• 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 #ifdef __KERNEL__
237 # define EXT4_BLOCK_SIZE(s)		((s)->s_blocksize)
238 #else
239 # define EXT4_BLOCK_SIZE(s)		(EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
240 #endif
241 #define	EXT4_ADDR_PER_BLOCK(s)		(EXT4_BLOCK_SIZE(s) / sizeof(__u32))
242 #define EXT4_CLUSTER_SIZE(s)		(EXT4_BLOCK_SIZE(s) << \
243 					 EXT4_SB(s)->s_cluster_bits)
244 #ifdef __KERNEL__
245 # define EXT4_BLOCK_SIZE_BITS(s)	((s)->s_blocksize_bits)
246 # define EXT4_CLUSTER_BITS(s)		(EXT4_SB(s)->s_cluster_bits)
247 #else
248 # define EXT4_BLOCK_SIZE_BITS(s)	((s)->s_log_block_size + 10)
249 #endif
250 #ifdef __KERNEL__
251 #define	EXT4_ADDR_PER_BLOCK_BITS(s)	(EXT4_SB(s)->s_addr_per_block_bits)
252 #define EXT4_INODE_SIZE(s)		(EXT4_SB(s)->s_inode_size)
253 #define EXT4_FIRST_INO(s)		(EXT4_SB(s)->s_first_ino)
254 #else
255 #define EXT4_INODE_SIZE(s)	(((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
256 				 EXT4_GOOD_OLD_INODE_SIZE : \
257 				 (s)->s_inode_size)
258 #define EXT4_FIRST_INO(s)	(((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
259 				 EXT4_GOOD_OLD_FIRST_INO : \
260 				 (s)->s_first_ino)
261 #endif
262 #define EXT4_BLOCK_ALIGN(size, blkbits)		ALIGN((size), (1 << (blkbits)))
263 
264 /* Translate a block number to a cluster number */
265 #define EXT4_B2C(sbi, blk)	((blk) >> (sbi)->s_cluster_bits)
266 /* Translate a cluster number to a block number */
267 #define EXT4_C2B(sbi, cluster)	((cluster) << (sbi)->s_cluster_bits)
268 /* Translate # of blks to # of clusters */
269 #define EXT4_NUM_B2C(sbi, blks)	(((blks) + (sbi)->s_cluster_ratio - 1) >> \
270 				 (sbi)->s_cluster_bits)
271 /* Mask out the low bits to get the starting block of the cluster */
272 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) &				\
273 				  ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
274 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) &				\
275 				  ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
276 /* Get the cluster offset */
277 #define EXT4_PBLK_COFF(s, pblk) ((pblk) &				\
278 				 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
279 #define EXT4_LBLK_COFF(s, lblk) ((lblk) &				\
280 				 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
281 
282 /*
283  * Structure of a blocks group descriptor
284  */
285 struct ext4_group_desc
286 {
287 	__le32	bg_block_bitmap_lo;	/* Blocks bitmap block */
288 	__le32	bg_inode_bitmap_lo;	/* Inodes bitmap block */
289 	__le32	bg_inode_table_lo;	/* Inodes table block */
290 	__le16	bg_free_blocks_count_lo;/* Free blocks count */
291 	__le16	bg_free_inodes_count_lo;/* Free inodes count */
292 	__le16	bg_used_dirs_count_lo;	/* Directories count */
293 	__le16	bg_flags;		/* EXT4_BG_flags (INODE_UNINIT, etc) */
294 	__le32  bg_exclude_bitmap_lo;   /* Exclude bitmap for snapshots */
295 	__le16  bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
296 	__le16  bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
297 	__le16  bg_itable_unused_lo;	/* Unused inodes count */
298 	__le16  bg_checksum;		/* crc16(sb_uuid+group+desc) */
299 	__le32	bg_block_bitmap_hi;	/* Blocks bitmap block MSB */
300 	__le32	bg_inode_bitmap_hi;	/* Inodes bitmap block MSB */
301 	__le32	bg_inode_table_hi;	/* Inodes table block MSB */
302 	__le16	bg_free_blocks_count_hi;/* Free blocks count MSB */
303 	__le16	bg_free_inodes_count_hi;/* Free inodes count MSB */
304 	__le16	bg_used_dirs_count_hi;	/* Directories count MSB */
305 	__le16  bg_itable_unused_hi;    /* Unused inodes count MSB */
306 	__le32  bg_exclude_bitmap_hi;   /* Exclude bitmap block MSB */
307 	__le16  bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
308 	__le16  bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
309 	__u32   bg_reserved;
310 };
311 
312 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END	\
313 	(offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
314 	 sizeof(__le16))
315 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END	\
316 	(offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
317 	 sizeof(__le16))
318 
319 /*
320  * Structure of a flex block group info
321  */
322 
323 struct flex_groups {
324 	atomic64_t	free_clusters;
325 	atomic_t	free_inodes;
326 	atomic_t	used_dirs;
327 };
328 
329 #define EXT4_BG_INODE_UNINIT	0x0001 /* Inode table/bitmap not in use */
330 #define EXT4_BG_BLOCK_UNINIT	0x0002 /* Block bitmap not in use */
331 #define EXT4_BG_INODE_ZEROED	0x0004 /* On-disk itable initialized to zero */
332 
333 /*
334  * Macro-instructions used to manage group descriptors
335  */
336 #define EXT4_MIN_DESC_SIZE		32
337 #define EXT4_MIN_DESC_SIZE_64BIT	64
338 #define	EXT4_MAX_DESC_SIZE		EXT4_MIN_BLOCK_SIZE
339 #define EXT4_DESC_SIZE(s)		(EXT4_SB(s)->s_desc_size)
340 #ifdef __KERNEL__
341 # define EXT4_BLOCKS_PER_GROUP(s)	(EXT4_SB(s)->s_blocks_per_group)
342 # define EXT4_CLUSTERS_PER_GROUP(s)	(EXT4_SB(s)->s_clusters_per_group)
343 # define EXT4_DESC_PER_BLOCK(s)		(EXT4_SB(s)->s_desc_per_block)
344 # define EXT4_INODES_PER_GROUP(s)	(EXT4_SB(s)->s_inodes_per_group)
345 # define EXT4_DESC_PER_BLOCK_BITS(s)	(EXT4_SB(s)->s_desc_per_block_bits)
346 #else
347 # define EXT4_BLOCKS_PER_GROUP(s)	((s)->s_blocks_per_group)
348 # define EXT4_DESC_PER_BLOCK(s)		(EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
349 # define EXT4_INODES_PER_GROUP(s)	((s)->s_inodes_per_group)
350 #endif
351 
352 /*
353  * Constants relative to the data blocks
354  */
355 #define	EXT4_NDIR_BLOCKS		12
356 #define	EXT4_IND_BLOCK			EXT4_NDIR_BLOCKS
357 #define	EXT4_DIND_BLOCK			(EXT4_IND_BLOCK + 1)
358 #define	EXT4_TIND_BLOCK			(EXT4_DIND_BLOCK + 1)
359 #define	EXT4_N_BLOCKS			(EXT4_TIND_BLOCK + 1)
360 
361 /*
362  * Inode flags
363  */
364 #define	EXT4_SECRM_FL			0x00000001 /* Secure deletion */
365 #define	EXT4_UNRM_FL			0x00000002 /* Undelete */
366 #define	EXT4_COMPR_FL			0x00000004 /* Compress file */
367 #define EXT4_SYNC_FL			0x00000008 /* Synchronous updates */
368 #define EXT4_IMMUTABLE_FL		0x00000010 /* Immutable file */
369 #define EXT4_APPEND_FL			0x00000020 /* writes to file may only append */
370 #define EXT4_NODUMP_FL			0x00000040 /* do not dump file */
371 #define EXT4_NOATIME_FL			0x00000080 /* do not update atime */
372 /* Reserved for compression usage... */
373 #define EXT4_DIRTY_FL			0x00000100
374 #define EXT4_COMPRBLK_FL		0x00000200 /* One or more compressed clusters */
375 #define EXT4_NOCOMPR_FL			0x00000400 /* Don't compress */
376 	/* nb: was previously EXT2_ECOMPR_FL */
377 #define EXT4_ENCRYPT_FL			0x00000800 /* encrypted file */
378 /* End compression flags --- maybe not all used */
379 #define EXT4_INDEX_FL			0x00001000 /* hash-indexed directory */
380 #define EXT4_IMAGIC_FL			0x00002000 /* AFS directory */
381 #define EXT4_JOURNAL_DATA_FL		0x00004000 /* file data should be journaled */
382 #define EXT4_NOTAIL_FL			0x00008000 /* file tail should not be merged */
383 #define EXT4_DIRSYNC_FL			0x00010000 /* dirsync behaviour (directories only) */
384 #define EXT4_TOPDIR_FL			0x00020000 /* Top of directory hierarchies*/
385 #define EXT4_HUGE_FILE_FL               0x00040000 /* Set to each huge file */
386 #define EXT4_EXTENTS_FL			0x00080000 /* Inode uses extents */
387 #define EXT4_EA_INODE_FL	        0x00200000 /* Inode used for large EA */
388 #define EXT4_EOFBLOCKS_FL		0x00400000 /* Blocks allocated beyond EOF */
389 #define EXT4_INLINE_DATA_FL		0x10000000 /* Inode has inline data. */
390 #define EXT4_RESERVED_FL		0x80000000 /* reserved for ext4 lib */
391 
392 #define EXT4_FL_USER_VISIBLE		0x004BDFFF /* User visible flags */
393 #define EXT4_FL_USER_MODIFIABLE		0x004380FF /* User modifiable flags */
394 
395 /* Flags that should be inherited by new inodes from their parent. */
396 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
397 			   EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
398 			   EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
399 			   EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
400 
401 /* Flags that are appropriate for regular files (all but dir-specific ones). */
402 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
403 
404 /* Flags that are appropriate for non-directories/regular files. */
405 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
406 
407 /* Mask out flags that are inappropriate for the given type of inode. */
ext4_mask_flags(umode_t mode,__u32 flags)408 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
409 {
410 	if (S_ISDIR(mode))
411 		return flags;
412 	else if (S_ISREG(mode))
413 		return flags & EXT4_REG_FLMASK;
414 	else
415 		return flags & EXT4_OTHER_FLMASK;
416 }
417 
418 /*
419  * Inode flags used for atomic set/get
420  */
421 enum {
422 	EXT4_INODE_SECRM	= 0,	/* Secure deletion */
423 	EXT4_INODE_UNRM		= 1,	/* Undelete */
424 	EXT4_INODE_COMPR	= 2,	/* Compress file */
425 	EXT4_INODE_SYNC		= 3,	/* Synchronous updates */
426 	EXT4_INODE_IMMUTABLE	= 4,	/* Immutable file */
427 	EXT4_INODE_APPEND	= 5,	/* writes to file may only append */
428 	EXT4_INODE_NODUMP	= 6,	/* do not dump file */
429 	EXT4_INODE_NOATIME	= 7,	/* do not update atime */
430 /* Reserved for compression usage... */
431 	EXT4_INODE_DIRTY	= 8,
432 	EXT4_INODE_COMPRBLK	= 9,	/* One or more compressed clusters */
433 	EXT4_INODE_NOCOMPR	= 10,	/* Don't compress */
434 	EXT4_INODE_ENCRYPT	= 11,	/* Encrypted file */
435 /* End compression flags --- maybe not all used */
436 	EXT4_INODE_INDEX	= 12,	/* hash-indexed directory */
437 	EXT4_INODE_IMAGIC	= 13,	/* AFS directory */
438 	EXT4_INODE_JOURNAL_DATA	= 14,	/* file data should be journaled */
439 	EXT4_INODE_NOTAIL	= 15,	/* file tail should not be merged */
440 	EXT4_INODE_DIRSYNC	= 16,	/* dirsync behaviour (directories only) */
441 	EXT4_INODE_TOPDIR	= 17,	/* Top of directory hierarchies*/
442 	EXT4_INODE_HUGE_FILE	= 18,	/* Set to each huge file */
443 	EXT4_INODE_EXTENTS	= 19,	/* Inode uses extents */
444 	EXT4_INODE_EA_INODE	= 21,	/* Inode used for large EA */
445 	EXT4_INODE_EOFBLOCKS	= 22,	/* Blocks allocated beyond EOF */
446 	EXT4_INODE_INLINE_DATA	= 28,	/* Data in inode. */
447 	EXT4_INODE_RESERVED	= 31,	/* reserved for ext4 lib */
448 };
449 
450 /*
451  * Since it's pretty easy to mix up bit numbers and hex values, we use a
452  * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
453  * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
454  * any extra space in the compiled kernel image, otherwise, the build will fail.
