xref: /linux/fs/ext4/ext4.h (revision ab93e0dd72c37d378dd936f031ffb83ff2bd87ce)
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/magic.h>
24 #include <linux/jbd2.h>
25 #include <linux/quota.h>
26 #include <linux/rwsem.h>
27 #include <linux/rbtree.h>
28 #include <linux/seqlock.h>
29 #include <linux/mutex.h>
30 #include <linux/timer.h>
31 #include <linux/wait.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 %lu: (%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 };
264 
265 /*
266  * Block validity checking, system zone rbtree.
267  */
268 struct ext4_system_blocks {
269 	struct rb_root root;
270 	struct rcu_head rcu;
271 };
272 
273 /*
274  * Flags for ext4_io_end->flags
275  */
276 #define EXT4_IO_END_UNWRITTEN	0x0001
277 #define EXT4_IO_END_FAILED	0x0002
278 
279 #define EXT4_IO_END_DEFER_COMPLETION (EXT4_IO_END_UNWRITTEN | EXT4_IO_END_FAILED)
280 
281 struct ext4_io_end_vec {
282 	struct list_head list;		/* list of io_end_vec */
283 	loff_t offset;			/* offset in the file */
284 	ssize_t size;			/* size of the extent */
285 };
286 
287 /*
288  * For converting unwritten extents on a work queue. 'handle' is used for
289  * buffered writeback.
290  */
291 typedef struct ext4_io_end {
292 	struct list_head	list;		/* per-file finished IO list */
293 	handle_t		*handle;	/* handle reserved for extent
294 						 * conversion */
295 	struct inode		*inode;		/* file being written to */
296 	struct bio		*bio;		/* Linked list of completed
297 						 * bios covering the extent */
298 	unsigned int		flag;		/* unwritten or not */
299 	refcount_t		count;		/* reference counter */
300 	struct list_head	list_vec;	/* list of ext4_io_end_vec */
301 } ext4_io_end_t;
302 
303 struct ext4_io_submit {
304 	struct writeback_control *io_wbc;
305 	struct bio		*io_bio;
306 	ext4_io_end_t		*io_end;
307 	sector_t		io_next_block;
308 };
309 
310 /*
311  * Special inodes numbers
312  */
313 #define	EXT4_BAD_INO		 1	/* Bad blocks inode */
314 #define EXT4_ROOT_INO		 2	/* Root inode */
315 #define EXT4_USR_QUOTA_INO	 3	/* User quota inode */
316 #define EXT4_GRP_QUOTA_INO	 4	/* Group quota inode */
317 #define EXT4_BOOT_LOADER_INO	 5	/* Boot loader inode */
318 #define EXT4_UNDEL_DIR_INO	 6	/* Undelete directory inode */
319 #define EXT4_RESIZE_INO		 7	/* Reserved group descriptors inode */
320 #define EXT4_JOURNAL_INO	 8	/* Journal inode */
321 
322 /* First non-reserved inode for old ext4 filesystems */
323 #define EXT4_GOOD_OLD_FIRST_INO	11
324 
325 /*
326  * Maximal count of links to a file
327  */
328 #define EXT4_LINK_MAX		65000
329 
330 /*
331  * Macro-instructions used to manage several block sizes
332  */
333 #define EXT4_MIN_BLOCK_SIZE		1024
334 #define	EXT4_MAX_BLOCK_SIZE		65536
335 #define EXT4_MIN_BLOCK_LOG_SIZE		10
336 #define EXT4_MAX_BLOCK_LOG_SIZE		16
337 #define EXT4_MAX_CLUSTER_LOG_SIZE	30
338 #ifdef __KERNEL__
339 # define EXT4_BLOCK_SIZE(s)		((s)->s_blocksize)
340 #else
341 # define EXT4_BLOCK_SIZE(s)		(EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
342 #endif
343 #define	EXT4_ADDR_PER_BLOCK(s)		(EXT4_BLOCK_SIZE(s) / sizeof(__u32))
344 #define EXT4_CLUSTER_SIZE(s)		(EXT4_BLOCK_SIZE(s) << \
345 					 EXT4_SB(s)->s_cluster_bits)
346 #ifdef __KERNEL__
347 # define EXT4_BLOCK_SIZE_BITS(s)	((s)->s_blocksize_bits)
348 # define EXT4_CLUSTER_BITS(s)		(EXT4_SB(s)->s_cluster_bits)
349 #else
350 # define EXT4_BLOCK_SIZE_BITS(s)	((s)->s_log_block_size + 10)
351 #endif
352 #ifdef __KERNEL__
353 #define	EXT4_ADDR_PER_BLOCK_BITS(s)	(EXT4_SB(s)->s_addr_per_block_bits)
354 #define EXT4_INODE_SIZE(s)		(EXT4_SB(s)->s_inode_size)
355 #define EXT4_FIRST_INO(s)		(EXT4_SB(s)->s_first_ino)
356 #else
357 #define EXT4_INODE_SIZE(s)	(((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
358 				 EXT4_GOOD_OLD_INODE_SIZE : \
359 				 (s)->s_inode_size)
360 #define EXT4_FIRST_INO(s)	(((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
361 				 EXT4_GOOD_OLD_FIRST_INO : \
362 				 (s)->s_first_ino)
363 #endif
364 #define EXT4_BLOCK_ALIGN(size, blkbits)		ALIGN((size), (1 << (blkbits)))
365 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \
366 	((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \
367 								  blkbits))
368 #define EXT4_B_TO_LBLK(inode, offset) \
369 	(round_up((offset), i_blocksize(inode)) >> (inode)->i_blkbits)
370 
371 /* Translate a block number to a cluster number */
372 #define EXT4_B2C(sbi, blk)	((blk) >> (sbi)->s_cluster_bits)
373 /* Translate a cluster number to a block number */
374 #define EXT4_C2B(sbi, cluster)	((cluster) << (sbi)->s_cluster_bits)
375 /* Translate # of blks to # of clusters */
376 #define EXT4_NUM_B2C(sbi, blks)	(((blks) + (sbi)->s_cluster_ratio - 1) >> \
377 				 (sbi)->s_cluster_bits)
378 /* Mask out the low bits to get the starting block of the cluster */
379 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) &				\
380 				  ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
381 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) &				\
382 				  ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
383 /* Fill in the low bits to get the last block of the cluster */
384 #define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) |				\
385 				    ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1))
386 /* Get the cluster offset */
387 #define EXT4_PBLK_COFF(s, pblk) ((pblk) &				\
388 				 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
389 #define EXT4_LBLK_COFF(s, lblk) ((lblk) &				\
390 				 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
391 
392 /*
393  * Structure of a blocks group descriptor
394  */
395 struct ext4_group_desc
396 {
397 	__le32	bg_block_bitmap_lo;	/* Blocks bitmap block */
398 	__le32	bg_inode_bitmap_lo;	/* Inodes bitmap block */
399 	__le32	bg_inode_table_lo;	/* Inodes table block */
400 	__le16	bg_free_blocks_count_lo;/* Free blocks count */
401 	__le16	bg_free_inodes_count_lo;/* Free inodes count */
402 	__le16	bg_used_dirs_count_lo;	/* Directories count */
403 	__le16	bg_flags;		/* EXT4_BG_flags (INODE_UNINIT, etc) */
404 	__le32  bg_exclude_bitmap_lo;   /* Exclude bitmap for snapshots */
405 	__le16  bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
406 	__le16  bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
407 	__le16  bg_itable_unused_lo;	/* Unused inodes count */
408 	__le16  bg_checksum;		/* crc16(sb_uuid+group+desc) */
409 	__le32	bg_block_bitmap_hi;	/* Blocks bitmap block MSB */
410 	__le32	bg_inode_bitmap_hi;	/* Inodes bitmap block MSB */
411 	__le32	bg_inode_table_hi;	/* Inodes table block MSB */
412 	__le16	bg_free_blocks_count_hi;/* Free blocks count MSB */
413 	__le16	bg_free_inodes_count_hi;/* Free inodes count MSB */
414 	__le16	bg_used_dirs_count_hi;	/* Directories count MSB */
415 	__le16  bg_itable_unused_hi;    /* Unused inodes count MSB */
416 	__le32  bg_exclude_bitmap_hi;   /* Exclude bitmap block MSB */
417 	__le16  bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
418 	__le16  bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
419 	__u32   bg_reserved;
420 };
421 
422 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END	\
423 	(offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
424 	 sizeof(__le16))
425 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END	\
426 	(offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
427 	 sizeof(__le16))
428 
429 /*
430  * Structure of a flex block group info
431  */
432 
433 struct flex_groups {
434 	atomic64_t	free_clusters;
435 	atomic_t	free_inodes;
436 	atomic_t	used_dirs;
437 };
438 
439 #define EXT4_BG_INODE_UNINIT	0x0001 /* Inode table/bitmap not in use */
440 #define EXT4_BG_BLOCK_UNINIT	0x0002 /* Block bitmap not in use */
441 #define EXT4_BG_INODE_ZEROED	0x0004 /* On-disk itable initialized to zero */
442 
443 /*
444  * Macro-instructions used to manage group descriptors
445  */
446 #define EXT4_MIN_DESC_SIZE		32
447 #define EXT4_MIN_DESC_SIZE_64BIT	64
448 #define	EXT4_MAX_DESC_SIZE		EXT4_MIN_BLOCK_SIZE
449 #define EXT4_DESC_SIZE(s)		(EXT4_SB(s)->s_desc_size)
450 #ifdef __KERNEL__
451 # define EXT4_BLOCKS_PER_GROUP(s)	(EXT4_SB(s)->s_blocks_per_group)
452 # define EXT4_CLUSTERS_PER_GROUP(s)	(EXT4_SB(s)->s_clusters_per_group)
453 # define EXT4_DESC_PER_BLOCK(s)		(EXT4_SB(s)->s_desc_per_block)
454 # define EXT4_INODES_PER_GROUP(s)	(EXT4_SB(s)->s_inodes_per_group)
455 # define EXT4_DESC_PER_BLOCK_BITS(s)	(EXT4_SB(s)->s_desc_per_block_bits)
456 #else
457 # define EXT4_BLOCKS_PER_GROUP(s)	((s)->s_blocks_per_group)
458 # define EXT4_DESC_PER_BLOCK(s)		(EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
459 # define EXT4_INODES_PER_GROUP(s)	((s)->s_inodes_per_group)
460 #endif
461 
462 /*
463  * Constants relative to the data blocks
464  */
465 #define	EXT4_NDIR_BLOCKS		12
466 #define	EXT4_IND_BLOCK			EXT4_NDIR_BLOCKS
467 #define	EXT4_DIND_BLOCK			(EXT4_IND_BLOCK + 1)
468 #define	EXT4_TIND_BLOCK			(EXT4_DIND_BLOCK + 1)
469 #define	EXT4_N_BLOCKS			(EXT4_TIND_BLOCK + 1)
470 
471 /*
472  * Inode flags
473  */
474 #define	EXT4_SECRM_FL			0x00000001 /* Secure deletion */
475 #define	EXT4_UNRM_FL			0x00000002 /* Undelete */
476 #define	EXT4_COMPR_FL			0x00000004 /* Compress file */
477 #define EXT4_SYNC_FL			0x00000008 /* Synchronous updates */
478 #define EXT4_IMMUTABLE_FL		0x00000010 /* Immutable file */
479 #define EXT4_APPEND_FL			0x00000020 /* writes to file may only append */
480 #define EXT4_NODUMP_FL			0x00000040 /* do not dump file */
481 #define EXT4_NOATIME_FL			0x00000080 /* do not update atime */
482 /* Reserved for compression usage... */
483 #define EXT4_DIRTY_FL			0x00000100
484 #define EXT4_COMPRBLK_FL		0x00000200 /* One or more compressed clusters */
485 #define EXT4_NOCOMPR_FL			0x00000400 /* Don't compress */
486 	/* nb: was previously EXT2_ECOMPR_FL */
487 #define EXT4_ENCRYPT_FL			0x00000800 /* encrypted file */
488 /* End compression flags --- maybe not all used */
489 #define EXT4_INDEX_FL			0x00001000 /* hash-indexed directory */
490 #define EXT4_IMAGIC_FL			0x00002000 /* AFS directory */
491 #define EXT4_JOURNAL_DATA_FL		0x00004000 /* file data should be journaled */
492 #define EXT4_NOTAIL_FL			0x00008000 /* file tail should not be merged */
493 #define EXT4_DIRSYNC_FL			0x00010000 /* dirsync behaviour (directories only) */
494 #define EXT4_TOPDIR_FL			0x00020000 /* Top of directory hierarchies*/
495 #define EXT4_HUGE_FILE_FL               0x00040000 /* Set to each huge file */
496 #define EXT4_EXTENTS_FL			0x00080000 /* Inode uses extents */
497 #define EXT4_VERITY_FL			0x00100000 /* Verity protected inode */
498 #define EXT4_EA_INODE_FL	        0x00200000 /* Inode used for large EA */
499 /* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */
500 
501 #define EXT4_DAX_FL			0x02000000 /* Inode is DAX */
502 
503 #define EXT4_INLINE_DATA_FL		0x10000000 /* Inode has inline data. */
504 #define EXT4_PROJINHERIT_FL		0x20000000 /* Create with parents projid */
505 #define EXT4_CASEFOLD_FL		0x40000000 /* Casefolded directory */
506 #define EXT4_RESERVED_FL		0x80000000 /* reserved for ext4 lib */
507 
508 /* User modifiable flags */
509 #define EXT4_FL_USER_MODIFIABLE		(EXT4_SECRM_FL | \
510 					 EXT4_UNRM_FL | \
511 					 EXT4_COMPR_FL | \
512 					 EXT4_SYNC_FL | \
513 					 EXT4_IMMUTABLE_FL | \
514 					 EXT4_APPEND_FL | \
515 					 EXT4_NODUMP_FL | \
516 					 EXT4_NOATIME_FL | \
517 					 EXT4_JOURNAL_DATA_FL | \
518 					 EXT4_NOTAIL_FL | \
519 					 EXT4_DIRSYNC_FL | \
520 					 EXT4_TOPDIR_FL | \
521 					 EXT4_EXTENTS_FL | \
522 					 0x00400000 /* EXT4_EOFBLOCKS_FL */ | \
523 					 EXT4_DAX_FL | \
524 					 EXT4_PROJINHERIT_FL | \
525 					 EXT4_CASEFOLD_FL)
526 
527 /* User visible flags */
528 #define EXT4_FL_USER_VISIBLE		(EXT4_FL_USER_MODIFIABLE | \
529 					 EXT4_DIRTY_FL | \
530 					 EXT4_COMPRBLK_FL | \
531 					 EXT4_NOCOMPR_FL | \
532 					 EXT4_ENCRYPT_FL | \
533 					 EXT4_INDEX_FL | \
534 					 EXT4_VERITY_FL | \
535 					 EXT4_INLINE_DATA_FL)
536 
537 /* Flags that should be inherited by new inodes from their parent. */
538 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
539 			   EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
540 			   EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
541 			   EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
542 			   EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL |\
543 			   EXT4_DAX_FL)
544 
545 /* Flags that are appropriate for regular files (all but dir-specific ones). */
546 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\
547 			   EXT4_PROJINHERIT_FL))
548 
549 /* Flags that are appropriate for non-directories/regular files. */
550 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
551 
552 /* The only flags that should be swapped */
553 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL)
554 
555 /* Flags which are mutually exclusive to DAX */
556 #define EXT4_DAX_MUT_EXCL (EXT4_VERITY_FL | EXT4_ENCRYPT_FL |\
557 			   EXT4_JOURNAL_DATA_FL | EXT4_INLINE_DATA_FL)
558 
559 /* Mask out flags that are inappropriate for the given type of inode. */
ext4_mask_flags(umode_t mode,__u32 flags)560 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
561 {
562 	if (S_ISDIR(mode))
563 		return flags;
564 	else if (S_ISREG(mode))
565 		return flags & EXT4_REG_FLMASK;
566 	else
567 		return flags & EXT4_OTHER_FLMASK;
568 }
569 
570 /*
571  * Inode flags used for atomic set/get
572  */
573 enum {
574 	EXT4_INODE_SECRM	= 0,	/* Secure deletion */
575 	EXT4_INODE_UNRM		= 1,	/* Undelete */
576 	EXT4_INODE_COMPR	= 2,	/* Compress file */
577 	EXT4_INODE_SYNC		= 3,	/* Synchronous updates */
578 	EXT4_INODE_IMMUTABLE	= 4,	/* Immutable file */
579 	EXT4_INODE_APPEND	= 5,	/* writes to file may only append */
580 	EXT4_INODE_NODUMP	= 6,	/* do not dump file */
581 	EXT4_INODE_NOATIME	= 7,	/* do not update atime */
582 /* Reserved for compression usage... */
583 	EXT4_INODE_DIRTY	= 8,
584 	EXT4_INODE_COMPRBLK	= 9,	/* One or more compressed clusters */
585 	EXT4_INODE_NOCOMPR	= 10,	/* Don't compress */
586 	EXT4_INODE_ENCRYPT	= 11,	/* Encrypted file */
587 /* End compression flags --- maybe not all used */
588 	EXT4_INODE_INDEX	= 12,	/* hash-indexed directory */
589 	EXT4_INODE_IMAGIC	= 13,	/* AFS directory */
590 	EXT4_INODE_JOURNAL_DATA	= 14,	/* file data should be journaled */
591 	EXT4_INODE_NOTAIL	= 15,	/* file tail should not be merged */
592 	EXT4_INODE_DIRSYNC	= 16,	/* dirsync behaviour (directories only) */
593 	EXT4_INODE_TOPDIR	= 17,	/* Top of directory hierarchies*/
594 	EXT4_INODE_HUGE_FILE	= 18,	/* Set to each huge file */
595 	EXT4_INODE_EXTENTS	= 19,	/* Inode uses extents */
596 	EXT4_INODE_VERITY	= 20,	/* Verity protected inode */
597 	EXT4_INODE_EA_INODE	= 21,	/* Inode used for large EA */
598 /* 22 was formerly EXT4_INODE_EOFBLOCKS */
599 	EXT4_INODE_DAX		= 25,	/* Inode is DAX */
600 	EXT4_INODE_INLINE_DATA	= 28,	/* Data in inode. */
601 	EXT4_INODE_PROJINHERIT	= 29,	/* Create with parents projid */
602 	EXT4_INODE_CASEFOLD	= 30,	/* Casefolded directory */
603 	EXT4_INODE_RESERVED	= 31,	/* reserved for ext4 lib */
604 };
605 
606 /*
607  * Since it's pretty easy to mix up bit numbers and hex values, we use a
608  * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
609  * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
610  * any extra space in the compiled kernel image, otherwise, the build will fail.
611  * It's important that these values are the same, since we are using
612  * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
613  * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
614  * values found in ext2, ext3 and ext4 filesystems, and of course the values
615  * defined in e2fsprogs.
