming all of the available space. For new inodes we always reserve
781  * enough space for the kernel's known extended fields, but for inodes
782  * created with an old kernel this might not have been the case. None of
783  * the extended inode fields is critical for correct filesystem operation.
784  * This macro checks if a certain field fits in the inode. Note that
785  * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
786  */
787 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field)   \
788         ((offsetof(typeof(*ext4_inode), field) +        \
789           sizeof((ext4_inode)->field))                  \
790         <= (EXT4_GOOD_OLD_INODE_SIZE +                  \
791             (einode)->i_extra_isize))                   \
792
793 /*
794  * We use an encoding that preserves the times for extra epoch "00":
795  *
796  * extra  msb of                         adjust for signed
797  * epoch  32-bit                         32-bit tv_sec to
798  * bits   time    decoded 64-bit tv_sec  64-bit tv_sec      valid time range
799  * 0 0    1    -0x80000000..-0x00000001  0x000000000 1901-12-13..1969-12-31
800  * 0 0    0    0x000000000..0x07fffffff  0x000000000 1970-01-01..2038-01-19
801  * 0 1    1    0x080000000..0x0ffffffff  0x100000000 2038-01-19..2106-02-07
802  * 0 1    0    0x100000000..0x17fffffff  0x100000000 2106-02-07..2174-02-25
803  * 1 0    1    0x180000000..0x1ffffffff  0x200000000 2174-02-25..2242-03-16
804  * 1 0    0    0x200000000..0x27fffffff  0x200000000 2242-03-16..2310-04-04
805  * 1 1    1    0x280000000..0x2ffffffff  0x300000000 2310-04-04..2378-04-22
806  * 1 1    0    0x300000000..0x37fffffff  0x300000000 2378-04-22..2446-05-10
807  *
808  * Note that previous versions of the kernel on 64-bit systems would
809  * incorrectly use extra epoch bits 1,1 for dates between 1901 and
810  * 1970.  e2fsck will correct this, assuming that it is run on the
811  * affected filesystem before 2242.
812  */
813
814 static inline __le32 ext4_encode_extra_time(struct timespec *time)
815 {
816         u32 extra = sizeof(time->tv_sec) > 4 ?
817                 ((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK : 0;
818         return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
819 }
820
821 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
822 {
823         if (unlikely(sizeof(time->tv_sec) > 4 &&
824                         (extra & cpu_to_le32(EXT4_EPOCH_MASK)))) {
825 #if LINUX_VERSION_CODE < KERNEL_VERSION(4,20,0)
826                 /* Handle legacy encoding of pre-1970 dates with epoch
827                  * bits 1,1.  We assume that by kernel version 4.20,
828                  * everyone will have run fsck over the affected
829                  * filesystems to correct the problem.  (This
830                  * backwards compatibility may be removed before this
831                  * time, at the discretion of the ext4 developers.)
832                  */
833                 u64 extra_bits = le32_to_cpu(extra) & EXT4_EPOCH_MASK;
834                 if (extra_bits == 3 && ((time->tv_sec) & 0x80000000) != 0)
835                         extra_bits = 0;
836                 time->tv_sec += extra_bits << 32;
837 #else
838                 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
839 #endif
840         }
841         time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
842 }
843
844 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode)                          \
845 do {                                                                           \
846         (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec);               \
847         if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     \
848                 (raw_inode)->xtime ## _extra =                                 \
849                                 ext4_encode_extra_time(&(inode)->xtime);       \
850 } while (0)
851
852 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode)                        \
853 do {                                                                           \
854         if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))                      \
855                 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec);      \
856         if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))            \
857                 (raw_inode)->xtime ## _extra =                                 \
858                                 ext4_encode_extra_time(&(einode)->xtime);      \
859 } while (0)
860
861 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode)                          \
862 do {                                                                           \
863         (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime);       \
864         if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     \
865                 ext4_decode_extra_time(&(inode)->xtime,                        \
866                                        raw_inode->xtime ## _extra);            \
867         else                                                                   \
868                 (inode)->xtime.tv_nsec = 0;                                    \
869 } while (0)
870
871 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode)                        \
872 do {                                                                           \
873         if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))                      \
874                 (einode)->xtime.tv_sec =                                       \
875                         (signed)le32_to_cpu((raw_inode)->xtime);               \
876         else                                                                   \
877                 (einode)->xtime.tv_sec = 0;                                    \
878         if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))            \
879                 ext4_decode_extra_time(&(einode)->xtime,                       \
880                                        raw_inode->xtime ## _extra);            \
881         else                                                                   \
882                 (einode)->xtime.tv_nsec = 0;                                   \
883 } while (0)
884
885 #define i_disk_version osd1.linux1.l_i_version
886
887 #if defined(__KERNEL__) || defined(__linux__)
888 #define i_reserved1     osd1.linux1.l_i_reserved1
889 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
890 #define i_blocks_high   osd2.linux2.l_i_blocks_high
891 #define i_uid_low       i_uid
892 #define i_gid_low       i_gid
893 #define i_uid_high      osd2.linux2.l_i_uid_high
894 #define i_gid_high      osd2.linux2.l_i_gid_high
895 #define i_checksum_lo   osd2.linux2.l_i_checksum_lo
896
897 #elif defined(__GNU__)
898
899 #define i_translator    osd1.hurd1.h_i_translator
900 #define i_uid_high      osd2.hurd2.h_i_uid_high
901 #define i_gid_high      osd2.hurd2.h_i_gid_high
902 #define i_author        osd2.hurd2.h_i_author
903
904 #elif defined(__masix__)
905
906 #define i_reserved1     osd1.masix1.m_i_reserved1
907 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
908 #define i_reserved2     osd2.masix2.m_i_reserved2
909
910 #endif /* defined(__KERNEL__) || defined(__linux__) */
911
912 #include "extents_status.h"
913
914 /*
915  * Lock subclasses for i_data_sem in the ext4_inode_info structure.
916  *
917  * These are needed to avoid lockdep false positives when we need to
918  * allocate blocks to the quota inode during ext4_map_blocks(), while
919  * holding i_data_sem for a normal (non-quota) inode.  Since we don't
920  * do quota tracking for the quota inode, this avoids deadlock (as
921  * well as infinite recursion, since it isn't turtles all the way
922  * down...)
923  *
924  *  I_DATA_SEM_NORMAL - Used for most inodes
925  *  I_DATA_SEM_OTHER  - Used by move_inode.c for the second normal inode
926  *                        where the second inode has larger inode number
927  *                        than the first
928  *  I_DATA_SEM_QUOTA  - Used for quota inodes only
929  */
930 enum {
931         I_DATA_SEM_NORMAL = 0,
932         I_DATA_SEM_OTHER,
933         I_DATA_SEM_QUOTA,
934 };
935
936
937 /*
938  * fourth extended file system inode data in memory
939  */
940 struct ext4_inode_info {
941         __le32  i_data[15];     /* unconverted */
942         __u32   i_dtime;
943         ext4_fsblk_t    i_file_acl;
944
945         /*
946          * i_block_group is the number of the block group which contains
947          * this file's inode.  Constant across the lifetime of the inode,
948          * it is ued for making block allocation decisions - we try to
949          * place a file's data blocks near its inode block, and new inodes
950          * near to their parent directory's inode.
951          */
952         ext4_group_t    i_block_group;
953         ext4_lblk_t     i_dir_start_lookup;
954 #if (BITS_PER_LONG < 64)
955         unsigned long   i_state_flags;          /* Dynamic state flags */
956 #endif
957         unsigned long   i_flags;
958
959         /*
960          * Extended attributes can be read independently of the main file
961          * data. Taking i_mutex even when reading would cause contention
962          * between readers of EAs and writers of regular file data, so
963          * instead we synchronize on xattr_sem when reading or changing
964          * EAs.
965          */
966         struct rw_semaphore xattr_sem;
967
968         struct list_head i_orphan;      /* unlinked but open inodes */
969
970         /*
971          * i_disksize keeps track of what the inode size is ON DISK, not
972          * in memory.  During truncate, i_size is set to the new size by
973          * the VFS prior to calling ext4_truncate(), but the filesystem won't
974          * set i_disksize to 0 until the truncate is actually under way.
975          *
976          * The intent is that i_disksize always represents the blocks which
977          * are used by this file.  This allows recovery to restart truncate
978          * on orphans if we crash during truncate.  We actually write i_disksize
979          * into the on-disk inode when writing inodes out, instead of i_size.
980          *
981          * The only time when i_disksize and i_size may be different is when
982          * a truncate is in progress.  The only things which change i_disksize
983          * are ext4_get_block (growth) and ext4_truncate (shrinkth).
984          */
985         loff_t  i_disksize;
986
987         /*
988          * i_data_sem is for serialising ext4_truncate() against
989          * ext4_getblock().  In the 2.4 ext2 design, great chunks of inode's
990          * data tree are chopped off during truncate. We can't do that in
991          * ext4 because whenever we perform intermediate commits during
992          * truncate, the inode and all the metadata blocks *must* be in a
993          * consistent state which allows truncation of the orphans to restart
994          * during recovery.  Hence we must fix the get_block-vs-truncate race
995          * by other means, so we have i_data_sem.
996          */
997         struct rw_semaphore i_data_sem;
998         /*
999          * i_mmap_sem is for serializing page faults with truncate / punch hole
1000          * operations. We have to make sure that new page cannot be faulted in
1001          * a section of the inode that is being punched. We cannot easily use
1002          * i_data_sem for this since we need protection for the whole punch
1003          * operation and i_data_sem ranks below transaction start so we have
1004          * to occasionally drop it.
1005          */
1006         struct rw_semaphore i_mmap_sem;
1007         struct inode vfs_inode;
1008         struct jbd2_inode *jinode;
1009
1010         spinlock_t i_raw_lock;  /* protects updates to the raw inode */
1011
1012         /*
1013          * File creation time. Its function is same as that of
1014          * struct timespec i_{a,c,m}time in the generic inode.
1015          */
1016         struct timespec i_crtime;
1017
1018         /* mballoc */
1019         struct list_head i_prealloc_list;
1020         spinlock_t i_prealloc_lock;
1021
1022         /* extents status tree */
1023         struct ext4_es_tree i_es_tree;
1024         rwlock_t i_es_lock;
1025         struct list_head i_es_list;
1026         unsigned int i_es_all_nr;       /* protected by i_es_lock */
1027         unsigned int i_es_shk_nr;       /* protected by i_es_lock */
1028         ext4_lblk_t i_es_shrink_lblk;   /* Offset where we start searching for
1029                                            extents to shrink. Protected by
1030                                            i_es_lock  */
1031
1032         /* ialloc */
1033         ext4_group_t    i_last_alloc_group;
1034
1035         /* allocation reservation info for delalloc */
1036         /* In case of bigalloc, these refer to clusters rather than blocks */
1037         unsigned int i_reserved_data_blocks;
1038         unsigned int i_reserved_meta_blocks;
1039         unsigned int i_allocated_meta_blocks;
1040         ext4_lblk_t i_da_metadata_calc_last_lblock;
1041         int i_da_metadata_calc_len;
1042
1043         /* on-disk additional length */
1044         __u16 i_extra_isize;
1045
1046         /* Indicate the inline data space. */
1047         u16 i_inline_off;
1048         u16 i_inline_size;
1049
1050 #ifdef CONFIG_QUOTA
1051         /* quota space reservation, managed internally by quota code */
1052         qsize_t i_reserved_quota;
1053 #endif
1054
1055         /* Lock protecting lists below */
1056         spinlock_t i_completed_io_lock;
1057         /*
1058          * Completed IOs that need unwritten extents handling and have
1059          * transaction reserved
1060          */
1061         struct list_head i_rsv_conversion_list;
1062         struct work_struct i_rsv_conversion_work;
1063         atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1064
1065         spinlock_t i_block_reservation_lock;
1066
1067         /*
1068          * Transactions that contain inode's metadata needed to complete
1069          * fsync and fdatasync, respectively.
