ext4: introduce new metadata buffer read helpers
[platform/kernel/linux-rpi.git] / fs / ext4 / ext4.h
1 // SPDX-License-Identifier: GPL-2.0
2 /*
3  *  ext4.h
4  *
5  * Copyright (C) 1992, 1993, 1994, 1995
6  * Remy Card (card@masi.ibp.fr)
7  * Laboratoire MASI - Institut Blaise Pascal
8  * Universite Pierre et Marie Curie (Paris VI)
9  *
10  *  from
11  *
12  *  linux/include/linux/minix_fs.h
13  *
14  *  Copyright (C) 1991, 1992  Linus Torvalds
15  */
16
17 #ifndef _EXT4_H
18 #define _EXT4_H
19
20 #include <linux/types.h>
21 #include <linux/blkdev.h>
22 #include <linux/magic.h>
23 #include <linux/jbd2.h>
24 #include <linux/quota.h>
25 #include <linux/rwsem.h>
26 #include <linux/rbtree.h>
27 #include <linux/seqlock.h>
28 #include <linux/mutex.h>
29 #include <linux/timer.h>
30 #include <linux/wait.h>
31 #include <linux/sched/signal.h>
32 #include <linux/blockgroup_lock.h>
33 #include <linux/percpu_counter.h>
34 #include <linux/ratelimit.h>
35 #include <crypto/hash.h>
36 #include <linux/falloc.h>
37 #include <linux/percpu-rwsem.h>
38 #include <linux/fiemap.h>
39 #ifdef __KERNEL__
40 #include <linux/compat.h>
41 #endif
42
43 #include <linux/fscrypt.h>
44 #include <linux/fsverity.h>
45
46 #include <linux/compiler.h>
47
48 /*
49  * The fourth extended filesystem constants/structures
50  */
51
52 /*
53  * with AGGRESSIVE_CHECK allocator runs consistency checks over
54  * structures. these checks slow things down a lot
55  */
56 #define AGGRESSIVE_CHECK__
57
58 /*
59  * with DOUBLE_CHECK defined mballoc creates persistent in-core
60  * bitmaps, maintains and uses them to check for double allocations
61  */
62 #define DOUBLE_CHECK__
63
64 /*
65  * Define EXT4FS_DEBUG to produce debug messages
66  */
67 #undef EXT4FS_DEBUG
68
69 /*
70  * Debug code
71  */
72 #ifdef EXT4FS_DEBUG
73 #define ext4_debug(f, a...)                                             \
74         do {                                                            \
75                 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:",        \
76                         __FILE__, __LINE__, __func__);                  \
77                 printk(KERN_DEBUG f, ## a);                             \
78         } while (0)
79 #else
80 #define ext4_debug(fmt, ...)    no_printk(fmt, ##__VA_ARGS__)
81 #endif
82
83  /*
84   * Turn on EXT_DEBUG to enable ext4_ext_show_path/leaf/move in extents.c
85   */
86 #define EXT_DEBUG__
87
88 /*
89  * Dynamic printk for controlled extents debugging.
90  */
91 #ifdef CONFIG_EXT4_DEBUG
92 #define ext_debug(ino, fmt, ...)                                        \
93         pr_debug("[%s/%d] EXT4-fs (%s): ino %lu: (%s, %d): %s:" fmt,    \
94                  current->comm, task_pid_nr(current),                   \
95                  ino->i_sb->s_id, ino->i_ino, __FILE__, __LINE__,       \
96                  __func__, ##__VA_ARGS__)
97 #else
98 #define ext_debug(ino, fmt, ...)        no_printk(fmt, ##__VA_ARGS__)
99 #endif
100
101 /* data type for block offset of block group */
102 typedef int ext4_grpblk_t;
103
104 /* data type for filesystem-wide blocks number */
105 typedef unsigned long long ext4_fsblk_t;
106
107 /* data type for file logical block number */
108 typedef __u32 ext4_lblk_t;
109
110 /* data type for block group number */
111 typedef unsigned int ext4_group_t;
112
113 enum SHIFT_DIRECTION {
114         SHIFT_LEFT = 0,
115         SHIFT_RIGHT,
116 };
117
118 /*
119  * Flags used in mballoc's allocation_context flags field.
120  *
121  * Also used to show what's going on for debugging purposes when the
122  * flag field is exported via the traceport interface
123  */
124
125 /* prefer goal again. length */
126 #define EXT4_MB_HINT_MERGE              0x0001
127 /* blocks already reserved */
128 #define EXT4_MB_HINT_RESERVED           0x0002
129 /* metadata is being allocated */
130 #define EXT4_MB_HINT_METADATA           0x0004
131 /* first blocks in the file */
132 #define EXT4_MB_HINT_FIRST              0x0008
133 /* search for the best chunk */
134 #define EXT4_MB_HINT_BEST               0x0010
135 /* data is being allocated */
136 #define EXT4_MB_HINT_DATA               0x0020
137 /* don't preallocate (for tails) */
138 #define EXT4_MB_HINT_NOPREALLOC         0x0040
139 /* allocate for locality group */
140 #define EXT4_MB_HINT_GROUP_ALLOC        0x0080
141 /* allocate goal blocks or none */
142 #define EXT4_MB_HINT_GOAL_ONLY          0x0100
143 /* goal is meaningful */
144 #define EXT4_MB_HINT_TRY_GOAL           0x0200
145 /* blocks already pre-reserved by delayed allocation */
146 #define EXT4_MB_DELALLOC_RESERVED       0x0400
147 /* We are doing stream allocation */
148 #define EXT4_MB_STREAM_ALLOC            0x0800
149 /* Use reserved root blocks if needed */
150 #define EXT4_MB_USE_ROOT_BLOCKS         0x1000
151 /* Use blocks from reserved pool */
152 #define EXT4_MB_USE_RESERVED            0x2000
153 /* Do strict check for free blocks while retrying block allocation */
154 #define EXT4_MB_STRICT_CHECK            0x4000
155
156 struct ext4_allocation_request {
157         /* target inode for block we're allocating */
158         struct inode *inode;
159         /* how many blocks we want to allocate */
160         unsigned int len;
161         /* logical block in target inode */
162         ext4_lblk_t logical;
163         /* the closest logical allocated block to the left */
164         ext4_lblk_t lleft;
165         /* the closest logical allocated block to the right */
166         ext4_lblk_t lright;
167         /* phys. target (a hint) */
168         ext4_fsblk_t goal;
169         /* phys. block for the closest logical allocated block to the left */
170         ext4_fsblk_t pleft;
171         /* phys. block for the closest logical allocated block to the right */
172         ext4_fsblk_t pright;
173         /* flags. see above EXT4_MB_HINT_* */
174         unsigned int flags;
175 };
176
177 /*
178  * Logical to physical block mapping, used by ext4_map_blocks()
179  *
180  * This structure is used to pass requests into ext4_map_blocks() as
181  * well as to store the information returned by ext4_map_blocks().  It
182  * takes less room on the stack than a struct buffer_head.
183  */
184 #define EXT4_MAP_NEW            BIT(BH_New)
185 #define EXT4_MAP_MAPPED         BIT(BH_Mapped)
186 #define EXT4_MAP_UNWRITTEN      BIT(BH_Unwritten)
187 #define EXT4_MAP_BOUNDARY       BIT(BH_Boundary)
188 #define EXT4_MAP_FLAGS          (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
189                                  EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
190
191 struct ext4_map_blocks {
192         ext4_fsblk_t m_pblk;
193         ext4_lblk_t m_lblk;
194         unsigned int m_len;
195         unsigned int m_flags;
196 };
197
198 /*
199  * Block validity checking, system zone rbtree.
200  */
201 struct ext4_system_blocks {
202         struct rb_root root;
203         struct rcu_head rcu;
204 };
205
206 /*
207  * Flags for ext4_io_end->flags
208  */
209 #define EXT4_IO_END_UNWRITTEN   0x0001
210
211 struct ext4_io_end_vec {
212         struct list_head list;          /* list of io_end_vec */
213         loff_t offset;                  /* offset in the file */
214         ssize_t size;                   /* size of the extent */
215 };
216
217 /*
218  * For converting unwritten extents on a work queue. 'handle' is used for
219  * buffered writeback.
220  */
221 typedef struct ext4_io_end {
222         struct list_head        list;           /* per-file finished IO list */
223         handle_t                *handle;        /* handle reserved for extent
224                                                  * conversion */
225         struct inode            *inode;         /* file being written to */
226         struct bio              *bio;           /* Linked list of completed
227                                                  * bios covering the extent */
228         unsigned int            flag;           /* unwritten or not */
229         atomic_t                count;          /* reference counter */
230         struct list_head        list_vec;       /* list of ext4_io_end_vec */
231 } ext4_io_end_t;
232
233 struct ext4_io_submit {
234         struct writeback_control *io_wbc;
235         struct bio              *io_bio;
236         ext4_io_end_t           *io_end;
237         sector_t                io_next_block;
238 };
239
240 /*
241  * Special inodes numbers
242  */
243 #define EXT4_BAD_INO             1      /* Bad blocks inode */
244 #define EXT4_ROOT_INO            2      /* Root inode */
245 #define EXT4_USR_QUOTA_INO       3      /* User quota inode */
246 #define EXT4_GRP_QUOTA_INO       4      /* Group quota inode */
247 #define EXT4_BOOT_LOADER_INO     5      /* Boot loader inode */
248 #define EXT4_UNDEL_DIR_INO       6      /* Undelete directory inode */
249 #define EXT4_RESIZE_INO          7      /* Reserved group descriptors inode */
250 #define EXT4_JOURNAL_INO         8      /* Journal inode */
251
252 /* First non-reserved inode for old ext4 filesystems */
253 #define EXT4_GOOD_OLD_FIRST_INO 11
254
255 /*
256  * Maximal count of links to a file
257  */
258 #define EXT4_LINK_MAX           65000
259
260 /*
261  * Macro-instructions used to manage several block sizes
262  */
263 #define EXT4_MIN_BLOCK_SIZE             1024
264 #define EXT4_MAX_BLOCK_SIZE             65536
265 #define EXT4_MIN_BLOCK_LOG_SIZE         10
266 #define EXT4_MAX_BLOCK_LOG_SIZE         16
267 #define EXT4_MAX_CLUSTER_LOG_SIZE       30
268 #ifdef __KERNEL__
269 # define EXT4_BLOCK_SIZE(s)             ((s)->s_blocksize)
270 #else
271 # define EXT4_BLOCK_SIZE(s)             (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
272 #endif
273 #define EXT4_ADDR_PER_BLOCK(s)          (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
274 #define EXT4_CLUSTER_SIZE(s)            (EXT4_BLOCK_SIZE(s) << \
275                                          EXT4_SB(s)->s_cluster_bits)
276 #ifdef __KERNEL__
277 # define EXT4_BLOCK_SIZE_BITS(s)        ((s)->s_blocksize_bits)
278 # define EXT4_CLUSTER_BITS(s)           (EXT4_SB(s)->s_cluster_bits)
279 #else
280 # define EXT4_BLOCK_SIZE_BITS(s)        ((s)->s_log_block_size + 10)
281 #endif
282 #ifdef __KERNEL__
283 #define EXT4_ADDR_PER_BLOCK_BITS(s)     (EXT4_SB(s)->s_addr_per_block_bits)
284 #define EXT4_INODE_SIZE(s)              (EXT4_SB(s)->s_inode_size)
285 #define EXT4_FIRST_INO(s)               (EXT4_SB(s)->s_first_ino)
286 #else
287 #define EXT4_INODE_SIZE(s)      (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
288                                  EXT4_GOOD_OLD_INODE_SIZE : \
289                                  (s)->s_inode_size)
290 #define EXT4_FIRST_INO(s)       (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
291                                  EXT4_GOOD_OLD_FIRST_INO : \
292                                  (s)->s_first_ino)
293 #endif
294 #define EXT4_BLOCK_ALIGN(size, blkbits)         ALIGN((size), (1 << (blkbits)))
295 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \
296         ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \
297                                                                   blkbits))
298
299 /* Translate a block number to a cluster number */
300 #define EXT4_B2C(sbi, blk)      ((blk) >> (sbi)->s_cluster_bits)
301 /* Translate a cluster number to a block number */
302 #define EXT4_C2B(sbi, cluster)  ((cluster) << (sbi)->s_cluster_bits)
303 /* Translate # of blks to # of clusters */
304 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
305                                  (sbi)->s_cluster_bits)
306 /* Mask out the low bits to get the starting block of the cluster */
307 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) &                              \
308                                   ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
309 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) &                              \
310                                   ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
311 /* Fill in the low bits to get the last block of the cluster */
312 #define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) |                            \
313                                     ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1))
314 /* Get the cluster offset */
315 #define EXT4_PBLK_COFF(s, pblk) ((pblk) &                               \
316                                  ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
317 #define EXT4_LBLK_COFF(s, lblk) ((lblk) &                               \
318                                  ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
319
320 /*
321  * Structure of a blocks group descriptor
322  */
323 struct ext4_group_desc
324 {
325         __le32  bg_block_bitmap_lo;     /* Blocks bitmap block */
326         __le32  bg_inode_bitmap_lo;     /* Inodes bitmap block */
327         __le32  bg_inode_table_lo;      /* Inodes table block */
328         __le16  bg_free_blocks_count_lo;/* Free blocks count */
329         __le16  bg_free_inodes_count_lo;/* Free inodes count */
330         __le16  bg_used_dirs_count_lo;  /* Directories count */
331         __le16  bg_flags;               /* EXT4_BG_flags (INODE_UNINIT, etc) */
332         __le32  bg_exclude_bitmap_lo;   /* Exclude bitmap for snapshots */
333         __le16  bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
334         __le16  bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
335         __le16  bg_itable_unused_lo;    /* Unused inodes count */
336         __le16  bg_checksum;            /* crc16(sb_uuid+group+desc) */
337         __le32  bg_block_bitmap_hi;     /* Blocks bitmap block MSB */
338         __le32  bg_inode_bitmap_hi;     /* Inodes bitmap block MSB */
339         __le32  bg_inode_table_hi;      /* Inodes table block MSB */
340         __le16  bg_free_blocks_count_hi;/* Free blocks count MSB */
341         __le16  bg_free_inodes_count_hi;/* Free inodes count MSB */
342         __le16  bg_used_dirs_count_hi;  /* Directories count MSB */
343         __le16  bg_itable_unused_hi;    /* Unused inodes count MSB */
344         __le32  bg_exclude_bitmap_hi;   /* Exclude bitmap block MSB */
345         __le16  bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
346         __le16  bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
347         __u32   bg_reserved;
348 };
349
350 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END        \
351         (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
352          sizeof(__le16))
353 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END        \
354         (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
355          sizeof(__le16))
356
357 /*
358  * Structure of a flex block group info
359  */
360
361 struct flex_groups {
362         atomic64_t      free_clusters;
363         atomic_t        free_inodes;
364         atomic_t        used_dirs;
365 };
366
367 #define EXT4_BG_INODE_UNINIT    0x0001 /* Inode table/bitmap not in use */
368 #define EXT4_BG_BLOCK_UNINIT    0x0002 /* Block bitmap not in use */
369 #define EXT4_BG_INODE_ZEROED    0x0004 /* On-disk itable initialized to zero */
370
371 /*
372  * Macro-instructions used to manage group descriptors
373  */
374 #define EXT4_MIN_DESC_SIZE              32
375 #define EXT4_MIN_DESC_SIZE_64BIT        64
376 #define EXT4_MAX_DESC_SIZE              EXT4_MIN_BLOCK_SIZE
377 #define EXT4_DESC_SIZE(s)               (EXT4_SB(s)->s_desc_size)
378 #ifdef __KERNEL__
379 # define EXT4_BLOCKS_PER_GROUP(s)       (EXT4_SB(s)->s_blocks_per_group)
380 # define EXT4_CLUSTERS_PER_GROUP(s)     (EXT4_SB(s)->s_clusters_per_group)
381 # define EXT4_DESC_PER_BLOCK(s)         (EXT4_SB(s)->s_desc_per_block)
382 # define EXT4_INODES_PER_GROUP(s)       (EXT4_SB(s)->s_inodes_per_group)
383 # define EXT4_DESC_PER_BLOCK_BITS(s)    (EXT4_SB(s)->s_desc_per_block_bits)
384 #else
385 # define EXT4_BLOCKS_PER_GROUP(s)       ((s)->s_blocks_per_group)
386 # define EXT4_DESC_PER_BLOCK(s)         (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
387 # define EXT4_INODES_PER_GROUP(s)       ((s)->s_inodes_per_group)
388 #endif
389
390 /*
391  * Constants relative to the data blocks
392  */
393 #define EXT4_NDIR_BLOCKS                12
394 #define EXT4_IND_BLOCK                  EXT4_NDIR_BLOCKS
395 #define EXT4_DIND_BLOCK                 (EXT4_IND_BLOCK + 1)
396 #define EXT4_TIND_BLOCK                 (EXT4_DIND_BLOCK + 1)
397 #define EXT4_N_BLOCKS                   (EXT4_TIND_BLOCK + 1)
398
399 /*
400  * Inode flags
401  */
402 #define EXT4_SECRM_FL                   0x00000001 /* Secure deletion */
403 #define EXT4_UNRM_FL                    0x00000002 /* Undelete */
404 #define EXT4_COMPR_FL                   0x00000004 /* Compress file */
405 #define EXT4_SYNC_FL                    0x00000008 /* Synchronous updates */
406 #define EXT4_IMMUTABLE_FL               0x00000010 /* Immutable file */
407 #define EXT4_APPEND_FL                  0x00000020 /* writes to file may only append */
408 #define EXT4_NODUMP_FL                  0x00000040 /* do not dump file */
409 #define EXT4_NOATIME_FL                 0x00000080 /* do not update atime */
410 /* Reserved for compression usage... */
411 #define EXT4_DIRTY_FL                   0x00000100
412 #define EXT4_COMPRBLK_FL                0x00000200 /* One or more compressed clusters */
413 #define EXT4_NOCOMPR_FL                 0x00000400 /* Don't compress */
414         /* nb: was previously EXT2_ECOMPR_FL */
415 #define EXT4_ENCRYPT_FL                 0x00000800 /* encrypted file */
416 /* End compression flags --- maybe not all used */
417 #define EXT4_INDEX_FL                   0x00001000 /* hash-indexed directory */
418 #define EXT4_IMAGIC_FL                  0x00002000 /* AFS directory */
419 #define EXT4_JOURNAL_DATA_FL            0x00004000 /* file data should be journaled */
420 #define EXT4_NOTAIL_FL                  0x00008000 /* file tail should not be merged */
421 #define EXT4_DIRSYNC_FL                 0x00010000 /* dirsync behaviour (directories only) */
422 #define EXT4_TOPDIR_FL                  0x00020000 /* Top of directory hierarchies*/
423 #define EXT4_HUGE_FILE_FL               0x00040000 /* Set to each huge file */
424 #define EXT4_EXTENTS_FL                 0x00080000 /* Inode uses extents */
425 #define EXT4_VERITY_FL                  0x00100000 /* Verity protected inode */
426 #define EXT4_EA_INODE_FL                0x00200000 /* Inode used for large EA */
427 /* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */
428
429 #define EXT4_DAX_FL                     0x02000000 /* Inode is DAX */
430
431 #define EXT4_INLINE_DATA_FL             0x10000000 /* Inode has inline data. */
432 #define EXT4_PROJINHERIT_FL             0x20000000 /* Create with parents projid */
433 #define EXT4_CASEFOLD_FL                0x40000000 /* Casefolded directory */
434 #define EXT4_RESERVED_FL                0x80000000 /* reserved for ext4 lib */
435
436 /* User modifiable flags */
437 #define EXT4_FL_USER_MODIFIABLE         (EXT4_SECRM_FL | \
438                                          EXT4_UNRM_FL | \
439                                          EXT4_COMPR_FL | \
440                                          EXT4_SYNC_FL | \
441                                          EXT4_IMMUTABLE_FL | \
442                                          EXT4_APPEND_FL | \
443                                          EXT4_NODUMP_FL | \
444                                          EXT4_NOATIME_FL | \
445                                          EXT4_JOURNAL_DATA_FL | \
446                                          EXT4_NOTAIL_FL | \
447                                          EXT4_DIRSYNC_FL | \
