xref: /linux/fs/ext4/ext4.h (revision 5c2a430e85994f4873ea5ec42091baa1153bc731)
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 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 /* blocks already reserved */
189 #define EXT4_MB_HINT_RESERVED		0x0002
190 /* metadata is being allocated */
191 #define EXT4_MB_HINT_METADATA		0x0004
192 /* first blocks in the file */
193 #define EXT4_MB_HINT_FIRST		0x0008
194 /* search for the best chunk */
195 #define EXT4_MB_HINT_BEST		0x0010
196 /* data is being allocated */
197 #define EXT4_MB_HINT_DATA		0x0020
198 /* don't preallocate (for tails) */
199 #define EXT4_MB_HINT_NOPREALLOC		0x0040
200 /* allocate for locality group */
201 #define EXT4_MB_HINT_GROUP_ALLOC	0x0080
202 /* allocate goal blocks or none */
203 #define EXT4_MB_HINT_GOAL_ONLY		0x0100
204 /* goal is meaningful */
205 #define EXT4_MB_HINT_TRY_GOAL		0x0200
206 /* blocks already pre-reserved by delayed allocation */
207 #define EXT4_MB_DELALLOC_RESERVED	0x0400
208 /* We are doing stream allocation */
209 #define EXT4_MB_STREAM_ALLOC		0x0800
210 /* Use reserved root blocks if needed */
211 #define EXT4_MB_USE_ROOT_BLOCKS		0x1000
212 /* Use blocks from reserved pool */
213 #define EXT4_MB_USE_RESERVED		0x2000
214 /* Do strict check for free blocks while retrying block allocation */
215 #define EXT4_MB_STRICT_CHECK		0x4000
216 /* Large fragment size list lookup succeeded at least once for
217  * CR_POWER2_ALIGNED */
218 #define EXT4_MB_CR_POWER2_ALIGNED_OPTIMIZED		0x8000
219 /* Avg fragment size rb tree lookup succeeded at least once for
220  * CR_GOAL_LEN_FAST */
221 #define EXT4_MB_CR_GOAL_LEN_FAST_OPTIMIZED		0x00010000
222 /* Avg fragment size rb tree lookup succeeded at least once for
223  * CR_BEST_AVAIL_LEN */
224 #define EXT4_MB_CR_BEST_AVAIL_LEN_OPTIMIZED		0x00020000
225 
226 struct ext4_allocation_request {
227 	/* target inode for block we're allocating */
228 	struct inode *inode;
229 	/* how many blocks we want to allocate */
230 	unsigned int len;
231 	/* logical block in target inode */
232 	ext4_lblk_t logical;
233 	/* the closest logical allocated block to the left */
234 	ext4_lblk_t lleft;
235 	/* the closest logical allocated block to the right */
236 	ext4_lblk_t lright;
237 	/* phys. target (a hint) */
238 	ext4_fsblk_t goal;
239 	/* phys. block for the closest logical allocated block to the left */
240 	ext4_fsblk_t pleft;
241 	/* phys. block for the closest logical allocated block to the right */
242 	ext4_fsblk_t pright;
243 	/* flags. see above EXT4_MB_HINT_* */
244 	unsigned int flags;
245 };
246 
247 /*
248  * Logical to physical block mapping, used by ext4_map_blocks()
249  *
250  * This structure is used to pass requests into ext4_map_blocks() as
251  * well as to store the information returned by ext4_map_blocks().  It
252  * takes less room on the stack than a struct buffer_head.
253  */
254 #define EXT4_MAP_NEW		BIT(BH_New)
255 #define EXT4_MAP_MAPPED		BIT(BH_Mapped)
256 #define EXT4_MAP_UNWRITTEN	BIT(BH_Unwritten)
257 #define EXT4_MAP_BOUNDARY	BIT(BH_Boundary)
258 #define EXT4_MAP_DELAYED	BIT(BH_Delay)
259 #define EXT4_MAP_FLAGS		(EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
260 				 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\
261 				 EXT4_MAP_DELAYED)
262 
263 struct ext4_map_blocks {
264 	ext4_fsblk_t m_pblk;
265 	ext4_lblk_t m_lblk;
266 	unsigned int m_len;
267 	unsigned int m_flags;
268 };
269 
270 /*
271  * Block validity checking, system zone rbtree.
272  */
273 struct ext4_system_blocks {
274 	struct rb_root root;
275 	struct rcu_head rcu;
276 };
277 
278 /*
279  * Flags for ext4_io_end->flags
280  */
281 #define EXT4_IO_END_UNWRITTEN	0x0001
282 #define EXT4_IO_END_FAILED	0x0002
283 
284 #define EXT4_IO_END_DEFER_COMPLETION (EXT4_IO_END_UNWRITTEN | EXT4_IO_END_FAILED)
285 
286 struct ext4_io_end_vec {
287 	struct list_head list;		/* list of io_end_vec */
288 	loff_t offset;			/* offset in the file */
289 	ssize_t size;			/* size of the extent */
290 };
291 
292 /*
293  * For converting unwritten extents on a work queue. 'handle' is used for
294  * buffered writeback.
295  */
296 typedef struct ext4_io_end {
297 	struct list_head	list;		/* per-file finished IO list */
298 	handle_t		*handle;	/* handle reserved for extent
299 						 * conversion */
300 	struct inode		*inode;		/* file being written to */
301 	struct bio		*bio;		/* Linked list of completed
302 						 * bios covering the extent */
303 	unsigned int		flag;		/* unwritten or not */
304 	refcount_t		count;		/* reference counter */
305 	struct list_head	list_vec;	/* list of ext4_io_end_vec */
306 } ext4_io_end_t;
307 
308 struct ext4_io_submit {
309 	struct writeback_control *io_wbc;
310 	struct bio		*io_bio;
311 	ext4_io_end_t		*io_end;
312 	sector_t		io_next_block;
313 };
314 
315 /*
316  * Special inodes numbers
317  */
318 #define	EXT4_BAD_INO		 1	/* Bad blocks inode */
319 #define EXT4_ROOT_INO		 2	/* Root inode */
320 #define EXT4_USR_QUOTA_INO	 3	/* User quota inode */
321 #define EXT4_GRP_QUOTA_INO	 4	/* Group quota inode */
322 #define EXT4_BOOT_LOADER_INO	 5	/* Boot loader inode */
323 #define EXT4_UNDEL_DIR_INO	 6	/* Undelete directory inode */
324 #define EXT4_RESIZE_INO		 7	/* Reserved group descriptors inode */
325 #define EXT4_JOURNAL_INO	 8	/* Journal inode */
326 
327 /* First non-reserved inode for old ext4 filesystems */
328 #define EXT4_GOOD_OLD_FIRST_INO	11
329 
330 /*
331  * Maximal count of links to a file
332  */
333 #define EXT4_LINK_MAX		65000
334 
335 /*
336  * Macro-instructions used to manage several block sizes
337  */
338 #define EXT4_MIN_BLOCK_SIZE		1024
339 #define	EXT4_MAX_BLOCK_SIZE		65536
340 #define EXT4_MIN_BLOCK_LOG_SIZE		10
341 #define EXT4_MAX_BLOCK_LOG_SIZE		16
342 #define EXT4_MAX_CLUSTER_LOG_SIZE	30
343 #ifdef __KERNEL__
344 # define EXT4_BLOCK_SIZE(s)		((s)->s_blocksize)
345 #else
346 # define EXT4_BLOCK_SIZE(s)		(EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
347 #endif
348 #define	EXT4_ADDR_PER_BLOCK(s)		(EXT4_BLOCK_SIZE(s) / sizeof(__u32))
349 #define EXT4_CLUSTER_SIZE(s)		(EXT4_BLOCK_SIZE(s) << \
350 					 EXT4_SB(s)->s_cluster_bits)
351 #ifdef __KERNEL__
352 # define EXT4_BLOCK_SIZE_BITS(s)	((s)->s_blocksize_bits)
353 # define EXT4_CLUSTER_BITS(s)		(EXT4_SB(s)->s_cluster_bits)
354 #else
355 # define EXT4_BLOCK_SIZE_BITS(s)	((s)->s_log_block_size + 10)
356 #endif
357 #ifdef __KERNEL__
358 #define	EXT4_ADDR_PER_BLOCK_BITS(s)	(EXT4_SB(s)->s_addr_per_block_bits)
359 #define EXT4_INODE_SIZE(s)		(EXT4_SB(s)->s_inode_size)
360 #define EXT4_FIRST_INO(s)		(EXT4_SB(s)->s_first_ino)
361 #else
362 #define EXT4_INODE_SIZE(s)	(((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
363 				 EXT4_GOOD_OLD_INODE_SIZE : \
364 				 (s)->s_inode_size)
365 #define EXT4_FIRST_INO(s)	(((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
366 				 EXT4_GOOD_OLD_FIRST_INO : \
367 				 (s)->s_first_ino)
368 #endif
369 #define EXT4_BLOCK_ALIGN(size, blkbits)		ALIGN((size), (1 << (blkbits)))
370 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \
371 	((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \
372 								  blkbits))
373 #define EXT4_B_TO_LBLK(inode, offset) \
374 	(round_up((offset), i_blocksize(inode)) >> (inode)->i_blkbits)
375 
376 /* Translate a block number to a cluster number */
377 #define EXT4_B2C(sbi, blk)	((blk) >> (sbi)->s_cluster_bits)
378 /* Translate a cluster number to a block number */
379 #define EXT4_C2B(sbi, cluster)	((cluster) << (sbi)->s_cluster_bits)
380 /* Translate # of blks to # of clusters */
381 #define EXT4_NUM_B2C(sbi, blks)	(((blks) + (sbi)->s_cluster_ratio - 1) >> \
382 				 (sbi)->s_cluster_bits)
383 /* Mask out the low bits to get the starting block of the cluster */
384 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) &				\
385 				  ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
386 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) &				\
387 				  ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
388 /* Fill in the low bits to get the last block of the cluster */
389 #define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) |				\
390 				    ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1))
391 /* Get the cluster offset */
392 #define EXT4_PBLK_COFF(s, pblk) ((pblk) &				\
393 				 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
394 #define EXT4_LBLK_COFF(s, lblk) ((lblk) &				\
395 				 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
396 
397 /*
398  * Structure of a blocks group descriptor
399  */
400 struct ext4_group_desc
401 {
402 	__le32	bg_block_bitmap_lo;	/* Blocks bitmap block */
403 	__le32	bg_inode_bitmap_lo;	/* Inodes bitmap block */
404 	__le32	bg_inode_table_lo;	/* Inodes table block */
405 	__le16	bg_free_blocks_count_lo;/* Free blocks count */
406 	__le16	bg_free_inodes_count_lo;/* Free inodes count */
407 	__le16	bg_used_dirs_count_lo;	/* Directories count */
408 	__le16	bg_flags;		/* EXT4_BG_flags (INODE_UNINIT, etc) */
409 	__le32  bg_exclude_bitmap_lo;   /* Exclude bitmap for snapshots */
410 	__le16  bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
411 	__le16  bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
412 	__le16  bg_itable_unused_lo;	/* Unused inodes count */
413 	__le16  bg_checksum;		/* crc16(sb_uuid+group+desc) */
414 	__le32	bg_block_bitmap_hi;	/* Blocks bitmap block MSB */
415 	__le32	bg_inode_bitmap_hi;	/* Inodes bitmap block MSB */
416 	__le32	bg_inode_table_hi;	/* Inodes table block MSB */
417 	__le16	bg_free_blocks_count_hi;/* Free blocks count MSB */
418 	__le16	bg_free_inodes_count_hi;/* Free inodes count MSB */
419 	__le16	bg_used_dirs_count_hi;	/* Directories count MSB */
420 	__le16  bg_itable_unused_hi;    /* Unused inodes count MSB */
421 	__le32  bg_exclude_bitmap_hi;   /* Exclude bitmap block MSB */
422 	__le16  bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
423 	__le16  bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
424 	__u32   bg_reserved;
425 };
426 
427 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END	\
428 	(offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
429 	 sizeof(__le16))
430 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END	\
431 	(offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
432 	 sizeof(__le16))
433 
434 /*
435  * Structure of a flex block group info
436  */
437 
438 struct flex_groups {
439 	atomic64_t	free_clusters;
440 	atomic_t	free_inodes;
441 	atomic_t	used_dirs;
442 };
443 
444 #define EXT4_BG_INODE_UNINIT	0x0001 /* Inode table/bitmap not in use */
445 #define EXT4_BG_BLOCK_UNINIT	0x0002 /* Block bitmap not in use */
446 #define EXT4_BG_INODE_ZEROED	0x0004 /* On-disk itable initialized to zero */
447 
448 /*
449  * Macro-instructions used to manage group descriptors
450  */
451 #define EXT4_MIN_DESC_SIZE		32
452 #define EXT4_MIN_DESC_SIZE_64BIT	64
453 #define	EXT4_MAX_DESC_SIZE		EXT4_MIN_BLOCK_SIZE
454 #define EXT4_DESC_SIZE(s)		(EXT4_SB(s)->s_desc_size)
455 #ifdef __KERNEL__
456 # define EXT4_BLOCKS_PER_GROUP(s)	(EXT4_SB(s)->s_blocks_per_group)
457 # define EXT4_CLUSTERS_PER_GROUP(s)	(EXT4_SB(s)->s_clusters_per_group)
458 # define EXT4_DESC_PER_BLOCK(s)		(EXT4_SB(s)->s_desc_per_block)
459 # define EXT4_INODES_PER_GROUP(s)	(EXT4_SB(s)->s_inodes_per_group)
460 # define EXT4_DESC_PER_BLOCK_BITS(s)	(EXT4_SB(s)->s_desc_per_block_bits)
461 #else
462 # define EXT4_BLOCKS_PER_GROUP(s)	((s)->s_blocks_per_group)
463 # define EXT4_DESC_PER_BLOCK(s)		(EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
464 # define EXT4_INODES_PER_GROUP(s)	((s)->s_inodes_per_group)
465 #endif
466 
467 /*
468  * Constants relative to the data blocks
469  */
470 #define	EXT4_NDIR_BLOCKS		12
471 #define	EXT4_IND_BLOCK			EXT4_NDIR_BLOCKS
472 #define	EXT4_DIND_BLOCK			(EXT4_IND_BLOCK + 1)
473 #define	EXT4_TIND_BLOCK			(EXT4_DIND_BLOCK + 1)
474 #define	EXT4_N_BLOCKS			(EXT4_TIND_BLOCK + 1)
475 
476 /*
477  * Inode flags
478  */
479 #define	EXT4_SECRM_FL			0x00000001 /* Secure deletion */
480 #define	EXT4_UNRM_FL			0x00000002 /* Undelete */
481 #define	EXT4_COMPR_FL			0x00000004 /* Compress file */
482 #define EXT4_SYNC_FL			0x00000008 /* Synchronous updates */
483 #define EXT4_IMMUTABLE_FL		0x00000010 /* Immutable file */
484 #define EXT4_APPEND_FL			0x00000020 /* writes to file may only append */
485 #define EXT4_NODUMP_FL			0x00000040 /* do not dump file */
486 #define EXT4_NOATIME_FL			0x00000080 /* do not update atime */
487 /* Reserved for compression usage... */
488 #define EXT4_DIRTY_FL			0x00000100
489 #define EXT4_COMPRBLK_FL		0x00000200 /* One or more compressed clusters */
490 #define EXT4_NOCOMPR_FL			0x00000400 /* Don't compress */
491 	/* nb: was previously EXT2_ECOMPR_FL */
492 #define EXT4_ENCRYPT_FL			0x00000800 /* encrypted file */
493 /* End compression flags --- maybe not all used */
494 #define EXT4_INDEX_FL			0x00001000 /* hash-indexed directory */
495 #define EXT4_IMAGIC_FL			0x00002000 /* AFS directory */
496 #define EXT4_JOURNAL_DATA_FL		0x00004000 /* file data should be journaled */
497 #define EXT4_NOTAIL_FL			0x00008000 /* file tail should not be merged */
498 #define EXT4_DIRSYNC_FL			0x00010000 /* dirsync behaviour (directories only) */
499 #define EXT4_TOPDIR_FL			0x00020000 /* Top of directory hierarchies*/
500 #define EXT4_HUGE_FILE_FL               0x00040000 /* Set to each huge file */
501 #define EXT4_EXTENTS_FL			0x00080000 /* Inode uses extents */
502 #define EXT4_VERITY_FL			0x00100000 /* Verity protected inode */
503 #define EXT4_EA_INODE_FL	        0x00200000 /* Inode used for large EA */
504 /* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */
505 
506 #define EXT4_DAX_FL			0x02000000 /* Inode is DAX */
507 
508 #define EXT4_INLINE_DATA_FL		0x10000000 /* Inode has inline data. */
509 #define EXT4_PROJINHERIT_FL		0x20000000 /* Create with parents projid */
510 #define EXT4_CASEFOLD_FL		0x40000000 /* Casefolded directory */
511 #define EXT4_RESERVED_FL		0x80000000 /* reserved for ext4 lib */
512 
513 /* User modifiable flags */
514 #define EXT4_FL_USER_MODIFIABLE		(EXT4_SECRM_FL | \
515 					 EXT4_UNRM_FL | \
516 					 EXT4_COMPR_FL | \
517 					 EXT4_SYNC_FL | \
518 					 EXT4_IMMUTABLE_FL | \
519 					 EXT4_APPEND_FL | \
520 					 EXT4_NODUMP_FL | \
521 					 EXT4_NOATIME_FL | \
522 					 EXT4_JOURNAL_DATA_FL | \
523 					 EXT4_NOTAIL_FL | \
524 					 EXT4_DIRSYNC_FL | \
525 					 EXT4_TOPDIR_FL | \
526 					 EXT4_EXTENTS_FL | \
527 					 0x00400000 /* EXT4_EOFBLOCKS_FL */ | \
528 					 EXT4_DAX_FL | \
529 					 EXT4_PROJINHERIT_FL | \
530 					 EXT4_CASEFOLD_FL)
531 
532 /* User visible flags */
533 #define EXT4_FL_USER_VISIBLE		(EXT4_FL_USER_MODIFIABLE | \
534 					 EXT4_DIRTY_FL | \
535 					 EXT4_COMPRBLK_FL | \
536 					 EXT4_NOCOMPR_FL | \
537 					 EXT4_ENCRYPT_FL | \
538 					 EXT4_INDEX_FL | \
539 					 EXT4_VERITY_FL | \
540 					 EXT4_INLINE_DATA_FL)
541 
542 /* Flags that should be inherited by new inodes from their parent. */
543 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
544 			   EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
545 			   EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
546 			   EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
547 			   EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL |\
548 			   EXT4_DAX_FL)
549 
550 /* Flags that are appropriate for regular files (all but dir-specific ones). */
551 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\
552 			   EXT4_PROJINHERIT_FL))
553 
554 /* Flags that are appropriate for non-directories/regular files. */
555 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
556 
557 /* The only flags that should be swapped */
558 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL)
559 
560 /* Flags which are mutually exclusive to DAX */
561 #define EXT4_DAX_MUT_EXCL (EXT4_VERITY_FL | EXT4_ENCRYPT_FL |\
562 			   EXT4_JOURNAL_DATA_FL | EXT4_INLINE_DATA_FL)
563 
564 /* Mask out flags that are inappropriate for the given type of inode. */
ext4_mask_flags(umode_t mode,__u32 flags)565 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
566 {
567 	if (S_ISDIR(mode))
568 		return flags;
569 	else if (S_ISREG(mode))
570 		return flags & EXT4_REG_FLMASK;
571 	else
572 		return flags & EXT4_OTHER_FLMASK;
573 }
574 
575 /*
576  * Inode flags used for atomic set/get
577  */
578 enum {
579 	EXT4_INODE_SECRM	= 0,	/* Secure deletion */
580 	EXT4_INODE_UNRM		= 1,	/* Undelete */
581 	EXT4_INODE_COMPR	= 2,	/* Compress file */
582 	EXT4_INODE_SYNC		= 3,	/* Synchronous updates */
583 	EXT4_INODE_IMMUTABLE	= 4,	/* Immutable file */
584 	EXT4_INODE_APPEND	= 5,	/* writes to file may only append */
585 	EXT4_INODE_NODUMP	= 6,	/* do not dump file */
586 	EXT4_INODE_NOATIME	= 7,	/* do not update atime */
587 /* Reserved for compression usage... */
588 	EXT4_INODE_DIRTY	= 8,
589 	EXT4_INODE_COMPRBLK	= 9,	/* One or more compressed clusters */
590 	EXT4_INODE_NOCOMPR	= 10,	/* Don't compress */
591 	EXT4_INODE_ENCRYPT	= 11,	/* Encrypted file */
592 /* End compression flags --- maybe not all used */
593 	EXT4_INODE_INDEX	= 12,	/* hash-indexed directory */
594 	EXT4_INODE_IMAGIC	= 13,	/* AFS directory */
595 	EXT4_INODE_JOURNAL_DATA	= 14,	/* file data should be journaled */
596 	EXT4_INODE_NOTAIL	= 15,	/* file tail should not be merged */
597 	EXT4_INODE_DIRSYNC	= 16,	/* dirsync behaviour (directories only) */
598 	EXT4_INODE_TOPDIR	= 17,	/* Top of directory hierarchies*/
599 	EXT4_INODE_HUGE_FILE	= 18,	/* Set to each huge file */
600 	EXT4_INODE_EXTENTS	= 19,	/* Inode uses extents */
601 	EXT4_INODE_VERITY	= 20,	/* Verity protected inode */
602 	EXT4_INODE_EA_INODE	= 21,	/* Inode used for large EA */
603 /* 22 was formerly EXT4_INODE_EOFBLOCKS */
604 	EXT4_INODE_DAX		= 25,	/* Inode is DAX */
605 	EXT4_INODE_INLINE_DATA	= 28,	/* Data in inode. */
606 	EXT4_INODE_PROJINHERIT	= 29,	/* Create with parents projid */
607 	EXT4_INODE_CASEFOLD	= 30,	/* Casefolded directory */
608 	EXT4_INODE_RESERVED	= 31,	/* reserved for ext4 lib */
609 };
610 
611 /*
612  * Since it's pretty easy to mix up bit numbers and hex values, we use a
613  * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
614  * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
615  * any extra space in the compiled kernel image, otherwise, the build will fail.
