• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 /*
2  *  ext4.h
3  *
4  * Copyright (C) 1992, 1993, 1994, 1995
5  * Remy Card (card@masi.ibp.fr)
6  * Laboratoire MASI - Institut Blaise Pascal
7  * Universite Pierre et Marie Curie (Paris VI)
8  *
9  *  from
10  *
11  *  linux/include/linux/minix_fs.h
12  *
13  *  Copyright (C) 1991, 1992  Linus Torvalds
14  */
15 
16 #ifndef _EXT4_H
17 #define _EXT4_H
18 
19 #include <linux/types.h>
20 #include <linux/blkdev.h>
21 #include <linux/magic.h>
22 #include <linux/jbd2.h>
23 #include <linux/quota.h>
24 #include <linux/rwsem.h>
25 #include <linux/rbtree.h>
26 #include <linux/seqlock.h>
27 #include <linux/mutex.h>
28 #include <linux/timer.h>
29 #include <linux/wait.h>
30 #include <linux/blockgroup_lock.h>
31 #include <linux/percpu_counter.h>
32 #include <linux/ratelimit.h>
33 #include <crypto/hash.h>
34 #include <linux/falloc.h>
35 #ifdef __KERNEL__
36 #include <linux/compat.h>
37 #endif
38 
39 /*
40  * The fourth extended filesystem constants/structures
41  */
42 
43 /*
44  * Define EXT4FS_DEBUG to produce debug messages
45  */
46 #undef EXT4FS_DEBUG
47 
48 /*
49  * Debug code
50  */
51 #ifdef EXT4FS_DEBUG
52 #define ext4_debug(f, a...)						\
53 	do {								\
54 		printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:",	\
55 			__FILE__, __LINE__, __func__);			\
56 		printk(KERN_DEBUG f, ## a);				\
57 	} while (0)
58 #else
59 #define ext4_debug(fmt, ...)	no_printk(fmt, ##__VA_ARGS__)
60 #endif
61 
62 /*
63  * Turn on EXT_DEBUG to get lots of info about extents operations.
64  */
65 #define EXT_DEBUG__
66 #ifdef EXT_DEBUG
67 #define ext_debug(fmt, ...)	printk(fmt, ##__VA_ARGS__)
68 #else
69 #define ext_debug(fmt, ...)	no_printk(fmt, ##__VA_ARGS__)
70 #endif
71 
72 #define EXT4_ERROR_INODE(inode, fmt, a...) \
73 	ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
74 
75 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...)			\
76 	ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
77 
78 #define EXT4_ERROR_FILE(file, block, fmt, a...)				\
79 	ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
80 
81 /* data type for block offset of block group */
82 typedef int ext4_grpblk_t;
83 
84 /* data type for filesystem-wide blocks number */
85 typedef unsigned long long ext4_fsblk_t;
86 
87 /* data type for file logical block number */
88 typedef __u32 ext4_lblk_t;
89 
90 /* data type for block group number */
91 typedef unsigned int ext4_group_t;
92 
93 /*
94  * Flags used in mballoc's allocation_context flags field.
95  *
96  * Also used to show what's going on for debugging purposes when the
97  * flag field is exported via the traceport interface
98  */
99 
100 /* prefer goal again. length */
101 #define EXT4_MB_HINT_MERGE		0x0001
102 /* blocks already reserved */
103 #define EXT4_MB_HINT_RESERVED		0x0002
104 /* metadata is being allocated */
105 #define EXT4_MB_HINT_METADATA		0x0004
106 /* first blocks in the file */
107 #define EXT4_MB_HINT_FIRST		0x0008
108 /* search for the best chunk */
109 #define EXT4_MB_HINT_BEST		0x0010
110 /* data is being allocated */
111 #define EXT4_MB_HINT_DATA		0x0020
112 /* don't preallocate (for tails) */
113 #define EXT4_MB_HINT_NOPREALLOC		0x0040
114 /* allocate for locality group */
115 #define EXT4_MB_HINT_GROUP_ALLOC	0x0080
116 /* allocate goal blocks or none */
117 #define EXT4_MB_HINT_GOAL_ONLY		0x0100
118 /* goal is meaningful */
119 #define EXT4_MB_HINT_TRY_GOAL		0x0200
120 /* blocks already pre-reserved by delayed allocation */
121 #define EXT4_MB_DELALLOC_RESERVED	0x0400
122 /* We are doing stream allocation */
123 #define EXT4_MB_STREAM_ALLOC		0x0800
124 /* Use reserved root blocks if needed */
125 #define EXT4_MB_USE_ROOT_BLOCKS		0x1000
126 /* Use blocks from reserved pool */
127 #define EXT4_MB_USE_RESERVED		0x2000
128 
129 struct ext4_allocation_request {
130 	/* target inode for block we're allocating */
131 	struct inode *inode;
132 	/* how many blocks we want to allocate */
133 	unsigned int len;
134 	/* logical block in target inode */
135 	ext4_lblk_t logical;
136 	/* the closest logical allocated block to the left */
137 	ext4_lblk_t lleft;
138 	/* the closest logical allocated block to the right */
139 	ext4_lblk_t lright;
140 	/* phys. target (a hint) */
141 	ext4_fsblk_t goal;
142 	/* phys. block for the closest logical allocated block to the left */
143 	ext4_fsblk_t pleft;
144 	/* phys. block for the closest logical allocated block to the right */
145 	ext4_fsblk_t pright;
146 	/* flags. see above EXT4_MB_HINT_* */
147 	unsigned int flags;
148 };
149 
150 /*
151  * Logical to physical block mapping, used by ext4_map_blocks()
152  *
153  * This structure is used to pass requests into ext4_map_blocks() as
154  * well as to store the information returned by ext4_map_blocks().  It
155  * takes less room on the stack than a struct buffer_head.
156  */
157 #define EXT4_MAP_NEW		(1 << BH_New)
158 #define EXT4_MAP_MAPPED		(1 << BH_Mapped)
159 #define EXT4_MAP_UNWRITTEN	(1 << BH_Unwritten)
160 #define EXT4_MAP_BOUNDARY	(1 << BH_Boundary)
161 /* Sometimes (in the bigalloc case, from ext4_da_get_block_prep) the caller of
162  * ext4_map_blocks wants to know whether or not the underlying cluster has
163  * already been accounted for. EXT4_MAP_FROM_CLUSTER conveys to the caller that
164  * the requested mapping was from previously mapped (or delayed allocated)
165  * cluster. We use BH_AllocFromCluster only for this flag. BH_AllocFromCluster
166  * should never appear on buffer_head's state flags.
167  */
168 #define EXT4_MAP_FROM_CLUSTER	(1 << BH_AllocFromCluster)
169 #define EXT4_MAP_FLAGS		(EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
170 				 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\
171 				 EXT4_MAP_FROM_CLUSTER)
172 
173 struct ext4_map_blocks {
174 	ext4_fsblk_t m_pblk;
175 	ext4_lblk_t m_lblk;
176 	unsigned int m_len;
177 	unsigned int m_flags;
178 };
179 
180 /*
181  * Flags for ext4_io_end->flags
182  */
183 #define	EXT4_IO_END_UNWRITTEN	0x0001
184 
185 /*
186  * For converting unwritten extents on a work queue. 'handle' is used for
187  * buffered writeback.
188  */
189 typedef struct ext4_io_end {
190 	struct list_head	list;		/* per-file finished IO list */
191 	handle_t		*handle;	/* handle reserved for extent
192 						 * conversion */
193 	struct inode		*inode;		/* file being written to */
194 	struct bio		*bio;		/* Linked list of completed
195 						 * bios covering the extent */
196 	unsigned int		flag;		/* unwritten or not */
197 	loff_t			offset;		/* offset in the file */
198 	ssize_t			size;		/* size of the extent */
199 	atomic_t		count;		/* reference counter */
200 } ext4_io_end_t;
201 
202 struct ext4_io_submit {
203 	int			io_op;
204 	struct bio		*io_bio;
205 	ext4_io_end_t		*io_end;
206 	sector_t		io_next_block;
207 };
208 
209 /*
210  * Special inodes numbers
211  */
212 #define	EXT4_BAD_INO		 1	/* Bad blocks inode */
213 #define EXT4_ROOT_INO		 2	/* Root inode */
214 #define EXT4_USR_QUOTA_INO	 3	/* User quota inode */
215 #define EXT4_GRP_QUOTA_INO	 4	/* Group quota inode */
216 #define EXT4_BOOT_LOADER_INO	 5	/* Boot loader inode */
217 #define EXT4_UNDEL_DIR_INO	 6	/* Undelete directory inode */
218 #define EXT4_RESIZE_INO		 7	/* Reserved group descriptors inode */
219 #define EXT4_JOURNAL_INO	 8	/* Journal inode */
220 
221 /* First non-reserved inode for old ext4 filesystems */
222 #define EXT4_GOOD_OLD_FIRST_INO	11
223 
224 /*
225  * Maximal count of links to a file
226  */
227 #define EXT4_LINK_MAX		65000
228 
229 /*
230  * Macro-instructions used to manage several block sizes
231  */
232 #define EXT4_MIN_BLOCK_SIZE		1024
233 #define	EXT4_MAX_BLOCK_SIZE		65536
234 #define EXT4_MIN_BLOCK_LOG_SIZE		10
235 #define EXT4_MAX_BLOCK_LOG_SIZE		16
236 #define EXT4_MAX_CLUSTER_LOG_SIZE	30
237 #ifdef __KERNEL__
238 # define EXT4_BLOCK_SIZE(s)		((s)->s_blocksize)
239 #else
240 # define EXT4_BLOCK_SIZE(s)		(EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
241 #endif
242 #define	EXT4_ADDR_PER_BLOCK(s)		(EXT4_BLOCK_SIZE(s) / sizeof(__u32))
243 #define EXT4_CLUSTER_SIZE(s)		(EXT4_BLOCK_SIZE(s) << \
244 					 EXT4_SB(s)->s_cluster_bits)
245 #ifdef __KERNEL__
246 # define EXT4_BLOCK_SIZE_BITS(s)	((s)->s_blocksize_bits)
247 # define EXT4_CLUSTER_BITS(s)		(EXT4_SB(s)->s_cluster_bits)
248 #else
249 # define EXT4_BLOCK_SIZE_BITS(s)	((s)->s_log_block_size + 10)
250 #endif
251 #ifdef __KERNEL__
252 #define	EXT4_ADDR_PER_BLOCK_BITS(s)	(EXT4_SB(s)->s_addr_per_block_bits)
253 #define EXT4_INODE_SIZE(s)		(EXT4_SB(s)->s_inode_size)
254 #define EXT4_FIRST_INO(s)		(EXT4_SB(s)->s_first_ino)
255 #else
256 #define EXT4_INODE_SIZE(s)	(((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
257 				 EXT4_GOOD_OLD_INODE_SIZE : \
258 				 (s)->s_inode_size)
259 #define EXT4_FIRST_INO(s)	(((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
260 				 EXT4_GOOD_OLD_FIRST_INO : \
261 				 (s)->s_first_ino)
262 #endif
263 #define EXT4_BLOCK_ALIGN(size, blkbits)		ALIGN((size), (1 << (blkbits)))
264 
265 /* Translate a block number to a cluster number */
266 #define EXT4_B2C(sbi, blk)	((blk) >> (sbi)->s_cluster_bits)
267 /* Translate a cluster number to a block number */
268 #define EXT4_C2B(sbi, cluster)	((cluster) << (sbi)->s_cluster_bits)
269 /* Translate # of blks to # of clusters */
270 #define EXT4_NUM_B2C(sbi, blks)	(((blks) + (sbi)->s_cluster_ratio - 1) >> \
271 				 (sbi)->s_cluster_bits)
272 /* Mask out the low bits to get the starting block of the cluster */
273 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) &				\
274 				  ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
275 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) &				\
276 				  ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
277 /* Get the cluster offset */
278 #define EXT4_PBLK_COFF(s, pblk) ((pblk) &				\
279 				 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
280 #define EXT4_LBLK_COFF(s, lblk) ((lblk) &				\
281 				 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
282 
283 /*
284  * Structure of a blocks group descriptor
285  */
286 struct ext4_group_desc
287 {
288 	__le32	bg_block_bitmap_lo;	/* Blocks bitmap block */
289 	__le32	bg_inode_bitmap_lo;	/* Inodes bitmap block */
290 	__le32	bg_inode_table_lo;	/* Inodes table block */
291 	__le16	bg_free_blocks_count_lo;/* Free blocks count */
292 	__le16	bg_free_inodes_count_lo;/* Free inodes count */
293 	__le16	bg_used_dirs_count_lo;	/* Directories count */
294 	__le16	bg_flags;		/* EXT4_BG_flags (INODE_UNINIT, etc) */
295 	__le32  bg_exclude_bitmap_lo;   /* Exclude bitmap for snapshots */
296 	__le16  bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
297 	__le16  bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
298 	__le16  bg_itable_unused_lo;	/* Unused inodes count */
299 	__le16  bg_checksum;		/* crc16(sb_uuid+group+desc) */
300 	__le32	bg_block_bitmap_hi;	/* Blocks bitmap block MSB */
301 	__le32	bg_inode_bitmap_hi;	/* Inodes bitmap block MSB */
302 	__le32	bg_inode_table_hi;	/* Inodes table block MSB */
303 	__le16	bg_free_blocks_count_hi;/* Free blocks count MSB */
304 	__le16	bg_free_inodes_count_hi;/* Free inodes count MSB */
305 	__le16	bg_used_dirs_count_hi;	/* Directories count MSB */
306 	__le16  bg_itable_unused_hi;    /* Unused inodes count MSB */
307 	__le32  bg_exclude_bitmap_hi;   /* Exclude bitmap block MSB */
308 	__le16  bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
309 	__le16  bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
310 	__u32   bg_reserved;
311 };
312 
313 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END	\
314 	(offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
315 	 sizeof(__le16))
316 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END	\
317 	(offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
318 	 sizeof(__le16))
319 
320 /*
321  * Structure of a flex block group info
322  */
323 
324 struct flex_groups {
325 	atomic64_t	free_clusters;
326 	atomic_t	free_inodes;
327 	atomic_t	used_dirs;
328 };
329 
330 #define EXT4_BG_INODE_UNINIT	0x0001 /* Inode table/bitmap not in use */
331 #define EXT4_BG_BLOCK_UNINIT	0x0002 /* Block bitmap not in use */
332 #define EXT4_BG_INODE_ZEROED	0x0004 /* On-disk itable initialized to zero */
333 
334 /*
335  * Macro-instructions used to manage group descriptors
336  */
337 #define EXT4_MIN_DESC_SIZE		32
338 #define EXT4_MIN_DESC_SIZE_64BIT	64
339 #define	EXT4_MAX_DESC_SIZE		EXT4_MIN_BLOCK_SIZE
340 #define EXT4_DESC_SIZE(s)		(EXT4_SB(s)->s_desc_size)
341 #ifdef __KERNEL__
342 # define EXT4_BLOCKS_PER_GROUP(s)	(EXT4_SB(s)->s_blocks_per_group)
343 # define EXT4_CLUSTERS_PER_GROUP(s)	(EXT4_SB(s)->s_clusters_per_group)
344 # define EXT4_DESC_PER_BLOCK(s)		(EXT4_SB(s)->s_desc_per_block)
345 # define EXT4_INODES_PER_GROUP(s)	(EXT4_SB(s)->s_inodes_per_group)
346 # define EXT4_DESC_PER_BLOCK_BITS(s)	(EXT4_SB(s)->s_desc_per_block_bits)
347 #else
348 # define EXT4_BLOCKS_PER_GROUP(s)	((s)->s_blocks_per_group)
349 # define EXT4_DESC_PER_BLOCK(s)		(EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
350 # define EXT4_INODES_PER_GROUP(s)	((s)->s_inodes_per_group)
351 #endif
352 
353 /*
354  * Constants relative to the data blocks
355  */
356 #define	EXT4_NDIR_BLOCKS		12
357 #define	EXT4_IND_BLOCK			EXT4_NDIR_BLOCKS
358 #define	EXT4_DIND_BLOCK			(EXT4_IND_BLOCK + 1)
359 #define	EXT4_TIND_BLOCK			(EXT4_DIND_BLOCK + 1)
360 #define	EXT4_N_BLOCKS			(EXT4_TIND_BLOCK + 1)
361 
362 /*
363  * Inode flags
364  */
365 #define	EXT4_SECRM_FL			0x00000001 /* Secure deletion */
366 #define	EXT4_UNRM_FL			0x00000002 /* Undelete */
367 #define	EXT4_COMPR_FL			0x00000004 /* Compress file */
368 #define EXT4_SYNC_FL			0x00000008 /* Synchronous updates */
369 #define EXT4_IMMUTABLE_FL		0x00000010 /* Immutable file */
370 #define EXT4_APPEND_FL			0x00000020 /* writes to file may only append */
371 #define EXT4_NODUMP_FL			0x00000040 /* do not dump file */
372 #define EXT4_NOATIME_FL			0x00000080 /* do not update atime */
373 /* Reserved for compression usage... */
374 #define EXT4_DIRTY_FL			0x00000100
375 #define EXT4_COMPRBLK_FL		0x00000200 /* One or more compressed clusters */
376 #define EXT4_NOCOMPR_FL			0x00000400 /* Don't compress */
377 	/* nb: was previously EXT2_ECOMPR_FL */
378 #define EXT4_ENCRYPT_FL			0x00000800 /* encrypted file */
379 /* End compression flags --- maybe not all used */
380 #define EXT4_INDEX_FL			0x00001000 /* hash-indexed directory */
381 #define EXT4_IMAGIC_FL			0x00002000 /* AFS directory */
382 #define EXT4_JOURNAL_DATA_FL		0x00004000 /* file data should be journaled */
383 #define EXT4_NOTAIL_FL			0x00008000 /* file tail should not be merged */
384 #define EXT4_DIRSYNC_FL			0x00010000 /* dirsync behaviour (directories only) */
385 #define EXT4_TOPDIR_FL			0x00020000 /* Top of directory hierarchies*/
386 #define EXT4_HUGE_FILE_FL               0x00040000 /* Set to each huge file */
387 #define EXT4_EXTENTS_FL			0x00080000 /* Inode uses extents */
388 #define EXT4_EA_INODE_FL	        0x00200000 /* Inode used for large EA */
389 #define EXT4_EOFBLOCKS_FL		0x00400000 /* Blocks allocated beyond EOF */
390 #define EXT4_INLINE_DATA_FL		0x10000000 /* Inode has inline data. */
391 #define EXT4_RESERVED_FL		0x80000000 /* reserved for ext4 lib */
392 
393 #define EXT4_FL_USER_VISIBLE		0x004BDFFF /* User visible flags */
394 #define EXT4_FL_USER_MODIFIABLE		0x004380FF /* User modifiable flags */
395 
396 /* Flags that should be inherited by new inodes from their parent. */
397 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
398 			   EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
399 			   EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
400 			   EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
401 
402 /* Flags that are appropriate for regular files (all but dir-specific ones). */
403 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
404 
405 /* Flags that are appropriate for non-directories/regular files. */
406 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
407 
408 /* Mask out flags that are inappropriate for the given type of inode. */
ext4_mask_flags(umode_t mode,__u32 flags)409 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
410 {
411 	if (S_ISDIR(mode))
412 		return flags;
413 	else if (S_ISREG(mode))
414 		return flags & EXT4_REG_FLMASK;
415 	else
416 		return flags & EXT4_OTHER_FLMASK;
417 }
418 
419 /*
420  * Inode flags used for atomic set/get
421  */
422 enum {
423 	EXT4_INODE_SECRM	= 0,	/* Secure deletion */
424 	EXT4_INODE_UNRM		= 1,	/* Undelete */
425 	EXT4_INODE_COMPR	= 2,	/* Compress file */
426 	EXT4_INODE_SYNC		= 3,	/* Synchronous updates */
427 	EXT4_INODE_IMMUTABLE	= 4,	/* Immutable file */
428 	EXT4_INODE_APPEND	= 5,	/* writes to file may only append */
429 	EXT4_INODE_NODUMP	= 6,	/* do not dump file */
430 	EXT4_INODE_NOATIME	= 7,	/* do not update atime */
431 /* Reserved for compression usage, co-opted for encryption usage */
432 	EXT4_INODE_DIRTY	= 8,
433 	EXT4_INODE_COMPRBLK	= 9,	/* One or more compressed clusters */
434 	EXT4_INODE_NOCOMPR	= 10,	/* Don't compress */
435 	EXT4_INODE_ENCRYPT	= 11,	/* Encrypted */
436 /* End compression flags --- maybe not all used */
437 	EXT4_INODE_INDEX	= 12,	/* hash-indexed directory */
438 	EXT4_INODE_IMAGIC	= 13,	/* AFS directory */
439 	EXT4_INODE_JOURNAL_DATA	= 14,	/* file data should be journaled */
440 	EXT4_INODE_NOTAIL	= 15,	/* file tail should not be merged */
441 	EXT4_INODE_DIRSYNC	= 16,	/* dirsync behaviour (directories only) */
442 	EXT4_INODE_TOPDIR	= 17,	/* Top of directory hierarchies*/
443 	EXT4_INODE_HUGE_FILE	= 18,	/* Set to each huge file */
444 	EXT4_INODE_EXTENTS	= 19,	/* Inode uses extents */
445 	EXT4_INODE_EA_INODE	= 21,	/* Inode used for large EA */
446 	EXT4_INODE_EOFBLOCKS	= 22,	/* Blocks allocated beyond EOF */
447 	EXT4_INODE_INLINE_DATA	= 28,	/* Data in inode. */
448 	EXT4_INODE_RESERVED	= 31,	/* reserved for ext4 lib */
449 };
450 
451 /*
452  * Since it's pretty easy to mix up bit numbers and hex values, we use a
453  * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
454  * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
455  * any extra space in the compiled kernel image, otherwise, the build will fail.
