• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 // SPDX-License-Identifier: GPL-2.0
2 /*
3  *  ext4.h
4  *
5  * Copyright (C) 1992, 1993, 1994, 1995
6  * Remy Card (card@masi.ibp.fr)
7  * Laboratoire MASI - Institut Blaise Pascal
8  * Universite Pierre et Marie Curie (Paris VI)
9  *
10  *  from
11  *
12  *  linux/include/linux/minix_fs.h
13  *
14  *  Copyright (C) 1991, 1992  Linus Torvalds
15  */
16 
17 #ifndef _EXT4_H
18 #define _EXT4_H
19 
20 #include <linux/types.h>
21 #include <linux/blkdev.h>
22 #include <linux/magic.h>
23 #include <linux/jbd2.h>
24 #include <linux/quota.h>
25 #include <linux/rwsem.h>
26 #include <linux/rbtree.h>
27 #include <linux/seqlock.h>
28 #include <linux/mutex.h>
29 #include <linux/timer.h>
30 #include <linux/version.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 <crypto/hash.h>
37 #include <linux/falloc.h>
38 #include <linux/percpu-rwsem.h>
39 #ifdef __KERNEL__
40 #include <linux/compat.h>
41 #endif
42 
43 #define FSCRYPT_NEED_OPS
44 #include <linux/fscrypt.h>
45 #include <linux/fsverity.h>
46 
47 #include <linux/compiler.h>
48 
49 /*
50  * The fourth extended filesystem constants/structures
51  */
52 
53 /*
54  * with AGGRESSIVE_CHECK allocator runs consistency checks over
55  * structures. these checks slow things down a lot
56  */
57 #define AGGRESSIVE_CHECK__
58 
59 /*
60  * with DOUBLE_CHECK defined mballoc creates persistent in-core
61  * bitmaps, maintains and uses them to check for double allocations
62  */
63 #define DOUBLE_CHECK__
64 
65 /*
66  * Define EXT4FS_DEBUG to produce debug messages
67  */
68 #undef EXT4FS_DEBUG
69 
70 /*
71  * Debug code
72  */
73 #ifdef EXT4FS_DEBUG
74 #define ext4_debug(f, a...)						\
75 	do {								\
76 		printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:",	\
77 			__FILE__, __LINE__, __func__);			\
78 		printk(KERN_DEBUG f, ## a);				\
79 	} while (0)
80 #else
81 #define ext4_debug(fmt, ...)	no_printk(fmt, ##__VA_ARGS__)
82 #endif
83 
84 /*
85  * Turn on EXT_DEBUG to get lots of info about extents operations.
86  */
87 #define EXT_DEBUG__
88 #ifdef EXT_DEBUG
89 #define ext_debug(fmt, ...)	printk(fmt, ##__VA_ARGS__)
90 #else
91 #define ext_debug(fmt, ...)	no_printk(fmt, ##__VA_ARGS__)
92 #endif
93 
94 /* data type for block offset of block group */
95 typedef int ext4_grpblk_t;
96 
97 /* data type for filesystem-wide blocks number */
98 typedef unsigned long long ext4_fsblk_t;
99 
100 /* data type for file logical block number */
101 typedef __u32 ext4_lblk_t;
102 
103 /* data type for block group number */
104 typedef unsigned int ext4_group_t;
105 
106 enum SHIFT_DIRECTION {
107 	SHIFT_LEFT = 0,
108 	SHIFT_RIGHT,
109 };
110 
111 /*
112  * Flags used in mballoc's allocation_context flags field.
113  *
114  * Also used to show what's going on for debugging purposes when the
115  * flag field is exported via the traceport interface
116  */
117 
118 /* prefer goal again. length */
119 #define EXT4_MB_HINT_MERGE		0x0001
120 /* blocks already reserved */
121 #define EXT4_MB_HINT_RESERVED		0x0002
122 /* metadata is being allocated */
123 #define EXT4_MB_HINT_METADATA		0x0004
124 /* first blocks in the file */
125 #define EXT4_MB_HINT_FIRST		0x0008
126 /* search for the best chunk */
127 #define EXT4_MB_HINT_BEST		0x0010
128 /* data is being allocated */
129 #define EXT4_MB_HINT_DATA		0x0020
130 /* don't preallocate (for tails) */
131 #define EXT4_MB_HINT_NOPREALLOC		0x0040
132 /* allocate for locality group */
133 #define EXT4_MB_HINT_GROUP_ALLOC	0x0080
134 /* allocate goal blocks or none */
135 #define EXT4_MB_HINT_GOAL_ONLY		0x0100
136 /* goal is meaningful */
137 #define EXT4_MB_HINT_TRY_GOAL		0x0200
138 /* blocks already pre-reserved by delayed allocation */
139 #define EXT4_MB_DELALLOC_RESERVED	0x0400
140 /* We are doing stream allocation */
141 #define EXT4_MB_STREAM_ALLOC		0x0800
142 /* Use reserved root blocks if needed */
143 #define EXT4_MB_USE_ROOT_BLOCKS		0x1000
144 /* Use blocks from reserved pool */
145 #define EXT4_MB_USE_RESERVED		0x2000
146 
147 struct ext4_allocation_request {
148 	/* target inode for block we're allocating */
149 	struct inode *inode;
150 	/* how many blocks we want to allocate */
151 	unsigned int len;
152 	/* logical block in target inode */
153 	ext4_lblk_t logical;
154 	/* the closest logical allocated block to the left */
155 	ext4_lblk_t lleft;
156 	/* the closest logical allocated block to the right */
157 	ext4_lblk_t lright;
158 	/* phys. target (a hint) */
159 	ext4_fsblk_t goal;
160 	/* phys. block for the closest logical allocated block to the left */
161 	ext4_fsblk_t pleft;
162 	/* phys. block for the closest logical allocated block to the right */
163 	ext4_fsblk_t pright;
164 	/* flags. see above EXT4_MB_HINT_* */
165 	unsigned int flags;
166 };
167 
168 /*
169  * Logical to physical block mapping, used by ext4_map_blocks()
170  *
171  * This structure is used to pass requests into ext4_map_blocks() as
172  * well as to store the information returned by ext4_map_blocks().  It
173  * takes less room on the stack than a struct buffer_head.
174  */
175 #define EXT4_MAP_NEW		(1 << BH_New)
176 #define EXT4_MAP_MAPPED		(1 << BH_Mapped)
177 #define EXT4_MAP_UNWRITTEN	(1 << BH_Unwritten)
178 #define EXT4_MAP_BOUNDARY	(1 << BH_Boundary)
179 #define EXT4_MAP_FLAGS		(EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
180 				 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
181 
182 struct ext4_map_blocks {
183 	ext4_fsblk_t m_pblk;
184 	ext4_lblk_t m_lblk;
185 	unsigned int m_len;
186 	unsigned int m_flags;
187 };
188 
189 /*
190  * Block validity checking, system zone rbtree.
191  */
192 struct ext4_system_blocks {
193 	struct rb_root root;
194 	struct rcu_head rcu;
195 };
196 
197 /*
198  * Flags for ext4_io_end->flags
199  */
200 #define	EXT4_IO_END_UNWRITTEN	0x0001
201 
202 /*
203  * For converting unwritten extents on a work queue. 'handle' is used for
204  * buffered writeback.
205  */
206 typedef struct ext4_io_end {
207 	struct list_head	list;		/* per-file finished IO list */
208 	handle_t		*handle;	/* handle reserved for extent
209 						 * conversion */
210 	struct inode		*inode;		/* file being written to */
211 	struct bio		*bio;		/* Linked list of completed
212 						 * bios covering the extent */
213 	unsigned int		flag;		/* unwritten or not */
214 	atomic_t		count;		/* reference counter */
215 	loff_t			offset;		/* offset in the file */
216 	ssize_t			size;		/* size of the extent */
217 } ext4_io_end_t;
218 
219 struct ext4_io_submit {
220 	struct writeback_control *io_wbc;
221 	struct bio		*io_bio;
222 	ext4_io_end_t		*io_end;
223 	sector_t		io_next_block;
224 };
225 
226 /*
227  * Special inodes numbers
228  */
229 #define	EXT4_BAD_INO		 1	/* Bad blocks inode */
230 #define EXT4_ROOT_INO		 2	/* Root inode */
231 #define EXT4_USR_QUOTA_INO	 3	/* User quota inode */
232 #define EXT4_GRP_QUOTA_INO	 4	/* Group quota inode */
233 #define EXT4_BOOT_LOADER_INO	 5	/* Boot loader inode */
234 #define EXT4_UNDEL_DIR_INO	 6	/* Undelete directory inode */
235 #define EXT4_RESIZE_INO		 7	/* Reserved group descriptors inode */
236 #define EXT4_JOURNAL_INO	 8	/* Journal inode */
237 
238 /* First non-reserved inode for old ext4 filesystems */
239 #define EXT4_GOOD_OLD_FIRST_INO	11
240 
241 /*
242  * Maximal count of links to a file
243  */
244 #define EXT4_LINK_MAX		65000
245 
246 /*
247  * Macro-instructions used to manage several block sizes
248  */
249 #define EXT4_MIN_BLOCK_SIZE		1024
250 #define	EXT4_MAX_BLOCK_SIZE		65536
251 #define EXT4_MIN_BLOCK_LOG_SIZE		10
252 #define EXT4_MAX_BLOCK_LOG_SIZE		16
253 #define EXT4_MAX_CLUSTER_LOG_SIZE	30
254 #ifdef __KERNEL__
255 # define EXT4_BLOCK_SIZE(s)		((s)->s_blocksize)
256 #else
257 # define EXT4_BLOCK_SIZE(s)		(EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
258 #endif
259 #define	EXT4_ADDR_PER_BLOCK(s)		(EXT4_BLOCK_SIZE(s) / sizeof(__u32))
260 #define EXT4_CLUSTER_SIZE(s)		(EXT4_BLOCK_SIZE(s) << \
261 					 EXT4_SB(s)->s_cluster_bits)
262 #ifdef __KERNEL__
263 # define EXT4_BLOCK_SIZE_BITS(s)	((s)->s_blocksize_bits)
264 # define EXT4_CLUSTER_BITS(s)		(EXT4_SB(s)->s_cluster_bits)
265 #else
266 # define EXT4_BLOCK_SIZE_BITS(s)	((s)->s_log_block_size + 10)
267 #endif
268 #ifdef __KERNEL__
269 #define	EXT4_ADDR_PER_BLOCK_BITS(s)	(EXT4_SB(s)->s_addr_per_block_bits)
270 #define EXT4_INODE_SIZE(s)		(EXT4_SB(s)->s_inode_size)
271 #define EXT4_FIRST_INO(s)		(EXT4_SB(s)->s_first_ino)
272 #else
273 #define EXT4_INODE_SIZE(s)	(((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
274 				 EXT4_GOOD_OLD_INODE_SIZE : \
275 				 (s)->s_inode_size)
276 #define EXT4_FIRST_INO(s)	(((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
277 				 EXT4_GOOD_OLD_FIRST_INO : \
278 				 (s)->s_first_ino)
279 #endif
280 #define EXT4_BLOCK_ALIGN(size, blkbits)		ALIGN((size), (1 << (blkbits)))
281 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \
282 	((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \
283 								  blkbits))
284 
285 /* Translate a block number to a cluster number */
286 #define EXT4_B2C(sbi, blk)	((blk) >> (sbi)->s_cluster_bits)
287 /* Translate a cluster number to a block number */
288 #define EXT4_C2B(sbi, cluster)	((cluster) << (sbi)->s_cluster_bits)
289 /* Translate # of blks to # of clusters */
290 #define EXT4_NUM_B2C(sbi, blks)	(((blks) + (sbi)->s_cluster_ratio - 1) >> \
291 				 (sbi)->s_cluster_bits)
292 /* Mask out the low bits to get the starting block of the cluster */
293 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) &				\
294 				  ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
295 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) &				\
296 				  ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
297 /* Fill in the low bits to get the last block of the cluster */
298 #define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) |				\
299 				    ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1))
300 /* Get the cluster offset */
301 #define EXT4_PBLK_COFF(s, pblk) ((pblk) &				\
302 				 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
303 #define EXT4_LBLK_COFF(s, lblk) ((lblk) &				\
304 				 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
305 
306 /*
307  * Structure of a blocks group descriptor
308  */
309 struct ext4_group_desc
310 {
311 	__le32	bg_block_bitmap_lo;	/* Blocks bitmap block */
312 	__le32	bg_inode_bitmap_lo;	/* Inodes bitmap block */
313 	__le32	bg_inode_table_lo;	/* Inodes table block */
314 	__le16	bg_free_blocks_count_lo;/* Free blocks count */
315 	__le16	bg_free_inodes_count_lo;/* Free inodes count */
316 	__le16	bg_used_dirs_count_lo;	/* Directories count */
317 	__le16	bg_flags;		/* EXT4_BG_flags (INODE_UNINIT, etc) */
318 	__le32  bg_exclude_bitmap_lo;   /* Exclude bitmap for snapshots */
319 	__le16  bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
320 	__le16  bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
321 	__le16  bg_itable_unused_lo;	/* Unused inodes count */
322 	__le16  bg_checksum;		/* crc16(sb_uuid+group+desc) */
323 	__le32	bg_block_bitmap_hi;	/* Blocks bitmap block MSB */
324 	__le32	bg_inode_bitmap_hi;	/* Inodes bitmap block MSB */
325 	__le32	bg_inode_table_hi;	/* Inodes table block MSB */
326 	__le16	bg_free_blocks_count_hi;/* Free blocks count MSB */
327 	__le16	bg_free_inodes_count_hi;/* Free inodes count MSB */
328 	__le16	bg_used_dirs_count_hi;	/* Directories count MSB */
329 	__le16  bg_itable_unused_hi;    /* Unused inodes count MSB */
330 	__le32  bg_exclude_bitmap_hi;   /* Exclude bitmap block MSB */
331 	__le16  bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
332 	__le16  bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
333 	__u32   bg_reserved;
334 };
335 
336 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END	\
337 	(offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
338 	 sizeof(__le16))
339 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END	\
340 	(offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
341 	 sizeof(__le16))
342 
343 /*
344  * Structure of a flex block group info
345  */
346 
347 struct flex_groups {
348 	atomic64_t	free_clusters;
349 	atomic_t	free_inodes;
350 	atomic_t	used_dirs;
351 };
352 
353 #define EXT4_BG_INODE_UNINIT	0x0001 /* Inode table/bitmap not in use */
354 #define EXT4_BG_BLOCK_UNINIT	0x0002 /* Block bitmap not in use */
355 #define EXT4_BG_INODE_ZEROED	0x0004 /* On-disk itable initialized to zero */
356 
357 /*
358  * Macro-instructions used to manage group descriptors
359  */
360 #define EXT4_MIN_DESC_SIZE		32
361 #define EXT4_MIN_DESC_SIZE_64BIT	64
362 #define	EXT4_MAX_DESC_SIZE		EXT4_MIN_BLOCK_SIZE
363 #define EXT4_DESC_SIZE(s)		(EXT4_SB(s)->s_desc_size)
364 #ifdef __KERNEL__
365 # define EXT4_BLOCKS_PER_GROUP(s)	(EXT4_SB(s)->s_blocks_per_group)
366 # define EXT4_CLUSTERS_PER_GROUP(s)	(EXT4_SB(s)->s_clusters_per_group)
367 # define EXT4_DESC_PER_BLOCK(s)		(EXT4_SB(s)->s_desc_per_block)
368 # define EXT4_INODES_PER_GROUP(s)	(EXT4_SB(s)->s_inodes_per_group)
369 # define EXT4_DESC_PER_BLOCK_BITS(s)	(EXT4_SB(s)->s_desc_per_block_bits)
370 #else
371 # define EXT4_BLOCKS_PER_GROUP(s)	((s)->s_blocks_per_group)
372 # define EXT4_DESC_PER_BLOCK(s)		(EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
373 # define EXT4_INODES_PER_GROUP(s)	((s)->s_inodes_per_group)
374 #endif
375 
376 /*
377  * Constants relative to the data blocks
378  */
379 #define	EXT4_NDIR_BLOCKS		12
380 #define	EXT4_IND_BLOCK			EXT4_NDIR_BLOCKS
381 #define	EXT4_DIND_BLOCK			(EXT4_IND_BLOCK + 1)
382 #define	EXT4_TIND_BLOCK			(EXT4_DIND_BLOCK + 1)
383 #define	EXT4_N_BLOCKS			(EXT4_TIND_BLOCK + 1)
384 
385 /*
386  * Inode flags
387  */
388 #define	EXT4_SECRM_FL			0x00000001 /* Secure deletion */
389 #define	EXT4_UNRM_FL			0x00000002 /* Undelete */
390 #define	EXT4_COMPR_FL			0x00000004 /* Compress file */
391 #define EXT4_SYNC_FL			0x00000008 /* Synchronous updates */
392 #define EXT4_IMMUTABLE_FL		0x00000010 /* Immutable file */
393 #define EXT4_APPEND_FL			0x00000020 /* writes to file may only append */
394 #define EXT4_NODUMP_FL			0x00000040 /* do not dump file */
395 #define EXT4_NOATIME_FL			0x00000080 /* do not update atime */
396 /* Reserved for compression usage... */
397 #define EXT4_DIRTY_FL			0x00000100
398 #define EXT4_COMPRBLK_FL		0x00000200 /* One or more compressed clusters */
399 #define EXT4_NOCOMPR_FL			0x00000400 /* Don't compress */
400 	/* nb: was previously EXT2_ECOMPR_FL */
401 #define EXT4_ENCRYPT_FL			0x00000800 /* encrypted file */
402 /* End compression flags --- maybe not all used */
403 #define EXT4_INDEX_FL			0x00001000 /* hash-indexed directory */
404 #define EXT4_IMAGIC_FL			0x00002000 /* AFS directory */
405 #define EXT4_JOURNAL_DATA_FL		0x00004000 /* file data should be journaled */
406 #define EXT4_NOTAIL_FL			0x00008000 /* file tail should not be merged */
407 #define EXT4_DIRSYNC_FL			0x00010000 /* dirsync behaviour (directories only) */
408 #define EXT4_TOPDIR_FL			0x00020000 /* Top of directory hierarchies*/
409 #define EXT4_HUGE_FILE_FL               0x00040000 /* Set to each huge file */
410 #define EXT4_EXTENTS_FL			0x00080000 /* Inode uses extents */
411 #define EXT4_VERITY_FL			0x00100000 /* Verity protected inode */
412 #define EXT4_EA_INODE_FL	        0x00200000 /* Inode used for large EA */
413 #define EXT4_EOFBLOCKS_FL		0x00400000 /* Blocks allocated beyond EOF */
414 #define EXT4_INLINE_DATA_FL		0x10000000 /* Inode has inline data. */
415 #define EXT4_PROJINHERIT_FL		0x20000000 /* Create with parents projid */
416 #define EXT4_CASEFOLD_FL		0x40000000 /* Casefolded file */
417 #define EXT4_RESERVED_FL		0x80000000 /* reserved for ext4 lib */
418 
419 #define EXT4_FL_USER_VISIBLE		0x705BDFFF /* User visible flags */
420 #define EXT4_FL_USER_MODIFIABLE		0x604BC0FF /* User modifiable flags */
421 
422 /* Flags we can manipulate with through EXT4_IOC_FSSETXATTR */
423 #define EXT4_FL_XFLAG_VISIBLE		(EXT4_SYNC_FL | \
424 					 EXT4_IMMUTABLE_FL | \
425 					 EXT4_APPEND_FL | \
426 					 EXT4_NODUMP_FL | \
427 					 EXT4_NOATIME_FL | \
428 					 EXT4_PROJINHERIT_FL)
429 
430 /* Flags that should be inherited by new inodes from their parent. */
431 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
432 			   EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
433 			   EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
434 			   EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
435 			   EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL)
436 
437 /* Flags that are appropriate for regular files (all but dir-specific ones). */
438 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\
439 			   EXT4_PROJINHERIT_FL))
440 
441 /* Flags that are appropriate for non-directories/regular files. */
442 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
443 
444 /* The only flags that should be swapped */
445 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL)
446 
447 /* Mask out flags that are inappropriate for the given type of inode. */
ext4_mask_flags(umode_t mode,__u32 flags)448 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
449 {
450 	if (S_ISDIR(mode))
451 		return flags;
452 	else if (S_ISREG(mode))
453 		return flags & EXT4_REG_FLMASK;
454 	else
455 		return flags & EXT4_OTHER_FLMASK;
456 }
457 
458 /*
459  * Inode flags used for atomic set/get
460  */
461 enum {
462 	EXT4_INODE_SECRM	= 0,	/* Secure deletion */
463 	EXT4_INODE_UNRM		= 1,	/* Undelete */
464 	EXT4_INODE_COMPR	= 2,	/* Compress file */
465 	EXT4_INODE_SYNC		= 3,	/* Synchronous updates */
466 	EXT4_INODE_IMMUTABLE	= 4,	/* Immutable file */
467 	EXT4_INODE_APPEND	= 5,	/* writes to file may only append */
468 	EXT4_INODE_NODUMP	= 6,	/* do not dump file */
469 	EXT4_INODE_NOATIME	= 7,	/* do not update atime */
470 /* Reserved for compression usage... */
471 	EXT4_INODE_DIRTY	= 8,
472 	EXT4_INODE_COMPRBLK	= 9,	/* One or more compressed clusters */
473 	EXT4_INODE_NOCOMPR	= 10,	/* Don't compress */
474 	EXT4_INODE_ENCRYPT	= 11,	/* Encrypted file */
475 /* End compression flags --- maybe not all used */
476 	EXT4_INODE_INDEX	= 12,	/* hash-indexed directory */
477 	EXT4_INODE_IMAGIC	= 13,	/* AFS directory */
478 	EXT4_INODE_JOURNAL_DATA	= 14,	/* file data should be journaled */
479 	EXT4_INODE_NOTAIL	= 15,	/* file tail should not be merged */
480 	EXT4_INODE_DIRSYNC	= 16,	/* dirsync behaviour (directories only) */
481 	EXT4_INODE_TOPDIR	= 17,	/* Top of directory hierarchies*/
482 	EXT4_INODE_HUGE_FILE	= 18,	/* Set to each huge file */
483 	EXT4_INODE_EXTENTS	= 19,	/* Inode uses extents */
484 	EXT4_INODE_VERITY	= 20,	/* Verity protected inode */
485 	EXT4_INODE_EA_INODE	= 21,	/* Inode used for large EA */
486 	EXT4_INODE_EOFBLOCKS	= 22,	/* Blocks allocated beyond EOF */
487 	EXT4_INODE_INLINE_DATA	= 28,	/* Data in inode. */
488 	EXT4_INODE_PROJINHERIT	= 29,	/* Create with parents projid */
489 	EXT4_INODE_RESERVED	= 31,	/* reserved for ext4 lib */
490 };
491 
492 /*
493  * Since it's pretty easy to mix up bit numbers and hex values, we use a
494  * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
495  * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
496  * any extra space in the compiled kernel image, otherwise, the build will fail.