448                                          EXT4_TOPDIR_FL | \
449                                          EXT4_EXTENTS_FL | \
450                                          0x00400000 /* EXT4_EOFBLOCKS_FL */ | \
451                                          EXT4_DAX_FL | \
452                                          EXT4_PROJINHERIT_FL | \
453                                          EXT4_CASEFOLD_FL)
454
455 /* User visible flags */
456 #define EXT4_FL_USER_VISIBLE            (EXT4_FL_USER_MODIFIABLE | \
457                                          EXT4_DIRTY_FL | \
458                                          EXT4_COMPRBLK_FL | \
459                                          EXT4_NOCOMPR_FL | \
460                                          EXT4_ENCRYPT_FL | \
461                                          EXT4_INDEX_FL | \
462                                          EXT4_VERITY_FL | \
463                                          EXT4_INLINE_DATA_FL)
464
465 /* Flags we can manipulate with through FS_IOC_FSSETXATTR */
466 #define EXT4_FL_XFLAG_VISIBLE           (EXT4_SYNC_FL | \
467                                          EXT4_IMMUTABLE_FL | \
468                                          EXT4_APPEND_FL | \
469                                          EXT4_NODUMP_FL | \
470                                          EXT4_NOATIME_FL | \
471                                          EXT4_PROJINHERIT_FL | \
472                                          EXT4_DAX_FL)
473
474 /* Flags that should be inherited by new inodes from their parent. */
475 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
476                            EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
477                            EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
478                            EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
479                            EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL |\
480                            EXT4_DAX_FL)
481
482 /* Flags that are appropriate for regular files (all but dir-specific ones). */
483 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\
484                            EXT4_PROJINHERIT_FL))
485
486 /* Flags that are appropriate for non-directories/regular files. */
487 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
488
489 /* The only flags that should be swapped */
490 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL)
491
492 /* Flags which are mutually exclusive to DAX */
493 #define EXT4_DAX_MUT_EXCL (EXT4_VERITY_FL | EXT4_ENCRYPT_FL |\
494                            EXT4_JOURNAL_DATA_FL | EXT4_INLINE_DATA_FL)
495
496 /* Mask out flags that are inappropriate for the given type of inode. */
497 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
498 {
499         if (S_ISDIR(mode))
500                 return flags;
501         else if (S_ISREG(mode))
502                 return flags & EXT4_REG_FLMASK;
503         else
504                 return flags & EXT4_OTHER_FLMASK;
505 }
506
507 /*
508  * Inode flags used for atomic set/get
509  */
510 enum {
511         EXT4_INODE_SECRM        = 0,    /* Secure deletion */
512         EXT4_INODE_UNRM         = 1,    /* Undelete */
513         EXT4_INODE_COMPR        = 2,    /* Compress file */
514         EXT4_INODE_SYNC         = 3,    /* Synchronous updates */
515         EXT4_INODE_IMMUTABLE    = 4,    /* Immutable file */
516         EXT4_INODE_APPEND       = 5,    /* writes to file may only append */
517         EXT4_INODE_NODUMP       = 6,    /* do not dump file */
518         EXT4_INODE_NOATIME      = 7,    /* do not update atime */
519 /* Reserved for compression usage... */
520         EXT4_INODE_DIRTY        = 8,
521         EXT4_INODE_COMPRBLK     = 9,    /* One or more compressed clusters */
522         EXT4_INODE_NOCOMPR      = 10,   /* Don't compress */
523         EXT4_INODE_ENCRYPT      = 11,   /* Encrypted file */
524 /* End compression flags --- maybe not all used */
525         EXT4_INODE_INDEX        = 12,   /* hash-indexed directory */
526         EXT4_INODE_IMAGIC       = 13,   /* AFS directory */
527         EXT4_INODE_JOURNAL_DATA = 14,   /* file data should be journaled */
528         EXT4_INODE_NOTAIL       = 15,   /* file tail should not be merged */
529         EXT4_INODE_DIRSYNC      = 16,   /* dirsync behaviour (directories only) */
530         EXT4_INODE_TOPDIR       = 17,   /* Top of directory hierarchies*/
531         EXT4_INODE_HUGE_FILE    = 18,   /* Set to each huge file */
532         EXT4_INODE_EXTENTS      = 19,   /* Inode uses extents */
533         EXT4_INODE_VERITY       = 20,   /* Verity protected inode */
534         EXT4_INODE_EA_INODE     = 21,   /* Inode used for large EA */
535 /* 22 was formerly EXT4_INODE_EOFBLOCKS */
536         EXT4_INODE_DAX          = 25,   /* Inode is DAX */
537         EXT4_INODE_INLINE_DATA  = 28,   /* Data in inode. */
538         EXT4_INODE_PROJINHERIT  = 29,   /* Create with parents projid */
539         EXT4_INODE_CASEFOLD     = 30,   /* Casefolded directory */
540         EXT4_INODE_RESERVED     = 31,   /* reserved for ext4 lib */
541 };
542
543 /*
544  * Since it's pretty easy to mix up bit numbers and hex values, we use a
545  * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
546  * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
547  * any extra space in the compiled kernel image, otherwise, the build will fail.
548  * It's important that these values are the same, since we are using
549  * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
550  * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
551  * values found in ext2, ext3 and ext4 filesystems, and of course the values
552  * defined in e2fsprogs.
553  *
554  * It's not paranoia if the Murphy's Law really *is* out to get you.  :-)
555  */
556 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
557 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
558
559 static inline void ext4_check_flag_values(void)
560 {
561         CHECK_FLAG_VALUE(SECRM);
562         CHECK_FLAG_VALUE(UNRM);
563         CHECK_FLAG_VALUE(COMPR);
564         CHECK_FLAG_VALUE(SYNC);
565         CHECK_FLAG_VALUE(IMMUTABLE);
566         CHECK_FLAG_VALUE(APPEND);
567         CHECK_FLAG_VALUE(NODUMP);
568         CHECK_FLAG_VALUE(NOATIME);
569         CHECK_FLAG_VALUE(DIRTY);
570         CHECK_FLAG_VALUE(COMPRBLK);
571         CHECK_FLAG_VALUE(NOCOMPR);
572         CHECK_FLAG_VALUE(ENCRYPT);
573         CHECK_FLAG_VALUE(INDEX);
574         CHECK_FLAG_VALUE(IMAGIC);
575         CHECK_FLAG_VALUE(JOURNAL_DATA);
576         CHECK_FLAG_VALUE(NOTAIL);
577         CHECK_FLAG_VALUE(DIRSYNC);
578         CHECK_FLAG_VALUE(TOPDIR);
579         CHECK_FLAG_VALUE(HUGE_FILE);
580         CHECK_FLAG_VALUE(EXTENTS);
581         CHECK_FLAG_VALUE(VERITY);
582         CHECK_FLAG_VALUE(EA_INODE);
583         CHECK_FLAG_VALUE(INLINE_DATA);
584         CHECK_FLAG_VALUE(PROJINHERIT);
585         CHECK_FLAG_VALUE(CASEFOLD);
586         CHECK_FLAG_VALUE(RESERVED);
587 }
588
589 /* Used to pass group descriptor data when online resize is done */
590 struct ext4_new_group_input {
591         __u32 group;            /* Group number for this data */
592         __u64 block_bitmap;     /* Absolute block number of block bitmap */
593         __u64 inode_bitmap;     /* Absolute block number of inode bitmap */
594         __u64 inode_table;      /* Absolute block number of inode table start */
595         __u32 blocks_count;     /* Total number of blocks in this group */
596         __u16 reserved_blocks;  /* Number of reserved blocks in this group */
597         __u16 unused;
598 };
599
600 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
601 struct compat_ext4_new_group_input {
602         u32 group;
603         compat_u64 block_bitmap;
604         compat_u64 inode_bitmap;
605         compat_u64 inode_table;
606         u32 blocks_count;
607         u16 reserved_blocks;
608         u16 unused;
609 };
610 #endif
611
612 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
613 struct ext4_new_group_data {
614         __u32 group;
615         __u64 block_bitmap;
616         __u64 inode_bitmap;
617         __u64 inode_table;
618         __u32 blocks_count;
619         __u16 reserved_blocks;
620         __u16 mdata_blocks;
621         __u32 free_clusters_count;
622 };
623
624 /* Indexes used to index group tables in ext4_new_group_data */
625 enum {
626         BLOCK_BITMAP = 0,       /* block bitmap */
627         INODE_BITMAP,           /* inode bitmap */
628         INODE_TABLE,            /* inode tables */
629         GROUP_TABLE_COUNT,
630 };
631
632 /*
633  * Flags used by ext4_map_blocks()
634  */
635         /* Allocate any needed blocks and/or convert an unwritten
636            extent to be an initialized ext4 */
637 #define EXT4_GET_BLOCKS_CREATE                  0x0001
638         /* Request the creation of an unwritten extent */
639 #define EXT4_GET_BLOCKS_UNWRIT_EXT              0x0002
640 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT       (EXT4_GET_BLOCKS_UNWRIT_EXT|\
641                                                  EXT4_GET_BLOCKS_CREATE)
642         /* Caller is from the delayed allocation writeout path
643          * finally doing the actual allocation of delayed blocks */
644 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE        0x0004
645         /* caller is from the direct IO path, request to creation of an
646         unwritten extents if not allocated, split the unwritten
647         extent if blocks has been preallocated already*/
648 #define EXT4_GET_BLOCKS_PRE_IO                  0x0008
649 #define EXT4_GET_BLOCKS_CONVERT                 0x0010
650 #define EXT4_GET_BLOCKS_IO_CREATE_EXT           (EXT4_GET_BLOCKS_PRE_IO|\
651                                          EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
652         /* Convert extent to initialized after IO complete */
653 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT          (EXT4_GET_BLOCKS_CONVERT|\
654                                          EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
655         /* Eventual metadata allocation (due to growing extent tree)
656          * should not fail, so try to use reserved blocks for that.*/
657 #define EXT4_GET_BLOCKS_METADATA_NOFAIL         0x0020
658         /* Don't normalize allocation size (used for fallocate) */
659 #define EXT4_GET_BLOCKS_NO_NORMALIZE            0x0040
660         /* Convert written extents to unwritten */
661 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN       0x0100
662         /* Write zeros to newly created written extents */
663 #define EXT4_GET_BLOCKS_ZERO                    0x0200
664 #define EXT4_GET_BLOCKS_CREATE_ZERO             (EXT4_GET_BLOCKS_CREATE |\
665                                         EXT4_GET_BLOCKS_ZERO)
666         /* Caller will submit data before dropping transaction handle. This
667          * allows jbd2 to avoid submitting data before commit. */
668 #define EXT4_GET_BLOCKS_IO_SUBMIT               0x0400
669
670 /*
671  * The bit position of these flags must not overlap with any of the
672  * EXT4_GET_BLOCKS_*.  They are used by ext4_find_extent(),
673  * read_extent_tree_block(), ext4_split_extent_at(),
674  * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
675  * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
676  * caching the extents when reading from the extent tree while a
677  * truncate or punch hole operation is in progress.
678  */
679 #define EXT4_EX_NOCACHE                         0x40000000
680 #define EXT4_EX_FORCE_CACHE                     0x20000000
681 #define EXT4_EX_NOFAIL                          0x10000000
682
683 /*
684  * Flags used by ext4_free_blocks
685  */
686 #define EXT4_FREE_BLOCKS_METADATA               0x0001
687 #define EXT4_FREE_BLOCKS_FORGET                 0x0002
688 #define EXT4_FREE_BLOCKS_VALIDATED              0x0004
689 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE         0x0008
690 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER   0x0010
691 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER    0x0020
692 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER      0x0040
693
694 /*
695  * ioctl commands
696  */
697 #define EXT4_IOC_GETVERSION             _IOR('f', 3, long)
698 #define EXT4_IOC_SETVERSION             _IOW('f', 4, long)
699 #define EXT4_IOC_GETVERSION_OLD         FS_IOC_GETVERSION
700 #define EXT4_IOC_SETVERSION_OLD         FS_IOC_SETVERSION
701 #define EXT4_IOC_GETRSVSZ               _IOR('f', 5, long)
702 #define EXT4_IOC_SETRSVSZ               _IOW('f', 6, long)
703 #define EXT4_IOC_GROUP_EXTEND           _IOW('f', 7, unsigned long)
704 #define EXT4_IOC_GROUP_ADD              _IOW('f', 8, struct ext4_new_group_input)
705 #define EXT4_IOC_MIGRATE                _IO('f', 9)
706  /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
707  /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
708 #define EXT4_IOC_ALLOC_DA_BLKS          _IO('f', 12)
709 #define EXT4_IOC_MOVE_EXT               _IOWR('f', 15, struct move_extent)
710 #define EXT4_IOC_RESIZE_FS              _IOW('f', 16, __u64)
711 #define EXT4_IOC_SWAP_BOOT              _IO('f', 17)
712 #define EXT4_IOC_PRECACHE_EXTENTS       _IO('f', 18)
713 /* ioctl codes 19--39 are reserved for fscrypt */
714 #define EXT4_IOC_CLEAR_ES_CACHE         _IO('f', 40)
715 #define EXT4_IOC_GETSTATE               _IOW('f', 41, __u32)
716 #define EXT4_IOC_GET_ES_CACHE           _IOWR('f', 42, struct fiemap)
717
718 #define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32)
719
720 /*
721  * Flags for going down operation
722  */
723 #define EXT4_GOING_FLAGS_DEFAULT                0x0     /* going down */
724 #define EXT4_GOING_FLAGS_LOGFLUSH               0x1     /* flush log but not data */
725 #define EXT4_GOING_FLAGS_NOLOGFLUSH             0x2     /* don't flush log nor data */
726
727 /*
728  * Flags returned by EXT4_IOC_GETSTATE
729  *
730  * We only expose to userspace a subset of the state flags in
731  * i_state_flags
732  */
733 #define EXT4_STATE_FLAG_EXT_PRECACHED   0x00000001
734 #define EXT4_STATE_FLAG_NEW             0x00000002
735 #define EXT4_STATE_FLAG_NEWENTRY        0x00000004
736 #define EXT4_STATE_FLAG_DA_ALLOC_CLOSE  0x00000008
737
738 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
739 /*
740  * ioctl commands in 32 bit emulation
741  */
742 #define EXT4_IOC32_GETVERSION           _IOR('f', 3, int)
743 #define EXT4_IOC32_SETVERSION           _IOW('f', 4, int)
744 #define EXT4_IOC32_GETRSVSZ             _IOR('f', 5, int)
745 #define EXT4_IOC32_SETRSVSZ             _IOW('f', 6, int)
746 #define EXT4_IOC32_GROUP_EXTEND         _IOW('f', 7, unsigned int)
747 #define EXT4_IOC32_GROUP_ADD            _IOW('f', 8, struct compat_ext4_new_group_input)
748 #define EXT4_IOC32_GETVERSION_OLD       FS_IOC32_GETVERSION
749 #define EXT4_IOC32_SETVERSION_OLD       FS_IOC32_SETVERSION
750 #endif
751
752 /*
753  * Returned by EXT4_IOC_GET_ES_CACHE as an additional possible flag.
754  * It indicates that the entry in extent status cache is for a hole.
755  */
756 #define EXT4_FIEMAP_EXTENT_HOLE         0x08000000
757
758 /* Max physical block we can address w/o extents */
759 #define EXT4_MAX_BLOCK_FILE_PHYS        0xFFFFFFFF
760
761 /* Max logical block we can support */
762 #define EXT4_MAX_LOGICAL_BLOCK          0xFFFFFFFE
763
764 /*
765  * Structure of an inode on the disk
766  */
767 struct ext4_inode {
768         __le16  i_mode;         /* File mode */
769         __le16  i_uid;          /* Low 16 bits of Owner Uid */
770         __le32  i_size_lo;      /* Size in bytes */
771         __le32  i_atime;        /* Access time */
772         __le32  i_ctime;        /* Inode Change time */
773         __le32  i_mtime;        /* Modification time */
774         __le32  i_dtime;        /* Deletion Time */
775         __le16  i_gid;          /* Low 16 bits of Group Id */
776         __le16  i_links_count;  /* Links count */
777         __le32  i_blocks_lo;    /* Blocks count */
778         __le32  i_flags;        /* File flags */
779         union {
780                 struct {
781                         __le32  l_i_version;
782                 } linux1;
783                 struct {
784                         __u32  h_i_translator;
785                 } hurd1;
786                 struct {
787                         __u32  m_i_reserved1;
788                 } masix1;
789         } osd1;                         /* OS dependent 1 */
790         __le32  i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
791         __le32  i_generation;   /* File version (for NFS) */
792         __le32  i_file_acl_lo;  /* File ACL */
793         __le32  i_size_high;
794         __le32  i_obso_faddr;   /* Obsoleted fragment address */
795         union {
796                 struct {
797                         __le16  l_i_blocks_high; /* were l_i_reserved1 */
798                         __le16  l_i_file_acl_high;
799                         __le16  l_i_uid_high;   /* these 2 fields */
800                         __le16  l_i_gid_high;   /* were reserved2[0] */
801                         __le16  l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
802                         __le16  l_i_reserved;
803                 } linux2;
804                 struct {
805                         __le16  h_i_reserved1;  /* Obsoleted fragment number/size which are removed in ext4 */
806                         __u16   h_i_mode_high;
807                         __u16   h_i_uid_high;
808                         __u16   h_i_gid_high;
809                         __u32   h_i_author;
810                 } hurd2;
811                 struct {
812                         __le16  h_i_reserved1;  /* Obsoleted fragment number/size which are removed in ext4 */
813                         __le16  m_i_file_acl_high;
814                         __u32   m_i_reserved2[2];
815                 } masix2;
816         } osd2;                         /* OS dependent 2 */
817         __le16  i_extra_isize;
818         __le16  i_checksum_hi;  /* crc32c(uuid+inum+inode) BE */
819         __le32  i_ctime_extra;  /* extra Change time      (nsec << 2 | epoch) */
820         __le32  i_mtime_extra;  /* extra Modification time(nsec << 2 | epoch) */
821         __le32  i_atime_extra;  /* extra Access time      (nsec << 2 | epoch) */
822         __le32  i_crtime;       /* File Creation time */
823         __le32  i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
824         __le32  i_version_hi;   /* high 32 bits for 64-bit version */
825         __le32  i_projid;       /* Project ID */
826 };
827
828 struct move_extent {
829         __u32 reserved;         /* should be zero */
830         __u32 donor_fd;         /* donor file descriptor */
831         __u64 orig_start;       /* logical start offset in block for orig */
832         __u64 donor_start;      /* logical start offset in block for donor */
833         __u64 len;              /* block length to be moved */
834         __u64 moved_len;        /* moved block length */
835 };
836
837 #define EXT4_EPOCH_BITS 2
838 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
839 #define EXT4_NSEC_MASK  (~0UL << EXT4_EPOCH_BITS)
840
841 /*
842  * Extended fields will fit into an inode if the filesystem was formatted
843  * with large inodes (-I 256 or larger) and there are not currently any EAs
844  * consuming all of the available space. For new inodes we always reserve
845  * enough space for the kernel's known extended fields, but for inodes
846  * created with an old kernel this might not have been the case. None of
847  * the extended inode fields is critical for correct filesystem operation.