455  * It's important that these values are the same, since we are using
456  * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
457  * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
458  * values found in ext2, ext3 and ext4 filesystems, and of course the values
459  * defined in e2fsprogs.
460  *
461  * It's not paranoia if the Murphy's Law really *is* out to get you.  :-)
462  */
463 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
464 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
465 
ext4_check_flag_values(void)466 static inline void ext4_check_flag_values(void)
467 {
468 	CHECK_FLAG_VALUE(SECRM);
469 	CHECK_FLAG_VALUE(UNRM);
470 	CHECK_FLAG_VALUE(COMPR);
471 	CHECK_FLAG_VALUE(SYNC);
472 	CHECK_FLAG_VALUE(IMMUTABLE);
473 	CHECK_FLAG_VALUE(APPEND);
474 	CHECK_FLAG_VALUE(NODUMP);
475 	CHECK_FLAG_VALUE(NOATIME);
476 	CHECK_FLAG_VALUE(DIRTY);
477 	CHECK_FLAG_VALUE(COMPRBLK);
478 	CHECK_FLAG_VALUE(NOCOMPR);
479 	CHECK_FLAG_VALUE(ENCRYPT);
480 	CHECK_FLAG_VALUE(INDEX);
481 	CHECK_FLAG_VALUE(IMAGIC);
482 	CHECK_FLAG_VALUE(JOURNAL_DATA);
483 	CHECK_FLAG_VALUE(NOTAIL);
484 	CHECK_FLAG_VALUE(DIRSYNC);
485 	CHECK_FLAG_VALUE(TOPDIR);
486 	CHECK_FLAG_VALUE(HUGE_FILE);
487 	CHECK_FLAG_VALUE(EXTENTS);
488 	CHECK_FLAG_VALUE(EA_INODE);
489 	CHECK_FLAG_VALUE(EOFBLOCKS);
490 	CHECK_FLAG_VALUE(INLINE_DATA);
491 	CHECK_FLAG_VALUE(RESERVED);
492 }
493 
494 /* Used to pass group descriptor data when online resize is done */
495 struct ext4_new_group_input {
496 	__u32 group;		/* Group number for this data */
497 	__u64 block_bitmap;	/* Absolute block number of block bitmap */
498 	__u64 inode_bitmap;	/* Absolute block number of inode bitmap */
499 	__u64 inode_table;	/* Absolute block number of inode table start */
500 	__u32 blocks_count;	/* Total number of blocks in this group */
501 	__u16 reserved_blocks;	/* Number of reserved blocks in this group */
502 	__u16 unused;
503 };
504 
505 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
506 struct compat_ext4_new_group_input {
507 	u32 group;
508 	compat_u64 block_bitmap;
509 	compat_u64 inode_bitmap;
510 	compat_u64 inode_table;
511 	u32 blocks_count;
512 	u16 reserved_blocks;
513 	u16 unused;
514 };
515 #endif
516 
517 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
518 struct ext4_new_group_data {
519 	__u32 group;
520 	__u64 block_bitmap;
521 	__u64 inode_bitmap;
522 	__u64 inode_table;
523 	__u32 blocks_count;
524 	__u16 reserved_blocks;
525 	__u16 unused;
526 	__u32 free_blocks_count;
527 };
528 
529 /* Indexes used to index group tables in ext4_new_group_data */
530 enum {
531 	BLOCK_BITMAP = 0,	/* block bitmap */
532 	INODE_BITMAP,		/* inode bitmap */
533 	INODE_TABLE,		/* inode tables */
534 	GROUP_TABLE_COUNT,
535 };
536 
537 /*
538  * Flags used by ext4_map_blocks()
539  */
540 	/* Allocate any needed blocks and/or convert an unwritten
541 	   extent to be an initialized ext4 */
542 #define EXT4_GET_BLOCKS_CREATE			0x0001
543 	/* Request the creation of an unwritten extent */
544 #define EXT4_GET_BLOCKS_UNWRIT_EXT		0x0002
545 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT	(EXT4_GET_BLOCKS_UNWRIT_EXT|\
546 						 EXT4_GET_BLOCKS_CREATE)
547 	/* Caller is from the delayed allocation writeout path
548 	 * finally doing the actual allocation of delayed blocks */
549 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE	0x0004
550 	/* caller is from the direct IO path, request to creation of an
551 	unwritten extents if not allocated, split the unwritten
552 	extent if blocks has been preallocated already*/
553 #define EXT4_GET_BLOCKS_PRE_IO			0x0008
554 #define EXT4_GET_BLOCKS_CONVERT			0x0010
555 #define EXT4_GET_BLOCKS_IO_CREATE_EXT		(EXT4_GET_BLOCKS_PRE_IO|\
556 					 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
557 	/* Convert extent to initialized after IO complete */
558 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT		(EXT4_GET_BLOCKS_CONVERT|\
559 					 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
560 	/* Eventual metadata allocation (due to growing extent tree)
561 	 * should not fail, so try to use reserved blocks for that.*/
562 #define EXT4_GET_BLOCKS_METADATA_NOFAIL		0x0020
563 	/* Don't normalize allocation size (used for fallocate) */
564 #define EXT4_GET_BLOCKS_NO_NORMALIZE		0x0040
565 	/* Request will not result in inode size update (user for fallocate) */
566 #define EXT4_GET_BLOCKS_KEEP_SIZE		0x0080
567 	/* Do not take i_data_sem locking in ext4_map_blocks */
568 #define EXT4_GET_BLOCKS_NO_LOCK			0x0100
569 	/* Do not put hole in extent cache */
570 #define EXT4_GET_BLOCKS_NO_PUT_HOLE		0x0200
571 	/* Convert written extents to unwritten */
572 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN	0x0400
573 
574 /*
575  * The bit position of these flags must not overlap with any of the
576  * EXT4_GET_BLOCKS_*.  They are used by ext4_find_extent(),
577  * read_extent_tree_block(), ext4_split_extent_at(),
578  * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
579  * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
580  * caching the extents when reading from the extent tree while a
581  * truncate or punch hole operation is in progress.
