1 // SPDX-License-Identifier: GPL-2.0
2 /*
3  *  ext4.h
4  *
5  * Copyright (C) 1992, 1993, 1994, 1995
6  * Remy Card (card@masi.ibp.fr)
7  * Laboratoire MASI - Institut Blaise Pascal
8  * Universite Pierre et Marie Curie (Paris VI)
9  *
10  *  from
11  *
12  *  linux/include/linux/minix_fs.h
13  *
14  *  Copyright (C) 1991, 1992  Linus Torvalds
15  */
16 
17 #ifndef _EXT4_H
18 #define _EXT4_H
19 
20 #include <linux/types.h>
21 #include <linux/blkdev.h>
22 #include <linux/magic.h>
23 #include <linux/jbd2.h>
24 #include <linux/quota.h>
25 #include <linux/rwsem.h>
26 #include <linux/rbtree.h>
27 #include <linux/seqlock.h>
28 #include <linux/mutex.h>
29 #include <linux/timer.h>
30 #include <linux/wait.h>
31 #include <linux/sched/signal.h>
32 #include <linux/blockgroup_lock.h>
33 #include <linux/percpu_counter.h>
34 #include <linux/ratelimit.h>
35 #include <crypto/hash.h>
36 #include <linux/falloc.h>
37 #include <linux/percpu-rwsem.h>
38 #include <linux/fiemap.h>
39 #ifdef __KERNEL__
40 #include <linux/compat.h>
41 #endif
42 
43 #include <linux/fscrypt.h>
44 #include <linux/fsverity.h>
45 
46 #include <linux/compiler.h>
47 
48 /*
49  * The fourth extended filesystem constants/structures
50  */
51 
52 /*
53  * with AGGRESSIVE_CHECK allocator runs consistency checks over
54  * structures. these checks slow things down a lot
55  */
56 #define AGGRESSIVE_CHECK__
57 
58 /*
59  * with DOUBLE_CHECK defined mballoc creates persistent in-core
60  * bitmaps, maintains and uses them to check for double allocations
61  */
62 #define DOUBLE_CHECK__
63 
64 /*
65  * Define EXT4FS_DEBUG to produce debug messages
66  */
67 #undef EXT4FS_DEBUG
68 
69 /*
70  * Debug code
71  */
72 #ifdef EXT4FS_DEBUG
73 #define ext4_debug(f, a...)						\
74 	do {								\
75 		printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:",	\
76 			__FILE__, __LINE__, __func__);			\
77 		printk(KERN_DEBUG f, ## a);				\
78 	} while (0)
79 #else
80 #define ext4_debug(fmt, ...)	no_printk(fmt, ##__VA_ARGS__)
81 #endif
82 
83  /*
84   * Turn on EXT_DEBUG to enable ext4_ext_show_path/leaf/move in extents.c
85   */
86 #define EXT_DEBUG__
87 
88 /*
89  * Dynamic printk for controlled extents debugging.
90  */
91 #ifdef CONFIG_EXT4_DEBUG
92 #define ext_debug(ino, fmt, ...)					\
93 	pr_debug("[%s/%d] EXT4-fs (%s): ino %lu: (%s, %d): %s:" fmt,	\
94 		 current->comm, task_pid_nr(current),			\
95 		 ino->i_sb->s_id, ino->i_ino, __FILE__, __LINE__,	\
96 		 __func__, ##__VA_ARGS__)
97 #else
98 #define ext_debug(ino, fmt, ...)	no_printk(fmt, ##__VA_ARGS__)
99 #endif
100 
101 /* data type for block offset of block group */
102 typedef int ext4_grpblk_t;
103 
104 /* data type for filesystem-wide blocks number */
105 typedef unsigned long long ext4_fsblk_t;
106 
107 /* data type for file logical block number */
108 typedef __u32 ext4_lblk_t;
109 
110 /* data type for block group number */
111 typedef unsigned int ext4_group_t;
112 
113 enum SHIFT_DIRECTION {
114 	SHIFT_LEFT = 0,
115 	SHIFT_RIGHT,
116 };
117 
118 /*
119  * Flags used in mballoc's allocation_context flags field.
120  *
121  * Also used to show what's going on for debugging purposes when the
122  * flag field is exported via the traceport interface
123  */
124 
125 /* prefer goal again. length */
126 #define EXT4_MB_HINT_MERGE		0x0001
127 /* blocks already reserved */
128 #define EXT4_MB_HINT_RESERVED		0x0002
129 /* metadata is being allocated */
130 #define EXT4_MB_HINT_METADATA		0x0004
131 /* first blocks in the file */
132 #define EXT4_MB_HINT_FIRST		0x0008
133 /* search for the best chunk */
134 #define EXT4_MB_HINT_BEST		0x0010
135 /* data is being allocated */
136 #define EXT4_MB_HINT_DATA		0x0020
137 /* don't preallocate (for tails) */
138 #define EXT4_MB_HINT_NOPREALLOC		0x0040
139 /* allocate for locality group */
140 #define EXT4_MB_HINT_GROUP_ALLOC	0x0080
141 /* allocate goal blocks or none */
142 #define EXT4_MB_HINT_GOAL_ONLY		0x0100
143 /* goal is meaningful */
144 #define EXT4_MB_HINT_TRY_GOAL		0x0200
145 /* blocks already pre-reserved by delayed allocation */
146 #define EXT4_MB_DELALLOC_RESERVED	0x0400
147 /* We are doing stream allocation */
148 #define EXT4_MB_STREAM_ALLOC		0x0800
149 /* Use reserved root blocks if needed */
150 #define EXT4_MB_USE_ROOT_BLOCKS		0x1000
151 /* Use blocks from reserved pool */
152 #define EXT4_MB_USE_RESERVED		0x2000
153 /* Do strict check for free blocks while retrying block allocation */
154 #define EXT4_MB_STRICT_CHECK		0x4000
155 
156 struct ext4_allocation_request {
157 	/* target inode for block we're allocating */
158 	struct inode *inode;
159 	/* how many blocks we want to allocate */
160 	unsigned int len;
161 	/* logical block in target inode */
162 	ext4_lblk_t logical;
163 	/* the closest logical allocated block to the left */
164 	ext4_lblk_t lleft;
165 	/* the closest logical allocated block to the right */
166 	ext4_lblk_t lright;
167 	/* phys. target (a hint) */
168 	ext4_fsblk_t goal;
169 	/* phys. block for the closest logical allocated block to the left */
170 	ext4_fsblk_t pleft;
171 	/* phys. block for the closest logical allocated block to the right */
172 	ext4_fsblk_t pright;
173 	/* flags. see above EXT4_MB_HINT_* */
174 	unsigned int flags;
175 };
176 
177 /*
178  * Logical to physical block mapping, used by ext4_map_blocks()
179  *
180  * This structure is used to pass requests into ext4_map_blocks() as
181  * well as to store the information returned by ext4_map_blocks().  It
182  * takes less room on the stack than a struct buffer_head.
183  */
184 #define EXT4_MAP_NEW		BIT(BH_New)
185 #define EXT4_MAP_MAPPED		BIT(BH_Mapped)
186 #define EXT4_MAP_UNWRITTEN	BIT(BH_Unwritten)
187 #define EXT4_MAP_BOUNDARY	BIT(BH_Boundary)
188 #define EXT4_MAP_FLAGS		(EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
189 				 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
190 
191 struct ext4_map_blocks {
192 	ext4_fsblk_t m_pblk;
193 	ext4_lblk_t m_lblk;
194 	unsigned int m_len;
195 	unsigned int m_flags;
196 };
197 
198 /*
199  * Block validity checking, system zone rbtree.
200  */
201 struct ext4_system_blocks {
202 	struct rb_root root;
203 	struct rcu_head rcu;
204 };
205 
206 /*
207  * Flags for ext4_io_end->flags
208  */
209 #define	EXT4_IO_END_UNWRITTEN	0x0001
210 
211 struct ext4_io_end_vec {
212 	struct list_head list;		/* list of io_end_vec */
213 	loff_t offset;			/* offset in the file */
214 	ssize_t size;			/* size of the extent */
215 };
216 
217 /*
218  * For converting unwritten extents on a work queue. 'handle' is used for
219  * buffered writeback.
220  */
221 typedef struct ext4_io_end {
222 	struct list_head	list;		/* per-file finished IO list */
223 	handle_t		*handle;	/* handle reserved for extent
224 						 * conversion */
225 	struct inode		*inode;		/* file being written to */
226 	struct bio		*bio;		/* Linked list of completed
227 						 * bios covering the extent */
228 	unsigned int		flag;		/* unwritten or not */
229 	atomic_t		count;		/* reference counter */
230 	struct list_head	list_vec;	/* list of ext4_io_end_vec */
231 } ext4_io_end_t;
232 
233 struct ext4_io_submit {
234 	struct writeback_control *io_wbc;
235 	struct bio		*io_bio;
236 	ext4_io_end_t		*io_end;
237 	sector_t		io_next_block;
238 };
239 
240 /*
241  * Special inodes numbers
242  */
243 #define	EXT4_BAD_INO		 1	/* Bad blocks inode */
244 #define EXT4_ROOT_INO		 2	/* Root inode */
245 #define EXT4_USR_QUOTA_INO	 3	/* User quota inode */
246 #define EXT4_GRP_QUOTA_INO	 4	/* Group quota inode */
247 #define EXT4_BOOT_LOADER_INO	 5	/* Boot loader inode */
248 #define EXT4_UNDEL_DIR_INO	 6	/* Undelete directory inode */
249 #define EXT4_RESIZE_INO		 7	/* Reserved group descriptors inode */
250 #define EXT4_JOURNAL_INO	 8	/* Journal inode */
251 
252 /* First non-reserved inode for old ext4 filesystems */
253 #define EXT4_GOOD_OLD_FIRST_INO	11
254 
255 /*
256  * Maximal count of links to a file
257  */
258 #define EXT4_LINK_MAX		65000
259 
260 /*
261  * Macro-instructions used to manage several block sizes
262  */
263 #define EXT4_MIN_BLOCK_SIZE		1024
264 #define	EXT4_MAX_BLOCK_SIZE		65536
265 #define EXT4_MIN_BLOCK_LOG_SIZE		10
266 #define EXT4_MAX_BLOCK_LOG_SIZE		16
267 #define EXT4_MAX_CLUSTER_LOG_SIZE	30
268 #ifdef __KERNEL__
269 # define EXT4_BLOCK_SIZE(s)		((s)->s_blocksize)
270 #else
271 # define EXT4_BLOCK_SIZE(s)		(EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
272 #endif
273 #define	EXT4_ADDR_PER_BLOCK(s)		(EXT4_BLOCK_SIZE(s) / sizeof(__u32))
274 #define EXT4_CLUSTER_SIZE(s)		(EXT4_BLOCK_SIZE(s) << \
275 					 EXT4_SB(s)->s_cluster_bits)
276 #ifdef __KERNEL__
277 # define EXT4_BLOCK_SIZE_BITS(s)	((s)->s_blocksize_bits)
278 # define EXT4_CLUSTER_BITS(s)		(EXT4_SB(s)->s_cluster_bits)
279 #else
280 # define EXT4_BLOCK_SIZE_BITS(s)	((s)->s_log_block_size + 10)
281 #endif
282 #ifdef __KERNEL__
283 #define	EXT4_ADDR_PER_BLOCK_BITS(s)	(EXT4_SB(s)->s_addr_per_block_bits)
284 #define EXT4_INODE_SIZE(s)		(EXT4_SB(s)->s_inode_size)
285 #define EXT4_FIRST_INO(s)		(EXT4_SB(s)->s_first_ino)
286 #else
287 #define EXT4_INODE_SIZE(s)	(((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
288 				 EXT4_GOOD_OLD_INODE_SIZE : \
289 				 (s)->s_inode_size)
290 #define EXT4_FIRST_INO(s)	(((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
291 				 EXT4_GOOD_OLD_FIRST_INO : \
292 				 (s)->s_first_ino)
293 #endif
294 #define EXT4_BLOCK_ALIGN(size, blkbits)		ALIGN((size), (1 << (blkbits)))
295 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \
296 	((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \
297 								  blkbits))
298 
299 /* Translate a block number to a cluster number */
300 #define EXT4_B2C(sbi, blk)	((blk) >> (sbi)->s_cluster_bits)
301 /* Translate a cluster number to a block number */
302 #define EXT4_C2B(sbi, cluster)	((cluster) << (sbi)->s_cluster_bits)
303 /* Translate # of blks to # of clusters */
304 #define EXT4_NUM_B2C(sbi, blks)	(((blks) + (sbi)->s_cluster_ratio - 1) >> \
305 				 (sbi)->s_cluster_bits)
306 /* Mask out the low bits to get the starting block of the cluster */
307 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) &				\
308 				  ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
309 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) &				\
310 				  ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
311 /* Fill in the low bits to get the last block of the cluster */
312 #define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) |				\
313 				    ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1))
314 /* Get the cluster offset */
315 #define EXT4_PBLK_COFF(s, pblk) ((pblk) &				\
316 				 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
317 #define EXT4_LBLK_COFF(s, lblk) ((lblk) &				\
318 				 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
319 
320 /*
321  * Structure of a blocks group descriptor
322  */
323 struct ext4_group_desc
324 {
325 	__le32	bg_block_bitmap_lo;	/* Blocks bitmap block */
326 	__le32	bg_inode_bitmap_lo;	/* Inodes bitmap block */
327 	__le32	bg_inode_table_lo;	/* Inodes table block */
328 	__le16	bg_free_blocks_count_lo;/* Free blocks count */
329 	__le16	bg_free_inodes_count_lo;/* Free inodes count */
330 	__le16	bg_used_dirs_count_lo;	/* Directories count */
331 	__le16	bg_flags;		/* EXT4_BG_flags (INODE_UNINIT, etc) */
332 	__le32  bg_exclude_bitmap_lo;   /* Exclude bitmap for snapshots */
333 	__le16  bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
334 	__le16  bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
335 	__le16  bg_itable_unused_lo;	/* Unused inodes count */
336 	__le16  bg_checksum;		/* crc16(sb_uuid+group+desc) */
337 	__le32	bg_block_bitmap_hi;	/* Blocks bitmap block MSB */
338 	__le32	bg_inode_bitmap_hi;	/* Inodes bitmap block MSB */
339 	__le32	bg_inode_table_hi;	/* Inodes table block MSB */
340 	__le16	bg_free_blocks_count_hi;/* Free blocks count MSB */
341 	__le16	bg_free_inodes_count_hi;/* Free inodes count MSB */
342 	__le16	bg_used_dirs_count_hi;	/* Directories count MSB */
343 	__le16  bg_itable_unused_hi;    /* Unused inodes count MSB */
344 	__le32  bg_exclude_bitmap_hi;   /* Exclude bitmap block MSB */
345 	__le16  bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
346 	__le16  bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
347 	__u32   bg_reserved;
348 };
349 
350 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END	\
351 	(offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
352 	 sizeof(__le16))
353 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END	\
354 	(offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
355 	 sizeof(__le16))
356 
357 /*
358  * Structure of a flex block group info
359  */
360 
361 struct flex_groups {
362 	atomic64_t	free_clusters;
363 	atomic_t	free_inodes;
364 	atomic_t	used_dirs;
365 };
366 
367 #define EXT4_BG_INODE_UNINIT	0x0001 /* Inode table/bitmap not in use */
368 #define EXT4_BG_BLOCK_UNINIT	0x0002 /* Block bitmap not in use */
369 #define EXT4_BG_INODE_ZEROED	0x0004 /* On-disk itable initialized to zero */
370 
371 /*
372  * Macro-instructions used to manage group descriptors
373  */
374 #define EXT4_MIN_DESC_SIZE		32
375 #define EXT4_MIN_DESC_SIZE_64BIT	64
376 #define	EXT4_MAX_DESC_SIZE		EXT4_MIN_BLOCK_SIZE
377 #define EXT4_DESC_SIZE(s)		(EXT4_SB(s)->s_desc_size)
378 #ifdef __KERNEL__
379 # define EXT4_BLOCKS_PER_GROUP(s)	(EXT4_SB(s)->s_blocks_per_group)
380 # define EXT4_CLUSTERS_PER_GROUP(s)	(EXT4_SB(s)->s_clusters_per_group)
381 # define EXT4_DESC_PER_BLOCK(s)		(EXT4_SB(s)->s_desc_per_block)
382 # define EXT4_INODES_PER_GROUP(s)	(EXT4_SB(s)->s_inodes_per_group)
383 # define EXT4_DESC_PER_BLOCK_BITS(s)	(EXT4_SB(s)->s_desc_per_block_bits)
384 #else
385 # define EXT4_BLOCKS_PER_GROUP(s)	((s)->s_blocks_per_group)
386 # define EXT4_DESC_PER_BLOCK(s)		(EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
387 # define EXT4_INODES_PER_GROUP(s)	((s)->s_inodes_per_group)
388 #endif
389 
390 /*
391  * Constants relative to the data blocks
392  */
393 #define	EXT4_NDIR_BLOCKS		12
394 #define	EXT4_IND_BLOCK			EXT4_NDIR_BLOCKS
395 #define	EXT4_DIND_BLOCK			(EXT4_IND_BLOCK + 1)
396 #define	EXT4_TIND_BLOCK			(EXT4_DIND_BLOCK + 1)
397 #define	EXT4_N_BLOCKS			(EXT4_TIND_BLOCK + 1)
398 
399 /*
400  * Inode flags
401  */
402 #define	EXT4_SECRM_FL			0x00000001 /* Secure deletion */
403 #define	EXT4_UNRM_FL			0x00000002 /* Undelete */
404 #define	EXT4_COMPR_FL			0x00000004 /* Compress file */
405 #define EXT4_SYNC_FL			0x00000008 /* Synchronous updates */
406 #define EXT4_IMMUTABLE_FL		0x00000010 /* Immutable file */
407 #define EXT4_APPEND_FL			0x00000020 /* writes to file may only append */
408 #define EXT4_NODUMP_FL			0x00000040 /* do not dump file */
409 #define EXT4_NOATIME_FL			0x00000080 /* do not update atime */
410 /* Reserved for compression usage... */
411 #define EXT4_DIRTY_FL			0x00000100
412 #define EXT4_COMPRBLK_FL		0x00000200 /* One or more compressed clusters */
413 #define EXT4_NOCOMPR_FL			0x00000400 /* Don't compress */
414 	/* nb: was previously EXT2_ECOMPR_FL */
415 #define EXT4_ENCRYPT_FL			0x00000800 /* encrypted file */
416 /* End compression flags --- maybe not all used */
417 #define EXT4_INDEX_FL			0x00001000 /* hash-indexed directory */
418 #define EXT4_IMAGIC_FL			0x00002000 /* AFS directory */
419 #define EXT4_JOURNAL_DATA_FL		0x00004000 /* file data should be journaled */
420 #define EXT4_NOTAIL_FL			0x00008000 /* file tail should not be merged */
421 #define EXT4_DIRSYNC_FL			0x00010000 /* dirsync behaviour (directories only) */
422 #define EXT4_TOPDIR_FL			0x00020000 /* Top of directory hierarchies*/
423 #define EXT4_HUGE_FILE_FL               0x00040000 /* Set to each huge file */
424 #define EXT4_EXTENTS_FL			0x00080000 /* Inode uses extents */
425 #define EXT4_VERITY_FL			0x00100000 /* Verity protected inode */
426 #define EXT4_EA_INODE_FL	        0x00200000 /* Inode used for large EA */
427 /* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */
428 
429 #define EXT4_DAX_FL			0x02000000 /* Inode is DAX */
430 
431 #define EXT4_INLINE_DATA_FL		0x10000000 /* Inode has inline data. */
432 #define EXT4_PROJINHERIT_FL		0x20000000 /* Create with parents projid */
433 #define EXT4_CASEFOLD_FL		0x40000000 /* Casefolded directory */
434 #define EXT4_RESERVED_FL		0x80000000 /* reserved for ext4 lib */
435 
436 /* User modifiable flags */
437 #define EXT4_FL_USER_MODIFIABLE		(EXT4_SECRM_FL | \
438 					 EXT4_UNRM_FL | \
439 					 EXT4_COMPR_FL | \
440 					 EXT4_SYNC_FL | \
441 					 EXT4_IMMUTABLE_FL | \
442 					 EXT4_APPEND_FL | \
443 					 EXT4_NODUMP_FL | \
444 					 EXT4_NOATIME_FL | \
445 					 EXT4_JOURNAL_DATA_FL | \
446 					 EXT4_NOTAIL_FL | \
447 					 EXT4_DIRSYNC_FL | \
448 					 EXT4_TOPDIR_FL | \
449 					 EXT4_EXTENTS_FL | \
450 					 0x00400000 /* EXT4_EOFBLOCKS_FL */ | \
451 					 EXT4_DAX_FL | \
452 					 EXT4_PROJINHERIT_FL | \
453 					 EXT4_CASEFOLD_FL)
454 
455 /* User visible flags */
456 #define EXT4_FL_USER_VISIBLE		(EXT4_FL_USER_MODIFIABLE | \
457 					 EXT4_DIRTY_FL | \
458 					 EXT4_COMPRBLK_FL | \
459 					 EXT4_NOCOMPR_FL | \
460 					 EXT4_ENCRYPT_FL | \
461 					 EXT4_INDEX_FL | \
462 					 EXT4_VERITY_FL | \
463 					 EXT4_INLINE_DATA_FL)
464 
465 /* Flags we can manipulate with through FS_IOC_FSSETXATTR */
466 #define EXT4_FL_XFLAG_VISIBLE		(EXT4_SYNC_FL | \
467 					 EXT4_IMMUTABLE_FL | \
468 					 EXT4_APPEND_FL | \
469 					 EXT4_NODUMP_FL | \
470 					 EXT4_NOATIME_FL | \
471 					 EXT4_PROJINHERIT_FL | \
472 					 EXT4_DAX_FL)
473 
474 /* Flags that should be inherited by new inodes from their parent. */
475 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
476 			   EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
477 			   EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
478 			   EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
479 			   EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL |\
480 			   EXT4_DAX_FL)
481 
482 /* Flags that are appropriate for regular files (all but dir-specific ones). */
483 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\
484 			   EXT4_PROJINHERIT_FL))
485 
486 /* Flags that are appropriate for non-directories/regular files. */
487 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
488 
489 /* The only flags that should be swapped */
490 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL)
491 
492 /* Flags which are mutually exclusive to DAX */
493 #define EXT4_DAX_MUT_EXCL (EXT4_VERITY_FL | EXT4_ENCRYPT_FL |\
494 			   EXT4_JOURNAL_DATA_FL | EXT4_INLINE_DATA_FL)
495 
496 /* Mask out flags that are inappropriate for the given type of inode. */
ext4_mask_flags(umode_t mode,__u32 flags)497 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
498 {
499 	if (S_ISDIR(mode))
500 		return flags;
501 	else if (S_ISREG(mode))
502 		return flags & EXT4_REG_FLMASK;
503 	else
504 		return flags & EXT4_OTHER_FLMASK;
505 }
506 
507 /*
508  * Inode flags used for atomic set/get
509  */
510 enum {
511 	EXT4_INODE_SECRM	= 0,	/* Secure deletion */
512 	EXT4_INODE_UNRM		= 1,	/* Undelete */
513 	EXT4_INODE_COMPR	= 2,	/* Compress file */
514 	EXT4_INODE_SYNC		= 3,	/* Synchronous updates */
515 	EXT4_INODE_IMMUTABLE	= 4,	/* Immutable file */
516 	EXT4_INODE_APPEND	= 5,	/* writes to file may only append */
517 	EXT4_INODE_NODUMP	= 6,	/* do not dump file */
518 	EXT4_INODE_NOATIME	= 7,	/* do not update atime */
519 /* Reserved for compression usage... */
520 	EXT4_INODE_DIRTY	= 8,
521 	EXT4_INODE_COMPRBLK	= 9,	/* One or more compressed clusters */
522 	EXT4_INODE_NOCOMPR	= 10,	/* Don't compress */
523 	EXT4_INODE_ENCRYPT	= 11,	/* Encrypted file */
524 /* End compression flags --- maybe not all used */
525 	EXT4_INODE_INDEX	= 12,	/* hash-indexed directory */
526 	EXT4_INODE_IMAGIC	= 13,	/* AFS directory */
527 	EXT4_INODE_JOURNAL_DATA	= 14,	/* file data should be journaled */
528 	EXT4_INODE_NOTAIL	= 15,	/* file tail should not be merged */
529 	EXT4_INODE_DIRSYNC	= 16,	/* dirsync behaviour (directories only) */
530 	EXT4_INODE_TOPDIR	= 17,	/* Top of directory hierarchies*/
531 	EXT4_INODE_HUGE_FILE	= 18,	/* Set to each huge file */
532 	EXT4_INODE_EXTENTS	= 19,	/* Inode uses extents */
533 	EXT4_INODE_VERITY	= 20,	/* Verity protected inode */
534 	EXT4_INODE_EA_INODE	= 21,	/* Inode used for large EA */
535 /* 22 was formerly EXT4_INODE_EOFBLOCKS */
536 	EXT4_INODE_DAX		= 25,	/* Inode is DAX */
537 	EXT4_INODE_INLINE_DATA	= 28,	/* Data in inode. */
538 	EXT4_INODE_PROJINHERIT	= 29,	/* Create with parents projid */
539 	EXT4_INODE_CASEFOLD	= 30,	/* Casefolded directory */
540 	EXT4_INODE_RESERVED	= 31,	/* reserved for ext4 lib */
541 };
542 
543 /*
544  * Since it's pretty easy to mix up bit numbers and hex values, we use a
545  * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
546  * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
547  * any extra space in the compiled kernel image, otherwise, the build will fail.
