xref: /openbmc/linux/fs/ext4/ext4.h (revision b737eecd)
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 #define ASSERT(assert)						\
102 do {									\
103 	if (unlikely(!(assert))) {					\
104 		printk(KERN_EMERG					\
105 		       "Assertion failure in %s() at %s:%d: '%s'\n",	\
106 		       __func__, __FILE__, __LINE__, #assert);		\
107 		BUG();							\
108 	}								\
109 } while (0)
110 
111 /* data type for block offset of block group */
112 typedef int ext4_grpblk_t;
113 
114 /* data type for filesystem-wide blocks number */
115 typedef unsigned long long ext4_fsblk_t;
116 
117 /* data type for file logical block number */
118 typedef __u32 ext4_lblk_t;
119 
120 /* data type for block group number */
121 typedef unsigned int ext4_group_t;
122 
123 enum SHIFT_DIRECTION {
124 	SHIFT_LEFT = 0,
125 	SHIFT_RIGHT,
126 };
127 
128 /*
129  * Flags used in mballoc's allocation_context flags field.
130  *
131  * Also used to show what's going on for debugging purposes when the
132  * flag field is exported via the traceport interface
133  */
134 
135 /* prefer goal again. length */
136 #define EXT4_MB_HINT_MERGE		0x0001
137 /* blocks already reserved */
138 #define EXT4_MB_HINT_RESERVED		0x0002
139 /* metadata is being allocated */
140 #define EXT4_MB_HINT_METADATA		0x0004
141 /* first blocks in the file */
142 #define EXT4_MB_HINT_FIRST		0x0008
143 /* search for the best chunk */
144 #define EXT4_MB_HINT_BEST		0x0010
145 /* data is being allocated */
146 #define EXT4_MB_HINT_DATA		0x0020
147 /* don't preallocate (for tails) */
148 #define EXT4_MB_HINT_NOPREALLOC		0x0040
149 /* allocate for locality group */
150 #define EXT4_MB_HINT_GROUP_ALLOC	0x0080
151 /* allocate goal blocks or none */
152 #define EXT4_MB_HINT_GOAL_ONLY		0x0100
153 /* goal is meaningful */
154 #define EXT4_MB_HINT_TRY_GOAL		0x0200
155 /* blocks already pre-reserved by delayed allocation */
156 #define EXT4_MB_DELALLOC_RESERVED	0x0400
157 /* We are doing stream allocation */
158 #define EXT4_MB_STREAM_ALLOC		0x0800
159 /* Use reserved root blocks if needed */
160 #define EXT4_MB_USE_ROOT_BLOCKS		0x1000
161 /* Use blocks from reserved pool */
162 #define EXT4_MB_USE_RESERVED		0x2000
163 /* Do strict check for free blocks while retrying block allocation */
164 #define EXT4_MB_STRICT_CHECK		0x4000
165 
166 struct ext4_allocation_request {
167 	/* target inode for block we're allocating */
168 	struct inode *inode;
169 	/* how many blocks we want to allocate */
170 	unsigned int len;
171 	/* logical block in target inode */
172 	ext4_lblk_t logical;
173 	/* the closest logical allocated block to the left */
174 	ext4_lblk_t lleft;
175 	/* the closest logical allocated block to the right */
176 	ext4_lblk_t lright;
177 	/* phys. target (a hint) */
178 	ext4_fsblk_t goal;
179 	/* phys. block for the closest logical allocated block to the left */
180 	ext4_fsblk_t pleft;
181 	/* phys. block for the closest logical allocated block to the right */
182 	ext4_fsblk_t pright;
183 	/* flags. see above EXT4_MB_HINT_* */
184 	unsigned int flags;
185 };
186 
187 /*
188  * Logical to physical block mapping, used by ext4_map_blocks()
189  *
190  * This structure is used to pass requests into ext4_map_blocks() as
191  * well as to store the information returned by ext4_map_blocks().  It
192  * takes less room on the stack than a struct buffer_head.
193  */
194 #define EXT4_MAP_NEW		BIT(BH_New)
195 #define EXT4_MAP_MAPPED		BIT(BH_Mapped)
196 #define EXT4_MAP_UNWRITTEN	BIT(BH_Unwritten)
197 #define EXT4_MAP_BOUNDARY	BIT(BH_Boundary)
198 #define EXT4_MAP_FLAGS		(EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
199 				 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
200 
201 struct ext4_map_blocks {
202 	ext4_fsblk_t m_pblk;
203 	ext4_lblk_t m_lblk;
204 	unsigned int m_len;
205 	unsigned int m_flags;
206 };
207 
208 /*
209  * Block validity checking, system zone rbtree.
210  */
211 struct ext4_system_blocks {
212 	struct rb_root root;
213 	struct rcu_head rcu;
214 };
215 
216 /*
217  * Flags for ext4_io_end->flags
218  */
219 #define	EXT4_IO_END_UNWRITTEN	0x0001
220 
221 struct ext4_io_end_vec {
222 	struct list_head list;		/* list of io_end_vec */
223 	loff_t offset;			/* offset in the file */
224 	ssize_t size;			/* size of the extent */
225 };
226 
227 /*
228  * For converting unwritten extents on a work queue. 'handle' is used for
229  * buffered writeback.
230  */
231 typedef struct ext4_io_end {
232 	struct list_head	list;		/* per-file finished IO list */
233 	handle_t		*handle;	/* handle reserved for extent
234 						 * conversion */
235 	struct inode		*inode;		/* file being written to */
236 	struct bio		*bio;		/* Linked list of completed
237 						 * bios covering the extent */
238 	unsigned int		flag;		/* unwritten or not */
239 	atomic_t		count;		/* reference counter */
240 	struct list_head	list_vec;	/* list of ext4_io_end_vec */
241 } ext4_io_end_t;
242 
243 struct ext4_io_submit {
244 	struct writeback_control *io_wbc;
245 	struct bio		*io_bio;
246 	ext4_io_end_t		*io_end;
247 	sector_t		io_next_block;
248 };
249 
250 /*
251  * Special inodes numbers
252  */
253 #define	EXT4_BAD_INO		 1	/* Bad blocks inode */
254 #define EXT4_ROOT_INO		 2	/* Root inode */
255 #define EXT4_USR_QUOTA_INO	 3	/* User quota inode */
256 #define EXT4_GRP_QUOTA_INO	 4	/* Group quota inode */
257 #define EXT4_BOOT_LOADER_INO	 5	/* Boot loader inode */
258 #define EXT4_UNDEL_DIR_INO	 6	/* Undelete directory inode */
259 #define EXT4_RESIZE_INO		 7	/* Reserved group descriptors inode */
260 #define EXT4_JOURNAL_INO	 8	/* Journal inode */
261 
262 /* First non-reserved inode for old ext4 filesystems */
263 #define EXT4_GOOD_OLD_FIRST_INO	11
264 
265 /*
266  * Maximal count of links to a file
267  */
268 #define EXT4_LINK_MAX		65000
269 
270 /*
271  * Macro-instructions used to manage several block sizes
272  */
273 #define EXT4_MIN_BLOCK_SIZE		1024
274 #define	EXT4_MAX_BLOCK_SIZE		65536
275 #define EXT4_MIN_BLOCK_LOG_SIZE		10
276 #define EXT4_MAX_BLOCK_LOG_SIZE		16
277 #define EXT4_MAX_CLUSTER_LOG_SIZE	30
278 #ifdef __KERNEL__
279 # define EXT4_BLOCK_SIZE(s)		((s)->s_blocksize)
280 #else
281 # define EXT4_BLOCK_SIZE(s)		(EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
282 #endif
283 #define	EXT4_ADDR_PER_BLOCK(s)		(EXT4_BLOCK_SIZE(s) / sizeof(__u32))
284 #define EXT4_CLUSTER_SIZE(s)		(EXT4_BLOCK_SIZE(s) << \
285 					 EXT4_SB(s)->s_cluster_bits)
286 #ifdef __KERNEL__
287 # define EXT4_BLOCK_SIZE_BITS(s)	((s)->s_blocksize_bits)
288 # define EXT4_CLUSTER_BITS(s)		(EXT4_SB(s)->s_cluster_bits)
289 #else
290 # define EXT4_BLOCK_SIZE_BITS(s)	((s)->s_log_block_size + 10)
291 #endif
292 #ifdef __KERNEL__
293 #define	EXT4_ADDR_PER_BLOCK_BITS(s)	(EXT4_SB(s)->s_addr_per_block_bits)
294 #define EXT4_INODE_SIZE(s)		(EXT4_SB(s)->s_inode_size)
295 #define EXT4_FIRST_INO(s)		(EXT4_SB(s)->s_first_ino)
296 #else
297 #define EXT4_INODE_SIZE(s)	(((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
298 				 EXT4_GOOD_OLD_INODE_SIZE : \
299 				 (s)->s_inode_size)
300 #define EXT4_FIRST_INO(s)	(((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
301 				 EXT4_GOOD_OLD_FIRST_INO : \
302 				 (s)->s_first_ino)
303 #endif
304 #define EXT4_BLOCK_ALIGN(size, blkbits)		ALIGN((size), (1 << (blkbits)))
305 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \
306 	((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \
307 								  blkbits))
308 
309 /* Translate a block number to a cluster number */
310 #define EXT4_B2C(sbi, blk)	((blk) >> (sbi)->s_cluster_bits)
311 /* Translate a cluster number to a block number */
312 #define EXT4_C2B(sbi, cluster)	((cluster) << (sbi)->s_cluster_bits)
313 /* Translate # of blks to # of clusters */
314 #define EXT4_NUM_B2C(sbi, blks)	(((blks) + (sbi)->s_cluster_ratio - 1) >> \
315 				 (sbi)->s_cluster_bits)
316 /* Mask out the low bits to get the starting block of the cluster */
317 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) &				\
318 				  ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
319 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) &				\
320 				  ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
321 /* Fill in the low bits to get the last block of the cluster */
322 #define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) |				\
323 				    ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1))
324 /* Get the cluster offset */
325 #define EXT4_PBLK_COFF(s, pblk) ((pblk) &				\
326 				 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
327 #define EXT4_LBLK_COFF(s, lblk) ((lblk) &				\
328 				 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
329 
330 /*
331  * Structure of a blocks group descriptor
332  */
333 struct ext4_group_desc
334 {
335 	__le32	bg_block_bitmap_lo;	/* Blocks bitmap block */
336 	__le32	bg_inode_bitmap_lo;	/* Inodes bitmap block */
337 	__le32	bg_inode_table_lo;	/* Inodes table block */
338 	__le16	bg_free_blocks_count_lo;/* Free blocks count */
339 	__le16	bg_free_inodes_count_lo;/* Free inodes count */
340 	__le16	bg_used_dirs_count_lo;	/* Directories count */
341 	__le16	bg_flags;		/* EXT4_BG_flags (INODE_UNINIT, etc) */
342 	__le32  bg_exclude_bitmap_lo;   /* Exclude bitmap for snapshots */
343 	__le16  bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
344 	__le16  bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
345 	__le16  bg_itable_unused_lo;	/* Unused inodes count */
346 	__le16  bg_checksum;		/* crc16(sb_uuid+group+desc) */
347 	__le32	bg_block_bitmap_hi;	/* Blocks bitmap block MSB */
348 	__le32	bg_inode_bitmap_hi;	/* Inodes bitmap block MSB */
349 	__le32	bg_inode_table_hi;	/* Inodes table block MSB */
350 	__le16	bg_free_blocks_count_hi;/* Free blocks count MSB */
351 	__le16	bg_free_inodes_count_hi;/* Free inodes count MSB */
352 	__le16	bg_used_dirs_count_hi;	/* Directories count MSB */
353 	__le16  bg_itable_unused_hi;    /* Unused inodes count MSB */
354 	__le32  bg_exclude_bitmap_hi;   /* Exclude bitmap block MSB */
355 	__le16  bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
356 	__le16  bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
357 	__u32   bg_reserved;
358 };
359 
360 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END	\
361 	(offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
362 	 sizeof(__le16))
363 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END	\
364 	(offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
365 	 sizeof(__le16))
366 
367 /*
368  * Structure of a flex block group info
369  */
370 
371 struct flex_groups {
372 	atomic64_t	free_clusters;
373 	atomic_t	free_inodes;
374 	atomic_t	used_dirs;
375 };
376 
377 #define EXT4_BG_INODE_UNINIT	0x0001 /* Inode table/bitmap not in use */
378 #define EXT4_BG_BLOCK_UNINIT	0x0002 /* Block bitmap not in use */
379 #define EXT4_BG_INODE_ZEROED	0x0004 /* On-disk itable initialized to zero */
380 
381 /*
382  * Macro-instructions used to manage group descriptors
383  */
384 #define EXT4_MIN_DESC_SIZE		32
385 #define EXT4_MIN_DESC_SIZE_64BIT	64
386 #define	EXT4_MAX_DESC_SIZE		EXT4_MIN_BLOCK_SIZE
387 #define EXT4_DESC_SIZE(s)		(EXT4_SB(s)->s_desc_size)
388 #ifdef __KERNEL__
389 # define EXT4_BLOCKS_PER_GROUP(s)	(EXT4_SB(s)->s_blocks_per_group)
390 # define EXT4_CLUSTERS_PER_GROUP(s)	(EXT4_SB(s)->s_clusters_per_group)
391 # define EXT4_DESC_PER_BLOCK(s)		(EXT4_SB(s)->s_desc_per_block)
392 # define EXT4_INODES_PER_GROUP(s)	(EXT4_SB(s)->s_inodes_per_group)
393 # define EXT4_DESC_PER_BLOCK_BITS(s)	(EXT4_SB(s)->s_desc_per_block_bits)
394 #else
395 # define EXT4_BLOCKS_PER_GROUP(s)	((s)->s_blocks_per_group)
396 # define EXT4_DESC_PER_BLOCK(s)		(EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
397 # define EXT4_INODES_PER_GROUP(s)	((s)->s_inodes_per_group)
398 #endif
399 
400 /*
401  * Constants relative to the data blocks
402  */
403 #define	EXT4_NDIR_BLOCKS		12
404 #define	EXT4_IND_BLOCK			EXT4_NDIR_BLOCKS
405 #define	EXT4_DIND_BLOCK			(EXT4_IND_BLOCK + 1)
406 #define	EXT4_TIND_BLOCK			(EXT4_DIND_BLOCK + 1)
407 #define	EXT4_N_BLOCKS			(EXT4_TIND_BLOCK + 1)
408 
409 /*
410  * Inode flags
411  */
412 #define	EXT4_SECRM_FL			0x00000001 /* Secure deletion */
413 #define	EXT4_UNRM_FL			0x00000002 /* Undelete */
414 #define	EXT4_COMPR_FL			0x00000004 /* Compress file */
415 #define EXT4_SYNC_FL			0x00000008 /* Synchronous updates */
416 #define EXT4_IMMUTABLE_FL		0x00000010 /* Immutable file */
417 #define EXT4_APPEND_FL			0x00000020 /* writes to file may only append */
418 #define EXT4_NODUMP_FL			0x00000040 /* do not dump file */
419 #define EXT4_NOATIME_FL			0x00000080 /* do not update atime */
420 /* Reserved for compression usage... */
421 #define EXT4_DIRTY_FL			0x00000100
422 #define EXT4_COMPRBLK_FL		0x00000200 /* One or more compressed clusters */
423 #define EXT4_NOCOMPR_FL			0x00000400 /* Don't compress */
424 	/* nb: was previously EXT2_ECOMPR_FL */
425 #define EXT4_ENCRYPT_FL			0x00000800 /* encrypted file */
426 /* End compression flags --- maybe not all used */
427 #define EXT4_INDEX_FL			0x00001000 /* hash-indexed directory */
428 #define EXT4_IMAGIC_FL			0x00002000 /* AFS directory */
429 #define EXT4_JOURNAL_DATA_FL		0x00004000 /* file data should be journaled */
430 #define EXT4_NOTAIL_FL			0x00008000 /* file tail should not be merged */
431 #define EXT4_DIRSYNC_FL			0x00010000 /* dirsync behaviour (directories only) */
432 #define EXT4_TOPDIR_FL			0x00020000 /* Top of directory hierarchies*/
433 #define EXT4_HUGE_FILE_FL               0x00040000 /* Set to each huge file */
434 #define EXT4_EXTENTS_FL			0x00080000 /* Inode uses extents */
435 #define EXT4_VERITY_FL			0x00100000 /* Verity protected inode */
436 #define EXT4_EA_INODE_FL	        0x00200000 /* Inode used for large EA */
437 /* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */
438 
439 #define EXT4_DAX_FL			0x02000000 /* Inode is DAX */
440 
441 #define EXT4_INLINE_DATA_FL		0x10000000 /* Inode has inline data. */
442 #define EXT4_PROJINHERIT_FL		0x20000000 /* Create with parents projid */
443 #define EXT4_CASEFOLD_FL		0x40000000 /* Casefolded directory */
444 #define EXT4_RESERVED_FL		0x80000000 /* reserved for ext4 lib */
445 
446 /* User modifiable flags */
447 #define EXT4_FL_USER_MODIFIABLE		(EXT4_SECRM_FL | \
448 					 EXT4_UNRM_FL | \
449 					 EXT4_COMPR_FL | \
450 					 EXT4_SYNC_FL | \
451 					 EXT4_IMMUTABLE_FL | \
452 					 EXT4_APPEND_FL | \
453 					 EXT4_NODUMP_FL | \
454 					 EXT4_NOATIME_FL | \
455 					 EXT4_JOURNAL_DATA_FL | \
456 					 EXT4_NOTAIL_FL | \
457 					 EXT4_DIRSYNC_FL | \
458 					 EXT4_TOPDIR_FL | \
459 					 EXT4_EXTENTS_FL | \
460 					 0x00400000 /* EXT4_EOFBLOCKS_FL */ | \
461 					 EXT4_DAX_FL | \
462 					 EXT4_PROJINHERIT_FL | \
463 					 EXT4_CASEFOLD_FL)
464 
465 /* User visible flags */
466 #define EXT4_FL_USER_VISIBLE		(EXT4_FL_USER_MODIFIABLE | \
467 					 EXT4_DIRTY_FL | \
468 					 EXT4_COMPRBLK_FL | \
469 					 EXT4_NOCOMPR_FL | \
470 					 EXT4_ENCRYPT_FL | \
471 					 EXT4_INDEX_FL | \
472 					 EXT4_VERITY_FL | \
473 					 EXT4_INLINE_DATA_FL)
474 
475 /* Flags we can manipulate with through FS_IOC_FSSETXATTR */
476 #define EXT4_FL_XFLAG_VISIBLE		(EXT4_SYNC_FL | \
477 					 EXT4_IMMUTABLE_FL | \
478 					 EXT4_APPEND_FL | \
479 					 EXT4_NODUMP_FL | \
480 					 EXT4_NOATIME_FL | \
481 					 EXT4_PROJINHERIT_FL | \
482 					 EXT4_DAX_FL)
483 
484 /* Flags that should be inherited by new inodes from their parent. */
485 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
486 			   EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
487 			   EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
488 			   EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
489 			   EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL |\
490 			   EXT4_DAX_FL)
491 
492 /* Flags that are appropriate for regular files (all but dir-specific ones). */
493 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\
494 			   EXT4_PROJINHERIT_FL))
495 
496 /* Flags that are appropriate for non-directories/regular files. */
497 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
498 
499 /* The only flags that should be swapped */
500 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL)
501 
502 /* Flags which are mutually exclusive to DAX */
503 #define EXT4_DAX_MUT_EXCL (EXT4_VERITY_FL | EXT4_ENCRYPT_FL |\
504 			   EXT4_JOURNAL_DATA_FL | EXT4_INLINE_DATA_FL)
505 
506 /* Mask out flags that are inappropriate for the given type of inode. */
507 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
508 {
509 	if (S_ISDIR(mode))
510 		return flags;
511 	else if (S_ISREG(mode))
512 		return flags & EXT4_REG_FLMASK;
513 	else
514 		return flags & EXT4_OTHER_FLMASK;
515 }
516 
517 /*
518  * Inode flags used for atomic set/get
519  */
520 enum {
521 	EXT4_INODE_SECRM	= 0,	/* Secure deletion */
522 	EXT4_INODE_UNRM		= 1,	/* Undelete */
523 	EXT4_INODE_COMPR	= 2,	/* Compress file */
524 	EXT4_INODE_SYNC		= 3,	/* Synchronous updates */
525 	EXT4_INODE_IMMUTABLE	= 4,	/* Immutable file */
526 	EXT4_INODE_APPEND	= 5,	/* writes to file may only append */
527 	EXT4_INODE_NODUMP	= 6,	/* do not dump file */
528 	EXT4_INODE_NOATIME	= 7,	/* do not update atime */
529 /* Reserved for compression usage... */
530 	EXT4_INODE_DIRTY	= 8,
531 	EXT4_INODE_COMPRBLK	= 9,	/* One or more compressed clusters */
532 	EXT4_INODE_NOCOMPR	= 10,	/* Don't compress */
533 	EXT4_INODE_ENCRYPT	= 11,	/* Encrypted file */
534 /* End compression flags --- maybe not all used */
535 	EXT4_INODE_INDEX	= 12,	/* hash-indexed directory */
536 	EXT4_INODE_IMAGIC	= 13,	/* AFS directory */
537 	EXT4_INODE_JOURNAL_DATA	= 14,	/* file data should be journaled */
538 	EXT4_INODE_NOTAIL	= 15,	/* file tail should not be merged */
539 	EXT4_INODE_DIRSYNC	= 16,	/* dirsync behaviour (directories only) */
540 	EXT4_INODE_TOPDIR	= 17,	/* Top of directory hierarchies*/
541 	EXT4_INODE_HUGE_FILE	= 18,	/* Set to each huge file */
542 	EXT4_INODE_EXTENTS	= 19,	/* Inode uses extents */
543 	EXT4_INODE_VERITY	= 20,	/* Verity protected inode */
544 	EXT4_INODE_EA_INODE	= 21,	/* Inode used for large EA */
545 /* 22 was formerly EXT4_INODE_EOFBLOCKS */
546 	EXT4_INODE_DAX		= 25,	/* Inode is DAX */
547 	EXT4_INODE_INLINE_DATA	= 28,	/* Data in inode. */
548 	EXT4_INODE_PROJINHERIT	= 29,	/* Create with parents projid */
549 	EXT4_INODE_CASEFOLD	= 30,	/* Casefolded directory */
550 	EXT4_INODE_RESERVED	= 31,	/* reserved for ext4 lib */
551 };
552 
553 /*
554  * Since it's pretty easy to mix up bit numbers and hex values, we use a
555  * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
556  * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
557  * any extra space in the compiled kernel image, otherwise, the build will fail.
