xref: /linux/fs/ext4/ext4.h (revision 9e8fc16dc6b474b9a87843c3d3ee12fc83f2bdcc)
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/refcount.h>
21 #include <linux/types.h>
22 #include <linux/blkdev.h>
23 #include <linux/iomap.h>
24 #include <linux/magic.h>
25 #include <linux/jbd2.h>
26 #include <linux/quota.h>
27 #include <linux/rwsem.h>
28 #include <linux/rbtree.h>
29 #include <linux/seqlock.h>
30 #include <linux/mutex.h>
31 #include <linux/timer.h>
32 #include <linux/sched/signal.h>
33 #include <linux/blockgroup_lock.h>
34 #include <linux/percpu_counter.h>
35 #include <linux/ratelimit.h>
36 #include <linux/crc32c.h>
37 #include <linux/falloc.h>
38 #include <linux/percpu-rwsem.h>
39 #include <linux/fiemap.h>
40 #ifdef __KERNEL__
41 #include <linux/compat.h>
42 #endif
43 #include <uapi/linux/ext4.h>
44 
45 #include <linux/fscrypt.h>
46 #include <linux/fsverity.h>
47 
48 #include <linux/compiler.h>
49 
50 /*
51  * The fourth extended filesystem constants/structures
52  */
53 
54 /*
55  * with AGGRESSIVE_CHECK allocator runs consistency checks over
56  * structures. these checks slow things down a lot
57  */
58 #define AGGRESSIVE_CHECK__
59 
60 /*
61  * with DOUBLE_CHECK defined mballoc creates persistent in-core
62  * bitmaps, maintains and uses them to check for double allocations
63  */
64 #define DOUBLE_CHECK__
65 
66 /*
67  * Define EXT4FS_DEBUG to produce debug messages
68  */
69 #undef EXT4FS_DEBUG
70 
71 /*
72  * Debug code
73  */
74 #ifdef EXT4FS_DEBUG
75 #define ext4_debug(f, a...)						\
76 	do {								\
77 		printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:",	\
78 			__FILE__, __LINE__, __func__);			\
79 		printk(KERN_DEBUG f, ## a);				\
80 	} while (0)
81 #else
82 #define ext4_debug(fmt, ...)	no_printk(fmt, ##__VA_ARGS__)
83 #endif
84 
85  /*
86   * Turn on EXT_DEBUG to enable ext4_ext_show_path/leaf/move in extents.c
87   */
88 #define EXT_DEBUG__
89 
90 /*
91  * Dynamic printk for controlled extents debugging.
92  */
93 #ifdef CONFIG_EXT4_DEBUG
94 #define ext_debug(ino, fmt, ...)					\
95 	pr_debug("[%s/%d] EXT4-fs (%s): ino %llu: (%s, %d): %s:" fmt,	\
96 		 current->comm, task_pid_nr(current),			\
97 		 ino->i_sb->s_id, ino->i_ino, __FILE__, __LINE__,	\
98 		 __func__, ##__VA_ARGS__)
99 #else
100 #define ext_debug(ino, fmt, ...)	no_printk(fmt, ##__VA_ARGS__)
101 #endif
102 
103 #define ASSERT(assert)						\
104 do {									\
105 	if (unlikely(!(assert))) {					\
106 		printk(KERN_EMERG					\
107 		       "Assertion failure in %s() at %s:%d: '%s'\n",	\
108 		       __func__, __FILE__, __LINE__, #assert);		\
109 		BUG();							\
110 	}								\
111 } while (0)
112 
113 /* data type for block offset of block group */
114 typedef int ext4_grpblk_t;
115 
116 /* data type for filesystem-wide blocks number */
117 typedef unsigned long long ext4_fsblk_t;
118 
119 /* data type for file logical block number */
120 typedef __u32 ext4_lblk_t;
121 
122 /* data type for block group number */
123 typedef unsigned int ext4_group_t;
124 
125 enum SHIFT_DIRECTION {
126 	SHIFT_LEFT = 0,
127 	SHIFT_RIGHT,
128 };
129 
130 /*
131  * For each criteria, mballoc has slightly different way of finding
132  * the required blocks nad usually, higher the criteria the slower the
133  * allocation.  We start at lower criterias and keep falling back to
134  * higher ones if we are not able to find any blocks.  Lower (earlier)
135  * criteria are faster.
136  */
137 enum criteria {
138 	/*
139 	 * Used when number of blocks needed is a power of 2. This
140 	 * doesn't trigger any disk IO except prefetch and is the
141 	 * fastest criteria.
142 	 */
143 	CR_POWER2_ALIGNED,
144 
145 	/*
146 	 * Tries to lookup in-memory data structures to find the most
147 	 * suitable group that satisfies goal request. No disk IO
148 	 * except block prefetch.
149 	 */
150 	CR_GOAL_LEN_FAST,
151 
152         /*
153 	 * Same as CR_GOAL_LEN_FAST but is allowed to reduce the goal
154          * length to the best available length for faster allocation.
155 	 */
156 	CR_BEST_AVAIL_LEN,
157 
158 	/*
159 	 * Reads each block group sequentially, performing disk IO if
160 	 * necessary, to find suitable block group. Tries to
161 	 * allocate goal length but might trim the request if nothing
162 	 * is found after enough tries.
163 	 */
164 	CR_GOAL_LEN_SLOW,
165 
166 	/*
167 	 * Finds the first free set of blocks and allocates
168 	 * those. This is only used in rare cases when
169 	 * CR_GOAL_LEN_SLOW also fails to allocate anything.
170 	 */
171 	CR_ANY_FREE,
172 
173 	/*
174 	 * Number of criterias defined.
175 	 */
176 	EXT4_MB_NUM_CRS
177 };
178 
179 /*
180  * Flags used in mballoc's allocation_context flags field.
181  *
182  * Also used to show what's going on for debugging purposes when the
183  * flag field is exported via the traceport interface
184  */
185 
186 /* prefer goal again. length */
187 #define EXT4_MB_HINT_MERGE		0x0001
188 /* first blocks in the file */
189 #define EXT4_MB_HINT_FIRST		0x0008
190 /* data is being allocated */
191 #define EXT4_MB_HINT_DATA		0x0020
192 /* don't preallocate (for tails) */
193 #define EXT4_MB_HINT_NOPREALLOC		0x0040
194 /* allocate for locality group */
195 #define EXT4_MB_HINT_GROUP_ALLOC	0x0080
196 /* allocate goal blocks or none */
197 #define EXT4_MB_HINT_GOAL_ONLY		0x0100
198 /* goal is meaningful */
199 #define EXT4_MB_HINT_TRY_GOAL		0x0200
200 /* blocks already pre-reserved by delayed allocation */
201 #define EXT4_MB_DELALLOC_RESERVED	0x0400
202 /* We are doing stream allocation */
203 #define EXT4_MB_STREAM_ALLOC		0x0800
204 /* Use reserved root blocks if needed */
205 #define EXT4_MB_USE_ROOT_BLOCKS		0x1000
206 /* Use blocks from reserved pool */
207 #define EXT4_MB_USE_RESERVED		0x2000
208 /* Do strict check for free blocks while retrying block allocation */
209 #define EXT4_MB_STRICT_CHECK		0x4000
210 
211 struct ext4_allocation_request {
212 	/* target inode for block we're allocating */
213 	struct inode *inode;
214 	/* how many blocks we want to allocate */
215 	unsigned int len;
216 	/* logical block in target inode */
217 	ext4_lblk_t logical;
218 	/* the closest logical allocated block to the left */
219 	ext4_lblk_t lleft;
220 	/* the closest logical allocated block to the right */
221 	ext4_lblk_t lright;
222 	/* phys. target (a hint) */
223 	ext4_fsblk_t goal;
224 	/* phys. block for the closest logical allocated block to the left */
225 	ext4_fsblk_t pleft;
226 	/* phys. block for the closest logical allocated block to the right */
227 	ext4_fsblk_t pright;
228 	/* flags. see above EXT4_MB_HINT_* */
229 	unsigned int flags;
230 };
231 
232 /*
233  * Logical to physical block mapping, used by ext4_map_blocks()
234  *
235  * This structure is used to pass requests into ext4_map_blocks() as
236  * well as to store the information returned by ext4_map_blocks().  It
237  * takes less room on the stack than a struct buffer_head.
238  */
239 #define EXT4_MAP_NEW		BIT(BH_New)
240 #define EXT4_MAP_MAPPED		BIT(BH_Mapped)
241 #define EXT4_MAP_UNWRITTEN	BIT(BH_Unwritten)
242 #define EXT4_MAP_BOUNDARY	BIT(BH_Boundary)
243 #define EXT4_MAP_DELAYED	BIT(BH_Delay)
244 /*
245  * This is for use in ext4_map_query_blocks() for a special case where we can
246  * have a physically and logically contiguous blocks split across two leaf
247  * nodes instead of a single extent. This is required in case of atomic writes
248  * to know whether the returned extent is last in leaf. If yes, then lookup for
249  * next in leaf block in ext4_map_query_blocks_next_in_leaf().
250  * - This is never going to be added to any buffer head state.
251  * - We use the next available bit after BH_BITMAP_UPTODATE.
252  */
253 #define EXT4_MAP_QUERY_LAST_IN_LEAF	BIT(BH_BITMAP_UPTODATE + 1)
254 #define EXT4_MAP_FLAGS		(EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
255 				 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\
256 				 EXT4_MAP_DELAYED | EXT4_MAP_QUERY_LAST_IN_LEAF)
257 
258 struct ext4_map_blocks {
259 	ext4_fsblk_t m_pblk;
260 	ext4_lblk_t m_lblk;
261 	unsigned int m_len;
262 	unsigned int m_flags;
263 	u64 m_seq;
264 };
265 
266 /*
267  * Block validity checking, system zone rbtree.
268  */
269 struct ext4_system_blocks {
270 	struct rb_root root;
271 	struct rcu_head rcu;
272 };
273 
274 /*
275  * Flags for ext4_io_end->flags
276  */
277 #define EXT4_IO_END_UNWRITTEN	0x0001
278 #define EXT4_IO_END_FAILED	0x0002
279 
280 #define EXT4_IO_END_DEFER_COMPLETION (EXT4_IO_END_UNWRITTEN | EXT4_IO_END_FAILED)
281 
282 struct ext4_io_end_vec {
283 	struct list_head list;		/* list of io_end_vec */
284 	loff_t offset;			/* offset in the file */
285 	ssize_t size;			/* size of the extent */
286 };
287 
288 /*
289  * For converting unwritten extents on a work queue. 'handle' is used for
290  * buffered writeback.
291  */
292 typedef struct ext4_io_end {
293 	struct list_head	list;		/* per-file finished IO list */
294 	handle_t		*handle;	/* handle reserved for extent
295 						 * conversion */
296 	struct inode		*inode;		/* file being written to */
297 	struct bio		*bio;		/* Linked list of completed
298 						 * bios covering the extent */
299 	unsigned int		flag;		/* unwritten or not */
300 	refcount_t		count;		/* reference counter */
301 	struct list_head	list_vec;	/* list of ext4_io_end_vec */
302 } ext4_io_end_t;
303 
304 struct ext4_io_submit {
305 	struct writeback_control *io_wbc;
306 	struct bio		*io_bio;
307 	ext4_io_end_t		*io_end;
308 	sector_t		io_next_block;
309 };
310 
311 /*
312  * Special inodes numbers
313  */
314 #define	EXT4_BAD_INO		 1	/* Bad blocks inode */
315 #define EXT4_ROOT_INO		 2	/* Root inode */
316 #define EXT4_USR_QUOTA_INO	 3	/* User quota inode */
317 #define EXT4_GRP_QUOTA_INO	 4	/* Group quota inode */
318 #define EXT4_BOOT_LOADER_INO	 5	/* Boot loader inode */
319 #define EXT4_UNDEL_DIR_INO	 6	/* Undelete directory inode */
320 #define EXT4_RESIZE_INO		 7	/* Reserved group descriptors inode */
321 #define EXT4_JOURNAL_INO	 8	/* Journal inode */
322 
323 /* First non-reserved inode for old ext4 filesystems */
324 #define EXT4_GOOD_OLD_FIRST_INO	11
325 
326 /*
327  * Maximal count of links to a file
328  */
329 #define EXT4_LINK_MAX		65000
330 
331 /*
332  * Macro-instructions used to manage several block sizes
333  */
334 #define EXT4_MIN_BLOCK_SIZE		1024
335 #define	EXT4_MAX_BLOCK_SIZE		65536
336 #define EXT4_MIN_BLOCK_LOG_SIZE		10
337 #define EXT4_MAX_BLOCK_LOG_SIZE		16
338 #define EXT4_MAX_CLUSTER_LOG_SIZE	28
339 #ifdef __KERNEL__
340 # define EXT4_BLOCK_SIZE(s)		((s)->s_blocksize)
341 #else
342 # define EXT4_BLOCK_SIZE(s)		(EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
343 #endif
344 #define	EXT4_ADDR_PER_BLOCK(s)		(EXT4_BLOCK_SIZE(s) / sizeof(__u32))
345 #define EXT4_CLUSTER_SIZE(s)		(EXT4_BLOCK_SIZE(s) << \
346 					 EXT4_SB(s)->s_cluster_bits)
347 #ifdef __KERNEL__
348 # define EXT4_BLOCK_SIZE_BITS(s)	((s)->s_blocksize_bits)
349 # define EXT4_CLUSTER_BITS(s)		(EXT4_SB(s)->s_cluster_bits)
350 #else
351 # define EXT4_BLOCK_SIZE_BITS(s)	((s)->s_log_block_size + 10)
352 #endif
353 #ifdef __KERNEL__
354 #define	EXT4_ADDR_PER_BLOCK_BITS(s)	(EXT4_SB(s)->s_addr_per_block_bits)
355 #define EXT4_INODE_SIZE(s)		(EXT4_SB(s)->s_inode_size)
356 #define EXT4_FIRST_INO(s)		(EXT4_SB(s)->s_first_ino)
357 #else
358 #define EXT4_INODE_SIZE(s)	(((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
359 				 EXT4_GOOD_OLD_INODE_SIZE : \
360 				 (s)->s_inode_size)
361 #define EXT4_FIRST_INO(s)	(((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
362 				 EXT4_GOOD_OLD_FIRST_INO : \
363 				 (s)->s_first_ino)
364 #endif
365 #define EXT4_BLOCK_ALIGN(size, blkbits)		ALIGN((size), (1 << (blkbits)))
366 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \
367 	((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \
368 								  blkbits))
369 #define EXT4_B_TO_LBLK(inode, offset) \
370 	(round_up((offset), i_blocksize(inode)) >> (inode)->i_blkbits)
371 #define EXT4_LBLK_TO_B(inode, lblk) ((loff_t)(lblk) << (inode)->i_blkbits)
372 
373 /* Translate a block number to a page index */
374 #define EXT4_LBLK_TO_PG(inode, lblk)	(EXT4_LBLK_TO_B((inode), (lblk)) >> \
375 					 PAGE_SHIFT)
376 /* Translate a page index to a block number */
377 #define EXT4_PG_TO_LBLK(inode, pnum)	(((loff_t)(pnum) << PAGE_SHIFT) >> \
378 					 (inode)->i_blkbits)
379 /* Translate a block number to a cluster number */
380 #define EXT4_B2C(sbi, blk)	((blk) >> (sbi)->s_cluster_bits)
381 /* Translate a cluster number to a block number */
382 #define EXT4_C2B(sbi, cluster)	((cluster) << (sbi)->s_cluster_bits)
383 /* Translate # of blks to # of clusters */
384 #define EXT4_NUM_B2C(sbi, blks)	(((blks) + (sbi)->s_cluster_ratio - 1) >> \
385 				 (sbi)->s_cluster_bits)
386 /* Mask out the low bits to get the starting block of the cluster */
387 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) &				\
388 				  ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
389 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) &				\
390 				  ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
391 /* Fill in the low bits to get the last block of the cluster */
392 #define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) |				\
393 				    ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1))
394 /* Get the cluster offset */
395 #define EXT4_PBLK_COFF(s, pblk) ((pblk) &				\
396 				 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
397 #define EXT4_LBLK_COFF(s, lblk) ((lblk) &				\
398 				 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
399 
400 /*
401  * Structure of a blocks group descriptor
402  */
403 struct ext4_group_desc
404 {
405 	__le32	bg_block_bitmap_lo;	/* Blocks bitmap block */
406 	__le32	bg_inode_bitmap_lo;	/* Inodes bitmap block */
407 	__le32	bg_inode_table_lo;	/* Inodes table block */
408 	__le16	bg_free_blocks_count_lo;/* Free blocks count */
409 	__le16	bg_free_inodes_count_lo;/* Free inodes count */
410 	__le16	bg_used_dirs_count_lo;	/* Directories count */
411 	__le16	bg_flags;		/* EXT4_BG_flags (INODE_UNINIT, etc) */
412 	__le32  bg_exclude_bitmap_lo;   /* Exclude bitmap for snapshots */
413 	__le16  bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
414 	__le16  bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
415 	__le16  bg_itable_unused_lo;	/* Unused inodes count */
416 	__le16  bg_checksum;		/* crc16(sb_uuid+group+desc) */
417 	__le32	bg_block_bitmap_hi;	/* Blocks bitmap block MSB */
418 	__le32	bg_inode_bitmap_hi;	/* Inodes bitmap block MSB */
419 	__le32	bg_inode_table_hi;	/* Inodes table block MSB */
420 	__le16	bg_free_blocks_count_hi;/* Free blocks count MSB */
421 	__le16	bg_free_inodes_count_hi;/* Free inodes count MSB */
422 	__le16	bg_used_dirs_count_hi;	/* Directories count MSB */
423 	__le16  bg_itable_unused_hi;    /* Unused inodes count MSB */
424 	__le32  bg_exclude_bitmap_hi;   /* Exclude bitmap block MSB */
425 	__le16  bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
426 	__le16  bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
427 	__u32   bg_reserved;
428 };
429 
430 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END	\
431 	(offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
432 	 sizeof(__le16))
433 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END	\
434 	(offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
435 	 sizeof(__le16))
436 
437 /*
438  * Structure of a flex block group info
439  */
440 
441 struct flex_groups {
442 	atomic64_t	free_clusters;
443 	atomic_t	free_inodes;
444 	atomic_t	used_dirs;
445 };
446 
447 #define EXT4_BG_INODE_UNINIT	0x0001 /* Inode table/bitmap not in use */
448 #define EXT4_BG_BLOCK_UNINIT	0x0002 /* Block bitmap not in use */
449 #define EXT4_BG_INODE_ZEROED	0x0004 /* On-disk itable initialized to zero */
450 
451 /*
452  * Macro-instructions used to manage group descriptors
453  */
454 #define EXT4_MIN_DESC_SIZE		32
455 #define EXT4_MIN_DESC_SIZE_64BIT	64
456 #define	EXT4_MAX_DESC_SIZE		EXT4_MIN_BLOCK_SIZE
457 #define EXT4_DESC_SIZE(s)		(EXT4_SB(s)->s_desc_size)
458 #ifdef __KERNEL__
459 # define EXT4_BLOCKS_PER_GROUP(s)	(EXT4_SB(s)->s_blocks_per_group)
460 # define EXT4_CLUSTERS_PER_GROUP(s)	(EXT4_SB(s)->s_clusters_per_group)
461 # define EXT4_DESC_PER_BLOCK(s)		(EXT4_SB(s)->s_desc_per_block)
462 # define EXT4_INODES_PER_GROUP(s)	(EXT4_SB(s)->s_inodes_per_group)
463 # define EXT4_DESC_PER_BLOCK_BITS(s)	(EXT4_SB(s)->s_desc_per_block_bits)
464 #else
465 # define EXT4_BLOCKS_PER_GROUP(s)	((s)->s_blocks_per_group)
466 # define EXT4_DESC_PER_BLOCK(s)		(EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
467 # define EXT4_INODES_PER_GROUP(s)	((s)->s_inodes_per_group)
468 #endif
469 
470 /*
471  * Constants relative to the data blocks
472  */
473 #define	EXT4_NDIR_BLOCKS		12
474 #define	EXT4_IND_BLOCK			EXT4_NDIR_BLOCKS
475 #define	EXT4_DIND_BLOCK			(EXT4_IND_BLOCK + 1)
476 #define	EXT4_TIND_BLOCK			(EXT4_DIND_BLOCK + 1)
477 #define	EXT4_N_BLOCKS			(EXT4_TIND_BLOCK + 1)
478 
479 /*
480  * Inode flags
481  */
482 #define	EXT4_SECRM_FL			0x00000001 /* Secure deletion */
483 #define	EXT4_UNRM_FL			0x00000002 /* Undelete */
484 #define	EXT4_COMPR_FL			0x00000004 /* Compress file */
485 #define EXT4_SYNC_FL			0x00000008 /* Synchronous updates */
486 #define EXT4_IMMUTABLE_FL		0x00000010 /* Immutable file */
487 #define EXT4_APPEND_FL			0x00000020 /* writes to file may only append */
488 #define EXT4_NODUMP_FL			0x00000040 /* do not dump file */
489 #define EXT4_NOATIME_FL			0x00000080 /* do not update atime */
490 /* Reserved for compression usage... */
491 #define EXT4_DIRTY_FL			0x00000100
492 #define EXT4_COMPRBLK_FL		0x00000200 /* One or more compressed clusters */
493 #define EXT4_NOCOMPR_FL			0x00000400 /* Don't compress */
494 	/* nb: was previously EXT2_ECOMPR_FL */
495 #define EXT4_ENCRYPT_FL			0x00000800 /* encrypted file */
496 /* End compression flags --- maybe not all used */
497 #define EXT4_INDEX_FL			0x00001000 /* hash-indexed directory */
498 #define EXT4_IMAGIC_FL			0x00002000 /* AFS directory */
499 #define EXT4_JOURNAL_DATA_FL		0x00004000 /* file data should be journaled */
500 #define EXT4_NOTAIL_FL			0x00008000 /* file tail should not be merged */
501 #define EXT4_DIRSYNC_FL			0x00010000 /* dirsync behaviour (directories only) */
502 #define EXT4_TOPDIR_FL			0x00020000 /* Top of directory hierarchies*/
503 #define EXT4_HUGE_FILE_FL               0x00040000 /* Set to each huge file */
504 #define EXT4_EXTENTS_FL			0x00080000 /* Inode uses extents */
505 #define EXT4_VERITY_FL			0x00100000 /* Verity protected inode */
506 #define EXT4_EA_INODE_FL	        0x00200000 /* Inode used for large EA */
507 /* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */
508 
509 #define EXT4_DAX_FL			0x02000000 /* Inode is DAX */
510 
511 #define EXT4_INLINE_DATA_FL		0x10000000 /* Inode has inline data. */
512 #define EXT4_PROJINHERIT_FL		0x20000000 /* Create with parents projid */
513 #define EXT4_CASEFOLD_FL		0x40000000 /* Casefolded directory */
514 #define EXT4_RESERVED_FL		0x80000000 /* reserved for ext4 lib */
515 
516 /* User modifiable flags */
517 #define EXT4_FL_USER_MODIFIABLE		(EXT4_SECRM_FL | \
518 					 EXT4_UNRM_FL | \
519 					 EXT4_COMPR_FL | \
520 					 EXT4_SYNC_FL | \
521 					 EXT4_IMMUTABLE_FL | \
522 					 EXT4_APPEND_FL | \
523 					 EXT4_NODUMP_FL | \
524 					 EXT4_NOATIME_FL | \
525 					 EXT4_JOURNAL_DATA_FL | \
526 					 EXT4_NOTAIL_FL | \
527 					 EXT4_DIRSYNC_FL | \
528 					 EXT4_TOPDIR_FL | \
529 					 EXT4_EXTENTS_FL | \
530 					 0x00400000 /* EXT4_EOFBLOCKS_FL */ | \
531 					 EXT4_DAX_FL | \
532 					 EXT4_PROJINHERIT_FL | \
533 					 EXT4_CASEFOLD_FL)
534 
535 /* User visible flags */
536 #define EXT4_FL_USER_VISIBLE		(EXT4_FL_USER_MODIFIABLE | \
537 					 EXT4_DIRTY_FL | \
538 					 EXT4_COMPRBLK_FL | \
539 					 EXT4_NOCOMPR_FL | \
540 					 EXT4_ENCRYPT_FL | \
541 					 EXT4_INDEX_FL | \
542 					 EXT4_VERITY_FL | \
543 					 EXT4_INLINE_DATA_FL)
544 
545 /* Flags that should be inherited by new inodes from their parent. */
546 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
547 			   EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
548 			   EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
549 			   EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
550 			   EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL |\
551 			   EXT4_DAX_FL)
552 
553 /* Flags that are appropriate for regular files (all but dir-specific ones). */
554 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\
555 			   EXT4_PROJINHERIT_FL))
556 
557 /* Flags that are appropriate for non-directories/regular files. */
558 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
559 
560 /* The only flags that should be swapped */
561 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL)
562 
563 /* Flags which are mutually exclusive to DAX */
564 #define EXT4_DAX_MUT_EXCL (EXT4_VERITY_FL | EXT4_ENCRYPT_FL |\
565 			   EXT4_JOURNAL_DATA_FL | EXT4_INLINE_DATA_FL)
566 
567 /* Mask out flags that are inappropriate for the given type of inode. */
568 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
569 {
570 	if (S_ISDIR(mode))
571 		return flags;
572 	else if (S_ISREG(mode))
573 		return flags & EXT4_REG_FLMASK;
574 	else
575 		return flags & EXT4_OTHER_FLMASK;
576 }
577 
578 /*
579  * Inode flags used for atomic set/get
580  */
581 enum {
582 	EXT4_INODE_SECRM	= 0,	/* Secure deletion */
583 	EXT4_INODE_UNRM		= 1,	/* Undelete */
584 	EXT4_INODE_COMPR	= 2,	/* Compress file */
585 	EXT4_INODE_SYNC		= 3,	/* Synchronous updates */
586 	EXT4_INODE_IMMUTABLE	= 4,	/* Immutable file */
587 	EXT4_INODE_APPEND	= 5,	/* writes to file may only append */
588 	EXT4_INODE_NODUMP	= 6,	/* do not dump file */
589 	EXT4_INODE_NOATIME	= 7,	/* do not update atime */
590 /* Reserved for compression usage... */
591 	EXT4_INODE_DIRTY	= 8,
592 	EXT4_INODE_COMPRBLK	= 9,	/* One or more compressed clusters */
593 	EXT4_INODE_NOCOMPR	= 10,	/* Don't compress */
594 	EXT4_INODE_ENCRYPT	= 11,	/* Encrypted file */
595 /* End compression flags --- maybe not all used */
596 	EXT4_INODE_INDEX	= 12,	/* hash-indexed directory */
597 	EXT4_INODE_IMAGIC	= 13,	/* AFS directory */
598 	EXT4_INODE_JOURNAL_DATA	= 14,	/* file data should be journaled */
599 	EXT4_INODE_NOTAIL	= 15,	/* file tail should not be merged */
600 	EXT4_INODE_DIRSYNC	= 16,	/* dirsync behaviour (directories only) */
601 	EXT4_INODE_TOPDIR	= 17,	/* Top of directory hierarchies*/
602 	EXT4_INODE_HUGE_FILE	= 18,	/* Set to each huge file */
603 	EXT4_INODE_EXTENTS	= 19,	/* Inode uses extents */
604 	EXT4_INODE_VERITY	= 20,	/* Verity protected inode */
605 	EXT4_INODE_EA_INODE	= 21,	/* Inode used for large EA */
606 /* 22 was formerly EXT4_INODE_EOFBLOCKS */
607 	EXT4_INODE_DAX		= 25,	/* Inode is DAX */
608 	EXT4_INODE_INLINE_DATA	= 28,	/* Data in inode. */
609 	EXT4_INODE_PROJINHERIT	= 29,	/* Create with parents projid */
610 	EXT4_INODE_CASEFOLD	= 30,	/* Casefolded directory */
611 	EXT4_INODE_RESERVED	= 31,	/* reserved for ext4 lib */
612 };
613 
614 /*
615  * Since it's pretty easy to mix up bit numbers and hex values, we use a
616  * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
617  * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
618  * any extra space in the compiled kernel image, otherwise, the build will fail.