548  * It's important that these values are the same, since we are using
549  * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
550  * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
551  * values found in ext2, ext3 and ext4 filesystems, and of course the values
552  * defined in e2fsprogs.
553  *
554  * It's not paranoia if the Murphy's Law really *is* out to get you.  :-)
555  */
556 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
557 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
558 
ext4_check_flag_values(void)559 static inline void ext4_check_flag_values(void)
560 {
561 	CHECK_FLAG_VALUE(SECRM);
562 	CHECK_FLAG_VALUE(UNRM);
563 	CHECK_FLAG_VALUE(COMPR);
564 	CHECK_FLAG_VALUE(SYNC);
565 	CHECK_FLAG_VALUE(IMMUTABLE);
566 	CHECK_FLAG_VALUE(APPEND);
567 	CHECK_FLAG_VALUE(NODUMP);
568 	CHECK_FLAG_VALUE(NOATIME);
569 	CHECK_FLAG_VALUE(DIRTY);
570 	CHECK_FLAG_VALUE(COMPRBLK);
571 	CHECK_FLAG_VALUE(NOCOMPR);
572 	CHECK_FLAG_VALUE(ENCRYPT);
573 	CHECK_FLAG_VALUE(INDEX);
574 	CHECK_FLAG_VALUE(IMAGIC);
575 	CHECK_FLAG_VALUE(JOURNAL_DATA);
576 	CHECK_FLAG_VALUE(NOTAIL);
577 	CHECK_FLAG_VALUE(DIRSYNC);
578 	CHECK_FLAG_VALUE(TOPDIR);
579 	CHECK_FLAG_VALUE(HUGE_FILE);
580 	CHECK_FLAG_VALUE(EXTENTS);
581 	CHECK_FLAG_VALUE(VERITY);
582 	CHECK_FLAG_VALUE(EA_INODE);
583 	CHECK_FLAG_VALUE(INLINE_DATA);
584 	CHECK_FLAG_VALUE(PROJINHERIT);
585 	CHECK_FLAG_VALUE(CASEFOLD);
586 	CHECK_FLAG_VALUE(RESERVED);
587 }
588 
589 /* Used to pass group descriptor data when online resize is done */
590 struct ext4_new_group_input {
591 	__u32 group;		/* Group number for this data */
592 	__u64 block_bitmap;	/* Absolute block number of block bitmap */
593 	__u64 inode_bitmap;	/* Absolute block number of inode bitmap */
594 	__u64 inode_table;	/* Absolute block number of inode table start */
595 	__u32 blocks_count;	/* Total number of blocks in this group */
596 	__u16 reserved_blocks;	/* Number of reserved blocks in this group */
597 	__u16 unused;
598 };
599 
600 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
601 struct compat_ext4_new_group_input {
602 	u32 group;
603 	compat_u64 block_bitmap;
604 	compat_u64 inode_bitmap;
605 	compat_u64 inode_table;
606 	u32 blocks_count;
607 	u16 reserved_blocks;
608 	u16 unused;
609 };
610 #endif
611 
612 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
613 struct ext4_new_group_data {
614 	__u32 group;
615 	__u64 block_bitmap;
616 	__u64 inode_bitmap;
617 	__u64 inode_table;
618 	__u32 blocks_count;
619 	__u16 reserved_blocks;
620 	__u16 mdata_blocks;
621 	__u32 free_clusters_count;
622 };
623 
624 /* Indexes used to index group tables in ext4_new_group_data */
625 enum {
626 	BLOCK_BITMAP = 0,	/* block bitmap */
627 	INODE_BITMAP,		/* inode bitmap */
628 	INODE_TABLE,		/* inode tables */
629 	GROUP_TABLE_COUNT,
630 };
631 
632 /*
633  * Flags used by ext4_map_blocks()
634  */
635 	/* Allocate any needed blocks and/or convert an unwritten
636 	   extent to be an initialized ext4 */
637 #define EXT4_GET_BLOCKS_CREATE			0x0001
638 	/* Request the creation of an unwritten extent */
639 #define EXT4_GET_BLOCKS_UNWRIT_EXT		0x0002
640 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT	(EXT4_GET_BLOCKS_UNWRIT_EXT|\
641 						 EXT4_GET_BLOCKS_CREATE)
642 	/* Caller is from the delayed allocation writeout path
643 	 * finally doing the actual allocation of delayed blocks */
644 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE	0x0004
645 	/* caller is from the direct IO path, request to creation of an
646 	unwritten extents if not allocated, split the unwritten
647 	extent if blocks has been preallocated already*/
648 #define EXT4_GET_BLOCKS_PRE_IO			0x0008
649 #define EXT4_GET_BLOCKS_CONVERT			0x0010
650 #define EXT4_GET_BLOCKS_IO_CREATE_EXT		(EXT4_GET_BLOCKS_PRE_IO|\
651 					 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
652 	/* Convert extent to initialized after IO complete */
653 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT		(EXT4_GET_BLOCKS_CONVERT|\
654 					 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
655 	/* Eventual metadata allocation (due to growing extent tree)
656 	 * should not fail, so try to use reserved blocks for that.*/
657 #define EXT4_GET_BLOCKS_METADATA_NOFAIL		0x0020
658 	/* Don't normalize allocation size (used for fallocate) */
659 #define EXT4_GET_BLOCKS_NO_NORMALIZE		0x0040
660 	/* Convert written extents to unwritten */
661 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN	0x0100
662 	/* Write zeros to newly created written extents */
663 #define EXT4_GET_BLOCKS_ZERO			0x0200
664 #define EXT4_GET_BLOCKS_CREATE_ZERO		(EXT4_GET_BLOCKS_CREATE |\
665 					EXT4_GET_BLOCKS_ZERO)
666 	/* Caller will submit data before dropping transaction handle. This
667 	 * allows jbd2 to avoid submitting data before commit. */
668 #define EXT4_GET_BLOCKS_IO_SUBMIT		0x0400
669 
670 /*
671  * The bit position of these flags must not overlap with any of the
672  * EXT4_GET_BLOCKS_*.  They are used by ext4_find_extent(),
673  * read_extent_tree_block(), ext4_split_extent_at(),
674  * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
675  * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
676  * caching the extents when reading from the extent tree while a
677  * truncate or punch hole operation is in progress.
678  */
679 #define EXT4_EX_NOCACHE				0x40000000
680 #define EXT4_EX_FORCE_CACHE			0x20000000
681 #define EXT4_EX_NOFAIL				0x10000000
682 
683 /*
684  * Flags used by ext4_free_blocks
685  */
686 #define EXT4_FREE_BLOCKS_METADATA		0x0001
687 #define EXT4_FREE_BLOCKS_FORGET			0x0002
688 #define EXT4_FREE_BLOCKS_VALIDATED		0x0004
689 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE		0x0008
690 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER	0x0010
691 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER	0x0020
692 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER      0x0040
693 
694 /*
695  * ioctl commands
696  */
697 #define	EXT4_IOC_GETVERSION		_IOR('f', 3, long)
698 #define	EXT4_IOC_SETVERSION		_IOW('f', 4, long)
699 #define	EXT4_IOC_GETVERSION_OLD		FS_IOC_GETVERSION
700 #define	EXT4_IOC_SETVERSION_OLD		FS_IOC_SETVERSION
701 #define EXT4_IOC_GETRSVSZ		_IOR('f', 5, long)
702 #define EXT4_IOC_SETRSVSZ		_IOW('f', 6, long)
703 #define EXT4_IOC_GROUP_EXTEND		_IOW('f', 7, unsigned long)
704 #define EXT4_IOC_GROUP_ADD		_IOW('f', 8, struct ext4_new_group_input)
705 #define EXT4_IOC_MIGRATE		_IO('f', 9)
706  /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
707  /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
708 #define EXT4_IOC_ALLOC_DA_BLKS		_IO('f', 12)
709 #define EXT4_IOC_MOVE_EXT		_IOWR('f', 15, struct move_extent)
710 #define EXT4_IOC_RESIZE_FS		_IOW('f', 16, __u64)
711 #define EXT4_IOC_SWAP_BOOT		_IO('f', 17)
712 #define EXT4_IOC_PRECACHE_EXTENTS	_IO('f', 18)
713 /* ioctl codes 19--39 are reserved for fscrypt */
714 #define EXT4_IOC_CLEAR_ES_CACHE		_IO('f', 40)
715 #define EXT4_IOC_GETSTATE		_IOW('f', 41, __u32)
716 #define EXT4_IOC_GET_ES_CACHE		_IOWR('f', 42, struct fiemap)
717 
718 #define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32)
719 
720 /*
721  * Flags for going down operation
722  */
723 #define EXT4_GOING_FLAGS_DEFAULT		0x0	/* going down */
724 #define EXT4_GOING_FLAGS_LOGFLUSH		0x1	/* flush log but not data */
725 #define EXT4_GOING_FLAGS_NOLOGFLUSH		0x2	/* don't flush log nor data */
726 
727 /*
728  * Flags returned by EXT4_IOC_GETSTATE
729  *
730  * We only expose to userspace a subset of the state flags in
731  * i_state_flags
732  */
733 #define EXT4_STATE_FLAG_EXT_PRECACHED	0x00000001
734 #define EXT4_STATE_FLAG_NEW		0x00000002
735 #define EXT4_STATE_FLAG_NEWENTRY	0x00000004
736 #define EXT4_STATE_FLAG_DA_ALLOC_CLOSE	0x00000008
737 
738 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
739 /*
740  * ioctl commands in 32 bit emulation
741  */
742 #define EXT4_IOC32_GETVERSION		_IOR('f', 3, int)
743 #define EXT4_IOC32_SETVERSION		_IOW('f', 4, int)
744 #define EXT4_IOC32_GETRSVSZ		_IOR('f', 5, int)
745 #define EXT4_IOC32_SETRSVSZ		_IOW('f', 6, int)
746 #define EXT4_IOC32_GROUP_EXTEND		_IOW('f', 7, unsigned int)
747 #define EXT4_IOC32_GROUP_ADD		_IOW('f', 8, struct compat_ext4_new_group_input)
748 #define EXT4_IOC32_GETVERSION_OLD	FS_IOC32_GETVERSION
749 #define EXT4_IOC32_SETVERSION_OLD	FS_IOC32_SETVERSION
750 #endif
751 
752 /*
753  * Returned by EXT4_IOC_GET_ES_CACHE as an additional possible flag.
754  * It indicates that the entry in extent status cache is for a hole.
755  */
756 #define EXT4_FIEMAP_EXTENT_HOLE		0x08000000
757 
758 /* Max physical block we can address w/o extents */
759 #define EXT4_MAX_BLOCK_FILE_PHYS	0xFFFFFFFF
760 
761 /* Max logical block we can support */
762 #define EXT4_MAX_LOGICAL_BLOCK		0xFFFFFFFE
763 
764 /*
765  * Structure of an inode on the disk
766  */
767 struct ext4_inode {
768 	__le16	i_mode;		/* File mode */
769 	__le16	i_uid;		/* Low 16 bits of Owner Uid */
770 	__le32	i_size_lo;	/* Size in bytes */
771 	__le32	i_atime;	/* Access time */
772 	__le32	i_ctime;	/* Inode Change time */
773 	__le32	i_mtime;	/* Modification time */
774 	__le32	i_dtime;	/* Deletion Time */
775 	__le16	i_gid;		/* Low 16 bits of Group Id */
776 	__le16	i_links_count;	/* Links count */
777 	__le32	i_blocks_lo;	/* Blocks count */
778 	__le32	i_flags;	/* File flags */
779 	union {
780 		struct {
781 			__le32  l_i_version;
782 		} linux1;
783 		struct {
784 			__u32  h_i_translator;
785 		} hurd1;
786 		struct {
787 			__u32  m_i_reserved1;
788 		} masix1;
789 	} osd1;				/* OS dependent 1 */
790 	__le32	i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
791 	__le32	i_generation;	/* File version (for NFS) */
792 	__le32	i_file_acl_lo;	/* File ACL */
793 	__le32	i_size_high;
794 	__le32	i_obso_faddr;	/* Obsoleted fragment address */
795 	union {
796 		struct {
797 			__le16	l_i_blocks_high; /* were l_i_reserved1 */
798 			__le16	l_i_file_acl_high;
799 			__le16	l_i_uid_high;	/* these 2 fields */
800 			__le16	l_i_gid_high;	/* were reserved2[0] */
801 			__le16	l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
802 			__le16	l_i_reserved;
803 		} linux2;
804 		struct {
805 			__le16	h_i_reserved1;	/* Obsoleted fragment number/size which are removed in ext4 */
806 			__u16	h_i_mode_high;
807 			__u16	h_i_uid_high;
808 			__u16	h_i_gid_high;
809 			__u32	h_i_author;
810 		} hurd2;
811 		struct {
812 			__le16	h_i_reserved1;	/* Obsoleted fragment number/size which are removed in ext4 */
813 			__le16	m_i_file_acl_high;
814 			__u32	m_i_reserved2[2];
815 		} masix2;
816 	} osd2;				/* OS dependent 2 */
817 	__le16	i_extra_isize;
818 	__le16	i_checksum_hi;	/* crc32c(uuid+inum+inode) BE */
819 	__le32  i_ctime_extra;  /* extra Change time      (nsec << 2 | epoch) */
820 	__le32  i_mtime_extra;  /* extra Modification time(nsec << 2 | epoch) */
821 	__le32  i_atime_extra;  /* extra Access time      (nsec << 2 | epoch) */
822 	__le32  i_crtime;       /* File Creation time */
823 	__le32  i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
824 	__le32  i_version_hi;	/* high 32 bits for 64-bit version */
825 	__le32	i_projid;	/* Project ID */
826 };
827 
828 struct move_extent {
829 	__u32 reserved;		/* should be zero */
830 	__u32 donor_fd;		/* donor file descriptor */
831 	__u64 orig_start;	/* logical start offset in block for orig */
832 	__u64 donor_start;	/* logical start offset in block for donor */
833 	__u64 len;		/* block length to be moved */
834 	__u64 moved_len;	/* moved block length */
835 };
836 
837 #define EXT4_EPOCH_BITS 2
838 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
839 #define EXT4_NSEC_MASK  (~0UL << EXT4_EPOCH_BITS)
840 
841 /*
842  * Extended fields will fit into an inode if the filesystem was formatted
843  * with large inodes (-I 256 or larger) and there are not currently any EAs
844  * consuming all of the available space. For new inodes we always reserve
845  * enough space for the kernel's known extended fields, but for inodes
846  * created with an old kernel this might not have been the case. None of
847  * the extended inode fields is critical for correct filesystem operation.