616  *
617  * It's not paranoia if the Murphy's Law really *is* out to get you.  :-)
618  */
619 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1U << EXT4_INODE_##FLAG))
620 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
621 
ext4_check_flag_values(void)622 static inline void ext4_check_flag_values(void)
623 {
624 	CHECK_FLAG_VALUE(SECRM);
625 	CHECK_FLAG_VALUE(UNRM);
626 	CHECK_FLAG_VALUE(COMPR);
627 	CHECK_FLAG_VALUE(SYNC);
628 	CHECK_FLAG_VALUE(IMMUTABLE);
629 	CHECK_FLAG_VALUE(APPEND);
630 	CHECK_FLAG_VALUE(NODUMP);
631 	CHECK_FLAG_VALUE(NOATIME);
632 	CHECK_FLAG_VALUE(DIRTY);
633 	CHECK_FLAG_VALUE(COMPRBLK);
634 	CHECK_FLAG_VALUE(NOCOMPR);
635 	CHECK_FLAG_VALUE(ENCRYPT);
636 	CHECK_FLAG_VALUE(INDEX);
637 	CHECK_FLAG_VALUE(IMAGIC);
638 	CHECK_FLAG_VALUE(JOURNAL_DATA);
639 	CHECK_FLAG_VALUE(NOTAIL);
640 	CHECK_FLAG_VALUE(DIRSYNC);
641 	CHECK_FLAG_VALUE(TOPDIR);
642 	CHECK_FLAG_VALUE(HUGE_FILE);
643 	CHECK_FLAG_VALUE(EXTENTS);
644 	CHECK_FLAG_VALUE(VERITY);
645 	CHECK_FLAG_VALUE(EA_INODE);
646 	CHECK_FLAG_VALUE(INLINE_DATA);
647 	CHECK_FLAG_VALUE(PROJINHERIT);
648 	CHECK_FLAG_VALUE(CASEFOLD);
649 	CHECK_FLAG_VALUE(RESERVED);
650 }
651 
652 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
653 struct compat_ext4_new_group_input {
654 	u32 group;
655 	compat_u64 block_bitmap;
656 	compat_u64 inode_bitmap;
657 	compat_u64 inode_table;
658 	u32 blocks_count;
659 	u16 reserved_blocks;
660 	u16 unused;
661 };
662 #endif
663 
664 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
665 struct ext4_new_group_data {
666 	__u32 group;
667 	__u64 block_bitmap;
668 	__u64 inode_bitmap;
669 	__u64 inode_table;
670 	__u32 blocks_count;
671 	__u16 reserved_blocks;
672 	__u16 mdata_blocks;
673 	__u32 free_clusters_count;
674 };
675 
676 /* Indexes used to index group tables in ext4_new_group_data */
677 enum {
678 	BLOCK_BITMAP = 0,	/* block bitmap */
679 	INODE_BITMAP,		/* inode bitmap */
680 	INODE_TABLE,		/* inode tables */
681 	GROUP_TABLE_COUNT,
682 };
683 
684 /*
685  * Flags used by ext4_map_blocks()
686  */
687 	/* Allocate any needed blocks and/or convert an unwritten
688 	   extent to be an initialized ext4 */
689 #define EXT4_GET_BLOCKS_CREATE			0x0001
690 	/* Request the creation of an unwritten extent */
691 #define EXT4_GET_BLOCKS_UNWRIT_EXT		0x0002
692 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT	(EXT4_GET_BLOCKS_UNWRIT_EXT|\
693 						 EXT4_GET_BLOCKS_CREATE)
694 	/* Caller is from the delayed allocation writeout path
695 	 * finally doing the actual allocation of delayed blocks */
696 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE	0x0004
697 	/* caller is from the direct IO path, request to creation of an
698 	unwritten extents if not allocated, split the unwritten
699 	extent if blocks has been preallocated already*/
700 #define EXT4_GET_BLOCKS_PRE_IO			0x0008
701 #define EXT4_GET_BLOCKS_CONVERT			0x0010
702 #define EXT4_GET_BLOCKS_IO_CREATE_EXT		(EXT4_GET_BLOCKS_PRE_IO|\
703 					 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
704 	/* Eventual metadata allocation (due to growing extent tree)
705 	 * should not fail, so try to use reserved blocks for that.*/
706 #define EXT4_GET_BLOCKS_METADATA_NOFAIL		0x0020
707 	/* Don't normalize allocation size (used for fallocate) */
708 #define EXT4_GET_BLOCKS_NO_NORMALIZE		0x0040
709 	/* Convert written extents to unwritten */
710 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN	0x0100
711 	/* Write zeros to newly created written extents */
712 #define EXT4_GET_BLOCKS_ZERO			0x0200
713 #define EXT4_GET_BLOCKS_CREATE_ZERO		(EXT4_GET_BLOCKS_CREATE |\
714 					EXT4_GET_BLOCKS_ZERO)
715 	/* Caller is in the context of data submission, such as writeback,
716 	 * fsync, etc. Especially, in the generic writeback path, caller will
717 	 * submit data before dropping transaction handle. This allows jbd2
718 	 * to avoid submitting data before commit. */
719 #define EXT4_GET_BLOCKS_IO_SUBMIT		0x0400
720 	/* Convert extent to initialized after IO complete */
721 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT		(EXT4_GET_BLOCKS_CONVERT |\
722 					 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT |\
723 					 EXT4_GET_BLOCKS_IO_SUBMIT)
724 	/* Caller is in the atomic contex, find extent if it has been cached */
725 #define EXT4_GET_BLOCKS_CACHED_NOWAIT		0x0800
726 /*
727  * Atomic write caller needs this to query in the slow path of mixed mapping
728  * case, when a contiguous extent can be split across two adjacent leaf nodes.
729  * Look EXT4_MAP_QUERY_LAST_IN_LEAF.
730  */
731 #define EXT4_GET_BLOCKS_QUERY_LAST_IN_LEAF	0x1000
732 
733 /*
734  * The bit position of these flags must not overlap with any of the
735  * EXT4_GET_BLOCKS_*.  They are used by ext4_find_extent(),
736  * read_extent_tree_block(), ext4_split_extent_at(),
737  * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
738  * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
739  * caching the extents when reading from the extent tree while a
740  * truncate or punch hole operation is in progress.
741  */
742 #define EXT4_EX_NOCACHE				0x40000000
743 #define EXT4_EX_FORCE_CACHE			0x20000000
744 #define EXT4_EX_NOFAIL				0x10000000
745 /*
746  * ext4_map_query_blocks() uses this filter mask to filter the flags needed to
747  * pass while lookup/querying of on disk extent tree.
748  */
749 #define EXT4_EX_QUERY_FILTER	(EXT4_EX_NOCACHE | EXT4_EX_FORCE_CACHE |\
750 				 EXT4_EX_NOFAIL |\
751 				 EXT4_GET_BLOCKS_QUERY_LAST_IN_LEAF)
752 
753 /*
754  * Flags used by ext4_free_blocks
755  */
756 #define EXT4_FREE_BLOCKS_METADATA		0x0001
757 #define EXT4_FREE_BLOCKS_FORGET			0x0002
758 #define EXT4_FREE_BLOCKS_VALIDATED		0x0004
759 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE		0x0008
760 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER	0x0010
761 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER	0x0020
762 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER      0x0040
763 
764 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
765 /*
766  * ioctl commands in 32 bit emulation
767  */
768 #define EXT4_IOC32_GETVERSION		_IOR('f', 3, int)
769 #define EXT4_IOC32_SETVERSION		_IOW('f', 4, int)
770 #define EXT4_IOC32_GETRSVSZ		_IOR('f', 5, int)
771 #define EXT4_IOC32_SETRSVSZ		_IOW('f', 6, int)
772 #define EXT4_IOC32_GROUP_EXTEND		_IOW('f', 7, unsigned int)
773 #define EXT4_IOC32_GROUP_ADD		_IOW('f', 8, struct compat_ext4_new_group_input)
774 #define EXT4_IOC32_GETVERSION_OLD	FS_IOC32_GETVERSION
775 #define EXT4_IOC32_SETVERSION_OLD	FS_IOC32_SETVERSION
776 #endif
777 
778 /* Max physical block we can address w/o extents */
779 #define EXT4_MAX_BLOCK_FILE_PHYS	0xFFFFFFFF
780 
781 /* Max logical block we can support */
782 #define EXT4_MAX_LOGICAL_BLOCK		0xFFFFFFFE
783 
784 /*
785  * Structure of an inode on the disk
786  */
787 struct ext4_inode {
788 	__le16	i_mode;		/* File mode */
789 	__le16	i_uid;		/* Low 16 bits of Owner Uid */
790 	__le32	i_size_lo;	/* Size in bytes */
791 	__le32	i_atime;	/* Access time */
792 	__le32	i_ctime;	/* Inode Change time */
793 	__le32	i_mtime;	/* Modification time */
794 	__le32	i_dtime;	/* Deletion Time */
795 	__le16	i_gid;		/* Low 16 bits of Group Id */
796 	__le16	i_links_count;	/* Links count */
797 	__le32	i_blocks_lo;	/* Blocks count */
798 	__le32	i_flags;	/* File flags */
799 	union {
800 		struct {
801 			__le32  l_i_version;
802 		} linux1;
803 		struct {
804 			__u32  h_i_translator;
805 		} hurd1;
806 		struct {
807 			__u32  m_i_reserved1;
808 		} masix1;
809 	} osd1;				/* OS dependent 1 */
810 	__le32	i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
811 	__le32	i_generation;	/* File version (for NFS) */
812 	__le32	i_file_acl_lo;	/* File ACL */
813 	__le32	i_size_high;
814 	__le32	i_obso_faddr;	/* Obsoleted fragment address */
815 	union {
816 		struct {
817 			__le16	l_i_blocks_high; /* were l_i_reserved1 */
818 			__le16	l_i_file_acl_high;
819 			__le16	l_i_uid_high;	/* these 2 fields */
820 			__le16	l_i_gid_high;	/* were reserved2[0] */
821 			__le16	l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
822 			__le16	l_i_reserved;
823 		} linux2;
824 		struct {
825 			__le16	h_i_reserved1;	/* Obsoleted fragment number/size which are removed in ext4 */
826 			__u16	h_i_mode_high;
827 			__u16	h_i_uid_high;
828 			__u16	h_i_gid_high;
829 			__u32	h_i_author;
830 		} hurd2;
831 		struct {
832 			__le16	h_i_reserved1;	/* Obsoleted fragment number/size which are removed in ext4 */
833 			__le16	m_i_file_acl_high;
834 			__u32	m_i_reserved2[2];
835 		} masix2;
836 	} osd2;				/* OS dependent 2 */
837 	__le16	i_extra_isize;
838 	__le16	i_checksum_hi;	/* crc32c(uuid+inum+inode) BE */
839 	__le32  i_ctime_extra;  /* extra Change time      (nsec << 2 | epoch) */
840 	__le32  i_mtime_extra;  /* extra Modification time(nsec << 2 | epoch) */
841 	__le32  i_atime_extra;  /* extra Access time      (nsec << 2 | epoch) */
842 	__le32  i_crtime;       /* File Creation time */
843 	__le32  i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
844 	__le32  i_version_hi;	/* high 32 bits for 64-bit version */
845 	__le32	i_projid;	/* Project ID */
846 };
847 
848 #define EXT4_EPOCH_BITS 2
849 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
850 #define EXT4_NSEC_MASK  (~0UL << EXT4_EPOCH_BITS)
851 
852 /*
853  * Extended fields will fit into an inode if the filesystem was formatted
854  * with large inodes (-I 256 or larger) and there are not currently any EAs
855  * consuming all of the available space. For new inodes we always reserve
856  * enough space for the kernel's known extended fields, but for inodes
857  * created with an old kernel this might not have been the case. None of
858  * the extended inode fields is critical for correct filesystem operation.
859  * This macro checks if a certain field fits in the inode. Note that
860  * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
861  */
862 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field)	\
863 	((offsetof(typeof(*ext4_inode), field) +	\
864 	  sizeof((ext4_inode)->field))			\
865 	<= (EXT4_GOOD_OLD_INODE_SIZE +			\
866 	    (einode)->i_extra_isize))			\
867 
868 /*
869  * We use an encoding that preserves the times for extra epoch "00":
870  *
871  * extra  msb of                         adjust for signed
872  * epoch  32-bit                         32-bit tv_sec to
873  * bits   time    decoded 64-bit tv_sec  64-bit tv_sec      valid time range
874  * 0 0    1    -0x80000000..-0x00000001  0x000000000 1901-12-13..1969-12-31
875  * 0 0    0    0x000000000..0x07fffffff  0x000000000 1970-01-01..2038-01-19
876  * 0 1    1    0x080000000..0x0ffffffff  0x100000000 2038-01-19..2106-02-07
877  * 0 1    0    0x100000000..0x17fffffff  0x100000000 2106-02-07..2174-02-25
878  * 1 0    1    0x180000000..0x1ffffffff  0x200000000 2174-02-25..2242-03-16
879  * 1 0    0    0x200000000..0x27fffffff  0x200000000 2242-03-16..2310-04-04
880  * 1 1    1    0x280000000..0x2ffffffff  0x300000000 2310-04-04..2378-04-22
881  * 1 1    0    0x300000000..0x37fffffff  0x300000000 2378-04-22..2446-05-10
882  *
883  * Note that previous versions of the kernel on 64-bit systems would
884  * incorrectly use extra epoch bits 1,1 for dates between 1901 and
885  * 1970.  e2fsck will correct this, assuming that it is run on the
886  * affected filesystem before 2242.
887  */
888 
ext4_encode_extra_time(struct timespec64 ts)889 static inline __le32 ext4_encode_extra_time(struct timespec64 ts)
890 {
891 	u32 extra = ((ts.tv_sec - (s32)ts.tv_sec) >> 32) & EXT4_EPOCH_MASK;
892 	return cpu_to_le32(extra | (ts.tv_nsec << EXT4_EPOCH_BITS));
893 }
894 
ext4_decode_extra_time(__le32 base,__le32 extra)895 static inline struct timespec64 ext4_decode_extra_time(__le32 base,
896 						       __le32 extra)
897 {
898 	struct timespec64 ts = { .tv_sec = (signed)le32_to_cpu(base) };
899 
900 	if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK)))
901 		ts.tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
902 	ts.tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
903 	return ts;
904 }
905 
906 #define EXT4_INODE_SET_XTIME_VAL(xtime, inode, raw_inode, ts)			\
907 do {										\
908 	if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {	\
909 		(raw_inode)->xtime = cpu_to_le32((ts).tv_sec);			\
910 		(raw_inode)->xtime ## _extra = ext4_encode_extra_time(ts);	\
911 	} else									\
912 		(raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (ts).tv_sec, S32_MIN, S32_MAX));	\
913 } while (0)
914 
915 #define EXT4_INODE_SET_ATIME(inode, raw_inode)						\
916 	EXT4_INODE_SET_XTIME_VAL(i_atime, inode, raw_inode, inode_get_atime(inode))
917 
918 #define EXT4_INODE_SET_MTIME(inode, raw_inode)						\
919 	EXT4_INODE_SET_XTIME_VAL(i_mtime, inode, raw_inode, inode_get_mtime(inode))
920 
921 #define EXT4_INODE_SET_CTIME(inode, raw_inode)						\
922 	EXT4_INODE_SET_XTIME_VAL(i_ctime, inode, raw_inode, inode_get_ctime(inode))
923 
924 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode)				\
925 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))			\
926 		EXT4_INODE_SET_XTIME_VAL(xtime, &((einode)->vfs_inode),		\
927 					 raw_inode, (einode)->xtime)
928 
929 #define EXT4_INODE_GET_XTIME_VAL(xtime, inode, raw_inode)			\
930 	(EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra) ?	\
931 		ext4_decode_extra_time((raw_inode)->xtime,				\
932 				       (raw_inode)->xtime ## _extra) :		\
933 		(struct timespec64) {						\
934 			.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime)	\
935 		})
936 
937 #define EXT4_INODE_GET_ATIME(inode, raw_inode)					\
938 do {										\
939 	inode_set_atime_to_ts(inode,						\
940 		EXT4_INODE_GET_XTIME_VAL(i_atime, inode, raw_inode));		\
941 } while (0)
942 
943 #define EXT4_INODE_GET_MTIME(inode, raw_inode)					\
944 do {										\
945 	inode_set_mtime_to_ts(inode,						\
946 		EXT4_INODE_GET_XTIME_VAL(i_mtime, inode, raw_inode));		\
947 } while (0)
948 
949 #define EXT4_INODE_GET_CTIME(inode, raw_inode)					\
950 do {										\
951 	inode_set_ctime_to_ts(inode,						\
952 		EXT4_INODE_GET_XTIME_VAL(i_ctime, inode, raw_inode));		\
953 } while (0)
954 
955 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode)				\
956 do {										\
957 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) 			\
958 		(einode)->xtime =						\
959 			EXT4_INODE_GET_XTIME_VAL(xtime, &(einode->vfs_inode),	\
960 						 raw_inode);			\
961 	else									\
962 		(einode)->xtime = (struct timespec64){0, 0};			\
963 } while (0)
964 
965 #define i_disk_version osd1.linux1.l_i_version
966 
967 #if defined(__KERNEL__) || defined(__linux__)
968 #define i_reserved1	osd1.linux1.l_i_reserved1
969 #define i_file_acl_high	osd2.linux2.l_i_file_acl_high
970 #define i_blocks_high	osd2.linux2.l_i_blocks_high
971 #define i_uid_low	i_uid
972 #define i_gid_low	i_gid
973 #define i_uid_high	osd2.linux2.l_i_uid_high
974 #define i_gid_high	osd2.linux2.l_i_gid_high
975 #define i_checksum_lo	osd2.linux2.l_i_checksum_lo
976 
977 #elif defined(__GNU__)
978 
979 #define i_translator	osd1.hurd1.h_i_translator
980 #define i_uid_high	osd2.hurd2.h_i_uid_high
981 #define i_gid_high	osd2.hurd2.h_i_gid_high
982 #define i_author	osd2.hurd2.h_i_author
983 
984 #elif defined(__masix__)
985 
986 #define i_reserved1	osd1.masix1.m_i_reserved1
987 #define i_file_acl_high	osd2.masix2.m_i_file_acl_high
988 #define i_reserved2	osd2.masix2.m_i_reserved2
989 
990 #endif /* defined(__KERNEL__) || defined(__linux__) */
991 
992 #include "extents_status.h"
993 #include "fast_commit.h"
994 
995 /*
996  * Lock subclasses for i_data_sem in the ext4_inode_info structure.
997  *
998  * These are needed to avoid lockdep false positives when we need to
999  * allocate blocks to the quota inode during ext4_map_blocks(), while
1000  * holding i_data_sem for a normal (non-quota) inode.  Since we don't
1001  * do quota tracking for the quota inode, this avoids deadlock (as
1002  * well as infinite recursion, since it isn't turtles all the way
1003  * down...)
1004  *
1005  *  I_DATA_SEM_NORMAL - Used for most inodes
1006  *  I_DATA_SEM_OTHER  - Used by move_inode.c for the second normal inode
1007  *			  where the second inode has larger inode number
1008  *			  than the first
1009  *  I_DATA_SEM_QUOTA  - Used for quota inodes only
1010  *  I_DATA_SEM_EA     - Used for ea_inodes only
1011  */
1012 enum {
1013 	I_DATA_SEM_NORMAL = 0,
1014 	I_DATA_SEM_OTHER,
1015 	I_DATA_SEM_QUOTA,
1016 	I_DATA_SEM_EA
1017 };
1018 
1019 
1020 /*
1021  * fourth extended file system inode data in memory
1022  */
1023 struct ext4_inode_info {
1024 	__le32	i_data[15];	/* unconverted */
1025 	__u32	i_dtime;
1026 	ext4_fsblk_t	i_file_acl;
1027 
1028 	/*
1029 	 * i_block_group is the number of the block group which contains
1030 	 * this file's inode.  Constant across the lifetime of the inode,
1031 	 * it is used for making block allocation decisions - we try to
1032 	 * place a file's data blocks near its inode block, and new inodes
1033 	 * near to their parent directory's inode.
1034 	 */
1035 	ext4_group_t	i_block_group;
1036 	ext4_lblk_t	i_dir_start_lookup;
1037 #if (BITS_PER_LONG < 64)
1038 	unsigned long	i_state_flags;		/* Dynamic state flags */
1039 #endif
1040 	unsigned long	i_flags;
1041 
1042 	/*
1043 	 * Extended attributes can be read independently of the main file
1044 	 * data. Taking i_rwsem even when reading would cause contention
1045 	 * between readers of EAs and writers of regular file data, so
1046 	 * instead we synchronize on xattr_sem when reading or changing
1047 	 * EAs.
1048 	 */
1049 	struct rw_semaphore xattr_sem;
1050 
1051 	/*
1052 	 * Inodes with EXT4_STATE_ORPHAN_FILE use i_orphan_idx. Otherwise
1053 	 * i_orphan is used.
1054 	 */
1055 	union {
1056 		struct list_head i_orphan;	/* unlinked but open inodes */
1057 		unsigned int i_orphan_idx;	/* Index in orphan file */
1058 	};
1059 
1060 	/* Fast commit related info */
1061 
1062 	/* For tracking dentry create updates */
1063 	struct list_head i_fc_dilist;
1064 	struct list_head i_fc_list;	/*
1065 					 * inodes that need fast commit
1066 					 * protected by sbi->s_fc_lock.
1067 					 */
1068 
1069 	/* Start of lblk range that needs to be committed in this fast commit */
1070 	ext4_lblk_t i_fc_lblk_start;
1071 
1072 	/* End of lblk range that needs to be committed in this fast commit */
1073 	ext4_lblk_t i_fc_lblk_len;
1074 
1075 	spinlock_t i_raw_lock;	/* protects updates to the raw inode */
1076 
1077 	/* Fast commit wait queue for this inode */
1078 	wait_queue_head_t i_fc_wait;
1079 
1080 	/*
1081 	 * Protect concurrent accesses on i_fc_lblk_start, i_fc_lblk_len
1082 	 * and inode's EXT4_FC_STATE_COMMITTING state bit.
1083 	 */
1084 	spinlock_t i_fc_lock;
1085 
1086 	/*
1087 	 * i_disksize keeps track of what the inode size is ON DISK, not
1088 	 * in memory.  During truncate, i_size is set to the new size by
1089 	 * the VFS prior to calling ext4_truncate(), but the filesystem won't
1090 	 * set i_disksize to 0 until the truncate is actually under way.
1091 	 *
1092 	 * The intent is that i_disksize always represents the blocks which
1093 	 * are used by this file.  This allows recovery to restart truncate
1094 	 * on orphans if we crash during truncate.  We actually write i_disksize
1095 	 * into the on-disk inode when writing inodes out, instead of i_size.
1096 	 *
1097 	 * The only time when i_disksize and i_size may be different is when
1098 	 * a truncate is in progress.  The only things which change i_disksize
1099 	 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
1100 	 */
1101 	loff_t	i_disksize;
1102 
1103 	/*
1104 	 * i_data_sem is for serialising ext4_truncate() against
1105 	 * ext4_getblock().  In the 2.4 ext2 design, great chunks of inode's
1106 	 * data tree are chopped off during truncate. We can't do that in
1107 	 * ext4 because whenever we perform intermediate commits during
1108 	 * truncate, the inode and all the metadata blocks *must* be in a
1109 	 * consistent state which allows truncation of the orphans to restart
1110 	 * during recovery.  Hence we must fix the get_block-vs-truncate race
1111 	 * by other means, so we have i_data_sem.