619  * It's important that these values are the same, since we are using
620  * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
621  * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
622  * values found in ext2, ext3 and ext4 filesystems, and of course the values
623  * defined in e2fsprogs.
624  *
625  * It's not paranoia if the Murphy's Law really *is* out to get you.  :-)
626  */
627 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1U << EXT4_INODE_##FLAG))
628 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
629 
630 static inline void ext4_check_flag_values(void)
631 {
632 	CHECK_FLAG_VALUE(SECRM);
633 	CHECK_FLAG_VALUE(UNRM);
634 	CHECK_FLAG_VALUE(COMPR);
635 	CHECK_FLAG_VALUE(SYNC);
636 	CHECK_FLAG_VALUE(IMMUTABLE);
637 	CHECK_FLAG_VALUE(APPEND);
638 	CHECK_FLAG_VALUE(NODUMP);
639 	CHECK_FLAG_VALUE(NOATIME);
640 	CHECK_FLAG_VALUE(DIRTY);
641 	CHECK_FLAG_VALUE(COMPRBLK);
642 	CHECK_FLAG_VALUE(NOCOMPR);
643 	CHECK_FLAG_VALUE(ENCRYPT);
644 	CHECK_FLAG_VALUE(INDEX);
645 	CHECK_FLAG_VALUE(IMAGIC);
646 	CHECK_FLAG_VALUE(JOURNAL_DATA);
647 	CHECK_FLAG_VALUE(NOTAIL);
648 	CHECK_FLAG_VALUE(DIRSYNC);
649 	CHECK_FLAG_VALUE(TOPDIR);
650 	CHECK_FLAG_VALUE(HUGE_FILE);
651 	CHECK_FLAG_VALUE(EXTENTS);
652 	CHECK_FLAG_VALUE(VERITY);
653 	CHECK_FLAG_VALUE(EA_INODE);
654 	CHECK_FLAG_VALUE(INLINE_DATA);
655 	CHECK_FLAG_VALUE(PROJINHERIT);
656 	CHECK_FLAG_VALUE(CASEFOLD);
657 	CHECK_FLAG_VALUE(RESERVED);
658 }
659 
660 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
661 struct compat_ext4_new_group_input {
662 	u32 group;
663 	compat_u64 block_bitmap;
664 	compat_u64 inode_bitmap;
665 	compat_u64 inode_table;
666 	u32 blocks_count;
667 	u16 reserved_blocks;
668 	u16 unused;
669 };
670 #endif
671 
672 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
673 struct ext4_new_group_data {
674 	__u32 group;
675 	__u64 block_bitmap;
676 	__u64 inode_bitmap;
677 	__u64 inode_table;
678 	__u32 blocks_count;
679 	__u16 reserved_blocks;
680 	__u16 mdata_blocks;
681 	__u32 free_clusters_count;
682 };
683 
684 /* Indexes used to index group tables in ext4_new_group_data */
685 enum {
686 	BLOCK_BITMAP = 0,	/* block bitmap */
687 	INODE_BITMAP,		/* inode bitmap */
688 	INODE_TABLE,		/* inode tables */
689 	GROUP_TABLE_COUNT,
690 };
691 
692 /*
693  * Flags used by ext4_map_blocks()
694  */
695 	/* Allocate any needed blocks and/or convert an unwritten
696 	   extent to be an initialized ext4 */
697 #define EXT4_GET_BLOCKS_CREATE			0x0001
698 	/* Request the creation of an unwritten extent */
699 #define EXT4_GET_BLOCKS_UNWRIT_EXT		0x0002
700 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT	(EXT4_GET_BLOCKS_UNWRIT_EXT|\
701 						 EXT4_GET_BLOCKS_CREATE)
702 	/* Caller is from the delayed allocation writeout path
703 	 * finally doing the actual allocation of delayed blocks */
704 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE	0x0004
705 	/*
706 	 * This means that we cannot merge newly allocated extents, and if we
707 	 * found an unwritten extent, we need to split it.
708 	 */
709 #define EXT4_GET_BLOCKS_SPLIT_NOMERGE		0x0008
710 	/* Convert unwritten extent to initialized. */
711 #define EXT4_GET_BLOCKS_CONVERT			0x0010
712 	/* Eventual metadata allocation (due to growing extent tree)
713 	 * should not fail, so try to use reserved blocks for that.*/
714 #define EXT4_GET_BLOCKS_METADATA_NOFAIL		0x0020
715 	/* Don't normalize allocation size (used for fallocate) */
716 #define EXT4_GET_BLOCKS_NO_NORMALIZE		0x0040
717 	/* Convert written extents to unwritten */
718 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN	0x0100
719 	/* Write zeros to newly created written extents */
720 #define EXT4_GET_BLOCKS_ZERO			0x0200
721 #define EXT4_GET_BLOCKS_CREATE_ZERO		(EXT4_GET_BLOCKS_CREATE |\
722 					EXT4_GET_BLOCKS_ZERO)
723 	/* Caller is in the context of data submission, such as writeback,
724 	 * fsync, etc. Especially, in the generic writeback path, caller will
725 	 * submit data before dropping transaction handle. This allows jbd2
726 	 * to avoid submitting data before commit. */
727 #define EXT4_GET_BLOCKS_IO_SUBMIT		0x0400
728 	/* Convert extent to initialized after IO complete */
729 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT		(EXT4_GET_BLOCKS_CONVERT |\
730 					 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT |\
731 					 EXT4_GET_BLOCKS_IO_SUBMIT)
732 	/* Caller is in the atomic contex, find extent if it has been cached */
733 #define EXT4_GET_BLOCKS_CACHED_NOWAIT		0x0800
734 /*
735  * Atomic write caller needs this to query in the slow path of mixed mapping
736  * case, when a contiguous extent can be split across two adjacent leaf nodes.
737  * Look EXT4_MAP_QUERY_LAST_IN_LEAF.
738  */
739 #define EXT4_GET_BLOCKS_QUERY_LAST_IN_LEAF	0x1000
740 
741 /*
742  * The bit position of these flags must not overlap with any of the
743  * EXT4_GET_BLOCKS_*.  They are used by ext4_find_extent(),
744  * read_extent_tree_block(), ext4_split_extent_at(),
745  * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
746  * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
747  * caching the extents when reading from the extent tree while a
748  * truncate or punch hole operation is in progress.
749  */
750 #define EXT4_EX_NOCACHE				0x40000000
751 #define EXT4_EX_FORCE_CACHE			0x20000000
752 #define EXT4_EX_NOFAIL				0x10000000
753 /*
754  * ext4_map_query_blocks() uses this filter mask to filter the flags needed to
755  * pass while lookup/querying of on disk extent tree.
756  */
757 #define EXT4_EX_QUERY_FILTER	(EXT4_EX_NOCACHE | EXT4_EX_FORCE_CACHE |\
758 				 EXT4_EX_NOFAIL |\
759 				 EXT4_GET_BLOCKS_QUERY_LAST_IN_LEAF)
760 
761 /*
762  * Flags used by ext4_free_blocks
763  */
764 #define EXT4_FREE_BLOCKS_METADATA		0x0001
765 #define EXT4_FREE_BLOCKS_FORGET			0x0002
766 #define EXT4_FREE_BLOCKS_VALIDATED		0x0004
767 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE		0x0008
768 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER	0x0010
769 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER	0x0020
770 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER      0x0040
771 
772 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
773 /*
774  * ioctl commands in 32 bit emulation
775  */
776 #define EXT4_IOC32_GETVERSION		_IOR('f', 3, int)
777 #define EXT4_IOC32_SETVERSION		_IOW('f', 4, int)
778 #define EXT4_IOC32_GETRSVSZ		_IOR('f', 5, int)
779 #define EXT4_IOC32_SETRSVSZ		_IOW('f', 6, int)
780 #define EXT4_IOC32_GROUP_EXTEND		_IOW('f', 7, unsigned int)
781 #define EXT4_IOC32_GROUP_ADD		_IOW('f', 8, struct compat_ext4_new_group_input)
782 #define EXT4_IOC32_GETVERSION_OLD	FS_IOC32_GETVERSION
783 #define EXT4_IOC32_SETVERSION_OLD	FS_IOC32_SETVERSION
784 #endif
785 
786 /* Max physical block we can address w/o extents */
787 #define EXT4_MAX_BLOCK_FILE_PHYS	0xFFFFFFFF
788 
789 /* Max logical block we can support */
790 #define EXT4_MAX_LOGICAL_BLOCK		0xFFFFFFFE
791 
792 /*
793  * Structure of an inode on the disk
794  */
795 struct ext4_inode {
796 	__le16	i_mode;		/* File mode */
797 	__le16	i_uid;		/* Low 16 bits of Owner Uid */
798 	__le32	i_size_lo;	/* Size in bytes */
799 	__le32	i_atime;	/* Access time */
800 	__le32	i_ctime;	/* Inode Change time */
801 	__le32	i_mtime;	/* Modification time */
802 	__le32	i_dtime;	/* Deletion Time */
803 	__le16	i_gid;		/* Low 16 bits of Group Id */
804 	__le16	i_links_count;	/* Links count */
805 	__le32	i_blocks_lo;	/* Blocks count */
806 	__le32	i_flags;	/* File flags */
807 	union {
808 		struct {
809 			__le32  l_i_version;
810 		} linux1;
811 		struct {
812 			__u32  h_i_translator;
813 		} hurd1;
814 		struct {
815 			__u32  m_i_reserved1;
816 		} masix1;
817 	} osd1;				/* OS dependent 1 */
818 	__le32	i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
819 	__le32	i_generation;	/* File version (for NFS) */
820 	__le32	i_file_acl_lo;	/* File ACL */
821 	__le32	i_size_high;
822 	__le32	i_obso_faddr;	/* Obsoleted fragment address */
823 	union {
824 		struct {
825 			__le16	l_i_blocks_high; /* were l_i_reserved1 */
826 			__le16	l_i_file_acl_high;
827 			__le16	l_i_uid_high;	/* these 2 fields */
828 			__le16	l_i_gid_high;	/* were reserved2[0] */
829 			__le16	l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
830 			__le16	l_i_reserved;
831 		} linux2;
832 		struct {
833 			__le16	h_i_reserved1;	/* Obsoleted fragment number/size which are removed in ext4 */
834 			__u16	h_i_mode_high;
835 			__u16	h_i_uid_high;
836 			__u16	h_i_gid_high;
837 			__u32	h_i_author;
838 		} hurd2;
839 		struct {
840 			__le16	h_i_reserved1;	/* Obsoleted fragment number/size which are removed in ext4 */
841 			__le16	m_i_file_acl_high;
842 			__u32	m_i_reserved2[2];
843 		} masix2;
844 	} osd2;				/* OS dependent 2 */
845 	__le16	i_extra_isize;
846 	__le16	i_checksum_hi;	/* crc32c(uuid+inum+inode) BE */
847 	__le32  i_ctime_extra;  /* extra Change time      (nsec << 2 | epoch) */
848 	__le32  i_mtime_extra;  /* extra Modification time(nsec << 2 | epoch) */
849 	__le32  i_atime_extra;  /* extra Access time      (nsec << 2 | epoch) */
850 	__le32  i_crtime;       /* File Creation time */
851 	__le32  i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
852 	__le32  i_version_hi;	/* high 32 bits for 64-bit version */
853 	__le32	i_projid;	/* Project ID */
854 };
855 
856 #define EXT4_EPOCH_BITS 2
857 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
858 #define EXT4_NSEC_MASK  (~0UL << EXT4_EPOCH_BITS)
859 
860 /*
861  * Extended fields will fit into an inode if the filesystem was formatted
862  * with large inodes (-I 256 or larger) and there are not currently any EAs
863  * consuming all of the available space. For new inodes we always reserve
864  * enough space for the kernel's known extended fields, but for inodes
865  * created with an old kernel this might not have been the case. None of
866  * the extended inode fields is critical for correct filesystem operation.
867  * This macro checks if a certain field fits in the inode. Note that
868  * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
869  */
870 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field)	\
871 	((offsetof(typeof(*ext4_inode), field) +	\
872 	  sizeof((ext4_inode)->field))			\
873 	<= (EXT4_GOOD_OLD_INODE_SIZE +			\
874 	    (einode)->i_extra_isize))			\
875 
876 /*
877  * We use an encoding that preserves the times for extra epoch "00":
878  *
879  * extra  msb of                         adjust for signed
880  * epoch  32-bit                         32-bit tv_sec to
881  * bits   time    decoded 64-bit tv_sec  64-bit tv_sec      valid time range
882  * 0 0    1    -0x80000000..-0x00000001  0x000000000 1901-12-13..1969-12-31
883  * 0 0    0    0x000000000..0x07fffffff  0x000000000 1970-01-01..2038-01-19
884  * 0 1    1    0x080000000..0x0ffffffff  0x100000000 2038-01-19..2106-02-07
885  * 0 1    0    0x100000000..0x17fffffff  0x100000000 2106-02-07..2174-02-25
886  * 1 0    1    0x180000000..0x1ffffffff  0x200000000 2174-02-25..2242-03-16
887  * 1 0    0    0x200000000..0x27fffffff  0x200000000 2242-03-16..2310-04-04
888  * 1 1    1    0x280000000..0x2ffffffff  0x300000000 2310-04-04..2378-04-22
889  * 1 1    0    0x300000000..0x37fffffff  0x300000000 2378-04-22..2446-05-10
890  *
891  * Note that previous versions of the kernel on 64-bit systems would
892  * incorrectly use extra epoch bits 1,1 for dates between 1901 and
893  * 1970.  e2fsck will correct this, assuming that it is run on the
894  * affected filesystem before 2242.
895  */
896 
897 static inline __le32 ext4_encode_extra_time(struct timespec64 ts)
898 {
899 	u32 extra = ((ts.tv_sec - (s32)ts.tv_sec) >> 32) & EXT4_EPOCH_MASK;
900 	return cpu_to_le32(extra | (ts.tv_nsec << EXT4_EPOCH_BITS));
901 }
902 
903 static inline struct timespec64 ext4_decode_extra_time(__le32 base,
904 						       __le32 extra)
905 {
906 	struct timespec64 ts = { .tv_sec = (signed)le32_to_cpu(base) };
907 
908 	if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK)))
909 		ts.tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
910 	ts.tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
911 	return ts;
912 }
913 
914 #define EXT4_INODE_SET_XTIME_VAL(xtime, inode, raw_inode, ts)			\
915 do {										\
916 	if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {	\
917 		(raw_inode)->xtime = cpu_to_le32((ts).tv_sec);			\
918 		(raw_inode)->xtime ## _extra = ext4_encode_extra_time(ts);	\
919 	} else									\
920 		(raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (ts).tv_sec, S32_MIN, S32_MAX));	\
921 } while (0)
922 
923 #define EXT4_INODE_SET_ATIME(inode, raw_inode)						\
924 	EXT4_INODE_SET_XTIME_VAL(i_atime, inode, raw_inode, inode_get_atime(inode))
925 
926 #define EXT4_INODE_SET_MTIME(inode, raw_inode)						\
927 	EXT4_INODE_SET_XTIME_VAL(i_mtime, inode, raw_inode, inode_get_mtime(inode))
928 
929 #define EXT4_INODE_SET_CTIME(inode, raw_inode)						\
930 	EXT4_INODE_SET_XTIME_VAL(i_ctime, inode, raw_inode, inode_get_ctime(inode))
931 
932 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode)				\
933 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))			\
934 		EXT4_INODE_SET_XTIME_VAL(xtime, &((einode)->vfs_inode),		\
935 					 raw_inode, (einode)->xtime)
936 
937 #define EXT4_INODE_GET_XTIME_VAL(xtime, inode, raw_inode)			\
938 	(EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra) ?	\
939 		ext4_decode_extra_time((raw_inode)->xtime,				\
940 				       (raw_inode)->xtime ## _extra) :		\
941 		(struct timespec64) {						\
942 			.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime)	\
943 		})
944 
945 #define EXT4_INODE_GET_ATIME(inode, raw_inode)					\
946 do {										\
947 	inode_set_atime_to_ts(inode,						\
948 		EXT4_INODE_GET_XTIME_VAL(i_atime, inode, raw_inode));		\
949 } while (0)
950 
951 #define EXT4_INODE_GET_MTIME(inode, raw_inode)					\
952 do {										\
953 	inode_set_mtime_to_ts(inode,						\
954 		EXT4_INODE_GET_XTIME_VAL(i_mtime, inode, raw_inode));		\
955 } while (0)
956 
957 #define EXT4_INODE_GET_CTIME(inode, raw_inode)					\
958 do {										\
959 	inode_set_ctime_to_ts(inode,						\
960 		EXT4_INODE_GET_XTIME_VAL(i_ctime, inode, raw_inode));		\
961 } while (0)
962 
963 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode)				\
964 do {										\
965 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) 			\
966 		(einode)->xtime =						\
967 			EXT4_INODE_GET_XTIME_VAL(xtime, &(einode->vfs_inode),	\
968 						 raw_inode);			\
969 	else									\
970 		(einode)->xtime = (struct timespec64){0, 0};			\
971 } while (0)
972 
973 #define i_disk_version osd1.linux1.l_i_version
974 
975 #if defined(__KERNEL__) || defined(__linux__)
976 #define i_reserved1	osd1.linux1.l_i_reserved1
977 #define i_file_acl_high	osd2.linux2.l_i_file_acl_high
978 #define i_blocks_high	osd2.linux2.l_i_blocks_high
979 #define i_uid_low	i_uid
980 #define i_gid_low	i_gid
981 #define i_uid_high	osd2.linux2.l_i_uid_high
982 #define i_gid_high	osd2.linux2.l_i_gid_high
983 #define i_checksum_lo	osd2.linux2.l_i_checksum_lo
984 
985 #elif defined(__GNU__)
986 
987 #define i_translator	osd1.hurd1.h_i_translator
988 #define i_uid_high	osd2.hurd2.h_i_uid_high
989 #define i_gid_high	osd2.hurd2.h_i_gid_high
990 #define i_author	osd2.hurd2.h_i_author
991 
992 #elif defined(__masix__)
993 
994 #define i_reserved1	osd1.masix1.m_i_reserved1
995 #define i_file_acl_high	osd2.masix2.m_i_file_acl_high
996 #define i_reserved2	osd2.masix2.m_i_reserved2
997 
998 #endif /* defined(__KERNEL__) || defined(__linux__) */
999 
1000 #include "extents_status.h"
1001 #include "fast_commit.h"
1002 
1003 /*
1004  * Lock subclasses for i_data_sem in the ext4_inode_info structure.