848  * This macro checks if a certain field fits in the inode. Note that
849  * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
850  */
851 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field)	\
852 	((offsetof(typeof(*ext4_inode), field) +	\
853 	  sizeof((ext4_inode)->field))			\
854 	<= (EXT4_GOOD_OLD_INODE_SIZE +			\
855 	    (einode)->i_extra_isize))			\
856 
857 /*
858  * We use an encoding that preserves the times for extra epoch "00":
859  *
860  * extra  msb of                         adjust for signed
861  * epoch  32-bit                         32-bit tv_sec to
862  * bits   time    decoded 64-bit tv_sec  64-bit tv_sec      valid time range
863  * 0 0    1    -0x80000000..-0x00000001  0x000000000 1901-12-13..1969-12-31
864  * 0 0    0    0x000000000..0x07fffffff  0x000000000 1970-01-01..2038-01-19
865  * 0 1    1    0x080000000..0x0ffffffff  0x100000000 2038-01-19..2106-02-07
866  * 0 1    0    0x100000000..0x17fffffff  0x100000000 2106-02-07..2174-02-25
867  * 1 0    1    0x180000000..0x1ffffffff  0x200000000 2174-02-25..2242-03-16
868  * 1 0    0    0x200000000..0x27fffffff  0x200000000 2242-03-16..2310-04-04
869  * 1 1    1    0x280000000..0x2ffffffff  0x300000000 2310-04-04..2378-04-22
870  * 1 1    0    0x300000000..0x37fffffff  0x300000000 2378-04-22..2446-05-10
871  *
872  * Note that previous versions of the kernel on 64-bit systems would
873  * incorrectly use extra epoch bits 1,1 for dates between 1901 and
874  * 1970.  e2fsck will correct this, assuming that it is run on the
875  * affected filesystem before 2242.
876  */
877 
ext4_encode_extra_time(struct timespec64 * time)878 static inline __le32 ext4_encode_extra_time(struct timespec64 *time)
879 {
880 	u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK;
881 	return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
882 }
883 
ext4_decode_extra_time(struct timespec64 * time,__le32 extra)884 static inline void ext4_decode_extra_time(struct timespec64 *time,
885 					  __le32 extra)
886 {
887 	if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK)))
888 		time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
889 	time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
890 }
891 
892 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode)				\
893 do {										\
894 	if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     {\
895 		(raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec);	\
896 		(raw_inode)->xtime ## _extra =					\
897 				ext4_encode_extra_time(&(inode)->xtime);	\
898 		}								\
899 	else	\
900 		(raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (inode)->xtime.tv_sec, S32_MIN, S32_MAX));	\
901 } while (0)
902 
903 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode)			       \
904 do {									       \
905 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))		       \
906 		(raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec);      \
907 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))	       \
908 		(raw_inode)->xtime ## _extra =				       \
909 				ext4_encode_extra_time(&(einode)->xtime);      \
910 } while (0)
911 
912 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode)				\
913 do {										\
914 	(inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime);	\
915 	if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {	\
916 		ext4_decode_extra_time(&(inode)->xtime,				\
917 				       raw_inode->xtime ## _extra);		\
918 		}								\
919 	else									\
920 		(inode)->xtime.tv_nsec = 0;					\
921 } while (0)
922 
923 
924 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode)			       \
925 do {									       \
926 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))		       \
927 		(einode)->xtime.tv_sec = 				       \
928 			(signed)le32_to_cpu((raw_inode)->xtime);	       \
929 	else								       \
930 		(einode)->xtime.tv_sec = 0;				       \
931 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))	       \
932 		ext4_decode_extra_time(&(einode)->xtime,		       \
933 				       raw_inode->xtime ## _extra);	       \
934 	else								       \
935 		(einode)->xtime.tv_nsec = 0;				       \
936 } while (0)
937 
938 #define i_disk_version osd1.linux1.l_i_version
939 
940 #if defined(__KERNEL__) || defined(__linux__)
941 #define i_reserved1	osd1.linux1.l_i_reserved1
942 #define i_file_acl_high	osd2.linux2.l_i_file_acl_high
943 #define i_blocks_high	osd2.linux2.l_i_blocks_high
944 #define i_uid_low	i_uid
945 #define i_gid_low	i_gid
946 #define i_uid_high	osd2.linux2.l_i_uid_high
947 #define i_gid_high	osd2.linux2.l_i_gid_high
948 #define i_checksum_lo	osd2.linux2.l_i_checksum_lo
949 
950 #elif defined(__GNU__)
951 
952 #define i_translator	osd1.hurd1.h_i_translator
953 #define i_uid_high	osd2.hurd2.h_i_uid_high
954 #define i_gid_high	osd2.hurd2.h_i_gid_high
955 #define i_author	osd2.hurd2.h_i_author
956 
957 #elif defined(__masix__)
958 
959 #define i_reserved1	osd1.masix1.m_i_reserved1
960 #define i_file_acl_high	osd2.masix2.m_i_file_acl_high
961 #define i_reserved2	osd2.masix2.m_i_reserved2
962 
963 #endif /* defined(__KERNEL__) || defined(__linux__) */
964 
965 #include "extents_status.h"
966 #include "fast_commit.h"
967 
968 /*
969  * Lock subclasses for i_data_sem in the ext4_inode_info structure.
970  *
971  * These are needed to avoid lockdep false positives when we need to
972  * allocate blocks to the quota inode during ext4_map_blocks(), while
973  * holding i_data_sem for a normal (non-quota) inode.  Since we don't
974  * do quota tracking for the quota inode, this avoids deadlock (as
975  * well as infinite recursion, since it isn't turtles all the way
976  * down...)
977  *
978  *  I_DATA_SEM_NORMAL - Used for most inodes
979  *  I_DATA_SEM_OTHER  - Used by move_inode.c for the second normal inode
980  *			  where the second inode has larger inode number
981  *			  than the first
982  *  I_DATA_SEM_QUOTA  - Used for quota inodes only
983  */
984 enum {
985 	I_DATA_SEM_NORMAL = 0,
986 	I_DATA_SEM_OTHER,
987 	I_DATA_SEM_QUOTA,
988 };
989 
990 
991 /*
992  * fourth extended file system inode data in memory
993  */
994 struct ext4_inode_info {
995 	__le32	i_data[15];	/* unconverted */
996 	__u32	i_dtime;
997 	ext4_fsblk_t	i_file_acl;
998 
999 	/*
1000 	 * i_block_group is the number of the block group which contains
1001 	 * this file's inode.  Constant across the lifetime of the inode,
1002 	 * it is used for making block allocation decisions - we try to
1003 	 * place a file's data blocks near its inode block, and new inodes
1004 	 * near to their parent directory's inode.
1005 	 */
1006 	ext4_group_t	i_block_group;
1007 	ext4_lblk_t	i_dir_start_lookup;
1008 #if (BITS_PER_LONG < 64)
1009 	unsigned long	i_state_flags;		/* Dynamic state flags */
1010 #endif
1011 	unsigned long	i_flags;
1012 
1013 	/*
1014 	 * Extended attributes can be read independently of the main file
1015 	 * data. Taking i_mutex even when reading would cause contention
1016 	 * between readers of EAs and writers of regular file data, so
1017 	 * instead we synchronize on xattr_sem when reading or changing
1018 	 * EAs.
1019 	 */
1020 	struct rw_semaphore xattr_sem;
1021 
1022 	struct list_head i_orphan;	/* unlinked but open inodes */
1023 
1024 	/* Fast commit related info */
1025 
1026 	struct list_head i_fc_list;	/*
1027 					 * inodes that need fast commit
1028 					 * protected by sbi->s_fc_lock.
1029 					 */
1030 
1031 	/* Start of lblk range that needs to be committed in this fast commit */
1032 	ext4_lblk_t i_fc_lblk_start;
1033 
1034 	/* End of lblk range that needs to be committed in this fast commit */
1035 	ext4_lblk_t i_fc_lblk_len;
1036 
1037 	/* Number of ongoing updates on this inode */
1038 	atomic_t  i_fc_updates;
1039 
1040 	/* Fast commit wait queue for this inode */
1041 	wait_queue_head_t i_fc_wait;
1042 
1043 	/* Protect concurrent accesses on i_fc_lblk_start, i_fc_lblk_len */
1044 	struct mutex i_fc_lock;
1045 
1046 	/*
1047 	 * i_disksize keeps track of what the inode size is ON DISK, not
1048 	 * in memory.  During truncate, i_size is set to the new size by
1049 	 * the VFS prior to calling ext4_truncate(), but the filesystem won't
1050 	 * set i_disksize to 0 until the truncate is actually under way.
1051 	 *
1052 	 * The intent is that i_disksize always represents the blocks which
1053 	 * are used by this file.  This allows recovery to restart truncate
1054 	 * on orphans if we crash during truncate.  We actually write i_disksize
1055 	 * into the on-disk inode when writing inodes out, instead of i_size.
1056 	 *
1057 	 * The only time when i_disksize and i_size may be different is when
1058 	 * a truncate is in progress.  The only things which change i_disksize
1059 	 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
1060 	 */
1061 	loff_t	i_disksize;
1062 
1063 	/*
1064 	 * i_data_sem is for serialising ext4_truncate() against
1065 	 * ext4_getblock().  In the 2.4 ext2 design, great chunks of inode's
1066 	 * data tree are chopped off during truncate. We can't do that in
1067 	 * ext4 because whenever we perform intermediate commits during
1068 	 * truncate, the inode and all the metadata blocks *must* be in a
1069 	 * consistent state which allows truncation of the orphans to restart
1070 	 * during recovery.  Hence we must fix the get_block-vs-truncate race
1071 	 * by other means, so we have i_data_sem.
1072 	 */
1073 	struct rw_semaphore i_data_sem;
1074 	/*
1075 	 * i_mmap_sem is for serializing page faults with truncate / punch hole
1076 	 * operations. We have to make sure that new page cannot be faulted in
1077 	 * a section of the inode that is being punched. We cannot easily use
1078 	 * i_data_sem for this since we need protection for the whole punch
1079 	 * operation and i_data_sem ranks below transaction start so we have
1080 	 * to occasionally drop it.
1081 	 */
1082 	struct rw_semaphore i_mmap_sem;
1083 	struct inode vfs_inode;
1084 	struct jbd2_inode *jinode;
1085 
1086 	spinlock_t i_raw_lock;	/* protects updates to the raw inode */
1087 
1088 	/*
1089 	 * File creation time. Its function is same as that of
1090 	 * struct timespec64 i_{a,c,m}time in the generic inode.
1091 	 */
1092 	struct timespec64 i_crtime;
1093 
1094 	/* mballoc */
1095 	atomic_t i_prealloc_active;
1096 	struct list_head i_prealloc_list;
1097 	spinlock_t i_prealloc_lock;
1098 
1099 	/* extents status tree */
1100 	struct ext4_es_tree i_es_tree;
1101 	rwlock_t i_es_lock;
1102 	struct list_head i_es_list;
1103 	unsigned int i_es_all_nr;	/* protected by i_es_lock */
1104 	unsigned int i_es_shk_nr;	/* protected by i_es_lock */
1105 	ext4_lblk_t i_es_shrink_lblk;	/* Offset where we start searching for
1106 					   extents to shrink. Protected by
1107 					   i_es_lock  */
1108 
1109 	/* ialloc */
1110 	ext4_group_t	i_last_alloc_group;
1111 
1112 	/* allocation reservation info for delalloc */
1113 	/* In case of bigalloc, this refer to clusters rather than blocks */
1114 	unsigned int i_reserved_data_blocks;
1115 
1116 	/* pending cluster reservations for bigalloc file systems */
1117 	struct ext4_pending_tree i_pending_tree;
1118 
1119 	/* on-disk additional length */
1120 	__u16 i_extra_isize;
1121 
1122 	/* Indicate the inline data space. */
1123 	u16 i_inline_off;
1124 	u16 i_inline_size;
1125 
1126 #ifdef CONFIG_QUOTA
1127 	/* quota space reservation, managed internally by quota code */
1128 	qsize_t i_reserved_quota;
1129 #endif
1130 
1131 	/* Lock protecting lists below */
1132 	spinlock_t i_completed_io_lock;
1133 	/*
1134 	 * Completed IOs that need unwritten extents handling and have
1135 	 * transaction reserved
1136 	 */
1137 	struct list_head i_rsv_conversion_list;
1138 	struct work_struct i_rsv_conversion_work;
1139 	atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1140 
1141 	spinlock_t i_block_reservation_lock;
1142 
1143 	/*
1144 	 * Transactions that contain inode's metadata needed to complete
1145 	 * fsync and fdatasync, respectively.
1146 	 */
1147 	tid_t i_sync_tid;
1148 	tid_t i_datasync_tid;
1149 
1150 #ifdef CONFIG_QUOTA
1151 	struct dquot *i_dquot[MAXQUOTAS];
1152 #endif
1153 
1154 	/* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1155 	__u32 i_csum_seed;
1156 
1157 	kprojid_t i_projid;
1158 };
1159 
1160 /*
1161  * File system states
1162  */
1163 #define	EXT4_VALID_FS			0x0001	/* Unmounted cleanly */
1164 #define	EXT4_ERROR_FS			0x0002	/* Errors detected */
1165 #define	EXT4_ORPHAN_FS			0x0004	/* Orphans being recovered */
1166 #define EXT4_FC_REPLAY			0x0020	/* Fast commit replay ongoing */
1167 
1168 /*
1169  * Misc. filesystem flags
1170  */
1171 #define EXT2_FLAGS_SIGNED_HASH		0x0001  /* Signed dirhash in use */
1172 #define EXT2_FLAGS_UNSIGNED_HASH	0x0002  /* Unsigned dirhash in use */
1173 #define EXT2_FLAGS_TEST_FILESYS		0x0004	/* to test development code */
1174 
1175 /*
1176  * Mount flags set via mount options or defaults
1177  */
1178 #define EXT4_MOUNT_NO_MBCACHE		0x00001 /* Do not use mbcache */
1179 #define EXT4_MOUNT_GRPID		0x00004	/* Create files with directory's group */
1180 #define EXT4_MOUNT_DEBUG		0x00008	/* Some debugging messages */
1181 #define EXT4_MOUNT_ERRORS_CONT		0x00010	/* Continue on errors */
1182 #define EXT4_MOUNT_ERRORS_RO		0x00020	/* Remount fs ro on errors */
1183 #define EXT4_MOUNT_ERRORS_PANIC		0x00040	/* Panic on errors */
1184 #define EXT4_MOUNT_ERRORS_MASK		0x00070
1185 #define EXT4_MOUNT_MINIX_DF		0x00080	/* Mimics the Minix statfs */
1186 #define EXT4_MOUNT_NOLOAD		0x00100	/* Don't use existing journal*/
1187 #ifdef CONFIG_FS_DAX
1188 #define EXT4_MOUNT_DAX_ALWAYS		0x00200	/* Direct Access */
1189 #else
1190 #define EXT4_MOUNT_DAX_ALWAYS		0
1191 #endif
1192 #define EXT4_MOUNT_DATA_FLAGS		0x00C00	/* Mode for data writes: */
1193 #define EXT4_MOUNT_JOURNAL_DATA		0x00400	/* Write data to journal */
1194 #define EXT4_MOUNT_ORDERED_DATA		0x00800	/* Flush data before commit */
1195 #define EXT4_MOUNT_WRITEBACK_DATA	0x00C00	/* No data ordering */
1196 #define EXT4_MOUNT_UPDATE_JOURNAL	0x01000	/* Update the journal format */
1197 #define EXT4_MOUNT_NO_UID32		0x02000  /* Disable 32-bit UIDs */
1198 #define EXT4_MOUNT_XATTR_USER		0x04000	/* Extended user attributes */
1199 #define EXT4_MOUNT_POSIX_ACL		0x08000	/* POSIX Access Control Lists */
1200 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC	0x10000	/* No auto delalloc mapping */
1201 #define EXT4_MOUNT_BARRIER		0x20000 /* Use block barriers */
1202 #define EXT4_MOUNT_QUOTA		0x40000 /* Some quota option set */
1203 #define EXT4_MOUNT_USRQUOTA		0x80000 /* "old" user quota,
1204 						 * enable enforcement for hidden
1205 						 * quota files */
1206 #define EXT4_MOUNT_GRPQUOTA		0x100000 /* "old" group quota, enable
1207 						  * enforcement for hidden quota
1208 						  * files */
1209 #define EXT4_MOUNT_PRJQUOTA		0x200000 /* Enable project quota
1210 						  * enforcement */
1211 #define EXT4_MOUNT_DIOREAD_NOLOCK	0x400000 /* Enable support for dio read nolocking */
1212 #define EXT4_MOUNT_JOURNAL_CHECKSUM	0x800000 /* Journal checksums */
1213 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT	0x1000000 /* Journal Async Commit */
1214 #define EXT4_MOUNT_WARN_ON_ERROR	0x2000000 /* Trigger WARN_ON on error */
1215 #define EXT4_MOUNT_PREFETCH_BLOCK_BITMAPS 0x4000000
1216 #define EXT4_MOUNT_DELALLOC		0x8000000 /* Delalloc support */
1217 #define EXT4_MOUNT_DATA_ERR_ABORT	0x10000000 /* Abort on file data write */
1218 #define EXT4_MOUNT_BLOCK_VALIDITY	0x20000000 /* Block validity checking */
1219 #define EXT4_MOUNT_DISCARD		0x40000000 /* Issue DISCARD requests */
1220 #define EXT4_MOUNT_INIT_INODE_TABLE	0x80000000 /* Initialize uninitialized itables */
1221 
1222 /*
1223  * Mount flags set either automatically (could not be set by mount option)
1224  * based on per file system feature or property or in special cases such as
1225  * distinguishing between explicit mount option definition and default.
1226  */
1227 #define EXT4_MOUNT2_EXPLICIT_DELALLOC	0x00000001 /* User explicitly
1228 						      specified delalloc */
1229 #define EXT4_MOUNT2_STD_GROUP_SIZE	0x00000002 /* We have standard group
1230 						      size of blocksize * 8
1231 						      blocks */
1232 #define EXT4_MOUNT2_HURD_COMPAT		0x00000004 /* Support HURD-castrated
1233 						      file systems */
1234 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM	0x00000008 /* User explicitly
1235 						specified journal checksum */
1236 
1237 #define EXT4_MOUNT2_JOURNAL_FAST_COMMIT	0x00000010 /* Journal fast commit */
1238 #define EXT4_MOUNT2_DAX_NEVER		0x00000020 /* Do not allow Direct Access */
1239 #define EXT4_MOUNT2_DAX_INODE		0x00000040 /* For printing options only */
1240 
1241 
1242 #define clear_opt(sb, opt)		EXT4_SB(sb)->s_mount_opt &= \
1243 						~EXT4_MOUNT_##opt
1244 #define set_opt(sb, opt)		EXT4_SB(sb)->s_mount_opt |= \
1245 						EXT4_MOUNT_##opt
1246 #define test_opt(sb, opt)		(EXT4_SB(sb)->s_mount_opt & \
1247 					 EXT4_MOUNT_##opt)
1248 
1249 #define clear_opt2(sb, opt)		EXT4_SB(sb)->s_mount_opt2 &= \
1250 						~EXT4_MOUNT2_##opt
1251 #define set_opt2(sb, opt)		EXT4_SB(sb)->s_mount_opt2 |= \
1252 						EXT4_MOUNT2_##opt
1253 #define test_opt2(sb, opt)		(EXT4_SB(sb)->s_mount_opt2 & \
1254 					 EXT4_MOUNT2_##opt)
1255 
1256 #define ext4_test_and_set_bit		__test_and_set_bit_le
1257 #define ext4_set_bit			__set_bit_le
1258 #define ext4_set_bit_atomic		ext2_set_bit_atomic
1259 #define ext4_test_and_clear_bit		__test_and_clear_bit_le
1260 #define ext4_clear_bit			__clear_bit_le
1261 #define ext4_clear_bit_atomic		ext2_clear_bit_atomic
1262 #define ext4_test_bit			test_bit_le
1263 #define ext4_find_next_zero_bit		find_next_zero_bit_le
1264 #define ext4_find_next_bit		find_next_bit_le
1265 
1266 extern void ext4_set_bits(void *bm, int cur, int len);
1267 
1268 /*
1269  * Maximal mount counts between two filesystem checks
1270  */
1271 #define EXT4_DFL_MAX_MNT_COUNT		20	/* Allow 20 mounts */
1272 #define EXT4_DFL_CHECKINTERVAL		0	/* Don't use interval check */
1273 
1274 /*
1275  * Behaviour when detecting errors
1276  */
1277 #define EXT4_ERRORS_CONTINUE		1	/* Continue execution */
1278 #define EXT4_ERRORS_RO			2	/* Remount fs read-only */
1279 #define EXT4_ERRORS_PANIC		3	/* Panic */
1280 #define EXT4_ERRORS_DEFAULT		EXT4_ERRORS_CONTINUE
1281 
1282 /* Metadata checksum algorithm codes */
1283 #define EXT4_CRC32C_CHKSUM		1
1284 
1285 /*
1286  * Structure of the super block
1287  */
1288 struct ext4_super_block {
1289 /*00*/	__le32	s_inodes_count;		/* Inodes count */
1290 	__le32	s_blocks_count_lo;	/* Blocks count */
1291 	__le32	s_r_blocks_count_lo;	/* Reserved blocks count */
1292 	__le32	s_free_blocks_count_lo;	/* Free blocks count */
1293 /*10*/	__le32	s_free_inodes_count;	/* Free inodes count */
1294 	__le32	s_first_data_block;	/* First Data Block */
1295 	__le32	s_log_block_size;	/* Block size */
1296 	__le32	s_log_cluster_size;	/* Allocation cluster size */
1297 /*20*/	__le32	s_blocks_per_group;	/* # Blocks per group */
1298 	__le32	s_clusters_per_group;	/* # Clusters per group */
1299 	__le32	s_inodes_per_group;	/* # Inodes per group */
1300 	__le32	s_mtime;		/* Mount time */
1301 /*30*/	__le32	s_wtime;		/* Write time */
1302 	__le16	s_mnt_count;		/* Mount count */
1303 	__le16	s_max_mnt_count;	/* Maximal mount count */
1304 	__le16	s_magic;		/* Magic signature */
1305 	__le16	s_state;		/* File system state */
1306 	__le16	s_errors;		/* Behaviour when detecting errors */
1307 	__le16	s_minor_rev_level;	/* minor revision level */
1308 /*40*/	__le32	s_lastcheck;		/* time of last check */
1309 	__le32	s_checkinterval;	/* max. time between checks */
1310 	__le32	s_creator_os;		/* OS */
1311 	__le32	s_rev_level;		/* Revision level */
1312 /*50*/	__le16	s_def_resuid;		/* Default uid for reserved blocks */
1313 	__le16	s_def_resgid;		/* Default gid for reserved blocks */
1314 	/*
1315 	 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1316 	 *
1317 	 * Note: the difference between the compatible feature set and
1318 	 * the incompatible feature set is that if there is a bit set
1319 	 * in the incompatible feature set that the kernel doesn't
1320 	 * know about, it should refuse to mount the filesystem.