1112 	 */
1113 	struct rw_semaphore i_data_sem;
1114 	struct inode vfs_inode;
1115 	struct jbd2_inode *jinode;
1116 
1117 	/*
1118 	 * File creation time. Its function is same as that of
1119 	 * struct timespec64 i_{a,c,m}time in the generic inode.
1120 	 */
1121 	struct timespec64 i_crtime;
1122 
1123 	/* mballoc */
1124 	atomic_t i_prealloc_active;
1125 
1126 	/* allocation reservation info for delalloc */
1127 	/* In case of bigalloc, this refer to clusters rather than blocks */
1128 	unsigned int i_reserved_data_blocks;
1129 	struct rb_root i_prealloc_node;
1130 	rwlock_t i_prealloc_lock;
1131 
1132 	/* extents status tree */
1133 	struct ext4_es_tree i_es_tree;
1134 	rwlock_t i_es_lock;
1135 	struct list_head i_es_list;
1136 	unsigned int i_es_all_nr;	/* protected by i_es_lock */
1137 	unsigned int i_es_shk_nr;	/* protected by i_es_lock */
1138 	ext4_lblk_t i_es_shrink_lblk;	/* Offset where we start searching for
1139 					   extents to shrink. Protected by
1140 					   i_es_lock  */
1141 
1142 	/* ialloc */
1143 	ext4_group_t	i_last_alloc_group;
1144 
1145 	/* pending cluster reservations for bigalloc file systems */
1146 	struct ext4_pending_tree i_pending_tree;
1147 
1148 	/* on-disk additional length */
1149 	__u16 i_extra_isize;
1150 
1151 	/* Indicate the inline data space. */
1152 	u16 i_inline_off;
1153 	u16 i_inline_size;
1154 
1155 #ifdef CONFIG_QUOTA
1156 	/* quota space reservation, managed internally by quota code */
1157 	qsize_t i_reserved_quota;
1158 #endif
1159 	spinlock_t i_block_reservation_lock;
1160 
1161 	/* Lock protecting lists below */
1162 	spinlock_t i_completed_io_lock;
1163 	/*
1164 	 * Completed IOs that need unwritten extents handling and have
1165 	 * transaction reserved
1166 	 */
1167 	struct list_head i_rsv_conversion_list;
1168 	struct work_struct i_rsv_conversion_work;
1169 
1170 	/*
1171 	 * Transactions that contain inode's metadata needed to complete
1172 	 * fsync and fdatasync, respectively.
1173 	 */
1174 	tid_t i_sync_tid;
1175 	tid_t i_datasync_tid;
1176 
1177 #ifdef CONFIG_QUOTA
1178 	struct dquot __rcu *i_dquot[MAXQUOTAS];
1179 #endif
1180 
1181 	/* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1182 	__u32 i_csum_seed;
1183 
1184 	kprojid_t i_projid;
1185 };
1186 
1187 /*
1188  * File system states
1189  */
1190 #define	EXT4_VALID_FS			0x0001	/* Unmounted cleanly */
1191 #define	EXT4_ERROR_FS			0x0002	/* Errors detected */
1192 #define	EXT4_ORPHAN_FS			0x0004	/* Orphans being recovered */
1193 #define EXT4_FC_REPLAY			0x0020	/* Fast commit replay ongoing */
1194 
1195 /*
1196  * Misc. filesystem flags
1197  */
1198 #define EXT2_FLAGS_SIGNED_HASH		0x0001  /* Signed dirhash in use */
1199 #define EXT2_FLAGS_UNSIGNED_HASH	0x0002  /* Unsigned dirhash in use */
1200 #define EXT2_FLAGS_TEST_FILESYS		0x0004	/* to test development code */
1201 
1202 /*
1203  * Mount flags set via mount options or defaults
1204  */
1205 #define EXT4_MOUNT_NO_MBCACHE		0x00001 /* Do not use mbcache */
1206 #define EXT4_MOUNT_GRPID		0x00004	/* Create files with directory's group */
1207 #define EXT4_MOUNT_DEBUG		0x00008	/* Some debugging messages */
1208 #define EXT4_MOUNT_ERRORS_CONT		0x00010	/* Continue on errors */
1209 #define EXT4_MOUNT_ERRORS_RO		0x00020	/* Remount fs ro on errors */
1210 #define EXT4_MOUNT_ERRORS_PANIC		0x00040	/* Panic on errors */
1211 #define EXT4_MOUNT_ERRORS_MASK		0x00070
1212 #define EXT4_MOUNT_MINIX_DF		0x00080	/* Mimics the Minix statfs */
1213 #define EXT4_MOUNT_NOLOAD		0x00100	/* Don't use existing journal*/
1214 #ifdef CONFIG_FS_DAX
1215 #define EXT4_MOUNT_DAX_ALWAYS		0x00200	/* Direct Access */
1216 #else
1217 #define EXT4_MOUNT_DAX_ALWAYS		0
1218 #endif
1219 #define EXT4_MOUNT_DATA_FLAGS		0x00C00	/* Mode for data writes: */
1220 #define EXT4_MOUNT_JOURNAL_DATA		0x00400	/* Write data to journal */
1221 #define EXT4_MOUNT_ORDERED_DATA		0x00800	/* Flush data before commit */
1222 #define EXT4_MOUNT_WRITEBACK_DATA	0x00C00	/* No data ordering */
1223 #define EXT4_MOUNT_UPDATE_JOURNAL	0x01000	/* Update the journal format */
1224 #define EXT4_MOUNT_NO_UID32		0x02000  /* Disable 32-bit UIDs */
1225 #define EXT4_MOUNT_XATTR_USER		0x04000	/* Extended user attributes */
1226 #define EXT4_MOUNT_POSIX_ACL		0x08000	/* POSIX Access Control Lists */
1227 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC	0x10000	/* No auto delalloc mapping */
1228 #define EXT4_MOUNT_BARRIER		0x20000 /* Use block barriers */
1229 #define EXT4_MOUNT_QUOTA		0x40000 /* Some quota option set */
1230 #define EXT4_MOUNT_USRQUOTA		0x80000 /* "old" user quota,
1231 						 * enable enforcement for hidden
1232 						 * quota files */
1233 #define EXT4_MOUNT_GRPQUOTA		0x100000 /* "old" group quota, enable
1234 						  * enforcement for hidden quota
1235 						  * files */
1236 #define EXT4_MOUNT_PRJQUOTA		0x200000 /* Enable project quota
1237 						  * enforcement */
1238 #define EXT4_MOUNT_DIOREAD_NOLOCK	0x400000 /* Enable support for dio read nolocking */
1239 #define EXT4_MOUNT_JOURNAL_CHECKSUM	0x800000 /* Journal checksums */
1240 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT	0x1000000 /* Journal Async Commit */
1241 #define EXT4_MOUNT_WARN_ON_ERROR	0x2000000 /* Trigger WARN_ON on error */
1242 #define EXT4_MOUNT_NO_PREFETCH_BLOCK_BITMAPS 0x4000000
1243 #define EXT4_MOUNT_DELALLOC		0x8000000 /* Delalloc support */
1244 #define EXT4_MOUNT_DATA_ERR_ABORT	0x10000000 /* Abort on file data write */
1245 #define EXT4_MOUNT_BLOCK_VALIDITY	0x20000000 /* Block validity checking */
1246 #define EXT4_MOUNT_DISCARD		0x40000000 /* Issue DISCARD requests */
1247 #define EXT4_MOUNT_INIT_INODE_TABLE	0x80000000 /* Initialize uninitialized itables */
1248 
1249 /*
1250  * Mount flags set either automatically (could not be set by mount option)
1251  * based on per file system feature or property or in special cases such as
1252  * distinguishing between explicit mount option definition and default.
1253  */
1254 #define EXT4_MOUNT2_EXPLICIT_DELALLOC	0x00000001 /* User explicitly
1255 						      specified delalloc */
1256 #define EXT4_MOUNT2_STD_GROUP_SIZE	0x00000002 /* We have standard group
1257 						      size of blocksize * 8
1258 						      blocks */
1259 #define EXT4_MOUNT2_HURD_COMPAT		0x00000004 /* Support HURD-castrated
1260 						      file systems */
1261 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM	0x00000008 /* User explicitly
1262 						specified journal checksum */
1263 
1264 #define EXT4_MOUNT2_JOURNAL_FAST_COMMIT	0x00000010 /* Journal fast commit */
1265 #define EXT4_MOUNT2_DAX_NEVER		0x00000020 /* Do not allow Direct Access */
1266 #define EXT4_MOUNT2_DAX_INODE		0x00000040 /* For printing options only */
1267 #define EXT4_MOUNT2_MB_OPTIMIZE_SCAN	0x00000080 /* Optimize group
1268 						    * scanning in mballoc
1269 						    */
1270 #define EXT4_MOUNT2_ABORT		0x00000100 /* Abort filesystem */
1271 
1272 #define clear_opt(sb, opt)		EXT4_SB(sb)->s_mount_opt &= \
1273 						~EXT4_MOUNT_##opt
1274 #define set_opt(sb, opt)		EXT4_SB(sb)->s_mount_opt |= \
1275 						EXT4_MOUNT_##opt
1276 #define test_opt(sb, opt)		(EXT4_SB(sb)->s_mount_opt & \
1277 					 EXT4_MOUNT_##opt)
1278 
1279 #define clear_opt2(sb, opt)		EXT4_SB(sb)->s_mount_opt2 &= \
1280 						~EXT4_MOUNT2_##opt
1281 #define set_opt2(sb, opt)		EXT4_SB(sb)->s_mount_opt2 |= \
1282 						EXT4_MOUNT2_##opt
1283 #define test_opt2(sb, opt)		(EXT4_SB(sb)->s_mount_opt2 & \
1284 					 EXT4_MOUNT2_##opt)
1285 
1286 #define ext4_test_and_set_bit		__test_and_set_bit_le
1287 #define ext4_set_bit			__set_bit_le
1288 #define ext4_test_and_clear_bit		__test_and_clear_bit_le
1289 #define ext4_clear_bit			__clear_bit_le
1290 #define ext4_test_bit			test_bit_le
1291 #define ext4_find_next_zero_bit		find_next_zero_bit_le
1292 #define ext4_find_next_bit		find_next_bit_le
1293 
1294 extern void mb_set_bits(void *bm, int cur, int len);
1295 
1296 /*
1297  * Maximal mount counts between two filesystem checks
1298  */
1299 #define EXT4_DFL_MAX_MNT_COUNT		20	/* Allow 20 mounts */
1300 #define EXT4_DFL_CHECKINTERVAL		0	/* Don't use interval check */
1301 
1302 /*
1303  * Behaviour when detecting errors
1304  */
1305 #define EXT4_ERRORS_CONTINUE		1	/* Continue execution */
1306 #define EXT4_ERRORS_RO			2	/* Remount fs read-only */
1307 #define EXT4_ERRORS_PANIC		3	/* Panic */
1308 #define EXT4_ERRORS_DEFAULT		EXT4_ERRORS_CONTINUE
1309 
1310 /* Metadata checksum algorithm codes */
1311 #define EXT4_CRC32C_CHKSUM		1
1312 
1313 #define EXT4_LABEL_MAX			16
1314 
1315 /*
1316  * Structure of the super block
1317  */
1318 struct ext4_super_block {
1319 /*00*/	__le32	s_inodes_count;		/* Inodes count */
1320 	__le32	s_blocks_count_lo;	/* Blocks count */
1321 	__le32	s_r_blocks_count_lo;	/* Reserved blocks count */
1322 	__le32	s_free_blocks_count_lo;	/* Free blocks count */
1323 /*10*/	__le32	s_free_inodes_count;	/* Free inodes count */
1324 	__le32	s_first_data_block;	/* First Data Block */
1325 	__le32	s_log_block_size;	/* Block size */
1326 	__le32	s_log_cluster_size;	/* Allocation cluster size */
1327 /*20*/	__le32	s_blocks_per_group;	/* # Blocks per group */
1328 	__le32	s_clusters_per_group;	/* # Clusters per group */
1329 	__le32	s_inodes_per_group;	/* # Inodes per group */
1330 	__le32	s_mtime;		/* Mount time */
1331 /*30*/	__le32	s_wtime;		/* Write time */
1332 	__le16	s_mnt_count;		/* Mount count */
1333 	__le16	s_max_mnt_count;	/* Maximal mount count */
1334 	__le16	s_magic;		/* Magic signature */
1335 	__le16	s_state;		/* File system state */
1336 	__le16	s_errors;		/* Behaviour when detecting errors */
1337 	__le16	s_minor_rev_level;	/* minor revision level */
1338 /*40*/	__le32	s_lastcheck;		/* time of last check */
1339 	__le32	s_checkinterval;	/* max. time between checks */
1340 	__le32	s_creator_os;		/* OS */
1341 	__le32	s_rev_level;		/* Revision level */
1342 /*50*/	__le16	s_def_resuid;		/* Default uid for reserved blocks */
1343 	__le16	s_def_resgid;		/* Default gid for reserved blocks */
1344 	/*
1345 	 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1346 	 *
1347 	 * Note: the difference between the compatible feature set and
1348 	 * the incompatible feature set is that if there is a bit set
1349 	 * in the incompatible feature set that the kernel doesn't
1350 	 * know about, it should refuse to mount the filesystem.
1351 	 *
1352 	 * e2fsck's requirements are more strict; if it doesn't know
1353 	 * about a feature in either the compatible or incompatible
1354 	 * feature set, it must abort and not try to meddle with
1355 	 * things it doesn't understand...
1356 	 */
1357 	__le32	s_first_ino;		/* First non-reserved inode */
1358 	__le16  s_inode_size;		/* size of inode structure */
1359 	__le16	s_block_group_nr;	/* block group # of this superblock */
1360 	__le32	s_feature_compat;	/* compatible feature set */
1361 /*60*/	__le32	s_feature_incompat;	/* incompatible feature set */
1362 	__le32	s_feature_ro_compat;	/* readonly-compatible feature set */
1363 /*68*/	__u8	s_uuid[16];		/* 128-bit uuid for volume */
1364 /*78*/	char	s_volume_name[EXT4_LABEL_MAX] __nonstring; /* volume name */
1365 /*88*/	char	s_last_mounted[64] __nonstring;	/* directory where last mounted */
1366 /*C8*/	__le32	s_algorithm_usage_bitmap; /* For compression */
1367 	/*
1368 	 * Performance hints.  Directory preallocation should only
1369 	 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1370 	 */
1371 	__u8	s_prealloc_blocks;	/* Nr of blocks to try to preallocate*/
1372 	__u8	s_prealloc_dir_blocks;	/* Nr to preallocate for dirs */
1373 	__le16	s_reserved_gdt_blocks;	/* Per group desc for online growth */
1374 	/*
1375 	 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1376 	 */
1377 /*D0*/	__u8	s_journal_uuid[16];	/* uuid of journal superblock */
1378 /*E0*/	__le32	s_journal_inum;		/* inode number of journal file */
1379 	__le32	s_journal_dev;		/* device number of journal file */
1380 	__le32	s_last_orphan;		/* start of list of inodes to delete */
1381 	__le32	s_hash_seed[4];		/* HTREE hash seed */
1382 	__u8	s_def_hash_version;	/* Default hash version to use */
1383 	__u8	s_jnl_backup_type;
1384 	__le16  s_desc_size;		/* size of group descriptor */
1385 /*100*/	__le32	s_default_mount_opts;
1386 	__le32	s_first_meta_bg;	/* First metablock block group */
1387 	__le32	s_mkfs_time;		/* When the filesystem was created */
1388 	__le32	s_jnl_blocks[17];	/* Backup of the journal inode */
1389 	/* 64bit support valid if EXT4_FEATURE_INCOMPAT_64BIT */
1390 /*150*/	__le32	s_blocks_count_hi;	/* Blocks count */
1391 	__le32	s_r_blocks_count_hi;	/* Reserved blocks count */
1392 	__le32	s_free_blocks_count_hi;	/* Free blocks count */
1393 	__le16	s_min_extra_isize;	/* All inodes have at least # bytes */
1394 	__le16	s_want_extra_isize; 	/* New inodes should reserve # bytes */
1395 	__le32	s_flags;		/* Miscellaneous flags */
1396 	__le16  s_raid_stride;		/* RAID stride */
1397 	__le16  s_mmp_update_interval;  /* # seconds to wait in MMP checking */
1398 	__le64  s_mmp_block;            /* Block for multi-mount protection */
1399 	__le32  s_raid_stripe_width;    /* blocks on all data disks (N*stride)*/
1400 	__u8	s_log_groups_per_flex;  /* FLEX_BG group size */
1401 	__u8	s_checksum_type;	/* metadata checksum algorithm used */
1402 	__u8	s_encryption_level;	/* versioning level for encryption */
1403 	__u8	s_reserved_pad;		/* Padding to next 32bits */
1404 	__le64	s_kbytes_written;	/* nr of lifetime kilobytes written */
1405 	__le32	s_snapshot_inum;	/* Inode number of active snapshot */
1406 	__le32	s_snapshot_id;		/* sequential ID of active snapshot */
1407 	__le64	s_snapshot_r_blocks_count; /* reserved blocks for active
1408 					      snapshot's future use */
1409 	__le32	s_snapshot_list;	/* inode number of the head of the
1410 					   on-disk snapshot list */
1411 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1412 	__le32	s_error_count;		/* number of fs errors */
1413 	__le32	s_first_error_time;	/* first time an error happened */
1414 	__le32	s_first_error_ino;	/* inode involved in first error */
1415 	__le64	s_first_error_block;	/* block involved of first error */
1416 	__u8	s_first_error_func[32] __nonstring;	/* function where the error happened */
1417 	__le32	s_first_error_line;	/* line number where error happened */
1418 	__le32	s_last_error_time;	/* most recent time of an error */
1419 	__le32	s_last_error_ino;	/* inode involved in last error */
1420 	__le32	s_last_error_line;	/* line number where error happened */
1421 	__le64	s_last_error_block;	/* block involved of last error */
1422 	__u8	s_last_error_func[32] __nonstring;	/* function where the error happened */
1423 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1424 	__u8	s_mount_opts[64];
1425 	__le32	s_usr_quota_inum;	/* inode for tracking user quota */
1426 	__le32	s_grp_quota_inum;	/* inode for tracking group quota */
1427 	__le32	s_overhead_clusters;	/* overhead blocks/clusters in fs */
1428 	__le32	s_backup_bgs[2];	/* groups with sparse_super2 SBs */
1429 	__u8	s_encrypt_algos[4];	/* Encryption algorithms in use  */
1430 	__u8	s_encrypt_pw_salt[16];	/* Salt used for string2key algorithm */
1431 	__le32	s_lpf_ino;		/* Location of the lost+found inode */
1432 	__le32	s_prj_quota_inum;	/* inode for tracking project quota */
1433 	__le32	s_checksum_seed;	/* crc32c(uuid) if csum_seed set */
1434 	__u8	s_wtime_hi;
1435 	__u8	s_mtime_hi;
1436 	__u8	s_mkfs_time_hi;
1437 	__u8	s_lastcheck_hi;
1438 	__u8	s_first_error_time_hi;
1439 	__u8	s_last_error_time_hi;
1440 	__u8	s_first_error_errcode;
1441 	__u8    s_last_error_errcode;
1442 	__le16  s_encoding;		/* Filename charset encoding */
1443 	__le16  s_encoding_flags;	/* Filename charset encoding flags */
1444 	__le32  s_orphan_file_inum;	/* Inode for tracking orphan inodes */
1445 	__le32	s_reserved[94];		/* Padding to the end of the block */
1446 	__le32	s_checksum;		/* crc32c(superblock) */
1447 };
1448 
1449 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1450 
1451 #ifdef __KERNEL__
1452 
1453 /* Number of quota types we support */
1454 #define EXT4_MAXQUOTAS 3
1455 
1456 #define EXT4_ENC_UTF8_12_1	1
1457 
1458 /* Types of ext4 journal triggers */
1459 enum ext4_journal_trigger_type {
1460 	EXT4_JTR_ORPHAN_FILE,
1461 	EXT4_JTR_NONE	/* This must be the last entry for indexing to work! */
1462 };
1463 
1464 #define EXT4_JOURNAL_TRIGGER_COUNT EXT4_JTR_NONE
1465 
1466 struct ext4_journal_trigger {
1467 	struct jbd2_buffer_trigger_type tr_triggers;
1468 	struct super_block *sb;
1469 };
1470 
EXT4_TRIGGER(struct jbd2_buffer_trigger_type * trigger)1471 static inline struct ext4_journal_trigger *EXT4_TRIGGER(
1472 				struct jbd2_buffer_trigger_type *trigger)
1473 {
1474 	return container_of(trigger, struct ext4_journal_trigger, tr_triggers);
1475 }
1476 
1477 #define EXT4_ORPHAN_BLOCK_MAGIC 0x0b10ca04
1478 
1479 /* Structure at the tail of orphan block */
1480 struct ext4_orphan_block_tail {
1481 	__le32 ob_magic;
1482 	__le32 ob_checksum;
1483 };
1484 
ext4_inodes_per_orphan_block(struct super_block * sb)1485 static inline int ext4_inodes_per_orphan_block(struct super_block *sb)
1486 {
1487 	return (sb->s_blocksize - sizeof(struct ext4_orphan_block_tail)) /
1488 			sizeof(u32);
1489 }
1490 
1491 struct ext4_orphan_block {
1492 	atomic_t ob_free_entries;	/* Number of free orphan entries in block */
1493 	struct buffer_head *ob_bh;	/* Buffer for orphan block */
1494 };
1495 
1496 /*
1497  * Info about orphan file.