456  * It's important that these values are the same, since we are using
457  * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
458  * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
459  * values found in ext2, ext3 and ext4 filesystems, and of course the values
460  * defined in e2fsprogs.
461  *
462  * It's not paranoia if the Murphy's Law really *is* out to get you.  :-)
463  */
464 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
465 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
466 
ext4_check_flag_values(void)467 static inline void ext4_check_flag_values(void)
468 {
469 	CHECK_FLAG_VALUE(SECRM);
470 	CHECK_FLAG_VALUE(UNRM);
471 	CHECK_FLAG_VALUE(COMPR);
472 	CHECK_FLAG_VALUE(SYNC);
473 	CHECK_FLAG_VALUE(IMMUTABLE);
474 	CHECK_FLAG_VALUE(APPEND);
475 	CHECK_FLAG_VALUE(NODUMP);
476 	CHECK_FLAG_VALUE(NOATIME);
477 	CHECK_FLAG_VALUE(DIRTY);
478 	CHECK_FLAG_VALUE(COMPRBLK);
479 	CHECK_FLAG_VALUE(NOCOMPR);
480 	CHECK_FLAG_VALUE(ENCRYPT);
481 	CHECK_FLAG_VALUE(INDEX);
482 	CHECK_FLAG_VALUE(IMAGIC);
483 	CHECK_FLAG_VALUE(JOURNAL_DATA);
484 	CHECK_FLAG_VALUE(NOTAIL);
485 	CHECK_FLAG_VALUE(DIRSYNC);
486 	CHECK_FLAG_VALUE(TOPDIR);
487 	CHECK_FLAG_VALUE(HUGE_FILE);
488 	CHECK_FLAG_VALUE(EXTENTS);
489 	CHECK_FLAG_VALUE(EA_INODE);
490 	CHECK_FLAG_VALUE(EOFBLOCKS);
491 	CHECK_FLAG_VALUE(INLINE_DATA);
492 	CHECK_FLAG_VALUE(RESERVED);
493 }
494 
495 /* Used to pass group descriptor data when online resize is done */
496 struct ext4_new_group_input {
497 	__u32 group;		/* Group number for this data */
498 	__u64 block_bitmap;	/* Absolute block number of block bitmap */
499 	__u64 inode_bitmap;	/* Absolute block number of inode bitmap */
500 	__u64 inode_table;	/* Absolute block number of inode table start */
501 	__u32 blocks_count;	/* Total number of blocks in this group */
502 	__u16 reserved_blocks;	/* Number of reserved blocks in this group */
503 	__u16 unused;
504 };
505 
506 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
507 struct compat_ext4_new_group_input {
508 	u32 group;
509 	compat_u64 block_bitmap;
510 	compat_u64 inode_bitmap;
511 	compat_u64 inode_table;
512 	u32 blocks_count;
513 	u16 reserved_blocks;
514 	u16 unused;
515 };
516 #endif
517 
518 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
519 struct ext4_new_group_data {
520 	__u32 group;
521 	__u64 block_bitmap;
522 	__u64 inode_bitmap;
523 	__u64 inode_table;
524 	__u32 blocks_count;
525 	__u16 reserved_blocks;
526 	__u16 unused;
527 	__u32 free_blocks_count;
528 };
529 
530 /* Indexes used to index group tables in ext4_new_group_data */
531 enum {
532 	BLOCK_BITMAP = 0,	/* block bitmap */
533 	INODE_BITMAP,		/* inode bitmap */
534 	INODE_TABLE,		/* inode tables */
535 	GROUP_TABLE_COUNT,
536 };
537 
538 /*
539  * Flags used by ext4_map_blocks()
540  */
541 	/* Allocate any needed blocks and/or convert an unwritten
542 	   extent to be an initialized ext4 */
543 #define EXT4_GET_BLOCKS_CREATE			0x0001
544 	/* Request the creation of an unwritten extent */
545 #define EXT4_GET_BLOCKS_UNWRIT_EXT		0x0002
546 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT	(EXT4_GET_BLOCKS_UNWRIT_EXT|\
547 						 EXT4_GET_BLOCKS_CREATE)
548 	/* Caller is from the delayed allocation writeout path
549 	 * finally doing the actual allocation of delayed blocks */
550 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE	0x0004
551 	/* caller is from the direct IO path, request to creation of an
552 	unwritten extents if not allocated, split the unwritten
553 	extent if blocks has been preallocated already*/
554 #define EXT4_GET_BLOCKS_PRE_IO			0x0008
555 #define EXT4_GET_BLOCKS_CONVERT			0x0010
556 #define EXT4_GET_BLOCKS_IO_CREATE_EXT		(EXT4_GET_BLOCKS_PRE_IO|\
557 					 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
558 	/* Convert extent to initialized after IO complete */
559 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT		(EXT4_GET_BLOCKS_CONVERT|\
560 					 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
561 	/* Eventual metadata allocation (due to growing extent tree)
562 	 * should not fail, so try to use reserved blocks for that.*/
563 #define EXT4_GET_BLOCKS_METADATA_NOFAIL		0x0020
564 	/* Don't normalize allocation size (used for fallocate) */
565 #define EXT4_GET_BLOCKS_NO_NORMALIZE		0x0040
566 	/* Request will not result in inode size update (user for fallocate) */
567 #define EXT4_GET_BLOCKS_KEEP_SIZE		0x0080
568 	/* Do not take i_data_sem locking in ext4_map_blocks */
569 #define EXT4_GET_BLOCKS_NO_LOCK			0x0100
570 	/* Do not put hole in extent cache */
571 #define EXT4_GET_BLOCKS_NO_PUT_HOLE		0x0200
572 	/* Convert written extents to unwritten */
573 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN	0x0400
574 
575 /*
576  * The bit position of these flags must not overlap with any of the
577  * EXT4_GET_BLOCKS_*.  They are used by ext4_find_extent(),
578  * read_extent_tree_block(), ext4_split_extent_at(),
579  * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
580  * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
581  * caching the extents when reading from the extent tree while a
582  * truncate or punch hole operation is in progress.
583  */
584 #define EXT4_EX_NOCACHE				0x40000000
585 #define EXT4_EX_FORCE_CACHE			0x20000000
586 
587 /*
588  * Flags used by ext4_free_blocks
589  */
590 #define EXT4_FREE_BLOCKS_METADATA	0x0001
591 #define EXT4_FREE_BLOCKS_FORGET		0x0002
592 #define EXT4_FREE_BLOCKS_VALIDATED	0x0004
593 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE	0x0008
594 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER	0x0010
595 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER	0x0020
596 
597 /* Encryption algorithms */
598 #define EXT4_ENCRYPTION_MODE_INVALID		0
599 #define EXT4_ENCRYPTION_MODE_AES_256_XTS	1
600 #define EXT4_ENCRYPTION_MODE_AES_256_GCM	2
601 #define EXT4_ENCRYPTION_MODE_AES_256_CBC	3
602 #define EXT4_ENCRYPTION_MODE_AES_256_CTS	4
603 
604 #include "ext4_crypto.h"
605 
606 /*
607  * ioctl commands
608  */
609 #define	EXT4_IOC_GETFLAGS		FS_IOC_GETFLAGS
610 #define	EXT4_IOC_SETFLAGS		FS_IOC_SETFLAGS
611 #define	EXT4_IOC_GETVERSION		_IOR('f', 3, long)
612 #define	EXT4_IOC_SETVERSION		_IOW('f', 4, long)
613 #define	EXT4_IOC_GETVERSION_OLD		FS_IOC_GETVERSION
614 #define	EXT4_IOC_SETVERSION_OLD		FS_IOC_SETVERSION
615 #define EXT4_IOC_GETRSVSZ		_IOR('f', 5, long)
616 #define EXT4_IOC_SETRSVSZ		_IOW('f', 6, long)
617 #define EXT4_IOC_GROUP_EXTEND		_IOW('f', 7, unsigned long)
618 #define EXT4_IOC_GROUP_ADD		_IOW('f', 8, struct ext4_new_group_input)
619 #define EXT4_IOC_MIGRATE		_IO('f', 9)
620  /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
621  /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
622 #define EXT4_IOC_ALLOC_DA_BLKS		_IO('f', 12)
623 #define EXT4_IOC_MOVE_EXT		_IOWR('f', 15, struct move_extent)
624 #define EXT4_IOC_RESIZE_FS		_IOW('f', 16, __u64)
625 #define EXT4_IOC_SWAP_BOOT		_IO('f', 17)
626 #define EXT4_IOC_PRECACHE_EXTENTS	_IO('f', 18)
627 #define EXT4_IOC_SET_ENCRYPTION_POLICY	_IOR('f', 19, struct ext4_encryption_policy)
628 #define EXT4_IOC_GET_ENCRYPTION_PWSALT	_IOW('f', 20, __u8[16])
629 #define EXT4_IOC_GET_ENCRYPTION_POLICY	_IOW('f', 21, struct ext4_encryption_policy)
630 
631 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
632 /*
633  * ioctl commands in 32 bit emulation
634  */
635 #define EXT4_IOC32_GETFLAGS		FS_IOC32_GETFLAGS
636 #define EXT4_IOC32_SETFLAGS		FS_IOC32_SETFLAGS
637 #define EXT4_IOC32_GETVERSION		_IOR('f', 3, int)
638 #define EXT4_IOC32_SETVERSION		_IOW('f', 4, int)
639 #define EXT4_IOC32_GETRSVSZ		_IOR('f', 5, int)
640 #define EXT4_IOC32_SETRSVSZ		_IOW('f', 6, int)
641 #define EXT4_IOC32_GROUP_EXTEND		_IOW('f', 7, unsigned int)
642 #define EXT4_IOC32_GROUP_ADD		_IOW('f', 8, struct compat_ext4_new_group_input)
643 #define EXT4_IOC32_GETVERSION_OLD	FS_IOC32_GETVERSION
644 #define EXT4_IOC32_SETVERSION_OLD	FS_IOC32_SETVERSION
645 #endif
646 
647 /* Max physical block we can address w/o extents */
648 #define EXT4_MAX_BLOCK_FILE_PHYS	0xFFFFFFFF
649 
650 /*
651  * Structure of an inode on the disk
652  */
653 struct ext4_inode {
654 	__le16	i_mode;		/* File mode */
655 	__le16	i_uid;		/* Low 16 bits of Owner Uid */
656 	__le32	i_size_lo;	/* Size in bytes */
657 	__le32	i_atime;	/* Access time */
658 	__le32	i_ctime;	/* Inode Change time */
659 	__le32	i_mtime;	/* Modification time */
660 	__le32	i_dtime;	/* Deletion Time */
661 	__le16	i_gid;		/* Low 16 bits of Group Id */
662 	__le16	i_links_count;	/* Links count */
663 	__le32	i_blocks_lo;	/* Blocks count */
664 	__le32	i_flags;	/* File flags */
665 	union {
666 		struct {
667 			__le32  l_i_version;
668 		} linux1;
669 		struct {
670 			__u32  h_i_translator;
671 		} hurd1;
672 		struct {
673 			__u32  m_i_reserved1;
674 		} masix1;
675 	} osd1;				/* OS dependent 1 */
676 	__le32	i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
677 	__le32	i_generation;	/* File version (for NFS) */
678 	__le32	i_file_acl_lo;	/* File ACL */
679 	__le32	i_size_high;
680 	__le32	i_obso_faddr;	/* Obsoleted fragment address */
681 	union {
682 		struct {
683 			__le16	l_i_blocks_high; /* were l_i_reserved1 */
684 			__le16	l_i_file_acl_high;
685 			__le16	l_i_uid_high;	/* these 2 fields */
686 			__le16	l_i_gid_high;	/* were reserved2[0] */
687 			__le16	l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
688 			__le16	l_i_reserved;
689 		} linux2;
690 		struct {
691 			__le16	h_i_reserved1;	/* Obsoleted fragment number/size which are removed in ext4 */
692 			__u16	h_i_mode_high;
693 			__u16	h_i_uid_high;
694 			__u16	h_i_gid_high;
695 			__u32	h_i_author;
696 		} hurd2;
697 		struct {
698 			__le16	h_i_reserved1;	/* Obsoleted fragment number/size which are removed in ext4 */
699 			__le16	m_i_file_acl_high;
700 			__u32	m_i_reserved2[2];
701 		} masix2;
702 	} osd2;				/* OS dependent 2 */
703 	__le16	i_extra_isize;
704 	__le16	i_checksum_hi;	/* crc32c(uuid+inum+inode) BE */
705 	__le32  i_ctime_extra;  /* extra Change time      (nsec << 2 | epoch) */
706 	__le32  i_mtime_extra;  /* extra Modification time(nsec << 2 | epoch) */
707 	__le32  i_atime_extra;  /* extra Access time      (nsec << 2 | epoch) */
708 	__le32  i_crtime;       /* File Creation time */
709 	__le32  i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
710 	__le32  i_version_hi;	/* high 32 bits for 64-bit version */
711 };
712 
713 struct move_extent {
714 	__u32 reserved;		/* should be zero */
715 	__u32 donor_fd;		/* donor file descriptor */
716 	__u64 orig_start;	/* logical start offset in block for orig */
717 	__u64 donor_start;	/* logical start offset in block for donor */
718 	__u64 len;		/* block length to be moved */
719 	__u64 moved_len;	/* moved block length */
720 };
721 
722 #define EXT4_EPOCH_BITS 2
723 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
724 #define EXT4_NSEC_MASK  (~0UL << EXT4_EPOCH_BITS)
725 
726 /*
727  * Extended fields will fit into an inode if the filesystem was formatted
728  * with large inodes (-I 256 or larger) and there are not currently any EAs
729  * consuming all of the available space. For new inodes we always reserve
730  * enough space for the kernel's known extended fields, but for inodes
731  * created with an old kernel this might not have been the case. None of
732  * the extended inode fields is critical for correct filesystem operation.