1005  *
1006  * These are needed to avoid lockdep false positives when we need to
1007  * allocate blocks to the quota inode during ext4_map_blocks(), while
1008  * holding i_data_sem for a normal (non-quota) inode.  Since we don't
1009  * do quota tracking for the quota inode, this avoids deadlock (as
1010  * well as infinite recursion, since it isn't turtles all the way
1011  * down...)
1012  *
1013  *  I_DATA_SEM_NORMAL - Used for most inodes
1014  *  I_DATA_SEM_OTHER  - Used by move_inode.c for the second normal inode
1015  *			  where the second inode has larger inode number
1016  *			  than the first
1017  *  I_DATA_SEM_QUOTA  - Used for quota inodes only
1018  *  I_DATA_SEM_EA     - Used for ea_inodes only
1019  *  I_DATA_SEM_JOURNAL - Used for journal inode only
1020  */
1021 enum {
1022 	I_DATA_SEM_NORMAL = 0,
1023 	I_DATA_SEM_OTHER,
1024 	I_DATA_SEM_QUOTA,
1025 	I_DATA_SEM_EA,
1026 	I_DATA_SEM_JOURNAL
1027 };
1028 
1029 struct ext4_fc_inode_snap;
1030 
1031 /*
1032  * Snapshot failure reasons for ext4_fc_lock_updates tracepoint.
1033  * Keep these stable for tooling.
1034  */
1035 enum ext4_fc_snap_err {
1036 	EXT4_FC_SNAP_ERR_NONE = 0,
1037 	EXT4_FC_SNAP_ERR_ES_MISS,
1038 	EXT4_FC_SNAP_ERR_ES_DELAYED,
1039 	EXT4_FC_SNAP_ERR_ES_OTHER,
1040 	EXT4_FC_SNAP_ERR_INODES_CAP,
1041 	EXT4_FC_SNAP_ERR_RANGES_CAP,
1042 	EXT4_FC_SNAP_ERR_NOMEM,
1043 	EXT4_FC_SNAP_ERR_INODE_LOC,
1044 };
1045 
1046 /*
1047  * fourth extended file system inode data in memory
1048  */
1049 struct ext4_inode_info {
1050 	__le32	i_data[15];	/* unconverted */
1051 	__u32	i_dtime;
1052 	ext4_fsblk_t	i_file_acl;
1053 
1054 	/*
1055 	 * i_block_group is the number of the block group which contains
1056 	 * this file's inode.  Constant across the lifetime of the inode,
1057 	 * it is used for making block allocation decisions - we try to
1058 	 * place a file's data blocks near its inode block, and new inodes
1059 	 * near to their parent directory's inode.
1060 	 */
1061 	ext4_group_t	i_block_group;
1062 	ext4_lblk_t	i_dir_start_lookup;
1063 #if (BITS_PER_LONG < 64)
1064 	unsigned long	i_state_flags;		/* Dynamic state flags */
1065 #endif
1066 	unsigned long	i_flags;
1067 
1068 	/*
1069 	 * Extended attributes can be read independently of the main file
1070 	 * data. Taking i_rwsem even when reading would cause contention
1071 	 * between readers of EAs and writers of regular file data, so
1072 	 * instead we synchronize on xattr_sem when reading or changing
1073 	 * EAs.
1074 	 *
1075 	 * EA inodes (EXT4_EA_INODE_FL) do not use xattr_sem; they reuse
1076 	 * the space for deferred iput linkage.
1077 	 */
1078 	union {
1079 		struct rw_semaphore xattr_sem;
1080 		struct llist_node i_ea_iput_node;
1081 	};
1082 
1083 	/*
1084 	 * Inodes with EXT4_STATE_ORPHAN_FILE use i_orphan_idx. Otherwise
1085 	 * i_orphan is used.
1086 	 */
1087 	union {
1088 		struct list_head i_orphan;	/* unlinked but open inodes */
1089 		unsigned int i_orphan_idx;	/* Index in orphan file */
1090 	};
1091 
1092 	/* Fast commit related info */
1093 
1094 	/* For tracking dentry create updates */
1095 	struct list_head i_fc_dilist;
1096 	struct list_head i_fc_list;	/*
1097 					 * inodes that need fast commit
1098 					 * protected by sbi->s_fc_lock.
1099 					 */
1100 
1101 	/* Start of lblk range that needs to be committed in this fast commit */
1102 	ext4_lblk_t i_fc_lblk_start;
1103 
1104 	/* End of lblk range that needs to be committed in this fast commit */
1105 	ext4_lblk_t i_fc_lblk_len;
1106 
1107 	/*
1108 	 * Commit-time fast commit snapshots.
1109 	 *
1110 	 * i_fc_snap is installed and freed under sbi->s_fc_lock. The fast
1111 	 * commit log writing path reads the snapshot under sbi->s_fc_lock while
1112 	 * serializing fast commit TLVs.
1113 	 *
1114 	 * The snapshot lifetime is bounded by EXT4_STATE_FC_COMMITTING and the
1115 	 * corresponding cleanup / eviction paths.
1116 	 *
1117 	 * i_fc_snap points to per-inode snapshot data for fast commit:
1118 	 * - a raw inode snapshot for EXT4_FC_TAG_INODE
1119 	 * - data range records for EXT4_FC_TAG_{ADD,DEL}_RANGE
1120 	 */
1121 	struct ext4_fc_inode_snap *i_fc_snap;
1122 
1123 	spinlock_t i_raw_lock;	/* protects updates to the raw inode */
1124 
1125 	/*
1126 	 * Protect concurrent accesses on i_fc_lblk_start, i_fc_lblk_len
1127 	 * and inode's EXT4_FC_STATE_COMMITTING state bit.
1128 	 */
1129 	spinlock_t i_fc_lock;
1130 
1131 	/*
1132 	 * i_disksize keeps track of what the inode size is ON DISK, not
1133 	 * in memory.  During truncate, i_size is set to the new size by
1134 	 * the VFS prior to calling ext4_truncate(), but the filesystem won't
1135 	 * set i_disksize to 0 until the truncate is actually under way.
1136 	 *
1137 	 * The intent is that i_disksize always represents the blocks which
1138 	 * are used by this file.  This allows recovery to restart truncate
1139 	 * on orphans if we crash during truncate.  We actually write i_disksize
1140 	 * into the on-disk inode when writing inodes out, instead of i_size.
1141 	 *
1142 	 * The only time when i_disksize and i_size may be different is when
1143 	 * a truncate is in progress.  The only things which change i_disksize
1144 	 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
1145 	 */
1146 	loff_t	i_disksize;
1147 
1148 	/*
1149 	 * i_data_sem is for serialising ext4_truncate() against
1150 	 * ext4_getblock().  In the 2.4 ext2 design, great chunks of inode's
1151 	 * data tree are chopped off during truncate. We can't do that in
1152 	 * ext4 because whenever we perform intermediate commits during
1153 	 * truncate, the inode and all the metadata blocks *must* be in a
1154 	 * consistent state which allows truncation of the orphans to restart
1155 	 * during recovery.  Hence we must fix the get_block-vs-truncate race
1156 	 * by other means, so we have i_data_sem.
1157 	 */
1158 	struct rw_semaphore i_data_sem;
1159 	struct inode vfs_inode;
1160 	struct jbd2_inode *jinode;
1161 	struct mapping_metadata_bhs *i_metadata_bhs;
1162 
1163 	/*
1164 	 * File creation time. Its function is same as that of
1165 	 * struct timespec64 i_{a,c,m}time in the generic inode.
1166 	 */
1167 	struct timespec64 i_crtime;
1168 
1169 	/* mballoc */
1170 	atomic_t i_prealloc_active;
1171 
1172 	/* allocation reservation info for delalloc */
1173 	/* In case of bigalloc, this refer to clusters rather than blocks */
1174 	unsigned int i_reserved_data_blocks;
1175 	struct rb_root i_prealloc_node;
1176 	rwlock_t i_prealloc_lock;
1177 
1178 	/* extents status tree */
1179 	struct ext4_es_tree i_es_tree;
1180 	rwlock_t i_es_lock;
1181 	struct list_head i_es_list;
1182 	unsigned int i_es_all_nr;	/* protected by i_es_lock */
1183 	unsigned int i_es_shk_nr;	/* protected by i_es_lock */
1184 	ext4_lblk_t i_es_shrink_lblk;	/* Offset where we start searching for
1185 					   extents to shrink. Protected by
1186 					   i_es_lock  */
1187 	u64 i_es_seq;			/* Change counter for extents.
1188 					   Protected by i_es_lock */
1189 
1190 	/* ialloc */
1191 	ext4_group_t	i_last_alloc_group;
1192 
1193 	/* pending cluster reservations for bigalloc file systems */
1194 	struct ext4_pending_tree i_pending_tree;
1195 
1196 	/* on-disk additional length */
1197 	__u16 i_extra_isize;
1198 
1199 	/* Indicate the inline data space. */
1200 	u16 i_inline_off;
1201 	u16 i_inline_size;
1202 
1203 #ifdef CONFIG_QUOTA
1204 	/* quota space reservation, managed internally by quota code */
1205 	qsize_t i_reserved_quota;
1206 #endif
1207 	spinlock_t i_block_reservation_lock;
1208 
1209 	/* Lock protecting lists below */
1210 	spinlock_t i_completed_io_lock;
1211 	/*
1212 	 * Completed IOs that need unwritten extents handling and have a
1213 	 * transaction reserved for the buffer_head writeback path, and
1214 	 * also used by the iomap writeback path to queue ioends needing
1215 	 * unwritten extents conversion, i_disksize update, etc.
1216 	 */
1217 	struct list_head i_rsv_conversion_list;
1218 	struct work_struct i_rsv_conversion_work;
1219 
1220 	/*
1221 	 * Transactions that contain inode's metadata needed to complete
1222 	 * fsync and fdatasync, respectively.
1223 	 */
1224 	tid_t i_sync_tid;
1225 	tid_t i_datasync_tid;
1226 
1227 #ifdef CONFIG_QUOTA
1228 	struct dquot __rcu *i_dquot[MAXQUOTAS];
1229 #endif
1230 
1231 	/* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1232 	__u32 i_csum_seed;
1233 
1234 	kprojid_t i_projid;
1235 
1236 #ifdef CONFIG_FS_ENCRYPTION
1237 	struct fscrypt_inode_info *i_crypt_info;
1238 #endif
1239 };
1240 
1241 /*
1242  * File system states
1243  */
1244 #define	EXT4_VALID_FS			0x0001	/* Unmounted cleanly */
1245 #define	EXT4_ERROR_FS			0x0002	/* Errors detected */
1246 #define	EXT4_ORPHAN_FS			0x0004	/* Orphans being recovered */
1247 #define EXT4_FC_REPLAY			0x0020	/* Fast commit replay ongoing */
1248 
1249 /*
1250  * Misc. filesystem flags
1251  */
1252 #define EXT2_FLAGS_SIGNED_HASH		0x0001  /* Signed dirhash in use */
1253 #define EXT2_FLAGS_UNSIGNED_HASH	0x0002  /* Unsigned dirhash in use */
1254 #define EXT2_FLAGS_TEST_FILESYS		0x0004	/* to test development code */
1255 
1256 /*
1257  * Mount flags set via mount options or defaults
1258  */
1259 #define EXT4_MOUNT_NO_MBCACHE		0x00001 /* Do not use mbcache */
1260 #define EXT4_MOUNT_GRPID		0x00004	/* Create files with directory's group */
1261 #define EXT4_MOUNT_DEBUG		0x00008	/* Some debugging messages */
1262 #define EXT4_MOUNT_ERRORS_CONT		0x00010	/* Continue on errors */
1263 #define EXT4_MOUNT_ERRORS_RO		0x00020	/* Remount fs ro on errors */
1264 #define EXT4_MOUNT_ERRORS_PANIC		0x00040	/* Panic on errors */
1265 #define EXT4_MOUNT_ERRORS_MASK		0x00070
1266 #define EXT4_MOUNT_MINIX_DF		0x00080	/* Mimics the Minix statfs */
1267 #define EXT4_MOUNT_NOLOAD		0x00100	/* Don't use existing journal*/
1268 #ifdef CONFIG_FS_DAX
1269 #define EXT4_MOUNT_DAX_ALWAYS		0x00200	/* Direct Access */
1270 #else
1271 #define EXT4_MOUNT_DAX_ALWAYS		0
1272 #endif
1273 #define EXT4_MOUNT_DATA_FLAGS		0x00C00	/* Mode for data writes: */
1274 #define EXT4_MOUNT_JOURNAL_DATA		0x00400	/* Write data to journal */
1275 #define EXT4_MOUNT_ORDERED_DATA		0x00800	/* Flush data before commit */
1276 #define EXT4_MOUNT_WRITEBACK_DATA	0x00C00	/* No data ordering */
1277 #define EXT4_MOUNT_UPDATE_JOURNAL	0x01000	/* Update the journal format */
1278 #define EXT4_MOUNT_NO_UID32		0x02000  /* Disable 32-bit UIDs */
1279 #define EXT4_MOUNT_XATTR_USER		0x04000	/* Extended user attributes */
1280 #define EXT4_MOUNT_POSIX_ACL		0x08000	/* POSIX Access Control Lists */
1281 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC	0x10000	/* No auto delalloc mapping */
1282 #define EXT4_MOUNT_BARRIER		0x20000 /* Use block barriers */
1283 #define EXT4_MOUNT_QUOTA		0x40000 /* Some quota option set */
1284 #define EXT4_MOUNT_USRQUOTA		0x80000 /* "old" user quota,
1285 						 * enable enforcement for hidden
1286 						 * quota files */
1287 #define EXT4_MOUNT_GRPQUOTA		0x100000 /* "old" group quota, enable
1288 						  * enforcement for hidden quota
1289 						  * files */
1290 #define EXT4_MOUNT_PRJQUOTA		0x200000 /* Enable project quota
1291 						  * enforcement */
1292 #define EXT4_MOUNT_DIOREAD_NOLOCK	0x400000 /* Enable support for dio read nolocking */
1293 #define EXT4_MOUNT_JOURNAL_CHECKSUM	0x800000 /* Journal checksums */
1294 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT	0x1000000 /* Journal Async Commit */
1295 #define EXT4_MOUNT_WARN_ON_ERROR	0x2000000 /* Trigger WARN_ON on error */
1296 #define EXT4_MOUNT_NO_PREFETCH_BLOCK_BITMAPS 0x4000000
1297 #define EXT4_MOUNT_DELALLOC		0x8000000 /* Delalloc support */
1298 #define EXT4_MOUNT_DATA_ERR_ABORT	0x10000000 /* Abort on file data write */
1299 #define EXT4_MOUNT_BLOCK_VALIDITY	0x20000000 /* Block validity checking */
1300 #define EXT4_MOUNT_DISCARD		0x40000000 /* Issue DISCARD requests */
1301 #define EXT4_MOUNT_INIT_INODE_TABLE	0x80000000 /* Initialize uninitialized itables */
1302 
1303 /*
1304  * Mount flags set either automatically (could not be set by mount option)
1305  * based on per file system feature or property or in special cases such as
1306  * distinguishing between explicit mount option definition and default.
1307  */
1308 #define EXT4_MOUNT2_EXPLICIT_DELALLOC	0x00000001 /* User explicitly
1309 						      specified delalloc */
1310 #define EXT4_MOUNT2_STD_GROUP_SIZE	0x00000002 /* We have standard group
1311 						      size of blocksize * 8
1312 						      blocks */
1313 #define EXT4_MOUNT2_HURD_COMPAT		0x00000004 /* Support HURD-castrated
1314 						      file systems */
1315 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM	0x00000008 /* User explicitly
1316 						specified journal checksum */
1317 
1318 #define EXT4_MOUNT2_JOURNAL_FAST_COMMIT	0x00000010 /* Journal fast commit */
1319 #define EXT4_MOUNT2_DAX_NEVER		0x00000020 /* Do not allow Direct Access */
1320 #define EXT4_MOUNT2_DAX_INODE		0x00000040 /* For printing options only */
1321 #define EXT4_MOUNT2_MB_OPTIMIZE_SCAN	0x00000080 /* Optimize group
1322 						    * scanning in mballoc
1323 						    */
1324 #define EXT4_MOUNT2_ABORT		0x00000100 /* Abort filesystem */
1325 #define EXT4_MOUNT2_BUFFERED_IOMAP	0x00000200 /* Use iomap for buffered I/O */
1326 
1327 #define clear_opt(sb, opt)		EXT4_SB(sb)->s_mount_opt &= \
1328 						~EXT4_MOUNT_##opt
1329 #define set_opt(sb, opt)		EXT4_SB(sb)->s_mount_opt |= \
1330 						EXT4_MOUNT_##opt
1331 #define test_opt(sb, opt)		(EXT4_SB(sb)->s_mount_opt & \
1332 					 EXT4_MOUNT_##opt)
1333 
1334 #define clear_opt2(sb, opt)		EXT4_SB(sb)->s_mount_opt2 &= \
1335 						~EXT4_MOUNT2_##opt
1336 #define set_opt2(sb, opt)		EXT4_SB(sb)->s_mount_opt2 |= \
1337 						EXT4_MOUNT2_##opt
1338 #define test_opt2(sb, opt)		(EXT4_SB(sb)->s_mount_opt2 & \
1339 					 EXT4_MOUNT2_##opt)
1340 
1341 #define ext4_test_and_set_bit		__test_and_set_bit_le
1342 #define ext4_set_bit			__set_bit_le
1343 #define ext4_test_and_clear_bit		__test_and_clear_bit_le
1344 #define ext4_clear_bit			__clear_bit_le
1345 #define ext4_test_bit			test_bit_le
1346 #define ext4_find_next_zero_bit		find_next_zero_bit_le
1347 #define ext4_find_next_bit		find_next_bit_le
1348 
1349 extern void mb_set_bits(void *bm, int cur, int len);
1350 
1351 /*
1352  * Maximal mount counts between two filesystem checks
1353  */
1354 #define EXT4_DFL_MAX_MNT_COUNT		20	/* Allow 20 mounts */
1355 #define EXT4_DFL_CHECKINTERVAL		0	/* Don't use interval check */
1356 
1357 /*
1358  * Behaviour when detecting errors
1359  */
1360 #define EXT4_ERRORS_CONTINUE		1	/* Continue execution */
1361 #define EXT4_ERRORS_RO			2	/* Remount fs read-only */
1362 #define EXT4_ERRORS_PANIC		3	/* Panic */
1363 #define EXT4_ERRORS_DEFAULT		EXT4_ERRORS_CONTINUE
1364 
1365 /* Metadata checksum algorithm codes */
1366 #define EXT4_CRC32C_CHKSUM		1
1367 
1368 #define EXT4_LABEL_MAX			16
1369 
1370 /*
1371  * Structure of the super block
1372  */
1373 struct ext4_super_block {
1374 /*00*/	__le32	s_inodes_count;		/* Inodes count */
1375 	__le32	s_blocks_count_lo;	/* Blocks count */
1376 	__le32	s_r_blocks_count_lo;	/* Reserved blocks count */
1377 	__le32	s_free_blocks_count_lo;	/* Free blocks count */
1378 /*10*/	__le32	s_free_inodes_count;	/* Free inodes count */
1379 	__le32	s_first_data_block;	/* First Data Block */
1380 	__le32	s_log_block_size;	/* Block size */
1381 	__le32	s_log_cluster_size;	/* Allocation cluster size */
1382 /*20*/	__le32	s_blocks_per_group;	/* # Blocks per group */
1383 	__le32	s_clusters_per_group;	/* # Clusters per group */
1384 	__le32	s_inodes_per_group;	/* # Inodes per group */
1385 	__le32	s_mtime;		/* Mount time */
1386 /*30*/	__le32	s_wtime;		/* Write time */
1387 	__le16	s_mnt_count;		/* Mount count */
1388 	__le16	s_max_mnt_count;	/* Maximal mount count */
1389 	__le16	s_magic;		/* Magic signature */
1390 	__le16	s_state;		/* File system state */
1391 	__le16	s_errors;		/* Behaviour when detecting errors */
1392 	__le16	s_minor_rev_level;	/* minor revision level */
1393 /*40*/	__le32	s_lastcheck;		/* time of last check */
1394 	__le32	s_checkinterval;	/* max. time between checks */
1395 	__le32	s_creator_os;		/* OS */
1396 	__le32	s_rev_level;		/* Revision level */
1397 /*50*/	__le16	s_def_resuid;		/* Default uid for reserved blocks */
1398 	__le16	s_def_resgid;		/* Default gid for reserved blocks */
1399 	/*
1400 	 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1401 	 *
1402 	 * Note: the difference between the compatible feature set and
1403 	 * the incompatible feature set is that if there is a bit set
1404 	 * in the incompatible feature set that the kernel doesn't
1405 	 * know about, it should refuse to mount the filesystem.
1406 	 *
1407 	 * e2fsck's requirements are more strict; if it doesn't know
1408 	 * about a feature in either the compatible or incompatible
1409 	 * feature set, it must abort and not try to meddle with
1410 	 * things it doesn't understand...