616  * It's important that these values are the same, since we are using
617  * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
618  * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
619  * values found in ext2, ext3 and ext4 filesystems, and of course the values
620  * defined in e2fsprogs.
621  *
622  * It's not paranoia if the Murphy's Law really *is* out to get you.  :-)
623  */
624 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1U << EXT4_INODE_##FLAG))
625 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
626 
ext4_check_flag_values(void)627 static inline void ext4_check_flag_values(void)
628 {
629 	CHECK_FLAG_VALUE(SECRM);
630 	CHECK_FLAG_VALUE(UNRM);
631 	CHECK_FLAG_VALUE(COMPR);
632 	CHECK_FLAG_VALUE(SYNC);
633 	CHECK_FLAG_VALUE(IMMUTABLE);
634 	CHECK_FLAG_VALUE(APPEND);
635 	CHECK_FLAG_VALUE(NODUMP);
636 	CHECK_FLAG_VALUE(NOATIME);
637 	CHECK_FLAG_VALUE(DIRTY);
638 	CHECK_FLAG_VALUE(COMPRBLK);
639 	CHECK_FLAG_VALUE(NOCOMPR);
640 	CHECK_FLAG_VALUE(ENCRYPT);
641 	CHECK_FLAG_VALUE(INDEX);
642 	CHECK_FLAG_VALUE(IMAGIC);
643 	CHECK_FLAG_VALUE(JOURNAL_DATA);
644 	CHECK_FLAG_VALUE(NOTAIL);
645 	CHECK_FLAG_VALUE(DIRSYNC);
646 	CHECK_FLAG_VALUE(TOPDIR);
647 	CHECK_FLAG_VALUE(HUGE_FILE);
648 	CHECK_FLAG_VALUE(EXTENTS);
649 	CHECK_FLAG_VALUE(VERITY);
650 	CHECK_FLAG_VALUE(EA_INODE);
651 	CHECK_FLAG_VALUE(INLINE_DATA);
652 	CHECK_FLAG_VALUE(PROJINHERIT);
653 	CHECK_FLAG_VALUE(CASEFOLD);
654 	CHECK_FLAG_VALUE(RESERVED);
655 }
656 
657 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
658 struct compat_ext4_new_group_input {
659 	u32 group;
660 	compat_u64 block_bitmap;
661 	compat_u64 inode_bitmap;
662 	compat_u64 inode_table;
663 	u32 blocks_count;
664 	u16 reserved_blocks;
665 	u16 unused;
666 };
667 #endif
668 
669 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
670 struct ext4_new_group_data {
671 	__u32 group;
672 	__u64 block_bitmap;
673 	__u64 inode_bitmap;
674 	__u64 inode_table;
675 	__u32 blocks_count;
676 	__u16 reserved_blocks;
677 	__u16 mdata_blocks;
678 	__u32 free_clusters_count;
679 };
680 
681 /* Indexes used to index group tables in ext4_new_group_data */
682 enum {
683 	BLOCK_BITMAP = 0,	/* block bitmap */
684 	INODE_BITMAP,		/* inode bitmap */
685 	INODE_TABLE,		/* inode tables */
686 	GROUP_TABLE_COUNT,
687 };
688 
689 /*
690  * Flags used by ext4_map_blocks()
691  */
692 	/* Allocate any needed blocks and/or convert an unwritten
693 	   extent to be an initialized ext4 */
694 #define EXT4_GET_BLOCKS_CREATE			0x0001
695 	/* Request the creation of an unwritten extent */
696 #define EXT4_GET_BLOCKS_UNWRIT_EXT		0x0002
697 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT	(EXT4_GET_BLOCKS_UNWRIT_EXT|\
698 						 EXT4_GET_BLOCKS_CREATE)
699 	/* Caller is from the delayed allocation writeout path
700 	 * finally doing the actual allocation of delayed blocks */
701 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE	0x0004
702 	/* caller is from the direct IO path, request to creation of an
703 	unwritten extents if not allocated, split the unwritten
704 	extent if blocks has been preallocated already*/
705 #define EXT4_GET_BLOCKS_PRE_IO			0x0008
706 #define EXT4_GET_BLOCKS_CONVERT			0x0010
707 #define EXT4_GET_BLOCKS_IO_CREATE_EXT		(EXT4_GET_BLOCKS_PRE_IO|\
708 					 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
709 	/* Convert extent to initialized after IO complete */
710 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT		(EXT4_GET_BLOCKS_CONVERT|\
711 					 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
712 	/* Eventual metadata allocation (due to growing extent tree)
713 	 * should not fail, so try to use reserved blocks for that.*/
714 #define EXT4_GET_BLOCKS_METADATA_NOFAIL		0x0020
715 	/* Don't normalize allocation size (used for fallocate) */
716 #define EXT4_GET_BLOCKS_NO_NORMALIZE		0x0040
717 	/* Convert written extents to unwritten */
718 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN	0x0100
719 	/* Write zeros to newly created written extents */
720 #define EXT4_GET_BLOCKS_ZERO			0x0200
721 #define EXT4_GET_BLOCKS_CREATE_ZERO		(EXT4_GET_BLOCKS_CREATE |\
722 					EXT4_GET_BLOCKS_ZERO)
723 	/* Caller will submit data before dropping transaction handle. This
724 	 * allows jbd2 to avoid submitting data before commit. */
725 #define EXT4_GET_BLOCKS_IO_SUBMIT		0x0400
726 	/* Caller is in the atomic contex, find extent if it has been cached */
727 #define EXT4_GET_BLOCKS_CACHED_NOWAIT		0x0800
728 
729 /*
730  * The bit position of these flags must not overlap with any of the
731  * EXT4_GET_BLOCKS_*.  They are used by ext4_find_extent(),
732  * read_extent_tree_block(), ext4_split_extent_at(),
733  * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
734  * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
735  * caching the extents when reading from the extent tree while a
736  * truncate or punch hole operation is in progress.
737  */
738 #define EXT4_EX_NOCACHE				0x40000000
739 #define EXT4_EX_FORCE_CACHE			0x20000000
740 #define EXT4_EX_NOFAIL				0x10000000
741 
742 /*
743  * Flags used by ext4_free_blocks
744  */
745 #define EXT4_FREE_BLOCKS_METADATA		0x0001
746 #define EXT4_FREE_BLOCKS_FORGET			0x0002
747 #define EXT4_FREE_BLOCKS_VALIDATED		0x0004
748 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE		0x0008
749 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER	0x0010
750 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER	0x0020
751 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER      0x0040
752 
753 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
754 /*
755  * ioctl commands in 32 bit emulation
756  */
757 #define EXT4_IOC32_GETVERSION		_IOR('f', 3, int)
758 #define EXT4_IOC32_SETVERSION		_IOW('f', 4, int)
759 #define EXT4_IOC32_GETRSVSZ		_IOR('f', 5, int)
760 #define EXT4_IOC32_SETRSVSZ		_IOW('f', 6, int)
761 #define EXT4_IOC32_GROUP_EXTEND		_IOW('f', 7, unsigned int)
762 #define EXT4_IOC32_GROUP_ADD		_IOW('f', 8, struct compat_ext4_new_group_input)
763 #define EXT4_IOC32_GETVERSION_OLD	FS_IOC32_GETVERSION
764 #define EXT4_IOC32_SETVERSION_OLD	FS_IOC32_SETVERSION
765 #endif
766 
767 /* Max physical block we can address w/o extents */
768 #define EXT4_MAX_BLOCK_FILE_PHYS	0xFFFFFFFF
769 
770 /* Max logical block we can support */
771 #define EXT4_MAX_LOGICAL_BLOCK		0xFFFFFFFE
772 
773 /*
774  * Structure of an inode on the disk
775  */
776 struct ext4_inode {
777 	__le16	i_mode;		/* File mode */
778 	__le16	i_uid;		/* Low 16 bits of Owner Uid */
779 	__le32	i_size_lo;	/* Size in bytes */
780 	__le32	i_atime;	/* Access time */
781 	__le32	i_ctime;	/* Inode Change time */
782 	__le32	i_mtime;	/* Modification time */
783 	__le32	i_dtime;	/* Deletion Time */
784 	__le16	i_gid;		/* Low 16 bits of Group Id */
785 	__le16	i_links_count;	/* Links count */
786 	__le32	i_blocks_lo;	/* Blocks count */
787 	__le32	i_flags;	/* File flags */
788 	union {
789 		struct {
790 			__le32  l_i_version;
791 		} linux1;
792 		struct {
793 			__u32  h_i_translator;
794 		} hurd1;
795 		struct {
796 			__u32  m_i_reserved1;
797 		} masix1;
798 	} osd1;				/* OS dependent 1 */
799 	__le32	i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
800 	__le32	i_generation;	/* File version (for NFS) */
801 	__le32	i_file_acl_lo;	/* File ACL */
802 	__le32	i_size_high;
803 	__le32	i_obso_faddr;	/* Obsoleted fragment address */
804 	union {
805 		struct {
806 			__le16	l_i_blocks_high; /* were l_i_reserved1 */
807 			__le16	l_i_file_acl_high;
808 			__le16	l_i_uid_high;	/* these 2 fields */
809 			__le16	l_i_gid_high;	/* were reserved2[0] */
810 			__le16	l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
811 			__le16	l_i_reserved;
812 		} linux2;
813 		struct {
814 			__le16	h_i_reserved1;	/* Obsoleted fragment number/size which are removed in ext4 */
815 			__u16	h_i_mode_high;
816 			__u16	h_i_uid_high;
817 			__u16	h_i_gid_high;
818 			__u32	h_i_author;
819 		} hurd2;
820 		struct {
821 			__le16	h_i_reserved1;	/* Obsoleted fragment number/size which are removed in ext4 */
822 			__le16	m_i_file_acl_high;
823 			__u32	m_i_reserved2[2];
824 		} masix2;
825 	} osd2;				/* OS dependent 2 */
826 	__le16	i_extra_isize;
827 	__le16	i_checksum_hi;	/* crc32c(uuid+inum+inode) BE */
828 	__le32  i_ctime_extra;  /* extra Change time      (nsec << 2 | epoch) */
829 	__le32  i_mtime_extra;  /* extra Modification time(nsec << 2 | epoch) */
830 	__le32  i_atime_extra;  /* extra Access time      (nsec << 2 | epoch) */
831 	__le32  i_crtime;       /* File Creation time */
832 	__le32  i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
833 	__le32  i_version_hi;	/* high 32 bits for 64-bit version */
834 	__le32	i_projid;	/* Project ID */
835 };
836 
837 #define EXT4_EPOCH_BITS 2
838 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
839 #define EXT4_NSEC_MASK  (~0UL << EXT4_EPOCH_BITS)
840 
841 /*
842  * Extended fields will fit into an inode if the filesystem was formatted
843  * with large inodes (-I 256 or larger) and there are not currently any EAs
844  * consuming all of the available space. For new inodes we always reserve
845  * enough space for the kernel's known extended fields, but for inodes
846  * created with an old kernel this might not have been the case. None of
847  * the extended inode fields is critical for correct filesystem operation.
848  * This macro checks if a certain field fits in the inode. Note that
849  * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
850  */
851 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field)	\
852 	((offsetof(typeof(*ext4_inode), field) +	\
853 	  sizeof((ext4_inode)->field))			\
854 	<= (EXT4_GOOD_OLD_INODE_SIZE +			\
855 	    (einode)->i_extra_isize))			\
856 
857 /*
858  * We use an encoding that preserves the times for extra epoch "00":
859  *
860  * extra  msb of                         adjust for signed
861  * epoch  32-bit                         32-bit tv_sec to
862  * bits   time    decoded 64-bit tv_sec  64-bit tv_sec      valid time range
863  * 0 0    1    -0x80000000..-0x00000001  0x000000000 1901-12-13..1969-12-31
864  * 0 0    0    0x000000000..0x07fffffff  0x000000000 1970-01-01..2038-01-19
865  * 0 1    1    0x080000000..0x0ffffffff  0x100000000 2038-01-19..2106-02-07
866  * 0 1    0    0x100000000..0x17fffffff  0x100000000 2106-02-07..2174-02-25
867  * 1 0    1    0x180000000..0x1ffffffff  0x200000000 2174-02-25..2242-03-16
868  * 1 0    0    0x200000000..0x27fffffff  0x200000000 2242-03-16..2310-04-04
869  * 1 1    1    0x280000000..0x2ffffffff  0x300000000 2310-04-04..2378-04-22
870  * 1 1    0    0x300000000..0x37fffffff  0x300000000 2378-04-22..2446-05-10
871  *
872  * Note that previous versions of the kernel on 64-bit systems would
873  * incorrectly use extra epoch bits 1,1 for dates between 1901 and
874  * 1970.  e2fsck will correct this, assuming that it is run on the
875  * affected filesystem before 2242.
876  */
877 
ext4_encode_extra_time(struct timespec64 ts)878 static inline __le32 ext4_encode_extra_time(struct timespec64 ts)
879 {
880 	u32 extra = ((ts.tv_sec - (s32)ts.tv_sec) >> 32) & EXT4_EPOCH_MASK;
881 	return cpu_to_le32(extra | (ts.tv_nsec << EXT4_EPOCH_BITS));
882 }
883 
ext4_decode_extra_time(__le32 base,__le32 extra)884 static inline struct timespec64 ext4_decode_extra_time(__le32 base,
885 						       __le32 extra)
886 {
887 	struct timespec64 ts = { .tv_sec = (signed)le32_to_cpu(base) };
888 
889 	if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK)))
890 		ts.tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
891 	ts.tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
892 	return ts;
893 }
894 
895 #define EXT4_INODE_SET_XTIME_VAL(xtime, inode, raw_inode, ts)			\
896 do {										\
897 	if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {	\
898 		(raw_inode)->xtime = cpu_to_le32((ts).tv_sec);			\
899 		(raw_inode)->xtime ## _extra = ext4_encode_extra_time(ts);	\
900 	} else									\
901 		(raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (ts).tv_sec, S32_MIN, S32_MAX));	\
902 } while (0)
903 
904 #define EXT4_INODE_SET_ATIME(inode, raw_inode)						\
905 	EXT4_INODE_SET_XTIME_VAL(i_atime, inode, raw_inode, inode_get_atime(inode))
906 
907 #define EXT4_INODE_SET_MTIME(inode, raw_inode)						\
908 	EXT4_INODE_SET_XTIME_VAL(i_mtime, inode, raw_inode, inode_get_mtime(inode))
909 
910 #define EXT4_INODE_SET_CTIME(inode, raw_inode)						\
911 	EXT4_INODE_SET_XTIME_VAL(i_ctime, inode, raw_inode, inode_get_ctime(inode))
912 
913 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode)				\
914 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))			\
915 		EXT4_INODE_SET_XTIME_VAL(xtime, &((einode)->vfs_inode),		\
916 					 raw_inode, (einode)->xtime)
917 
918 #define EXT4_INODE_GET_XTIME_VAL(xtime, inode, raw_inode)			\
919 	(EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra) ?	\
920 		ext4_decode_extra_time((raw_inode)->xtime,				\
921 				       (raw_inode)->xtime ## _extra) :		\
922 		(struct timespec64) {						\
923 			.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime)	\
924 		})
925 
926 #define EXT4_INODE_GET_ATIME(inode, raw_inode)					\
927 do {										\
928 	inode_set_atime_to_ts(inode,						\
929 		EXT4_INODE_GET_XTIME_VAL(i_atime, inode, raw_inode));		\
930 } while (0)
931 
932 #define EXT4_INODE_GET_MTIME(inode, raw_inode)					\
933 do {										\
934 	inode_set_mtime_to_ts(inode,						\
935 		EXT4_INODE_GET_XTIME_VAL(i_mtime, inode, raw_inode));		\
936 } while (0)
937 
938 #define EXT4_INODE_GET_CTIME(inode, raw_inode)					\
939 do {										\
940 	inode_set_ctime_to_ts(inode,						\
941 		EXT4_INODE_GET_XTIME_VAL(i_ctime, inode, raw_inode));		\
942 } while (0)
943 
944 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode)				\
945 do {										\
946 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) 			\
947 		(einode)->xtime =						\
948 			EXT4_INODE_GET_XTIME_VAL(xtime, &(einode->vfs_inode),	\
949 						 raw_inode);			\
950 	else									\
951 		(einode)->xtime = (struct timespec64){0, 0};			\
952 } while (0)
953 
954 #define i_disk_version osd1.linux1.l_i_version
955 
956 #if defined(__KERNEL__) || defined(__linux__)
957 #define i_reserved1	osd1.linux1.l_i_reserved1
958 #define i_file_acl_high	osd2.linux2.l_i_file_acl_high
959 #define i_blocks_high	osd2.linux2.l_i_blocks_high
960 #define i_uid_low	i_uid
961 #define i_gid_low	i_gid
962 #define i_uid_high	osd2.linux2.l_i_uid_high
963 #define i_gid_high	osd2.linux2.l_i_gid_high
964 #define i_checksum_lo	osd2.linux2.l_i_checksum_lo
965 
966 #elif defined(__GNU__)
967 
968 #define i_translator	osd1.hurd1.h_i_translator
969 #define i_uid_high	osd2.hurd2.h_i_uid_high
970 #define i_gid_high	osd2.hurd2.h_i_gid_high
971 #define i_author	osd2.hurd2.h_i_author
972 
973 #elif defined(__masix__)
974 
975 #define i_reserved1	osd1.masix1.m_i_reserved1
976 #define i_file_acl_high	osd2.masix2.m_i_file_acl_high
977 #define i_reserved2	osd2.masix2.m_i_reserved2
978 
979 #endif /* defined(__KERNEL__) || defined(__linux__) */
980 
981 #include "extents_status.h"
982 #include "fast_commit.h"
983 
984 /*
985  * Lock subclasses for i_data_sem in the ext4_inode_info structure.
986  *
987  * These are needed to avoid lockdep false positives when we need to
988  * allocate blocks to the quota inode during ext4_map_blocks(), while
989  * holding i_data_sem for a normal (non-quota) inode.  Since we don't
990  * do quota tracking for the quota inode, this avoids deadlock (as
991  * well as infinite recursion, since it isn't turtles all the way
992  * down...)
993  *
994  *  I_DATA_SEM_NORMAL - Used for most inodes
995  *  I_DATA_SEM_OTHER  - Used by move_inode.c for the second normal inode
996  *			  where the second inode has larger inode number
997  *			  than the first
998  *  I_DATA_SEM_QUOTA  - Used for quota inodes only
999  *  I_DATA_SEM_EA     - Used for ea_inodes only
1000  */
1001 enum {
1002 	I_DATA_SEM_NORMAL = 0,
1003 	I_DATA_SEM_OTHER,
1004 	I_DATA_SEM_QUOTA,
1005 	I_DATA_SEM_EA
1006 };
1007 
1008 
1009 /*
1010  * fourth extended file system inode data in memory
1011  */
1012 struct ext4_inode_info {
1013 	__le32	i_data[15];	/* unconverted */
1014 	__u32	i_dtime;
1015 	ext4_fsblk_t	i_file_acl;
1016 
1017 	/*
1018 	 * i_block_group is the number of the block group which contains
1019 	 * this file's inode.  Constant across the lifetime of the inode,
1020 	 * it is used for making block allocation decisions - we try to
1021 	 * place a file's data blocks near its inode block, and new inodes
1022 	 * near to their parent directory's inode.
1023 	 */
1024 	ext4_group_t	i_block_group;
1025 	ext4_lblk_t	i_dir_start_lookup;
1026 #if (BITS_PER_LONG < 64)
1027 	unsigned long	i_state_flags;		/* Dynamic state flags */
1028 #endif
1029 	unsigned long	i_flags;
1030 
1031 	/*
1032 	 * Extended attributes can be read independently of the main file
1033 	 * data. Taking i_rwsem even when reading would cause contention
1034 	 * between readers of EAs and writers of regular file data, so
1035 	 * instead we synchronize on xattr_sem when reading or changing
1036 	 * EAs.
1037 	 */
1038 	struct rw_semaphore xattr_sem;
1039 
1040 	/*
1041 	 * Inodes with EXT4_STATE_ORPHAN_FILE use i_orphan_idx. Otherwise
1042 	 * i_orphan is used.