733  * This macro checks if a certain field fits in the inode. Note that
734  * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
735  */
736 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field)	\
737 	((offsetof(typeof(*ext4_inode), field) +	\
738 	  sizeof((ext4_inode)->field))			\
739 	<= (EXT4_GOOD_OLD_INODE_SIZE +			\
740 	    (einode)->i_extra_isize))			\
741 
ext4_encode_extra_time(struct timespec * time)742 static inline __le32 ext4_encode_extra_time(struct timespec *time)
743 {
744        return cpu_to_le32((sizeof(time->tv_sec) > 4 ?
745 			   (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) |
746                           ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK));
747 }
748 
ext4_decode_extra_time(struct timespec * time,__le32 extra)749 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
750 {
751        if (sizeof(time->tv_sec) > 4)
752 	       time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK)
753 			       << 32;
754        time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
755 }
756 
757 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode)			       \
758 do {									       \
759 	(raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec);	       \
760 	if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     \
761 		(raw_inode)->xtime ## _extra =				       \
762 				ext4_encode_extra_time(&(inode)->xtime);       \
763 } while (0)
764 
765 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode)			       \
766 do {									       \
767 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))		       \
768 		(raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec);      \
769 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))	       \
770 		(raw_inode)->xtime ## _extra =				       \
771 				ext4_encode_extra_time(&(einode)->xtime);      \
772 } while (0)
773 
774 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode)			       \
775 do {									       \
776 	(inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime);       \
777 	if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     \
778 		ext4_decode_extra_time(&(inode)->xtime,			       \
779 				       raw_inode->xtime ## _extra);	       \
780 	else								       \
781 		(inode)->xtime.tv_nsec = 0;				       \
782 } while (0)
783 
784 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode)			       \
785 do {									       \
786 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))		       \
787 		(einode)->xtime.tv_sec = 				       \
788 			(signed)le32_to_cpu((raw_inode)->xtime);	       \
789 	else								       \
790 		(einode)->xtime.tv_sec = 0;				       \
791 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))	       \
792 		ext4_decode_extra_time(&(einode)->xtime,		       \
793 				       raw_inode->xtime ## _extra);	       \
794 	else								       \
795 		(einode)->xtime.tv_nsec = 0;				       \
796 } while (0)
797 
798 #define i_disk_version osd1.linux1.l_i_version
799 
800 #if defined(__KERNEL__) || defined(__linux__)
801 #define i_reserved1	osd1.linux1.l_i_reserved1
802 #define i_file_acl_high	osd2.linux2.l_i_file_acl_high
803 #define i_blocks_high	osd2.linux2.l_i_blocks_high
804 #define i_uid_low	i_uid
805 #define i_gid_low	i_gid
806 #define i_uid_high	osd2.linux2.l_i_uid_high
807 #define i_gid_high	osd2.linux2.l_i_gid_high
808 #define i_checksum_lo	osd2.linux2.l_i_checksum_lo
809 
810 #elif defined(__GNU__)
811 
812 #define i_translator	osd1.hurd1.h_i_translator
813 #define i_uid_high	osd2.hurd2.h_i_uid_high
814 #define i_gid_high	osd2.hurd2.h_i_gid_high
815 #define i_author	osd2.hurd2.h_i_author
816 
817 #elif defined(__masix__)
818 
819 #define i_reserved1	osd1.masix1.m_i_reserved1
820 #define i_file_acl_high	osd2.masix2.m_i_file_acl_high
821 #define i_reserved2	osd2.masix2.m_i_reserved2
822 
823 #endif /* defined(__KERNEL__) || defined(__linux__) */
824 
825 #include "extents_status.h"
826 
827 /*
828  * Lock subclasses for i_data_sem in the ext4_inode_info structure.
829  *
830  * These are needed to avoid lockdep false positives when we need to
831  * allocate blocks to the quota inode during ext4_map_blocks(), while
832  * holding i_data_sem for a normal (non-quota) inode.  Since we don't
833  * do quota tracking for the quota inode, this avoids deadlock (as
834  * well as infinite recursion, since it isn't turtles all the way
835  * down...)
836  *
837  *  I_DATA_SEM_NORMAL - Used for most inodes
838  *  I_DATA_SEM_OTHER  - Used by move_inode.c for the second normal inode
839  *			  where the second inode has larger inode number
840  *			  than the first
841  *  I_DATA_SEM_QUOTA  - Used for quota inodes only
842  */
843 enum {
844 	I_DATA_SEM_NORMAL = 0,
845 	I_DATA_SEM_OTHER,
846 	I_DATA_SEM_QUOTA,
847 };
848 
849 
850 /*
851  * fourth extended file system inode data in memory
852  */
853 struct ext4_inode_info {
854 	__le32	i_data[15];	/* unconverted */
855 	__u32	i_dtime;
856 	ext4_fsblk_t	i_file_acl;
857 
858 	/*
859 	 * i_block_group is the number of the block group which contains
860 	 * this file's inode.  Constant across the lifetime of the inode,
861 	 * it is ued for making block allocation decisions - we try to
862 	 * place a file's data blocks near its inode block, and new inodes
863 	 * near to their parent directory's inode.
864 	 */
865 	ext4_group_t	i_block_group;
866 	ext4_lblk_t	i_dir_start_lookup;
867 #if (BITS_PER_LONG < 64)
868 	unsigned long	i_state_flags;		/* Dynamic state flags */
869 #endif
870 	unsigned long	i_flags;
871 
872 	/*
873 	 * Extended attributes can be read independently of the main file
874 	 * data. Taking i_mutex even when reading would cause contention
875 	 * between readers of EAs and writers of regular file data, so
876 	 * instead we synchronize on xattr_sem when reading or changing
877 	 * EAs.
878 	 */
879 	struct rw_semaphore xattr_sem;
880 
881 	struct list_head i_orphan;	/* unlinked but open inodes */
882 
883 	/*
884 	 * i_disksize keeps track of what the inode size is ON DISK, not
885 	 * in memory.  During truncate, i_size is set to the new size by
886 	 * the VFS prior to calling ext4_truncate(), but the filesystem won't
887 	 * set i_disksize to 0 until the truncate is actually under way.
888 	 *
889 	 * The intent is that i_disksize always represents the blocks which
890 	 * are used by this file.  This allows recovery to restart truncate
891 	 * on orphans if we crash during truncate.  We actually write i_disksize
892 	 * into the on-disk inode when writing inodes out, instead of i_size.
893 	 *
894 	 * The only time when i_disksize and i_size may be different is when
895 	 * a truncate is in progress.  The only things which change i_disksize
896 	 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
897 	 */
898 	loff_t	i_disksize;
899 
900 	/*
901 	 * i_data_sem is for serialising ext4_truncate() against
902 	 * ext4_getblock().  In the 2.4 ext2 design, great chunks of inode's
903 	 * data tree are chopped off during truncate. We can't do that in
904 	 * ext4 because whenever we perform intermediate commits during
905 	 * truncate, the inode and all the metadata blocks *must* be in a
906 	 * consistent state which allows truncation of the orphans to restart
907 	 * during recovery.  Hence we must fix the get_block-vs-truncate race
908 	 * by other means, so we have i_data_sem.
909 	 */
910 	struct rw_semaphore i_data_sem;
911 	struct inode vfs_inode;
912 	struct jbd2_inode *jinode;
913 
914 	spinlock_t i_raw_lock;	/* protects updates to the raw inode */
915 
916 	/*
917 	 * File creation time. Its function is same as that of
918 	 * struct timespec i_{a,c,m}time in the generic inode.
919 	 */
920 	struct timespec i_crtime;
921 
922 	/* mballoc */
923 	struct list_head i_prealloc_list;
924 	spinlock_t i_prealloc_lock;
925 
926 	/* extents status tree */
927 	struct ext4_es_tree i_es_tree;
928 	rwlock_t i_es_lock;
929 	struct list_head i_es_lru;
930 	unsigned int i_es_all_nr;	/* protected by i_es_lock */
931 	unsigned int i_es_lru_nr;	/* protected by i_es_lock */
932 	unsigned long i_touch_when;	/* jiffies of last accessing */
933 
934 	/* ialloc */
935 	ext4_group_t	i_last_alloc_group;
936 
937 	/* allocation reservation info for delalloc */
938 	/* In case of bigalloc, these refer to clusters rather than blocks */
939 	unsigned int i_reserved_data_blocks;
940 	unsigned int i_reserved_meta_blocks;
941 	unsigned int i_allocated_meta_blocks;
942 	ext4_lblk_t i_da_metadata_calc_last_lblock;
943 	int i_da_metadata_calc_len;
944 
945 	/* on-disk additional length */
946 	__u16 i_extra_isize;
947 
948 	/* Indicate the inline data space. */
949 	u16 i_inline_off;
950 	u16 i_inline_size;
951 
952 #ifdef CONFIG_QUOTA
953 	/* quota space reservation, managed internally by quota code */
954 	qsize_t i_reserved_quota;
955 #endif
956 
957 	/* Lock protecting lists below */
958 	spinlock_t i_completed_io_lock;
959 	/*
960 	 * Completed IOs that need unwritten extents handling and have
961 	 * transaction reserved
962 	 */
963 	struct list_head i_rsv_conversion_list;
964 	/*
965 	 * Completed IOs that need unwritten extents handling and don't have
966 	 * transaction reserved
967 	 */
968 	atomic_t i_ioend_count;	/* Number of outstanding io_end structs */
969 	atomic_t i_unwritten; /* Nr. of inflight conversions pending */
970 	struct work_struct i_rsv_conversion_work;
971 
972 	spinlock_t i_block_reservation_lock;
973 
974 	/*
975 	 * Transactions that contain inode's metadata needed to complete
976 	 * fsync and fdatasync, respectively.
977 	 */
978 	tid_t i_sync_tid;
979 	tid_t i_datasync_tid;
980 
981 	/* Precomputed uuid+inum+igen checksum for seeding inode checksums */
982 	__u32 i_csum_seed;
983 
984 #ifdef CONFIG_EXT4_FS_ENCRYPTION
985 	/* Encryption params */
986 	struct ext4_crypt_info *i_crypt_info;
987 #endif
988 };
989 
990 /*
991  * File system states
992  */
993 #define	EXT4_VALID_FS			0x0001	/* Unmounted cleanly */
994 #define	EXT4_ERROR_FS			0x0002	/* Errors detected */
995 #define	EXT4_ORPHAN_FS			0x0004	/* Orphans being recovered */
996 
997 /*
998  * Misc. filesystem flags
999  */
1000 #define EXT2_FLAGS_SIGNED_HASH		0x0001  /* Signed dirhash in use */
1001 #define EXT2_FLAGS_UNSIGNED_HASH	0x0002  /* Unsigned dirhash in use */
1002 #define EXT2_FLAGS_TEST_FILESYS		0x0004	/* to test development code */
1003 
1004 /*
1005  * Mount flags set via mount options or defaults
1006  */
1007 #define EXT4_MOUNT_GRPID		0x00004	/* Create files with directory's group */
1008 #define EXT4_MOUNT_DEBUG		0x00008	/* Some debugging messages */
1009 #define EXT4_MOUNT_ERRORS_CONT		0x00010	/* Continue on errors */
1010 #define EXT4_MOUNT_ERRORS_RO		0x00020	/* Remount fs ro on errors */
1011 #define EXT4_MOUNT_ERRORS_PANIC		0x00040	/* Panic on errors */
1012 #define EXT4_MOUNT_ERRORS_MASK		0x00070
1013 #define EXT4_MOUNT_MINIX_DF		0x00080	/* Mimics the Minix statfs */
1014 #define EXT4_MOUNT_NOLOAD		0x00100	/* Don't use existing journal*/
1015 #define EXT4_MOUNT_DATA_FLAGS		0x00C00	/* Mode for data writes: */
1016 #define EXT4_MOUNT_JOURNAL_DATA		0x00400	/* Write data to journal */
1017 #define EXT4_MOUNT_ORDERED_DATA		0x00800	/* Flush data before commit */
1018 #define EXT4_MOUNT_WRITEBACK_DATA	0x00C00	/* No data ordering */
1019 #define EXT4_MOUNT_UPDATE_JOURNAL	0x01000	/* Update the journal format */
1020 #define EXT4_MOUNT_NO_UID32		0x02000  /* Disable 32-bit UIDs */
1021 #define EXT4_MOUNT_XATTR_USER		0x04000	/* Extended user attributes */
1022 #define EXT4_MOUNT_POSIX_ACL		0x08000	/* POSIX Access Control Lists */
1023 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC	0x10000	/* No auto delalloc mapping */
1024 #define EXT4_MOUNT_BARRIER		0x20000 /* Use block barriers */
1025 #define EXT4_MOUNT_QUOTA		0x80000 /* Some quota option set */
1026 #define EXT4_MOUNT_USRQUOTA		0x100000 /* "old" user quota */
1027 #define EXT4_MOUNT_GRPQUOTA		0x200000 /* "old" group quota */
1028 #define EXT4_MOUNT_DIOREAD_NOLOCK	0x400000 /* Enable support for dio read nolocking */
1029 #define EXT4_MOUNT_JOURNAL_CHECKSUM	0x800000 /* Journal checksums */
1030 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT	0x1000000 /* Journal Async Commit */
1031 #define EXT4_MOUNT_DELALLOC		0x8000000 /* Delalloc support */
1032 #define EXT4_MOUNT_DATA_ERR_ABORT	0x10000000 /* Abort on file data write */
1033 #define EXT4_MOUNT_BLOCK_VALIDITY	0x20000000 /* Block validity checking */
1034 #define EXT4_MOUNT_DISCARD		0x40000000 /* Issue DISCARD requests */
1035 #define EXT4_MOUNT_INIT_INODE_TABLE	0x80000000 /* Initialize uninitialized itables */
1036 
1037 /*
1038  * Mount flags set either automatically (could not be set by mount option)
1039  * based on per file system feature or property or in special cases such as
1040  * distinguishing between explicit mount option definition and default.