558  * It's important that these values are the same, since we are using
559  * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
560  * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
561  * values found in ext2, ext3 and ext4 filesystems, and of course the values
562  * defined in e2fsprogs.
563  *
564  * It's not paranoia if the Murphy's Law really *is* out to get you.  :-)
565  */
566 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
567 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
568 
569 static inline void ext4_check_flag_values(void)
570 {
571 	CHECK_FLAG_VALUE(SECRM);
572 	CHECK_FLAG_VALUE(UNRM);
573 	CHECK_FLAG_VALUE(COMPR);
574 	CHECK_FLAG_VALUE(SYNC);
575 	CHECK_FLAG_VALUE(IMMUTABLE);
576 	CHECK_FLAG_VALUE(APPEND);
577 	CHECK_FLAG_VALUE(NODUMP);
578 	CHECK_FLAG_VALUE(NOATIME);
579 	CHECK_FLAG_VALUE(DIRTY);
580 	CHECK_FLAG_VALUE(COMPRBLK);
581 	CHECK_FLAG_VALUE(NOCOMPR);
582 	CHECK_FLAG_VALUE(ENCRYPT);
583 	CHECK_FLAG_VALUE(INDEX);
584 	CHECK_FLAG_VALUE(IMAGIC);
585 	CHECK_FLAG_VALUE(JOURNAL_DATA);
586 	CHECK_FLAG_VALUE(NOTAIL);
587 	CHECK_FLAG_VALUE(DIRSYNC);
588 	CHECK_FLAG_VALUE(TOPDIR);
589 	CHECK_FLAG_VALUE(HUGE_FILE);
590 	CHECK_FLAG_VALUE(EXTENTS);
591 	CHECK_FLAG_VALUE(VERITY);
592 	CHECK_FLAG_VALUE(EA_INODE);
593 	CHECK_FLAG_VALUE(INLINE_DATA);
594 	CHECK_FLAG_VALUE(PROJINHERIT);
595 	CHECK_FLAG_VALUE(CASEFOLD);
596 	CHECK_FLAG_VALUE(RESERVED);
597 }
598 
599 /* Used to pass group descriptor data when online resize is done */
600 struct ext4_new_group_input {
601 	__u32 group;		/* Group number for this data */
602 	__u64 block_bitmap;	/* Absolute block number of block bitmap */
603 	__u64 inode_bitmap;	/* Absolute block number of inode bitmap */
604 	__u64 inode_table;	/* Absolute block number of inode table start */
605 	__u32 blocks_count;	/* Total number of blocks in this group */
606 	__u16 reserved_blocks;	/* Number of reserved blocks in this group */
607 	__u16 unused;
608 };
609 
610 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
611 struct compat_ext4_new_group_input {
612 	u32 group;
613 	compat_u64 block_bitmap;
614 	compat_u64 inode_bitmap;
615 	compat_u64 inode_table;
616 	u32 blocks_count;
617 	u16 reserved_blocks;
618 	u16 unused;
619 };
620 #endif
621 
622 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
623 struct ext4_new_group_data {
624 	__u32 group;
625 	__u64 block_bitmap;
626 	__u64 inode_bitmap;
627 	__u64 inode_table;
628 	__u32 blocks_count;
629 	__u16 reserved_blocks;
630 	__u16 mdata_blocks;
631 	__u32 free_clusters_count;
632 };
633 
634 /* Indexes used to index group tables in ext4_new_group_data */
635 enum {
636 	BLOCK_BITMAP = 0,	/* block bitmap */
637 	INODE_BITMAP,		/* inode bitmap */
638 	INODE_TABLE,		/* inode tables */
639 	GROUP_TABLE_COUNT,
640 };
641 
642 /*
643  * Flags used by ext4_map_blocks()
644  */
645 	/* Allocate any needed blocks and/or convert an unwritten
646 	   extent to be an initialized ext4 */
647 #define EXT4_GET_BLOCKS_CREATE			0x0001
648 	/* Request the creation of an unwritten extent */
649 #define EXT4_GET_BLOCKS_UNWRIT_EXT		0x0002
650 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT	(EXT4_GET_BLOCKS_UNWRIT_EXT|\
651 						 EXT4_GET_BLOCKS_CREATE)
652 	/* Caller is from the delayed allocation writeout path
653 	 * finally doing the actual allocation of delayed blocks */
654 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE	0x0004
655 	/* caller is from the direct IO path, request to creation of an
656 	unwritten extents if not allocated, split the unwritten
657 	extent if blocks has been preallocated already*/
658 #define EXT4_GET_BLOCKS_PRE_IO			0x0008
659 #define EXT4_GET_BLOCKS_CONVERT			0x0010
660 #define EXT4_GET_BLOCKS_IO_CREATE_EXT		(EXT4_GET_BLOCKS_PRE_IO|\
661 					 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
662 	/* Convert extent to initialized after IO complete */
663 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT		(EXT4_GET_BLOCKS_CONVERT|\
664 					 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
665 	/* Eventual metadata allocation (due to growing extent tree)
666 	 * should not fail, so try to use reserved blocks for that.*/
667 #define EXT4_GET_BLOCKS_METADATA_NOFAIL		0x0020
668 	/* Don't normalize allocation size (used for fallocate) */
669 #define EXT4_GET_BLOCKS_NO_NORMALIZE		0x0040
670 	/* Convert written extents to unwritten */
671 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN	0x0100
672 	/* Write zeros to newly created written extents */
673 #define EXT4_GET_BLOCKS_ZERO			0x0200
674 #define EXT4_GET_BLOCKS_CREATE_ZERO		(EXT4_GET_BLOCKS_CREATE |\
675 					EXT4_GET_BLOCKS_ZERO)
676 	/* Caller will submit data before dropping transaction handle. This
677 	 * allows jbd2 to avoid submitting data before commit. */
678 #define EXT4_GET_BLOCKS_IO_SUBMIT		0x0400
679 
680 /*
681  * The bit position of these flags must not overlap with any of the
682  * EXT4_GET_BLOCKS_*.  They are used by ext4_find_extent(),
683  * read_extent_tree_block(), ext4_split_extent_at(),
684  * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
685  * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
686  * caching the extents when reading from the extent tree while a
687  * truncate or punch hole operation is in progress.
688  */
689 #define EXT4_EX_NOCACHE				0x40000000
690 #define EXT4_EX_FORCE_CACHE			0x20000000
691 #define EXT4_EX_NOFAIL				0x10000000
692 
693 /*
694  * Flags used by ext4_free_blocks
695  */
696 #define EXT4_FREE_BLOCKS_METADATA		0x0001
697 #define EXT4_FREE_BLOCKS_FORGET			0x0002
698 #define EXT4_FREE_BLOCKS_VALIDATED		0x0004
699 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE		0x0008
700 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER	0x0010
701 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER	0x0020
702 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER      0x0040
703 
704 /*
705  * ioctl commands
706  */
707 #define	EXT4_IOC_GETVERSION		_IOR('f', 3, long)
708 #define	EXT4_IOC_SETVERSION		_IOW('f', 4, long)
709 #define	EXT4_IOC_GETVERSION_OLD		FS_IOC_GETVERSION
710 #define	EXT4_IOC_SETVERSION_OLD		FS_IOC_SETVERSION
711 #define EXT4_IOC_GETRSVSZ		_IOR('f', 5, long)
712 #define EXT4_IOC_SETRSVSZ		_IOW('f', 6, long)
713 #define EXT4_IOC_GROUP_EXTEND		_IOW('f', 7, unsigned long)
714 #define EXT4_IOC_GROUP_ADD		_IOW('f', 8, struct ext4_new_group_input)
715 #define EXT4_IOC_MIGRATE		_IO('f', 9)
716  /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
717  /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
718 #define EXT4_IOC_ALLOC_DA_BLKS		_IO('f', 12)
719 #define EXT4_IOC_MOVE_EXT		_IOWR('f', 15, struct move_extent)
720 #define EXT4_IOC_RESIZE_FS		_IOW('f', 16, __u64)
721 #define EXT4_IOC_SWAP_BOOT		_IO('f', 17)
722 #define EXT4_IOC_PRECACHE_EXTENTS	_IO('f', 18)
723 /* ioctl codes 19--39 are reserved for fscrypt */
724 #define EXT4_IOC_CLEAR_ES_CACHE		_IO('f', 40)
725 #define EXT4_IOC_GETSTATE		_IOW('f', 41, __u32)
726 #define EXT4_IOC_GET_ES_CACHE		_IOWR('f', 42, struct fiemap)
727 
728 #define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32)
729 
730 /*
731  * Flags for going down operation
732  */
733 #define EXT4_GOING_FLAGS_DEFAULT		0x0	/* going down */
734 #define EXT4_GOING_FLAGS_LOGFLUSH		0x1	/* flush log but not data */
735 #define EXT4_GOING_FLAGS_NOLOGFLUSH		0x2	/* don't flush log nor data */
736 
737 /*
738  * Flags returned by EXT4_IOC_GETSTATE
739  *
740  * We only expose to userspace a subset of the state flags in
741  * i_state_flags
742  */
743 #define EXT4_STATE_FLAG_EXT_PRECACHED	0x00000001
744 #define EXT4_STATE_FLAG_NEW		0x00000002
745 #define EXT4_STATE_FLAG_NEWENTRY	0x00000004
746 #define EXT4_STATE_FLAG_DA_ALLOC_CLOSE	0x00000008
747 
748 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
749 /*
750  * ioctl commands in 32 bit emulation
751  */
752 #define EXT4_IOC32_GETVERSION		_IOR('f', 3, int)
753 #define EXT4_IOC32_SETVERSION		_IOW('f', 4, int)
754 #define EXT4_IOC32_GETRSVSZ		_IOR('f', 5, int)
755 #define EXT4_IOC32_SETRSVSZ		_IOW('f', 6, int)
756 #define EXT4_IOC32_GROUP_EXTEND		_IOW('f', 7, unsigned int)
757 #define EXT4_IOC32_GROUP_ADD		_IOW('f', 8, struct compat_ext4_new_group_input)
758 #define EXT4_IOC32_GETVERSION_OLD	FS_IOC32_GETVERSION
759 #define EXT4_IOC32_SETVERSION_OLD	FS_IOC32_SETVERSION
760 #endif
761 
762 /*
763  * Returned by EXT4_IOC_GET_ES_CACHE as an additional possible flag.
764  * It indicates that the entry in extent status cache is for a hole.
765  */
766 #define EXT4_FIEMAP_EXTENT_HOLE		0x08000000
767 
768 /* Max physical block we can address w/o extents */
769 #define EXT4_MAX_BLOCK_FILE_PHYS	0xFFFFFFFF
770 
771 /* Max logical block we can support */
772 #define EXT4_MAX_LOGICAL_BLOCK		0xFFFFFFFE
773 
774 /*
775  * Structure of an inode on the disk
776  */
777 struct ext4_inode {
778 	__le16	i_mode;		/* File mode */
779 	__le16	i_uid;		/* Low 16 bits of Owner Uid */
780 	__le32	i_size_lo;	/* Size in bytes */
781 	__le32	i_atime;	/* Access time */
782 	__le32	i_ctime;	/* Inode Change time */
783 	__le32	i_mtime;	/* Modification time */
784 	__le32	i_dtime;	/* Deletion Time */
785 	__le16	i_gid;		/* Low 16 bits of Group Id */
786 	__le16	i_links_count;	/* Links count */
787 	__le32	i_blocks_lo;	/* Blocks count */
788 	__le32	i_flags;	/* File flags */
789 	union {
790 		struct {
791 			__le32  l_i_version;
792 		} linux1;
793 		struct {
794 			__u32  h_i_translator;
795 		} hurd1;
796 		struct {
797 			__u32  m_i_reserved1;
798 		} masix1;
799 	} osd1;				/* OS dependent 1 */
800 	__le32	i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
801 	__le32	i_generation;	/* File version (for NFS) */
802 	__le32	i_file_acl_lo;	/* File ACL */
803 	__le32	i_size_high;
804 	__le32	i_obso_faddr;	/* Obsoleted fragment address */
805 	union {
806 		struct {
807 			__le16	l_i_blocks_high; /* were l_i_reserved1 */
808 			__le16	l_i_file_acl_high;
809 			__le16	l_i_uid_high;	/* these 2 fields */
810 			__le16	l_i_gid_high;	/* were reserved2[0] */
811 			__le16	l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
812 			__le16	l_i_reserved;
813 		} linux2;
814 		struct {
815 			__le16	h_i_reserved1;	/* Obsoleted fragment number/size which are removed in ext4 */
816 			__u16	h_i_mode_high;
817 			__u16	h_i_uid_high;
818 			__u16	h_i_gid_high;
819 			__u32	h_i_author;
820 		} hurd2;
821 		struct {
822 			__le16	h_i_reserved1;	/* Obsoleted fragment number/size which are removed in ext4 */
823 			__le16	m_i_file_acl_high;
824 			__u32	m_i_reserved2[2];
825 		} masix2;
826 	} osd2;				/* OS dependent 2 */
827 	__le16	i_extra_isize;
828 	__le16	i_checksum_hi;	/* crc32c(uuid+inum+inode) BE */
829 	__le32  i_ctime_extra;  /* extra Change time      (nsec << 2 | epoch) */
830 	__le32  i_mtime_extra;  /* extra Modification time(nsec << 2 | epoch) */
831 	__le32  i_atime_extra;  /* extra Access time      (nsec << 2 | epoch) */
832 	__le32  i_crtime;       /* File Creation time */
833 	__le32  i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
834 	__le32  i_version_hi;	/* high 32 bits for 64-bit version */
835 	__le32	i_projid;	/* Project ID */
836 };
837 
838 struct move_extent {
839 	__u32 reserved;		/* should be zero */
840 	__u32 donor_fd;		/* donor file descriptor */
841 	__u64 orig_start;	/* logical start offset in block for orig */
842 	__u64 donor_start;	/* logical start offset in block for donor */
843 	__u64 len;		/* block length to be moved */
844 	__u64 moved_len;	/* moved block length */
845 };
846 
847 #define EXT4_EPOCH_BITS 2
848 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
849 #define EXT4_NSEC_MASK  (~0UL << EXT4_EPOCH_BITS)
850 
851 /*
852  * Extended fields will fit into an inode if the filesystem was formatted
853  * with large inodes (-I 256 or larger) and there are not currently any EAs
854  * consuming all of the available space. For new inodes we always reserve
855  * enough space for the kernel's known extended fields, but for inodes
856  * created with an old kernel this might not have been the case. None of
857  * the extended inode fields is critical for correct filesystem operation.
858  * This macro checks if a certain field fits in the inode. Note that
859  * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
860  */
861 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field)	\
862 	((offsetof(typeof(*ext4_inode), field) +	\
863 	  sizeof((ext4_inode)->field))			\
864 	<= (EXT4_GOOD_OLD_INODE_SIZE +			\
865 	    (einode)->i_extra_isize))			\
866 
867 /*
868  * We use an encoding that preserves the times for extra epoch "00":
869  *
870  * extra  msb of                         adjust for signed
871  * epoch  32-bit                         32-bit tv_sec to
872  * bits   time    decoded 64-bit tv_sec  64-bit tv_sec      valid time range
873  * 0 0    1    -0x80000000..-0x00000001  0x000000000 1901-12-13..1969-12-31
874  * 0 0    0    0x000000000..0x07fffffff  0x000000000 1970-01-01..2038-01-19
875  * 0 1    1    0x080000000..0x0ffffffff  0x100000000 2038-01-19..2106-02-07
876  * 0 1    0    0x100000000..0x17fffffff  0x100000000 2106-02-07..2174-02-25
877  * 1 0    1    0x180000000..0x1ffffffff  0x200000000 2174-02-25..2242-03-16
878  * 1 0    0    0x200000000..0x27fffffff  0x200000000 2242-03-16..2310-04-04
879  * 1 1    1    0x280000000..0x2ffffffff  0x300000000 2310-04-04..2378-04-22
880  * 1 1    0    0x300000000..0x37fffffff  0x300000000 2378-04-22..2446-05-10
881  *
882  * Note that previous versions of the kernel on 64-bit systems would
883  * incorrectly use extra epoch bits 1,1 for dates between 1901 and
884  * 1970.  e2fsck will correct this, assuming that it is run on the
885  * affected filesystem before 2242.