497  * It's important that these values are the same, since we are using
498  * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
499  * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
500  * values found in ext2, ext3 and ext4 filesystems, and of course the values
501  * defined in e2fsprogs.
502  *
503  * It's not paranoia if the Murphy's Law really *is* out to get you.  :-)
504  */
505 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1U << EXT4_INODE_##FLAG))
506 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
507 
ext4_check_flag_values(void)508 static inline void ext4_check_flag_values(void)
509 {
510 	CHECK_FLAG_VALUE(SECRM);
511 	CHECK_FLAG_VALUE(UNRM);
512 	CHECK_FLAG_VALUE(COMPR);
513 	CHECK_FLAG_VALUE(SYNC);
514 	CHECK_FLAG_VALUE(IMMUTABLE);
515 	CHECK_FLAG_VALUE(APPEND);
516 	CHECK_FLAG_VALUE(NODUMP);
517 	CHECK_FLAG_VALUE(NOATIME);
518 	CHECK_FLAG_VALUE(DIRTY);
519 	CHECK_FLAG_VALUE(COMPRBLK);
520 	CHECK_FLAG_VALUE(NOCOMPR);
521 	CHECK_FLAG_VALUE(ENCRYPT);
522 	CHECK_FLAG_VALUE(INDEX);
523 	CHECK_FLAG_VALUE(IMAGIC);
524 	CHECK_FLAG_VALUE(JOURNAL_DATA);
525 	CHECK_FLAG_VALUE(NOTAIL);
526 	CHECK_FLAG_VALUE(DIRSYNC);
527 	CHECK_FLAG_VALUE(TOPDIR);
528 	CHECK_FLAG_VALUE(HUGE_FILE);
529 	CHECK_FLAG_VALUE(EXTENTS);
530 	CHECK_FLAG_VALUE(VERITY);
531 	CHECK_FLAG_VALUE(EA_INODE);
532 	CHECK_FLAG_VALUE(EOFBLOCKS);
533 	CHECK_FLAG_VALUE(INLINE_DATA);
534 	CHECK_FLAG_VALUE(PROJINHERIT);
535 	CHECK_FLAG_VALUE(RESERVED);
536 }
537 
538 /* Used to pass group descriptor data when online resize is done */
539 struct ext4_new_group_input {
540 	__u32 group;		/* Group number for this data */
541 	__u64 block_bitmap;	/* Absolute block number of block bitmap */
542 	__u64 inode_bitmap;	/* Absolute block number of inode bitmap */
543 	__u64 inode_table;	/* Absolute block number of inode table start */
544 	__u32 blocks_count;	/* Total number of blocks in this group */
545 	__u16 reserved_blocks;	/* Number of reserved blocks in this group */
546 	__u16 unused;
547 };
548 
549 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
550 struct compat_ext4_new_group_input {
551 	u32 group;
552 	compat_u64 block_bitmap;
553 	compat_u64 inode_bitmap;
554 	compat_u64 inode_table;
555 	u32 blocks_count;
556 	u16 reserved_blocks;
557 	u16 unused;
558 };
559 #endif
560 
561 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
562 struct ext4_new_group_data {
563 	__u32 group;
564 	__u64 block_bitmap;
565 	__u64 inode_bitmap;
566 	__u64 inode_table;
567 	__u32 blocks_count;
568 	__u16 reserved_blocks;
569 	__u16 mdata_blocks;
570 	__u32 free_clusters_count;
571 };
572 
573 /* Indexes used to index group tables in ext4_new_group_data */
574 enum {
575 	BLOCK_BITMAP = 0,	/* block bitmap */
576 	INODE_BITMAP,		/* inode bitmap */
577 	INODE_TABLE,		/* inode tables */
578 	GROUP_TABLE_COUNT,
579 };
580 
581 /*
582  * Flags used by ext4_map_blocks()
583  */
584 	/* Allocate any needed blocks and/or convert an unwritten
585 	   extent to be an initialized ext4 */
586 #define EXT4_GET_BLOCKS_CREATE			0x0001
587 	/* Request the creation of an unwritten extent */
588 #define EXT4_GET_BLOCKS_UNWRIT_EXT		0x0002
589 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT	(EXT4_GET_BLOCKS_UNWRIT_EXT|\
590 						 EXT4_GET_BLOCKS_CREATE)
591 	/* Caller is from the delayed allocation writeout path
592 	 * finally doing the actual allocation of delayed blocks */
593 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE	0x0004
594 	/* caller is from the direct IO path, request to creation of an
595 	unwritten extents if not allocated, split the unwritten
596 	extent if blocks has been preallocated already*/
597 #define EXT4_GET_BLOCKS_PRE_IO			0x0008
598 #define EXT4_GET_BLOCKS_CONVERT			0x0010
599 #define EXT4_GET_BLOCKS_IO_CREATE_EXT		(EXT4_GET_BLOCKS_PRE_IO|\
600 					 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
601 	/* Convert extent to initialized after IO complete */
602 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT		(EXT4_GET_BLOCKS_CONVERT|\
603 					 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
604 	/* Eventual metadata allocation (due to growing extent tree)
605 	 * should not fail, so try to use reserved blocks for that.*/
606 #define EXT4_GET_BLOCKS_METADATA_NOFAIL		0x0020
607 	/* Don't normalize allocation size (used for fallocate) */
608 #define EXT4_GET_BLOCKS_NO_NORMALIZE		0x0040
609 	/* Request will not result in inode size update (user for fallocate) */
610 #define EXT4_GET_BLOCKS_KEEP_SIZE		0x0080
611 	/* Convert written extents to unwritten */
612 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN	0x0100
613 	/* Write zeros to newly created written extents */
614 #define EXT4_GET_BLOCKS_ZERO			0x0200
615 #define EXT4_GET_BLOCKS_CREATE_ZERO		(EXT4_GET_BLOCKS_CREATE |\
616 					EXT4_GET_BLOCKS_ZERO)
617 	/* Caller will submit data before dropping transaction handle. This
618 	 * allows jbd2 to avoid submitting data before commit. */
619 #define EXT4_GET_BLOCKS_IO_SUBMIT		0x0400
620 
621 /*
622  * The bit position of these flags must not overlap with any of the
623  * EXT4_GET_BLOCKS_*.  They are used by ext4_find_extent(),
624  * read_extent_tree_block(), ext4_split_extent_at(),
625  * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
626  * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
627  * caching the extents when reading from the extent tree while a
628  * truncate or punch hole operation is in progress.
629  */
630 #define EXT4_EX_NOCACHE				0x40000000
631 #define EXT4_EX_FORCE_CACHE			0x20000000
632 
633 /*
634  * Flags used by ext4_free_blocks
635  */
636 #define EXT4_FREE_BLOCKS_METADATA		0x0001
637 #define EXT4_FREE_BLOCKS_FORGET			0x0002
638 #define EXT4_FREE_BLOCKS_VALIDATED		0x0004
639 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE		0x0008
640 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER	0x0010
641 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER	0x0020
642 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER      0x0040
643 
644 /*
645  * ioctl commands
646  */
647 #define	EXT4_IOC_GETFLAGS		FS_IOC_GETFLAGS
648 #define	EXT4_IOC_SETFLAGS		FS_IOC_SETFLAGS
649 #define	EXT4_IOC_GETVERSION		_IOR('f', 3, long)
650 #define	EXT4_IOC_SETVERSION		_IOW('f', 4, long)
651 #define	EXT4_IOC_GETVERSION_OLD		FS_IOC_GETVERSION
652 #define	EXT4_IOC_SETVERSION_OLD		FS_IOC_SETVERSION
653 #define EXT4_IOC_GETRSVSZ		_IOR('f', 5, long)
654 #define EXT4_IOC_SETRSVSZ		_IOW('f', 6, long)
655 #define EXT4_IOC_GROUP_EXTEND		_IOW('f', 7, unsigned long)
656 #define EXT4_IOC_GROUP_ADD		_IOW('f', 8, struct ext4_new_group_input)
657 #define EXT4_IOC_MIGRATE		_IO('f', 9)
658  /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
659  /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
660 #define EXT4_IOC_ALLOC_DA_BLKS		_IO('f', 12)
661 #define EXT4_IOC_MOVE_EXT		_IOWR('f', 15, struct move_extent)
662 #define EXT4_IOC_RESIZE_FS		_IOW('f', 16, __u64)
663 #define EXT4_IOC_SWAP_BOOT		_IO('f', 17)
664 #define EXT4_IOC_PRECACHE_EXTENTS	_IO('f', 18)
665 #define EXT4_IOC_SET_ENCRYPTION_POLICY	FS_IOC_SET_ENCRYPTION_POLICY
666 #define EXT4_IOC_GET_ENCRYPTION_PWSALT	FS_IOC_GET_ENCRYPTION_PWSALT
667 #define EXT4_IOC_GET_ENCRYPTION_POLICY	FS_IOC_GET_ENCRYPTION_POLICY
668 /* ioctl codes 19--39 are reserved for fscrypt */
669 #define EXT4_IOC_CLEAR_ES_CACHE		_IO('f', 40)
670 #define EXT4_IOC_GETSTATE		_IOW('f', 41, __u32)
671 #define EXT4_IOC_GET_ES_CACHE		_IOWR('f', 42, struct fiemap)
672 
673 #define EXT4_IOC_FSGETXATTR		FS_IOC_FSGETXATTR
674 #define EXT4_IOC_FSSETXATTR		FS_IOC_FSSETXATTR
675 
676 #define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32)
677 
678 /*
679  * Flags for going down operation
680  */
681 #define EXT4_GOING_FLAGS_DEFAULT		0x0	/* going down */
682 #define EXT4_GOING_FLAGS_LOGFLUSH		0x1	/* flush log but not data */
683 #define EXT4_GOING_FLAGS_NOLOGFLUSH		0x2	/* don't flush log nor data */
684 
685 /*
686  * Flags returned by EXT4_IOC_GETSTATE
687  *
688  * We only expose to userspace a subset of the state flags in
689  * i_state_flags
690  */
691 #define EXT4_STATE_FLAG_EXT_PRECACHED	0x00000001
692 #define EXT4_STATE_FLAG_NEW		0x00000002
693 #define EXT4_STATE_FLAG_NEWENTRY	0x00000004
694 #define EXT4_STATE_FLAG_DA_ALLOC_CLOSE	0x00000008
695 
696 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
697 /*
698  * ioctl commands in 32 bit emulation
699  */
700 #define EXT4_IOC32_GETFLAGS		FS_IOC32_GETFLAGS
701 #define EXT4_IOC32_SETFLAGS		FS_IOC32_SETFLAGS
702 #define EXT4_IOC32_GETVERSION		_IOR('f', 3, int)
703 #define EXT4_IOC32_SETVERSION		_IOW('f', 4, int)
704 #define EXT4_IOC32_GETRSVSZ		_IOR('f', 5, int)
705 #define EXT4_IOC32_SETRSVSZ		_IOW('f', 6, int)
706 #define EXT4_IOC32_GROUP_EXTEND		_IOW('f', 7, unsigned int)
707 #define EXT4_IOC32_GROUP_ADD		_IOW('f', 8, struct compat_ext4_new_group_input)
708 #define EXT4_IOC32_GETVERSION_OLD	FS_IOC32_GETVERSION
709 #define EXT4_IOC32_SETVERSION_OLD	FS_IOC32_SETVERSION
710 #endif
711 
712 /*
713  * Returned by EXT4_IOC_GET_ES_CACHE as an additional possible flag.
714  * It indicates that the entry in extent status cache is for a hole.
715  */
716 #define EXT4_FIEMAP_EXTENT_HOLE		0x08000000
717 
718 /* Max physical block we can address w/o extents */
719 #define EXT4_MAX_BLOCK_FILE_PHYS	0xFFFFFFFF
720 
721 /* Max logical block we can support */
722 #define EXT4_MAX_LOGICAL_BLOCK		0xFFFFFFFE
723 
724 /*
725  * Structure of an inode on the disk
726  */
727 struct ext4_inode {
728 	__le16	i_mode;		/* File mode */
729 	__le16	i_uid;		/* Low 16 bits of Owner Uid */
730 	__le32	i_size_lo;	/* Size in bytes */
731 	__le32	i_atime;	/* Access time */
732 	__le32	i_ctime;	/* Inode Change time */
733 	__le32	i_mtime;	/* Modification time */
734 	__le32	i_dtime;	/* Deletion Time */
735 	__le16	i_gid;		/* Low 16 bits of Group Id */
736 	__le16	i_links_count;	/* Links count */
737 	__le32	i_blocks_lo;	/* Blocks count */
738 	__le32	i_flags;	/* File flags */
739 	union {
740 		struct {
741 			__le32  l_i_version;
742 		} linux1;
743 		struct {
744 			__u32  h_i_translator;
745 		} hurd1;
746 		struct {
747 			__u32  m_i_reserved1;
748 		} masix1;
749 	} osd1;				/* OS dependent 1 */
750 	__le32	i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
751 	__le32	i_generation;	/* File version (for NFS) */
752 	__le32	i_file_acl_lo;	/* File ACL */
753 	__le32	i_size_high;
754 	__le32	i_obso_faddr;	/* Obsoleted fragment address */
755 	union {
756 		struct {
757 			__le16	l_i_blocks_high; /* were l_i_reserved1 */
758 			__le16	l_i_file_acl_high;
759 			__le16	l_i_uid_high;	/* these 2 fields */
760 			__le16	l_i_gid_high;	/* were reserved2[0] */
761 			__le16	l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
762 			__le16	l_i_reserved;
763 		} linux2;
764 		struct {
765 			__le16	h_i_reserved1;	/* Obsoleted fragment number/size which are removed in ext4 */
766 			__u16	h_i_mode_high;
767 			__u16	h_i_uid_high;
768 			__u16	h_i_gid_high;
769 			__u32	h_i_author;
770 		} hurd2;
771 		struct {
772 			__le16	h_i_reserved1;	/* Obsoleted fragment number/size which are removed in ext4 */
773 			__le16	m_i_file_acl_high;
774 			__u32	m_i_reserved2[2];
775 		} masix2;
776 	} osd2;				/* OS dependent 2 */
777 	__le16	i_extra_isize;
778 	__le16	i_checksum_hi;	/* crc32c(uuid+inum+inode) BE */
779 	__le32  i_ctime_extra;  /* extra Change time      (nsec << 2 | epoch) */
780 	__le32  i_mtime_extra;  /* extra Modification time(nsec << 2 | epoch) */
781 	__le32  i_atime_extra;  /* extra Access time      (nsec << 2 | epoch) */
782 	__le32  i_crtime;       /* File Creation time */
783 	__le32  i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
784 	__le32  i_version_hi;	/* high 32 bits for 64-bit version */
785 	__le32	i_projid;	/* Project ID */
786 };
787 
788 struct move_extent {
789 	__u32 reserved;		/* should be zero */
790 	__u32 donor_fd;		/* donor file descriptor */
791 	__u64 orig_start;	/* logical start offset in block for orig */
792 	__u64 donor_start;	/* logical start offset in block for donor */
793 	__u64 len;		/* block length to be moved */
794 	__u64 moved_len;	/* moved block length */
795 };
796 
797 #define EXT4_EPOCH_BITS 2
798 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
799 #define EXT4_NSEC_MASK  (~0UL << EXT4_EPOCH_BITS)
800 
801 /*
802  * Extended fields will fit into an inode if the filesystem was formatted
803  * with large inodes (-I 256 or larger) and there are not currently any EAs
804  * consuming all of the available space. For new inodes we always reserve
805  * enough space for the kernel's known extended fields, but for inodes
806  * created with an old kernel this might not have been the case. None of
807  * the extended inode fields is critical for correct filesystem operation.