1041  */
1042 #define EXT4_MOUNT2_EXPLICIT_DELALLOC	0x00000001 /* User explicitly
1043 						      specified delalloc */
1044 #define EXT4_MOUNT2_STD_GROUP_SIZE	0x00000002 /* We have standard group
1045 						      size of blocksize * 8
1046 						      blocks */
1047 #define EXT4_MOUNT2_HURD_COMPAT		0x00000004 /* Support HURD-castrated
1048 						      file systems */
1049 
1050 #define clear_opt(sb, opt)		EXT4_SB(sb)->s_mount_opt &= \
1051 						~EXT4_MOUNT_##opt
1052 #define set_opt(sb, opt)		EXT4_SB(sb)->s_mount_opt |= \
1053 						EXT4_MOUNT_##opt
1054 #define test_opt(sb, opt)		(EXT4_SB(sb)->s_mount_opt & \
1055 					 EXT4_MOUNT_##opt)
1056 
1057 #define clear_opt2(sb, opt)		EXT4_SB(sb)->s_mount_opt2 &= \
1058 						~EXT4_MOUNT2_##opt
1059 #define set_opt2(sb, opt)		EXT4_SB(sb)->s_mount_opt2 |= \
1060 						EXT4_MOUNT2_##opt
1061 #define test_opt2(sb, opt)		(EXT4_SB(sb)->s_mount_opt2 & \
1062 					 EXT4_MOUNT2_##opt)
1063 
1064 #define ext4_test_and_set_bit		__test_and_set_bit_le
1065 #define ext4_set_bit			__set_bit_le
1066 #define ext4_set_bit_atomic		ext2_set_bit_atomic
1067 #define ext4_test_and_clear_bit		__test_and_clear_bit_le
1068 #define ext4_clear_bit			__clear_bit_le
1069 #define ext4_clear_bit_atomic		ext2_clear_bit_atomic
1070 #define ext4_test_bit			test_bit_le
1071 #define ext4_find_next_zero_bit		find_next_zero_bit_le
1072 #define ext4_find_next_bit		find_next_bit_le
1073 
1074 extern void ext4_set_bits(void *bm, int cur, int len);
1075 
1076 /*
1077  * Maximal mount counts between two filesystem checks
1078  */
1079 #define EXT4_DFL_MAX_MNT_COUNT		20	/* Allow 20 mounts */
1080 #define EXT4_DFL_CHECKINTERVAL		0	/* Don't use interval check */
1081 
1082 /*
1083  * Behaviour when detecting errors
1084  */
1085 #define EXT4_ERRORS_CONTINUE		1	/* Continue execution */
1086 #define EXT4_ERRORS_RO			2	/* Remount fs read-only */
1087 #define EXT4_ERRORS_PANIC		3	/* Panic */
1088 #define EXT4_ERRORS_DEFAULT		EXT4_ERRORS_CONTINUE
1089 
1090 /* Metadata checksum algorithm codes */
1091 #define EXT4_CRC32C_CHKSUM		1
1092 
1093 /*
1094  * Structure of the super block
1095  */
1096 struct ext4_super_block {
1097 /*00*/	__le32	s_inodes_count;		/* Inodes count */
1098 	__le32	s_blocks_count_lo;	/* Blocks count */
1099 	__le32	s_r_blocks_count_lo;	/* Reserved blocks count */
1100 	__le32	s_free_blocks_count_lo;	/* Free blocks count */
1101 /*10*/	__le32	s_free_inodes_count;	/* Free inodes count */
1102 	__le32	s_first_data_block;	/* First Data Block */
1103 	__le32	s_log_block_size;	/* Block size */
1104 	__le32	s_log_cluster_size;	/* Allocation cluster size */
1105 /*20*/	__le32	s_blocks_per_group;	/* # Blocks per group */
1106 	__le32	s_clusters_per_group;	/* # Clusters per group */
1107 	__le32	s_inodes_per_group;	/* # Inodes per group */
1108 	__le32	s_mtime;		/* Mount time */
1109 /*30*/	__le32	s_wtime;		/* Write time */
1110 	__le16	s_mnt_count;		/* Mount count */
1111 	__le16	s_max_mnt_count;	/* Maximal mount count */
1112 	__le16	s_magic;		/* Magic signature */
1113 	__le16	s_state;		/* File system state */
1114 	__le16	s_errors;		/* Behaviour when detecting errors */
1115 	__le16	s_minor_rev_level;	/* minor revision level */
1116 /*40*/	__le32	s_lastcheck;		/* time of last check */
1117 	__le32	s_checkinterval;	/* max. time between checks */
1118 	__le32	s_creator_os;		/* OS */
1119 	__le32	s_rev_level;		/* Revision level */
1120 /*50*/	__le16	s_def_resuid;		/* Default uid for reserved blocks */
1121 	__le16	s_def_resgid;		/* Default gid for reserved blocks */
1122 	/*
1123 	 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1124 	 *
1125 	 * Note: the difference between the compatible feature set and
1126 	 * the incompatible feature set is that if there is a bit set
1127 	 * in the incompatible feature set that the kernel doesn't
1128 	 * know about, it should refuse to mount the filesystem.
1129 	 *
1130 	 * e2fsck's requirements are more strict; if it doesn't know
1131 	 * about a feature in either the compatible or incompatible
1132 	 * feature set, it must abort and not try to meddle with
1133 	 * things it doesn't understand...
1134 	 */
1135 	__le32	s_first_ino;		/* First non-reserved inode */
1136 	__le16  s_inode_size;		/* size of inode structure */
1137 	__le16	s_block_group_nr;	/* block group # of this superblock */
1138 	__le32	s_feature_compat;	/* compatible feature set */
1139 /*60*/	__le32	s_feature_incompat;	/* incompatible feature set */
1140 	__le32	s_feature_ro_compat;	/* readonly-compatible feature set */
1141 /*68*/	__u8	s_uuid[16];		/* 128-bit uuid for volume */
1142 /*78*/	char	s_volume_name[16];	/* volume name */
1143 /*88*/	char	s_last_mounted[64];	/* directory where last mounted */
1144 /*C8*/	__le32	s_algorithm_usage_bitmap; /* For compression */
1145 	/*
1146 	 * Performance hints.  Directory preallocation should only
1147 	 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1148 	 */
1149 	__u8	s_prealloc_blocks;	/* Nr of blocks to try to preallocate*/
1150 	__u8	s_prealloc_dir_blocks;	/* Nr to preallocate for dirs */
1151 	__le16	s_reserved_gdt_blocks;	/* Per group desc for online growth */
1152 	/*
1153 	 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1154 	 */
1155 /*D0*/	__u8	s_journal_uuid[16];	/* uuid of journal superblock */
1156 /*E0*/	__le32	s_journal_inum;		/* inode number of journal file */
1157 	__le32	s_journal_dev;		/* device number of journal file */
1158 	__le32	s_last_orphan;		/* start of list of inodes to delete */
1159 	__le32	s_hash_seed[4];		/* HTREE hash seed */
1160 	__u8	s_def_hash_version;	/* Default hash version to use */
1161 	__u8	s_jnl_backup_type;
1162 	__le16  s_desc_size;		/* size of group descriptor */
1163 /*100*/	__le32	s_default_mount_opts;
1164 	__le32	s_first_meta_bg;	/* First metablock block group */
1165 	__le32	s_mkfs_time;		/* When the filesystem was created */
1166 	__le32	s_jnl_blocks[17];	/* Backup of the journal inode */
1167 	/* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1168 /*150*/	__le32	s_blocks_count_hi;	/* Blocks count */
1169 	__le32	s_r_blocks_count_hi;	/* Reserved blocks count */
1170 	__le32	s_free_blocks_count_hi;	/* Free blocks count */
1171 	__le16	s_min_extra_isize;	/* All inodes have at least # bytes */
1172 	__le16	s_want_extra_isize; 	/* New inodes should reserve # bytes */
1173 	__le32	s_flags;		/* Miscellaneous flags */
1174 	__le16  s_raid_stride;		/* RAID stride */
1175 	__le16  s_mmp_update_interval;  /* # seconds to wait in MMP checking */
1176 	__le64  s_mmp_block;            /* Block for multi-mount protection */
1177 	__le32  s_raid_stripe_width;    /* blocks on all data disks (N*stride)*/
1178 	__u8	s_log_groups_per_flex;  /* FLEX_BG group size */
1179 	__u8	s_checksum_type;	/* metadata checksum algorithm used */
1180 	__u8	s_encryption_level;	/* versioning level for encryption */
1181 	__u8	s_reserved_pad;		/* Padding to next 32bits */
1182 	__le64	s_kbytes_written;	/* nr of lifetime kilobytes written */
1183 	__le32	s_snapshot_inum;	/* Inode number of active snapshot */
1184 	__le32	s_snapshot_id;		/* sequential ID of active snapshot */
1185 	__le64	s_snapshot_r_blocks_count; /* reserved blocks for active
1186 					      snapshot's future use */
1187 	__le32	s_snapshot_list;	/* inode number of the head of the
1188 					   on-disk snapshot list */
1189 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1190 	__le32	s_error_count;		/* number of fs errors */
1191 	__le32	s_first_error_time;	/* first time an error happened */
1192 	__le32	s_first_error_ino;	/* inode involved in first error */
1193 	__le64	s_first_error_block;	/* block involved of first error */
1194 	__u8	s_first_error_func[32];	/* function where the error happened */
1195 	__le32	s_first_error_line;	/* line number where error happened */
1196 	__le32	s_last_error_time;	/* most recent time of an error */
1197 	__le32	s_last_error_ino;	/* inode involved in last error */
1198 	__le32	s_last_error_line;	/* line number where error happened */
1199 	__le64	s_last_error_block;	/* block involved of last error */
1200 	__u8	s_last_error_func[32];	/* function where the error happened */
1201 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1202 	__u8	s_mount_opts[64];
1203 	__le32	s_usr_quota_inum;	/* inode for tracking user quota */
1204 	__le32	s_grp_quota_inum;	/* inode for tracking group quota */
1205 	__le32	s_overhead_clusters;	/* overhead blocks/clusters in fs */
1206 	__le32	s_backup_bgs[2];	/* groups with sparse_super2 SBs */
1207 	__u8	s_encrypt_algos[4];	/* Encryption algorithms in use  */
1208 	__u8	s_encrypt_pw_salt[16];	/* Salt used for string2key algorithm */
1209 	__le32	s_lpf_ino;		/* Location of the lost+found inode */
1210 	__le32	s_reserved[100];	/* Padding to the end of the block */
1211 	__le32	s_checksum;		/* crc32c(superblock) */
1212 };
1213 
1214 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1215 
1216 #ifdef __KERNEL__
1217 
1218 /*
1219  * run-time mount flags
1220  */
1221 #define EXT4_MF_MNTDIR_SAMPLED		0x0001
1222 #define EXT4_MF_FS_ABORTED		0x0002	/* Fatal error detected */
1223 #define EXT4_MF_TEST_DUMMY_ENCRYPTION	0x0004
1224 
1225 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1226 #define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \
1227 						EXT4_MF_TEST_DUMMY_ENCRYPTION))
1228 #else
1229 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1230 #endif
1231 
1232 /* Number of quota types we support */
1233 #define EXT4_MAXQUOTAS 2
1234 
1235 /*
1236  * fourth extended-fs super-block data in memory
1237  */
1238 struct ext4_sb_info {
1239 	unsigned long s_desc_size;	/* Size of a group descriptor in bytes */
1240 	unsigned long s_inodes_per_block;/* Number of inodes per block */
1241 	unsigned long s_blocks_per_group;/* Number of blocks in a group */
1242 	unsigned long s_clusters_per_group; /* Number of clusters in a group */
1243 	unsigned long s_inodes_per_group;/* Number of inodes in a group */
1244 	unsigned long s_itb_per_group;	/* Number of inode table blocks per group */
1245 	unsigned long s_gdb_count;	/* Number of group descriptor blocks */
1246 	unsigned long s_desc_per_block;	/* Number of group descriptors per block */
1247 	ext4_group_t s_groups_count;	/* Number of groups in the fs */
1248 	ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1249 	unsigned long s_overhead;  /* # of fs overhead clusters */
1250 	unsigned int s_cluster_ratio;	/* Number of blocks per cluster */
1251 	unsigned int s_cluster_bits;	/* log2 of s_cluster_ratio */
1252 	loff_t s_bitmap_maxbytes;	/* max bytes for bitmap files */
1253 	struct buffer_head * s_sbh;	/* Buffer containing the super block */
1254 	struct ext4_super_block *s_es;	/* Pointer to the super block in the buffer */
1255 	struct buffer_head **s_group_desc;
1256 	unsigned int s_mount_opt;
1257 	unsigned int s_mount_opt2;
1258 	unsigned int s_mount_flags;
1259 	unsigned int s_def_mount_opt;
1260 	ext4_fsblk_t s_sb_block;
1261 	atomic64_t s_resv_clusters;
1262 	kuid_t s_resuid;
1263 	kgid_t s_resgid;
1264 	unsigned short s_mount_state;
1265 	unsigned short s_pad;
1266 	int s_addr_per_block_bits;
1267 	int s_desc_per_block_bits;
1268 	int s_inode_size;
1269 	int s_first_ino;
1270 	unsigned int s_inode_readahead_blks;
1271 	unsigned int s_inode_goal;
1272 	spinlock_t s_next_gen_lock;
1273 	u32 s_next_generation;
1274 	u32 s_hash_seed[4];
1275 	int s_def_hash_version;
1276 	int s_hash_unsigned;	/* 3 if hash should be signed, 0 if not */
1277 	struct percpu_counter s_freeclusters_counter;
1278 	struct percpu_counter s_freeinodes_counter;
1279 	struct percpu_counter s_dirs_counter;
1280 	struct percpu_counter s_dirtyclusters_counter;
1281 	struct blockgroup_lock *s_blockgroup_lock;
1282 	struct proc_dir_entry *s_proc;
1283 	struct kobject s_kobj;
1284 	struct completion s_kobj_unregister;
1285 	struct super_block *s_sb;
1286 
1287 	/* Journaling */
1288 	struct journal_s *s_journal;
1289 	struct list_head s_orphan;
1290 	struct mutex s_orphan_lock;
1291 	unsigned long s_resize_flags;		/* Flags indicating if there
1292 						   is a resizer */
1293 	unsigned long s_commit_interval;
1294 	u32 s_max_batch_time;
1295 	u32 s_min_batch_time;
1296 	struct block_device *journal_bdev;
1297 #ifdef CONFIG_QUOTA
1298 	char *s_qf_names[EXT4_MAXQUOTAS];	/* Names of quota files with journalled quota */
1299 	int s_jquota_fmt;			/* Format of quota to use */
1300 #endif
1301 	unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1302 	struct rb_root system_blks;
1303 
1304 #ifdef EXTENTS_STATS
1305 	/* ext4 extents stats */
1306 	unsigned long s_ext_min;
1307 	unsigned long s_ext_max;
1308 	unsigned long s_depth_max;
1309 	spinlock_t s_ext_stats_lock;
1310 	unsigned long s_ext_blocks;
1311 	unsigned long s_ext_extents;
1312 #endif
1313 
1314 	/* for buddy allocator */
1315 	struct ext4_group_info ***s_group_info;
1316 	struct inode *s_buddy_cache;
1317 	spinlock_t s_md_lock;
1318 	unsigned short *s_mb_offsets;
1319 	unsigned int *s_mb_maxs;
1320 	unsigned int s_group_info_size;
1321 
1322 	/* tunables */
1323 	unsigned long s_stripe;
1324 	unsigned int s_mb_stream_request;
1325 	unsigned int s_mb_max_to_scan;
1326 	unsigned int s_mb_min_to_scan;
1327 	unsigned int s_mb_stats;
1328 	unsigned int s_mb_order2_reqs;
1329 	unsigned int s_mb_group_prealloc;
1330 	unsigned int s_max_dir_size_kb;
1331 	/* where last allocation was done - for stream allocation */
1332 	unsigned long s_mb_last_group;
1333 	unsigned long s_mb_last_start;
1334 
1335 	/* stats for buddy allocator */
1336 	atomic_t s_bal_reqs;	/* number of reqs with len > 1 */
1337 	atomic_t s_bal_success;	/* we found long enough chunks */
1338 	atomic_t s_bal_allocated;	/* in blocks */
1339 	atomic_t s_bal_ex_scanned;	/* total extents scanned */
1340 	atomic_t s_bal_goals;	/* goal hits */
1341 	atomic_t s_bal_breaks;	/* too long searches */
1342 	atomic_t s_bal_2orders;	/* 2^order hits */
1343 	spinlock_t s_bal_lock;
1344 	unsigned long s_mb_buddies_generated;
1345 	unsigned long long s_mb_generation_time;
1346 	atomic_t s_mb_lost_chunks;
1347 	atomic_t s_mb_preallocated;
1348 	atomic_t s_mb_discarded;
1349 	atomic_t s_lock_busy;
1350 
1351 	/* locality groups */
1352 	struct ext4_locality_group __percpu *s_locality_groups;
1353 
1354 	/* for write statistics */
1355 	unsigned long s_sectors_written_start;
1356 	u64 s_kbytes_written;
1357 
1358 	/* the size of zero-out chunk */
1359 	unsigned int s_extent_max_zeroout_kb;
1360 
1361 	unsigned int s_log_groups_per_flex;
1362 	struct flex_groups *s_flex_groups;
1363 	ext4_group_t s_flex_groups_allocated;
1364 
1365 	/* workqueue for reserved extent conversions (buffered io) */
1366 	struct workqueue_struct *rsv_conversion_wq;
1367 
1368 	/* timer for periodic error stats printing */
1369 	struct timer_list s_err_report;
1370 
1371 	/* Lazy inode table initialization info */
1372 	struct ext4_li_request *s_li_request;
1373 	/* Wait multiplier for lazy initialization thread */
1374 	unsigned int s_li_wait_mult;
1375 
1376 	/* Kernel thread for multiple mount protection */
1377 	struct task_struct *s_mmp_tsk;
1378 
1379 	/* record the last minlen when FITRIM is called. */
1380 	atomic_t s_last_trim_minblks;
1381 
1382 	/* Reference to checksum algorithm driver via cryptoapi */