886  */
887 
888 static inline __le32 ext4_encode_extra_time(struct timespec64 *time)
889 {
890 	u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK;
891 	return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
892 }
893 
894 static inline void ext4_decode_extra_time(struct timespec64 *time,
895 					  __le32 extra)
896 {
897 	if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK)))
898 		time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
899 	time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
900 }
901 
902 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode)				\
903 do {										\
904 	if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     {\
905 		(raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec);	\
906 		(raw_inode)->xtime ## _extra =					\
907 				ext4_encode_extra_time(&(inode)->xtime);	\
908 		}								\
909 	else	\
910 		(raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (inode)->xtime.tv_sec, S32_MIN, S32_MAX));	\
911 } while (0)
912 
913 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode)			       \
914 do {									       \
915 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))		       \
916 		(raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec);      \
917 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))	       \
918 		(raw_inode)->xtime ## _extra =				       \
919 				ext4_encode_extra_time(&(einode)->xtime);      \
920 } while (0)
921 
922 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode)				\
923 do {										\
924 	(inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime);	\
925 	if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {	\
926 		ext4_decode_extra_time(&(inode)->xtime,				\
927 				       raw_inode->xtime ## _extra);		\
928 		}								\
929 	else									\
930 		(inode)->xtime.tv_nsec = 0;					\
931 } while (0)
932 
933 
934 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode)			       \
935 do {									       \
936 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))		       \
937 		(einode)->xtime.tv_sec = 				       \
938 			(signed)le32_to_cpu((raw_inode)->xtime);	       \
939 	else								       \
940 		(einode)->xtime.tv_sec = 0;				       \
941 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))	       \
942 		ext4_decode_extra_time(&(einode)->xtime,		       \
943 				       raw_inode->xtime ## _extra);	       \
944 	else								       \
945 		(einode)->xtime.tv_nsec = 0;				       \
946 } while (0)
947 
948 #define i_disk_version osd1.linux1.l_i_version
949 
950 #if defined(__KERNEL__) || defined(__linux__)
951 #define i_reserved1	osd1.linux1.l_i_reserved1
952 #define i_file_acl_high	osd2.linux2.l_i_file_acl_high
953 #define i_blocks_high	osd2.linux2.l_i_blocks_high
954 #define i_uid_low	i_uid
955 #define i_gid_low	i_gid
956 #define i_uid_high	osd2.linux2.l_i_uid_high
957 #define i_gid_high	osd2.linux2.l_i_gid_high
958 #define i_checksum_lo	osd2.linux2.l_i_checksum_lo
959 
960 #elif defined(__GNU__)
961 
962 #define i_translator	osd1.hurd1.h_i_translator
963 #define i_uid_high	osd2.hurd2.h_i_uid_high
964 #define i_gid_high	osd2.hurd2.h_i_gid_high
965 #define i_author	osd2.hurd2.h_i_author
966 
967 #elif defined(__masix__)
968 
969 #define i_reserved1	osd1.masix1.m_i_reserved1
970 #define i_file_acl_high	osd2.masix2.m_i_file_acl_high
971 #define i_reserved2	osd2.masix2.m_i_reserved2
972 
973 #endif /* defined(__KERNEL__) || defined(__linux__) */
974 
975 #include "extents_status.h"
976 #include "fast_commit.h"
977 
978 /*
979  * Lock subclasses for i_data_sem in the ext4_inode_info structure.
980  *
981  * These are needed to avoid lockdep false positives when we need to
982  * allocate blocks to the quota inode during ext4_map_blocks(), while
983  * holding i_data_sem for a normal (non-quota) inode.  Since we don't
984  * do quota tracking for the quota inode, this avoids deadlock (as
985  * well as infinite recursion, since it isn't turtles all the way
986  * down...)
987  *
988  *  I_DATA_SEM_NORMAL - Used for most inodes
989  *  I_DATA_SEM_OTHER  - Used by move_inode.c for the second normal inode
990  *			  where the second inode has larger inode number
991  *			  than the first
992  *  I_DATA_SEM_QUOTA  - Used for quota inodes only
993  */
994 enum {
995 	I_DATA_SEM_NORMAL = 0,
996 	I_DATA_SEM_OTHER,
997 	I_DATA_SEM_QUOTA,
998 };
999 
1000 
1001 /*
1002  * fourth extended file system inode data in memory
1003  */
1004 struct ext4_inode_info {
1005 	__le32	i_data[15];	/* unconverted */
1006 	__u32	i_dtime;
1007 	ext4_fsblk_t	i_file_acl;
1008 
1009 	/*
1010 	 * i_block_group is the number of the block group which contains
1011 	 * this file's inode.  Constant across the lifetime of the inode,
1012 	 * it is used for making block allocation decisions - we try to
1013 	 * place a file's data blocks near its inode block, and new inodes
1014 	 * near to their parent directory's inode.
1015 	 */
1016 	ext4_group_t	i_block_group;
1017 	ext4_lblk_t	i_dir_start_lookup;
1018 #if (BITS_PER_LONG < 64)
1019 	unsigned long	i_state_flags;		/* Dynamic state flags */
1020 #endif
1021 	unsigned long	i_flags;
1022 
1023 	/*
1024 	 * Extended attributes can be read independently of the main file
1025 	 * data. Taking i_mutex even when reading would cause contention
1026 	 * between readers of EAs and writers of regular file data, so
1027 	 * instead we synchronize on xattr_sem when reading or changing
1028 	 * EAs.
1029 	 */
1030 	struct rw_semaphore xattr_sem;
1031 
1032 	struct list_head i_orphan;	/* unlinked but open inodes */
1033 
1034 	/* Fast commit related info */
1035 
1036 	struct list_head i_fc_list;	/*
1037 					 * inodes that need fast commit
1038 					 * protected by sbi->s_fc_lock.
1039 					 */
1040 
1041 	/* Start of lblk range that needs to be committed in this fast commit */
1042 	ext4_lblk_t i_fc_lblk_start;
1043 
1044 	/* End of lblk range that needs to be committed in this fast commit */
1045 	ext4_lblk_t i_fc_lblk_len;
1046 
1047 	/* Number of ongoing updates on this inode */
1048 	atomic_t  i_fc_updates;
1049 
1050 	/* Fast commit wait queue for this inode */
1051 	wait_queue_head_t i_fc_wait;
1052 
1053 	/* Protect concurrent accesses on i_fc_lblk_start, i_fc_lblk_len */
1054 	struct mutex i_fc_lock;
1055 
1056 	/*
1057 	 * i_disksize keeps track of what the inode size is ON DISK, not
1058 	 * in memory.  During truncate, i_size is set to the new size by
1059 	 * the VFS prior to calling ext4_truncate(), but the filesystem won't
1060 	 * set i_disksize to 0 until the truncate is actually under way.
1061 	 *
1062 	 * The intent is that i_disksize always represents the blocks which
1063 	 * are used by this file.  This allows recovery to restart truncate
1064 	 * on orphans if we crash during truncate.  We actually write i_disksize
1065 	 * into the on-disk inode when writing inodes out, instead of i_size.
1066 	 *
1067 	 * The only time when i_disksize and i_size may be different is when
1068 	 * a truncate is in progress.  The only things which change i_disksize
1069 	 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
1070 	 */
1071 	loff_t	i_disksize;
1072 
1073 	/*
1074 	 * i_data_sem is for serialising ext4_truncate() against
1075 	 * ext4_getblock().  In the 2.4 ext2 design, great chunks of inode's
1076 	 * data tree are chopped off during truncate. We can't do that in
1077 	 * ext4 because whenever we perform intermediate commits during
1078 	 * truncate, the inode and all the metadata blocks *must* be in a
1079 	 * consistent state which allows truncation of the orphans to restart
1080 	 * during recovery.  Hence we must fix the get_block-vs-truncate race
1081 	 * by other means, so we have i_data_sem.
1082 	 */
1083 	struct rw_semaphore i_data_sem;
1084 	/*
1085 	 * i_mmap_sem is for serializing page faults with truncate / punch hole
1086 	 * operations. We have to make sure that new page cannot be faulted in
1087 	 * a section of the inode that is being punched. We cannot easily use
1088 	 * i_data_sem for this since we need protection for the whole punch
1089 	 * operation and i_data_sem ranks below transaction start so we have
1090 	 * to occasionally drop it.
1091 	 */
1092 	struct rw_semaphore i_mmap_sem;
1093 	struct inode vfs_inode;
1094 	struct jbd2_inode *jinode;
1095 
1096 	spinlock_t i_raw_lock;	/* protects updates to the raw inode */
1097 
1098 	/*
1099 	 * File creation time. Its function is same as that of
1100 	 * struct timespec64 i_{a,c,m}time in the generic inode.
1101 	 */
1102 	struct timespec64 i_crtime;
1103 
1104 	/* mballoc */
1105 	atomic_t i_prealloc_active;
1106 	struct list_head i_prealloc_list;
1107 	spinlock_t i_prealloc_lock;
1108 
1109 	/* extents status tree */
1110 	struct ext4_es_tree i_es_tree;
1111 	rwlock_t i_es_lock;
1112 	struct list_head i_es_list;
1113 	unsigned int i_es_all_nr;	/* protected by i_es_lock */
1114 	unsigned int i_es_shk_nr;	/* protected by i_es_lock */
1115 	ext4_lblk_t i_es_shrink_lblk;	/* Offset where we start searching for
1116 					   extents to shrink. Protected by
1117 					   i_es_lock  */
1118 
1119 	/* ialloc */
1120 	ext4_group_t	i_last_alloc_group;
1121 
1122 	/* allocation reservation info for delalloc */
1123 	/* In case of bigalloc, this refer to clusters rather than blocks */
1124 	unsigned int i_reserved_data_blocks;
1125 
1126 	/* pending cluster reservations for bigalloc file systems */
1127 	struct ext4_pending_tree i_pending_tree;
1128 
1129 	/* on-disk additional length */
1130 	__u16 i_extra_isize;
1131 
1132 	/* Indicate the inline data space. */
1133 	u16 i_inline_off;
1134 	u16 i_inline_size;
1135 
1136 #ifdef CONFIG_QUOTA
1137 	/* quota space reservation, managed internally by quota code */
1138 	qsize_t i_reserved_quota;
1139 #endif
1140 
1141 	/* Lock protecting lists below */
1142 	spinlock_t i_completed_io_lock;
1143 	/*
1144 	 * Completed IOs that need unwritten extents handling and have
1145 	 * transaction reserved
1146 	 */
1147 	struct list_head i_rsv_conversion_list;
1148 	struct work_struct i_rsv_conversion_work;
1149 	atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1150 
1151 	spinlock_t i_block_reservation_lock;
1152 
1153 	/*
1154 	 * Transactions that contain inode's metadata needed to complete
1155 	 * fsync and fdatasync, respectively.
1156 	 */
1157 	tid_t i_sync_tid;
1158 	tid_t i_datasync_tid;
1159 
1160 #ifdef CONFIG_QUOTA
1161 	struct dquot *i_dquot[MAXQUOTAS];
1162 #endif
1163 
1164 	/* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1165 	__u32 i_csum_seed;
1166 
1167 	kprojid_t i_projid;
1168 };
1169 
1170 /*
1171  * File system states
1172  */
1173 #define	EXT4_VALID_FS			0x0001	/* Unmounted cleanly */
1174 #define	EXT4_ERROR_FS			0x0002	/* Errors detected */
1175 #define	EXT4_ORPHAN_FS			0x0004	/* Orphans being recovered */
1176 #define EXT4_FC_REPLAY			0x0020	/* Fast commit replay ongoing */
1177 
1178 /*
1179  * Misc. filesystem flags
1180  */
1181 #define EXT2_FLAGS_SIGNED_HASH		0x0001  /* Signed dirhash in use */
1182 #define EXT2_FLAGS_UNSIGNED_HASH	0x0002  /* Unsigned dirhash in use */
1183 #define EXT2_FLAGS_TEST_FILESYS		0x0004	/* to test development code */
1184 
1185 /*
1186  * Mount flags set via mount options or defaults
1187  */
1188 #define EXT4_MOUNT_NO_MBCACHE		0x00001 /* Do not use mbcache */
1189 #define EXT4_MOUNT_GRPID		0x00004	/* Create files with directory's group */
1190 #define EXT4_MOUNT_DEBUG		0x00008	/* Some debugging messages */
1191 #define EXT4_MOUNT_ERRORS_CONT		0x00010	/* Continue on errors */
1192 #define EXT4_MOUNT_ERRORS_RO		0x00020	/* Remount fs ro on errors */
1193 #define EXT4_MOUNT_ERRORS_PANIC		0x00040	/* Panic on errors */
1194 #define EXT4_MOUNT_ERRORS_MASK		0x00070
1195 #define EXT4_MOUNT_MINIX_DF		0x00080	/* Mimics the Minix statfs */
1196 #define EXT4_MOUNT_NOLOAD		0x00100	/* Don't use existing journal*/
1197 #ifdef CONFIG_FS_DAX
1198 #define EXT4_MOUNT_DAX_ALWAYS		0x00200	/* Direct Access */
1199 #else
1200 #define EXT4_MOUNT_DAX_ALWAYS		0
1201 #endif
1202 #define EXT4_MOUNT_DATA_FLAGS		0x00C00	/* Mode for data writes: */
1203 #define EXT4_MOUNT_JOURNAL_DATA		0x00400	/* Write data to journal */
1204 #define EXT4_MOUNT_ORDERED_DATA		0x00800	/* Flush data before commit */
1205 #define EXT4_MOUNT_WRITEBACK_DATA	0x00C00	/* No data ordering */
1206 #define EXT4_MOUNT_UPDATE_JOURNAL	0x01000	/* Update the journal format */
1207 #define EXT4_MOUNT_NO_UID32		0x02000  /* Disable 32-bit UIDs */
1208 #define EXT4_MOUNT_XATTR_USER		0x04000	/* Extended user attributes */
1209 #define EXT4_MOUNT_POSIX_ACL		0x08000	/* POSIX Access Control Lists */
1210 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC	0x10000	/* No auto delalloc mapping */
1211 #define EXT4_MOUNT_BARRIER		0x20000 /* Use block barriers */
1212 #define EXT4_MOUNT_QUOTA		0x40000 /* Some quota option set */
1213 #define EXT4_MOUNT_USRQUOTA		0x80000 /* "old" user quota,
1214 						 * enable enforcement for hidden
1215 						 * quota files */
1216 #define EXT4_MOUNT_GRPQUOTA		0x100000 /* "old" group quota, enable
1217 						  * enforcement for hidden quota
1218 						  * files */
1219 #define EXT4_MOUNT_PRJQUOTA		0x200000 /* Enable project quota
1220 						  * enforcement */
1221 #define EXT4_MOUNT_DIOREAD_NOLOCK	0x400000 /* Enable support for dio read nolocking */
1222 #define EXT4_MOUNT_JOURNAL_CHECKSUM	0x800000 /* Journal checksums */
1223 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT	0x1000000 /* Journal Async Commit */
1224 #define EXT4_MOUNT_WARN_ON_ERROR	0x2000000 /* Trigger WARN_ON on error */
1225 #define EXT4_MOUNT_PREFETCH_BLOCK_BITMAPS 0x4000000
1226 #define EXT4_MOUNT_DELALLOC		0x8000000 /* Delalloc support */
1227 #define EXT4_MOUNT_DATA_ERR_ABORT	0x10000000 /* Abort on file data write */
1228 #define EXT4_MOUNT_BLOCK_VALIDITY	0x20000000 /* Block validity checking */
1229 #define EXT4_MOUNT_DISCARD		0x40000000 /* Issue DISCARD requests */
1230 #define EXT4_MOUNT_INIT_INODE_TABLE	0x80000000 /* Initialize uninitialized itables */
1231 
1232 /*
1233  * Mount flags set either automatically (could not be set by mount option)
1234  * based on per file system feature or property or in special cases such as
1235  * distinguishing between explicit mount option definition and default.
1236  */
1237 #define EXT4_MOUNT2_EXPLICIT_DELALLOC	0x00000001 /* User explicitly
1238 						      specified delalloc */
1239 #define EXT4_MOUNT2_STD_GROUP_SIZE	0x00000002 /* We have standard group
1240 						      size of blocksize * 8
1241 						      blocks */
1242 #define EXT4_MOUNT2_HURD_COMPAT		0x00000004 /* Support HURD-castrated
1243 						      file systems */
1244 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM	0x00000008 /* User explicitly
1245 						specified journal checksum */
1246 
1247 #define EXT4_MOUNT2_JOURNAL_FAST_COMMIT	0x00000010 /* Journal fast commit */
1248 #define EXT4_MOUNT2_DAX_NEVER		0x00000020 /* Do not allow Direct Access */
1249 #define EXT4_MOUNT2_DAX_INODE		0x00000040 /* For printing options only */
1250 
1251 
1252 #define clear_opt(sb, opt)		EXT4_SB(sb)->s_mount_opt &= \
1253 						~EXT4_MOUNT_##opt
1254 #define set_opt(sb, opt)		EXT4_SB(sb)->s_mount_opt |= \
1255 						EXT4_MOUNT_##opt
1256 #define test_opt(sb, opt)		(EXT4_SB(sb)->s_mount_opt & \
1257 					 EXT4_MOUNT_##opt)
1258 
1259 #define clear_opt2(sb, opt)		EXT4_SB(sb)->s_mount_opt2 &= \
1260 						~EXT4_MOUNT2_##opt
1261 #define set_opt2(sb, opt)		EXT4_SB(sb)->s_mount_opt2 |= \
1262 						EXT4_MOUNT2_##opt
1263 #define test_opt2(sb, opt)		(EXT4_SB(sb)->s_mount_opt2 & \
1264 					 EXT4_MOUNT2_##opt)
1265 
1266 #define ext4_test_and_set_bit		__test_and_set_bit_le
1267 #define ext4_set_bit			__set_bit_le
1268 #define ext4_set_bit_atomic		ext2_set_bit_atomic
1269 #define ext4_test_and_clear_bit		__test_and_clear_bit_le
1270 #define ext4_clear_bit			__clear_bit_le
1271 #define ext4_clear_bit_atomic		ext2_clear_bit_atomic
1272 #define ext4_test_bit			test_bit_le
1273 #define ext4_find_next_zero_bit		find_next_zero_bit_le
1274 #define ext4_find_next_bit		find_next_bit_le
1275 
1276 extern void ext4_set_bits(void *bm, int cur, int len);
1277 
1278 /*
1279  * Maximal mount counts between two filesystem checks
1280  */
1281 #define EXT4_DFL_MAX_MNT_COUNT		20	/* Allow 20 mounts */
1282 #define EXT4_DFL_CHECKINTERVAL		0	/* Don't use interval check */
1283 
1284 /*
1285  * Behaviour when detecting errors
1286  */
1287 #define EXT4_ERRORS_CONTINUE		1	/* Continue execution */
1288 #define EXT4_ERRORS_RO			2	/* Remount fs read-only */
1289 #define EXT4_ERRORS_PANIC		3	/* Panic */
1290 #define EXT4_ERRORS_DEFAULT		EXT4_ERRORS_CONTINUE
1291 
1292 /* Metadata checksum algorithm codes */
1293 #define EXT4_CRC32C_CHKSUM		1
1294 
1295 /*
1296  * Structure of the super block
1297  */
1298 struct ext4_super_block {
1299 /*00*/	__le32	s_inodes_count;		/* Inodes count */
1300 	__le32	s_blocks_count_lo;	/* Blocks count */
1301 	__le32	s_r_blocks_count_lo;	/* Reserved blocks count */
1302 	__le32	s_free_blocks_count_lo;	/* Free blocks count */
1303 /*10*/	__le32	s_free_inodes_count;	/* Free inodes count */
1304 	__le32	s_first_data_block;	/* First Data Block */
1305 	__le32	s_log_block_size;	/* Block size */
1306 	__le32	s_log_cluster_size;	/* Allocation cluster size */
1307 /*20*/	__le32	s_blocks_per_group;	/* # Blocks per group */
1308 	__le32	s_clusters_per_group;	/* # Clusters per group */
1309 	__le32	s_inodes_per_group;	/* # Inodes per group */
1310 	__le32	s_mtime;		/* Mount time */
1311 /*30*/	__le32	s_wtime;		/* Write time */
1312 	__le16	s_mnt_count;		/* Mount count */
1313 	__le16	s_max_mnt_count;	/* Maximal mount count */
1314 	__le16	s_magic;		/* Magic signature */
1315 	__le16	s_state;		/* File system state */
1316 	__le16	s_errors;		/* Behaviour when detecting errors */
1317 	__le16	s_minor_rev_level;	/* minor revision level */
1318 /*40*/	__le32	s_lastcheck;		/* time of last check */
1319 	__le32	s_checkinterval;	/* max. time between checks */
1320 	__le32	s_creator_os;		/* OS */
1321 	__le32	s_rev_level;		/* Revision level */
1322 /*50*/	__le16	s_def_resuid;		/* Default uid for reserved blocks */
1323 	__le16	s_def_resgid;		/* Default gid for reserved blocks */
1324 	/*
1325 	 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1326 	 *
1327 	 * Note: the difference between the compatible feature set and
1328 	 * the incompatible feature set is that if there is a bit set
1329 	 * in the incompatible feature set that the kernel doesn't
1330 	 * know about, it should refuse to mount the filesystem.