848  * This macro checks if a certain field fits in the inode. Note that
849  * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
850  */
851 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field)   \
852         ((offsetof(typeof(*ext4_inode), field) +        \
853           sizeof((ext4_inode)->field))                  \
854         <= (EXT4_GOOD_OLD_INODE_SIZE +                  \
855             (einode)->i_extra_isize))                   \
856
857 /*
858  * We use an encoding that preserves the times for extra epoch "00":
859  *
860  * extra  msb of                         adjust for signed
861  * epoch  32-bit                         32-bit tv_sec to
862  * bits   time    decoded 64-bit tv_sec  64-bit tv_sec      valid time range
863  * 0 0    1    -0x80000000..-0x00000001  0x000000000 1901-12-13..1969-12-31
864  * 0 0    0    0x000000000..0x07fffffff  0x000000000 1970-01-01..2038-01-19
865  * 0 1    1    0x080000000..0x0ffffffff  0x100000000 2038-01-19..2106-02-07
866  * 0 1    0    0x100000000..0x17fffffff  0x100000000 2106-02-07..2174-02-25
867  * 1 0    1    0x180000000..0x1ffffffff  0x200000000 2174-02-25..2242-03-16
868  * 1 0    0    0x200000000..0x27fffffff  0x200000000 2242-03-16..2310-04-04
869  * 1 1    1    0x280000000..0x2ffffffff  0x300000000 2310-04-04..2378-04-22
870  * 1 1    0    0x300000000..0x37fffffff  0x300000000 2378-04-22..2446-05-10
871  *
872  * Note that previous versions of the kernel on 64-bit systems would
873  * incorrectly use extra epoch bits 1,1 for dates between 1901 and
874  * 1970.  e2fsck will correct this, assuming that it is run on the
875  * affected filesystem before 2242.
876  */
877
878 static inline __le32 ext4_encode_extra_time(struct timespec64 *time)
879 {
880         u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK;
881         return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
882 }
883
884 static inline void ext4_decode_extra_time(struct timespec64 *time,
885                                           __le32 extra)
886 {
887         if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK)))
888                 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
889         time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
890 }
891
892 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode)                           \
893 do {                                                                            \
894         if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     {\
895                 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec);        \
896                 (raw_inode)->xtime ## _extra =                                  \
897                                 ext4_encode_extra_time(&(inode)->xtime);        \
898                 }                                                               \
899         else    \
900                 (raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (inode)->xtime.tv_sec, S32_MIN, S32_MAX));    \
901 } while (0)
902
903 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode)                        \
904 do {                                                                           \
905         if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))                      \
906                 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec);      \
907         if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))            \
908                 (raw_inode)->xtime ## _extra =                                 \
909                                 ext4_encode_extra_time(&(einode)->xtime);      \
910 } while (0)
911
912 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode)                           \
913 do {                                                                            \
914         (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime);        \
915         if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {    \
916                 ext4_decode_extra_time(&(inode)->xtime,                         \
917                                        raw_inode->xtime ## _extra);             \
918                 }                                                               \
919         else                                                                    \
920                 (inode)->xtime.tv_nsec = 0;                                     \
921 } while (0)
922
923
924 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode)                        \
925 do {                                                                           \
926         if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))                      \
927                 (einode)->xtime.tv_sec =                                       \
928                         (signed)le32_to_cpu((raw_inode)->xtime);               \
929         else                                                                   \
930                 (einode)->xtime.tv_sec = 0;                                    \
931         if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))            \
932                 ext4_decode_extra_time(&(einode)->xtime,                       \
933                                        raw_inode->xtime ## _extra);            \
934         else                                                                   \
935                 (einode)->xtime.tv_nsec = 0;                                   \
936 } while (0)
937
938 #define i_disk_version osd1.linux1.l_i_version
939
940 #if defined(__KERNEL__) || defined(__linux__)
941 #define i_reserved1     osd1.linux1.l_i_reserved1
942 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
943 #define i_blocks_high   osd2.linux2.l_i_blocks_high
944 #define i_uid_low       i_uid
945 #define i_gid_low       i_gid
946 #define i_uid_high      osd2.linux2.l_i_uid_high
947 #define i_gid_high      osd2.linux2.l_i_gid_high
948 #define i_checksum_lo   osd2.linux2.l_i_checksum_lo
949
950 #elif defined(__GNU__)
951
952 #define i_translator    osd1.hurd1.h_i_translator
953 #define i_uid_high      osd2.hurd2.h_i_uid_high
954 #define i_gid_high      osd2.hurd2.h_i_gid_high
955 #define i_author        osd2.hurd2.h_i_author
956
957 #elif defined(__masix__)
958
959 #define i_reserved1     osd1.masix1.m_i_reserved1
960 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
961 #define i_reserved2     osd2.masix2.m_i_reserved2
962
963 #endif /* defined(__KERNEL__) || defined(__linux__) */
964
965 #include "extents_status.h"
966
967 /*
968  * Lock subclasses for i_data_sem in the ext4_inode_info structure.
969  *
970  * These are needed to avoid lockdep false positives when we need to
971  * allocate blocks to the quota inode during ext4_map_blocks(), while
972  * holding i_data_sem for a normal (non-quota) inode.  Since we don't
973  * do quota tracking for the quota inode, this avoids deadlock (as
974  * well as infinite recursion, since it isn't turtles all the way
975  * down...)
976  *
977  *  I_DATA_SEM_NORMAL - Used for most inodes
978  *  I_DATA_SEM_OTHER  - Used by move_inode.c for the second normal inode
979  *                        where the second inode has larger inode number
980  *                        than the first
981  *  I_DATA_SEM_QUOTA  - Used for quota inodes only
982  */
983 enum {
984         I_DATA_SEM_NORMAL = 0,
985         I_DATA_SEM_OTHER,
986         I_DATA_SEM_QUOTA,
987 };
988
989
990 /*
991  * fourth extended file system inode data in memory
992  */
993 struct ext4_inode_info {
994         __le32  i_data[15];     /* unconverted */
995         __u32   i_dtime;
996         ext4_fsblk_t    i_file_acl;
997
998         /*
999          * i_block_group is the number of the block group which contains
1000          * this file's inode.  Constant across the lifetime of the inode,
1001          * it is used for making block allocation decisions - we try to
1002          * place a file's data blocks near its inode block, and new inodes
1003          * near to their parent directory's inode.
1004          */
1005         ext4_group_t    i_block_group;
1006         ext4_lblk_t     i_dir_start_lookup;
1007 #if (BITS_PER_LONG < 64)
1008         unsigned long   i_state_flags;          /* Dynamic state flags */
1009 #endif
1010         unsigned long   i_flags;
1011
1012         /*
1013          * Extended attributes can be read independently of the main file
1014          * data. Taking i_mutex even when reading would cause contention
1015          * between readers of EAs and writers of regular file data, so
1016          * instead we synchronize on xattr_sem when reading or changing
1017          * EAs.
1018          */
1019         struct rw_semaphore xattr_sem;
1020
1021         struct list_head i_orphan;      /* unlinked but open inodes */
1022
1023         /*
1024          * i_disksize keeps track of what the inode size is ON DISK, not
1025          * in memory.  During truncate, i_size is set to the new size by
1026          * the VFS prior to calling ext4_truncate(), but the filesystem won't
1027          * set i_disksize to 0 until the truncate is actually under way.
1028          *
1029          * The intent is that i_disksize always represents the blocks which
1030          * are used by this file.  This allows recovery to restart truncate
1031          * on orphans if we crash during truncate.  We actually write i_disksize
1032          * into the on-disk inode when writing inodes out, instead of i_size.
1033          *
1034          * The only time when i_disksize and i_size may be different is when
1035          * a truncate is in progress.  The only things which change i_disksize
1036          * are ext4_get_block (growth) and ext4_truncate (shrinkth).
1037          */
1038         loff_t  i_disksize;
1039
1040         /*
1041          * i_data_sem is for serialising ext4_truncate() against
1042          * ext4_getblock().  In the 2.4 ext2 design, great chunks of inode's
1043          * data tree are chopped off during truncate. We can't do that in
1044          * ext4 because whenever we perform intermediate commits during
1045          * truncate, the inode and all the metadata blocks *must* be in a
1046          * consistent state which allows truncation of the orphans to restart
1047          * during recovery.  Hence we must fix the get_block-vs-truncate race
1048          * by other means, so we have i_data_sem.
1049          */
1050         struct rw_semaphore i_data_sem;
1051         /*
1052          * i_mmap_sem is for serializing page faults with truncate / punch hole
1053          * operations. We have to make sure that new page cannot be faulted in
1054          * a section of the inode that is being punched. We cannot easily use
1055          * i_data_sem for this since we need protection for the whole punch
1056          * operation and i_data_sem ranks below transaction start so we have
1057          * to occasionally drop it.
1058          */
1059         struct rw_semaphore i_mmap_sem;
1060         struct inode vfs_inode;
1061         struct jbd2_inode *jinode;
1062
1063         spinlock_t i_raw_lock;  /* protects updates to the raw inode */
1064
1065         /*
1066          * File creation time. Its function is same as that of
1067          * struct timespec64 i_{a,c,m}time in the generic inode.
1068          */
1069         struct timespec64 i_crtime;
1070
1071         /* mballoc */
1072         atomic_t i_prealloc_active;
1073         struct list_head i_prealloc_list;
1074         spinlock_t i_prealloc_lock;
1075
1076         /* extents status tree */
1077         struct ext4_es_tree i_es_tree;
1078         rwlock_t i_es_lock;
1079         struct list_head i_es_list;
1080         unsigned int i_es_all_nr;       /* protected by i_es_lock */
1081         unsigned int i_es_shk_nr;       /* protected by i_es_lock */
1082         ext4_lblk_t i_es_shrink_lblk;   /* Offset where we start searching for
1083                                            extents to shrink. Protected by
1084                                            i_es_lock  */
1085
1086         /* ialloc */
1087         ext4_group_t    i_last_alloc_group;
1088
1089         /* allocation reservation info for delalloc */
1090         /* In case of bigalloc, this refer to clusters rather than blocks */
1091         unsigned int i_reserved_data_blocks;
1092
1093         /* pending cluster reservations for bigalloc file systems */
1094         struct ext4_pending_tree i_pending_tree;
1095
1096         /* on-disk additional length */
1097         __u16 i_extra_isize;
1098
1099         /* Indicate the inline data space. */
1100         u16 i_inline_off;
1101         u16 i_inline_size;
1102
1103 #ifdef CONFIG_QUOTA
1104         /* quota space reservation, managed internally by quota code */
1105         qsize_t i_reserved_quota;
1106 #endif
1107
1108         /* Lock protecting lists below */
1109         spinlock_t i_completed_io_lock;
1110         /*
1111          * Completed IOs that need unwritten extents handling and have
1112          * transaction reserved
1113          */
1114         struct list_head i_rsv_conversion_list;
1115         struct work_struct i_rsv_conversion_work;
1116         atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1117
1118         spinlock_t i_block_reservation_lock;
1119
1120         /*
1121          * Transactions that contain inode's metadata needed to complete
1122          * fsync and fdatasync, respectively.
1123          */
1124         tid_t i_sync_tid;
1125         tid_t i_datasync_tid;
1126
1127 #ifdef CONFIG_QUOTA
1128         struct dquot *i_dquot[MAXQUOTAS];
1129 #endif
1130
1131         /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1132         __u32 i_csum_seed;
1133
1134         kprojid_t i_projid;
1135 };
1136
1137 /*
1138  * File system states
1139  */
1140 #define EXT4_VALID_FS                   0x0001  /* Unmounted cleanly */
1141 #define EXT4_ERROR_FS                   0x0002  /* Errors detected */
1142 #define EXT4_ORPHAN_FS                  0x0004  /* Orphans being recovered */
1143
1144 /*
1145  * Misc. filesystem flags
1146  */
1147 #define EXT2_FLAGS_SIGNED_HASH          0x0001  /* Signed dirhash in use */
1148 #define EXT2_FLAGS_UNSIGNED_HASH        0x0002  /* Unsigned dirhash in use */
1149 #define EXT2_FLAGS_TEST_FILESYS         0x0004  /* to test development code */
1150
1151 /*
1152  * Mount flags set via mount options or defaults
1153  */
1154 #define EXT4_MOUNT_NO_MBCACHE           0x00001 /* Do not use mbcache */
1155 #define EXT4_MOUNT_GRPID                0x00004 /* Create files with directory's group */
1156 #define EXT4_MOUNT_DEBUG                0x00008 /* Some debugging messages */
1157 #define EXT4_MOUNT_ERRORS_CONT          0x00010 /* Continue on errors */
1158 #define EXT4_MOUNT_ERRORS_RO            0x00020 /* Remount fs ro on errors */
1159 #define EXT4_MOUNT_ERRORS_PANIC         0x00040 /* Panic on errors */
1160 #define EXT4_MOUNT_ERRORS_MASK          0x00070
1161 #define EXT4_MOUNT_MINIX_DF             0x00080 /* Mimics the Minix statfs */
1162 #define EXT4_MOUNT_NOLOAD               0x00100 /* Don't use existing journal*/
1163 #ifdef CONFIG_FS_DAX
1164 #define EXT4_MOUNT_DAX_ALWAYS           0x00200 /* Direct Access */
1165 #else
1166 #define EXT4_MOUNT_DAX_ALWAYS           0
1167 #endif
1168 #define EXT4_MOUNT_DATA_FLAGS           0x00C00 /* Mode for data writes: */
1169 #define EXT4_MOUNT_JOURNAL_DATA         0x00400 /* Write data to journal */
1170 #define EXT4_MOUNT_ORDERED_DATA         0x00800 /* Flush data before commit */
1171 #define EXT4_MOUNT_WRITEBACK_DATA       0x00C00 /* No data ordering */
1172 #define EXT4_MOUNT_UPDATE_JOURNAL       0x01000 /* Update the journal format */
1173 #define EXT4_MOUNT_NO_UID32             0x02000  /* Disable 32-bit UIDs */
1174 #define EXT4_MOUNT_XATTR_USER           0x04000 /* Extended user attributes */
1175 #define EXT4_MOUNT_POSIX_ACL            0x08000 /* POSIX Access Control Lists */
1176 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC     0x10000 /* No auto delalloc mapping */
1177 #define EXT4_MOUNT_BARRIER              0x20000 /* Use block barriers */
1178 #define EXT4_MOUNT_QUOTA                0x40000 /* Some quota option set */
1179 #define EXT4_MOUNT_USRQUOTA             0x80000 /* "old" user quota,
1180                                                  * enable enforcement for hidden
1181                                                  * quota files */
1182 #define EXT4_MOUNT_GRPQUOTA             0x100000 /* "old" group quota, enable
1183                                                   * enforcement for hidden quota
1184                                                   * files */
1185 #define EXT4_MOUNT_PRJQUOTA             0x200000 /* Enable project quota
1186                                                   * enforcement */
1187 #define EXT4_MOUNT_DIOREAD_NOLOCK       0x400000 /* Enable support for dio read nolocking */
1188 #define EXT4_MOUNT_JOURNAL_CHECKSUM     0x800000 /* Journal checksums */
1189 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1190 #define EXT4_MOUNT_WARN_ON_ERROR        0x2000000 /* Trigger WARN_ON on error */
1191 #define EXT4_MOUNT_PREFETCH_BLOCK_BITMAPS 0x4000000
1192 #define EXT4_MOUNT_DELALLOC             0x8000000 /* Delalloc support */
1193 #define EXT4_MOUNT_DATA_ERR_ABORT       0x10000000 /* Abort on file data write */
1194 #define EXT4_MOUNT_BLOCK_VALIDITY       0x20000000 /* Block validity checking */
1195 #define EXT4_MOUNT_DISCARD              0x40000000 /* Issue DISCARD requests */
1196 #define EXT4_MOUNT_INIT_INODE_TABLE     0x80000000 /* Initialize uninitialized itables */
1197
1198 /*
1199  * Mount flags set either automatically (could not be set by mount option)
1200  * based on per file system feature or property or in special cases such as
1201  * distinguishing between explicit mount option definition and default.