1321 	 *
1322 	 * e2fsck's requirements are more strict; if it doesn't know
1323 	 * about a feature in either the compatible or incompatible
1324 	 * feature set, it must abort and not try to meddle with
1325 	 * things it doesn't understand...
1326 	 */
1327 	__le32	s_first_ino;		/* First non-reserved inode */
1328 	__le16  s_inode_size;		/* size of inode structure */
1329 	__le16	s_block_group_nr;	/* block group # of this superblock */
1330 	__le32	s_feature_compat;	/* compatible feature set */
1331 /*60*/	__le32	s_feature_incompat;	/* incompatible feature set */
1332 	__le32	s_feature_ro_compat;	/* readonly-compatible feature set */
1333 /*68*/	__u8	s_uuid[16];		/* 128-bit uuid for volume */
1334 /*78*/	char	s_volume_name[16];	/* volume name */
1335 /*88*/	char	s_last_mounted[64] __nonstring;	/* directory where last mounted */
1336 /*C8*/	__le32	s_algorithm_usage_bitmap; /* For compression */
1337 	/*
1338 	 * Performance hints.  Directory preallocation should only
1339 	 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1340 	 */
1341 	__u8	s_prealloc_blocks;	/* Nr of blocks to try to preallocate*/
1342 	__u8	s_prealloc_dir_blocks;	/* Nr to preallocate for dirs */
1343 	__le16	s_reserved_gdt_blocks;	/* Per group desc for online growth */
1344 	/*
1345 	 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1346 	 */
1347 /*D0*/	__u8	s_journal_uuid[16];	/* uuid of journal superblock */
1348 /*E0*/	__le32	s_journal_inum;		/* inode number of journal file */
1349 	__le32	s_journal_dev;		/* device number of journal file */
1350 	__le32	s_last_orphan;		/* start of list of inodes to delete */
1351 	__le32	s_hash_seed[4];		/* HTREE hash seed */
1352 	__u8	s_def_hash_version;	/* Default hash version to use */
1353 	__u8	s_jnl_backup_type;
1354 	__le16  s_desc_size;		/* size of group descriptor */
1355 /*100*/	__le32	s_default_mount_opts;
1356 	__le32	s_first_meta_bg;	/* First metablock block group */
1357 	__le32	s_mkfs_time;		/* When the filesystem was created */
1358 	__le32	s_jnl_blocks[17];	/* Backup of the journal inode */
1359 	/* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1360 /*150*/	__le32	s_blocks_count_hi;	/* Blocks count */
1361 	__le32	s_r_blocks_count_hi;	/* Reserved blocks count */
1362 	__le32	s_free_blocks_count_hi;	/* Free blocks count */
1363 	__le16	s_min_extra_isize;	/* All inodes have at least # bytes */
1364 	__le16	s_want_extra_isize; 	/* New inodes should reserve # bytes */
1365 	__le32	s_flags;		/* Miscellaneous flags */
1366 	__le16  s_raid_stride;		/* RAID stride */
1367 	__le16  s_mmp_update_interval;  /* # seconds to wait in MMP checking */
1368 	__le64  s_mmp_block;            /* Block for multi-mount protection */
1369 	__le32  s_raid_stripe_width;    /* blocks on all data disks (N*stride)*/
1370 	__u8	s_log_groups_per_flex;  /* FLEX_BG group size */
1371 	__u8	s_checksum_type;	/* metadata checksum algorithm used */
1372 	__u8	s_encryption_level;	/* versioning level for encryption */
1373 	__u8	s_reserved_pad;		/* Padding to next 32bits */
1374 	__le64	s_kbytes_written;	/* nr of lifetime kilobytes written */
1375 	__le32	s_snapshot_inum;	/* Inode number of active snapshot */
1376 	__le32	s_snapshot_id;		/* sequential ID of active snapshot */
1377 	__le64	s_snapshot_r_blocks_count; /* reserved blocks for active
1378 					      snapshot's future use */
1379 	__le32	s_snapshot_list;	/* inode number of the head of the
1380 					   on-disk snapshot list */
1381 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1382 	__le32	s_error_count;		/* number of fs errors */
1383 	__le32	s_first_error_time;	/* first time an error happened */
1384 	__le32	s_first_error_ino;	/* inode involved in first error */
1385 	__le64	s_first_error_block;	/* block involved of first error */
1386 	__u8	s_first_error_func[32] __nonstring;	/* function where the error happened */
1387 	__le32	s_first_error_line;	/* line number where error happened */
1388 	__le32	s_last_error_time;	/* most recent time of an error */
1389 	__le32	s_last_error_ino;	/* inode involved in last error */
1390 	__le32	s_last_error_line;	/* line number where error happened */
1391 	__le64	s_last_error_block;	/* block involved of last error */
1392 	__u8	s_last_error_func[32] __nonstring;	/* function where the error happened */
1393 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1394 	__u8	s_mount_opts[64];
1395 	__le32	s_usr_quota_inum;	/* inode for tracking user quota */
1396 	__le32	s_grp_quota_inum;	/* inode for tracking group quota */
1397 	__le32	s_overhead_clusters;	/* overhead blocks/clusters in fs */
1398 	__le32	s_backup_bgs[2];	/* groups with sparse_super2 SBs */
1399 	__u8	s_encrypt_algos[4];	/* Encryption algorithms in use  */
1400 	__u8	s_encrypt_pw_salt[16];	/* Salt used for string2key algorithm */
1401 	__le32	s_lpf_ino;		/* Location of the lost+found inode */
1402 	__le32	s_prj_quota_inum;	/* inode for tracking project quota */
1403 	__le32	s_checksum_seed;	/* crc32c(uuid) if csum_seed set */
1404 	__u8	s_wtime_hi;
1405 	__u8	s_mtime_hi;
1406 	__u8	s_mkfs_time_hi;
1407 	__u8	s_lastcheck_hi;
1408 	__u8	s_first_error_time_hi;
1409 	__u8	s_last_error_time_hi;
1410 	__u8	s_first_error_errcode;
1411 	__u8    s_last_error_errcode;
1412 	__le16  s_encoding;		/* Filename charset encoding */
1413 	__le16  s_encoding_flags;	/* Filename charset encoding flags */
1414 	__le32	s_reserved[95];		/* Padding to the end of the block */
1415 	__le32	s_checksum;		/* crc32c(superblock) */
1416 };
1417 
1418 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1419 
1420 #ifdef __KERNEL__
1421 
1422 #ifdef CONFIG_FS_ENCRYPTION
1423 #define DUMMY_ENCRYPTION_ENABLED(sbi) ((sbi)->s_dummy_enc_policy.policy != NULL)
1424 #else
1425 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1426 #endif
1427 
1428 /* Number of quota types we support */
1429 #define EXT4_MAXQUOTAS 3
1430 
1431 #define EXT4_ENC_UTF8_12_1	1
1432 
1433 /*
1434  * fourth extended-fs super-block data in memory
1435  */
1436 struct ext4_sb_info {
1437 	unsigned long s_desc_size;	/* Size of a group descriptor in bytes */
1438 	unsigned long s_inodes_per_block;/* Number of inodes per block */
1439 	unsigned long s_blocks_per_group;/* Number of blocks in a group */
1440 	unsigned long s_clusters_per_group; /* Number of clusters in a group */
1441 	unsigned long s_inodes_per_group;/* Number of inodes in a group */
1442 	unsigned long s_itb_per_group;	/* Number of inode table blocks per group */
1443 	unsigned long s_gdb_count;	/* Number of group descriptor blocks */
1444 	unsigned long s_desc_per_block;	/* Number of group descriptors per block */
1445 	ext4_group_t s_groups_count;	/* Number of groups in the fs */
1446 	ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1447 	unsigned long s_overhead;  /* # of fs overhead clusters */
1448 	unsigned int s_cluster_ratio;	/* Number of blocks per cluster */
1449 	unsigned int s_cluster_bits;	/* log2 of s_cluster_ratio */
1450 	loff_t s_bitmap_maxbytes;	/* max bytes for bitmap files */
1451 	struct buffer_head * s_sbh;	/* Buffer containing the super block */
1452 	struct ext4_super_block *s_es;	/* Pointer to the super block in the buffer */
1453 	struct buffer_head * __rcu *s_group_desc;
1454 	unsigned int s_mount_opt;
1455 	unsigned int s_mount_opt2;
1456 	unsigned long s_mount_flags;
1457 	unsigned int s_def_mount_opt;
1458 	ext4_fsblk_t s_sb_block;
1459 	atomic64_t s_resv_clusters;
1460 	kuid_t s_resuid;
1461 	kgid_t s_resgid;
1462 	unsigned short s_mount_state;
1463 	unsigned short s_pad;
1464 	int s_addr_per_block_bits;
1465 	int s_desc_per_block_bits;
1466 	int s_inode_size;
1467 	int s_first_ino;
1468 	unsigned int s_inode_readahead_blks;
1469 	unsigned int s_inode_goal;
1470 	u32 s_hash_seed[4];
1471 	int s_def_hash_version;
1472 	int s_hash_unsigned;	/* 3 if hash should be signed, 0 if not */
1473 	struct percpu_counter s_freeclusters_counter;
1474 	struct percpu_counter s_freeinodes_counter;
1475 	struct percpu_counter s_dirs_counter;
1476 	struct percpu_counter s_dirtyclusters_counter;
1477 	struct blockgroup_lock *s_blockgroup_lock;
1478 	struct proc_dir_entry *s_proc;
1479 	struct kobject s_kobj;
1480 	struct completion s_kobj_unregister;
1481 	struct super_block *s_sb;
1482 
1483 	/* Journaling */
1484 	struct journal_s *s_journal;
1485 	struct list_head s_orphan;
1486 	struct mutex s_orphan_lock;
1487 	unsigned long s_ext4_flags;		/* Ext4 superblock flags */
1488 	unsigned long s_commit_interval;
1489 	u32 s_max_batch_time;
1490 	u32 s_min_batch_time;
1491 	struct block_device *s_journal_bdev;
1492 #ifdef CONFIG_QUOTA
1493 	/* Names of quota files with journalled quota */
1494 	char __rcu *s_qf_names[EXT4_MAXQUOTAS];
1495 	int s_jquota_fmt;			/* Format of quota to use */
1496 #endif
1497 	unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1498 	struct ext4_system_blocks __rcu *s_system_blks;
1499 
1500 #ifdef EXTENTS_STATS
1501 	/* ext4 extents stats */
1502 	unsigned long s_ext_min;
1503 	unsigned long s_ext_max;
1504 	unsigned long s_depth_max;
1505 	spinlock_t s_ext_stats_lock;
1506 	unsigned long s_ext_blocks;
1507 	unsigned long s_ext_extents;
1508 #endif
1509 
1510 	/* for buddy allocator */
1511 	struct ext4_group_info ** __rcu *s_group_info;
1512 	struct inode *s_buddy_cache;
1513 	spinlock_t s_md_lock;
1514 	unsigned short *s_mb_offsets;
1515 	unsigned int *s_mb_maxs;
1516 	unsigned int s_group_info_size;
1517 	unsigned int s_mb_free_pending;
1518 	struct list_head s_freed_data_list;	/* List of blocks to be freed
1519 						   after commit completed */
1520 
1521 	/* tunables */
1522 	unsigned long s_stripe;
1523 	unsigned int s_mb_stream_request;
1524 	unsigned int s_mb_max_to_scan;
1525 	unsigned int s_mb_min_to_scan;
1526 	unsigned int s_mb_stats;
1527 	unsigned int s_mb_order2_reqs;
1528 	unsigned int s_mb_group_prealloc;
1529 	unsigned int s_mb_max_inode_prealloc;
1530 	unsigned int s_max_dir_size_kb;
1531 	/* where last allocation was done - for stream allocation */
1532 	unsigned long s_mb_last_group;
1533 	unsigned long s_mb_last_start;
1534 	unsigned int s_mb_prefetch;
1535 	unsigned int s_mb_prefetch_limit;
1536 
1537 	/* stats for buddy allocator */
1538 	atomic_t s_bal_reqs;	/* number of reqs with len > 1 */
1539 	atomic_t s_bal_success;	/* we found long enough chunks */
1540 	atomic_t s_bal_allocated;	/* in blocks */
1541 	atomic_t s_bal_ex_scanned;	/* total extents scanned */
1542 	atomic_t s_bal_goals;	/* goal hits */
1543 	atomic_t s_bal_breaks;	/* too long searches */
1544 	atomic_t s_bal_2orders;	/* 2^order hits */
1545 	spinlock_t s_bal_lock;
1546 	unsigned long s_mb_buddies_generated;
1547 	unsigned long long s_mb_generation_time;
1548 	atomic_t s_mb_lost_chunks;
1549 	atomic_t s_mb_preallocated;
1550 	atomic_t s_mb_discarded;
1551 	atomic_t s_lock_busy;
1552 
1553 	/* locality groups */
1554 	struct ext4_locality_group __percpu *s_locality_groups;
1555 
1556 	/* for write statistics */
1557 	unsigned long s_sectors_written_start;
1558 	u64 s_kbytes_written;
1559 
1560 	/* the size of zero-out chunk */
1561 	unsigned int s_extent_max_zeroout_kb;
1562 
1563 	unsigned int s_log_groups_per_flex;
1564 	struct flex_groups * __rcu *s_flex_groups;
1565 	ext4_group_t s_flex_groups_allocated;
1566 
1567 	/* workqueue for reserved extent conversions (buffered io) */
1568 	struct workqueue_struct *rsv_conversion_wq;
1569 
1570 	/* timer for periodic error stats printing */
1571 	struct timer_list s_err_report;
1572 
1573 	/* Lazy inode table initialization info */
1574 	struct ext4_li_request *s_li_request;
1575 	/* Wait multiplier for lazy initialization thread */
1576 	unsigned int s_li_wait_mult;
1577 
1578 	/* Kernel thread for multiple mount protection */
1579 	struct task_struct *s_mmp_tsk;
1580 
1581 	/* record the last minlen when FITRIM is called. */
1582 	atomic_t s_last_trim_minblks;
1583 
1584 	/* Reference to checksum algorithm driver via cryptoapi */
1585 	struct crypto_shash *s_chksum_driver;
1586 
1587 	/* Precomputed FS UUID checksum for seeding other checksums */
1588 	__u32 s_csum_seed;
1589 
1590 	/* Reclaim extents from extent status tree */
1591 	struct shrinker s_es_shrinker;
1592 	struct list_head s_es_list;	/* List of inodes with reclaimable extents */
1593 	long s_es_nr_inode;
1594 	struct ext4_es_stats s_es_stats;
1595 	struct mb_cache *s_ea_block_cache;
1596 	struct mb_cache *s_ea_inode_cache;
1597 	spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1598 
1599 	/* Ratelimit ext4 messages. */
1600 	struct ratelimit_state s_err_ratelimit_state;
1601 	struct ratelimit_state s_warning_ratelimit_state;
1602 	struct ratelimit_state s_msg_ratelimit_state;
1603 	atomic_t s_warning_count;
1604 	atomic_t s_msg_count;
1605 
1606 	/* Encryption policy for '-o test_dummy_encryption' */
1607 	struct fscrypt_dummy_policy s_dummy_enc_policy;
1608 
1609 	/*
1610 	 * Barrier between writepages ops and changing any inode's JOURNAL_DATA
1611 	 * or EXTENTS flag.
1612 	 */
1613 	struct percpu_rw_semaphore s_writepages_rwsem;
1614 	struct dax_device *s_daxdev;
1615 #ifdef CONFIG_EXT4_DEBUG
1616 	unsigned long s_simulate_fail;
1617 #endif
1618 	/* Record the errseq of the backing block device */
1619 	errseq_t s_bdev_wb_err;
1620 	spinlock_t s_bdev_wb_lock;
1621 
1622 	/* Ext4 fast commit stuff */
1623 	atomic_t s_fc_subtid;
1624 	atomic_t s_fc_ineligible_updates;
1625 	/*
1626 	 * After commit starts, the main queue gets locked, and the further
1627 	 * updates get added in the staging queue.
1628 	 */
1629 #define FC_Q_MAIN	0
1630 #define FC_Q_STAGING	1
1631 	struct list_head s_fc_q[2];	/* Inodes staged for fast commit
1632 					 * that have data changes in them.
1633 					 */
1634 	struct list_head s_fc_dentry_q[2];	/* directory entry updates */
1635 	unsigned int s_fc_bytes;
1636 	/*
1637 	 * Main fast commit lock. This lock protects accesses to the
1638 	 * following fields:
1639 	 * ei->i_fc_list, s_fc_dentry_q, s_fc_q, s_fc_bytes, s_fc_bh.
1640 	 */
1641 	spinlock_t s_fc_lock;
1642 	struct buffer_head *s_fc_bh;
1643 	struct ext4_fc_stats s_fc_stats;
1644 	u64 s_fc_avg_commit_time;
1645 #ifdef CONFIG_EXT4_DEBUG
1646 	int s_fc_debug_max_replay;
1647 #endif
1648 	struct ext4_fc_replay_state s_fc_replay_state;
1649 };
1650 
EXT4_SB(struct super_block * sb)1651 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1652 {
1653 	return sb->s_fs_info;
1654 }
EXT4_I(struct inode * inode)1655 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1656 {
1657 	return container_of(inode, struct ext4_inode_info, vfs_inode);
1658 }
1659 
ext4_valid_inum(struct super_block * sb,unsigned long ino)1660 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1661 {
1662 	return ino == EXT4_ROOT_INO ||
1663 		(ino >= EXT4_FIRST_INO(sb) &&
1664 		 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1665 }
1666 
1667 /*
1668  * Returns: sbi->field[index]
1669  * Used to access an array element from the following sbi fields which require
1670  * rcu protection to avoid dereferencing an invalid pointer due to reassignment
1671  * - s_group_desc
1672  * - s_group_info
1673  * - s_flex_group
1674  */
1675 #define sbi_array_rcu_deref(sbi, field, index)				   \
1676 ({									   \
1677 	typeof(*((sbi)->field)) _v;					   \
1678 	rcu_read_lock();						   \
1679 	_v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index];	   \
1680 	rcu_read_unlock();						   \
1681 	_v;								   \
1682 })
1683 
1684 /*
1685  * run-time mount flags
1686  */
1687 enum {
1688 	EXT4_MF_MNTDIR_SAMPLED,
1689 	EXT4_MF_FS_ABORTED,	/* Fatal error detected */
1690 	EXT4_MF_FC_INELIGIBLE,	/* Fast commit ineligible */
1691 	EXT4_MF_FC_COMMITTING	/* File system underoing a fast
1692 				 * commit.