1383 	struct crypto_shash *s_chksum_driver;
1384 
1385 	/* Precomputed FS UUID checksum for seeding other checksums */
1386 	__u32 s_csum_seed;
1387 
1388 	/* Reclaim extents from extent status tree */
1389 	struct shrinker s_es_shrinker;
1390 	struct list_head s_es_lru;
1391 	struct ext4_es_stats s_es_stats;
1392 	struct mb_cache *s_mb_cache;
1393 	spinlock_t s_es_lru_lock ____cacheline_aligned_in_smp;
1394 
1395 	/* Ratelimit ext4 messages. */
1396 	struct ratelimit_state s_err_ratelimit_state;
1397 	struct ratelimit_state s_warning_ratelimit_state;
1398 	struct ratelimit_state s_msg_ratelimit_state;
1399 };
1400 
EXT4_SB(struct super_block * sb)1401 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1402 {
1403 	return sb->s_fs_info;
1404 }
EXT4_I(struct inode * inode)1405 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1406 {
1407 	return container_of(inode, struct ext4_inode_info, vfs_inode);
1408 }
1409 
ext4_current_time(struct inode * inode)1410 static inline struct timespec ext4_current_time(struct inode *inode)
1411 {
1412 	return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1413 		current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1414 }
1415 
ext4_valid_inum(struct super_block * sb,unsigned long ino)1416 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1417 {
1418 	return ino == EXT4_ROOT_INO ||
1419 		ino == EXT4_USR_QUOTA_INO ||
1420 		ino == EXT4_GRP_QUOTA_INO ||
1421 		ino == EXT4_BOOT_LOADER_INO ||
1422 		ino == EXT4_JOURNAL_INO ||
1423 		ino == EXT4_RESIZE_INO ||
1424 		(ino >= EXT4_FIRST_INO(sb) &&
1425 		 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1426 }
1427 
ext4_set_io_unwritten_flag(struct inode * inode,struct ext4_io_end * io_end)1428 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
1429 					      struct ext4_io_end *io_end)
1430 {
1431 	if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
1432 		io_end->flag |= EXT4_IO_END_UNWRITTEN;
1433 		atomic_inc(&EXT4_I(inode)->i_unwritten);
1434 	}
1435 }
1436 
ext4_inode_aio(struct inode * inode)1437 static inline ext4_io_end_t *ext4_inode_aio(struct inode *inode)
1438 {
1439 	return inode->i_private;
1440 }
1441 
ext4_inode_aio_set(struct inode * inode,ext4_io_end_t * io)1442 static inline void ext4_inode_aio_set(struct inode *inode, ext4_io_end_t *io)
1443 {
1444 	inode->i_private = io;
1445 }
1446 
1447 /*
1448  * Inode dynamic state flags
1449  */
1450 enum {
1451 	EXT4_STATE_JDATA,		/* journaled data exists */
1452 	EXT4_STATE_NEW,			/* inode is newly created */
1453 	EXT4_STATE_XATTR,		/* has in-inode xattrs */
1454 	EXT4_STATE_NO_EXPAND,		/* No space for expansion */
1455 	EXT4_STATE_DA_ALLOC_CLOSE,	/* Alloc DA blks on close */
1456 	EXT4_STATE_EXT_MIGRATE,		/* Inode is migrating */
1457 	EXT4_STATE_DIO_UNWRITTEN,	/* need convert on dio done*/
1458 	EXT4_STATE_NEWENTRY,		/* File just added to dir */
1459 	EXT4_STATE_DIOREAD_LOCK,	/* Disable support for dio read
1460 					   nolocking */
1461 	EXT4_STATE_MAY_INLINE_DATA,	/* may have in-inode data */
1462 	EXT4_STATE_ORDERED_MODE,	/* data=ordered mode */
1463 	EXT4_STATE_EXT_PRECACHED,	/* extents have been precached */
1464 };
1465 
1466 #define EXT4_INODE_BIT_FNS(name, field, offset)				\
1467 static inline int ext4_test_inode_##name(struct inode *inode, int bit)	\
1468 {									\
1469 	return test_bit(bit + (offset), &EXT4_I(inode)->i_##field);	\
1470 }									\
1471 static inline void ext4_set_inode_##name(struct inode *inode, int bit)	\
1472 {									\
1473 	set_bit(bit + (offset), &EXT4_I(inode)->i_##field);		\
1474 }									\
1475 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1476 {									\
1477 	clear_bit(bit + (offset), &EXT4_I(inode)->i_##field);		\
1478 }
1479 
1480 /* Add these declarations here only so that these functions can be
1481  * found by name.  Otherwise, they are very hard to locate. */
1482 static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1483 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1484 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1485 EXT4_INODE_BIT_FNS(flag, flags, 0)
1486 
1487 /* Add these declarations here only so that these functions can be
1488  * found by name.  Otherwise, they are very hard to locate. */
1489 static inline int ext4_test_inode_state(struct inode *inode, int bit);
1490 static inline void ext4_set_inode_state(struct inode *inode, int bit);
1491 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1492 #if (BITS_PER_LONG < 64)
1493 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1494 
ext4_clear_state_flags(struct ext4_inode_info * ei)1495 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1496 {
1497 	(ei)->i_state_flags = 0;
1498 }
1499 #else
1500 EXT4_INODE_BIT_FNS(state, flags, 32)
1501 
ext4_clear_state_flags(struct ext4_inode_info * ei)1502 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1503 {
1504 	/* We depend on the fact that callers will set i_flags */
1505 }
1506 #endif
1507 #else
1508 /* Assume that user mode programs are passing in an ext4fs superblock, not
1509  * a kernel struct super_block.  This will allow us to call the feature-test
1510  * macros from user land. */
1511 #define EXT4_SB(sb)	(sb)
1512 #endif
1513 
1514 /*
1515  * Returns true if the inode is inode is encrypted
1516  */
ext4_encrypted_inode(struct inode * inode)1517 static inline int ext4_encrypted_inode(struct inode *inode)
1518 {
1519 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1520 	return ext4_test_inode_flag(inode, EXT4_INODE_ENCRYPT);
1521 #else
1522 	return 0;
1523 #endif
1524 }
1525 
1526 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1527 
1528 /*
1529  * Codes for operating systems
1530  */
1531 #define EXT4_OS_LINUX		0
1532 #define EXT4_OS_HURD		1
1533 #define EXT4_OS_MASIX		2
1534 #define EXT4_OS_FREEBSD		3
1535 #define EXT4_OS_LITES		4
1536 
1537 /*
1538  * Revision levels
1539  */
1540 #define EXT4_GOOD_OLD_REV	0	/* The good old (original) format */
1541 #define EXT4_DYNAMIC_REV	1	/* V2 format w/ dynamic inode sizes */
1542 
1543 #define EXT4_CURRENT_REV	EXT4_GOOD_OLD_REV
1544 #define EXT4_MAX_SUPP_REV	EXT4_DYNAMIC_REV
1545 
1546 #define EXT4_GOOD_OLD_INODE_SIZE 128
1547 
1548 /*
1549  * Feature set definitions
1550  */
1551 
1552 #define EXT4_HAS_COMPAT_FEATURE(sb,mask)			\
1553 	((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1554 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask)			\
1555 	((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1556 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask)			\
1557 	((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1558 #define EXT4_SET_COMPAT_FEATURE(sb,mask)			\
1559 	EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1560 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask)			\
1561 	EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1562 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask)			\
1563 	EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1564 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask)			\
1565 	EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1566 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask)			\
1567 	EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1568 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask)			\
1569 	EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1570 
1571 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC	0x0001
1572 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES	0x0002
1573 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL		0x0004
1574 #define EXT4_FEATURE_COMPAT_EXT_ATTR		0x0008
1575 #define EXT4_FEATURE_COMPAT_RESIZE_INODE	0x0010
1576 #define EXT4_FEATURE_COMPAT_DIR_INDEX		0x0020
1577 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2	0x0200
1578 
1579 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER	0x0001
1580 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE	0x0002
1581 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR	0x0004
1582 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE        0x0008
1583 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM		0x0010
1584 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK	0x0020
1585 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE	0x0040
1586 #define EXT4_FEATURE_RO_COMPAT_QUOTA		0x0100
1587 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC		0x0200
1588 /*
1589  * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM).  When
1590  * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1591  * all other data structures' checksums.  However, the METADATA_CSUM and
1592  * GDT_CSUM bits are mutually exclusive.
1593  */
1594 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM	0x0400
1595 
1596 #define EXT4_FEATURE_INCOMPAT_COMPRESSION	0x0001
1597 #define EXT4_FEATURE_INCOMPAT_FILETYPE		0x0002
1598 #define EXT4_FEATURE_INCOMPAT_RECOVER		0x0004 /* Needs recovery */
1599 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV	0x0008 /* Journal device */
1600 #define EXT4_FEATURE_INCOMPAT_META_BG		0x0010
1601 #define EXT4_FEATURE_INCOMPAT_EXTENTS		0x0040 /* extents support */
1602 #define EXT4_FEATURE_INCOMPAT_64BIT		0x0080
1603 #define EXT4_FEATURE_INCOMPAT_MMP               0x0100
1604 #define EXT4_FEATURE_INCOMPAT_FLEX_BG		0x0200
1605 #define EXT4_FEATURE_INCOMPAT_EA_INODE		0x0400 /* EA in inode */
1606 #define EXT4_FEATURE_INCOMPAT_DIRDATA		0x1000 /* data in dirent */
1607 #define EXT4_FEATURE_INCOMPAT_BG_USE_META_CSUM	0x2000 /* use crc32c for bg */
1608 #define EXT4_FEATURE_INCOMPAT_LARGEDIR		0x4000 /* >2GB or 3-lvl htree */
1609 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA	0x8000 /* data in inode */
1610 #define EXT4_FEATURE_INCOMPAT_ENCRYPT		0x10000
1611 
1612 #define EXT2_FEATURE_COMPAT_SUPP	EXT4_FEATURE_COMPAT_EXT_ATTR
1613 #define EXT2_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
1614 					 EXT4_FEATURE_INCOMPAT_META_BG)
1615 #define EXT2_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1616 					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1617 					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1618 
1619 #define EXT3_FEATURE_COMPAT_SUPP	EXT4_FEATURE_COMPAT_EXT_ATTR
1620 #define EXT3_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
1621 					 EXT4_FEATURE_INCOMPAT_RECOVER| \
1622 					 EXT4_FEATURE_INCOMPAT_META_BG)
1623 #define EXT3_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1624 					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1625 					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1626 
1627 #define EXT4_FEATURE_COMPAT_SUPP	EXT2_FEATURE_COMPAT_EXT_ATTR
1628 #define EXT4_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
1629 					 EXT4_FEATURE_INCOMPAT_RECOVER| \
1630 					 EXT4_FEATURE_INCOMPAT_META_BG| \
1631 					 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1632 					 EXT4_FEATURE_INCOMPAT_64BIT| \
1633 					 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1634 					 EXT4_FEATURE_INCOMPAT_MMP | \
1635 					 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
1636 					 EXT4_FEATURE_INCOMPAT_ENCRYPT)
1637 #define EXT4_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1638 					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1639 					 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1640 					 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1641 					 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1642 					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1643 					 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1644 					 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1645 					 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1646 					 EXT4_FEATURE_RO_COMPAT_QUOTA)
1647 
1648 /*
1649  * Default values for user and/or group using reserved blocks
1650  */
1651 #define	EXT4_DEF_RESUID		0
1652 #define	EXT4_DEF_RESGID		0
1653 
1654 #define EXT4_DEF_INODE_READAHEAD_BLKS	32
1655 
1656 /*
1657  * Default mount options
1658  */
1659 #define EXT4_DEFM_DEBUG		0x0001
1660 #define EXT4_DEFM_BSDGROUPS	0x0002
1661 #define EXT4_DEFM_XATTR_USER	0x0004
1662 #define EXT4_DEFM_ACL		0x0008
1663 #define EXT4_DEFM_UID16		0x0010
1664 #define EXT4_DEFM_JMODE		0x0060
1665 #define EXT4_DEFM_JMODE_DATA	0x0020
1666 #define EXT4_DEFM_JMODE_ORDERED	0x0040
1667 #define EXT4_DEFM_JMODE_WBACK	0x0060
1668 #define EXT4_DEFM_NOBARRIER	0x0100
1669 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1670 #define EXT4_DEFM_DISCARD	0x0400
1671 #define EXT4_DEFM_NODELALLOC	0x0800
1672 
1673 /*
1674  * Default journal batch times
1675  */
1676 #define EXT4_DEF_MIN_BATCH_TIME	0
1677 #define EXT4_DEF_MAX_BATCH_TIME	15000 /* 15ms */
1678 
1679 /*
1680  * Minimum number of groups in a flexgroup before we separate out
1681  * directories into the first block group of a flexgroup
1682  */
1683 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME	4
1684 
1685 /*
1686  * Structure of a directory entry
1687  */
1688 #define EXT4_NAME_LEN 255
1689 
1690 struct ext4_dir_entry {
1691 	__le32	inode;			/* Inode number */
1692 	__le16	rec_len;		/* Directory entry length */
1693 	__le16	name_len;		/* Name length */
1694 	char	name[EXT4_NAME_LEN];	/* File name */
1695 };
1696 
1697 /*
1698  * The new version of the directory entry.  Since EXT4 structures are
1699  * stored in intel byte order, and the name_len field could never be
1700  * bigger than 255 chars, it's safe to reclaim the extra byte for the
1701  * file_type field.
1702  */
1703 struct ext4_dir_entry_2 {
1704 	__le32	inode;			/* Inode number */
1705 	__le16	rec_len;		/* Directory entry length */
1706 	__u8	name_len;		/* Name length */
1707 	__u8	file_type;
1708 	char	name[EXT4_NAME_LEN];	/* File name */
1709 };
1710 
1711 /*
1712  * This is a bogus directory entry at the end of each leaf block that
1713  * records checksums.
1714  */
1715 struct ext4_dir_entry_tail {
1716 	__le32	det_reserved_zero1;	/* Pretend to be unused */
1717 	__le16	det_rec_len;		/* 12 */
1718 	__u8	det_reserved_zero2;	/* Zero name length */
1719 	__u8	det_reserved_ft;	/* 0xDE, fake file type */
1720 	__le32	det_checksum;		/* crc32c(uuid+inum+dirblock) */
1721 };
1722 
1723 #define EXT4_DIRENT_TAIL(block, blocksize) \
1724 	((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1725 					((blocksize) - \
1726 					 sizeof(struct ext4_dir_entry_tail))))
1727 
1728 /*
1729  * Ext4 directory file types.  Only the low 3 bits are used.  The
1730  * other bits are reserved for now.
1731  */
1732 #define EXT4_FT_UNKNOWN		0
1733 #define EXT4_FT_REG_FILE	1
1734 #define EXT4_FT_DIR		2
1735 #define EXT4_FT_CHRDEV		3
1736 #define EXT4_FT_BLKDEV		4
1737 #define EXT4_FT_FIFO		5
1738 #define EXT4_FT_SOCK		6
1739 #define EXT4_FT_SYMLINK		7
1740 
1741 #define EXT4_FT_MAX		8
1742 
1743 #define EXT4_FT_DIR_CSUM	0xDE
1744 
1745 /*
1746  * EXT4_DIR_PAD defines the directory entries boundaries
1747  *
1748  * NOTE: It must be a multiple of 4
1749  */
1750 #define EXT4_DIR_PAD			4
1751 #define EXT4_DIR_ROUND			(EXT4_DIR_PAD - 1)
1752 #define EXT4_DIR_REC_LEN(name_len)	(((name_len) + 8 + EXT4_DIR_ROUND) & \
1753 					 ~EXT4_DIR_ROUND)
1754 #define EXT4_MAX_REC_LEN		((1<<16)-1)
1755 
1756 /*
1757  * If we ever get support for fs block sizes > page_size, we'll need
1758  * to remove the #if statements in the next two functions...