1070          */
1071         tid_t i_sync_tid;
1072         tid_t i_datasync_tid;
1073
1074 #ifdef CONFIG_QUOTA
1075         struct dquot *i_dquot[MAXQUOTAS];
1076 #endif
1077
1078         /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1079         __u32 i_csum_seed;
1080
1081 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1082         /* Encryption params */
1083         struct ext4_crypt_info *i_crypt_info;
1084 #endif
1085         kprojid_t i_projid;
1086 };
1087
1088 /*
1089  * File system states
1090  */
1091 #define EXT4_VALID_FS                   0x0001  /* Unmounted cleanly */
1092 #define EXT4_ERROR_FS                   0x0002  /* Errors detected */
1093 #define EXT4_ORPHAN_FS                  0x0004  /* Orphans being recovered */
1094
1095 /*
1096  * Misc. filesystem flags
1097  */
1098 #define EXT2_FLAGS_SIGNED_HASH          0x0001  /* Signed dirhash in use */
1099 #define EXT2_FLAGS_UNSIGNED_HASH        0x0002  /* Unsigned dirhash in use */
1100 #define EXT2_FLAGS_TEST_FILESYS         0x0004  /* to test development code */
1101
1102 /*
1103  * Mount flags set via mount options or defaults
1104  */
1105 #define EXT4_MOUNT_GRPID                0x00004 /* Create files with directory's group */
1106 #define EXT4_MOUNT_DEBUG                0x00008 /* Some debugging messages */
1107 #define EXT4_MOUNT_ERRORS_CONT          0x00010 /* Continue on errors */
1108 #define EXT4_MOUNT_ERRORS_RO            0x00020 /* Remount fs ro on errors */
1109 #define EXT4_MOUNT_ERRORS_PANIC         0x00040 /* Panic on errors */
1110 #define EXT4_MOUNT_ERRORS_MASK          0x00070
1111 #define EXT4_MOUNT_MINIX_DF             0x00080 /* Mimics the Minix statfs */
1112 #define EXT4_MOUNT_NOLOAD               0x00100 /* Don't use existing journal*/
1113 #ifdef CONFIG_FS_DAX
1114 #define EXT4_MOUNT_DAX                  0x00200 /* Direct Access */
1115 #else
1116 #define EXT4_MOUNT_DAX                  0
1117 #endif
1118 #define EXT4_MOUNT_DATA_FLAGS           0x00C00 /* Mode for data writes: */
1119 #define EXT4_MOUNT_JOURNAL_DATA         0x00400 /* Write data to journal */
1120 #define EXT4_MOUNT_ORDERED_DATA         0x00800 /* Flush data before commit */
1121 #define EXT4_MOUNT_WRITEBACK_DATA       0x00C00 /* No data ordering */
1122 #define EXT4_MOUNT_UPDATE_JOURNAL       0x01000 /* Update the journal format */
1123 #define EXT4_MOUNT_NO_UID32             0x02000  /* Disable 32-bit UIDs */
1124 #define EXT4_MOUNT_XATTR_USER           0x04000 /* Extended user attributes */
1125 #define EXT4_MOUNT_POSIX_ACL            0x08000 /* POSIX Access Control Lists */
1126 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC     0x10000 /* No auto delalloc mapping */
1127 #define EXT4_MOUNT_BARRIER              0x20000 /* Use block barriers */
1128 #define EXT4_MOUNT_QUOTA                0x80000 /* Some quota option set */
1129 #define EXT4_MOUNT_USRQUOTA             0x100000 /* "old" user quota */
1130 #define EXT4_MOUNT_GRPQUOTA             0x200000 /* "old" group quota */
1131 #define EXT4_MOUNT_DIOREAD_NOLOCK       0x400000 /* Enable support for dio read nolocking */
1132 #define EXT4_MOUNT_JOURNAL_CHECKSUM     0x800000 /* Journal checksums */
1133 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1134 #define EXT4_MOUNT_DELALLOC             0x8000000 /* Delalloc support */
1135 #define EXT4_MOUNT_DATA_ERR_ABORT       0x10000000 /* Abort on file data write */
1136 #define EXT4_MOUNT_BLOCK_VALIDITY       0x20000000 /* Block validity checking */
1137 #define EXT4_MOUNT_DISCARD              0x40000000 /* Issue DISCARD requests */
1138 #define EXT4_MOUNT_INIT_INODE_TABLE     0x80000000 /* Initialize uninitialized itables */
1139
1140 /*
1141  * Mount flags set either automatically (could not be set by mount option)
1142  * based on per file system feature or property or in special cases such as
1143  * distinguishing between explicit mount option definition and default.
1144  */
1145 #define EXT4_MOUNT2_EXPLICIT_DELALLOC   0x00000001 /* User explicitly
1146                                                       specified delalloc */
1147 #define EXT4_MOUNT2_STD_GROUP_SIZE      0x00000002 /* We have standard group
1148                                                       size of blocksize * 8
1149                                                       blocks */
1150 #define EXT4_MOUNT2_HURD_COMPAT         0x00000004 /* Support HURD-castrated
1151                                                       file systems */
1152
1153 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM   0x00000008 /* User explicitly
1154                                                 specified journal checksum */
1155
1156 #define clear_opt(sb, opt)              EXT4_SB(sb)->s_mount_opt &= \
1157                                                 ~EXT4_MOUNT_##opt
1158 #define set_opt(sb, opt)                EXT4_SB(sb)->s_mount_opt |= \
1159                                                 EXT4_MOUNT_##opt
1160 #define test_opt(sb, opt)               (EXT4_SB(sb)->s_mount_opt & \
1161                                          EXT4_MOUNT_##opt)
1162
1163 #define clear_opt2(sb, opt)             EXT4_SB(sb)->s_mount_opt2 &= \
1164                                                 ~EXT4_MOUNT2_##opt
1165 #define set_opt2(sb, opt)               EXT4_SB(sb)->s_mount_opt2 |= \
1166                                                 EXT4_MOUNT2_##opt
1167 #define test_opt2(sb, opt)              (EXT4_SB(sb)->s_mount_opt2 & \
1168                                          EXT4_MOUNT2_##opt)
1169
1170 #define ext4_test_and_set_bit           __test_and_set_bit_le
1171 #define ext4_set_bit                    __set_bit_le
1172 #define ext4_set_bit_atomic             ext2_set_bit_atomic
1173 #define ext4_test_and_clear_bit         __test_and_clear_bit_le
1174 #define ext4_clear_bit                  __clear_bit_le
1175 #define ext4_clear_bit_atomic           ext2_clear_bit_atomic
1176 #define ext4_test_bit                   test_bit_le
1177 #define ext4_find_next_zero_bit         find_next_zero_bit_le
1178 #define ext4_find_next_bit              find_next_bit_le
1179
1180 extern void ext4_set_bits(void *bm, int cur, int len);
1181
1182 /*
1183  * Maximal mount counts between two filesystem checks
1184  */
1185 #define EXT4_DFL_MAX_MNT_COUNT          20      /* Allow 20 mounts */
1186 #define EXT4_DFL_CHECKINTERVAL          0       /* Don't use interval check */
1187
1188 /*
1189  * Behaviour when detecting errors
1190  */
1191 #define EXT4_ERRORS_CONTINUE            1       /* Continue execution */
1192 #define EXT4_ERRORS_RO                  2       /* Remount fs read-only */
1193 #define EXT4_ERRORS_PANIC               3       /* Panic */
1194 #define EXT4_ERRORS_DEFAULT             EXT4_ERRORS_CONTINUE
1195
1196 /* Metadata checksum algorithm codes */
1197 #define EXT4_CRC32C_CHKSUM              1
1198
1199 /*
1200  * Structure of the super block
1201  */
1202 struct ext4_super_block {
1203 /*00*/  __le32  s_inodes_count;         /* Inodes count */
1204         __le32  s_blocks_count_lo;      /* Blocks count */
1205         __le32  s_r_blocks_count_lo;    /* Reserved blocks count */
1206         __le32  s_free_blocks_count_lo; /* Free blocks count */
1207 /*10*/  __le32  s_free_inodes_count;    /* Free inodes count */
1208         __le32  s_first_data_block;     /* First Data Block */
1209         __le32  s_log_block_size;       /* Block size */
1210         __le32  s_log_cluster_size;     /* Allocation cluster size */
1211 /*20*/  __le32  s_blocks_per_group;     /* # Blocks per group */
1212         __le32  s_clusters_per_group;   /* # Clusters per group */
1213         __le32  s_inodes_per_group;     /* # Inodes per group */
1214         __le32  s_mtime;                /* Mount time */
1215 /*30*/  __le32  s_wtime;                /* Write time */
1216         __le16  s_mnt_count;            /* Mount count */
1217         __le16  s_max_mnt_count;        /* Maximal mount count */
1218         __le16  s_magic;                /* Magic signature */
1219         __le16  s_state;                /* File system state */
1220         __le16  s_errors;               /* Behaviour when detecting errors */
1221         __le16  s_minor_rev_level;      /* minor revision level */
1222 /*40*/  __le32  s_lastcheck;            /* time of last check */
1223         __le32  s_checkinterval;        /* max. time between checks */
1224         __le32  s_creator_os;           /* OS */
1225         __le32  s_rev_level;            /* Revision level */
1226 /*50*/  __le16  s_def_resuid;           /* Default uid for reserved blocks */
1227         __le16  s_def_resgid;           /* Default gid for reserved blocks */
1228         /*
1229          * These fields are for EXT4_DYNAMIC_REV superblocks only.
1230          *
1231          * Note: the difference between the compatible feature set and
1232          * the incompatible feature set is that if there is a bit set
1233          * in the incompatible feature set that the kernel doesn't
1234          * know about, it should refuse to mount the filesystem.
1235          *
1236          * e2fsck's requirements are more strict; if it doesn't know
1237          * about a feature in either the compatible or incompatible
1238          * feature set, it must abort and not try to meddle with
1239          * things it doesn't understand...