1331 	 *
1332 	 * e2fsck's requirements are more strict; if it doesn't know
1333 	 * about a feature in either the compatible or incompatible
1334 	 * feature set, it must abort and not try to meddle with
1335 	 * things it doesn't understand...
1336 	 */
1337 	__le32	s_first_ino;		/* First non-reserved inode */
1338 	__le16  s_inode_size;		/* size of inode structure */
1339 	__le16	s_block_group_nr;	/* block group # of this superblock */
1340 	__le32	s_feature_compat;	/* compatible feature set */
1341 /*60*/	__le32	s_feature_incompat;	/* incompatible feature set */
1342 	__le32	s_feature_ro_compat;	/* readonly-compatible feature set */
1343 /*68*/	__u8	s_uuid[16];		/* 128-bit uuid for volume */
1344 /*78*/	char	s_volume_name[16];	/* volume name */
1345 /*88*/	char	s_last_mounted[64] __nonstring;	/* directory where last mounted */
1346 /*C8*/	__le32	s_algorithm_usage_bitmap; /* For compression */
1347 	/*
1348 	 * Performance hints.  Directory preallocation should only
1349 	 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1350 	 */
1351 	__u8	s_prealloc_blocks;	/* Nr of blocks to try to preallocate*/
1352 	__u8	s_prealloc_dir_blocks;	/* Nr to preallocate for dirs */
1353 	__le16	s_reserved_gdt_blocks;	/* Per group desc for online growth */
1354 	/*
1355 	 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1356 	 */
1357 /*D0*/	__u8	s_journal_uuid[16];	/* uuid of journal superblock */
1358 /*E0*/	__le32	s_journal_inum;		/* inode number of journal file */
1359 	__le32	s_journal_dev;		/* device number of journal file */
1360 	__le32	s_last_orphan;		/* start of list of inodes to delete */
1361 	__le32	s_hash_seed[4];		/* HTREE hash seed */
1362 	__u8	s_def_hash_version;	/* Default hash version to use */
1363 	__u8	s_jnl_backup_type;
1364 	__le16  s_desc_size;		/* size of group descriptor */
1365 /*100*/	__le32	s_default_mount_opts;
1366 	__le32	s_first_meta_bg;	/* First metablock block group */
1367 	__le32	s_mkfs_time;		/* When the filesystem was created */
1368 	__le32	s_jnl_blocks[17];	/* Backup of the journal inode */
1369 	/* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1370 /*150*/	__le32	s_blocks_count_hi;	/* Blocks count */
1371 	__le32	s_r_blocks_count_hi;	/* Reserved blocks count */
1372 	__le32	s_free_blocks_count_hi;	/* Free blocks count */
1373 	__le16	s_min_extra_isize;	/* All inodes have at least # bytes */
1374 	__le16	s_want_extra_isize; 	/* New inodes should reserve # bytes */
1375 	__le32	s_flags;		/* Miscellaneous flags */
1376 	__le16  s_raid_stride;		/* RAID stride */
1377 	__le16  s_mmp_update_interval;  /* # seconds to wait in MMP checking */
1378 	__le64  s_mmp_block;            /* Block for multi-mount protection */
1379 	__le32  s_raid_stripe_width;    /* blocks on all data disks (N*stride)*/
1380 	__u8	s_log_groups_per_flex;  /* FLEX_BG group size */
1381 	__u8	s_checksum_type;	/* metadata checksum algorithm used */
1382 	__u8	s_encryption_level;	/* versioning level for encryption */
1383 	__u8	s_reserved_pad;		/* Padding to next 32bits */
1384 	__le64	s_kbytes_written;	/* nr of lifetime kilobytes written */
1385 	__le32	s_snapshot_inum;	/* Inode number of active snapshot */
1386 	__le32	s_snapshot_id;		/* sequential ID of active snapshot */
1387 	__le64	s_snapshot_r_blocks_count; /* reserved blocks for active
1388 					      snapshot's future use */
1389 	__le32	s_snapshot_list;	/* inode number of the head of the
1390 					   on-disk snapshot list */
1391 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1392 	__le32	s_error_count;		/* number of fs errors */
1393 	__le32	s_first_error_time;	/* first time an error happened */
1394 	__le32	s_first_error_ino;	/* inode involved in first error */
1395 	__le64	s_first_error_block;	/* block involved of first error */
1396 	__u8	s_first_error_func[32] __nonstring;	/* function where the error happened */
1397 	__le32	s_first_error_line;	/* line number where error happened */
1398 	__le32	s_last_error_time;	/* most recent time of an error */
1399 	__le32	s_last_error_ino;	/* inode involved in last error */
1400 	__le32	s_last_error_line;	/* line number where error happened */
1401 	__le64	s_last_error_block;	/* block involved of last error */
1402 	__u8	s_last_error_func[32] __nonstring;	/* function where the error happened */
1403 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1404 	__u8	s_mount_opts[64];
1405 	__le32	s_usr_quota_inum;	/* inode for tracking user quota */
1406 	__le32	s_grp_quota_inum;	/* inode for tracking group quota */
1407 	__le32	s_overhead_clusters;	/* overhead blocks/clusters in fs */
1408 	__le32	s_backup_bgs[2];	/* groups with sparse_super2 SBs */
1409 	__u8	s_encrypt_algos[4];	/* Encryption algorithms in use  */
1410 	__u8	s_encrypt_pw_salt[16];	/* Salt used for string2key algorithm */
1411 	__le32	s_lpf_ino;		/* Location of the lost+found inode */
1412 	__le32	s_prj_quota_inum;	/* inode for tracking project quota */
1413 	__le32	s_checksum_seed;	/* crc32c(uuid) if csum_seed set */
1414 	__u8	s_wtime_hi;
1415 	__u8	s_mtime_hi;
1416 	__u8	s_mkfs_time_hi;
1417 	__u8	s_lastcheck_hi;
1418 	__u8	s_first_error_time_hi;
1419 	__u8	s_last_error_time_hi;
1420 	__u8	s_first_error_errcode;
1421 	__u8    s_last_error_errcode;
1422 	__le16  s_encoding;		/* Filename charset encoding */
1423 	__le16  s_encoding_flags;	/* Filename charset encoding flags */
1424 	__le32	s_reserved[95];		/* Padding to the end of the block */
1425 	__le32	s_checksum;		/* crc32c(superblock) */
1426 };
1427 
1428 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1429 
1430 #ifdef __KERNEL__
1431 
1432 #ifdef CONFIG_FS_ENCRYPTION
1433 #define DUMMY_ENCRYPTION_ENABLED(sbi) ((sbi)->s_dummy_enc_policy.policy != NULL)
1434 #else
1435 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1436 #endif
1437 
1438 /* Number of quota types we support */
1439 #define EXT4_MAXQUOTAS 3
1440 
1441 #define EXT4_ENC_UTF8_12_1	1
1442 
1443 /*
1444  * fourth extended-fs super-block data in memory
1445  */
1446 struct ext4_sb_info {
1447 	unsigned long s_desc_size;	/* Size of a group descriptor in bytes */
1448 	unsigned long s_inodes_per_block;/* Number of inodes per block */
1449 	unsigned long s_blocks_per_group;/* Number of blocks in a group */
1450 	unsigned long s_clusters_per_group; /* Number of clusters in a group */
1451 	unsigned long s_inodes_per_group;/* Number of inodes in a group */
1452 	unsigned long s_itb_per_group;	/* Number of inode table blocks per group */
1453 	unsigned long s_gdb_count;	/* Number of group descriptor blocks */
1454 	unsigned long s_desc_per_block;	/* Number of group descriptors per block */
1455 	ext4_group_t s_groups_count;	/* Number of groups in the fs */
1456 	ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1457 	unsigned long s_overhead;  /* # of fs overhead clusters */
1458 	unsigned int s_cluster_ratio;	/* Number of blocks per cluster */
1459 	unsigned int s_cluster_bits;	/* log2 of s_cluster_ratio */
1460 	loff_t s_bitmap_maxbytes;	/* max bytes for bitmap files */
1461 	struct buffer_head * s_sbh;	/* Buffer containing the super block */
1462 	struct ext4_super_block *s_es;	/* Pointer to the super block in the buffer */
1463 	struct buffer_head * __rcu *s_group_desc;
1464 	unsigned int s_mount_opt;
1465 	unsigned int s_mount_opt2;
1466 	unsigned long s_mount_flags;
1467 	unsigned int s_def_mount_opt;
1468 	ext4_fsblk_t s_sb_block;
1469 	atomic64_t s_resv_clusters;
1470 	kuid_t s_resuid;
1471 	kgid_t s_resgid;
1472 	unsigned short s_mount_state;
1473 	unsigned short s_pad;
1474 	int s_addr_per_block_bits;
1475 	int s_desc_per_block_bits;
1476 	int s_inode_size;
1477 	int s_first_ino;
1478 	unsigned int s_inode_readahead_blks;
1479 	unsigned int s_inode_goal;
1480 	u32 s_hash_seed[4];
1481 	int s_def_hash_version;
1482 	int s_hash_unsigned;	/* 3 if hash should be signed, 0 if not */
1483 	struct percpu_counter s_freeclusters_counter;
1484 	struct percpu_counter s_freeinodes_counter;
1485 	struct percpu_counter s_dirs_counter;
1486 	struct percpu_counter s_dirtyclusters_counter;
1487 	struct percpu_counter s_sra_exceeded_retry_limit;
1488 	struct blockgroup_lock *s_blockgroup_lock;
1489 	struct proc_dir_entry *s_proc;
1490 	struct kobject s_kobj;
1491 	struct completion s_kobj_unregister;
1492 	struct super_block *s_sb;
1493 
1494 	/* Journaling */
1495 	struct journal_s *s_journal;
1496 	struct list_head s_orphan;
1497 	struct mutex s_orphan_lock;
1498 	unsigned long s_ext4_flags;		/* Ext4 superblock flags */
1499 	unsigned long s_commit_interval;
1500 	u32 s_max_batch_time;
1501 	u32 s_min_batch_time;
1502 	struct block_device *s_journal_bdev;
1503 #ifdef CONFIG_QUOTA
1504 	/* Names of quota files with journalled quota */
1505 	char __rcu *s_qf_names[EXT4_MAXQUOTAS];
1506 	int s_jquota_fmt;			/* Format of quota to use */
1507 #endif
1508 	unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1509 	struct ext4_system_blocks __rcu *s_system_blks;
1510 
1511 #ifdef EXTENTS_STATS
1512 	/* ext4 extents stats */
1513 	unsigned long s_ext_min;
1514 	unsigned long s_ext_max;
1515 	unsigned long s_depth_max;
1516 	spinlock_t s_ext_stats_lock;
1517 	unsigned long s_ext_blocks;
1518 	unsigned long s_ext_extents;
1519 #endif
1520 
1521 	/* for buddy allocator */
1522 	struct ext4_group_info ** __rcu *s_group_info;
1523 	struct inode *s_buddy_cache;
1524 	spinlock_t s_md_lock;
1525 	unsigned short *s_mb_offsets;
1526 	unsigned int *s_mb_maxs;
1527 	unsigned int s_group_info_size;
1528 	unsigned int s_mb_free_pending;
1529 	struct list_head s_freed_data_list;	/* List of blocks to be freed
1530 						   after commit completed */
1531 
1532 	/* tunables */
1533 	unsigned long s_stripe;
1534 	unsigned int s_mb_stream_request;
1535 	unsigned int s_mb_max_to_scan;
1536 	unsigned int s_mb_min_to_scan;
1537 	unsigned int s_mb_stats;
1538 	unsigned int s_mb_order2_reqs;
1539 	unsigned int s_mb_group_prealloc;
1540 	unsigned int s_mb_max_inode_prealloc;
1541 	unsigned int s_max_dir_size_kb;
1542 	/* where last allocation was done - for stream allocation */
1543 	unsigned long s_mb_last_group;
1544 	unsigned long s_mb_last_start;
1545 	unsigned int s_mb_prefetch;
1546 	unsigned int s_mb_prefetch_limit;
1547 
1548 	/* stats for buddy allocator */
1549 	atomic_t s_bal_reqs;	/* number of reqs with len > 1 */
1550 	atomic_t s_bal_success;	/* we found long enough chunks */
1551 	atomic_t s_bal_allocated;	/* in blocks */
1552 	atomic_t s_bal_ex_scanned;	/* total extents scanned */
1553 	atomic_t s_bal_goals;	/* goal hits */
1554 	atomic_t s_bal_breaks;	/* too long searches */
1555 	atomic_t s_bal_2orders;	/* 2^order hits */
1556 	spinlock_t s_bal_lock;
1557 	unsigned long s_mb_buddies_generated;
1558 	unsigned long long s_mb_generation_time;
1559 	atomic_t s_mb_lost_chunks;
1560 	atomic_t s_mb_preallocated;
1561 	atomic_t s_mb_discarded;
1562 	atomic_t s_lock_busy;
1563 
1564 	/* locality groups */
1565 	struct ext4_locality_group __percpu *s_locality_groups;
1566 
1567 	/* for write statistics */
1568 	unsigned long s_sectors_written_start;
1569 	u64 s_kbytes_written;
1570 
1571 	/* the size of zero-out chunk */
1572 	unsigned int s_extent_max_zeroout_kb;
1573 
1574 	unsigned int s_log_groups_per_flex;
1575 	struct flex_groups * __rcu *s_flex_groups;
1576 	ext4_group_t s_flex_groups_allocated;
1577 
1578 	/* workqueue for reserved extent conversions (buffered io) */
1579 	struct workqueue_struct *rsv_conversion_wq;
1580 
1581 	/* timer for periodic error stats printing */
1582 	struct timer_list s_err_report;
1583 
1584 	/* Lazy inode table initialization info */
1585 	struct ext4_li_request *s_li_request;
1586 	/* Wait multiplier for lazy initialization thread */
1587 	unsigned int s_li_wait_mult;
1588 
1589 	/* Kernel thread for multiple mount protection */
1590 	struct task_struct *s_mmp_tsk;
1591 
1592 	/* record the last minlen when FITRIM is called. */
1593 	atomic_t s_last_trim_minblks;
1594 
1595 	/* Reference to checksum algorithm driver via cryptoapi */
1596 	struct crypto_shash *s_chksum_driver;
1597 
1598 	/* Precomputed FS UUID checksum for seeding other checksums */
1599 	__u32 s_csum_seed;
1600 
1601 	/* Reclaim extents from extent status tree */
1602 	struct shrinker s_es_shrinker;
1603 	struct list_head s_es_list;	/* List of inodes with reclaimable extents */
1604 	long s_es_nr_inode;
1605 	struct ext4_es_stats s_es_stats;
1606 	struct mb_cache *s_ea_block_cache;
1607 	struct mb_cache *s_ea_inode_cache;
1608 	spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1609 
1610 	/* Ratelimit ext4 messages. */
1611 	struct ratelimit_state s_err_ratelimit_state;
1612 	struct ratelimit_state s_warning_ratelimit_state;
1613 	struct ratelimit_state s_msg_ratelimit_state;
1614 	atomic_t s_warning_count;
1615 	atomic_t s_msg_count;
1616 
1617 	/* Encryption policy for '-o test_dummy_encryption' */
1618 	struct fscrypt_dummy_policy s_dummy_enc_policy;
1619 
1620 	/*
1621 	 * Barrier between writepages ops and changing any inode's JOURNAL_DATA
1622 	 * or EXTENTS flag.
1623 	 */
1624 	struct percpu_rw_semaphore s_writepages_rwsem;
1625 	struct dax_device *s_daxdev;
1626 #ifdef CONFIG_EXT4_DEBUG
1627 	unsigned long s_simulate_fail;
1628 #endif
1629 	/* Record the errseq of the backing block device */
1630 	errseq_t s_bdev_wb_err;
1631 	spinlock_t s_bdev_wb_lock;
1632 
1633 	/* Information about errors that happened during this mount */
1634 	spinlock_t s_error_lock;
1635 	int s_add_error_count;
1636 	int s_first_error_code;
1637 	__u32 s_first_error_line;
1638 	__u32 s_first_error_ino;
1639 	__u64 s_first_error_block;
1640 	const char *s_first_error_func;
1641 	time64_t s_first_error_time;
1642 	int s_last_error_code;
1643 	__u32 s_last_error_line;
1644 	__u32 s_last_error_ino;
1645 	__u64 s_last_error_block;
1646 	const char *s_last_error_func;
1647 	time64_t s_last_error_time;
1648 	/*
1649 	 * If we are in a context where we cannot update error information in
1650 	 * the on-disk superblock, we queue this work to do it.
1651 	 */
1652 	struct work_struct s_error_work;
1653 
1654 	/* Ext4 fast commit stuff */
1655 	atomic_t s_fc_subtid;
1656 	atomic_t s_fc_ineligible_updates;
1657 	/*
1658 	 * After commit starts, the main queue gets locked, and the further
1659 	 * updates get added in the staging queue.
1660 	 */
1661 #define FC_Q_MAIN	0
1662 #define FC_Q_STAGING	1
1663 	struct list_head s_fc_q[2];	/* Inodes staged for fast commit
1664 					 * that have data changes in them.
1665 					 */
1666 	struct list_head s_fc_dentry_q[2];	/* directory entry updates */
1667 	unsigned int s_fc_bytes;
1668 	/*
1669 	 * Main fast commit lock. This lock protects accesses to the
1670 	 * following fields:
1671 	 * ei->i_fc_list, s_fc_dentry_q, s_fc_q, s_fc_bytes, s_fc_bh.
1672 	 */
1673 	spinlock_t s_fc_lock;
1674 	struct buffer_head *s_fc_bh;
1675 	struct ext4_fc_stats s_fc_stats;
1676 	u64 s_fc_avg_commit_time;
1677 #ifdef CONFIG_EXT4_DEBUG
1678 	int s_fc_debug_max_replay;
1679 #endif
1680 	struct ext4_fc_replay_state s_fc_replay_state;
1681 };
1682 
1683 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1684 {
1685 	return sb->s_fs_info;
1686 }
1687 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1688 {
1689 	return container_of(inode, struct ext4_inode_info, vfs_inode);
1690 }
1691 
1692 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1693 {
1694 	return ino == EXT4_ROOT_INO ||
1695 		(ino >= EXT4_FIRST_INO(sb) &&
1696 		 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1697 }
1698 
1699 /*
1700  * Returns: sbi->field[index]
1701  * Used to access an array element from the following sbi fields which require
1702  * rcu protection to avoid dereferencing an invalid pointer due to reassignment
1703  * - s_group_desc
1704  * - s_group_info
1705  * - s_flex_group
1706  */
1707 #define sbi_array_rcu_deref(sbi, field, index)				   \
1708 ({									   \
1709 	typeof(*((sbi)->field)) _v;					   \
1710 	rcu_read_lock();						   \
1711 	_v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index];	   \
1712 	rcu_read_unlock();						   \
1713 	_v;								   \
1714 })
1715 
1716 /*
1717  * run-time mount flags
1718  */
1719 enum {
1720 	EXT4_MF_MNTDIR_SAMPLED,
1721 	EXT4_MF_FS_ABORTED,	/* Fatal error detected */
1722 	EXT4_MF_FC_INELIGIBLE,	/* Fast commit ineligible */
1723 	EXT4_MF_FC_COMMITTING	/* File system underoing a fast
1724 				 * commit.