1759  */
1760 static inline unsigned int
ext4_rec_len_from_disk(__le16 dlen,unsigned blocksize)1761 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1762 {
1763 	unsigned len = le16_to_cpu(dlen);
1764 
1765 #if (PAGE_CACHE_SIZE >= 65536)
1766 	if (len == EXT4_MAX_REC_LEN || len == 0)
1767 		return blocksize;
1768 	return (len & 65532) | ((len & 3) << 16);
1769 #else
1770 	return len;
1771 #endif
1772 }
1773 
ext4_rec_len_to_disk(unsigned len,unsigned blocksize)1774 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1775 {
1776 	if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1777 		BUG();
1778 #if (PAGE_CACHE_SIZE >= 65536)
1779 	if (len < 65536)
1780 		return cpu_to_le16(len);
1781 	if (len == blocksize) {
1782 		if (blocksize == 65536)
1783 			return cpu_to_le16(EXT4_MAX_REC_LEN);
1784 		else
1785 			return cpu_to_le16(0);
1786 	}
1787 	return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1788 #else
1789 	return cpu_to_le16(len);
1790 #endif
1791 }
1792 
1793 /*
1794  * Hash Tree Directory indexing
1795  * (c) Daniel Phillips, 2001
1796  */
1797 
1798 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1799 				      EXT4_FEATURE_COMPAT_DIR_INDEX) && \
1800 		    ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
1801 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1802 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1803 
1804 /* Legal values for the dx_root hash_version field: */
1805 
1806 #define DX_HASH_LEGACY		0
1807 #define DX_HASH_HALF_MD4	1
1808 #define DX_HASH_TEA		2
1809 #define DX_HASH_LEGACY_UNSIGNED	3
1810 #define DX_HASH_HALF_MD4_UNSIGNED	4
1811 #define DX_HASH_TEA_UNSIGNED		5
1812 
ext4_chksum(struct ext4_sb_info * sbi,u32 crc,const void * address,unsigned int length)1813 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
1814 			      const void *address, unsigned int length)
1815 {
1816 	struct {
1817 		struct shash_desc shash;
1818 		char ctx[4];
1819 	} desc;
1820 	int err;
1821 
1822 	BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
1823 
1824 	desc.shash.tfm = sbi->s_chksum_driver;
1825 	desc.shash.flags = 0;
1826 	*(u32 *)desc.ctx = crc;
1827 
1828 	err = crypto_shash_update(&desc.shash, address, length);
1829 	BUG_ON(err);
1830 
1831 	return *(u32 *)desc.ctx;
1832 }
1833 
1834 #ifdef __KERNEL__
1835 
1836 /* hash info structure used by the directory hash */
1837 struct dx_hash_info
1838 {
1839 	u32		hash;
1840 	u32		minor_hash;
1841 	int		hash_version;
1842 	u32		*seed;
1843 };
1844 
1845 
1846 /* 32 and 64 bit signed EOF for dx directories */
1847 #define EXT4_HTREE_EOF_32BIT   ((1UL  << (32 - 1)) - 1)
1848 #define EXT4_HTREE_EOF_64BIT   ((1ULL << (64 - 1)) - 1)
1849 
1850 
1851 /*
1852  * Control parameters used by ext4_htree_next_block
1853  */
1854 #define HASH_NB_ALWAYS		1
1855 
1856 struct ext4_filename {
1857 	const struct qstr *usr_fname;
1858 	struct ext4_str disk_name;
1859 	struct dx_hash_info hinfo;
1860 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1861 	struct ext4_str crypto_buf;
1862 #endif
1863 };
1864 
1865 #define fname_name(p) ((p)->disk_name.name)
1866 #define fname_len(p)  ((p)->disk_name.len)
1867 
1868 /*
1869  * Describe an inode's exact location on disk and in memory
1870  */
1871 struct ext4_iloc
1872 {
1873 	struct buffer_head *bh;
1874 	unsigned long offset;
1875 	ext4_group_t block_group;
1876 };
1877 
ext4_raw_inode(struct ext4_iloc * iloc)1878 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
1879 {
1880 	return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
1881 }
1882 
1883 /*
1884  * This structure is stuffed into the struct file's private_data field
1885  * for directories.  It is where we put information so that we can do
1886  * readdir operations in hash tree order.
1887  */
1888 struct dir_private_info {
1889 	struct rb_root	root;
1890 	struct rb_node	*curr_node;
1891 	struct fname	*extra_fname;
1892 	loff_t		last_pos;
1893 	__u32		curr_hash;
1894 	__u32		curr_minor_hash;
1895 	__u32		next_hash;
1896 };
1897 
1898 /* calculate the first block number of the group */
1899 static inline ext4_fsblk_t
ext4_group_first_block_no(struct super_block * sb,ext4_group_t group_no)1900 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
1901 {
1902 	return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
1903 		le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
1904 }
1905 
1906 /*
1907  * Special error return code only used by dx_probe() and its callers.
1908  */
1909 #define ERR_BAD_DX_DIR	(-(MAX_ERRNO - 1))
1910 
1911 /*
1912  * Timeout and state flag for lazy initialization inode thread.
1913  */
1914 #define EXT4_DEF_LI_WAIT_MULT			10
1915 #define EXT4_DEF_LI_MAX_START_DELAY		5
1916 #define EXT4_LAZYINIT_QUIT			0x0001
1917 #define EXT4_LAZYINIT_RUNNING			0x0002
1918 
1919 /*
1920  * Lazy inode table initialization info
1921  */
1922 struct ext4_lazy_init {
1923 	unsigned long		li_state;
1924 	struct list_head	li_request_list;
1925 	struct mutex		li_list_mtx;
1926 };
1927 
1928 struct ext4_li_request {
1929 	struct super_block	*lr_super;
1930 	struct ext4_sb_info	*lr_sbi;
1931 	ext4_group_t		lr_next_group;
1932 	struct list_head	lr_request;
1933 	unsigned long		lr_next_sched;
1934 	unsigned long		lr_timeout;
1935 };
1936 
1937 struct ext4_features {
1938 	struct kobject f_kobj;
1939 	struct completion f_kobj_unregister;
1940 };
1941 
1942 /*
1943  * This structure will be used for multiple mount protection. It will be
1944  * written into the block number saved in the s_mmp_block field in the
1945  * superblock. Programs that check MMP should assume that if
1946  * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
1947  * to use the filesystem, regardless of how old the timestamp is.
1948  */
1949 #define EXT4_MMP_MAGIC     0x004D4D50U /* ASCII for MMP */
1950 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
1951 #define EXT4_MMP_SEQ_FSCK  0xE24D4D50U /* mmp_seq value when being fscked */
1952 #define EXT4_MMP_SEQ_MAX   0xE24D4D4FU /* maximum valid mmp_seq value */
1953 
1954 struct mmp_struct {
1955 	__le32	mmp_magic;		/* Magic number for MMP */
1956 	__le32	mmp_seq;		/* Sequence no. updated periodically */
1957 
1958 	/*
1959 	 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
1960 	 * purposes and do not affect the correctness of the algorithm
1961 	 */
1962 	__le64	mmp_time;		/* Time last updated */
1963 	char	mmp_nodename[64];	/* Node which last updated MMP block */
1964 	char	mmp_bdevname[32];	/* Bdev which last updated MMP block */
1965 
1966 	/*
1967 	 * mmp_check_interval is used to verify if the MMP block has been
1968 	 * updated on the block device. The value is updated based on the
1969 	 * maximum time to write the MMP block during an update cycle.
1970 	 */
1971 	__le16	mmp_check_interval;
1972 
1973 	__le16	mmp_pad1;
1974 	__le32	mmp_pad2[226];
1975 	__le32	mmp_checksum;		/* crc32c(uuid+mmp_block) */
1976 };
1977 
1978 /* arguments passed to the mmp thread */
1979 struct mmpd_data {
1980 	struct buffer_head *bh; /* bh from initial read_mmp_block() */
1981 	struct super_block *sb;  /* super block of the fs */
1982 };
1983 
1984 /*
1985  * Check interval multiplier
1986  * The MMP block is written every update interval and initially checked every
1987  * update interval x the multiplier (the value is then adapted based on the
1988  * write latency). The reason is that writes can be delayed under load and we
1989  * don't want readers to incorrectly assume that the filesystem is no longer
1990  * in use.
1991  */
1992 #define EXT4_MMP_CHECK_MULT		2UL
1993 
1994 /*
1995  * Minimum interval for MMP checking in seconds.
1996  */
1997 #define EXT4_MMP_MIN_CHECK_INTERVAL	5UL
1998 
1999 /*
2000  * Maximum interval for MMP checking in seconds.
2001  */
2002 #define EXT4_MMP_MAX_CHECK_INTERVAL	300UL
2003 
2004 /*
2005  * Function prototypes
2006  */
2007 
2008 /*
2009  * Ok, these declarations are also in <linux/kernel.h> but none of the
2010  * ext4 source programs needs to include it so they are duplicated here.
2011  */
2012 # define NORET_TYPE	/**/
2013 # define ATTRIB_NORET	__attribute__((noreturn))
2014 # define NORET_AND	noreturn,
2015 
2016 /* bitmap.c */
2017 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2018 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2019 				struct ext4_group_desc *gdp,
2020 				struct buffer_head *bh, int sz);
2021 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2022 				  struct ext4_group_desc *gdp,
2023 				  struct buffer_head *bh, int sz);
2024 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2025 				struct ext4_group_desc *gdp,
2026 				struct buffer_head *bh);
2027 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2028 				  struct ext4_group_desc *gdp,
2029 				  struct buffer_head *bh);
2030 
2031 /* balloc.c */
2032 extern void ext4_get_group_no_and_offset(struct super_block *sb,
2033 					 ext4_fsblk_t blocknr,
2034 					 ext4_group_t *blockgrpp,
2035 					 ext4_grpblk_t *offsetp);
2036 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2037 					  ext4_fsblk_t block);
2038 
2039 extern unsigned int ext4_block_group(struct super_block *sb,
2040 			ext4_fsblk_t blocknr);
2041 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2042 			ext4_fsblk_t blocknr);
2043 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2044 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2045 			ext4_group_t group);
2046 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2047 					 ext4_fsblk_t goal,
2048 					 unsigned int flags,
2049 					 unsigned long *count,
2050 					 int *errp);
2051 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2052 				    s64 nclusters, unsigned int flags);
2053 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2054 extern void ext4_check_blocks_bitmap(struct super_block *);
2055 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2056 						    ext4_group_t block_group,
2057 						    struct buffer_head ** bh);
2058 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2059 
2060 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2061 						ext4_group_t block_group);
2062 extern int ext4_wait_block_bitmap(struct super_block *sb,
2063 				  ext4_group_t block_group,
2064 				  struct buffer_head *bh);
2065 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2066 						  ext4_group_t block_group);
2067 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2068 					      ext4_group_t block_group,
2069 					      struct ext4_group_desc *gdp);
2070 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2071 
2072 /* crypto_policy.c */
2073 int ext4_is_child_context_consistent_with_parent(struct inode *parent,
2074 						 struct inode *child);
2075 int ext4_inherit_context(struct inode *parent, struct inode *child);
2076 void ext4_to_hex(char *dst, char *src, size_t src_size);
2077 int ext4_process_policy(const struct ext4_encryption_policy *policy,
2078 			struct inode *inode);
2079 int ext4_get_policy(struct inode *inode,
2080 		    struct ext4_encryption_policy *policy);
2081 
2082 /* crypto.c */
2083 extern struct kmem_cache *ext4_crypt_info_cachep;
2084 bool ext4_valid_contents_enc_mode(uint32_t mode);
2085 uint32_t ext4_validate_encryption_key_size(uint32_t mode, uint32_t size);
2086 extern struct workqueue_struct *ext4_read_workqueue;
2087 struct ext4_crypto_ctx *ext4_get_crypto_ctx(struct inode *inode,
2088 					    gfp_t gfp_flags);
2089 void ext4_release_crypto_ctx(struct ext4_crypto_ctx *ctx);
2090 void ext4_restore_control_page(struct page *data_page);
2091 struct page *ext4_encrypt(struct inode *inode,
2092 			  struct page *plaintext_page,
2093 			  gfp_t gfp_flags);
2094 int ext4_decrypt(struct page *page);
2095 int ext4_encrypted_zeroout(struct inode *inode, struct ext4_extent *ex);
2096 extern const struct dentry_operations ext4_encrypted_d_ops;
2097 
2098 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2099 int ext4_init_crypto(void);
2100 void ext4_exit_crypto(void);
ext4_sb_has_crypto(struct super_block * sb)2101 static inline int ext4_sb_has_crypto(struct super_block *sb)
2102 {
2103 	return EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_ENCRYPT);
2104 }
2105 #else
ext4_init_crypto(void)2106 static inline int ext4_init_crypto(void) { return 0; }
ext4_exit_crypto(void)2107 static inline void ext4_exit_crypto(void) { }
ext4_sb_has_crypto(struct super_block * sb)2108 static inline int ext4_sb_has_crypto(struct super_block *sb)
2109 {
2110 	return 0;
2111 }
2112 #endif
2113 
2114 /* crypto_fname.c */
2115 bool ext4_valid_filenames_enc_mode(uint32_t mode);
2116 u32 ext4_fname_crypto_round_up(u32 size, u32 blksize);
2117 unsigned ext4_fname_encrypted_size(struct inode *inode, u32 ilen);
2118 int ext4_fname_crypto_alloc_buffer(struct inode *inode,
2119 				   u32 ilen, struct ext4_str *crypto_str);
2120 int _ext4_fname_disk_to_usr(struct inode *inode,
2121 			    struct dx_hash_info *hinfo,
2122 			    const struct ext4_str *iname,
2123 			    struct ext4_str *oname);
2124 int ext4_fname_disk_to_usr(struct inode *inode,
2125 			   struct dx_hash_info *hinfo,
2126 			   const struct ext4_dir_entry_2 *de,
2127 			   struct ext4_str *oname);
2128 int ext4_fname_usr_to_disk(struct inode *inode,
2129 			   const struct qstr *iname,
2130 			   struct ext4_str *oname);
2131 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2132 void ext4_fname_crypto_free_buffer(struct ext4_str *crypto_str);
2133 int ext4_fname_setup_filename(struct inode *dir, const struct qstr *iname,
2134 			      int lookup, struct ext4_filename *fname);
2135 void ext4_fname_free_filename(struct ext4_filename *fname);
2136 #else
2137 static inline
ext4_setup_fname_crypto(struct inode * inode)2138 int ext4_setup_fname_crypto(struct inode *inode)
2139 {
2140 	return 0;
2141 }
ext4_fname_crypto_free_buffer(struct ext4_str * p)2142 static inline void ext4_fname_crypto_free_buffer(struct ext4_str *p) { }
ext4_fname_setup_filename(struct inode * dir,const struct qstr * iname,int lookup,struct ext4_filename * fname)2143 static inline int ext4_fname_setup_filename(struct inode *dir,
2144 				     const struct qstr *iname,
2145 				     int lookup, struct ext4_filename *fname)
2146 {
2147 	fname->usr_fname = iname;
2148 	fname->disk_name.name = (unsigned char *) iname->name;
2149 	fname->disk_name.len = iname->len;
2150 	return 0;
2151 }
ext4_fname_free_filename(struct ext4_filename * fname)2152 static inline void ext4_fname_free_filename(struct ext4_filename *fname) { }
2153 #endif
2154 
2155 
2156 /* crypto_key.c */
2157 void ext4_free_crypt_info(struct ext4_crypt_info *ci);
2158 void ext4_free_encryption_info(struct inode *inode, struct ext4_crypt_info *ci);
2159 
2160 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2161 int ext4_has_encryption_key(struct inode *inode);
2162 
2163 int ext4_get_encryption_info(struct inode *inode);
2164 
ext4_encryption_info(struct inode * inode)2165 static inline struct ext4_crypt_info *ext4_encryption_info(struct inode *inode)
2166 {
2167 	return EXT4_I(inode)->i_crypt_info;
2168 }
2169 
2170 #else
ext4_has_encryption_key(struct inode * inode)2171 static inline int ext4_has_encryption_key(struct inode *inode)
2172 {
2173 	return 0;
2174 }
ext4_get_encryption_info(struct inode * inode)2175 static inline int ext4_get_encryption_info(struct inode *inode)
2176 {
2177 	return 0;
2178 }
ext4_encryption_info(struct inode * inode)2179 static inline struct ext4_crypt_info *ext4_encryption_info(struct inode *inode)
2180 {
2181 	return NULL;
2182 }
2183 #endif
2184 
2185 
2186 /* dir.c */
2187 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2188 				  struct file *,
2189 				  struct ext4_dir_entry_2 *,
2190 				  struct buffer_head *, char *, int,
2191 				  unsigned int);
2192 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset)	\
2193 	unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2194 					(de), (bh), (buf), (size), (offset)))
2195 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2196 				__u32 minor_hash,
2197 				struct ext4_dir_entry_2 *dirent,
2198 				struct ext4_str *ent_name);
2199 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2200 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2201 			     struct buffer_head *bh,
2202 			     void *buf, int buf_size,
2203 			     struct ext4_filename *fname,
2204 			     struct ext4_dir_entry_2 **dest_de);
2205 int ext4_insert_dentry(struct inode *dir,
2206 		       struct inode *inode,
2207 		       struct ext4_dir_entry_2 *de,
2208 		       int buf_size,
2209 		       struct ext4_filename *fname);
ext4_update_dx_flag(struct inode * inode)2210 static inline void ext4_update_dx_flag(struct inode *inode)
2211 {
2212 	if (!EXT4_HAS_COMPAT_FEATURE(inode->i_sb,
2213 				     EXT4_FEATURE_COMPAT_DIR_INDEX))
2214 		ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2215 }
2216 static unsigned char ext4_filetype_table[] = {
2217 	DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2218 };
2219 
get_dtype(struct super_block * sb,int filetype)2220 static inline  unsigned char get_dtype(struct super_block *sb, int filetype)
2221 {
2222 	if (!EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE) ||
2223 	    (filetype >= EXT4_FT_MAX))
2224 		return DT_UNKNOWN;
2225 
2226 	return ext4_filetype_table[filetype];
2227 }
2228 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2229 			     void *buf, int buf_size);
2230 
2231 /* fsync.c */
2232 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2233 
2234 /* hash.c */
2235 extern int ext4fs_dirhash(const char *name, int len, struct
2236 			  dx_hash_info *hinfo);
2237 
2238 /* ialloc.c */
2239 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2240 				      const struct qstr *qstr, __u32 goal,
2241 				      uid_t *owner, int handle_type,