1240          */
1241         __le32  s_first_ino;            /* First non-reserved inode */
1242         __le16  s_inode_size;           /* size of inode structure */
1243         __le16  s_block_group_nr;       /* block group # of this superblock */
1244         __le32  s_feature_compat;       /* compatible feature set */
1245 /*60*/  __le32  s_feature_incompat;     /* incompatible feature set */
1246         __le32  s_feature_ro_compat;    /* readonly-compatible feature set */
1247 /*68*/  __u8    s_uuid[16];             /* 128-bit uuid for volume */
1248 /*78*/  char    s_volume_name[16];      /* volume name */
1249 /*88*/  char    s_last_mounted[64];     /* directory where last mounted */
1250 /*C8*/  __le32  s_algorithm_usage_bitmap; /* For compression */
1251         /*
1252          * Performance hints.  Directory preallocation should only
1253          * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1254          */
1255         __u8    s_prealloc_blocks;      /* Nr of blocks to try to preallocate*/
1256         __u8    s_prealloc_dir_blocks;  /* Nr to preallocate for dirs */
1257         __le16  s_reserved_gdt_blocks;  /* Per group desc for online growth */
1258         /*
1259          * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1260          */
1261 /*D0*/  __u8    s_journal_uuid[16];     /* uuid of journal superblock */
1262 /*E0*/  __le32  s_journal_inum;         /* inode number of journal file */
1263         __le32  s_journal_dev;          /* device number of journal file */
1264         __le32  s_last_orphan;          /* start of list of inodes to delete */
1265         __le32  s_hash_seed[4];         /* HTREE hash seed */
1266         __u8    s_def_hash_version;     /* Default hash version to use */
1267         __u8    s_jnl_backup_type;
1268         __le16  s_desc_size;            /* size of group descriptor */
1269 /*100*/ __le32  s_default_mount_opts;
1270         __le32  s_first_meta_bg;        /* First metablock block group */
1271         __le32  s_mkfs_time;            /* When the filesystem was created */
1272         __le32  s_jnl_blocks[17];       /* Backup of the journal inode */
1273         /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1274 /*150*/ __le32  s_blocks_count_hi;      /* Blocks count */
1275         __le32  s_r_blocks_count_hi;    /* Reserved blocks count */
1276         __le32  s_free_blocks_count_hi; /* Free blocks count */
1277         __le16  s_min_extra_isize;      /* All inodes have at least # bytes */
1278         __le16  s_want_extra_isize;     /* New inodes should reserve # bytes */
1279         __le32  s_flags;                /* Miscellaneous flags */
1280         __le16  s_raid_stride;          /* RAID stride */
1281         __le16  s_mmp_update_interval;  /* # seconds to wait in MMP checking */
1282         __le64  s_mmp_block;            /* Block for multi-mount protection */
1283         __le32  s_raid_stripe_width;    /* blocks on all data disks (N*stride)*/
1284         __u8    s_log_groups_per_flex;  /* FLEX_BG group size */
1285         __u8    s_checksum_type;        /* metadata checksum algorithm used */
1286         __u8    s_encryption_level;     /* versioning level for encryption */
1287         __u8    s_reserved_pad;         /* Padding to next 32bits */
1288         __le64  s_kbytes_written;       /* nr of lifetime kilobytes written */
1289         __le32  s_snapshot_inum;        /* Inode number of active snapshot */
1290         __le32  s_snapshot_id;          /* sequential ID of active snapshot */
1291         __le64  s_snapshot_r_blocks_count; /* reserved blocks for active
1292                                               snapshot's future use */
1293         __le32  s_snapshot_list;        /* inode number of the head of the
1294                                            on-disk snapshot list */
1295 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1296         __le32  s_error_count;          /* number of fs errors */
1297         __le32  s_first_error_time;     /* first time an error happened */
1298         __le32  s_first_error_ino;      /* inode involved in first error */
1299         __le64  s_first_error_block;    /* block involved of first error */
1300         __u8    s_first_error_func[32]; /* function where the error happened */
1301         __le32  s_first_error_line;     /* line number where error happened */
1302         __le32  s_last_error_time;      /* most recent time of an error */
1303         __le32  s_last_error_ino;       /* inode involved in last error */
1304         __le32  s_last_error_line;      /* line number where error happened */
1305         __le64  s_last_error_block;     /* block involved of last error */
1306         __u8    s_last_error_func[32];  /* function where the error happened */
1307 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1308         __u8    s_mount_opts[64];
1309         __le32  s_usr_quota_inum;       /* inode for tracking user quota */
1310         __le32  s_grp_quota_inum;       /* inode for tracking group quota */
1311         __le32  s_overhead_clusters;    /* overhead blocks/clusters in fs */
1312         __le32  s_backup_bgs[2];        /* groups with sparse_super2 SBs */
1313         __u8    s_encrypt_algos[4];     /* Encryption algorithms in use  */
1314         __u8    s_encrypt_pw_salt[16];  /* Salt used for string2key algorithm */
1315         __le32  s_lpf_ino;              /* Location of the lost+found inode */
1316         __le32  s_prj_quota_inum;       /* inode for tracking project quota */
1317         __le32  s_checksum_seed;        /* crc32c(uuid) if csum_seed set */
1318         __le32  s_reserved[98];         /* Padding to the end of the block */
1319         __le32  s_checksum;             /* crc32c(superblock) */
1320 };
1321
1322 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1323
1324 #ifdef __KERNEL__
1325
1326 /*
1327  * run-time mount flags
1328  */
1329 #define EXT4_MF_MNTDIR_SAMPLED          0x0001
1330 #define EXT4_MF_FS_ABORTED              0x0002  /* Fatal error detected */
1331 #define EXT4_MF_TEST_DUMMY_ENCRYPTION   0x0004
1332
1333 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1334 #define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \
1335                                                 EXT4_MF_TEST_DUMMY_ENCRYPTION))
1336 #else
1337 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1338 #endif
1339
1340 /* Number of quota types we support */
1341 #define EXT4_MAXQUOTAS 3
1342
1343 /*
1344  * fourth extended-fs super-block data in memory
1345  */
1346 struct ext4_sb_info {
1347         unsigned long s_desc_size;      /* Size of a group descriptor in bytes */
1348         unsigned long s_inodes_per_block;/* Number of inodes per block */
1349         unsigned long s_blocks_per_group;/* Number of blocks in a group */
1350         unsigned long s_clusters_per_group; /* Number of clusters in a group */
1351         unsigned long s_inodes_per_group;/* Number of inodes in a group */
1352         unsigned long s_itb_per_group;  /* Number of inode table blocks per group */
1353         unsigned long s_gdb_count;      /* Number of group descriptor blocks */
1354         unsigned long s_desc_per_block; /* Number of group descriptors per block */
1355         ext4_group_t s_groups_count;    /* Number of groups in the fs */
1356         ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1357         unsigned long s_overhead;  /* # of fs overhead clusters */
1358         unsigned int s_cluster_ratio;   /* Number of blocks per cluster */
1359         unsigned int s_cluster_bits;    /* log2 of s_cluster_ratio */
1360         loff_t s_bitmap_maxbytes;       /* max bytes for bitmap files */
1361         struct buffer_head * s_sbh;     /* Buffer containing the super block */
1362         struct ext4_super_block *s_es;  /* Pointer to the super block in the buffer */
1363         struct buffer_head **s_group_desc;
1364         unsigned int s_mount_opt;
1365         unsigned int s_mount_opt2;
1366         unsigned int s_mount_flags;
1367         unsigned int s_def_mount_opt;
1368         ext4_fsblk_t s_sb_block;
1369         atomic64_t s_resv_clusters;
1370         kuid_t s_resuid;
1371         kgid_t s_resgid;
1372         unsigned short s_mount_state;
1373         unsigned short s_pad;
1374         int s_addr_per_block_bits;
1375         int s_desc_per_block_bits;
1376         int s_inode_size;
1377         int s_first_ino;
1378         unsigned int s_inode_readahead_blks;
1379         unsigned int s_inode_goal;
1380         spinlock_t s_next_gen_lock;
1381         u32 s_next_generation;
1382         u32 s_hash_seed[4];
1383         int s_def_hash_version;
1384         int s_hash_unsigned;    /* 3 if hash should be signed, 0 if not */
1385         struct percpu_counter s_freeclusters_counter;
1386         struct percpu_counter s_freeinodes_counter;
1387         struct percpu_counter s_dirs_counter;
1388         struct percpu_counter s_dirtyclusters_counter;
1389         struct blockgroup_lock *s_blockgroup_lock;
1390         struct proc_dir_entry *s_proc;
1391         struct kobject s_kobj;
1392         struct completion s_kobj_unregister;
1393         struct super_block *s_sb;
1394
1395         /* Journaling */
1396         struct journal_s *s_journal;
1397         struct list_head s_orphan;
1398         struct mutex s_orphan_lock;
1399         unsigned long s_resize_flags;           /* Flags indicating if there
1400                                                    is a resizer */
1401         unsigned long s_commit_interval;
1402         u32 s_max_batch_time;
1403         u32 s_min_batch_time;
1404         struct block_device *journal_bdev;
1405 #ifdef CONFIG_QUOTA
1406         char *s_qf_names[EXT4_MAXQUOTAS];       /* Names of quota files with journalled quota */
1407         int s_jquota_fmt;                       /* Format of quota to use */
1408 #endif
1409         unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1410         struct rb_root system_blks;
1411
1412 #ifdef EXTENTS_STATS
1413         /* ext4 extents stats */
1414         unsigned long s_ext_min;
1415         unsigned long s_ext_max;
1416         unsigned long s_depth_max;
1417         spinlock_t s_ext_stats_lock;
1418         unsigned long s_ext_blocks;
1419         unsigned long s_ext_extents;
1420 #endif
1421
1422         /* for buddy allocator */
1423         struct ext4_group_info ***s_group_info;
1424         struct inode *s_buddy_cache;
1425         spinlock_t s_md_lock;
1426         unsigned short *s_mb_offsets;
1427         unsigned int *s_mb_maxs;
1428         unsigned int s_group_info_size;
1429
1430         /* tunables */
1431         unsigned long s_stripe;
1432         unsigned int s_mb_stream_request;
1433         unsigned int s_mb_max_to_scan;
1434         unsigned int s_mb_min_to_scan;
1435         unsigned int s_mb_stats;
1436         unsigned int s_mb_order2_reqs;
1437         unsigned int s_mb_group_prealloc;
1438         unsigned int s_max_dir_size_kb;
1439         /* where last allocation was done - for stream allocation */
1440         unsigned long s_mb_last_group;
1441         unsigned long s_mb_last_start;
1442
1443         /* stats for buddy allocator */
1444         atomic_t s_bal_reqs;    /* number of reqs with len > 1 */
1445         atomic_t s_bal_success; /* we found long enough chunks */
1446         atomic_t s_bal_allocated;       /* in blocks */
1447         atomic_t s_bal_ex_scanned;      /* total extents scanned */
1448         atomic_t s_bal_goals;   /* goal hits */
1449         atomic_t s_bal_breaks;  /* too long searches */
1450         atomic_t s_bal_2orders; /* 2^order hits */
1451         spinlock_t s_bal_lock;
1452         unsigned long s_mb_buddies_generated;
1453         unsigned long long s_mb_generation_time;
1454         atomic_t s_mb_lost_chunks;
1455         atomic_t s_mb_preallocated;
1456         atomic_t s_mb_discarded;
1457         atomic_t s_lock_busy;
1458
1459         /* locality groups */
1460         struct ext4_locality_group __percpu *s_locality_groups;
1461
1462         /* for write statistics */
1463         unsigned long s_sectors_written_start;
1464         u64 s_kbytes_written;
1465
1466         /* the size of zero-out chunk */
1467         unsigned int s_extent_max_zeroout_kb;
1468
1469         unsigned int s_log_groups_per_flex;
1470         struct flex_groups *s_flex_groups;
1471         ext4_group_t s_flex_groups_allocated;
1472
1473         /* workqueue for reserved extent conversions (buffered io) */
1474         struct workqueue_struct *rsv_conversion_wq;
1475
1476         /* timer for periodic error stats printing */
1477         struct timer_list s_err_report;
1478
1479         /* Lazy inode table initialization info */
1480         struct ext4_li_request *s_li_request;
1481         /* Wait multiplier for lazy initialization thread */
1482         unsigned int s_li_wait_mult;
1483
1484         /* Kernel thread for multiple mount protection */
1485         struct task_struct *s_mmp_tsk;
1486
1487         /* record the last minlen when FITRIM is called. */
1488         atomic_t s_last_trim_minblks;
1489
1490         /* Reference to checksum algorithm driver via cryptoapi */
1491         struct crypto_shash *s_chksum_driver;
1492
1493         /* Precomputed FS UUID checksum for seeding other checksums */
1494         __u32 s_csum_seed;
1495
1496         /* Reclaim extents from extent status tree */
1497         struct shrinker s_es_shrinker;
1498         struct list_head s_es_list;     /* List of inodes with reclaimable extents */
1499         long s_es_nr_inode;
1500         struct ext4_es_stats s_es_stats;
1501         struct mb_cache *s_mb_cache;
1502         spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1503
1504         /* Ratelimit ext4 messages. */
1505         struct ratelimit_state s_err_ratelimit_state;
1506         struct ratelimit_state s_warning_ratelimit_state;
1507         struct ratelimit_state s_msg_ratelimit_state;
1508 };
1509
1510 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1511 {
1512         return sb->s_fs_info;
1513 }
1514 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1515 {
1516         return container_of(inode, struct ext4_inode_info, vfs_inode);
1517 }
1518
1519 static inline struct timespec ext4_current_time(struct inode *inode)
1520 {
1521         return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1522                 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1523 }
1524
1525 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1526 {
1527         return ino == EXT4_ROOT_INO ||
1528                 ino == EXT4_USR_QUOTA_INO ||
1529                 ino == EXT4_GRP_QUOTA_INO ||
1530                 ino == EXT4_BOOT_LOADER_INO ||
1531                 ino == EXT4_JOURNAL_INO ||
1532                 ino == EXT4_RESIZE_INO ||
1533                 (ino >= EXT4_FIRST_INO(sb) &&
1534                  ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1535 }
1536
1537 /*
1538  * Inode dynamic state flags
1539  */
1540 enum {
1541         EXT4_STATE_JDATA,               /* journaled data exists */
1542         EXT4_STATE_NEW,                 /* inode is newly created */
1543         EXT4_STATE_XATTR,               /* has in-inode xattrs */
1544         EXT4_STATE_NO_EXPAND,           /* No space for expansion */
1545         EXT4_STATE_DA_ALLOC_CLOSE,      /* Alloc DA blks on close */
1546         EXT4_STATE_EXT_MIGRATE,         /* Inode is migrating */
1547         EXT4_STATE_DIO_UNWRITTEN,       /* need convert on dio done*/
1548         EXT4_STATE_NEWENTRY,            /* File just added to dir */
1549         EXT4_STATE_DIOREAD_LOCK,        /* Disable support for dio read
1550                                            nolocking */
1551         EXT4_STATE_MAY_INLINE_DATA,     /* may have in-inode data */
1552         EXT4_STATE_EXT_PRECACHED,       /* extents have been precached */
1553 };
1554
1555 #define EXT4_INODE_BIT_FNS(name, field, offset)                         \
1556 static inline int ext4_test_inode_##name(struct inode *inode, int bit)  \
1557 {                                                                       \
1558         return test_bit(bit + (offset), &EXT4_I(inode)->i_##field);     \
1559 }                                                                       \
1560 static inline void ext4_set_inode_##name(struct inode *inode, int bit)  \
1561 {                                                                       \
1562         set_bit(bit + (offset), &EXT4_I(inode)->i_##field);             \
1563 }                                                                       \
1564 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1565 {                                                                       \
1566         clear_bit(bit + (offset), &EXT4_I(inode)->i_##field);           \
1567 }
1568
1569 /* Add these declarations here only so that these functions can be
1570  * found by name.  Otherwise, they are very hard to locate. */
1571 static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1572 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1573 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1574 EXT4_INODE_BIT_FNS(flag, flags, 0)
1575