1725 				 */
1726 };
1727 
1728 static inline void ext4_set_mount_flag(struct super_block *sb, int bit)
1729 {
1730 	set_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1731 }
1732 
1733 static inline void ext4_clear_mount_flag(struct super_block *sb, int bit)
1734 {
1735 	clear_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1736 }
1737 
1738 static inline int ext4_test_mount_flag(struct super_block *sb, int bit)
1739 {
1740 	return test_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1741 }
1742 
1743 
1744 /*
1745  * Simulate_fail codes
1746  */
1747 #define EXT4_SIM_BBITMAP_EIO	1
1748 #define EXT4_SIM_BBITMAP_CRC	2
1749 #define EXT4_SIM_IBITMAP_EIO	3
1750 #define EXT4_SIM_IBITMAP_CRC	4
1751 #define EXT4_SIM_INODE_EIO	5
1752 #define EXT4_SIM_INODE_CRC	6
1753 #define EXT4_SIM_DIRBLOCK_EIO	7
1754 #define EXT4_SIM_DIRBLOCK_CRC	8
1755 
1756 static inline bool ext4_simulate_fail(struct super_block *sb,
1757 				     unsigned long code)
1758 {
1759 #ifdef CONFIG_EXT4_DEBUG
1760 	struct ext4_sb_info *sbi = EXT4_SB(sb);
1761 
1762 	if (unlikely(sbi->s_simulate_fail == code)) {
1763 		sbi->s_simulate_fail = 0;
1764 		return true;
1765 	}
1766 #endif
1767 	return false;
1768 }
1769 
1770 static inline void ext4_simulate_fail_bh(struct super_block *sb,
1771 					 struct buffer_head *bh,
1772 					 unsigned long code)
1773 {
1774 	if (!IS_ERR(bh) && ext4_simulate_fail(sb, code))
1775 		clear_buffer_uptodate(bh);
1776 }
1777 
1778 /*
1779  * Error number codes for s_{first,last}_error_errno
1780  *
1781  * Linux errno numbers are architecture specific, so we need to translate
1782  * them into something which is architecture independent.   We don't define
1783  * codes for all errno's; just the ones which are most likely to be the cause
1784  * of an ext4_error() call.
1785  */
1786 #define EXT4_ERR_UNKNOWN	 1
1787 #define EXT4_ERR_EIO		 2
1788 #define EXT4_ERR_ENOMEM		 3
1789 #define EXT4_ERR_EFSBADCRC	 4
1790 #define EXT4_ERR_EFSCORRUPTED	 5
1791 #define EXT4_ERR_ENOSPC		 6
1792 #define EXT4_ERR_ENOKEY		 7
1793 #define EXT4_ERR_EROFS		 8
1794 #define EXT4_ERR_EFBIG		 9
1795 #define EXT4_ERR_EEXIST		10
1796 #define EXT4_ERR_ERANGE		11
1797 #define EXT4_ERR_EOVERFLOW	12
1798 #define EXT4_ERR_EBUSY		13
1799 #define EXT4_ERR_ENOTDIR	14
1800 #define EXT4_ERR_ENOTEMPTY	15
1801 #define EXT4_ERR_ESHUTDOWN	16
1802 #define EXT4_ERR_EFAULT		17
1803 
1804 /*
1805  * Inode dynamic state flags
1806  */
1807 enum {
1808 	EXT4_STATE_JDATA,		/* journaled data exists */
1809 	EXT4_STATE_NEW,			/* inode is newly created */
1810 	EXT4_STATE_XATTR,		/* has in-inode xattrs */
1811 	EXT4_STATE_NO_EXPAND,		/* No space for expansion */
1812 	EXT4_STATE_DA_ALLOC_CLOSE,	/* Alloc DA blks on close */
1813 	EXT4_STATE_EXT_MIGRATE,		/* Inode is migrating */
1814 	EXT4_STATE_NEWENTRY,		/* File just added to dir */
1815 	EXT4_STATE_MAY_INLINE_DATA,	/* may have in-inode data */
1816 	EXT4_STATE_EXT_PRECACHED,	/* extents have been precached */
1817 	EXT4_STATE_LUSTRE_EA_INODE,	/* Lustre-style ea_inode */
1818 	EXT4_STATE_VERITY_IN_PROGRESS,	/* building fs-verity Merkle tree */
1819 	EXT4_STATE_FC_COMMITTING,	/* Fast commit ongoing */
1820 };
1821 
1822 #define EXT4_INODE_BIT_FNS(name, field, offset)				\
1823 static inline int ext4_test_inode_##name(struct inode *inode, int bit)	\
1824 {									\
1825 	return test_bit(bit + (offset), &EXT4_I(inode)->i_##field);	\
1826 }									\
1827 static inline void ext4_set_inode_##name(struct inode *inode, int bit)	\
1828 {									\
1829 	set_bit(bit + (offset), &EXT4_I(inode)->i_##field);		\
1830 }									\
1831 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1832 {									\
1833 	clear_bit(bit + (offset), &EXT4_I(inode)->i_##field);		\
1834 }
1835 
1836 /* Add these declarations here only so that these functions can be
1837  * found by name.  Otherwise, they are very hard to locate. */
1838 static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1839 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1840 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1841 EXT4_INODE_BIT_FNS(flag, flags, 0)
1842 
1843 /* Add these declarations here only so that these functions can be
1844  * found by name.  Otherwise, they are very hard to locate. */
1845 static inline int ext4_test_inode_state(struct inode *inode, int bit);
1846 static inline void ext4_set_inode_state(struct inode *inode, int bit);
1847 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1848 #if (BITS_PER_LONG < 64)
1849 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1850 
1851 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1852 {
1853 	(ei)->i_state_flags = 0;
1854 }
1855 #else
1856 EXT4_INODE_BIT_FNS(state, flags, 32)
1857 
1858 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1859 {
1860 	/* We depend on the fact that callers will set i_flags */
1861 }
1862 #endif
1863 #else
1864 /* Assume that user mode programs are passing in an ext4fs superblock, not
1865  * a kernel struct super_block.  This will allow us to call the feature-test
1866  * macros from user land. */
1867 #define EXT4_SB(sb)	(sb)
1868 #endif
1869 
1870 static inline bool ext4_verity_in_progress(struct inode *inode)
1871 {
1872 	return IS_ENABLED(CONFIG_FS_VERITY) &&
1873 	       ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS);
1874 }
1875 
1876 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1877 
1878 /*
1879  * Codes for operating systems
1880  */
1881 #define EXT4_OS_LINUX		0
1882 #define EXT4_OS_HURD		1
1883 #define EXT4_OS_MASIX		2
1884 #define EXT4_OS_FREEBSD		3
1885 #define EXT4_OS_LITES		4
1886 
1887 /*
1888  * Revision levels
1889  */
1890 #define EXT4_GOOD_OLD_REV	0	/* The good old (original) format */
1891 #define EXT4_DYNAMIC_REV	1	/* V2 format w/ dynamic inode sizes */
1892 
1893 #define EXT4_MAX_SUPP_REV	EXT4_DYNAMIC_REV
1894 
1895 #define EXT4_GOOD_OLD_INODE_SIZE 128
1896 
1897 #define EXT4_EXTRA_TIMESTAMP_MAX	(((s64)1 << 34) - 1  + S32_MIN)
1898 #define EXT4_NON_EXTRA_TIMESTAMP_MAX	S32_MAX
1899 #define EXT4_TIMESTAMP_MIN		S32_MIN
1900 
1901 /*
1902  * Feature set definitions
1903  */
1904 
1905 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC	0x0001
1906 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES	0x0002
1907 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL		0x0004
1908 #define EXT4_FEATURE_COMPAT_EXT_ATTR		0x0008
1909 #define EXT4_FEATURE_COMPAT_RESIZE_INODE	0x0010
1910 #define EXT4_FEATURE_COMPAT_DIR_INDEX		0x0020
1911 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2	0x0200
1912 /*
1913  * The reason why "FAST_COMMIT" is a compat feature is that, FS becomes
1914  * incompatible only if fast commit blocks are present in the FS. Since we
1915  * clear the journal (and thus the fast commit blocks), we don't mark FS as
1916  * incompatible. We also have a JBD2 incompat feature, which gets set when
1917  * there are fast commit blocks present in the journal.
1918  */
1919 #define EXT4_FEATURE_COMPAT_FAST_COMMIT		0x0400
1920 #define EXT4_FEATURE_COMPAT_STABLE_INODES	0x0800
1921 
1922 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER	0x0001
1923 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE	0x0002
1924 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR	0x0004
1925 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE        0x0008
1926 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM		0x0010
1927 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK	0x0020
1928 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE	0x0040
1929 #define EXT4_FEATURE_RO_COMPAT_QUOTA		0x0100
1930 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC		0x0200
1931 /*
1932  * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM).  When
1933  * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1934  * all other data structures' checksums.  However, the METADATA_CSUM and
1935  * GDT_CSUM bits are mutually exclusive.
1936  */
1937 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM	0x0400
1938 #define EXT4_FEATURE_RO_COMPAT_READONLY		0x1000
1939 #define EXT4_FEATURE_RO_COMPAT_PROJECT		0x2000
1940 #define EXT4_FEATURE_RO_COMPAT_VERITY		0x8000
1941 
1942 #define EXT4_FEATURE_INCOMPAT_COMPRESSION	0x0001
1943 #define EXT4_FEATURE_INCOMPAT_FILETYPE		0x0002
1944 #define EXT4_FEATURE_INCOMPAT_RECOVER		0x0004 /* Needs recovery */
1945 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV	0x0008 /* Journal device */
1946 #define EXT4_FEATURE_INCOMPAT_META_BG		0x0010
1947 #define EXT4_FEATURE_INCOMPAT_EXTENTS		0x0040 /* extents support */
1948 #define EXT4_FEATURE_INCOMPAT_64BIT		0x0080
1949 #define EXT4_FEATURE_INCOMPAT_MMP               0x0100
1950 #define EXT4_FEATURE_INCOMPAT_FLEX_BG		0x0200
1951 #define EXT4_FEATURE_INCOMPAT_EA_INODE		0x0400 /* EA in inode */
1952 #define EXT4_FEATURE_INCOMPAT_DIRDATA		0x1000 /* data in dirent */
1953 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED		0x2000
1954 #define EXT4_FEATURE_INCOMPAT_LARGEDIR		0x4000 /* >2GB or 3-lvl htree */
1955 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA	0x8000 /* data in inode */
1956 #define EXT4_FEATURE_INCOMPAT_ENCRYPT		0x10000
1957 #define EXT4_FEATURE_INCOMPAT_CASEFOLD		0x20000
1958 
1959 extern void ext4_update_dynamic_rev(struct super_block *sb);
1960 
1961 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
1962 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1963 { \
1964 	return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1965 		cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
1966 } \
1967 static inline void ext4_set_feature_##name(struct super_block *sb) \
1968 { \
1969 	ext4_update_dynamic_rev(sb); \
1970 	EXT4_SB(sb)->s_es->s_feature_compat |= \
1971 		cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1972 } \
1973 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1974 { \
1975 	EXT4_SB(sb)->s_es->s_feature_compat &= \
1976 		~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1977 }
1978 
1979 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
1980 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1981 { \
1982 	return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1983 		cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
1984 } \
1985 static inline void ext4_set_feature_##name(struct super_block *sb) \
1986 { \
1987 	ext4_update_dynamic_rev(sb); \
1988 	EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
1989 		cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1990 } \
1991 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1992 { \
1993 	EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
1994 		~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1995 }
1996 
1997 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
1998 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1999 { \
2000 	return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
2001 		cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
2002 } \
2003 static inline void ext4_set_feature_##name(struct super_block *sb) \
2004 { \
2005 	ext4_update_dynamic_rev(sb); \
2006 	EXT4_SB(sb)->s_es->s_feature_incompat |= \
2007 		cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
2008 } \
2009 static inline void ext4_clear_feature_##name(struct super_block *sb) \
2010 { \
2011 	EXT4_SB(sb)->s_es->s_feature_incompat &= \
2012 		~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
2013 }
2014 
2015 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc,		DIR_PREALLOC)
2016 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes,	IMAGIC_INODES)
2017 EXT4_FEATURE_COMPAT_FUNCS(journal,		HAS_JOURNAL)
2018 EXT4_FEATURE_COMPAT_FUNCS(xattr,		EXT_ATTR)
2019 EXT4_FEATURE_COMPAT_FUNCS(resize_inode,		RESIZE_INODE)
2020 EXT4_FEATURE_COMPAT_FUNCS(dir_index,		DIR_INDEX)
2021 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2,	SPARSE_SUPER2)
2022 EXT4_FEATURE_COMPAT_FUNCS(fast_commit,		FAST_COMMIT)
2023 EXT4_FEATURE_COMPAT_FUNCS(stable_inodes,	STABLE_INODES)
2024 
2025 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super,	SPARSE_SUPER)
2026 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file,	LARGE_FILE)
2027 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir,		BTREE_DIR)
2028 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file,		HUGE_FILE)
2029 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum,		GDT_CSUM)
2030 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink,		DIR_NLINK)
2031 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize,	EXTRA_ISIZE)
2032 EXT4_FEATURE_RO_COMPAT_FUNCS(quota,		QUOTA)
2033 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc,		BIGALLOC)
2034 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum,	METADATA_CSUM)
2035 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly,		READONLY)
2036 EXT4_FEATURE_RO_COMPAT_FUNCS(project,		PROJECT)
2037 EXT4_FEATURE_RO_COMPAT_FUNCS(verity,		VERITY)
2038 
2039 EXT4_FEATURE_INCOMPAT_FUNCS(compression,	COMPRESSION)
2040 EXT4_FEATURE_INCOMPAT_FUNCS(filetype,		FILETYPE)
2041 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery,	RECOVER)
2042 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev,	JOURNAL_DEV)
2043 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg,		META_BG)
2044 EXT4_FEATURE_INCOMPAT_FUNCS(extents,		EXTENTS)
2045 EXT4_FEATURE_INCOMPAT_FUNCS(64bit,		64BIT)
2046 EXT4_FEATURE_INCOMPAT_FUNCS(mmp,		MMP)
2047 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg,		FLEX_BG)
2048 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode,		EA_INODE)
2049 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata,		DIRDATA)
2050 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed,		CSUM_SEED)
2051 EXT4_FEATURE_INCOMPAT_FUNCS(largedir,		LARGEDIR)
2052 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data,	INLINE_DATA)
2053 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt,		ENCRYPT)
2054 EXT4_FEATURE_INCOMPAT_FUNCS(casefold,		CASEFOLD)
2055 
2056 #define EXT2_FEATURE_COMPAT_SUPP	EXT4_FEATURE_COMPAT_EXT_ATTR
2057 #define EXT2_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
2058 					 EXT4_FEATURE_INCOMPAT_META_BG)
2059 #define EXT2_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2060 					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2061 					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
2062 
2063 #define EXT3_FEATURE_COMPAT_SUPP	EXT4_FEATURE_COMPAT_EXT_ATTR
2064 #define EXT3_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
2065 					 EXT4_FEATURE_INCOMPAT_RECOVER| \
2066 					 EXT4_FEATURE_INCOMPAT_META_BG)
2067 #define EXT3_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2068 					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2069 					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
2070 
2071 #define EXT4_FEATURE_COMPAT_SUPP	EXT4_FEATURE_COMPAT_EXT_ATTR
2072 #define EXT4_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
2073 					 EXT4_FEATURE_INCOMPAT_RECOVER| \
2074 					 EXT4_FEATURE_INCOMPAT_META_BG| \
2075 					 EXT4_FEATURE_INCOMPAT_EXTENTS| \
2076 					 EXT4_FEATURE_INCOMPAT_64BIT| \
2077 					 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
2078 					 EXT4_FEATURE_INCOMPAT_EA_INODE| \
2079 					 EXT4_FEATURE_INCOMPAT_MMP | \
2080 					 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
2081 					 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
2082 					 EXT4_FEATURE_INCOMPAT_CASEFOLD | \
2083 					 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \
2084 					 EXT4_FEATURE_INCOMPAT_LARGEDIR)
2085 #define EXT4_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2086 					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2087 					 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
2088 					 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
2089 					 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
2090 					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
2091 					 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
2092 					 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
2093 					 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
2094 					 EXT4_FEATURE_RO_COMPAT_QUOTA |\
2095 					 EXT4_FEATURE_RO_COMPAT_PROJECT |\
2096 					 EXT4_FEATURE_RO_COMPAT_VERITY)
2097 
2098 #define EXTN_FEATURE_FUNCS(ver) \
2099 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
2100 { \
2101 	return ((EXT4_SB(sb)->s_es->s_feature_compat & \
2102 		cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
2103 } \
2104 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
2105 { \
2106 	return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
2107 		cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
2108 } \
2109 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
2110 { \
2111 	return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
2112 		cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
2113 }
2114 
2115 EXTN_FEATURE_FUNCS(2)
2116 EXTN_FEATURE_FUNCS(3)
2117 EXTN_FEATURE_FUNCS(4)
2118 
2119 static inline bool ext4_has_compat_features(struct super_block *sb)
2120 {
2121 	return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
2122 }
2123 static inline bool ext4_has_ro_compat_features(struct super_block *sb)
2124 {
2125 	return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
2126 }
2127 static inline bool ext4_has_incompat_features(struct super_block *sb)
2128 {
2129 	return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
2130 }
2131 
2132 /*
2133  * Superblock flags
2134  */
2135 #define EXT4_FLAGS_RESIZING	0
2136 #define EXT4_FLAGS_SHUTDOWN	1
2137 #define EXT4_FLAGS_BDEV_IS_DAX	2
2138 
2139 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi)
2140 {
2141 	return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags);
2142 }
2143 
2144 
2145 /*
2146  * Default values for user and/or group using reserved blocks
2147  */
2148 #define	EXT4_DEF_RESUID		0
2149 #define	EXT4_DEF_RESGID		0
2150 
2151 /*
2152  * Default project ID
2153  */
2154 #define	EXT4_DEF_PROJID		0
2155 
2156 #define EXT4_DEF_INODE_READAHEAD_BLKS	32
2157 
2158 /*
2159  * Default mount options
2160  */
2161 #define EXT4_DEFM_DEBUG		0x0001
2162 #define EXT4_DEFM_BSDGROUPS	0x0002
2163 #define EXT4_DEFM_XATTR_USER	0x0004
2164 #define EXT4_DEFM_ACL		0x0008
2165 #define EXT4_DEFM_UID16		0x0010
2166 #define EXT4_DEFM_JMODE		0x0060
2167 #define EXT4_DEFM_JMODE_DATA	0x0020
2168 #define EXT4_DEFM_JMODE_ORDERED	0x0040
2169 #define EXT4_DEFM_JMODE_WBACK	0x0060
2170 #define EXT4_DEFM_NOBARRIER	0x0100
2171 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
2172 #define EXT4_DEFM_DISCARD	0x0400
2173 #define EXT4_DEFM_NODELALLOC	0x0800
2174 
2175 /*
2176  * Default journal batch times
2177  */
2178 #define EXT4_DEF_MIN_BATCH_TIME	0
2179 #define EXT4_DEF_MAX_BATCH_TIME	15000 /* 15ms */
2180 
2181 /*
2182  * Minimum number of groups in a flexgroup before we separate out
2183  * directories into the first block group of a flexgroup
2184  */
2185 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME	4
2186 
2187 /*
2188  * Structure of a directory entry
2189  */
2190 #define EXT4_NAME_LEN 255
2191 
2192 struct ext4_dir_entry {
2193 	__le32	inode;			/* Inode number */
2194 	__le16	rec_len;		/* Directory entry length */
2195 	__le16	name_len;		/* Name length */
2196 	char	name[EXT4_NAME_LEN];	/* File name */
2197 };
2198 
2199 /*
2200  * The new version of the directory entry.  Since EXT4 structures are
2201  * stored in intel byte order, and the name_len field could never be
2202  * bigger than 255 chars, it's safe to reclaim the extra byte for the
2203  * file_type field.