1693 				 */
1694 };
1695 
ext4_set_mount_flag(struct super_block * sb,int bit)1696 static inline void ext4_set_mount_flag(struct super_block *sb, int bit)
1697 {
1698 	set_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1699 }
1700 
ext4_clear_mount_flag(struct super_block * sb,int bit)1701 static inline void ext4_clear_mount_flag(struct super_block *sb, int bit)
1702 {
1703 	clear_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1704 }
1705 
ext4_test_mount_flag(struct super_block * sb,int bit)1706 static inline int ext4_test_mount_flag(struct super_block *sb, int bit)
1707 {
1708 	return test_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1709 }
1710 
1711 
1712 /*
1713  * Simulate_fail codes
1714  */
1715 #define EXT4_SIM_BBITMAP_EIO	1
1716 #define EXT4_SIM_BBITMAP_CRC	2
1717 #define EXT4_SIM_IBITMAP_EIO	3
1718 #define EXT4_SIM_IBITMAP_CRC	4
1719 #define EXT4_SIM_INODE_EIO	5
1720 #define EXT4_SIM_INODE_CRC	6
1721 #define EXT4_SIM_DIRBLOCK_EIO	7
1722 #define EXT4_SIM_DIRBLOCK_CRC	8
1723 
ext4_simulate_fail(struct super_block * sb,unsigned long code)1724 static inline bool ext4_simulate_fail(struct super_block *sb,
1725 				     unsigned long code)
1726 {
1727 #ifdef CONFIG_EXT4_DEBUG
1728 	struct ext4_sb_info *sbi = EXT4_SB(sb);
1729 
1730 	if (unlikely(sbi->s_simulate_fail == code)) {
1731 		sbi->s_simulate_fail = 0;
1732 		return true;
1733 	}
1734 #endif
1735 	return false;
1736 }
1737 
ext4_simulate_fail_bh(struct super_block * sb,struct buffer_head * bh,unsigned long code)1738 static inline void ext4_simulate_fail_bh(struct super_block *sb,
1739 					 struct buffer_head *bh,
1740 					 unsigned long code)
1741 {
1742 	if (!IS_ERR(bh) && ext4_simulate_fail(sb, code))
1743 		clear_buffer_uptodate(bh);
1744 }
1745 
1746 /*
1747  * Error number codes for s_{first,last}_error_errno
1748  *
1749  * Linux errno numbers are architecture specific, so we need to translate
1750  * them into something which is architecture independent.   We don't define
1751  * codes for all errno's; just the ones which are most likely to be the cause
1752  * of an ext4_error() call.
1753  */
1754 #define EXT4_ERR_UNKNOWN	 1
1755 #define EXT4_ERR_EIO		 2
1756 #define EXT4_ERR_ENOMEM		 3
1757 #define EXT4_ERR_EFSBADCRC	 4
1758 #define EXT4_ERR_EFSCORRUPTED	 5
1759 #define EXT4_ERR_ENOSPC		 6
1760 #define EXT4_ERR_ENOKEY		 7
1761 #define EXT4_ERR_EROFS		 8
1762 #define EXT4_ERR_EFBIG		 9
1763 #define EXT4_ERR_EEXIST		10
1764 #define EXT4_ERR_ERANGE		11
1765 #define EXT4_ERR_EOVERFLOW	12
1766 #define EXT4_ERR_EBUSY		13
1767 #define EXT4_ERR_ENOTDIR	14
1768 #define EXT4_ERR_ENOTEMPTY	15
1769 #define EXT4_ERR_ESHUTDOWN	16
1770 #define EXT4_ERR_EFAULT		17
1771 
1772 /*
1773  * Inode dynamic state flags
1774  */
1775 enum {
1776 	EXT4_STATE_JDATA,		/* journaled data exists */
1777 	EXT4_STATE_NEW,			/* inode is newly created */
1778 	EXT4_STATE_XATTR,		/* has in-inode xattrs */
1779 	EXT4_STATE_NO_EXPAND,		/* No space for expansion */
1780 	EXT4_STATE_DA_ALLOC_CLOSE,	/* Alloc DA blks on close */
1781 	EXT4_STATE_EXT_MIGRATE,		/* Inode is migrating */
1782 	EXT4_STATE_NEWENTRY,		/* File just added to dir */
1783 	EXT4_STATE_MAY_INLINE_DATA,	/* may have in-inode data */
1784 	EXT4_STATE_EXT_PRECACHED,	/* extents have been precached */
1785 	EXT4_STATE_LUSTRE_EA_INODE,	/* Lustre-style ea_inode */
1786 	EXT4_STATE_VERITY_IN_PROGRESS,	/* building fs-verity Merkle tree */
1787 	EXT4_STATE_FC_COMMITTING,	/* Fast commit ongoing */
1788 };
1789 
1790 #define EXT4_INODE_BIT_FNS(name, field, offset)				\
1791 static inline int ext4_test_inode_##name(struct inode *inode, int bit)	\
1792 {									\
1793 	return test_bit(bit + (offset), &EXT4_I(inode)->i_##field);	\
1794 }									\
1795 static inline void ext4_set_inode_##name(struct inode *inode, int bit)	\
1796 {									\
1797 	set_bit(bit + (offset), &EXT4_I(inode)->i_##field);		\
1798 }									\
1799 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1800 {									\
1801 	clear_bit(bit + (offset), &EXT4_I(inode)->i_##field);		\
1802 }
1803 
1804 /* Add these declarations here only so that these functions can be
1805  * found by name.  Otherwise, they are very hard to locate. */
1806 static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1807 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1808 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1809 EXT4_INODE_BIT_FNS(flag, flags, 0)
1810 
1811 /* Add these declarations here only so that these functions can be
1812  * found by name.  Otherwise, they are very hard to locate. */
1813 static inline int ext4_test_inode_state(struct inode *inode, int bit);
1814 static inline void ext4_set_inode_state(struct inode *inode, int bit);
1815 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1816 #if (BITS_PER_LONG < 64)
1817 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1818 
ext4_clear_state_flags(struct ext4_inode_info * ei)1819 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1820 {
1821 	(ei)->i_state_flags = 0;
1822 }
1823 #else
1824 EXT4_INODE_BIT_FNS(state, flags, 32)
1825 
ext4_clear_state_flags(struct ext4_inode_info * ei)1826 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1827 {
1828 	/* We depend on the fact that callers will set i_flags */
1829 }
1830 #endif
1831 #else
1832 /* Assume that user mode programs are passing in an ext4fs superblock, not
1833  * a kernel struct super_block.  This will allow us to call the feature-test
1834  * macros from user land. */
1835 #define EXT4_SB(sb)	(sb)
1836 #endif
1837 
ext4_verity_in_progress(struct inode * inode)1838 static inline bool ext4_verity_in_progress(struct inode *inode)
1839 {
1840 	return IS_ENABLED(CONFIG_FS_VERITY) &&
1841 	       ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS);
1842 }
1843 
1844 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1845 
1846 /*
1847  * Codes for operating systems
1848  */
1849 #define EXT4_OS_LINUX		0
1850 #define EXT4_OS_HURD		1
1851 #define EXT4_OS_MASIX		2
1852 #define EXT4_OS_FREEBSD		3
1853 #define EXT4_OS_LITES		4
1854 
1855 /*
1856  * Revision levels
1857  */
1858 #define EXT4_GOOD_OLD_REV	0	/* The good old (original) format */
1859 #define EXT4_DYNAMIC_REV	1	/* V2 format w/ dynamic inode sizes */
1860 
1861 #define EXT4_CURRENT_REV	EXT4_GOOD_OLD_REV
1862 #define EXT4_MAX_SUPP_REV	EXT4_DYNAMIC_REV
1863 
1864 #define EXT4_GOOD_OLD_INODE_SIZE 128
1865 
1866 #define EXT4_EXTRA_TIMESTAMP_MAX	(((s64)1 << 34) - 1  + S32_MIN)
1867 #define EXT4_NON_EXTRA_TIMESTAMP_MAX	S32_MAX
1868 #define EXT4_TIMESTAMP_MIN		S32_MIN
1869 
1870 /*
1871  * Feature set definitions
1872  */
1873 
1874 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC	0x0001
1875 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES	0x0002
1876 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL		0x0004
1877 #define EXT4_FEATURE_COMPAT_EXT_ATTR		0x0008
1878 #define EXT4_FEATURE_COMPAT_RESIZE_INODE	0x0010
1879 #define EXT4_FEATURE_COMPAT_DIR_INDEX		0x0020
1880 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2	0x0200
1881 /*
1882  * The reason why "FAST_COMMIT" is a compat feature is that, FS becomes
1883  * incompatible only if fast commit blocks are present in the FS. Since we
1884  * clear the journal (and thus the fast commit blocks), we don't mark FS as
1885  * incompatible. We also have a JBD2 incompat feature, which gets set when
1886  * there are fast commit blocks present in the journal.
1887  */
1888 #define EXT4_FEATURE_COMPAT_FAST_COMMIT		0x0400
1889 #define EXT4_FEATURE_COMPAT_STABLE_INODES	0x0800
1890 
1891 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER	0x0001
1892 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE	0x0002
1893 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR	0x0004
1894 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE        0x0008
1895 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM		0x0010
1896 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK	0x0020
1897 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE	0x0040
1898 #define EXT4_FEATURE_RO_COMPAT_QUOTA		0x0100
1899 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC		0x0200
1900 /*
1901  * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM).  When
1902  * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1903  * all other data structures' checksums.  However, the METADATA_CSUM and
1904  * GDT_CSUM bits are mutually exclusive.
1905  */
1906 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM	0x0400
1907 #define EXT4_FEATURE_RO_COMPAT_READONLY		0x1000
1908 #define EXT4_FEATURE_RO_COMPAT_PROJECT		0x2000
1909 #define EXT4_FEATURE_RO_COMPAT_VERITY		0x8000
1910 
1911 #define EXT4_FEATURE_INCOMPAT_COMPRESSION	0x0001
1912 #define EXT4_FEATURE_INCOMPAT_FILETYPE		0x0002
1913 #define EXT4_FEATURE_INCOMPAT_RECOVER		0x0004 /* Needs recovery */
1914 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV	0x0008 /* Journal device */
1915 #define EXT4_FEATURE_INCOMPAT_META_BG		0x0010
1916 #define EXT4_FEATURE_INCOMPAT_EXTENTS		0x0040 /* extents support */
1917 #define EXT4_FEATURE_INCOMPAT_64BIT		0x0080
1918 #define EXT4_FEATURE_INCOMPAT_MMP               0x0100
1919 #define EXT4_FEATURE_INCOMPAT_FLEX_BG		0x0200
1920 #define EXT4_FEATURE_INCOMPAT_EA_INODE		0x0400 /* EA in inode */
1921 #define EXT4_FEATURE_INCOMPAT_DIRDATA		0x1000 /* data in dirent */
1922 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED		0x2000
1923 #define EXT4_FEATURE_INCOMPAT_LARGEDIR		0x4000 /* >2GB or 3-lvl htree */
1924 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA	0x8000 /* data in inode */
1925 #define EXT4_FEATURE_INCOMPAT_ENCRYPT		0x10000
1926 #define EXT4_FEATURE_INCOMPAT_CASEFOLD		0x20000
1927 
1928 extern void ext4_update_dynamic_rev(struct super_block *sb);
1929 
1930 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
1931 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1932 { \
1933 	return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1934 		cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
1935 } \
1936 static inline void ext4_set_feature_##name(struct super_block *sb) \
1937 { \
1938 	ext4_update_dynamic_rev(sb); \
1939 	EXT4_SB(sb)->s_es->s_feature_compat |= \
1940 		cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1941 } \
1942 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1943 { \
1944 	EXT4_SB(sb)->s_es->s_feature_compat &= \
1945 		~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1946 }
1947 
1948 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
1949 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1950 { \
1951 	return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1952 		cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
1953 } \
1954 static inline void ext4_set_feature_##name(struct super_block *sb) \
1955 { \
1956 	ext4_update_dynamic_rev(sb); \
1957 	EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
1958 		cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1959 } \
1960 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1961 { \
1962 	EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
1963 		~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1964 }
1965 
1966 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
1967 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1968 { \
1969 	return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1970 		cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
1971 } \
1972 static inline void ext4_set_feature_##name(struct super_block *sb) \
1973 { \
1974 	ext4_update_dynamic_rev(sb); \
1975 	EXT4_SB(sb)->s_es->s_feature_incompat |= \
1976 		cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1977 } \
1978 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1979 { \
1980 	EXT4_SB(sb)->s_es->s_feature_incompat &= \
1981 		~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1982 }
1983 
EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc,DIR_PREALLOC)1984 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc,		DIR_PREALLOC)
1985 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes,	IMAGIC_INODES)
1986 EXT4_FEATURE_COMPAT_FUNCS(journal,		HAS_JOURNAL)
1987 EXT4_FEATURE_COMPAT_FUNCS(xattr,		EXT_ATTR)
1988 EXT4_FEATURE_COMPAT_FUNCS(resize_inode,		RESIZE_INODE)
1989 EXT4_FEATURE_COMPAT_FUNCS(dir_index,		DIR_INDEX)
1990 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2,	SPARSE_SUPER2)
1991 EXT4_FEATURE_COMPAT_FUNCS(fast_commit,		FAST_COMMIT)
1992 EXT4_FEATURE_COMPAT_FUNCS(stable_inodes,	STABLE_INODES)
1993 
1994 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super,	SPARSE_SUPER)
1995 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file,	LARGE_FILE)
1996 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir,		BTREE_DIR)
1997 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file,		HUGE_FILE)
1998 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum,		GDT_CSUM)
1999 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink,		DIR_NLINK)
2000 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize,	EXTRA_ISIZE)
2001 EXT4_FEATURE_RO_COMPAT_FUNCS(quota,		QUOTA)
2002 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc,		BIGALLOC)
2003 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum,	METADATA_CSUM)
2004 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly,		READONLY)
2005 EXT4_FEATURE_RO_COMPAT_FUNCS(project,		PROJECT)
2006 EXT4_FEATURE_RO_COMPAT_FUNCS(verity,		VERITY)
2007 
2008 EXT4_FEATURE_INCOMPAT_FUNCS(compression,	COMPRESSION)
2009 EXT4_FEATURE_INCOMPAT_FUNCS(filetype,		FILETYPE)
2010 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery,	RECOVER)
2011 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev,	JOURNAL_DEV)
2012 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg,		META_BG)
2013 EXT4_FEATURE_INCOMPAT_FUNCS(extents,		EXTENTS)
2014 EXT4_FEATURE_INCOMPAT_FUNCS(64bit,		64BIT)
2015 EXT4_FEATURE_INCOMPAT_FUNCS(mmp,		MMP)
2016 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg,		FLEX_BG)
2017 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode,		EA_INODE)
2018 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata,		DIRDATA)
2019 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed,		CSUM_SEED)
2020 EXT4_FEATURE_INCOMPAT_FUNCS(largedir,		LARGEDIR)
2021 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data,	INLINE_DATA)
2022 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt,		ENCRYPT)
2023 EXT4_FEATURE_INCOMPAT_FUNCS(casefold,		CASEFOLD)
2024 
2025 #define EXT2_FEATURE_COMPAT_SUPP	EXT4_FEATURE_COMPAT_EXT_ATTR
2026 #define EXT2_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
2027 					 EXT4_FEATURE_INCOMPAT_META_BG)
2028 #define EXT2_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2029 					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2030 					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
2031 
2032 #define EXT3_FEATURE_COMPAT_SUPP	EXT4_FEATURE_COMPAT_EXT_ATTR
2033 #define EXT3_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
2034 					 EXT4_FEATURE_INCOMPAT_RECOVER| \
2035 					 EXT4_FEATURE_INCOMPAT_META_BG)
2036 #define EXT3_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2037 					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2038 					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
2039 
2040 #define EXT4_FEATURE_COMPAT_SUPP	EXT4_FEATURE_COMPAT_EXT_ATTR
2041 #define EXT4_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
2042 					 EXT4_FEATURE_INCOMPAT_RECOVER| \
2043 					 EXT4_FEATURE_INCOMPAT_META_BG| \
2044 					 EXT4_FEATURE_INCOMPAT_EXTENTS| \
2045 					 EXT4_FEATURE_INCOMPAT_64BIT| \
2046 					 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
2047 					 EXT4_FEATURE_INCOMPAT_EA_INODE| \
2048 					 EXT4_FEATURE_INCOMPAT_MMP | \
2049 					 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
2050 					 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
2051 					 EXT4_FEATURE_INCOMPAT_CASEFOLD | \
2052 					 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \
2053 					 EXT4_FEATURE_INCOMPAT_LARGEDIR)
2054 #define EXT4_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2055 					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2056 					 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
2057 					 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
2058 					 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
2059 					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
2060 					 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
2061 					 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
2062 					 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
2063 					 EXT4_FEATURE_RO_COMPAT_QUOTA |\
2064 					 EXT4_FEATURE_RO_COMPAT_PROJECT |\
2065 					 EXT4_FEATURE_RO_COMPAT_VERITY)
2066 
2067 #define EXTN_FEATURE_FUNCS(ver) \
2068 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
2069 { \
2070 	return ((EXT4_SB(sb)->s_es->s_feature_compat & \
2071 		cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
2072 } \
2073 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
2074 { \
2075 	return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
2076 		cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
2077 } \
2078 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
2079 { \
2080 	return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
2081 		cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
2082 }
2083 
2084 EXTN_FEATURE_FUNCS(2)
2085 EXTN_FEATURE_FUNCS(3)
2086 EXTN_FEATURE_FUNCS(4)
2087 
2088 static inline bool ext4_has_compat_features(struct super_block *sb)
2089 {
2090 	return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
2091 }
ext4_has_ro_compat_features(struct super_block * sb)2092 static inline bool ext4_has_ro_compat_features(struct super_block *sb)
2093 {
2094 	return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
2095 }
ext4_has_incompat_features(struct super_block * sb)2096 static inline bool ext4_has_incompat_features(struct super_block *sb)
2097 {
2098 	return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
2099 }
2100 
2101 /*
2102  * Superblock flags
2103  */
2104 #define EXT4_FLAGS_RESIZING	0
2105 #define EXT4_FLAGS_SHUTDOWN	1
2106 #define EXT4_FLAGS_BDEV_IS_DAX	2
2107 
ext4_forced_shutdown(struct ext4_sb_info * sbi)2108 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi)
2109 {
2110 	return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags);
2111 }
2112 
2113 
2114 /*
2115  * Default values for user and/or group using reserved blocks
2116  */
2117 #define	EXT4_DEF_RESUID		0
2118 #define	EXT4_DEF_RESGID		0
2119 
2120 /*
2121  * Default project ID
2122  */
2123 #define	EXT4_DEF_PROJID		0
2124 
2125 #define EXT4_DEF_INODE_READAHEAD_BLKS	32
2126 
2127 /*
2128  * Default mount options
2129  */
2130 #define EXT4_DEFM_DEBUG		0x0001
2131 #define EXT4_DEFM_BSDGROUPS	0x0002
2132 #define EXT4_DEFM_XATTR_USER	0x0004
2133 #define EXT4_DEFM_ACL		0x0008
2134 #define EXT4_DEFM_UID16		0x0010
2135 #define EXT4_DEFM_JMODE		0x0060
2136 #define EXT4_DEFM_JMODE_DATA	0x0020
2137 #define EXT4_DEFM_JMODE_ORDERED	0x0040
2138 #define EXT4_DEFM_JMODE_WBACK	0x0060
2139 #define EXT4_DEFM_NOBARRIER	0x0100
2140 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
2141 #define EXT4_DEFM_DISCARD	0x0400
2142 #define EXT4_DEFM_NODELALLOC	0x0800
2143 
2144 /*
2145  * Default journal batch times
2146  */
2147 #define EXT4_DEF_MIN_BATCH_TIME	0
2148 #define EXT4_DEF_MAX_BATCH_TIME	15000 /* 15ms */
2149 
2150 /*
2151  * Minimum number of groups in a flexgroup before we separate out
2152  * directories into the first block group of a flexgroup
2153  */
2154 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME	4
2155 
2156 /*
2157  * Structure of a directory entry
2158  */
2159 #define EXT4_NAME_LEN 255
2160 
2161 struct ext4_dir_entry {
2162 	__le32	inode;			/* Inode number */
2163 	__le16	rec_len;		/* Directory entry length */
2164 	__le16	name_len;		/* Name length */
2165 	char	name[EXT4_NAME_LEN];	/* File name */
2166 };
2167 
2168 /*
2169  * The new version of the directory entry.  Since EXT4 structures are
2170  * stored in intel byte order, and the name_len field could never be
2171  * bigger than 255 chars, it's safe to reclaim the extra byte for the
2172  * file_type field.