1043 	 */
1044 	union {
1045 		struct list_head i_orphan;	/* unlinked but open inodes */
1046 		unsigned int i_orphan_idx;	/* Index in orphan file */
1047 	};
1048 
1049 	/* Fast commit related info */
1050 
1051 	/* For tracking dentry create updates */
1052 	struct list_head i_fc_dilist;
1053 	struct list_head i_fc_list;	/*
1054 					 * inodes that need fast commit
1055 					 * protected by sbi->s_fc_lock.
1056 					 */
1057 
1058 	/* Start of lblk range that needs to be committed in this fast commit */
1059 	ext4_lblk_t i_fc_lblk_start;
1060 
1061 	/* End of lblk range that needs to be committed in this fast commit */
1062 	ext4_lblk_t i_fc_lblk_len;
1063 
1064 	/* Number of ongoing updates on this inode */
1065 	atomic_t  i_fc_updates;
1066 
1067 	spinlock_t i_raw_lock;	/* protects updates to the raw inode */
1068 
1069 	/* Fast commit wait queue for this inode */
1070 	wait_queue_head_t i_fc_wait;
1071 
1072 	/* Protect concurrent accesses on i_fc_lblk_start, i_fc_lblk_len */
1073 	struct mutex i_fc_lock;
1074 
1075 	/*
1076 	 * i_disksize keeps track of what the inode size is ON DISK, not
1077 	 * in memory.  During truncate, i_size is set to the new size by
1078 	 * the VFS prior to calling ext4_truncate(), but the filesystem won't
1079 	 * set i_disksize to 0 until the truncate is actually under way.
1080 	 *
1081 	 * The intent is that i_disksize always represents the blocks which
1082 	 * are used by this file.  This allows recovery to restart truncate
1083 	 * on orphans if we crash during truncate.  We actually write i_disksize
1084 	 * into the on-disk inode when writing inodes out, instead of i_size.
1085 	 *
1086 	 * The only time when i_disksize and i_size may be different is when
1087 	 * a truncate is in progress.  The only things which change i_disksize
1088 	 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
1089 	 */
1090 	loff_t	i_disksize;
1091 
1092 	/*
1093 	 * i_data_sem is for serialising ext4_truncate() against
1094 	 * ext4_getblock().  In the 2.4 ext2 design, great chunks of inode's
1095 	 * data tree are chopped off during truncate. We can't do that in
1096 	 * ext4 because whenever we perform intermediate commits during
1097 	 * truncate, the inode and all the metadata blocks *must* be in a
1098 	 * consistent state which allows truncation of the orphans to restart
1099 	 * during recovery.  Hence we must fix the get_block-vs-truncate race
1100 	 * by other means, so we have i_data_sem.
1101 	 */
1102 	struct rw_semaphore i_data_sem;
1103 	struct inode vfs_inode;
1104 	struct jbd2_inode *jinode;
1105 
1106 	/*
1107 	 * File creation time. Its function is same as that of
1108 	 * struct timespec64 i_{a,c,m}time in the generic inode.
1109 	 */
1110 	struct timespec64 i_crtime;
1111 
1112 	/* mballoc */
1113 	atomic_t i_prealloc_active;
1114 
1115 	/* allocation reservation info for delalloc */
1116 	/* In case of bigalloc, this refer to clusters rather than blocks */
1117 	unsigned int i_reserved_data_blocks;
1118 	struct rb_root i_prealloc_node;
1119 	rwlock_t i_prealloc_lock;
1120 
1121 	/* extents status tree */
1122 	struct ext4_es_tree i_es_tree;
1123 	rwlock_t i_es_lock;
1124 	struct list_head i_es_list;
1125 	unsigned int i_es_all_nr;	/* protected by i_es_lock */
1126 	unsigned int i_es_shk_nr;	/* protected by i_es_lock */
1127 	ext4_lblk_t i_es_shrink_lblk;	/* Offset where we start searching for
1128 					   extents to shrink. Protected by
1129 					   i_es_lock  */
1130 
1131 	/* ialloc */
1132 	ext4_group_t	i_last_alloc_group;
1133 
1134 	/* pending cluster reservations for bigalloc file systems */
1135 	struct ext4_pending_tree i_pending_tree;
1136 
1137 	/* on-disk additional length */
1138 	__u16 i_extra_isize;
1139 
1140 	/* Indicate the inline data space. */
1141 	u16 i_inline_off;
1142 	u16 i_inline_size;
1143 
1144 #ifdef CONFIG_QUOTA
1145 	/* quota space reservation, managed internally by quota code */
1146 	qsize_t i_reserved_quota;
1147 #endif
1148 	spinlock_t i_block_reservation_lock;
1149 
1150 	/* Lock protecting lists below */
1151 	spinlock_t i_completed_io_lock;
1152 	/*
1153 	 * Completed IOs that need unwritten extents handling and have
1154 	 * transaction reserved
1155 	 */
1156 	struct list_head i_rsv_conversion_list;
1157 	struct work_struct i_rsv_conversion_work;
1158 
1159 	/*
1160 	 * Transactions that contain inode's metadata needed to complete
1161 	 * fsync and fdatasync, respectively.
1162 	 */
1163 	tid_t i_sync_tid;
1164 	tid_t i_datasync_tid;
1165 
1166 #ifdef CONFIG_QUOTA
1167 	struct dquot __rcu *i_dquot[MAXQUOTAS];
1168 #endif
1169 
1170 	/* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1171 	__u32 i_csum_seed;
1172 
1173 	kprojid_t i_projid;
1174 };
1175 
1176 /*
1177  * File system states
1178  */
1179 #define	EXT4_VALID_FS			0x0001	/* Unmounted cleanly */
1180 #define	EXT4_ERROR_FS			0x0002	/* Errors detected */
1181 #define	EXT4_ORPHAN_FS			0x0004	/* Orphans being recovered */
1182 #define EXT4_FC_REPLAY			0x0020	/* Fast commit replay ongoing */
1183 
1184 /*
1185  * Misc. filesystem flags
1186  */
1187 #define EXT2_FLAGS_SIGNED_HASH		0x0001  /* Signed dirhash in use */
1188 #define EXT2_FLAGS_UNSIGNED_HASH	0x0002  /* Unsigned dirhash in use */
1189 #define EXT2_FLAGS_TEST_FILESYS		0x0004	/* to test development code */
1190 
1191 /*
1192  * Mount flags set via mount options or defaults
1193  */
1194 #define EXT4_MOUNT_NO_MBCACHE		0x00001 /* Do not use mbcache */
1195 #define EXT4_MOUNT_GRPID		0x00004	/* Create files with directory's group */
1196 #define EXT4_MOUNT_DEBUG		0x00008	/* Some debugging messages */
1197 #define EXT4_MOUNT_ERRORS_CONT		0x00010	/* Continue on errors */
1198 #define EXT4_MOUNT_ERRORS_RO		0x00020	/* Remount fs ro on errors */
1199 #define EXT4_MOUNT_ERRORS_PANIC		0x00040	/* Panic on errors */
1200 #define EXT4_MOUNT_ERRORS_MASK		0x00070
1201 #define EXT4_MOUNT_MINIX_DF		0x00080	/* Mimics the Minix statfs */
1202 #define EXT4_MOUNT_NOLOAD		0x00100	/* Don't use existing journal*/
1203 #ifdef CONFIG_FS_DAX
1204 #define EXT4_MOUNT_DAX_ALWAYS		0x00200	/* Direct Access */
1205 #else
1206 #define EXT4_MOUNT_DAX_ALWAYS		0
1207 #endif
1208 #define EXT4_MOUNT_DATA_FLAGS		0x00C00	/* Mode for data writes: */
1209 #define EXT4_MOUNT_JOURNAL_DATA		0x00400	/* Write data to journal */
1210 #define EXT4_MOUNT_ORDERED_DATA		0x00800	/* Flush data before commit */
1211 #define EXT4_MOUNT_WRITEBACK_DATA	0x00C00	/* No data ordering */
1212 #define EXT4_MOUNT_UPDATE_JOURNAL	0x01000	/* Update the journal format */
1213 #define EXT4_MOUNT_NO_UID32		0x02000  /* Disable 32-bit UIDs */
1214 #define EXT4_MOUNT_XATTR_USER		0x04000	/* Extended user attributes */
1215 #define EXT4_MOUNT_POSIX_ACL		0x08000	/* POSIX Access Control Lists */
1216 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC	0x10000	/* No auto delalloc mapping */
1217 #define EXT4_MOUNT_BARRIER		0x20000 /* Use block barriers */
1218 #define EXT4_MOUNT_QUOTA		0x40000 /* Some quota option set */
1219 #define EXT4_MOUNT_USRQUOTA		0x80000 /* "old" user quota,
1220 						 * enable enforcement for hidden
1221 						 * quota files */
1222 #define EXT4_MOUNT_GRPQUOTA		0x100000 /* "old" group quota, enable
1223 						  * enforcement for hidden quota
1224 						  * files */
1225 #define EXT4_MOUNT_PRJQUOTA		0x200000 /* Enable project quota
1226 						  * enforcement */
1227 #define EXT4_MOUNT_DIOREAD_NOLOCK	0x400000 /* Enable support for dio read nolocking */
1228 #define EXT4_MOUNT_JOURNAL_CHECKSUM	0x800000 /* Journal checksums */
1229 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT	0x1000000 /* Journal Async Commit */
1230 #define EXT4_MOUNT_WARN_ON_ERROR	0x2000000 /* Trigger WARN_ON on error */
1231 #define EXT4_MOUNT_NO_PREFETCH_BLOCK_BITMAPS 0x4000000
1232 #define EXT4_MOUNT_DELALLOC		0x8000000 /* Delalloc support */
1233 #define EXT4_MOUNT_DATA_ERR_ABORT	0x10000000 /* Abort on file data write */
1234 #define EXT4_MOUNT_BLOCK_VALIDITY	0x20000000 /* Block validity checking */
1235 #define EXT4_MOUNT_DISCARD		0x40000000 /* Issue DISCARD requests */
1236 #define EXT4_MOUNT_INIT_INODE_TABLE	0x80000000 /* Initialize uninitialized itables */
1237 
1238 /*
1239  * Mount flags set either automatically (could not be set by mount option)
1240  * based on per file system feature or property or in special cases such as
1241  * distinguishing between explicit mount option definition and default.
1242  */
1243 #define EXT4_MOUNT2_EXPLICIT_DELALLOC	0x00000001 /* User explicitly
1244 						      specified delalloc */
1245 #define EXT4_MOUNT2_STD_GROUP_SIZE	0x00000002 /* We have standard group
1246 						      size of blocksize * 8
1247 						      blocks */
1248 #define EXT4_MOUNT2_HURD_COMPAT		0x00000004 /* Support HURD-castrated
1249 						      file systems */
1250 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM	0x00000008 /* User explicitly
1251 						specified journal checksum */
1252 
1253 #define EXT4_MOUNT2_JOURNAL_FAST_COMMIT	0x00000010 /* Journal fast commit */
1254 #define EXT4_MOUNT2_DAX_NEVER		0x00000020 /* Do not allow Direct Access */
1255 #define EXT4_MOUNT2_DAX_INODE		0x00000040 /* For printing options only */
1256 #define EXT4_MOUNT2_MB_OPTIMIZE_SCAN	0x00000080 /* Optimize group
1257 						    * scanning in mballoc
1258 						    */
1259 #define EXT4_MOUNT2_ABORT		0x00000100 /* Abort filesystem */
1260 
1261 #define clear_opt(sb, opt)		EXT4_SB(sb)->s_mount_opt &= \
1262 						~EXT4_MOUNT_##opt
1263 #define set_opt(sb, opt)		EXT4_SB(sb)->s_mount_opt |= \
1264 						EXT4_MOUNT_##opt
1265 #define test_opt(sb, opt)		(EXT4_SB(sb)->s_mount_opt & \
1266 					 EXT4_MOUNT_##opt)
1267 
1268 #define clear_opt2(sb, opt)		EXT4_SB(sb)->s_mount_opt2 &= \
1269 						~EXT4_MOUNT2_##opt
1270 #define set_opt2(sb, opt)		EXT4_SB(sb)->s_mount_opt2 |= \
1271 						EXT4_MOUNT2_##opt
1272 #define test_opt2(sb, opt)		(EXT4_SB(sb)->s_mount_opt2 & \
1273 					 EXT4_MOUNT2_##opt)
1274 
1275 #define ext4_test_and_set_bit		__test_and_set_bit_le
1276 #define ext4_set_bit			__set_bit_le
1277 #define ext4_test_and_clear_bit		__test_and_clear_bit_le
1278 #define ext4_clear_bit			__clear_bit_le
1279 #define ext4_test_bit			test_bit_le
1280 #define ext4_find_next_zero_bit		find_next_zero_bit_le
1281 #define ext4_find_next_bit		find_next_bit_le
1282 
1283 extern void mb_set_bits(void *bm, int cur, int len);
1284 
1285 /*
1286  * Maximal mount counts between two filesystem checks
1287  */
1288 #define EXT4_DFL_MAX_MNT_COUNT		20	/* Allow 20 mounts */
1289 #define EXT4_DFL_CHECKINTERVAL		0	/* Don't use interval check */
1290 
1291 /*
1292  * Behaviour when detecting errors
1293  */
1294 #define EXT4_ERRORS_CONTINUE		1	/* Continue execution */
1295 #define EXT4_ERRORS_RO			2	/* Remount fs read-only */
1296 #define EXT4_ERRORS_PANIC		3	/* Panic */
1297 #define EXT4_ERRORS_DEFAULT		EXT4_ERRORS_CONTINUE
1298 
1299 /* Metadata checksum algorithm codes */
1300 #define EXT4_CRC32C_CHKSUM		1
1301 
1302 #define EXT4_LABEL_MAX			16
1303 
1304 /*
1305  * Structure of the super block
1306  */
1307 struct ext4_super_block {
1308 /*00*/	__le32	s_inodes_count;		/* Inodes count */
1309 	__le32	s_blocks_count_lo;	/* Blocks count */
1310 	__le32	s_r_blocks_count_lo;	/* Reserved blocks count */
1311 	__le32	s_free_blocks_count_lo;	/* Free blocks count */
1312 /*10*/	__le32	s_free_inodes_count;	/* Free inodes count */
1313 	__le32	s_first_data_block;	/* First Data Block */
1314 	__le32	s_log_block_size;	/* Block size */
1315 	__le32	s_log_cluster_size;	/* Allocation cluster size */
1316 /*20*/	__le32	s_blocks_per_group;	/* # Blocks per group */
1317 	__le32	s_clusters_per_group;	/* # Clusters per group */
1318 	__le32	s_inodes_per_group;	/* # Inodes per group */
1319 	__le32	s_mtime;		/* Mount time */
1320 /*30*/	__le32	s_wtime;		/* Write time */
1321 	__le16	s_mnt_count;		/* Mount count */
1322 	__le16	s_max_mnt_count;	/* Maximal mount count */
1323 	__le16	s_magic;		/* Magic signature */
1324 	__le16	s_state;		/* File system state */
1325 	__le16	s_errors;		/* Behaviour when detecting errors */
1326 	__le16	s_minor_rev_level;	/* minor revision level */
1327 /*40*/	__le32	s_lastcheck;		/* time of last check */
1328 	__le32	s_checkinterval;	/* max. time between checks */
1329 	__le32	s_creator_os;		/* OS */
1330 	__le32	s_rev_level;		/* Revision level */
1331 /*50*/	__le16	s_def_resuid;		/* Default uid for reserved blocks */
1332 	__le16	s_def_resgid;		/* Default gid for reserved blocks */
1333 	/*
1334 	 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1335 	 *
1336 	 * Note: the difference between the compatible feature set and
1337 	 * the incompatible feature set is that if there is a bit set
1338 	 * in the incompatible feature set that the kernel doesn't
1339 	 * know about, it should refuse to mount the filesystem.
1340 	 *
1341 	 * e2fsck's requirements are more strict; if it doesn't know
1342 	 * about a feature in either the compatible or incompatible
1343 	 * feature set, it must abort and not try to meddle with
1344 	 * things it doesn't understand...
1345 	 */
1346 	__le32	s_first_ino;		/* First non-reserved inode */
1347 	__le16  s_inode_size;		/* size of inode structure */
1348 	__le16	s_block_group_nr;	/* block group # of this superblock */
1349 	__le32	s_feature_compat;	/* compatible feature set */
1350 /*60*/	__le32	s_feature_incompat;	/* incompatible feature set */
1351 	__le32	s_feature_ro_compat;	/* readonly-compatible feature set */
1352 /*68*/	__u8	s_uuid[16];		/* 128-bit uuid for volume */
1353 /*78*/	char	s_volume_name[EXT4_LABEL_MAX] __nonstring; /* volume name */
1354 /*88*/	char	s_last_mounted[64] __nonstring;	/* directory where last mounted */
1355 /*C8*/	__le32	s_algorithm_usage_bitmap; /* For compression */
1356 	/*
1357 	 * Performance hints.  Directory preallocation should only
1358 	 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1359 	 */
1360 	__u8	s_prealloc_blocks;	/* Nr of blocks to try to preallocate*/
1361 	__u8	s_prealloc_dir_blocks;	/* Nr to preallocate for dirs */
1362 	__le16	s_reserved_gdt_blocks;	/* Per group desc for online growth */
1363 	/*
1364 	 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1365 	 */
1366 /*D0*/	__u8	s_journal_uuid[16];	/* uuid of journal superblock */
1367 /*E0*/	__le32	s_journal_inum;		/* inode number of journal file */
1368 	__le32	s_journal_dev;		/* device number of journal file */
1369 	__le32	s_last_orphan;		/* start of list of inodes to delete */
1370 	__le32	s_hash_seed[4];		/* HTREE hash seed */
1371 	__u8	s_def_hash_version;	/* Default hash version to use */
1372 	__u8	s_jnl_backup_type;
1373 	__le16  s_desc_size;		/* size of group descriptor */
1374 /*100*/	__le32	s_default_mount_opts;
1375 	__le32	s_first_meta_bg;	/* First metablock block group */
1376 	__le32	s_mkfs_time;		/* When the filesystem was created */
1377 	__le32	s_jnl_blocks[17];	/* Backup of the journal inode */
1378 	/* 64bit support valid if EXT4_FEATURE_INCOMPAT_64BIT */
1379 /*150*/	__le32	s_blocks_count_hi;	/* Blocks count */
1380 	__le32	s_r_blocks_count_hi;	/* Reserved blocks count */
1381 	__le32	s_free_blocks_count_hi;	/* Free blocks count */
1382 	__le16	s_min_extra_isize;	/* All inodes have at least # bytes */
1383 	__le16	s_want_extra_isize; 	/* New inodes should reserve # bytes */
1384 	__le32	s_flags;		/* Miscellaneous flags */
1385 	__le16  s_raid_stride;		/* RAID stride */
1386 	__le16  s_mmp_update_interval;  /* # seconds to wait in MMP checking */
1387 	__le64  s_mmp_block;            /* Block for multi-mount protection */
1388 	__le32  s_raid_stripe_width;    /* blocks on all data disks (N*stride)*/
1389 	__u8	s_log_groups_per_flex;  /* FLEX_BG group size */
1390 	__u8	s_checksum_type;	/* metadata checksum algorithm used */
1391 	__u8	s_encryption_level;	/* versioning level for encryption */
1392 	__u8	s_reserved_pad;		/* Padding to next 32bits */
1393 	__le64	s_kbytes_written;	/* nr of lifetime kilobytes written */
1394 	__le32	s_snapshot_inum;	/* Inode number of active snapshot */
1395 	__le32	s_snapshot_id;		/* sequential ID of active snapshot */
1396 	__le64	s_snapshot_r_blocks_count; /* reserved blocks for active
1397 					      snapshot's future use */
1398 	__le32	s_snapshot_list;	/* inode number of the head of the
1399 					   on-disk snapshot list */
1400 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1401 	__le32	s_error_count;		/* number of fs errors */
1402 	__le32	s_first_error_time;	/* first time an error happened */
1403 	__le32	s_first_error_ino;	/* inode involved in first error */
1404 	__le64	s_first_error_block;	/* block involved of first error */
1405 	__u8	s_first_error_func[32] __nonstring;	/* function where the error happened */
1406 	__le32	s_first_error_line;	/* line number where error happened */
1407 	__le32	s_last_error_time;	/* most recent time of an error */
1408 	__le32	s_last_error_ino;	/* inode involved in last error */
1409 	__le32	s_last_error_line;	/* line number where error happened */
1410 	__le64	s_last_error_block;	/* block involved of last error */
1411 	__u8	s_last_error_func[32] __nonstring;	/* function where the error happened */
1412 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1413 	__u8	s_mount_opts[64];
1414 	__le32	s_usr_quota_inum;	/* inode for tracking user quota */
1415 	__le32	s_grp_quota_inum;	/* inode for tracking group quota */
1416 	__le32	s_overhead_clusters;	/* overhead blocks/clusters in fs */
1417 	__le32	s_backup_bgs[2];	/* groups with sparse_super2 SBs */
1418 	__u8	s_encrypt_algos[4];	/* Encryption algorithms in use  */
1419 	__u8	s_encrypt_pw_salt[16];	/* Salt used for string2key algorithm */
1420 	__le32	s_lpf_ino;		/* Location of the lost+found inode */
1421 	__le32	s_prj_quota_inum;	/* inode for tracking project quota */
1422 	__le32	s_checksum_seed;	/* crc32c(uuid) if csum_seed set */
1423 	__u8	s_wtime_hi;
1424 	__u8	s_mtime_hi;
1425 	__u8	s_mkfs_time_hi;
1426 	__u8	s_lastcheck_hi;
1427 	__u8	s_first_error_time_hi;
1428 	__u8	s_last_error_time_hi;
1429 	__u8	s_first_error_errcode;
1430 	__u8    s_last_error_errcode;
1431 	__le16  s_encoding;		/* Filename charset encoding */
1432 	__le16  s_encoding_flags;	/* Filename charset encoding flags */
1433 	__le32  s_orphan_file_inum;	/* Inode for tracking orphan inodes */
1434 	__le32	s_reserved[94];		/* Padding to the end of the block */
1435 	__le32	s_checksum;		/* crc32c(superblock) */
1436 };
1437 
1438 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1439 
1440 #ifdef __KERNEL__
1441 
1442 /* Number of quota types we support */
1443 #define EXT4_MAXQUOTAS 3
1444 
1445 #define EXT4_ENC_UTF8_12_1	1
1446 
1447 /* Types of ext4 journal triggers */
1448 enum ext4_journal_trigger_type {
1449 	EXT4_JTR_ORPHAN_FILE,
1450 	EXT4_JTR_NONE	/* This must be the last entry for indexing to work! */
1451 };
1452 
1453 #define EXT4_JOURNAL_TRIGGER_COUNT EXT4_JTR_NONE
1454 
1455 struct ext4_journal_trigger {
1456 	struct jbd2_buffer_trigger_type tr_triggers;
1457 	struct super_block *sb;
1458 };
1459 
EXT4_TRIGGER(struct jbd2_buffer_trigger_type * trigger)1460 static inline struct ext4_journal_trigger *EXT4_TRIGGER(
1461 				struct jbd2_buffer_trigger_type *trigger)
1462 {
1463 	return container_of(trigger, struct ext4_journal_trigger, tr_triggers);
1464 }
1465 
1466 #define EXT4_ORPHAN_BLOCK_MAGIC 0x0b10ca04
1467 
1468 /* Structure at the tail of orphan block */
1469 struct ext4_orphan_block_tail {
1470 	__le32 ob_magic;
1471 	__le32 ob_checksum;
1472 };
1473 
ext4_inodes_per_orphan_block(struct super_block * sb)1474 static inline int ext4_inodes_per_orphan_block(struct super_block *sb)
1475 {
1476 	return (sb->s_blocksize - sizeof(struct ext4_orphan_block_tail)) /
1477 			sizeof(u32);
1478 }
1479 
1480 struct ext4_orphan_block {
1481 	atomic_t ob_free_entries;	/* Number of free orphan entries in block */
1482 	struct buffer_head *ob_bh;	/* Buffer for orphan block */
1483 };
1484 
1485 /*
1486  * Info about orphan file.