1411 	 */
1412 	__le32	s_first_ino;		/* First non-reserved inode */
1413 	__le16  s_inode_size;		/* size of inode structure */
1414 	__le16	s_block_group_nr;	/* block group # of this superblock */
1415 	__le32	s_feature_compat;	/* compatible feature set */
1416 /*60*/	__le32	s_feature_incompat;	/* incompatible feature set */
1417 	__le32	s_feature_ro_compat;	/* readonly-compatible feature set */
1418 /*68*/	__u8	s_uuid[16];		/* 128-bit uuid for volume */
1419 /*78*/	char	s_volume_name[EXT4_LABEL_MAX] __nonstring; /* volume name */
1420 /*88*/	char	s_last_mounted[64] __nonstring;	/* directory where last mounted */
1421 /*C8*/	__le32	s_algorithm_usage_bitmap; /* For compression */
1422 	/*
1423 	 * Performance hints.  Directory preallocation should only
1424 	 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1425 	 */
1426 	__u8	s_prealloc_blocks;	/* Nr of blocks to try to preallocate*/
1427 	__u8	s_prealloc_dir_blocks;	/* Nr to preallocate for dirs */
1428 	__le16	s_reserved_gdt_blocks;	/* Per group desc for online growth */
1429 	/*
1430 	 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1431 	 */
1432 /*D0*/	__u8	s_journal_uuid[16];	/* uuid of journal superblock */
1433 /*E0*/	__le32	s_journal_inum;		/* inode number of journal file */
1434 	__le32	s_journal_dev;		/* device number of journal file */
1435 	__le32	s_last_orphan;		/* start of list of inodes to delete */
1436 	__le32	s_hash_seed[4];		/* HTREE hash seed */
1437 	__u8	s_def_hash_version;	/* Default hash version to use */
1438 	__u8	s_jnl_backup_type;
1439 	__le16  s_desc_size;		/* size of group descriptor */
1440 /*100*/	__le32	s_default_mount_opts;
1441 	__le32	s_first_meta_bg;	/* First metablock block group */
1442 	__le32	s_mkfs_time;		/* When the filesystem was created */
1443 	__le32	s_jnl_blocks[17];	/* Backup of the journal inode */
1444 	/* 64bit support valid if EXT4_FEATURE_INCOMPAT_64BIT */
1445 /*150*/	__le32	s_blocks_count_hi;	/* Blocks count */
1446 	__le32	s_r_blocks_count_hi;	/* Reserved blocks count */
1447 	__le32	s_free_blocks_count_hi;	/* Free blocks count */
1448 	__le16	s_min_extra_isize;	/* All inodes have at least # bytes */
1449 	__le16	s_want_extra_isize; 	/* New inodes should reserve # bytes */
1450 	__le32	s_flags;		/* Miscellaneous flags */
1451 	__le16  s_raid_stride;		/* RAID stride */
1452 	__le16  s_mmp_update_interval;  /* # seconds to wait in MMP checking */
1453 	__le64  s_mmp_block;            /* Block for multi-mount protection */
1454 	__le32  s_raid_stripe_width;    /* blocks on all data disks (N*stride)*/
1455 	__u8	s_log_groups_per_flex;  /* FLEX_BG group size */
1456 	__u8	s_checksum_type;	/* metadata checksum algorithm used */
1457 	__u8	s_encryption_level;	/* versioning level for encryption */
1458 	__u8	s_reserved_pad;		/* Padding to next 32bits */
1459 	__le64	s_kbytes_written;	/* nr of lifetime kilobytes written */
1460 	__le32	s_snapshot_inum;	/* Inode number of active snapshot */
1461 	__le32	s_snapshot_id;		/* sequential ID of active snapshot */
1462 	__le64	s_snapshot_r_blocks_count; /* reserved blocks for active
1463 					      snapshot's future use */
1464 	__le32	s_snapshot_list;	/* inode number of the head of the
1465 					   on-disk snapshot list */
1466 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1467 	__le32	s_error_count;		/* number of fs errors */
1468 	__le32	s_first_error_time;	/* first time an error happened */
1469 	__le32	s_first_error_ino;	/* inode involved in first error */
1470 	__le64	s_first_error_block;	/* block involved of first error */
1471 	__u8	s_first_error_func[32] __nonstring;	/* function where the error happened */
1472 	__le32	s_first_error_line;	/* line number where error happened */
1473 	__le32	s_last_error_time;	/* most recent time of an error */
1474 	__le32	s_last_error_ino;	/* inode involved in last error */
1475 	__le32	s_last_error_line;	/* line number where error happened */
1476 	__le64	s_last_error_block;	/* block involved of last error */
1477 	__u8	s_last_error_func[32] __nonstring;	/* function where the error happened */
1478 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1479 	__u8	s_mount_opts[64];
1480 	__le32	s_usr_quota_inum;	/* inode for tracking user quota */
1481 	__le32	s_grp_quota_inum;	/* inode for tracking group quota */
1482 	__le32	s_overhead_clusters;	/* overhead blocks/clusters in fs */
1483 	__le32	s_backup_bgs[2];	/* groups with sparse_super2 SBs */
1484 	__u8	s_encrypt_algos[4];	/* Encryption algorithms in use  */
1485 	__u8	s_encrypt_pw_salt[16];	/* Salt used for string2key algorithm */
1486 	__le32	s_lpf_ino;		/* Location of the lost+found inode */
1487 	__le32	s_prj_quota_inum;	/* inode for tracking project quota */
1488 	__le32	s_checksum_seed;	/* crc32c(uuid) if csum_seed set */
1489 	__u8	s_wtime_hi;
1490 	__u8	s_mtime_hi;
1491 	__u8	s_mkfs_time_hi;
1492 	__u8	s_lastcheck_hi;
1493 	__u8	s_first_error_time_hi;
1494 	__u8	s_last_error_time_hi;
1495 	__u8	s_first_error_errcode;
1496 	__u8    s_last_error_errcode;
1497 	__le16  s_encoding;		/* Filename charset encoding */
1498 	__le16  s_encoding_flags;	/* Filename charset encoding flags */
1499 	__le32  s_orphan_file_inum;	/* Inode for tracking orphan inodes */
1500 	__le16	s_def_resuid_hi;
1501 	__le16	s_def_resgid_hi;
1502 	__le32	s_reserved[93];		/* Padding to the end of the block */
1503 	__le32	s_checksum;		/* crc32c(superblock) */
1504 };
1505 
1506 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1507 
1508 #ifdef __KERNEL__
1509 
1510 /* Number of quota types we support */
1511 #define EXT4_MAXQUOTAS 3
1512 
1513 #define EXT4_ENC_UTF8_12_1	1
1514 
1515 /* Types of ext4 journal triggers */
1516 enum ext4_journal_trigger_type {
1517 	EXT4_JTR_ORPHAN_FILE,
1518 	EXT4_JTR_NONE	/* This must be the last entry for indexing to work! */
1519 };
1520 
1521 #define EXT4_JOURNAL_TRIGGER_COUNT EXT4_JTR_NONE
1522 
1523 struct ext4_journal_trigger {
1524 	struct jbd2_buffer_trigger_type tr_triggers;
1525 	struct super_block *sb;
1526 };
1527 
1528 static inline struct ext4_journal_trigger *EXT4_TRIGGER(
1529 				struct jbd2_buffer_trigger_type *trigger)
1530 {
1531 	return container_of(trigger, struct ext4_journal_trigger, tr_triggers);
1532 }
1533 
1534 #define EXT4_ORPHAN_BLOCK_MAGIC 0x0b10ca04
1535 
1536 /* Structure at the tail of orphan block */
1537 struct ext4_orphan_block_tail {
1538 	__le32 ob_magic;
1539 	__le32 ob_checksum;
1540 };
1541 
1542 static inline int ext4_inodes_per_orphan_block(struct super_block *sb)
1543 {
1544 	return (sb->s_blocksize - sizeof(struct ext4_orphan_block_tail)) /
1545 			sizeof(u32);
1546 }
1547 
1548 struct ext4_orphan_block {
1549 	atomic_t ob_free_entries;	/* Number of free orphan entries in block */
1550 	struct buffer_head *ob_bh;	/* Buffer for orphan block */
1551 };
1552 
1553 /*
1554  * Info about orphan file.
1555  */
1556 struct ext4_orphan_info {
1557 	int of_blocks;			/* Number of orphan blocks in a file */
1558 	__u32 of_csum_seed;		/* Checksum seed for orphan file */
1559 	struct ext4_orphan_block *of_binfo;	/* Array with info about orphan
1560 						 * file blocks */
1561 };
1562 
1563 /*
1564  * Ext4 fast commit snapshot statistics.
1565  *
1566  * These are best-effort counters intended for debugging / performance
1567  * introspection; they are not exact under concurrent updates.
1568  */
1569 struct ext4_fc_snap_stats {
1570 	atomic64_t lock_updates_ns_total;
1571 	atomic64_t lock_updates_ns_max;
1572 	atomic64_t lock_updates_samples;
1573 
1574 	atomic64_t snap_inodes;
1575 	atomic64_t snap_ranges;
1576 
1577 	atomic64_t snap_fail_es_miss;
1578 	atomic64_t snap_fail_es_delayed;
1579 	atomic64_t snap_fail_es_other;
1580 
1581 	atomic64_t snap_fail_inodes_cap;
1582 	atomic64_t snap_fail_ranges_cap;
1583 	atomic64_t snap_fail_nomem;
1584 	atomic64_t snap_fail_inode_loc;
1585 
1586 	/*
1587 	 * Missing inode snapshots during log writing should never happen.
1588 	 * Keep this counter to help catch unexpected regressions.
1589 	 */
1590 	atomic64_t snap_fail_no_snap;
1591 };
1592 
1593 /*
1594  * fourth extended-fs super-block data in memory
1595  */
1596 struct ext4_sb_info {
1597 	unsigned long s_desc_size;	/* Size of a group descriptor in bytes */
1598 	unsigned long s_inodes_per_block;/* Number of inodes per block */
1599 	unsigned long s_blocks_per_group;/* Number of blocks in a group */
1600 	unsigned long s_clusters_per_group; /* Number of clusters in a group */
1601 	unsigned long s_inodes_per_group;/* Number of inodes in a group */
1602 	unsigned long s_itb_per_group;	/* Number of inode table blocks per group */
1603 	unsigned long s_gdb_count;	/* Number of group descriptor blocks */
1604 	unsigned long s_desc_per_block;	/* Number of group descriptors per block */
1605 	ext4_group_t s_groups_count;	/* Number of groups in the fs */
1606 	ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1607 	unsigned long s_overhead;  /* # of fs overhead clusters */
1608 	unsigned int s_cluster_ratio;	/* Number of blocks per cluster */
1609 	unsigned int s_cluster_bits;	/* log2 of s_cluster_ratio */
1610 	loff_t s_bitmap_maxbytes;	/* max bytes for bitmap files */
1611 	struct buffer_head * s_sbh;	/* Buffer containing the super block */
1612 	struct ext4_super_block *s_es;	/* Pointer to the super block in the buffer */
1613 	/* Array of bh's for the block group descriptors */
1614 	struct buffer_head * __rcu *s_group_desc;
1615 	unsigned int s_mount_opt;
1616 	unsigned int s_mount_opt2;
1617 	unsigned long s_mount_flags;
1618 	unsigned int s_def_mount_opt;
1619 	unsigned int s_def_mount_opt2;
1620 	ext4_fsblk_t s_sb_block;
1621 	atomic64_t s_resv_clusters;
1622 	kuid_t s_resuid;
1623 	kgid_t s_resgid;
1624 	unsigned short s_mount_state;
1625 	unsigned short s_pad;
1626 	int s_addr_per_block_bits;
1627 	int s_desc_per_block_bits;
1628 	int s_inode_size;
1629 	int s_first_ino;
1630 	unsigned int s_inode_readahead_blks;
1631 	unsigned int s_inode_goal;
1632 	u32 s_hash_seed[4];
1633 	int s_def_hash_version;
1634 	int s_hash_unsigned;	/* 3 if hash should be unsigned, 0 if not */
1635 	struct percpu_counter s_freeclusters_counter;
1636 	struct percpu_counter s_freeinodes_counter;
1637 	struct percpu_counter s_dirs_counter;
1638 	struct percpu_counter s_dirtyclusters_counter;
1639 	struct percpu_counter s_sra_exceeded_retry_limit;
1640 	struct blockgroup_lock *s_blockgroup_lock;
1641 	struct proc_dir_entry *s_proc;
1642 	struct kobject s_kobj;
1643 	struct completion s_kobj_unregister;
1644 	struct mutex s_error_notify_mutex; /* protects sysfs_notify vs kobject_del */
1645 	struct super_block *s_sb;
1646 	struct buffer_head *s_mmp_bh;
1647 
1648 	/* Journaling */
1649 	struct journal_s *s_journal;
1650 	unsigned long s_ext4_flags;		/* Ext4 superblock flags */
1651 	struct mutex s_orphan_lock;	/* Protects on disk list changes */
1652 	struct list_head s_orphan;	/* List of orphaned inodes in on disk
1653 					   list */
1654 	struct ext4_orphan_info s_orphan_info;
1655 	unsigned long s_commit_interval;
1656 	u32 s_max_batch_time;
1657 	u32 s_min_batch_time;
1658 	struct file *s_journal_bdev_file;
1659 #ifdef CONFIG_QUOTA
1660 	/* Names of quota files with journalled quota */
1661 	char __rcu *s_qf_names[EXT4_MAXQUOTAS];
1662 	int s_jquota_fmt;			/* Format of quota to use */
1663 #endif
1664 	unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1665 	struct ext4_system_blocks __rcu *s_system_blks;
1666 
1667 #ifdef EXTENTS_STATS
1668 	/* ext4 extents stats */
1669 	unsigned long s_ext_min;
1670 	unsigned long s_ext_max;
1671 	unsigned long s_depth_max;
1672 	spinlock_t s_ext_stats_lock;
1673 	unsigned long s_ext_blocks;
1674 	unsigned long s_ext_extents;
1675 #endif
1676 
1677 	/* for buddy allocator */
1678 	struct ext4_group_info ** __rcu *s_group_info;
1679 	struct inode *s_buddy_cache;
1680 	spinlock_t s_md_lock;
1681 	unsigned short *s_mb_offsets;
1682 	unsigned int *s_mb_maxs;
1683 	unsigned int s_group_info_size;
1684 	atomic_t s_mb_free_pending;
1685 	struct list_head s_freed_data_list[2];	/* List of blocks to be freed
1686 						   after commit completed */
1687 	struct list_head s_discard_list;
1688 	struct work_struct s_discard_work;
1689 	atomic_t s_retry_alloc_pending;
1690 	struct xarray *s_mb_avg_fragment_size;
1691 	struct xarray *s_mb_largest_free_orders;
1692 
1693 	/* tunables */
1694 	unsigned long s_stripe;
1695 	unsigned int s_mb_max_linear_groups;
1696 	unsigned int s_mb_stream_request;
1697 	unsigned int s_mb_max_to_scan;
1698 	unsigned int s_mb_min_to_scan;
1699 	unsigned int s_mb_stats;
1700 	unsigned int s_mb_order2_reqs;
1701 	unsigned int s_mb_group_prealloc;
1702 	unsigned int s_max_dir_size_kb;
1703 	unsigned int s_mb_prefetch;
1704 	unsigned int s_mb_prefetch_limit;
1705 	unsigned int s_mb_best_avail_max_trim_order;
1706 	unsigned int s_sb_update_sec;
1707 	unsigned int s_sb_update_kb;
1708 
1709 	/* where last allocation was done - for stream allocation */
1710 	ext4_group_t *s_mb_last_groups;
1711 	unsigned int s_mb_nr_global_goals;
1712 
1713 	/* stats for buddy allocator */
1714 	atomic_t s_bal_reqs;	/* number of reqs with len > 1 */
1715 	atomic_t s_bal_success;	/* we found long enough chunks */
1716 	atomic_t s_bal_allocated;	/* in blocks */
1717 	atomic_t s_bal_ex_scanned;	/* total extents scanned */
1718 	atomic_t s_bal_cX_ex_scanned[EXT4_MB_NUM_CRS];	/* total extents scanned */
1719 	atomic_t s_bal_groups_scanned;	/* number of groups scanned */
1720 	atomic_t s_bal_goals;	/* goal hits */
1721 	atomic_t s_bal_stream_goals;	/* stream allocation global goal hits */
1722 	atomic_t s_bal_len_goals;	/* len goal hits */
1723 	atomic_t s_bal_breaks;	/* too long searches */
1724 	atomic_t s_bal_2orders;	/* 2^order hits */
1725 	atomic64_t s_bal_cX_groups_considered[EXT4_MB_NUM_CRS];
1726 	atomic64_t s_bal_cX_hits[EXT4_MB_NUM_CRS];
1727 	atomic64_t s_bal_cX_failed[EXT4_MB_NUM_CRS];		/* cX loop didn't find blocks */
1728 	atomic_t s_mb_buddies_generated;	/* number of buddies generated */
1729 	atomic64_t s_mb_generation_time;
1730 	atomic_t s_mb_lost_chunks;
1731 	atomic_t s_mb_preallocated;
1732 	atomic_t s_mb_discarded;
1733 	atomic_t s_lock_busy;
1734 
1735 	/* locality groups */
1736 	struct ext4_locality_group __percpu *s_locality_groups;
1737 
1738 	/* for write statistics */
1739 	unsigned long s_sectors_written_start;
1740 	u64 s_kbytes_written;
1741 
1742 	/* the size of zero-out chunk */
1743 	unsigned int s_extent_max_zeroout_kb;
1744 
1745 	unsigned int s_log_groups_per_flex;
1746 	struct flex_groups * __rcu *s_flex_groups;
1747 	ext4_group_t s_flex_groups_allocated;
1748 
1749 	/* workqueue for reserved extent conversions (buffered io) */
1750 	struct workqueue_struct *rsv_conversion_wq;
1751 
1752 	/* timer for periodic error stats printing */
1753 	struct timer_list s_err_report;
1754 	/* timeout in seconds for s_err_report; 0 disables the timer. */
1755 	unsigned long s_err_report_sec;
1756 
1757 	/* Lazy inode table initialization info */
1758 	struct ext4_li_request *s_li_request;
1759 	/* Wait multiplier for lazy initialization thread */
1760 	unsigned int s_li_wait_mult;
1761 
1762 	/* Kernel thread for multiple mount protection */
1763 	struct task_struct *s_mmp_tsk;
1764 
1765 	/* record the last minlen when FITRIM is called. */
1766 	unsigned long s_last_trim_minblks;
1767 
1768 	/* minimum folio order of a page cache allocation */
1769 	u16 s_min_folio_order;
1770 	/* supported maximum folio order, 0 means not supported */
1771 	u16 s_max_folio_order;
1772 
1773 	/* Precomputed FS UUID checksum for seeding other checksums */
1774 	__u32 s_csum_seed;
1775 
1776 	/* Reclaim extents from extent status tree */
1777 	struct shrinker *s_es_shrinker;
1778 	struct list_head s_es_list;	/* List of inodes with reclaimable extents */
1779 	long s_es_nr_inode;
1780 	struct ext4_es_stats s_es_stats;
1781 	/* ext4_get_link() reads the sbi in rcu pathwalk */
1782 	struct rcu_head s_rcu;
1783 	struct mb_cache *s_ea_block_cache;
1784 	struct mb_cache *s_ea_inode_cache;
1785 
1786 	/* Deferred iput for EA inodes to avoid lock ordering issues */
1787 	struct llist_head s_ea_inode_to_free;
1788 	struct delayed_work s_ea_inode_work;
1789 
1790 	spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1791 
1792 	/* Journal triggers for checksum computation */
1793 	struct ext4_journal_trigger s_journal_triggers[EXT4_JOURNAL_TRIGGER_COUNT];
1794 
1795 	/* Ratelimit ext4 messages. */
1796 	struct ratelimit_state s_err_ratelimit_state;
1797 	struct ratelimit_state s_warning_ratelimit_state;
1798 	struct ratelimit_state s_msg_ratelimit_state;
1799 	atomic_t s_warning_count;
1800 	atomic_t s_msg_count;
1801 
1802 	/* Encryption policy for '-o test_dummy_encryption' */
1803 	struct fscrypt_dummy_policy s_dummy_enc_policy;
1804 
1805 	/*
1806 	 * Barrier between writepages ops and changing any inode's JOURNAL_DATA
1807 	 * or EXTENTS flag or between writepages ops and changing DELALLOC or
1808 	 * DIOREAD_NOLOCK mount options on remount.
1809 	 */
1810 	struct percpu_rw_semaphore s_writepages_rwsem;
1811 	struct dax_device *s_daxdev;
1812 	u64 s_dax_part_off;
1813 #ifdef CONFIG_EXT4_DEBUG
1814 	unsigned long s_simulate_fail;
1815 #endif
1816 	/* Record the errseq of the backing block device */
1817 	errseq_t s_bdev_wb_err;
1818 	spinlock_t s_bdev_wb_lock;
1819 
1820 	/* Information about errors that happened during this mount */
1821 	spinlock_t s_error_lock;
1822 	int s_add_error_count;
1823 	int s_first_error_code;
1824 	__u32 s_first_error_line;
1825 	__u32 s_first_error_ino;
1826 	__u64 s_first_error_block;
1827 	const char *s_first_error_func;
1828 	time64_t s_first_error_time;
1829 	int s_last_error_code;
1830 	__u32 s_last_error_line;
1831 	__u32 s_last_error_ino;
1832 	__u64 s_last_error_block;
1833 	const char *s_last_error_func;
1834 	time64_t s_last_error_time;
1835 	/*
1836 	 * If we are in a context where we cannot update the on-disk
1837 	 * superblock, we queue the work here.  This is used to update
1838 	 * the error information in the superblock, and for periodic
1839 	 * updates of the superblock called from the commit callback
1840 	 * function.
1841 	 */
1842 	struct work_struct s_sb_upd_work;
1843 
1844 	/* Atomic write unit values in bytes */
1845 	unsigned int s_awu_min;
1846 	unsigned int s_awu_max;
1847 
1848 	/* Ext4 fast commit sub transaction ID */
1849 	atomic_t s_fc_subtid;
1850 
1851 	/*
1852 	 * After commit starts, the main queue gets locked, and the further
1853 	 * updates get added in the staging queue.
1854 	 */
1855 #define FC_Q_MAIN	0
1856 #define FC_Q_STAGING	1
1857 	struct list_head s_fc_q[2];	/* Inodes staged for fast commit
1858 					 * that have data changes in them.
1859 					 */
1860 	struct list_head s_fc_dentry_q[2];	/* directory entry updates */
1861 	unsigned int s_fc_bytes;
1862 	/*
1863 	 * Main fast commit lock. This lock protects accesses to the
1864 	 * following fields:
1865 	 * ei->i_fc_list, s_fc_dentry_q, s_fc_q, s_fc_bytes, s_fc_bh.
1866 	 *
1867 	 * s_fc_lock can be taken from reclaim context (inode eviction) and is
1868 	 * thus reclaim unsafe. Use ext4_fc_lock()/ext4_fc_unlock() helpers
1869 	 * when acquiring / releasing the lock.
1870 	 */
1871 	struct mutex s_fc_lock;
1872 	struct buffer_head *s_fc_bh;
1873 	struct ext4_fc_stats s_fc_stats;
1874 	struct ext4_fc_snap_stats s_fc_snap_stats;
1875 	tid_t s_fc_ineligible_tid;
1876 #ifdef CONFIG_EXT4_DEBUG
1877 	int s_fc_debug_max_replay;
1878 #endif
1879 	struct ext4_fc_replay_state s_fc_replay_state;
1880 };
1881 
1882 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1883 {
1884 	return sb->s_fs_info;
1885 }
1886 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1887 {
1888 	return container_of(inode, struct ext4_inode_info, vfs_inode);
1889 }
1890 
1891 static inline int ext4_writepages_down_read(struct super_block *sb)
1892 {
1893 	percpu_down_read(&EXT4_SB(sb)->s_writepages_rwsem);
1894 	return memalloc_nofs_save();
1895 }
1896 
1897 static inline void ext4_writepages_up_read(struct super_block *sb, int ctx)
1898 {
1899 	memalloc_nofs_restore(ctx);
1900 	percpu_up_read(&EXT4_SB(sb)->s_writepages_rwsem);
1901 }
1902 
1903 static inline int ext4_writepages_down_write(struct super_block *sb)
1904 {
1905 	percpu_down_write(&EXT4_SB(sb)->s_writepages_rwsem);
1906 	return memalloc_nofs_save();
1907 }
1908 
1909 static inline void ext4_writepages_up_write(struct super_block *sb, int ctx)
1910 {
1911 	memalloc_nofs_restore(ctx);
1912 	percpu_up_write(&EXT4_SB(sb)->s_writepages_rwsem);
1913 }
1914 
1915 static inline int ext4_fc_lock(struct super_block *sb)
1916 {
1917 	mutex_lock(&EXT4_SB(sb)->s_fc_lock);
1918 	return memalloc_nofs_save();
1919 }
1920 
1921 static inline void ext4_fc_unlock(struct super_block *sb, int ctx)
1922 {
1923 	memalloc_nofs_restore(ctx);
1924 	mutex_unlock(&EXT4_SB(sb)->s_fc_lock);
1925 }
1926 
1927 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1928 {
1929 	return ino == EXT4_ROOT_INO ||
1930 		(ino >= EXT4_FIRST_INO(sb) &&
1931 		 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1932 }
1933 
1934 static inline int ext4_get_resuid(struct ext4_super_block *es)
1935 {
1936 	return le16_to_cpu(es->s_def_resuid) |
1937 		le16_to_cpu(es->s_def_resuid_hi) << 16;
1938 }
1939 
1940 static inline int ext4_get_resgid(struct ext4_super_block *es)
1941 {
1942 	return le16_to_cpu(es->s_def_resgid) |
1943 		le16_to_cpu(es->s_def_resgid_hi) << 16;
1944 }
1945 
1946 /*
1947  * Returns: sbi->field[index]
1948  * Used to access an array element from the following sbi fields which require
1949  * rcu protection to avoid dereferencing an invalid pointer due to reassignment
1950  * - s_group_desc
1951  * - s_group_info
1952  * - s_flex_group
1953  */
1954 #define sbi_array_rcu_deref(sbi, field, index)				   \
1955 ({									   \
1956 	typeof(*((sbi)->field)) _v;					   \
1957 	rcu_read_lock();						   \
1958 	_v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index];	   \
1959 	rcu_read_unlock();						   \
1960 	_v;								   \
1961 })
1962 
1963 /*
1964  * run-time mount flags
1965  */
1966 enum {
1967 	EXT4_MF_MNTDIR_SAMPLED,
1968 	EXT4_MF_FC_INELIGIBLE,	/* Fast commit ineligible */
1969 	EXT4_MF_JOURNAL_DESTROY	/* Journal is in process of destroying */
1970 };
1971 
1972 static inline void ext4_set_mount_flag(struct super_block *sb, int bit)
1973 {
1974 	set_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1975 }
1976 
1977 static inline void ext4_clear_mount_flag(struct super_block *sb, int bit)
1978 {
1979 	clear_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1980 }
1981 
1982 static inline int ext4_test_mount_flag(struct super_block *sb, int bit)
1983 {
1984 	return test_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1985 }
1986 
1987 
1988 /*
1989  * Simulate_fail codes
1990  */
1991 #define EXT4_SIM_BBITMAP_EIO	1
1992 #define EXT4_SIM_BBITMAP_CRC	2
1993 #define EXT4_SIM_IBITMAP_EIO	3
1994 #define EXT4_SIM_IBITMAP_CRC	4
1995 #define EXT4_SIM_INODE_EIO	5
1996 #define EXT4_SIM_INODE_CRC	6
1997 #define EXT4_SIM_DIRBLOCK_EIO	7
1998 #define EXT4_SIM_DIRBLOCK_CRC	8
1999 
2000 static inline bool ext4_simulate_fail(struct super_block *sb,
2001 				     unsigned long code)
2002 {
2003 #ifdef CONFIG_EXT4_DEBUG
2004 	struct ext4_sb_info *sbi = EXT4_SB(sb);
2005 
2006 	if (unlikely(sbi->s_simulate_fail == code)) {
2007 		sbi->s_simulate_fail = 0;
2008 		return true;
2009 	}
2010 #endif
2011 	return false;
2012 }
2013 
2014 /*
2015  * Error number codes for s_{first,last}_error_errno
2016  *
2017  * Linux errno numbers are architecture specific, so we need to translate
2018  * them into something which is architecture independent.   We don't define
2019  * codes for all errno's; just the ones which are most likely to be the cause
2020  * of an ext4_error() call.