1202  */
1203 #define EXT4_MOUNT2_EXPLICIT_DELALLOC   0x00000001 /* User explicitly
1204                                                       specified delalloc */
1205 #define EXT4_MOUNT2_STD_GROUP_SIZE      0x00000002 /* We have standard group
1206                                                       size of blocksize * 8
1207                                                       blocks */
1208 #define EXT4_MOUNT2_HURD_COMPAT         0x00000004 /* Support HURD-castrated
1209                                                       file systems */
1210 #define EXT4_MOUNT2_DAX_NEVER           0x00000008 /* Do not allow Direct Access */
1211 #define EXT4_MOUNT2_DAX_INODE           0x00000010 /* For printing options only */
1212
1213 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM   0x00000008 /* User explicitly
1214                                                 specified journal checksum */
1215
1216 #define clear_opt(sb, opt)              EXT4_SB(sb)->s_mount_opt &= \
1217                                                 ~EXT4_MOUNT_##opt
1218 #define set_opt(sb, opt)                EXT4_SB(sb)->s_mount_opt |= \
1219                                                 EXT4_MOUNT_##opt
1220 #define test_opt(sb, opt)               (EXT4_SB(sb)->s_mount_opt & \
1221                                          EXT4_MOUNT_##opt)
1222
1223 #define clear_opt2(sb, opt)             EXT4_SB(sb)->s_mount_opt2 &= \
1224                                                 ~EXT4_MOUNT2_##opt
1225 #define set_opt2(sb, opt)               EXT4_SB(sb)->s_mount_opt2 |= \
1226                                                 EXT4_MOUNT2_##opt
1227 #define test_opt2(sb, opt)              (EXT4_SB(sb)->s_mount_opt2 & \
1228                                          EXT4_MOUNT2_##opt)
1229
1230 #define ext4_test_and_set_bit           __test_and_set_bit_le
1231 #define ext4_set_bit                    __set_bit_le
1232 #define ext4_set_bit_atomic             ext2_set_bit_atomic
1233 #define ext4_test_and_clear_bit         __test_and_clear_bit_le
1234 #define ext4_clear_bit                  __clear_bit_le
1235 #define ext4_clear_bit_atomic           ext2_clear_bit_atomic
1236 #define ext4_test_bit                   test_bit_le
1237 #define ext4_find_next_zero_bit         find_next_zero_bit_le
1238 #define ext4_find_next_bit              find_next_bit_le
1239
1240 extern void ext4_set_bits(void *bm, int cur, int len);
1241
1242 /*
1243  * Maximal mount counts between two filesystem checks
1244  */
1245 #define EXT4_DFL_MAX_MNT_COUNT          20      /* Allow 20 mounts */
1246 #define EXT4_DFL_CHECKINTERVAL          0       /* Don't use interval check */
1247
1248 /*
1249  * Behaviour when detecting errors
1250  */
1251 #define EXT4_ERRORS_CONTINUE            1       /* Continue execution */
1252 #define EXT4_ERRORS_RO                  2       /* Remount fs read-only */
1253 #define EXT4_ERRORS_PANIC               3       /* Panic */
1254 #define EXT4_ERRORS_DEFAULT             EXT4_ERRORS_CONTINUE
1255
1256 /* Metadata checksum algorithm codes */
1257 #define EXT4_CRC32C_CHKSUM              1
1258
1259 /*
1260  * Structure of the super block
1261  */
1262 struct ext4_super_block {
1263 /*00*/  __le32  s_inodes_count;         /* Inodes count */
1264         __le32  s_blocks_count_lo;      /* Blocks count */
1265         __le32  s_r_blocks_count_lo;    /* Reserved blocks count */
1266         __le32  s_free_blocks_count_lo; /* Free blocks count */
1267 /*10*/  __le32  s_free_inodes_count;    /* Free inodes count */
1268         __le32  s_first_data_block;     /* First Data Block */
1269         __le32  s_log_block_size;       /* Block size */
1270         __le32  s_log_cluster_size;     /* Allocation cluster size */
1271 /*20*/  __le32  s_blocks_per_group;     /* # Blocks per group */
1272         __le32  s_clusters_per_group;   /* # Clusters per group */
1273         __le32  s_inodes_per_group;     /* # Inodes per group */
1274         __le32  s_mtime;                /* Mount time */
1275 /*30*/  __le32  s_wtime;                /* Write time */
1276         __le16  s_mnt_count;            /* Mount count */
1277         __le16  s_max_mnt_count;        /* Maximal mount count */
1278         __le16  s_magic;                /* Magic signature */
1279         __le16  s_state;                /* File system state */
1280         __le16  s_errors;               /* Behaviour when detecting errors */
1281         __le16  s_minor_rev_level;      /* minor revision level */
1282 /*40*/  __le32  s_lastcheck;            /* time of last check */
1283         __le32  s_checkinterval;        /* max. time between checks */
1284         __le32  s_creator_os;           /* OS */
1285         __le32  s_rev_level;            /* Revision level */
1286 /*50*/  __le16  s_def_resuid;           /* Default uid for reserved blocks */
1287         __le16  s_def_resgid;           /* Default gid for reserved blocks */
1288         /*
1289          * These fields are for EXT4_DYNAMIC_REV superblocks only.
1290          *
1291          * Note: the difference between the compatible feature set and
1292          * the incompatible feature set is that if there is a bit set
1293          * in the incompatible feature set that the kernel doesn't
1294          * know about, it should refuse to mount the filesystem.
1295          *
1296          * e2fsck's requirements are more strict; if it doesn't know
1297          * about a feature in either the compatible or incompatible
1298          * feature set, it must abort and not try to meddle with
1299          * things it doesn't understand...
1300          */
1301         __le32  s_first_ino;            /* First non-reserved inode */
1302         __le16  s_inode_size;           /* size of inode structure */
1303         __le16  s_block_group_nr;       /* block group # of this superblock */
1304         __le32  s_feature_compat;       /* compatible feature set */
1305 /*60*/  __le32  s_feature_incompat;     /* incompatible feature set */
1306         __le32  s_feature_ro_compat;    /* readonly-compatible feature set */
1307 /*68*/  __u8    s_uuid[16];             /* 128-bit uuid for volume */
1308 /*78*/  char    s_volume_name[16];      /* volume name */
1309 /*88*/  char    s_last_mounted[64] __nonstring; /* directory where last mounted */
1310 /*C8*/  __le32  s_algorithm_usage_bitmap; /* For compression */
1311         /*
1312          * Performance hints.  Directory preallocation should only
1313          * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1314          */
1315         __u8    s_prealloc_blocks;      /* Nr of blocks to try to preallocate*/
1316         __u8    s_prealloc_dir_blocks;  /* Nr to preallocate for dirs */
1317         __le16  s_reserved_gdt_blocks;  /* Per group desc for online growth */
1318         /*
1319          * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1320          */
1321 /*D0*/  __u8    s_journal_uuid[16];     /* uuid of journal superblock */
1322 /*E0*/  __le32  s_journal_inum;         /* inode number of journal file */
1323         __le32  s_journal_dev;          /* device number of journal file */
1324         __le32  s_last_orphan;          /* start of list of inodes to delete */
1325         __le32  s_hash_seed[4];         /* HTREE hash seed */
1326         __u8    s_def_hash_version;     /* Default hash version to use */
1327         __u8    s_jnl_backup_type;
1328         __le16  s_desc_size;            /* size of group descriptor */
1329 /*100*/ __le32  s_default_mount_opts;
1330         __le32  s_first_meta_bg;        /* First metablock block group */
1331         __le32  s_mkfs_time;            /* When the filesystem was created */
1332         __le32  s_jnl_blocks[17];       /* Backup of the journal inode */
1333         /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1334 /*150*/ __le32  s_blocks_count_hi;      /* Blocks count */
1335         __le32  s_r_blocks_count_hi;    /* Reserved blocks count */
1336         __le32  s_free_blocks_count_hi; /* Free blocks count */
1337         __le16  s_min_extra_isize;      /* All inodes have at least # bytes */
1338         __le16  s_want_extra_isize;     /* New inodes should reserve # bytes */
1339         __le32  s_flags;                /* Miscellaneous flags */
1340         __le16  s_raid_stride;          /* RAID stride */
1341         __le16  s_mmp_update_interval;  /* # seconds to wait in MMP checking */
1342         __le64  s_mmp_block;            /* Block for multi-mount protection */
1343         __le32  s_raid_stripe_width;    /* blocks on all data disks (N*stride)*/
1344         __u8    s_log_groups_per_flex;  /* FLEX_BG group size */
1345         __u8    s_checksum_type;        /* metadata checksum algorithm used */
1346         __u8    s_encryption_level;     /* versioning level for encryption */
1347         __u8    s_reserved_pad;         /* Padding to next 32bits */
1348         __le64  s_kbytes_written;       /* nr of lifetime kilobytes written */
1349         __le32  s_snapshot_inum;        /* Inode number of active snapshot */
1350         __le32  s_snapshot_id;          /* sequential ID of active snapshot */
1351         __le64  s_snapshot_r_blocks_count; /* reserved blocks for active
1352                                               snapshot's future use */
1353         __le32  s_snapshot_list;        /* inode number of the head of the
1354                                            on-disk snapshot list */
1355 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1356         __le32  s_error_count;          /* number of fs errors */
1357         __le32  s_first_error_time;     /* first time an error happened */
1358         __le32  s_first_error_ino;      /* inode involved in first error */
1359         __le64  s_first_error_block;    /* block involved of first error */
1360         __u8    s_first_error_func[32] __nonstring;     /* function where the error happened */
1361         __le32  s_first_error_line;     /* line number where error happened */
1362         __le32  s_last_error_time;      /* most recent time of an error */
1363         __le32  s_last_error_ino;       /* inode involved in last error */
1364         __le32  s_last_error_line;      /* line number where error happened */
1365         __le64  s_last_error_block;     /* block involved of last error */
1366         __u8    s_last_error_func[32] __nonstring;      /* function where the error happened */
1367 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1368         __u8    s_mount_opts[64];
1369         __le32  s_usr_quota_inum;       /* inode for tracking user quota */
1370         __le32  s_grp_quota_inum;       /* inode for tracking group quota */
1371         __le32  s_overhead_clusters;    /* overhead blocks/clusters in fs */
1372         __le32  s_backup_bgs[2];        /* groups with sparse_super2 SBs */
1373         __u8    s_encrypt_algos[4];     /* Encryption algorithms in use  */
1374         __u8    s_encrypt_pw_salt[16];  /* Salt used for string2key algorithm */
1375         __le32  s_lpf_ino;              /* Location of the lost+found inode */
1376         __le32  s_prj_quota_inum;       /* inode for tracking project quota */
1377         __le32  s_checksum_seed;        /* crc32c(uuid) if csum_seed set */
1378         __u8    s_wtime_hi;
1379         __u8    s_mtime_hi;
1380         __u8    s_mkfs_time_hi;
1381         __u8    s_lastcheck_hi;
1382         __u8    s_first_error_time_hi;
1383         __u8    s_last_error_time_hi;
1384         __u8    s_first_error_errcode;
1385         __u8    s_last_error_errcode;
1386         __le16  s_encoding;             /* Filename charset encoding */
1387         __le16  s_encoding_flags;       /* Filename charset encoding flags */
1388         __le32  s_reserved[95];         /* Padding to the end of the block */
1389         __le32  s_checksum;             /* crc32c(superblock) */
1390 };
1391
1392 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1393
1394 #ifdef __KERNEL__
1395
1396 /*
1397  * run-time mount flags
1398  */
1399 #define EXT4_MF_MNTDIR_SAMPLED          0x0001
1400 #define EXT4_MF_FS_ABORTED              0x0002  /* Fatal error detected */
1401
1402 #ifdef CONFIG_FS_ENCRYPTION
1403 #define DUMMY_ENCRYPTION_ENABLED(sbi) ((sbi)->s_dummy_enc_ctx.ctx != NULL)
1404 #else
1405 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1406 #endif
1407
1408 /* Number of quota types we support */
1409 #define EXT4_MAXQUOTAS 3
1410
1411 #define EXT4_ENC_UTF8_12_1      1
1412
1413 /*
1414  * Flags for ext4_sb_info.s_encoding_flags.
1415  */
1416 #define EXT4_ENC_STRICT_MODE_FL (1 << 0)
1417
1418 #define ext4_has_strict_mode(sbi) \
1419         (sbi->s_encoding_flags & EXT4_ENC_STRICT_MODE_FL)
1420
1421 /*
1422  * fourth extended-fs super-block data in memory
1423  */
1424 struct ext4_sb_info {
1425         unsigned long s_desc_size;      /* Size of a group descriptor in bytes */
1426         unsigned long s_inodes_per_block;/* Number of inodes per block */
1427         unsigned long s_blocks_per_group;/* Number of blocks in a group */
1428         unsigned long s_clusters_per_group; /* Number of clusters in a group */
1429         unsigned long s_inodes_per_group;/* Number of inodes in a group */
1430         unsigned long s_itb_per_group;  /* Number of inode table blocks per group */
1431         unsigned long s_gdb_count;      /* Number of group descriptor blocks */
1432         unsigned long s_desc_per_block; /* Number of group descriptors per block */
1433         ext4_group_t s_groups_count;    /* Number of groups in the fs */
1434         ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1435         unsigned long s_overhead;  /* # of fs overhead clusters */
1436         unsigned int s_cluster_ratio;   /* Number of blocks per cluster */
1437         unsigned int s_cluster_bits;    /* log2 of s_cluster_ratio */
1438         loff_t s_bitmap_maxbytes;       /* max bytes for bitmap files */
1439         struct buffer_head * s_sbh;     /* Buffer containing the super block */
1440         struct ext4_super_block *s_es;  /* Pointer to the super block in the buffer */
1441         struct buffer_head * __rcu *s_group_desc;
1442         unsigned int s_mount_opt;
1443         unsigned int s_mount_opt2;
1444         unsigned int s_mount_flags;
1445         unsigned int s_def_mount_opt;
1446         ext4_fsblk_t s_sb_block;
1447         atomic64_t s_resv_clusters;
1448         kuid_t s_resuid;
1449         kgid_t s_resgid;
1450         unsigned short s_mount_state;
1451         unsigned short s_pad;
1452         int s_addr_per_block_bits;
1453         int s_desc_per_block_bits;
1454         int s_inode_size;
1455         int s_first_ino;
1456         unsigned int s_inode_readahead_blks;
1457         unsigned int s_inode_goal;
1458         u32 s_hash_seed[4];
1459         int s_def_hash_version;
1460         int s_hash_unsigned;    /* 3 if hash should be signed, 0 if not */
1461         struct percpu_counter s_freeclusters_counter;
1462         struct percpu_counter s_freeinodes_counter;
1463         struct percpu_counter s_dirs_counter;
1464         struct percpu_counter s_dirtyclusters_counter;
1465         struct blockgroup_lock *s_blockgroup_lock;
1466         struct proc_dir_entry *s_proc;
1467         struct kobject s_kobj;
1468         struct completion s_kobj_unregister;
1469         struct super_block *s_sb;
1470 #ifdef CONFIG_UNICODE
1471         struct unicode_map *s_encoding;
1472         __u16 s_encoding_flags;
1473 #endif
1474
1475         /* Journaling */
1476         struct journal_s *s_journal;
1477         struct list_head s_orphan;
1478         struct mutex s_orphan_lock;
1479         unsigned long s_ext4_flags;             /* Ext4 superblock flags */
1480         unsigned long s_commit_interval;
1481         u32 s_max_batch_time;
1482         u32 s_min_batch_time;
1483         struct block_device *s_journal_bdev;
1484 #ifdef CONFIG_QUOTA
1485         /* Names of quota files with journalled quota */
1486         char __rcu *s_qf_names[EXT4_MAXQUOTAS];
1487         int s_jquota_fmt;                       /* Format of quota to use */
1488 #endif
1489         unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1490         struct ext4_system_blocks __rcu *s_system_blks;
1491
1492 #ifdef EXTENTS_STATS
1493         /* ext4 extents stats */
1494         unsigned long s_ext_min;
1495         unsigned long s_ext_max;
1496         unsigned long s_depth_max;
1497         spinlock_t s_ext_stats_lock;
1498         unsigned long s_ext_blocks;
1499         unsigned long s_ext_extents;
1500 #endif
1501
1502         /* for buddy allocator */
1503         struct ext4_group_info ** __rcu *s_group_info;
1504         struct inode *s_buddy_cache;
1505         spinlock_t s_md_lock;
1506         unsigned short *s_mb_offsets;
1507         unsigned int *s_mb_maxs;
1508         unsigned int s_group_info_size;
1509         unsigned int s_mb_free_pending;
1510         struct list_head s_freed_data_list;     /* List of blocks to be freed
1511                                                    after commit completed */
1512
1513         /* tunables */
1514         unsigned long s_stripe;
1515         unsigned int s_mb_stream_request;
1516         unsigned int s_mb_max_to_scan;
1517         unsigned int s_mb_min_to_scan;
1518         unsigned int s_mb_stats;
1519         unsigned int s_mb_order2_reqs;
1520         unsigned int s_mb_group_prealloc;
1521         unsigned int s_mb_max_inode_prealloc;
1522         unsigned int s_max_dir_size_kb;
1523         /* where last allocation was done - for stream allocation */
1524         unsigned long s_mb_last_group;
1525         unsigned long s_mb_last_start;
1526         unsigned int s_mb_prefetch;
1527         unsigned int s_mb_prefetch_limit;
1528
1529         /* stats for buddy allocator */
1530         atomic_t s_bal_reqs;    /* number of reqs with len > 1 */
1531         atomic_t s_bal_success; /* we found long enough chunks */
1532         atomic_t s_bal_allocated;       /* in blocks */
1533         atomic_t s_bal_ex_scanned;      /* total extents scanned */
1534         atomic_t s_bal_goals;   /* goal hits */
1535         atomic_t s_bal_breaks;  /* too long searches */
1536         atomic_t s_bal_2orders; /* 2^order hits */
1537         spinlock_t s_bal_lock;
1538         unsigned long s_mb_buddies_generated;
1539         unsigned long long s_mb_generation_time;
1540         atomic_t s_mb_lost_chunks;
1541         atomic_t s_mb_preallocated;
1542         atomic_t s_mb_discarded;
1543         atomic_t s_lock_busy;
1544
1545         /* locality groups */
1546         struct ext4_locality_group __percpu *s_locality_groups;
1547
1548         /* for write statistics */
1549         unsigned long s_sectors_written_start;
1550         u64 s_kbytes_written;
1551
1552         /* the size of zero-out chunk */
1553         unsigned int s_extent_max_zeroout_kb;
1554
1555         unsigned int s_log_groups_per_flex;
1556         struct flex_groups * __rcu *s_flex_groups;
1557         ext4_group_t s_flex_groups_allocated;
1558
1559         /* workqueue for reserved extent conversions (buffered io) */
1560         struct workqueue_struct *rsv_conversion_wq;
1561
1562         /* timer for periodic error stats printing */
1563         struct timer_list s_err_report;
1564
1565         /* Lazy inode table initialization info */
1566         struct ext4_li_request *s_li_request;
1567         /* Wait multiplier for lazy initialization thread */
1568         unsigned int s_li_wait_mult;
1569
1570         /* Kernel thread for multiple mount protection */
1571         struct task_struct *s_mmp_tsk;
1572
1573         /* record the last minlen when FITRIM is called. */
1574         atomic_t s_last_trim_minblks;
1575
1576         /* Reference to checksum algorithm driver via cryptoapi */
1577         struct crypto_shash *s_chksum_driver;
1578
1579         /* Precomputed FS UUID checksum for seeding other checksums */
1580         __u32 s_csum_seed;
1581
1582         /* Reclaim extents from extent status tree */
1583         struct shrinker s_es_shrinker;
1584         struct list_head s_es_list;     /* List of inodes with reclaimable extents */
1585         long s_es_nr_inode;
1586         struct ext4_es_stats s_es_stats;
1587         struct mb_cache *s_ea_block_cache;
1588         struct mb_cache *s_ea_inode_cache;
1589         spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1590
1591         /* Ratelimit ext4 messages. */
1592         struct ratelimit_state s_err_ratelimit_state;
1593         struct ratelimit_state s_warning_ratelimit_state;
1594         struct ratelimit_state s_msg_ratelimit_state;
1595         atomic_t s_warning_count;
1596         atomic_t s_msg_count;
1597
1598         /* Encryption context for '-o test_dummy_encryption' */
1599         struct fscrypt_dummy_context s_dummy_enc_ctx;
1600
1601         /*
1602          * Barrier between writepages ops and changing any inode's JOURNAL_DATA
1603          * or EXTENTS flag.