582  */
583 #define EXT4_EX_NOCACHE				0x40000000
584 #define EXT4_EX_FORCE_CACHE			0x20000000
585 
586 /*
587  * Flags used by ext4_free_blocks
588  */
589 #define EXT4_FREE_BLOCKS_METADATA	0x0001
590 #define EXT4_FREE_BLOCKS_FORGET		0x0002
591 #define EXT4_FREE_BLOCKS_VALIDATED	0x0004
592 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE	0x0008
593 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER	0x0010
594 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER	0x0020
595 
596 /* Encryption algorithms */
597 #define EXT4_ENCRYPTION_MODE_INVALID		0
598 #define EXT4_ENCRYPTION_MODE_AES_256_XTS	1
599 #define EXT4_ENCRYPTION_MODE_AES_256_GCM	2
600 #define EXT4_ENCRYPTION_MODE_AES_256_CBC	3
601 #define EXT4_ENCRYPTION_MODE_AES_256_CTS	4
602 
603 #include "ext4_crypto.h"
604 
605 /*
606  * ioctl commands
607  */
608 #define	EXT4_IOC_GETFLAGS		FS_IOC_GETFLAGS
609 #define	EXT4_IOC_SETFLAGS		FS_IOC_SETFLAGS
610 #define	EXT4_IOC_GETVERSION		_IOR('f', 3, long)
611 #define	EXT4_IOC_SETVERSION		_IOW('f', 4, long)
612 #define	EXT4_IOC_GETVERSION_OLD		FS_IOC_GETVERSION
613 #define	EXT4_IOC_SETVERSION_OLD		FS_IOC_SETVERSION
614 #define EXT4_IOC_GETRSVSZ		_IOR('f', 5, long)
615 #define EXT4_IOC_SETRSVSZ		_IOW('f', 6, long)
616 #define EXT4_IOC_GROUP_EXTEND		_IOW('f', 7, unsigned long)
617 #define EXT4_IOC_GROUP_ADD		_IOW('f', 8, struct ext4_new_group_input)
618 #define EXT4_IOC_MIGRATE		_IO('f', 9)
619  /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
620  /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
621 #define EXT4_IOC_ALLOC_DA_BLKS		_IO('f', 12)
622 #define EXT4_IOC_MOVE_EXT		_IOWR('f', 15, struct move_extent)
623 #define EXT4_IOC_RESIZE_FS		_IOW('f', 16, __u64)
624 #define EXT4_IOC_SWAP_BOOT		_IO('f', 17)
625 #define EXT4_IOC_PRECACHE_EXTENTS	_IO('f', 18)
626 #define EXT4_IOC_SET_ENCRYPTION_POLICY	_IOR('f', 19, struct ext4_encryption_policy)
627 #define EXT4_IOC_GET_ENCRYPTION_PWSALT	_IOW('f', 20, __u8[16])
628 #define EXT4_IOC_GET_ENCRYPTION_POLICY	_IOW('f', 21, struct ext4_encryption_policy)
629 
630 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
631 /*
632  * ioctl commands in 32 bit emulation
633  */
634 #define EXT4_IOC32_GETFLAGS		FS_IOC32_GETFLAGS
635 #define EXT4_IOC32_SETFLAGS		FS_IOC32_SETFLAGS
636 #define EXT4_IOC32_GETVERSION		_IOR('f', 3, int)
637 #define EXT4_IOC32_SETVERSION		_IOW('f', 4, int)
638 #define EXT4_IOC32_GETRSVSZ		_IOR('f', 5, int)
639 #define EXT4_IOC32_SETRSVSZ		_IOW('f', 6, int)
640 #define EXT4_IOC32_GROUP_EXTEND		_IOW('f', 7, unsigned int)
641 #define EXT4_IOC32_GROUP_ADD		_IOW('f', 8, struct compat_ext4_new_group_input)
642 #define EXT4_IOC32_GETVERSION_OLD	FS_IOC32_GETVERSION
643 #define EXT4_IOC32_SETVERSION_OLD	FS_IOC32_SETVERSION
644 #endif
645 
646 /* Max physical block we can address w/o extents */
647 #define EXT4_MAX_BLOCK_FILE_PHYS	0xFFFFFFFF
648 
649 /*
650  * Structure of an inode on the disk
651  */
652 struct ext4_inode {
653 	__le16	i_mode;		/* File mode */
654 	__le16	i_uid;		/* Low 16 bits of Owner Uid */
655 	__le32	i_size_lo;	/* Size in bytes */
656 	__le32	i_atime;	/* Access time */
657 	__le32	i_ctime;	/* Inode Change time */
658 	__le32	i_mtime;	/* Modification time */
659 	__le32	i_dtime;	/* Deletion Time */
660 	__le16	i_gid;		/* Low 16 bits of Group Id */
661 	__le16	i_links_count;	/* Links count */
662 	__le32	i_blocks_lo;	/* Blocks count */
663 	__le32	i_flags;	/* File flags */
664 	union {
665 		struct {
666 			__le32  l_i_version;
667 		} linux1;
668 		struct {
669 			__u32  h_i_translator;
670 		} hurd1;
671 		struct {
672 			__u32  m_i_reserved1;
673 		} masix1;
674 	} osd1;				/* OS dependent 1 */
675 	__le32	i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
676 	__le32	i_generation;	/* File version (for NFS) */
677 	__le32	i_file_acl_lo;	/* File ACL */
678 	__le32	i_size_high;
679 	__le32	i_obso_faddr;	/* Obsoleted fragment address */
680 	union {
681 		struct {
682 			__le16	l_i_blocks_high; /* were l_i_reserved1 */
683 			__le16	l_i_file_acl_high;
684 			__le16	l_i_uid_high;	/* these 2 fields */
685 			__le16	l_i_gid_high;	/* were reserved2[0] */
686 			__le16	l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
687 			__le16	l_i_reserved;
688 		} linux2;
689 		struct {
690 			__le16	h_i_reserved1;	/* Obsoleted fragment number/size which are removed in ext4 */
691 			__u16	h_i_mode_high;
692 			__u16	h_i_uid_high;
693 			__u16	h_i_gid_high;
694 			__u32	h_i_author;
695 		} hurd2;
696 		struct {
697 			__le16	h_i_reserved1;	/* Obsoleted fragment number/size which are removed in ext4 */
698 			__le16	m_i_file_acl_high;
699 			__u32	m_i_reserved2[2];
700 		} masix2;
701 	} osd2;				/* OS dependent 2 */
702 	__le16	i_extra_isize;
703 	__le16	i_checksum_hi;	/* crc32c(uuid+inum+inode) BE */
704 	__le32  i_ctime_extra;  /* extra Change time      (nsec << 2 | epoch) */
705 	__le32  i_mtime_extra;  /* extra Modification time(nsec << 2 | epoch) */
706 	__le32  i_atime_extra;  /* extra Access time      (nsec << 2 | epoch) */
707 	__le32  i_crtime;       /* File Creation time */
708 	__le32  i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
709 	__le32  i_version_hi;	/* high 32 bits for 64-bit version */
710 };
711 
712 struct move_extent {
713 	__u32 reserved;		/* should be zero */
714 	__u32 donor_fd;		/* donor file descriptor */
715 	__u64 orig_start;	/* logical start offset in block for orig */
716 	__u64 donor_start;	/* logical start offset in block for donor */
717 	__u64 len;		/* block length to be moved */
718 	__u64 moved_len;	/* moved block length */
719 };
720 
721 #define EXT4_EPOCH_BITS 2
722 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
723 #define EXT4_NSEC_MASK  (~0UL << EXT4_EPOCH_BITS)
724 
725 /*
726  * Extended fields will fit into an inode if the filesystem was formatted
727  * with large inodes (-I 256 or larger) and there are not currently any EAs
728  * consuming all of the available space. For new inodes we always reserve
729  * enough space for the kernel's known extended fields, but for inodes
730  * created with an old kernel this might not have been the case. None of
731  * the extended inode fields is critical for correct filesystem operation.
732  * This macro checks if a certain field fits in the inode. Note that
733  * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
734  */
735 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field)	\
736 	((offsetof(typeof(*ext4_inode), field) +	\
737 	  sizeof((ext4_inode)->field))			\
738 	<= (EXT4_GOOD_OLD_INODE_SIZE +			\
739 	    (einode)->i_extra_isize))			\
740 
ext4_encode_extra_time(struct timespec * time)741 static inline __le32 ext4_encode_extra_time(struct timespec *time)
742 {
743        return cpu_to_le32((sizeof(time->tv_sec) > 4 ?
744 			   (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) |
745                           ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK));
746 }
747 
ext4_decode_extra_time(struct timespec * time,__le32 extra)748 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
749 {
750        if (sizeof(time->tv_sec) > 4)
751 	       time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK)
752 			       << 32;
753        time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
754 }
755 
756 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode)			       \
757 do {									       \
758 	(raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec);	       \
759 	if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     \
760 		(raw_inode)->xtime ## _extra =				       \
761 				ext4_encode_extra_time(&(inode)->xtime);       \
762 } while (0)
763 
764 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode)			       \
765 do {									       \
766 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))		       \
767 		(raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec);      \
768 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))	       \
769 		(raw_inode)->xtime ## _extra =				       \
770 				ext4_encode_extra_time(&(einode)->xtime);      \
771 } while (0)
772 
773 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode)			       \
774 do {									       \
775 	(inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime);       \
776 	if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     \
777 		ext4_decode_extra_time(&(inode)->xtime,			       \
778 				       raw_inode->xtime ## _extra);	       \
779 	else								       \
780 		(inode)->xtime.tv_nsec = 0;				       \
781 } while (0)
782 
783 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode)			       \
784 do {									       \
785 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))		       \
786 		(einode)->xtime.tv_sec = 				       \
787 			(signed)le32_to_cpu((raw_inode)->xtime);	       \
788 	else								       \
789 		(einode)->xtime.tv_sec = 0;				       \
790 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))	       \
791 		ext4_decode_extra_time(&(einode)->xtime,		       \
792 				       raw_inode->xtime ## _extra);	       \
793 	else								       \
794 		(einode)->xtime.tv_nsec = 0;				       \
795 } while (0)
796 
797 #define i_disk_version osd1.linux1.l_i_version
798 
799 #if defined(__KERNEL__) || defined(__linux__)
800 #define i_reserved1	osd1.linux1.l_i_reserved1
801 #define i_file_acl_high	osd2.linux2.l_i_file_acl_high
802 #define i_blocks_high	osd2.linux2.l_i_blocks_high
803 #define i_uid_low	i_uid
804 #define i_gid_low	i_gid
805 #define i_uid_high	osd2.linux2.l_i_uid_high
806 #define i_gid_high	osd2.linux2.l_i_gid_high
807 #define i_checksum_lo	osd2.linux2.l_i_checksum_lo
808 
809 #elif defined(__GNU__)
810 
811 #define i_translator	osd1.hurd1.h_i_translator
812 #define i_uid_high	osd2.hurd2.h_i_uid_high
813 #define i_gid_high	osd2.hurd2.h_i_gid_high
814 #define i_author	osd2.hurd2.h_i_author
815 
816 #elif defined(__masix__)
817 
818 #define i_reserved1	osd1.masix1.m_i_reserved1
819 #define i_file_acl_high	osd2.masix2.m_i_file_acl_high
820 #define i_reserved2	osd2.masix2.m_i_reserved2
821 
822 #endif /* defined(__KERNEL__) || defined(__linux__) */
823 
824 #include "extents_status.h"
825 
826 /*
827  * fourth extended file system inode data in memory
828  */
829 struct ext4_inode_info {
830 	__le32	i_data[15];	/* unconverted */
831 	__u32	i_dtime;
832 	ext4_fsblk_t	i_file_acl;
833 
834 	/*
835 	 * i_block_group is the number of the block group which contains
836 	 * this file's inode.  Constant across the lifetime of the inode,
837 	 * it is ued for making block allocation decisions - we try to
838 	 * place a file's data blocks near its inode block, and new inodes
839 	 * near to their parent directory's inode.
840 	 */
841 	ext4_group_t	i_block_group;
842 	ext4_lblk_t	i_dir_start_lookup;
843 #if (BITS_PER_LONG < 64)
844 	unsigned long	i_state_flags;		/* Dynamic state flags */
845 #endif
846 	unsigned long	i_flags;
847 
848 	/*
849 	 * Extended attributes can be read independently of the main file
850 	 * data. Taking i_mutex even when reading would cause contention
851 	 * between readers of EAs and writers of regular file data, so
852 	 * instead we synchronize on xattr_sem when reading or changing
853 	 * EAs.
854 	 */
855 	struct rw_semaphore xattr_sem;
856 
857 	struct list_head i_orphan;	/* unlinked but open inodes */
858 
859 	/*
860 	 * i_disksize keeps track of what the inode size is ON DISK, not
861 	 * in memory.  During truncate, i_size is set to the new size by
862 	 * the VFS prior to calling ext4_truncate(), but the filesystem won't
863 	 * set i_disksize to 0 until the truncate is actually under way.