808  * This macro checks if a certain field fits in the inode. Note that
809  * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
810  */
811 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field)	\
812 	((offsetof(typeof(*ext4_inode), field) +	\
813 	  sizeof((ext4_inode)->field))			\
814 	<= (EXT4_GOOD_OLD_INODE_SIZE +			\
815 	    (einode)->i_extra_isize))			\
816 
817 /*
818  * We use an encoding that preserves the times for extra epoch "00":
819  *
820  * extra  msb of                         adjust for signed
821  * epoch  32-bit                         32-bit tv_sec to
822  * bits   time    decoded 64-bit tv_sec  64-bit tv_sec      valid time range
823  * 0 0    1    -0x80000000..-0x00000001  0x000000000 1901-12-13..1969-12-31
824  * 0 0    0    0x000000000..0x07fffffff  0x000000000 1970-01-01..2038-01-19
825  * 0 1    1    0x080000000..0x0ffffffff  0x100000000 2038-01-19..2106-02-07
826  * 0 1    0    0x100000000..0x17fffffff  0x100000000 2106-02-07..2174-02-25
827  * 1 0    1    0x180000000..0x1ffffffff  0x200000000 2174-02-25..2242-03-16
828  * 1 0    0    0x200000000..0x27fffffff  0x200000000 2242-03-16..2310-04-04
829  * 1 1    1    0x280000000..0x2ffffffff  0x300000000 2310-04-04..2378-04-22
830  * 1 1    0    0x300000000..0x37fffffff  0x300000000 2378-04-22..2446-05-10
831  *
832  * Note that previous versions of the kernel on 64-bit systems would
833  * incorrectly use extra epoch bits 1,1 for dates between 1901 and
834  * 1970.  e2fsck will correct this, assuming that it is run on the
835  * affected filesystem before 2242.
836  */
837 
ext4_encode_extra_time(struct timespec64 * time)838 static inline __le32 ext4_encode_extra_time(struct timespec64 *time)
839 {
840 	u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK;
841 	return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
842 }
843 
ext4_decode_extra_time(struct timespec64 * time,__le32 extra)844 static inline void ext4_decode_extra_time(struct timespec64 *time,
845 					  __le32 extra)
846 {
847 	if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK)))
848 		time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
849 	time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
850 }
851 
852 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode)				\
853 do {										\
854 	if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     {\
855 		(raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec);	\
856 		(raw_inode)->xtime ## _extra =					\
857 				ext4_encode_extra_time(&(inode)->xtime);	\
858 		}								\
859 	else	\
860 		(raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (inode)->xtime.tv_sec, S32_MIN, S32_MAX));	\
861 } while (0)
862 
863 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode)			       \
864 do {									       \
865 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))		       \
866 		(raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec);      \
867 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))	       \
868 		(raw_inode)->xtime ## _extra =				       \
869 				ext4_encode_extra_time(&(einode)->xtime);      \
870 } while (0)
871 
872 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode)				\
873 do {										\
874 	(inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime);	\
875 	if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {	\
876 		ext4_decode_extra_time(&(inode)->xtime,				\
877 				       raw_inode->xtime ## _extra);		\
878 		}								\
879 	else									\
880 		(inode)->xtime.tv_nsec = 0;					\
881 } while (0)
882 
883 
884 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode)			       \
885 do {									       \
886 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))		       \
887 		(einode)->xtime.tv_sec = 				       \
888 			(signed)le32_to_cpu((raw_inode)->xtime);	       \
889 	else								       \
890 		(einode)->xtime.tv_sec = 0;				       \
891 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))	       \
892 		ext4_decode_extra_time(&(einode)->xtime,		       \
893 				       raw_inode->xtime ## _extra);	       \
894 	else								       \
895 		(einode)->xtime.tv_nsec = 0;				       \
896 } while (0)
897 
898 #define i_disk_version osd1.linux1.l_i_version
899 
900 #if defined(__KERNEL__) || defined(__linux__)
901 #define i_reserved1	osd1.linux1.l_i_reserved1
902 #define i_file_acl_high	osd2.linux2.l_i_file_acl_high
903 #define i_blocks_high	osd2.linux2.l_i_blocks_high
904 #define i_uid_low	i_uid
905 #define i_gid_low	i_gid
906 #define i_uid_high	osd2.linux2.l_i_uid_high
907 #define i_gid_high	osd2.linux2.l_i_gid_high
908 #define i_checksum_lo	osd2.linux2.l_i_checksum_lo
909 
910 #elif defined(__GNU__)
911 
912 #define i_translator	osd1.hurd1.h_i_translator
913 #define i_uid_high	osd2.hurd2.h_i_uid_high
914 #define i_gid_high	osd2.hurd2.h_i_gid_high
915 #define i_author	osd2.hurd2.h_i_author
916 
917 #elif defined(__masix__)
918 
919 #define i_reserved1	osd1.masix1.m_i_reserved1
920 #define i_file_acl_high	osd2.masix2.m_i_file_acl_high
921 #define i_reserved2	osd2.masix2.m_i_reserved2
922 
923 #endif /* defined(__KERNEL__) || defined(__linux__) */
924 
925 #include "extents_status.h"
926 
927 /*
928  * Lock subclasses for i_data_sem in the ext4_inode_info structure.
929  *
930  * These are needed to avoid lockdep false positives when we need to
931  * allocate blocks to the quota inode during ext4_map_blocks(), while
932  * holding i_data_sem for a normal (non-quota) inode.  Since we don't
933  * do quota tracking for the quota inode, this avoids deadlock (as
934  * well as infinite recursion, since it isn't turtles all the way
935  * down...)
936  *
937  *  I_DATA_SEM_NORMAL - Used for most inodes
938  *  I_DATA_SEM_OTHER  - Used by move_inode.c for the second normal inode
939  *			  where the second inode has larger inode number
940  *			  than the first
941  *  I_DATA_SEM_QUOTA  - Used for quota inodes only
942  *  I_DATA_SEM_EA     - Used for ea_inodes only
943  */
944 enum {
945 	I_DATA_SEM_NORMAL = 0,
946 	I_DATA_SEM_OTHER,
947 	I_DATA_SEM_QUOTA,
948 	I_DATA_SEM_EA
949 };
950 
951 
952 /*
953  * fourth extended file system inode data in memory
954  */
955 struct ext4_inode_info {
956 	__le32	i_data[15];	/* unconverted */
957 	__u32	i_dtime;
958 	ext4_fsblk_t	i_file_acl;
959 
960 	/*
961 	 * i_block_group is the number of the block group which contains
962 	 * this file's inode.  Constant across the lifetime of the inode,
963 	 * it is used for making block allocation decisions - we try to
964 	 * place a file's data blocks near its inode block, and new inodes
965 	 * near to their parent directory's inode.
966 	 */
967 	ext4_group_t	i_block_group;
968 	ext4_lblk_t	i_dir_start_lookup;
969 #if (BITS_PER_LONG < 64)
970 	unsigned long	i_state_flags;		/* Dynamic state flags */
971 #endif
972 	unsigned long	i_flags;
973 
974 	/*
975 	 * Extended attributes can be read independently of the main file
976 	 * data. Taking i_mutex even when reading would cause contention
977 	 * between readers of EAs and writers of regular file data, so
978 	 * instead we synchronize on xattr_sem when reading or changing
979 	 * EAs.
980 	 */
981 	struct rw_semaphore xattr_sem;
982 
983 	struct list_head i_orphan;	/* unlinked but open inodes */
984 
985 	/*
986 	 * i_disksize keeps track of what the inode size is ON DISK, not
987 	 * in memory.  During truncate, i_size is set to the new size by
988 	 * the VFS prior to calling ext4_truncate(), but the filesystem won't
989 	 * set i_disksize to 0 until the truncate is actually under way.
990 	 *
991 	 * The intent is that i_disksize always represents the blocks which
992 	 * are used by this file.  This allows recovery to restart truncate
993 	 * on orphans if we crash during truncate.  We actually write i_disksize
994 	 * into the on-disk inode when writing inodes out, instead of i_size.
995 	 *
996 	 * The only time when i_disksize and i_size may be different is when
997 	 * a truncate is in progress.  The only things which change i_disksize
998 	 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
999 	 */
1000 	loff_t	i_disksize;
1001 
1002 	/*
1003 	 * i_data_sem is for serialising ext4_truncate() against
1004 	 * ext4_getblock().  In the 2.4 ext2 design, great chunks of inode's
1005 	 * data tree are chopped off during truncate. We can't do that in
1006 	 * ext4 because whenever we perform intermediate commits during
1007 	 * truncate, the inode and all the metadata blocks *must* be in a
1008 	 * consistent state which allows truncation of the orphans to restart
1009 	 * during recovery.  Hence we must fix the get_block-vs-truncate race
1010 	 * by other means, so we have i_data_sem.
1011 	 */
1012 	struct rw_semaphore i_data_sem;
1013 	/*
1014 	 * i_mmap_sem is for serializing page faults with truncate / punch hole
1015 	 * operations. We have to make sure that new page cannot be faulted in
1016 	 * a section of the inode that is being punched. We cannot easily use
1017 	 * i_data_sem for this since we need protection for the whole punch
1018 	 * operation and i_data_sem ranks below transaction start so we have
1019 	 * to occasionally drop it.
1020 	 */
1021 	struct rw_semaphore i_mmap_sem;
1022 	struct inode vfs_inode;
1023 	struct jbd2_inode *jinode;
1024 
1025 	spinlock_t i_raw_lock;	/* protects updates to the raw inode */
1026 
1027 	/*
1028 	 * File creation time. Its function is same as that of
1029 	 * struct timespec64 i_{a,c,m}time in the generic inode.
1030 	 */
1031 	struct timespec64 i_crtime;
1032 
1033 	/* mballoc */
1034 	struct list_head i_prealloc_list;
1035 	spinlock_t i_prealloc_lock;
1036 
1037 	/* extents status tree */
1038 	struct ext4_es_tree i_es_tree;
1039 	rwlock_t i_es_lock;
1040 	struct list_head i_es_list;
1041 	unsigned int i_es_all_nr;	/* protected by i_es_lock */
1042 	unsigned int i_es_shk_nr;	/* protected by i_es_lock */
1043 	ext4_lblk_t i_es_shrink_lblk;	/* Offset where we start searching for
1044 					   extents to shrink. Protected by
1045 					   i_es_lock  */
1046 
1047 	/* ialloc */
1048 	ext4_group_t	i_last_alloc_group;
1049 
1050 	/* allocation reservation info for delalloc */
1051 	/* In case of bigalloc, this refer to clusters rather than blocks */
1052 	unsigned int i_reserved_data_blocks;
1053 	ext4_lblk_t i_da_metadata_calc_last_lblock;
1054 	int i_da_metadata_calc_len;
1055 
1056 	/* pending cluster reservations for bigalloc file systems */
1057 	struct ext4_pending_tree i_pending_tree;
1058 
1059 	/* on-disk additional length */
1060 	__u16 i_extra_isize;
1061 
1062 	/* Indicate the inline data space. */
1063 	u16 i_inline_off;
1064 	u16 i_inline_size;
1065 
1066 #ifdef CONFIG_QUOTA
1067 	/* quota space reservation, managed internally by quota code */
1068 	qsize_t i_reserved_quota;
1069 #endif
1070 
1071 	/* Lock protecting lists below */
1072 	spinlock_t i_completed_io_lock;
1073 	/*
1074 	 * Completed IOs that need unwritten extents handling and have
1075 	 * transaction reserved
1076 	 */
1077 	struct list_head i_rsv_conversion_list;
1078 	struct work_struct i_rsv_conversion_work;
1079 	atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1080 
1081 	spinlock_t i_block_reservation_lock;
1082 
1083 	/*
1084 	 * Transactions that contain inode's metadata needed to complete
1085 	 * fsync and fdatasync, respectively.
1086 	 */
1087 	tid_t i_sync_tid;
1088 	tid_t i_datasync_tid;
1089 
1090 #ifdef CONFIG_QUOTA
1091 	struct dquot *i_dquot[MAXQUOTAS];
1092 #endif
1093 
1094 	/* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1095 	__u32 i_csum_seed;
1096 
1097 	kprojid_t i_projid;
1098 };
1099 
1100 /*
1101  * File system states
1102  */
1103 #define	EXT4_VALID_FS			0x0001	/* Unmounted cleanly */
1104 #define	EXT4_ERROR_FS			0x0002	/* Errors detected */
1105 #define	EXT4_ORPHAN_FS			0x0004	/* Orphans being recovered */
1106 
1107 /*
1108  * Misc. filesystem flags
1109  */
1110 #define EXT2_FLAGS_SIGNED_HASH		0x0001  /* Signed dirhash in use */
1111 #define EXT2_FLAGS_UNSIGNED_HASH	0x0002  /* Unsigned dirhash in use */
1112 #define EXT2_FLAGS_TEST_FILESYS		0x0004	/* to test development code */
1113 
1114 /*
1115  * Mount flags set via mount options or defaults
1116  */
1117 #define EXT4_MOUNT_NO_MBCACHE		0x00001 /* Do not use mbcache */
1118 #define EXT4_MOUNT_GRPID		0x00004	/* Create files with directory's group */
1119 #define EXT4_MOUNT_DEBUG		0x00008	/* Some debugging messages */
1120 #define EXT4_MOUNT_ERRORS_CONT		0x00010	/* Continue on errors */
1121 #define EXT4_MOUNT_ERRORS_RO		0x00020	/* Remount fs ro on errors */
1122 #define EXT4_MOUNT_ERRORS_PANIC		0x00040	/* Panic on errors */
1123 #define EXT4_MOUNT_ERRORS_MASK		0x00070
1124 #define EXT4_MOUNT_MINIX_DF		0x00080	/* Mimics the Minix statfs */
1125 #define EXT4_MOUNT_NOLOAD		0x00100	/* Don't use existing journal*/
1126 #ifdef CONFIG_FS_DAX
1127 #define EXT4_MOUNT_DAX			0x00200	/* Direct Access */
1128 #else
1129 #define EXT4_MOUNT_DAX			0
1130 #endif
1131 #define EXT4_MOUNT_DATA_FLAGS		0x00C00	/* Mode for data writes: */
1132 #define EXT4_MOUNT_JOURNAL_DATA		0x00400	/* Write data to journal */
1133 #define EXT4_MOUNT_ORDERED_DATA		0x00800	/* Flush data before commit */
1134 #define EXT4_MOUNT_WRITEBACK_DATA	0x00C00	/* No data ordering */
1135 #define EXT4_MOUNT_UPDATE_JOURNAL	0x01000	/* Update the journal format */
1136 #define EXT4_MOUNT_NO_UID32		0x02000  /* Disable 32-bit UIDs */
1137 #define EXT4_MOUNT_XATTR_USER		0x04000	/* Extended user attributes */
1138 #define EXT4_MOUNT_POSIX_ACL		0x08000	/* POSIX Access Control Lists */
1139 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC	0x10000	/* No auto delalloc mapping */
1140 #define EXT4_MOUNT_BARRIER		0x20000 /* Use block barriers */
1141 #define EXT4_MOUNT_QUOTA		0x40000 /* Some quota option set */
1142 #define EXT4_MOUNT_USRQUOTA		0x80000 /* "old" user quota,
1143 						 * enable enforcement for hidden
1144 						 * quota files */
1145 #define EXT4_MOUNT_GRPQUOTA		0x100000 /* "old" group quota, enable
1146 						  * enforcement for hidden quota
1147 						  * files */
1148 #define EXT4_MOUNT_PRJQUOTA		0x200000 /* Enable project quota
1149 						  * enforcement */
1150 #define EXT4_MOUNT_DIOREAD_NOLOCK	0x400000 /* Enable support for dio read nolocking */
1151 #define EXT4_MOUNT_JOURNAL_CHECKSUM	0x800000 /* Journal checksums */
1152 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT	0x1000000 /* Journal Async Commit */
1153 #define EXT4_MOUNT_WARN_ON_ERROR	0x2000000 /* Trigger WARN_ON on error */
1154 #define EXT4_MOUNT_INLINECRYPT		0x4000000 /* Inline encryption support */
1155 #define EXT4_MOUNT_DELALLOC		0x8000000 /* Delalloc support */
1156 #define EXT4_MOUNT_DATA_ERR_ABORT	0x10000000 /* Abort on file data write */
1157 #define EXT4_MOUNT_BLOCK_VALIDITY	0x20000000 /* Block validity checking */
1158 #define EXT4_MOUNT_DISCARD		0x40000000 /* Issue DISCARD requests */
1159 #define EXT4_MOUNT_INIT_INODE_TABLE	0x80000000 /* Initialize uninitialized itables */
1160 
1161 /*
1162  * Mount flags set either automatically (could not be set by mount option)
1163  * based on per file system feature or property or in special cases such as
1164  * distinguishing between explicit mount option definition and default.