1604          */
1605         struct percpu_rw_semaphore s_writepages_rwsem;
1606         struct dax_device *s_daxdev;
1607 #ifdef CONFIG_EXT4_DEBUG
1608         unsigned long s_simulate_fail;
1609 #endif
1610         /* Record the errseq of the backing block device */
1611         errseq_t s_bdev_wb_err;
1612         spinlock_t s_bdev_wb_lock;
1613 };
1614
1615 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1616 {
1617         return sb->s_fs_info;
1618 }
1619 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1620 {
1621         return container_of(inode, struct ext4_inode_info, vfs_inode);
1622 }
1623
1624 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1625 {
1626         return ino == EXT4_ROOT_INO ||
1627                 (ino >= EXT4_FIRST_INO(sb) &&
1628                  ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1629 }
1630
1631 /*
1632  * Returns: sbi->field[index]
1633  * Used to access an array element from the following sbi fields which require
1634  * rcu protection to avoid dereferencing an invalid pointer due to reassignment
1635  * - s_group_desc
1636  * - s_group_info
1637  * - s_flex_group
1638  */
1639 #define sbi_array_rcu_deref(sbi, field, index)                             \
1640 ({                                                                         \
1641         typeof(*((sbi)->field)) _v;                                        \
1642         rcu_read_lock();                                                   \
1643         _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index];          \
1644         rcu_read_unlock();                                                 \
1645         _v;                                                                \
1646 })
1647
1648 /*
1649  * Simulate_fail codes
1650  */
1651 #define EXT4_SIM_BBITMAP_EIO    1
1652 #define EXT4_SIM_BBITMAP_CRC    2
1653 #define EXT4_SIM_IBITMAP_EIO    3
1654 #define EXT4_SIM_IBITMAP_CRC    4
1655 #define EXT4_SIM_INODE_EIO      5
1656 #define EXT4_SIM_INODE_CRC      6
1657 #define EXT4_SIM_DIRBLOCK_EIO   7
1658 #define EXT4_SIM_DIRBLOCK_CRC   8
1659
1660 static inline bool ext4_simulate_fail(struct super_block *sb,
1661                                      unsigned long code)
1662 {
1663 #ifdef CONFIG_EXT4_DEBUG
1664         struct ext4_sb_info *sbi = EXT4_SB(sb);
1665
1666         if (unlikely(sbi->s_simulate_fail == code)) {
1667                 sbi->s_simulate_fail = 0;
1668                 return true;
1669         }
1670 #endif
1671         return false;
1672 }
1673
1674 static inline void ext4_simulate_fail_bh(struct super_block *sb,
1675                                          struct buffer_head *bh,
1676                                          unsigned long code)
1677 {
1678         if (!IS_ERR(bh) && ext4_simulate_fail(sb, code))
1679                 clear_buffer_uptodate(bh);
1680 }
1681
1682 /*
1683  * Error number codes for s_{first,last}_error_errno
1684  *
1685  * Linux errno numbers are architecture specific, so we need to translate
1686  * them into something which is architecture independent.   We don't define
1687  * codes for all errno's; just the ones which are most likely to be the cause
1688  * of an ext4_error() call.
1689  */
1690 #define EXT4_ERR_UNKNOWN         1
1691 #define EXT4_ERR_EIO             2
1692 #define EXT4_ERR_ENOMEM          3
1693 #define EXT4_ERR_EFSBADCRC       4
1694 #define EXT4_ERR_EFSCORRUPTED    5
1695 #define EXT4_ERR_ENOSPC          6
1696 #define EXT4_ERR_ENOKEY          7
1697 #define EXT4_ERR_EROFS           8
1698 #define EXT4_ERR_EFBIG           9
1699 #define EXT4_ERR_EEXIST         10
1700 #define EXT4_ERR_ERANGE         11
1701 #define EXT4_ERR_EOVERFLOW      12
1702 #define EXT4_ERR_EBUSY          13
1703 #define EXT4_ERR_ENOTDIR        14
1704 #define EXT4_ERR_ENOTEMPTY      15
1705 #define EXT4_ERR_ESHUTDOWN      16
1706 #define EXT4_ERR_EFAULT         17
1707
1708 /*
1709  * Inode dynamic state flags
1710  */
1711 enum {
1712         EXT4_STATE_JDATA,               /* journaled data exists */
1713         EXT4_STATE_NEW,                 /* inode is newly created */
1714         EXT4_STATE_XATTR,               /* has in-inode xattrs */
1715         EXT4_STATE_NO_EXPAND,           /* No space for expansion */
1716         EXT4_STATE_DA_ALLOC_CLOSE,      /* Alloc DA blks on close */
1717         EXT4_STATE_EXT_MIGRATE,         /* Inode is migrating */
1718         EXT4_STATE_NEWENTRY,            /* File just added to dir */
1719         EXT4_STATE_MAY_INLINE_DATA,     /* may have in-inode data */
1720         EXT4_STATE_EXT_PRECACHED,       /* extents have been precached */
1721         EXT4_STATE_LUSTRE_EA_INODE,     /* Lustre-style ea_inode */
1722         EXT4_STATE_VERITY_IN_PROGRESS,  /* building fs-verity Merkle tree */
1723 };
1724
1725 #define EXT4_INODE_BIT_FNS(name, field, offset)                         \
1726 static inline int ext4_test_inode_##name(struct inode *inode, int bit)  \
1727 {                                                                       \
1728         return test_bit(bit + (offset), &EXT4_I(inode)->i_##field);     \
1729 }                                                                       \
1730 static inline void ext4_set_inode_##name(struct inode *inode, int bit)  \
1731 {                                                                       \
1732         set_bit(bit + (offset), &EXT4_I(inode)->i_##field);             \
1733 }                                                                       \
1734 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1735 {                                                                       \
1736         clear_bit(bit + (offset), &EXT4_I(inode)->i_##field);           \
1737 }
1738
1739 /* Add these declarations here only so that these functions can be
1740  * found by name.  Otherwise, they are very hard to locate. */
1741 static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1742 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1743 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1744 EXT4_INODE_BIT_FNS(flag, flags, 0)
1745
1746 /* Add these declarations here only so that these functions can be
1747  * found by name.  Otherwise, they are very hard to locate. */
1748 static inline int ext4_test_inode_state(struct inode *inode, int bit);
1749 static inline void ext4_set_inode_state(struct inode *inode, int bit);
1750 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1751 #if (BITS_PER_LONG < 64)
1752 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1753
1754 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1755 {
1756         (ei)->i_state_flags = 0;
1757 }
1758 #else
1759 EXT4_INODE_BIT_FNS(state, flags, 32)
1760
1761 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1762 {
1763         /* We depend on the fact that callers will set i_flags */
1764 }
1765 #endif
1766 #else
1767 /* Assume that user mode programs are passing in an ext4fs superblock, not
1768  * a kernel struct super_block.  This will allow us to call the feature-test
1769  * macros from user land. */
1770 #define EXT4_SB(sb)     (sb)
1771 #endif
1772
1773 static inline bool ext4_verity_in_progress(struct inode *inode)
1774 {
1775         return IS_ENABLED(CONFIG_FS_VERITY) &&
1776                ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS);
1777 }
1778
1779 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1780
1781 /*
1782  * Codes for operating systems
1783  */
1784 #define EXT4_OS_LINUX           0
1785 #define EXT4_OS_HURD            1
1786 #define EXT4_OS_MASIX           2
1787 #define EXT4_OS_FREEBSD         3
1788 #define EXT4_OS_LITES           4
1789
1790 /*
1791  * Revision levels
1792  */
1793 #define EXT4_GOOD_OLD_REV       0       /* The good old (original) format */
1794 #define EXT4_DYNAMIC_REV        1       /* V2 format w/ dynamic inode sizes */
1795
1796 #define EXT4_CURRENT_REV        EXT4_GOOD_OLD_REV
1797 #define EXT4_MAX_SUPP_REV       EXT4_DYNAMIC_REV
1798
1799 #define EXT4_GOOD_OLD_INODE_SIZE 128
1800
1801 #define EXT4_EXTRA_TIMESTAMP_MAX        (((s64)1 << 34) - 1  + S32_MIN)
1802 #define EXT4_NON_EXTRA_TIMESTAMP_MAX    S32_MAX
1803 #define EXT4_TIMESTAMP_MIN              S32_MIN
1804
1805 /*
1806  * Feature set definitions
1807  */
1808
1809 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC        0x0001
1810 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES       0x0002
1811 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL         0x0004
1812 #define EXT4_FEATURE_COMPAT_EXT_ATTR            0x0008
1813 #define EXT4_FEATURE_COMPAT_RESIZE_INODE        0x0010
1814 #define EXT4_FEATURE_COMPAT_DIR_INDEX           0x0020
1815 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2       0x0200
1816 #define EXT4_FEATURE_COMPAT_STABLE_INODES       0x0800
1817
1818 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER     0x0001
1819 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE       0x0002
1820 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR        0x0004
1821 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE        0x0008
1822 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM         0x0010
1823 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK        0x0020
1824 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE      0x0040
1825 #define EXT4_FEATURE_RO_COMPAT_QUOTA            0x0100
1826 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC         0x0200
1827 /*
1828  * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM).  When
1829  * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1830  * all other data structures' checksums.  However, the METADATA_CSUM and
1831  * GDT_CSUM bits are mutually exclusive.
1832  */
1833 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM    0x0400
1834 #define EXT4_FEATURE_RO_COMPAT_READONLY         0x1000
1835 #define EXT4_FEATURE_RO_COMPAT_PROJECT          0x2000
1836 #define EXT4_FEATURE_RO_COMPAT_VERITY           0x8000
1837
1838 #define EXT4_FEATURE_INCOMPAT_COMPRESSION       0x0001
1839 #define EXT4_FEATURE_INCOMPAT_FILETYPE          0x0002
1840 #define EXT4_FEATURE_INCOMPAT_RECOVER           0x0004 /* Needs recovery */
1841 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV       0x0008 /* Journal device */
1842 #define EXT4_FEATURE_INCOMPAT_META_BG           0x0010
1843 #define EXT4_FEATURE_INCOMPAT_EXTENTS           0x0040 /* extents support */
1844 #define EXT4_FEATURE_INCOMPAT_64BIT             0x0080
1845 #define EXT4_FEATURE_INCOMPAT_MMP               0x0100
1846 #define EXT4_FEATURE_INCOMPAT_FLEX_BG           0x0200
1847 #define EXT4_FEATURE_INCOMPAT_EA_INODE          0x0400 /* EA in inode */
1848 #define EXT4_FEATURE_INCOMPAT_DIRDATA           0x1000 /* data in dirent */
1849 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED         0x2000
1850 #define EXT4_FEATURE_INCOMPAT_LARGEDIR          0x4000 /* >2GB or 3-lvl htree */
1851 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA       0x8000 /* data in inode */
1852 #define EXT4_FEATURE_INCOMPAT_ENCRYPT           0x10000
1853 #define EXT4_FEATURE_INCOMPAT_CASEFOLD          0x20000
1854
1855 extern void ext4_update_dynamic_rev(struct super_block *sb);
1856
1857 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
1858 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1859 { \
1860         return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1861                 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
1862 } \
1863 static inline void ext4_set_feature_##name(struct super_block *sb) \
1864 { \
1865         ext4_update_dynamic_rev(sb); \
1866         EXT4_SB(sb)->s_es->s_feature_compat |= \
1867                 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1868 } \
1869 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1870 { \
1871         EXT4_SB(sb)->s_es->s_feature_compat &= \
1872                 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1873 }
1874
1875 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
1876 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1877 { \
1878         return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1879                 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
1880 } \
1881 static inline void ext4_set_feature_##name(struct super_block *sb) \
1882 { \
1883         ext4_update_dynamic_rev(sb); \
1884         EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
1885                 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1886 } \
1887 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1888 { \
1889         EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
1890                 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1891 }
1892
1893 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
1894 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1895 { \
1896         return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1897                 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
1898 } \
1899 static inline void ext4_set_feature_##name(struct super_block *sb) \
1900 { \
1901         ext4_update_dynamic_rev(sb); \
1902         EXT4_SB(sb)->s_es->s_feature_incompat |= \
1903                 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1904 } \
1905 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1906 { \
1907         EXT4_SB(sb)->s_es->s_feature_incompat &= \
1908                 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1909 }
1910
1911 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc,         DIR_PREALLOC)
1912 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes,        IMAGIC_INODES)
1913 EXT4_FEATURE_COMPAT_FUNCS(journal,              HAS_JOURNAL)
1914 EXT4_FEATURE_COMPAT_FUNCS(xattr,                EXT_ATTR)
1915 EXT4_FEATURE_COMPAT_FUNCS(resize_inode,         RESIZE_INODE)
1916 EXT4_FEATURE_COMPAT_FUNCS(dir_index,            DIR_INDEX)
1917 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2,        SPARSE_SUPER2)
1918 EXT4_FEATURE_COMPAT_FUNCS(stable_inodes,        STABLE_INODES)
1919
1920 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super,      SPARSE_SUPER)
1921 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file,        LARGE_FILE)
1922 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir,         BTREE_DIR)
1923 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file,         HUGE_FILE)
1924 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum,          GDT_CSUM)
1925 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink,         DIR_NLINK)
1926 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize,       EXTRA_ISIZE)
1927 EXT4_FEATURE_RO_COMPAT_FUNCS(quota,             QUOTA)
1928 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc,          BIGALLOC)
1929 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum,     METADATA_CSUM)
1930 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly,          READONLY)
1931 EXT4_FEATURE_RO_COMPAT_FUNCS(project,           PROJECT)
1932 EXT4_FEATURE_RO_COMPAT_FUNCS(verity,            VERITY)
1933
1934 EXT4_FEATURE_INCOMPAT_FUNCS(compression,        COMPRESSION)
1935 EXT4_FEATURE_INCOMPAT_FUNCS(filetype,           FILETYPE)
1936 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery,     RECOVER)
1937 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev,        JOURNAL_DEV)
1938 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg,            META_BG)
1939 EXT4_FEATURE_INCOMPAT_FUNCS(extents,            EXTENTS)
1940 EXT4_FEATURE_INCOMPAT_FUNCS(64bit,              64BIT)
1941 EXT4_FEATURE_INCOMPAT_FUNCS(mmp,                MMP)
1942 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg,            FLEX_BG)
1943 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode,           EA_INODE)
1944 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata,            DIRDATA)
1945 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed,          CSUM_SEED)
1946 EXT4_FEATURE_INCOMPAT_FUNCS(largedir,           LARGEDIR)
1947 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data,        INLINE_DATA)
1948 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt,            ENCRYPT)
1949 EXT4_FEATURE_INCOMPAT_FUNCS(casefold,           CASEFOLD)
1950
1951 #define EXT2_FEATURE_COMPAT_SUPP        EXT4_FEATURE_COMPAT_EXT_ATTR
1952 #define EXT2_FEATURE_INCOMPAT_SUPP      (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1953                                          EXT4_FEATURE_INCOMPAT_META_BG)
1954 #define EXT2_FEATURE_RO_COMPAT_SUPP     (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1955                                          EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1956                                          EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1957
1958 #define EXT3_FEATURE_COMPAT_SUPP        EXT4_FEATURE_COMPAT_EXT_ATTR
1959 #define EXT3_FEATURE_INCOMPAT_SUPP      (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1960                                          EXT4_FEATURE_INCOMPAT_RECOVER| \
1961                                          EXT4_FEATURE_INCOMPAT_META_BG)
1962 #define EXT3_FEATURE_RO_COMPAT_SUPP     (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1963                                          EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1964                                          EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1965
1966 #define EXT4_FEATURE_COMPAT_SUPP        EXT4_FEATURE_COMPAT_EXT_ATTR
1967 #define EXT4_FEATURE_INCOMPAT_SUPP      (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1968                                          EXT4_FEATURE_INCOMPAT_RECOVER| \
1969                                          EXT4_FEATURE_INCOMPAT_META_BG| \
1970                                          EXT4_FEATURE_INCOMPAT_EXTENTS| \
1971                                          EXT4_FEATURE_INCOMPAT_64BIT| \
1972                                          EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1973                                          EXT4_FEATURE_INCOMPAT_EA_INODE| \
1974                                          EXT4_FEATURE_INCOMPAT_MMP | \
1975                                          EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
1976                                          EXT4_FEATURE_INCOMPAT_ENCRYPT | \
1977                                          EXT4_FEATURE_INCOMPAT_CASEFOLD | \
1978                                          EXT4_FEATURE_INCOMPAT_CSUM_SEED | \
1979                                          EXT4_FEATURE_INCOMPAT_LARGEDIR)
1980 #define EXT4_FEATURE_RO_COMPAT_SUPP     (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1981                                          EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1982                                          EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1983                                          EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1984                                          EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1985                                          EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1986                                          EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1987                                          EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1988                                          EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1989                                          EXT4_FEATURE_RO_COMPAT_QUOTA |\
1990                                          EXT4_FEATURE_RO_COMPAT_PROJECT |\
1991                                          EXT4_FEATURE_RO_COMPAT_VERITY)
1992
1993 #define EXTN_FEATURE_FUNCS(ver) \
1994 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
1995 { \
1996         return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1997                 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
1998 } \
1999 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
2000 { \
2001         return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
2002                 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
2003 } \
2004 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
2005 { \
2006         return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
2007                 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
2008 }
2009
2010 EXTN_FEATURE_FUNCS(2)
2011 EXTN_FEATURE_FUNCS(3)
2012 EXTN_FEATURE_FUNCS(4)
2013
2014 static inline bool ext4_has_compat_features(struct super_block *sb)
2015 {
2016         return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
2017 }
2018 static inline bool ext4_has_ro_compat_features(struct super_block *sb)
2019 {
2020         return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
2021 }
2022 static inline bool ext4_has_incompat_features(struct super_block *sb)
2023 {
2024         return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
2025 }
2026
2027 /*
2028  * Superblock flags
2029  */
2030 #define EXT4_FLAGS_RESIZING     0
2031 #define EXT4_FLAGS_SHUTDOWN     1
2032 #define EXT4_FLAGS_BDEV_IS_DAX  2
2033
2034 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi)
2035 {
2036         return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags);
2037 }
2038
2039
2040 /*
2041  * Default values for user and/or group using reserved blocks
2042  */
2043 #define EXT4_DEF_RESUID         0
2044 #define EXT4_DEF_RESGID         0
2045
2046 /*
2047  * Default project ID
2048  */
2049 #define EXT4_DEF_PROJID         0
2050
2051 #define EXT4_DEF_INODE_READAHEAD_BLKS   32
2052
2053 /*
2054  * Default mount options
2055  */
2056 #define EXT4_DEFM_DEBUG         0x0001
2057 #define EXT4_DEFM_BSDGROUPS     0x0002
2058 #define EXT4_DEFM_XATTR_USER    0x0004
2059 #define EXT4_DEFM_ACL           0x0008
2060 #define EXT4_DEFM_UID16         0x0010
2061 #define EXT4_DEFM_JMODE         0x0060
2062 #define EXT4_DEFM_JMODE_DATA    0x0020
2063 #define EXT4_DEFM_JMODE_ORDERED 0x0040
2064 #define EXT4_DEFM_JMODE_WBACK   0x0060
2065 #define EXT4_DEFM_NOBARRIER     0x0100
2066 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
2067 #define EXT4_DEFM_DISCARD       0x0400
2068 #define EXT4_DEFM_NODELALLOC    0x0800
2069
2070 /*
2071  * Default journal batch times
2072  */
2073 #define EXT4_DEF_MIN_BATCH_TIME 0
2074 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
2075
2076 /*
2077  * Minimum number of groups in a flexgroup before we separate out
2078  * directories into the first block group of a flexgroup
2079  */
2080 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
2081
2082 /*
2083  * Structure of a directory entry
2084  */
2085 #define EXT4_NAME_LEN 255
2086
2087 struct ext4_dir_entry {
2088         __le32  inode;                  /* Inode number */
2089         __le16  rec_len;                /* Directory entry length */
2090         __le16  name_len;               /* Name length */
2091         char    name[EXT4_NAME_LEN];    /* File name */
2092 };
2093
2094 /*
2095  * The new version of the directory entry.  Since EXT4 structures are
2096  * stored in intel byte order, and the name_len field could never be
2097  * bigger than 255 chars, it's safe to reclaim the extra byte for the
2098  * file_type field.