1498  */
1499 struct ext4_orphan_info {
1500 	int of_blocks;			/* Number of orphan blocks in a file */
1501 	__u32 of_csum_seed;		/* Checksum seed for orphan file */
1502 	struct ext4_orphan_block *of_binfo;	/* Array with info about orphan
1503 						 * file blocks */
1504 };
1505 
1506 /*
1507  * fourth extended-fs super-block data in memory
1508  */
1509 struct ext4_sb_info {
1510 	unsigned long s_desc_size;	/* Size of a group descriptor in bytes */
1511 	unsigned long s_inodes_per_block;/* Number of inodes per block */
1512 	unsigned long s_blocks_per_group;/* Number of blocks in a group */
1513 	unsigned long s_clusters_per_group; /* Number of clusters in a group */
1514 	unsigned long s_inodes_per_group;/* Number of inodes in a group */
1515 	unsigned long s_itb_per_group;	/* Number of inode table blocks per group */
1516 	unsigned long s_gdb_count;	/* Number of group descriptor blocks */
1517 	unsigned long s_desc_per_block;	/* Number of group descriptors per block */
1518 	ext4_group_t s_groups_count;	/* Number of groups in the fs */
1519 	ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1520 	unsigned long s_overhead;  /* # of fs overhead clusters */
1521 	unsigned int s_cluster_ratio;	/* Number of blocks per cluster */
1522 	unsigned int s_cluster_bits;	/* log2 of s_cluster_ratio */
1523 	loff_t s_bitmap_maxbytes;	/* max bytes for bitmap files */
1524 	struct buffer_head * s_sbh;	/* Buffer containing the super block */
1525 	struct ext4_super_block *s_es;	/* Pointer to the super block in the buffer */
1526 	/* Array of bh's for the block group descriptors */
1527 	struct buffer_head * __rcu *s_group_desc;
1528 	unsigned int s_mount_opt;
1529 	unsigned int s_mount_opt2;
1530 	unsigned long s_mount_flags;
1531 	unsigned int s_def_mount_opt;
1532 	unsigned int s_def_mount_opt2;
1533 	ext4_fsblk_t s_sb_block;
1534 	atomic64_t s_resv_clusters;
1535 	kuid_t s_resuid;
1536 	kgid_t s_resgid;
1537 	unsigned short s_mount_state;
1538 	unsigned short s_pad;
1539 	int s_addr_per_block_bits;
1540 	int s_desc_per_block_bits;
1541 	int s_inode_size;
1542 	int s_first_ino;
1543 	unsigned int s_inode_readahead_blks;
1544 	unsigned int s_inode_goal;
1545 	u32 s_hash_seed[4];
1546 	int s_def_hash_version;
1547 	int s_hash_unsigned;	/* 3 if hash should be unsigned, 0 if not */
1548 	struct percpu_counter s_freeclusters_counter;
1549 	struct percpu_counter s_freeinodes_counter;
1550 	struct percpu_counter s_dirs_counter;
1551 	struct percpu_counter s_dirtyclusters_counter;
1552 	struct percpu_counter s_sra_exceeded_retry_limit;
1553 	struct blockgroup_lock *s_blockgroup_lock;
1554 	struct proc_dir_entry *s_proc;
1555 	struct kobject s_kobj;
1556 	struct completion s_kobj_unregister;
1557 	struct super_block *s_sb;
1558 	struct buffer_head *s_mmp_bh;
1559 
1560 	/* Journaling */
1561 	struct journal_s *s_journal;
1562 	unsigned long s_ext4_flags;		/* Ext4 superblock flags */
1563 	struct mutex s_orphan_lock;	/* Protects on disk list changes */
1564 	struct list_head s_orphan;	/* List of orphaned inodes in on disk
1565 					   list */
1566 	struct ext4_orphan_info s_orphan_info;
1567 	unsigned long s_commit_interval;
1568 	u32 s_max_batch_time;
1569 	u32 s_min_batch_time;
1570 	struct file *s_journal_bdev_file;
1571 #ifdef CONFIG_QUOTA
1572 	/* Names of quota files with journalled quota */
1573 	char __rcu *s_qf_names[EXT4_MAXQUOTAS];
1574 	int s_jquota_fmt;			/* Format of quota to use */
1575 #endif
1576 	unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1577 	struct ext4_system_blocks __rcu *s_system_blks;
1578 
1579 #ifdef EXTENTS_STATS
1580 	/* ext4 extents stats */
1581 	unsigned long s_ext_min;
1582 	unsigned long s_ext_max;
1583 	unsigned long s_depth_max;
1584 	spinlock_t s_ext_stats_lock;
1585 	unsigned long s_ext_blocks;
1586 	unsigned long s_ext_extents;
1587 #endif
1588 
1589 	/* for buddy allocator */
1590 	struct ext4_group_info ** __rcu *s_group_info;
1591 	struct inode *s_buddy_cache;
1592 	spinlock_t s_md_lock;
1593 	unsigned short *s_mb_offsets;
1594 	unsigned int *s_mb_maxs;
1595 	unsigned int s_group_info_size;
1596 	atomic_t s_mb_free_pending;
1597 	struct list_head s_freed_data_list[2];	/* List of blocks to be freed
1598 						   after commit completed */
1599 	struct list_head s_discard_list;
1600 	struct work_struct s_discard_work;
1601 	atomic_t s_retry_alloc_pending;
1602 	struct xarray *s_mb_avg_fragment_size;
1603 	struct xarray *s_mb_largest_free_orders;
1604 
1605 	/* tunables */
1606 	unsigned long s_stripe;
1607 	unsigned int s_mb_max_linear_groups;
1608 	unsigned int s_mb_stream_request;
1609 	unsigned int s_mb_max_to_scan;
1610 	unsigned int s_mb_min_to_scan;
1611 	unsigned int s_mb_stats;
1612 	unsigned int s_mb_order2_reqs;
1613 	unsigned int s_mb_group_prealloc;
1614 	unsigned int s_max_dir_size_kb;
1615 	unsigned int s_mb_prefetch;
1616 	unsigned int s_mb_prefetch_limit;
1617 	unsigned int s_mb_best_avail_max_trim_order;
1618 	unsigned int s_sb_update_sec;
1619 	unsigned int s_sb_update_kb;
1620 
1621 	/* where last allocation was done - for stream allocation */
1622 	ext4_group_t *s_mb_last_groups;
1623 	unsigned int s_mb_nr_global_goals;
1624 
1625 	/* stats for buddy allocator */
1626 	atomic_t s_bal_reqs;	/* number of reqs with len > 1 */
1627 	atomic_t s_bal_success;	/* we found long enough chunks */
1628 	atomic_t s_bal_allocated;	/* in blocks */
1629 	atomic_t s_bal_ex_scanned;	/* total extents scanned */
1630 	atomic_t s_bal_cX_ex_scanned[EXT4_MB_NUM_CRS];	/* total extents scanned */
1631 	atomic_t s_bal_groups_scanned;	/* number of groups scanned */
1632 	atomic_t s_bal_goals;	/* goal hits */
1633 	atomic_t s_bal_stream_goals;	/* stream allocation global goal hits */
1634 	atomic_t s_bal_len_goals;	/* len goal hits */
1635 	atomic_t s_bal_breaks;	/* too long searches */
1636 	atomic_t s_bal_2orders;	/* 2^order hits */
1637 	atomic64_t s_bal_cX_groups_considered[EXT4_MB_NUM_CRS];
1638 	atomic64_t s_bal_cX_hits[EXT4_MB_NUM_CRS];
1639 	atomic64_t s_bal_cX_failed[EXT4_MB_NUM_CRS];		/* cX loop didn't find blocks */
1640 	atomic_t s_mb_buddies_generated;	/* number of buddies generated */
1641 	atomic64_t s_mb_generation_time;
1642 	atomic_t s_mb_lost_chunks;
1643 	atomic_t s_mb_preallocated;
1644 	atomic_t s_mb_discarded;
1645 	atomic_t s_lock_busy;
1646 
1647 	/* locality groups */
1648 	struct ext4_locality_group __percpu *s_locality_groups;
1649 
1650 	/* for write statistics */
1651 	unsigned long s_sectors_written_start;
1652 	u64 s_kbytes_written;
1653 
1654 	/* the size of zero-out chunk */
1655 	unsigned int s_extent_max_zeroout_kb;
1656 
1657 	unsigned int s_log_groups_per_flex;
1658 	struct flex_groups * __rcu *s_flex_groups;
1659 	ext4_group_t s_flex_groups_allocated;
1660 
1661 	/* workqueue for reserved extent conversions (buffered io) */
1662 	struct workqueue_struct *rsv_conversion_wq;
1663 
1664 	/* timer for periodic error stats printing */
1665 	struct timer_list s_err_report;
1666 
1667 	/* Lazy inode table initialization info */
1668 	struct ext4_li_request *s_li_request;
1669 	/* Wait multiplier for lazy initialization thread */
1670 	unsigned int s_li_wait_mult;
1671 
1672 	/* Kernel thread for multiple mount protection */
1673 	struct task_struct *s_mmp_tsk;
1674 
1675 	/* record the last minlen when FITRIM is called. */
1676 	unsigned long s_last_trim_minblks;
1677 
1678 	/* Precomputed FS UUID checksum for seeding other checksums */
1679 	__u32 s_csum_seed;
1680 
1681 	/* Reclaim extents from extent status tree */
1682 	struct shrinker *s_es_shrinker;
1683 	struct list_head s_es_list;	/* List of inodes with reclaimable extents */
1684 	long s_es_nr_inode;
1685 	struct ext4_es_stats s_es_stats;
1686 	struct mb_cache *s_ea_block_cache;
1687 	struct mb_cache *s_ea_inode_cache;
1688 	spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1689 
1690 	/* Journal triggers for checksum computation */
1691 	struct ext4_journal_trigger s_journal_triggers[EXT4_JOURNAL_TRIGGER_COUNT];
1692 
1693 	/* Ratelimit ext4 messages. */
1694 	struct ratelimit_state s_err_ratelimit_state;
1695 	struct ratelimit_state s_warning_ratelimit_state;
1696 	struct ratelimit_state s_msg_ratelimit_state;
1697 	atomic_t s_warning_count;
1698 	atomic_t s_msg_count;
1699 
1700 	/* Encryption policy for '-o test_dummy_encryption' */
1701 	struct fscrypt_dummy_policy s_dummy_enc_policy;
1702 
1703 	/*
1704 	 * Barrier between writepages ops and changing any inode's JOURNAL_DATA
1705 	 * or EXTENTS flag or between writepages ops and changing DELALLOC or
1706 	 * DIOREAD_NOLOCK mount options on remount.
1707 	 */
1708 	struct percpu_rw_semaphore s_writepages_rwsem;
1709 	struct dax_device *s_daxdev;
1710 	u64 s_dax_part_off;
1711 #ifdef CONFIG_EXT4_DEBUG
1712 	unsigned long s_simulate_fail;
1713 #endif
1714 	/* Record the errseq of the backing block device */
1715 	errseq_t s_bdev_wb_err;
1716 	spinlock_t s_bdev_wb_lock;
1717 
1718 	/* Information about errors that happened during this mount */
1719 	spinlock_t s_error_lock;
1720 	int s_add_error_count;
1721 	int s_first_error_code;
1722 	__u32 s_first_error_line;
1723 	__u32 s_first_error_ino;
1724 	__u64 s_first_error_block;
1725 	const char *s_first_error_func;
1726 	time64_t s_first_error_time;
1727 	int s_last_error_code;
1728 	__u32 s_last_error_line;
1729 	__u32 s_last_error_ino;
1730 	__u64 s_last_error_block;
1731 	const char *s_last_error_func;
1732 	time64_t s_last_error_time;
1733 	/*
1734 	 * If we are in a context where we cannot update the on-disk
1735 	 * superblock, we queue the work here.  This is used to update
1736 	 * the error information in the superblock, and for periodic
1737 	 * updates of the superblock called from the commit callback
1738 	 * function.
1739 	 */
1740 	struct work_struct s_sb_upd_work;
1741 
1742 	/* Atomic write unit values in bytes */
1743 	unsigned int s_awu_min;
1744 	unsigned int s_awu_max;
1745 
1746 	/* Ext4 fast commit sub transaction ID */
1747 	atomic_t s_fc_subtid;
1748 
1749 	/*
1750 	 * After commit starts, the main queue gets locked, and the further
1751 	 * updates get added in the staging queue.
1752 	 */
1753 #define FC_Q_MAIN	0
1754 #define FC_Q_STAGING	1
1755 	struct list_head s_fc_q[2];	/* Inodes staged for fast commit
1756 					 * that have data changes in them.
1757 					 */
1758 	struct list_head s_fc_dentry_q[2];	/* directory entry updates */
1759 	unsigned int s_fc_bytes;
1760 	/*
1761 	 * Main fast commit lock. This lock protects accesses to the
1762 	 * following fields:
1763 	 * ei->i_fc_list, s_fc_dentry_q, s_fc_q, s_fc_bytes, s_fc_bh.
1764 	 */
1765 	struct mutex s_fc_lock;
1766 	struct buffer_head *s_fc_bh;
1767 	struct ext4_fc_stats s_fc_stats;
1768 	tid_t s_fc_ineligible_tid;
1769 #ifdef CONFIG_EXT4_DEBUG
1770 	int s_fc_debug_max_replay;
1771 #endif
1772 	struct ext4_fc_replay_state s_fc_replay_state;
1773 };
1774 
EXT4_SB(struct super_block * sb)1775 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1776 {
1777 	return sb->s_fs_info;
1778 }
EXT4_I(struct inode * inode)1779 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1780 {
1781 	return container_of(inode, struct ext4_inode_info, vfs_inode);
1782 }
1783 
ext4_writepages_down_read(struct super_block * sb)1784 static inline int ext4_writepages_down_read(struct super_block *sb)
1785 {
1786 	percpu_down_read(&EXT4_SB(sb)->s_writepages_rwsem);
1787 	return memalloc_nofs_save();
1788 }
1789 
ext4_writepages_up_read(struct super_block * sb,int ctx)1790 static inline void ext4_writepages_up_read(struct super_block *sb, int ctx)
1791 {
1792 	memalloc_nofs_restore(ctx);
1793 	percpu_up_read(&EXT4_SB(sb)->s_writepages_rwsem);
1794 }
1795 
ext4_writepages_down_write(struct super_block * sb)1796 static inline int ext4_writepages_down_write(struct super_block *sb)
1797 {
1798 	percpu_down_write(&EXT4_SB(sb)->s_writepages_rwsem);
1799 	return memalloc_nofs_save();
1800 }
1801 
ext4_writepages_up_write(struct super_block * sb,int ctx)1802 static inline void ext4_writepages_up_write(struct super_block *sb, int ctx)
1803 {
1804 	memalloc_nofs_restore(ctx);
1805 	percpu_up_write(&EXT4_SB(sb)->s_writepages_rwsem);
1806 }
1807 
ext4_valid_inum(struct super_block * sb,unsigned long ino)1808 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1809 {
1810 	return ino == EXT4_ROOT_INO ||
1811 		(ino >= EXT4_FIRST_INO(sb) &&
1812 		 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1813 }
1814 
1815 /*
1816  * Returns: sbi->field[index]
1817  * Used to access an array element from the following sbi fields which require
1818  * rcu protection to avoid dereferencing an invalid pointer due to reassignment
1819  * - s_group_desc
1820  * - s_group_info
1821  * - s_flex_group
1822  */
1823 #define sbi_array_rcu_deref(sbi, field, index)				   \
1824 ({									   \
1825 	typeof(*((sbi)->field)) _v;					   \
1826 	rcu_read_lock();						   \
1827 	_v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index];	   \
1828 	rcu_read_unlock();						   \
1829 	_v;								   \
1830 })
1831 
1832 /*
1833  * run-time mount flags
1834  */
1835 enum {
1836 	EXT4_MF_MNTDIR_SAMPLED,
1837 	EXT4_MF_FC_INELIGIBLE,	/* Fast commit ineligible */
1838 	EXT4_MF_JOURNAL_DESTROY	/* Journal is in process of destroying */
1839 };
1840 
ext4_set_mount_flag(struct super_block * sb,int bit)1841 static inline void ext4_set_mount_flag(struct super_block *sb, int bit)
1842 {
1843 	set_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1844 }
1845 
ext4_clear_mount_flag(struct super_block * sb,int bit)1846 static inline void ext4_clear_mount_flag(struct super_block *sb, int bit)
1847 {
1848 	clear_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1849 }
1850 
ext4_test_mount_flag(struct super_block * sb,int bit)1851 static inline int ext4_test_mount_flag(struct super_block *sb, int bit)
1852 {
1853 	return test_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1854 }
1855 
1856 
1857 /*
1858  * Simulate_fail codes
1859  */
1860 #define EXT4_SIM_BBITMAP_EIO	1
1861 #define EXT4_SIM_BBITMAP_CRC	2
1862 #define EXT4_SIM_IBITMAP_EIO	3
1863 #define EXT4_SIM_IBITMAP_CRC	4
1864 #define EXT4_SIM_INODE_EIO	5
1865 #define EXT4_SIM_INODE_CRC	6
1866 #define EXT4_SIM_DIRBLOCK_EIO	7
1867 #define EXT4_SIM_DIRBLOCK_CRC	8
1868 
ext4_simulate_fail(struct super_block * sb,unsigned long code)1869 static inline bool ext4_simulate_fail(struct super_block *sb,
1870 				     unsigned long code)
1871 {
1872 #ifdef CONFIG_EXT4_DEBUG
1873 	struct ext4_sb_info *sbi = EXT4_SB(sb);
1874 
1875 	if (unlikely(sbi->s_simulate_fail == code)) {
1876 		sbi->s_simulate_fail = 0;
1877 		return true;
1878 	}
1879 #endif
1880 	return false;
1881 }
1882 
1883 /*
1884  * Error number codes for s_{first,last}_error_errno
1885  *
1886  * Linux errno numbers are architecture specific, so we need to translate
1887  * them into something which is architecture independent.   We don't define
1888  * codes for all errno's; just the ones which are most likely to be the cause
1889  * of an ext4_error() call.