864 	 *
865 	 * The intent is that i_disksize always represents the blocks which
866 	 * are used by this file.  This allows recovery to restart truncate
867 	 * on orphans if we crash during truncate.  We actually write i_disksize
868 	 * into the on-disk inode when writing inodes out, instead of i_size.
869 	 *
870 	 * The only time when i_disksize and i_size may be different is when
871 	 * a truncate is in progress.  The only things which change i_disksize
872 	 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
873 	 */
874 	loff_t	i_disksize;
875 
876 	/*
877 	 * i_data_sem is for serialising ext4_truncate() against
878 	 * ext4_getblock().  In the 2.4 ext2 design, great chunks of inode's
879 	 * data tree are chopped off during truncate. We can't do that in
880 	 * ext4 because whenever we perform intermediate commits during
881 	 * truncate, the inode and all the metadata blocks *must* be in a
882 	 * consistent state which allows truncation of the orphans to restart
883 	 * during recovery.  Hence we must fix the get_block-vs-truncate race
884 	 * by other means, so we have i_data_sem.
885 	 */
886 	struct rw_semaphore i_data_sem;
887 	struct inode vfs_inode;
888 	struct jbd2_inode *jinode;
889 
890 	spinlock_t i_raw_lock;	/* protects updates to the raw inode */
891 
892 	/*
893 	 * File creation time. Its function is same as that of
894 	 * struct timespec i_{a,c,m}time in the generic inode.
895 	 */
896 	struct timespec i_crtime;
897 
898 	/* mballoc */
899 	struct list_head i_prealloc_list;
900 	spinlock_t i_prealloc_lock;
901 
902 	/* extents status tree */
903 	struct ext4_es_tree i_es_tree;
904 	rwlock_t i_es_lock;
905 	struct list_head i_es_lru;
906 	unsigned int i_es_all_nr;	/* protected by i_es_lock */
907 	unsigned int i_es_lru_nr;	/* protected by i_es_lock */
908 	unsigned long i_touch_when;	/* jiffies of last accessing */
909 
910 	/* ialloc */
911 	ext4_group_t	i_last_alloc_group;
912 
913 	/* allocation reservation info for delalloc */
914 	/* In case of bigalloc, these refer to clusters rather than blocks */
915 	unsigned int i_reserved_data_blocks;
916 	unsigned int i_reserved_meta_blocks;
917 	unsigned int i_allocated_meta_blocks;
918 	ext4_lblk_t i_da_metadata_calc_last_lblock;
919 	int i_da_metadata_calc_len;
920 
921 	/* on-disk additional length */
922 	__u16 i_extra_isize;
923 
924 	/* Indicate the inline data space. */
925 	u16 i_inline_off;
926 	u16 i_inline_size;
927 
928 #ifdef CONFIG_QUOTA
929 	/* quota space reservation, managed internally by quota code */
930 	qsize_t i_reserved_quota;
931 #endif
932 
933 	/* Lock protecting lists below */
934 	spinlock_t i_completed_io_lock;
935 	/*
936 	 * Completed IOs that need unwritten extents handling and have
937 	 * transaction reserved
938 	 */
939 	struct list_head i_rsv_conversion_list;
940 	/*
941 	 * Completed IOs that need unwritten extents handling and don't have
942 	 * transaction reserved
943 	 */
944 	atomic_t i_ioend_count;	/* Number of outstanding io_end structs */
945 	atomic_t i_unwritten; /* Nr. of inflight conversions pending */
946 	struct work_struct i_rsv_conversion_work;
947 
948 	spinlock_t i_block_reservation_lock;
949 
950 	/*
951 	 * Transactions that contain inode's metadata needed to complete
952 	 * fsync and fdatasync, respectively.
953 	 */
954 	tid_t i_sync_tid;
955 	tid_t i_datasync_tid;
956 
957 	/* Precomputed uuid+inum+igen checksum for seeding inode checksums */
958 	__u32 i_csum_seed;
959 
960 #ifdef CONFIG_EXT4_FS_ENCRYPTION
961 	/* Encryption params */
962 	struct ext4_crypt_info *i_crypt_info;
963 #endif
964 };
965 
966 /*
967  * File system states
968  */
969 #define	EXT4_VALID_FS			0x0001	/* Unmounted cleanly */
970 #define	EXT4_ERROR_FS			0x0002	/* Errors detected */
971 #define	EXT4_ORPHAN_FS			0x0004	/* Orphans being recovered */
972 
973 /*
974  * Misc. filesystem flags
975  */
976 #define EXT2_FLAGS_SIGNED_HASH		0x0001  /* Signed dirhash in use */
977 #define EXT2_FLAGS_UNSIGNED_HASH	0x0002  /* Unsigned dirhash in use */
978 #define EXT2_FLAGS_TEST_FILESYS		0x0004	/* to test development code */
979 
980 /*
981  * Mount flags set via mount options or defaults
982  */
983 #define EXT4_MOUNT_GRPID		0x00004	/* Create files with directory's group */
984 #define EXT4_MOUNT_DEBUG		0x00008	/* Some debugging messages */
985 #define EXT4_MOUNT_ERRORS_CONT		0x00010	/* Continue on errors */
986 #define EXT4_MOUNT_ERRORS_RO		0x00020	/* Remount fs ro on errors */
987 #define EXT4_MOUNT_ERRORS_PANIC		0x00040	/* Panic on errors */
988 #define EXT4_MOUNT_ERRORS_MASK		0x00070
989 #define EXT4_MOUNT_MINIX_DF		0x00080	/* Mimics the Minix statfs */
990 #define EXT4_MOUNT_NOLOAD		0x00100	/* Don't use existing journal*/
991 #define EXT4_MOUNT_DATA_FLAGS		0x00C00	/* Mode for data writes: */
992 #define EXT4_MOUNT_JOURNAL_DATA		0x00400	/* Write data to journal */
993 #define EXT4_MOUNT_ORDERED_DATA		0x00800	/* Flush data before commit */
994 #define EXT4_MOUNT_WRITEBACK_DATA	0x00C00	/* No data ordering */
995 #define EXT4_MOUNT_UPDATE_JOURNAL	0x01000	/* Update the journal format */
996 #define EXT4_MOUNT_NO_UID32		0x02000  /* Disable 32-bit UIDs */
997 #define EXT4_MOUNT_XATTR_USER		0x04000	/* Extended user attributes */
998 #define EXT4_MOUNT_POSIX_ACL		0x08000	/* POSIX Access Control Lists */
999 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC	0x10000	/* No auto delalloc mapping */
1000 #define EXT4_MOUNT_BARRIER		0x20000 /* Use block barriers */
1001 #define EXT4_MOUNT_QUOTA		0x80000 /* Some quota option set */
1002 #define EXT4_MOUNT_USRQUOTA		0x100000 /* "old" user quota */
1003 #define EXT4_MOUNT_GRPQUOTA		0x200000 /* "old" group quota */
1004 #define EXT4_MOUNT_DIOREAD_NOLOCK	0x400000 /* Enable support for dio read nolocking */
1005 #define EXT4_MOUNT_JOURNAL_CHECKSUM	0x800000 /* Journal checksums */
1006 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT	0x1000000 /* Journal Async Commit */
1007 #define EXT4_MOUNT_DELALLOC		0x8000000 /* Delalloc support */
1008 #define EXT4_MOUNT_DATA_ERR_ABORT	0x10000000 /* Abort on file data write */
1009 #define EXT4_MOUNT_BLOCK_VALIDITY	0x20000000 /* Block validity checking */
1010 #define EXT4_MOUNT_DISCARD		0x40000000 /* Issue DISCARD requests */
1011 #define EXT4_MOUNT_INIT_INODE_TABLE	0x80000000 /* Initialize uninitialized itables */
1012 
1013 /*
1014  * Mount flags set either automatically (could not be set by mount option)
1015  * based on per file system feature or property or in special cases such as
1016  * distinguishing between explicit mount option definition and default.
1017  */
1018 #define EXT4_MOUNT2_EXPLICIT_DELALLOC	0x00000001 /* User explicitly
1019 						      specified delalloc */
1020 #define EXT4_MOUNT2_STD_GROUP_SIZE	0x00000002 /* We have standard group
1021 						      size of blocksize * 8
1022 						      blocks */
1023 #define EXT4_MOUNT2_HURD_COMPAT		0x00000004 /* Support HURD-castrated
1024 						      file systems */
1025 
1026 #define clear_opt(sb, opt)		EXT4_SB(sb)->s_mount_opt &= \
1027 						~EXT4_MOUNT_##opt
1028 #define set_opt(sb, opt)		EXT4_SB(sb)->s_mount_opt |= \
1029 						EXT4_MOUNT_##opt
1030 #define test_opt(sb, opt)		(EXT4_SB(sb)->s_mount_opt & \
1031 					 EXT4_MOUNT_##opt)
1032 
1033 #define clear_opt2(sb, opt)		EXT4_SB(sb)->s_mount_opt2 &= \
1034 						~EXT4_MOUNT2_##opt
1035 #define set_opt2(sb, opt)		EXT4_SB(sb)->s_mount_opt2 |= \
1036 						EXT4_MOUNT2_##opt
1037 #define test_opt2(sb, opt)		(EXT4_SB(sb)->s_mount_opt2 & \
1038 					 EXT4_MOUNT2_##opt)
1039 
1040 #define ext4_test_and_set_bit		__test_and_set_bit_le
1041 #define ext4_set_bit			__set_bit_le
1042 #define ext4_set_bit_atomic		ext2_set_bit_atomic
1043 #define ext4_test_and_clear_bit		__test_and_clear_bit_le
1044 #define ext4_clear_bit			__clear_bit_le
1045 #define ext4_clear_bit_atomic		ext2_clear_bit_atomic
1046 #define ext4_test_bit			test_bit_le
1047 #define ext4_find_next_zero_bit		find_next_zero_bit_le
1048 #define ext4_find_next_bit		find_next_bit_le
1049 
1050 extern void ext4_set_bits(void *bm, int cur, int len);
1051 
1052 /*
1053  * Maximal mount counts between two filesystem checks
1054  */
1055 #define EXT4_DFL_MAX_MNT_COUNT		20	/* Allow 20 mounts */
1056 #define EXT4_DFL_CHECKINTERVAL		0	/* Don't use interval check */
1057 
1058 /*
1059  * Behaviour when detecting errors
1060  */
1061 #define EXT4_ERRORS_CONTINUE		1	/* Continue execution */
1062 #define EXT4_ERRORS_RO			2	/* Remount fs read-only */
1063 #define EXT4_ERRORS_PANIC		3	/* Panic */
1064 #define EXT4_ERRORS_DEFAULT		EXT4_ERRORS_CONTINUE
1065 
1066 /* Metadata checksum algorithm codes */
1067 #define EXT4_CRC32C_CHKSUM		1
1068 
1069 /*
1070  * Structure of the super block
1071  */
1072 struct ext4_super_block {
1073 /*00*/	__le32	s_inodes_count;		/* Inodes count */
1074 	__le32	s_blocks_count_lo;	/* Blocks count */
1075 	__le32	s_r_blocks_count_lo;	/* Reserved blocks count */
1076 	__le32	s_free_blocks_count_lo;	/* Free blocks count */
1077 /*10*/	__le32	s_free_inodes_count;	/* Free inodes count */
1078 	__le32	s_first_data_block;	/* First Data Block */
1079 	__le32	s_log_block_size;	/* Block size */
1080 	__le32	s_log_cluster_size;	/* Allocation cluster size */
1081 /*20*/	__le32	s_blocks_per_group;	/* # Blocks per group */
1082 	__le32	s_clusters_per_group;	/* # Clusters per group */
1083 	__le32	s_inodes_per_group;	/* # Inodes per group */
1084 	__le32	s_mtime;		/* Mount time */
1085 /*30*/	__le32	s_wtime;		/* Write time */
1086 	__le16	s_mnt_count;		/* Mount count */
1087 	__le16	s_max_mnt_count;	/* Maximal mount count */
1088 	__le16	s_magic;		/* Magic signature */
1089 	__le16	s_state;		/* File system state */
1090 	__le16	s_errors;		/* Behaviour when detecting errors */
1091 	__le16	s_minor_rev_level;	/* minor revision level */
1092 /*40*/	__le32	s_lastcheck;		/* time of last check */
1093 	__le32	s_checkinterval;	/* max. time between checks */
1094 	__le32	s_creator_os;		/* OS */
1095 	__le32	s_rev_level;		/* Revision level */
1096 /*50*/	__le16	s_def_resuid;		/* Default uid for reserved blocks */
1097 	__le16	s_def_resgid;		/* Default gid for reserved blocks */
1098 	/*
1099 	 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1100 	 *
1101 	 * Note: the difference between the compatible feature set and
1102 	 * the incompatible feature set is that if there is a bit set
1103 	 * in the incompatible feature set that the kernel doesn't
1104 	 * know about, it should refuse to mount the filesystem.