1576 /* Add these declarations here only so that these functions can be
1577  * found by name.  Otherwise, they are very hard to locate. */
1578 static inline int ext4_test_inode_state(struct inode *inode, int bit);
1579 static inline void ext4_set_inode_state(struct inode *inode, int bit);
1580 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1581 #if (BITS_PER_LONG < 64)
1582 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1583
1584 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1585 {
1586         (ei)->i_state_flags = 0;
1587 }
1588 #else
1589 EXT4_INODE_BIT_FNS(state, flags, 32)
1590
1591 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1592 {
1593         /* We depend on the fact that callers will set i_flags */
1594 }
1595 #endif
1596 #else
1597 /* Assume that user mode programs are passing in an ext4fs superblock, not
1598  * a kernel struct super_block.  This will allow us to call the feature-test
1599  * macros from user land. */
1600 #define EXT4_SB(sb)     (sb)
1601 #endif
1602
1603 /*
1604  * Returns true if the inode is inode is encrypted
1605  */
1606 static inline int ext4_encrypted_inode(struct inode *inode)
1607 {
1608 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1609         return ext4_test_inode_flag(inode, EXT4_INODE_ENCRYPT);
1610 #else
1611         return 0;
1612 #endif
1613 }
1614
1615 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1616
1617 /*
1618  * Codes for operating systems
1619  */
1620 #define EXT4_OS_LINUX           0
1621 #define EXT4_OS_HURD            1
1622 #define EXT4_OS_MASIX           2
1623 #define EXT4_OS_FREEBSD         3
1624 #define EXT4_OS_LITES           4
1625
1626 /*
1627  * Revision levels
1628  */
1629 #define EXT4_GOOD_OLD_REV       0       /* The good old (original) format */
1630 #define EXT4_DYNAMIC_REV        1       /* V2 format w/ dynamic inode sizes */
1631
1632 #define EXT4_CURRENT_REV        EXT4_GOOD_OLD_REV
1633 #define EXT4_MAX_SUPP_REV       EXT4_DYNAMIC_REV
1634
1635 #define EXT4_GOOD_OLD_INODE_SIZE 128
1636
1637 /*
1638  * Feature set definitions
1639  */
1640
1641 /* Use the ext4_{has,set,clear}_feature_* helpers; these will be removed */
1642 #define EXT4_HAS_COMPAT_FEATURE(sb,mask)                        \
1643         ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1644 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask)                     \
1645         ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1646 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask)                      \
1647         ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1648 #define EXT4_SET_COMPAT_FEATURE(sb,mask)                        \
1649         EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1650 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask)                     \
1651         EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1652 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask)                      \
1653         EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1654 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask)                      \
1655         EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1656 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask)                   \
1657         EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1658 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask)                    \
1659         EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1660
1661 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC        0x0001
1662 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES       0x0002
1663 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL         0x0004
1664 #define EXT4_FEATURE_COMPAT_EXT_ATTR            0x0008
1665 #define EXT4_FEATURE_COMPAT_RESIZE_INODE        0x0010
1666 #define EXT4_FEATURE_COMPAT_DIR_INDEX           0x0020
1667 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2       0x0200
1668
1669 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER     0x0001
1670 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE       0x0002
1671 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR        0x0004
1672 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE        0x0008
1673 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM         0x0010
1674 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK        0x0020
1675 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE      0x0040
1676 #define EXT4_FEATURE_RO_COMPAT_QUOTA            0x0100
1677 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC         0x0200
1678 /*
1679  * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM).  When
1680  * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1681  * all other data structures' checksums.  However, the METADATA_CSUM and
1682  * GDT_CSUM bits are mutually exclusive.
1683  */
1684 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM    0x0400
1685 #define EXT4_FEATURE_RO_COMPAT_READONLY         0x1000
1686 #define EXT4_FEATURE_RO_COMPAT_PROJECT          0x2000
1687
1688 #define EXT4_FEATURE_INCOMPAT_COMPRESSION       0x0001
1689 #define EXT4_FEATURE_INCOMPAT_FILETYPE          0x0002
1690 #define EXT4_FEATURE_INCOMPAT_RECOVER           0x0004 /* Needs recovery */
1691 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV       0x0008 /* Journal device */
1692 #define EXT4_FEATURE_INCOMPAT_META_BG           0x0010
1693 #define EXT4_FEATURE_INCOMPAT_EXTENTS           0x0040 /* extents support */
1694 #define EXT4_FEATURE_INCOMPAT_64BIT             0x0080
1695 #define EXT4_FEATURE_INCOMPAT_MMP               0x0100
1696 #define EXT4_FEATURE_INCOMPAT_FLEX_BG           0x0200
1697 #define EXT4_FEATURE_INCOMPAT_EA_INODE          0x0400 /* EA in inode */
1698 #define EXT4_FEATURE_INCOMPAT_DIRDATA           0x1000 /* data in dirent */
1699 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED         0x2000
1700 #define EXT4_FEATURE_INCOMPAT_LARGEDIR          0x4000 /* >2GB or 3-lvl htree */
1701 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA       0x8000 /* data in inode */
1702 #define EXT4_FEATURE_INCOMPAT_ENCRYPT           0x10000
1703
1704 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
1705 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1706 { \
1707         return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1708                 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
1709 } \
1710 static inline void ext4_set_feature_##name(struct super_block *sb) \
1711 { \
1712         EXT4_SB(sb)->s_es->s_feature_compat |= \
1713                 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1714 } \
1715 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1716 { \
1717         EXT4_SB(sb)->s_es->s_feature_compat &= \
1718                 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1719 }
1720
1721 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
1722 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1723 { \
1724         return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1725                 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
1726 } \
1727 static inline void ext4_set_feature_##name(struct super_block *sb) \
1728 { \
1729         EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
1730                 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1731 } \
1732 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1733 { \
1734         EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
1735                 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1736 }
1737
1738 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
1739 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1740 { \
1741         return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1742                 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
1743 } \
1744 static inline void ext4_set_feature_##name(struct super_block *sb) \
1745 { \
1746         EXT4_SB(sb)->s_es->s_feature_incompat |= \
1747                 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1748 } \
1749 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1750 { \
1751         EXT4_SB(sb)->s_es->s_feature_incompat &= \
1752                 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1753 }
1754
1755 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc,         DIR_PREALLOC)
1756 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes,        IMAGIC_INODES)
1757 EXT4_FEATURE_COMPAT_FUNCS(journal,              HAS_JOURNAL)
1758 EXT4_FEATURE_COMPAT_FUNCS(xattr,                EXT_ATTR)
1759 EXT4_FEATURE_COMPAT_FUNCS(resize_inode,         RESIZE_INODE)
1760 EXT4_FEATURE_COMPAT_FUNCS(dir_index,            DIR_INDEX)
1761 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2,        SPARSE_SUPER2)
1762
1763 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super,      SPARSE_SUPER)
1764 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file,        LARGE_FILE)
1765 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir,         BTREE_DIR)
1766 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file,         HUGE_FILE)
1767 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum,          GDT_CSUM)
1768 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink,         DIR_NLINK)
1769 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize,       EXTRA_ISIZE)
1770 EXT4_FEATURE_RO_COMPAT_FUNCS(quota,             QUOTA)
1771 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc,          BIGALLOC)
1772 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum,     METADATA_CSUM)
1773 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly,          READONLY)
1774 EXT4_FEATURE_RO_COMPAT_FUNCS(project,           PROJECT)
1775
1776 EXT4_FEATURE_INCOMPAT_FUNCS(compression,        COMPRESSION)
1777 EXT4_FEATURE_INCOMPAT_FUNCS(filetype,           FILETYPE)
1778 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery,     RECOVER)
1779 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev,        JOURNAL_DEV)
1780 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg,            META_BG)
1781 EXT4_FEATURE_INCOMPAT_FUNCS(extents,            EXTENTS)
1782 EXT4_FEATURE_INCOMPAT_FUNCS(64bit,              64BIT)
1783 EXT4_FEATURE_INCOMPAT_FUNCS(mmp,                MMP)
1784 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg,            FLEX_BG)
1785 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode,           EA_INODE)
1786 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata,            DIRDATA)
1787 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed,          CSUM_SEED)
1788 EXT4_FEATURE_INCOMPAT_FUNCS(largedir,           LARGEDIR)
1789 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data,        INLINE_DATA)
1790 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt,            ENCRYPT)
1791
1792 #define EXT2_FEATURE_COMPAT_SUPP        EXT4_FEATURE_COMPAT_EXT_ATTR
1793 #define EXT2_FEATURE_INCOMPAT_SUPP      (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1794                                          EXT4_FEATURE_INCOMPAT_META_BG)
1795 #define EXT2_FEATURE_RO_COMPAT_SUPP     (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1796                                          EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1797                                          EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1798
1799 #define EXT3_FEATURE_COMPAT_SUPP        EXT4_FEATURE_COMPAT_EXT_ATTR
1800 #define EXT3_FEATURE_INCOMPAT_SUPP      (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1801                                          EXT4_FEATURE_INCOMPAT_RECOVER| \
1802                                          EXT4_FEATURE_INCOMPAT_META_BG)
1803 #define EXT3_FEATURE_RO_COMPAT_SUPP     (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1804                                          EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1805                                          EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1806
1807 #define EXT4_FEATURE_COMPAT_SUPP        EXT4_FEATURE_COMPAT_EXT_ATTR
1808 #define EXT4_FEATURE_INCOMPAT_SUPP      (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1809                                          EXT4_FEATURE_INCOMPAT_RECOVER| \
1810                                          EXT4_FEATURE_INCOMPAT_META_BG| \
1811                                          EXT4_FEATURE_INCOMPAT_EXTENTS| \
1812                                          EXT4_FEATURE_INCOMPAT_64BIT| \
1813                                          EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1814                                          EXT4_FEATURE_INCOMPAT_MMP | \
1815                                          EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
1816                                          EXT4_FEATURE_INCOMPAT_ENCRYPT | \
1817                                          EXT4_FEATURE_INCOMPAT_CSUM_SEED)
1818 #define EXT4_FEATURE_RO_COMPAT_SUPP     (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1819                                          EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1820                                          EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1821                                          EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1822                                          EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1823                                          EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1824                                          EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1825                                          EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1826                                          EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1827                                          EXT4_FEATURE_RO_COMPAT_QUOTA |\
1828                                          EXT4_FEATURE_RO_COMPAT_PROJECT)
1829
1830 #define EXTN_FEATURE_FUNCS(ver) \
1831 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
1832 { \
1833         return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1834                 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
1835 } \
1836 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
1837 { \
1838         return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1839                 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
1840 } \
1841 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
1842 { \
1843         return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1844                 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
1845 }
1846
1847 EXTN_FEATURE_FUNCS(2)
1848 EXTN_FEATURE_FUNCS(3)
1849 EXTN_FEATURE_FUNCS(4)
1850
1851 static inline bool ext4_has_compat_features(struct super_block *sb)
1852 {
1853         return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
1854 }
1855 static inline bool ext4_has_ro_compat_features(struct super_block *sb)
1856 {
1857         return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
1858 }
1859 static inline bool ext4_has_incompat_features(struct super_block *sb)
1860 {
1861         return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
1862 }
1863
1864 /*
1865  * Default values for user and/or group using reserved blocks
1866  */
1867 #define EXT4_DEF_RESUID         0
1868 #define EXT4_DEF_RESGID         0
1869
1870 /*
1871  * Default project ID
1872  */
1873 #define EXT4_DEF_PROJID         0
1874
1875 #define EXT4_DEF_INODE_READAHEAD_BLKS   32
1876
1877 /*
1878  * Default mount options
1879  */
1880 #define EXT4_DEFM_DEBUG         0x0001
1881 #define EXT4_DEFM_BSDGROUPS     0x0002
1882 #define EXT4_DEFM_XATTR_USER    0x0004
1883 #define EXT4_DEFM_ACL           0x0008
1884 #define EXT4_DEFM_UID16         0x0010
1885 #define EXT4_DEFM_JMODE         0x0060
1886 #define EXT4_DEFM_JMODE_DATA    0x0020
1887 #define EXT4_DEFM_JMODE_ORDERED 0x0040
1888 #define EXT4_DEFM_JMODE_WBACK   0x0060
1889 #define EXT4_DEFM_NOBARRIER     0x0100
1890 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1891 #define EXT4_DEFM_DISCARD       0x0400
1892 #define EXT4_DEFM_NODELALLOC    0x0800
1893
1894 /*
1895  * Default journal batch times
1896  */
1897 #define EXT4_DEF_MIN_BATCH_TIME 0
1898 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1899
1900 /*
1901  * Minimum number of groups in a flexgroup before we separate out
1902  * directories into the first block group of a flexgroup
1903  */
1904 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1905
1906 /*
1907  * Structure of a directory entry
1908  */
1909 #define EXT4_NAME_LEN 255
1910
1911 struct ext4_dir_entry {
1912         __le32  inode;                  /* Inode number */
1913         __le16  rec_len;                /* Directory entry length */
1914         __le16  name_len;               /* Name length */
1915         char    name[EXT4_NAME_LEN];    /* File name */
1916 };
1917
1918 /*
1919  * The new version of the directory entry.  Since EXT4 structures are
1920  * stored in intel byte order, and the name_len field could never be
1921  * bigger than 255 chars, it's safe to reclaim the extra byte for the
1922  * file_type field.