2204  */
2205 struct ext4_dir_entry_2 {
2206 	__le32	inode;			/* Inode number */
2207 	__le16	rec_len;		/* Directory entry length */
2208 	__u8	name_len;		/* Name length */
2209 	__u8	file_type;		/* See file type macros EXT4_FT_* below */
2210 	char	name[EXT4_NAME_LEN];	/* File name */
2211 };
2212 
2213 /*
2214  * This is a bogus directory entry at the end of each leaf block that
2215  * records checksums.
2216  */
2217 struct ext4_dir_entry_tail {
2218 	__le32	det_reserved_zero1;	/* Pretend to be unused */
2219 	__le16	det_rec_len;		/* 12 */
2220 	__u8	det_reserved_zero2;	/* Zero name length */
2221 	__u8	det_reserved_ft;	/* 0xDE, fake file type */
2222 	__le32	det_checksum;		/* crc32c(uuid+inum+dirblock) */
2223 };
2224 
2225 #define EXT4_DIRENT_TAIL(block, blocksize) \
2226 	((struct ext4_dir_entry_tail *)(((void *)(block)) + \
2227 					((blocksize) - \
2228 					 sizeof(struct ext4_dir_entry_tail))))
2229 
2230 /*
2231  * Ext4 directory file types.  Only the low 3 bits are used.  The
2232  * other bits are reserved for now.
2233  */
2234 #define EXT4_FT_UNKNOWN		0
2235 #define EXT4_FT_REG_FILE	1
2236 #define EXT4_FT_DIR		2
2237 #define EXT4_FT_CHRDEV		3
2238 #define EXT4_FT_BLKDEV		4
2239 #define EXT4_FT_FIFO		5
2240 #define EXT4_FT_SOCK		6
2241 #define EXT4_FT_SYMLINK		7
2242 
2243 #define EXT4_FT_MAX		8
2244 
2245 #define EXT4_FT_DIR_CSUM	0xDE
2246 
2247 /*
2248  * EXT4_DIR_PAD defines the directory entries boundaries
2249  *
2250  * NOTE: It must be a multiple of 4
2251  */
2252 #define EXT4_DIR_PAD			4
2253 #define EXT4_DIR_ROUND			(EXT4_DIR_PAD - 1)
2254 #define EXT4_DIR_REC_LEN(name_len)	(((name_len) + 8 + EXT4_DIR_ROUND) & \
2255 					 ~EXT4_DIR_ROUND)
2256 #define EXT4_MAX_REC_LEN		((1<<16)-1)
2257 
2258 /*
2259  * If we ever get support for fs block sizes > page_size, we'll need
2260  * to remove the #if statements in the next two functions...
2261  */
2262 static inline unsigned int
2263 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
2264 {
2265 	unsigned len = le16_to_cpu(dlen);
2266 
2267 #if (PAGE_SIZE >= 65536)
2268 	if (len == EXT4_MAX_REC_LEN || len == 0)
2269 		return blocksize;
2270 	return (len & 65532) | ((len & 3) << 16);
2271 #else
2272 	return len;
2273 #endif
2274 }
2275 
2276 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
2277 {
2278 	if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
2279 		BUG();
2280 #if (PAGE_SIZE >= 65536)
2281 	if (len < 65536)
2282 		return cpu_to_le16(len);
2283 	if (len == blocksize) {
2284 		if (blocksize == 65536)
2285 			return cpu_to_le16(EXT4_MAX_REC_LEN);
2286 		else
2287 			return cpu_to_le16(0);
2288 	}
2289 	return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
2290 #else
2291 	return cpu_to_le16(len);
2292 #endif
2293 }
2294 
2295 /*
2296  * Hash Tree Directory indexing
2297  * (c) Daniel Phillips, 2001
2298  */
2299 
2300 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
2301 		    ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2302 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \
2303 		    !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir)))
2304 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2305 
2306 /* Legal values for the dx_root hash_version field: */
2307 
2308 #define DX_HASH_LEGACY			0
2309 #define DX_HASH_HALF_MD4		1
2310 #define DX_HASH_TEA			2
2311 #define DX_HASH_LEGACY_UNSIGNED		3
2312 #define DX_HASH_HALF_MD4_UNSIGNED	4
2313 #define DX_HASH_TEA_UNSIGNED		5
2314 
2315 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2316 			      const void *address, unsigned int length)
2317 {
2318 	struct {
2319 		struct shash_desc shash;
2320 		char ctx[4];
2321 	} desc;
2322 
2323 	BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2324 
2325 	desc.shash.tfm = sbi->s_chksum_driver;
2326 	*(u32 *)desc.ctx = crc;
2327 
2328 	BUG_ON(crypto_shash_update(&desc.shash, address, length));
2329 
2330 	return *(u32 *)desc.ctx;
2331 }
2332 
2333 #ifdef __KERNEL__
2334 
2335 /* hash info structure used by the directory hash */
2336 struct dx_hash_info
2337 {
2338 	u32		hash;
2339 	u32		minor_hash;
2340 	int		hash_version;
2341 	u32		*seed;
2342 };
2343 
2344 
2345 /* 32 and 64 bit signed EOF for dx directories */
2346 #define EXT4_HTREE_EOF_32BIT   ((1UL  << (32 - 1)) - 1)
2347 #define EXT4_HTREE_EOF_64BIT   ((1ULL << (64 - 1)) - 1)
2348 
2349 
2350 /*
2351  * Control parameters used by ext4_htree_next_block
2352  */
2353 #define HASH_NB_ALWAYS		1
2354 
2355 struct ext4_filename {
2356 	const struct qstr *usr_fname;
2357 	struct fscrypt_str disk_name;
2358 	struct dx_hash_info hinfo;
2359 #ifdef CONFIG_FS_ENCRYPTION
2360 	struct fscrypt_str crypto_buf;
2361 #endif
2362 #ifdef CONFIG_UNICODE
2363 	struct fscrypt_str cf_name;
2364 #endif
2365 };
2366 
2367 #define fname_name(p) ((p)->disk_name.name)
2368 #define fname_len(p)  ((p)->disk_name.len)
2369 
2370 /*
2371  * Describe an inode's exact location on disk and in memory
2372  */
2373 struct ext4_iloc
2374 {
2375 	struct buffer_head *bh;
2376 	unsigned long offset;
2377 	ext4_group_t block_group;
2378 };
2379 
2380 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2381 {
2382 	return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2383 }
2384 
2385 static inline bool ext4_is_quota_file(struct inode *inode)
2386 {
2387 	return IS_NOQUOTA(inode) &&
2388 	       !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL);
2389 }
2390 
2391 /*
2392  * This structure is stuffed into the struct file's private_data field
2393  * for directories.  It is where we put information so that we can do
2394  * readdir operations in hash tree order.
2395  */
2396 struct dir_private_info {
2397 	struct rb_root	root;
2398 	struct rb_node	*curr_node;
2399 	struct fname	*extra_fname;
2400 	loff_t		last_pos;
2401 	__u32		curr_hash;
2402 	__u32		curr_minor_hash;
2403 	__u32		next_hash;
2404 };
2405 
2406 /* calculate the first block number of the group */
2407 static inline ext4_fsblk_t
2408 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2409 {
2410 	return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2411 		le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2412 }
2413 
2414 /*
2415  * Special error return code only used by dx_probe() and its callers.
2416  */
2417 #define ERR_BAD_DX_DIR	(-(MAX_ERRNO - 1))
2418 
2419 /* htree levels for ext4 */
2420 #define	EXT4_HTREE_LEVEL_COMPAT	2
2421 #define	EXT4_HTREE_LEVEL	3
2422 
2423 static inline int ext4_dir_htree_level(struct super_block *sb)
2424 {
2425 	return ext4_has_feature_largedir(sb) ?
2426 		EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT;
2427 }
2428 
2429 /*
2430  * Timeout and state flag for lazy initialization inode thread.
2431  */
2432 #define EXT4_DEF_LI_WAIT_MULT			10
2433 #define EXT4_DEF_LI_MAX_START_DELAY		5
2434 #define EXT4_LAZYINIT_QUIT			0x0001
2435 #define EXT4_LAZYINIT_RUNNING			0x0002
2436 
2437 /*
2438  * Lazy inode table initialization info
2439  */
2440 struct ext4_lazy_init {
2441 	unsigned long		li_state;
2442 	struct list_head	li_request_list;
2443 	struct mutex		li_list_mtx;
2444 };
2445 
2446 enum ext4_li_mode {
2447 	EXT4_LI_MODE_PREFETCH_BBITMAP,
2448 	EXT4_LI_MODE_ITABLE,
2449 };
2450 
2451 struct ext4_li_request {
2452 	struct super_block	*lr_super;
2453 	enum ext4_li_mode	lr_mode;
2454 	ext4_group_t		lr_first_not_zeroed;
2455 	ext4_group_t		lr_next_group;
2456 	struct list_head	lr_request;
2457 	unsigned long		lr_next_sched;
2458 	unsigned long		lr_timeout;
2459 };
2460 
2461 struct ext4_features {
2462 	struct kobject f_kobj;
2463 	struct completion f_kobj_unregister;
2464 };
2465 
2466 /*
2467  * This structure will be used for multiple mount protection. It will be
2468  * written into the block number saved in the s_mmp_block field in the
2469  * superblock. Programs that check MMP should assume that if
2470  * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2471  * to use the filesystem, regardless of how old the timestamp is.
2472  */
2473 #define EXT4_MMP_MAGIC     0x004D4D50U /* ASCII for MMP */
2474 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2475 #define EXT4_MMP_SEQ_FSCK  0xE24D4D50U /* mmp_seq value when being fscked */
2476 #define EXT4_MMP_SEQ_MAX   0xE24D4D4FU /* maximum valid mmp_seq value */
2477 
2478 struct mmp_struct {
2479 	__le32	mmp_magic;		/* Magic number for MMP */
2480 	__le32	mmp_seq;		/* Sequence no. updated periodically */
2481 
2482 	/*
2483 	 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2484 	 * purposes and do not affect the correctness of the algorithm
2485 	 */
2486 	__le64	mmp_time;		/* Time last updated */
2487 	char	mmp_nodename[64];	/* Node which last updated MMP block */
2488 	char	mmp_bdevname[32];	/* Bdev which last updated MMP block */
2489 
2490 	/*
2491 	 * mmp_check_interval is used to verify if the MMP block has been
2492 	 * updated on the block device. The value is updated based on the
2493 	 * maximum time to write the MMP block during an update cycle.
2494 	 */
2495 	__le16	mmp_check_interval;
2496 
2497 	__le16	mmp_pad1;
2498 	__le32	mmp_pad2[226];
2499 	__le32	mmp_checksum;		/* crc32c(uuid+mmp_block) */
2500 };
2501 
2502 /* arguments passed to the mmp thread */
2503 struct mmpd_data {
2504 	struct buffer_head *bh; /* bh from initial read_mmp_block() */
2505 	struct super_block *sb;  /* super block of the fs */
2506 };
2507 
2508 /*
2509  * Check interval multiplier
2510  * The MMP block is written every update interval and initially checked every
2511  * update interval x the multiplier (the value is then adapted based on the
2512  * write latency). The reason is that writes can be delayed under load and we
2513  * don't want readers to incorrectly assume that the filesystem is no longer
2514  * in use.
2515  */
2516 #define EXT4_MMP_CHECK_MULT		2UL
2517 
2518 /*
2519  * Minimum interval for MMP checking in seconds.
2520  */
2521 #define EXT4_MMP_MIN_CHECK_INTERVAL	5UL
2522 
2523 /*
2524  * Maximum interval for MMP checking in seconds.
2525  */
2526 #define EXT4_MMP_MAX_CHECK_INTERVAL	300UL
2527 
2528 /*
2529  * Function prototypes
2530  */
2531 
2532 /*
2533  * Ok, these declarations are also in <linux/kernel.h> but none of the
2534  * ext4 source programs needs to include it so they are duplicated here.
2535  */
2536 # define NORET_TYPE	/**/
2537 # define ATTRIB_NORET	__attribute__((noreturn))
2538 # define NORET_AND	noreturn,
2539 
2540 /* bitmap.c */
2541 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2542 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2543 				struct ext4_group_desc *gdp,
2544 				struct buffer_head *bh, int sz);
2545 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2546 				  struct ext4_group_desc *gdp,
2547 				  struct buffer_head *bh, int sz);
2548 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2549 				struct ext4_group_desc *gdp,
2550 				struct buffer_head *bh);
2551 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2552 				  struct ext4_group_desc *gdp,
2553 				  struct buffer_head *bh);
2554 
2555 /* balloc.c */
2556 extern void ext4_get_group_no_and_offset(struct super_block *sb,
2557 					 ext4_fsblk_t blocknr,
2558 					 ext4_group_t *blockgrpp,
2559 					 ext4_grpblk_t *offsetp);
2560 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2561 					  ext4_fsblk_t block);
2562 
2563 extern unsigned int ext4_block_group(struct super_block *sb,
2564 			ext4_fsblk_t blocknr);
2565 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2566 			ext4_fsblk_t blocknr);
2567 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2568 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2569 			ext4_group_t group);
2570 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2571 					 ext4_fsblk_t goal,
2572 					 unsigned int flags,
2573 					 unsigned long *count,
2574 					 int *errp);
2575 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2576 				    s64 nclusters, unsigned int flags);
2577 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2578 extern void ext4_check_blocks_bitmap(struct super_block *);
2579 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2580 						    ext4_group_t block_group,
2581 						    struct buffer_head ** bh);
2582 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2583 
2584 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2585 						ext4_group_t block_group,
2586 						bool ignore_locked);
2587 extern int ext4_wait_block_bitmap(struct super_block *sb,
2588 				  ext4_group_t block_group,
2589 				  struct buffer_head *bh);
2590 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2591 						  ext4_group_t block_group);
2592 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2593 					      ext4_group_t block_group,
2594 					      struct ext4_group_desc *gdp);
2595 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2596 
2597 #ifdef CONFIG_UNICODE
2598 extern void ext4_fname_setup_ci_filename(struct inode *dir,
2599 					 const struct qstr *iname,
2600 					 struct fscrypt_str *fname);
2601 #endif
2602 
2603 #ifdef CONFIG_FS_ENCRYPTION
2604 static inline void ext4_fname_from_fscrypt_name(struct ext4_filename *dst,
2605 						const struct fscrypt_name *src)
2606 {
2607 	memset(dst, 0, sizeof(*dst));
2608 
2609 	dst->usr_fname = src->usr_fname;
2610 	dst->disk_name = src->disk_name;
2611 	dst->hinfo.hash = src->hash;
2612 	dst->hinfo.minor_hash = src->minor_hash;
2613 	dst->crypto_buf = src->crypto_buf;
2614 }
2615 
2616 static inline int ext4_fname_setup_filename(struct inode *dir,
2617 					    const struct qstr *iname,
2618 					    int lookup,
2619 					    struct ext4_filename *fname)
2620 {
2621 	struct fscrypt_name name;
2622 	int err;
2623 
2624 	err = fscrypt_setup_filename(dir, iname, lookup, &name);
2625 	if (err)
2626 		return err;
2627 
2628 	ext4_fname_from_fscrypt_name(fname, &name);
2629 
2630 #ifdef CONFIG_UNICODE
2631 	ext4_fname_setup_ci_filename(dir, iname, &fname->cf_name);
2632 #endif
2633 	return 0;
2634 }
2635 
2636 static inline int ext4_fname_prepare_lookup(struct inode *dir,
2637 					    struct dentry *dentry,
2638 					    struct ext4_filename *fname)
2639 {
2640 	struct fscrypt_name name;
2641 	int err;
2642 
2643 	err = fscrypt_prepare_lookup(dir, dentry, &name);
2644 	if (err)
2645 		return err;
2646 
2647 	ext4_fname_from_fscrypt_name(fname, &name);
2648 
2649 #ifdef CONFIG_UNICODE
2650 	ext4_fname_setup_ci_filename(dir, &dentry->d_name, &fname->cf_name);
2651 #endif
2652 	return 0;
2653 }
2654 
2655 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2656 {
2657 	struct fscrypt_name name;
2658 
2659 	name.crypto_buf = fname->crypto_buf;
2660 	fscrypt_free_filename(&name);
2661 
2662 	fname->crypto_buf.name = NULL;
2663 	fname->usr_fname = NULL;
2664 	fname->disk_name.name = NULL;
2665 
2666 #ifdef CONFIG_UNICODE
2667 	kfree(fname->cf_name.name);
2668 	fname->cf_name.name = NULL;
2669 #endif
2670 }
2671 #else /* !CONFIG_FS_ENCRYPTION */
2672 static inline int ext4_fname_setup_filename(struct inode *dir,
2673 					    const struct qstr *iname,
2674 					    int lookup,
2675 					    struct ext4_filename *fname)
2676 {
2677 	fname->usr_fname = iname;
2678 	fname->disk_name.name = (unsigned char *) iname->name;
2679 	fname->disk_name.len = iname->len;
2680 
2681 #ifdef CONFIG_UNICODE
2682 	ext4_fname_setup_ci_filename(dir, iname, &fname->cf_name);
2683 #endif
2684 
2685 	return 0;
2686 }
2687 
2688 static inline int ext4_fname_prepare_lookup(struct inode *dir,
2689 					    struct dentry *dentry,
2690 					    struct ext4_filename *fname)
2691 {
2692 	return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname);
2693 }
2694 
2695 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2696 {
2697 #ifdef CONFIG_UNICODE
2698 	kfree(fname->cf_name.name);
2699 	fname->cf_name.name = NULL;
2700 #endif
2701 }
2702 #endif /* !CONFIG_FS_ENCRYPTION */
2703 
2704 /* dir.c */
2705 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2706 				  struct file *,
2707 				  struct ext4_dir_entry_2 *,
2708 				  struct buffer_head *, char *, int,
2709 				  unsigned int);
2710 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset)	\
2711 	unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2712 					(de), (bh), (buf), (size), (offset)))
2713 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2714 				__u32 minor_hash,
2715 				struct ext4_dir_entry_2 *dirent,
2716 				struct fscrypt_str *ent_name);
2717 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2718 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2719 			     struct buffer_head *bh,
2720 			     void *buf, int buf_size,
2721 			     struct ext4_filename *fname,
2722 			     struct ext4_dir_entry_2 **dest_de);
2723 void ext4_insert_dentry(struct inode *inode,
2724 			struct ext4_dir_entry_2 *de,
2725 			int buf_size,
2726 			struct ext4_filename *fname);
2727 static inline void ext4_update_dx_flag(struct inode *inode)
2728 {
2729 	if (!ext4_has_feature_dir_index(inode->i_sb) &&
2730 	    ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) {
2731 		/* ext4_iget() should have caught this... */
2732 		WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb));
2733 		ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2734 	}
2735 }
2736 static const unsigned char ext4_filetype_table[] = {
2737 	DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2738 };
2739 
2740 static inline  unsigned char get_dtype(struct super_block *sb, int filetype)
2741 {
2742 	if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2743 		return DT_UNKNOWN;
2744 
2745 	return ext4_filetype_table[filetype];
2746 }
2747 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2748 			     void *buf, int buf_size);
2749 
2750 /* fsync.c */
2751 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2752 
2753 /* hash.c */
2754 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len,
2755 			  struct dx_hash_info *hinfo);
2756 
2757 /* ialloc.c */
2758 extern int ext4_mark_inode_used(struct super_block *sb, int ino);
2759 extern struct inode *__ext4_new_inode(struct user_namespace *, handle_t *,
2760 				      struct inode *, umode_t,
2761 				      const struct qstr *qstr, __u32 goal,
2762 				      uid_t *owner, __u32 i_flags,
2763 				      int handle_type, unsigned int line_no,
2764 				      int nblocks);
2765 
2766 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags)          \
2767 	__ext4_new_inode(&init_user_ns, (handle), (dir), (mode), (qstr),       \
2768 			 (goal), (owner), i_flags, 0, 0, 0)
2769 #define ext4_new_inode_start_handle(mnt_userns, dir, mode, qstr, goal, owner, \
2770 				    type, nblocks)		    \
2771 	__ext4_new_inode((mnt_userns), NULL, (dir), (mode), (qstr), (goal), (owner), \
2772 			 0, (type), __LINE__, (nblocks))
2773 
2774 
2775 extern void ext4_free_inode(handle_t *, struct inode *);
2776 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2777 extern unsigned long ext4_count_free_inodes(struct super_block *);
2778 extern unsigned long ext4_count_dirs(struct super_block *);
2779 extern void ext4_check_inodes_bitmap(struct super_block *);
2780 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2781 extern int ext4_init_inode_table(struct super_block *sb,
2782 				 ext4_group_t group, int barrier);
2783 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2784 
2785 /* fast_commit.c */
2786 int ext4_fc_info_show(struct seq_file *seq, void *v);
2787 void ext4_fc_init(struct super_block *sb, journal_t *journal);
2788 void ext4_fc_init_inode(struct inode *inode);
2789 void ext4_fc_track_range(handle_t *handle, struct inode *inode, ext4_lblk_t start,
2790 			 ext4_lblk_t end);
2791 void __ext4_fc_track_unlink(handle_t *handle, struct inode *inode,
2792 	struct dentry *dentry);
2793 void __ext4_fc_track_link(handle_t *handle, struct inode *inode,
2794 	struct dentry *dentry);
2795 void ext4_fc_track_unlink(handle_t *handle, struct dentry *dentry);
2796 void ext4_fc_track_link(handle_t *handle, struct dentry *dentry);
2797 void __ext4_fc_track_create(handle_t *handle, struct inode *inode,
2798 			    struct dentry *dentry);
2799 void ext4_fc_track_create(handle_t *handle, struct dentry *dentry);
2800 void ext4_fc_track_inode(handle_t *handle, struct inode *inode);