1890  */
1891 #define EXT4_ERR_UNKNOWN	 1
1892 #define EXT4_ERR_EIO		 2
1893 #define EXT4_ERR_ENOMEM		 3
1894 #define EXT4_ERR_EFSBADCRC	 4
1895 #define EXT4_ERR_EFSCORRUPTED	 5
1896 #define EXT4_ERR_ENOSPC		 6
1897 #define EXT4_ERR_ENOKEY		 7
1898 #define EXT4_ERR_EROFS		 8
1899 #define EXT4_ERR_EFBIG		 9
1900 #define EXT4_ERR_EEXIST		10
1901 #define EXT4_ERR_ERANGE		11
1902 #define EXT4_ERR_EOVERFLOW	12
1903 #define EXT4_ERR_EBUSY		13
1904 #define EXT4_ERR_ENOTDIR	14
1905 #define EXT4_ERR_ENOTEMPTY	15
1906 #define EXT4_ERR_ESHUTDOWN	16
1907 #define EXT4_ERR_EFAULT		17
1908 
1909 /*
1910  * Inode dynamic state flags
1911  */
1912 enum {
1913 	EXT4_STATE_NEW,			/* inode is newly created */
1914 	EXT4_STATE_XATTR,		/* has in-inode xattrs */
1915 	EXT4_STATE_NO_EXPAND,		/* No space for expansion */
1916 	EXT4_STATE_DA_ALLOC_CLOSE,	/* Alloc DA blks on close */
1917 	EXT4_STATE_EXT_MIGRATE,		/* Inode is migrating */
1918 	EXT4_STATE_NEWENTRY,		/* File just added to dir */
1919 	EXT4_STATE_MAY_INLINE_DATA,	/* may have in-inode data */
1920 	EXT4_STATE_EXT_PRECACHED,	/* extents have been precached */
1921 	EXT4_STATE_LUSTRE_EA_INODE,	/* Lustre-style ea_inode */
1922 	EXT4_STATE_VERITY_IN_PROGRESS,	/* building fs-verity Merkle tree */
1923 	EXT4_STATE_FC_COMMITTING,	/* Fast commit ongoing */
1924 	EXT4_STATE_FC_FLUSHING_DATA,	/* Fast commit flushing data */
1925 	EXT4_STATE_ORPHAN_FILE,		/* Inode orphaned in orphan file */
1926 };
1927 
1928 #define EXT4_INODE_BIT_FNS(name, field, offset)				\
1929 static inline int ext4_test_inode_##name(struct inode *inode, int bit)	\
1930 {									\
1931 	return test_bit(bit + (offset), &EXT4_I(inode)->i_##field);	\
1932 }									\
1933 static inline void ext4_set_inode_##name(struct inode *inode, int bit)	\
1934 {									\
1935 	set_bit(bit + (offset), &EXT4_I(inode)->i_##field);		\
1936 }									\
1937 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1938 {									\
1939 	clear_bit(bit + (offset), &EXT4_I(inode)->i_##field);		\
1940 }
1941 
1942 /* Add these declarations here only so that these functions can be
1943  * found by name.  Otherwise, they are very hard to locate. */
1944 static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1945 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1946 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1947 EXT4_INODE_BIT_FNS(flag, flags, 0)
1948 
1949 /* Add these declarations here only so that these functions can be
1950  * found by name.  Otherwise, they are very hard to locate. */
1951 static inline int ext4_test_inode_state(struct inode *inode, int bit);
1952 static inline void ext4_set_inode_state(struct inode *inode, int bit);
1953 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1954 #if (BITS_PER_LONG < 64)
1955 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1956 
ext4_clear_state_flags(struct ext4_inode_info * ei)1957 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1958 {
1959 	(ei)->i_state_flags = 0;
1960 }
1961 #else
1962 EXT4_INODE_BIT_FNS(state, flags, 32)
1963 
ext4_clear_state_flags(struct ext4_inode_info * ei)1964 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1965 {
1966 	/* We depend on the fact that callers will set i_flags */
1967 }
1968 #endif
1969 #else
1970 /* Assume that user mode programs are passing in an ext4fs superblock, not
1971  * a kernel struct super_block.  This will allow us to call the feature-test
1972  * macros from user land. */
1973 #define EXT4_SB(sb)	(sb)
1974 #endif
1975 
ext4_verity_in_progress(struct inode * inode)1976 static inline bool ext4_verity_in_progress(struct inode *inode)
1977 {
1978 	return IS_ENABLED(CONFIG_FS_VERITY) &&
1979 	       ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS);
1980 }
1981 
1982 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1983 
1984 /*
1985  * Codes for operating systems
1986  */
1987 #define EXT4_OS_LINUX		0
1988 #define EXT4_OS_HURD		1
1989 #define EXT4_OS_MASIX		2
1990 #define EXT4_OS_FREEBSD		3
1991 #define EXT4_OS_LITES		4
1992 
1993 /*
1994  * Revision levels
1995  */
1996 #define EXT4_GOOD_OLD_REV	0	/* The good old (original) format */
1997 #define EXT4_DYNAMIC_REV	1	/* V2 format w/ dynamic inode sizes */
1998 
1999 #define EXT4_MAX_SUPP_REV	EXT4_DYNAMIC_REV
2000 
2001 #define EXT4_GOOD_OLD_INODE_SIZE 128
2002 
2003 #define EXT4_EXTRA_TIMESTAMP_MAX	(((s64)1 << 34) - 1  + S32_MIN)
2004 #define EXT4_NON_EXTRA_TIMESTAMP_MAX	S32_MAX
2005 #define EXT4_TIMESTAMP_MIN		S32_MIN
2006 
2007 /*
2008  * Feature set definitions
2009  */
2010 
2011 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC	0x0001
2012 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES	0x0002
2013 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL		0x0004
2014 #define EXT4_FEATURE_COMPAT_EXT_ATTR		0x0008
2015 #define EXT4_FEATURE_COMPAT_RESIZE_INODE	0x0010
2016 #define EXT4_FEATURE_COMPAT_DIR_INDEX		0x0020
2017 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2	0x0200
2018 /*
2019  * The reason why "FAST_COMMIT" is a compat feature is that, FS becomes
2020  * incompatible only if fast commit blocks are present in the FS. Since we
2021  * clear the journal (and thus the fast commit blocks), we don't mark FS as
2022  * incompatible. We also have a JBD2 incompat feature, which gets set when
2023  * there are fast commit blocks present in the journal.
2024  */
2025 #define EXT4_FEATURE_COMPAT_FAST_COMMIT		0x0400
2026 #define EXT4_FEATURE_COMPAT_STABLE_INODES	0x0800
2027 #define EXT4_FEATURE_COMPAT_ORPHAN_FILE		0x1000	/* Orphan file exists */
2028 
2029 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER	0x0001
2030 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE	0x0002
2031 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR	0x0004
2032 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE        0x0008
2033 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM		0x0010
2034 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK	0x0020
2035 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE	0x0040
2036 #define EXT4_FEATURE_RO_COMPAT_QUOTA		0x0100
2037 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC		0x0200
2038 /*
2039  * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM).  When
2040  * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
2041  * all other data structures' checksums.  However, the METADATA_CSUM and
2042  * GDT_CSUM bits are mutually exclusive.
2043  */
2044 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM	0x0400
2045 #define EXT4_FEATURE_RO_COMPAT_READONLY		0x1000
2046 #define EXT4_FEATURE_RO_COMPAT_PROJECT		0x2000
2047 #define EXT4_FEATURE_RO_COMPAT_VERITY		0x8000
2048 #define EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT	0x10000 /* Orphan file may be
2049 							   non-empty */
2050 
2051 #define EXT4_FEATURE_INCOMPAT_COMPRESSION	0x0001
2052 #define EXT4_FEATURE_INCOMPAT_FILETYPE		0x0002
2053 #define EXT4_FEATURE_INCOMPAT_RECOVER		0x0004 /* Needs recovery */
2054 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV	0x0008 /* Journal device */
2055 #define EXT4_FEATURE_INCOMPAT_META_BG		0x0010
2056 #define EXT4_FEATURE_INCOMPAT_EXTENTS		0x0040 /* extents support */
2057 #define EXT4_FEATURE_INCOMPAT_64BIT		0x0080
2058 #define EXT4_FEATURE_INCOMPAT_MMP               0x0100
2059 #define EXT4_FEATURE_INCOMPAT_FLEX_BG		0x0200
2060 #define EXT4_FEATURE_INCOMPAT_EA_INODE		0x0400 /* EA in inode */
2061 #define EXT4_FEATURE_INCOMPAT_DIRDATA		0x1000 /* data in dirent */
2062 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED		0x2000
2063 #define EXT4_FEATURE_INCOMPAT_LARGEDIR		0x4000 /* >2GB or 3-lvl htree */
2064 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA	0x8000 /* data in inode */
2065 #define EXT4_FEATURE_INCOMPAT_ENCRYPT		0x10000
2066 #define EXT4_FEATURE_INCOMPAT_CASEFOLD		0x20000
2067 
2068 extern void ext4_update_dynamic_rev(struct super_block *sb);
2069 
2070 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
2071 static inline bool ext4_has_feature_##name(struct super_block *sb) \
2072 { \
2073 	return ((EXT4_SB(sb)->s_es->s_feature_compat & \
2074 		cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
2075 } \
2076 static inline void ext4_set_feature_##name(struct super_block *sb) \
2077 { \
2078 	ext4_update_dynamic_rev(sb); \
2079 	EXT4_SB(sb)->s_es->s_feature_compat |= \
2080 		cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
2081 } \
2082 static inline void ext4_clear_feature_##name(struct super_block *sb) \
2083 { \
2084 	EXT4_SB(sb)->s_es->s_feature_compat &= \
2085 		~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
2086 }
2087 
2088 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
2089 static inline bool ext4_has_feature_##name(struct super_block *sb) \
2090 { \
2091 	return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
2092 		cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
2093 } \
2094 static inline void ext4_set_feature_##name(struct super_block *sb) \
2095 { \
2096 	ext4_update_dynamic_rev(sb); \
2097 	EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
2098 		cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
2099 } \
2100 static inline void ext4_clear_feature_##name(struct super_block *sb) \
2101 { \
2102 	EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
2103 		~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
2104 }
2105 
2106 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
2107 static inline bool ext4_has_feature_##name(struct super_block *sb) \
2108 { \
2109 	return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
2110 		cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
2111 } \
2112 static inline void ext4_set_feature_##name(struct super_block *sb) \
2113 { \
2114 	ext4_update_dynamic_rev(sb); \
2115 	EXT4_SB(sb)->s_es->s_feature_incompat |= \
2116 		cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
2117 } \
2118 static inline void ext4_clear_feature_##name(struct super_block *sb) \
2119 { \
2120 	EXT4_SB(sb)->s_es->s_feature_incompat &= \
2121 		~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
2122 }
2123 
EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc,DIR_PREALLOC)2124 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc,		DIR_PREALLOC)
2125 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes,	IMAGIC_INODES)
2126 EXT4_FEATURE_COMPAT_FUNCS(journal,		HAS_JOURNAL)
2127 EXT4_FEATURE_COMPAT_FUNCS(xattr,		EXT_ATTR)
2128 EXT4_FEATURE_COMPAT_FUNCS(resize_inode,		RESIZE_INODE)
2129 EXT4_FEATURE_COMPAT_FUNCS(dir_index,		DIR_INDEX)
2130 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2,	SPARSE_SUPER2)
2131 EXT4_FEATURE_COMPAT_FUNCS(fast_commit,		FAST_COMMIT)
2132 EXT4_FEATURE_COMPAT_FUNCS(stable_inodes,	STABLE_INODES)
2133 EXT4_FEATURE_COMPAT_FUNCS(orphan_file,		ORPHAN_FILE)
2134 
2135 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super,	SPARSE_SUPER)
2136 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file,	LARGE_FILE)
2137 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir,		BTREE_DIR)
2138 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file,		HUGE_FILE)
2139 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum,		GDT_CSUM)
2140 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink,		DIR_NLINK)
2141 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize,	EXTRA_ISIZE)
2142 EXT4_FEATURE_RO_COMPAT_FUNCS(quota,		QUOTA)
2143 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc,		BIGALLOC)
2144 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum,	METADATA_CSUM)
2145 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly,		READONLY)
2146 EXT4_FEATURE_RO_COMPAT_FUNCS(project,		PROJECT)
2147 EXT4_FEATURE_RO_COMPAT_FUNCS(verity,		VERITY)
2148 EXT4_FEATURE_RO_COMPAT_FUNCS(orphan_present,	ORPHAN_PRESENT)
2149 
2150 EXT4_FEATURE_INCOMPAT_FUNCS(compression,	COMPRESSION)
2151 EXT4_FEATURE_INCOMPAT_FUNCS(filetype,		FILETYPE)
2152 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery,	RECOVER)
2153 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev,	JOURNAL_DEV)
2154 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg,		META_BG)
2155 EXT4_FEATURE_INCOMPAT_FUNCS(extents,		EXTENTS)
2156 EXT4_FEATURE_INCOMPAT_FUNCS(64bit,		64BIT)
2157 EXT4_FEATURE_INCOMPAT_FUNCS(mmp,		MMP)
2158 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg,		FLEX_BG)
2159 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode,		EA_INODE)
2160 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata,		DIRDATA)
2161 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed,		CSUM_SEED)
2162 EXT4_FEATURE_INCOMPAT_FUNCS(largedir,		LARGEDIR)
2163 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data,	INLINE_DATA)
2164 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt,		ENCRYPT)
2165 EXT4_FEATURE_INCOMPAT_FUNCS(casefold,		CASEFOLD)
2166 
2167 #define EXT2_FEATURE_COMPAT_SUPP	EXT4_FEATURE_COMPAT_EXT_ATTR
2168 #define EXT2_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
2169 					 EXT4_FEATURE_INCOMPAT_META_BG)
2170 #define EXT2_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2171 					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2172 					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
2173 
2174 #define EXT3_FEATURE_COMPAT_SUPP	EXT4_FEATURE_COMPAT_EXT_ATTR
2175 #define EXT3_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
2176 					 EXT4_FEATURE_INCOMPAT_RECOVER| \
2177 					 EXT4_FEATURE_INCOMPAT_META_BG)
2178 #define EXT3_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2179 					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2180 					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
2181 
2182 #define EXT4_FEATURE_COMPAT_SUPP	(EXT4_FEATURE_COMPAT_EXT_ATTR| \
2183 					 EXT4_FEATURE_COMPAT_ORPHAN_FILE)
2184 #define EXT4_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
2185 					 EXT4_FEATURE_INCOMPAT_RECOVER| \
2186 					 EXT4_FEATURE_INCOMPAT_META_BG| \
2187 					 EXT4_FEATURE_INCOMPAT_EXTENTS| \
2188 					 EXT4_FEATURE_INCOMPAT_64BIT| \
2189 					 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
2190 					 EXT4_FEATURE_INCOMPAT_EA_INODE| \
2191 					 EXT4_FEATURE_INCOMPAT_MMP | \
2192 					 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
2193 					 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
2194 					 EXT4_FEATURE_INCOMPAT_CASEFOLD | \
2195 					 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \
2196 					 EXT4_FEATURE_INCOMPAT_LARGEDIR)
2197 #define EXT4_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2198 					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2199 					 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
2200 					 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
2201 					 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
2202 					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
2203 					 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
2204 					 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
2205 					 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
2206 					 EXT4_FEATURE_RO_COMPAT_QUOTA |\
2207 					 EXT4_FEATURE_RO_COMPAT_PROJECT |\
2208 					 EXT4_FEATURE_RO_COMPAT_VERITY |\
2209 					 EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT)
2210 
2211 #define EXTN_FEATURE_FUNCS(ver) \
2212 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
2213 { \
2214 	return ((EXT4_SB(sb)->s_es->s_feature_compat & \
2215 		cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
2216 } \
2217 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
2218 { \
2219 	return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
2220 		cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
2221 } \
2222 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
2223 { \
2224 	return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
2225 		cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
2226 }
2227 
2228 EXTN_FEATURE_FUNCS(2)
2229 EXTN_FEATURE_FUNCS(3)
2230 EXTN_FEATURE_FUNCS(4)
2231 
2232 static inline bool ext4_has_compat_features(struct super_block *sb)
2233 {
2234 	return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
2235 }
ext4_has_ro_compat_features(struct super_block * sb)2236 static inline bool ext4_has_ro_compat_features(struct super_block *sb)
2237 {
2238 	return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
2239 }
ext4_has_incompat_features(struct super_block * sb)2240 static inline bool ext4_has_incompat_features(struct super_block *sb)
2241 {
2242 	return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
2243 }
2244 
2245 extern int ext4_feature_set_ok(struct super_block *sb, int readonly);
2246 
2247 /*
2248  * Superblock flags
2249  */
2250 enum {
2251 	EXT4_FLAGS_RESIZING,	/* Avoid superblock update and resize race */
2252 	EXT4_FLAGS_SHUTDOWN,	/* Prevent access to the file system */
2253 	EXT4_FLAGS_BDEV_IS_DAX,	/* Current block device support DAX */
2254 	EXT4_FLAGS_EMERGENCY_RO,/* Emergency read-only due to fs errors */
2255 };
2256 
ext4_forced_shutdown(struct super_block * sb)2257 static inline int ext4_forced_shutdown(struct super_block *sb)
2258 {
2259 	return test_bit(EXT4_FLAGS_SHUTDOWN, &EXT4_SB(sb)->s_ext4_flags);
2260 }
2261 
ext4_emergency_ro(struct super_block * sb)2262 static inline int ext4_emergency_ro(struct super_block *sb)
2263 {
2264 	return test_bit(EXT4_FLAGS_EMERGENCY_RO, &EXT4_SB(sb)->s_ext4_flags);
2265 }
2266 
ext4_emergency_state(struct super_block * sb)2267 static inline int ext4_emergency_state(struct super_block *sb)
2268 {
2269 	if (unlikely(ext4_forced_shutdown(sb)))
2270 		return -EIO;
2271 	if (unlikely(ext4_emergency_ro(sb)))
2272 		return -EROFS;
2273 	return 0;
2274 }
2275 
2276 /*
2277  * Default values for user and/or group using reserved blocks
2278  */
2279 #define	EXT4_DEF_RESUID		0
2280 #define	EXT4_DEF_RESGID		0
2281 
2282 /*
2283  * Default project ID
2284  */
2285 #define	EXT4_DEF_PROJID		0
2286 
2287 #define EXT4_DEF_INODE_READAHEAD_BLKS	32
2288 
2289 /*
2290  * Default mount options
2291  */
2292 #define EXT4_DEFM_DEBUG		0x0001
2293 #define EXT4_DEFM_BSDGROUPS	0x0002
2294 #define EXT4_DEFM_XATTR_USER	0x0004
2295 #define EXT4_DEFM_ACL		0x0008
2296 #define EXT4_DEFM_UID16		0x0010
2297 #define EXT4_DEFM_JMODE		0x0060
2298 #define EXT4_DEFM_JMODE_DATA	0x0020
2299 #define EXT4_DEFM_JMODE_ORDERED	0x0040
2300 #define EXT4_DEFM_JMODE_WBACK	0x0060
2301 #define EXT4_DEFM_NOBARRIER	0x0100
2302 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
2303 #define EXT4_DEFM_DISCARD	0x0400
2304 #define EXT4_DEFM_NODELALLOC	0x0800
2305 
2306 /*
2307  * Default journal batch times and ioprio.
2308  */
2309 #define EXT4_DEF_MIN_BATCH_TIME	0
2310 #define EXT4_DEF_MAX_BATCH_TIME	15000 /* 15ms */
2311 #define EXT4_DEF_JOURNAL_IOPRIO (IOPRIO_PRIO_VALUE(IOPRIO_CLASS_BE, 3))
2312 
2313 
2314 /*
2315  * Default values for superblock update
2316  */
2317 #define EXT4_DEF_SB_UPDATE_INTERVAL_SEC (3600) /* seconds (1 hour) */
2318 #define EXT4_DEF_SB_UPDATE_INTERVAL_KB (16384) /* kilobytes (16MB) */
2319 
2320 
2321 /*
2322  * Minimum number of groups in a flexgroup before we separate out
2323  * directories into the first block group of a flexgroup
2324  */
2325 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME	4
2326 
2327 /*
2328  * Structure of a directory entry
2329  */
2330 #define EXT4_NAME_LEN 255
2331 /*
2332  * Base length of the ext4 directory entry excluding the name length
2333  */
2334 #define EXT4_BASE_DIR_LEN (sizeof(struct ext4_dir_entry_2) - EXT4_NAME_LEN)
2335 
2336 struct ext4_dir_entry {
2337 	__le32	inode;			/* Inode number */
2338 	__le16	rec_len;		/* Directory entry length */
2339 	__le16	name_len;		/* Name length */
2340 	char	name[EXT4_NAME_LEN];	/* File name */
2341 };
2342 
2343 
2344 /*
2345  * Encrypted Casefolded entries require saving the hash on disk. This structure
2346  * followed ext4_dir_entry_2's name[name_len] at the next 4 byte aligned
2347  * boundary.
2348  */
2349 struct ext4_dir_entry_hash {
2350 	__le32 hash;
2351 	__le32 minor_hash;
2352 };
2353 
2354 /*
2355  * The new version of the directory entry.  Since EXT4 structures are
2356  * stored in intel byte order, and the name_len field could never be
2357  * bigger than 255 chars, it's safe to reclaim the extra byte for the
2358  * file_type field.
2359  */
2360 struct ext4_dir_entry_2 {
2361 	__le32	inode;			/* Inode number */
2362 	__le16	rec_len;		/* Directory entry length */
2363 	__u8	name_len;		/* Name length */
2364 	__u8	file_type;		/* See file type macros EXT4_FT_* below */
2365 	char	name[EXT4_NAME_LEN];	/* File name */
2366 };
2367 
2368 /*
2369  * Access the hashes at the end of ext4_dir_entry_2
2370  */
2371 #define EXT4_DIRENT_HASHES(entry) \
2372 	((struct ext4_dir_entry_hash *) \
2373 		(((void *)(entry)) + \
2374 		((8 + (entry)->name_len + EXT4_DIR_ROUND) & ~EXT4_DIR_ROUND)))
2375 #define EXT4_DIRENT_HASH(entry) le32_to_cpu(EXT4_DIRENT_HASHES(entry)->hash)
2376 #define EXT4_DIRENT_MINOR_HASH(entry) \
2377 		le32_to_cpu(EXT4_DIRENT_HASHES(entry)->minor_hash)
2378 
ext4_hash_in_dirent(const struct inode * inode)2379 static inline bool ext4_hash_in_dirent(const struct inode *inode)
2380 {
2381 	return IS_CASEFOLDED(inode) && IS_ENCRYPTED(inode);
2382 }
2383 
2384 /*
2385  * This is a bogus directory entry at the end of each leaf block that
2386  * records checksums.
2387  */
2388 struct ext4_dir_entry_tail {
2389 	__le32	det_reserved_zero1;	/* Pretend to be unused */
2390 	__le16	det_rec_len;		/* 12 */
2391 	__u8	det_reserved_zero2;	/* Zero name length */
2392 	__u8	det_reserved_ft;	/* 0xDE, fake file type */
2393 	__le32	det_checksum;		/* crc32c(uuid+inum+dirblock) */
2394 };
2395 
2396 #define EXT4_DIRENT_TAIL(block, blocksize) \
2397 	((struct ext4_dir_entry_tail *)(((void *)(block)) + \
2398 					((blocksize) - \
2399 					 sizeof(struct ext4_dir_entry_tail))))
2400 
2401 /*
2402  * Ext4 directory file types.  Only the low 3 bits are used.  The
2403  * other bits are reserved for now.