1923  */
1924 struct ext4_dir_entry_2 {
1925         __le32  inode;                  /* Inode number */
1926         __le16  rec_len;                /* Directory entry length */
1927         __u8    name_len;               /* Name length */
1928         __u8    file_type;
1929         char    name[EXT4_NAME_LEN];    /* File name */
1930 };
1931
1932 /*
1933  * This is a bogus directory entry at the end of each leaf block that
1934  * records checksums.
1935  */
1936 struct ext4_dir_entry_tail {
1937         __le32  det_reserved_zero1;     /* Pretend to be unused */
1938         __le16  det_rec_len;            /* 12 */
1939         __u8    det_reserved_zero2;     /* Zero name length */
1940         __u8    det_reserved_ft;        /* 0xDE, fake file type */
1941         __le32  det_checksum;           /* crc32c(uuid+inum+dirblock) */
1942 };
1943
1944 #define EXT4_DIRENT_TAIL(block, blocksize) \
1945         ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1946                                         ((blocksize) - \
1947                                          sizeof(struct ext4_dir_entry_tail))))
1948
1949 /*
1950  * Ext4 directory file types.  Only the low 3 bits are used.  The
1951  * other bits are reserved for now.
1952  */
1953 #define EXT4_FT_UNKNOWN         0
1954 #define EXT4_FT_REG_FILE        1
1955 #define EXT4_FT_DIR             2
1956 #define EXT4_FT_CHRDEV          3
1957 #define EXT4_FT_BLKDEV          4
1958 #define EXT4_FT_FIFO            5
1959 #define EXT4_FT_SOCK            6
1960 #define EXT4_FT_SYMLINK         7
1961
1962 #define EXT4_FT_MAX             8
1963
1964 #define EXT4_FT_DIR_CSUM        0xDE
1965
1966 /*
1967  * EXT4_DIR_PAD defines the directory entries boundaries
1968  *
1969  * NOTE: It must be a multiple of 4
1970  */
1971 #define EXT4_DIR_PAD                    4
1972 #define EXT4_DIR_ROUND                  (EXT4_DIR_PAD - 1)
1973 #define EXT4_DIR_REC_LEN(name_len)      (((name_len) + 8 + EXT4_DIR_ROUND) & \
1974                                          ~EXT4_DIR_ROUND)
1975 #define EXT4_MAX_REC_LEN                ((1<<16)-1)
1976
1977 /*
1978  * If we ever get support for fs block sizes > page_size, we'll need
1979  * to remove the #if statements in the next two functions...
1980  */
1981 static inline unsigned int
1982 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1983 {
1984         unsigned len = le16_to_cpu(dlen);
1985
1986 #if (PAGE_SIZE >= 65536)
1987         if (len == EXT4_MAX_REC_LEN || len == 0)
1988                 return blocksize;
1989         return (len & 65532) | ((len & 3) << 16);
1990 #else
1991         return len;
1992 #endif
1993 }
1994
1995 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1996 {
1997         if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1998                 BUG();
1999 #if (PAGE_SIZE >= 65536)
2000         if (len < 65536)
2001                 return cpu_to_le16(len);
2002         if (len == blocksize) {
2003                 if (blocksize == 65536)
2004                         return cpu_to_le16(EXT4_MAX_REC_LEN);
2005                 else
2006                         return cpu_to_le16(0);
2007         }
2008         return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
2009 #else
2010         return cpu_to_le16(len);
2011 #endif
2012 }
2013
2014 /*
2015  * Hash Tree Directory indexing
2016  * (c) Daniel Phillips, 2001
2017  */
2018
2019 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
2020                     ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2021 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
2022 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2023
2024 /* Legal values for the dx_root hash_version field: */
2025
2026 #define DX_HASH_LEGACY          0
2027 #define DX_HASH_HALF_MD4        1
2028 #define DX_HASH_TEA             2
2029 #define DX_HASH_LEGACY_UNSIGNED 3
2030 #define DX_HASH_HALF_MD4_UNSIGNED       4
2031 #define DX_HASH_TEA_UNSIGNED            5
2032
2033 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2034                               const void *address, unsigned int length)
2035 {
2036         struct {
2037                 struct shash_desc shash;
2038                 char ctx[4];
2039         } desc;
2040         int err;
2041
2042         BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2043
2044         desc.shash.tfm = sbi->s_chksum_driver;
2045         desc.shash.flags = 0;
2046         *(u32 *)desc.ctx = crc;
2047
2048         err = crypto_shash_update(&desc.shash, address, length);
2049         BUG_ON(err);
2050
2051         return *(u32 *)desc.ctx;
2052 }
2053
2054 #ifdef __KERNEL__
2055
2056 /* hash info structure used by the directory hash */
2057 struct dx_hash_info
2058 {
2059         u32             hash;
2060         u32             minor_hash;
2061         int             hash_version;
2062         u32             *seed;
2063 };
2064
2065
2066 /* 32 and 64 bit signed EOF for dx directories */
2067 #define EXT4_HTREE_EOF_32BIT   ((1UL  << (32 - 1)) - 1)
2068 #define EXT4_HTREE_EOF_64BIT   ((1ULL << (64 - 1)) - 1)
2069
2070
2071 /*
2072  * Control parameters used by ext4_htree_next_block
2073  */
2074 #define HASH_NB_ALWAYS          1
2075
2076 struct ext4_filename {
2077         const struct qstr *usr_fname;
2078         struct ext4_str disk_name;
2079         struct dx_hash_info hinfo;
2080 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2081         struct ext4_str crypto_buf;
2082 #endif
2083 };
2084
2085 #define fname_name(p) ((p)->disk_name.name)
2086 #define fname_len(p)  ((p)->disk_name.len)
2087
2088 /*
2089  * Describe an inode's exact location on disk and in memory
2090  */
2091 struct ext4_iloc
2092 {
2093         struct buffer_head *bh;
2094         unsigned long offset;
2095         ext4_group_t block_group;
2096 };
2097
2098 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2099 {
2100         return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2101 }
2102
2103 /*
2104  * This structure is stuffed into the struct file's private_data field
2105  * for directories.  It is where we put information so that we can do
2106  * readdir operations in hash tree order.
2107  */
2108 struct dir_private_info {
2109         struct rb_root  root;
2110         struct rb_node  *curr_node;
2111         struct fname    *extra_fname;
2112         loff_t          last_pos;
2113         __u32           curr_hash;
2114         __u32           curr_minor_hash;
2115         __u32           next_hash;
2116 };
2117
2118 /* calculate the first block number of the group */
2119 static inline ext4_fsblk_t
2120 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2121 {
2122         return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2123                 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2124 }
2125
2126 /*
2127  * Special error return code only used by dx_probe() and its callers.
2128  */
2129 #define ERR_BAD_DX_DIR  (-(MAX_ERRNO - 1))
2130
2131 /*
2132  * Timeout and state flag for lazy initialization inode thread.
2133  */
2134 #define EXT4_DEF_LI_WAIT_MULT                   10
2135 #define EXT4_DEF_LI_MAX_START_DELAY             5
2136 #define EXT4_LAZYINIT_QUIT                      0x0001
2137 #define EXT4_LAZYINIT_RUNNING                   0x0002
2138
2139 /*
2140  * Lazy inode table initialization info
2141  */
2142 struct ext4_lazy_init {
2143         unsigned long           li_state;
2144         struct list_head        li_request_list;
2145         struct mutex            li_list_mtx;
2146 };
2147
2148 struct ext4_li_request {
2149         struct super_block      *lr_super;
2150         struct ext4_sb_info     *lr_sbi;
2151         ext4_group_t            lr_next_group;
2152         struct list_head        lr_request;
2153         unsigned long           lr_next_sched;
2154         unsigned long           lr_timeout;
2155 };
2156
2157 struct ext4_features {
2158         struct kobject f_kobj;
2159         struct completion f_kobj_unregister;
2160 };
2161
2162 /*
2163  * This structure will be used for multiple mount protection. It will be
2164  * written into the block number saved in the s_mmp_block field in the
2165  * superblock. Programs that check MMP should assume that if
2166  * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2167  * to use the filesystem, regardless of how old the timestamp is.
2168  */
2169 #define EXT4_MMP_MAGIC     0x004D4D50U /* ASCII for MMP */
2170 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2171 #define EXT4_MMP_SEQ_FSCK  0xE24D4D50U /* mmp_seq value when being fscked */
2172 #define EXT4_MMP_SEQ_MAX   0xE24D4D4FU /* maximum valid mmp_seq value */
2173
2174 struct mmp_struct {
2175         __le32  mmp_magic;              /* Magic number for MMP */
2176         __le32  mmp_seq;                /* Sequence no. updated periodically */
2177
2178         /*
2179          * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2180          * purposes and do not affect the correctness of the algorithm
2181          */
2182         __le64  mmp_time;               /* Time last updated */
2183         char    mmp_nodename[64];       /* Node which last updated MMP block */
2184         char    mmp_bdevname[32];       /* Bdev which last updated MMP block */
2185
2186         /*
2187          * mmp_check_interval is used to verify if the MMP block has been
2188          * updated on the block device. The value is updated based on the
2189          * maximum time to write the MMP block during an update cycle.
2190          */
2191         __le16  mmp_check_interval;
2192
2193         __le16  mmp_pad1;
2194         __le32  mmp_pad2[226];
2195         __le32  mmp_checksum;           /* crc32c(uuid+mmp_block) */
2196 };
2197
2198 /* arguments passed to the mmp thread */
2199 struct mmpd_data {
2200         struct buffer_head *bh; /* bh from initial read_mmp_block() */
2201         struct super_block *sb;  /* super block of the fs */
2202 };
2203
2204 /*
2205  * Check interval multiplier
2206  * The MMP block is written every update interval and initially checked every
2207  * update interval x the multiplier (the value is then adapted based on the
2208  * write latency). The reason is that writes can be delayed under load and we
2209  * don't want readers to incorrectly assume that the filesystem is no longer
2210  * in use.
2211  */
2212 #define EXT4_MMP_CHECK_MULT             2UL
2213
2214 /*
2215  * Minimum interval for MMP checking in seconds.
2216  */
2217 #define EXT4_MMP_MIN_CHECK_INTERVAL     5UL
2218
2219 /*
2220  * Maximum interval for MMP checking in seconds.