2801 void ext4_fc_mark_ineligible(struct super_block *sb, int reason);
2802 void ext4_fc_start_ineligible(struct super_block *sb, int reason);
2803 void ext4_fc_stop_ineligible(struct super_block *sb);
2804 void ext4_fc_start_update(struct inode *inode);
2805 void ext4_fc_stop_update(struct inode *inode);
2806 void ext4_fc_del(struct inode *inode);
2807 bool ext4_fc_replay_check_excluded(struct super_block *sb, ext4_fsblk_t block);
2808 void ext4_fc_replay_cleanup(struct super_block *sb);
2809 int ext4_fc_commit(journal_t *journal, tid_t commit_tid);
2810 int __init ext4_fc_init_dentry_cache(void);
2811 
2812 /* mballoc.c */
2813 extern const struct seq_operations ext4_mb_seq_groups_ops;
2814 extern long ext4_mb_stats;
2815 extern long ext4_mb_max_to_scan;
2816 extern int ext4_mb_init(struct super_block *);
2817 extern int ext4_mb_release(struct super_block *);
2818 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2819 				struct ext4_allocation_request *, int *);
2820 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2821 extern void ext4_discard_preallocations(struct inode *, unsigned int);
2822 extern int __init ext4_init_mballoc(void);
2823 extern void ext4_exit_mballoc(void);
2824 extern ext4_group_t ext4_mb_prefetch(struct super_block *sb,
2825 				     ext4_group_t group,
2826 				     unsigned int nr, int *cnt);
2827 extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group,
2828 				  unsigned int nr);
2829 
2830 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2831 			     struct buffer_head *bh, ext4_fsblk_t block,
2832 			     unsigned long count, int flags);
2833 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2834 				   ext4_group_t ngroups);
2835 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2836 		ext4_group_t i, struct ext4_group_desc *desc);
2837 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2838 				ext4_fsblk_t block, unsigned long count);
2839 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2840 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid);
2841 extern void ext4_mb_mark_bb(struct super_block *sb, ext4_fsblk_t block,
2842 		       int len, int state);
2843 
2844 /* inode.c */
2845 void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw,
2846 			 struct ext4_inode_info *ei);
2847 int ext4_inode_is_fast_symlink(struct inode *inode);
2848 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2849 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2850 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count,
2851 		     bool wait, struct buffer_head **bhs);
2852 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2853 			     struct buffer_head *bh_result, int create);
2854 int ext4_get_block(struct inode *inode, sector_t iblock,
2855 		   struct buffer_head *bh_result, int create);
2856 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2857 			   struct buffer_head *bh, int create);
2858 int ext4_walk_page_buffers(handle_t *handle,
2859 			   struct buffer_head *head,
2860 			   unsigned from,
2861 			   unsigned to,
2862 			   int *partial,
2863 			   int (*fn)(handle_t *handle,
2864 				     struct buffer_head *bh));
2865 int do_journal_get_write_access(handle_t *handle,
2866 				struct buffer_head *bh);
2867 #define FALL_BACK_TO_NONDELALLOC 1
2868 #define CONVERT_INLINE_DATA	 2
2869 
2870 typedef enum {
2871 	EXT4_IGET_NORMAL =	0,
2872 	EXT4_IGET_SPECIAL =	0x0001, /* OK to iget a system inode */
2873 	EXT4_IGET_HANDLE = 	0x0002	/* Inode # is from a handle */
2874 } ext4_iget_flags;
2875 
2876 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
2877 				 ext4_iget_flags flags, const char *function,
2878 				 unsigned int line);
2879 
2880 #define ext4_iget(sb, ino, flags) \
2881 	__ext4_iget((sb), (ino), (flags), __func__, __LINE__)
2882 
2883 extern int  ext4_write_inode(struct inode *, struct writeback_control *);
2884 extern int  ext4_setattr(struct user_namespace *, struct dentry *,
2885 			 struct iattr *);
2886 extern int  ext4_getattr(struct user_namespace *, const struct path *,
2887 			 struct kstat *, u32, unsigned int);
2888 extern void ext4_evict_inode(struct inode *);
2889 extern void ext4_clear_inode(struct inode *);
2890 extern int  ext4_file_getattr(struct user_namespace *, const struct path *,
2891 			      struct kstat *, u32, unsigned int);
2892 extern int  ext4_sync_inode(handle_t *, struct inode *);
2893 extern void ext4_dirty_inode(struct inode *, int);
2894 extern int ext4_change_inode_journal_flag(struct inode *, int);
2895 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2896 extern int ext4_get_fc_inode_loc(struct super_block *sb, unsigned long ino,
2897 			  struct ext4_iloc *iloc);
2898 extern int ext4_inode_attach_jinode(struct inode *inode);
2899 extern int ext4_can_truncate(struct inode *inode);
2900 extern int ext4_truncate(struct inode *);
2901 extern int ext4_break_layouts(struct inode *);
2902 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2903 extern void ext4_set_inode_flags(struct inode *, bool init);
2904 extern int ext4_alloc_da_blocks(struct inode *inode);
2905 extern void ext4_set_aops(struct inode *inode);
2906 extern int ext4_writepage_trans_blocks(struct inode *);
2907 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2908 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2909 			     loff_t lstart, loff_t lend);
2910 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf);
2911 extern vm_fault_t ext4_filemap_fault(struct vm_fault *vmf);
2912 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2913 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
2914 extern void ext4_da_release_space(struct inode *inode, int to_free);
2915 extern void ext4_da_update_reserve_space(struct inode *inode,
2916 					int used, int quota_claim);
2917 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
2918 			      ext4_fsblk_t pblk, ext4_lblk_t len);
2919 
2920 /* indirect.c */
2921 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2922 				struct ext4_map_blocks *map, int flags);
2923 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2924 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2925 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2926 				 ext4_lblk_t start, ext4_lblk_t end);
2927 
2928 /* ioctl.c */
2929 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2930 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2931 extern void ext4_reset_inode_seed(struct inode *inode);
2932 
2933 /* migrate.c */
2934 extern int ext4_ext_migrate(struct inode *);
2935 extern int ext4_ind_migrate(struct inode *inode);
2936 
2937 /* namei.c */
2938 extern int ext4_init_new_dir(handle_t *handle, struct inode *dir,
2939 			     struct inode *inode);
2940 extern int ext4_dirblock_csum_verify(struct inode *inode,
2941 				     struct buffer_head *bh);
2942 extern int ext4_orphan_add(handle_t *, struct inode *);
2943 extern int ext4_orphan_del(handle_t *, struct inode *);
2944 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2945 				__u32 start_minor_hash, __u32 *next_hash);
2946 extern int ext4_search_dir(struct buffer_head *bh,
2947 			   char *search_buf,
2948 			   int buf_size,
2949 			   struct inode *dir,
2950 			   struct ext4_filename *fname,
2951 			   unsigned int offset,
2952 			   struct ext4_dir_entry_2 **res_dir);
2953 extern int ext4_generic_delete_entry(struct inode *dir,
2954 				     struct ext4_dir_entry_2 *de_del,
2955 				     struct buffer_head *bh,
2956 				     void *entry_buf,
2957 				     int buf_size,
2958 				     int csum_size);
2959 extern bool ext4_empty_dir(struct inode *inode);
2960 
2961 /* resize.c */
2962 extern void ext4_kvfree_array_rcu(void *to_free);
2963 extern int ext4_group_add(struct super_block *sb,
2964 				struct ext4_new_group_data *input);
2965 extern int ext4_group_extend(struct super_block *sb,
2966 				struct ext4_super_block *es,
2967 				ext4_fsblk_t n_blocks_count);
2968 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2969 
2970 /* super.c */
2971 extern struct buffer_head *ext4_sb_bread(struct super_block *sb,
2972 					 sector_t block, int op_flags);
2973 extern struct buffer_head *ext4_sb_bread_unmovable(struct super_block *sb,
2974 						   sector_t block);
2975 extern void ext4_read_bh_nowait(struct buffer_head *bh, int op_flags,
2976 				bh_end_io_t *end_io);
2977 extern int ext4_read_bh(struct buffer_head *bh, int op_flags,
2978 			bh_end_io_t *end_io);
2979 extern int ext4_read_bh_lock(struct buffer_head *bh, int op_flags, bool wait);
2980 extern void ext4_sb_breadahead_unmovable(struct super_block *sb, sector_t block);
2981 extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
2982 extern int ext4_calculate_overhead(struct super_block *sb);
2983 extern void ext4_superblock_csum_set(struct super_block *sb);
2984 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2985 				    ext4_group_t ngroup);
2986 extern const char *ext4_decode_error(struct super_block *sb, int errno,
2987 				     char nbuf[16]);
2988 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb,
2989 					     ext4_group_t block_group,
2990 					     unsigned int flags);
2991 
2992 extern __printf(7, 8)
2993 void __ext4_error(struct super_block *, const char *, unsigned int, bool,
2994 		  int, __u64, const char *, ...);
2995 extern __printf(6, 7)
2996 void __ext4_error_inode(struct inode *, const char *, unsigned int,
2997 			ext4_fsblk_t, int, const char *, ...);
2998 extern __printf(5, 6)
2999 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
3000 		     const char *, ...);
3001 extern void __ext4_std_error(struct super_block *, const char *,
3002 			     unsigned int, int);
3003 extern __printf(4, 5)
3004 void __ext4_warning(struct super_block *, const char *, unsigned int,
3005 		    const char *, ...);
3006 extern __printf(4, 5)
3007 void __ext4_warning_inode(const struct inode *inode, const char *function,
3008 			  unsigned int line, const char *fmt, ...);
3009 extern __printf(3, 4)
3010 void __ext4_msg(struct super_block *, const char *, const char *, ...);
3011 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
3012 			   const char *, unsigned int, const char *);
3013 extern __printf(7, 8)
3014 void __ext4_grp_locked_error(const char *, unsigned int,
3015 			     struct super_block *, ext4_group_t,
3016 			     unsigned long, ext4_fsblk_t,
3017 			     const char *, ...);
3018 
3019 #define EXT4_ERROR_INODE(inode, fmt, a...) \
3020 	ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
3021 
3022 #define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...)			\
3023 	__ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a)
3024 
3025 #define ext4_error_inode_block(inode, block, err, fmt, a...)		\
3026 	__ext4_error_inode((inode), __func__, __LINE__, (block), (err),	\
3027 			   (fmt), ## a)
3028 
3029 #define EXT4_ERROR_FILE(file, block, fmt, a...)				\
3030 	ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
3031 
3032 #define ext4_abort(sb, err, fmt, a...)					\
3033 	__ext4_error((sb), __func__, __LINE__, true, (err), 0, (fmt), ## a)
3034 
3035 #ifdef CONFIG_PRINTK
3036 
3037 #define ext4_error_inode(inode, func, line, block, fmt, ...)		\
3038 	__ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__)
3039 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...)	\
3040 	__ext4_error_inode((inode), (func), (line), (block), 		\
3041 			   (err), (fmt), ##__VA_ARGS__)
3042 #define ext4_error_file(file, func, line, block, fmt, ...)		\
3043 	__ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
3044 #define ext4_error(sb, fmt, ...)					\
3045 	__ext4_error((sb), __func__, __LINE__, false, 0, 0, (fmt),	\
3046 		##__VA_ARGS__)
3047 #define ext4_error_err(sb, err, fmt, ...)				\
3048 	__ext4_error((sb), __func__, __LINE__, false, (err), 0, (fmt),	\
3049 		##__VA_ARGS__)
3050 #define ext4_warning(sb, fmt, ...)					\
3051 	__ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
3052 #define ext4_warning_inode(inode, fmt, ...)				\
3053 	__ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
3054 #define ext4_msg(sb, level, fmt, ...)				\
3055 	__ext4_msg(sb, level, fmt, ##__VA_ARGS__)
3056 #define dump_mmp_msg(sb, mmp, msg)					\
3057 	__dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
3058 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...)		\
3059 	__ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
3060 				fmt, ##__VA_ARGS__)
3061 
3062 #else
3063 
3064 #define ext4_error_inode(inode, func, line, block, fmt, ...)		\
3065 do {									\
3066 	no_printk(fmt, ##__VA_ARGS__);					\
3067 	__ext4_error_inode(inode, "", 0, block, 0, " ");		\
3068 } while (0)
3069 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...)	\
3070 do {									\
3071 	no_printk(fmt, ##__VA_ARGS__);					\
3072 	__ext4_error_inode(inode, "", 0, block, err, " ");		\
3073 } while (0)
3074 #define ext4_error_file(file, func, line, block, fmt, ...)		\
3075 do {									\
3076 	no_printk(fmt, ##__VA_ARGS__);					\
3077 	__ext4_error_file(file, "", 0, block, " ");			\
3078 } while (0)
3079 #define ext4_error(sb, fmt, ...)					\
3080 do {									\
3081 	no_printk(fmt, ##__VA_ARGS__);					\
3082 	__ext4_error(sb, "", 0, false, 0, 0, " ");			\
3083 } while (0)
3084 #define ext4_error_err(sb, err, fmt, ...)				\
3085 do {									\
3086 	no_printk(fmt, ##__VA_ARGS__);					\
3087 	__ext4_error(sb, "", 0, false, err, 0, " ");			\
3088 } while (0)
3089 #define ext4_warning(sb, fmt, ...)					\
3090 do {									\
3091 	no_printk(fmt, ##__VA_ARGS__);					\
3092 	__ext4_warning(sb, "", 0, " ");					\
3093 } while (0)
3094 #define ext4_warning_inode(inode, fmt, ...)				\
3095 do {									\
3096 	no_printk(fmt, ##__VA_ARGS__);					\
3097 	__ext4_warning_inode(inode, "", 0, " ");			\
3098 } while (0)
3099 #define ext4_msg(sb, level, fmt, ...)					\
3100 do {									\
3101 	no_printk(fmt, ##__VA_ARGS__);					\
3102 	__ext4_msg(sb, "", " ");					\
3103 } while (0)
3104 #define dump_mmp_msg(sb, mmp, msg)					\
3105 	__dump_mmp_msg(sb, mmp, "", 0, "")
3106 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...)		\
3107 do {									\
3108 	no_printk(fmt, ##__VA_ARGS__);				\
3109 	__ext4_grp_locked_error("", 0, sb, grp, ino, block, " ");	\
3110 } while (0)
3111 
3112 #endif
3113 
3114 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
3115 				      struct ext4_group_desc *bg);
3116 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
3117 				      struct ext4_group_desc *bg);
3118 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
3119 				     struct ext4_group_desc *bg);
3120 extern __u32 ext4_free_group_clusters(struct super_block *sb,
3121 				      struct ext4_group_desc *bg);
3122 extern __u32 ext4_free_inodes_count(struct super_block *sb,
3123 				 struct ext4_group_desc *bg);
3124 extern __u32 ext4_used_dirs_count(struct super_block *sb,
3125 				struct ext4_group_desc *bg);
3126 extern __u32 ext4_itable_unused_count(struct super_block *sb,
3127 				   struct ext4_group_desc *bg);
3128 extern void ext4_block_bitmap_set(struct super_block *sb,
3129 				  struct ext4_group_desc *bg, ext4_fsblk_t blk);
3130 extern void ext4_inode_bitmap_set(struct super_block *sb,
3131 				  struct ext4_group_desc *bg, ext4_fsblk_t blk);
3132 extern void ext4_inode_table_set(struct super_block *sb,
3133 				 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3134 extern void ext4_free_group_clusters_set(struct super_block *sb,
3135 					 struct ext4_group_desc *bg,
3136 					 __u32 count);
3137 extern void ext4_free_inodes_set(struct super_block *sb,
3138 				struct ext4_group_desc *bg, __u32 count);
3139 extern void ext4_used_dirs_set(struct super_block *sb,
3140 				struct ext4_group_desc *bg, __u32 count);
3141 extern void ext4_itable_unused_set(struct super_block *sb,
3142 				   struct ext4_group_desc *bg, __u32 count);
3143 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
3144 				       struct ext4_group_desc *gdp);
3145 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
3146 				     struct ext4_group_desc *gdp);
3147 extern int ext4_register_li_request(struct super_block *sb,
3148 				    ext4_group_t first_not_zeroed);
3149 
3150 static inline int ext4_has_metadata_csum(struct super_block *sb)
3151 {
3152 	WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
3153 		     !EXT4_SB(sb)->s_chksum_driver);
3154 
3155 	return ext4_has_feature_metadata_csum(sb) &&
3156 	       (EXT4_SB(sb)->s_chksum_driver != NULL);
3157 }
3158 
3159 static inline int ext4_has_group_desc_csum(struct super_block *sb)
3160 {
3161 	return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb);
3162 }
3163 
3164 #define ext4_read_incompat_64bit_val(es, name) \
3165 	(((es)->s_feature_incompat & cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT) \
3166 		? (ext4_fsblk_t)le32_to_cpu(es->name##_hi) << 32 : 0) | \
3167 		le32_to_cpu(es->name##_lo))
3168 
3169 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
3170 {
3171 	return ext4_read_incompat_64bit_val(es, s_blocks_count);
3172 }
3173 
3174 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
3175 {
3176 	return ext4_read_incompat_64bit_val(es, s_r_blocks_count);
3177 }
3178 
3179 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
3180 {
3181 	return ext4_read_incompat_64bit_val(es, s_free_blocks_count);
3182 }
3183 
3184 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
3185 					 ext4_fsblk_t blk)
3186 {
3187 	es->s_blocks_count_lo = cpu_to_le32((u32)blk);
3188 	es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
3189 }
3190 
3191 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
3192 					      ext4_fsblk_t blk)
3193 {
3194 	es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
3195 	es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
3196 }
3197 
3198 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
3199 					   ext4_fsblk_t blk)
3200 {
3201 	es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
3202 	es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
3203 }
3204 
3205 static inline loff_t ext4_isize(struct super_block *sb,
3206 				struct ext4_inode *raw_inode)
3207 {
3208 	if (ext4_has_feature_largedir(sb) ||
3209 	    S_ISREG(le16_to_cpu(raw_inode->i_mode)))
3210 		return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
3211 			le32_to_cpu(raw_inode->i_size_lo);
3212 
3213 	return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
3214 }
3215 
3216 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
3217 {
3218 	raw_inode->i_size_lo = cpu_to_le32(i_size);
3219 	raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
3220 }
3221 
3222 static inline
3223 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
3224 					    ext4_group_t group)
3225 {
3226 	 struct ext4_group_info **grp_info;
3227 	 long indexv, indexh;
3228 	 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
3229 	 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
3230 	 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
3231 	 grp_info = sbi_array_rcu_deref(EXT4_SB(sb), s_group_info, indexv);
3232 	 return grp_info[indexh];
3233 }
3234 
3235 /*
3236  * Reading s_groups_count requires using smp_rmb() afterwards.  See
3237  * the locking protocol documented in the comments of ext4_group_add()
3238  * in resize.c
3239  */
3240 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
3241 {
3242 	ext4_group_t	ngroups = EXT4_SB(sb)->s_groups_count;
3243 
3244 	smp_rmb();
3245 	return ngroups;
3246 }
3247 
3248 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
3249 					     ext4_group_t block_group)
3250 {
3251 	return block_group >> sbi->s_log_groups_per_flex;
3252 }
3253 
3254 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
3255 {
3256 	return 1 << sbi->s_log_groups_per_flex;
3257 }
3258 
3259 #define ext4_std_error(sb, errno)				\
3260 do {								\
3261 	if ((errno))						\
3262 		__ext4_std_error((sb), __func__, __LINE__, (errno));	\
3263 } while (0)
3264 
3265 #ifdef CONFIG_SMP
3266 /* Each CPU can accumulate percpu_counter_batch clusters in their local
3267  * counters. So we need to make sure we have free clusters more
3268  * than percpu_counter_batch  * nr_cpu_ids. Also add a window of 4 times.