2021  */
2022 #define EXT4_ERR_UNKNOWN	 1
2023 #define EXT4_ERR_EIO		 2
2024 #define EXT4_ERR_ENOMEM		 3
2025 #define EXT4_ERR_EFSBADCRC	 4
2026 #define EXT4_ERR_EFSCORRUPTED	 5
2027 #define EXT4_ERR_ENOSPC		 6
2028 #define EXT4_ERR_ENOKEY		 7
2029 #define EXT4_ERR_EROFS		 8
2030 #define EXT4_ERR_EFBIG		 9
2031 #define EXT4_ERR_EEXIST		10
2032 #define EXT4_ERR_ERANGE		11
2033 #define EXT4_ERR_EOVERFLOW	12
2034 #define EXT4_ERR_EBUSY		13
2035 #define EXT4_ERR_ENOTDIR	14
2036 #define EXT4_ERR_ENOTEMPTY	15
2037 #define EXT4_ERR_ESHUTDOWN	16
2038 #define EXT4_ERR_EFAULT		17
2039 
2040 /*
2041  * Inode dynamic state flags
2042  */
2043 enum {
2044 	EXT4_STATE_NEW,			/* inode is newly created */
2045 	EXT4_STATE_XATTR,		/* has in-inode xattrs */
2046 	EXT4_STATE_NO_EXPAND,		/* No space for expansion */
2047 	EXT4_STATE_DA_ALLOC_CLOSE,	/* Alloc DA blks on close */
2048 	EXT4_STATE_EXT_MIGRATE,		/* Inode is migrating */
2049 	EXT4_STATE_NEWENTRY,		/* File just added to dir */
2050 	EXT4_STATE_MAY_INLINE_DATA,	/* may have in-inode data */
2051 	EXT4_STATE_EXT_PRECACHED,	/* extents have been precached */
2052 	EXT4_STATE_LUSTRE_EA_INODE,	/* Lustre-style ea_inode */
2053 	EXT4_STATE_VERITY_IN_PROGRESS,	/* building fs-verity Merkle tree */
2054 	EXT4_STATE_FC_COMMITTING,	/* Fast commit ongoing */
2055 	EXT4_STATE_FC_FLUSHING_DATA,	/* Fast commit flushing data */
2056 	EXT4_STATE_ORPHAN_FILE,		/* Inode orphaned in orphan file */
2057 	EXT4_STATE_FC_REQUEUE,		/* Inode modified during fast commit */
2058 	EXT4_STATE_BUFFERED_IOMAP,	/* Inode use iomap for buffered IO */
2059 	EXT4_STATE_DISKSIZE_GROW_PENDING,
2060 					/* Has zeroed EOF block straddles
2061 					 * i_disksize awaiting writeback */
2062 };
2063 
2064 #define EXT4_INODE_BIT_FNS(name, field, offset)				\
2065 static inline int ext4_test_inode_##name(struct inode *inode, int bit)	\
2066 {									\
2067 	return test_bit(bit + (offset), &EXT4_I(inode)->i_##field);	\
2068 }									\
2069 static inline void ext4_set_inode_##name(struct inode *inode, int bit)	\
2070 {									\
2071 	set_bit(bit + (offset), &EXT4_I(inode)->i_##field);		\
2072 }									\
2073 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
2074 {									\
2075 	clear_bit(bit + (offset), &EXT4_I(inode)->i_##field);		\
2076 }
2077 
2078 /* Add these declarations here only so that these functions can be
2079  * found by name.  Otherwise, they are very hard to locate. */
2080 static inline int ext4_test_inode_flag(struct inode *inode, int bit);
2081 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
2082 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
2083 EXT4_INODE_BIT_FNS(flag, flags, 0)
2084 
2085 /* Add these declarations here only so that these functions can be
2086  * found by name.  Otherwise, they are very hard to locate. */
2087 static inline int ext4_test_inode_state(struct inode *inode, int bit);
2088 static inline void ext4_set_inode_state(struct inode *inode, int bit);
2089 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
2090 static inline unsigned long *ext4_inode_state_wait_word(struct inode *inode);
2091 static inline int ext4_inode_state_wait_bit(int bit);
2092 #if (BITS_PER_LONG < 64)
2093 EXT4_INODE_BIT_FNS(state, state_flags, 0)
2094 
2095 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
2096 {
2097 	(ei)->i_state_flags = 0;
2098 }
2099 #else
2100 EXT4_INODE_BIT_FNS(state, flags, 32)
2101 
2102 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
2103 {
2104 	/* We depend on the fact that callers will set i_flags */
2105 }
2106 #endif
2107 
2108 static inline unsigned long *ext4_inode_state_wait_word(struct inode *inode)
2109 {
2110 #if (BITS_PER_LONG < 64)
2111 	return &EXT4_I(inode)->i_state_flags;
2112 #else
2113 	return &EXT4_I(inode)->i_flags;
2114 #endif
2115 }
2116 
2117 static inline int ext4_inode_state_wait_bit(int bit)
2118 {
2119 #if (BITS_PER_LONG < 64)
2120 	return bit;
2121 #else
2122 	return bit + 32;
2123 #endif
2124 }
2125 #else
2126 /* Assume that user mode programs are passing in an ext4fs superblock, not
2127  * a kernel struct super_block.  This will allow us to call the feature-test
2128  * macros from user land. */
2129 #define EXT4_SB(sb)	(sb)
2130 #endif
2131 
2132 static inline bool ext4_verity_in_progress(struct inode *inode)
2133 {
2134 	return IS_ENABLED(CONFIG_FS_VERITY) &&
2135 	       ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS);
2136 }
2137 
2138 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
2139 
2140 /*
2141  * Check whether the inode is tracked as orphan (either in orphan file or
2142  * orphan list).
2143  */
2144 static inline bool ext4_inode_orphan_tracked(struct inode *inode)
2145 {
2146 	return ext4_test_inode_state(inode, EXT4_STATE_ORPHAN_FILE) ||
2147 		!list_empty(&EXT4_I(inode)->i_orphan);
2148 }
2149 
2150 static inline struct mapping_metadata_bhs *ext4_i_metadata_bhs(
2151 							struct inode *inode)
2152 {
2153 	/*
2154 	 * i_metadata_bhs is set in ext4_inode_attach_mmb() using cmpxchg().
2155 	 * We use READ_ONCE when accessing i_metadata_bhs to make sure we get
2156 	 * consistent view for all accesses.
2157 	 */
2158 	return READ_ONCE(EXT4_I(inode)->i_metadata_bhs);
2159 }
2160 
2161 /* Whether the inode pass through the iomap infrastructure for buffered I/O */
2162 static inline bool ext4_inode_buffered_iomap(struct inode *inode)
2163 {
2164 	return ext4_test_inode_state(inode, EXT4_STATE_BUFFERED_IOMAP);
2165 }
2166 
2167 /*
2168  * Codes for operating systems
2169  */
2170 #define EXT4_OS_LINUX		0
2171 #define EXT4_OS_HURD		1
2172 #define EXT4_OS_MASIX		2
2173 #define EXT4_OS_FREEBSD		3
2174 #define EXT4_OS_LITES		4
2175 
2176 /*
2177  * Revision levels
2178  */
2179 #define EXT4_GOOD_OLD_REV	0	/* The good old (original) format */
2180 #define EXT4_DYNAMIC_REV	1	/* V2 format w/ dynamic inode sizes */
2181 
2182 #define EXT4_MAX_SUPP_REV	EXT4_DYNAMIC_REV
2183 
2184 #define EXT4_GOOD_OLD_INODE_SIZE 128
2185 
2186 #define EXT4_EXTRA_TIMESTAMP_MAX	(((s64)1 << 34) - 1  + S32_MIN)
2187 #define EXT4_NON_EXTRA_TIMESTAMP_MAX	S32_MAX
2188 #define EXT4_TIMESTAMP_MIN		S32_MIN
2189 
2190 /*
2191  * Feature set definitions
2192  */
2193 
2194 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC	0x0001
2195 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES	0x0002
2196 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL		0x0004
2197 #define EXT4_FEATURE_COMPAT_EXT_ATTR		0x0008
2198 #define EXT4_FEATURE_COMPAT_RESIZE_INODE	0x0010
2199 #define EXT4_FEATURE_COMPAT_DIR_INDEX		0x0020
2200 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2	0x0200
2201 /*
2202  * The reason why "FAST_COMMIT" is a compat feature is that, FS becomes
2203  * incompatible only if fast commit blocks are present in the FS. Since we
2204  * clear the journal (and thus the fast commit blocks), we don't mark FS as
2205  * incompatible. We also have a JBD2 incompat feature, which gets set when
2206  * there are fast commit blocks present in the journal.
2207  */
2208 #define EXT4_FEATURE_COMPAT_FAST_COMMIT		0x0400
2209 #define EXT4_FEATURE_COMPAT_STABLE_INODES	0x0800
2210 #define EXT4_FEATURE_COMPAT_ORPHAN_FILE		0x1000	/* Orphan file exists */
2211 
2212 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER	0x0001
2213 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE	0x0002
2214 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR	0x0004
2215 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE        0x0008
2216 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM		0x0010
2217 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK	0x0020
2218 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE	0x0040
2219 #define EXT4_FEATURE_RO_COMPAT_QUOTA		0x0100
2220 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC		0x0200
2221 /*
2222  * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM).  When
2223  * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
2224  * all other data structures' checksums.  However, the METADATA_CSUM and
2225  * GDT_CSUM bits are mutually exclusive.
2226  */
2227 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM	0x0400
2228 #define EXT4_FEATURE_RO_COMPAT_READONLY		0x1000
2229 #define EXT4_FEATURE_RO_COMPAT_PROJECT		0x2000
2230 #define EXT4_FEATURE_RO_COMPAT_VERITY		0x8000
2231 #define EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT	0x10000 /* Orphan file may be
2232 							   non-empty */
2233 
2234 #define EXT4_FEATURE_INCOMPAT_COMPRESSION	0x0001
2235 #define EXT4_FEATURE_INCOMPAT_FILETYPE		0x0002
2236 #define EXT4_FEATURE_INCOMPAT_RECOVER		0x0004 /* Needs recovery */
2237 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV	0x0008 /* Journal device */
2238 #define EXT4_FEATURE_INCOMPAT_META_BG		0x0010
2239 #define EXT4_FEATURE_INCOMPAT_EXTENTS		0x0040 /* extents support */
2240 #define EXT4_FEATURE_INCOMPAT_64BIT		0x0080
2241 #define EXT4_FEATURE_INCOMPAT_MMP               0x0100
2242 #define EXT4_FEATURE_INCOMPAT_FLEX_BG		0x0200
2243 #define EXT4_FEATURE_INCOMPAT_EA_INODE		0x0400 /* EA in inode */
2244 #define EXT4_FEATURE_INCOMPAT_DIRDATA		0x1000 /* data in dirent */
2245 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED		0x2000
2246 #define EXT4_FEATURE_INCOMPAT_LARGEDIR		0x4000 /* >2GB or 3-lvl htree */
2247 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA	0x8000 /* data in inode */
2248 #define EXT4_FEATURE_INCOMPAT_ENCRYPT		0x10000
2249 #define EXT4_FEATURE_INCOMPAT_CASEFOLD		0x20000
2250 
2251 extern void ext4_update_dynamic_rev(struct super_block *sb);
2252 
2253 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
2254 static inline bool ext4_has_feature_##name(struct super_block *sb) \
2255 { \
2256 	return ((EXT4_SB(sb)->s_es->s_feature_compat & \
2257 		cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
2258 } \
2259 static inline void ext4_set_feature_##name(struct super_block *sb) \
2260 { \
2261 	ext4_update_dynamic_rev(sb); \
2262 	EXT4_SB(sb)->s_es->s_feature_compat |= \
2263 		cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
2264 } \
2265 static inline void ext4_clear_feature_##name(struct super_block *sb) \
2266 { \
2267 	EXT4_SB(sb)->s_es->s_feature_compat &= \
2268 		~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
2269 }
2270 
2271 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
2272 static inline bool ext4_has_feature_##name(struct super_block *sb) \
2273 { \
2274 	return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
2275 		cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
2276 } \
2277 static inline void ext4_set_feature_##name(struct super_block *sb) \
2278 { \
2279 	ext4_update_dynamic_rev(sb); \
2280 	EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
2281 		cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
2282 } \
2283 static inline void ext4_clear_feature_##name(struct super_block *sb) \
2284 { \
2285 	EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
2286 		~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
2287 }
2288 
2289 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
2290 static inline bool ext4_has_feature_##name(struct super_block *sb) \
2291 { \
2292 	return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
2293 		cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
2294 } \
2295 static inline void ext4_set_feature_##name(struct super_block *sb) \
2296 { \
2297 	ext4_update_dynamic_rev(sb); \
2298 	EXT4_SB(sb)->s_es->s_feature_incompat |= \
2299 		cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
2300 } \
2301 static inline void ext4_clear_feature_##name(struct super_block *sb) \
2302 { \
2303 	EXT4_SB(sb)->s_es->s_feature_incompat &= \
2304 		~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
2305 }
2306 
2307 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc,		DIR_PREALLOC)
2308 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes,	IMAGIC_INODES)
2309 EXT4_FEATURE_COMPAT_FUNCS(journal,		HAS_JOURNAL)
2310 EXT4_FEATURE_COMPAT_FUNCS(xattr,		EXT_ATTR)
2311 EXT4_FEATURE_COMPAT_FUNCS(resize_inode,		RESIZE_INODE)
2312 EXT4_FEATURE_COMPAT_FUNCS(dir_index,		DIR_INDEX)
2313 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2,	SPARSE_SUPER2)
2314 EXT4_FEATURE_COMPAT_FUNCS(fast_commit,		FAST_COMMIT)
2315 EXT4_FEATURE_COMPAT_FUNCS(stable_inodes,	STABLE_INODES)
2316 EXT4_FEATURE_COMPAT_FUNCS(orphan_file,		ORPHAN_FILE)
2317 
2318 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super,	SPARSE_SUPER)
2319 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file,	LARGE_FILE)
2320 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir,		BTREE_DIR)
2321 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file,		HUGE_FILE)
2322 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum,		GDT_CSUM)
2323 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink,		DIR_NLINK)
2324 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize,	EXTRA_ISIZE)
2325 EXT4_FEATURE_RO_COMPAT_FUNCS(quota,		QUOTA)
2326 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc,		BIGALLOC)
2327 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum,	METADATA_CSUM)
2328 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly,		READONLY)
2329 EXT4_FEATURE_RO_COMPAT_FUNCS(project,		PROJECT)
2330 EXT4_FEATURE_RO_COMPAT_FUNCS(verity,		VERITY)
2331 EXT4_FEATURE_RO_COMPAT_FUNCS(orphan_present,	ORPHAN_PRESENT)
2332 
2333 EXT4_FEATURE_INCOMPAT_FUNCS(compression,	COMPRESSION)
2334 EXT4_FEATURE_INCOMPAT_FUNCS(filetype,		FILETYPE)
2335 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery,	RECOVER)
2336 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev,	JOURNAL_DEV)
2337 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg,		META_BG)
2338 EXT4_FEATURE_INCOMPAT_FUNCS(extents,		EXTENTS)
2339 EXT4_FEATURE_INCOMPAT_FUNCS(64bit,		64BIT)
2340 EXT4_FEATURE_INCOMPAT_FUNCS(mmp,		MMP)
2341 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg,		FLEX_BG)
2342 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode,		EA_INODE)
2343 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata,		DIRDATA)
2344 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed,		CSUM_SEED)
2345 EXT4_FEATURE_INCOMPAT_FUNCS(largedir,		LARGEDIR)
2346 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data,	INLINE_DATA)
2347 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt,		ENCRYPT)
2348 EXT4_FEATURE_INCOMPAT_FUNCS(casefold,		CASEFOLD)
2349 
2350 #define EXT2_FEATURE_COMPAT_SUPP	EXT4_FEATURE_COMPAT_EXT_ATTR
2351 #define EXT2_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
2352 					 EXT4_FEATURE_INCOMPAT_META_BG)
2353 #define EXT2_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2354 					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2355 					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
2356 
2357 #define EXT3_FEATURE_COMPAT_SUPP	EXT4_FEATURE_COMPAT_EXT_ATTR
2358 #define EXT3_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
2359 					 EXT4_FEATURE_INCOMPAT_RECOVER| \
2360 					 EXT4_FEATURE_INCOMPAT_META_BG)
2361 #define EXT3_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2362 					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2363 					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
2364 
2365 #define EXT4_FEATURE_COMPAT_SUPP	(EXT4_FEATURE_COMPAT_EXT_ATTR| \
2366 					 EXT4_FEATURE_COMPAT_ORPHAN_FILE)
2367 #define EXT4_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
2368 					 EXT4_FEATURE_INCOMPAT_RECOVER| \
2369 					 EXT4_FEATURE_INCOMPAT_META_BG| \
2370 					 EXT4_FEATURE_INCOMPAT_EXTENTS| \
2371 					 EXT4_FEATURE_INCOMPAT_64BIT| \
2372 					 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
2373 					 EXT4_FEATURE_INCOMPAT_EA_INODE| \
2374 					 EXT4_FEATURE_INCOMPAT_MMP | \
2375 					 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
2376 					 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
2377 					 EXT4_FEATURE_INCOMPAT_CASEFOLD | \
2378 					 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \
2379 					 EXT4_FEATURE_INCOMPAT_LARGEDIR)
2380 #define EXT4_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2381 					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2382 					 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
2383 					 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
2384 					 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
2385 					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
2386 					 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
2387 					 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
2388 					 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
2389 					 EXT4_FEATURE_RO_COMPAT_QUOTA |\
2390 					 EXT4_FEATURE_RO_COMPAT_PROJECT |\
2391 					 EXT4_FEATURE_RO_COMPAT_VERITY |\
2392 					 EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT)
2393 
2394 #define EXTN_FEATURE_FUNCS(ver) \
2395 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
2396 { \
2397 	return ((EXT4_SB(sb)->s_es->s_feature_compat & \
2398 		cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
2399 } \
2400 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
2401 { \
2402 	return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
2403 		cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
2404 } \
2405 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
2406 { \
2407 	return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
2408 		cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
2409 }
2410 
2411 EXTN_FEATURE_FUNCS(2)
2412 EXTN_FEATURE_FUNCS(3)
2413 EXTN_FEATURE_FUNCS(4)
2414 
2415 static inline bool ext4_has_compat_features(struct super_block *sb)
2416 {
2417 	return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
2418 }
2419 static inline bool ext4_has_ro_compat_features(struct super_block *sb)
2420 {
2421 	return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
2422 }
2423 static inline bool ext4_has_incompat_features(struct super_block *sb)
2424 {
2425 	return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
2426 }
2427 
2428 extern int ext4_feature_set_ok(struct super_block *sb, int readonly);
2429 
2430 /*
2431  * Superblock flags
2432  */
2433 enum {
2434 	EXT4_FLAGS_RESIZING,	/* Avoid superblock update and resize race */
2435 	EXT4_FLAGS_SHUTDOWN,	/* Prevent access to the file system */
2436 	EXT4_FLAGS_BDEV_IS_DAX,	/* Current block device support DAX */
2437 	EXT4_FLAGS_EMERGENCY_RO,/* Emergency read-only due to fs errors */
2438 };
2439 
2440 static inline int ext4_forced_shutdown(struct super_block *sb)
2441 {
2442 	return test_bit(EXT4_FLAGS_SHUTDOWN, &EXT4_SB(sb)->s_ext4_flags);
2443 }
2444 
2445 static inline int ext4_emergency_ro(struct super_block *sb)
2446 {
2447 	return test_bit(EXT4_FLAGS_EMERGENCY_RO, &EXT4_SB(sb)->s_ext4_flags);
2448 }
2449 
2450 static inline int ext4_emergency_state(struct super_block *sb)
2451 {
2452 	if (unlikely(ext4_forced_shutdown(sb)))
2453 		return -EIO;
2454 	if (unlikely(ext4_emergency_ro(sb)))
2455 		return -EROFS;
2456 	return 0;
2457 }
2458 
2459 /*
2460  * Default values for user and/or group using reserved blocks
2461  */
2462 #define	EXT4_DEF_RESUID		0
2463 #define	EXT4_DEF_RESGID		0
2464 
2465 /*
2466  * Default project ID
2467  */
2468 #define	EXT4_DEF_PROJID		0
2469 
2470 #define EXT4_DEF_INODE_READAHEAD_BLKS	32
2471 
2472 /*
2473  * Default mount options
2474  */
2475 #define EXT4_DEFM_DEBUG		0x0001
2476 #define EXT4_DEFM_BSDGROUPS	0x0002
2477 #define EXT4_DEFM_XATTR_USER	0x0004
2478 #define EXT4_DEFM_ACL		0x0008
2479 #define EXT4_DEFM_UID16		0x0010
2480 #define EXT4_DEFM_JMODE		0x0060
2481 #define EXT4_DEFM_JMODE_DATA	0x0020
2482 #define EXT4_DEFM_JMODE_ORDERED	0x0040
2483 #define EXT4_DEFM_JMODE_WBACK	0x0060
2484 #define EXT4_DEFM_NOBARRIER	0x0100
2485 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
2486 #define EXT4_DEFM_DISCARD	0x0400
2487 #define EXT4_DEFM_NODELALLOC	0x0800
2488 
2489 /*
2490  * Default journal batch times and ioprio.