1165  */
1166 #define EXT4_MOUNT2_EXPLICIT_DELALLOC	0x00000001 /* User explicitly
1167 						      specified delalloc */
1168 #define EXT4_MOUNT2_STD_GROUP_SIZE	0x00000002 /* We have standard group
1169 						      size of blocksize * 8
1170 						      blocks */
1171 #define EXT4_MOUNT2_HURD_COMPAT		0x00000004 /* Support HURD-castrated
1172 						      file systems */
1173 
1174 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM	0x00000008 /* User explicitly
1175 						specified journal checksum */
1176 
1177 #define clear_opt(sb, opt)		EXT4_SB(sb)->s_mount_opt &= \
1178 						~EXT4_MOUNT_##opt
1179 #define set_opt(sb, opt)		EXT4_SB(sb)->s_mount_opt |= \
1180 						EXT4_MOUNT_##opt
1181 #define test_opt(sb, opt)		(EXT4_SB(sb)->s_mount_opt & \
1182 					 EXT4_MOUNT_##opt)
1183 
1184 #define clear_opt2(sb, opt)		EXT4_SB(sb)->s_mount_opt2 &= \
1185 						~EXT4_MOUNT2_##opt
1186 #define set_opt2(sb, opt)		EXT4_SB(sb)->s_mount_opt2 |= \
1187 						EXT4_MOUNT2_##opt
1188 #define test_opt2(sb, opt)		(EXT4_SB(sb)->s_mount_opt2 & \
1189 					 EXT4_MOUNT2_##opt)
1190 
1191 #define ext4_test_and_set_bit		__test_and_set_bit_le
1192 #define ext4_set_bit			__set_bit_le
1193 #define ext4_set_bit_atomic		ext2_set_bit_atomic
1194 #define ext4_test_and_clear_bit		__test_and_clear_bit_le
1195 #define ext4_clear_bit			__clear_bit_le
1196 #define ext4_clear_bit_atomic		ext2_clear_bit_atomic
1197 #define ext4_test_bit			test_bit_le
1198 #define ext4_find_next_zero_bit		find_next_zero_bit_le
1199 #define ext4_find_next_bit		find_next_bit_le
1200 
1201 extern void ext4_set_bits(void *bm, int cur, int len);
1202 
1203 /*
1204  * Maximal mount counts between two filesystem checks
1205  */
1206 #define EXT4_DFL_MAX_MNT_COUNT		20	/* Allow 20 mounts */
1207 #define EXT4_DFL_CHECKINTERVAL		0	/* Don't use interval check */
1208 
1209 /*
1210  * Behaviour when detecting errors
1211  */
1212 #define EXT4_ERRORS_CONTINUE		1	/* Continue execution */
1213 #define EXT4_ERRORS_RO			2	/* Remount fs read-only */
1214 #define EXT4_ERRORS_PANIC		3	/* Panic */
1215 #define EXT4_ERRORS_DEFAULT		EXT4_ERRORS_CONTINUE
1216 
1217 /* Metadata checksum algorithm codes */
1218 #define EXT4_CRC32C_CHKSUM		1
1219 
1220 /*
1221  * Structure of the super block
1222  */
1223 struct ext4_super_block {
1224 /*00*/	__le32	s_inodes_count;		/* Inodes count */
1225 	__le32	s_blocks_count_lo;	/* Blocks count */
1226 	__le32	s_r_blocks_count_lo;	/* Reserved blocks count */
1227 	__le32	s_free_blocks_count_lo;	/* Free blocks count */
1228 /*10*/	__le32	s_free_inodes_count;	/* Free inodes count */
1229 	__le32	s_first_data_block;	/* First Data Block */
1230 	__le32	s_log_block_size;	/* Block size */
1231 	__le32	s_log_cluster_size;	/* Allocation cluster size */
1232 /*20*/	__le32	s_blocks_per_group;	/* # Blocks per group */
1233 	__le32	s_clusters_per_group;	/* # Clusters per group */
1234 	__le32	s_inodes_per_group;	/* # Inodes per group */
1235 	__le32	s_mtime;		/* Mount time */
1236 /*30*/	__le32	s_wtime;		/* Write time */
1237 	__le16	s_mnt_count;		/* Mount count */
1238 	__le16	s_max_mnt_count;	/* Maximal mount count */
1239 	__le16	s_magic;		/* Magic signature */
1240 	__le16	s_state;		/* File system state */
1241 	__le16	s_errors;		/* Behaviour when detecting errors */
1242 	__le16	s_minor_rev_level;	/* minor revision level */
1243 /*40*/	__le32	s_lastcheck;		/* time of last check */
1244 	__le32	s_checkinterval;	/* max. time between checks */
1245 	__le32	s_creator_os;		/* OS */
1246 	__le32	s_rev_level;		/* Revision level */
1247 /*50*/	__le16	s_def_resuid;		/* Default uid for reserved blocks */
1248 	__le16	s_def_resgid;		/* Default gid for reserved blocks */
1249 	/*
1250 	 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1251 	 *
1252 	 * Note: the difference between the compatible feature set and
1253 	 * the incompatible feature set is that if there is a bit set
1254 	 * in the incompatible feature set that the kernel doesn't
1255 	 * know about, it should refuse to mount the filesystem.
1256 	 *
1257 	 * e2fsck's requirements are more strict; if it doesn't know
1258 	 * about a feature in either the compatible or incompatible
1259 	 * feature set, it must abort and not try to meddle with
1260 	 * things it doesn't understand...
1261 	 */
1262 	__le32	s_first_ino;		/* First non-reserved inode */
1263 	__le16  s_inode_size;		/* size of inode structure */
1264 	__le16	s_block_group_nr;	/* block group # of this superblock */
1265 	__le32	s_feature_compat;	/* compatible feature set */
1266 /*60*/	__le32	s_feature_incompat;	/* incompatible feature set */
1267 	__le32	s_feature_ro_compat;	/* readonly-compatible feature set */
1268 /*68*/	__u8	s_uuid[16];		/* 128-bit uuid for volume */
1269 /*78*/	char	s_volume_name[16];	/* volume name */
1270 /*88*/	char	s_last_mounted[64] __nonstring;	/* directory where last mounted */
1271 /*C8*/	__le32	s_algorithm_usage_bitmap; /* For compression */
1272 	/*
1273 	 * Performance hints.  Directory preallocation should only
1274 	 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1275 	 */
1276 	__u8	s_prealloc_blocks;	/* Nr of blocks to try to preallocate*/
1277 	__u8	s_prealloc_dir_blocks;	/* Nr to preallocate for dirs */
1278 	__le16	s_reserved_gdt_blocks;	/* Per group desc for online growth */
1279 	/*
1280 	 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1281 	 */
1282 /*D0*/	__u8	s_journal_uuid[16];	/* uuid of journal superblock */
1283 /*E0*/	__le32	s_journal_inum;		/* inode number of journal file */
1284 	__le32	s_journal_dev;		/* device number of journal file */
1285 	__le32	s_last_orphan;		/* start of list of inodes to delete */
1286 	__le32	s_hash_seed[4];		/* HTREE hash seed */
1287 	__u8	s_def_hash_version;	/* Default hash version to use */
1288 	__u8	s_jnl_backup_type;
1289 	__le16  s_desc_size;		/* size of group descriptor */
1290 /*100*/	__le32	s_default_mount_opts;
1291 	__le32	s_first_meta_bg;	/* First metablock block group */
1292 	__le32	s_mkfs_time;		/* When the filesystem was created */
1293 	__le32	s_jnl_blocks[17];	/* Backup of the journal inode */
1294 	/* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1295 /*150*/	__le32	s_blocks_count_hi;	/* Blocks count */
1296 	__le32	s_r_blocks_count_hi;	/* Reserved blocks count */
1297 	__le32	s_free_blocks_count_hi;	/* Free blocks count */
1298 	__le16	s_min_extra_isize;	/* All inodes have at least # bytes */
1299 	__le16	s_want_extra_isize; 	/* New inodes should reserve # bytes */
1300 	__le32	s_flags;		/* Miscellaneous flags */
1301 	__le16  s_raid_stride;		/* RAID stride */
1302 	__le16  s_mmp_update_interval;  /* # seconds to wait in MMP checking */
1303 	__le64  s_mmp_block;            /* Block for multi-mount protection */
1304 	__le32  s_raid_stripe_width;    /* blocks on all data disks (N*stride)*/
1305 	__u8	s_log_groups_per_flex;  /* FLEX_BG group size */
1306 	__u8	s_checksum_type;	/* metadata checksum algorithm used */
1307 	__u8	s_encryption_level;	/* versioning level for encryption */
1308 	__u8	s_reserved_pad;		/* Padding to next 32bits */
1309 	__le64	s_kbytes_written;	/* nr of lifetime kilobytes written */
1310 	__le32	s_snapshot_inum;	/* Inode number of active snapshot */
1311 	__le32	s_snapshot_id;		/* sequential ID of active snapshot */
1312 	__le64	s_snapshot_r_blocks_count; /* reserved blocks for active
1313 					      snapshot's future use */
1314 	__le32	s_snapshot_list;	/* inode number of the head of the
1315 					   on-disk snapshot list */
1316 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1317 	__le32	s_error_count;		/* number of fs errors */
1318 	__le32	s_first_error_time;	/* first time an error happened */
1319 	__le32	s_first_error_ino;	/* inode involved in first error */
1320 	__le64	s_first_error_block;	/* block involved of first error */
1321 	__u8	s_first_error_func[32] __nonstring;	/* function where the error happened */
1322 	__le32	s_first_error_line;	/* line number where error happened */
1323 	__le32	s_last_error_time;	/* most recent time of an error */
1324 	__le32	s_last_error_ino;	/* inode involved in last error */
1325 	__le32	s_last_error_line;	/* line number where error happened */
1326 	__le64	s_last_error_block;	/* block involved of last error */
1327 	__u8	s_last_error_func[32] __nonstring;	/* function where the error happened */
1328 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1329 	__u8	s_mount_opts[64];
1330 	__le32	s_usr_quota_inum;	/* inode for tracking user quota */
1331 	__le32	s_grp_quota_inum;	/* inode for tracking group quota */
1332 	__le32	s_overhead_clusters;	/* overhead blocks/clusters in fs */
1333 	__le32	s_backup_bgs[2];	/* groups with sparse_super2 SBs */
1334 	__u8	s_encrypt_algos[4];	/* Encryption algorithms in use  */
1335 	__u8	s_encrypt_pw_salt[16];	/* Salt used for string2key algorithm */
1336 	__le32	s_lpf_ino;		/* Location of the lost+found inode */
1337 	__le32	s_prj_quota_inum;	/* inode for tracking project quota */
1338 	__le32	s_checksum_seed;	/* crc32c(uuid) if csum_seed set */
1339 	__u8	s_wtime_hi;
1340 	__u8	s_mtime_hi;
1341 	__u8	s_mkfs_time_hi;
1342 	__u8	s_lastcheck_hi;
1343 	__u8	s_first_error_time_hi;
1344 	__u8	s_last_error_time_hi;
1345 	__u8	s_pad[2];
1346 	__le16  s_encoding;		/* Filename charset encoding */
1347 	__le16  s_encoding_flags;	/* Filename charset encoding flags */
1348 	__le32	s_reserved[95];		/* Padding to the end of the block */
1349 	__le32	s_checksum;		/* crc32c(superblock) */
1350 };
1351 
1352 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1353 
1354 #ifdef __KERNEL__
1355 
1356 /*
1357  * run-time mount flags
1358  */
1359 #define EXT4_MF_MNTDIR_SAMPLED		0x0001
1360 #define EXT4_MF_FS_ABORTED		0x0002	/* Fatal error detected */
1361 
1362 #ifdef CONFIG_FS_ENCRYPTION
1363 #define DUMMY_ENCRYPTION_ENABLED(sbi) ((sbi)->s_dummy_enc_ctx.ctx != NULL)
1364 #else
1365 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1366 #endif
1367 
1368 /* Number of quota types we support */
1369 #define EXT4_MAXQUOTAS 3
1370 
1371 #define EXT4_ENC_UTF8_12_1	1
1372 
1373 /*
1374  * fourth extended-fs super-block data in memory
1375  */
1376 struct ext4_sb_info {
1377 	unsigned long s_desc_size;	/* Size of a group descriptor in bytes */
1378 	unsigned long s_inodes_per_block;/* Number of inodes per block */
1379 	unsigned long s_blocks_per_group;/* Number of blocks in a group */
1380 	unsigned long s_clusters_per_group; /* Number of clusters in a group */
1381 	unsigned long s_inodes_per_group;/* Number of inodes in a group */
1382 	unsigned long s_itb_per_group;	/* Number of inode table blocks per group */
1383 	unsigned long s_gdb_count;	/* Number of group descriptor blocks */
1384 	unsigned long s_desc_per_block;	/* Number of group descriptors per block */
1385 	ext4_group_t s_groups_count;	/* Number of groups in the fs */
1386 	ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1387 	unsigned long s_overhead;  /* # of fs overhead clusters */
1388 	unsigned int s_cluster_ratio;	/* Number of blocks per cluster */
1389 	unsigned int s_cluster_bits;	/* log2 of s_cluster_ratio */
1390 	loff_t s_bitmap_maxbytes;	/* max bytes for bitmap files */
1391 	struct buffer_head * s_sbh;	/* Buffer containing the super block */
1392 	struct ext4_super_block *s_es;	/* Pointer to the super block in the buffer */
1393 	struct buffer_head * __rcu *s_group_desc;
1394 	unsigned int s_mount_opt;
1395 	unsigned int s_mount_opt2;
1396 	unsigned int s_mount_flags;
1397 	unsigned int s_def_mount_opt;
1398 	ext4_fsblk_t s_sb_block;
1399 	atomic64_t s_resv_clusters;
1400 	kuid_t s_resuid;
1401 	kgid_t s_resgid;
1402 	unsigned short s_mount_state;
1403 	unsigned short s_pad;
1404 	int s_addr_per_block_bits;
1405 	int s_desc_per_block_bits;
1406 	int s_inode_size;
1407 	int s_first_ino;
1408 	unsigned int s_inode_readahead_blks;
1409 	unsigned int s_inode_goal;
1410 	u32 s_hash_seed[4];
1411 	int s_def_hash_version;
1412 	int s_hash_unsigned;	/* 3 if hash should be signed, 0 if not */
1413 	struct percpu_counter s_freeclusters_counter;
1414 	struct percpu_counter s_freeinodes_counter;
1415 	struct percpu_counter s_dirs_counter;
1416 	struct percpu_counter s_dirtyclusters_counter;
1417 	struct percpu_counter s_sra_exceeded_retry_limit;
1418 	struct blockgroup_lock *s_blockgroup_lock;
1419 	struct proc_dir_entry *s_proc;
1420 	struct kobject s_kobj;
1421 	struct completion s_kobj_unregister;
1422 	struct super_block *s_sb;
1423 
1424 	/* Journaling */
1425 	struct journal_s *s_journal;
1426 	struct list_head s_orphan;
1427 	struct mutex s_orphan_lock;
1428 	unsigned long s_ext4_flags;		/* Ext4 superblock flags */
1429 	unsigned long s_commit_interval;
1430 	u32 s_max_batch_time;
1431 	u32 s_min_batch_time;
1432 	struct block_device *s_journal_bdev;
1433 #ifdef CONFIG_QUOTA
1434 	/* Names of quota files with journalled quota */
1435 	char __rcu *s_qf_names[EXT4_MAXQUOTAS];
1436 	int s_jquota_fmt;			/* Format of quota to use */
1437 #endif
1438 	unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1439 	struct ext4_system_blocks __rcu *system_blks;
1440 
1441 #ifdef EXTENTS_STATS
1442 	/* ext4 extents stats */
1443 	unsigned long s_ext_min;
1444 	unsigned long s_ext_max;
1445 	unsigned long s_depth_max;
1446 	spinlock_t s_ext_stats_lock;
1447 	unsigned long s_ext_blocks;
1448 	unsigned long s_ext_extents;
1449 #endif
1450 
1451 	/* for buddy allocator */
1452 	struct ext4_group_info ** __rcu *s_group_info;
1453 	struct inode *s_buddy_cache;
1454 	spinlock_t s_md_lock;
1455 	unsigned short *s_mb_offsets;
1456 	unsigned int *s_mb_maxs;
1457 	unsigned int s_group_info_size;
1458 	unsigned int s_mb_free_pending;
1459 	struct list_head s_freed_data_list;	/* List of blocks to be freed
1460 						   after commit completed */
1461 
1462 	/* tunables */
1463 	unsigned long s_stripe;
1464 	unsigned int s_mb_stream_request;
1465 	unsigned int s_mb_max_to_scan;
1466 	unsigned int s_mb_min_to_scan;
1467 	unsigned int s_mb_stats;
1468 	unsigned int s_mb_order2_reqs;
1469 	unsigned int s_mb_group_prealloc;
1470 	unsigned int s_max_dir_size_kb;
1471 	/* where last allocation was done - for stream allocation */
1472 	unsigned long s_mb_last_group;
1473 	unsigned long s_mb_last_start;
1474 
1475 	/* stats for buddy allocator */
1476 	atomic_t s_bal_reqs;	/* number of reqs with len > 1 */
1477 	atomic_t s_bal_success;	/* we found long enough chunks */
1478 	atomic_t s_bal_allocated;	/* in blocks */
1479 	atomic_t s_bal_ex_scanned;	/* total extents scanned */
1480 	atomic_t s_bal_goals;	/* goal hits */
1481 	atomic_t s_bal_breaks;	/* too long searches */
1482 	atomic_t s_bal_2orders;	/* 2^order hits */
1483 	spinlock_t s_bal_lock;
1484 	unsigned long s_mb_buddies_generated;
1485 	unsigned long long s_mb_generation_time;
1486 	atomic_t s_mb_lost_chunks;
1487 	atomic_t s_mb_preallocated;
1488 	atomic_t s_mb_discarded;
1489 	atomic_t s_lock_busy;
1490 
1491 	/* locality groups */
1492 	struct ext4_locality_group __percpu *s_locality_groups;
1493 
1494 	/* for write statistics */
1495 	unsigned long s_sectors_written_start;
1496 	u64 s_kbytes_written;
1497 
1498 	/* the size of zero-out chunk */
1499 	unsigned int s_extent_max_zeroout_kb;
1500 
1501 	unsigned int s_log_groups_per_flex;
1502 	struct flex_groups * __rcu *s_flex_groups;
1503 	ext4_group_t s_flex_groups_allocated;
1504 
1505 	/* workqueue for reserved extent conversions (buffered io) */
1506 	struct workqueue_struct *rsv_conversion_wq;
1507 
1508 	/* timer for periodic error stats printing */
1509 	struct timer_list s_err_report;
1510 
1511 	/* Lazy inode table initialization info */
1512 	struct ext4_li_request *s_li_request;
1513 	/* Wait multiplier for lazy initialization thread */
1514 	unsigned int s_li_wait_mult;
1515 
1516 	/* Kernel thread for multiple mount protection */
1517 	struct task_struct *s_mmp_tsk;
1518 
1519 	/* record the last minlen when FITRIM is called. */
1520 	unsigned long s_last_trim_minblks;
1521 
1522 	/* Reference to checksum algorithm driver via cryptoapi */
1523 	struct crypto_shash *s_chksum_driver;
1524 
1525 	/* Precomputed FS UUID checksum for seeding other checksums */
1526 	__u32 s_csum_seed;
1527 
1528 	/* Reclaim extents from extent status tree */
1529 	struct shrinker s_es_shrinker;
1530 	struct list_head s_es_list;	/* List of inodes with reclaimable extents */
1531 	long s_es_nr_inode;
1532 	struct ext4_es_stats s_es_stats;
1533 	struct mb_cache *s_ea_block_cache;
1534 	struct mb_cache *s_ea_inode_cache;
1535 	spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1536 
1537 	/* Ratelimit ext4 messages. */
1538 	struct ratelimit_state s_err_ratelimit_state;
1539 	struct ratelimit_state s_warning_ratelimit_state;
1540 	struct ratelimit_state s_msg_ratelimit_state;
1541 
1542 	/* Encryption context for '-o test_dummy_encryption' */
1543 	struct fscrypt_dummy_context s_dummy_enc_ctx;
1544 
1545 	/*
1546 	 * Barrier between writepages ops and changing any inode's JOURNAL_DATA
1547 	 * or EXTENTS flag.