2221  */
2222 #define EXT4_MMP_MAX_CHECK_INTERVAL     300UL
2223
2224 /*
2225  * Function prototypes
2226  */
2227
2228 /*
2229  * Ok, these declarations are also in <linux/kernel.h> but none of the
2230  * ext4 source programs needs to include it so they are duplicated here.
2231  */
2232 # define NORET_TYPE     /**/
2233 # define ATTRIB_NORET   __attribute__((noreturn))
2234 # define NORET_AND      noreturn,
2235
2236 /* bitmap.c */
2237 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2238 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2239                                 struct ext4_group_desc *gdp,
2240                                 struct buffer_head *bh, int sz);
2241 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2242                                   struct ext4_group_desc *gdp,
2243                                   struct buffer_head *bh, int sz);
2244 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2245                                 struct ext4_group_desc *gdp,
2246                                 struct buffer_head *bh);
2247 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2248                                   struct ext4_group_desc *gdp,
2249                                   struct buffer_head *bh);
2250
2251 /* balloc.c */
2252 extern void ext4_get_group_no_and_offset(struct super_block *sb,
2253                                          ext4_fsblk_t blocknr,
2254                                          ext4_group_t *blockgrpp,
2255                                          ext4_grpblk_t *offsetp);
2256 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2257                                           ext4_fsblk_t block);
2258
2259 extern unsigned int ext4_block_group(struct super_block *sb,
2260                         ext4_fsblk_t blocknr);
2261 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2262                         ext4_fsblk_t blocknr);
2263 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2264 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2265                         ext4_group_t group);
2266 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2267                                          ext4_fsblk_t goal,
2268                                          unsigned int flags,
2269                                          unsigned long *count,
2270                                          int *errp);
2271 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2272                                     s64 nclusters, unsigned int flags);
2273 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2274 extern void ext4_check_blocks_bitmap(struct super_block *);
2275 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2276                                                     ext4_group_t block_group,
2277                                                     struct buffer_head ** bh);
2278 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2279
2280 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2281                                                 ext4_group_t block_group);
2282 extern int ext4_wait_block_bitmap(struct super_block *sb,
2283                                   ext4_group_t block_group,
2284                                   struct buffer_head *bh);
2285 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2286                                                   ext4_group_t block_group);
2287 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2288                                               ext4_group_t block_group,
2289                                               struct ext4_group_desc *gdp);
2290 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2291
2292 /* crypto_policy.c */
2293 int ext4_is_child_context_consistent_with_parent(struct inode *parent,
2294                                                  struct inode *child);
2295 int ext4_inherit_context(struct inode *parent, struct inode *child);
2296 void ext4_to_hex(char *dst, char *src, size_t src_size);
2297 int ext4_process_policy(const struct ext4_encryption_policy *policy,
2298                         struct inode *inode);
2299 int ext4_get_policy(struct inode *inode,
2300                     struct ext4_encryption_policy *policy);
2301
2302 /* crypto.c */
2303 extern struct kmem_cache *ext4_crypt_info_cachep;
2304 bool ext4_valid_contents_enc_mode(uint32_t mode);
2305 uint32_t ext4_validate_encryption_key_size(uint32_t mode, uint32_t size);
2306 extern struct workqueue_struct *ext4_read_workqueue;
2307 struct ext4_crypto_ctx *ext4_get_crypto_ctx(struct inode *inode,
2308                                             gfp_t gfp_flags);
2309 void ext4_release_crypto_ctx(struct ext4_crypto_ctx *ctx);
2310 void ext4_restore_control_page(struct page *data_page);
2311 struct page *ext4_encrypt(struct inode *inode,
2312                           struct page *plaintext_page,
2313                           gfp_t gfp_flags);
2314 int ext4_decrypt(struct page *page);
2315 int ext4_encrypted_zeroout(struct inode *inode, ext4_lblk_t lblk,
2316                            ext4_fsblk_t pblk, ext4_lblk_t len);
2317 extern const struct dentry_operations ext4_encrypted_d_ops;
2318
2319 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2320 int ext4_init_crypto(void);
2321 void ext4_exit_crypto(void);
2322 static inline int ext4_sb_has_crypto(struct super_block *sb)
2323 {
2324         return ext4_has_feature_encrypt(sb);
2325 }
2326 #else
2327 static inline int ext4_init_crypto(void) { return 0; }
2328 static inline void ext4_exit_crypto(void) { }
2329 static inline int ext4_sb_has_crypto(struct super_block *sb)
2330 {
2331         return 0;
2332 }
2333 #endif
2334
2335 /* crypto_fname.c */
2336 bool ext4_valid_filenames_enc_mode(uint32_t mode);
2337 u32 ext4_fname_crypto_round_up(u32 size, u32 blksize);
2338 unsigned ext4_fname_encrypted_size(struct inode *inode, u32 ilen);
2339 int ext4_fname_crypto_alloc_buffer(struct inode *inode,
2340                                    u32 ilen, struct ext4_str *crypto_str);
2341 int _ext4_fname_disk_to_usr(struct inode *inode,
2342                             struct dx_hash_info *hinfo,
2343                             const struct ext4_str *iname,
2344                             struct ext4_str *oname);
2345 int ext4_fname_disk_to_usr(struct inode *inode,
2346                            struct dx_hash_info *hinfo,
2347                            const struct ext4_dir_entry_2 *de,
2348                            struct ext4_str *oname);
2349 int ext4_fname_usr_to_disk(struct inode *inode,
2350                            const struct qstr *iname,
2351                            struct ext4_str *oname);
2352 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2353 void ext4_fname_crypto_free_buffer(struct ext4_str *crypto_str);
2354 int ext4_fname_setup_filename(struct inode *dir, const struct qstr *iname,
2355                               int lookup, struct ext4_filename *fname);
2356 void ext4_fname_free_filename(struct ext4_filename *fname);
2357 #else
2358 static inline
2359 int ext4_setup_fname_crypto(struct inode *inode)
2360 {
2361         return 0;
2362 }
2363 static inline void ext4_fname_crypto_free_buffer(struct ext4_str *p) { }
2364 static inline int ext4_fname_setup_filename(struct inode *dir,
2365                                      const struct qstr *iname,
2366                                      int lookup, struct ext4_filename *fname)
2367 {
2368         fname->usr_fname = iname;
2369         fname->disk_name.name = (unsigned char *) iname->name;
2370         fname->disk_name.len = iname->len;
2371         return 0;
2372 }
2373 static inline void ext4_fname_free_filename(struct ext4_filename *fname) { }
2374 #endif
2375
2376
2377 /* crypto_key.c */
2378 void ext4_free_crypt_info(struct ext4_crypt_info *ci);
2379 void ext4_free_encryption_info(struct inode *inode, struct ext4_crypt_info *ci);
2380 int _ext4_get_encryption_info(struct inode *inode);
2381
2382 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2383 int ext4_has_encryption_key(struct inode *inode);
2384
2385 static inline int ext4_get_encryption_info(struct inode *inode)
2386 {
2387         struct ext4_crypt_info *ci = EXT4_I(inode)->i_crypt_info;
2388
2389         if (!ci ||
2390             (ci->ci_keyring_key &&
2391              (ci->ci_keyring_key->flags & ((1 << KEY_FLAG_INVALIDATED) |
2392                                            (1 << KEY_FLAG_REVOKED) |
2393                                            (1 << KEY_FLAG_DEAD)))))
2394                 return _ext4_get_encryption_info(inode);
2395         return 0;
2396 }
2397
2398 static inline struct ext4_crypt_info *ext4_encryption_info(struct inode *inode)
2399 {
2400         return EXT4_I(inode)->i_crypt_info;
2401 }
2402
2403 #else
2404 static inline int ext4_has_encryption_key(struct inode *inode)
2405 {
2406         return 0;
2407 }
2408 static inline int ext4_get_encryption_info(struct inode *inode)
2409 {
2410         return 0;
2411 }
2412 static inline struct ext4_crypt_info *ext4_encryption_info(struct inode *inode)
2413 {
2414         return NULL;
2415 }
2416 #endif
2417
2418
2419 /* dir.c */
2420 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2421                                   struct file *,
2422                                   struct ext4_dir_entry_2 *,
2423                                   struct buffer_head *, char *, int,
2424                                   unsigned int);
2425 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset)      \
2426         unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2427                                         (de), (bh), (buf), (size), (offset)))
2428 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2429                                 __u32 minor_hash,
2430                                 struct ext4_dir_entry_2 *dirent,
2431                                 struct ext4_str *ent_name);
2432 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2433 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2434                              struct buffer_head *bh,
2435                              void *buf, int buf_size,
2436                              struct ext4_filename *fname,
2437                              struct ext4_dir_entry_2 **dest_de);
2438 int ext4_insert_dentry(struct inode *dir,
2439                        struct inode *inode,
2440                        struct ext4_dir_entry_2 *de,
2441                        int buf_size,
2442                        struct ext4_filename *fname);
2443 static inline void ext4_update_dx_flag(struct inode *inode)
2444 {
2445         if (!ext4_has_feature_dir_index(inode->i_sb))
2446                 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2447 }
2448 static unsigned char ext4_filetype_table[] = {
2449         DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2450 };
2451
2452 static inline  unsigned char get_dtype(struct super_block *sb, int filetype)
2453 {
2454         if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2455                 return DT_UNKNOWN;
2456
2457         return ext4_filetype_table[filetype];
2458 }
2459 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2460                              void *buf, int buf_size);
2461
2462 /* fsync.c */
2463 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2464
2465 /* hash.c */
2466 extern int ext4fs_dirhash(const char *name, int len, struct
2467                           dx_hash_info *hinfo);
2468
2469 /* ialloc.c */
2470 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2471                                       const struct qstr *qstr, __u32 goal,
2472                                       uid_t *owner, int handle_type,
2473                                       unsigned int line_no, int nblocks);
2474
2475 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner) \
2476         __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2477                          0, 0, 0)
2478 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2479                                     type, nblocks)                  \
2480         __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2481                          (type), __LINE__, (nblocks))
2482
2483
2484 extern void ext4_free_inode(handle_t *, struct inode *);
2485 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2486 extern unsigned long ext4_count_free_inodes(struct super_block *);
2487 extern unsigned long ext4_count_dirs(struct super_block *);
2488 extern void ext4_check_inodes_bitmap(struct super_block *);
2489 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2490 extern int ext4_init_inode_table(struct super_block *sb,
2491                                  ext4_group_t group, int barrier);
2492 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2493
2494 /* mballoc.c */
2495 extern const struct file_operations ext4_seq_mb_groups_fops;
2496 extern long ext4_mb_stats;
2497 extern long ext4_mb_max_to_scan;
2498 extern int ext4_mb_init(struct super_block *);
2499 extern int ext4_mb_release(struct super_block *);
2500 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2501                                 struct ext4_allocation_request *, int *);
2502 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2503 extern void ext4_discard_preallocations(struct inode *);
2504 extern int __init ext4_init_mballoc(void);
2505 extern void ext4_exit_mballoc(void);
2506 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2507                              struct buffer_head *bh, ext4_fsblk_t block,
2508                              unsigned long count, int flags);
2509 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2510                                    ext4_group_t ngroups);
2511 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2512                 ext4_group_t i, struct ext4_group_desc *desc);
2513 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2514                                 ext4_fsblk_t block, unsigned long count);
2515 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2516
2517 /* inode.c */
2518 int ext4_inode_is_fast_symlink(struct inode *inode);
2519 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2520 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2521 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2522                              struct buffer_head *bh_result, int create);
2523 int ext4_dax_mmap_get_block(struct inode *inode, sector_t iblock,
2524                             struct buffer_head *bh_result, int create);
2525 int ext4_get_block(struct inode *inode, sector_t iblock,
2526                    struct buffer_head *bh_result, int create);
2527 int ext4_dio_get_block(struct inode *inode, sector_t iblock,
2528                        struct buffer_head *bh_result, int create);
2529 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2530                            struct buffer_head *bh, int create);
2531 int ext4_walk_page_buffers(handle_t *handle,
2532                            struct buffer_head *head,
2533                            unsigned from,
2534                            unsigned to,
2535                            int *partial,
2536                            int (*fn)(handle_t *handle,
2537                                      struct buffer_head *bh));
2538 int do_journal_get_write_access(handle_t *handle,
2539                                 struct buffer_head *bh);
2540 #define FALL_BACK_TO_NONDELALLOC 1
2541 #define CONVERT_INLINE_DATA      2
2542
2543 extern struct inode *ext4_iget(struct super_block *, unsigned long);
2544 extern struct inode *ext4_iget_normal(struct super_block *, unsigned long);
2545 extern int  ext4_write_inode(struct inode *, struct writeback_control *);
2546 extern int  ext4_setattr(struct dentry *, struct iattr *);
2547 extern int  ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
2548                                 struct kstat *stat);
2549 extern void ext4_evict_inode(struct inode *);
2550 extern void ext4_clear_inode(struct inode *);
2551 extern int  ext4_sync_inode(handle_t *, struct inode *);
2552 extern void ext4_dirty_inode(struct inode *, int);
2553 extern int ext4_change_inode_journal_flag(struct inode *, int);
2554 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2555 extern int ext4_inode_attach_jinode(struct inode *inode);
2556 extern int ext4_can_truncate(struct inode *inode);
2557 extern void ext4_truncate(struct inode *);
2558 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2559 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
2560 extern void ext4_set_inode_flags(struct inode *);
2561 extern void ext4_get_inode_flags(struct ext4_inode_info *);
2562 extern int ext4_alloc_da_blocks(struct inode *inode);
2563 extern void ext4_set_aops(struct inode *inode);
2564 extern int ext4_writepage_trans_blocks(struct inode *);
2565 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2566 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2567                              loff_t lstart, loff_t lend);
2568 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
2569 extern int ext4_filemap_fault(struct vm_area_struct *vma, struct vm_fault *vmf);