2491  */
2492 #define EXT4_DEF_MIN_BATCH_TIME	0
2493 #define EXT4_DEF_MAX_BATCH_TIME	15000 /* 15ms */
2494 #define EXT4_DEF_JOURNAL_IOPRIO (IOPRIO_PRIO_VALUE(IOPRIO_CLASS_BE, 3))
2495 
2496 
2497 /*
2498  * Default values for superblock update
2499  */
2500 #define EXT4_DEF_SB_UPDATE_INTERVAL_SEC (3600) /* seconds (1 hour) */
2501 #define EXT4_DEF_SB_UPDATE_INTERVAL_KB (16384) /* kilobytes (16MB) */
2502 
2503 /*
2504  * Minimum number of groups in a flexgroup before we separate out
2505  * directories into the first block group of a flexgroup
2506  */
2507 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME	4
2508 
2509 /*
2510  * Structure of a directory entry
2511  */
2512 #define EXT4_NAME_LEN 255
2513 /*
2514  * Base length of the ext4 directory entry excluding the name length
2515  */
2516 #define EXT4_BASE_DIR_LEN (sizeof(struct ext4_dir_entry_2) - EXT4_NAME_LEN)
2517 
2518 struct ext4_dir_entry {
2519 	__le32	inode;			/* Inode number */
2520 	__le16	rec_len;		/* Directory entry length */
2521 	__le16	name_len;		/* Name length */
2522 	char	name[EXT4_NAME_LEN];	/* File name */
2523 };
2524 
2525 
2526 /*
2527  * Encrypted Casefolded entries require saving the hash on disk. This structure
2528  * followed ext4_dir_entry_2's name[name_len] at the next 4 byte aligned
2529  * boundary.
2530  */
2531 struct ext4_dir_entry_hash {
2532 	__le32 hash;
2533 	__le32 minor_hash;
2534 };
2535 
2536 /*
2537  * The new version of the directory entry.  Since EXT4 structures are
2538  * stored in intel byte order, and the name_len field could never be
2539  * bigger than 255 chars, it's safe to reclaim the extra byte for the
2540  * file_type field.
2541  */
2542 struct ext4_dir_entry_2 {
2543 	__le32	inode;			/* Inode number */
2544 	__le16	rec_len;		/* Directory entry length */
2545 	__u8	name_len;		/* Name length */
2546 	__u8	file_type;		/* See file type macros EXT4_FT_* below */
2547 	char	name[EXT4_NAME_LEN];	/* File name */
2548 };
2549 
2550 /*
2551  * Access the hashes at the end of ext4_dir_entry_2
2552  */
2553 #define EXT4_DIRENT_HASHES(entry) \
2554 	((struct ext4_dir_entry_hash *) \
2555 		(((void *)(entry)) + \
2556 		((8 + (entry)->name_len + EXT4_DIR_ROUND) & ~EXT4_DIR_ROUND)))
2557 #define EXT4_DIRENT_HASH(entry) le32_to_cpu(EXT4_DIRENT_HASHES(entry)->hash)
2558 #define EXT4_DIRENT_MINOR_HASH(entry) \
2559 		le32_to_cpu(EXT4_DIRENT_HASHES(entry)->minor_hash)
2560 
2561 static inline bool ext4_hash_in_dirent(const struct inode *inode)
2562 {
2563 	return IS_CASEFOLDED(inode) && IS_ENCRYPTED(inode);
2564 }
2565 
2566 /*
2567  * This is a bogus directory entry at the end of each leaf block that
2568  * records checksums.
2569  */
2570 struct ext4_dir_entry_tail {
2571 	__le32	det_reserved_zero1;	/* Pretend to be unused */
2572 	__le16	det_rec_len;		/* 12 */
2573 	__u8	det_reserved_zero2;	/* Zero name length */
2574 	__u8	det_reserved_ft;	/* 0xDE, fake file type */
2575 	__le32	det_checksum;		/* crc32c(uuid+inum+dirblock) */
2576 };
2577 
2578 #define EXT4_DIRENT_TAIL(block, blocksize) \
2579 	((struct ext4_dir_entry_tail *)(((void *)(block)) + \
2580 					((blocksize) - \
2581 					 sizeof(struct ext4_dir_entry_tail))))
2582 
2583 /*
2584  * Ext4 directory file types.  Only the low 3 bits are used.  The
2585  * other bits are reserved for now.
2586  */
2587 #define EXT4_FT_UNKNOWN		0
2588 #define EXT4_FT_REG_FILE	1
2589 #define EXT4_FT_DIR		2
2590 #define EXT4_FT_CHRDEV		3
2591 #define EXT4_FT_BLKDEV		4
2592 #define EXT4_FT_FIFO		5
2593 #define EXT4_FT_SOCK		6
2594 #define EXT4_FT_SYMLINK		7
2595 
2596 #define EXT4_FT_MAX		8
2597 
2598 #define EXT4_FT_DIR_CSUM	0xDE
2599 
2600 /*
2601  * EXT4_DIR_PAD defines the directory entries boundaries
2602  *
2603  * NOTE: It must be a multiple of 4
2604  */
2605 #define EXT4_DIR_PAD			4
2606 #define EXT4_DIR_ROUND			(EXT4_DIR_PAD - 1)
2607 #define EXT4_MAX_REC_LEN		((1<<16)-1)
2608 
2609 /*
2610  * The rec_len is dependent on the type of directory. Directories that are
2611  * casefolded and encrypted need to store the hash as well, so we add room for
2612  * ext4_extended_dir_entry_2. For all entries related to '.' or '..' you should
2613  * pass NULL for dir, as those entries do not use the extra fields.
2614  */
2615 static inline unsigned int ext4_dir_rec_len(__u8 name_len,
2616 						const struct inode *dir)
2617 {
2618 	int rec_len = (name_len + 8 + EXT4_DIR_ROUND);
2619 
2620 	if (dir && ext4_hash_in_dirent(dir))
2621 		rec_len += sizeof(struct ext4_dir_entry_hash);
2622 	return (rec_len & ~EXT4_DIR_ROUND);
2623 }
2624 
2625 static inline unsigned int
2626 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
2627 {
2628 	unsigned len = le16_to_cpu(dlen);
2629 
2630 	if (len == EXT4_MAX_REC_LEN || len == 0)
2631 		return blocksize;
2632 	return (len & 65532) | ((len & 3) << 16);
2633 }
2634 
2635 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
2636 {
2637 	BUG_ON((len > blocksize) || (blocksize > (1 << 18)) || (len & 3));
2638 	if (len < 65536)
2639 		return cpu_to_le16(len);
2640 	if (len == blocksize) {
2641 		if (blocksize == 65536)
2642 			return cpu_to_le16(EXT4_MAX_REC_LEN);
2643 		else
2644 			return cpu_to_le16(0);
2645 	}
2646 	return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
2647 }
2648 
2649 /*
2650  * Hash Tree Directory indexing
2651  * (c) Daniel Phillips, 2001
2652  */
2653 
2654 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
2655 		    ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2656 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \
2657 		    !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir)))
2658 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2659 
2660 /* Legal values for the dx_root hash_version field: */
2661 
2662 #define DX_HASH_LEGACY			0
2663 #define DX_HASH_HALF_MD4		1
2664 #define DX_HASH_TEA			2
2665 #define DX_HASH_LEGACY_UNSIGNED		3
2666 #define DX_HASH_HALF_MD4_UNSIGNED	4
2667 #define DX_HASH_TEA_UNSIGNED		5
2668 #define DX_HASH_SIPHASH			6
2669 #define DX_HASH_LAST 			DX_HASH_SIPHASH
2670 
2671 static inline u32 ext4_chksum(u32 crc, const void *address, unsigned int length)
2672 {
2673 	return crc32c(crc, address, length);
2674 }
2675 
2676 #ifdef __KERNEL__
2677 
2678 /* hash info structure used by the directory hash */
2679 struct dx_hash_info
2680 {
2681 	u32		hash;
2682 	u32		minor_hash;
2683 	int		hash_version;
2684 	u32		*seed;
2685 };
2686 
2687 
2688 /* 32 and 64 bit signed EOF for dx directories */
2689 #define EXT4_HTREE_EOF_32BIT   ((1UL  << (32 - 1)) - 1)
2690 #define EXT4_HTREE_EOF_64BIT   ((1ULL << (64 - 1)) - 1)
2691 
2692 
2693 /*
2694  * Control parameters used by ext4_htree_next_block
2695  */
2696 #define HASH_NB_ALWAYS		1
2697 
2698 struct ext4_filename {
2699 	const struct qstr *usr_fname;
2700 	struct fscrypt_str disk_name;
2701 	struct dx_hash_info hinfo;
2702 #ifdef CONFIG_FS_ENCRYPTION
2703 	struct fscrypt_str crypto_buf;
2704 #endif
2705 #if IS_ENABLED(CONFIG_UNICODE)
2706 	struct qstr cf_name;
2707 #endif
2708 };
2709 
2710 #define fname_name(p) ((p)->disk_name.name)
2711 #define fname_usr_name(p) ((p)->usr_fname->name)
2712 #define fname_len(p)  ((p)->disk_name.len)
2713 
2714 /*
2715  * Describe an inode's exact location on disk and in memory
2716  */
2717 struct ext4_iloc
2718 {
2719 	struct buffer_head *bh;
2720 	unsigned long offset;
2721 	ext4_group_t block_group;
2722 };
2723 
2724 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2725 {
2726 	return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2727 }
2728 
2729 static inline bool ext4_is_quota_file(struct inode *inode)
2730 {
2731 	return IS_NOQUOTA(inode) &&
2732 	       !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL);
2733 }
2734 
2735 /*
2736  * This structure is stuffed into the struct file's private_data field
2737  * for directories.  It is where we put information so that we can do
2738  * readdir operations in hash tree order.
2739  */
2740 struct dir_private_info {
2741 	struct rb_root	root;
2742 	struct rb_node	*curr_node;
2743 	struct fname	*extra_fname;
2744 	loff_t		last_pos;
2745 	__u32		curr_hash;
2746 	__u32		curr_minor_hash;
2747 	__u32		next_hash;
2748 	u64		cookie;
2749 	bool		initialized;
2750 };
2751 
2752 /* calculate the first block number of the group */
2753 static inline ext4_fsblk_t
2754 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2755 {
2756 	return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2757 		le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2758 }
2759 
2760 /*
2761  * Special error return code only used by dx_probe() and its callers.
2762  */
2763 #define ERR_BAD_DX_DIR	(-(MAX_ERRNO - 1))
2764 
2765 /* htree levels for ext4 */
2766 #define	EXT4_HTREE_LEVEL_COMPAT	2
2767 #define	EXT4_HTREE_LEVEL	3
2768 
2769 static inline int ext4_dir_htree_level(struct super_block *sb)
2770 {
2771 	return ext4_has_feature_largedir(sb) ?
2772 		EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT;
2773 }
2774 
2775 /*
2776  * Timeout and state flag for lazy initialization inode thread.
2777  */
2778 #define EXT4_DEF_LI_WAIT_MULT			10
2779 #define EXT4_DEF_LI_MAX_START_DELAY		5
2780 #define EXT4_LAZYINIT_QUIT			0x0001
2781 #define EXT4_LAZYINIT_RUNNING			0x0002
2782 
2783 /*
2784  * Lazy inode table initialization info
2785  */
2786 struct ext4_lazy_init {
2787 	unsigned long		li_state;
2788 	struct list_head	li_request_list;
2789 	struct mutex		li_list_mtx;
2790 };
2791 
2792 enum ext4_li_mode {
2793 	EXT4_LI_MODE_PREFETCH_BBITMAP,
2794 	EXT4_LI_MODE_ITABLE,
2795 };
2796 
2797 struct ext4_li_request {
2798 	struct super_block	*lr_super;
2799 	enum ext4_li_mode	lr_mode;
2800 	ext4_group_t		lr_first_not_zeroed;
2801 	ext4_group_t		lr_next_group;
2802 	struct list_head	lr_request;
2803 	unsigned long		lr_next_sched;
2804 	unsigned long		lr_timeout;
2805 };
2806 
2807 struct ext4_features {
2808 	struct kobject f_kobj;
2809 	struct completion f_kobj_unregister;
2810 };
2811 
2812 /*
2813  * This structure will be used for multiple mount protection. It will be
2814  * written into the block number saved in the s_mmp_block field in the
2815  * superblock. Programs that check MMP should assume that if
2816  * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2817  * to use the filesystem, regardless of how old the timestamp is.
2818  */
2819 #define EXT4_MMP_MAGIC     0x004D4D50U /* ASCII for MMP */
2820 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2821 #define EXT4_MMP_SEQ_FSCK  0xE24D4D50U /* mmp_seq value when being fscked */
2822 #define EXT4_MMP_SEQ_MAX   0xE24D4D4FU /* maximum valid mmp_seq value */
2823 
2824 struct mmp_struct {
2825 	__le32	mmp_magic;		/* Magic number for MMP */
2826 	__le32	mmp_seq;		/* Sequence no. updated periodically */
2827 
2828 	/*
2829 	 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2830 	 * purposes and do not affect the correctness of the algorithm
2831 	 */
2832 	__le64	mmp_time;		/* Time last updated */
2833 	char	mmp_nodename[64];	/* Node which last updated MMP block */
2834 	char	mmp_bdevname[32];	/* Bdev which last updated MMP block */
2835 
2836 	/*
2837 	 * mmp_check_interval is used to verify if the MMP block has been
2838 	 * updated on the block device. The value is updated based on the
2839 	 * maximum time to write the MMP block during an update cycle.
2840 	 */
2841 	__le16	mmp_check_interval;
2842 
2843 	__le16	mmp_pad1;
2844 	__le32	mmp_pad2[226];
2845 	__le32	mmp_checksum;		/* crc32c(uuid+mmp_block) */
2846 };
2847 
2848 /* arguments passed to the mmp thread */
2849 struct mmpd_data {
2850 	struct buffer_head *bh; /* bh from initial read_mmp_block() */
2851 	struct super_block *sb;  /* super block of the fs */
2852 };
2853 
2854 /*
2855  * Check interval multiplier
2856  * The MMP block is written every update interval and initially checked every
2857  * update interval x the multiplier (the value is then adapted based on the
2858  * write latency). The reason is that writes can be delayed under load and we
2859  * don't want readers to incorrectly assume that the filesystem is no longer
2860  * in use.
2861  */
2862 #define EXT4_MMP_CHECK_MULT		2UL
2863 
2864 /*
2865  * Minimum interval for MMP checking in seconds.
2866  */
2867 #define EXT4_MMP_MIN_CHECK_INTERVAL	5UL
2868 
2869 /*
2870  * Maximum interval for MMP checking in seconds.
2871  */
2872 #define EXT4_MMP_MAX_CHECK_INTERVAL	300UL
2873 
2874 /*
2875  * Function prototypes
2876  */
2877 
2878 /*
2879  * Ok, these declarations are also in <linux/kernel.h> but none of the
2880  * ext4 source programs needs to include it so they are duplicated here.
2881  */
2882 # define NORET_TYPE	/**/
2883 # define ATTRIB_NORET	__attribute__((noreturn))
2884 # define NORET_AND	noreturn,
2885 
2886 /* bitmap.c */
2887 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2888 void ext4_inode_bitmap_csum_set(struct super_block *sb,
2889 				struct ext4_group_desc *gdp,
2890 				struct buffer_head *bh);
2891 int ext4_inode_bitmap_csum_verify(struct super_block *sb,
2892 				  struct ext4_group_desc *gdp,
2893 				  struct buffer_head *bh);
2894 void ext4_block_bitmap_csum_set(struct super_block *sb,
2895 				struct ext4_group_desc *gdp,
2896 				struct buffer_head *bh);
2897 int ext4_block_bitmap_csum_verify(struct super_block *sb,
2898 				  struct ext4_group_desc *gdp,
2899 				  struct buffer_head *bh);
2900 
2901 /* balloc.c */
2902 extern void ext4_get_group_no_and_offset(struct super_block *sb,
2903 					 ext4_fsblk_t blocknr,
2904 					 ext4_group_t *blockgrpp,
2905 					 ext4_grpblk_t *offsetp);
2906 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2907 					  ext4_fsblk_t block);
2908 
2909 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2910 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2911 			ext4_group_t group);
2912 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2913 					 ext4_fsblk_t goal,
2914 					 unsigned int flags,
2915 					 unsigned long *count,
2916 					 int *errp);
2917 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2918 				    s64 nclusters, unsigned int flags);
2919 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2920 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2921 						    ext4_group_t block_group,
2922 						    struct buffer_head ** bh);
2923 extern struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2924 						   ext4_group_t group);
2925 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2926 
2927 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2928 						ext4_group_t block_group,
2929 						bool ignore_locked);
2930 extern int ext4_wait_block_bitmap(struct super_block *sb,
2931 				  ext4_group_t block_group,
2932 				  struct buffer_head *bh);
2933 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2934 						  ext4_group_t block_group);
2935 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2936 					      ext4_group_t block_group,
2937 					      struct ext4_group_desc *gdp);
2938 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2939 
2940 #if IS_ENABLED(CONFIG_UNICODE)
2941 extern int ext4_fname_setup_ci_filename(struct inode *dir,
2942 					const struct qstr *iname,
2943 					struct ext4_filename *fname);
2944 
2945 static inline void ext4_fname_free_ci_filename(struct ext4_filename *fname)
2946 {
2947 	kfree(fname->cf_name.name);
2948 	fname->cf_name.name = NULL;
2949 }
2950 #else
2951 static inline int ext4_fname_setup_ci_filename(struct inode *dir,
2952 					       const struct qstr *iname,
2953 					       struct ext4_filename *fname)
2954 {
2955 	return 0;
2956 }
2957 
2958 static inline void ext4_fname_free_ci_filename(struct ext4_filename *fname)
2959 {
2960 }
2961 #endif
2962 
2963 /* ext4 encryption related stuff goes here crypto.c */
2964 #ifdef CONFIG_FS_ENCRYPTION
2965 extern const struct fscrypt_operations ext4_cryptops;
2966 
2967 int ext4_fname_setup_filename(struct inode *dir, const struct qstr *iname,
2968 			      int lookup, struct ext4_filename *fname);
2969 
2970 int ext4_fname_prepare_lookup(struct inode *dir, struct dentry *dentry,
2971 			      struct ext4_filename *fname);
2972 
2973 void ext4_fname_free_filename(struct ext4_filename *fname);
2974 
2975 int ext4_ioctl_get_encryption_pwsalt(struct file *filp, void __user *arg);
2976 
2977 #else /* !CONFIG_FS_ENCRYPTION */
2978 static inline int ext4_fname_setup_filename(struct inode *dir,
2979 					    const struct qstr *iname,
2980 					    int lookup,
2981 					    struct ext4_filename *fname)
2982 {
2983 	fname->usr_fname = iname;
2984 	fname->disk_name.name = (unsigned char *) iname->name;
2985 	fname->disk_name.len = iname->len;
2986 
2987 	return ext4_fname_setup_ci_filename(dir, iname, fname);
2988 }
2989 
2990 static inline int ext4_fname_prepare_lookup(struct inode *dir,
2991 					    struct dentry *dentry,
2992 					    struct ext4_filename *fname)
2993 {
2994 	return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname);
2995 }
2996 
2997 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2998 {
2999 	ext4_fname_free_ci_filename(fname);
3000 }
3001 
3002 static inline int ext4_ioctl_get_encryption_pwsalt(struct file *filp,
3003 						   void __user *arg)
3004 {
3005 	return -EOPNOTSUPP;
3006 }
3007 #endif /* !CONFIG_FS_ENCRYPTION */
3008 
3009 /* dir.c */
3010 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
3011 				  struct file *,
3012 				  struct ext4_dir_entry_2 *,
3013 				  struct buffer_head *, char *, int,
3014 				  unsigned int);
3015 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
3016 	unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
3017 				(de), (bh), (buf), (size), (offset)))
3018 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
3019 				__u32 minor_hash,
3020 				struct ext4_dir_entry_2 *dirent,
3021 				struct fscrypt_str *ent_name);
3022 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
3023 extern int ext4_find_dest_de(struct inode *dir, struct buffer_head *bh,
3024 			     void *buf, int buf_size,
3025 			     struct ext4_filename *fname,
3026 			     struct ext4_dir_entry_2 **dest_de);
3027 void ext4_insert_dentry(struct inode *dir, struct inode *inode,
3028 			struct ext4_dir_entry_2 *de,
3029 			int buf_size,
3030 			struct ext4_filename *fname);
3031 static inline void ext4_update_dx_flag(struct inode *inode)
3032 {
3033 	if (!ext4_has_feature_dir_index(inode->i_sb) &&
3034 	    ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) {
3035 		/* ext4_iget() should have caught this... */
3036 		WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb));
3037 		ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
3038 	}
3039 }
3040 static const unsigned char ext4_filetype_table[] = {
3041 	DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
3042 };
3043 
3044 static inline  unsigned char get_dtype(struct super_block *sb, int filetype)
3045 {
3046 	if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
3047 		return DT_UNKNOWN;
3048 
3049 	return ext4_filetype_table[filetype];
3050 }
3051 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
3052 			     void *buf, int buf_size);
3053 
3054 /* fsync.c */
3055 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
3056 
3057 /* hash.c */
3058 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len,
3059 			  struct dx_hash_info *hinfo);
3060 
3061 /* ialloc.c */
3062 extern int ext4_mark_inode_used(struct super_block *sb, int ino);
3063 extern struct inode *__ext4_new_inode(const struct mnt_idmap *, handle_t *,
3064 				      struct inode *, umode_t,
3065 				      const struct qstr *qstr, __u32 goal,
3066 				      uid_t *owner, __u32 i_flags,
3067 				      int handle_type, unsigned int line_no,
3068 				      int nblocks);
3069 
3070 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags)          \
3071 	__ext4_new_inode(&nop_mnt_idmap, (handle), (dir), (mode), (qstr),      \
3072 			 (goal), (owner), i_flags, 0, 0, 0)
3073 #define ext4_new_inode_start_handle(idmap, dir, mode, qstr, goal, owner, \
3074 				    type, nblocks)		    \
3075 	__ext4_new_inode((idmap), NULL, (dir), (mode), (qstr), (goal), (owner), \
3076 			 0, (type), __LINE__, (nblocks))
3077 
3078 
3079 extern void ext4_free_inode(handle_t *, struct inode *);
3080 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
3081 extern unsigned long ext4_count_free_inodes(struct super_block *);
3082 extern unsigned long ext4_count_dirs(struct super_block *);
3083 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
3084 extern int ext4_init_inode_table(struct super_block *sb,
3085 				 ext4_group_t group, int barrier);
3086 void ext4_end_bitmap_read(struct bio *bio);
3087 
3088 /* fast_commit.c */
3089 int ext4_fc_info_show(struct seq_file *seq, void *v);
3090 void ext4_fc_init(struct super_block *sb, journal_t *journal);
3091 void ext4_fc_init_inode(struct inode *inode);
3092 void ext4_fc_track_range(handle_t *handle, struct inode *inode, ext4_lblk_t start,
3093 			 ext4_lblk_t end);
3094 void __ext4_fc_track_unlink(handle_t *handle, struct inode *inode,
3095 	struct dentry *dentry);
3096 void __ext4_fc_track_link(handle_t *handle, struct inode *inode,
3097 	struct dentry *dentry);
3098 void ext4_fc_track_unlink(handle_t *handle, struct dentry *dentry);
3099 void ext4_fc_track_link(handle_t *handle, struct inode *inode,
3100 			struct dentry *dentry);
3101 void __ext4_fc_track_create(handle_t *handle, struct inode *inode,