1487  */
1488 struct ext4_orphan_info {
1489 	int of_blocks;			/* Number of orphan blocks in a file */
1490 	__u32 of_csum_seed;		/* Checksum seed for orphan file */
1491 	struct ext4_orphan_block *of_binfo;	/* Array with info about orphan
1492 						 * file blocks */
1493 };
1494 
1495 /*
1496  * fourth extended-fs super-block data in memory
1497  */
1498 struct ext4_sb_info {
1499 	unsigned long s_desc_size;	/* Size of a group descriptor in bytes */
1500 	unsigned long s_inodes_per_block;/* Number of inodes per block */
1501 	unsigned long s_blocks_per_group;/* Number of blocks in a group */
1502 	unsigned long s_clusters_per_group; /* Number of clusters in a group */
1503 	unsigned long s_inodes_per_group;/* Number of inodes in a group */
1504 	unsigned long s_itb_per_group;	/* Number of inode table blocks per group */
1505 	unsigned long s_gdb_count;	/* Number of group descriptor blocks */
1506 	unsigned long s_desc_per_block;	/* Number of group descriptors per block */
1507 	ext4_group_t s_groups_count;	/* Number of groups in the fs */
1508 	ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1509 	unsigned long s_overhead;  /* # of fs overhead clusters */
1510 	unsigned int s_cluster_ratio;	/* Number of blocks per cluster */
1511 	unsigned int s_cluster_bits;	/* log2 of s_cluster_ratio */
1512 	loff_t s_bitmap_maxbytes;	/* max bytes for bitmap files */
1513 	struct buffer_head * s_sbh;	/* Buffer containing the super block */
1514 	struct ext4_super_block *s_es;	/* Pointer to the super block in the buffer */
1515 	/* Array of bh's for the block group descriptors */
1516 	struct buffer_head * __rcu *s_group_desc;
1517 	unsigned int s_mount_opt;
1518 	unsigned int s_mount_opt2;
1519 	unsigned long s_mount_flags;
1520 	unsigned int s_def_mount_opt;
1521 	unsigned int s_def_mount_opt2;
1522 	ext4_fsblk_t s_sb_block;
1523 	atomic64_t s_resv_clusters;
1524 	kuid_t s_resuid;
1525 	kgid_t s_resgid;
1526 	unsigned short s_mount_state;
1527 	unsigned short s_pad;
1528 	int s_addr_per_block_bits;
1529 	int s_desc_per_block_bits;
1530 	int s_inode_size;
1531 	int s_first_ino;
1532 	unsigned int s_inode_readahead_blks;
1533 	unsigned int s_inode_goal;
1534 	u32 s_hash_seed[4];
1535 	int s_def_hash_version;
1536 	int s_hash_unsigned;	/* 3 if hash should be unsigned, 0 if not */
1537 	struct percpu_counter s_freeclusters_counter;
1538 	struct percpu_counter s_freeinodes_counter;
1539 	struct percpu_counter s_dirs_counter;
1540 	struct percpu_counter s_dirtyclusters_counter;
1541 	struct percpu_counter s_sra_exceeded_retry_limit;
1542 	struct blockgroup_lock *s_blockgroup_lock;
1543 	struct proc_dir_entry *s_proc;
1544 	struct kobject s_kobj;
1545 	struct completion s_kobj_unregister;
1546 	struct super_block *s_sb;
1547 	struct buffer_head *s_mmp_bh;
1548 
1549 	/* Journaling */
1550 	struct journal_s *s_journal;
1551 	unsigned long s_ext4_flags;		/* Ext4 superblock flags */
1552 	struct mutex s_orphan_lock;	/* Protects on disk list changes */
1553 	struct list_head s_orphan;	/* List of orphaned inodes in on disk
1554 					   list */
1555 	struct ext4_orphan_info s_orphan_info;
1556 	unsigned long s_commit_interval;
1557 	u32 s_max_batch_time;
1558 	u32 s_min_batch_time;
1559 	struct file *s_journal_bdev_file;
1560 #ifdef CONFIG_QUOTA
1561 	/* Names of quota files with journalled quota */
1562 	char __rcu *s_qf_names[EXT4_MAXQUOTAS];
1563 	int s_jquota_fmt;			/* Format of quota to use */
1564 #endif
1565 	unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1566 	struct ext4_system_blocks __rcu *s_system_blks;
1567 
1568 #ifdef EXTENTS_STATS
1569 	/* ext4 extents stats */
1570 	unsigned long s_ext_min;
1571 	unsigned long s_ext_max;
1572 	unsigned long s_depth_max;
1573 	spinlock_t s_ext_stats_lock;
1574 	unsigned long s_ext_blocks;
1575 	unsigned long s_ext_extents;
1576 #endif
1577 
1578 	/* for buddy allocator */
1579 	struct ext4_group_info ** __rcu *s_group_info;
1580 	struct inode *s_buddy_cache;
1581 	spinlock_t s_md_lock;
1582 	unsigned short *s_mb_offsets;
1583 	unsigned int *s_mb_maxs;
1584 	unsigned int s_group_info_size;
1585 	unsigned int s_mb_free_pending;
1586 	struct list_head s_freed_data_list[2];	/* List of blocks to be freed
1587 						   after commit completed */
1588 	struct list_head s_discard_list;
1589 	struct work_struct s_discard_work;
1590 	atomic_t s_retry_alloc_pending;
1591 	struct list_head *s_mb_avg_fragment_size;
1592 	rwlock_t *s_mb_avg_fragment_size_locks;
1593 	struct list_head *s_mb_largest_free_orders;
1594 	rwlock_t *s_mb_largest_free_orders_locks;
1595 
1596 	/* tunables */
1597 	unsigned long s_stripe;
1598 	unsigned int s_mb_max_linear_groups;
1599 	unsigned int s_mb_stream_request;
1600 	unsigned int s_mb_max_to_scan;
1601 	unsigned int s_mb_min_to_scan;
1602 	unsigned int s_mb_stats;
1603 	unsigned int s_mb_order2_reqs;
1604 	unsigned int s_mb_group_prealloc;
1605 	unsigned int s_max_dir_size_kb;
1606 	/* where last allocation was done - for stream allocation */
1607 	unsigned long s_mb_last_group;
1608 	unsigned long s_mb_last_start;
1609 	unsigned int s_mb_prefetch;
1610 	unsigned int s_mb_prefetch_limit;
1611 	unsigned int s_mb_best_avail_max_trim_order;
1612 	unsigned int s_sb_update_sec;
1613 	unsigned int s_sb_update_kb;
1614 
1615 	/* stats for buddy allocator */
1616 	atomic_t s_bal_reqs;	/* number of reqs with len > 1 */
1617 	atomic_t s_bal_success;	/* we found long enough chunks */
1618 	atomic_t s_bal_allocated;	/* in blocks */
1619 	atomic_t s_bal_ex_scanned;	/* total extents scanned */
1620 	atomic_t s_bal_cX_ex_scanned[EXT4_MB_NUM_CRS];	/* total extents scanned */
1621 	atomic_t s_bal_groups_scanned;	/* number of groups scanned */
1622 	atomic_t s_bal_goals;	/* goal hits */
1623 	atomic_t s_bal_len_goals;	/* len goal hits */
1624 	atomic_t s_bal_breaks;	/* too long searches */
1625 	atomic_t s_bal_2orders;	/* 2^order hits */
1626 	atomic_t s_bal_p2_aligned_bad_suggestions;
1627 	atomic_t s_bal_goal_fast_bad_suggestions;
1628 	atomic_t s_bal_best_avail_bad_suggestions;
1629 	atomic64_t s_bal_cX_groups_considered[EXT4_MB_NUM_CRS];
1630 	atomic64_t s_bal_cX_hits[EXT4_MB_NUM_CRS];
1631 	atomic64_t s_bal_cX_failed[EXT4_MB_NUM_CRS];		/* cX loop didn't find blocks */
1632 	atomic_t s_mb_buddies_generated;	/* number of buddies generated */
1633 	atomic64_t s_mb_generation_time;
1634 	atomic_t s_mb_lost_chunks;
1635 	atomic_t s_mb_preallocated;
1636 	atomic_t s_mb_discarded;
1637 	atomic_t s_lock_busy;
1638 
1639 	/* locality groups */
1640 	struct ext4_locality_group __percpu *s_locality_groups;
1641 
1642 	/* for write statistics */
1643 	unsigned long s_sectors_written_start;
1644 	u64 s_kbytes_written;
1645 
1646 	/* the size of zero-out chunk */
1647 	unsigned int s_extent_max_zeroout_kb;
1648 
1649 	unsigned int s_log_groups_per_flex;
1650 	struct flex_groups * __rcu *s_flex_groups;
1651 	ext4_group_t s_flex_groups_allocated;
1652 
1653 	/* workqueue for reserved extent conversions (buffered io) */
1654 	struct workqueue_struct *rsv_conversion_wq;
1655 
1656 	/* timer for periodic error stats printing */
1657 	struct timer_list s_err_report;
1658 
1659 	/* Lazy inode table initialization info */
1660 	struct ext4_li_request *s_li_request;
1661 	/* Wait multiplier for lazy initialization thread */
1662 	unsigned int s_li_wait_mult;
1663 
1664 	/* Kernel thread for multiple mount protection */
1665 	struct task_struct *s_mmp_tsk;
1666 
1667 	/* record the last minlen when FITRIM is called. */
1668 	unsigned long s_last_trim_minblks;
1669 
1670 	/* Precomputed FS UUID checksum for seeding other checksums */
1671 	__u32 s_csum_seed;
1672 
1673 	/* Reclaim extents from extent status tree */
1674 	struct shrinker *s_es_shrinker;
1675 	struct list_head s_es_list;	/* List of inodes with reclaimable extents */
1676 	long s_es_nr_inode;
1677 	struct ext4_es_stats s_es_stats;
1678 	struct mb_cache *s_ea_block_cache;
1679 	struct mb_cache *s_ea_inode_cache;
1680 	spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1681 
1682 	/* Journal triggers for checksum computation */
1683 	struct ext4_journal_trigger s_journal_triggers[EXT4_JOURNAL_TRIGGER_COUNT];
1684 
1685 	/* Ratelimit ext4 messages. */
1686 	struct ratelimit_state s_err_ratelimit_state;
1687 	struct ratelimit_state s_warning_ratelimit_state;
1688 	struct ratelimit_state s_msg_ratelimit_state;
1689 	atomic_t s_warning_count;
1690 	atomic_t s_msg_count;
1691 
1692 	/* Encryption policy for '-o test_dummy_encryption' */
1693 	struct fscrypt_dummy_policy s_dummy_enc_policy;
1694 
1695 	/*
1696 	 * Barrier between writepages ops and changing any inode's JOURNAL_DATA
1697 	 * or EXTENTS flag or between writepages ops and changing DELALLOC or
1698 	 * DIOREAD_NOLOCK mount options on remount.
1699 	 */
1700 	struct percpu_rw_semaphore s_writepages_rwsem;
1701 	struct dax_device *s_daxdev;
1702 	u64 s_dax_part_off;
1703 #ifdef CONFIG_EXT4_DEBUG
1704 	unsigned long s_simulate_fail;
1705 #endif
1706 	/* Record the errseq of the backing block device */
1707 	errseq_t s_bdev_wb_err;
1708 	spinlock_t s_bdev_wb_lock;
1709 
1710 	/* Information about errors that happened during this mount */
1711 	spinlock_t s_error_lock;
1712 	int s_add_error_count;
1713 	int s_first_error_code;
1714 	__u32 s_first_error_line;
1715 	__u32 s_first_error_ino;
1716 	__u64 s_first_error_block;
1717 	const char *s_first_error_func;
1718 	time64_t s_first_error_time;
1719 	int s_last_error_code;
1720 	__u32 s_last_error_line;
1721 	__u32 s_last_error_ino;
1722 	__u64 s_last_error_block;
1723 	const char *s_last_error_func;
1724 	time64_t s_last_error_time;
1725 	/*
1726 	 * If we are in a context where we cannot update the on-disk
1727 	 * superblock, we queue the work here.  This is used to update
1728 	 * the error information in the superblock, and for periodic
1729 	 * updates of the superblock called from the commit callback
1730 	 * function.
1731 	 */
1732 	struct work_struct s_sb_upd_work;
1733 
1734 	/* Atomic write unit values in bytes */
1735 	unsigned int s_awu_min;
1736 	unsigned int s_awu_max;
1737 
1738 	/* Ext4 fast commit sub transaction ID */
1739 	atomic_t s_fc_subtid;
1740 
1741 	/*
1742 	 * After commit starts, the main queue gets locked, and the further
1743 	 * updates get added in the staging queue.
1744 	 */
1745 #define FC_Q_MAIN	0
1746 #define FC_Q_STAGING	1
1747 	struct list_head s_fc_q[2];	/* Inodes staged for fast commit
1748 					 * that have data changes in them.
1749 					 */
1750 	struct list_head s_fc_dentry_q[2];	/* directory entry updates */
1751 	unsigned int s_fc_bytes;
1752 	/*
1753 	 * Main fast commit lock. This lock protects accesses to the
1754 	 * following fields:
1755 	 * ei->i_fc_list, s_fc_dentry_q, s_fc_q, s_fc_bytes, s_fc_bh.
1756 	 */
1757 	spinlock_t s_fc_lock;
1758 	struct buffer_head *s_fc_bh;
1759 	struct ext4_fc_stats s_fc_stats;
1760 	tid_t s_fc_ineligible_tid;
1761 #ifdef CONFIG_EXT4_DEBUG
1762 	int s_fc_debug_max_replay;
1763 #endif
1764 	struct ext4_fc_replay_state s_fc_replay_state;
1765 };
1766 
EXT4_SB(struct super_block * sb)1767 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1768 {
1769 	return sb->s_fs_info;
1770 }
EXT4_I(struct inode * inode)1771 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1772 {
1773 	return container_of(inode, struct ext4_inode_info, vfs_inode);
1774 }
1775 
ext4_writepages_down_read(struct super_block * sb)1776 static inline int ext4_writepages_down_read(struct super_block *sb)
1777 {
1778 	percpu_down_read(&EXT4_SB(sb)->s_writepages_rwsem);
1779 	return memalloc_nofs_save();
1780 }
1781 
ext4_writepages_up_read(struct super_block * sb,int ctx)1782 static inline void ext4_writepages_up_read(struct super_block *sb, int ctx)
1783 {
1784 	memalloc_nofs_restore(ctx);
1785 	percpu_up_read(&EXT4_SB(sb)->s_writepages_rwsem);
1786 }
1787 
ext4_writepages_down_write(struct super_block * sb)1788 static inline int ext4_writepages_down_write(struct super_block *sb)
1789 {
1790 	percpu_down_write(&EXT4_SB(sb)->s_writepages_rwsem);
1791 	return memalloc_nofs_save();
1792 }
1793 
ext4_writepages_up_write(struct super_block * sb,int ctx)1794 static inline void ext4_writepages_up_write(struct super_block *sb, int ctx)
1795 {
1796 	memalloc_nofs_restore(ctx);
1797 	percpu_up_write(&EXT4_SB(sb)->s_writepages_rwsem);
1798 }
1799 
ext4_valid_inum(struct super_block * sb,unsigned long ino)1800 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1801 {
1802 	return ino == EXT4_ROOT_INO ||
1803 		(ino >= EXT4_FIRST_INO(sb) &&
1804 		 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1805 }
1806 
1807 /*
1808  * Returns: sbi->field[index]
1809  * Used to access an array element from the following sbi fields which require
1810  * rcu protection to avoid dereferencing an invalid pointer due to reassignment
1811  * - s_group_desc
1812  * - s_group_info
1813  * - s_flex_group
1814  */
1815 #define sbi_array_rcu_deref(sbi, field, index)				   \
1816 ({									   \
1817 	typeof(*((sbi)->field)) _v;					   \
1818 	rcu_read_lock();						   \
1819 	_v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index];	   \
1820 	rcu_read_unlock();						   \
1821 	_v;								   \
1822 })
1823 
1824 /*
1825  * run-time mount flags
1826  */
1827 enum {
1828 	EXT4_MF_MNTDIR_SAMPLED,
1829 	EXT4_MF_FC_INELIGIBLE,	/* Fast commit ineligible */
1830 	EXT4_MF_JOURNAL_DESTROY	/* Journal is in process of destroying */
1831 };
1832 
ext4_set_mount_flag(struct super_block * sb,int bit)1833 static inline void ext4_set_mount_flag(struct super_block *sb, int bit)
1834 {
1835 	set_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1836 }
1837 
ext4_clear_mount_flag(struct super_block * sb,int bit)1838 static inline void ext4_clear_mount_flag(struct super_block *sb, int bit)
1839 {
1840 	clear_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1841 }
1842 
ext4_test_mount_flag(struct super_block * sb,int bit)1843 static inline int ext4_test_mount_flag(struct super_block *sb, int bit)
1844 {
1845 	return test_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1846 }
1847 
1848 
1849 /*
1850  * Simulate_fail codes
1851  */
1852 #define EXT4_SIM_BBITMAP_EIO	1
1853 #define EXT4_SIM_BBITMAP_CRC	2
1854 #define EXT4_SIM_IBITMAP_EIO	3
1855 #define EXT4_SIM_IBITMAP_CRC	4
1856 #define EXT4_SIM_INODE_EIO	5
1857 #define EXT4_SIM_INODE_CRC	6
1858 #define EXT4_SIM_DIRBLOCK_EIO	7
1859 #define EXT4_SIM_DIRBLOCK_CRC	8
1860 
ext4_simulate_fail(struct super_block * sb,unsigned long code)1861 static inline bool ext4_simulate_fail(struct super_block *sb,
1862 				     unsigned long code)
1863 {
1864 #ifdef CONFIG_EXT4_DEBUG
1865 	struct ext4_sb_info *sbi = EXT4_SB(sb);
1866 
1867 	if (unlikely(sbi->s_simulate_fail == code)) {
1868 		sbi->s_simulate_fail = 0;
1869 		return true;
1870 	}
1871 #endif
1872 	return false;
1873 }
1874 
1875 /*
1876  * Error number codes for s_{first,last}_error_errno
1877  *
1878  * Linux errno numbers are architecture specific, so we need to translate
1879  * them into something which is architecture independent.   We don't define
1880  * codes for all errno's; just the ones which are most likely to be the cause
1881  * of an ext4_error() call.