3269  */
3270 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
3271 #else
3272 #define EXT4_FREECLUSTERS_WATERMARK 0
3273 #endif
3274 
3275 /* Update i_disksize. Requires i_mutex to avoid races with truncate */
3276 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
3277 {
3278 	WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
3279 		     !inode_is_locked(inode));
3280 	down_write(&EXT4_I(inode)->i_data_sem);
3281 	if (newsize > EXT4_I(inode)->i_disksize)
3282 		WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize);
3283 	up_write(&EXT4_I(inode)->i_data_sem);
3284 }
3285 
3286 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
3287 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
3288 {
3289 	int changed = 0;
3290 
3291 	if (newsize > inode->i_size) {
3292 		i_size_write(inode, newsize);
3293 		changed = 1;
3294 	}
3295 	if (newsize > EXT4_I(inode)->i_disksize) {
3296 		ext4_update_i_disksize(inode, newsize);
3297 		changed |= 2;
3298 	}
3299 	return changed;
3300 }
3301 
3302 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
3303 				      loff_t len);
3304 
3305 struct ext4_group_info {
3306 	unsigned long   bb_state;
3307 #ifdef AGGRESSIVE_CHECK
3308 	unsigned long	bb_check_counter;
3309 #endif
3310 	struct rb_root  bb_free_root;
3311 	ext4_grpblk_t	bb_first_free;	/* first free block */
3312 	ext4_grpblk_t	bb_free;	/* total free blocks */
3313 	ext4_grpblk_t	bb_fragments;	/* nr of freespace fragments */
3314 	ext4_grpblk_t	bb_largest_free_order;/* order of largest frag in BG */
3315 	struct          list_head bb_prealloc_list;
3316 #ifdef DOUBLE_CHECK
3317 	void            *bb_bitmap;
3318 #endif
3319 	struct rw_semaphore alloc_sem;
3320 	ext4_grpblk_t	bb_counters[];	/* Nr of free power-of-two-block
3321 					 * regions, index is order.
3322 					 * bb_counters[3] = 5 means
3323 					 * 5 free 8-block regions. */
3324 };
3325 
3326 #define EXT4_GROUP_INFO_NEED_INIT_BIT		0
3327 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT		1
3328 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT	2
3329 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT	3
3330 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT		\
3331 	(1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT)
3332 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT		\
3333 	(1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT)
3334 #define EXT4_GROUP_INFO_BBITMAP_READ_BIT	4
3335 
3336 #define EXT4_MB_GRP_NEED_INIT(grp)	\
3337 	(test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
3338 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp)	\
3339 	(test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3340 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp)	\
3341 	(test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3342 
3343 #define EXT4_MB_GRP_WAS_TRIMMED(grp)	\
3344 	(test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3345 #define EXT4_MB_GRP_SET_TRIMMED(grp)	\
3346 	(set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3347 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp)	\
3348 	(clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3349 #define EXT4_MB_GRP_TEST_AND_SET_READ(grp)	\
3350 	(test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state)))
3351 
3352 #define EXT4_MAX_CONTENTION		8
3353 #define EXT4_CONTENTION_THRESHOLD	2
3354 
3355 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
3356 					      ext4_group_t group)
3357 {
3358 	return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
3359 }
3360 
3361 /*
3362  * Returns true if the filesystem is busy enough that attempts to
3363  * access the block group locks has run into contention.
3364  */
3365 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
3366 {
3367 	return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
3368 }
3369 
3370 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
3371 {
3372 	spinlock_t *lock = ext4_group_lock_ptr(sb, group);
3373 	if (spin_trylock(lock))
3374 		/*
3375 		 * We're able to grab the lock right away, so drop the
3376 		 * lock contention counter.
3377 		 */
3378 		atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
3379 	else {
3380 		/*
3381 		 * The lock is busy, so bump the contention counter,
3382 		 * and then wait on the spin lock.
3383 		 */
3384 		atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
3385 				  EXT4_MAX_CONTENTION);
3386 		spin_lock(lock);
3387 	}
3388 }
3389 
3390 static inline void ext4_unlock_group(struct super_block *sb,
3391 					ext4_group_t group)
3392 {
3393 	spin_unlock(ext4_group_lock_ptr(sb, group));
3394 }
3395 
3396 #ifdef CONFIG_QUOTA
3397 static inline bool ext4_quota_capable(struct super_block *sb)
3398 {
3399 	return (test_opt(sb, QUOTA) || ext4_has_feature_quota(sb));
3400 }
3401 
3402 static inline bool ext4_is_quota_journalled(struct super_block *sb)
3403 {
3404 	struct ext4_sb_info *sbi = EXT4_SB(sb);
3405 
3406 	return (ext4_has_feature_quota(sb) ||
3407 		sbi->s_qf_names[USRQUOTA] || sbi->s_qf_names[GRPQUOTA]);
3408 }
3409 #endif
3410 
3411 /*
3412  * Block validity checking
3413  */
3414 #define ext4_check_indirect_blockref(inode, bh)				\
3415 	ext4_check_blockref(__func__, __LINE__, inode,			\
3416 			    (__le32 *)(bh)->b_data,			\
3417 			    EXT4_ADDR_PER_BLOCK((inode)->i_sb))
3418 
3419 #define ext4_ind_check_inode(inode)					\
3420 	ext4_check_blockref(__func__, __LINE__, inode,			\
3421 			    EXT4_I(inode)->i_data,			\
3422 			    EXT4_NDIR_BLOCKS)
3423 
3424 /*
3425  * Inodes and files operations
3426  */
3427 
3428 /* dir.c */
3429 extern const struct file_operations ext4_dir_operations;
3430 
3431 /* file.c */
3432 extern const struct inode_operations ext4_file_inode_operations;
3433 extern const struct file_operations ext4_file_operations;
3434 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
3435 
3436 /* inline.c */
3437 extern int ext4_get_max_inline_size(struct inode *inode);
3438 extern int ext4_find_inline_data_nolock(struct inode *inode);
3439 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
3440 				 unsigned int len);
3441 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
3442 
3443 extern int ext4_readpage_inline(struct inode *inode, struct page *page);
3444 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3445 					 struct inode *inode,
3446 					 loff_t pos, unsigned len,
3447 					 unsigned flags,
3448 					 struct page **pagep);
3449 extern int ext4_write_inline_data_end(struct inode *inode,
3450 				      loff_t pos, unsigned len,
3451 				      unsigned copied,
3452 				      struct page *page);
3453 extern struct buffer_head *
3454 ext4_journalled_write_inline_data(struct inode *inode,
3455 				  unsigned len,
3456 				  struct page *page);
3457 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
3458 					   struct inode *inode,
3459 					   loff_t pos, unsigned len,
3460 					   unsigned flags,
3461 					   struct page **pagep,
3462 					   void **fsdata);
3463 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
3464 					 unsigned len, unsigned copied,
3465 					 struct page *page);
3466 extern int ext4_try_add_inline_entry(handle_t *handle,
3467 				     struct ext4_filename *fname,
3468 				     struct inode *dir, struct inode *inode);
3469 extern int ext4_try_create_inline_dir(handle_t *handle,
3470 				      struct inode *parent,
3471 				      struct inode *inode);
3472 extern int ext4_read_inline_dir(struct file *filp,
3473 				struct dir_context *ctx,
3474 				int *has_inline_data);
3475 extern int ext4_inlinedir_to_tree(struct file *dir_file,
3476 				  struct inode *dir, ext4_lblk_t block,
3477 				  struct dx_hash_info *hinfo,
3478 				  __u32 start_hash, __u32 start_minor_hash,
3479 				  int *has_inline_data);
3480 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3481 					struct ext4_filename *fname,
3482 					struct ext4_dir_entry_2 **res_dir,
3483 					int *has_inline_data);
3484 extern int ext4_delete_inline_entry(handle_t *handle,
3485 				    struct inode *dir,
3486 				    struct ext4_dir_entry_2 *de_del,
3487 				    struct buffer_head *bh,
3488 				    int *has_inline_data);
3489 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3490 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3491 					struct ext4_dir_entry_2 **parent_de,
3492 					int *retval);
3493 extern int ext4_inline_data_fiemap(struct inode *inode,
3494 				   struct fiemap_extent_info *fieinfo,
3495 				   int *has_inline, __u64 start, __u64 len);
3496 
3497 struct iomap;
3498 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap);
3499 
3500 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3501 
3502 extern int ext4_convert_inline_data(struct inode *inode);
3503 
3504 static inline int ext4_has_inline_data(struct inode *inode)
3505 {
3506 	return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3507 	       EXT4_I(inode)->i_inline_off;
3508 }
3509 
3510 /* namei.c */
3511 extern const struct inode_operations ext4_dir_inode_operations;
3512 extern const struct inode_operations ext4_special_inode_operations;
3513 extern struct dentry *ext4_get_parent(struct dentry *child);
3514 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3515 				 struct ext4_dir_entry_2 *de,
3516 				 int blocksize, int csum_size,
3517 				 unsigned int parent_ino, int dotdot_real_len);
3518 extern void ext4_initialize_dirent_tail(struct buffer_head *bh,
3519 					unsigned int blocksize);
3520 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode,
3521 				      struct buffer_head *bh);
3522 extern int ext4_ci_compare(const struct inode *parent,
3523 			   const struct qstr *fname,
3524 			   const struct qstr *entry, bool quick);
3525 extern int __ext4_unlink(handle_t *handle, struct inode *dir, const struct qstr *d_name,
3526 			 struct inode *inode);
3527 extern int __ext4_link(struct inode *dir, struct inode *inode,
3528 		       struct dentry *dentry);
3529 
3530 #define S_SHIFT 12
3531 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = {
3532 	[S_IFREG >> S_SHIFT]	= EXT4_FT_REG_FILE,
3533 	[S_IFDIR >> S_SHIFT]	= EXT4_FT_DIR,
3534 	[S_IFCHR >> S_SHIFT]	= EXT4_FT_CHRDEV,
3535 	[S_IFBLK >> S_SHIFT]	= EXT4_FT_BLKDEV,
3536 	[S_IFIFO >> S_SHIFT]	= EXT4_FT_FIFO,
3537 	[S_IFSOCK >> S_SHIFT]	= EXT4_FT_SOCK,
3538 	[S_IFLNK >> S_SHIFT]	= EXT4_FT_SYMLINK,
3539 };
3540 
3541 static inline void ext4_set_de_type(struct super_block *sb,
3542 				struct ext4_dir_entry_2 *de,
3543 				umode_t mode) {
3544 	if (ext4_has_feature_filetype(sb))
3545 		de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3546 }
3547 
3548 /* readpages.c */
3549 extern int ext4_mpage_readpages(struct inode *inode,
3550 		struct readahead_control *rac, struct page *page);
3551 extern int __init ext4_init_post_read_processing(void);
3552 extern void ext4_exit_post_read_processing(void);
3553 
3554 /* symlink.c */
3555 extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3556 extern const struct inode_operations ext4_symlink_inode_operations;
3557 extern const struct inode_operations ext4_fast_symlink_inode_operations;
3558 
3559 /* sysfs.c */
3560 extern int ext4_register_sysfs(struct super_block *sb);
3561 extern void ext4_unregister_sysfs(struct super_block *sb);
3562 extern int __init ext4_init_sysfs(void);
3563 extern void ext4_exit_sysfs(void);
3564 
3565 /* block_validity */
3566 extern void ext4_release_system_zone(struct super_block *sb);
3567 extern int ext4_setup_system_zone(struct super_block *sb);
3568 extern int __init ext4_init_system_zone(void);
3569 extern void ext4_exit_system_zone(void);
3570 extern int ext4_inode_block_valid(struct inode *inode,
3571 				  ext4_fsblk_t start_blk,
3572 				  unsigned int count);
3573 extern int ext4_check_blockref(const char *, unsigned int,
3574 			       struct inode *, __le32 *, unsigned int);
3575 
3576 /* extents.c */
3577 struct ext4_ext_path;
3578 struct ext4_extent;
3579 
3580 /*
3581  * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3582  * __le32.
3583  */
3584 #define EXT_MAX_BLOCKS	0xffffffff
3585 
3586 extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode);
3587 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3588 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3589 			       struct ext4_map_blocks *map, int flags);
3590 extern int ext4_ext_truncate(handle_t *, struct inode *);
3591 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3592 				 ext4_lblk_t end);
3593 extern void ext4_ext_init(struct super_block *);
3594 extern void ext4_ext_release(struct super_block *);
3595 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3596 			  loff_t len);
3597 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3598 					  loff_t offset, ssize_t len);
3599 extern int ext4_convert_unwritten_io_end_vec(handle_t *handle,
3600 					     ext4_io_end_t *io_end);
3601 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3602 			   struct ext4_map_blocks *map, int flags);
3603 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3604 						   int num,
3605 						   struct ext4_ext_path *path);
3606 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3607 				  struct ext4_ext_path **,
3608 				  struct ext4_extent *, int);
3609 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3610 					      struct ext4_ext_path **,
3611 					      int flags);
3612 extern void ext4_ext_drop_refs(struct ext4_ext_path *);
3613 extern int ext4_ext_check_inode(struct inode *inode);
3614 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3615 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3616 			__u64 start, __u64 len);
3617 extern int ext4_get_es_cache(struct inode *inode,
3618 			     struct fiemap_extent_info *fieinfo,
3619 			     __u64 start, __u64 len);
3620 extern int ext4_ext_precache(struct inode *inode);
3621 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3622 				struct inode *inode2, ext4_lblk_t lblk1,
3623 			     ext4_lblk_t lblk2,  ext4_lblk_t count,
3624 			     int mark_unwritten,int *err);
3625 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu);
3626 extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode,
3627 				       int check_cred, int restart_cred,
3628 				       int revoke_cred);
3629 extern void ext4_ext_replay_shrink_inode(struct inode *inode, ext4_lblk_t end);
3630 extern int ext4_ext_replay_set_iblocks(struct inode *inode);
3631 extern int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start,
3632 		int len, int unwritten, ext4_fsblk_t pblk);
3633 extern int ext4_ext_clear_bb(struct inode *inode);
3634 
3635 
3636 /* move_extent.c */
3637 extern void ext4_double_down_write_data_sem(struct inode *first,
3638 					    struct inode *second);
3639 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3640 					  struct inode *donor_inode);
3641 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3642 			     __u64 start_orig, __u64 start_donor,
3643 			     __u64 len, __u64 *moved_len);
3644 
3645 /* page-io.c */
3646 extern int __init ext4_init_pageio(void);
3647 extern void ext4_exit_pageio(void);
3648 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3649 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3650 extern int ext4_put_io_end(ext4_io_end_t *io_end);
3651 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3652 extern void ext4_io_submit_init(struct ext4_io_submit *io,
3653 				struct writeback_control *wbc);
3654 extern void ext4_end_io_rsv_work(struct work_struct *work);
3655 extern void ext4_io_submit(struct ext4_io_submit *io);
3656 extern int ext4_bio_write_page(struct ext4_io_submit *io,
3657 			       struct page *page,
3658 			       int len,
3659 			       bool keep_towrite);
3660 extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end);
3661 extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end);
3662 
3663 /* mmp.c */
3664 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3665 
3666 /* verity.c */
3667 extern const struct fsverity_operations ext4_verityops;
3668 
3669 /*
3670  * Add new method to test whether block and inode bitmaps are properly
3671  * initialized. With uninit_bg reading the block from disk is not enough
3672  * to mark the bitmap uptodate. We need to also zero-out the bitmap
3673  */
3674 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3675 
3676 static inline int bitmap_uptodate(struct buffer_head *bh)
3677 {
3678 	return (buffer_uptodate(bh) &&
3679 			test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3680 }
3681 static inline void set_bitmap_uptodate(struct buffer_head *bh)
3682 {
3683 	set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3684 }
3685 
3686 #define in_range(b, first, len)	((b) >= (first) && (b) <= (first) + (len) - 1)
3687 
3688 /* For ioend & aio unwritten conversion wait queues */
3689 #define EXT4_WQ_HASH_SZ		37
3690 #define ext4_ioend_wq(v)   (&ext4__ioend_wq[((unsigned long)(v)) %\
3691 					    EXT4_WQ_HASH_SZ])
3692 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3693 
3694 extern int ext4_resize_begin(struct super_block *sb);
3695 extern void ext4_resize_end(struct super_block *sb);
3696 
3697 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
3698 					      struct ext4_io_end *io_end)
3699 {
3700 	if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
3701 		io_end->flag |= EXT4_IO_END_UNWRITTEN;
3702 		atomic_inc(&EXT4_I(inode)->i_unwritten);
3703 	}
3704 }
3705 
3706 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3707 {
3708 	struct inode *inode = io_end->inode;
3709 
3710 	if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
3711 		io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3712 		/* Wake up anyone waiting on unwritten extent conversion */
3713 		if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
3714 			wake_up_all(ext4_ioend_wq(inode));
3715 	}
3716 }
3717 
3718 extern const struct iomap_ops ext4_iomap_ops;
3719 extern const struct iomap_ops ext4_iomap_overwrite_ops;
3720 extern const struct iomap_ops ext4_iomap_report_ops;
3721 
3722 static inline int ext4_buffer_uptodate(struct buffer_head *bh)
3723 {
3724 	/*
3725 	 * If the buffer has the write error flag, we have failed
3726 	 * to write out data in the block.  In this  case, we don't
3727 	 * have to read the block because we may read the old data
3728 	 * successfully.
3729 	 */
3730 	if (!buffer_uptodate(bh) && buffer_write_io_error(bh))
3731 		set_buffer_uptodate(bh);
3732 	return buffer_uptodate(bh);
3733 }
3734 
3735 #endif	/* __KERNEL__ */
3736 
3737 #define EFSBADCRC	EBADMSG		/* Bad CRC detected */
3738 #define EFSCORRUPTED	EUCLEAN		/* Filesystem is corrupted */
3739 
3740 #endif	/* _EXT4_H */
3741