2099  */
2100 struct ext4_dir_entry_2 {
2101         __le32  inode;                  /* Inode number */
2102         __le16  rec_len;                /* Directory entry length */
2103         __u8    name_len;               /* Name length */
2104         __u8    file_type;              /* See file type macros EXT4_FT_* below */
2105         char    name[EXT4_NAME_LEN];    /* File name */
2106 };
2107
2108 /*
2109  * This is a bogus directory entry at the end of each leaf block that
2110  * records checksums.
2111  */
2112 struct ext4_dir_entry_tail {
2113         __le32  det_reserved_zero1;     /* Pretend to be unused */
2114         __le16  det_rec_len;            /* 12 */
2115         __u8    det_reserved_zero2;     /* Zero name length */
2116         __u8    det_reserved_ft;        /* 0xDE, fake file type */
2117         __le32  det_checksum;           /* crc32c(uuid+inum+dirblock) */
2118 };
2119
2120 #define EXT4_DIRENT_TAIL(block, blocksize) \
2121         ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
2122                                         ((blocksize) - \
2123                                          sizeof(struct ext4_dir_entry_tail))))
2124
2125 /*
2126  * Ext4 directory file types.  Only the low 3 bits are used.  The
2127  * other bits are reserved for now.
2128  */
2129 #define EXT4_FT_UNKNOWN         0
2130 #define EXT4_FT_REG_FILE        1
2131 #define EXT4_FT_DIR             2
2132 #define EXT4_FT_CHRDEV          3
2133 #define EXT4_FT_BLKDEV          4
2134 #define EXT4_FT_FIFO            5
2135 #define EXT4_FT_SOCK            6
2136 #define EXT4_FT_SYMLINK         7
2137
2138 #define EXT4_FT_MAX             8
2139
2140 #define EXT4_FT_DIR_CSUM        0xDE
2141
2142 /*
2143  * EXT4_DIR_PAD defines the directory entries boundaries
2144  *
2145  * NOTE: It must be a multiple of 4
2146  */
2147 #define EXT4_DIR_PAD                    4
2148 #define EXT4_DIR_ROUND                  (EXT4_DIR_PAD - 1)
2149 #define EXT4_DIR_REC_LEN(name_len)      (((name_len) + 8 + EXT4_DIR_ROUND) & \
2150                                          ~EXT4_DIR_ROUND)
2151 #define EXT4_MAX_REC_LEN                ((1<<16)-1)
2152
2153 /*
2154  * If we ever get support for fs block sizes > page_size, we'll need
2155  * to remove the #if statements in the next two functions...
2156  */
2157 static inline unsigned int
2158 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
2159 {
2160         unsigned len = le16_to_cpu(dlen);
2161
2162 #if (PAGE_SIZE >= 65536)
2163         if (len == EXT4_MAX_REC_LEN || len == 0)
2164                 return blocksize;
2165         return (len & 65532) | ((len & 3) << 16);
2166 #else
2167         return len;
2168 #endif
2169 }
2170
2171 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
2172 {
2173         if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
2174                 BUG();
2175 #if (PAGE_SIZE >= 65536)
2176         if (len < 65536)
2177                 return cpu_to_le16(len);
2178         if (len == blocksize) {
2179                 if (blocksize == 65536)
2180                         return cpu_to_le16(EXT4_MAX_REC_LEN);
2181                 else
2182                         return cpu_to_le16(0);
2183         }
2184         return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
2185 #else
2186         return cpu_to_le16(len);
2187 #endif
2188 }
2189
2190 /*
2191  * Hash Tree Directory indexing
2192  * (c) Daniel Phillips, 2001
2193  */
2194
2195 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
2196                     ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2197 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \
2198                     !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir)))
2199 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2200
2201 /* Legal values for the dx_root hash_version field: */
2202
2203 #define DX_HASH_LEGACY                  0
2204 #define DX_HASH_HALF_MD4                1
2205 #define DX_HASH_TEA                     2
2206 #define DX_HASH_LEGACY_UNSIGNED         3
2207 #define DX_HASH_HALF_MD4_UNSIGNED       4
2208 #define DX_HASH_TEA_UNSIGNED            5
2209
2210 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2211                               const void *address, unsigned int length)
2212 {
2213         struct {
2214                 struct shash_desc shash;
2215                 char ctx[4];
2216         } desc;
2217
2218         BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2219
2220         desc.shash.tfm = sbi->s_chksum_driver;
2221         *(u32 *)desc.ctx = crc;
2222
2223         BUG_ON(crypto_shash_update(&desc.shash, address, length));
2224
2225         return *(u32 *)desc.ctx;
2226 }
2227
2228 #ifdef __KERNEL__
2229
2230 /* hash info structure used by the directory hash */
2231 struct dx_hash_info
2232 {
2233         u32             hash;
2234         u32             minor_hash;
2235         int             hash_version;
2236         u32             *seed;
2237 };
2238
2239
2240 /* 32 and 64 bit signed EOF for dx directories */
2241 #define EXT4_HTREE_EOF_32BIT   ((1UL  << (32 - 1)) - 1)
2242 #define EXT4_HTREE_EOF_64BIT   ((1ULL << (64 - 1)) - 1)
2243
2244
2245 /*
2246  * Control parameters used by ext4_htree_next_block
2247  */
2248 #define HASH_NB_ALWAYS          1
2249
2250 struct ext4_filename {
2251         const struct qstr *usr_fname;
2252         struct fscrypt_str disk_name;
2253         struct dx_hash_info hinfo;
2254 #ifdef CONFIG_FS_ENCRYPTION
2255         struct fscrypt_str crypto_buf;
2256 #endif
2257 #ifdef CONFIG_UNICODE
2258         struct fscrypt_str cf_name;
2259 #endif
2260 };
2261
2262 #define fname_name(p) ((p)->disk_name.name)
2263 #define fname_len(p)  ((p)->disk_name.len)
2264
2265 /*
2266  * Describe an inode's exact location on disk and in memory
2267  */
2268 struct ext4_iloc
2269 {
2270         struct buffer_head *bh;
2271         unsigned long offset;
2272         ext4_group_t block_group;
2273 };
2274
2275 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2276 {
2277         return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2278 }
2279
2280 static inline bool ext4_is_quota_file(struct inode *inode)
2281 {
2282         return IS_NOQUOTA(inode) &&
2283                !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL);
2284 }
2285
2286 /*
2287  * This structure is stuffed into the struct file's private_data field
2288  * for directories.  It is where we put information so that we can do
2289  * readdir operations in hash tree order.
2290  */
2291 struct dir_private_info {
2292         struct rb_root  root;
2293         struct rb_node  *curr_node;
2294         struct fname    *extra_fname;
2295         loff_t          last_pos;
2296         __u32           curr_hash;
2297         __u32           curr_minor_hash;
2298         __u32           next_hash;
2299 };
2300
2301 /* calculate the first block number of the group */
2302 static inline ext4_fsblk_t
2303 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2304 {
2305         return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2306                 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2307 }
2308
2309 /*
2310  * Special error return code only used by dx_probe() and its callers.
2311  */
2312 #define ERR_BAD_DX_DIR  (-(MAX_ERRNO - 1))
2313
2314 /* htree levels for ext4 */
2315 #define EXT4_HTREE_LEVEL_COMPAT 2
2316 #define EXT4_HTREE_LEVEL        3
2317
2318 static inline int ext4_dir_htree_level(struct super_block *sb)
2319 {
2320         return ext4_has_feature_largedir(sb) ?
2321                 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT;
2322 }
2323
2324 /*
2325  * Timeout and state flag for lazy initialization inode thread.
2326  */
2327 #define EXT4_DEF_LI_WAIT_MULT                   10
2328 #define EXT4_DEF_LI_MAX_START_DELAY             5
2329 #define EXT4_LAZYINIT_QUIT                      0x0001
2330 #define EXT4_LAZYINIT_RUNNING                   0x0002
2331
2332 /*
2333  * Lazy inode table initialization info
2334  */
2335 struct ext4_lazy_init {
2336         unsigned long           li_state;
2337         struct list_head        li_request_list;
2338         struct mutex            li_list_mtx;
2339 };
2340
2341 enum ext4_li_mode {
2342         EXT4_LI_MODE_PREFETCH_BBITMAP,
2343         EXT4_LI_MODE_ITABLE,
2344 };
2345
2346 struct ext4_li_request {
2347         struct super_block      *lr_super;
2348         enum ext4_li_mode       lr_mode;
2349         ext4_group_t            lr_first_not_zeroed;
2350         ext4_group_t            lr_next_group;
2351         struct list_head        lr_request;
2352         unsigned long           lr_next_sched;
2353         unsigned long           lr_timeout;
2354 };
2355
2356 struct ext4_features {
2357         struct kobject f_kobj;
2358         struct completion f_kobj_unregister;
2359 };
2360
2361 /*
2362  * This structure will be used for multiple mount protection. It will be
2363  * written into the block number saved in the s_mmp_block field in the
2364  * superblock. Programs that check MMP should assume that if
2365  * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2366  * to use the filesystem, regardless of how old the timestamp is.
2367  */
2368 #define EXT4_MMP_MAGIC     0x004D4D50U /* ASCII for MMP */
2369 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2370 #define EXT4_MMP_SEQ_FSCK  0xE24D4D50U /* mmp_seq value when being fscked */
2371 #define EXT4_MMP_SEQ_MAX   0xE24D4D4FU /* maximum valid mmp_seq value */
2372
2373 struct mmp_struct {
2374         __le32  mmp_magic;              /* Magic number for MMP */
2375         __le32  mmp_seq;                /* Sequence no. updated periodically */
2376
2377         /*
2378          * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2379          * purposes and do not affect the correctness of the algorithm
2380          */
2381         __le64  mmp_time;               /* Time last updated */
2382         char    mmp_nodename[64];       /* Node which last updated MMP block */
2383         char    mmp_bdevname[32];       /* Bdev which last updated MMP block */
2384
2385         /*
2386          * mmp_check_interval is used to verify if the MMP block has been
2387          * updated on the block device. The value is updated based on the
2388          * maximum time to write the MMP block during an update cycle.
2389          */
2390         __le16  mmp_check_interval;
2391
2392         __le16  mmp_pad1;
2393         __le32  mmp_pad2[226];
2394         __le32  mmp_checksum;           /* crc32c(uuid+mmp_block) */
2395 };
2396
2397 /* arguments passed to the mmp thread */
2398 struct mmpd_data {
2399         struct buffer_head *bh; /* bh from initial read_mmp_block() */
2400         struct super_block *sb;  /* super block of the fs */
2401 };
2402
2403 /*
2404  * Check interval multiplier
2405  * The MMP block is written every update interval and initially checked every
2406  * update interval x the multiplier (the value is then adapted based on the
2407  * write latency). The reason is that writes can be delayed under load and we
2408  * don't want readers to incorrectly assume that the filesystem is no longer
2409  * in use.
2410  */
2411 #define EXT4_MMP_CHECK_MULT             2UL
2412
2413 /*
2414  * Minimum interval for MMP checking in seconds.
2415  */
2416 #define EXT4_MMP_MIN_CHECK_INTERVAL     5UL
2417
2418 /*
2419  * Maximum interval for MMP checking in seconds.
2420  */
2421 #define EXT4_MMP_MAX_CHECK_INTERVAL     300UL
2422
2423 /*
2424  * Function prototypes
2425  */
2426
2427 /*
2428  * Ok, these declarations are also in <linux/kernel.h> but none of the
2429  * ext4 source programs needs to include it so they are duplicated here.