2242 				      unsigned int line_no, int nblocks);
2243 
2244 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner) \
2245 	__ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2246 			 0, 0, 0)
2247 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2248 				    type, nblocks)		    \
2249 	__ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2250 			 (type), __LINE__, (nblocks))
2251 
2252 
2253 extern void ext4_free_inode(handle_t *, struct inode *);
2254 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2255 extern unsigned long ext4_count_free_inodes(struct super_block *);
2256 extern unsigned long ext4_count_dirs(struct super_block *);
2257 extern void ext4_check_inodes_bitmap(struct super_block *);
2258 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2259 extern int ext4_init_inode_table(struct super_block *sb,
2260 				 ext4_group_t group, int barrier);
2261 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2262 
2263 /* mballoc.c */
2264 extern long ext4_mb_stats;
2265 extern long ext4_mb_max_to_scan;
2266 extern int ext4_mb_init(struct super_block *);
2267 extern int ext4_mb_release(struct super_block *);
2268 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2269 				struct ext4_allocation_request *, int *);
2270 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2271 extern void ext4_discard_preallocations(struct inode *);
2272 extern int __init ext4_init_mballoc(void);
2273 extern void ext4_exit_mballoc(void);
2274 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2275 			     struct buffer_head *bh, ext4_fsblk_t block,
2276 			     unsigned long count, int flags);
2277 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2278 				   ext4_group_t ngroups);
2279 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2280 		ext4_group_t i, struct ext4_group_desc *desc);
2281 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2282 				ext4_fsblk_t block, unsigned long count);
2283 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *,
2284 				unsigned long blkdev_flags);
2285 
2286 /* inode.c */
2287 int ext4_inode_is_fast_symlink(struct inode *inode);
2288 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2289 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2290 int ext4_get_block_write(struct inode *inode, sector_t iblock,
2291 			 struct buffer_head *bh_result, int create);
2292 int ext4_get_block(struct inode *inode, sector_t iblock,
2293 				struct buffer_head *bh_result, int create);
2294 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2295 			   struct buffer_head *bh, int create);
2296 int ext4_walk_page_buffers(handle_t *handle,
2297 			   struct buffer_head *head,
2298 			   unsigned from,
2299 			   unsigned to,
2300 			   int *partial,
2301 			   int (*fn)(handle_t *handle,
2302 				     struct buffer_head *bh));
2303 int do_journal_get_write_access(handle_t *handle,
2304 				struct buffer_head *bh);
2305 #define FALL_BACK_TO_NONDELALLOC 1
2306 #define CONVERT_INLINE_DATA	 2
2307 
2308 extern struct inode *ext4_iget(struct super_block *, unsigned long);
2309 extern struct inode *ext4_iget_normal(struct super_block *, unsigned long);
2310 extern int  ext4_write_inode(struct inode *, struct writeback_control *);
2311 extern int  ext4_setattr(struct dentry *, struct iattr *);
2312 extern int  ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
2313 				struct kstat *stat);
2314 extern void ext4_evict_inode(struct inode *);
2315 extern void ext4_clear_inode(struct inode *);
2316 extern int  ext4_sync_inode(handle_t *, struct inode *);
2317 extern void ext4_dirty_inode(struct inode *, int);
2318 extern int ext4_change_inode_journal_flag(struct inode *, int);
2319 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2320 extern int ext4_inode_attach_jinode(struct inode *inode);
2321 extern int ext4_can_truncate(struct inode *inode);
2322 extern void ext4_truncate(struct inode *);
2323 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2324 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
2325 extern void ext4_set_inode_flags(struct inode *);
2326 extern void ext4_get_inode_flags(struct ext4_inode_info *);
2327 extern int ext4_alloc_da_blocks(struct inode *inode);
2328 extern void ext4_set_aops(struct inode *inode);
2329 extern int ext4_writepage_trans_blocks(struct inode *);
2330 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2331 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2332 			     loff_t lstart, loff_t lend);
2333 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
2334 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2335 extern void ext4_da_update_reserve_space(struct inode *inode,
2336 					int used, int quota_claim);
2337 
2338 /* indirect.c */
2339 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2340 				struct ext4_map_blocks *map, int flags);
2341 extern ssize_t ext4_ind_direct_IO(int rw, struct kiocb *iocb,
2342 				struct iov_iter *iter, loff_t offset);
2343 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2344 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2345 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2346 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2347 				 ext4_lblk_t start, ext4_lblk_t end);
2348 
2349 /* ioctl.c */
2350 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2351 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2352 
2353 /* migrate.c */
2354 extern int ext4_ext_migrate(struct inode *);
2355 extern int ext4_ind_migrate(struct inode *inode);
2356 
2357 /* namei.c */
2358 extern int ext4_dirent_csum_verify(struct inode *inode,
2359 				   struct ext4_dir_entry *dirent);
2360 extern int ext4_orphan_add(handle_t *, struct inode *);
2361 extern int ext4_orphan_del(handle_t *, struct inode *);
2362 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2363 				__u32 start_minor_hash, __u32 *next_hash);
2364 extern int ext4_search_dir(struct buffer_head *bh,
2365 			   char *search_buf,
2366 			   int buf_size,
2367 			   struct inode *dir,
2368 			   struct ext4_filename *fname,
2369 			   const struct qstr *d_name,
2370 			   unsigned int offset,
2371 			   struct ext4_dir_entry_2 **res_dir);
2372 extern int ext4_generic_delete_entry(handle_t *handle,
2373 				     struct inode *dir,
2374 				     struct ext4_dir_entry_2 *de_del,
2375 				     struct buffer_head *bh,
2376 				     void *entry_buf,
2377 				     int buf_size,
2378 				     int csum_size);
2379 extern int ext4_empty_dir(struct inode *inode);
2380 
2381 /* resize.c */
2382 extern int ext4_group_add(struct super_block *sb,
2383 				struct ext4_new_group_data *input);
2384 extern int ext4_group_extend(struct super_block *sb,
2385 				struct ext4_super_block *es,
2386 				ext4_fsblk_t n_blocks_count);
2387 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2388 
2389 /* super.c */
2390 extern int ext4_calculate_overhead(struct super_block *sb);
2391 extern void ext4_superblock_csum_set(struct super_block *sb);
2392 extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2393 extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2394 extern void ext4_kvfree(void *ptr);
2395 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2396 				    ext4_group_t ngroup);
2397 extern const char *ext4_decode_error(struct super_block *sb, int errno,
2398 				     char nbuf[16]);
2399 
2400 extern __printf(4, 5)
2401 void __ext4_error(struct super_block *, const char *, unsigned int,
2402 		  const char *, ...);
2403 extern __printf(5, 6)
2404 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2405 		      const char *, ...);
2406 extern __printf(5, 6)
2407 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2408 		     const char *, ...);
2409 extern void __ext4_std_error(struct super_block *, const char *,
2410 			     unsigned int, int);
2411 extern __printf(4, 5)
2412 void __ext4_abort(struct super_block *, const char *, unsigned int,
2413 		  const char *, ...);
2414 extern __printf(4, 5)
2415 void __ext4_warning(struct super_block *, const char *, unsigned int,
2416 		    const char *, ...);
2417 extern __printf(3, 4)
2418 void __ext4_msg(struct super_block *, const char *, const char *, ...);
2419 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2420 			   const char *, unsigned int, const char *);
2421 extern __printf(7, 8)
2422 void __ext4_grp_locked_error(const char *, unsigned int,
2423 			     struct super_block *, ext4_group_t,
2424 			     unsigned long, ext4_fsblk_t,
2425 			     const char *, ...);
2426 
2427 #ifdef CONFIG_PRINTK
2428 
2429 #define ext4_error_inode(inode, func, line, block, fmt, ...)		\
2430 	__ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__)
2431 #define ext4_error_file(file, func, line, block, fmt, ...)		\
2432 	__ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2433 #define ext4_error(sb, fmt, ...)					\
2434 	__ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2435 #define ext4_abort(sb, fmt, ...)					\
2436 	__ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2437 #define ext4_warning(sb, fmt, ...)					\
2438 	__ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2439 #define ext4_msg(sb, level, fmt, ...)				\
2440 	__ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2441 #define dump_mmp_msg(sb, mmp, msg)					\
2442 	__dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2443 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...)		\
2444 	__ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2445 				fmt, ##__VA_ARGS__)
2446 
2447 #else
2448 
2449 #define ext4_error_inode(inode, func, line, block, fmt, ...)		\
2450 do {									\
2451 	no_printk(fmt, ##__VA_ARGS__);					\
2452 	__ext4_error_inode(inode, "", 0, block, " ");			\
2453 } while (0)
2454 #define ext4_error_file(file, func, line, block, fmt, ...)		\
2455 do {									\
2456 	no_printk(fmt, ##__VA_ARGS__);					\
2457 	__ext4_error_file(file, "", 0, block, " ");			\
2458 } while (0)
2459 #define ext4_error(sb, fmt, ...)					\
2460 do {									\
2461 	no_printk(fmt, ##__VA_ARGS__);					\
2462 	__ext4_error(sb, "", 0, " ");					\
2463 } while (0)
2464 #define ext4_abort(sb, fmt, ...)					\
2465 do {									\
2466 	no_printk(fmt, ##__VA_ARGS__);					\
2467 	__ext4_abort(sb, "", 0, " ");					\
2468 } while (0)
2469 #define ext4_warning(sb, fmt, ...)					\
2470 do {									\
2471 	no_printk(fmt, ##__VA_ARGS__);					\
2472 	__ext4_warning(sb, "", 0, " ");					\
2473 } while (0)
2474 #define ext4_msg(sb, level, fmt, ...)					\
2475 do {									\
2476 	no_printk(fmt, ##__VA_ARGS__);					\
2477 	__ext4_msg(sb, "", " ");					\
2478 } while (0)
2479 #define dump_mmp_msg(sb, mmp, msg)					\
2480 	__dump_mmp_msg(sb, mmp, "", 0, "")
2481 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...)		\
2482 do {									\
2483 	no_printk(fmt, ##__VA_ARGS__);				\
2484 	__ext4_grp_locked_error("", 0, sb, grp, ino, block, " ");	\
2485 } while (0)
2486 
2487 #endif
2488 
2489 extern void ext4_update_dynamic_rev(struct super_block *sb);
2490 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2491 					__u32 compat);
2492 extern int ext4_update_rocompat_feature(handle_t *handle,
2493 					struct super_block *sb,	__u32 rocompat);
2494 extern int ext4_update_incompat_feature(handle_t *handle,
2495 					struct super_block *sb,	__u32 incompat);
2496 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2497 				      struct ext4_group_desc *bg);
2498 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2499 				      struct ext4_group_desc *bg);
2500 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2501 				     struct ext4_group_desc *bg);
2502 extern __u32 ext4_free_group_clusters(struct super_block *sb,
2503 				      struct ext4_group_desc *bg);
2504 extern __u32 ext4_free_inodes_count(struct super_block *sb,
2505 				 struct ext4_group_desc *bg);
2506 extern __u32 ext4_used_dirs_count(struct super_block *sb,
2507 				struct ext4_group_desc *bg);
2508 extern __u32 ext4_itable_unused_count(struct super_block *sb,
2509 				   struct ext4_group_desc *bg);
2510 extern void ext4_block_bitmap_set(struct super_block *sb,
2511 				  struct ext4_group_desc *bg, ext4_fsblk_t blk);
2512 extern void ext4_inode_bitmap_set(struct super_block *sb,
2513 				  struct ext4_group_desc *bg, ext4_fsblk_t blk);
2514 extern void ext4_inode_table_set(struct super_block *sb,
2515 				 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2516 extern void ext4_free_group_clusters_set(struct super_block *sb,
2517 					 struct ext4_group_desc *bg,
2518 					 __u32 count);
2519 extern void ext4_free_inodes_set(struct super_block *sb,
2520 				struct ext4_group_desc *bg, __u32 count);
2521 extern void ext4_used_dirs_set(struct super_block *sb,
2522 				struct ext4_group_desc *bg, __u32 count);
2523 extern void ext4_itable_unused_set(struct super_block *sb,
2524 				   struct ext4_group_desc *bg, __u32 count);
2525 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2526 				       struct ext4_group_desc *gdp);
2527 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2528 				     struct ext4_group_desc *gdp);
2529 extern int ext4_register_li_request(struct super_block *sb,
2530 				    ext4_group_t first_not_zeroed);
2531 
ext4_has_group_desc_csum(struct super_block * sb)2532 static inline int ext4_has_group_desc_csum(struct super_block *sb)
2533 {
2534 	return EXT4_HAS_RO_COMPAT_FEATURE(sb,
2535 					  EXT4_FEATURE_RO_COMPAT_GDT_CSUM) ||
2536 	       (EXT4_SB(sb)->s_chksum_driver != NULL);
2537 }
2538 
ext4_has_metadata_csum(struct super_block * sb)2539 static inline int ext4_has_metadata_csum(struct super_block *sb)
2540 {
2541 	WARN_ON_ONCE(EXT4_HAS_RO_COMPAT_FEATURE(sb,
2542 			EXT4_FEATURE_RO_COMPAT_METADATA_CSUM) &&
2543 		     !EXT4_SB(sb)->s_chksum_driver);
2544 
2545 	return (EXT4_SB(sb)->s_chksum_driver != NULL);
2546 }
ext4_blocks_count(struct ext4_super_block * es)2547 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2548 {
2549 	return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2550 		le32_to_cpu(es->s_blocks_count_lo);
2551 }
2552 
ext4_r_blocks_count(struct ext4_super_block * es)2553 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2554 {
2555 	return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2556 		le32_to_cpu(es->s_r_blocks_count_lo);
2557 }
2558 
ext4_free_blocks_count(struct ext4_super_block * es)2559 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2560 {
2561 	return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2562 		le32_to_cpu(es->s_free_blocks_count_lo);
2563 }
2564 
ext4_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)2565 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2566 					 ext4_fsblk_t blk)
2567 {
2568 	es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2569 	es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2570 }
2571 
ext4_free_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)2572 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2573 					      ext4_fsblk_t blk)
2574 {
2575 	es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2576 	es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2577 }
2578 
ext4_r_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)2579 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2580 					   ext4_fsblk_t blk)
2581 {
2582 	es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2583 	es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2584 }
2585 
ext4_isize(struct ext4_inode * raw_inode)2586 static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
2587 {
2588 	if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2589 		return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2590 			le32_to_cpu(raw_inode->i_size_lo);
2591 	else
2592 		return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2593 }
2594 
ext4_isize_set(struct ext4_inode * raw_inode,loff_t i_size)2595 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2596 {
2597 	raw_inode->i_size_lo = cpu_to_le32(i_size);
2598 	raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2599 }
2600 
2601 static inline
ext4_get_group_info(struct super_block * sb,ext4_group_t group)2602 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2603 					    ext4_group_t group)
2604 {
2605 	 struct ext4_group_info ***grp_info;
2606 	 long indexv, indexh;
2607 	 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
2608 	 grp_info = EXT4_SB(sb)->s_group_info;
2609 	 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2610 	 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2611 	 return grp_info[indexv][indexh];
2612 }
2613 
2614 /*
2615  * Reading s_groups_count requires using smp_rmb() afterwards.  See
2616  * the locking protocol documented in the comments of ext4_group_add()
2617  * in resize.c
2618  */
ext4_get_groups_count(struct super_block * sb)2619 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2620 {
2621 	ext4_group_t	ngroups = EXT4_SB(sb)->s_groups_count;
2622 
2623 	smp_rmb();
2624 	return ngroups;
2625 }
2626 
ext4_flex_group(struct ext4_sb_info * sbi,ext4_group_t block_group)2627 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2628 					     ext4_group_t block_group)
2629 {
2630 	return block_group >> sbi->s_log_groups_per_flex;
2631 }
2632 
ext4_flex_bg_size(struct ext4_sb_info * sbi)2633 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2634 {
2635 	return 1 << sbi->s_log_groups_per_flex;
2636 }
2637 
2638 #define ext4_std_error(sb, errno)				\
2639 do {								\
2640 	if ((errno))						\
2641 		__ext4_std_error((sb), __func__, __LINE__, (errno));	\
2642 } while (0)
2643 
2644 #ifdef CONFIG_SMP
2645 /* Each CPU can accumulate percpu_counter_batch clusters in their local
2646  * counters. So we need to make sure we have free clusters more
2647  * than percpu_counter_batch  * nr_cpu_ids. Also add a window of 4 times.