2570 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2571 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
2572 extern void ext4_da_update_reserve_space(struct inode *inode,
2573                                         int used, int quota_claim);
2574 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
2575                               ext4_fsblk_t pblk, ext4_lblk_t len);
2576 extern int ext4_get_next_extent(struct inode *inode, ext4_lblk_t lblk,
2577                                 unsigned int map_len,
2578                                 struct extent_status *result);
2579
2580 /* indirect.c */
2581 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2582                                 struct ext4_map_blocks *map, int flags);
2583 extern ssize_t ext4_ind_direct_IO(struct kiocb *iocb, struct iov_iter *iter,
2584                                   loff_t offset);
2585 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2586 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2587 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2588 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2589                                  ext4_lblk_t start, ext4_lblk_t end);
2590
2591 /* ioctl.c */
2592 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2593 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2594
2595 /* migrate.c */
2596 extern int ext4_ext_migrate(struct inode *);
2597 extern int ext4_ind_migrate(struct inode *inode);
2598
2599 /* namei.c */
2600 extern int ext4_dirent_csum_verify(struct inode *inode,
2601                                    struct ext4_dir_entry *dirent);
2602 extern int ext4_orphan_add(handle_t *, struct inode *);
2603 extern int ext4_orphan_del(handle_t *, struct inode *);
2604 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2605                                 __u32 start_minor_hash, __u32 *next_hash);
2606 extern int ext4_search_dir(struct buffer_head *bh,
2607                            char *search_buf,
2608                            int buf_size,
2609                            struct inode *dir,
2610                            struct ext4_filename *fname,
2611                            const struct qstr *d_name,
2612                            unsigned int offset,
2613                            struct ext4_dir_entry_2 **res_dir);
2614 extern int ext4_generic_delete_entry(handle_t *handle,
2615                                      struct inode *dir,
2616                                      struct ext4_dir_entry_2 *de_del,
2617                                      struct buffer_head *bh,
2618                                      void *entry_buf,
2619                                      int buf_size,
2620                                      int csum_size);
2621 extern int ext4_empty_dir(struct inode *inode);
2622
2623 /* resize.c */
2624 extern int ext4_group_add(struct super_block *sb,
2625                                 struct ext4_new_group_data *input);
2626 extern int ext4_group_extend(struct super_block *sb,
2627                                 struct ext4_super_block *es,
2628                                 ext4_fsblk_t n_blocks_count);
2629 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2630
2631 /* super.c */
2632 extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
2633 extern int ext4_calculate_overhead(struct super_block *sb);
2634 extern void ext4_superblock_csum_set(struct super_block *sb);
2635 extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2636 extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2637 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2638                                     ext4_group_t ngroup);
2639 extern const char *ext4_decode_error(struct super_block *sb, int errno,
2640                                      char nbuf[16]);
2641
2642 extern __printf(4, 5)
2643 void __ext4_error(struct super_block *, const char *, unsigned int,
2644                   const char *, ...);
2645 extern __printf(5, 6)
2646 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2647                       const char *, ...);
2648 extern __printf(5, 6)
2649 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2650                      const char *, ...);
2651 extern void __ext4_std_error(struct super_block *, const char *,
2652                              unsigned int, int);
2653 extern __printf(4, 5)
2654 void __ext4_abort(struct super_block *, const char *, unsigned int,
2655                   const char *, ...);
2656 extern __printf(4, 5)
2657 void __ext4_warning(struct super_block *, const char *, unsigned int,
2658                     const char *, ...);
2659 extern __printf(4, 5)
2660 void __ext4_warning_inode(const struct inode *inode, const char *function,
2661                           unsigned int line, const char *fmt, ...);
2662 extern __printf(3, 4)
2663 void __ext4_msg(struct super_block *, const char *, const char *, ...);
2664 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2665                            const char *, unsigned int, const char *);
2666 extern __printf(7, 8)
2667 void __ext4_grp_locked_error(const char *, unsigned int,
2668                              struct super_block *, ext4_group_t,
2669                              unsigned long, ext4_fsblk_t,
2670                              const char *, ...);
2671
2672 #define EXT4_ERROR_INODE(inode, fmt, a...) \
2673         ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
2674
2675 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...)                 \
2676         ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
2677
2678 #define EXT4_ERROR_FILE(file, block, fmt, a...)                         \
2679         ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
2680
2681 #ifdef CONFIG_PRINTK
2682
2683 #define ext4_error_inode(inode, func, line, block, fmt, ...)            \
2684         __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__)
2685 #define ext4_error_file(file, func, line, block, fmt, ...)              \
2686         __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2687 #define ext4_error(sb, fmt, ...)                                        \
2688         __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2689 #define ext4_abort(sb, fmt, ...)                                        \
2690         __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2691 #define ext4_warning(sb, fmt, ...)                                      \
2692         __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2693 #define ext4_warning_inode(inode, fmt, ...)                             \
2694         __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
2695 #define ext4_msg(sb, level, fmt, ...)                           \
2696         __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2697 #define dump_mmp_msg(sb, mmp, msg)                                      \
2698         __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2699 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...)            \
2700         __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2701                                 fmt, ##__VA_ARGS__)
2702
2703 #else
2704
2705 #define ext4_error_inode(inode, func, line, block, fmt, ...)            \
2706 do {                                                                    \
2707         no_printk(fmt, ##__VA_ARGS__);                                  \
2708         __ext4_error_inode(inode, "", 0, block, " ");                   \
2709 } while (0)
2710 #define ext4_error_file(file, func, line, block, fmt, ...)              \
2711 do {                                                                    \
2712         no_printk(fmt, ##__VA_ARGS__);                                  \
2713         __ext4_error_file(file, "", 0, block, " ");                     \
2714 } while (0)
2715 #define ext4_error(sb, fmt, ...)                                        \
2716 do {                                                                    \
2717         no_printk(fmt, ##__VA_ARGS__);                                  \
2718         __ext4_error(sb, "", 0, " ");                                   \
2719 } while (0)
2720 #define ext4_abort(sb, fmt, ...)                                        \
2721 do {                                                                    \
2722         no_printk(fmt, ##__VA_ARGS__);                                  \
2723         __ext4_abort(sb, "", 0, " ");                                   \
2724 } while (0)
2725 #define ext4_warning(sb, fmt, ...)                                      \
2726 do {                                                                    \
2727         no_printk(fmt, ##__VA_ARGS__);                                  \
2728         __ext4_warning(sb, "", 0, " ");                                 \
2729 } while (0)
2730 #define ext4_warning_inode(inode, fmt, ...)                             \
2731 do {                                                                    \
2732         no_printk(fmt, ##__VA_ARGS__);                                  \
2733         __ext4_warning_inode(inode, "", 0, " ");                        \
2734 } while (0)
2735 #define ext4_msg(sb, level, fmt, ...)                                   \
2736 do {                                                                    \
2737         no_printk(fmt, ##__VA_ARGS__);                                  \
2738         __ext4_msg(sb, "", " ");                                        \
2739 } while (0)
2740 #define dump_mmp_msg(sb, mmp, msg)                                      \
2741         __dump_mmp_msg(sb, mmp, "", 0, "")
2742 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...)            \
2743 do {                                                                    \
2744         no_printk(fmt, ##__VA_ARGS__);                          \
2745         __ext4_grp_locked_error("", 0, sb, grp, ino, block, " ");       \
2746 } while (0)
2747
2748 #endif
2749
2750 extern void ext4_update_dynamic_rev(struct super_block *sb);
2751 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2752                                         __u32 compat);
2753 extern int ext4_update_rocompat_feature(handle_t *handle,
2754                                         struct super_block *sb, __u32 rocompat);
2755 extern int ext4_update_incompat_feature(handle_t *handle,
2756                                         struct super_block *sb, __u32 incompat);
2757 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2758                                       struct ext4_group_desc *bg);
2759 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2760                                       struct ext4_group_desc *bg);
2761 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2762                                      struct ext4_group_desc *bg);
2763 extern __u32 ext4_free_group_clusters(struct super_block *sb,
2764                                       struct ext4_group_desc *bg);
2765 extern __u32 ext4_free_inodes_count(struct super_block *sb,
2766                                  struct ext4_group_desc *bg);
2767 extern __u32 ext4_used_dirs_count(struct super_block *sb,
2768                                 struct ext4_group_desc *bg);
2769 extern __u32 ext4_itable_unused_count(struct super_block *sb,
2770                                    struct ext4_group_desc *bg);
2771 extern void ext4_block_bitmap_set(struct super_block *sb,
2772                                   struct ext4_group_desc *bg, ext4_fsblk_t blk);
2773 extern void ext4_inode_bitmap_set(struct super_block *sb,
2774                                   struct ext4_group_desc *bg, ext4_fsblk_t blk);
2775 extern void ext4_inode_table_set(struct super_block *sb,
2776                                  struct ext4_group_desc *bg, ext4_fsblk_t blk);
2777 extern void ext4_free_group_clusters_set(struct super_block *sb,
2778                                          struct ext4_group_desc *bg,
2779                                          __u32 count);
2780 extern void ext4_free_inodes_set(struct super_block *sb,
2781                                 struct ext4_group_desc *bg, __u32 count);
2782 extern void ext4_used_dirs_set(struct super_block *sb,
2783                                 struct ext4_group_desc *bg, __u32 count);
2784 extern void ext4_itable_unused_set(struct super_block *sb,
2785                                    struct ext4_group_desc *bg, __u32 count);
2786 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2787                                        struct ext4_group_desc *gdp);
2788 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2789                                      struct ext4_group_desc *gdp);
2790 extern int ext4_register_li_request(struct super_block *sb,
2791                                     ext4_group_t first_not_zeroed);
2792
2793 static inline int ext4_has_group_desc_csum(struct super_block *sb)
2794 {
2795         return ext4_has_feature_gdt_csum(sb) ||
2796                EXT4_SB(sb)->s_chksum_driver != NULL;
2797 }
2798
2799 static inline int ext4_has_metadata_csum(struct super_block *sb)
2800 {
2801         WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
2802                      !EXT4_SB(sb)->s_chksum_driver);
2803
2804         return (EXT4_SB(sb)->s_chksum_driver != NULL);
2805 }
2806 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2807 {
2808         return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2809                 le32_to_cpu(es->s_blocks_count_lo);
2810 }
2811
2812 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2813 {
2814         return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2815                 le32_to_cpu(es->s_r_blocks_count_lo);
2816 }
2817
2818 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2819 {
2820         return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2821                 le32_to_cpu(es->s_free_blocks_count_lo);
2822 }
2823
2824 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2825                                          ext4_fsblk_t blk)
2826 {
2827         es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2828         es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2829 }
2830
2831 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2832                                               ext4_fsblk_t blk)
2833 {
2834         es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2835         es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2836 }
2837
2838 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2839                                            ext4_fsblk_t blk)
2840 {
2841         es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2842         es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2843 }
2844
2845 static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
2846 {
2847         if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2848                 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2849                         le32_to_cpu(raw_inode->i_size_lo);
2850         else
2851                 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2852 }
2853
2854 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2855 {
2856         raw_inode->i_size_lo = cpu_to_le32(i_size);
2857         raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2858 }
2859
2860 static inline
2861 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2862                                             ext4_group_t group)
2863 {
2864          struct ext4_group_info ***grp_info;
2865          long indexv, indexh;
2866          BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
2867          grp_info = EXT4_SB(sb)->s_group_info;
2868          indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2869          indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2870          return grp_info[indexv][indexh];
2871 }
2872
2873 /*
2874  * Reading s_groups_count requires using smp_rmb() afterwards.  See
2875  * the locking protocol documented in the comments of ext4_group_add()
2876  * in resize.c
2877  */
2878 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2879 {
2880         ext4_group_t    ngroups = EXT4_SB(sb)->s_groups_count;
2881
2882         smp_rmb();
2883         return ngroups;
2884 }
2885
2886 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2887                                              ext4_group_t block_group)
2888 {
2889         return block_group >> sbi->s_log_groups_per_flex;
2890 }
2891
2892 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2893 {
2894         return 1 << sbi->s_log_groups_per_flex;
2895 }
2896
2897 #define ext4_std_error(sb, errno)                               \
2898 do {                                                            \
2899         if ((errno))                                            \
2900                 __ext4_std_error((sb), __func__, __LINE__, (errno));    \
2901 } while (0)
2902
2903 #ifdef CONFIG_SMP
2904 /* Each CPU can accumulate percpu_counter_batch clusters in their local
2905  * counters. So we need to make sure we have free clusters more
2906  * than percpu_counter_batch  * nr_cpu_ids. Also add a window of 4 times.