1105 	 *
1106 	 * e2fsck's requirements are more strict; if it doesn't know
1107 	 * about a feature in either the compatible or incompatible
1108 	 * feature set, it must abort and not try to meddle with
1109 	 * things it doesn't understand...
1110 	 */
1111 	__le32	s_first_ino;		/* First non-reserved inode */
1112 	__le16  s_inode_size;		/* size of inode structure */
1113 	__le16	s_block_group_nr;	/* block group # of this superblock */
1114 	__le32	s_feature_compat;	/* compatible feature set */
1115 /*60*/	__le32	s_feature_incompat;	/* incompatible feature set */
1116 	__le32	s_feature_ro_compat;	/* readonly-compatible feature set */
1117 /*68*/	__u8	s_uuid[16];		/* 128-bit uuid for volume */
1118 /*78*/	char	s_volume_name[16];	/* volume name */
1119 /*88*/	char	s_last_mounted[64];	/* directory where last mounted */
1120 /*C8*/	__le32	s_algorithm_usage_bitmap; /* For compression */
1121 	/*
1122 	 * Performance hints.  Directory preallocation should only
1123 	 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1124 	 */
1125 	__u8	s_prealloc_blocks;	/* Nr of blocks to try to preallocate*/
1126 	__u8	s_prealloc_dir_blocks;	/* Nr to preallocate for dirs */
1127 	__le16	s_reserved_gdt_blocks;	/* Per group desc for online growth */
1128 	/*
1129 	 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1130 	 */
1131 /*D0*/	__u8	s_journal_uuid[16];	/* uuid of journal superblock */
1132 /*E0*/	__le32	s_journal_inum;		/* inode number of journal file */
1133 	__le32	s_journal_dev;		/* device number of journal file */
1134 	__le32	s_last_orphan;		/* start of list of inodes to delete */
1135 	__le32	s_hash_seed[4];		/* HTREE hash seed */
1136 	__u8	s_def_hash_version;	/* Default hash version to use */
1137 	__u8	s_jnl_backup_type;
1138 	__le16  s_desc_size;		/* size of group descriptor */
1139 /*100*/	__le32	s_default_mount_opts;
1140 	__le32	s_first_meta_bg;	/* First metablock block group */
1141 	__le32	s_mkfs_time;		/* When the filesystem was created */
1142 	__le32	s_jnl_blocks[17];	/* Backup of the journal inode */
1143 	/* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1144 /*150*/	__le32	s_blocks_count_hi;	/* Blocks count */
1145 	__le32	s_r_blocks_count_hi;	/* Reserved blocks count */
1146 	__le32	s_free_blocks_count_hi;	/* Free blocks count */
1147 	__le16	s_min_extra_isize;	/* All inodes have at least # bytes */
1148 	__le16	s_want_extra_isize; 	/* New inodes should reserve # bytes */
1149 	__le32	s_flags;		/* Miscellaneous flags */
1150 	__le16  s_raid_stride;		/* RAID stride */
1151 	__le16  s_mmp_update_interval;  /* # seconds to wait in MMP checking */
1152 	__le64  s_mmp_block;            /* Block for multi-mount protection */
1153 	__le32  s_raid_stripe_width;    /* blocks on all data disks (N*stride)*/
1154 	__u8	s_log_groups_per_flex;  /* FLEX_BG group size */
1155 	__u8	s_checksum_type;	/* metadata checksum algorithm used */
1156 	__u8	s_encryption_level;	/* versioning level for encryption */
1157 	__u8	s_reserved_pad;		/* Padding to next 32bits */
1158 	__le64	s_kbytes_written;	/* nr of lifetime kilobytes written */
1159 	__le32	s_snapshot_inum;	/* Inode number of active snapshot */
1160 	__le32	s_snapshot_id;		/* sequential ID of active snapshot */
1161 	__le64	s_snapshot_r_blocks_count; /* reserved blocks for active
1162 					      snapshot's future use */
1163 	__le32	s_snapshot_list;	/* inode number of the head of the
1164 					   on-disk snapshot list */
1165 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1166 	__le32	s_error_count;		/* number of fs errors */
1167 	__le32	s_first_error_time;	/* first time an error happened */
1168 	__le32	s_first_error_ino;	/* inode involved in first error */
1169 	__le64	s_first_error_block;	/* block involved of first error */
1170 	__u8	s_first_error_func[32];	/* function where the error happened */
1171 	__le32	s_first_error_line;	/* line number where error happened */
1172 	__le32	s_last_error_time;	/* most recent time of an error */
1173 	__le32	s_last_error_ino;	/* inode involved in last error */
1174 	__le32	s_last_error_line;	/* line number where error happened */
1175 	__le64	s_last_error_block;	/* block involved of last error */
1176 	__u8	s_last_error_func[32];	/* function where the error happened */
1177 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1178 	__u8	s_mount_opts[64];
1179 	__le32	s_usr_quota_inum;	/* inode for tracking user quota */
1180 	__le32	s_grp_quota_inum;	/* inode for tracking group quota */
1181 	__le32	s_overhead_clusters;	/* overhead blocks/clusters in fs */
1182 	__le32	s_backup_bgs[2];	/* groups with sparse_super2 SBs */
1183 	__u8	s_encrypt_algos[4];	/* Encryption algorithms in use  */
1184 	__u8	s_encrypt_pw_salt[16];	/* Salt used for string2key algorithm */
1185 	__le32	s_lpf_ino;		/* Location of the lost+found inode */
1186 	__le32	s_reserved[100];	/* Padding to the end of the block */
1187 	__le32	s_checksum;		/* crc32c(superblock) */
1188 };
1189 
1190 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1191 
1192 #ifdef __KERNEL__
1193 
1194 /*
1195  * run-time mount flags
1196  */
1197 #define EXT4_MF_MNTDIR_SAMPLED		0x0001
1198 #define EXT4_MF_FS_ABORTED		0x0002	/* Fatal error detected */
1199 #define EXT4_MF_TEST_DUMMY_ENCRYPTION	0x0004
1200 
1201 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1202 #define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \
1203 						EXT4_MF_TEST_DUMMY_ENCRYPTION))
1204 #else
1205 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1206 #endif
1207 
1208 /* Number of quota types we support */
1209 #define EXT4_MAXQUOTAS 2
1210 
1211 /*
1212  * fourth extended-fs super-block data in memory
1213  */
1214 struct ext4_sb_info {
1215 	unsigned long s_desc_size;	/* Size of a group descriptor in bytes */
1216 	unsigned long s_inodes_per_block;/* Number of inodes per block */
1217 	unsigned long s_blocks_per_group;/* Number of blocks in a group */
1218 	unsigned long s_clusters_per_group; /* Number of clusters in a group */
1219 	unsigned long s_inodes_per_group;/* Number of inodes in a group */
1220 	unsigned long s_itb_per_group;	/* Number of inode table blocks per group */
1221 	unsigned long s_gdb_count;	/* Number of group descriptor blocks */
1222 	unsigned long s_desc_per_block;	/* Number of group descriptors per block */
1223 	ext4_group_t s_groups_count;	/* Number of groups in the fs */
1224 	ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1225 	unsigned long s_overhead;  /* # of fs overhead clusters */
1226 	unsigned int s_cluster_ratio;	/* Number of blocks per cluster */
1227 	unsigned int s_cluster_bits;	/* log2 of s_cluster_ratio */
1228 	loff_t s_bitmap_maxbytes;	/* max bytes for bitmap files */
1229 	struct buffer_head * s_sbh;	/* Buffer containing the super block */
1230 	struct ext4_super_block *s_es;	/* Pointer to the super block in the buffer */
1231 	struct buffer_head **s_group_desc;
1232 	unsigned int s_mount_opt;
1233 	unsigned int s_mount_opt2;
1234 	unsigned int s_mount_flags;
1235 	unsigned int s_def_mount_opt;
1236 	ext4_fsblk_t s_sb_block;
1237 	atomic64_t s_resv_clusters;
1238 	kuid_t s_resuid;
1239 	kgid_t s_resgid;
1240 	unsigned short s_mount_state;
1241 	unsigned short s_pad;
1242 	int s_addr_per_block_bits;
1243 	int s_desc_per_block_bits;
1244 	int s_inode_size;
1245 	int s_first_ino;
1246 	unsigned int s_inode_readahead_blks;
1247 	unsigned int s_inode_goal;
1248 	spinlock_t s_next_gen_lock;
1249 	u32 s_next_generation;
1250 	u32 s_hash_seed[4];
1251 	int s_def_hash_version;
1252 	int s_hash_unsigned;	/* 3 if hash should be signed, 0 if not */
1253 	struct percpu_counter s_freeclusters_counter;
1254 	struct percpu_counter s_freeinodes_counter;
1255 	struct percpu_counter s_dirs_counter;
1256 	struct percpu_counter s_dirtyclusters_counter;
1257 	struct blockgroup_lock *s_blockgroup_lock;
1258 	struct proc_dir_entry *s_proc;
1259 	struct kobject s_kobj;
1260 	struct completion s_kobj_unregister;
1261 	struct super_block *s_sb;
1262 
1263 	/* Journaling */
1264 	struct journal_s *s_journal;
1265 	struct list_head s_orphan;
1266 	struct mutex s_orphan_lock;
1267 	unsigned long s_resize_flags;		/* Flags indicating if there
1268 						   is a resizer */
1269 	unsigned long s_commit_interval;
1270 	u32 s_max_batch_time;
1271 	u32 s_min_batch_time;
1272 	struct block_device *journal_bdev;
1273 #ifdef CONFIG_QUOTA
1274 	char *s_qf_names[EXT4_MAXQUOTAS];	/* Names of quota files with journalled quota */
1275 	int s_jquota_fmt;			/* Format of quota to use */
1276 #endif
1277 	unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1278 	struct rb_root system_blks;
1279 
1280 #ifdef EXTENTS_STATS
1281 	/* ext4 extents stats */
1282 	unsigned long s_ext_min;
1283 	unsigned long s_ext_max;
1284 	unsigned long s_depth_max;
1285 	spinlock_t s_ext_stats_lock;
1286 	unsigned long s_ext_blocks;
1287 	unsigned long s_ext_extents;
1288 #endif
1289 
1290 	/* for buddy allocator */
1291 	struct ext4_group_info ***s_group_info;
1292 	struct inode *s_buddy_cache;
1293 	spinlock_t s_md_lock;
1294 	unsigned short *s_mb_offsets;
1295 	unsigned int *s_mb_maxs;
1296 	unsigned int s_group_info_size;
1297 
1298 	/* tunables */
1299 	unsigned long s_stripe;
1300 	unsigned int s_mb_stream_request;
1301 	unsigned int s_mb_max_to_scan;
1302 	unsigned int s_mb_min_to_scan;
1303 	unsigned int s_mb_stats;
1304 	unsigned int s_mb_order2_reqs;
1305 	unsigned int s_mb_group_prealloc;
1306 	unsigned int s_max_dir_size_kb;
1307 	/* where last allocation was done - for stream allocation */
1308 	unsigned long s_mb_last_group;
1309 	unsigned long s_mb_last_start;
1310 
1311 	/* stats for buddy allocator */
1312 	atomic_t s_bal_reqs;	/* number of reqs with len > 1 */
1313 	atomic_t s_bal_success;	/* we found long enough chunks */
1314 	atomic_t s_bal_allocated;	/* in blocks */
1315 	atomic_t s_bal_ex_scanned;	/* total extents scanned */
1316 	atomic_t s_bal_goals;	/* goal hits */
1317 	atomic_t s_bal_breaks;	/* too long searches */
1318 	atomic_t s_bal_2orders;	/* 2^order hits */