1548 	 */
1549 	struct percpu_rw_semaphore s_writepages_rwsem;
1550 	struct dax_device *s_daxdev;
1551 };
1552 
EXT4_SB(struct super_block * sb)1553 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1554 {
1555 	return sb->s_fs_info;
1556 }
EXT4_I(struct inode * inode)1557 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1558 {
1559 	return container_of(inode, struct ext4_inode_info, vfs_inode);
1560 }
1561 
ext4_valid_inum(struct super_block * sb,unsigned long ino)1562 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1563 {
1564 	return ino == EXT4_ROOT_INO ||
1565 		(ino >= EXT4_FIRST_INO(sb) &&
1566 		 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1567 }
1568 
1569 /*
1570  * Returns: sbi->field[index]
1571  * Used to access an array element from the following sbi fields which require
1572  * rcu protection to avoid dereferencing an invalid pointer due to reassignment
1573  * - s_group_desc
1574  * - s_group_info
1575  * - s_flex_group
1576  */
1577 #define sbi_array_rcu_deref(sbi, field, index)				   \
1578 ({									   \
1579 	typeof(*((sbi)->field)) _v;					   \
1580 	rcu_read_lock();						   \
1581 	_v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index];	   \
1582 	rcu_read_unlock();						   \
1583 	_v;								   \
1584 })
1585 
1586 /*
1587  * Inode dynamic state flags
1588  */
1589 enum {
1590 	EXT4_STATE_JDATA,		/* journaled data exists */
1591 	EXT4_STATE_NEW,			/* inode is newly created */
1592 	EXT4_STATE_XATTR,		/* has in-inode xattrs */
1593 	EXT4_STATE_NO_EXPAND,		/* No space for expansion */
1594 	EXT4_STATE_DA_ALLOC_CLOSE,	/* Alloc DA blks on close */
1595 	EXT4_STATE_EXT_MIGRATE,		/* Inode is migrating */
1596 	EXT4_STATE_DIO_UNWRITTEN,	/* need convert on dio done*/
1597 	EXT4_STATE_NEWENTRY,		/* File just added to dir */
1598 	EXT4_STATE_MAY_INLINE_DATA,	/* may have in-inode data */
1599 	EXT4_STATE_EXT_PRECACHED,	/* extents have been precached */
1600 	EXT4_STATE_LUSTRE_EA_INODE,	/* Lustre-style ea_inode */
1601 	EXT4_STATE_VERITY_IN_PROGRESS,	/* building fs-verity Merkle tree */
1602 };
1603 
1604 #define EXT4_INODE_BIT_FNS(name, field, offset)				\
1605 static inline int ext4_test_inode_##name(struct inode *inode, int bit)	\
1606 {									\
1607 	return test_bit(bit + (offset), &EXT4_I(inode)->i_##field);	\
1608 }									\
1609 static inline void ext4_set_inode_##name(struct inode *inode, int bit)	\
1610 {									\
1611 	set_bit(bit + (offset), &EXT4_I(inode)->i_##field);		\
1612 }									\
1613 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1614 {									\
1615 	clear_bit(bit + (offset), &EXT4_I(inode)->i_##field);		\
1616 }
1617 
1618 /* Add these declarations here only so that these functions can be
1619  * found by name.  Otherwise, they are very hard to locate. */
1620 static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1621 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1622 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1623 EXT4_INODE_BIT_FNS(flag, flags, 0)
1624 
1625 /* Add these declarations here only so that these functions can be
1626  * found by name.  Otherwise, they are very hard to locate. */
1627 static inline int ext4_test_inode_state(struct inode *inode, int bit);
1628 static inline void ext4_set_inode_state(struct inode *inode, int bit);
1629 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1630 #if (BITS_PER_LONG < 64)
1631 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1632 
ext4_clear_state_flags(struct ext4_inode_info * ei)1633 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1634 {
1635 	(ei)->i_state_flags = 0;
1636 }
1637 #else
1638 EXT4_INODE_BIT_FNS(state, flags, 32)
1639 
ext4_clear_state_flags(struct ext4_inode_info * ei)1640 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1641 {
1642 	/* We depend on the fact that callers will set i_flags */
1643 }
1644 #endif
1645 #else
1646 /* Assume that user mode programs are passing in an ext4fs superblock, not
1647  * a kernel struct super_block.  This will allow us to call the feature-test
1648  * macros from user land. */
1649 #define EXT4_SB(sb)	(sb)
1650 #endif
1651 
ext4_verity_in_progress(struct inode * inode)1652 static inline bool ext4_verity_in_progress(struct inode *inode)
1653 {
1654 	return IS_ENABLED(CONFIG_FS_VERITY) &&
1655 	       ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS);
1656 }
1657 
1658 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1659 
1660 /*
1661  * Codes for operating systems
1662  */
1663 #define EXT4_OS_LINUX		0
1664 #define EXT4_OS_HURD		1
1665 #define EXT4_OS_MASIX		2
1666 #define EXT4_OS_FREEBSD		3
1667 #define EXT4_OS_LITES		4
1668 
1669 /*
1670  * Revision levels
1671  */
1672 #define EXT4_GOOD_OLD_REV	0	/* The good old (original) format */
1673 #define EXT4_DYNAMIC_REV	1	/* V2 format w/ dynamic inode sizes */
1674 
1675 #define EXT4_CURRENT_REV	EXT4_GOOD_OLD_REV
1676 #define EXT4_MAX_SUPP_REV	EXT4_DYNAMIC_REV
1677 
1678 #define EXT4_GOOD_OLD_INODE_SIZE 128
1679 
1680 #define EXT4_EXTRA_TIMESTAMP_MAX	(((s64)1 << 34) - 1  + S32_MIN)
1681 #define EXT4_NON_EXTRA_TIMESTAMP_MAX	S32_MAX
1682 #define EXT4_TIMESTAMP_MIN		S32_MIN
1683 
1684 /*
1685  * Feature set definitions
1686  */
1687 
1688 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC	0x0001
1689 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES	0x0002
1690 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL		0x0004
1691 #define EXT4_FEATURE_COMPAT_EXT_ATTR		0x0008
1692 #define EXT4_FEATURE_COMPAT_RESIZE_INODE	0x0010
1693 #define EXT4_FEATURE_COMPAT_DIR_INDEX		0x0020
1694 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2	0x0200
1695 #define EXT4_FEATURE_COMPAT_STABLE_INODES	0x0800
1696 
1697 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER	0x0001
1698 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE	0x0002
1699 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR	0x0004
1700 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE        0x0008
1701 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM		0x0010
1702 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK	0x0020
1703 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE	0x0040
1704 #define EXT4_FEATURE_RO_COMPAT_QUOTA		0x0100
1705 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC		0x0200
1706 /*
1707  * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM).  When
1708  * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1709  * all other data structures' checksums.  However, the METADATA_CSUM and
1710  * GDT_CSUM bits are mutually exclusive.
1711  */
1712 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM	0x0400
1713 #define EXT4_FEATURE_RO_COMPAT_READONLY		0x1000
1714 #define EXT4_FEATURE_RO_COMPAT_PROJECT		0x2000
1715 #define EXT4_FEATURE_RO_COMPAT_VERITY		0x8000
1716 
1717 #define EXT4_FEATURE_INCOMPAT_COMPRESSION	0x0001
1718 #define EXT4_FEATURE_INCOMPAT_FILETYPE		0x0002
1719 #define EXT4_FEATURE_INCOMPAT_RECOVER		0x0004 /* Needs recovery */
1720 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV	0x0008 /* Journal device */
1721 #define EXT4_FEATURE_INCOMPAT_META_BG		0x0010
1722 #define EXT4_FEATURE_INCOMPAT_EXTENTS		0x0040 /* extents support */
1723 #define EXT4_FEATURE_INCOMPAT_64BIT		0x0080
1724 #define EXT4_FEATURE_INCOMPAT_MMP               0x0100
1725 #define EXT4_FEATURE_INCOMPAT_FLEX_BG		0x0200
1726 #define EXT4_FEATURE_INCOMPAT_EA_INODE		0x0400 /* EA in inode */
1727 #define EXT4_FEATURE_INCOMPAT_DIRDATA		0x1000 /* data in dirent */
1728 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED		0x2000
1729 #define EXT4_FEATURE_INCOMPAT_LARGEDIR		0x4000 /* >2GB or 3-lvl htree */
1730 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA	0x8000 /* data in inode */
1731 #define EXT4_FEATURE_INCOMPAT_ENCRYPT		0x10000
1732 #define EXT4_FEATURE_INCOMPAT_CASEFOLD		0x20000
1733 
1734 extern void ext4_update_dynamic_rev(struct super_block *sb);
1735 
1736 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
1737 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1738 { \
1739 	return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1740 		cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
1741 } \
1742 static inline void ext4_set_feature_##name(struct super_block *sb) \
1743 { \
1744 	ext4_update_dynamic_rev(sb); \
1745 	EXT4_SB(sb)->s_es->s_feature_compat |= \
1746 		cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1747 } \
1748 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1749 { \
1750 	EXT4_SB(sb)->s_es->s_feature_compat &= \
1751 		~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1752 }
1753 
1754 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
1755 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1756 { \
1757 	return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1758 		cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
1759 } \
1760 static inline void ext4_set_feature_##name(struct super_block *sb) \
1761 { \
1762 	ext4_update_dynamic_rev(sb); \
1763 	EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
1764 		cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1765 } \
1766 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1767 { \
1768 	EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
1769 		~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1770 }
1771 
1772 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
1773 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1774 { \
1775 	return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1776 		cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
1777 } \
1778 static inline void ext4_set_feature_##name(struct super_block *sb) \
1779 { \
1780 	ext4_update_dynamic_rev(sb); \
1781 	EXT4_SB(sb)->s_es->s_feature_incompat |= \
1782 		cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1783 } \
1784 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1785 { \
1786 	EXT4_SB(sb)->s_es->s_feature_incompat &= \
1787 		~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1788 }
1789 
EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc,DIR_PREALLOC)1790 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc,		DIR_PREALLOC)
1791 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes,	IMAGIC_INODES)
1792 EXT4_FEATURE_COMPAT_FUNCS(journal,		HAS_JOURNAL)
1793 EXT4_FEATURE_COMPAT_FUNCS(xattr,		EXT_ATTR)
1794 EXT4_FEATURE_COMPAT_FUNCS(resize_inode,		RESIZE_INODE)
1795 EXT4_FEATURE_COMPAT_FUNCS(dir_index,		DIR_INDEX)
1796 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2,	SPARSE_SUPER2)
1797 EXT4_FEATURE_COMPAT_FUNCS(stable_inodes,	STABLE_INODES)
1798 
1799 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super,	SPARSE_SUPER)
1800 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file,	LARGE_FILE)
1801 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir,		BTREE_DIR)
1802 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file,		HUGE_FILE)
1803 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum,		GDT_CSUM)
1804 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink,		DIR_NLINK)
1805 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize,	EXTRA_ISIZE)
1806 EXT4_FEATURE_RO_COMPAT_FUNCS(quota,		QUOTA)
1807 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc,		BIGALLOC)
1808 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum,	METADATA_CSUM)
1809 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly,		READONLY)
1810 EXT4_FEATURE_RO_COMPAT_FUNCS(project,		PROJECT)
1811 EXT4_FEATURE_RO_COMPAT_FUNCS(verity,		VERITY)
1812 
1813 EXT4_FEATURE_INCOMPAT_FUNCS(compression,	COMPRESSION)
1814 EXT4_FEATURE_INCOMPAT_FUNCS(filetype,		FILETYPE)
1815 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery,	RECOVER)
1816 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev,	JOURNAL_DEV)
1817 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg,		META_BG)
1818 EXT4_FEATURE_INCOMPAT_FUNCS(extents,		EXTENTS)
1819 EXT4_FEATURE_INCOMPAT_FUNCS(64bit,		64BIT)
1820 EXT4_FEATURE_INCOMPAT_FUNCS(mmp,		MMP)
1821 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg,		FLEX_BG)
1822 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode,		EA_INODE)
1823 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata,		DIRDATA)
1824 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed,		CSUM_SEED)
1825 EXT4_FEATURE_INCOMPAT_FUNCS(largedir,		LARGEDIR)
1826 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data,	INLINE_DATA)
1827 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt,		ENCRYPT)
1828 EXT4_FEATURE_INCOMPAT_FUNCS(casefold,		CASEFOLD)
1829 
1830 #define EXT2_FEATURE_COMPAT_SUPP	EXT4_FEATURE_COMPAT_EXT_ATTR
1831 #define EXT2_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
1832 					 EXT4_FEATURE_INCOMPAT_META_BG)
1833 #define EXT2_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1834 					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1835 					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1836 
1837 #define EXT3_FEATURE_COMPAT_SUPP	EXT4_FEATURE_COMPAT_EXT_ATTR
1838 #define EXT3_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
1839 					 EXT4_FEATURE_INCOMPAT_RECOVER| \
1840 					 EXT4_FEATURE_INCOMPAT_META_BG)
1841 #define EXT3_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1842 					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1843 					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1844 
1845 #define EXT4_FEATURE_COMPAT_SUPP	EXT4_FEATURE_COMPAT_EXT_ATTR
1846 #define EXT4_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
1847 					 EXT4_FEATURE_INCOMPAT_RECOVER| \
1848 					 EXT4_FEATURE_INCOMPAT_META_BG| \
1849 					 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1850 					 EXT4_FEATURE_INCOMPAT_64BIT| \
1851 					 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1852 					 EXT4_FEATURE_INCOMPAT_EA_INODE| \
1853 					 EXT4_FEATURE_INCOMPAT_MMP | \
1854 					 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
1855 					 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
1856 					 EXT4_FEATURE_INCOMPAT_CASEFOLD | \
1857 					 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \
1858 					 EXT4_FEATURE_INCOMPAT_LARGEDIR)
1859 #define EXT4_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1860 					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1861 					 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1862 					 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1863 					 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1864 					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1865 					 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1866 					 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1867 					 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1868 					 EXT4_FEATURE_RO_COMPAT_QUOTA |\
1869 					 EXT4_FEATURE_RO_COMPAT_PROJECT |\
1870 					 EXT4_FEATURE_RO_COMPAT_VERITY)
1871 
1872 #define EXTN_FEATURE_FUNCS(ver) \
1873 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
1874 { \
1875 	return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1876 		cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
1877 } \
1878 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
1879 { \
1880 	return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1881 		cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
1882 } \
1883 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
1884 { \
1885 	return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1886 		cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
1887 }
1888 
1889 EXTN_FEATURE_FUNCS(2)
1890 EXTN_FEATURE_FUNCS(3)
1891 EXTN_FEATURE_FUNCS(4)
1892 
1893 static inline bool ext4_has_compat_features(struct super_block *sb)
1894 {
1895 	return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
1896 }
ext4_has_ro_compat_features(struct super_block * sb)1897 static inline bool ext4_has_ro_compat_features(struct super_block *sb)
1898 {
1899 	return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
1900 }
ext4_has_incompat_features(struct super_block * sb)1901 static inline bool ext4_has_incompat_features(struct super_block *sb)
1902 {
1903 	return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
1904 }
1905 
1906 /*
1907  * Superblock flags
1908  */
1909 #define EXT4_FLAGS_RESIZING	0
1910 #define EXT4_FLAGS_SHUTDOWN	1
1911 
ext4_forced_shutdown(struct ext4_sb_info * sbi)1912 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi)
1913 {
1914 	return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags);
1915 }
1916 
1917 
1918 /*
1919  * Default values for user and/or group using reserved blocks
1920  */
1921 #define	EXT4_DEF_RESUID		0
1922 #define	EXT4_DEF_RESGID		0
1923 
1924 /*
1925  * Default project ID
1926  */
1927 #define	EXT4_DEF_PROJID		0
1928 
1929 #define EXT4_DEF_INODE_READAHEAD_BLKS	32
1930 
1931 /*
1932  * Default mount options
1933  */
1934 #define EXT4_DEFM_DEBUG		0x0001
1935 #define EXT4_DEFM_BSDGROUPS	0x0002
1936 #define EXT4_DEFM_XATTR_USER	0x0004
1937 #define EXT4_DEFM_ACL		0x0008
1938 #define EXT4_DEFM_UID16		0x0010
1939 #define EXT4_DEFM_JMODE		0x0060
1940 #define EXT4_DEFM_JMODE_DATA	0x0020
1941 #define EXT4_DEFM_JMODE_ORDERED	0x0040
1942 #define EXT4_DEFM_JMODE_WBACK	0x0060
1943 #define EXT4_DEFM_NOBARRIER	0x0100
1944 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1945 #define EXT4_DEFM_DISCARD	0x0400
1946 #define EXT4_DEFM_NODELALLOC	0x0800
1947 
1948 /*
1949  * Default journal batch times
1950  */
1951 #define EXT4_DEF_MIN_BATCH_TIME	0
1952 #define EXT4_DEF_MAX_BATCH_TIME	15000 /* 15ms */
1953 
1954 /*
1955  * Minimum number of groups in a flexgroup before we separate out
1956  * directories into the first block group of a flexgroup
1957  */
1958 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME	4
1959 
1960 /*
1961  * Structure of a directory entry
1962  */
1963 #define EXT4_NAME_LEN 255
1964 /*
1965  * Base length of the ext4 directory entry excluding the name length
1966  */
1967 #define EXT4_BASE_DIR_LEN (sizeof(struct ext4_dir_entry_2) - EXT4_NAME_LEN)
1968 
1969 struct ext4_dir_entry {
1970 	__le32	inode;			/* Inode number */
1971 	__le16	rec_len;		/* Directory entry length */
1972 	__le16	name_len;		/* Name length */
1973 	char	name[EXT4_NAME_LEN];	/* File name */
1974 };
1975 
1976 
1977 /*
1978  * Encrypted Casefolded entries require saving the hash on disk. This structure
1979  * followed ext4_dir_entry_2's name[name_len] at the next 4 byte aligned
1980  * boundary.
1981  */
1982 struct ext4_dir_entry_hash {
1983 	__le32 hash;
1984 	__le32 minor_hash;
1985 };
1986 
1987 /*
1988  * The new version of the directory entry.  Since EXT4 structures are
1989  * stored in intel byte order, and the name_len field could never be
1990  * bigger than 255 chars, it's safe to reclaim the extra byte for the
1991  * file_type field.