2404  */
2405 #define EXT4_FT_UNKNOWN		0
2406 #define EXT4_FT_REG_FILE	1
2407 #define EXT4_FT_DIR		2
2408 #define EXT4_FT_CHRDEV		3
2409 #define EXT4_FT_BLKDEV		4
2410 #define EXT4_FT_FIFO		5
2411 #define EXT4_FT_SOCK		6
2412 #define EXT4_FT_SYMLINK		7
2413 
2414 #define EXT4_FT_MAX		8
2415 
2416 #define EXT4_FT_DIR_CSUM	0xDE
2417 
2418 /*
2419  * EXT4_DIR_PAD defines the directory entries boundaries
2420  *
2421  * NOTE: It must be a multiple of 4
2422  */
2423 #define EXT4_DIR_PAD			4
2424 #define EXT4_DIR_ROUND			(EXT4_DIR_PAD - 1)
2425 #define EXT4_MAX_REC_LEN		((1<<16)-1)
2426 
2427 /*
2428  * The rec_len is dependent on the type of directory. Directories that are
2429  * casefolded and encrypted need to store the hash as well, so we add room for
2430  * ext4_extended_dir_entry_2. For all entries related to '.' or '..' you should
2431  * pass NULL for dir, as those entries do not use the extra fields.
2432  */
ext4_dir_rec_len(__u8 name_len,const struct inode * dir)2433 static inline unsigned int ext4_dir_rec_len(__u8 name_len,
2434 						const struct inode *dir)
2435 {
2436 	int rec_len = (name_len + 8 + EXT4_DIR_ROUND);
2437 
2438 	if (dir && ext4_hash_in_dirent(dir))
2439 		rec_len += sizeof(struct ext4_dir_entry_hash);
2440 	return (rec_len & ~EXT4_DIR_ROUND);
2441 }
2442 
2443 /*
2444  * If we ever get support for fs block sizes > page_size, we'll need
2445  * to remove the #if statements in the next two functions...
2446  */
2447 static inline unsigned int
ext4_rec_len_from_disk(__le16 dlen,unsigned blocksize)2448 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
2449 {
2450 	unsigned len = le16_to_cpu(dlen);
2451 
2452 #if (PAGE_SIZE >= 65536)
2453 	if (len == EXT4_MAX_REC_LEN || len == 0)
2454 		return blocksize;
2455 	return (len & 65532) | ((len & 3) << 16);
2456 #else
2457 	return len;
2458 #endif
2459 }
2460 
ext4_rec_len_to_disk(unsigned len,unsigned blocksize)2461 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
2462 {
2463 	BUG_ON((len > blocksize) || (blocksize > (1 << 18)) || (len & 3));
2464 #if (PAGE_SIZE >= 65536)
2465 	if (len < 65536)
2466 		return cpu_to_le16(len);
2467 	if (len == blocksize) {
2468 		if (blocksize == 65536)
2469 			return cpu_to_le16(EXT4_MAX_REC_LEN);
2470 		else
2471 			return cpu_to_le16(0);
2472 	}
2473 	return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
2474 #else
2475 	return cpu_to_le16(len);
2476 #endif
2477 }
2478 
2479 /*
2480  * Hash Tree Directory indexing
2481  * (c) Daniel Phillips, 2001
2482  */
2483 
2484 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
2485 		    ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2486 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \
2487 		    !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir)))
2488 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2489 
2490 /* Legal values for the dx_root hash_version field: */
2491 
2492 #define DX_HASH_LEGACY			0
2493 #define DX_HASH_HALF_MD4		1
2494 #define DX_HASH_TEA			2
2495 #define DX_HASH_LEGACY_UNSIGNED		3
2496 #define DX_HASH_HALF_MD4_UNSIGNED	4
2497 #define DX_HASH_TEA_UNSIGNED		5
2498 #define DX_HASH_SIPHASH			6
2499 #define DX_HASH_LAST 			DX_HASH_SIPHASH
2500 
ext4_chksum(u32 crc,const void * address,unsigned int length)2501 static inline u32 ext4_chksum(u32 crc, const void *address, unsigned int length)
2502 {
2503 	return crc32c(crc, address, length);
2504 }
2505 
2506 #ifdef __KERNEL__
2507 
2508 /* hash info structure used by the directory hash */
2509 struct dx_hash_info
2510 {
2511 	u32		hash;
2512 	u32		minor_hash;
2513 	int		hash_version;
2514 	u32		*seed;
2515 };
2516 
2517 
2518 /* 32 and 64 bit signed EOF for dx directories */
2519 #define EXT4_HTREE_EOF_32BIT   ((1UL  << (32 - 1)) - 1)
2520 #define EXT4_HTREE_EOF_64BIT   ((1ULL << (64 - 1)) - 1)
2521 
2522 
2523 /*
2524  * Control parameters used by ext4_htree_next_block
2525  */
2526 #define HASH_NB_ALWAYS		1
2527 
2528 struct ext4_filename {
2529 	const struct qstr *usr_fname;
2530 	struct fscrypt_str disk_name;
2531 	struct dx_hash_info hinfo;
2532 #ifdef CONFIG_FS_ENCRYPTION
2533 	struct fscrypt_str crypto_buf;
2534 #endif
2535 #if IS_ENABLED(CONFIG_UNICODE)
2536 	struct qstr cf_name;
2537 #endif
2538 };
2539 
2540 #define fname_name(p) ((p)->disk_name.name)
2541 #define fname_usr_name(p) ((p)->usr_fname->name)
2542 #define fname_len(p)  ((p)->disk_name.len)
2543 
2544 /*
2545  * Describe an inode's exact location on disk and in memory
2546  */
2547 struct ext4_iloc
2548 {
2549 	struct buffer_head *bh;
2550 	unsigned long offset;
2551 	ext4_group_t block_group;
2552 };
2553 
ext4_raw_inode(struct ext4_iloc * iloc)2554 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2555 {
2556 	return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2557 }
2558 
ext4_is_quota_file(struct inode * inode)2559 static inline bool ext4_is_quota_file(struct inode *inode)
2560 {
2561 	return IS_NOQUOTA(inode) &&
2562 	       !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL);
2563 }
2564 
2565 /*
2566  * This structure is stuffed into the struct file's private_data field
2567  * for directories.  It is where we put information so that we can do
2568  * readdir operations in hash tree order.
2569  */
2570 struct dir_private_info {
2571 	struct rb_root	root;
2572 	struct rb_node	*curr_node;
2573 	struct fname	*extra_fname;
2574 	loff_t		last_pos;
2575 	__u32		curr_hash;
2576 	__u32		curr_minor_hash;
2577 	__u32		next_hash;
2578 	u64		cookie;
2579 	bool		initialized;
2580 };
2581 
2582 /* calculate the first block number of the group */
2583 static inline ext4_fsblk_t
ext4_group_first_block_no(struct super_block * sb,ext4_group_t group_no)2584 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2585 {
2586 	return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2587 		le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2588 }
2589 
2590 /*
2591  * Special error return code only used by dx_probe() and its callers.
2592  */
2593 #define ERR_BAD_DX_DIR	(-(MAX_ERRNO - 1))
2594 
2595 /* htree levels for ext4 */
2596 #define	EXT4_HTREE_LEVEL_COMPAT	2
2597 #define	EXT4_HTREE_LEVEL	3
2598 
ext4_dir_htree_level(struct super_block * sb)2599 static inline int ext4_dir_htree_level(struct super_block *sb)
2600 {
2601 	return ext4_has_feature_largedir(sb) ?
2602 		EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT;
2603 }
2604 
2605 /*
2606  * Timeout and state flag for lazy initialization inode thread.
2607  */
2608 #define EXT4_DEF_LI_WAIT_MULT			10
2609 #define EXT4_DEF_LI_MAX_START_DELAY		5
2610 #define EXT4_LAZYINIT_QUIT			0x0001
2611 #define EXT4_LAZYINIT_RUNNING			0x0002
2612 
2613 /*
2614  * Lazy inode table initialization info
2615  */
2616 struct ext4_lazy_init {
2617 	unsigned long		li_state;
2618 	struct list_head	li_request_list;
2619 	struct mutex		li_list_mtx;
2620 };
2621 
2622 enum ext4_li_mode {
2623 	EXT4_LI_MODE_PREFETCH_BBITMAP,
2624 	EXT4_LI_MODE_ITABLE,
2625 };
2626 
2627 struct ext4_li_request {
2628 	struct super_block	*lr_super;
2629 	enum ext4_li_mode	lr_mode;
2630 	ext4_group_t		lr_first_not_zeroed;
2631 	ext4_group_t		lr_next_group;
2632 	struct list_head	lr_request;
2633 	unsigned long		lr_next_sched;
2634 	unsigned long		lr_timeout;
2635 };
2636 
2637 struct ext4_features {
2638 	struct kobject f_kobj;
2639 	struct completion f_kobj_unregister;
2640 };
2641 
2642 /*
2643  * This structure will be used for multiple mount protection. It will be
2644  * written into the block number saved in the s_mmp_block field in the
2645  * superblock. Programs that check MMP should assume that if
2646  * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2647  * to use the filesystem, regardless of how old the timestamp is.
2648  */
2649 #define EXT4_MMP_MAGIC     0x004D4D50U /* ASCII for MMP */
2650 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2651 #define EXT4_MMP_SEQ_FSCK  0xE24D4D50U /* mmp_seq value when being fscked */
2652 #define EXT4_MMP_SEQ_MAX   0xE24D4D4FU /* maximum valid mmp_seq value */
2653 
2654 struct mmp_struct {
2655 	__le32	mmp_magic;		/* Magic number for MMP */
2656 	__le32	mmp_seq;		/* Sequence no. updated periodically */
2657 
2658 	/*
2659 	 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2660 	 * purposes and do not affect the correctness of the algorithm
2661 	 */
2662 	__le64	mmp_time;		/* Time last updated */
2663 	char	mmp_nodename[64];	/* Node which last updated MMP block */
2664 	char	mmp_bdevname[32];	/* Bdev which last updated MMP block */
2665 
2666 	/*
2667 	 * mmp_check_interval is used to verify if the MMP block has been
2668 	 * updated on the block device. The value is updated based on the
2669 	 * maximum time to write the MMP block during an update cycle.
2670 	 */
2671 	__le16	mmp_check_interval;
2672 
2673 	__le16	mmp_pad1;
2674 	__le32	mmp_pad2[226];
2675 	__le32	mmp_checksum;		/* crc32c(uuid+mmp_block) */
2676 };
2677 
2678 /* arguments passed to the mmp thread */
2679 struct mmpd_data {
2680 	struct buffer_head *bh; /* bh from initial read_mmp_block() */
2681 	struct super_block *sb;  /* super block of the fs */
2682 };
2683 
2684 /*
2685  * Check interval multiplier
2686  * The MMP block is written every update interval and initially checked every
2687  * update interval x the multiplier (the value is then adapted based on the
2688  * write latency). The reason is that writes can be delayed under load and we
2689  * don't want readers to incorrectly assume that the filesystem is no longer
2690  * in use.
2691  */
2692 #define EXT4_MMP_CHECK_MULT		2UL
2693 
2694 /*
2695  * Minimum interval for MMP checking in seconds.
2696  */
2697 #define EXT4_MMP_MIN_CHECK_INTERVAL	5UL
2698 
2699 /*
2700  * Maximum interval for MMP checking in seconds.
2701  */
2702 #define EXT4_MMP_MAX_CHECK_INTERVAL	300UL
2703 
2704 /*
2705  * Function prototypes
2706  */
2707 
2708 /*
2709  * Ok, these declarations are also in <linux/kernel.h> but none of the
2710  * ext4 source programs needs to include it so they are duplicated here.
2711  */
2712 # define NORET_TYPE	/**/
2713 # define ATTRIB_NORET	__attribute__((noreturn))
2714 # define NORET_AND	noreturn,
2715 
2716 /* bitmap.c */
2717 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2718 void ext4_inode_bitmap_csum_set(struct super_block *sb,
2719 				struct ext4_group_desc *gdp,
2720 				struct buffer_head *bh);
2721 int ext4_inode_bitmap_csum_verify(struct super_block *sb,
2722 				  struct ext4_group_desc *gdp,
2723 				  struct buffer_head *bh);
2724 void ext4_block_bitmap_csum_set(struct super_block *sb,
2725 				struct ext4_group_desc *gdp,
2726 				struct buffer_head *bh);
2727 int ext4_block_bitmap_csum_verify(struct super_block *sb,
2728 				  struct ext4_group_desc *gdp,
2729 				  struct buffer_head *bh);
2730 
2731 /* balloc.c */
2732 extern void ext4_get_group_no_and_offset(struct super_block *sb,
2733 					 ext4_fsblk_t blocknr,
2734 					 ext4_group_t *blockgrpp,
2735 					 ext4_grpblk_t *offsetp);
2736 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2737 					  ext4_fsblk_t block);
2738 
2739 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2740 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2741 			ext4_group_t group);
2742 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2743 					 ext4_fsblk_t goal,
2744 					 unsigned int flags,
2745 					 unsigned long *count,
2746 					 int *errp);
2747 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2748 				    s64 nclusters, unsigned int flags);
2749 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2750 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2751 						    ext4_group_t block_group,
2752 						    struct buffer_head ** bh);
2753 extern struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2754 						   ext4_group_t group);
2755 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2756 
2757 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2758 						ext4_group_t block_group,
2759 						bool ignore_locked);
2760 extern int ext4_wait_block_bitmap(struct super_block *sb,
2761 				  ext4_group_t block_group,
2762 				  struct buffer_head *bh);
2763 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2764 						  ext4_group_t block_group);
2765 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2766 					      ext4_group_t block_group,
2767 					      struct ext4_group_desc *gdp);
2768 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2769 
2770 #if IS_ENABLED(CONFIG_UNICODE)
2771 extern int ext4_fname_setup_ci_filename(struct inode *dir,
2772 					const struct qstr *iname,
2773 					struct ext4_filename *fname);
2774 
ext4_fname_free_ci_filename(struct ext4_filename * fname)2775 static inline void ext4_fname_free_ci_filename(struct ext4_filename *fname)
2776 {
2777 	kfree(fname->cf_name.name);
2778 	fname->cf_name.name = NULL;
2779 }
2780 #else
ext4_fname_setup_ci_filename(struct inode * dir,const struct qstr * iname,struct ext4_filename * fname)2781 static inline int ext4_fname_setup_ci_filename(struct inode *dir,
2782 					       const struct qstr *iname,
2783 					       struct ext4_filename *fname)
2784 {
2785 	return 0;
2786 }
2787 
ext4_fname_free_ci_filename(struct ext4_filename * fname)2788 static inline void ext4_fname_free_ci_filename(struct ext4_filename *fname)
2789 {
2790 }
2791 #endif
2792 
2793 /* ext4 encryption related stuff goes here crypto.c */
2794 #ifdef CONFIG_FS_ENCRYPTION
2795 extern const struct fscrypt_operations ext4_cryptops;
2796 
2797 int ext4_fname_setup_filename(struct inode *dir, const struct qstr *iname,
2798 			      int lookup, struct ext4_filename *fname);
2799 
2800 int ext4_fname_prepare_lookup(struct inode *dir, struct dentry *dentry,
2801 			      struct ext4_filename *fname);
2802 
2803 void ext4_fname_free_filename(struct ext4_filename *fname);
2804 
2805 int ext4_ioctl_get_encryption_pwsalt(struct file *filp, void __user *arg);
2806 
2807 #else /* !CONFIG_FS_ENCRYPTION */
ext4_fname_setup_filename(struct inode * dir,const struct qstr * iname,int lookup,struct ext4_filename * fname)2808 static inline int ext4_fname_setup_filename(struct inode *dir,
2809 					    const struct qstr *iname,
2810 					    int lookup,
2811 					    struct ext4_filename *fname)
2812 {
2813 	fname->usr_fname = iname;
2814 	fname->disk_name.name = (unsigned char *) iname->name;
2815 	fname->disk_name.len = iname->len;
2816 
2817 	return ext4_fname_setup_ci_filename(dir, iname, fname);
2818 }
2819 
ext4_fname_prepare_lookup(struct inode * dir,struct dentry * dentry,struct ext4_filename * fname)2820 static inline int ext4_fname_prepare_lookup(struct inode *dir,
2821 					    struct dentry *dentry,
2822 					    struct ext4_filename *fname)
2823 {
2824 	return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname);
2825 }
2826 
ext4_fname_free_filename(struct ext4_filename * fname)2827 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2828 {
2829 	ext4_fname_free_ci_filename(fname);
2830 }
2831 
ext4_ioctl_get_encryption_pwsalt(struct file * filp,void __user * arg)2832 static inline int ext4_ioctl_get_encryption_pwsalt(struct file *filp,
2833 						   void __user *arg)
2834 {
2835 	return -EOPNOTSUPP;
2836 }
2837 #endif /* !CONFIG_FS_ENCRYPTION */
2838 
2839 /* dir.c */
2840 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2841 				  struct file *,
2842 				  struct ext4_dir_entry_2 *,
2843 				  struct buffer_head *, char *, int,
2844 				  unsigned int);
2845 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2846 	unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2847 				(de), (bh), (buf), (size), (offset)))
2848 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2849 				__u32 minor_hash,
2850 				struct ext4_dir_entry_2 *dirent,
2851 				struct fscrypt_str *ent_name);
2852 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2853 extern int ext4_find_dest_de(struct inode *dir, struct buffer_head *bh,
2854 			     void *buf, int buf_size,
2855 			     struct ext4_filename *fname,
2856 			     struct ext4_dir_entry_2 **dest_de);
2857 void ext4_insert_dentry(struct inode *dir, struct inode *inode,
2858 			struct ext4_dir_entry_2 *de,
2859 			int buf_size,
2860 			struct ext4_filename *fname);
ext4_update_dx_flag(struct inode * inode)2861 static inline void ext4_update_dx_flag(struct inode *inode)
2862 {
2863 	if (!ext4_has_feature_dir_index(inode->i_sb) &&
2864 	    ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) {
2865 		/* ext4_iget() should have caught this... */
2866 		WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb));
2867 		ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2868 	}
2869 }
2870 static const unsigned char ext4_filetype_table[] = {
2871 	DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2872 };
2873 
get_dtype(struct super_block * sb,int filetype)2874 static inline  unsigned char get_dtype(struct super_block *sb, int filetype)
2875 {
2876 	if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2877 		return DT_UNKNOWN;
2878 
2879 	return ext4_filetype_table[filetype];
2880 }
2881 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2882 			     void *buf, int buf_size);
2883 
2884 /* fsync.c */
2885 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2886 
2887 /* hash.c */
2888 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len,
2889 			  struct dx_hash_info *hinfo);
2890 
2891 /* ialloc.c */
2892 extern int ext4_mark_inode_used(struct super_block *sb, int ino);
2893 extern struct inode *__ext4_new_inode(struct mnt_idmap *, handle_t *,
2894 				      struct inode *, umode_t,
2895 				      const struct qstr *qstr, __u32 goal,
2896 				      uid_t *owner, __u32 i_flags,
2897 				      int handle_type, unsigned int line_no,
2898 				      int nblocks);
2899 
2900 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags)          \
2901 	__ext4_new_inode(&nop_mnt_idmap, (handle), (dir), (mode), (qstr),      \
2902 			 (goal), (owner), i_flags, 0, 0, 0)
2903 #define ext4_new_inode_start_handle(idmap, dir, mode, qstr, goal, owner, \
2904 				    type, nblocks)		    \
2905 	__ext4_new_inode((idmap), NULL, (dir), (mode), (qstr), (goal), (owner), \
2906 			 0, (type), __LINE__, (nblocks))
2907 
2908 
2909 extern void ext4_free_inode(handle_t *, struct inode *);
2910 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2911 extern unsigned long ext4_count_free_inodes(struct super_block *);
2912 extern unsigned long ext4_count_dirs(struct super_block *);
2913 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2914 extern int ext4_init_inode_table(struct super_block *sb,
2915 				 ext4_group_t group, int barrier);
2916 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2917 
2918 /* fast_commit.c */
2919 int ext4_fc_info_show(struct seq_file *seq, void *v);
2920 void ext4_fc_init(struct super_block *sb, journal_t *journal);
2921 void ext4_fc_init_inode(struct inode *inode);
2922 void ext4_fc_track_range(handle_t *handle, struct inode *inode, ext4_lblk_t start,
2923 			 ext4_lblk_t end);
2924 void __ext4_fc_track_unlink(handle_t *handle, struct inode *inode,
2925 	struct dentry *dentry);
2926 void __ext4_fc_track_link(handle_t *handle, struct inode *inode,
2927 	struct dentry *dentry);
2928 void ext4_fc_track_unlink(handle_t *handle, struct dentry *dentry);
2929 void ext4_fc_track_link(handle_t *handle, struct dentry *dentry);
2930 void __ext4_fc_track_create(handle_t *handle, struct inode *inode,
2931 			    struct dentry *dentry);
2932 void ext4_fc_track_create(handle_t *handle, struct dentry *dentry);
2933 void ext4_fc_track_inode(handle_t *handle, struct inode *inode);
2934 void ext4_fc_mark_ineligible(struct super_block *sb, int reason, handle_t *handle);
2935 void ext4_fc_del(struct inode *inode);
2936 bool ext4_fc_replay_check_excluded(struct super_block *sb, ext4_fsblk_t block);
2937 void ext4_fc_replay_cleanup(struct super_block *sb);
2938 int ext4_fc_commit(journal_t *journal, tid_t commit_tid);
2939 int __init ext4_fc_init_dentry_cache(void);
2940 void ext4_fc_destroy_dentry_cache(void);
2941 int ext4_fc_record_regions(struct super_block *sb, int ino,
2942 			   ext4_lblk_t lblk, ext4_fsblk_t pblk,
2943 			   int len, int replay);
2944 
2945 /* mballoc.c */
2946 extern const struct seq_operations ext4_mb_seq_groups_ops;
2947 extern const struct seq_operations ext4_mb_seq_structs_summary_ops;
2948 extern int ext4_seq_mb_stats_show(struct seq_file *seq, void *offset);
2949 extern int ext4_mb_init(struct super_block *);
2950 extern void ext4_mb_release(struct super_block *);
2951 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2952 				struct ext4_allocation_request *, int *);
2953 extern void ext4_discard_preallocations(struct inode *);
2954 extern int __init ext4_init_mballoc(void);
2955 extern void ext4_exit_mballoc(void);
2956 extern ext4_group_t ext4_mb_prefetch(struct super_block *sb,