2173  */
2174 struct ext4_dir_entry_2 {
2175 	__le32	inode;			/* Inode number */
2176 	__le16	rec_len;		/* Directory entry length */
2177 	__u8	name_len;		/* Name length */
2178 	__u8	file_type;		/* See file type macros EXT4_FT_* below */
2179 	char	name[EXT4_NAME_LEN];	/* File name */
2180 };
2181 
2182 /*
2183  * This is a bogus directory entry at the end of each leaf block that
2184  * records checksums.
2185  */
2186 struct ext4_dir_entry_tail {
2187 	__le32	det_reserved_zero1;	/* Pretend to be unused */
2188 	__le16	det_rec_len;		/* 12 */
2189 	__u8	det_reserved_zero2;	/* Zero name length */
2190 	__u8	det_reserved_ft;	/* 0xDE, fake file type */
2191 	__le32	det_checksum;		/* crc32c(uuid+inum+dirblock) */
2192 };
2193 
2194 #define EXT4_DIRENT_TAIL(block, blocksize) \
2195 	((struct ext4_dir_entry_tail *)(((void *)(block)) + \
2196 					((blocksize) - \
2197 					 sizeof(struct ext4_dir_entry_tail))))
2198 
2199 /*
2200  * Ext4 directory file types.  Only the low 3 bits are used.  The
2201  * other bits are reserved for now.
2202  */
2203 #define EXT4_FT_UNKNOWN		0
2204 #define EXT4_FT_REG_FILE	1
2205 #define EXT4_FT_DIR		2
2206 #define EXT4_FT_CHRDEV		3
2207 #define EXT4_FT_BLKDEV		4
2208 #define EXT4_FT_FIFO		5
2209 #define EXT4_FT_SOCK		6
2210 #define EXT4_FT_SYMLINK		7
2211 
2212 #define EXT4_FT_MAX		8
2213 
2214 #define EXT4_FT_DIR_CSUM	0xDE
2215 
2216 /*
2217  * EXT4_DIR_PAD defines the directory entries boundaries
2218  *
2219  * NOTE: It must be a multiple of 4
2220  */
2221 #define EXT4_DIR_PAD			4
2222 #define EXT4_DIR_ROUND			(EXT4_DIR_PAD - 1)
2223 #define EXT4_DIR_REC_LEN(name_len)	(((name_len) + 8 + EXT4_DIR_ROUND) & \
2224 					 ~EXT4_DIR_ROUND)
2225 #define EXT4_MAX_REC_LEN		((1<<16)-1)
2226 
2227 /*
2228  * If we ever get support for fs block sizes > page_size, we'll need
2229  * to remove the #if statements in the next two functions...
2230  */
2231 static inline unsigned int
ext4_rec_len_from_disk(__le16 dlen,unsigned blocksize)2232 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
2233 {
2234 	unsigned len = le16_to_cpu(dlen);
2235 
2236 #if (PAGE_SIZE >= 65536)
2237 	if (len == EXT4_MAX_REC_LEN || len == 0)
2238 		return blocksize;
2239 	return (len & 65532) | ((len & 3) << 16);
2240 #else
2241 	return len;
2242 #endif
2243 }
2244 
ext4_rec_len_to_disk(unsigned len,unsigned blocksize)2245 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
2246 {
2247 	if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
2248 		BUG();
2249 #if (PAGE_SIZE >= 65536)
2250 	if (len < 65536)
2251 		return cpu_to_le16(len);
2252 	if (len == blocksize) {
2253 		if (blocksize == 65536)
2254 			return cpu_to_le16(EXT4_MAX_REC_LEN);
2255 		else
2256 			return cpu_to_le16(0);
2257 	}
2258 	return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
2259 #else
2260 	return cpu_to_le16(len);
2261 #endif
2262 }
2263 
2264 /*
2265  * Hash Tree Directory indexing
2266  * (c) Daniel Phillips, 2001
2267  */
2268 
2269 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
2270 		    ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2271 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \
2272 		    !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir)))
2273 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2274 
2275 /* Legal values for the dx_root hash_version field: */
2276 
2277 #define DX_HASH_LEGACY			0
2278 #define DX_HASH_HALF_MD4		1
2279 #define DX_HASH_TEA			2
2280 #define DX_HASH_LEGACY_UNSIGNED		3
2281 #define DX_HASH_HALF_MD4_UNSIGNED	4
2282 #define DX_HASH_TEA_UNSIGNED		5
2283 
ext4_chksum(struct ext4_sb_info * sbi,u32 crc,const void * address,unsigned int length)2284 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2285 			      const void *address, unsigned int length)
2286 {
2287 	struct {
2288 		struct shash_desc shash;
2289 		char ctx[4];
2290 	} desc;
2291 
2292 	BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2293 
2294 	desc.shash.tfm = sbi->s_chksum_driver;
2295 	*(u32 *)desc.ctx = crc;
2296 
2297 	BUG_ON(crypto_shash_update(&desc.shash, address, length));
2298 
2299 	return *(u32 *)desc.ctx;
2300 }
2301 
2302 #ifdef __KERNEL__
2303 
2304 /* hash info structure used by the directory hash */
2305 struct dx_hash_info
2306 {
2307 	u32		hash;
2308 	u32		minor_hash;
2309 	int		hash_version;
2310 	u32		*seed;
2311 };
2312 
2313 
2314 /* 32 and 64 bit signed EOF for dx directories */
2315 #define EXT4_HTREE_EOF_32BIT   ((1UL  << (32 - 1)) - 1)
2316 #define EXT4_HTREE_EOF_64BIT   ((1ULL << (64 - 1)) - 1)
2317 
2318 
2319 /*
2320  * Control parameters used by ext4_htree_next_block
2321  */
2322 #define HASH_NB_ALWAYS		1
2323 
2324 struct ext4_filename {
2325 	const struct qstr *usr_fname;
2326 	struct fscrypt_str disk_name;
2327 	struct dx_hash_info hinfo;
2328 #ifdef CONFIG_FS_ENCRYPTION
2329 	struct fscrypt_str crypto_buf;
2330 #endif
2331 #ifdef CONFIG_UNICODE
2332 	struct fscrypt_str cf_name;
2333 #endif
2334 };
2335 
2336 #define fname_name(p) ((p)->disk_name.name)
2337 #define fname_len(p)  ((p)->disk_name.len)
2338 
2339 /*
2340  * Describe an inode's exact location on disk and in memory
2341  */
2342 struct ext4_iloc
2343 {
2344 	struct buffer_head *bh;
2345 	unsigned long offset;
2346 	ext4_group_t block_group;
2347 };
2348 
ext4_raw_inode(struct ext4_iloc * iloc)2349 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2350 {
2351 	return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2352 }
2353 
ext4_is_quota_file(struct inode * inode)2354 static inline bool ext4_is_quota_file(struct inode *inode)
2355 {
2356 	return IS_NOQUOTA(inode) &&
2357 	       !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL);
2358 }
2359 
2360 /*
2361  * This structure is stuffed into the struct file's private_data field
2362  * for directories.  It is where we put information so that we can do
2363  * readdir operations in hash tree order.
2364  */
2365 struct dir_private_info {
2366 	struct rb_root	root;
2367 	struct rb_node	*curr_node;
2368 	struct fname	*extra_fname;
2369 	loff_t		last_pos;
2370 	__u32		curr_hash;
2371 	__u32		curr_minor_hash;
2372 	__u32		next_hash;
2373 };
2374 
2375 /* calculate the first block number of the group */
2376 static inline ext4_fsblk_t
ext4_group_first_block_no(struct super_block * sb,ext4_group_t group_no)2377 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2378 {
2379 	return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2380 		le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2381 }
2382 
2383 /*
2384  * Special error return code only used by dx_probe() and its callers.
2385  */
2386 #define ERR_BAD_DX_DIR	(-(MAX_ERRNO - 1))
2387 
2388 /* htree levels for ext4 */
2389 #define	EXT4_HTREE_LEVEL_COMPAT	2
2390 #define	EXT4_HTREE_LEVEL	3
2391 
ext4_dir_htree_level(struct super_block * sb)2392 static inline int ext4_dir_htree_level(struct super_block *sb)
2393 {
2394 	return ext4_has_feature_largedir(sb) ?
2395 		EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT;
2396 }
2397 
2398 /*
2399  * Timeout and state flag for lazy initialization inode thread.
2400  */
2401 #define EXT4_DEF_LI_WAIT_MULT			10
2402 #define EXT4_DEF_LI_MAX_START_DELAY		5
2403 #define EXT4_LAZYINIT_QUIT			0x0001
2404 #define EXT4_LAZYINIT_RUNNING			0x0002
2405 
2406 /*
2407  * Lazy inode table initialization info
2408  */
2409 struct ext4_lazy_init {
2410 	unsigned long		li_state;
2411 	struct list_head	li_request_list;
2412 	struct mutex		li_list_mtx;
2413 };
2414 
2415 enum ext4_li_mode {
2416 	EXT4_LI_MODE_PREFETCH_BBITMAP,
2417 	EXT4_LI_MODE_ITABLE,
2418 };
2419 
2420 struct ext4_li_request {
2421 	struct super_block	*lr_super;
2422 	enum ext4_li_mode	lr_mode;
2423 	ext4_group_t		lr_first_not_zeroed;
2424 	ext4_group_t		lr_next_group;
2425 	struct list_head	lr_request;
2426 	unsigned long		lr_next_sched;
2427 	unsigned long		lr_timeout;
2428 };
2429 
2430 struct ext4_features {
2431 	struct kobject f_kobj;
2432 	struct completion f_kobj_unregister;
2433 };
2434 
2435 /*
2436  * This structure will be used for multiple mount protection. It will be
2437  * written into the block number saved in the s_mmp_block field in the
2438  * superblock. Programs that check MMP should assume that if
2439  * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2440  * to use the filesystem, regardless of how old the timestamp is.
2441  */
2442 #define EXT4_MMP_MAGIC     0x004D4D50U /* ASCII for MMP */
2443 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2444 #define EXT4_MMP_SEQ_FSCK  0xE24D4D50U /* mmp_seq value when being fscked */
2445 #define EXT4_MMP_SEQ_MAX   0xE24D4D4FU /* maximum valid mmp_seq value */
2446 
2447 struct mmp_struct {
2448 	__le32	mmp_magic;		/* Magic number for MMP */
2449 	__le32	mmp_seq;		/* Sequence no. updated periodically */
2450 
2451 	/*
2452 	 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2453 	 * purposes and do not affect the correctness of the algorithm
2454 	 */
2455 	__le64	mmp_time;		/* Time last updated */
2456 	char	mmp_nodename[64];	/* Node which last updated MMP block */
2457 	char	mmp_bdevname[32];	/* Bdev which last updated MMP block */
2458 
2459 	/*
2460 	 * mmp_check_interval is used to verify if the MMP block has been
2461 	 * updated on the block device. The value is updated based on the
2462 	 * maximum time to write the MMP block during an update cycle.
2463 	 */
2464 	__le16	mmp_check_interval;
2465 
2466 	__le16	mmp_pad1;
2467 	__le32	mmp_pad2[226];
2468 	__le32	mmp_checksum;		/* crc32c(uuid+mmp_block) */
2469 };
2470 
2471 /* arguments passed to the mmp thread */
2472 struct mmpd_data {
2473 	struct buffer_head *bh; /* bh from initial read_mmp_block() */
2474 	struct super_block *sb;  /* super block of the fs */
2475 };
2476 
2477 /*
2478  * Check interval multiplier
2479  * The MMP block is written every update interval and initially checked every
2480  * update interval x the multiplier (the value is then adapted based on the
2481  * write latency). The reason is that writes can be delayed under load and we
2482  * don't want readers to incorrectly assume that the filesystem is no longer
2483  * in use.
2484  */
2485 #define EXT4_MMP_CHECK_MULT		2UL
2486 
2487 /*
2488  * Minimum interval for MMP checking in seconds.
2489  */
2490 #define EXT4_MMP_MIN_CHECK_INTERVAL	5UL
2491 
2492 /*
2493  * Maximum interval for MMP checking in seconds.
2494  */
2495 #define EXT4_MMP_MAX_CHECK_INTERVAL	300UL
2496 
2497 /*
2498  * Function prototypes
2499  */
2500 
2501 /*
2502  * Ok, these declarations are also in <linux/kernel.h> but none of the
2503  * ext4 source programs needs to include it so they are duplicated here.
2504  */
2505 # define NORET_TYPE	/**/
2506 # define ATTRIB_NORET	__attribute__((noreturn))
2507 # define NORET_AND	noreturn,
2508 
2509 /* bitmap.c */
2510 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2511 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2512 				struct ext4_group_desc *gdp,
2513 				struct buffer_head *bh, int sz);
2514 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2515 				  struct ext4_group_desc *gdp,
2516 				  struct buffer_head *bh, int sz);
2517 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2518 				struct ext4_group_desc *gdp,
2519 				struct buffer_head *bh);
2520 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2521 				  struct ext4_group_desc *gdp,
2522 				  struct buffer_head *bh);
2523 
2524 /* balloc.c */
2525 extern void ext4_get_group_no_and_offset(struct super_block *sb,
2526 					 ext4_fsblk_t blocknr,
2527 					 ext4_group_t *blockgrpp,
2528 					 ext4_grpblk_t *offsetp);
2529 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2530 					  ext4_fsblk_t block);
2531 
2532 extern unsigned int ext4_block_group(struct super_block *sb,
2533 			ext4_fsblk_t blocknr);
2534 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2535 			ext4_fsblk_t blocknr);
2536 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2537 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2538 			ext4_group_t group);
2539 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2540 					 ext4_fsblk_t goal,
2541 					 unsigned int flags,
2542 					 unsigned long *count,
2543 					 int *errp);
2544 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2545 				    s64 nclusters, unsigned int flags);
2546 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2547 extern void ext4_check_blocks_bitmap(struct super_block *);
2548 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2549 						    ext4_group_t block_group,
2550 						    struct buffer_head ** bh);
2551 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2552 
2553 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2554 						ext4_group_t block_group,
2555 						bool ignore_locked);
2556 extern int ext4_wait_block_bitmap(struct super_block *sb,
2557 				  ext4_group_t block_group,
2558 				  struct buffer_head *bh);
2559 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2560 						  ext4_group_t block_group);
2561 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2562 					      ext4_group_t block_group,
2563 					      struct ext4_group_desc *gdp);
2564 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2565 
2566 #ifdef CONFIG_UNICODE
2567 extern void ext4_fname_setup_ci_filename(struct inode *dir,
2568 					 const struct qstr *iname,
2569 					 struct fscrypt_str *fname);
2570 #endif
2571 
2572 #ifdef CONFIG_FS_ENCRYPTION
ext4_fname_from_fscrypt_name(struct ext4_filename * dst,const struct fscrypt_name * src)2573 static inline void ext4_fname_from_fscrypt_name(struct ext4_filename *dst,
2574 						const struct fscrypt_name *src)
2575 {
2576 	memset(dst, 0, sizeof(*dst));
2577 
2578 	dst->usr_fname = src->usr_fname;
2579 	dst->disk_name = src->disk_name;
2580 	dst->hinfo.hash = src->hash;
2581 	dst->hinfo.minor_hash = src->minor_hash;
2582 	dst->crypto_buf = src->crypto_buf;
2583 }
2584 
ext4_fname_setup_filename(struct inode * dir,const struct qstr * iname,int lookup,struct ext4_filename * fname)2585 static inline int ext4_fname_setup_filename(struct inode *dir,
2586 					    const struct qstr *iname,
2587 					    int lookup,
2588 					    struct ext4_filename *fname)
2589 {
2590 	struct fscrypt_name name;
2591 	int err;
2592 
2593 	err = fscrypt_setup_filename(dir, iname, lookup, &name);
2594 	if (err)
2595 		return err;
2596 
2597 	ext4_fname_from_fscrypt_name(fname, &name);
2598 
2599 #ifdef CONFIG_UNICODE
2600 	ext4_fname_setup_ci_filename(dir, iname, &fname->cf_name);
2601 #endif
2602 	return 0;
2603 }
2604 
ext4_fname_prepare_lookup(struct inode * dir,struct dentry * dentry,struct ext4_filename * fname)2605 static inline int ext4_fname_prepare_lookup(struct inode *dir,
2606 					    struct dentry *dentry,
2607 					    struct ext4_filename *fname)
2608 {
2609 	struct fscrypt_name name;
2610 	int err;
2611 
2612 	err = fscrypt_prepare_lookup(dir, dentry, &name);
2613 	if (err)
2614 		return err;
2615 
2616 	ext4_fname_from_fscrypt_name(fname, &name);
2617 
2618 #ifdef CONFIG_UNICODE
2619 	ext4_fname_setup_ci_filename(dir, &dentry->d_name, &fname->cf_name);
2620 #endif
2621 	return 0;
2622 }
2623 
ext4_fname_free_filename(struct ext4_filename * fname)2624 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2625 {
2626 	struct fscrypt_name name;
2627 
2628 	name.crypto_buf = fname->crypto_buf;
2629 	fscrypt_free_filename(&name);
2630 
2631 	fname->crypto_buf.name = NULL;
2632 	fname->usr_fname = NULL;
2633 	fname->disk_name.name = NULL;
2634 
2635 #ifdef CONFIG_UNICODE
2636 	kfree(fname->cf_name.name);
2637 	fname->cf_name.name = NULL;
2638 #endif
2639 }
2640 #else /* !CONFIG_FS_ENCRYPTION */
ext4_fname_setup_filename(struct inode * dir,const struct qstr * iname,int lookup,struct ext4_filename * fname)2641 static inline int ext4_fname_setup_filename(struct inode *dir,
2642 					    const struct qstr *iname,
2643 					    int lookup,
2644 					    struct ext4_filename *fname)
2645 {
2646 	fname->usr_fname = iname;
2647 	fname->disk_name.name = (unsigned char *) iname->name;
2648 	fname->disk_name.len = iname->len;
2649 
2650 #ifdef CONFIG_UNICODE
2651 	ext4_fname_setup_ci_filename(dir, iname, &fname->cf_name);
2652 #endif
2653 
2654 	return 0;
2655 }
2656 
ext4_fname_prepare_lookup(struct inode * dir,struct dentry * dentry,struct ext4_filename * fname)2657 static inline int ext4_fname_prepare_lookup(struct inode *dir,
2658 					    struct dentry *dentry,
2659 					    struct ext4_filename *fname)
2660 {
2661 	return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname);
2662 }
2663 
ext4_fname_free_filename(struct ext4_filename * fname)2664 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2665 {
2666 #ifdef CONFIG_UNICODE
2667 	kfree(fname->cf_name.name);
2668 	fname->cf_name.name = NULL;
2669 #endif
2670 }
2671 #endif /* !CONFIG_FS_ENCRYPTION */
2672 
2673 /* dir.c */
2674 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2675 				  struct file *,
2676 				  struct ext4_dir_entry_2 *,
2677 				  struct buffer_head *, char *, int,
2678 				  unsigned int);
2679 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset)	\
2680 	unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2681 					(de), (bh), (buf), (size), (offset)))
2682 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2683 				__u32 minor_hash,
2684 				struct ext4_dir_entry_2 *dirent,
2685 				struct fscrypt_str *ent_name);
2686 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2687 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2688 			     struct buffer_head *bh,
2689 			     void *buf, int buf_size,
2690 			     struct ext4_filename *fname,
2691 			     struct ext4_dir_entry_2 **dest_de);
2692 void ext4_insert_dentry(struct inode *inode,
2693 			struct ext4_dir_entry_2 *de,
2694 			int buf_size,
2695 			struct ext4_filename *fname);
ext4_update_dx_flag(struct inode * inode)2696 static inline void ext4_update_dx_flag(struct inode *inode)
2697 {
2698 	if (!ext4_has_feature_dir_index(inode->i_sb) &&
2699 	    ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) {
2700 		/* ext4_iget() should have caught this... */
2701 		WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb));
2702 		ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2703 	}
2704 }
2705 static const unsigned char ext4_filetype_table[] = {
2706 	DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2707 };
2708 
get_dtype(struct super_block * sb,int filetype)2709 static inline  unsigned char get_dtype(struct super_block *sb, int filetype)
2710 {
2711 	if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2712 		return DT_UNKNOWN;
2713 
2714 	return ext4_filetype_table[filetype];
2715 }
2716 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2717 			     void *buf, int buf_size);
2718 
2719 /* fsync.c */
2720 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2721 
2722 /* hash.c */
2723 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len,
2724 			  struct dx_hash_info *hinfo);
2725 
2726 /* ialloc.c */
2727 extern int ext4_mark_inode_used(struct super_block *sb, int ino);
2728 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2729 				      const struct qstr *qstr, __u32 goal,
2730 				      uid_t *owner, __u32 i_flags,
2731 				      int handle_type, unsigned int line_no,
2732 				      int nblocks);
2733 
2734 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \
2735 	__ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2736 			 i_flags, 0, 0, 0)
2737 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2738 				    type, nblocks)		    \
2739 	__ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2740 			 0, (type), __LINE__, (nblocks))
2741 
2742 
2743 extern void ext4_free_inode(handle_t *, struct inode *);
2744 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2745 extern unsigned long ext4_count_free_inodes(struct super_block *);
2746 extern unsigned long ext4_count_dirs(struct super_block *);
2747 extern void ext4_check_inodes_bitmap(struct super_block *);
2748 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2749 extern int ext4_init_inode_table(struct super_block *sb,
2750 				 ext4_group_t group, int barrier);
2751 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2752 
2753 /* fast_commit.c */
2754 int ext4_fc_info_show(struct seq_file *seq, void *v);
2755 void ext4_fc_init(struct super_block *sb, journal_t *journal);
2756 void ext4_fc_init_inode(struct inode *inode);
2757 void ext4_fc_track_range(handle_t *handle, struct inode *inode, ext4_lblk_t start,
2758 			 ext4_lblk_t end);
2759 void __ext4_fc_track_unlink(handle_t *handle, struct inode *inode,
2760 	struct dentry *dentry);
2761 void __ext4_fc_track_link(handle_t *handle, struct inode *inode,
2762 	struct dentry *dentry);
2763 void ext4_fc_track_unlink(handle_t *handle, struct dentry *dentry);
2764 void ext4_fc_track_link(handle_t *handle, struct dentry *dentry);
2765 void ext4_fc_track_create(handle_t *handle, struct dentry *dentry);
2766 void ext4_fc_track_inode(handle_t *handle, struct inode *inode);
2767 void ext4_fc_mark_ineligible(struct super_block *sb, int reason);