1992  */
1993 struct ext4_dir_entry_2 {
1994 	__le32	inode;			/* Inode number */
1995 	__le16	rec_len;		/* Directory entry length */
1996 	__u8	name_len;		/* Name length */
1997 	__u8	file_type;
1998 	char	name[EXT4_NAME_LEN];	/* File name */
1999 };
2000 
2001 /*
2002  * Access the hashes at the end of ext4_dir_entry_2
2003  */
2004 #define EXT4_DIRENT_HASHES(entry) \
2005 	((struct ext4_dir_entry_hash *) \
2006 		(((void *)(entry)) + \
2007 		((8 + (entry)->name_len + EXT4_DIR_ROUND) & ~EXT4_DIR_ROUND)))
2008 #define EXT4_DIRENT_HASH(entry) le32_to_cpu(EXT4_DIRENT_HASHES(de)->hash)
2009 #define EXT4_DIRENT_MINOR_HASH(entry) \
2010 		le32_to_cpu(EXT4_DIRENT_HASHES(de)->minor_hash)
2011 
ext4_hash_in_dirent(const struct inode * inode)2012 static inline bool ext4_hash_in_dirent(const struct inode *inode)
2013 {
2014 	return IS_CASEFOLDED(inode) && IS_ENCRYPTED(inode);
2015 }
2016 
2017 /*
2018  * This is a bogus directory entry at the end of each leaf block that
2019  * records checksums.
2020  */
2021 struct ext4_dir_entry_tail {
2022 	__le32	det_reserved_zero1;	/* Pretend to be unused */
2023 	__le16	det_rec_len;		/* 12 */
2024 	__u8	det_reserved_zero2;	/* Zero name length */
2025 	__u8	det_reserved_ft;	/* 0xDE, fake file type */
2026 	__le32	det_checksum;		/* crc32c(uuid+inum+dirblock) */
2027 };
2028 
2029 #define EXT4_DIRENT_TAIL(block, blocksize) \
2030 	((struct ext4_dir_entry_tail *)(((void *)(block)) + \
2031 					((blocksize) - \
2032 					 sizeof(struct ext4_dir_entry_tail))))
2033 
2034 /*
2035  * Ext4 directory file types.  Only the low 3 bits are used.  The
2036  * other bits are reserved for now.
2037  */
2038 #define EXT4_FT_UNKNOWN		0
2039 #define EXT4_FT_REG_FILE	1
2040 #define EXT4_FT_DIR		2
2041 #define EXT4_FT_CHRDEV		3
2042 #define EXT4_FT_BLKDEV		4
2043 #define EXT4_FT_FIFO		5
2044 #define EXT4_FT_SOCK		6
2045 #define EXT4_FT_SYMLINK		7
2046 
2047 #define EXT4_FT_MAX		8
2048 
2049 #define EXT4_FT_DIR_CSUM	0xDE
2050 
2051 /*
2052  * EXT4_DIR_PAD defines the directory entries boundaries
2053  *
2054  * NOTE: It must be a multiple of 4
2055  */
2056 #define EXT4_DIR_PAD			4
2057 #define EXT4_DIR_ROUND			(EXT4_DIR_PAD - 1)
2058 #define EXT4_MAX_REC_LEN		((1<<16)-1)
2059 
2060 /*
2061  * The rec_len is dependent on the type of directory. Directories that are
2062  * casefolded and encrypted need to store the hash as well, so we add room for
2063  * ext4_extended_dir_entry_2. For all entries related to '.' or '..' you should
2064  * pass NULL for dir, as those entries do not use the extra fields.
2065  */
ext4_dir_rec_len(__u8 name_len,const struct inode * dir)2066 static inline unsigned int ext4_dir_rec_len(__u8 name_len,
2067 						const struct inode *dir)
2068 {
2069 	int rec_len = (name_len + 8 + EXT4_DIR_ROUND);
2070 
2071 	if (dir && ext4_hash_in_dirent(dir))
2072 		rec_len += sizeof(struct ext4_dir_entry_hash);
2073 	return (rec_len & ~EXT4_DIR_ROUND);
2074 }
2075 
2076 /*
2077  * If we ever get support for fs block sizes > page_size, we'll need
2078  * to remove the #if statements in the next two functions...
2079  */
2080 static inline unsigned int
ext4_rec_len_from_disk(__le16 dlen,unsigned blocksize)2081 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
2082 {
2083 	unsigned len = le16_to_cpu(dlen);
2084 
2085 #if (PAGE_SIZE >= 65536)
2086 	if (len == EXT4_MAX_REC_LEN || len == 0)
2087 		return blocksize;
2088 	return (len & 65532) | ((len & 3) << 16);
2089 #else
2090 	return len;
2091 #endif
2092 }
2093 
ext4_rec_len_to_disk(unsigned len,unsigned blocksize)2094 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
2095 {
2096 	if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
2097 		BUG();
2098 #if (PAGE_SIZE >= 65536)
2099 	if (len < 65536)
2100 		return cpu_to_le16(len);
2101 	if (len == blocksize) {
2102 		if (blocksize == 65536)
2103 			return cpu_to_le16(EXT4_MAX_REC_LEN);
2104 		else
2105 			return cpu_to_le16(0);
2106 	}
2107 	return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
2108 #else
2109 	return cpu_to_le16(len);
2110 #endif
2111 }
2112 
2113 /*
2114  * Hash Tree Directory indexing
2115  * (c) Daniel Phillips, 2001
2116  */
2117 
2118 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
2119 		    ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2120 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \
2121 		    !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir)))
2122 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2123 
2124 /* Legal values for the dx_root hash_version field: */
2125 
2126 #define DX_HASH_LEGACY			0
2127 #define DX_HASH_HALF_MD4		1
2128 #define DX_HASH_TEA			2
2129 #define DX_HASH_LEGACY_UNSIGNED		3
2130 #define DX_HASH_HALF_MD4_UNSIGNED	4
2131 #define DX_HASH_TEA_UNSIGNED		5
2132 #define DX_HASH_SIPHASH			6
2133 
ext4_chksum(struct ext4_sb_info * sbi,u32 crc,const void * address,unsigned int length)2134 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2135 			      const void *address, unsigned int length)
2136 {
2137 	struct {
2138 		struct shash_desc shash;
2139 		char ctx[4];
2140 	} desc;
2141 
2142 	BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2143 
2144 	desc.shash.tfm = sbi->s_chksum_driver;
2145 	*(u32 *)desc.ctx = crc;
2146 
2147 	BUG_ON(crypto_shash_update(&desc.shash, address, length));
2148 
2149 	return *(u32 *)desc.ctx;
2150 }
2151 
2152 #ifdef __KERNEL__
2153 
2154 /* hash info structure used by the directory hash */
2155 struct dx_hash_info
2156 {
2157 	u32		hash;
2158 	u32		minor_hash;
2159 	int		hash_version;
2160 	u32		*seed;
2161 };
2162 
2163 
2164 /* 32 and 64 bit signed EOF for dx directories */
2165 #define EXT4_HTREE_EOF_32BIT   ((1UL  << (32 - 1)) - 1)
2166 #define EXT4_HTREE_EOF_64BIT   ((1ULL << (64 - 1)) - 1)
2167 
2168 
2169 /*
2170  * Control parameters used by ext4_htree_next_block
2171  */
2172 #define HASH_NB_ALWAYS		1
2173 
2174 struct ext4_filename {
2175 	const struct qstr *usr_fname;
2176 	struct fscrypt_str disk_name;
2177 	struct dx_hash_info hinfo;
2178 #ifdef CONFIG_FS_ENCRYPTION
2179 	struct fscrypt_str crypto_buf;
2180 #endif
2181 #ifdef CONFIG_UNICODE
2182 	struct fscrypt_str cf_name;
2183 #endif
2184 };
2185 
2186 #define fname_name(p) ((p)->disk_name.name)
2187 #define fname_usr_name(p) ((p)->usr_fname->name)
2188 #define fname_len(p)  ((p)->disk_name.len)
2189 
2190 /*
2191  * Describe an inode's exact location on disk and in memory
2192  */
2193 struct ext4_iloc
2194 {
2195 	struct buffer_head *bh;
2196 	unsigned long offset;
2197 	ext4_group_t block_group;
2198 };
2199 
ext4_raw_inode(struct ext4_iloc * iloc)2200 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2201 {
2202 	return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2203 }
2204 
ext4_is_quota_file(struct inode * inode)2205 static inline bool ext4_is_quota_file(struct inode *inode)
2206 {
2207 	return IS_NOQUOTA(inode) &&
2208 	       !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL);
2209 }
2210 
2211 /*
2212  * This structure is stuffed into the struct file's private_data field
2213  * for directories.  It is where we put information so that we can do
2214  * readdir operations in hash tree order.
2215  */
2216 struct dir_private_info {
2217 	struct rb_root	root;
2218 	struct rb_node	*curr_node;
2219 	struct fname	*extra_fname;
2220 	loff_t		last_pos;
2221 	__u32		curr_hash;
2222 	__u32		curr_minor_hash;
2223 	__u32		next_hash;
2224 };
2225 
2226 /* calculate the first block number of the group */
2227 static inline ext4_fsblk_t
ext4_group_first_block_no(struct super_block * sb,ext4_group_t group_no)2228 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2229 {
2230 	return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2231 		le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2232 }
2233 
2234 /*
2235  * Special error return code only used by dx_probe() and its callers.
2236  */
2237 #define ERR_BAD_DX_DIR	(-(MAX_ERRNO - 1))
2238 
2239 /* htree levels for ext4 */
2240 #define	EXT4_HTREE_LEVEL_COMPAT	2
2241 #define	EXT4_HTREE_LEVEL	3
2242 
ext4_dir_htree_level(struct super_block * sb)2243 static inline int ext4_dir_htree_level(struct super_block *sb)
2244 {
2245 	return ext4_has_feature_largedir(sb) ?
2246 		EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT;
2247 }
2248 
2249 /*
2250  * Timeout and state flag for lazy initialization inode thread.
2251  */
2252 #define EXT4_DEF_LI_WAIT_MULT			10
2253 #define EXT4_DEF_LI_MAX_START_DELAY		5
2254 #define EXT4_LAZYINIT_QUIT			0x0001
2255 #define EXT4_LAZYINIT_RUNNING			0x0002
2256 
2257 /*
2258  * Lazy inode table initialization info
2259  */
2260 struct ext4_lazy_init {
2261 	unsigned long		li_state;
2262 	struct list_head	li_request_list;
2263 	struct mutex		li_list_mtx;
2264 };
2265 
2266 struct ext4_li_request {
2267 	struct super_block	*lr_super;
2268 	struct ext4_sb_info	*lr_sbi;
2269 	ext4_group_t		lr_next_group;
2270 	struct list_head	lr_request;
2271 	unsigned long		lr_next_sched;
2272 	unsigned long		lr_timeout;
2273 };
2274 
2275 struct ext4_features {
2276 	struct kobject f_kobj;
2277 	struct completion f_kobj_unregister;
2278 };
2279 
2280 /*
2281  * This structure will be used for multiple mount protection. It will be
2282  * written into the block number saved in the s_mmp_block field in the
2283  * superblock. Programs that check MMP should assume that if
2284  * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2285  * to use the filesystem, regardless of how old the timestamp is.
2286  */
2287 #define EXT4_MMP_MAGIC     0x004D4D50U /* ASCII for MMP */
2288 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2289 #define EXT4_MMP_SEQ_FSCK  0xE24D4D50U /* mmp_seq value when being fscked */
2290 #define EXT4_MMP_SEQ_MAX   0xE24D4D4FU /* maximum valid mmp_seq value */
2291 
2292 struct mmp_struct {
2293 	__le32	mmp_magic;		/* Magic number for MMP */
2294 	__le32	mmp_seq;		/* Sequence no. updated periodically */
2295 
2296 	/*
2297 	 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2298 	 * purposes and do not affect the correctness of the algorithm
2299 	 */
2300 	__le64	mmp_time;		/* Time last updated */
2301 	char	mmp_nodename[64];	/* Node which last updated MMP block */
2302 	char	mmp_bdevname[32];	/* Bdev which last updated MMP block */
2303 
2304 	/*
2305 	 * mmp_check_interval is used to verify if the MMP block has been
2306 	 * updated on the block device. The value is updated based on the
2307 	 * maximum time to write the MMP block during an update cycle.
2308 	 */
2309 	__le16	mmp_check_interval;
2310 
2311 	__le16	mmp_pad1;
2312 	__le32	mmp_pad2[226];
2313 	__le32	mmp_checksum;		/* crc32c(uuid+mmp_block) */
2314 };
2315 
2316 /* arguments passed to the mmp thread */
2317 struct mmpd_data {
2318 	struct buffer_head *bh; /* bh from initial read_mmp_block() */
2319 	struct super_block *sb;  /* super block of the fs */
2320 };
2321 
2322 /*
2323  * Check interval multiplier
2324  * The MMP block is written every update interval and initially checked every
2325  * update interval x the multiplier (the value is then adapted based on the
2326  * write latency). The reason is that writes can be delayed under load and we
2327  * don't want readers to incorrectly assume that the filesystem is no longer
2328  * in use.
2329  */
2330 #define EXT4_MMP_CHECK_MULT		2UL
2331 
2332 /*
2333  * Minimum interval for MMP checking in seconds.
2334  */
2335 #define EXT4_MMP_MIN_CHECK_INTERVAL	5UL
2336 
2337 /*
2338  * Maximum interval for MMP checking in seconds.
2339  */
2340 #define EXT4_MMP_MAX_CHECK_INTERVAL	300UL
2341 
2342 /*
2343  * Function prototypes
2344  */
2345 
2346 /*
2347  * Ok, these declarations are also in <linux/kernel.h> but none of the
2348  * ext4 source programs needs to include it so they are duplicated here.
2349  */
2350 # define NORET_TYPE	/**/
2351 # define ATTRIB_NORET	__attribute__((noreturn))
2352 # define NORET_AND	noreturn,
2353 
2354 /* bitmap.c */
2355 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2356 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2357 				struct ext4_group_desc *gdp,
2358 				struct buffer_head *bh, int sz);
2359 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2360 				  struct ext4_group_desc *gdp,
2361 				  struct buffer_head *bh, int sz);
2362 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2363 				struct ext4_group_desc *gdp,
2364 				struct buffer_head *bh);
2365 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2366 				  struct ext4_group_desc *gdp,
2367 				  struct buffer_head *bh);
2368 
2369 /* balloc.c */
2370 extern void ext4_get_group_no_and_offset(struct super_block *sb,
2371 					 ext4_fsblk_t blocknr,
2372 					 ext4_group_t *blockgrpp,
2373 					 ext4_grpblk_t *offsetp);
2374 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2375 					  ext4_fsblk_t block);
2376 
2377 extern unsigned int ext4_block_group(struct super_block *sb,
2378 			ext4_fsblk_t blocknr);
2379 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2380 			ext4_fsblk_t blocknr);
2381 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2382 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2383 			ext4_group_t group);
2384 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2385 					 ext4_fsblk_t goal,
2386 					 unsigned int flags,
2387 					 unsigned long *count,
2388 					 int *errp);
2389 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2390 				    s64 nclusters, unsigned int flags);
2391 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2392 extern void ext4_check_blocks_bitmap(struct super_block *);
2393 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2394 						    ext4_group_t block_group,
2395 						    struct buffer_head ** bh);
2396 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2397 
2398 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2399 						ext4_group_t block_group);
2400 extern int ext4_wait_block_bitmap(struct super_block *sb,
2401 				  ext4_group_t block_group,
2402 				  struct buffer_head *bh);
2403 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2404 						  ext4_group_t block_group);
2405 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2406 					      ext4_group_t block_group,
2407 					      struct ext4_group_desc *gdp);
2408 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2409 
2410 #ifdef CONFIG_UNICODE
2411 extern int ext4_fname_setup_ci_filename(struct inode *dir,
2412 					 const struct qstr *iname,
2413 					 struct ext4_filename *fname);
2414 #endif
2415 
2416 #ifdef CONFIG_FS_ENCRYPTION
ext4_fname_from_fscrypt_name(struct ext4_filename * dst,const struct fscrypt_name * src)2417 static inline void ext4_fname_from_fscrypt_name(struct ext4_filename *dst,
2418 						const struct fscrypt_name *src)
2419 {
2420 	memset(dst, 0, sizeof(*dst));
2421 
2422 	dst->usr_fname = src->usr_fname;
2423 	dst->disk_name = src->disk_name;
2424 	dst->hinfo.hash = src->hash;
2425 	dst->hinfo.minor_hash = src->minor_hash;
2426 	dst->crypto_buf = src->crypto_buf;
2427 }
2428 
ext4_fname_setup_filename(struct inode * dir,const struct qstr * iname,int lookup,struct ext4_filename * fname)2429 static inline int ext4_fname_setup_filename(struct inode *dir,
2430 					    const struct qstr *iname,
2431 					    int lookup,
2432 					    struct ext4_filename *fname)
2433 {
2434 	struct fscrypt_name name;
2435 	int err;
2436 
2437 	err = fscrypt_setup_filename(dir, iname, lookup, &name);
2438 	if (err)
2439 		return err;
2440 
2441 	ext4_fname_from_fscrypt_name(fname, &name);
2442 
2443 #ifdef CONFIG_UNICODE
2444 	err = ext4_fname_setup_ci_filename(dir, iname, fname);
2445 #endif
2446 	return err;
2447 }
2448 
ext4_fname_prepare_lookup(struct inode * dir,struct dentry * dentry,struct ext4_filename * fname)2449 static inline int ext4_fname_prepare_lookup(struct inode *dir,
2450 					    struct dentry *dentry,
2451 					    struct ext4_filename *fname)
2452 {
2453 	struct fscrypt_name name;
2454 	int err;
2455 
2456 	err = fscrypt_prepare_lookup(dir, dentry, &name);
2457 	if (err)
2458 		return err;
2459 
2460 	ext4_fname_from_fscrypt_name(fname, &name);
2461 
2462 #ifdef CONFIG_UNICODE
2463 	err = ext4_fname_setup_ci_filename(dir, &dentry->d_name, fname);
2464 #endif
2465 	return err;
2466 }
2467 
ext4_fname_free_filename(struct ext4_filename * fname)2468 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2469 {
2470 	struct fscrypt_name name;
2471 
2472 	name.crypto_buf = fname->crypto_buf;
2473 	fscrypt_free_filename(&name);
2474 
2475 	fname->crypto_buf.name = NULL;
2476 	fname->usr_fname = NULL;
2477 	fname->disk_name.name = NULL;
2478 
2479 #ifdef CONFIG_UNICODE
2480 	kfree(fname->cf_name.name);
2481 	fname->cf_name.name = NULL;
2482 #endif
2483 }
2484 #else /* !CONFIG_FS_ENCRYPTION */
ext4_fname_setup_filename(struct inode * dir,const struct qstr * iname,int lookup,struct ext4_filename * fname)2485 static inline int ext4_fname_setup_filename(struct inode *dir,
2486 					    const struct qstr *iname,
2487 					    int lookup,
2488 					    struct ext4_filename *fname)
2489 {
2490 	int err = 0;
2491 	fname->usr_fname = iname;
2492 	fname->disk_name.name = (unsigned char *) iname->name;
2493 	fname->disk_name.len = iname->len;
2494 
2495 #ifdef CONFIG_UNICODE
2496 	err = ext4_fname_setup_ci_filename(dir, iname, fname);
2497 #endif
2498 
2499 	return err;
2500 }
2501 
ext4_fname_prepare_lookup(struct inode * dir,struct dentry * dentry,struct ext4_filename * fname)2502 static inline int ext4_fname_prepare_lookup(struct inode *dir,
2503 					    struct dentry *dentry,
2504 					    struct ext4_filename *fname)
2505 {
2506 	return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname);
2507 }
2508 
ext4_fname_free_filename(struct ext4_filename * fname)2509 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2510 {
2511 #ifdef CONFIG_UNICODE
2512 	kfree(fname->cf_name.name);
2513 	fname->cf_name.name = NULL;
2514 #endif
2515 }
2516 #endif /* !CONFIG_FS_ENCRYPTION */
2517 
2518 /* dir.c */
2519 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2520 				  struct file *,
2521 				  struct ext4_dir_entry_2 *,
2522 				  struct buffer_head *, char *, int,
2523 				  unsigned int);
2524 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2525 	unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2526 				(de), (bh), (buf), (size), (offset)))
2527 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2528 				__u32 minor_hash,
2529 				struct ext4_dir_entry_2 *dirent,
2530 				struct fscrypt_str *ent_name);
2531 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2532 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2533 			     struct buffer_head *bh,
2534 			     void *buf, int buf_size,
2535 			     struct ext4_filename *fname,
2536 			     struct ext4_dir_entry_2 **dest_de);
2537 void ext4_insert_dentry(struct inode *dir, struct inode *inode,
2538 			struct ext4_dir_entry_2 *de,
2539 			int buf_size,
2540 			struct ext4_filename *fname);
ext4_update_dx_flag(struct inode * inode)2541 static inline void ext4_update_dx_flag(struct inode *inode)
2542 {
2543 	if (!ext4_has_feature_dir_index(inode->i_sb) &&
2544 	    ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) {
2545 		/* ext4_iget() should have caught this... */
2546 		WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb));
2547 		ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2548 	}
2549 }
2550 static const unsigned char ext4_filetype_table[] = {
2551 	DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2552 };
2553 
get_dtype(struct super_block * sb,int filetype)2554 static inline  unsigned char get_dtype(struct super_block *sb, int filetype)
2555 {
2556 	if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2557 		return DT_UNKNOWN;
2558 
2559 	return ext4_filetype_table[filetype];
2560 }
2561 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2562 			     void *buf, int buf_size);
2563 
2564 /* fsync.c */
2565 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2566 
2567 /* hash.c */
2568 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len,
2569 			  struct dx_hash_info *hinfo);
2570 
2571 /* ialloc.c */