1319 	spinlock_t s_bal_lock;
1320 	unsigned long s_mb_buddies_generated;
1321 	unsigned long long s_mb_generation_time;
1322 	atomic_t s_mb_lost_chunks;
1323 	atomic_t s_mb_preallocated;
1324 	atomic_t s_mb_discarded;
1325 	atomic_t s_lock_busy;
1326 
1327 	/* locality groups */
1328 	struct ext4_locality_group __percpu *s_locality_groups;
1329 
1330 	/* for write statistics */
1331 	unsigned long s_sectors_written_start;
1332 	u64 s_kbytes_written;
1333 
1334 	/* the size of zero-out chunk */
1335 	unsigned int s_extent_max_zeroout_kb;
1336 
1337 	unsigned int s_log_groups_per_flex;
1338 	struct flex_groups *s_flex_groups;
1339 	ext4_group_t s_flex_groups_allocated;
1340 
1341 	/* workqueue for reserved extent conversions (buffered io) */
1342 	struct workqueue_struct *rsv_conversion_wq;
1343 
1344 	/* timer for periodic error stats printing */
1345 	struct timer_list s_err_report;
1346 
1347 	/* Lazy inode table initialization info */
1348 	struct ext4_li_request *s_li_request;
1349 	/* Wait multiplier for lazy initialization thread */
1350 	unsigned int s_li_wait_mult;
1351 
1352 	/* Kernel thread for multiple mount protection */
1353 	struct task_struct *s_mmp_tsk;
1354 
1355 	/* record the last minlen when FITRIM is called. */
1356 	atomic_t s_last_trim_minblks;
1357 
1358 	/* Reference to checksum algorithm driver via cryptoapi */
1359 	struct crypto_shash *s_chksum_driver;
1360 
1361 	/* Precomputed FS UUID checksum for seeding other checksums */
1362 	__u32 s_csum_seed;
1363 
1364 	/* Reclaim extents from extent status tree */
1365 	struct shrinker s_es_shrinker;
1366 	struct list_head s_es_lru;
1367 	struct ext4_es_stats s_es_stats;
1368 	struct mb_cache *s_mb_cache;
1369 	spinlock_t s_es_lru_lock ____cacheline_aligned_in_smp;
1370 
1371 	/* Ratelimit ext4 messages. */
1372 	struct ratelimit_state s_err_ratelimit_state;
1373 	struct ratelimit_state s_warning_ratelimit_state;
1374 	struct ratelimit_state s_msg_ratelimit_state;
1375 };
1376 
EXT4_SB(struct super_block * sb)1377 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1378 {
1379 	return sb->s_fs_info;
1380 }
EXT4_I(struct inode * inode)1381 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1382 {
1383 	return container_of(inode, struct ext4_inode_info, vfs_inode);
1384 }
1385 
ext4_current_time(struct inode * inode)1386 static inline struct timespec ext4_current_time(struct inode *inode)
1387 {
1388 	return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1389 		current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1390 }
1391 
ext4_valid_inum(struct super_block * sb,unsigned long ino)1392 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1393 {
1394 	return ino == EXT4_ROOT_INO ||
1395 		ino == EXT4_USR_QUOTA_INO ||
1396 		ino == EXT4_GRP_QUOTA_INO ||
1397 		ino == EXT4_BOOT_LOADER_INO ||
1398 		ino == EXT4_JOURNAL_INO ||
1399 		ino == EXT4_RESIZE_INO ||
1400 		(ino >= EXT4_FIRST_INO(sb) &&
1401 		 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1402 }
1403 
ext4_set_io_unwritten_flag(struct inode * inode,struct ext4_io_end * io_end)1404 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
1405 					      struct ext4_io_end *io_end)
1406 {
1407 	if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
1408 		io_end->flag |= EXT4_IO_END_UNWRITTEN;
1409 		atomic_inc(&EXT4_I(inode)->i_unwritten);
1410 	}
1411 }
1412 
ext4_inode_aio(struct inode * inode)1413 static inline ext4_io_end_t *ext4_inode_aio(struct inode *inode)
1414 {
1415 	return inode->i_private;
1416 }
1417 
ext4_inode_aio_set(struct inode * inode,ext4_io_end_t * io)1418 static inline void ext4_inode_aio_set(struct inode *inode, ext4_io_end_t *io)
1419 {
1420 	inode->i_private = io;
1421 }
1422 
1423 /*
1424  * Inode dynamic state flags
1425  */
1426 enum {
1427 	EXT4_STATE_JDATA,		/* journaled data exists */
1428 	EXT4_STATE_NEW,			/* inode is newly created */
1429 	EXT4_STATE_XATTR,		/* has in-inode xattrs */
1430 	EXT4_STATE_NO_EXPAND,		/* No space for expansion */
1431 	EXT4_STATE_DA_ALLOC_CLOSE,	/* Alloc DA blks on close */
1432 	EXT4_STATE_EXT_MIGRATE,		/* Inode is migrating */
1433 	EXT4_STATE_DIO_UNWRITTEN,	/* need convert on dio done*/
1434 	EXT4_STATE_NEWENTRY,		/* File just added to dir */
1435 	EXT4_STATE_DIOREAD_LOCK,	/* Disable support for dio read
1436 					   nolocking */
1437 	EXT4_STATE_MAY_INLINE_DATA,	/* may have in-inode data */
1438 	EXT4_STATE_ORDERED_MODE,	/* data=ordered mode */
1439 	EXT4_STATE_EXT_PRECACHED,	/* extents have been precached */
1440 };
1441 
1442 #define EXT4_INODE_BIT_FNS(name, field, offset)				\
1443 static inline int ext4_test_inode_##name(struct inode *inode, int bit)	\
1444 {									\
1445 	return test_bit(bit + (offset), &EXT4_I(inode)->i_##field);	\
1446 }									\
1447 static inline void ext4_set_inode_##name(struct inode *inode, int bit)	\
1448 {									\
1449 	set_bit(bit + (offset), &EXT4_I(inode)->i_##field);		\
1450 }									\
1451 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1452 {									\
1453 	clear_bit(bit + (offset), &EXT4_I(inode)->i_##field);		\
1454 }
1455 
1456 /* Add these declarations here only so that these functions can be
1457  * found by name.  Otherwise, they are very hard to locate. */
1458 static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1459 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1460 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1461 EXT4_INODE_BIT_FNS(flag, flags, 0)
1462 
1463 /* Add these declarations here only so that these functions can be
1464  * found by name.  Otherwise, they are very hard to locate. */
1465 static inline int ext4_test_inode_state(struct inode *inode, int bit);
1466 static inline void ext4_set_inode_state(struct inode *inode, int bit);
1467 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1468 #if (BITS_PER_LONG < 64)
1469 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1470 
ext4_clear_state_flags(struct ext4_inode_info * ei)1471 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1472 {
1473 	(ei)->i_state_flags = 0;
1474 }
1475 #else
1476 EXT4_INODE_BIT_FNS(state, flags, 32)
1477 
ext4_clear_state_flags(struct ext4_inode_info * ei)1478 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1479 {
1480 	/* We depend on the fact that callers will set i_flags */
1481 }
1482 #endif
1483 #else
1484 /* Assume that user mode programs are passing in an ext4fs superblock, not
1485  * a kernel struct super_block.  This will allow us to call the feature-test
1486  * macros from user land. */
1487 #define EXT4_SB(sb)	(sb)
1488 #endif
1489 
1490 /*
1491  * Returns true if the inode is inode is encrypted
1492  */
ext4_encrypted_inode(struct inode * inode)1493 static inline int ext4_encrypted_inode(struct inode *inode)
1494 {
1495 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1496 	return ext4_test_inode_flag(inode, EXT4_INODE_ENCRYPT);
1497 #else
1498 	return 0;
1499 #endif
1500 }
1501 
1502 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1503 
1504 /*
1505  * Codes for operating systems
1506  */
1507 #define EXT4_OS_LINUX		0
1508 #define EXT4_OS_HURD		1
1509 #define EXT4_OS_MASIX		2
1510 #define EXT4_OS_FREEBSD		3
1511 #define EXT4_OS_LITES		4
1512 
1513 /*
1514  * Revision levels
1515  */
1516 #define EXT4_GOOD_OLD_REV	0	/* The good old (original) format */
1517 #define EXT4_DYNAMIC_REV	1	/* V2 format w/ dynamic inode sizes */
1518 
1519 #define EXT4_CURRENT_REV	EXT4_GOOD_OLD_REV
1520 #define EXT4_MAX_SUPP_REV	EXT4_DYNAMIC_REV
1521 
1522 #define EXT4_GOOD_OLD_INODE_SIZE 128
1523 
1524 /*
1525  * Feature set definitions
1526  */
1527 
1528 #define EXT4_HAS_COMPAT_FEATURE(sb,mask)			\
1529 	((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1530 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask)			\
1531 	((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1532 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask)			\
1533 	((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1534 #define EXT4_SET_COMPAT_FEATURE(sb,mask)			\
1535 	EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1536 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask)			\
1537 	EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1538 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask)			\
1539 	EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1540 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask)			\
1541 	EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1542 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask)			\
1543 	EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1544 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask)			\
1545 	EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1546 
1547 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC	0x0001
1548 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES	0x0002
1549 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL		0x0004
1550 #define EXT4_FEATURE_COMPAT_EXT_ATTR		0x0008
1551 #define EXT4_FEATURE_COMPAT_RESIZE_INODE	0x0010
1552 #define EXT4_FEATURE_COMPAT_DIR_INDEX		0x0020
1553 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2	0x0200
1554 
1555 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER	0x0001
1556 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE	0x0002
1557 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR	0x0004
1558 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE        0x0008
1559 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM		0x0010
1560 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK	0x0020
1561 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE	0x0040
1562 #define EXT4_FEATURE_RO_COMPAT_QUOTA		0x0100
1563 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC		0x0200
1564 /*
1565  * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM).  When
1566  * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1567  * all other data structures' checksums.  However, the METADATA_CSUM and
1568  * GDT_CSUM bits are mutually exclusive.