2430  */
2431 # define NORET_TYPE     /**/
2432 # define ATTRIB_NORET   __attribute__((noreturn))
2433 # define NORET_AND      noreturn,
2434
2435 /* bitmap.c */
2436 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2437 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2438                                 struct ext4_group_desc *gdp,
2439                                 struct buffer_head *bh, int sz);
2440 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2441                                   struct ext4_group_desc *gdp,
2442                                   struct buffer_head *bh, int sz);
2443 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2444                                 struct ext4_group_desc *gdp,
2445                                 struct buffer_head *bh);
2446 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2447                                   struct ext4_group_desc *gdp,
2448                                   struct buffer_head *bh);
2449
2450 /* balloc.c */
2451 extern void ext4_get_group_no_and_offset(struct super_block *sb,
2452                                          ext4_fsblk_t blocknr,
2453                                          ext4_group_t *blockgrpp,
2454                                          ext4_grpblk_t *offsetp);
2455 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2456                                           ext4_fsblk_t block);
2457
2458 extern unsigned int ext4_block_group(struct super_block *sb,
2459                         ext4_fsblk_t blocknr);
2460 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2461                         ext4_fsblk_t blocknr);
2462 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2463 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2464                         ext4_group_t group);
2465 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2466                                          ext4_fsblk_t goal,
2467                                          unsigned int flags,
2468                                          unsigned long *count,
2469                                          int *errp);
2470 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2471                                     s64 nclusters, unsigned int flags);
2472 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2473 extern void ext4_check_blocks_bitmap(struct super_block *);
2474 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2475                                                     ext4_group_t block_group,
2476                                                     struct buffer_head ** bh);
2477 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2478
2479 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2480                                                 ext4_group_t block_group,
2481                                                 bool ignore_locked);
2482 extern int ext4_wait_block_bitmap(struct super_block *sb,
2483                                   ext4_group_t block_group,
2484                                   struct buffer_head *bh);
2485 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2486                                                   ext4_group_t block_group);
2487 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2488                                               ext4_group_t block_group,
2489                                               struct ext4_group_desc *gdp);
2490 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2491
2492 #ifdef CONFIG_UNICODE
2493 extern void ext4_fname_setup_ci_filename(struct inode *dir,
2494                                          const struct qstr *iname,
2495                                          struct fscrypt_str *fname);
2496 #endif
2497
2498 #ifdef CONFIG_FS_ENCRYPTION
2499 static inline void ext4_fname_from_fscrypt_name(struct ext4_filename *dst,
2500                                                 const struct fscrypt_name *src)
2501 {
2502         memset(dst, 0, sizeof(*dst));
2503
2504         dst->usr_fname = src->usr_fname;
2505         dst->disk_name = src->disk_name;
2506         dst->hinfo.hash = src->hash;
2507         dst->hinfo.minor_hash = src->minor_hash;
2508         dst->crypto_buf = src->crypto_buf;
2509 }
2510
2511 static inline int ext4_fname_setup_filename(struct inode *dir,
2512                                             const struct qstr *iname,
2513                                             int lookup,
2514                                             struct ext4_filename *fname)
2515 {
2516         struct fscrypt_name name;
2517         int err;
2518
2519         err = fscrypt_setup_filename(dir, iname, lookup, &name);
2520         if (err)
2521                 return err;
2522
2523         ext4_fname_from_fscrypt_name(fname, &name);
2524
2525 #ifdef CONFIG_UNICODE
2526         ext4_fname_setup_ci_filename(dir, iname, &fname->cf_name);
2527 #endif
2528         return 0;
2529 }
2530
2531 static inline int ext4_fname_prepare_lookup(struct inode *dir,
2532                                             struct dentry *dentry,
2533                                             struct ext4_filename *fname)
2534 {
2535         struct fscrypt_name name;
2536         int err;
2537
2538         err = fscrypt_prepare_lookup(dir, dentry, &name);
2539         if (err)
2540                 return err;
2541
2542         ext4_fname_from_fscrypt_name(fname, &name);
2543
2544 #ifdef CONFIG_UNICODE
2545         ext4_fname_setup_ci_filename(dir, &dentry->d_name, &fname->cf_name);
2546 #endif
2547         return 0;
2548 }
2549
2550 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2551 {
2552         struct fscrypt_name name;
2553
2554         name.crypto_buf = fname->crypto_buf;
2555         fscrypt_free_filename(&name);
2556
2557         fname->crypto_buf.name = NULL;
2558         fname->usr_fname = NULL;
2559         fname->disk_name.name = NULL;
2560
2561 #ifdef CONFIG_UNICODE
2562         kfree(fname->cf_name.name);
2563         fname->cf_name.name = NULL;
2564 #endif
2565 }
2566 #else /* !CONFIG_FS_ENCRYPTION */
2567 static inline int ext4_fname_setup_filename(struct inode *dir,
2568                                             const struct qstr *iname,
2569                                             int lookup,
2570                                             struct ext4_filename *fname)
2571 {
2572         fname->usr_fname = iname;
2573         fname->disk_name.name = (unsigned char *) iname->name;
2574         fname->disk_name.len = iname->len;
2575
2576 #ifdef CONFIG_UNICODE
2577         ext4_fname_setup_ci_filename(dir, iname, &fname->cf_name);
2578 #endif
2579
2580         return 0;
2581 }
2582
2583 static inline int ext4_fname_prepare_lookup(struct inode *dir,
2584                                             struct dentry *dentry,
2585                                             struct ext4_filename *fname)
2586 {
2587         return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname);
2588 }
2589
2590 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2591 {
2592 #ifdef CONFIG_UNICODE
2593         kfree(fname->cf_name.name);
2594         fname->cf_name.name = NULL;
2595 #endif
2596 }
2597 #endif /* !CONFIG_FS_ENCRYPTION */
2598
2599 /* dir.c */
2600 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2601                                   struct file *,
2602                                   struct ext4_dir_entry_2 *,
2603                                   struct buffer_head *, char *, int,
2604                                   unsigned int);
2605 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset)      \
2606         unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2607                                         (de), (bh), (buf), (size), (offset)))
2608 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2609                                 __u32 minor_hash,
2610                                 struct ext4_dir_entry_2 *dirent,
2611                                 struct fscrypt_str *ent_name);
2612 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2613 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2614                              struct buffer_head *bh,
2615                              void *buf, int buf_size,
2616                              struct ext4_filename *fname,
2617                              struct ext4_dir_entry_2 **dest_de);
2618 void ext4_insert_dentry(struct inode *inode,
2619                         struct ext4_dir_entry_2 *de,
2620                         int buf_size,
2621                         struct ext4_filename *fname);
2622 static inline void ext4_update_dx_flag(struct inode *inode)
2623 {
2624         if (!ext4_has_feature_dir_index(inode->i_sb)) {
2625                 /* ext4_iget() should have caught this... */
2626                 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb));
2627                 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2628         }
2629 }
2630 static const unsigned char ext4_filetype_table[] = {
2631         DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2632 };
2633
2634 static inline  unsigned char get_dtype(struct super_block *sb, int filetype)
2635 {
2636         if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2637                 return DT_UNKNOWN;
2638
2639         return ext4_filetype_table[filetype];
2640 }
2641 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2642                              void *buf, int buf_size);
2643
2644 /* fsync.c */
2645 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2646
2647 /* hash.c */
2648 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len,
2649                           struct dx_hash_info *hinfo);
2650
2651 /* ialloc.c */
2652 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2653                                       const struct qstr *qstr, __u32 goal,
2654                                       uid_t *owner, __u32 i_flags,
2655                                       int handle_type, unsigned int line_no,
2656                                       int nblocks);
2657
2658 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \
2659         __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2660                          i_flags, 0, 0, 0)
2661 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2662                                     type, nblocks)                  \
2663         __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2664                          0, (type), __LINE__, (nblocks))
2665
2666
2667 extern void ext4_free_inode(handle_t *, struct inode *);
2668 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2669 extern unsigned long ext4_count_free_inodes(struct super_block *);
2670 extern unsigned long ext4_count_dirs(struct super_block *);
2671 extern void ext4_check_inodes_bitmap(struct super_block *);
2672 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2673 extern int ext4_init_inode_table(struct super_block *sb,
2674                                  ext4_group_t group, int barrier);
2675 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2676
2677 /* mballoc.c */
2678 extern const struct seq_operations ext4_mb_seq_groups_ops;
2679 extern long ext4_mb_stats;
2680 extern long ext4_mb_max_to_scan;
2681 extern int ext4_mb_init(struct super_block *);
2682 extern int ext4_mb_release(struct super_block *);
2683 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2684                                 struct ext4_allocation_request *, int *);
2685 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2686 extern void ext4_discard_preallocations(struct inode *, unsigned int);
2687 extern int __init ext4_init_mballoc(void);
2688 extern void ext4_exit_mballoc(void);
2689 extern ext4_group_t ext4_mb_prefetch(struct super_block *sb,
2690                                      ext4_group_t group,
2691                                      unsigned int nr, int *cnt);
2692 extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group,
2693                                   unsigned int nr);
2694
2695 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2696                              struct buffer_head *bh, ext4_fsblk_t block,
2697                              unsigned long count, int flags);
2698 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2699                                    ext4_group_t ngroups);
2700 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2701                 ext4_group_t i, struct ext4_group_desc *desc);
2702 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2703                                 ext4_fsblk_t block, unsigned long count);
2704 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2705 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid);
2706
2707 /* inode.c */
2708 int ext4_inode_is_fast_symlink(struct inode *inode);
2709 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2710 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2711 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count,
2712                      bool wait, struct buffer_head **bhs);
2713 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2714                              struct buffer_head *bh_result, int create);
2715 int ext4_get_block(struct inode *inode, sector_t iblock,
2716                    struct buffer_head *bh_result, int create);
2717 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2718                            struct buffer_head *bh, int create);
2719 int ext4_walk_page_buffers(handle_t *handle,
2720                            struct buffer_head *head,
2721                            unsigned from,
2722                            unsigned to,
2723                            int *partial,
2724                            int (*fn)(handle_t *handle,
2725                                      struct buffer_head *bh));
2726 int do_journal_get_write_access(handle_t *handle,
2727                                 struct buffer_head *bh);
2728 #define FALL_BACK_TO_NONDELALLOC 1
2729 #define CONVERT_INLINE_DATA      2
2730
2731 typedef enum {
2732         EXT4_IGET_NORMAL =      0,
2733         EXT4_IGET_SPECIAL =     0x0001, /* OK to iget a system inode */
2734         EXT4_IGET_HANDLE =      0x0002  /* Inode # is from a handle */
2735 } ext4_iget_flags;
2736
2737 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
2738                                  ext4_iget_flags flags, const char *function,
2739                                  unsigned int line);
2740
2741 #define ext4_iget(sb, ino, flags) \
2742         __ext4_iget((sb), (ino), (flags), __func__, __LINE__)
2743
2744 extern int  ext4_write_inode(struct inode *, struct writeback_control *);
2745 extern int  ext4_setattr(struct dentry *, struct iattr *);
2746 extern int  ext4_getattr(const struct path *, struct kstat *, u32, unsigned int);
2747 extern void ext4_evict_inode(struct inode *);
2748 extern void ext4_clear_inode(struct inode *);
2749 extern int  ext4_file_getattr(const struct path *, struct kstat *, u32, unsigned int);
2750 extern int  ext4_sync_inode(handle_t *, struct inode *);
2751 extern void ext4_dirty_inode(struct inode *, int);
2752 extern int ext4_change_inode_journal_flag(struct inode *, int);
2753 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2754 extern int ext4_inode_attach_jinode(struct inode *inode);
2755 extern int ext4_can_truncate(struct inode *inode);
2756 extern int ext4_truncate(struct inode *);
2757 extern int ext4_break_layouts(struct inode *);
2758 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2759 extern void ext4_set_inode_flags(struct inode *, bool init);
2760 extern int ext4_alloc_da_blocks(struct inode *inode);
2761 extern void ext4_set_aops(struct inode *inode);
2762 extern int ext4_writepage_trans_blocks(struct inode *);
2763 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2764 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2765                              loff_t lstart, loff_t lend);
2766 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf);
2767 extern vm_fault_t ext4_filemap_fault(struct vm_fault *vmf);
2768 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2769 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
2770 extern void ext4_da_release_space(struct inode *inode, int to_free);
2771 extern void ext4_da_update_reserve_space(struct inode *inode,
2772                                         int used, int quota_claim);
2773 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
2774                               ext4_fsblk_t pblk, ext4_lblk_t len);
2775
2776 /* indirect.c */
2777 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2778                                 struct ext4_map_blocks *map, int flags);
2779 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2780 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2781 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2782                                  ext4_lblk_t start, ext4_lblk_t end);
2783
2784 /* ioctl.c */
2785 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2786 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2787
2788 /* migrate.c */
2789 extern int ext4_ext_migrate(struct inode *);
2790 extern int ext4_ind_migrate(struct inode *inode);
2791
2792 /* namei.c */
2793 extern int ext4_dirblock_csum_verify(struct inode *inode,
2794                                      struct buffer_head *bh);
2795 extern int ext4_orphan_add(handle_t *, struct inode *);
2796 extern int ext4_orphan_del(handle_t *, struct inode *);
2797 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2798                                 __u32 start_minor_hash, __u32 *next_hash);
2799 extern int ext4_search_dir(struct buffer_head *bh,
2800                            char *search_buf,
2801                            int buf_size,
2802                            struct inode *dir,
2803                            struct ext4_filename *fname,
2804                            unsigned int offset,
2805                            struct ext4_dir_entry_2 **res_dir);
2806 extern int ext4_generic_delete_entry(struct inode *dir,
2807                                      struct ext4_dir_entry_2 *de_del,
2808                                      struct buffer_head *bh,
2809                                      void *entry_buf,
2810                                      int buf_size,
2811                                      int csum_size);
2812 extern bool ext4_empty_dir(struct inode *inode);
2813
2814 /* resize.c */
2815 extern void ext4_kvfree_array_rcu(void *to_free);
2816 extern int ext4_group_add(struct super_block *sb,
2817                                 struct ext4_new_group_data *input);
2818 extern int ext4_group_extend(struct super_block *sb,
2819                                 struct ext4_super_block *es,
2820                                 ext4_fsblk_t n_blocks_count);
2821 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2822
2823 /* super.c */
2824 extern struct buffer_head *ext4_sb_bread(struct super_block *sb,
2825                                          sector_t block, int op_flags);
2826 extern void ext4_read_bh_nowait(struct buffer_head *bh, int op_flags,
2827                                 bh_end_io_t *end_io);
2828 extern int ext4_read_bh(struct buffer_head *bh, int op_flags,
2829                         bh_end_io_t *end_io);
2830 extern int ext4_read_bh_lock(struct buffer_head *bh, int op_flags, bool wait);
2831 extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
2832 extern int ext4_calculate_overhead(struct super_block *sb);
2833 extern void ext4_superblock_csum_set(struct super_block *sb);
2834 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2835                                     ext4_group_t ngroup);
2836 extern const char *ext4_decode_error(struct super_block *sb, int errno,
2837                                      char nbuf[16]);
2838 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb,
2839                                              ext4_group_t block_group,
2840                                              unsigned int flags);
2841
2842 extern __printf(6, 7)
2843 void __ext4_error(struct super_block *, const char *, unsigned int, int, __u64,
2844                   const char *, ...);
2845 extern __printf(6, 7)
2846 void __ext4_error_inode(struct inode *, const char *, unsigned int,
2847                         ext4_fsblk_t, int, const char *, ...);
2848 extern __printf(5, 6)
2849 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2850                      const char *, ...);
2851 extern void __ext4_std_error(struct super_block *, const char *,
2852                              unsigned int, int);
2853 extern __printf(5, 6)
2854 void __ext4_abort(struct super_block *, const char *, unsigned int, int,
2855                   const char *, ...);
2856 extern __printf(4, 5)
2857 void __ext4_warning(struct super_block *, const char *, unsigned int,
2858                     const char *, ...);
2859 extern __printf(4, 5)
2860 void __ext4_warning_inode(const struct inode *inode, const char *function,
2861                           unsigned int line, const char *fmt, ...);
2862 extern __printf(3, 4)
2863 void __ext4_msg(struct super_block *, const char *, const char *, ...);
2864 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2865                            const char *, unsigned int, const char *);
2866 extern __printf(7, 8)
2867 void __ext4_grp_locked_error(const char *, unsigned int,
2868                              struct super_block *, ext4_group_t,
2869                              unsigned long, ext4_fsblk_t,
2870                              const char *, ...);
2871
2872 #define EXT4_ERROR_INODE(inode, fmt, a...) \
2873         ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
2874
2875 #define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...)                     \
2876         __ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a)
2877
2878 #define ext4_error_inode_block(inode, block, err, fmt, a...)            \
2879         __ext4_error_inode((inode), __func__, __LINE__, (block), (err), \
2880                            (fmt), ## a)
2881
2882 #define EXT4_ERROR_FILE(file, block, fmt, a...)                         \
2883         ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
2884
2885 #ifdef CONFIG_PRINTK
2886
2887 #define ext4_error_inode(inode, func, line, block, fmt, ...)            \
2888         __ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__)
2889 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...)   \
2890         __ext4_error_inode((inode), (func), (line), (block),            \
2891                            (err), (fmt), ##__VA_ARGS__)
2892 #define ext4_error_file(file, func, line, block, fmt, ...)              \
2893         __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2894 #define ext4_error(sb, fmt, ...)                                        \
2895         __ext4_error((sb), __func__, __LINE__, 0, 0, (fmt), ##__VA_ARGS__)
2896 #define ext4_error_err(sb, err, fmt, ...)                               \
2897         __ext4_error((sb), __func__, __LINE__, (err), 0, (fmt), ##__VA_ARGS__)
2898 #define ext4_abort(sb, err, fmt, ...)                                   \
2899         __ext4_abort((sb), __func__, __LINE__, (err), (fmt), ##__VA_ARGS__)
2900 #define ext4_warning(sb, fmt, ...)                                      \
2901         __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2902 #define ext4_warning_inode(inode, fmt, ...)                             \
2903         __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
2904 #define ext4_msg(sb, level, fmt, ...)                           \
2905         __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2906 #define dump_mmp_msg(sb, mmp, msg)                                      \
2907         __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2908 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...)            \
2909         __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2910                                 fmt, ##__VA_ARGS__)
2911
2912 #else
2913
2914 #define ext4_error_inode(inode, func, line, block, fmt, ...)            \
2915 do {                                                                    \
2916         no_printk(fmt, ##__VA_ARGS__);                                  \
2917         __ext4_error_inode(inode, "", 0, block, 0, " ");                \
2918 } while (0)
2919 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...)   \
2920 do {                                                                    \
2921         no_printk(fmt, ##__VA_ARGS__);                                  \
2922         __ext4_error_inode(inode, "", 0, block, err, " ");              \
2923 } while (0)
2924 #define ext4_error_file(file, func, line, block, fmt, ...)              \
2925 do {                                                                    \
2926         no_printk(fmt, ##__VA_ARGS__);                                  \
2927         __ext4_error_file(file, "", 0, block, " ");                     \
2928 } while (0)
2929 #define ext4_error(sb, fmt, ...)                                        \
2930 do {                                                                    \
2931         no_printk(fmt, ##__VA_ARGS__);                                  \
2932         __ext4_error(sb, "", 0, 0, 0, " ");                             \
2933 } while (0)
2934 #define ext4_error_err(sb, err, fmt, ...)                               \
2935 do {                                                                    \
2936         no_printk(fmt, ##__VA_ARGS__);                                  \
2937         __ext4_error(sb, "", 0, err, 0, " ");                           \
2938 } while (0)
2939 #define ext4_abort(sb, err, fmt, ...)                                   \
2940 do {                                                                    \
2941         no_printk(fmt, ##__VA_ARGS__);                                  \
2942         __ext4_abort(sb, "", 0, err, " ");                              \
2943 } while (0)
2944 #define ext4_warning(sb, fmt, ...)                                      \
2945 do {                                                                    \
2946         no_printk(fmt, ##__VA_ARGS__);                                  \
2947         __ext4_warning(sb, "", 0, " ");                                 \
2948 } while (0)
2949 #define ext4_warning_inode(inode, fmt, ...)                             \
2950 do {                                                                    \
2951         no_printk(fmt, ##__VA_ARGS__);                                  \
2952         __ext4_warning_inode(inode, "", 0, " ");                        \
2953 } while (0)
2954 #define ext4_msg(sb, level, fmt, ...)                                   \
2955 do {                                                                    \
2956         no_printk(fmt, ##__VA_ARGS__);                                  \
2957         __ext4_msg(sb, "", " ");                                        \
2958 } while (0)
2959 #define dump_mmp_msg(sb, mmp, msg)                                      \
2960         __dump_mmp_msg(sb, mmp, "", 0, "")
2961 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...)            \
2962 do {                                                                    \
2963         no_printk(fmt, ##__VA_ARGS__);                          \
2964         __ext4_grp_locked_error("", 0, sb, grp, ino, block, " ");       \
2965 } while (0)
2966
2967 #endif
2968
2969 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2970                                       struct ext4_group_desc *bg);
2971 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2972                                       struct ext4_group_desc *bg);
2973 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2974                                      struct ext4_group_desc *bg);
2975 extern __u32 ext4_free_group_clusters(struct super_block *sb,
2976                                       struct ext4_group_desc *bg);
2977 extern __u32 ext4_free_inodes_count(struct super_block *sb,
2978                                  struct ext4_group_desc *bg);
2979 extern __u32 ext4_used_dirs_count(struct super_block *sb,
2980                                 struct ext4_group_desc *bg);
2981 extern __u32 ext4_itable_unused_count(struct super_block *sb,
2982                                    struct ext4_group_desc *bg);
2983 extern void ext4_block_bitmap_set(struct super_block *sb,
2984                                   struct ext4_group_desc *bg, ext4_fsblk_t blk);
2985 extern void ext4_inode_bitmap_set(struct super_block *sb,
2986                                   struct ext4_group_desc *bg, ext4_fsblk_t blk);
2987 extern void ext4_inode_table_set(struct super_block *sb,
2988                                  struct ext4_group_desc *bg, ext4_fsblk_t blk);
2989 extern void ext4_free_group_clusters_set(struct super_block *sb,
2990                                          struct ext4_group_desc *bg,
2991                                          __u32 count);
2992 extern void ext4_free_inodes_set(struct super_block *sb,
2993                                 struct ext4_group_desc *bg, __u32 count);
2994 extern void ext4_used_dirs_set(struct super_block *sb,
2995                                 struct ext4_group_desc *bg, __u32 count);
2996 extern void ext4_itable_unused_set(struct super_block *sb,
2997                                    struct ext4_group_desc *bg, __u32 count);
2998 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2999                                        struct ext4_group_desc *gdp);
3000 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
3001                                      struct ext4_group_desc *gdp);
3002 extern int ext4_register_li_request(struct super_block *sb,
3003                                     ext4_group_t first_not_zeroed);
3004
3005 static inline int ext4_has_metadata_csum(struct super_block *sb)
3006 {
3007         WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
3008                      !EXT4_SB(sb)->s_chksum_driver);
3009
3010         return ext4_has_feature_metadata_csum(sb) &&
3011                (EXT4_SB(sb)->s_chksum_driver != NULL);
3012 }
3013
3014 static inline int ext4_has_group_desc_csum(struct super_block *sb)
3015 {
3016         return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb);
3017 }
3018
3019 #define ext4_read_incompat_64bit_val(es, name) \
3020         (((es)->s_feature_incompat & cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT) \
3021                 ? (ext4_fsblk_t)le32_to_cpu(es->name##_hi) << 32 : 0) | \
3022                 le32_to_cpu(es->name##_lo))
3023
3024 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
3025 {
3026         return ext4_read_incompat_64bit_val(es, s_blocks_count);
3027 }
3028
3029 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
3030 {
3031         return ext4_read_incompat_64bit_val(es, s_r_blocks_count);
3032 }
3033
3034 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
3035 {
3036         return ext4_read_incompat_64bit_val(es, s_free_blocks_count);
3037 }
3038
3039 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
3040                                          ext4_fsblk_t blk)
3041 {
3042         es->s_blocks_count_lo = cpu_to_le32((u32)blk);
3043         es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
3044 }
3045
3046 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
3047                                               ext4_fsblk_t blk)
3048 {
3049         es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
3050         es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
3051 }
3052
3053 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
3054                                            ext4_fsblk_t blk)
3055 {
3056         es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
3057         es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
3058 }
3059
3060 static inline loff_t ext4_isize(struct super_block *sb,
3061                                 struct ext4_inode *raw_inode)
3062 {
3063         if (ext4_has_feature_largedir(sb) ||
3064             S_ISREG(le16_to_cpu(raw_inode->i_mode)))
3065                 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
3066                         le32_to_cpu(raw_inode->i_size_lo);
3067
3068         return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
3069 }
3070
3071 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
3072 {
3073         raw_inode->i_size_lo = cpu_to_le32(i_size);
3074         raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
3075 }
3076
3077 static inline
3078 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
3079                                             ext4_group_t group)
3080 {
3081          struct ext4_group_info **grp_info;
3082          long indexv, indexh;
3083          BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
3084          indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
3085          indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
3086          grp_info = sbi_array_rcu_deref(EXT4_SB(sb), s_group_info, indexv);
3087          return grp_info[indexh];
3088 }
3089
3090 /*
3091  * Reading s_groups_count requires using smp_rmb() afterwards.  See
3092  * the locking protocol documented in the comments of ext4_group_add()
3093  * in resize.c
3094  */
3095 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
3096 {
3097         ext4_group_t    ngroups = EXT4_SB(sb)->s_groups_count;
3098
3099         smp_rmb();
3100         return ngroups;
3101 }
3102
3103 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
3104                                              ext4_group_t block_group)
3105 {
3106         return block_group >> sbi->s_log_groups_per_flex;
3107 }
3108
3109 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
3110 {
3111         return 1 << sbi->s_log_groups_per_flex;
3112 }
3113
3114 #define ext4_std_error(sb, errno)                               \
3115 do {                                                            \
3116         if ((errno))                                            \
3117                 __ext4_std_error((sb), __func__, __LINE__, (errno));    \
3118 } while (0)
3119
3120 #ifdef CONFIG_SMP
3121 /* Each CPU can accumulate percpu_counter_batch clusters in their local
3122  * counters. So we need to make sure we have free clusters more
3123  * than percpu_counter_batch  * nr_cpu_ids. Also add a window of 4 times.