2572 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2573 				      const struct qstr *qstr, __u32 goal,
2574 				      uid_t *owner, __u32 i_flags,
2575 				      int handle_type, unsigned int line_no,
2576 				      int nblocks);
2577 
2578 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \
2579 	__ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2580 			 i_flags, 0, 0, 0)
2581 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2582 				    type, nblocks)		    \
2583 	__ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2584 			 0, (type), __LINE__, (nblocks))
2585 
2586 
2587 extern void ext4_free_inode(handle_t *, struct inode *);
2588 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2589 extern unsigned long ext4_count_free_inodes(struct super_block *);
2590 extern unsigned long ext4_count_dirs(struct super_block *);
2591 extern void ext4_check_inodes_bitmap(struct super_block *);
2592 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2593 extern int ext4_init_inode_table(struct super_block *sb,
2594 				 ext4_group_t group, int barrier);
2595 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2596 
2597 /* mballoc.c */
2598 extern const struct seq_operations ext4_mb_seq_groups_ops;
2599 extern long ext4_mb_stats;
2600 extern long ext4_mb_max_to_scan;
2601 extern int ext4_mb_init(struct super_block *);
2602 extern int ext4_mb_release(struct super_block *);
2603 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2604 				struct ext4_allocation_request *, int *);
2605 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2606 extern void ext4_discard_preallocations(struct inode *);
2607 extern int __init ext4_init_mballoc(void);
2608 extern void ext4_exit_mballoc(void);
2609 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2610 			     struct buffer_head *bh, ext4_fsblk_t block,
2611 			     unsigned long count, int flags);
2612 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2613 				   ext4_group_t ngroups);
2614 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2615 		ext4_group_t i, struct ext4_group_desc *desc);
2616 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2617 				ext4_fsblk_t block, unsigned long count);
2618 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2619 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid);
2620 
2621 /* inode.c */
2622 int ext4_inode_is_fast_symlink(struct inode *inode);
2623 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2624 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2625 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count,
2626 		     bool wait, struct buffer_head **bhs);
2627 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2628 			     struct buffer_head *bh_result, int create);
2629 int ext4_get_block(struct inode *inode, sector_t iblock,
2630 		   struct buffer_head *bh_result, int create);
2631 int ext4_dio_get_block(struct inode *inode, sector_t iblock,
2632 		       struct buffer_head *bh_result, int create);
2633 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2634 			   struct buffer_head *bh, int create);
2635 int ext4_walk_page_buffers(handle_t *handle,
2636 			   struct buffer_head *head,
2637 			   unsigned from,
2638 			   unsigned to,
2639 			   int *partial,
2640 			   int (*fn)(handle_t *handle,
2641 				     struct buffer_head *bh));
2642 int do_journal_get_write_access(handle_t *handle,
2643 				struct buffer_head *bh);
2644 #define FALL_BACK_TO_NONDELALLOC 1
2645 #define CONVERT_INLINE_DATA	 2
2646 
2647 typedef enum {
2648 	EXT4_IGET_NORMAL =	0,
2649 	EXT4_IGET_SPECIAL =	0x0001, /* OK to iget a system inode */
2650 	EXT4_IGET_HANDLE = 	0x0002,	/* Inode # is from a handle */
2651 	EXT4_IGET_BAD =		0x0004, /* Allow to iget a bad inode */
2652 	EXT4_IGET_EA_INODE =	0x0008	/* Inode should contain an EA value */
2653 } ext4_iget_flags;
2654 
2655 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
2656 				 ext4_iget_flags flags, const char *function,
2657 				 unsigned int line);
2658 
2659 #define ext4_iget(sb, ino, flags) \
2660 	__ext4_iget((sb), (ino), (flags), __func__, __LINE__)
2661 
2662 extern int  ext4_write_inode(struct inode *, struct writeback_control *);
2663 extern int  ext4_setattr(struct dentry *, struct iattr *);
2664 extern int  ext4_getattr(const struct path *, struct kstat *, u32, unsigned int);
2665 extern void ext4_evict_inode(struct inode *);
2666 extern void ext4_clear_inode(struct inode *);
2667 extern int  ext4_file_getattr(const struct path *, struct kstat *, u32, unsigned int);
2668 extern int  ext4_sync_inode(handle_t *, struct inode *);
2669 extern void ext4_dirty_inode(struct inode *, int);
2670 extern int ext4_change_inode_journal_flag(struct inode *, int);
2671 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2672 extern int ext4_inode_attach_jinode(struct inode *inode);
2673 extern int ext4_can_truncate(struct inode *inode);
2674 extern int ext4_truncate(struct inode *);
2675 extern int ext4_break_layouts(struct inode *);
2676 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2677 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
2678 extern void ext4_set_inode_flags(struct inode *);
2679 extern int ext4_alloc_da_blocks(struct inode *inode);
2680 extern void ext4_set_aops(struct inode *inode);
2681 extern int ext4_writepage_trans_blocks(struct inode *);
2682 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2683 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2684 			     loff_t lstart, loff_t lend);
2685 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf);
2686 extern vm_fault_t ext4_filemap_fault(struct vm_fault *vmf);
2687 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2688 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
2689 extern void ext4_da_release_space(struct inode *inode, int to_free);
2690 extern void ext4_da_update_reserve_space(struct inode *inode,
2691 					int used, int quota_claim);
2692 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
2693 			      ext4_fsblk_t pblk, ext4_lblk_t len);
2694 
2695 /* indirect.c */
2696 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2697 				struct ext4_map_blocks *map, int flags);
2698 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2699 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2700 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2701 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2702 				 ext4_lblk_t start, ext4_lblk_t end);
2703 
2704 /* ioctl.c */
2705 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2706 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2707 
2708 /* migrate.c */
2709 extern int ext4_ext_migrate(struct inode *);
2710 extern int ext4_ind_migrate(struct inode *inode);
2711 
2712 /* namei.c */
2713 extern int ext4_dirblock_csum_verify(struct inode *inode,
2714 				     struct buffer_head *bh);
2715 extern int ext4_orphan_add(handle_t *, struct inode *);
2716 extern int ext4_orphan_del(handle_t *, struct inode *);
2717 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2718 				__u32 start_minor_hash, __u32 *next_hash);
2719 extern int ext4_search_dir(struct buffer_head *bh,
2720 			   char *search_buf,
2721 			   int buf_size,
2722 			   struct inode *dir,
2723 			   struct ext4_filename *fname,
2724 			   unsigned int offset,
2725 			   struct ext4_dir_entry_2 **res_dir);
2726 extern int ext4_generic_delete_entry(handle_t *handle,
2727 				     struct inode *dir,
2728 				     struct ext4_dir_entry_2 *de_del,
2729 				     struct buffer_head *bh,
2730 				     void *entry_buf,
2731 				     int buf_size,
2732 				     int csum_size);
2733 extern bool ext4_empty_dir(struct inode *inode);
2734 
2735 /* resize.c */
2736 extern void ext4_kvfree_array_rcu(void *to_free);
2737 extern int ext4_group_add(struct super_block *sb,
2738 				struct ext4_new_group_data *input);
2739 extern int ext4_group_extend(struct super_block *sb,
2740 				struct ext4_super_block *es,
2741 				ext4_fsblk_t n_blocks_count);
2742 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2743 
2744 /* super.c */
2745 extern struct buffer_head *ext4_sb_bread(struct super_block *sb,
2746 					 sector_t block, int op_flags);
2747 extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
2748 extern int ext4_calculate_overhead(struct super_block *sb);
2749 extern void ext4_superblock_csum_set(struct super_block *sb);
2750 extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2751 extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2752 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2753 				    ext4_group_t ngroup);
2754 extern const char *ext4_decode_error(struct super_block *sb, int errno,
2755 				     char nbuf[16]);
2756 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb,
2757 					     ext4_group_t block_group,
2758 					     unsigned int flags);
2759 
2760 extern __printf(4, 5)
2761 void __ext4_error(struct super_block *, const char *, unsigned int,
2762 		  const char *, ...);
2763 extern __printf(5, 6)
2764 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2765 		      const char *, ...);
2766 extern __printf(5, 6)
2767 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2768 		     const char *, ...);
2769 extern void __ext4_std_error(struct super_block *, const char *,
2770 			     unsigned int, int);
2771 extern __printf(4, 5)
2772 void __ext4_abort(struct super_block *, const char *, unsigned int,
2773 		  const char *, ...);
2774 extern __printf(4, 5)
2775 void __ext4_warning(struct super_block *, const char *, unsigned int,
2776 		    const char *, ...);
2777 extern __printf(4, 5)
2778 void __ext4_warning_inode(const struct inode *inode, const char *function,
2779 			  unsigned int line, const char *fmt, ...);
2780 extern __printf(3, 4)
2781 void __ext4_msg(struct super_block *, const char *, const char *, ...);
2782 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2783 			   const char *, unsigned int, const char *);
2784 extern __printf(7, 8)
2785 void __ext4_grp_locked_error(const char *, unsigned int,
2786 			     struct super_block *, ext4_group_t,
2787 			     unsigned long, ext4_fsblk_t,
2788 			     const char *, ...);
2789 
2790 #define EXT4_ERROR_INODE(inode, fmt, a...) \
2791 	ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
2792 
2793 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...)			\
2794 	ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
2795 
2796 #define EXT4_ERROR_FILE(file, block, fmt, a...)				\
2797 	ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
2798 
2799 #ifdef CONFIG_PRINTK
2800 
2801 #define ext4_error_inode(inode, func, line, block, fmt, ...)		\
2802 	__ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__)
2803 #define ext4_error_file(file, func, line, block, fmt, ...)		\
2804 	__ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2805 #define ext4_error(sb, fmt, ...)					\
2806 	__ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2807 #define ext4_abort(sb, fmt, ...)					\
2808 	__ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2809 #define ext4_warning(sb, fmt, ...)					\
2810 	__ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2811 #define ext4_warning_inode(inode, fmt, ...)				\
2812 	__ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
2813 #define ext4_msg(sb, level, fmt, ...)				\
2814 	__ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2815 #define dump_mmp_msg(sb, mmp, msg)					\
2816 	__dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2817 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...)		\
2818 	__ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2819 				fmt, ##__VA_ARGS__)
2820 
2821 #else
2822 
2823 #define ext4_error_inode(inode, func, line, block, fmt, ...)		\
2824 do {									\
2825 	no_printk(fmt, ##__VA_ARGS__);					\
2826 	__ext4_error_inode(inode, "", 0, block, " ");			\
2827 } while (0)
2828 #define ext4_error_file(file, func, line, block, fmt, ...)		\
2829 do {									\
2830 	no_printk(fmt, ##__VA_ARGS__);					\
2831 	__ext4_error_file(file, "", 0, block, " ");			\
2832 } while (0)
2833 #define ext4_error(sb, fmt, ...)					\
2834 do {									\
2835 	no_printk(fmt, ##__VA_ARGS__);					\
2836 	__ext4_error(sb, "", 0, " ");					\
2837 } while (0)
2838 #define ext4_abort(sb, fmt, ...)					\
2839 do {									\
2840 	no_printk(fmt, ##__VA_ARGS__);					\
2841 	__ext4_abort(sb, "", 0, " ");					\
2842 } while (0)
2843 #define ext4_warning(sb, fmt, ...)					\
2844 do {									\
2845 	no_printk(fmt, ##__VA_ARGS__);					\
2846 	__ext4_warning(sb, "", 0, " ");					\
2847 } while (0)
2848 #define ext4_warning_inode(inode, fmt, ...)				\
2849 do {									\
2850 	no_printk(fmt, ##__VA_ARGS__);					\
2851 	__ext4_warning_inode(inode, "", 0, " ");			\
2852 } while (0)
2853 #define ext4_msg(sb, level, fmt, ...)					\
2854 do {									\
2855 	no_printk(fmt, ##__VA_ARGS__);					\
2856 	__ext4_msg(sb, "", " ");					\
2857 } while (0)
2858 #define dump_mmp_msg(sb, mmp, msg)					\
2859 	__dump_mmp_msg(sb, mmp, "", 0, "")
2860 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...)		\
2861 do {									\
2862 	no_printk(fmt, ##__VA_ARGS__);				\
2863 	__ext4_grp_locked_error("", 0, sb, grp, ino, block, " ");	\
2864 } while (0)
2865 
2866 #endif
2867 
2868 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2869 					__u32 compat);
2870 extern int ext4_update_rocompat_feature(handle_t *handle,
2871 					struct super_block *sb,	__u32 rocompat);
2872 extern int ext4_update_incompat_feature(handle_t *handle,
2873 					struct super_block *sb,	__u32 incompat);
2874 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2875 				      struct ext4_group_desc *bg);
2876 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2877 				      struct ext4_group_desc *bg);
2878 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2879 				     struct ext4_group_desc *bg);
2880 extern __u32 ext4_free_group_clusters(struct super_block *sb,
2881 				      struct ext4_group_desc *bg);
2882 extern __u32 ext4_free_inodes_count(struct super_block *sb,
2883 				 struct ext4_group_desc *bg);
2884 extern __u32 ext4_used_dirs_count(struct super_block *sb,
2885 				struct ext4_group_desc *bg);
2886 extern __u32 ext4_itable_unused_count(struct super_block *sb,
2887 				   struct ext4_group_desc *bg);
2888 extern void ext4_block_bitmap_set(struct super_block *sb,
2889 				  struct ext4_group_desc *bg, ext4_fsblk_t blk);
2890 extern void ext4_inode_bitmap_set(struct super_block *sb,
2891 				  struct ext4_group_desc *bg, ext4_fsblk_t blk);
2892 extern void ext4_inode_table_set(struct super_block *sb,
2893 				 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2894 extern void ext4_free_group_clusters_set(struct super_block *sb,
2895 					 struct ext4_group_desc *bg,
2896 					 __u32 count);
2897 extern void ext4_free_inodes_set(struct super_block *sb,
2898 				struct ext4_group_desc *bg, __u32 count);
2899 extern void ext4_used_dirs_set(struct super_block *sb,
2900 				struct ext4_group_desc *bg, __u32 count);
2901 extern void ext4_itable_unused_set(struct super_block *sb,
2902 				   struct ext4_group_desc *bg, __u32 count);
2903 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2904 				       struct ext4_group_desc *gdp);
2905 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2906 				     struct ext4_group_desc *gdp);
2907 extern int ext4_register_li_request(struct super_block *sb,
2908 				    ext4_group_t first_not_zeroed);
2909 
ext4_has_metadata_csum(struct super_block * sb)2910 static inline int ext4_has_metadata_csum(struct super_block *sb)
2911 {
2912 	WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
2913 		     !EXT4_SB(sb)->s_chksum_driver);
2914 
2915 	return ext4_has_feature_metadata_csum(sb) &&
2916 	       (EXT4_SB(sb)->s_chksum_driver != NULL);
2917 }
2918 
ext4_has_group_desc_csum(struct super_block * sb)2919 static inline int ext4_has_group_desc_csum(struct super_block *sb)
2920 {
2921 	return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb);
2922 }
2923 
ext4_blocks_count(struct ext4_super_block * es)2924 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2925 {
2926 	return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2927 		le32_to_cpu(es->s_blocks_count_lo);
2928 }
2929 
ext4_r_blocks_count(struct ext4_super_block * es)2930 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2931 {
2932 	return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2933 		le32_to_cpu(es->s_r_blocks_count_lo);
2934 }
2935 
ext4_free_blocks_count(struct ext4_super_block * es)2936 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2937 {
2938 	return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2939 		le32_to_cpu(es->s_free_blocks_count_lo);
2940 }
2941 
ext4_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)2942 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2943 					 ext4_fsblk_t blk)
2944 {
2945 	es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2946 	es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2947 }
2948 
ext4_free_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)2949 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2950 					      ext4_fsblk_t blk)
2951 {
2952 	es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2953 	es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2954 }
2955 
ext4_r_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)2956 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2957 					   ext4_fsblk_t blk)
2958 {
2959 	es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2960 	es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2961 }
2962 
ext4_isize(struct super_block * sb,struct ext4_inode * raw_inode)2963 static inline loff_t ext4_isize(struct super_block *sb,
2964 				struct ext4_inode *raw_inode)
2965 {
2966 	if (ext4_has_feature_largedir(sb) ||
2967 	    S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2968 		return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2969 			le32_to_cpu(raw_inode->i_size_lo);
2970 
2971 	return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2972 }
2973 
ext4_isize_set(struct ext4_inode * raw_inode,loff_t i_size)2974 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2975 {
2976 	raw_inode->i_size_lo = cpu_to_le32(i_size);
2977 	raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2978 }
2979 
2980 static inline
ext4_get_group_info(struct super_block * sb,ext4_group_t group)2981 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2982 					    ext4_group_t group)
2983 {
2984 	 struct ext4_group_info **grp_info;
2985 	 long indexv, indexh;
2986 	 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
2987 	 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2988 	 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2989 	 grp_info = sbi_array_rcu_deref(EXT4_SB(sb), s_group_info, indexv);
2990 	 return grp_info[indexh];
2991 }
2992 
2993 /*
2994  * Reading s_groups_count requires using smp_rmb() afterwards.  See
2995  * the locking protocol documented in the comments of ext4_group_add()
2996  * in resize.c
2997  */
ext4_get_groups_count(struct super_block * sb)2998 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2999 {
3000 	ext4_group_t	ngroups = EXT4_SB(sb)->s_groups_count;
3001 
3002 	smp_rmb();
3003 	return ngroups;
3004 }
3005 
ext4_flex_group(struct ext4_sb_info * sbi,ext4_group_t block_group)3006 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
3007 					     ext4_group_t block_group)
3008 {
3009 	return block_group >> sbi->s_log_groups_per_flex;
3010 }
3011 
ext4_flex_bg_size(struct ext4_sb_info * sbi)3012 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
3013 {
3014 	return 1 << sbi->s_log_groups_per_flex;
3015 }
3016 
3017 #define ext4_std_error(sb, errno)				\
3018 do {								\
3019 	if ((errno))						\
3020 		__ext4_std_error((sb), __func__, __LINE__, (errno));	\
3021 } while (0)
3022 
3023 #ifdef CONFIG_SMP
3024 /* Each CPU can accumulate percpu_counter_batch clusters in their local
3025  * counters. So we need to make sure we have free clusters more
3026  * than percpu_counter_batch  * nr_cpu_ids. Also add a window of 4 times.