1569  */
1570 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM	0x0400
1571 
1572 #define EXT4_FEATURE_INCOMPAT_COMPRESSION	0x0001
1573 #define EXT4_FEATURE_INCOMPAT_FILETYPE		0x0002
1574 #define EXT4_FEATURE_INCOMPAT_RECOVER		0x0004 /* Needs recovery */
1575 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV	0x0008 /* Journal device */
1576 #define EXT4_FEATURE_INCOMPAT_META_BG		0x0010
1577 #define EXT4_FEATURE_INCOMPAT_EXTENTS		0x0040 /* extents support */
1578 #define EXT4_FEATURE_INCOMPAT_64BIT		0x0080
1579 #define EXT4_FEATURE_INCOMPAT_MMP               0x0100
1580 #define EXT4_FEATURE_INCOMPAT_FLEX_BG		0x0200
1581 #define EXT4_FEATURE_INCOMPAT_EA_INODE		0x0400 /* EA in inode */
1582 #define EXT4_FEATURE_INCOMPAT_DIRDATA		0x1000 /* data in dirent */
1583 #define EXT4_FEATURE_INCOMPAT_BG_USE_META_CSUM	0x2000 /* use crc32c for bg */
1584 #define EXT4_FEATURE_INCOMPAT_LARGEDIR		0x4000 /* >2GB or 3-lvl htree */
1585 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA	0x8000 /* data in inode */
1586 #define EXT4_FEATURE_INCOMPAT_ENCRYPT		0x10000
1587 
1588 #define EXT2_FEATURE_COMPAT_SUPP	EXT4_FEATURE_COMPAT_EXT_ATTR
1589 #define EXT2_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
1590 					 EXT4_FEATURE_INCOMPAT_META_BG)
1591 #define EXT2_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1592 					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1593 					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1594 
1595 #define EXT3_FEATURE_COMPAT_SUPP	EXT4_FEATURE_COMPAT_EXT_ATTR
1596 #define EXT3_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
1597 					 EXT4_FEATURE_INCOMPAT_RECOVER| \
1598 					 EXT4_FEATURE_INCOMPAT_META_BG)
1599 #define EXT3_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1600 					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1601 					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1602 
1603 #define EXT4_FEATURE_COMPAT_SUPP	EXT2_FEATURE_COMPAT_EXT_ATTR
1604 #define EXT4_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
1605 					 EXT4_FEATURE_INCOMPAT_RECOVER| \
1606 					 EXT4_FEATURE_INCOMPAT_META_BG| \
1607 					 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1608 					 EXT4_FEATURE_INCOMPAT_64BIT| \
1609 					 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1610 					 EXT4_FEATURE_INCOMPAT_MMP | \
1611 					 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
1612 					 EXT4_FEATURE_INCOMPAT_ENCRYPT)
1613 #define EXT4_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1614 					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1615 					 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1616 					 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1617 					 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1618 					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1619 					 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1620 					 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1621 					 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1622 					 EXT4_FEATURE_RO_COMPAT_QUOTA)
1623 
1624 /*
1625  * Default values for user and/or group using reserved blocks
1626  */
1627 #define	EXT4_DEF_RESUID		0
1628 #define	EXT4_DEF_RESGID		0
1629 
1630 #define EXT4_DEF_INODE_READAHEAD_BLKS	32
1631 
1632 /*
1633  * Default mount options
1634  */
1635 #define EXT4_DEFM_DEBUG		0x0001
1636 #define EXT4_DEFM_BSDGROUPS	0x0002
1637 #define EXT4_DEFM_XATTR_USER	0x0004
1638 #define EXT4_DEFM_ACL		0x0008
1639 #define EXT4_DEFM_UID16		0x0010
1640 #define EXT4_DEFM_JMODE		0x0060
1641 #define EXT4_DEFM_JMODE_DATA	0x0020
1642 #define EXT4_DEFM_JMODE_ORDERED	0x0040
1643 #define EXT4_DEFM_JMODE_WBACK	0x0060
1644 #define EXT4_DEFM_NOBARRIER	0x0100
1645 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1646 #define EXT4_DEFM_DISCARD	0x0400
1647 #define EXT4_DEFM_NODELALLOC	0x0800
1648 
1649 /*
1650  * Default journal batch times
1651  */
1652 #define EXT4_DEF_MIN_BATCH_TIME	0
1653 #define EXT4_DEF_MAX_BATCH_TIME	15000 /* 15ms */
1654 
1655 /*
1656  * Minimum number of groups in a flexgroup before we separate out
1657  * directories into the first block group of a flexgroup
1658  */
1659 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME	4
1660 
1661 /*
1662  * Structure of a directory entry
1663  */
1664 #define EXT4_NAME_LEN 255
1665 
1666 struct ext4_dir_entry {
1667 	__le32	inode;			/* Inode number */
1668 	__le16	rec_len;		/* Directory entry length */
1669 	__le16	name_len;		/* Name length */
1670 	char	name[EXT4_NAME_LEN];	/* File name */
1671 };
1672 
1673 /*
1674  * The new version of the directory entry.  Since EXT4 structures are
1675  * stored in intel byte order, and the name_len field could never be
1676  * bigger than 255 chars, it's safe to reclaim the extra byte for the
1677  * file_type field.
1678  */
1679 struct ext4_dir_entry_2 {
1680 	__le32	inode;			/* Inode number */
1681 	__le16	rec_len;		/* Directory entry length */
1682 	__u8	name_len;		/* Name length */
1683 	__u8	file_type;
1684 	char	name[EXT4_NAME_LEN];	/* File name */
1685 };
1686 
1687 /*
1688  * This is a bogus directory entry at the end of each leaf block that
1689  * records checksums.
1690  */
1691 struct ext4_dir_entry_tail {
1692 	__le32	det_reserved_zero1;	/* Pretend to be unused */
1693 	__le16	det_rec_len;		/* 12 */
1694 	__u8	det_reserved_zero2;	/* Zero name length */
1695 	__u8	det_reserved_ft;	/* 0xDE, fake file type */
1696 	__le32	det_checksum;		/* crc32c(uuid+inum+dirblock) */
1697 };
1698 
1699 #define EXT4_DIRENT_TAIL(block, blocksize) \
1700 	((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1701 					((blocksize) - \
1702 					 sizeof(struct ext4_dir_entry_tail))))
1703 
1704 /*
1705  * Ext4 directory file types.  Only the low 3 bits are used.  The
1706  * other bits are reserved for now.
1707  */
1708 #define EXT4_FT_UNKNOWN		0
1709 #define EXT4_FT_REG_FILE	1
1710 #define EXT4_FT_DIR		2
1711 #define EXT4_FT_CHRDEV		3
1712 #define EXT4_FT_BLKDEV		4
1713 #define EXT4_FT_FIFO		5
1714 #define EXT4_FT_SOCK		6
1715 #define EXT4_FT_SYMLINK		7
1716 
1717 #define EXT4_FT_MAX		8
1718 
1719 #define EXT4_FT_DIR_CSUM	0xDE
1720 
1721 /*
1722  * EXT4_DIR_PAD defines the directory entries boundaries
1723  *
1724  * NOTE: It must be a multiple of 4
1725  */
1726 #define EXT4_DIR_PAD			4
1727 #define EXT4_DIR_ROUND			(EXT4_DIR_PAD - 1)
1728 #define EXT4_DIR_REC_LEN(name_len)	(((name_len) + 8 + EXT4_DIR_ROUND) & \
1729 					 ~EXT4_DIR_ROUND)
1730 #define EXT4_MAX_REC_LEN		((1<<16)-1)
1731 
1732 /*
1733  * If we ever get support for fs block sizes > page_size, we'll need
1734  * to remove the #if statements in the next two functions...