2768 void ext4_fc_start_ineligible(struct super_block *sb, int reason);
2769 void ext4_fc_stop_ineligible(struct super_block *sb);
2770 void ext4_fc_start_update(struct inode *inode);
2771 void ext4_fc_stop_update(struct inode *inode);
2772 void ext4_fc_del(struct inode *inode);
2773 bool ext4_fc_replay_check_excluded(struct super_block *sb, ext4_fsblk_t block);
2774 void ext4_fc_replay_cleanup(struct super_block *sb);
2775 int ext4_fc_commit(journal_t *journal, tid_t commit_tid);
2776 int __init ext4_fc_init_dentry_cache(void);
2777 
2778 /* mballoc.c */
2779 extern const struct seq_operations ext4_mb_seq_groups_ops;
2780 extern long ext4_mb_stats;
2781 extern long ext4_mb_max_to_scan;
2782 extern int ext4_mb_init(struct super_block *);
2783 extern int ext4_mb_release(struct super_block *);
2784 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2785 				struct ext4_allocation_request *, int *);
2786 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2787 extern void ext4_discard_preallocations(struct inode *, unsigned int);
2788 extern int __init ext4_init_mballoc(void);
2789 extern void ext4_exit_mballoc(void);
2790 extern ext4_group_t ext4_mb_prefetch(struct super_block *sb,
2791 				     ext4_group_t group,
2792 				     unsigned int nr, int *cnt);
2793 extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group,
2794 				  unsigned int nr);
2795 
2796 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2797 			     struct buffer_head *bh, ext4_fsblk_t block,
2798 			     unsigned long count, int flags);
2799 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2800 				   ext4_group_t ngroups);
2801 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2802 		ext4_group_t i, struct ext4_group_desc *desc);
2803 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2804 				ext4_fsblk_t block, unsigned long count);
2805 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2806 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid);
2807 extern void ext4_mb_mark_bb(struct super_block *sb, ext4_fsblk_t block,
2808 		       int len, int state);
2809 
2810 /* inode.c */
2811 void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw,
2812 			 struct ext4_inode_info *ei);
2813 int ext4_inode_is_fast_symlink(struct inode *inode);
2814 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2815 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2816 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count,
2817 		     bool wait, struct buffer_head **bhs);
2818 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2819 			     struct buffer_head *bh_result, int create);
2820 int ext4_get_block(struct inode *inode, sector_t iblock,
2821 		   struct buffer_head *bh_result, int create);
2822 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2823 			   struct buffer_head *bh, int create);
2824 int ext4_walk_page_buffers(handle_t *handle,
2825 			   struct buffer_head *head,
2826 			   unsigned from,
2827 			   unsigned to,
2828 			   int *partial,
2829 			   int (*fn)(handle_t *handle,
2830 				     struct buffer_head *bh));
2831 int do_journal_get_write_access(handle_t *handle,
2832 				struct buffer_head *bh);
2833 #define FALL_BACK_TO_NONDELALLOC 1
2834 #define CONVERT_INLINE_DATA	 2
2835 
2836 typedef enum {
2837 	EXT4_IGET_NORMAL =	0,
2838 	EXT4_IGET_SPECIAL =	0x0001, /* OK to iget a system inode */
2839 	EXT4_IGET_HANDLE = 	0x0002	/* Inode # is from a handle */
2840 } ext4_iget_flags;
2841 
2842 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
2843 				 ext4_iget_flags flags, const char *function,
2844 				 unsigned int line);
2845 
2846 #define ext4_iget(sb, ino, flags) \
2847 	__ext4_iget((sb), (ino), (flags), __func__, __LINE__)
2848 
2849 extern int  ext4_write_inode(struct inode *, struct writeback_control *);
2850 extern int  ext4_setattr(struct dentry *, struct iattr *);
2851 extern int  ext4_getattr(const struct path *, struct kstat *, u32, unsigned int);
2852 extern void ext4_evict_inode(struct inode *);
2853 extern void ext4_clear_inode(struct inode *);
2854 extern int  ext4_file_getattr(const struct path *, struct kstat *, u32, unsigned int);
2855 extern int  ext4_sync_inode(handle_t *, struct inode *);
2856 extern void ext4_dirty_inode(struct inode *, int);
2857 extern int ext4_change_inode_journal_flag(struct inode *, int);
2858 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2859 extern int ext4_get_fc_inode_loc(struct super_block *sb, unsigned long ino,
2860 			  struct ext4_iloc *iloc);
2861 extern int ext4_inode_attach_jinode(struct inode *inode);
2862 extern int ext4_can_truncate(struct inode *inode);
2863 extern int ext4_truncate(struct inode *);
2864 extern int ext4_break_layouts(struct inode *);
2865 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2866 extern void ext4_set_inode_flags(struct inode *, bool init);
2867 extern int ext4_alloc_da_blocks(struct inode *inode);
2868 extern void ext4_set_aops(struct inode *inode);
2869 extern int ext4_writepage_trans_blocks(struct inode *);
2870 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2871 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2872 			     loff_t lstart, loff_t lend);
2873 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf);
2874 extern vm_fault_t ext4_filemap_fault(struct vm_fault *vmf);
2875 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2876 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
2877 extern void ext4_da_release_space(struct inode *inode, int to_free);
2878 extern void ext4_da_update_reserve_space(struct inode *inode,
2879 					int used, int quota_claim);
2880 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
2881 			      ext4_fsblk_t pblk, ext4_lblk_t len);
2882 
2883 /* indirect.c */
2884 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2885 				struct ext4_map_blocks *map, int flags);
2886 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2887 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2888 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2889 				 ext4_lblk_t start, ext4_lblk_t end);
2890 
2891 /* ioctl.c */
2892 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2893 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2894 extern void ext4_reset_inode_seed(struct inode *inode);
2895 
2896 /* migrate.c */
2897 extern int ext4_ext_migrate(struct inode *);
2898 extern int ext4_ind_migrate(struct inode *inode);
2899 
2900 /* namei.c */
2901 extern int ext4_init_new_dir(handle_t *handle, struct inode *dir,
2902 			     struct inode *inode);
2903 extern int ext4_dirblock_csum_verify(struct inode *inode,
2904 				     struct buffer_head *bh);
2905 extern int ext4_orphan_add(handle_t *, struct inode *);
2906 extern int ext4_orphan_del(handle_t *, struct inode *);
2907 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2908 				__u32 start_minor_hash, __u32 *next_hash);
2909 extern int ext4_search_dir(struct buffer_head *bh,
2910 			   char *search_buf,
2911 			   int buf_size,
2912 			   struct inode *dir,
2913 			   struct ext4_filename *fname,
2914 			   unsigned int offset,
2915 			   struct ext4_dir_entry_2 **res_dir);
2916 extern int ext4_generic_delete_entry(struct inode *dir,
2917 				     struct ext4_dir_entry_2 *de_del,
2918 				     struct buffer_head *bh,
2919 				     void *entry_buf,
2920 				     int buf_size,
2921 				     int csum_size);
2922 extern bool ext4_empty_dir(struct inode *inode);
2923 
2924 /* resize.c */
2925 extern void ext4_kvfree_array_rcu(void *to_free);
2926 extern int ext4_group_add(struct super_block *sb,
2927 				struct ext4_new_group_data *input);
2928 extern int ext4_group_extend(struct super_block *sb,
2929 				struct ext4_super_block *es,
2930 				ext4_fsblk_t n_blocks_count);
2931 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2932 
2933 /* super.c */
2934 extern struct buffer_head *ext4_sb_bread(struct super_block *sb,
2935 					 sector_t block, int op_flags);
2936 extern struct buffer_head *ext4_sb_bread_unmovable(struct super_block *sb,
2937 						   sector_t block);
2938 extern void ext4_read_bh_nowait(struct buffer_head *bh, int op_flags,
2939 				bh_end_io_t *end_io);
2940 extern int ext4_read_bh(struct buffer_head *bh, int op_flags,
2941 			bh_end_io_t *end_io);
2942 extern int ext4_read_bh_lock(struct buffer_head *bh, int op_flags, bool wait);
2943 extern void ext4_sb_breadahead_unmovable(struct super_block *sb, sector_t block);
2944 extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
2945 extern int ext4_calculate_overhead(struct super_block *sb);
2946 extern void ext4_superblock_csum_set(struct super_block *sb);
2947 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2948 				    ext4_group_t ngroup);
2949 extern const char *ext4_decode_error(struct super_block *sb, int errno,
2950 				     char nbuf[16]);
2951 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb,
2952 					     ext4_group_t block_group,
2953 					     unsigned int flags);
2954 
2955 extern __printf(6, 7)
2956 void __ext4_error(struct super_block *, const char *, unsigned int, int, __u64,
2957 		  const char *, ...);
2958 extern __printf(6, 7)
2959 void __ext4_error_inode(struct inode *, const char *, unsigned int,
2960 			ext4_fsblk_t, int, const char *, ...);
2961 extern __printf(5, 6)
2962 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2963 		     const char *, ...);
2964 extern void __ext4_std_error(struct super_block *, const char *,
2965 			     unsigned int, int);
2966 extern __printf(5, 6)
2967 void __ext4_abort(struct super_block *, const char *, unsigned int, int,
2968 		  const char *, ...);
2969 extern __printf(4, 5)
2970 void __ext4_warning(struct super_block *, const char *, unsigned int,
2971 		    const char *, ...);
2972 extern __printf(4, 5)
2973 void __ext4_warning_inode(const struct inode *inode, const char *function,
2974 			  unsigned int line, const char *fmt, ...);
2975 extern __printf(3, 4)
2976 void __ext4_msg(struct super_block *, const char *, const char *, ...);
2977 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2978 			   const char *, unsigned int, const char *);
2979 extern __printf(7, 8)
2980 void __ext4_grp_locked_error(const char *, unsigned int,
2981 			     struct super_block *, ext4_group_t,
2982 			     unsigned long, ext4_fsblk_t,
2983 			     const char *, ...);
2984 
2985 #define EXT4_ERROR_INODE(inode, fmt, a...) \
2986 	ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
2987 
2988 #define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...)			\
2989 	__ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a)
2990 
2991 #define ext4_error_inode_block(inode, block, err, fmt, a...)		\
2992 	__ext4_error_inode((inode), __func__, __LINE__, (block), (err),	\
2993 			   (fmt), ## a)
2994 
2995 #define EXT4_ERROR_FILE(file, block, fmt, a...)				\
2996 	ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
2997 
2998 #ifdef CONFIG_PRINTK
2999 
3000 #define ext4_error_inode(inode, func, line, block, fmt, ...)		\
3001 	__ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__)
3002 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...)	\
3003 	__ext4_error_inode((inode), (func), (line), (block), 		\
3004 			   (err), (fmt), ##__VA_ARGS__)
3005 #define ext4_error_file(file, func, line, block, fmt, ...)		\
3006 	__ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
3007 #define ext4_error(sb, fmt, ...)					\
3008 	__ext4_error((sb), __func__, __LINE__, 0, 0, (fmt), ##__VA_ARGS__)
3009 #define ext4_error_err(sb, err, fmt, ...)				\
3010 	__ext4_error((sb), __func__, __LINE__, (err), 0, (fmt), ##__VA_ARGS__)
3011 #define ext4_abort(sb, err, fmt, ...)					\
3012 	__ext4_abort((sb), __func__, __LINE__, (err), (fmt), ##__VA_ARGS__)
3013 #define ext4_warning(sb, fmt, ...)					\
3014 	__ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
3015 #define ext4_warning_inode(inode, fmt, ...)				\
3016 	__ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
3017 #define ext4_msg(sb, level, fmt, ...)				\
3018 	__ext4_msg(sb, level, fmt, ##__VA_ARGS__)
3019 #define dump_mmp_msg(sb, mmp, msg)					\
3020 	__dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
3021 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...)		\
3022 	__ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
3023 				fmt, ##__VA_ARGS__)
3024 
3025 #else
3026 
3027 #define ext4_error_inode(inode, func, line, block, fmt, ...)		\
3028 do {									\
3029 	no_printk(fmt, ##__VA_ARGS__);					\
3030 	__ext4_error_inode(inode, "", 0, block, 0, " ");		\
3031 } while (0)
3032 #define ext4_error_inode_err(inode, func, line, block, err, fmt, ...)	\
3033 do {									\
3034 	no_printk(fmt, ##__VA_ARGS__);					\
3035 	__ext4_error_inode(inode, "", 0, block, err, " ");		\
3036 } while (0)
3037 #define ext4_error_file(file, func, line, block, fmt, ...)		\
3038 do {									\
3039 	no_printk(fmt, ##__VA_ARGS__);					\
3040 	__ext4_error_file(file, "", 0, block, " ");			\
3041 } while (0)
3042 #define ext4_error(sb, fmt, ...)					\
3043 do {									\
3044 	no_printk(fmt, ##__VA_ARGS__);					\
3045 	__ext4_error(sb, "", 0, 0, 0, " ");				\
3046 } while (0)
3047 #define ext4_error_err(sb, err, fmt, ...)				\
3048 do {									\
3049 	no_printk(fmt, ##__VA_ARGS__);					\
3050 	__ext4_error(sb, "", 0, err, 0, " ");				\
3051 } while (0)
3052 #define ext4_abort(sb, err, fmt, ...)					\
3053 do {									\
3054 	no_printk(fmt, ##__VA_ARGS__);					\
3055 	__ext4_abort(sb, "", 0, err, " ");				\
3056 } while (0)
3057 #define ext4_warning(sb, fmt, ...)					\
3058 do {									\
3059 	no_printk(fmt, ##__VA_ARGS__);					\
3060 	__ext4_warning(sb, "", 0, " ");					\
3061 } while (0)
3062 #define ext4_warning_inode(inode, fmt, ...)				\
3063 do {									\
3064 	no_printk(fmt, ##__VA_ARGS__);					\
3065 	__ext4_warning_inode(inode, "", 0, " ");			\
3066 } while (0)
3067 #define ext4_msg(sb, level, fmt, ...)					\
3068 do {									\
3069 	no_printk(fmt, ##__VA_ARGS__);					\
3070 	__ext4_msg(sb, "", " ");					\
3071 } while (0)
3072 #define dump_mmp_msg(sb, mmp, msg)					\
3073 	__dump_mmp_msg(sb, mmp, "", 0, "")
3074 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...)		\
3075 do {									\
3076 	no_printk(fmt, ##__VA_ARGS__);				\
3077 	__ext4_grp_locked_error("", 0, sb, grp, ino, block, " ");	\
3078 } while (0)
3079 
3080 #endif
3081 
3082 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
3083 				      struct ext4_group_desc *bg);
3084 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
3085 				      struct ext4_group_desc *bg);
3086 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
3087 				     struct ext4_group_desc *bg);
3088 extern __u32 ext4_free_group_clusters(struct super_block *sb,
3089 				      struct ext4_group_desc *bg);
3090 extern __u32 ext4_free_inodes_count(struct super_block *sb,
3091 				 struct ext4_group_desc *bg);
3092 extern __u32 ext4_used_dirs_count(struct super_block *sb,
3093 				struct ext4_group_desc *bg);
3094 extern __u32 ext4_itable_unused_count(struct super_block *sb,
3095 				   struct ext4_group_desc *bg);
3096 extern void ext4_block_bitmap_set(struct super_block *sb,
3097 				  struct ext4_group_desc *bg, ext4_fsblk_t blk);
3098 extern void ext4_inode_bitmap_set(struct super_block *sb,
3099 				  struct ext4_group_desc *bg, ext4_fsblk_t blk);
3100 extern void ext4_inode_table_set(struct super_block *sb,
3101 				 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3102 extern void ext4_free_group_clusters_set(struct super_block *sb,
3103 					 struct ext4_group_desc *bg,
3104 					 __u32 count);
3105 extern void ext4_free_inodes_set(struct super_block *sb,
3106 				struct ext4_group_desc *bg, __u32 count);
3107 extern void ext4_used_dirs_set(struct super_block *sb,
3108 				struct ext4_group_desc *bg, __u32 count);
3109 extern void ext4_itable_unused_set(struct super_block *sb,
3110 				   struct ext4_group_desc *bg, __u32 count);
3111 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
3112 				       struct ext4_group_desc *gdp);
3113 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
3114 				     struct ext4_group_desc *gdp);
3115 extern int ext4_register_li_request(struct super_block *sb,
3116 				    ext4_group_t first_not_zeroed);
3117 
ext4_has_metadata_csum(struct super_block * sb)3118 static inline int ext4_has_metadata_csum(struct super_block *sb)
3119 {
3120 	WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
3121 		     !EXT4_SB(sb)->s_chksum_driver);
3122 
3123 	return ext4_has_feature_metadata_csum(sb) &&
3124 	       (EXT4_SB(sb)->s_chksum_driver != NULL);
3125 }
3126 
ext4_has_group_desc_csum(struct super_block * sb)3127 static inline int ext4_has_group_desc_csum(struct super_block *sb)
3128 {
3129 	return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb);
3130 }
3131 
3132 #define ext4_read_incompat_64bit_val(es, name) \
3133 	(((es)->s_feature_incompat & cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT) \
3134 		? (ext4_fsblk_t)le32_to_cpu(es->name##_hi) << 32 : 0) | \
3135 		le32_to_cpu(es->name##_lo))
3136 
ext4_blocks_count(struct ext4_super_block * es)3137 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
3138 {
3139 	return ext4_read_incompat_64bit_val(es, s_blocks_count);
3140 }
3141 
ext4_r_blocks_count(struct ext4_super_block * es)3142 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
3143 {
3144 	return ext4_read_incompat_64bit_val(es, s_r_blocks_count);
3145 }
3146 
ext4_free_blocks_count(struct ext4_super_block * es)3147 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
3148 {
3149 	return ext4_read_incompat_64bit_val(es, s_free_blocks_count);
3150 }
3151 
ext4_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)3152 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
3153 					 ext4_fsblk_t blk)
3154 {
3155 	es->s_blocks_count_lo = cpu_to_le32((u32)blk);
3156 	es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
3157 }
3158 
ext4_free_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)3159 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
3160 					      ext4_fsblk_t blk)
3161 {
3162 	es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
3163 	es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
3164 }
3165 
ext4_r_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)3166 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
3167 					   ext4_fsblk_t blk)
3168 {
3169 	es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
3170 	es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
3171 }
3172 
ext4_isize(struct super_block * sb,struct ext4_inode * raw_inode)3173 static inline loff_t ext4_isize(struct super_block *sb,
3174 				struct ext4_inode *raw_inode)
3175 {
3176 	if (ext4_has_feature_largedir(sb) ||
3177 	    S_ISREG(le16_to_cpu(raw_inode->i_mode)))
3178 		return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
3179 			le32_to_cpu(raw_inode->i_size_lo);
3180 
3181 	return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
3182 }
3183 
ext4_isize_set(struct ext4_inode * raw_inode,loff_t i_size)3184 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
3185 {
3186 	raw_inode->i_size_lo = cpu_to_le32(i_size);
3187 	raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
3188 }
3189 
3190 static inline
ext4_get_group_info(struct super_block * sb,ext4_group_t group)3191 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
3192 					    ext4_group_t group)
3193 {
3194 	 struct ext4_group_info **grp_info;
3195 	 long indexv, indexh;
3196 	 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
3197 	 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
3198 	 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
3199 	 grp_info = sbi_array_rcu_deref(EXT4_SB(sb), s_group_info, indexv);
3200 	 return grp_info[indexh];
3201 }
3202 
3203 /*
3204  * Reading s_groups_count requires using smp_rmb() afterwards.  See
3205  * the locking protocol documented in the comments of ext4_group_add()
3206  * in resize.c
3207  */
ext4_get_groups_count(struct super_block * sb)3208 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
3209 {
3210 	ext4_group_t	ngroups = EXT4_SB(sb)->s_groups_count;
3211 
3212 	smp_rmb();
3213 	return ngroups;
3214 }
3215 
ext4_flex_group(struct ext4_sb_info * sbi,ext4_group_t block_group)3216 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
3217 					     ext4_group_t block_group)
3218 {
3219 	return block_group >> sbi->s_log_groups_per_flex;
3220 }
3221 
ext4_flex_bg_size(struct ext4_sb_info * sbi)3222 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
3223 {
3224 	return 1 << sbi->s_log_groups_per_flex;
3225 }
3226 
3227 #define ext4_std_error(sb, errno)				\
3228 do {								\
3229 	if ((errno))						\
3230 		__ext4_std_error((sb), __func__, __LINE__, (errno));	\
3231 } while (0)
3232 
3233 #ifdef CONFIG_SMP
3234 /* Each CPU can accumulate percpu_counter_batch clusters in their local
3235  * counters. So we need to make sure we have free clusters more
3236  * than percpu_counter_batch  * nr_cpu_ids. Also add a window of 4 times.