3124  */
3125 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
3126 #else
3127 #define EXT4_FREECLUSTERS_WATERMARK 0
3128 #endif
3129
3130 /* Update i_disksize. Requires i_mutex to avoid races with truncate */
3131 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
3132 {
3133         WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
3134                      !inode_is_locked(inode));
3135         down_write(&EXT4_I(inode)->i_data_sem);
3136         if (newsize > EXT4_I(inode)->i_disksize)
3137                 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize);
3138         up_write(&EXT4_I(inode)->i_data_sem);
3139 }
3140
3141 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
3142 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
3143 {
3144         int changed = 0;
3145
3146         if (newsize > inode->i_size) {
3147                 i_size_write(inode, newsize);
3148                 changed = 1;
3149         }
3150         if (newsize > EXT4_I(inode)->i_disksize) {
3151                 ext4_update_i_disksize(inode, newsize);
3152                 changed |= 2;
3153         }
3154         return changed;
3155 }
3156
3157 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
3158                                       loff_t len);
3159
3160 struct ext4_group_info {
3161         unsigned long   bb_state;
3162 #ifdef AGGRESSIVE_CHECK
3163         unsigned long   bb_check_counter;
3164 #endif
3165         struct rb_root  bb_free_root;
3166         ext4_grpblk_t   bb_first_free;  /* first free block */
3167         ext4_grpblk_t   bb_free;        /* total free blocks */
3168         ext4_grpblk_t   bb_fragments;   /* nr of freespace fragments */
3169         ext4_grpblk_t   bb_largest_free_order;/* order of largest frag in BG */
3170         struct          list_head bb_prealloc_list;
3171 #ifdef DOUBLE_CHECK
3172         void            *bb_bitmap;
3173 #endif
3174         struct rw_semaphore alloc_sem;
3175         ext4_grpblk_t   bb_counters[];  /* Nr of free power-of-two-block
3176                                          * regions, index is order.
3177                                          * bb_counters[3] = 5 means
3178                                          * 5 free 8-block regions. */
3179 };
3180
3181 #define EXT4_GROUP_INFO_NEED_INIT_BIT           0
3182 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT         1
3183 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT     2
3184 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT     3
3185 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT         \
3186         (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT)
3187 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT         \
3188         (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT)
3189 #define EXT4_GROUP_INFO_BBITMAP_READ_BIT        4
3190
3191 #define EXT4_MB_GRP_NEED_INIT(grp)      \
3192         (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
3193 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp)        \
3194         (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3195 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp)        \
3196         (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3197
3198 #define EXT4_MB_GRP_WAS_TRIMMED(grp)    \
3199         (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3200 #define EXT4_MB_GRP_SET_TRIMMED(grp)    \
3201         (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3202 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp)  \
3203         (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3204 #define EXT4_MB_GRP_TEST_AND_SET_READ(grp)      \
3205         (test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state)))
3206
3207 #define EXT4_MAX_CONTENTION             8
3208 #define EXT4_CONTENTION_THRESHOLD       2
3209
3210 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
3211                                               ext4_group_t group)
3212 {
3213         return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
3214 }
3215
3216 /*
3217  * Returns true if the filesystem is busy enough that attempts to
3218  * access the block group locks has run into contention.
3219  */
3220 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
3221 {
3222         return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
3223 }
3224
3225 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
3226 {
3227         spinlock_t *lock = ext4_group_lock_ptr(sb, group);
3228         if (spin_trylock(lock))
3229                 /*
3230                  * We're able to grab the lock right away, so drop the
3231                  * lock contention counter.
3232                  */
3233                 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
3234         else {
3235                 /*
3236                  * The lock is busy, so bump the contention counter,
3237                  * and then wait on the spin lock.
3238                  */
3239                 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
3240                                   EXT4_MAX_CONTENTION);
3241                 spin_lock(lock);
3242         }
3243 }
3244
3245 static inline void ext4_unlock_group(struct super_block *sb,
3246                                         ext4_group_t group)
3247 {
3248         spin_unlock(ext4_group_lock_ptr(sb, group));
3249 }
3250
3251 /*
3252  * Block validity checking
3253  */
3254 #define ext4_check_indirect_blockref(inode, bh)                         \
3255         ext4_check_blockref(__func__, __LINE__, inode,                  \
3256                             (__le32 *)(bh)->b_data,                     \
3257                             EXT4_ADDR_PER_BLOCK((inode)->i_sb))
3258
3259 #define ext4_ind_check_inode(inode)                                     \
3260         ext4_check_blockref(__func__, __LINE__, inode,                  \
3261                             EXT4_I(inode)->i_data,                      \
3262                             EXT4_NDIR_BLOCKS)
3263
3264 /*
3265  * Inodes and files operations
3266  */
3267
3268 /* dir.c */
3269 extern const struct file_operations ext4_dir_operations;
3270
3271 #ifdef CONFIG_UNICODE
3272 extern const struct dentry_operations ext4_dentry_ops;
3273 #endif
3274
3275 /* file.c */
3276 extern const struct inode_operations ext4_file_inode_operations;
3277 extern const struct file_operations ext4_file_operations;
3278 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
3279
3280 /* inline.c */
3281 extern int ext4_get_max_inline_size(struct inode *inode);
3282 extern int ext4_find_inline_data_nolock(struct inode *inode);
3283 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
3284                                  unsigned int len);
3285 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
3286
3287 extern int ext4_readpage_inline(struct inode *inode, struct page *page);
3288 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3289                                          struct inode *inode,
3290                                          loff_t pos, unsigned len,
3291                                          unsigned flags,
3292                                          struct page **pagep);
3293 extern int ext4_write_inline_data_end(struct inode *inode,
3294                                       loff_t pos, unsigned len,
3295                                       unsigned copied,
3296                                       struct page *page);
3297 extern struct buffer_head *
3298 ext4_journalled_write_inline_data(struct inode *inode,
3299                                   unsigned len,
3300                                   struct page *page);
3301 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
3302                                            struct inode *inode,
3303                                            loff_t pos, unsigned len,
3304                                            unsigned flags,
3305                                            struct page **pagep,
3306                                            void **fsdata);
3307 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
3308                                          unsigned len, unsigned copied,
3309                                          struct page *page);
3310 extern int ext4_try_add_inline_entry(handle_t *handle,
3311                                      struct ext4_filename *fname,
3312                                      struct inode *dir, struct inode *inode);
3313 extern int ext4_try_create_inline_dir(handle_t *handle,
3314                                       struct inode *parent,
3315                                       struct inode *inode);
3316 extern int ext4_read_inline_dir(struct file *filp,
3317                                 struct dir_context *ctx,
3318                                 int *has_inline_data);
3319 extern int ext4_inlinedir_to_tree(struct file *dir_file,
3320                                   struct inode *dir, ext4_lblk_t block,
3321                                   struct dx_hash_info *hinfo,
3322                                   __u32 start_hash, __u32 start_minor_hash,
3323                                   int *has_inline_data);
3324 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3325                                         struct ext4_filename *fname,
3326                                         struct ext4_dir_entry_2 **res_dir,
3327                                         int *has_inline_data);
3328 extern int ext4_delete_inline_entry(handle_t *handle,
3329                                     struct inode *dir,
3330                                     struct ext4_dir_entry_2 *de_del,
3331                                     struct buffer_head *bh,
3332                                     int *has_inline_data);
3333 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3334 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3335                                         struct ext4_dir_entry_2 **parent_de,
3336                                         int *retval);
3337 extern int ext4_inline_data_fiemap(struct inode *inode,
3338                                    struct fiemap_extent_info *fieinfo,
3339                                    int *has_inline, __u64 start, __u64 len);
3340
3341 struct iomap;
3342 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap);
3343
3344 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3345
3346 extern int ext4_convert_inline_data(struct inode *inode);
3347
3348 static inline int ext4_has_inline_data(struct inode *inode)
3349 {
3350         return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3351                EXT4_I(inode)->i_inline_off;
3352 }
3353
3354 /* namei.c */
3355 extern const struct inode_operations ext4_dir_inode_operations;
3356 extern const struct inode_operations ext4_special_inode_operations;
3357 extern struct dentry *ext4_get_parent(struct dentry *child);
3358 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3359                                  struct ext4_dir_entry_2 *de,
3360                                  int blocksize, int csum_size,
3361                                  unsigned int parent_ino, int dotdot_real_len);
3362 extern void ext4_initialize_dirent_tail(struct buffer_head *bh,
3363                                         unsigned int blocksize);
3364 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode,
3365                                       struct buffer_head *bh);
3366 extern int ext4_ci_compare(const struct inode *parent,
3367                            const struct qstr *fname,
3368                            const struct qstr *entry, bool quick);
3369
3370 #define S_SHIFT 12
3371 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = {
3372         [S_IFREG >> S_SHIFT]    = EXT4_FT_REG_FILE,
3373         [S_IFDIR >> S_SHIFT]    = EXT4_FT_DIR,
3374         [S_IFCHR >> S_SHIFT]    = EXT4_FT_CHRDEV,
3375         [S_IFBLK >> S_SHIFT]    = EXT4_FT_BLKDEV,
3376         [S_IFIFO >> S_SHIFT]    = EXT4_FT_FIFO,
3377         [S_IFSOCK >> S_SHIFT]   = EXT4_FT_SOCK,
3378         [S_IFLNK >> S_SHIFT]    = EXT4_FT_SYMLINK,
3379 };
3380
3381 static inline void ext4_set_de_type(struct super_block *sb,
3382                                 struct ext4_dir_entry_2 *de,
3383                                 umode_t mode) {
3384         if (ext4_has_feature_filetype(sb))
3385                 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3386 }
3387
3388 /* readpages.c */
3389 extern int ext4_mpage_readpages(struct inode *inode,
3390                 struct readahead_control *rac, struct page *page);
3391 extern int __init ext4_init_post_read_processing(void);
3392 extern void ext4_exit_post_read_processing(void);
3393
3394 /* symlink.c */
3395 extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3396 extern const struct inode_operations ext4_symlink_inode_operations;
3397 extern const struct inode_operations ext4_fast_symlink_inode_operations;
3398
3399 /* sysfs.c */
3400 extern int ext4_register_sysfs(struct super_block *sb);
3401 extern void ext4_unregister_sysfs(struct super_block *sb);
3402 extern int __init ext4_init_sysfs(void);
3403 extern void ext4_exit_sysfs(void);
3404
3405 /* block_validity */
3406 extern void ext4_release_system_zone(struct super_block *sb);
3407 extern int ext4_setup_system_zone(struct super_block *sb);
3408 extern int __init ext4_init_system_zone(void);
3409 extern void ext4_exit_system_zone(void);
3410 extern int ext4_inode_block_valid(struct inode *inode,
3411                                   ext4_fsblk_t start_blk,
3412                                   unsigned int count);
3413 extern int ext4_check_blockref(const char *, unsigned int,
3414                                struct inode *, __le32 *, unsigned int);
3415
3416 /* extents.c */
3417 struct ext4_ext_path;
3418 struct ext4_extent;
3419
3420 /*
3421  * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3422  * __le32.
3423  */
3424 #define EXT_MAX_BLOCKS  0xffffffff
3425
3426 extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode);
3427 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3428 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3429                                struct ext4_map_blocks *map, int flags);
3430 extern int ext4_ext_truncate(handle_t *, struct inode *);
3431 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3432                                  ext4_lblk_t end);
3433 extern void ext4_ext_init(struct super_block *);
3434 extern void ext4_ext_release(struct super_block *);
3435 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3436                           loff_t len);
3437 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3438                                           loff_t offset, ssize_t len);
3439 extern int ext4_convert_unwritten_io_end_vec(handle_t *handle,
3440                                              ext4_io_end_t *io_end);
3441 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3442                            struct ext4_map_blocks *map, int flags);
3443 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3444                                                    int num,
3445                                                    struct ext4_ext_path *path);
3446 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3447                                   struct ext4_ext_path **,
3448                                   struct ext4_extent *, int);
3449 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3450                                               struct ext4_ext_path **,
3451                                               int flags);
3452 extern void ext4_ext_drop_refs(struct ext4_ext_path *);
3453 extern int ext4_ext_check_inode(struct inode *inode);
3454 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3455 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3456                         __u64 start, __u64 len);
3457 extern int ext4_get_es_cache(struct inode *inode,
3458                              struct fiemap_extent_info *fieinfo,
3459                              __u64 start, __u64 len);
3460 extern int ext4_ext_precache(struct inode *inode);
3461 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3462                                 struct inode *inode2, ext4_lblk_t lblk1,
3463                              ext4_lblk_t lblk2,  ext4_lblk_t count,
3464                              int mark_unwritten,int *err);
3465 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu);
3466 extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode,
3467                                        int check_cred, int restart_cred,
3468                                        int revoke_cred);
3469
3470
3471 /* move_extent.c */
3472 extern void ext4_double_down_write_data_sem(struct inode *first,
3473                                             struct inode *second);
3474 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3475                                           struct inode *donor_inode);
3476 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3477                              __u64 start_orig, __u64 start_donor,
3478                              __u64 len, __u64 *moved_len);
3479
3480 /* page-io.c */
3481 extern int __init ext4_init_pageio(void);
3482 extern void ext4_exit_pageio(void);
3483 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3484 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3485 extern int ext4_put_io_end(ext4_io_end_t *io_end);
3486 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3487 extern void ext4_io_submit_init(struct ext4_io_submit *io,
3488                                 struct writeback_control *wbc);
3489 extern void ext4_end_io_rsv_work(struct work_struct *work);
3490 extern void ext4_io_submit(struct ext4_io_submit *io);
3491 extern int ext4_bio_write_page(struct ext4_io_submit *io,
3492                                struct page *page,
3493                                int len,
3494                                struct writeback_control *wbc,
3495                                bool keep_towrite);
3496 extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end);
3497 extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end);
3498
3499 /* mmp.c */
3500 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3501
3502 /* verity.c */
3503 extern const struct fsverity_operations ext4_verityops;
3504
3505 /*
3506  * Add new method to test whether block and inode bitmaps are properly
3507  * initialized. With uninit_bg reading the block from disk is not enough
3508  * to mark the bitmap uptodate. We need to also zero-out the bitmap
3509  */
3510 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3511
3512 static inline int bitmap_uptodate(struct buffer_head *bh)
3513 {
3514         return (buffer_uptodate(bh) &&
3515                         test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3516 }
3517 static inline void set_bitmap_uptodate(struct buffer_head *bh)
3518 {
3519         set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3520 }
3521
3522 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
3523
3524 /* For ioend & aio unwritten conversion wait queues */
3525 #define EXT4_WQ_HASH_SZ         37
3526 #define ext4_ioend_wq(v)   (&ext4__ioend_wq[((unsigned long)(v)) %\
3527                                             EXT4_WQ_HASH_SZ])
3528 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3529
3530 extern int ext4_resize_begin(struct super_block *sb);
3531 extern void ext4_resize_end(struct super_block *sb);
3532
3533 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
3534                                               struct ext4_io_end *io_end)
3535 {
3536         if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
3537                 io_end->flag |= EXT4_IO_END_UNWRITTEN;
3538                 atomic_inc(&EXT4_I(inode)->i_unwritten);
3539         }
3540 }
3541
3542 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3543 {
3544         struct inode *inode = io_end->inode;
3545
3546         if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
3547                 io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3548                 /* Wake up anyone waiting on unwritten extent conversion */
3549                 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
3550                         wake_up_all(ext4_ioend_wq(inode));
3551         }
3552 }
3553
3554 extern const struct iomap_ops ext4_iomap_ops;
3555 extern const struct iomap_ops ext4_iomap_overwrite_ops;
3556 extern const struct iomap_ops ext4_iomap_report_ops;
3557
3558 static inline int ext4_buffer_uptodate(struct buffer_head *bh)
3559 {
3560         /*
3561          * If the buffer has the write error flag, we have failed
3562          * to write out data in the block.  In this  case, we don't
3563          * have to read the block because we may read the old data
3564          * successfully.
3565          */
3566         if (!buffer_uptodate(bh) && buffer_write_io_error(bh))
3567                 set_buffer_uptodate(bh);
3568         return buffer_uptodate(bh);
3569 }
3570
3571 #endif  /* __KERNEL__ */
3572
3573 #define EFSBADCRC       EBADMSG         /* Bad CRC detected */
3574 #define EFSCORRUPTED    EUCLEAN         /* Filesystem is corrupted */
3575
3576 #endif  /* _EXT4_H */