3027  */
3028 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
3029 #else
3030 #define EXT4_FREECLUSTERS_WATERMARK 0
3031 #endif
3032 
3033 /* Update i_disksize. Requires i_mutex to avoid races with truncate */
ext4_update_i_disksize(struct inode * inode,loff_t newsize)3034 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
3035 {
3036 	WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
3037 		     !inode_is_locked(inode));
3038 	down_write(&EXT4_I(inode)->i_data_sem);
3039 	if (newsize > EXT4_I(inode)->i_disksize)
3040 		WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize);
3041 	up_write(&EXT4_I(inode)->i_data_sem);
3042 }
3043 
3044 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
ext4_update_inode_size(struct inode * inode,loff_t newsize)3045 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
3046 {
3047 	int changed = 0;
3048 
3049 	if (newsize > inode->i_size) {
3050 		i_size_write(inode, newsize);
3051 		changed = 1;
3052 	}
3053 	if (newsize > EXT4_I(inode)->i_disksize) {
3054 		ext4_update_i_disksize(inode, newsize);
3055 		changed |= 2;
3056 	}
3057 	return changed;
3058 }
3059 
3060 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
3061 				      loff_t len);
3062 
3063 struct ext4_group_info {
3064 	unsigned long   bb_state;
3065 	struct rb_root  bb_free_root;
3066 	ext4_grpblk_t	bb_first_free;	/* first free block */
3067 	ext4_grpblk_t	bb_free;	/* total free blocks */
3068 	ext4_grpblk_t	bb_fragments;	/* nr of freespace fragments */
3069 	ext4_grpblk_t	bb_largest_free_order;/* order of largest frag in BG */
3070 	struct          list_head bb_prealloc_list;
3071 #ifdef DOUBLE_CHECK
3072 	void            *bb_bitmap;
3073 #endif
3074 	struct rw_semaphore alloc_sem;
3075 	ext4_grpblk_t	bb_counters[];	/* Nr of free power-of-two-block
3076 					 * regions, index is order.
3077 					 * bb_counters[3] = 5 means
3078 					 * 5 free 8-block regions. */
3079 };
3080 
3081 #define EXT4_GROUP_INFO_NEED_INIT_BIT		0
3082 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT		1
3083 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT	2
3084 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT	3
3085 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT		\
3086 	(1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT)
3087 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT		\
3088 	(1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT)
3089 
3090 #define EXT4_MB_GRP_NEED_INIT(grp)	\
3091 	(test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
3092 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp)	\
3093 	(test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3094 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp)	\
3095 	(test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3096 
3097 #define EXT4_MB_GRP_WAS_TRIMMED(grp)	\
3098 	(test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3099 #define EXT4_MB_GRP_SET_TRIMMED(grp)	\
3100 	(set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3101 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp)	\
3102 	(clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3103 
3104 #define EXT4_MAX_CONTENTION		8
3105 #define EXT4_CONTENTION_THRESHOLD	2
3106 
ext4_group_lock_ptr(struct super_block * sb,ext4_group_t group)3107 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
3108 					      ext4_group_t group)
3109 {
3110 	return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
3111 }
3112 
3113 /*
3114  * Returns true if the filesystem is busy enough that attempts to
3115  * access the block group locks has run into contention.
3116  */
ext4_fs_is_busy(struct ext4_sb_info * sbi)3117 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
3118 {
3119 	return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
3120 }
3121 
ext4_lock_group(struct super_block * sb,ext4_group_t group)3122 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
3123 {
3124 	spinlock_t *lock = ext4_group_lock_ptr(sb, group);
3125 	if (spin_trylock(lock))
3126 		/*
3127 		 * We're able to grab the lock right away, so drop the
3128 		 * lock contention counter.
3129 		 */
3130 		atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
3131 	else {
3132 		/*
3133 		 * The lock is busy, so bump the contention counter,
3134 		 * and then wait on the spin lock.
3135 		 */
3136 		atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
3137 				  EXT4_MAX_CONTENTION);
3138 		spin_lock(lock);
3139 	}
3140 }
3141 
ext4_unlock_group(struct super_block * sb,ext4_group_t group)3142 static inline void ext4_unlock_group(struct super_block *sb,
3143 					ext4_group_t group)
3144 {
3145 	spin_unlock(ext4_group_lock_ptr(sb, group));
3146 }
3147 
3148 /*
3149  * Block validity checking
3150  */
3151 #define ext4_check_indirect_blockref(inode, bh)				\
3152 	ext4_check_blockref(__func__, __LINE__, inode,			\
3153 			    (__le32 *)(bh)->b_data,			\
3154 			    EXT4_ADDR_PER_BLOCK((inode)->i_sb))
3155 
3156 #define ext4_ind_check_inode(inode)					\
3157 	ext4_check_blockref(__func__, __LINE__, inode,			\
3158 			    EXT4_I(inode)->i_data,			\
3159 			    EXT4_NDIR_BLOCKS)
3160 
3161 /*
3162  * Inodes and files operations
3163  */
3164 
3165 /* dir.c */
3166 extern const struct file_operations ext4_dir_operations;
3167 
3168 #ifdef CONFIG_UNICODE
3169 extern const struct dentry_operations ext4_dentry_ops;
3170 #endif
3171 
3172 /* file.c */
3173 extern const struct inode_operations ext4_file_inode_operations;
3174 extern const struct file_operations ext4_file_operations;
3175 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
3176 
3177 /* inline.c */
3178 extern int ext4_get_max_inline_size(struct inode *inode);
3179 extern int ext4_find_inline_data_nolock(struct inode *inode);
3180 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
3181 				 unsigned int len);
3182 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
3183 
3184 extern int ext4_readpage_inline(struct inode *inode, struct page *page);
3185 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3186 					 struct inode *inode,
3187 					 loff_t pos, unsigned len,
3188 					 unsigned flags,
3189 					 struct page **pagep);
3190 extern int ext4_write_inline_data_end(struct inode *inode,
3191 				      loff_t pos, unsigned len,
3192 				      unsigned copied,
3193 				      struct page *page);
3194 extern struct buffer_head *
3195 ext4_journalled_write_inline_data(struct inode *inode,
3196 				  unsigned len,
3197 				  struct page *page);
3198 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
3199 					   struct inode *inode,
3200 					   loff_t pos, unsigned len,
3201 					   unsigned flags,
3202 					   struct page **pagep,
3203 					   void **fsdata);
3204 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
3205 					 unsigned len, unsigned copied,
3206 					 struct page *page);
3207 extern int ext4_try_add_inline_entry(handle_t *handle,
3208 				     struct ext4_filename *fname,
3209 				     struct inode *dir, struct inode *inode);
3210 extern int ext4_try_create_inline_dir(handle_t *handle,
3211 				      struct inode *parent,
3212 				      struct inode *inode);
3213 extern int ext4_read_inline_dir(struct file *filp,
3214 				struct dir_context *ctx,
3215 				int *has_inline_data);
3216 extern int ext4_inlinedir_to_tree(struct file *dir_file,
3217 				  struct inode *dir, ext4_lblk_t block,
3218 				  struct dx_hash_info *hinfo,
3219 				  __u32 start_hash, __u32 start_minor_hash,
3220 				  int *has_inline_data);
3221 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3222 					struct ext4_filename *fname,
3223 					struct ext4_dir_entry_2 **res_dir,
3224 					int *has_inline_data);
3225 extern int ext4_delete_inline_entry(handle_t *handle,
3226 				    struct inode *dir,
3227 				    struct ext4_dir_entry_2 *de_del,
3228 				    struct buffer_head *bh,
3229 				    int *has_inline_data);
3230 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3231 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3232 					struct ext4_dir_entry_2 **parent_de,
3233 					int *retval);
3234 extern int ext4_inline_data_fiemap(struct inode *inode,
3235 				   struct fiemap_extent_info *fieinfo,
3236 				   int *has_inline, __u64 start, __u64 len);
3237 
3238 struct iomap;
3239 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap);
3240 
3241 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3242 
3243 extern int ext4_convert_inline_data(struct inode *inode);
3244 
ext4_has_inline_data(struct inode * inode)3245 static inline int ext4_has_inline_data(struct inode *inode)
3246 {
3247 	return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3248 	       EXT4_I(inode)->i_inline_off;
3249 }
3250 
3251 /* namei.c */
3252 extern const struct inode_operations ext4_dir_inode_operations;
3253 extern const struct inode_operations ext4_special_inode_operations;
3254 extern struct dentry *ext4_get_parent(struct dentry *child);
3255 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3256 				 struct ext4_dir_entry_2 *de,
3257 				 int blocksize, int csum_size,
3258 				 unsigned int parent_ino, int dotdot_real_len);
3259 extern void ext4_initialize_dirent_tail(struct buffer_head *bh,
3260 					unsigned int blocksize);
3261 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode,
3262 				      struct buffer_head *bh);
3263 
3264 #define S_SHIFT 12
3265 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = {
3266 	[S_IFREG >> S_SHIFT]	= EXT4_FT_REG_FILE,
3267 	[S_IFDIR >> S_SHIFT]	= EXT4_FT_DIR,
3268 	[S_IFCHR >> S_SHIFT]	= EXT4_FT_CHRDEV,
3269 	[S_IFBLK >> S_SHIFT]	= EXT4_FT_BLKDEV,
3270 	[S_IFIFO >> S_SHIFT]	= EXT4_FT_FIFO,
3271 	[S_IFSOCK >> S_SHIFT]	= EXT4_FT_SOCK,
3272 	[S_IFLNK >> S_SHIFT]	= EXT4_FT_SYMLINK,
3273 };
3274 
ext4_set_de_type(struct super_block * sb,struct ext4_dir_entry_2 * de,umode_t mode)3275 static inline void ext4_set_de_type(struct super_block *sb,
3276 				struct ext4_dir_entry_2 *de,
3277 				umode_t mode) {
3278 	if (ext4_has_feature_filetype(sb))
3279 		de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3280 }
3281 
3282 /* readpages.c */
3283 extern int ext4_mpage_readpages(struct address_space *mapping,
3284 				struct list_head *pages, struct page *page,
3285 				unsigned nr_pages, bool is_readahead);
3286 extern int __init ext4_init_post_read_processing(void);
3287 extern void ext4_exit_post_read_processing(void);
3288 
3289 /* symlink.c */
3290 extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3291 extern const struct inode_operations ext4_symlink_inode_operations;
3292 extern const struct inode_operations ext4_fast_symlink_inode_operations;
3293 
3294 /* sysfs.c */
3295 extern int ext4_register_sysfs(struct super_block *sb);
3296 extern void ext4_unregister_sysfs(struct super_block *sb);
3297 extern int __init ext4_init_sysfs(void);
3298 extern void ext4_exit_sysfs(void);
3299 
3300 /* block_validity */
3301 extern void ext4_release_system_zone(struct super_block *sb);
3302 extern int ext4_setup_system_zone(struct super_block *sb);
3303 extern int __init ext4_init_system_zone(void);
3304 extern void ext4_exit_system_zone(void);
3305 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
3306 				 ext4_fsblk_t start_blk,
3307 				 unsigned int count);
3308 extern int ext4_check_blockref(const char *, unsigned int,
3309 			       struct inode *, __le32 *, unsigned int);
3310 
3311 /* extents.c */
3312 struct ext4_ext_path;
3313 struct ext4_extent;
3314 
3315 /*
3316  * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3317  * __le32.
3318  */
3319 #define EXT_MAX_BLOCKS	0xffffffff
3320 
3321 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
3322 extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
3323 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3324 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3325 			       struct ext4_map_blocks *map, int flags);
3326 extern int ext4_ext_truncate(handle_t *, struct inode *);
3327 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3328 				 ext4_lblk_t end);
3329 extern void ext4_ext_init(struct super_block *);
3330 extern void ext4_ext_release(struct super_block *);
3331 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3332 			  loff_t len);
3333 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3334 					  loff_t offset, ssize_t len);
3335 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3336 			   struct ext4_map_blocks *map, int flags);
3337 extern int ext4_ext_calc_metadata_amount(struct inode *inode,
3338 					 ext4_lblk_t lblocks);
3339 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3340 						   int num,
3341 						   struct ext4_ext_path *path);
3342 extern int ext4_can_extents_be_merged(struct inode *inode,
3343 				      struct ext4_extent *ex1,
3344 				      struct ext4_extent *ex2);
3345 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3346 				  struct ext4_ext_path **,
3347 				  struct ext4_extent *, int);
3348 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3349 					      struct ext4_ext_path **,
3350 					      int flags);
3351 extern void ext4_ext_drop_refs(struct ext4_ext_path *);
3352 extern int ext4_ext_check_inode(struct inode *inode);
3353 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3354 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3355 			__u64 start, __u64 len);
3356 extern int ext4_get_es_cache(struct inode *inode,
3357 			     struct fiemap_extent_info *fieinfo,
3358 			     __u64 start, __u64 len);
3359 extern int ext4_ext_precache(struct inode *inode);
3360 extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len);
3361 extern int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len);
3362 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3363 				struct inode *inode2, ext4_lblk_t lblk1,
3364 			     ext4_lblk_t lblk2,  ext4_lblk_t count,
3365 			     int mark_unwritten,int *err);
3366 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu);
3367 
3368 /* move_extent.c */
3369 extern void ext4_double_down_write_data_sem(struct inode *first,
3370 					    struct inode *second);
3371 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3372 					  struct inode *donor_inode);
3373 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3374 			     __u64 start_orig, __u64 start_donor,
3375 			     __u64 len, __u64 *moved_len);
3376 
3377 /* page-io.c */
3378 extern int __init ext4_init_pageio(void);
3379 extern void ext4_exit_pageio(void);
3380 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3381 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3382 extern int ext4_put_io_end(ext4_io_end_t *io_end);
3383 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3384 extern void ext4_io_submit_init(struct ext4_io_submit *io,
3385 				struct writeback_control *wbc);
3386 extern void ext4_end_io_rsv_work(struct work_struct *work);
3387 extern void ext4_io_submit(struct ext4_io_submit *io);
3388 extern int ext4_bio_write_page(struct ext4_io_submit *io,
3389 			       struct page *page,
3390 			       int len,
3391 			       struct writeback_control *wbc,
3392 			       bool keep_towrite);
3393 
3394 /* mmp.c */
3395 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3396 
3397 /* verity.c */
3398 extern const struct fsverity_operations ext4_verityops;
3399 
3400 /*
3401  * Add new method to test whether block and inode bitmaps are properly
3402  * initialized. With uninit_bg reading the block from disk is not enough
3403  * to mark the bitmap uptodate. We need to also zero-out the bitmap
3404  */
3405 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3406 
bitmap_uptodate(struct buffer_head * bh)3407 static inline int bitmap_uptodate(struct buffer_head *bh)
3408 {
3409 	return (buffer_uptodate(bh) &&
3410 			test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3411 }
set_bitmap_uptodate(struct buffer_head * bh)3412 static inline void set_bitmap_uptodate(struct buffer_head *bh)
3413 {
3414 	set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3415 }
3416 
3417 #define in_range(b, first, len)	((b) >= (first) && (b) <= (first) + (len) - 1)
3418 
3419 /* For ioend & aio unwritten conversion wait queues */
3420 #define EXT4_WQ_HASH_SZ		37
3421 #define ext4_ioend_wq(v)   (&ext4__ioend_wq[((unsigned long)(v)) %\
3422 					    EXT4_WQ_HASH_SZ])
3423 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3424 
3425 extern int ext4_resize_begin(struct super_block *sb);
3426 extern void ext4_resize_end(struct super_block *sb);
3427 
ext4_set_io_unwritten_flag(struct inode * inode,struct ext4_io_end * io_end)3428 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
3429 					      struct ext4_io_end *io_end)
3430 {
3431 	if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
3432 		io_end->flag |= EXT4_IO_END_UNWRITTEN;
3433 		atomic_inc(&EXT4_I(inode)->i_unwritten);
3434 	}
3435 }
3436 
ext4_clear_io_unwritten_flag(ext4_io_end_t * io_end)3437 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3438 {
3439 	struct inode *inode = io_end->inode;
3440 
3441 	if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
3442 		io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3443 		/* Wake up anyone waiting on unwritten extent conversion */
3444 		if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
3445 			wake_up_all(ext4_ioend_wq(inode));
3446 	}
3447 }
3448 
3449 extern const struct iomap_ops ext4_iomap_ops;
3450 
ext4_buffer_uptodate(struct buffer_head * bh)3451 static inline int ext4_buffer_uptodate(struct buffer_head *bh)
3452 {
3453 	/*
3454 	 * If the buffer has the write error flag, we have failed
3455 	 * to write out data in the block.  In this  case, we don't
3456 	 * have to read the block because we may read the old data
3457 	 * successfully.
3458 	 */
3459 	if (!buffer_uptodate(bh) && buffer_write_io_error(bh))
3460 		set_buffer_uptodate(bh);
3461 	return buffer_uptodate(bh);
3462 }
3463 
3464 #endif	/* __KERNEL__ */
3465 
3466 #define EFSBADCRC	EBADMSG		/* Bad CRC detected */
3467 #define EFSCORRUPTED	EUCLEAN		/* Filesystem is corrupted */
3468 
3469 #endif	/* _EXT4_H */
3470