1882  */
1883 #define EXT4_ERR_UNKNOWN	 1
1884 #define EXT4_ERR_EIO		 2
1885 #define EXT4_ERR_ENOMEM		 3
1886 #define EXT4_ERR_EFSBADCRC	 4
1887 #define EXT4_ERR_EFSCORRUPTED	 5
1888 #define EXT4_ERR_ENOSPC		 6
1889 #define EXT4_ERR_ENOKEY		 7
1890 #define EXT4_ERR_EROFS		 8
1891 #define EXT4_ERR_EFBIG		 9
1892 #define EXT4_ERR_EEXIST		10
1893 #define EXT4_ERR_ERANGE		11
1894 #define EXT4_ERR_EOVERFLOW	12
1895 #define EXT4_ERR_EBUSY		13
1896 #define EXT4_ERR_ENOTDIR	14
1897 #define EXT4_ERR_ENOTEMPTY	15
1898 #define EXT4_ERR_ESHUTDOWN	16
1899 #define EXT4_ERR_EFAULT		17
1900 
1901 /*
1902  * Inode dynamic state flags
1903  */
1904 enum {
1905 	EXT4_STATE_NEW,			/* inode is newly created */
1906 	EXT4_STATE_XATTR,		/* has in-inode xattrs */
1907 	EXT4_STATE_NO_EXPAND,		/* No space for expansion */
1908 	EXT4_STATE_DA_ALLOC_CLOSE,	/* Alloc DA blks on close */
1909 	EXT4_STATE_EXT_MIGRATE,		/* Inode is migrating */
1910 	EXT4_STATE_NEWENTRY,		/* File just added to dir */
1911 	EXT4_STATE_MAY_INLINE_DATA,	/* may have in-inode data */
1912 	EXT4_STATE_EXT_PRECACHED,	/* extents have been precached */
1913 	EXT4_STATE_LUSTRE_EA_INODE,	/* Lustre-style ea_inode */
1914 	EXT4_STATE_VERITY_IN_PROGRESS,	/* building fs-verity Merkle tree */
1915 	EXT4_STATE_FC_COMMITTING,	/* Fast commit ongoing */
1916 	EXT4_STATE_ORPHAN_FILE,		/* Inode orphaned in orphan file */
1917 };
1918 
1919 #define EXT4_INODE_BIT_FNS(name, field, offset)				\
1920 static inline int ext4_test_inode_##name(struct inode *inode, int bit)	\
1921 {									\
1922 	return test_bit(bit + (offset), &EXT4_I(inode)->i_##field);	\
1923 }									\
1924 static inline void ext4_set_inode_##name(struct inode *inode, int bit)	\
1925 {									\
1926 	set_bit(bit + (offset), &EXT4_I(inode)->i_##field);		\
1927 }									\
1928 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1929 {									\
1930 	clear_bit(bit + (offset), &EXT4_I(inode)->i_##field);		\
1931 }
1932 
1933 /* Add these declarations here only so that these functions can be
1934  * found by name.  Otherwise, they are very hard to locate. */
1935 static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1936 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1937 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1938 EXT4_INODE_BIT_FNS(flag, flags, 0)
1939 
1940 /* Add these declarations here only so that these functions can be
1941  * found by name.  Otherwise, they are very hard to locate. */
1942 static inline int ext4_test_inode_state(struct inode *inode, int bit);
1943 static inline void ext4_set_inode_state(struct inode *inode, int bit);
1944 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1945 #if (BITS_PER_LONG < 64)
1946 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1947 
ext4_clear_state_flags(struct ext4_inode_info * ei)1948 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1949 {
1950 	(ei)->i_state_flags = 0;
1951 }
1952 #else
1953 EXT4_INODE_BIT_FNS(state, flags, 32)
1954 
ext4_clear_state_flags(struct ext4_inode_info * ei)1955 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1956 {
1957 	/* We depend on the fact that callers will set i_flags */
1958 }
1959 #endif
1960 #else
1961 /* Assume that user mode programs are passing in an ext4fs superblock, not
1962  * a kernel struct super_block.  This will allow us to call the feature-test
1963  * macros from user land. */
1964 #define EXT4_SB(sb)	(sb)
1965 #endif
1966 
ext4_verity_in_progress(struct inode * inode)1967 static inline bool ext4_verity_in_progress(struct inode *inode)
1968 {
1969 	return IS_ENABLED(CONFIG_FS_VERITY) &&
1970 	       ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS);
1971 }
1972 
1973 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1974 
1975 /*
1976  * Codes for operating systems
1977  */
1978 #define EXT4_OS_LINUX		0
1979 #define EXT4_OS_HURD		1
1980 #define EXT4_OS_MASIX		2
1981 #define EXT4_OS_FREEBSD		3
1982 #define EXT4_OS_LITES		4
1983 
1984 /*
1985  * Revision levels
1986  */
1987 #define EXT4_GOOD_OLD_REV	0	/* The good old (original) format */
1988 #define EXT4_DYNAMIC_REV	1	/* V2 format w/ dynamic inode sizes */
1989 
1990 #define EXT4_MAX_SUPP_REV	EXT4_DYNAMIC_REV
1991 
1992 #define EXT4_GOOD_OLD_INODE_SIZE 128
1993 
1994 #define EXT4_EXTRA_TIMESTAMP_MAX	(((s64)1 << 34) - 1  + S32_MIN)
1995 #define EXT4_NON_EXTRA_TIMESTAMP_MAX	S32_MAX
1996 #define EXT4_TIMESTAMP_MIN		S32_MIN
1997 
1998 /*
1999  * Feature set definitions
2000  */
2001 
2002 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC	0x0001
2003 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES	0x0002
2004 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL		0x0004
2005 #define EXT4_FEATURE_COMPAT_EXT_ATTR		0x0008
2006 #define EXT4_FEATURE_COMPAT_RESIZE_INODE	0x0010
2007 #define EXT4_FEATURE_COMPAT_DIR_INDEX		0x0020
2008 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2	0x0200
2009 /*
2010  * The reason why "FAST_COMMIT" is a compat feature is that, FS becomes
2011  * incompatible only if fast commit blocks are present in the FS. Since we
2012  * clear the journal (and thus the fast commit blocks), we don't mark FS as
2013  * incompatible. We also have a JBD2 incompat feature, which gets set when
2014  * there are fast commit blocks present in the journal.
2015  */
2016 #define EXT4_FEATURE_COMPAT_FAST_COMMIT		0x0400
2017 #define EXT4_FEATURE_COMPAT_STABLE_INODES	0x0800
2018 #define EXT4_FEATURE_COMPAT_ORPHAN_FILE		0x1000	/* Orphan file exists */
2019 
2020 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER	0x0001
2021 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE	0x0002
2022 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR	0x0004
2023 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE        0x0008
2024 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM		0x0010
2025 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK	0x0020
2026 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE	0x0040
2027 #define EXT4_FEATURE_RO_COMPAT_QUOTA		0x0100
2028 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC		0x0200
2029 /*
2030  * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM).  When
2031  * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
2032  * all other data structures' checksums.  However, the METADATA_CSUM and
2033  * GDT_CSUM bits are mutually exclusive.
2034  */
2035 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM	0x0400
2036 #define EXT4_FEATURE_RO_COMPAT_READONLY		0x1000
2037 #define EXT4_FEATURE_RO_COMPAT_PROJECT		0x2000
2038 #define EXT4_FEATURE_RO_COMPAT_VERITY		0x8000
2039 #define EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT	0x10000 /* Orphan file may be
2040 							   non-empty */
2041 
2042 #define EXT4_FEATURE_INCOMPAT_COMPRESSION	0x0001
2043 #define EXT4_FEATURE_INCOMPAT_FILETYPE		0x0002
2044 #define EXT4_FEATURE_INCOMPAT_RECOVER		0x0004 /* Needs recovery */
2045 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV	0x0008 /* Journal device */
2046 #define EXT4_FEATURE_INCOMPAT_META_BG		0x0010
2047 #define EXT4_FEATURE_INCOMPAT_EXTENTS		0x0040 /* extents support */
2048 #define EXT4_FEATURE_INCOMPAT_64BIT		0x0080
2049 #define EXT4_FEATURE_INCOMPAT_MMP               0x0100
2050 #define EXT4_FEATURE_INCOMPAT_FLEX_BG		0x0200
2051 #define EXT4_FEATURE_INCOMPAT_EA_INODE		0x0400 /* EA in inode */
2052 #define EXT4_FEATURE_INCOMPAT_DIRDATA		0x1000 /* data in dirent */
2053 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED		0x2000
2054 #define EXT4_FEATURE_INCOMPAT_LARGEDIR		0x4000 /* >2GB or 3-lvl htree */
2055 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA	0x8000 /* data in inode */
2056 #define EXT4_FEATURE_INCOMPAT_ENCRYPT		0x10000
2057 #define EXT4_FEATURE_INCOMPAT_CASEFOLD		0x20000
2058 
2059 extern void ext4_update_dynamic_rev(struct super_block *sb);
2060 
2061 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
2062 static inline bool ext4_has_feature_##name(struct super_block *sb) \
2063 { \
2064 	return ((EXT4_SB(sb)->s_es->s_feature_compat & \
2065 		cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
2066 } \
2067 static inline void ext4_set_feature_##name(struct super_block *sb) \
2068 { \
2069 	ext4_update_dynamic_rev(sb); \
2070 	EXT4_SB(sb)->s_es->s_feature_compat |= \
2071 		cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
2072 } \
2073 static inline void ext4_clear_feature_##name(struct super_block *sb) \
2074 { \
2075 	EXT4_SB(sb)->s_es->s_feature_compat &= \
2076 		~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
2077 }
2078 
2079 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
2080 static inline bool ext4_has_feature_##name(struct super_block *sb) \
2081 { \
2082 	return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
2083 		cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
2084 } \
2085 static inline void ext4_set_feature_##name(struct super_block *sb) \
2086 { \
2087 	ext4_update_dynamic_rev(sb); \
2088 	EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
2089 		cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
2090 } \
2091 static inline void ext4_clear_feature_##name(struct super_block *sb) \
2092 { \
2093 	EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
2094 		~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
2095 }
2096 
2097 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
2098 static inline bool ext4_has_feature_##name(struct super_block *sb) \
2099 { \
2100 	return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
2101 		cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
2102 } \
2103 static inline void ext4_set_feature_##name(struct super_block *sb) \
2104 { \
2105 	ext4_update_dynamic_rev(sb); \
2106 	EXT4_SB(sb)->s_es->s_feature_incompat |= \
2107 		cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
2108 } \
2109 static inline void ext4_clear_feature_##name(struct super_block *sb) \
2110 { \
2111 	EXT4_SB(sb)->s_es->s_feature_incompat &= \
2112 		~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
2113 }
2114 
EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc,DIR_PREALLOC)2115 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc,		DIR_PREALLOC)
2116 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes,	IMAGIC_INODES)
2117 EXT4_FEATURE_COMPAT_FUNCS(journal,		HAS_JOURNAL)
2118 EXT4_FEATURE_COMPAT_FUNCS(xattr,		EXT_ATTR)
2119 EXT4_FEATURE_COMPAT_FUNCS(resize_inode,		RESIZE_INODE)
2120 EXT4_FEATURE_COMPAT_FUNCS(dir_index,		DIR_INDEX)
2121 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2,	SPARSE_SUPER2)
2122 EXT4_FEATURE_COMPAT_FUNCS(fast_commit,		FAST_COMMIT)
2123 EXT4_FEATURE_COMPAT_FUNCS(stable_inodes,	STABLE_INODES)
2124 EXT4_FEATURE_COMPAT_FUNCS(orphan_file,		ORPHAN_FILE)
2125 
2126 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super,	SPARSE_SUPER)
2127 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file,	LARGE_FILE)
2128 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir,		BTREE_DIR)
2129 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file,		HUGE_FILE)
2130 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum,		GDT_CSUM)
2131 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink,		DIR_NLINK)
2132 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize,	EXTRA_ISIZE)
2133 EXT4_FEATURE_RO_COMPAT_FUNCS(quota,		QUOTA)
2134 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc,		BIGALLOC)
2135 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum,	METADATA_CSUM)
2136 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly,		READONLY)
2137 EXT4_FEATURE_RO_COMPAT_FUNCS(project,		PROJECT)
2138 EXT4_FEATURE_RO_COMPAT_FUNCS(verity,		VERITY)
2139 EXT4_FEATURE_RO_COMPAT_FUNCS(orphan_present,	ORPHAN_PRESENT)
2140 
2141 EXT4_FEATURE_INCOMPAT_FUNCS(compression,	COMPRESSION)
2142 EXT4_FEATURE_INCOMPAT_FUNCS(filetype,		FILETYPE)
2143 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery,	RECOVER)
2144 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev,	JOURNAL_DEV)
2145 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg,		META_BG)
2146 EXT4_FEATURE_INCOMPAT_FUNCS(extents,		EXTENTS)
2147 EXT4_FEATURE_INCOMPAT_FUNCS(64bit,		64BIT)
2148 EXT4_FEATURE_INCOMPAT_FUNCS(mmp,		MMP)
2149 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg,		FLEX_BG)
2150 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode,		EA_INODE)
2151 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata,		DIRDATA)
2152 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed,		CSUM_SEED)
2153 EXT4_FEATURE_INCOMPAT_FUNCS(largedir,		LARGEDIR)
2154 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data,	INLINE_DATA)
2155 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt,		ENCRYPT)
2156 EXT4_FEATURE_INCOMPAT_FUNCS(casefold,		CASEFOLD)
2157 
2158 #define EXT2_FEATURE_COMPAT_SUPP	EXT4_FEATURE_COMPAT_EXT_ATTR
2159 #define EXT2_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
2160 					 EXT4_FEATURE_INCOMPAT_META_BG)
2161 #define EXT2_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2162 					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2163 					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
2164 
2165 #define EXT3_FEATURE_COMPAT_SUPP	EXT4_FEATURE_COMPAT_EXT_ATTR
2166 #define EXT3_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
2167 					 EXT4_FEATURE_INCOMPAT_RECOVER| \
2168 					 EXT4_FEATURE_INCOMPAT_META_BG)
2169 #define EXT3_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2170 					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2171 					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
2172 
2173 #define EXT4_FEATURE_COMPAT_SUPP	(EXT4_FEATURE_COMPAT_EXT_ATTR| \
2174 					 EXT4_FEATURE_COMPAT_ORPHAN_FILE)
2175 #define EXT4_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
2176 					 EXT4_FEATURE_INCOMPAT_RECOVER| \
2177 					 EXT4_FEATURE_INCOMPAT_META_BG| \
2178 					 EXT4_FEATURE_INCOMPAT_EXTENTS| \
2179 					 EXT4_FEATURE_INCOMPAT_64BIT| \
2180 					 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
2181 					 EXT4_FEATURE_INCOMPAT_EA_INODE| \
2182 					 EXT4_FEATURE_INCOMPAT_MMP | \
2183 					 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
2184 					 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
2185 					 EXT4_FEATURE_INCOMPAT_CASEFOLD | \
2186 					 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \
2187 					 EXT4_FEATURE_INCOMPAT_LARGEDIR)
2188 #define EXT4_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2189 					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2190 					 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
2191 					 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
2192 					 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
2193 					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
2194 					 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
2195 					 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
2196 					 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
2197 					 EXT4_FEATURE_RO_COMPAT_QUOTA |\
2198 					 EXT4_FEATURE_RO_COMPAT_PROJECT |\
2199 					 EXT4_FEATURE_RO_COMPAT_VERITY |\
2200 					 EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT)
2201 
2202 #define EXTN_FEATURE_FUNCS(ver) \
2203 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
2204 { \
2205 	return ((EXT4_SB(sb)->s_es->s_feature_compat & \
2206 		cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
2207 } \
2208 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
2209 { \
2210 	return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
2211 		cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
2212 } \
2213 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
2214 { \
2215 	return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
2216 		cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
2217 }
2218 
2219 EXTN_FEATURE_FUNCS(2)
2220 EXTN_FEATURE_FUNCS(3)
2221 EXTN_FEATURE_FUNCS(4)
2222 
2223 static inline bool ext4_has_compat_features(struct super_block *sb)
2224 {
2225 	return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
2226 }
ext4_has_ro_compat_features(struct super_block * sb)2227 static inline bool ext4_has_ro_compat_features(struct super_block *sb)
2228 {
2229 	return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
2230 }
ext4_has_incompat_features(struct super_block * sb)2231 static inline bool ext4_has_incompat_features(struct super_block *sb)
2232 {
2233 	return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
2234 }
2235 
2236 extern int ext4_feature_set_ok(struct super_block *sb, int readonly);
2237 
2238 /*
2239  * Superblock flags
2240  */
2241 enum {
2242 	EXT4_FLAGS_RESIZING,	/* Avoid superblock update and resize race */
2243 	EXT4_FLAGS_SHUTDOWN,	/* Prevent access to the file system */
2244 	EXT4_FLAGS_BDEV_IS_DAX,	/* Current block device support DAX */
2245 	EXT4_FLAGS_EMERGENCY_RO,/* Emergency read-only due to fs errors */
2246 };
2247 
ext4_forced_shutdown(struct super_block * sb)2248 static inline int ext4_forced_shutdown(struct super_block *sb)
2249 {
2250 	return test_bit(EXT4_FLAGS_SHUTDOWN, &EXT4_SB(sb)->s_ext4_flags);
2251 }
2252 
ext4_emergency_ro(struct super_block * sb)2253 static inline int ext4_emergency_ro(struct super_block *sb)
2254 {
2255 	return test_bit(EXT4_FLAGS_EMERGENCY_RO, &EXT4_SB(sb)->s_ext4_flags);
2256 }
2257 
ext4_emergency_state(struct super_block * sb)2258 static inline int ext4_emergency_state(struct super_block *sb)
2259 {
2260 	if (unlikely(ext4_forced_shutdown(sb)))
2261 		return -EIO;
2262 	if (unlikely(ext4_emergency_ro(sb)))
2263 		return -EROFS;
2264 	return 0;
2265 }
2266 
2267 /*
2268  * Default values for user and/or group using reserved blocks
2269  */
2270 #define	EXT4_DEF_RESUID		0
2271 #define	EXT4_DEF_RESGID		0
2272 
2273 /*
2274  * Default project ID
2275  */
2276 #define	EXT4_DEF_PROJID		0
2277 
2278 #define EXT4_DEF_INODE_READAHEAD_BLKS	32
2279 
2280 /*
2281  * Default mount options
2282  */
2283 #define EXT4_DEFM_DEBUG		0x0001
2284 #define EXT4_DEFM_BSDGROUPS	0x0002
2285 #define EXT4_DEFM_XATTR_USER	0x0004
2286 #define EXT4_DEFM_ACL		0x0008
2287 #define EXT4_DEFM_UID16		0x0010
2288 #define EXT4_DEFM_JMODE		0x0060
2289 #define EXT4_DEFM_JMODE_DATA	0x0020
2290 #define EXT4_DEFM_JMODE_ORDERED	0x0040
2291 #define EXT4_DEFM_JMODE_WBACK	0x0060
2292 #define EXT4_DEFM_NOBARRIER	0x0100
2293 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
2294 #define EXT4_DEFM_DISCARD	0x0400
2295 #define EXT4_DEFM_NODELALLOC	0x0800
2296 
2297 /*
2298  * Default journal batch times
2299  */
2300 #define EXT4_DEF_MIN_BATCH_TIME	0
2301 #define EXT4_DEF_MAX_BATCH_TIME	15000 /* 15ms */
2302 
2303 /*
2304  * Default values for superblock update
2305  */
2306 #define EXT4_DEF_SB_UPDATE_INTERVAL_SEC (3600) /* seconds (1 hour) */
2307 #define EXT4_DEF_SB_UPDATE_INTERVAL_KB (16384) /* kilobytes (16MB) */
2308 
2309 
2310 /*
2311  * Minimum number of groups in a flexgroup before we separate out
2312  * directories into the first block group of a flexgroup
2313  */
2314 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME	4
2315 
2316 /*
2317  * Structure of a directory entry
2318  */
2319 #define EXT4_NAME_LEN 255
2320 /*
2321  * Base length of the ext4 directory entry excluding the name length
2322  */
2323 #define EXT4_BASE_DIR_LEN (sizeof(struct ext4_dir_entry_2) - EXT4_NAME_LEN)
2324 
2325 struct ext4_dir_entry {
2326 	__le32	inode;			/* Inode number */
2327 	__le16	rec_len;		/* Directory entry length */
2328 	__le16	name_len;		/* Name length */
2329 	char	name[EXT4_NAME_LEN];	/* File name */
2330 };
2331 
2332 
2333 /*
2334  * Encrypted Casefolded entries require saving the hash on disk. This structure
2335  * followed ext4_dir_entry_2's name[name_len] at the next 4 byte aligned
2336  * boundary.
2337  */
2338 struct ext4_dir_entry_hash {
2339 	__le32 hash;
2340 	__le32 minor_hash;
2341 };
2342 
2343 /*
2344  * The new version of the directory entry.  Since EXT4 structures are
2345  * stored in intel byte order, and the name_len field could never be
2346  * bigger than 255 chars, it's safe to reclaim the extra byte for the
2347  * file_type field.