2957 				     ext4_group_t group,
2958 				     unsigned int nr, int *cnt);
2959 extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group,
2960 				  unsigned int nr);
2961 
2962 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2963 			     struct buffer_head *bh, ext4_fsblk_t block,
2964 			     unsigned long count, int flags);
2965 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2966 				   ext4_group_t ngroups);
2967 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2968 		ext4_group_t i, struct ext4_group_desc *desc);
2969 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2970 				ext4_fsblk_t block, unsigned long count);
2971 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2972 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid);
2973 extern void ext4_mb_mark_bb(struct super_block *sb, ext4_fsblk_t block,
2974 			    int len, bool state);
ext4_mb_cr_expensive(enum criteria cr)2975 static inline bool ext4_mb_cr_expensive(enum criteria cr)
2976 {
2977 	return cr >= CR_GOAL_LEN_SLOW;
2978 }
2979 
2980 /* inode.c */
2981 void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw,
2982 			 struct ext4_inode_info *ei);
2983 int ext4_inode_is_fast_symlink(struct inode *inode);
2984 void ext4_check_map_extents_env(struct inode *inode);
2985 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2986 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2987 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count,
2988 		     bool wait, struct buffer_head **bhs);
2989 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2990 			     struct buffer_head *bh_result, int create);
2991 int ext4_get_block(struct inode *inode, sector_t iblock,
2992 		   struct buffer_head *bh_result, int create);
2993 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2994 			   struct buffer_head *bh, int create);
2995 int ext4_walk_page_buffers(handle_t *handle,
2996 			   struct inode *inode,
2997 			   struct buffer_head *head,
2998 			   unsigned from,
2999 			   unsigned to,
3000 			   int *partial,
3001 			   int (*fn)(handle_t *handle, struct inode *inode,
3002 				     struct buffer_head *bh));
3003 int do_journal_get_write_access(handle_t *handle, struct inode *inode,
3004 				struct buffer_head *bh);
3005 void ext4_set_inode_mapping_order(struct inode *inode);
3006 #define FALL_BACK_TO_NONDELALLOC 1
3007 #define CONVERT_INLINE_DATA	 2
3008 
3009 typedef enum {
3010 	EXT4_IGET_NORMAL =	0,
3011 	EXT4_IGET_SPECIAL =	0x0001, /* OK to iget a system inode */
3012 	EXT4_IGET_HANDLE = 	0x0002,	/* Inode # is from a handle */
3013 	EXT4_IGET_BAD =		0x0004, /* Allow to iget a bad inode */
3014 	EXT4_IGET_EA_INODE =	0x0008	/* Inode should contain an EA value */
3015 } ext4_iget_flags;
3016 
3017 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
3018 				 ext4_iget_flags flags, const char *function,
3019 				 unsigned int line);
3020 
3021 #define ext4_iget(sb, ino, flags) \
3022 	__ext4_iget((sb), (ino), (flags), __func__, __LINE__)
3023 
3024 extern int  ext4_write_inode(struct inode *, struct writeback_control *);
3025 extern int  ext4_setattr(struct mnt_idmap *, struct dentry *,
3026 			 struct iattr *);
3027 extern u32  ext4_dio_alignment(struct inode *inode);
3028 extern int  ext4_getattr(struct mnt_idmap *, const struct path *,
3029 			 struct kstat *, u32, unsigned int);
3030 extern void ext4_evict_inode(struct inode *);
3031 extern void ext4_clear_inode(struct inode *);
3032 extern int  ext4_file_getattr(struct mnt_idmap *, const struct path *,
3033 			      struct kstat *, u32, unsigned int);
3034 extern void ext4_dirty_inode(struct inode *, int);
3035 extern int ext4_change_inode_journal_flag(struct inode *, int);
3036 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
3037 extern int ext4_get_fc_inode_loc(struct super_block *sb, unsigned long ino,
3038 			  struct ext4_iloc *iloc);
3039 extern int ext4_inode_attach_jinode(struct inode *inode);
3040 extern int ext4_can_truncate(struct inode *inode);
3041 extern int ext4_truncate(struct inode *);
3042 extern int ext4_break_layouts(struct inode *);
3043 extern int ext4_truncate_page_cache_block_range(struct inode *inode,
3044 						loff_t start, loff_t end);
3045 extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length);
3046 extern void ext4_set_inode_flags(struct inode *, bool init);
3047 extern int ext4_alloc_da_blocks(struct inode *inode);
3048 extern void ext4_set_aops(struct inode *inode);
3049 extern int ext4_normal_submit_inode_data_buffers(struct jbd2_inode *jinode);
3050 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
3051 extern int ext4_chunk_trans_extent(struct inode *inode, int nrblocks);
3052 extern int ext4_meta_trans_blocks(struct inode *inode, int lblocks,
3053 				  int pextents);
3054 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
3055 			     loff_t lstart, loff_t lend);
3056 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf);
3057 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
3058 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
3059 extern void ext4_da_release_space(struct inode *inode, int to_free);
3060 extern void ext4_da_update_reserve_space(struct inode *inode,
3061 					int used, int quota_claim);
3062 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
3063 			      ext4_fsblk_t pblk, ext4_lblk_t len);
3064 
is_special_ino(struct super_block * sb,unsigned long ino)3065 static inline bool is_special_ino(struct super_block *sb, unsigned long ino)
3066 {
3067 	struct ext4_super_block *es = EXT4_SB(sb)->s_es;
3068 
3069 	return (ino < EXT4_FIRST_INO(sb) && ino != EXT4_ROOT_INO) ||
3070 		ino == le32_to_cpu(es->s_usr_quota_inum) ||
3071 		ino == le32_to_cpu(es->s_grp_quota_inum) ||
3072 		ino == le32_to_cpu(es->s_prj_quota_inum) ||
3073 		ino == le32_to_cpu(es->s_orphan_file_inum);
3074 }
3075 
3076 /* indirect.c */
3077 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
3078 				struct ext4_map_blocks *map, int flags);
3079 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
3080 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
3081 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
3082 				 ext4_lblk_t start, ext4_lblk_t end);
3083 
3084 /* ioctl.c */
3085 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
3086 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
3087 int ext4_fileattr_set(struct mnt_idmap *idmap,
3088 		      struct dentry *dentry, struct file_kattr *fa);
3089 int ext4_fileattr_get(struct dentry *dentry, struct file_kattr *fa);
3090 extern void ext4_reset_inode_seed(struct inode *inode);
3091 int ext4_update_overhead(struct super_block *sb, bool force);
3092 int ext4_force_shutdown(struct super_block *sb, u32 flags);
3093 
3094 /* migrate.c */
3095 extern int ext4_ext_migrate(struct inode *);
3096 extern int ext4_ind_migrate(struct inode *inode);
3097 
3098 /* namei.c */
3099 extern int ext4_init_new_dir(handle_t *handle, struct inode *dir,
3100 			     struct inode *inode);
3101 extern int ext4_dirblock_csum_verify(struct inode *inode,
3102 				     struct buffer_head *bh);
3103 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
3104 				__u32 start_minor_hash, __u32 *next_hash);
3105 extern int ext4_search_dir(struct buffer_head *bh,
3106 			   char *search_buf,
3107 			   int buf_size,
3108 			   struct inode *dir,
3109 			   struct ext4_filename *fname,
3110 			   unsigned int offset,
3111 			   struct ext4_dir_entry_2 **res_dir);
3112 extern int ext4_generic_delete_entry(struct inode *dir,
3113 				     struct ext4_dir_entry_2 *de_del,
3114 				     struct buffer_head *bh,
3115 				     void *entry_buf,
3116 				     int buf_size,
3117 				     int csum_size);
3118 extern bool ext4_empty_dir(struct inode *inode);
3119 
3120 /* resize.c */
3121 extern void ext4_kvfree_array_rcu(void *to_free);
3122 extern int ext4_group_add(struct super_block *sb,
3123 				struct ext4_new_group_data *input);
3124 extern int ext4_group_extend(struct super_block *sb,
3125 				struct ext4_super_block *es,
3126 				ext4_fsblk_t n_blocks_count);
3127 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
3128 extern unsigned int ext4_list_backups(struct super_block *sb,
3129 				      unsigned int *three, unsigned int *five,
3130 				      unsigned int *seven);
3131 
3132 /* super.c */
3133 extern struct buffer_head *ext4_sb_bread(struct super_block *sb,
3134 					 sector_t block, blk_opf_t op_flags);
3135 extern struct buffer_head *ext4_sb_bread_unmovable(struct super_block *sb,
3136 						   sector_t block);
3137 extern void ext4_read_bh_nowait(struct buffer_head *bh, blk_opf_t op_flags,
3138 				bh_end_io_t *end_io, bool simu_fail);
3139 extern int ext4_read_bh(struct buffer_head *bh, blk_opf_t op_flags,
3140 			bh_end_io_t *end_io, bool simu_fail);
3141 extern int ext4_read_bh_lock(struct buffer_head *bh, blk_opf_t op_flags, bool wait);
3142 extern void ext4_sb_breadahead_unmovable(struct super_block *sb, sector_t block);
3143 extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
3144 extern int ext4_calculate_overhead(struct super_block *sb);
3145 extern __le32 ext4_superblock_csum(struct ext4_super_block *es);
3146 extern void ext4_superblock_csum_set(struct super_block *sb);
3147 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
3148 				    ext4_group_t ngroup);
3149 extern const char *ext4_decode_error(struct super_block *sb, int errno,
3150 				     char nbuf[16]);
3151 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb,
3152 					     ext4_group_t block_group,
3153 					     unsigned int flags);
3154 extern unsigned int ext4_num_base_meta_blocks(struct super_block *sb,
3155 					      ext4_group_t block_group);
3156 
3157 extern __printf(7, 8)
3158 void __ext4_error(struct super_block *, const char *, unsigned int, bool,
3159 		  int, __u64, const char *, ...);
3160 extern __printf(6, 7)
3161 void __ext4_error_inode(struct inode *, const char *, unsigned int,
3162 			ext4_fsblk_t, int, const char *, ...);
3163 extern __printf(5, 6)
3164 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
3165 		     const char *, ...);
3166 extern void __ext4_std_error(struct super_block *, const char *,
3167 			     unsigned int, int);
3168 extern __printf(4, 5)
3169 void __ext4_warning(struct super_block *, const char *, unsigned int,
3170 		    const char *, ...);
3171 extern __printf(4, 5)
3172 void __ext4_warning_inode(const struct inode *inode, const char *function,
3173 			  unsigned int line, const char *fmt, ...);
3174 extern __printf(3, 4)
3175 void __ext4_msg(struct super_block *, const char *, const char *, ...);
3176 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
3177 			   const char *, unsigned int, const char *);
3178 extern __printf(7, 8)
3179 void __ext4_grp_locked_error(const char *, unsigned int,
3180 			     struct super_block *, ext4_group_t,
3181 			     unsigned long, ext4_fsblk_t,
3182 			     const char *, ...);
3183 
3184 #define EXT4_ERROR_INODE(inode, fmt, a...) \
3185 	ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
3186 
3187 #define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...)			\
3188 	__ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a)
3189 
3190 #define ext4_error_inode_block(inode, block, err, fmt, a...)		\
3191 	__ext4_error_inode((inode), __func__, __LINE__, (block), (err),	\
3192 			   (fmt), ## a)
3193 
3194 #define EXT4_ERROR_FILE(file, block, fmt, a...)				\
3195 	ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
3196 
3197 #define ext4_abort(sb, err, fmt, a...)					\
3198 	__ext4_error((sb), __func__, __LINE__, true, (err), 0, (fmt), ## a)
3199 
3200 #ifdef CONFIG_PRINTK
3201 
3202 #define ext4_error_inode(inode, func, line, block, fmt, ...)		\
3203 	__ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__)
3204 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...)	\
3205 	__ext4_error_inode((inode), (func), (line), (block), 		\
3206 			   (err), (fmt), ##__VA_ARGS__)
3207 #define ext4_error_file(file, func, line, block, fmt, ...)		\
3208 	__ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
3209 #define ext4_error(sb, fmt, ...)					\
3210 	__ext4_error((sb), __func__, __LINE__, false, 0, 0, (fmt),	\
3211 		##__VA_ARGS__)
3212 #define ext4_error_err(sb, err, fmt, ...)				\
3213 	__ext4_error((sb), __func__, __LINE__, false, (err), 0, (fmt),	\
3214 		##__VA_ARGS__)
3215 #define ext4_warning(sb, fmt, ...)					\
3216 	__ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
3217 #define ext4_warning_inode(inode, fmt, ...)				\
3218 	__ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
3219 #define ext4_msg(sb, level, fmt, ...)				\
3220 	__ext4_msg(sb, level, fmt, ##__VA_ARGS__)
3221 #define dump_mmp_msg(sb, mmp, msg)					\
3222 	__dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
3223 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...)		\
3224 	__ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
3225 				fmt, ##__VA_ARGS__)
3226 
3227 #else
3228 
3229 #define ext4_error_inode(inode, func, line, block, fmt, ...)		\
3230 do {									\
3231 	no_printk(fmt, ##__VA_ARGS__);					\
3232 	__ext4_error_inode(inode, "", 0, block, 0, " ");		\
3233 } while (0)
3234 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...)	\
3235 do {									\
3236 	no_printk(fmt, ##__VA_ARGS__);					\
3237 	__ext4_error_inode(inode, "", 0, block, err, " ");		\
3238 } while (0)
3239 #define ext4_error_file(file, func, line, block, fmt, ...)		\
3240 do {									\
3241 	no_printk(fmt, ##__VA_ARGS__);					\
3242 	__ext4_error_file(file, "", 0, block, " ");			\
3243 } while (0)
3244 #define ext4_error(sb, fmt, ...)					\
3245 do {									\
3246 	no_printk(fmt, ##__VA_ARGS__);					\
3247 	__ext4_error(sb, "", 0, false, 0, 0, " ");			\
3248 } while (0)
3249 #define ext4_error_err(sb, err, fmt, ...)				\
3250 do {									\
3251 	no_printk(fmt, ##__VA_ARGS__);					\
3252 	__ext4_error(sb, "", 0, false, err, 0, " ");			\
3253 } while (0)
3254 #define ext4_warning(sb, fmt, ...)					\
3255 do {									\
3256 	no_printk(fmt, ##__VA_ARGS__);					\
3257 	__ext4_warning(sb, "", 0, " ");					\
3258 } while (0)
3259 #define ext4_warning_inode(inode, fmt, ...)				\
3260 do {									\
3261 	no_printk(fmt, ##__VA_ARGS__);					\
3262 	__ext4_warning_inode(inode, "", 0, " ");			\
3263 } while (0)
3264 #define ext4_msg(sb, level, fmt, ...)					\
3265 do {									\
3266 	no_printk(fmt, ##__VA_ARGS__);					\
3267 	__ext4_msg(sb, "", " ");					\
3268 } while (0)
3269 #define dump_mmp_msg(sb, mmp, msg)					\
3270 	__dump_mmp_msg(sb, mmp, "", 0, "")
3271 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...)		\
3272 do {									\
3273 	no_printk(fmt, ##__VA_ARGS__);				\
3274 	__ext4_grp_locked_error("", 0, sb, grp, ino, block, " ");	\
3275 } while (0)
3276 
3277 #endif
3278 
3279 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
3280 				      struct ext4_group_desc *bg);
3281 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
3282 				      struct ext4_group_desc *bg);
3283 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
3284 				     struct ext4_group_desc *bg);
3285 extern __u32 ext4_free_group_clusters(struct super_block *sb,
3286 				      struct ext4_group_desc *bg);
3287 extern __u32 ext4_free_inodes_count(struct super_block *sb,
3288 				 struct ext4_group_desc *bg);
3289 extern __u32 ext4_used_dirs_count(struct super_block *sb,
3290 				struct ext4_group_desc *bg);
3291 extern __u32 ext4_itable_unused_count(struct super_block *sb,
3292 				   struct ext4_group_desc *bg);
3293 extern void ext4_block_bitmap_set(struct super_block *sb,
3294 				  struct ext4_group_desc *bg, ext4_fsblk_t blk);
3295 extern void ext4_inode_bitmap_set(struct super_block *sb,
3296 				  struct ext4_group_desc *bg, ext4_fsblk_t blk);
3297 extern void ext4_inode_table_set(struct super_block *sb,
3298 				 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3299 extern void ext4_free_group_clusters_set(struct super_block *sb,
3300 					 struct ext4_group_desc *bg,
3301 					 __u32 count);
3302 extern void ext4_free_inodes_set(struct super_block *sb,
3303 				struct ext4_group_desc *bg, __u32 count);
3304 extern void ext4_used_dirs_set(struct super_block *sb,
3305 				struct ext4_group_desc *bg, __u32 count);
3306 extern void ext4_itable_unused_set(struct super_block *sb,
3307 				   struct ext4_group_desc *bg, __u32 count);
3308 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
3309 				       struct ext4_group_desc *gdp);
3310 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
3311 				     struct ext4_group_desc *gdp);
3312 extern int ext4_register_li_request(struct super_block *sb,
3313 				    ext4_group_t first_not_zeroed);
3314 
ext4_has_group_desc_csum(struct super_block * sb)3315 static inline int ext4_has_group_desc_csum(struct super_block *sb)
3316 {
3317 	return ext4_has_feature_gdt_csum(sb) ||
3318 	       ext4_has_feature_metadata_csum(sb);
3319 }
3320 
3321 #define ext4_read_incompat_64bit_val(es, name) \
3322 	(((es)->s_feature_incompat & cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT) \
3323 		? (ext4_fsblk_t)le32_to_cpu(es->name##_hi) << 32 : 0) | \
3324 		le32_to_cpu(es->name##_lo))
3325 
ext4_blocks_count(struct ext4_super_block * es)3326 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
3327 {
3328 	return ext4_read_incompat_64bit_val(es, s_blocks_count);
3329 }
3330 
ext4_r_blocks_count(struct ext4_super_block * es)3331 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
3332 {
3333 	return ext4_read_incompat_64bit_val(es, s_r_blocks_count);
3334 }
3335 
ext4_free_blocks_count(struct ext4_super_block * es)3336 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
3337 {
3338 	return ext4_read_incompat_64bit_val(es, s_free_blocks_count);
3339 }
3340 
ext4_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)3341 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
3342 					 ext4_fsblk_t blk)
3343 {
3344 	es->s_blocks_count_lo = cpu_to_le32((u32)blk);
3345 	es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
3346 }
3347 
ext4_free_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)3348 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
3349 					      ext4_fsblk_t blk)
3350 {
3351 	es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
3352 	es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
3353 }
3354 
ext4_r_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)3355 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
3356 					   ext4_fsblk_t blk)
3357 {
3358 	es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
3359 	es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
3360 }
3361 
ext4_isize(struct super_block * sb,struct ext4_inode * raw_inode)3362 static inline loff_t ext4_isize(struct super_block *sb,
3363 				struct ext4_inode *raw_inode)
3364 {
3365 	if (ext4_has_feature_largedir(sb) ||
3366 	    S_ISREG(le16_to_cpu(raw_inode->i_mode)))
3367 		return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
3368 			le32_to_cpu(raw_inode->i_size_lo);
3369 
3370 	return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
3371 }
3372 
ext4_isize_set(struct ext4_inode * raw_inode,loff_t i_size)3373 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
3374 {
3375 	raw_inode->i_size_lo = cpu_to_le32(i_size);
3376 	raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
3377 }
3378 
3379 /*
3380  * Reading s_groups_count requires using smp_rmb() afterwards.  See
3381  * the locking protocol documented in the comments of ext4_group_add()
3382  * in resize.c
3383  */
ext4_get_groups_count(struct super_block * sb)3384 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
3385 {
3386 	ext4_group_t	ngroups = EXT4_SB(sb)->s_groups_count;
3387 
3388 	smp_rmb();
3389 	return ngroups;
3390 }
3391 
ext4_flex_group(struct ext4_sb_info * sbi,ext4_group_t block_group)3392 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
3393 					     ext4_group_t block_group)
3394 {
3395 	return block_group >> sbi->s_log_groups_per_flex;
3396 }
3397 
ext4_flex_bg_size(struct ext4_sb_info * sbi)3398 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
3399 {
3400 	return 1 << sbi->s_log_groups_per_flex;
3401 }
3402 
ext4_get_maxbytes(struct inode * inode)3403 static inline loff_t ext4_get_maxbytes(struct inode *inode)
3404 {
3405 	if (ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS))
3406 		return inode->i_sb->s_maxbytes;
3407 	return EXT4_SB(inode->i_sb)->s_bitmap_maxbytes;
3408 }
3409 
3410 #define ext4_std_error(sb, errno)				\
3411 do {								\
3412 	if ((errno))						\
3413 		__ext4_std_error((sb), __func__, __LINE__, (errno));	\
3414 } while (0)
3415 
3416 #ifdef CONFIG_SMP
3417 /* Each CPU can accumulate percpu_counter_batch clusters in their local
3418  * counters. So we need to make sure we have free clusters more
3419  * than percpu_counter_batch  * nr_cpu_ids. Also add a window of 4 times.