3237  */
3238 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
3239 #else
3240 #define EXT4_FREECLUSTERS_WATERMARK 0
3241 #endif
3242 
3243 /* Update i_disksize. Requires i_mutex to avoid races with truncate */
ext4_update_i_disksize(struct inode * inode,loff_t newsize)3244 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
3245 {
3246 	WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
3247 		     !inode_is_locked(inode));
3248 	down_write(&EXT4_I(inode)->i_data_sem);
3249 	if (newsize > EXT4_I(inode)->i_disksize)
3250 		WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize);
3251 	up_write(&EXT4_I(inode)->i_data_sem);
3252 }
3253 
3254 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
ext4_update_inode_size(struct inode * inode,loff_t newsize)3255 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
3256 {
3257 	int changed = 0;
3258 
3259 	if (newsize > inode->i_size) {
3260 		i_size_write(inode, newsize);
3261 		changed = 1;
3262 	}
3263 	if (newsize > EXT4_I(inode)->i_disksize) {
3264 		ext4_update_i_disksize(inode, newsize);
3265 		changed |= 2;
3266 	}
3267 	return changed;
3268 }
3269 
3270 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
3271 				      loff_t len);
3272 
3273 struct ext4_group_info {
3274 	unsigned long   bb_state;
3275 #ifdef AGGRESSIVE_CHECK
3276 	unsigned long	bb_check_counter;
3277 #endif
3278 	struct rb_root  bb_free_root;
3279 	ext4_grpblk_t	bb_first_free;	/* first free block */
3280 	ext4_grpblk_t	bb_free;	/* total free blocks */
3281 	ext4_grpblk_t	bb_fragments;	/* nr of freespace fragments */
3282 	ext4_grpblk_t	bb_largest_free_order;/* order of largest frag in BG */
3283 	struct          list_head bb_prealloc_list;
3284 #ifdef DOUBLE_CHECK
3285 	void            *bb_bitmap;
3286 #endif
3287 	struct rw_semaphore alloc_sem;
3288 	ext4_grpblk_t	bb_counters[];	/* Nr of free power-of-two-block
3289 					 * regions, index is order.
3290 					 * bb_counters[3] = 5 means
3291 					 * 5 free 8-block regions. */
3292 };
3293 
3294 #define EXT4_GROUP_INFO_NEED_INIT_BIT		0
3295 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT		1
3296 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT	2
3297 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT	3
3298 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT		\
3299 	(1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT)
3300 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT		\
3301 	(1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT)
3302 #define EXT4_GROUP_INFO_BBITMAP_READ_BIT	4
3303 
3304 #define EXT4_MB_GRP_NEED_INIT(grp)	\
3305 	(test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
3306 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp)	\
3307 	(test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3308 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp)	\
3309 	(test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3310 
3311 #define EXT4_MB_GRP_WAS_TRIMMED(grp)	\
3312 	(test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3313 #define EXT4_MB_GRP_SET_TRIMMED(grp)	\
3314 	(set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3315 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp)	\
3316 	(clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3317 #define EXT4_MB_GRP_TEST_AND_SET_READ(grp)	\
3318 	(test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state)))
3319 
3320 #define EXT4_MAX_CONTENTION		8
3321 #define EXT4_CONTENTION_THRESHOLD	2
3322 
ext4_group_lock_ptr(struct super_block * sb,ext4_group_t group)3323 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
3324 					      ext4_group_t group)
3325 {
3326 	return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
3327 }
3328 
3329 /*
3330  * Returns true if the filesystem is busy enough that attempts to
3331  * access the block group locks has run into contention.
3332  */
ext4_fs_is_busy(struct ext4_sb_info * sbi)3333 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
3334 {
3335 	return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
3336 }
3337 
ext4_lock_group(struct super_block * sb,ext4_group_t group)3338 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
3339 {
3340 	spinlock_t *lock = ext4_group_lock_ptr(sb, group);
3341 	if (spin_trylock(lock))
3342 		/*
3343 		 * We're able to grab the lock right away, so drop the
3344 		 * lock contention counter.
3345 		 */
3346 		atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
3347 	else {
3348 		/*
3349 		 * The lock is busy, so bump the contention counter,
3350 		 * and then wait on the spin lock.
3351 		 */
3352 		atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
3353 				  EXT4_MAX_CONTENTION);
3354 		spin_lock(lock);
3355 	}
3356 }
3357 
ext4_unlock_group(struct super_block * sb,ext4_group_t group)3358 static inline void ext4_unlock_group(struct super_block *sb,
3359 					ext4_group_t group)
3360 {
3361 	spin_unlock(ext4_group_lock_ptr(sb, group));
3362 }
3363 
3364 /*
3365  * Block validity checking
3366  */
3367 #define ext4_check_indirect_blockref(inode, bh)				\
3368 	ext4_check_blockref(__func__, __LINE__, inode,			\
3369 			    (__le32 *)(bh)->b_data,			\
3370 			    EXT4_ADDR_PER_BLOCK((inode)->i_sb))
3371 
3372 #define ext4_ind_check_inode(inode)					\
3373 	ext4_check_blockref(__func__, __LINE__, inode,			\
3374 			    EXT4_I(inode)->i_data,			\
3375 			    EXT4_NDIR_BLOCKS)
3376 
3377 /*
3378  * Inodes and files operations
3379  */
3380 
3381 /* dir.c */
3382 extern const struct file_operations ext4_dir_operations;
3383 
3384 #ifdef CONFIG_UNICODE
3385 extern const struct dentry_operations ext4_dentry_ops;
3386 #endif
3387 
3388 /* file.c */
3389 extern const struct inode_operations ext4_file_inode_operations;
3390 extern const struct file_operations ext4_file_operations;
3391 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
3392 
3393 /* inline.c */
3394 extern int ext4_get_max_inline_size(struct inode *inode);
3395 extern int ext4_find_inline_data_nolock(struct inode *inode);
3396 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
3397 				 unsigned int len);
3398 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
3399 
3400 extern int ext4_readpage_inline(struct inode *inode, struct page *page);
3401 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3402 					 struct inode *inode,
3403 					 loff_t pos, unsigned len,
3404 					 unsigned flags,
3405 					 struct page **pagep);
3406 extern int ext4_write_inline_data_end(struct inode *inode,
3407 				      loff_t pos, unsigned len,
3408 				      unsigned copied,
3409 				      struct page *page);
3410 extern struct buffer_head *
3411 ext4_journalled_write_inline_data(struct inode *inode,
3412 				  unsigned len,
3413 				  struct page *page);
3414 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
3415 					   struct inode *inode,
3416 					   loff_t pos, unsigned len,
3417 					   unsigned flags,
3418 					   struct page **pagep,
3419 					   void **fsdata);
3420 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
3421 					 unsigned len, unsigned copied,
3422 					 struct page *page);
3423 extern int ext4_try_add_inline_entry(handle_t *handle,
3424 				     struct ext4_filename *fname,
3425 				     struct inode *dir, struct inode *inode);
3426 extern int ext4_try_create_inline_dir(handle_t *handle,
3427 				      struct inode *parent,
3428 				      struct inode *inode);
3429 extern int ext4_read_inline_dir(struct file *filp,
3430 				struct dir_context *ctx,
3431 				int *has_inline_data);
3432 extern int ext4_inlinedir_to_tree(struct file *dir_file,
3433 				  struct inode *dir, ext4_lblk_t block,
3434 				  struct dx_hash_info *hinfo,
3435 				  __u32 start_hash, __u32 start_minor_hash,
3436 				  int *has_inline_data);
3437 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3438 					struct ext4_filename *fname,
3439 					struct ext4_dir_entry_2 **res_dir,
3440 					int *has_inline_data);
3441 extern int ext4_delete_inline_entry(handle_t *handle,
3442 				    struct inode *dir,
3443 				    struct ext4_dir_entry_2 *de_del,
3444 				    struct buffer_head *bh,
3445 				    int *has_inline_data);
3446 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3447 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3448 					struct ext4_dir_entry_2 **parent_de,
3449 					int *retval);
3450 extern int ext4_inline_data_fiemap(struct inode *inode,
3451 				   struct fiemap_extent_info *fieinfo,
3452 				   int *has_inline, __u64 start, __u64 len);
3453 
3454 struct iomap;
3455 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap);
3456 
3457 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3458 
3459 extern int ext4_convert_inline_data(struct inode *inode);
3460 
ext4_has_inline_data(struct inode * inode)3461 static inline int ext4_has_inline_data(struct inode *inode)
3462 {
3463 	return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3464 	       EXT4_I(inode)->i_inline_off;
3465 }
3466 
3467 /* namei.c */
3468 extern const struct inode_operations ext4_dir_inode_operations;
3469 extern const struct inode_operations ext4_special_inode_operations;
3470 extern struct dentry *ext4_get_parent(struct dentry *child);
3471 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3472 				 struct ext4_dir_entry_2 *de,
3473 				 int blocksize, int csum_size,
3474 				 unsigned int parent_ino, int dotdot_real_len);
3475 extern void ext4_initialize_dirent_tail(struct buffer_head *bh,
3476 					unsigned int blocksize);
3477 extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode,
3478 				      struct buffer_head *bh);
3479 extern int ext4_ci_compare(const struct inode *parent,
3480 			   const struct qstr *fname,
3481 			   const struct qstr *entry, bool quick);
3482 extern int __ext4_unlink(handle_t *handle, struct inode *dir, const struct qstr *d_name,
3483 			 struct inode *inode);
3484 extern int __ext4_link(struct inode *dir, struct inode *inode,
3485 		       struct dentry *dentry);
3486 
3487 #define S_SHIFT 12
3488 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = {
3489 	[S_IFREG >> S_SHIFT]	= EXT4_FT_REG_FILE,
3490 	[S_IFDIR >> S_SHIFT]	= EXT4_FT_DIR,
3491 	[S_IFCHR >> S_SHIFT]	= EXT4_FT_CHRDEV,
3492 	[S_IFBLK >> S_SHIFT]	= EXT4_FT_BLKDEV,
3493 	[S_IFIFO >> S_SHIFT]	= EXT4_FT_FIFO,
3494 	[S_IFSOCK >> S_SHIFT]	= EXT4_FT_SOCK,
3495 	[S_IFLNK >> S_SHIFT]	= EXT4_FT_SYMLINK,
3496 };
3497 
ext4_set_de_type(struct super_block * sb,struct ext4_dir_entry_2 * de,umode_t mode)3498 static inline void ext4_set_de_type(struct super_block *sb,
3499 				struct ext4_dir_entry_2 *de,
3500 				umode_t mode) {
3501 	if (ext4_has_feature_filetype(sb))
3502 		de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3503 }
3504 
3505 /* readpages.c */
3506 extern int ext4_mpage_readpages(struct inode *inode,
3507 		struct readahead_control *rac, struct page *page);
3508 extern int __init ext4_init_post_read_processing(void);
3509 extern void ext4_exit_post_read_processing(void);
3510 
3511 /* symlink.c */
3512 extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3513 extern const struct inode_operations ext4_symlink_inode_operations;
3514 extern const struct inode_operations ext4_fast_symlink_inode_operations;
3515 
3516 /* sysfs.c */
3517 extern int ext4_register_sysfs(struct super_block *sb);
3518 extern void ext4_unregister_sysfs(struct super_block *sb);
3519 extern int __init ext4_init_sysfs(void);
3520 extern void ext4_exit_sysfs(void);
3521 
3522 /* block_validity */
3523 extern void ext4_release_system_zone(struct super_block *sb);
3524 extern int ext4_setup_system_zone(struct super_block *sb);
3525 extern int __init ext4_init_system_zone(void);
3526 extern void ext4_exit_system_zone(void);
3527 extern int ext4_inode_block_valid(struct inode *inode,
3528 				  ext4_fsblk_t start_blk,
3529 				  unsigned int count);
3530 extern int ext4_check_blockref(const char *, unsigned int,
3531 			       struct inode *, __le32 *, unsigned int);
3532 
3533 /* extents.c */
3534 struct ext4_ext_path;
3535 struct ext4_extent;
3536 
3537 /*
3538  * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3539  * __le32.
3540  */
3541 #define EXT_MAX_BLOCKS	0xffffffff
3542 
3543 extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode);
3544 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3545 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3546 			       struct ext4_map_blocks *map, int flags);
3547 extern int ext4_ext_truncate(handle_t *, struct inode *);
3548 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3549 				 ext4_lblk_t end);
3550 extern void ext4_ext_init(struct super_block *);
3551 extern void ext4_ext_release(struct super_block *);
3552 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3553 			  loff_t len);
3554 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3555 					  loff_t offset, ssize_t len);
3556 extern int ext4_convert_unwritten_io_end_vec(handle_t *handle,
3557 					     ext4_io_end_t *io_end);
3558 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3559 			   struct ext4_map_blocks *map, int flags);
3560 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3561 						   int num,
3562 						   struct ext4_ext_path *path);
3563 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3564 				  struct ext4_ext_path **,
3565 				  struct ext4_extent *, int);
3566 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3567 					      struct ext4_ext_path **,
3568 					      int flags);
3569 extern void ext4_ext_drop_refs(struct ext4_ext_path *);
3570 extern int ext4_ext_check_inode(struct inode *inode);
3571 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3572 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3573 			__u64 start, __u64 len);
3574 extern int ext4_get_es_cache(struct inode *inode,
3575 			     struct fiemap_extent_info *fieinfo,
3576 			     __u64 start, __u64 len);
3577 extern int ext4_ext_precache(struct inode *inode);
3578 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3579 				struct inode *inode2, ext4_lblk_t lblk1,
3580 			     ext4_lblk_t lblk2,  ext4_lblk_t count,
3581 			     int mark_unwritten,int *err);
3582 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu);
3583 extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode,
3584 				       int check_cred, int restart_cred,
3585 				       int revoke_cred);
3586 extern void ext4_ext_replay_shrink_inode(struct inode *inode, ext4_lblk_t end);
3587 extern int ext4_ext_replay_set_iblocks(struct inode *inode);
3588 extern int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start,
3589 		int len, int unwritten, ext4_fsblk_t pblk);
3590 extern int ext4_ext_clear_bb(struct inode *inode);
3591 
3592 
3593 /* move_extent.c */
3594 extern void ext4_double_down_write_data_sem(struct inode *first,
3595 					    struct inode *second);
3596 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3597 					  struct inode *donor_inode);
3598 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3599 			     __u64 start_orig, __u64 start_donor,
3600 			     __u64 len, __u64 *moved_len);
3601 
3602 /* page-io.c */
3603 extern int __init ext4_init_pageio(void);
3604 extern void ext4_exit_pageio(void);
3605 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3606 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3607 extern int ext4_put_io_end(ext4_io_end_t *io_end);
3608 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3609 extern void ext4_io_submit_init(struct ext4_io_submit *io,
3610 				struct writeback_control *wbc);
3611 extern void ext4_end_io_rsv_work(struct work_struct *work);
3612 extern void ext4_io_submit(struct ext4_io_submit *io);
3613 extern int ext4_bio_write_page(struct ext4_io_submit *io,
3614 			       struct page *page,
3615 			       int len,
3616 			       struct writeback_control *wbc,
3617 			       bool keep_towrite);
3618 extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end);
3619 extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end);
3620 
3621 /* mmp.c */
3622 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3623 
3624 /* verity.c */
3625 extern const struct fsverity_operations ext4_verityops;
3626 
3627 /*
3628  * Add new method to test whether block and inode bitmaps are properly
3629  * initialized. With uninit_bg reading the block from disk is not enough
3630  * to mark the bitmap uptodate. We need to also zero-out the bitmap
3631  */
3632 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3633 
bitmap_uptodate(struct buffer_head * bh)3634 static inline int bitmap_uptodate(struct buffer_head *bh)
3635 {
3636 	return (buffer_uptodate(bh) &&
3637 			test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3638 }
set_bitmap_uptodate(struct buffer_head * bh)3639 static inline void set_bitmap_uptodate(struct buffer_head *bh)
3640 {
3641 	set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3642 }
3643 
3644 #define in_range(b, first, len)	((b) >= (first) && (b) <= (first) + (len) - 1)
3645 
3646 /* For ioend & aio unwritten conversion wait queues */
3647 #define EXT4_WQ_HASH_SZ		37
3648 #define ext4_ioend_wq(v)   (&ext4__ioend_wq[((unsigned long)(v)) %\
3649 					    EXT4_WQ_HASH_SZ])
3650 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3651 
3652 extern int ext4_resize_begin(struct super_block *sb);
3653 extern void ext4_resize_end(struct super_block *sb);
3654 
ext4_set_io_unwritten_flag(struct inode * inode,struct ext4_io_end * io_end)3655 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
3656 					      struct ext4_io_end *io_end)
3657 {
3658 	if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
3659 		io_end->flag |= EXT4_IO_END_UNWRITTEN;
3660 		atomic_inc(&EXT4_I(inode)->i_unwritten);
3661 	}
3662 }
3663 
ext4_clear_io_unwritten_flag(ext4_io_end_t * io_end)3664 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3665 {
3666 	struct inode *inode = io_end->inode;
3667 
3668 	if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
3669 		io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3670 		/* Wake up anyone waiting on unwritten extent conversion */
3671 		if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
3672 			wake_up_all(ext4_ioend_wq(inode));
3673 	}
3674 }
3675 
3676 extern const struct iomap_ops ext4_iomap_ops;
3677 extern const struct iomap_ops ext4_iomap_overwrite_ops;
3678 extern const struct iomap_ops ext4_iomap_report_ops;
3679 
ext4_buffer_uptodate(struct buffer_head * bh)3680 static inline int ext4_buffer_uptodate(struct buffer_head *bh)
3681 {
3682 	/*
3683 	 * If the buffer has the write error flag, we have failed
3684 	 * to write out data in the block.  In this  case, we don't
3685 	 * have to read the block because we may read the old data
3686 	 * successfully.
3687 	 */
3688 	if (!buffer_uptodate(bh) && buffer_write_io_error(bh))
3689 		set_buffer_uptodate(bh);
3690 	return buffer_uptodate(bh);
3691 }
3692 
3693 #endif	/* __KERNEL__ */
3694 
3695 #define EFSBADCRC	EBADMSG		/* Bad CRC detected */
3696 #define EFSCORRUPTED	EUCLEAN		/* Filesystem is corrupted */
3697 
3698 #endif	/* _EXT4_H */
3699