2348  */
2349 struct ext4_dir_entry_2 {
2350 	__le32	inode;			/* Inode number */
2351 	__le16	rec_len;		/* Directory entry length */
2352 	__u8	name_len;		/* Name length */
2353 	__u8	file_type;		/* See file type macros EXT4_FT_* below */
2354 	char	name[EXT4_NAME_LEN];	/* File name */
2355 };
2356 
2357 /*
2358  * Access the hashes at the end of ext4_dir_entry_2
2359  */
2360 #define EXT4_DIRENT_HASHES(entry) \
2361 	((struct ext4_dir_entry_hash *) \
2362 		(((void *)(entry)) + \
2363 		((8 + (entry)->name_len + EXT4_DIR_ROUND) & ~EXT4_DIR_ROUND)))
2364 #define EXT4_DIRENT_HASH(entry) le32_to_cpu(EXT4_DIRENT_HASHES(entry)->hash)
2365 #define EXT4_DIRENT_MINOR_HASH(entry) \
2366 		le32_to_cpu(EXT4_DIRENT_HASHES(entry)->minor_hash)
2367 
ext4_hash_in_dirent(const struct inode * inode)2368 static inline bool ext4_hash_in_dirent(const struct inode *inode)
2369 {
2370 	return IS_CASEFOLDED(inode) && IS_ENCRYPTED(inode);
2371 }
2372 
2373 /*
2374  * This is a bogus directory entry at the end of each leaf block that
2375  * records checksums.
2376  */
2377 struct ext4_dir_entry_tail {
2378 	__le32	det_reserved_zero1;	/* Pretend to be unused */
2379 	__le16	det_rec_len;		/* 12 */
2380 	__u8	det_reserved_zero2;	/* Zero name length */
2381 	__u8	det_reserved_ft;	/* 0xDE, fake file type */
2382 	__le32	det_checksum;		/* crc32c(uuid+inum+dirblock) */
2383 };
2384 
2385 #define EXT4_DIRENT_TAIL(block, blocksize) \
2386 	((struct ext4_dir_entry_tail *)(((void *)(block)) + \
2387 					((blocksize) - \
2388 					 sizeof(struct ext4_dir_entry_tail))))
2389 
2390 /*
2391  * Ext4 directory file types.  Only the low 3 bits are used.  The
2392  * other bits are reserved for now.
2393  */
2394 #define EXT4_FT_UNKNOWN		0
2395 #define EXT4_FT_REG_FILE	1
2396 #define EXT4_FT_DIR		2
2397 #define EXT4_FT_CHRDEV		3
2398 #define EXT4_FT_BLKDEV		4
2399 #define EXT4_FT_FIFO		5
2400 #define EXT4_FT_SOCK		6
2401 #define EXT4_FT_SYMLINK		7
2402 
2403 #define EXT4_FT_MAX		8
2404 
2405 #define EXT4_FT_DIR_CSUM	0xDE
2406 
2407 /*
2408  * EXT4_DIR_PAD defines the directory entries boundaries
2409  *
2410  * NOTE: It must be a multiple of 4
2411  */
2412 #define EXT4_DIR_PAD			4
2413 #define EXT4_DIR_ROUND			(EXT4_DIR_PAD - 1)
2414 #define EXT4_MAX_REC_LEN		((1<<16)-1)
2415 
2416 /*
2417  * The rec_len is dependent on the type of directory. Directories that are
2418  * casefolded and encrypted need to store the hash as well, so we add room for
2419  * ext4_extended_dir_entry_2. For all entries related to '.' or '..' you should
2420  * pass NULL for dir, as those entries do not use the extra fields.
2421  */
ext4_dir_rec_len(__u8 name_len,const struct inode * dir)2422 static inline unsigned int ext4_dir_rec_len(__u8 name_len,
2423 						const struct inode *dir)
2424 {
2425 	int rec_len = (name_len + 8 + EXT4_DIR_ROUND);
2426 
2427 	if (dir && ext4_hash_in_dirent(dir))
2428 		rec_len += sizeof(struct ext4_dir_entry_hash);
2429 	return (rec_len & ~EXT4_DIR_ROUND);
2430 }
2431 
2432 /*
2433  * If we ever get support for fs block sizes > page_size, we'll need
2434  * to remove the #if statements in the next two functions...
2435  */
2436 static inline unsigned int
ext4_rec_len_from_disk(__le16 dlen,unsigned blocksize)2437 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
2438 {
2439 	unsigned len = le16_to_cpu(dlen);
2440 
2441 #if (PAGE_SIZE >= 65536)
2442 	if (len == EXT4_MAX_REC_LEN || len == 0)
2443 		return blocksize;
2444 	return (len & 65532) | ((len & 3) << 16);
2445 #else
2446 	return len;
2447 #endif
2448 }
2449 
ext4_rec_len_to_disk(unsigned len,unsigned blocksize)2450 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
2451 {
2452 	BUG_ON((len > blocksize) || (blocksize > (1 << 18)) || (len & 3));
2453 #if (PAGE_SIZE >= 65536)
2454 	if (len < 65536)
2455 		return cpu_to_le16(len);
2456 	if (len == blocksize) {
2457 		if (blocksize == 65536)
2458 			return cpu_to_le16(EXT4_MAX_REC_LEN);
2459 		else
2460 			return cpu_to_le16(0);
2461 	}
2462 	return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
2463 #else
2464 	return cpu_to_le16(len);
2465 #endif
2466 }
2467 
2468 /*
2469  * Hash Tree Directory indexing
2470  * (c) Daniel Phillips, 2001
2471  */
2472 
2473 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
2474 		    ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2475 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \
2476 		    !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir)))
2477 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2478 
2479 /* Legal values for the dx_root hash_version field: */
2480 
2481 #define DX_HASH_LEGACY			0
2482 #define DX_HASH_HALF_MD4		1
2483 #define DX_HASH_TEA			2
2484 #define DX_HASH_LEGACY_UNSIGNED		3
2485 #define DX_HASH_HALF_MD4_UNSIGNED	4
2486 #define DX_HASH_TEA_UNSIGNED		5
2487 #define DX_HASH_SIPHASH			6
2488 #define DX_HASH_LAST 			DX_HASH_SIPHASH
2489 
ext4_chksum(struct ext4_sb_info * sbi,u32 crc,const void * address,unsigned int length)2490 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2491 			      const void *address, unsigned int length)
2492 {
2493 	return crc32c(crc, address, length);
2494 }
2495 
2496 #ifdef __KERNEL__
2497 
2498 /* hash info structure used by the directory hash */
2499 struct dx_hash_info
2500 {
2501 	u32		hash;
2502 	u32		minor_hash;
2503 	int		hash_version;
2504 	u32		*seed;
2505 };
2506 
2507 
2508 /* 32 and 64 bit signed EOF for dx directories */
2509 #define EXT4_HTREE_EOF_32BIT   ((1UL  << (32 - 1)) - 1)
2510 #define EXT4_HTREE_EOF_64BIT   ((1ULL << (64 - 1)) - 1)
2511 
2512 
2513 /*
2514  * Control parameters used by ext4_htree_next_block
2515  */
2516 #define HASH_NB_ALWAYS		1
2517 
2518 struct ext4_filename {
2519 	const struct qstr *usr_fname;
2520 	struct fscrypt_str disk_name;
2521 	struct dx_hash_info hinfo;
2522 #ifdef CONFIG_FS_ENCRYPTION
2523 	struct fscrypt_str crypto_buf;
2524 #endif
2525 #if IS_ENABLED(CONFIG_UNICODE)
2526 	struct qstr cf_name;
2527 #endif
2528 };
2529 
2530 #define fname_name(p) ((p)->disk_name.name)
2531 #define fname_usr_name(p) ((p)->usr_fname->name)
2532 #define fname_len(p)  ((p)->disk_name.len)
2533 
2534 /*
2535  * Describe an inode's exact location on disk and in memory
2536  */
2537 struct ext4_iloc
2538 {
2539 	struct buffer_head *bh;
2540 	unsigned long offset;
2541 	ext4_group_t block_group;
2542 };
2543 
ext4_raw_inode(struct ext4_iloc * iloc)2544 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2545 {
2546 	return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2547 }
2548 
ext4_is_quota_file(struct inode * inode)2549 static inline bool ext4_is_quota_file(struct inode *inode)
2550 {
2551 	return IS_NOQUOTA(inode) &&
2552 	       !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL);
2553 }
2554 
2555 /*
2556  * This structure is stuffed into the struct file's private_data field
2557  * for directories.  It is where we put information so that we can do
2558  * readdir operations in hash tree order.
2559  */
2560 struct dir_private_info {
2561 	struct rb_root	root;
2562 	struct rb_node	*curr_node;
2563 	struct fname	*extra_fname;
2564 	loff_t		last_pos;
2565 	__u32		curr_hash;
2566 	__u32		curr_minor_hash;
2567 	__u32		next_hash;
2568 	u64		cookie;
2569 	bool		initialized;
2570 };
2571 
2572 /* calculate the first block number of the group */
2573 static inline ext4_fsblk_t
ext4_group_first_block_no(struct super_block * sb,ext4_group_t group_no)2574 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2575 {
2576 	return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2577 		le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2578 }
2579 
2580 /*
2581  * Special error return code only used by dx_probe() and its callers.
2582  */
2583 #define ERR_BAD_DX_DIR	(-(MAX_ERRNO - 1))
2584 
2585 /* htree levels for ext4 */
2586 #define	EXT4_HTREE_LEVEL_COMPAT	2
2587 #define	EXT4_HTREE_LEVEL	3
2588 
ext4_dir_htree_level(struct super_block * sb)2589 static inline int ext4_dir_htree_level(struct super_block *sb)
2590 {
2591 	return ext4_has_feature_largedir(sb) ?
2592 		EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT;
2593 }
2594 
2595 /*
2596  * Timeout and state flag for lazy initialization inode thread.
2597  */
2598 #define EXT4_DEF_LI_WAIT_MULT			10
2599 #define EXT4_DEF_LI_MAX_START_DELAY		5
2600 #define EXT4_LAZYINIT_QUIT			0x0001
2601 #define EXT4_LAZYINIT_RUNNING			0x0002
2602 
2603 /*
2604  * Lazy inode table initialization info
2605  */
2606 struct ext4_lazy_init {
2607 	unsigned long		li_state;
2608 	struct list_head	li_request_list;
2609 	struct mutex		li_list_mtx;
2610 };
2611 
2612 enum ext4_li_mode {
2613 	EXT4_LI_MODE_PREFETCH_BBITMAP,
2614 	EXT4_LI_MODE_ITABLE,
2615 };
2616 
2617 struct ext4_li_request {
2618 	struct super_block	*lr_super;
2619 	enum ext4_li_mode	lr_mode;
2620 	ext4_group_t		lr_first_not_zeroed;
2621 	ext4_group_t		lr_next_group;
2622 	struct list_head	lr_request;
2623 	unsigned long		lr_next_sched;
2624 	unsigned long		lr_timeout;
2625 };
2626 
2627 struct ext4_features {
2628 	struct kobject f_kobj;
2629 	struct completion f_kobj_unregister;
2630 };
2631 
2632 /*
2633  * This structure will be used for multiple mount protection. It will be
2634  * written into the block number saved in the s_mmp_block field in the
2635  * superblock. Programs that check MMP should assume that if
2636  * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2637  * to use the filesystem, regardless of how old the timestamp is.
2638  */
2639 #define EXT4_MMP_MAGIC     0x004D4D50U /* ASCII for MMP */
2640 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2641 #define EXT4_MMP_SEQ_FSCK  0xE24D4D50U /* mmp_seq value when being fscked */
2642 #define EXT4_MMP_SEQ_MAX   0xE24D4D4FU /* maximum valid mmp_seq value */
2643 
2644 struct mmp_struct {
2645 	__le32	mmp_magic;		/* Magic number for MMP */
2646 	__le32	mmp_seq;		/* Sequence no. updated periodically */
2647 
2648 	/*
2649 	 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2650 	 * purposes and do not affect the correctness of the algorithm
2651 	 */
2652 	__le64	mmp_time;		/* Time last updated */
2653 	char	mmp_nodename[64];	/* Node which last updated MMP block */
2654 	char	mmp_bdevname[32];	/* Bdev which last updated MMP block */
2655 
2656 	/*
2657 	 * mmp_check_interval is used to verify if the MMP block has been
2658 	 * updated on the block device. The value is updated based on the
2659 	 * maximum time to write the MMP block during an update cycle.
2660 	 */
2661 	__le16	mmp_check_interval;
2662 
2663 	__le16	mmp_pad1;
2664 	__le32	mmp_pad2[226];
2665 	__le32	mmp_checksum;		/* crc32c(uuid+mmp_block) */
2666 };
2667 
2668 /* arguments passed to the mmp thread */
2669 struct mmpd_data {
2670 	struct buffer_head *bh; /* bh from initial read_mmp_block() */
2671 	struct super_block *sb;  /* super block of the fs */
2672 };
2673 
2674 /*
2675  * Check interval multiplier
2676  * The MMP block is written every update interval and initially checked every
2677  * update interval x the multiplier (the value is then adapted based on the
2678  * write latency). The reason is that writes can be delayed under load and we
2679  * don't want readers to incorrectly assume that the filesystem is no longer
2680  * in use.
2681  */
2682 #define EXT4_MMP_CHECK_MULT		2UL
2683 
2684 /*
2685  * Minimum interval for MMP checking in seconds.
2686  */
2687 #define EXT4_MMP_MIN_CHECK_INTERVAL	5UL
2688 
2689 /*
2690  * Maximum interval for MMP checking in seconds.
2691  */
2692 #define EXT4_MMP_MAX_CHECK_INTERVAL	300UL
2693 
2694 /*
2695  * Function prototypes
2696  */
2697 
2698 /*
2699  * Ok, these declarations are also in <linux/kernel.h> but none of the
2700  * ext4 source programs needs to include it so they are duplicated here.
2701  */
2702 # define NORET_TYPE	/**/
2703 # define ATTRIB_NORET	__attribute__((noreturn))
2704 # define NORET_AND	noreturn,
2705 
2706 /* bitmap.c */
2707 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2708 void ext4_inode_bitmap_csum_set(struct super_block *sb,
2709 				struct ext4_group_desc *gdp,
2710 				struct buffer_head *bh);
2711 int ext4_inode_bitmap_csum_verify(struct super_block *sb,
2712 				  struct ext4_group_desc *gdp,
2713 				  struct buffer_head *bh);
2714 void ext4_block_bitmap_csum_set(struct super_block *sb,
2715 				struct ext4_group_desc *gdp,
2716 				struct buffer_head *bh);
2717 int ext4_block_bitmap_csum_verify(struct super_block *sb,
2718 				  struct ext4_group_desc *gdp,
2719 				  struct buffer_head *bh);
2720 
2721 /* balloc.c */
2722 extern void ext4_get_group_no_and_offset(struct super_block *sb,
2723 					 ext4_fsblk_t blocknr,
2724 					 ext4_group_t *blockgrpp,
2725 					 ext4_grpblk_t *offsetp);
2726 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2727 					  ext4_fsblk_t block);
2728 
2729 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2730 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2731 			ext4_group_t group);
2732 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2733 					 ext4_fsblk_t goal,
2734 					 unsigned int flags,
2735 					 unsigned long *count,
2736 					 int *errp);
2737 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2738 				    s64 nclusters, unsigned int flags);
2739 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2740 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2741 						    ext4_group_t block_group,
2742 						    struct buffer_head ** bh);
2743 extern struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2744 						   ext4_group_t group);
2745 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2746 
2747 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2748 						ext4_group_t block_group,
2749 						bool ignore_locked);
2750 extern int ext4_wait_block_bitmap(struct super_block *sb,
2751 				  ext4_group_t block_group,
2752 				  struct buffer_head *bh);
2753 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2754 						  ext4_group_t block_group);
2755 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2756 					      ext4_group_t block_group,
2757 					      struct ext4_group_desc *gdp);
2758 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2759 
2760 #if IS_ENABLED(CONFIG_UNICODE)
2761 extern int ext4_fname_setup_ci_filename(struct inode *dir,
2762 					const struct qstr *iname,
2763 					struct ext4_filename *fname);
2764 
ext4_fname_free_ci_filename(struct ext4_filename * fname)2765 static inline void ext4_fname_free_ci_filename(struct ext4_filename *fname)
2766 {
2767 	kfree(fname->cf_name.name);
2768 	fname->cf_name.name = NULL;
2769 }
2770 #else
ext4_fname_setup_ci_filename(struct inode * dir,const struct qstr * iname,struct ext4_filename * fname)2771 static inline int ext4_fname_setup_ci_filename(struct inode *dir,
2772 					       const struct qstr *iname,
2773 					       struct ext4_filename *fname)
2774 {
2775 	return 0;
2776 }
2777 
ext4_fname_free_ci_filename(struct ext4_filename * fname)2778 static inline void ext4_fname_free_ci_filename(struct ext4_filename *fname)
2779 {
2780 }
2781 #endif
2782 
2783 /* ext4 encryption related stuff goes here crypto.c */
2784 #ifdef CONFIG_FS_ENCRYPTION
2785 extern const struct fscrypt_operations ext4_cryptops;
2786 
2787 int ext4_fname_setup_filename(struct inode *dir, const struct qstr *iname,
2788 			      int lookup, struct ext4_filename *fname);
2789 
2790 int ext4_fname_prepare_lookup(struct inode *dir, struct dentry *dentry,
2791 			      struct ext4_filename *fname);
2792 
2793 void ext4_fname_free_filename(struct ext4_filename *fname);
2794 
2795 int ext4_ioctl_get_encryption_pwsalt(struct file *filp, void __user *arg);
2796 
2797 #else /* !CONFIG_FS_ENCRYPTION */
ext4_fname_setup_filename(struct inode * dir,const struct qstr * iname,int lookup,struct ext4_filename * fname)2798 static inline int ext4_fname_setup_filename(struct inode *dir,
2799 					    const struct qstr *iname,
2800 					    int lookup,
2801 					    struct ext4_filename *fname)
2802 {
2803 	fname->usr_fname = iname;
2804 	fname->disk_name.name = (unsigned char *) iname->name;
2805 	fname->disk_name.len = iname->len;
2806 
2807 	return ext4_fname_setup_ci_filename(dir, iname, fname);
2808 }
2809 
ext4_fname_prepare_lookup(struct inode * dir,struct dentry * dentry,struct ext4_filename * fname)2810 static inline int ext4_fname_prepare_lookup(struct inode *dir,
2811 					    struct dentry *dentry,
2812 					    struct ext4_filename *fname)
2813 {
2814 	return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname);
2815 }
2816 
ext4_fname_free_filename(struct ext4_filename * fname)2817 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2818 {
2819 	ext4_fname_free_ci_filename(fname);
2820 }
2821 
ext4_ioctl_get_encryption_pwsalt(struct file * filp,void __user * arg)2822 static inline int ext4_ioctl_get_encryption_pwsalt(struct file *filp,
2823 						   void __user *arg)
2824 {
2825 	return -EOPNOTSUPP;
2826 }
2827 #endif /* !CONFIG_FS_ENCRYPTION */
2828 
2829 /* dir.c */
2830 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2831 				  struct file *,
2832 				  struct ext4_dir_entry_2 *,
2833 				  struct buffer_head *, char *, int,
2834 				  unsigned int);
2835 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2836 	unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2837 				(de), (bh), (buf), (size), (offset)))
2838 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2839 				__u32 minor_hash,
2840 				struct ext4_dir_entry_2 *dirent,
2841 				struct fscrypt_str *ent_name);
2842 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2843 extern int ext4_find_dest_de(struct inode *dir, struct buffer_head *bh,
2844 			     void *buf, int buf_size,
2845 			     struct ext4_filename *fname,
2846 			     struct ext4_dir_entry_2 **dest_de);
2847 void ext4_insert_dentry(struct inode *dir, struct inode *inode,
2848 			struct ext4_dir_entry_2 *de,
2849 			int buf_size,
2850 			struct ext4_filename *fname);
ext4_update_dx_flag(struct inode * inode)2851 static inline void ext4_update_dx_flag(struct inode *inode)
2852 {
2853 	if (!ext4_has_feature_dir_index(inode->i_sb) &&
2854 	    ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) {
2855 		/* ext4_iget() should have caught this... */
2856 		WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb));
2857 		ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2858 	}
2859 }
2860 static const unsigned char ext4_filetype_table[] = {
2861 	DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2862 };
2863 
get_dtype(struct super_block * sb,int filetype)2864 static inline  unsigned char get_dtype(struct super_block *sb, int filetype)
2865 {
2866 	if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2867 		return DT_UNKNOWN;
2868 
2869 	return ext4_filetype_table[filetype];
2870 }
2871 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2872 			     void *buf, int buf_size);
2873 
2874 /* fsync.c */
2875 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2876 
2877 /* hash.c */
2878 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len,
2879 			  struct dx_hash_info *hinfo);
2880 
2881 /* ialloc.c */
2882 extern int ext4_mark_inode_used(struct super_block *sb, int ino);
2883 extern struct inode *__ext4_new_inode(struct mnt_idmap *, handle_t *,
2884 				      struct inode *, umode_t,
2885 				      const struct qstr *qstr, __u32 goal,
2886 				      uid_t *owner, __u32 i_flags,
2887 				      int handle_type, unsigned int line_no,
2888 				      int nblocks);
2889 
2890 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags)          \
2891 	__ext4_new_inode(&nop_mnt_idmap, (handle), (dir), (mode), (qstr),      \
2892 			 (goal), (owner), i_flags, 0, 0, 0)
2893 #define ext4_new_inode_start_handle(idmap, dir, mode, qstr, goal, owner, \
2894 				    type, nblocks)		    \
2895 	__ext4_new_inode((idmap), NULL, (dir), (mode), (qstr), (goal), (owner), \
2896 			 0, (type), __LINE__, (nblocks))
2897 
2898 
2899 extern void ext4_free_inode(handle_t *, struct inode *);
2900 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2901 extern unsigned long ext4_count_free_inodes(struct super_block *);
2902 extern unsigned long ext4_count_dirs(struct super_block *);
2903 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2904 extern int ext4_init_inode_table(struct super_block *sb,
2905 				 ext4_group_t group, int barrier);
2906 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2907 
2908 /* fast_commit.c */
2909 int ext4_fc_info_show(struct seq_file *seq, void *v);
2910 void ext4_fc_init(struct super_block *sb, journal_t *journal);
2911 void ext4_fc_init_inode(struct inode *inode);
2912 void ext4_fc_track_range(handle_t *handle, struct inode *inode, ext4_lblk_t start,
2913 			 ext4_lblk_t end);
2914 void __ext4_fc_track_unlink(handle_t *handle, struct inode *inode,
2915 	struct dentry *dentry);
2916 void __ext4_fc_track_link(handle_t *handle, struct inode *inode,
2917 	struct dentry *dentry);
2918 void ext4_fc_track_unlink(handle_t *handle, struct dentry *dentry);
2919 void ext4_fc_track_link(handle_t *handle, struct dentry *dentry);
2920 void __ext4_fc_track_create(handle_t *handle, struct inode *inode,
2921 			    struct dentry *dentry);
2922 void ext4_fc_track_create(handle_t *handle, struct dentry *dentry);