3102 			    struct dentry *dentry);
3103 void ext4_fc_track_create(handle_t *handle, struct dentry *dentry);
3104 void ext4_fc_track_inode(handle_t *handle, struct inode *inode);
3105 void ext4_fc_mark_ineligible(struct super_block *sb, int reason, handle_t *handle);
3106 void ext4_fc_del(struct inode *inode);
3107 bool ext4_fc_replay_check_excluded(struct super_block *sb, ext4_fsblk_t block);
3108 void ext4_fc_replay_cleanup(struct super_block *sb);
3109 int ext4_fc_commit(journal_t *journal, tid_t commit_tid);
3110 int __init ext4_fc_init_dentry_cache(void);
3111 void ext4_fc_destroy_dentry_cache(void);
3112 int ext4_fc_record_regions(struct super_block *sb, int ino,
3113 			   ext4_lblk_t lblk, ext4_fsblk_t pblk,
3114 			   int len, int replay);
3115 
3116 /* mballoc.c */
3117 extern const struct seq_operations ext4_mb_seq_groups_ops;
3118 extern const struct seq_operations ext4_mb_seq_structs_summary_ops;
3119 extern int ext4_seq_mb_stats_show(struct seq_file *seq, void *offset);
3120 extern int ext4_mb_init(struct super_block *);
3121 extern void ext4_mb_release(struct super_block *);
3122 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
3123 				struct ext4_allocation_request *, int *);
3124 extern void ext4_discard_preallocations(struct inode *);
3125 extern int __init ext4_init_mballoc(void);
3126 extern void ext4_exit_mballoc(void);
3127 extern ext4_group_t ext4_mb_prefetch(struct super_block *sb,
3128 				     ext4_group_t group,
3129 				     unsigned int nr, int *cnt);
3130 extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group,
3131 				  unsigned int nr);
3132 
3133 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
3134 			     struct buffer_head *bh, ext4_fsblk_t block,
3135 			     unsigned long count, int flags);
3136 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
3137 				   ext4_group_t ngroups);
3138 extern int ext4_mb_add_groupinfo(struct super_block *sb,
3139 		ext4_group_t i, struct ext4_group_desc *desc);
3140 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
3141 				ext4_fsblk_t block, unsigned long count);
3142 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
3143 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid);
3144 extern void ext4_mb_mark_bb(struct super_block *sb, ext4_fsblk_t block,
3145 			    int len, bool state);
3146 static inline bool ext4_mb_cr_expensive(enum criteria cr)
3147 {
3148 	return cr >= CR_GOAL_LEN_SLOW;
3149 }
3150 
3151 /* inode.c */
3152 void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw,
3153 			 struct ext4_inode_info *ei);
3154 int ext4_inode_is_fast_symlink(struct inode *inode);
3155 void ext4_check_map_extents_env(struct inode *inode);
3156 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
3157 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
3158 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count,
3159 		     bool wait, struct buffer_head **bhs);
3160 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
3161 			     struct buffer_head *bh_result, int create);
3162 int ext4_get_block(struct inode *inode, sector_t iblock,
3163 		   struct buffer_head *bh_result, int create);
3164 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
3165 			   struct buffer_head *bh, int create);
3166 int ext4_walk_page_buffers(handle_t *handle,
3167 			   struct inode *inode,
3168 			   struct buffer_head *head,
3169 			   unsigned from,
3170 			   unsigned to,
3171 			   int *partial,
3172 			   int (*fn)(handle_t *handle, struct inode *inode,
3173 				     struct buffer_head *bh));
3174 int do_journal_get_write_access(handle_t *handle, struct inode *inode,
3175 				struct buffer_head *bh);
3176 void ext4_set_inode_mapping_order(struct inode *inode);
3177 void ext4_enable_buffered_iomap(struct inode *inode);
3178 int ext4_nonda_switch(struct super_block *sb);
3179 #define FALL_BACK_TO_NONDELALLOC 1
3180 #define EXT4_WRITE_DATA_INLINE	 2
3181 
3182 typedef enum {
3183 	EXT4_IGET_NORMAL =	0,
3184 	EXT4_IGET_SPECIAL =	0x0001, /* OK to iget a system inode */
3185 	EXT4_IGET_HANDLE = 	0x0002,	/* Inode # is from a handle */
3186 	EXT4_IGET_BAD =		0x0004, /* Allow to iget a bad inode */
3187 	EXT4_IGET_EA_INODE =	0x0008,	/* Inode should contain an EA value */
3188 	EXT4_IGET_NOWAIT =	0x0010	/* Non-blocking lookup (skip if freeing) */
3189 } ext4_iget_flags;
3190 
3191 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
3192 				 ext4_iget_flags flags, const char *function,
3193 				 unsigned int line);
3194 
3195 #define ext4_iget(sb, ino, flags) \
3196 	__ext4_iget((sb), (ino), (flags), __func__, __LINE__)
3197 
3198 extern int  ext4_write_inode(struct inode *, struct writeback_control *);
3199 extern int  ext4_sync_inode_metadata(struct inode *, struct writeback_control *);
3200 extern int  ext4_setattr(const struct mnt_idmap *, struct dentry *,
3201 			 struct iattr *);
3202 extern u32  ext4_dio_alignment(struct inode *inode);
3203 extern int  ext4_getattr(const struct mnt_idmap *, const struct path *,
3204 			 struct kstat *, u32, unsigned int);
3205 extern void ext4_evict_inode(struct inode *);
3206 extern void ext4_clear_inode(struct inode *);
3207 extern int  ext4_file_getattr(const struct mnt_idmap *, const struct path *,
3208 			      struct kstat *, u32, unsigned int);
3209 extern void ext4_dirty_inode(struct inode *, int);
3210 extern int ext4_change_inode_journal_flag(struct inode *, int);
3211 int ext4_get_inode_loc(struct inode *inode, struct ext4_iloc *iloc);
3212 int ext4_get_inode_loc_noio(struct inode *inode, struct ext4_iloc *iloc);
3213 int ext4_get_fc_inode_loc(struct super_block *sb, unsigned long ino,
3214 			  struct ext4_iloc *iloc);
3215 extern int ext4_inode_attach_jinode(struct inode *inode);
3216 extern int ext4_can_truncate(struct inode *inode);
3217 extern int ext4_truncate(struct inode *);
3218 extern int ext4_break_layouts(struct inode *);
3219 extern int ext4_truncate_page_cache_block_range(struct inode *inode,
3220 						loff_t start, loff_t end);
3221 extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length);
3222 extern void ext4_set_inode_flags(struct inode *, bool init);
3223 extern int ext4_alloc_da_blocks(struct inode *inode);
3224 extern void ext4_set_aops(struct inode *inode);
3225 extern int ext4_normal_submit_inode_data_buffers(struct jbd2_inode *jinode);
3226 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
3227 extern int ext4_chunk_trans_extent(struct inode *inode, int nrblocks);
3228 extern int ext4_meta_trans_blocks(struct inode *inode, int lblocks,
3229 				  int pextents, int alloc_extents);
3230 void ext4_iomap_clear_disksize_pending(struct inode *inode);
3231 void ext4_iomap_wait_disksize_pending(struct inode *inode);
3232 unsigned int ext4_iomap_get_disksize_pending_range(struct inode *inode,
3233 						   loff_t *start);
3234 extern int ext4_iomap_sync_zeroed_eof(struct inode *inode,
3235 				      loff_t offset, loff_t end);
3236 extern int ext4_block_zero_eof(struct inode *inode, loff_t from, loff_t end);
3237 
3238 #define EXT4_PARTIAL_ZERO_START	0x1
3239 #define EXT4_PARTIAL_ZERO_END	0x2
3240 extern int ext4_zero_partial_blocks(struct inode *inode, loff_t lstart,
3241 				    loff_t length, unsigned int *partial_zeroed);
3242 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf);
3243 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
3244 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
3245 extern void ext4_da_release_space(struct inode *inode, int to_free);
3246 extern void ext4_da_update_reserve_space(struct inode *inode,
3247 					int used, int quota_claim);
3248 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
3249 			      ext4_fsblk_t pblk, ext4_lblk_t len);
3250 
3251 static inline bool is_special_ino(struct super_block *sb, unsigned long ino)
3252 {
3253 	struct ext4_super_block *es = EXT4_SB(sb)->s_es;
3254 
3255 	return (ino < EXT4_FIRST_INO(sb) && ino != EXT4_ROOT_INO) ||
3256 		ino == le32_to_cpu(es->s_usr_quota_inum) ||
3257 		ino == le32_to_cpu(es->s_grp_quota_inum) ||
3258 		ino == le32_to_cpu(es->s_prj_quota_inum) ||
3259 		ino == le32_to_cpu(es->s_orphan_file_inum);
3260 }
3261 
3262 /* indirect.c */
3263 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
3264 				struct ext4_map_blocks *map, int flags);
3265 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
3266 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
3267 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
3268 				 ext4_lblk_t start, ext4_lblk_t end);
3269 
3270 /* ioctl.c */
3271 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
3272 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
3273 int ext4_fileattr_set(const struct mnt_idmap *idmap,
3274 		      struct dentry *dentry, struct file_kattr *fa);
3275 int ext4_fileattr_get(struct dentry *dentry, struct file_kattr *fa);
3276 extern void ext4_reset_inode_seed(struct inode *inode);
3277 int ext4_update_overhead(struct super_block *sb, bool force);
3278 int ext4_force_shutdown(struct super_block *sb, u32 flags);
3279 
3280 /* migrate.c */
3281 extern int ext4_ext_migrate(struct inode *);
3282 extern int ext4_ind_migrate(struct inode *inode);
3283 
3284 /* namei.c */
3285 extern int ext4_init_new_dir(handle_t *handle, struct inode *dir,
3286 			     struct inode *inode);
3287 extern int ext4_dirblock_csum_verify(struct inode *inode,
3288 				     struct buffer_head *bh);
3289 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
3290 				__u32 start_minor_hash, __u32 *next_hash);
3291 extern int ext4_search_dir(struct buffer_head *bh,
3292 			   char *search_buf,
3293 			   int buf_size,
3294 			   struct inode *dir,
3295 			   struct ext4_filename *fname,
3296 			   unsigned int offset,
3297 			   struct ext4_dir_entry_2 **res_dir);
3298 extern int ext4_generic_delete_entry(struct inode *dir,
3299 				     struct ext4_dir_entry_2 *de_del,
3300 				     struct buffer_head *bh,
3301 				     void *entry_buf,
3302 				     int buf_size,
3303 				     int csum_size);
3304 extern bool ext4_empty_dir(struct inode *inode);
3305 
3306 /* resize.c */
3307 extern void ext4_kvfree_array_rcu(void *to_free);
3308 extern int ext4_group_add(struct super_block *sb,
3309 				struct ext4_new_group_data *input);
3310 extern int ext4_group_extend(struct super_block *sb,
3311 				struct ext4_super_block *es,
3312 				ext4_fsblk_t n_blocks_count);
3313 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
3314 extern unsigned int ext4_list_backups(struct super_block *sb,
3315 				      unsigned int *three, unsigned int *five,
3316 				      unsigned int *seven);
3317 
3318 /* super.c */
3319 extern struct buffer_head *ext4_sb_bread(struct super_block *sb,
3320 					 sector_t block, blk_opf_t op_flags);
3321 extern struct buffer_head *ext4_sb_bread_unmovable(struct super_block *sb,
3322 						   sector_t block);
3323 extern struct buffer_head *ext4_sb_bread_nofail(struct super_block *sb,
3324 						sector_t block);
3325 void ext4_read_bh_nowait(struct buffer_head *bh, blk_opf_t op_flags,
3326 		bio_end_io_t end_io, bool simu_fail);
3327 int ext4_read_bh(struct buffer_head *bh, blk_opf_t op_flags,
3328 		bio_end_io_t end_io, bool simu_fail);
3329 extern int ext4_read_bh_lock(struct buffer_head *bh, blk_opf_t op_flags, bool wait);
3330 extern void ext4_sb_breadahead_unmovable(struct super_block *sb, sector_t block);
3331 extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
3332 extern int ext4_calculate_overhead(struct super_block *sb);
3333 extern __le32 ext4_superblock_csum(struct ext4_super_block *es);
3334 extern void ext4_superblock_csum_set(struct super_block *sb);
3335 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
3336 				    ext4_group_t ngroup);
3337 extern const char *ext4_decode_error(struct super_block *sb, int errno,
3338 				     char nbuf[16]);
3339 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb,
3340 					     ext4_group_t block_group,
3341 					     unsigned int flags);
3342 extern unsigned int ext4_num_base_meta_blocks(struct super_block *sb,
3343 					      ext4_group_t block_group);
3344 extern void print_daily_error_info(struct timer_list *t);
3345 
3346 extern __printf(7, 8)
3347 void __ext4_error(struct super_block *, const char *, unsigned int, bool,
3348 		  int, __u64, const char *, ...);
3349 extern __printf(6, 7)
3350 void __ext4_error_inode(struct inode *, const char *, unsigned int,
3351 			ext4_fsblk_t, int, const char *, ...);
3352 extern __printf(5, 6)
3353 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
3354 		     const char *, ...);
3355 extern void __ext4_std_error(struct super_block *, const char *,
3356 			     unsigned int, int);
3357 extern __printf(4, 5)
3358 void __ext4_warning(struct super_block *, const char *, unsigned int,
3359 		    const char *, ...);
3360 extern __printf(4, 5)
3361 void __ext4_warning_inode(const struct inode *inode, const char *function,
3362 			  unsigned int line, const char *fmt, ...);
3363 extern __printf(3, 4)
3364 void __ext4_msg(struct super_block *, const char *, const char *, ...);
3365 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
3366 			   const char *, unsigned int, const char *);
3367 extern __printf(7, 8)
3368 void __ext4_grp_locked_error(const char *, unsigned int,
3369 			     struct super_block *, ext4_group_t,
3370 			     u64, ext4_fsblk_t,
3371 			     const char *, ...);
3372 
3373 #define EXT4_ERROR_INODE(inode, fmt, a...) \
3374 	ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
3375 
3376 #define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...)			\
3377 	__ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a)
3378 
3379 #define ext4_error_inode_block(inode, block, err, fmt, a...)		\
3380 	__ext4_error_inode((inode), __func__, __LINE__, (block), (err),	\
3381 			   (fmt), ## a)
3382 
3383 #define EXT4_ERROR_FILE(file, block, fmt, a...)				\
3384 	ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
3385 
3386 #define ext4_abort(sb, err, fmt, a...)					\
3387 	__ext4_error((sb), __func__, __LINE__, true, (err), 0, (fmt), ## a)
3388 
3389 #ifdef CONFIG_PRINTK
3390 
3391 #define ext4_error_inode(inode, func, line, block, fmt, ...)		\
3392 	__ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__)
3393 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...)	\
3394 	__ext4_error_inode((inode), (func), (line), (block), 		\
3395 			   (err), (fmt), ##__VA_ARGS__)
3396 #define ext4_error_file(file, func, line, block, fmt, ...)		\
3397 	__ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
3398 #define ext4_error(sb, fmt, ...)					\
3399 	__ext4_error((sb), __func__, __LINE__, false, 0, 0, (fmt),	\
3400 		##__VA_ARGS__)
3401 #define ext4_error_err(sb, err, fmt, ...)				\
3402 	__ext4_error((sb), __func__, __LINE__, false, (err), 0, (fmt),	\
3403 		##__VA_ARGS__)
3404 #define ext4_warning(sb, fmt, ...)					\
3405 	__ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
3406 #define ext4_warning_inode(inode, fmt, ...)				\
3407 	__ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
3408 #define ext4_msg(sb, level, fmt, ...)				\
3409 	__ext4_msg(sb, level, fmt, ##__VA_ARGS__)
3410 #define dump_mmp_msg(sb, mmp, msg)					\
3411 	__dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
3412 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...)		\
3413 	__ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
3414 				fmt, ##__VA_ARGS__)
3415 
3416 #else
3417 
3418 #define ext4_error_inode(inode, func, line, block, fmt, ...)		\
3419 do {									\
3420 	no_printk(fmt, ##__VA_ARGS__);					\
3421 	__ext4_error_inode(inode, "", 0, block, 0, " ");		\
3422 } while (0)
3423 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...)	\
3424 do {									\
3425 	no_printk(fmt, ##__VA_ARGS__);					\
3426 	__ext4_error_inode(inode, "", 0, block, err, " ");		\
3427 } while (0)
3428 #define ext4_error_file(file, func, line, block, fmt, ...)		\
3429 do {									\
3430 	no_printk(fmt, ##__VA_ARGS__);					\
3431 	__ext4_error_file(file, "", 0, block, " ");			\
3432 } while (0)
3433 #define ext4_error(sb, fmt, ...)					\
3434 do {									\
3435 	no_printk(fmt, ##__VA_ARGS__);					\
3436 	__ext4_error(sb, "", 0, false, 0, 0, " ");			\
3437 } while (0)
3438 #define ext4_error_err(sb, err, fmt, ...)				\
3439 do {									\
3440 	no_printk(fmt, ##__VA_ARGS__);					\
3441 	__ext4_error(sb, "", 0, false, err, 0, " ");			\
3442 } while (0)
3443 #define ext4_warning(sb, fmt, ...)					\
3444 do {									\
3445 	no_printk(fmt, ##__VA_ARGS__);					\
3446 	__ext4_warning(sb, "", 0, " ");					\
3447 } while (0)
3448 #define ext4_warning_inode(inode, fmt, ...)				\
3449 do {									\
3450 	no_printk(fmt, ##__VA_ARGS__);					\
3451 	__ext4_warning_inode(inode, "", 0, " ");			\
3452 } while (0)
3453 #define ext4_msg(sb, level, fmt, ...)					\
3454 do {									\
3455 	no_printk(fmt, ##__VA_ARGS__);					\
3456 	__ext4_msg(sb, "", " ");					\
3457 } while (0)
3458 #define dump_mmp_msg(sb, mmp, msg)					\
3459 	__dump_mmp_msg(sb, mmp, "", 0, "")
3460 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...)		\
3461 do {									\
3462 	no_printk(fmt, ##__VA_ARGS__);				\
3463 	__ext4_grp_locked_error("", 0, sb, grp, ino, block, " ");	\
3464 } while (0)
3465 
3466 #endif
3467 
3468 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
3469 				      struct ext4_group_desc *bg);
3470 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
3471 				      struct ext4_group_desc *bg);
3472 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
3473 				     struct ext4_group_desc *bg);
3474 extern __u32 ext4_free_group_clusters(struct super_block *sb,
3475 				      struct ext4_group_desc *bg);
3476 extern __u32 ext4_free_inodes_count(struct super_block *sb,
3477 				 struct ext4_group_desc *bg);
3478 extern __u32 ext4_used_dirs_count(struct super_block *sb,
3479 				struct ext4_group_desc *bg);
3480 extern __u32 ext4_itable_unused_count(struct super_block *sb,
3481 				   struct ext4_group_desc *bg);
3482 extern void ext4_block_bitmap_set(struct super_block *sb,
3483 				  struct ext4_group_desc *bg, ext4_fsblk_t blk);
3484 extern void ext4_inode_bitmap_set(struct super_block *sb,
3485 				  struct ext4_group_desc *bg, ext4_fsblk_t blk);
3486 extern void ext4_inode_table_set(struct super_block *sb,
3487 				 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3488 extern void ext4_free_group_clusters_set(struct super_block *sb,
3489 					 struct ext4_group_desc *bg,
3490 					 __u32 count);
3491 extern void ext4_free_inodes_set(struct super_block *sb,
3492 				struct ext4_group_desc *bg, __u32 count);
3493 extern void ext4_used_dirs_set(struct super_block *sb,
3494 				struct ext4_group_desc *bg, __u32 count);
3495 extern void ext4_itable_unused_set(struct super_block *sb,
3496 				   struct ext4_group_desc *bg, __u32 count);
3497 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
3498 				       struct ext4_group_desc *gdp);
3499 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
3500 				     struct ext4_group_desc *gdp);
3501 extern int ext4_register_li_request(struct super_block *sb,
3502 				    ext4_group_t first_not_zeroed);
3503 
3504 static inline int ext4_has_group_desc_csum(struct super_block *sb)
3505 {
3506 	return ext4_has_feature_gdt_csum(sb) ||
3507 	       ext4_has_feature_metadata_csum(sb);
3508 }
3509 
3510 #define ext4_read_incompat_64bit_val(es, name) \
3511 	(((es)->s_feature_incompat & cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT) \
3512 		? (ext4_fsblk_t)le32_to_cpu(es->name##_hi) << 32 : 0) | \
3513 		le32_to_cpu(es->name##_lo))
3514 
3515 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
3516 {
3517 	return ext4_read_incompat_64bit_val(es, s_blocks_count);
3518 }
3519 
3520 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
3521 {
3522 	return ext4_read_incompat_64bit_val(es, s_r_blocks_count);
3523 }
3524 
3525 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
3526 {
3527 	return ext4_read_incompat_64bit_val(es, s_free_blocks_count);
3528 }
3529 
3530 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
3531 					 ext4_fsblk_t blk)
3532 {
3533 	es->s_blocks_count_lo = cpu_to_le32((u32)blk);
3534 	es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
3535 }
3536 
3537 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
3538 					      ext4_fsblk_t blk)
3539 {
3540 	es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
3541 	es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
3542 }
3543 
3544 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
3545 					   ext4_fsblk_t blk)
3546 {
3547 	es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
3548 	es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
3549 }
3550 
3551 static inline loff_t ext4_isize(struct super_block *sb,
3552 				struct ext4_inode *raw_inode)
3553 {
3554 	if (ext4_has_feature_largedir(sb) ||
3555 	    S_ISREG(le16_to_cpu(raw_inode->i_mode)))
3556 		return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
3557 			le32_to_cpu(raw_inode->i_size_lo);
3558 
3559 	return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
3560 }
3561 
3562 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
3563 {
3564 	raw_inode->i_size_lo = cpu_to_le32(i_size);
3565 	raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
3566 }
3567 
3568 /*
3569  * Reading s_groups_count uses acquire semantics.