3420  */
3421 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
3422 #else
3423 #define EXT4_FREECLUSTERS_WATERMARK 0
3424 #endif
3425 
3426 /* Update i_disksize. Requires i_rwsem to avoid races with truncate */
ext4_update_i_disksize(struct inode * inode,loff_t newsize)3427 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
3428 {
3429 	WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
3430 		     !inode_is_locked(inode));
3431 	down_write(&EXT4_I(inode)->i_data_sem);
3432 	if (newsize > EXT4_I(inode)->i_disksize)
3433 		WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize);
3434 	up_write(&EXT4_I(inode)->i_data_sem);
3435 }
3436 
3437 /* Update i_size, i_disksize. Requires i_rwsem to avoid races with truncate */
ext4_update_inode_size(struct inode * inode,loff_t newsize)3438 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
3439 {
3440 	int changed = 0;
3441 
3442 	if (newsize > inode->i_size) {
3443 		i_size_write(inode, newsize);
3444 		changed = 1;
3445 	}
3446 	if (newsize > EXT4_I(inode)->i_disksize) {
3447 		ext4_update_i_disksize(inode, newsize);
3448 		changed |= 2;
3449 	}
3450 	return changed;
3451 }
3452 
3453 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
3454 				      loff_t len);
3455 
3456 struct ext4_group_info {
3457 	unsigned long   bb_state;
3458 #ifdef AGGRESSIVE_CHECK
3459 	unsigned long	bb_check_counter;
3460 #endif
3461 	struct rb_root  bb_free_root;
3462 	ext4_grpblk_t	bb_first_free;	/* first free block */
3463 	ext4_grpblk_t	bb_free;	/* total free blocks */
3464 	ext4_grpblk_t	bb_fragments;	/* nr of freespace fragments */
3465 	int		bb_avg_fragment_size_order;	/* order of average
3466 							   fragment in BG */
3467 	ext4_grpblk_t	bb_largest_free_order;/* order of largest frag in BG */
3468 	ext4_group_t	bb_group;	/* Group number */
3469 	struct          list_head bb_prealloc_list;
3470 #ifdef DOUBLE_CHECK
3471 	void            *bb_bitmap;
3472 #endif
3473 	struct rw_semaphore alloc_sem;
3474 	ext4_grpblk_t	bb_counters[];	/* Nr of free power-of-two-block
3475 					 * regions, index is order.
3476 					 * bb_counters[3] = 5 means
3477 					 * 5 free 8-block regions. */
3478 };
3479 
3480 #define EXT4_GROUP_INFO_NEED_INIT_BIT		0
3481 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT		1
3482 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT	2
3483 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT	3
3484 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT		\
3485 	(1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT)
3486 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT		\
3487 	(1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT)
3488 #define EXT4_GROUP_INFO_BBITMAP_READ_BIT	4
3489 
3490 #define EXT4_MB_GRP_NEED_INIT(grp)	\
3491 	(test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
3492 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp)	\
3493 	(test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3494 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp)	\
3495 	(test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3496 
3497 #define EXT4_MB_GRP_WAS_TRIMMED(grp)	\
3498 	(test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3499 #define EXT4_MB_GRP_SET_TRIMMED(grp)	\
3500 	(set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3501 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp)	\
3502 	(clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3503 #define EXT4_MB_GRP_TEST_AND_SET_READ(grp)	\
3504 	(test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state)))
3505 
3506 #define EXT4_MAX_CONTENTION		8
3507 #define EXT4_CONTENTION_THRESHOLD	2
3508 
ext4_group_lock_ptr(struct super_block * sb,ext4_group_t group)3509 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
3510 					      ext4_group_t group)
3511 {
3512 	return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
3513 }
3514 
3515 /*
3516  * Returns true if the filesystem is busy enough that attempts to
3517  * access the block group locks has run into contention.
3518  */
ext4_fs_is_busy(struct ext4_sb_info * sbi)3519 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
3520 {
3521 	return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
3522 }
3523 
ext4_try_lock_group(struct super_block * sb,ext4_group_t group)3524 static inline bool ext4_try_lock_group(struct super_block *sb, ext4_group_t group)
3525 {
3526 	if (!spin_trylock(ext4_group_lock_ptr(sb, group)))
3527 		return false;
3528 	/*
3529 	 * We're able to grab the lock right away, so drop the lock
3530 	 * contention counter.
3531 	 */
3532 	atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
3533 	return true;
3534 }
3535 
ext4_lock_group(struct super_block * sb,ext4_group_t group)3536 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
3537 {
3538 	if (!ext4_try_lock_group(sb, group)) {
3539 		/*
3540 		 * The lock is busy, so bump the contention counter,
3541 		 * and then wait on the spin lock.
3542 		 */
3543 		atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
3544 				  EXT4_MAX_CONTENTION);
3545 		spin_lock(ext4_group_lock_ptr(sb, group));
3546 	}
3547 }
3548 
ext4_unlock_group(struct super_block * sb,ext4_group_t group)3549 static inline void ext4_unlock_group(struct super_block *sb,
3550 					ext4_group_t group)
3551 {
3552 	spin_unlock(ext4_group_lock_ptr(sb, group));
3553 }
3554 
3555 #ifdef CONFIG_QUOTA
ext4_quota_capable(struct super_block * sb)3556 static inline bool ext4_quota_capable(struct super_block *sb)
3557 {
3558 	return (test_opt(sb, QUOTA) || ext4_has_feature_quota(sb));
3559 }
3560 
ext4_is_quota_journalled(struct super_block * sb)3561 static inline bool ext4_is_quota_journalled(struct super_block *sb)
3562 {
3563 	struct ext4_sb_info *sbi = EXT4_SB(sb);
3564 
3565 	return (ext4_has_feature_quota(sb) ||
3566 		sbi->s_qf_names[USRQUOTA] || sbi->s_qf_names[GRPQUOTA]);
3567 }
3568 int ext4_enable_quotas(struct super_block *sb);
3569 #endif
3570 
3571 /*
3572  * Block validity checking
3573  */
3574 #define ext4_check_indirect_blockref(inode, bh)				\
3575 	ext4_check_blockref(__func__, __LINE__, inode,			\
3576 			    (__le32 *)(bh)->b_data,			\
3577 			    EXT4_ADDR_PER_BLOCK((inode)->i_sb))
3578 
3579 #define ext4_ind_check_inode(inode)					\
3580 	ext4_check_blockref(__func__, __LINE__, inode,			\
3581 			    EXT4_I(inode)->i_data,			\
3582 			    EXT4_NDIR_BLOCKS)
3583 
3584 /*
3585  * Inodes and files operations
3586  */
3587 
3588 /* dir.c */
3589 extern const struct file_operations ext4_dir_operations;
3590 
3591 /* file.c */
3592 extern const struct inode_operations ext4_file_inode_operations;
3593 extern const struct file_operations ext4_file_operations;
3594 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
3595 
3596 /* inline.c */
3597 extern int ext4_get_max_inline_size(struct inode *inode);
3598 extern int ext4_find_inline_data_nolock(struct inode *inode);
3599 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
3600 extern void ext4_update_final_de(void *de_buf, int old_size, int new_size);
3601 
3602 int ext4_readpage_inline(struct inode *inode, struct folio *folio);
3603 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3604 					 struct inode *inode,
3605 					 loff_t pos, unsigned len,
3606 					 struct folio **foliop);
3607 int ext4_write_inline_data_end(struct inode *inode, loff_t pos, unsigned len,
3608 			       unsigned copied, struct folio *folio);
3609 extern int ext4_generic_write_inline_data(struct address_space *mapping,
3610 					  struct inode *inode,
3611 					  loff_t pos, unsigned len,
3612 					  struct folio **foliop,
3613 					  void **fsdata, bool da);
3614 extern int ext4_try_add_inline_entry(handle_t *handle,
3615 				     struct ext4_filename *fname,
3616 				     struct inode *dir, struct inode *inode);
3617 extern int ext4_try_create_inline_dir(handle_t *handle,
3618 				      struct inode *parent,
3619 				      struct inode *inode);
3620 extern int ext4_read_inline_dir(struct file *filp,
3621 				struct dir_context *ctx,
3622 				int *has_inline_data);
3623 extern int ext4_inlinedir_to_tree(struct file *dir_file,
3624 				  struct inode *dir, ext4_lblk_t block,
3625 				  struct dx_hash_info *hinfo,
3626 				  __u32 start_hash, __u32 start_minor_hash,
3627 				  int *has_inline_data);
3628 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3629 					struct ext4_filename *fname,
3630 					struct ext4_dir_entry_2 **res_dir,
3631 					int *has_inline_data);
3632 extern int ext4_delete_inline_entry(handle_t *handle,
3633 				    struct inode *dir,
3634 				    struct ext4_dir_entry_2 *de_del,
3635 				    struct buffer_head *bh,
3636 				    int *has_inline_data);
3637 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3638 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3639 					struct ext4_dir_entry_2 **parent_de,
3640 					int *retval);
3641 extern void *ext4_read_inline_link(struct inode *inode);
3642 
3643 struct iomap;
3644 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap);
3645 
3646 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3647 
3648 extern int ext4_convert_inline_data(struct inode *inode);
3649 
ext4_has_inline_data(struct inode * inode)3650 static inline int ext4_has_inline_data(struct inode *inode)
3651 {
3652 	return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3653 	       EXT4_I(inode)->i_inline_off;
3654 }
3655 
3656 /* namei.c */
3657 extern const struct inode_operations ext4_dir_inode_operations;
3658 extern const struct inode_operations ext4_special_inode_operations;
3659 extern struct dentry *ext4_get_parent(struct dentry *child);
3660 extern int ext4_init_dirblock(handle_t *handle, struct inode *inode,
3661 			      struct buffer_head *dir_block,
3662 			      unsigned int parent_ino, void *inline_buf,
3663 			      int inline_size);
3664 extern void ext4_initialize_dirent_tail(struct buffer_head *bh,
3665 					unsigned int blocksize);
3666 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode,
3667 				      struct buffer_head *bh);
3668 extern int __ext4_unlink(struct inode *dir, const struct qstr *d_name,
3669 			 struct inode *inode, struct dentry *dentry);
3670 extern int __ext4_link(struct inode *dir, struct inode *inode,
3671 		       struct dentry *dentry);
3672 
3673 #define S_SHIFT 12
3674 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = {
3675 	[S_IFREG >> S_SHIFT]	= EXT4_FT_REG_FILE,
3676 	[S_IFDIR >> S_SHIFT]	= EXT4_FT_DIR,
3677 	[S_IFCHR >> S_SHIFT]	= EXT4_FT_CHRDEV,
3678 	[S_IFBLK >> S_SHIFT]	= EXT4_FT_BLKDEV,
3679 	[S_IFIFO >> S_SHIFT]	= EXT4_FT_FIFO,
3680 	[S_IFSOCK >> S_SHIFT]	= EXT4_FT_SOCK,
3681 	[S_IFLNK >> S_SHIFT]	= EXT4_FT_SYMLINK,
3682 };
3683 
ext4_set_de_type(struct super_block * sb,struct ext4_dir_entry_2 * de,umode_t mode)3684 static inline void ext4_set_de_type(struct super_block *sb,
3685 				struct ext4_dir_entry_2 *de,
3686 				umode_t mode) {
3687 	if (ext4_has_feature_filetype(sb))
3688 		de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3689 }
3690 
3691 /* readpages.c */
3692 extern int ext4_mpage_readpages(struct inode *inode,
3693 		struct readahead_control *rac, struct folio *folio);
3694 extern int __init ext4_init_post_read_processing(void);
3695 extern void ext4_exit_post_read_processing(void);
3696 
3697 /* symlink.c */
3698 extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3699 extern const struct inode_operations ext4_symlink_inode_operations;
3700 extern const struct inode_operations ext4_fast_symlink_inode_operations;
3701 
3702 /* sysfs.c */
3703 extern void ext4_notify_error_sysfs(struct ext4_sb_info *sbi);
3704 extern int ext4_register_sysfs(struct super_block *sb);
3705 extern void ext4_unregister_sysfs(struct super_block *sb);
3706 extern int __init ext4_init_sysfs(void);
3707 extern void ext4_exit_sysfs(void);
3708 
3709 /* block_validity */
3710 extern void ext4_release_system_zone(struct super_block *sb);
3711 extern int ext4_setup_system_zone(struct super_block *sb);
3712 extern int __init ext4_init_system_zone(void);
3713 extern void ext4_exit_system_zone(void);
3714 extern int ext4_inode_block_valid(struct inode *inode,
3715 				  ext4_fsblk_t start_blk,
3716 				  unsigned int count);
3717 extern int ext4_check_blockref(const char *, unsigned int,
3718 			       struct inode *, __le32 *, unsigned int);
3719 extern int ext4_sb_block_valid(struct super_block *sb, struct inode *inode,
3720 				ext4_fsblk_t start_blk, unsigned int count);
3721 
3722 
3723 /* extents.c */
3724 struct ext4_ext_path;
3725 struct ext4_extent;
3726 
3727 /*
3728  * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3729  * __le32.
3730  */
3731 #define EXT_MAX_BLOCKS	0xffffffff
3732 
3733 extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode);
3734 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3735 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3736 			       struct ext4_map_blocks *map, int flags);
3737 extern int ext4_ext_truncate(handle_t *, struct inode *);
3738 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3739 				 ext4_lblk_t end);
3740 extern void ext4_ext_init(struct super_block *);
3741 extern void ext4_ext_release(struct super_block *);
3742 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3743 			  loff_t len);
3744 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3745 					  loff_t offset, ssize_t len);
3746 extern int ext4_convert_unwritten_extents_atomic(handle_t *handle,
3747 			struct inode *inode, loff_t offset, ssize_t len);
3748 extern int ext4_convert_unwritten_io_end_vec(handle_t *handle,
3749 					     ext4_io_end_t *io_end);
3750 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3751 			   struct ext4_map_blocks *map, int flags);
3752 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3753 						   int num,
3754 						   struct ext4_ext_path *path);
3755 extern struct ext4_ext_path *ext4_ext_insert_extent(
3756 				handle_t *handle, struct inode *inode,
3757 				struct ext4_ext_path *path,
3758 				struct ext4_extent *newext, int gb_flags);
3759 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3760 					      struct ext4_ext_path *,
3761 					      int flags);
3762 extern void ext4_free_ext_path(struct ext4_ext_path *);
3763 extern int ext4_ext_check_inode(struct inode *inode);
3764 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3765 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3766 			__u64 start, __u64 len);
3767 extern int ext4_get_es_cache(struct inode *inode,
3768 			     struct fiemap_extent_info *fieinfo,
3769 			     __u64 start, __u64 len);
3770 extern int ext4_ext_precache(struct inode *inode);
3771 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3772 				struct inode *inode2, ext4_lblk_t lblk1,
3773 			     ext4_lblk_t lblk2,  ext4_lblk_t count,
3774 			     int mark_unwritten,int *err);
3775 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu);
3776 extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode,
3777 				       int check_cred, int restart_cred,
3778 				       int revoke_cred);
3779 extern void ext4_ext_replay_shrink_inode(struct inode *inode, ext4_lblk_t end);
3780 extern int ext4_ext_replay_set_iblocks(struct inode *inode);
3781 extern int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start,
3782 		int len, int unwritten, ext4_fsblk_t pblk);
3783 extern int ext4_ext_clear_bb(struct inode *inode);
3784 
3785 
3786 /* move_extent.c */
3787 extern void ext4_double_down_write_data_sem(struct inode *first,
3788 					    struct inode *second);
3789 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3790 					  struct inode *donor_inode);
3791 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3792 			     __u64 start_orig, __u64 start_donor,
3793 			     __u64 len, __u64 *moved_len);
3794 
3795 /* page-io.c */
3796 extern int __init ext4_init_pageio(void);
3797 extern void ext4_exit_pageio(void);
3798 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3799 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3800 extern int ext4_put_io_end(ext4_io_end_t *io_end);
3801 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3802 extern void ext4_io_submit_init(struct ext4_io_submit *io,
3803 				struct writeback_control *wbc);
3804 extern void ext4_end_io_rsv_work(struct work_struct *work);
3805 extern void ext4_io_submit(struct ext4_io_submit *io);
3806 int ext4_bio_write_folio(struct ext4_io_submit *io, struct folio *page,
3807 		size_t len);
3808 extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end);
3809 extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end);
3810 
3811 /* mmp.c */
3812 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3813 
3814 /* mmp.c */
3815 extern void ext4_stop_mmpd(struct ext4_sb_info *sbi);
3816 
3817 /* verity.c */
3818 extern const struct fsverity_operations ext4_verityops;
3819 
3820 /* orphan.c */
3821 extern int ext4_orphan_add(handle_t *, struct inode *);
3822 extern int ext4_orphan_del(handle_t *, struct inode *);
3823 extern void ext4_orphan_cleanup(struct super_block *sb,
3824 				struct ext4_super_block *es);
3825 extern void ext4_release_orphan_info(struct super_block *sb);
3826 extern int ext4_init_orphan_info(struct super_block *sb);
3827 extern int ext4_orphan_file_empty(struct super_block *sb);
3828 extern void ext4_orphan_file_block_trigger(
3829 				struct jbd2_buffer_trigger_type *triggers,
3830 				struct buffer_head *bh,
3831 				void *data, size_t size);
3832 
3833 /*
3834  * Add new method to test whether block and inode bitmaps are properly
3835  * initialized. With uninit_bg reading the block from disk is not enough
3836  * to mark the bitmap uptodate. We need to also zero-out the bitmap
3837  */
3838 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3839 
bitmap_uptodate(struct buffer_head * bh)3840 static inline int bitmap_uptodate(struct buffer_head *bh)
3841 {
3842 	return (buffer_uptodate(bh) &&
3843 			test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3844 }
set_bitmap_uptodate(struct buffer_head * bh)3845 static inline void set_bitmap_uptodate(struct buffer_head *bh)
3846 {
3847 	set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3848 }
3849 
3850 extern int ext4_resize_begin(struct super_block *sb);
3851 extern int ext4_resize_end(struct super_block *sb, bool update_backups);
3852 
ext4_set_io_unwritten_flag(struct ext4_io_end * io_end)3853 static inline void ext4_set_io_unwritten_flag(struct ext4_io_end *io_end)
3854 {
3855 	if (!(io_end->flag & EXT4_IO_END_UNWRITTEN))
3856 		io_end->flag |= EXT4_IO_END_UNWRITTEN;
3857 }
3858 
ext4_clear_io_unwritten_flag(ext4_io_end_t * io_end)3859 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3860 {
3861 	if (io_end->flag & EXT4_IO_END_UNWRITTEN)
3862 		io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3863 }
3864 
3865 extern const struct iomap_ops ext4_iomap_ops;
3866 extern const struct iomap_ops ext4_iomap_overwrite_ops;
3867 extern const struct iomap_ops ext4_iomap_report_ops;
3868 
ext4_buffer_uptodate(struct buffer_head * bh)3869 static inline int ext4_buffer_uptodate(struct buffer_head *bh)
3870 {
3871 	/*
3872 	 * If the buffer has the write error flag, we have failed
3873 	 * to write out data in the block.  In this  case, we don't
3874 	 * have to read the block because we may read the old data
3875 	 * successfully.
3876 	 */
3877 	if (buffer_write_io_error(bh))
3878 		set_buffer_uptodate(bh);
3879 	return buffer_uptodate(bh);
3880 }
3881 
ext4_inode_can_atomic_write(struct inode * inode)3882 static inline bool ext4_inode_can_atomic_write(struct inode *inode)
3883 {
3884 
3885 	return S_ISREG(inode->i_mode) &&
3886 		ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS) &&
3887 		EXT4_SB(inode->i_sb)->s_awu_min > 0;
3888 }
3889 
3890 extern int ext4_block_write_begin(handle_t *handle, struct folio *folio,
3891 				  loff_t pos, unsigned len,
3892 				  get_block_t *get_block);
3893 #endif	/* __KERNEL__ */
3894 
3895 #define EFSBADCRC	EBADMSG		/* Bad CRC detected */
3896 #define EFSCORRUPTED	EUCLEAN		/* Filesystem is corrupted */
3897 
3898 #endif	/* _EXT4_H */
3899