2923 void ext4_fc_track_inode(handle_t *handle, struct inode *inode);
2924 void ext4_fc_mark_ineligible(struct super_block *sb, int reason, handle_t *handle);
2925 void ext4_fc_start_update(struct inode *inode);
2926 void ext4_fc_stop_update(struct inode *inode);
2927 void ext4_fc_del(struct inode *inode);
2928 bool ext4_fc_replay_check_excluded(struct super_block *sb, ext4_fsblk_t block);
2929 void ext4_fc_replay_cleanup(struct super_block *sb);
2930 int ext4_fc_commit(journal_t *journal, tid_t commit_tid);
2931 int __init ext4_fc_init_dentry_cache(void);
2932 void ext4_fc_destroy_dentry_cache(void);
2933 int ext4_fc_record_regions(struct super_block *sb, int ino,
2934 			   ext4_lblk_t lblk, ext4_fsblk_t pblk,
2935 			   int len, int replay);
2936 
2937 /* mballoc.c */
2938 extern const struct seq_operations ext4_mb_seq_groups_ops;
2939 extern const struct seq_operations ext4_mb_seq_structs_summary_ops;
2940 extern int ext4_seq_mb_stats_show(struct seq_file *seq, void *offset);
2941 extern int ext4_mb_init(struct super_block *);
2942 extern void ext4_mb_release(struct super_block *);
2943 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2944 				struct ext4_allocation_request *, int *);
2945 extern void ext4_discard_preallocations(struct inode *);
2946 extern int __init ext4_init_mballoc(void);
2947 extern void ext4_exit_mballoc(void);
2948 extern ext4_group_t ext4_mb_prefetch(struct super_block *sb,
2949 				     ext4_group_t group,
2950 				     unsigned int nr, int *cnt);
2951 extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group,
2952 				  unsigned int nr);
2953 
2954 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2955 			     struct buffer_head *bh, ext4_fsblk_t block,
2956 			     unsigned long count, int flags);
2957 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2958 				   ext4_group_t ngroups);
2959 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2960 		ext4_group_t i, struct ext4_group_desc *desc);
2961 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2962 				ext4_fsblk_t block, unsigned long count);
2963 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2964 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid);
2965 extern void ext4_mb_mark_bb(struct super_block *sb, ext4_fsblk_t block,
2966 			    int len, bool state);
ext4_mb_cr_expensive(enum criteria cr)2967 static inline bool ext4_mb_cr_expensive(enum criteria cr)
2968 {
2969 	return cr >= CR_GOAL_LEN_SLOW;
2970 }
2971 
2972 /* inode.c */
2973 void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw,
2974 			 struct ext4_inode_info *ei);
2975 int ext4_inode_is_fast_symlink(struct inode *inode);
2976 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2977 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2978 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count,
2979 		     bool wait, struct buffer_head **bhs);
2980 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2981 			     struct buffer_head *bh_result, int create);
2982 int ext4_get_block(struct inode *inode, sector_t iblock,
2983 		   struct buffer_head *bh_result, int create);
2984 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2985 			   struct buffer_head *bh, int create);
2986 int ext4_walk_page_buffers(handle_t *handle,
2987 			   struct inode *inode,
2988 			   struct buffer_head *head,
2989 			   unsigned from,
2990 			   unsigned to,
2991 			   int *partial,
2992 			   int (*fn)(handle_t *handle, struct inode *inode,
2993 				     struct buffer_head *bh));
2994 int do_journal_get_write_access(handle_t *handle, struct inode *inode,
2995 				struct buffer_head *bh);
2996 #define FALL_BACK_TO_NONDELALLOC 1
2997 #define CONVERT_INLINE_DATA	 2
2998 
2999 typedef enum {
3000 	EXT4_IGET_NORMAL =	0,
3001 	EXT4_IGET_SPECIAL =	0x0001, /* OK to iget a system inode */
3002 	EXT4_IGET_HANDLE = 	0x0002,	/* Inode # is from a handle */
3003 	EXT4_IGET_BAD =		0x0004, /* Allow to iget a bad inode */
3004 	EXT4_IGET_EA_INODE =	0x0008	/* Inode should contain an EA value */
3005 } ext4_iget_flags;
3006 
3007 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
3008 				 ext4_iget_flags flags, const char *function,
3009 				 unsigned int line);
3010 
3011 #define ext4_iget(sb, ino, flags) \
3012 	__ext4_iget((sb), (ino), (flags), __func__, __LINE__)
3013 
3014 extern int  ext4_write_inode(struct inode *, struct writeback_control *);
3015 extern int  ext4_setattr(struct mnt_idmap *, struct dentry *,
3016 			 struct iattr *);
3017 extern u32  ext4_dio_alignment(struct inode *inode);
3018 extern int  ext4_getattr(struct mnt_idmap *, const struct path *,
3019 			 struct kstat *, u32, unsigned int);
3020 extern void ext4_evict_inode(struct inode *);
3021 extern void ext4_clear_inode(struct inode *);
3022 extern int  ext4_file_getattr(struct mnt_idmap *, const struct path *,
3023 			      struct kstat *, u32, unsigned int);
3024 extern void ext4_dirty_inode(struct inode *, int);
3025 extern int ext4_change_inode_journal_flag(struct inode *, int);
3026 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
3027 extern int ext4_get_fc_inode_loc(struct super_block *sb, unsigned long ino,
3028 			  struct ext4_iloc *iloc);
3029 extern int ext4_inode_attach_jinode(struct inode *inode);
3030 extern int ext4_can_truncate(struct inode *inode);
3031 extern int ext4_truncate(struct inode *);
3032 extern int ext4_break_layouts(struct inode *);
3033 extern int ext4_truncate_page_cache_block_range(struct inode *inode,
3034 						loff_t start, loff_t end);
3035 extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length);
3036 extern void ext4_set_inode_flags(struct inode *, bool init);
3037 extern int ext4_alloc_da_blocks(struct inode *inode);
3038 extern void ext4_set_aops(struct inode *inode);
3039 extern int ext4_writepage_trans_blocks(struct inode *);
3040 extern int ext4_normal_submit_inode_data_buffers(struct jbd2_inode *jinode);
3041 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
3042 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
3043 			     loff_t lstart, loff_t lend);
3044 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf);
3045 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
3046 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
3047 extern void ext4_da_release_space(struct inode *inode, int to_free);
3048 extern void ext4_da_update_reserve_space(struct inode *inode,
3049 					int used, int quota_claim);
3050 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
3051 			      ext4_fsblk_t pblk, ext4_lblk_t len);
3052 
3053 /* indirect.c */
3054 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
3055 				struct ext4_map_blocks *map, int flags);
3056 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
3057 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
3058 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
3059 				 ext4_lblk_t start, ext4_lblk_t end);
3060 
3061 /* ioctl.c */
3062 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
3063 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
3064 int ext4_fileattr_set(struct mnt_idmap *idmap,
3065 		      struct dentry *dentry, struct fileattr *fa);
3066 int ext4_fileattr_get(struct dentry *dentry, struct fileattr *fa);
3067 extern void ext4_reset_inode_seed(struct inode *inode);
3068 int ext4_update_overhead(struct super_block *sb, bool force);
3069 int ext4_force_shutdown(struct super_block *sb, u32 flags);
3070 
3071 /* migrate.c */
3072 extern int ext4_ext_migrate(struct inode *);
3073 extern int ext4_ind_migrate(struct inode *inode);
3074 
3075 /* namei.c */
3076 extern int ext4_init_new_dir(handle_t *handle, struct inode *dir,
3077 			     struct inode *inode);
3078 extern int ext4_dirblock_csum_verify(struct inode *inode,
3079 				     struct buffer_head *bh);
3080 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
3081 				__u32 start_minor_hash, __u32 *next_hash);
3082 extern int ext4_search_dir(struct buffer_head *bh,
3083 			   char *search_buf,
3084 			   int buf_size,
3085 			   struct inode *dir,
3086 			   struct ext4_filename *fname,
3087 			   unsigned int offset,
3088 			   struct ext4_dir_entry_2 **res_dir);
3089 extern int ext4_generic_delete_entry(struct inode *dir,
3090 				     struct ext4_dir_entry_2 *de_del,
3091 				     struct buffer_head *bh,
3092 				     void *entry_buf,
3093 				     int buf_size,
3094 				     int csum_size);
3095 extern bool ext4_empty_dir(struct inode *inode);
3096 
3097 /* resize.c */
3098 extern void ext4_kvfree_array_rcu(void *to_free);
3099 extern int ext4_group_add(struct super_block *sb,
3100 				struct ext4_new_group_data *input);
3101 extern int ext4_group_extend(struct super_block *sb,
3102 				struct ext4_super_block *es,
3103 				ext4_fsblk_t n_blocks_count);
3104 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
3105 extern unsigned int ext4_list_backups(struct super_block *sb,
3106 				      unsigned int *three, unsigned int *five,
3107 				      unsigned int *seven);
3108 
3109 /* super.c */
3110 extern struct buffer_head *ext4_sb_bread(struct super_block *sb,
3111 					 sector_t block, blk_opf_t op_flags);
3112 extern struct buffer_head *ext4_sb_bread_unmovable(struct super_block *sb,
3113 						   sector_t block);
3114 extern void ext4_read_bh_nowait(struct buffer_head *bh, blk_opf_t op_flags,
3115 				bh_end_io_t *end_io, bool simu_fail);
3116 extern int ext4_read_bh(struct buffer_head *bh, blk_opf_t op_flags,
3117 			bh_end_io_t *end_io, bool simu_fail);
3118 extern int ext4_read_bh_lock(struct buffer_head *bh, blk_opf_t op_flags, bool wait);
3119 extern void ext4_sb_breadahead_unmovable(struct super_block *sb, sector_t block);
3120 extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
3121 extern int ext4_calculate_overhead(struct super_block *sb);
3122 extern __le32 ext4_superblock_csum(struct super_block *sb,
3123 				   struct ext4_super_block *es);
3124 extern void ext4_superblock_csum_set(struct super_block *sb);
3125 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
3126 				    ext4_group_t ngroup);
3127 extern const char *ext4_decode_error(struct super_block *sb, int errno,
3128 				     char nbuf[16]);
3129 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb,
3130 					     ext4_group_t block_group,
3131 					     unsigned int flags);
3132 extern unsigned int ext4_num_base_meta_blocks(struct super_block *sb,
3133 					      ext4_group_t block_group);
3134 
3135 extern __printf(7, 8)
3136 void __ext4_error(struct super_block *, const char *, unsigned int, bool,
3137 		  int, __u64, const char *, ...);
3138 extern __printf(6, 7)
3139 void __ext4_error_inode(struct inode *, const char *, unsigned int,
3140 			ext4_fsblk_t, int, const char *, ...);
3141 extern __printf(5, 6)
3142 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
3143 		     const char *, ...);
3144 extern void __ext4_std_error(struct super_block *, const char *,
3145 			     unsigned int, int);
3146 extern __printf(4, 5)
3147 void __ext4_warning(struct super_block *, const char *, unsigned int,
3148 		    const char *, ...);
3149 extern __printf(4, 5)
3150 void __ext4_warning_inode(const struct inode *inode, const char *function,
3151 			  unsigned int line, const char *fmt, ...);
3152 extern __printf(3, 4)
3153 void __ext4_msg(struct super_block *, const char *, const char *, ...);
3154 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
3155 			   const char *, unsigned int, const char *);
3156 extern __printf(7, 8)
3157 void __ext4_grp_locked_error(const char *, unsigned int,
3158 			     struct super_block *, ext4_group_t,
3159 			     unsigned long, ext4_fsblk_t,
3160 			     const char *, ...);
3161 
3162 #define EXT4_ERROR_INODE(inode, fmt, a...) \
3163 	ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
3164 
3165 #define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...)			\
3166 	__ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a)
3167 
3168 #define ext4_error_inode_block(inode, block, err, fmt, a...)		\
3169 	__ext4_error_inode((inode), __func__, __LINE__, (block), (err),	\
3170 			   (fmt), ## a)
3171 
3172 #define EXT4_ERROR_FILE(file, block, fmt, a...)				\
3173 	ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
3174 
3175 #define ext4_abort(sb, err, fmt, a...)					\
3176 	__ext4_error((sb), __func__, __LINE__, true, (err), 0, (fmt), ## a)
3177 
3178 #ifdef CONFIG_PRINTK
3179 
3180 #define ext4_error_inode(inode, func, line, block, fmt, ...)		\
3181 	__ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__)
3182 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...)	\
3183 	__ext4_error_inode((inode), (func), (line), (block), 		\
3184 			   (err), (fmt), ##__VA_ARGS__)
3185 #define ext4_error_file(file, func, line, block, fmt, ...)		\
3186 	__ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
3187 #define ext4_error(sb, fmt, ...)					\
3188 	__ext4_error((sb), __func__, __LINE__, false, 0, 0, (fmt),	\
3189 		##__VA_ARGS__)
3190 #define ext4_error_err(sb, err, fmt, ...)				\
3191 	__ext4_error((sb), __func__, __LINE__, false, (err), 0, (fmt),	\
3192 		##__VA_ARGS__)
3193 #define ext4_warning(sb, fmt, ...)					\
3194 	__ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
3195 #define ext4_warning_inode(inode, fmt, ...)				\
3196 	__ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
3197 #define ext4_msg(sb, level, fmt, ...)				\
3198 	__ext4_msg(sb, level, fmt, ##__VA_ARGS__)
3199 #define dump_mmp_msg(sb, mmp, msg)					\
3200 	__dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
3201 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...)		\
3202 	__ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
3203 				fmt, ##__VA_ARGS__)
3204 
3205 #else
3206 
3207 #define ext4_error_inode(inode, func, line, block, fmt, ...)		\
3208 do {									\
3209 	no_printk(fmt, ##__VA_ARGS__);					\
3210 	__ext4_error_inode(inode, "", 0, block, 0, " ");		\
3211 } while (0)
3212 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...)	\
3213 do {									\
3214 	no_printk(fmt, ##__VA_ARGS__);					\
3215 	__ext4_error_inode(inode, "", 0, block, err, " ");		\
3216 } while (0)
3217 #define ext4_error_file(file, func, line, block, fmt, ...)		\
3218 do {									\
3219 	no_printk(fmt, ##__VA_ARGS__);					\
3220 	__ext4_error_file(file, "", 0, block, " ");			\
3221 } while (0)
3222 #define ext4_error(sb, fmt, ...)					\
3223 do {									\
3224 	no_printk(fmt, ##__VA_ARGS__);					\
3225 	__ext4_error(sb, "", 0, false, 0, 0, " ");			\
3226 } while (0)
3227 #define ext4_error_err(sb, err, fmt, ...)				\
3228 do {									\
3229 	no_printk(fmt, ##__VA_ARGS__);					\
3230 	__ext4_error(sb, "", 0, false, err, 0, " ");			\
3231 } while (0)
3232 #define ext4_warning(sb, fmt, ...)					\
3233 do {									\
3234 	no_printk(fmt, ##__VA_ARGS__);					\
3235 	__ext4_warning(sb, "", 0, " ");					\
3236 } while (0)
3237 #define ext4_warning_inode(inode, fmt, ...)				\
3238 do {									\
3239 	no_printk(fmt, ##__VA_ARGS__);					\
3240 	__ext4_warning_inode(inode, "", 0, " ");			\
3241 } while (0)
3242 #define ext4_msg(sb, level, fmt, ...)					\
3243 do {									\
3244 	no_printk(fmt, ##__VA_ARGS__);					\
3245 	__ext4_msg(sb, "", " ");					\
3246 } while (0)
3247 #define dump_mmp_msg(sb, mmp, msg)					\
3248 	__dump_mmp_msg(sb, mmp, "", 0, "")
3249 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...)		\
3250 do {									\
3251 	no_printk(fmt, ##__VA_ARGS__);				\
3252 	__ext4_grp_locked_error("", 0, sb, grp, ino, block, " ");	\
3253 } while (0)
3254 
3255 #endif
3256 
3257 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
3258 				      struct ext4_group_desc *bg);
3259 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
3260 				      struct ext4_group_desc *bg);
3261 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
3262 				     struct ext4_group_desc *bg);
3263 extern __u32 ext4_free_group_clusters(struct super_block *sb,
3264 				      struct ext4_group_desc *bg);
3265 extern __u32 ext4_free_inodes_count(struct super_block *sb,
3266 				 struct ext4_group_desc *bg);
3267 extern __u32 ext4_used_dirs_count(struct super_block *sb,
3268 				struct ext4_group_desc *bg);
3269 extern __u32 ext4_itable_unused_count(struct super_block *sb,
3270 				   struct ext4_group_desc *bg);
3271 extern void ext4_block_bitmap_set(struct super_block *sb,
3272 				  struct ext4_group_desc *bg, ext4_fsblk_t blk);
3273 extern void ext4_inode_bitmap_set(struct super_block *sb,
3274 				  struct ext4_group_desc *bg, ext4_fsblk_t blk);
3275 extern void ext4_inode_table_set(struct super_block *sb,
3276 				 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3277 extern void ext4_free_group_clusters_set(struct super_block *sb,
3278 					 struct ext4_group_desc *bg,
3279 					 __u32 count);
3280 extern void ext4_free_inodes_set(struct super_block *sb,
3281 				struct ext4_group_desc *bg, __u32 count);
3282 extern void ext4_used_dirs_set(struct super_block *sb,
3283 				struct ext4_group_desc *bg, __u32 count);
3284 extern void ext4_itable_unused_set(struct super_block *sb,
3285 				   struct ext4_group_desc *bg, __u32 count);
3286 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
3287 				       struct ext4_group_desc *gdp);
3288 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
3289 				     struct ext4_group_desc *gdp);
3290 extern int ext4_register_li_request(struct super_block *sb,
3291 				    ext4_group_t first_not_zeroed);
3292 
ext4_has_group_desc_csum(struct super_block * sb)3293 static inline int ext4_has_group_desc_csum(struct super_block *sb)
3294 {
3295 	return ext4_has_feature_gdt_csum(sb) ||
3296 	       ext4_has_feature_metadata_csum(sb);
3297 }
3298 
3299 #define ext4_read_incompat_64bit_val(es, name) \
3300 	(((es)->s_feature_incompat & cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT) \
3301 		? (ext4_fsblk_t)le32_to_cpu(es->name##_hi) << 32 : 0) | \
3302 		le32_to_cpu(es->name##_lo))
3303 
ext4_blocks_count(struct ext4_super_block * es)3304 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
3305 {
3306 	return ext4_read_incompat_64bit_val(es, s_blocks_count);
3307 }
3308 
ext4_r_blocks_count(struct ext4_super_block * es)3309 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
3310 {
3311 	return ext4_read_incompat_64bit_val(es, s_r_blocks_count);
3312 }
3313 
ext4_free_blocks_count(struct ext4_super_block * es)3314 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
3315 {
3316 	return ext4_read_incompat_64bit_val(es, s_free_blocks_count);
3317 }
3318 
ext4_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)3319 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
3320 					 ext4_fsblk_t blk)
3321 {
3322 	es->s_blocks_count_lo = cpu_to_le32((u32)blk);
3323 	es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
3324 }
3325 
ext4_free_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)3326 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
3327 					      ext4_fsblk_t blk)
3328 {
3329 	es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
3330 	es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
3331 }
3332 
ext4_r_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)3333 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
3334 					   ext4_fsblk_t blk)
3335 {
3336 	es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
3337 	es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
3338 }
3339 
ext4_isize(struct super_block * sb,struct ext4_inode * raw_inode)3340 static inline loff_t ext4_isize(struct super_block *sb,
3341 				struct ext4_inode *raw_inode)
3342 {
3343 	if (ext4_has_feature_largedir(sb) ||
3344 	    S_ISREG(le16_to_cpu(raw_inode->i_mode)))
3345 		return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
3346 			le32_to_cpu(raw_inode->i_size_lo);
3347 
3348 	return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
3349 }
3350 
ext4_isize_set(struct ext4_inode * raw_inode,loff_t i_size)3351 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
3352 {
3353 	raw_inode->i_size_lo = cpu_to_le32(i_size);
3354 	raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
3355 }
3356 
3357 /*
3358  * Reading s_groups_count requires using smp_rmb() afterwards.  See
3359  * the locking protocol documented in the comments of ext4_group_add()
3360  * in resize.c
3361  */
ext4_get_groups_count(struct super_block * sb)3362 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
3363 {
3364 	ext4_group_t	ngroups = EXT4_SB(sb)->s_groups_count;
3365 
3366 	smp_rmb();
3367 	return ngroups;
3368 }
3369 
ext4_flex_group(struct ext4_sb_info * sbi,ext4_group_t block_group)3370 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
3371 					     ext4_group_t block_group)
3372 {
3373 	return block_group >> sbi->s_log_groups_per_flex;
3374 }
3375 
ext4_flex_bg_size(struct ext4_sb_info * sbi)3376 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
3377 {
3378 	return 1 << sbi->s_log_groups_per_flex;
3379 }
3380 
3381 #define ext4_std_error(sb, errno)				\
3382 do {								\
3383 	if ((errno))						\
3384 		__ext4_std_error((sb), __func__, __LINE__, (errno));	\
3385 } while (0)
3386 
3387 #ifdef CONFIG_SMP
3388 /* Each CPU can accumulate percpu_counter_batch clusters in their local
3389  * counters. So we need to make sure we have free clusters more
3390  * than percpu_counter_batch  * nr_cpu_ids. Also add a window of 4 times.