2907  */
2908 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2909 #else
2910 #define EXT4_FREECLUSTERS_WATERMARK 0
2911 #endif
2912
2913 /* Update i_disksize. Requires i_mutex to avoid races with truncate */
2914 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2915 {
2916         WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
2917                      !inode_is_locked(inode));
2918         down_write(&EXT4_I(inode)->i_data_sem);
2919         if (newsize > EXT4_I(inode)->i_disksize)
2920                 EXT4_I(inode)->i_disksize = newsize;
2921         up_write(&EXT4_I(inode)->i_data_sem);
2922 }
2923
2924 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
2925 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
2926 {
2927         int changed = 0;
2928
2929         if (newsize > inode->i_size) {
2930                 i_size_write(inode, newsize);
2931                 changed = 1;
2932         }
2933         if (newsize > EXT4_I(inode)->i_disksize) {
2934                 ext4_update_i_disksize(inode, newsize);
2935                 changed |= 2;
2936         }
2937         return changed;
2938 }
2939
2940 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
2941                                       loff_t len);
2942
2943 struct ext4_group_info {
2944         unsigned long   bb_state;
2945         struct rb_root  bb_free_root;
2946         ext4_grpblk_t   bb_first_free;  /* first free block */
2947         ext4_grpblk_t   bb_free;        /* total free blocks */
2948         ext4_grpblk_t   bb_fragments;   /* nr of freespace fragments */
2949         ext4_grpblk_t   bb_largest_free_order;/* order of largest frag in BG */
2950         struct          list_head bb_prealloc_list;
2951 #ifdef DOUBLE_CHECK
2952         void            *bb_bitmap;
2953 #endif
2954         struct rw_semaphore alloc_sem;
2955         ext4_grpblk_t   bb_counters[];  /* Nr of free power-of-two-block
2956                                          * regions, index is order.
2957                                          * bb_counters[3] = 5 means
2958                                          * 5 free 8-block regions. */
2959 };
2960
2961 #define EXT4_GROUP_INFO_NEED_INIT_BIT           0
2962 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT         1
2963 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT     2
2964 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT     3
2965
2966 #define EXT4_MB_GRP_NEED_INIT(grp)      \
2967         (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2968 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp)        \
2969         (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2970 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp)        \
2971         (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2972
2973 #define EXT4_MB_GRP_WAS_TRIMMED(grp)    \
2974         (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2975 #define EXT4_MB_GRP_SET_TRIMMED(grp)    \
2976         (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2977 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp)  \
2978         (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2979
2980 #define EXT4_MAX_CONTENTION             8
2981 #define EXT4_CONTENTION_THRESHOLD       2
2982
2983 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2984                                               ext4_group_t group)
2985 {
2986         return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2987 }
2988
2989 /*
2990  * Returns true if the filesystem is busy enough that attempts to
2991  * access the block group locks has run into contention.
2992  */
2993 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2994 {
2995         return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2996 }
2997
2998 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2999 {
3000         spinlock_t *lock = ext4_group_lock_ptr(sb, group);
3001         if (spin_trylock(lock))
3002                 /*
3003                  * We're able to grab the lock right away, so drop the
3004                  * lock contention counter.
3005                  */
3006                 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
3007         else {
3008                 /*
3009                  * The lock is busy, so bump the contention counter,
3010                  * and then wait on the spin lock.
3011                  */
3012                 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
3013                                   EXT4_MAX_CONTENTION);
3014                 spin_lock(lock);
3015         }
3016 }
3017
3018 static inline void ext4_unlock_group(struct super_block *sb,
3019                                         ext4_group_t group)
3020 {
3021         spin_unlock(ext4_group_lock_ptr(sb, group));
3022 }
3023
3024 /*
3025  * Block validity checking
3026  */
3027 #define ext4_check_indirect_blockref(inode, bh)                         \
3028         ext4_check_blockref(__func__, __LINE__, inode,                  \
3029                             (__le32 *)(bh)->b_data,                     \
3030                             EXT4_ADDR_PER_BLOCK((inode)->i_sb))
3031
3032 #define ext4_ind_check_inode(inode)                                     \
3033         ext4_check_blockref(__func__, __LINE__, inode,                  \
3034                             EXT4_I(inode)->i_data,                      \
3035                             EXT4_NDIR_BLOCKS)
3036
3037 /*
3038  * Inodes and files operations
3039  */
3040
3041 /* dir.c */
3042 extern const struct file_operations ext4_dir_operations;
3043
3044 /* file.c */
3045 extern const struct inode_operations ext4_file_inode_operations;
3046 extern const struct file_operations ext4_file_operations;
3047 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
3048
3049 /* inline.c */
3050 extern int ext4_get_max_inline_size(struct inode *inode);
3051 extern int ext4_find_inline_data_nolock(struct inode *inode);
3052 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
3053                                  unsigned int len);
3054 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
3055
3056 extern int ext4_readpage_inline(struct inode *inode, struct page *page);
3057 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3058                                          struct inode *inode,
3059                                          loff_t pos, unsigned len,
3060                                          unsigned flags,
3061                                          struct page **pagep);
3062 extern int ext4_write_inline_data_end(struct inode *inode,
3063                                       loff_t pos, unsigned len,
3064                                       unsigned copied,
3065                                       struct page *page);
3066 extern struct buffer_head *
3067 ext4_journalled_write_inline_data(struct inode *inode,
3068                                   unsigned len,
3069                                   struct page *page);
3070 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
3071                                            struct inode *inode,
3072                                            loff_t pos, unsigned len,
3073                                            unsigned flags,
3074                                            struct page **pagep,
3075                                            void **fsdata);
3076 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
3077                                          unsigned len, unsigned copied,
3078                                          struct page *page);
3079 extern int ext4_try_add_inline_entry(handle_t *handle,
3080                                      struct ext4_filename *fname,
3081                                      struct inode *dir, struct inode *inode);
3082 extern int ext4_try_create_inline_dir(handle_t *handle,
3083                                       struct inode *parent,
3084                                       struct inode *inode);
3085 extern int ext4_read_inline_dir(struct file *filp,
3086                                 struct dir_context *ctx,
3087                                 int *has_inline_data);
3088 extern int htree_inlinedir_to_tree(struct file *dir_file,
3089                                    struct inode *dir, ext4_lblk_t block,
3090                                    struct dx_hash_info *hinfo,
3091                                    __u32 start_hash, __u32 start_minor_hash,
3092                                    int *has_inline_data);
3093 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3094                                         struct ext4_filename *fname,
3095                                         const struct qstr *d_name,
3096                                         struct ext4_dir_entry_2 **res_dir,
3097                                         int *has_inline_data);
3098 extern int ext4_delete_inline_entry(handle_t *handle,
3099                                     struct inode *dir,
3100                                     struct ext4_dir_entry_2 *de_del,
3101                                     struct buffer_head *bh,
3102                                     int *has_inline_data);
3103 extern int empty_inline_dir(struct inode *dir, int *has_inline_data);
3104 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3105                                         struct ext4_dir_entry_2 **parent_de,
3106                                         int *retval);
3107 extern int ext4_inline_data_fiemap(struct inode *inode,
3108                                    struct fiemap_extent_info *fieinfo,
3109                                    int *has_inline, __u64 start, __u64 len);
3110 extern int ext4_try_to_evict_inline_data(handle_t *handle,
3111                                          struct inode *inode,
3112                                          int needed);
3113 extern void ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3114
3115 extern int ext4_convert_inline_data(struct inode *inode);
3116
3117 static inline int ext4_has_inline_data(struct inode *inode)
3118 {
3119         return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3120                EXT4_I(inode)->i_inline_off;
3121 }
3122
3123 /* namei.c */
3124 extern const struct inode_operations ext4_dir_inode_operations;
3125 extern const struct inode_operations ext4_special_inode_operations;
3126 extern struct dentry *ext4_get_parent(struct dentry *child);
3127 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3128                                  struct ext4_dir_entry_2 *de,
3129                                  int blocksize, int csum_size,
3130                                  unsigned int parent_ino, int dotdot_real_len);
3131 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t,
3132                                    unsigned int blocksize);
3133 extern int ext4_handle_dirty_dirent_node(handle_t *handle,
3134                                          struct inode *inode,
3135                                          struct buffer_head *bh);
3136 #define S_SHIFT 12
3137 static unsigned char ext4_type_by_mode[S_IFMT >> S_SHIFT] = {
3138         [S_IFREG >> S_SHIFT]    = EXT4_FT_REG_FILE,
3139         [S_IFDIR >> S_SHIFT]    = EXT4_FT_DIR,
3140         [S_IFCHR >> S_SHIFT]    = EXT4_FT_CHRDEV,
3141         [S_IFBLK >> S_SHIFT]    = EXT4_FT_BLKDEV,
3142         [S_IFIFO >> S_SHIFT]    = EXT4_FT_FIFO,
3143         [S_IFSOCK >> S_SHIFT]   = EXT4_FT_SOCK,
3144         [S_IFLNK >> S_SHIFT]    = EXT4_FT_SYMLINK,
3145 };
3146
3147 static inline void ext4_set_de_type(struct super_block *sb,
3148                                 struct ext4_dir_entry_2 *de,
3149                                 umode_t mode) {
3150         if (ext4_has_feature_filetype(sb))
3151                 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3152 }
3153
3154 /* readpages.c */
3155 extern int ext4_mpage_readpages(struct address_space *mapping,
3156                                 struct list_head *pages, struct page *page,
3157                                 unsigned nr_pages);
3158
3159 /* symlink.c */
3160 extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3161 extern const struct inode_operations ext4_symlink_inode_operations;
3162 extern const struct inode_operations ext4_fast_symlink_inode_operations;
3163
3164 /* sysfs.c */
3165 extern int ext4_register_sysfs(struct super_block *sb);
3166 extern void ext4_unregister_sysfs(struct super_block *sb);
3167 extern int __init ext4_init_sysfs(void);
3168 extern void ext4_exit_sysfs(void);
3169
3170 /* block_validity */
3171 extern void ext4_release_system_zone(struct super_block *sb);
3172 extern int ext4_setup_system_zone(struct super_block *sb);
3173 extern int __init ext4_init_system_zone(void);
3174 extern void ext4_exit_system_zone(void);
3175 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
3176                                  ext4_fsblk_t start_blk,
3177                                  unsigned int count);
3178 extern int ext4_check_blockref(const char *, unsigned int,
3179                                struct inode *, __le32 *, unsigned int);
3180
3181 /* extents.c */
3182 struct ext4_ext_path;
3183 struct ext4_extent;
3184
3185 /*
3186  * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3187  * __le32.
3188  */
3189 #define EXT_MAX_BLOCKS  0xffffffff
3190
3191 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
3192 extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
<