2648  */
2649 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2650 #else
2651 #define EXT4_FREECLUSTERS_WATERMARK 0
2652 #endif
2653 
2654 /* Update i_disksize. Requires i_mutex to avoid races with truncate */
ext4_update_i_disksize(struct inode * inode,loff_t newsize)2655 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2656 {
2657 	WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
2658 		     !mutex_is_locked(&inode->i_mutex));
2659 	down_write(&EXT4_I(inode)->i_data_sem);
2660 	if (newsize > EXT4_I(inode)->i_disksize)
2661 		EXT4_I(inode)->i_disksize = newsize;
2662 	up_write(&EXT4_I(inode)->i_data_sem);
2663 }
2664 
2665 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
ext4_update_inode_size(struct inode * inode,loff_t newsize)2666 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
2667 {
2668 	int changed = 0;
2669 
2670 	if (newsize > inode->i_size) {
2671 		i_size_write(inode, newsize);
2672 		changed = 1;
2673 	}
2674 	if (newsize > EXT4_I(inode)->i_disksize) {
2675 		ext4_update_i_disksize(inode, newsize);
2676 		changed |= 2;
2677 	}
2678 	return changed;
2679 }
2680 
2681 struct ext4_group_info {
2682 	unsigned long   bb_state;
2683 	struct rb_root  bb_free_root;
2684 	ext4_grpblk_t	bb_first_free;	/* first free block */
2685 	ext4_grpblk_t	bb_free;	/* total free blocks */
2686 	ext4_grpblk_t	bb_fragments;	/* nr of freespace fragments */
2687 	ext4_grpblk_t	bb_largest_free_order;/* order of largest frag in BG */
2688 	struct          list_head bb_prealloc_list;
2689 #ifdef DOUBLE_CHECK
2690 	void            *bb_bitmap;
2691 #endif
2692 	struct rw_semaphore alloc_sem;
2693 	ext4_grpblk_t	bb_counters[];	/* Nr of free power-of-two-block
2694 					 * regions, index is order.
2695 					 * bb_counters[3] = 5 means
2696 					 * 5 free 8-block regions. */
2697 };
2698 
2699 #define EXT4_GROUP_INFO_NEED_INIT_BIT		0
2700 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT		1
2701 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT	2
2702 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT	3
2703 
2704 #define EXT4_MB_GRP_NEED_INIT(grp)	\
2705 	(test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2706 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp)	\
2707 	(test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2708 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp)	\
2709 	(test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2710 
2711 #define EXT4_MB_GRP_WAS_TRIMMED(grp)	\
2712 	(test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2713 #define EXT4_MB_GRP_SET_TRIMMED(grp)	\
2714 	(set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2715 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp)	\
2716 	(clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2717 
2718 #define EXT4_MAX_CONTENTION		8
2719 #define EXT4_CONTENTION_THRESHOLD	2
2720 
ext4_group_lock_ptr(struct super_block * sb,ext4_group_t group)2721 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2722 					      ext4_group_t group)
2723 {
2724 	return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2725 }
2726 
2727 /*
2728  * Returns true if the filesystem is busy enough that attempts to
2729  * access the block group locks has run into contention.
2730  */
ext4_fs_is_busy(struct ext4_sb_info * sbi)2731 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2732 {
2733 	return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2734 }
2735 
ext4_lock_group(struct super_block * sb,ext4_group_t group)2736 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2737 {
2738 	spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2739 	if (spin_trylock(lock))
2740 		/*
2741 		 * We're able to grab the lock right away, so drop the
2742 		 * lock contention counter.
2743 		 */
2744 		atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2745 	else {
2746 		/*
2747 		 * The lock is busy, so bump the contention counter,
2748 		 * and then wait on the spin lock.
2749 		 */
2750 		atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2751 				  EXT4_MAX_CONTENTION);
2752 		spin_lock(lock);
2753 	}
2754 }
2755 
ext4_unlock_group(struct super_block * sb,ext4_group_t group)2756 static inline void ext4_unlock_group(struct super_block *sb,
2757 					ext4_group_t group)
2758 {
2759 	spin_unlock(ext4_group_lock_ptr(sb, group));
2760 }
2761 
2762 /*
2763  * Block validity checking
2764  */
2765 #define ext4_check_indirect_blockref(inode, bh)				\
2766 	ext4_check_blockref(__func__, __LINE__, inode,			\
2767 			    (__le32 *)(bh)->b_data,			\
2768 			    EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2769 
2770 #define ext4_ind_check_inode(inode)					\
2771 	ext4_check_blockref(__func__, __LINE__, inode,			\
2772 			    EXT4_I(inode)->i_data,			\
2773 			    EXT4_NDIR_BLOCKS)
2774 
2775 /*
2776  * Inodes and files operations
2777  */
2778 
2779 /* dir.c */
2780 extern const struct file_operations ext4_dir_operations;
2781 
2782 /* file.c */
2783 extern const struct inode_operations ext4_file_inode_operations;
2784 extern const struct file_operations ext4_file_operations;
2785 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
2786 
2787 /* inline.c */
2788 extern int ext4_get_max_inline_size(struct inode *inode);
2789 extern int ext4_find_inline_data_nolock(struct inode *inode);
2790 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
2791 				 unsigned int len);
2792 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
2793 
2794 extern int ext4_readpage_inline(struct inode *inode, struct page *page);
2795 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
2796 					 struct inode *inode,
2797 					 loff_t pos, unsigned len,
2798 					 unsigned flags,
2799 					 struct page **pagep);
2800 extern int ext4_write_inline_data_end(struct inode *inode,
2801 				      loff_t pos, unsigned len,
2802 				      unsigned copied,
2803 				      struct page *page);
2804 extern struct buffer_head *
2805 ext4_journalled_write_inline_data(struct inode *inode,
2806 				  unsigned len,
2807 				  struct page *page);
2808 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
2809 					   struct inode *inode,
2810 					   loff_t pos, unsigned len,
2811 					   unsigned flags,
2812 					   struct page **pagep,
2813 					   void **fsdata);
2814 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
2815 					 unsigned len, unsigned copied,
2816 					 struct page *page);
2817 extern int ext4_try_add_inline_entry(handle_t *handle,
2818 				     struct ext4_filename *fname,
2819 				     struct dentry *dentry,
2820 				     struct inode *inode);
2821 extern int ext4_try_create_inline_dir(handle_t *handle,
2822 				      struct inode *parent,
2823 				      struct inode *inode);
2824 extern int ext4_read_inline_dir(struct file *filp,
2825 				struct dir_context *ctx,
2826 				int *has_inline_data);
2827 extern int htree_inlinedir_to_tree(struct file *dir_file,
2828 				   struct inode *dir, ext4_lblk_t block,
2829 				   struct dx_hash_info *hinfo,
2830 				   __u32 start_hash, __u32 start_minor_hash,
2831 				   int *has_inline_data);
2832 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
2833 					struct ext4_filename *fname,
2834 					const struct qstr *d_name,
2835 					struct ext4_dir_entry_2 **res_dir,
2836 					int *has_inline_data);
2837 extern int ext4_delete_inline_entry(handle_t *handle,
2838 				    struct inode *dir,
2839 				    struct ext4_dir_entry_2 *de_del,
2840 				    struct buffer_head *bh,
2841 				    int *has_inline_data);
2842 extern int empty_inline_dir(struct inode *dir, int *has_inline_data);
2843 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
2844 					struct ext4_dir_entry_2 **parent_de,
2845 					int *retval);
2846 extern int ext4_inline_data_fiemap(struct inode *inode,
2847 				   struct fiemap_extent_info *fieinfo,
2848 				   int *has_inline);
2849 extern int ext4_try_to_evict_inline_data(handle_t *handle,
2850 					 struct inode *inode,
2851 					 int needed);
2852 extern void ext4_inline_data_truncate(struct inode *inode, int *has_inline);
2853 
2854 extern int ext4_convert_inline_data(struct inode *inode);
2855 
ext4_has_inline_data(struct inode * inode)2856 static inline int ext4_has_inline_data(struct inode *inode)
2857 {
2858 	return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
2859 	       EXT4_I(inode)->i_inline_off;
2860 }
2861 
2862 /* namei.c */
2863 extern const struct inode_operations ext4_dir_inode_operations;
2864 extern const struct inode_operations ext4_special_inode_operations;
2865 extern struct dentry *ext4_get_parent(struct dentry *child);
2866 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
2867 				 struct ext4_dir_entry_2 *de,
2868 				 int blocksize, int csum_size,
2869 				 unsigned int parent_ino, int dotdot_real_len);
2870 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t,
2871 				   unsigned int blocksize);
2872 extern int ext4_handle_dirty_dirent_node(handle_t *handle,
2873 					 struct inode *inode,
2874 					 struct buffer_head *bh);
2875 #define S_SHIFT 12
2876 static unsigned char ext4_type_by_mode[S_IFMT >> S_SHIFT] = {
2877 	[S_IFREG >> S_SHIFT]	= EXT4_FT_REG_FILE,
2878 	[S_IFDIR >> S_SHIFT]	= EXT4_FT_DIR,
2879 	[S_IFCHR >> S_SHIFT]	= EXT4_FT_CHRDEV,
2880 	[S_IFBLK >> S_SHIFT]	= EXT4_FT_BLKDEV,
2881 	[S_IFIFO >> S_SHIFT]	= EXT4_FT_FIFO,
2882 	[S_IFSOCK >> S_SHIFT]	= EXT4_FT_SOCK,
2883 	[S_IFLNK >> S_SHIFT]	= EXT4_FT_SYMLINK,
2884 };
2885 
ext4_set_de_type(struct super_block * sb,struct ext4_dir_entry_2 * de,umode_t mode)2886 static inline void ext4_set_de_type(struct super_block *sb,
2887 				struct ext4_dir_entry_2 *de,
2888 				umode_t mode) {
2889 	if (EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE))
2890 		de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
2891 }
2892 
2893 /* readpages.c */
2894 extern int ext4_mpage_readpages(struct address_space *mapping,
2895 				struct list_head *pages, struct page *page,
2896 				unsigned nr_pages);
2897 
2898 /* symlink.c */
2899 extern const struct inode_operations ext4_symlink_inode_operations;
2900 extern const struct inode_operations ext4_fast_symlink_inode_operations;
2901 
2902 /* block_validity */
2903 extern void ext4_release_system_zone(struct super_block *sb);
2904 extern int ext4_setup_system_zone(struct super_block *sb);
2905 extern int __init ext4_init_system_zone(void);
2906 extern void ext4_exit_system_zone(void);
2907 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
2908 				 ext4_fsblk_t start_blk,
2909 				 unsigned int count);
2910 extern int ext4_check_blockref(const char *, unsigned int,
2911 			       struct inode *, __le32 *, unsigned int);
2912 
2913 /* extents.c */
2914 struct ext4_ext_path;
2915 struct ext4_extent;
2916 
2917 /*
2918  * Maximum number of logical blocks in a file; ext4_extent's ee_block is
2919  * __le32.
2920  */
2921 #define EXT_MAX_BLOCKS	0xffffffff
2922 
2923 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
2924 extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
2925 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
2926 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
2927 			       struct ext4_map_blocks *map, int flags);
2928 extern void ext4_ext_truncate(handle_t *, struct inode *);
2929 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
2930 				 ext4_lblk_t end);
2931 extern void ext4_ext_init(struct super_block *);
2932 extern void ext4_ext_release(struct super_block *);
2933 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
2934 			  loff_t len);
2935 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
2936 					  loff_t offset, ssize_t len);
2937 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
2938 			   struct ext4_map_blocks *map, int flags);
2939 extern int ext4_ext_calc_metadata_amount(struct inode *inode,
2940 					 ext4_lblk_t lblocks);
2941 extern int ext4_extent_tree_init(handle_t *, struct inode *);
2942 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
2943 						   int num,
2944 						   struct ext4_ext_path *path);
2945 extern int ext4_can_extents_be_merged(struct inode *inode,
2946 				      struct ext4_extent *ex1,
2947 				      struct ext4_extent *ex2);
2948 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
2949 				  struct ext4_ext_path **,
2950 				  struct ext4_extent *, int);
2951 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
2952 					      struct ext4_ext_path **,
2953 					      int flags);
2954 extern void ext4_ext_drop_refs(struct ext4_ext_path *);
2955 extern int ext4_ext_check_inode(struct inode *inode);
2956 extern int ext4_find_delalloc_range(struct inode *inode,
2957 				    ext4_lblk_t lblk_start,
2958 				    ext4_lblk_t lblk_end);
2959 extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk);
2960 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
2961 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
2962 			__u64 start, __u64 len);
2963 extern int ext4_ext_precache(struct inode *inode);
2964 extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len);
2965 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
2966 				struct inode *inode2, ext4_lblk_t lblk1,
2967 			     ext4_lblk_t lblk2,  ext4_lblk_t count,
2968 			     int mark_unwritten,int *err);
2969 
2970 /* move_extent.c */
2971 extern void ext4_double_down_write_data_sem(struct inode *first,
2972 					    struct inode *second);
2973 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
2974 					  struct inode *donor_inode);
2975 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
2976 			     __u64 start_orig, __u64 start_donor,
2977 			     __u64 len, __u64 *moved_len);
2978 
2979 /* page-io.c */
2980 extern int __init ext4_init_pageio(void);
2981 extern void ext4_exit_pageio(void);
2982 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
2983 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
2984 extern int ext4_put_io_end(ext4_io_end_t *io_end);
2985 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
2986 extern void ext4_io_submit_init(struct ext4_io_submit *io,
2987 				struct writeback_control *wbc);
2988 extern void ext4_end_io_rsv_work(struct work_struct *work);
2989 extern void ext4_io_submit(struct ext4_io_submit *io);
2990 extern int ext4_bio_write_page(struct ext4_io_submit *io,
2991 			       struct page *page,
2992 			       int len,
2993 			       struct writeback_control *wbc,
2994 			       bool keep_towrite);
2995 
2996 /* mmp.c */
2997 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
2998 
2999 /*
3000  * Note that these flags will never ever appear in a buffer_head's state flag.
3001  * See EXT4_MAP_... to see where this is used.
3002  */
3003 enum ext4_state_bits {
3004 	BH_AllocFromCluster	/* allocated blocks were part of already
3005 				 * allocated cluster. */
3006 	= BH_JBDPrivateStart
3007 };
3008 
3009 /*
3010  * Add new method to test whether block and inode bitmaps are properly
3011  * initialized. With uninit_bg reading the block from disk is not enough
3012  * to mark the bitmap uptodate. We need to also zero-out the bitmap
3013  */
3014 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3015 
bitmap_uptodate(struct buffer_head * bh)3016 static inline int bitmap_uptodate(struct buffer_head *bh)
3017 {
3018 	return (buffer_uptodate(bh) &&
3019 			test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3020 }
set_bitmap_uptodate(struct buffer_head * bh)3021 static inline void set_bitmap_uptodate(struct buffer_head *bh)
3022 {
3023 	set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3024 }
3025 
3026 /*
3027  * Disable DIO read nolock optimization, so new dioreaders will be forced
3028  * to grab i_mutex
3029  */
ext4_inode_block_unlocked_dio(struct inode * inode)3030 static inline void ext4_inode_block_unlocked_dio(struct inode *inode)
3031 {
3032 	ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
3033 	smp_mb();
3034 }
ext4_inode_resume_unlocked_dio(struct inode * inode)3035 static inline void ext4_inode_resume_unlocked_dio(struct inode *inode)
3036 {
3037 	smp_mb();
3038 	ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
3039 }
3040 
3041 #define in_range(b, first, len)	((b) >= (first) && (b) <= (first) + (len) - 1)
3042 
3043 /* For ioend & aio unwritten conversion wait queues */
3044 #define EXT4_WQ_HASH_SZ		37
3045 #define ext4_ioend_wq(v)   (&ext4__ioend_wq[((unsigned long)(v)) %\
3046 					    EXT4_WQ_HASH_SZ])
3047 #define ext4_aio_mutex(v)  (&ext4__aio_mutex[((unsigned long)(v)) %\
3048 					     EXT4_WQ_HASH_SZ])
3049 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3050 extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ];
3051 
3052 #define EXT4_RESIZING	0
3053 extern int ext4_resize_begin(struct super_block *sb);
3054 extern void ext4_resize_end(struct super_block *sb);
3055 
3056 #endif	/* __KERNEL__ */
3057 
3058 #endif	/* _EXT4_H */
3059