3391  */
3392 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
3393 #else
3394 #define EXT4_FREECLUSTERS_WATERMARK 0
3395 #endif
3396 
3397 /* Update i_disksize. Requires i_rwsem to avoid races with truncate */
ext4_update_i_disksize(struct inode * inode,loff_t newsize)3398 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
3399 {
3400 	WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
3401 		     !inode_is_locked(inode));
3402 	down_write(&EXT4_I(inode)->i_data_sem);
3403 	if (newsize > EXT4_I(inode)->i_disksize)
3404 		WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize);
3405 	up_write(&EXT4_I(inode)->i_data_sem);
3406 }
3407 
3408 /* Update i_size, i_disksize. Requires i_rwsem to avoid races with truncate */
ext4_update_inode_size(struct inode * inode,loff_t newsize)3409 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
3410 {
3411 	int changed = 0;
3412 
3413 	if (newsize > inode->i_size) {
3414 		i_size_write(inode, newsize);
3415 		changed = 1;
3416 	}
3417 	if (newsize > EXT4_I(inode)->i_disksize) {
3418 		ext4_update_i_disksize(inode, newsize);
3419 		changed |= 2;
3420 	}
3421 	return changed;
3422 }
3423 
3424 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
3425 				      loff_t len);
3426 
3427 struct ext4_group_info {
3428 	unsigned long   bb_state;
3429 #ifdef AGGRESSIVE_CHECK
3430 	unsigned long	bb_check_counter;
3431 #endif
3432 	struct rb_root  bb_free_root;
3433 	ext4_grpblk_t	bb_first_free;	/* first free block */
3434 	ext4_grpblk_t	bb_free;	/* total free blocks */
3435 	ext4_grpblk_t	bb_fragments;	/* nr of freespace fragments */
3436 	int		bb_avg_fragment_size_order;	/* order of average
3437 							   fragment in BG */
3438 	ext4_grpblk_t	bb_largest_free_order;/* order of largest frag in BG */
3439 	ext4_group_t	bb_group;	/* Group number */
3440 	struct          list_head bb_prealloc_list;
3441 #ifdef DOUBLE_CHECK
3442 	void            *bb_bitmap;
3443 #endif
3444 	struct rw_semaphore alloc_sem;
3445 	struct list_head bb_avg_fragment_size_node;
3446 	struct list_head bb_largest_free_order_node;
3447 	ext4_grpblk_t	bb_counters[];	/* Nr of free power-of-two-block
3448 					 * regions, index is order.
3449 					 * bb_counters[3] = 5 means
3450 					 * 5 free 8-block regions. */
3451 };
3452 
3453 #define EXT4_GROUP_INFO_NEED_INIT_BIT		0
3454 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT		1
3455 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT	2
3456 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT	3
3457 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT		\
3458 	(1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT)
3459 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT		\
3460 	(1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT)
3461 #define EXT4_GROUP_INFO_BBITMAP_READ_BIT	4
3462 
3463 #define EXT4_MB_GRP_NEED_INIT(grp)	\
3464 	(test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
3465 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp)	\
3466 	(test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3467 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp)	\
3468 	(test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3469 
3470 #define EXT4_MB_GRP_WAS_TRIMMED(grp)	\
3471 	(test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3472 #define EXT4_MB_GRP_SET_TRIMMED(grp)	\
3473 	(set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3474 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp)	\
3475 	(clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3476 #define EXT4_MB_GRP_TEST_AND_SET_READ(grp)	\
3477 	(test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state)))
3478 
3479 #define EXT4_MAX_CONTENTION		8
3480 #define EXT4_CONTENTION_THRESHOLD	2
3481 
ext4_group_lock_ptr(struct super_block * sb,ext4_group_t group)3482 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
3483 					      ext4_group_t group)
3484 {
3485 	return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
3486 }
3487 
3488 /*
3489  * Returns true if the filesystem is busy enough that attempts to
3490  * access the block group locks has run into contention.
3491  */
ext4_fs_is_busy(struct ext4_sb_info * sbi)3492 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
3493 {
3494 	return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
3495 }
3496 
ext4_lock_group(struct super_block * sb,ext4_group_t group)3497 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
3498 {
3499 	spinlock_t *lock = ext4_group_lock_ptr(sb, group);
3500 	if (spin_trylock(lock))
3501 		/*
3502 		 * We're able to grab the lock right away, so drop the
3503 		 * lock contention counter.
3504 		 */
3505 		atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
3506 	else {
3507 		/*
3508 		 * The lock is busy, so bump the contention counter,
3509 		 * and then wait on the spin lock.
3510 		 */
3511 		atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
3512 				  EXT4_MAX_CONTENTION);
3513 		spin_lock(lock);
3514 	}
3515 }
3516 
ext4_unlock_group(struct super_block * sb,ext4_group_t group)3517 static inline void ext4_unlock_group(struct super_block *sb,
3518 					ext4_group_t group)
3519 {
3520 	spin_unlock(ext4_group_lock_ptr(sb, group));
3521 }
3522 
3523 #ifdef CONFIG_QUOTA
ext4_quota_capable(struct super_block * sb)3524 static inline bool ext4_quota_capable(struct super_block *sb)
3525 {
3526 	return (test_opt(sb, QUOTA) || ext4_has_feature_quota(sb));
3527 }
3528 
ext4_is_quota_journalled(struct super_block * sb)3529 static inline bool ext4_is_quota_journalled(struct super_block *sb)
3530 {
3531 	struct ext4_sb_info *sbi = EXT4_SB(sb);
3532 
3533 	return (ext4_has_feature_quota(sb) ||
3534 		sbi->s_qf_names[USRQUOTA] || sbi->s_qf_names[GRPQUOTA]);
3535 }
3536 int ext4_enable_quotas(struct super_block *sb);
3537 #endif
3538 
3539 /*
3540  * Block validity checking
3541  */
3542 #define ext4_check_indirect_blockref(inode, bh)				\
3543 	ext4_check_blockref(__func__, __LINE__, inode,			\
3544 			    (__le32 *)(bh)->b_data,			\
3545 			    EXT4_ADDR_PER_BLOCK((inode)->i_sb))
3546 
3547 #define ext4_ind_check_inode(inode)					\
3548 	ext4_check_blockref(__func__, __LINE__, inode,			\
3549 			    EXT4_I(inode)->i_data,			\
3550 			    EXT4_NDIR_BLOCKS)
3551 
3552 /*
3553  * Inodes and files operations
3554  */
3555 
3556 /* dir.c */
3557 extern const struct file_operations ext4_dir_operations;
3558 
3559 /* file.c */
3560 extern const struct inode_operations ext4_file_inode_operations;
3561 extern const struct file_operations ext4_file_operations;
3562 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
3563 
3564 /* inline.c */
3565 extern int ext4_get_max_inline_size(struct inode *inode);
3566 extern int ext4_find_inline_data_nolock(struct inode *inode);
3567 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
3568 
3569 int ext4_readpage_inline(struct inode *inode, struct folio *folio);
3570 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3571 					 struct inode *inode,
3572 					 loff_t pos, unsigned len,
3573 					 struct folio **foliop);
3574 int ext4_write_inline_data_end(struct inode *inode, loff_t pos, unsigned len,
3575 			       unsigned copied, struct folio *folio);
3576 extern int ext4_generic_write_inline_data(struct address_space *mapping,
3577 					  struct inode *inode,
3578 					  loff_t pos, unsigned len,
3579 					  struct folio **foliop,
3580 					  void **fsdata, bool da);
3581 extern int ext4_try_add_inline_entry(handle_t *handle,
3582 				     struct ext4_filename *fname,
3583 				     struct inode *dir, struct inode *inode);
3584 extern int ext4_try_create_inline_dir(handle_t *handle,
3585 				      struct inode *parent,
3586 				      struct inode *inode);
3587 extern int ext4_read_inline_dir(struct file *filp,
3588 				struct dir_context *ctx,
3589 				int *has_inline_data);
3590 extern int ext4_inlinedir_to_tree(struct file *dir_file,
3591 				  struct inode *dir, ext4_lblk_t block,
3592 				  struct dx_hash_info *hinfo,
3593 				  __u32 start_hash, __u32 start_minor_hash,
3594 				  int *has_inline_data);
3595 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3596 					struct ext4_filename *fname,
3597 					struct ext4_dir_entry_2 **res_dir,
3598 					int *has_inline_data);
3599 extern int ext4_delete_inline_entry(handle_t *handle,
3600 				    struct inode *dir,
3601 				    struct ext4_dir_entry_2 *de_del,
3602 				    struct buffer_head *bh,
3603 				    int *has_inline_data);
3604 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3605 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3606 					struct ext4_dir_entry_2 **parent_de,
3607 					int *retval);
3608 extern void *ext4_read_inline_link(struct inode *inode);
3609 
3610 struct iomap;
3611 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap);
3612 
3613 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3614 
3615 extern int ext4_convert_inline_data(struct inode *inode);
3616 
ext4_has_inline_data(struct inode * inode)3617 static inline int ext4_has_inline_data(struct inode *inode)
3618 {
3619 	return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3620 	       EXT4_I(inode)->i_inline_off;
3621 }
3622 
3623 /* namei.c */
3624 extern const struct inode_operations ext4_dir_inode_operations;
3625 extern const struct inode_operations ext4_special_inode_operations;
3626 extern struct dentry *ext4_get_parent(struct dentry *child);
3627 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3628 				 struct ext4_dir_entry_2 *de,
3629 				 int blocksize, int csum_size,
3630 				 unsigned int parent_ino, int dotdot_real_len);
3631 extern void ext4_initialize_dirent_tail(struct buffer_head *bh,
3632 					unsigned int blocksize);
3633 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode,
3634 				      struct buffer_head *bh);
3635 extern int __ext4_unlink(struct inode *dir, const struct qstr *d_name,
3636 			 struct inode *inode, struct dentry *dentry);
3637 extern int __ext4_link(struct inode *dir, struct inode *inode,
3638 		       struct dentry *dentry);
3639 
3640 #define S_SHIFT 12
3641 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = {
3642 	[S_IFREG >> S_SHIFT]	= EXT4_FT_REG_FILE,
3643 	[S_IFDIR >> S_SHIFT]	= EXT4_FT_DIR,
3644 	[S_IFCHR >> S_SHIFT]	= EXT4_FT_CHRDEV,
3645 	[S_IFBLK >> S_SHIFT]	= EXT4_FT_BLKDEV,
3646 	[S_IFIFO >> S_SHIFT]	= EXT4_FT_FIFO,
3647 	[S_IFSOCK >> S_SHIFT]	= EXT4_FT_SOCK,
3648 	[S_IFLNK >> S_SHIFT]	= EXT4_FT_SYMLINK,
3649 };
3650 
ext4_set_de_type(struct super_block * sb,struct ext4_dir_entry_2 * de,umode_t mode)3651 static inline void ext4_set_de_type(struct super_block *sb,
3652 				struct ext4_dir_entry_2 *de,
3653 				umode_t mode) {
3654 	if (ext4_has_feature_filetype(sb))
3655 		de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3656 }
3657 
3658 /* readpages.c */
3659 extern int ext4_mpage_readpages(struct inode *inode,
3660 		struct readahead_control *rac, struct folio *folio);
3661 extern int __init ext4_init_post_read_processing(void);
3662 extern void ext4_exit_post_read_processing(void);
3663 
3664 /* symlink.c */
3665 extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3666 extern const struct inode_operations ext4_symlink_inode_operations;
3667 extern const struct inode_operations ext4_fast_symlink_inode_operations;
3668 
3669 /* sysfs.c */
3670 extern void ext4_notify_error_sysfs(struct ext4_sb_info *sbi);
3671 extern int ext4_register_sysfs(struct super_block *sb);
3672 extern void ext4_unregister_sysfs(struct super_block *sb);
3673 extern int __init ext4_init_sysfs(void);
3674 extern void ext4_exit_sysfs(void);
3675 
3676 /* block_validity */
3677 extern void ext4_release_system_zone(struct super_block *sb);
3678 extern int ext4_setup_system_zone(struct super_block *sb);
3679 extern int __init ext4_init_system_zone(void);
3680 extern void ext4_exit_system_zone(void);
3681 extern int ext4_inode_block_valid(struct inode *inode,
3682 				  ext4_fsblk_t start_blk,
3683 				  unsigned int count);
3684 extern int ext4_check_blockref(const char *, unsigned int,
3685 			       struct inode *, __le32 *, unsigned int);
3686 extern int ext4_sb_block_valid(struct super_block *sb, struct inode *inode,
3687 				ext4_fsblk_t start_blk, unsigned int count);
3688 
3689 
3690 /* extents.c */
3691 struct ext4_ext_path;
3692 struct ext4_extent;
3693 
3694 /*
3695  * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3696  * __le32.
3697  */
3698 #define EXT_MAX_BLOCKS	0xffffffff
3699 
3700 extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode);
3701 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3702 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3703 			       struct ext4_map_blocks *map, int flags);
3704 extern int ext4_ext_truncate(handle_t *, struct inode *);
3705 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3706 				 ext4_lblk_t end);
3707 extern void ext4_ext_init(struct super_block *);
3708 extern void ext4_ext_release(struct super_block *);
3709 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3710 			  loff_t len);
3711 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3712 					  loff_t offset, ssize_t len);
3713 extern int ext4_convert_unwritten_io_end_vec(handle_t *handle,
3714 					     ext4_io_end_t *io_end);
3715 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3716 			   struct ext4_map_blocks *map, int flags);
3717 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3718 						   int num,
3719 						   struct ext4_ext_path *path);
3720 extern struct ext4_ext_path *ext4_ext_insert_extent(
3721 				handle_t *handle, struct inode *inode,
3722 				struct ext4_ext_path *path,
3723 				struct ext4_extent *newext, int gb_flags);
3724 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3725 					      struct ext4_ext_path *,
3726 					      int flags);
3727 extern void ext4_free_ext_path(struct ext4_ext_path *);
3728 extern int ext4_ext_check_inode(struct inode *inode);
3729 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3730 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3731 			__u64 start, __u64 len);
3732 extern int ext4_get_es_cache(struct inode *inode,
3733 			     struct fiemap_extent_info *fieinfo,
3734 			     __u64 start, __u64 len);
3735 extern int ext4_ext_precache(struct inode *inode);
3736 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3737 				struct inode *inode2, ext4_lblk_t lblk1,
3738 			     ext4_lblk_t lblk2,  ext4_lblk_t count,
3739 			     int mark_unwritten,int *err);
3740 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu);
3741 extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode,
3742 				       int check_cred, int restart_cred,
3743 				       int revoke_cred);
3744 extern void ext4_ext_replay_shrink_inode(struct inode *inode, ext4_lblk_t end);
3745 extern int ext4_ext_replay_set_iblocks(struct inode *inode);
3746 extern int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start,
3747 		int len, int unwritten, ext4_fsblk_t pblk);
3748 extern int ext4_ext_clear_bb(struct inode *inode);
3749 
3750 
3751 /* move_extent.c */
3752 extern void ext4_double_down_write_data_sem(struct inode *first,
3753 					    struct inode *second);
3754 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3755 					  struct inode *donor_inode);
3756 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3757 			     __u64 start_orig, __u64 start_donor,
3758 			     __u64 len, __u64 *moved_len);
3759 
3760 /* page-io.c */
3761 extern int __init ext4_init_pageio(void);
3762 extern void ext4_exit_pageio(void);
3763 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3764 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3765 extern int ext4_put_io_end(ext4_io_end_t *io_end);
3766 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3767 extern void ext4_io_submit_init(struct ext4_io_submit *io,
3768 				struct writeback_control *wbc);
3769 extern void ext4_end_io_rsv_work(struct work_struct *work);
3770 extern void ext4_io_submit(struct ext4_io_submit *io);
3771 int ext4_bio_write_folio(struct ext4_io_submit *io, struct folio *page,
3772 		size_t len);
3773 extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end);
3774 extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end);
3775 
3776 /* mmp.c */
3777 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3778 
3779 /* mmp.c */
3780 extern void ext4_stop_mmpd(struct ext4_sb_info *sbi);
3781 
3782 /* verity.c */
3783 extern const struct fsverity_operations ext4_verityops;
3784 
3785 /* orphan.c */
3786 extern int ext4_orphan_add(handle_t *, struct inode *);
3787 extern int ext4_orphan_del(handle_t *, struct inode *);
3788 extern void ext4_orphan_cleanup(struct super_block *sb,
3789 				struct ext4_super_block *es);
3790 extern void ext4_release_orphan_info(struct super_block *sb);
3791 extern int ext4_init_orphan_info(struct super_block *sb);
3792 extern int ext4_orphan_file_empty(struct super_block *sb);
3793 extern void ext4_orphan_file_block_trigger(
3794 				struct jbd2_buffer_trigger_type *triggers,
3795 				struct buffer_head *bh,
3796 				void *data, size_t size);
3797 
3798 /*
3799  * Add new method to test whether block and inode bitmaps are properly
3800  * initialized. With uninit_bg reading the block from disk is not enough
3801  * to mark the bitmap uptodate. We need to also zero-out the bitmap
3802  */
3803 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3804 
bitmap_uptodate(struct buffer_head * bh)3805 static inline int bitmap_uptodate(struct buffer_head *bh)
3806 {
3807 	return (buffer_uptodate(bh) &&
3808 			test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3809 }
set_bitmap_uptodate(struct buffer_head * bh)3810 static inline void set_bitmap_uptodate(struct buffer_head *bh)
3811 {
3812 	set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3813 }
3814 
3815 extern int ext4_resize_begin(struct super_block *sb);
3816 extern int ext4_resize_end(struct super_block *sb, bool update_backups);
3817 
ext4_set_io_unwritten_flag(struct ext4_io_end * io_end)3818 static inline void ext4_set_io_unwritten_flag(struct ext4_io_end *io_end)
3819 {
3820 	if (!(io_end->flag & EXT4_IO_END_UNWRITTEN))
3821 		io_end->flag |= EXT4_IO_END_UNWRITTEN;
3822 }
3823 
ext4_clear_io_unwritten_flag(ext4_io_end_t * io_end)3824 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3825 {
3826 	if (io_end->flag & EXT4_IO_END_UNWRITTEN)
3827 		io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3828 }
3829 
3830 extern const struct iomap_ops ext4_iomap_ops;
3831 extern const struct iomap_ops ext4_iomap_overwrite_ops;
3832 extern const struct iomap_ops ext4_iomap_report_ops;
3833 
ext4_buffer_uptodate(struct buffer_head * bh)3834 static inline int ext4_buffer_uptodate(struct buffer_head *bh)
3835 {
3836 	/*
3837 	 * If the buffer has the write error flag, we have failed
3838 	 * to write out data in the block.  In this  case, we don't
3839 	 * have to read the block because we may read the old data
3840 	 * successfully.
3841 	 */
3842 	if (buffer_write_io_error(bh))
3843 		set_buffer_uptodate(bh);
3844 	return buffer_uptodate(bh);
3845 }
3846 
ext4_inode_can_atomic_write(struct inode * inode)3847 static inline bool ext4_inode_can_atomic_write(struct inode *inode)
3848 {
3849 
3850 	return S_ISREG(inode->i_mode) && EXT4_SB(inode->i_sb)->s_awu_min > 0;
3851 }
3852 
3853 extern int ext4_block_write_begin(handle_t *handle, struct folio *folio,
3854 				  loff_t pos, unsigned len,
3855 				  get_block_t *get_block);
3856 #endif	/* __KERNEL__ */
3857 
3858 #define EFSBADCRC	EBADMSG		/* Bad CRC detected */
3859 #define EFSCORRUPTED	EUCLEAN		/* Filesystem is corrupted */
3860 
3861 #endif	/* _EXT4_H */
3862