3570  */
3571 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
3572 {
3573 	/* Pairs with smp_store_release() in ext4_update_super() */
3574 	return smp_load_acquire(&EXT4_SB(sb)->s_groups_count);
3575 }
3576 
3577 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
3578 					     ext4_group_t block_group)
3579 {
3580 	return block_group >> sbi->s_log_groups_per_flex;
3581 }
3582 
3583 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
3584 {
3585 	return 1 << sbi->s_log_groups_per_flex;
3586 }
3587 
3588 static inline loff_t ext4_get_maxbytes(struct inode *inode)
3589 {
3590 	if (ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS))
3591 		return inode->i_sb->s_maxbytes;
3592 	return EXT4_SB(inode->i_sb)->s_bitmap_maxbytes;
3593 }
3594 
3595 #define ext4_std_error(sb, errno)				\
3596 do {								\
3597 	if ((errno))						\
3598 		__ext4_std_error((sb), __func__, __LINE__, (errno));	\
3599 } while (0)
3600 
3601 #ifdef CONFIG_SMP
3602 /* Each CPU can accumulate percpu_counter_batch clusters in their local
3603  * counters. So we need to make sure we have free clusters more
3604  * than percpu_counter_batch  * nr_cpu_ids. Also add a window of 4 times.
3605  */
3606 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
3607 #else
3608 #define EXT4_FREECLUSTERS_WATERMARK 0
3609 #endif
3610 
3611 /*
3612  * Update i_disksize. Requires i_rwsem to avoid races with truncate.
3613  *
3614  * In the iomap buffered I/O path, the EXT4_STATE_DISKSIZE_GROW_PENDING
3615  * inode state bit indicates that the zeroed EOF partial block which
3616  * straddles i_disksize is still waiting writeback.  In that case,
3617  * i_disksize will be updated after the pending zeroed data has been
3618  * written out.
3619  */
3620 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
3621 {
3622 	WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
3623 		     !inode_is_locked(inode));
3624 	down_write(&EXT4_I(inode)->i_data_sem);
3625 	if (newsize > EXT4_I(inode)->i_disksize &&
3626 	    !ext4_test_inode_state(inode, EXT4_STATE_DISKSIZE_GROW_PENDING))
3627 		WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize);
3628 	up_write(&EXT4_I(inode)->i_data_sem);
3629 }
3630 
3631 static inline void __ext4_set_i_disksize(struct inode *inode, loff_t newsize)
3632 {
3633 	WARN_ON_ONCE(!rwsem_is_locked(&EXT4_I(inode)->i_data_sem));
3634 
3635 	if (newsize < EXT4_I(inode)->i_disksize ||
3636 	    !ext4_test_inode_state(inode, EXT4_STATE_DISKSIZE_GROW_PENDING))
3637 		WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize);
3638 }
3639 
3640 /*
3641  * Update i_size and i_disksize to @newsize.  Requires i_rwsem to avoid
3642  * races with truncate.
3643  *
3644  * In the iomap buffered I/O path, i_disksize is updated only if no zeroed
3645  * pending block straddles i_disksize (EXT4_STATE_DISKSIZE_GROW_PENDING
3646  * clear), otherwise the ioend worker for the pending block will advance
3647  * i_disksize once the pending block is written back.  Both updates happen
3648  * under i_data_sem so that the writeback ioend worker can always see the
3649  * latest i_size under the same semaphore.
3650  *
3651  * Returns 0 if nothing changed, 1 if i_size was raised, 2 if i_disksize
3652  * was raised, or 3 if both were.
3653  */
3654 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
3655 {
3656 	int changed = 0;
3657 
3658 	if (newsize <= inode->i_size && newsize <= EXT4_I(inode)->i_disksize)
3659 		return 0;
3660 
3661 	down_write(&EXT4_I(inode)->i_data_sem);
3662 	if (newsize > inode->i_size) {
3663 		i_size_write(inode, newsize);
3664 		changed = 1;
3665 	}
3666 	if (newsize > EXT4_I(inode)->i_disksize &&
3667 	    !ext4_test_inode_state(inode, EXT4_STATE_DISKSIZE_GROW_PENDING)) {
3668 		WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize);
3669 		changed |= 2;
3670 	}
3671 	up_write(&EXT4_I(inode)->i_data_sem);
3672 	return changed;
3673 }
3674 
3675 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
3676 				      loff_t len);
3677 
3678 struct ext4_group_info {
3679 	unsigned long   bb_state;
3680 #ifdef AGGRESSIVE_CHECK
3681 	unsigned long	bb_check_counter;
3682 #endif
3683 	struct rb_root  bb_free_root;
3684 	ext4_grpblk_t	bb_first_free;	/* first free block */
3685 	ext4_grpblk_t	bb_free;	/* total free blocks */
3686 	ext4_grpblk_t	bb_fragments;	/* nr of freespace fragments */
3687 	int		bb_avg_fragment_size_order;	/* order of average
3688 							   fragment in BG */
3689 	ext4_grpblk_t	bb_largest_free_order;/* order of largest frag in BG */
3690 	ext4_group_t	bb_group;	/* Group number */
3691 	struct          list_head bb_prealloc_list;
3692 #ifdef DOUBLE_CHECK
3693 	void            *bb_bitmap;
3694 #endif
3695 	struct rw_semaphore alloc_sem;
3696 	ext4_grpblk_t	bb_counters[];	/* Nr of free power-of-two-block
3697 					 * regions, index is order.
3698 					 * bb_counters[3] = 5 means
3699 					 * 5 free 8-block regions. */
3700 };
3701 
3702 #define EXT4_GROUP_INFO_NEED_INIT_BIT		0
3703 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT		1
3704 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT	2
3705 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT	3
3706 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT		\
3707 	(1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT)
3708 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT		\
3709 	(1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT)
3710 #define EXT4_GROUP_INFO_BBITMAP_READ_BIT	4
3711 
3712 #define EXT4_MB_GRP_NEED_INIT(grp)	\
3713 	(test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
3714 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp)	\
3715 	(test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3716 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp)	\
3717 	(test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3718 
3719 #define EXT4_MB_GRP_WAS_TRIMMED(grp)	\
3720 	(test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3721 #define EXT4_MB_GRP_SET_TRIMMED(grp)	\
3722 	(set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3723 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp)	\
3724 	(clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3725 #define EXT4_MB_GRP_TEST_AND_SET_READ(grp)	\
3726 	(ext4_mb_grp_test_and_set_read((grp)))
3727 
3728 static inline int ext4_mb_grp_test_and_set_read(struct ext4_group_info *grp)
3729 {
3730 	return (test_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &grp->bb_state) ||
3731 		test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &grp->bb_state));
3732 }
3733 
3734 #define EXT4_MAX_CONTENTION		8
3735 #define EXT4_CONTENTION_THRESHOLD	2
3736 
3737 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
3738 					      ext4_group_t group)
3739 {
3740 	return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
3741 }
3742 
3743 /*
3744  * Returns true if the filesystem is busy enough that attempts to
3745  * access the block group locks has run into contention.
3746  */
3747 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
3748 {
3749 	return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
3750 }
3751 
3752 static inline bool ext4_try_lock_group(struct super_block *sb, ext4_group_t group)
3753 {
3754 	if (!spin_trylock(ext4_group_lock_ptr(sb, group)))
3755 		return false;
3756 	/*
3757 	 * We're able to grab the lock right away, so drop the lock
3758 	 * contention counter.
3759 	 */
3760 	atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
3761 	return true;
3762 }
3763 
3764 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
3765 {
3766 	if (!ext4_try_lock_group(sb, group)) {
3767 		/*
3768 		 * The lock is busy, so bump the contention counter,
3769 		 * and then wait on the spin lock.
3770 		 */
3771 		atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
3772 				  EXT4_MAX_CONTENTION);
3773 		spin_lock(ext4_group_lock_ptr(sb, group));
3774 	}
3775 }
3776 
3777 static inline void ext4_unlock_group(struct super_block *sb,
3778 					ext4_group_t group)
3779 {
3780 	spin_unlock(ext4_group_lock_ptr(sb, group));
3781 }
3782 
3783 #ifdef CONFIG_QUOTA
3784 static inline bool ext4_quota_capable(struct super_block *sb)
3785 {
3786 	return (test_opt(sb, QUOTA) || ext4_has_feature_quota(sb));
3787 }
3788 
3789 static inline bool ext4_is_quota_journalled(struct super_block *sb)
3790 {
3791 	struct ext4_sb_info *sbi = EXT4_SB(sb);
3792 
3793 	return (ext4_has_feature_quota(sb) ||
3794 		sbi->s_qf_names[USRQUOTA] || sbi->s_qf_names[GRPQUOTA]);
3795 }
3796 int ext4_enable_quotas(struct super_block *sb);
3797 #endif
3798 
3799 /*
3800  * Block validity checking
3801  */
3802 #define ext4_check_indirect_blockref(inode, bh)				\
3803 	ext4_check_blockref(__func__, __LINE__, inode,			\
3804 			    (__le32 *)(bh)->b_data,			\
3805 			    EXT4_ADDR_PER_BLOCK((inode)->i_sb))
3806 
3807 #define ext4_ind_check_inode(inode)					\
3808 	ext4_check_blockref(__func__, __LINE__, inode,			\
3809 			    EXT4_I(inode)->i_data,			\
3810 			    EXT4_NDIR_BLOCKS)
3811 
3812 /*
3813  * Inodes and files operations
3814  */
3815 
3816 /* dir.c */
3817 extern const struct file_operations ext4_dir_operations;
3818 
3819 /* file.c */
3820 extern const struct inode_operations ext4_file_inode_operations;
3821 extern const struct file_operations ext4_file_operations;
3822 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
3823 
3824 /* inline.c */
3825 extern int ext4_get_max_inline_size(struct inode *inode);
3826 extern int ext4_find_inline_data_nolock(struct inode *inode);
3827 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
3828 extern void ext4_update_final_de(void *de_buf, int old_size, int new_size);
3829 
3830 int ext4_readpage_inline(struct inode *inode, struct folio *folio);
3831 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3832 					 struct inode *inode,
3833 					 loff_t pos, unsigned len,
3834 					 struct folio **foliop);
3835 int ext4_write_inline_data_end(struct inode *inode, loff_t pos, unsigned len,
3836 			       unsigned copied, struct folio *folio);
3837 extern int ext4_generic_write_inline_data(struct address_space *mapping,
3838 					  struct inode *inode,
3839 					  loff_t pos, unsigned len,
3840 					  struct folio **foliop,
3841 					  bool da);
3842 extern int ext4_try_add_inline_entry(handle_t *handle,
3843 				     struct ext4_filename *fname,
3844 				     struct inode *dir, struct inode *inode);
3845 extern int ext4_try_create_inline_dir(handle_t *handle,
3846 				      struct inode *parent,
3847 				      struct inode *inode);
3848 extern int ext4_read_inline_dir(struct file *filp,
3849 				struct dir_context *ctx,
3850 				int *has_inline_data);
3851 extern int ext4_inlinedir_to_tree(struct file *dir_file,
3852 				  struct inode *dir, ext4_lblk_t block,
3853 				  struct dx_hash_info *hinfo,
3854 				  __u32 start_hash, __u32 start_minor_hash,
3855 				  int *has_inline_data);
3856 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3857 					struct ext4_filename *fname,
3858 					struct ext4_dir_entry_2 **res_dir,
3859 					int *has_inline_data);
3860 extern int ext4_delete_inline_entry(handle_t *handle,
3861 				    struct inode *dir,
3862 				    struct ext4_dir_entry_2 *de_del,
3863 				    struct buffer_head *bh,
3864 				    int *has_inline_data);
3865 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3866 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3867 					struct ext4_dir_entry_2 **parent_de,
3868 					int *retval);
3869 extern void *ext4_read_inline_link(struct inode *inode);
3870 
3871 struct iomap;
3872 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap);
3873 
3874 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3875 
3876 extern int ext4_convert_inline_data(struct inode *inode);
3877 
3878 static inline int ext4_has_inline_data(struct inode *inode)
3879 {
3880 	return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3881 	       EXT4_I(inode)->i_inline_off;
3882 }
3883 
3884 /* namei.c */
3885 extern const struct inode_operations ext4_dir_inode_operations;
3886 extern const struct inode_operations ext4_special_inode_operations;
3887 extern struct dentry *ext4_get_parent(struct dentry *child);
3888 extern int ext4_init_dirblock(handle_t *handle, struct inode *inode,
3889 			      struct buffer_head *dir_block,
3890 			      unsigned int parent_ino, void *inline_buf,
3891 			      int inline_size);
3892 extern void ext4_initialize_dirent_tail(struct buffer_head *bh,
3893 					unsigned int blocksize);
3894 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode,
3895 				      struct buffer_head *bh);
3896 extern int __ext4_unlink(struct inode *dir, const struct qstr *d_name,
3897 			 struct inode *inode, struct dentry *dentry);
3898 extern int __ext4_link(struct inode *dir, struct inode *inode,
3899 		       const struct qstr *d_name, struct dentry *dentry);
3900 
3901 #define S_SHIFT 12
3902 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = {
3903 	[S_IFREG >> S_SHIFT]	= EXT4_FT_REG_FILE,
3904 	[S_IFDIR >> S_SHIFT]	= EXT4_FT_DIR,
3905 	[S_IFCHR >> S_SHIFT]	= EXT4_FT_CHRDEV,
3906 	[S_IFBLK >> S_SHIFT]	= EXT4_FT_BLKDEV,
3907 	[S_IFIFO >> S_SHIFT]	= EXT4_FT_FIFO,
3908 	[S_IFSOCK >> S_SHIFT]	= EXT4_FT_SOCK,
3909 	[S_IFLNK >> S_SHIFT]	= EXT4_FT_SYMLINK,
3910 };
3911 
3912 static inline void ext4_set_de_type(struct super_block *sb,
3913 				struct ext4_dir_entry_2 *de,
3914 				umode_t mode) {
3915 	if (ext4_has_feature_filetype(sb))
3916 		de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3917 }
3918 
3919 /* readpages.c */
3920 int ext4_read_folio(struct file *file, struct folio *folio);
3921 void ext4_readahead(struct readahead_control *rac);
3922 int __init ext4_init_verity_caches(void);
3923 void ext4_exit_verity_caches(void);
3924 
3925 /* symlink.c */
3926 extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3927 extern const struct inode_operations ext4_symlink_inode_operations;
3928 extern const struct inode_operations ext4_fast_symlink_inode_operations;
3929 
3930 /* sysfs.c */
3931 extern void ext4_notify_error_sysfs(struct ext4_sb_info *sbi);
3932 extern int ext4_register_sysfs(struct super_block *sb);
3933 extern void ext4_unregister_sysfs(struct super_block *sb);
3934 extern int __init ext4_init_sysfs(void);
3935 extern void ext4_exit_sysfs(void);
3936 
3937 /* block_validity */
3938 extern void ext4_release_system_zone(struct super_block *sb);
3939 extern int ext4_setup_system_zone(struct super_block *sb);
3940 extern int __init ext4_init_system_zone(void);
3941 extern void ext4_exit_system_zone(void);
3942 extern int ext4_inode_block_valid(struct inode *inode,
3943 				  ext4_fsblk_t start_blk,
3944 				  unsigned int count);
3945 extern int ext4_check_blockref(const char *, unsigned int,
3946 			       struct inode *, __le32 *, unsigned int);
3947 extern int ext4_sb_block_valid(struct super_block *sb, struct inode *inode,
3948 				ext4_fsblk_t start_blk, unsigned int count);
3949 
3950 
3951 /* extents.c */
3952 struct ext4_ext_path;
3953 struct ext4_extent;
3954 
3955 /*
3956  * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3957  * __le32.
3958  */
3959 #define EXT_MAX_BLOCKS	0xffffffff
3960 
3961 extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode);
3962 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3963 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3964 			       struct ext4_map_blocks *map, int flags);
3965 ext4_lblk_t ext4_determine_insert_hole(struct inode *inode, ext4_lblk_t lblk,
3966 				       ext4_lblk_t hole_start,
3967 				       ext4_lblk_t hole_len);
3968 extern int ext4_ext_truncate(handle_t *, struct inode *);
3969 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3970 				 ext4_lblk_t end);
3971 extern void ext4_ext_init(struct super_block *);
3972 extern void ext4_ext_release(struct super_block *);
3973 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3974 			  loff_t len);
3975 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3976 					  loff_t offset, ssize_t len,
3977 					  ext4_lblk_t *converted);
3978 extern int ext4_convert_unwritten_extents_atomic(handle_t *handle,
3979 			struct inode *inode, loff_t offset, ssize_t len);
3980 extern int ext4_convert_unwritten_io_end_vec(handle_t *handle,
3981 					     ext4_io_end_t *io_end);
3982 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3983 			   struct ext4_map_blocks *map, int flags);
3984 extern int ext4_map_query_blocks(handle_t *handle, struct inode *inode,
3985 				  struct ext4_map_blocks *map, int flags);
3986 extern int ext4_map_create_blocks(handle_t *handle, struct inode *inode,
3987 				  struct ext4_map_blocks *map, int flags);
3988 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3989 						   int num,
3990 						   struct ext4_ext_path *path);
3991 extern struct ext4_ext_path *ext4_ext_insert_extent(
3992 				handle_t *handle, struct inode *inode,
3993 				struct ext4_ext_path *path,
3994 				struct ext4_extent *newext, int gb_flags);
3995 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3996 					      struct ext4_ext_path *,
3997 					      int flags);
3998 extern void ext4_free_ext_path(struct ext4_ext_path *);
3999 extern int ext4_ext_check_inode(struct inode *inode);
4000 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
4001 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
4002 			__u64 start, __u64 len);
4003 extern int ext4_get_es_cache(struct inode *inode,
4004 			     struct fiemap_extent_info *fieinfo,
4005 			     __u64 start, __u64 len);
4006 extern int ext4_ext_precache(struct inode *inode);
4007 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
4008 				struct inode *inode2, ext4_lblk_t lblk1,
4009 			     ext4_lblk_t lblk2,  ext4_lblk_t count,
4010 			     int mark_unwritten,int *err);
4011 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu);
4012 extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode,
4013 				       int check_cred, int restart_cred,
4014 				       int revoke_cred);
4015 extern void ext4_ext_replay_shrink_inode(struct inode *inode, ext4_lblk_t end);
4016 extern int ext4_ext_replay_set_iblocks(struct inode *inode);
4017 extern int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start,
4018 		int len, int unwritten, ext4_fsblk_t pblk);
4019 extern int ext4_ext_clear_bb(struct inode *inode);
4020 
4021 
4022 /* move_extent.c */
4023 extern void ext4_double_down_write_data_sem(struct inode *first,
4024 					    struct inode *second);
4025 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
4026 					  struct inode *donor_inode);
4027 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
4028 			     __u64 start_orig, __u64 start_donor,
4029 			     __u64 len, __u64 *moved_len);
4030 
4031 /* page-io.c */
4032 /*
4033  * The I/O range covers the zeroed EOF block that straddles i_disksize
4034  * and will advance it upon completion.
4035  */
4036 #define EXT4_IOMAP_IOEND_DISKSIZE_GROW_IO	1UL
4037 
4038 extern int __init ext4_init_pageio(void);
4039 extern void ext4_exit_pageio(void);
4040 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
4041 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
4042 extern int ext4_put_io_end(ext4_io_end_t *io_end);
4043 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
4044 extern void ext4_io_submit_init(struct ext4_io_submit *io,
4045 				struct writeback_control *wbc);
4046 extern void ext4_end_io_rsv_work(struct work_struct *work);
4047 extern void ext4_io_submit(struct ext4_io_submit *io);
4048 void ext4_bio_write_folio(struct ext4_io_submit *io, struct folio *page,
4049 		size_t len);
4050 extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end);
4051 extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end);
4052 extern void ext4_iomap_end_io(struct work_struct *work);
4053 extern void ext4_iomap_end_bio(struct bio *bio);
4054 
4055 /* mmp.c */
4056 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
4057 
4058 /* mmp.c */
4059 extern void ext4_stop_mmpd(struct ext4_sb_info *sbi);
4060 
4061 /* verity.c */
4062 extern const struct fsverity_operations ext4_verityops;
4063 
4064 /* orphan.c */
4065 extern int ext4_orphan_add(handle_t *, struct inode *);
4066 extern int ext4_orphan_del(handle_t *, struct inode *);
4067 extern void ext4_orphan_cleanup(struct super_block *sb,
4068 				struct ext4_super_block *es);
4069 extern void ext4_release_orphan_info(struct super_block *sb);
4070 extern int ext4_init_orphan_info(struct super_block *sb);
4071 extern int ext4_orphan_file_empty(struct super_block *sb);
4072 extern void ext4_orphan_file_block_trigger(
4073 				struct jbd2_buffer_trigger_type *triggers,
4074 				struct buffer_head *bh,
4075 				void *data, size_t size);
4076 
4077 /*
4078  * Add new method to test whether block and inode bitmaps are properly
4079  * initialized. With uninit_bg reading the block from disk is not enough
4080  * to mark the bitmap uptodate. We need to also zero-out the bitmap
4081  */
4082 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
4083 
4084 static inline int bitmap_uptodate(struct buffer_head *bh)
4085 {
4086 	return (buffer_uptodate(bh) &&
4087 			test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
4088 }
4089 static inline void set_bitmap_uptodate(struct buffer_head *bh)
4090 {
4091 	set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
4092 }
4093 
4094 extern int ext4_resize_begin(struct super_block *sb);
4095 extern int ext4_resize_end(struct super_block *sb, bool update_backups);
4096 
4097 static inline void ext4_set_io_unwritten_flag(struct ext4_io_end *io_end)
4098 {
4099 	if (!(io_end->flag & EXT4_IO_END_UNWRITTEN))
4100 		io_end->flag |= EXT4_IO_END_UNWRITTEN;
4101 }
4102 
4103 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
4104 {
4105 	if (io_end->flag & EXT4_IO_END_UNWRITTEN)
4106 		io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
4107 }
4108 
4109 DECLARE_IOMAP_ITER_NEXT(ext4_iomap_next);
4110 DECLARE_IOMAP_ITER_NEXT(ext4_iomap_next_report);
4111 DECLARE_IOMAP_ITER_NEXT(ext4_iomap_buffered_write_next);
4112 DECLARE_IOMAP_ITER_NEXT(ext4_iomap_buffered_da_write_next);
4113 
4114 extern const struct iomap_write_ops ext4_iomap_write_ops;
4115 
4116 int ext4_iomap_begin(struct inode *inode, loff_t offset, loff_t length,
4117 		unsigned flags, struct iomap *iomap, struct iomap *srcmap);
4118 
4119 static inline int ext4_buffer_uptodate(struct buffer_head *bh)
4120 {
4121 	/*
4122 	 * If the buffer has the write error flag, we have failed
4123 	 * to write out data in the block.  In this  case, we don't
4124 	 * have to read the block because we may read the old data
4125 	 * successfully.
4126 	 */
4127 	if (buffer_write_io_error(bh))
4128 		set_buffer_uptodate(bh);
4129 	return buffer_uptodate(bh);
4130 }
4131 
4132 static inline bool ext4_inode_can_atomic_write(struct inode *inode)
4133 {
4134 
4135 	return S_ISREG(inode->i_mode) &&
4136 		ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS) &&
4137 		EXT4_SB(inode->i_sb)->s_awu_min > 0;
4138 }
4139 
4140 extern int ext4_block_write_begin(handle_t *handle, struct folio *folio,
4141 				  loff_t pos, unsigned len,
4142 				  get_block_t *get_block);
4143 
4144 #if IS_ENABLED(CONFIG_EXT4_KUNIT_TESTS)
4145 #define EXPORT_SYMBOL_FOR_EXT4_TEST(sym) \
4146 	EXPORT_SYMBOL_FOR_MODULES(sym, "ext4-test")
4147 #endif
4148 #endif	/* __KERNEL__ */
4149 
4150 #endif	/* _EXT4_H */
4151