1735  */
1736 static inline unsigned int
ext4_rec_len_from_disk(__le16 dlen,unsigned blocksize)1737 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1738 {
1739 	unsigned len = le16_to_cpu(dlen);
1740 
1741 #if (PAGE_CACHE_SIZE >= 65536)
1742 	if (len == EXT4_MAX_REC_LEN || len == 0)
1743 		return blocksize;
1744 	return (len & 65532) | ((len & 3) << 16);
1745 #else
1746 	return len;
1747 #endif
1748 }
1749 
ext4_rec_len_to_disk(unsigned len,unsigned blocksize)1750 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1751 {
1752 	if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1753 		BUG();
1754 #if (PAGE_CACHE_SIZE >= 65536)
1755 	if (len < 65536)
1756 		return cpu_to_le16(len);
1757 	if (len == blocksize) {
1758 		if (blocksize == 65536)
1759 			return cpu_to_le16(EXT4_MAX_REC_LEN);
1760 		else
1761 			return cpu_to_le16(0);
1762 	}
1763 	return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1764 #else
1765 	return cpu_to_le16(len);
1766 #endif
1767 }
1768 
1769 /*
1770  * Hash Tree Directory indexing
1771  * (c) Daniel Phillips, 2001
1772  */
1773 
1774 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1775 				      EXT4_FEATURE_COMPAT_DIR_INDEX) && \
1776 		    ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
1777 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1778 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1779 
1780 /* Legal values for the dx_root hash_version field: */
1781 
1782 #define DX_HASH_LEGACY		0
1783 #define DX_HASH_HALF_MD4	1
1784 #define DX_HASH_TEA		2
1785 #define DX_HASH_LEGACY_UNSIGNED	3
1786 #define DX_HASH_HALF_MD4_UNSIGNED	4
1787 #define DX_HASH_TEA_UNSIGNED		5
1788 
ext4_chksum(struct ext4_sb_info * sbi,u32 crc,const void * address,unsigned int length)1789 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
1790 			      const void *address, unsigned int length)
1791 {
1792 	struct {
1793 		struct shash_desc shash;
1794 		char ctx[4];
1795 	} desc;
1796 	int err;
1797 
1798 	BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
1799 
1800 	desc.shash.tfm = sbi->s_chksum_driver;
1801 	desc.shash.flags = 0;
1802 	*(u32 *)desc.ctx = crc;
1803 
1804 	err = crypto_shash_update(&desc.shash, address, length);
1805 	BUG_ON(err);
1806 
1807 	return *(u32 *)desc.ctx;
1808 }
1809 
1810 #ifdef __KERNEL__
1811 
1812 /* hash info structure used by the directory hash */
1813 struct dx_hash_info
1814 {
1815 	u32		hash;
1816 	u32		minor_hash;
1817 	int		hash_version;
1818 	u32		*seed;
1819 };
1820 
1821 
1822 /* 32 and 64 bit signed EOF for dx directories */
1823 #define EXT4_HTREE_EOF_32BIT   ((1UL  << (32 - 1)) - 1)
1824 #define EXT4_HTREE_EOF_64BIT   ((1ULL << (64 - 1)) - 1)
1825 
1826 
1827 /*
1828  * Control parameters used by ext4_htree_next_block
1829  */
1830 #define HASH_NB_ALWAYS		1
1831 
1832 struct ext4_filename {
1833 	const struct qstr *usr_fname;
1834 	struct ext4_str disk_name;
1835 	struct dx_hash_info hinfo;
1836 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1837 	struct ext4_str crypto_buf;
1838 #endif
1839 };
1840 
1841 #define fname_name(p) ((p)->disk_name.name)
1842 #define fname_len(p)  ((p)->disk_name.len)
1843 
1844 /*
1845  * Describe an inode's exact location on disk and in memory
1846  */
1847 struct ext4_iloc
1848 {
1849 	struct buffer_head *bh;
1850 	unsigned long offset;
1851 	ext4_group_t block_group;
1852 };
1853 
ext4_raw_inode(struct ext4_iloc * iloc)1854 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
1855 {
1856 	return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
1857 }
1858 
1859 /*
1860  * This structure is stuffed into the struct file's private_data field
1861  * for directories.  It is where we put information so that we can do
1862  * readdir operations in hash tree order.
1863  */
1864 struct dir_private_info {
1865 	struct rb_root	root;
1866 	struct rb_node	*curr_node;
1867 	struct fname	*extra_fname;
1868 	loff_t		last_pos;
1869 	__u32		curr_hash;
1870 	__u32		curr_minor_hash;
1871 	__u32		next_hash;
1872 };
1873 
1874 /* calculate the first block number of the group */
1875 static inline ext4_fsblk_t
ext4_group_first_block_no(struct super_block * sb,ext4_group_t group_no)1876 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
1877 {
1878 	return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
1879 		le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
1880 }
1881 
1882 /*
1883  * Special error return code only used by dx_probe() and its callers.
1884  */
1885 #define ERR_BAD_DX_DIR	(-(MAX_ERRNO - 1))
1886 
1887 /*
1888  * Timeout and state flag for lazy initialization inode thread.
1889  */
1890 #define EXT4_DEF_LI_WAIT_MULT			10
1891 #define EXT4_DEF_LI_MAX_START_DELAY		5
1892 #define EXT4_LAZYINIT_QUIT			0x0001
1893 #define EXT4_LAZYINIT_RUNNING			0x0002
1894 
1895 /*
1896  * Lazy inode table initialization info
1897  */
1898 struct ext4_lazy_init {
1899 	unsigned long		li_state;
1900 	struct list_head	li_request_list;
1901 	struct mutex		li_list_mtx;
1902 };
1903 
1904 struct ext4_li_request {
1905 	struct super_block	*lr_super;
1906 	struct ext4_sb_info	*lr_sbi;
1907 	ext4_group_t		lr_next_group;
1908 	struct list_head	lr_request;
1909 	unsigned long		lr_next_sched;
1910 	unsigned long		lr_timeout;
1911 };
1912 
1913 struct ext4_features {
1914 	struct kobject f_kobj;
1915 	struct completion f_kobj_unregister;
1916 };
1917 
1918 /*
1919  * This structure will be used for multiple mount protection. It will be
1920  * written into the block number saved in the s_mmp_block field in the
1921  * superblock. Programs that check MMP should assume that if
1922  * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
1923  * to use the filesystem, regardless of how old the timestamp is.
1924  */
1925 #define EXT4_MMP_MAGIC     0x004D4D50U /* ASCII for MMP */
1926 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
1927 #define EXT4_MMP_SEQ_FSCK  0xE24D4D50U /* mmp_seq value when being fscked */
1928 #define EXT4_MMP_SEQ_MAX   0xE24D4D4FU /* maximum valid mmp_seq value */
1929 
1930 struct mmp_struct {
1931 	__le32	mmp_magic;		/* Magic number for MMP */
1932 	__le32	mmp_seq;		/* Sequence no. updated periodically */
1933 
1934 	/*
1935 	 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
1936 	 * purposes and do not affect the correctness of the algorithm
1937 	 */
1938 	__le64	mmp_time;		/* Time last updated */
1939 	char	mmp_nodename[64];	/* Node which last updated MMP block */
1940 	char	mmp_bdevname[32];	/* Bdev which last updated MMP block */
1941 
1942 	/*
1943 	 * mmp_check_interval is used to verify if the MMP block has been
1944 	 * updated on the block device. The value is updated based on the
1945 	 * maximum time to write the MMP block during an update cycle.
1946 	 */
1947 	__le16	mmp_check_interval;
1948 
1949 	__le16	mmp_pad1;
1950 	__le32	mmp_pad2[226];
1951 	__le32	mmp_checksum;		/* crc32c(uuid+mmp_block) */
1952 };
1953 
1954 /* arguments passed to the mmp thread */
1955 struct mmpd_data {
1956 	struct buffer_head *bh; /* bh from initial read_mmp_block() */
1957 	struct super_block *sb;  /* super block of the fs */
1958 };
1959 
1960 /*
1961  * Check interval multiplier
1962  * The MMP block is written every update interval and initially checked every
1963  * update interval x the multiplier (the value is then adapted based on the
1964  * write latency). The reason is that writes can be delayed under load and we
1965  * don't want readers to incorrectly assume that the filesystem is no longer
1966  * in use.
1967  */
1968 #define EXT4_MMP_CHECK_MULT		2UL
1969 
1970 /*
1971  * Minimum interval for MMP checking in seconds.
1972  */
1973 #define EXT4_MMP_MIN_CHECK_INTERVAL	5UL
1974 
1975 /*
1976  * Maximum interval for MMP checking in seconds.
1977  */
1978 #define EXT4_MMP_MAX_CHECK_INTERVAL	300UL
1979 
1980 /*
1981  * Function prototypes
1982  */
1983 
1984 /*
1985  * Ok, these declarations are also in <linux/kernel.h> but none of the
1986  * ext4 source programs needs to include it so they are duplicated here.
1987  */
1988 # define NORET_TYPE	/**/
1989 # define ATTRIB_NORET	__attribute__((noreturn))
1990 # define NORET_AND	noreturn,
1991 
1992 /* bitmap.c */
1993 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
1994 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
1995 				struct ext4_group_desc *gdp,
1996 				struct buffer_head *bh, int sz);
1997 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
1998 				  struct ext4_group_desc *gdp,
1999 				  struct buffer_head *bh, int sz);
2000 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2001 				struct ext4_group_desc *gdp,
2002 				struct buffer_head *bh);
2003 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2004 				  struct ext4_group_desc *gdp,
2005 				  struct buffer_head *bh);
2006 
2007 /* balloc.c */
2008 extern void ext4_get_group_no_and_offset(struct super_block *sb,
2009 					 ext4_fsblk_t blocknr,
2010 					 ext4_group_t *blockgrpp,
2011 					 ext4_grpblk_t *offsetp);
2012 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2013 					  ext4_fsblk_t block);
2014 
2015 extern unsigned int ext4_block_group(struct super_block *sb,
2016 			ext4_fsblk_t blocknr);
2017 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2018 			ext4_fsblk_t blocknr);
2019 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2020 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2021 			ext4_group_t group);
2022 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2023 					 ext4_fsblk_t goal,
2024 					 unsigned int flags,
2025 					 unsigned long *count,
2026 					 int *errp);
2027 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2028 				    s64 nclusters, unsigned int flags);
2029 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2030 extern void ext4_check_blocks_bitmap(struct super_block *);
2031 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2032 						    ext4_group_t block_group,
2033 						    struct buffer_head ** bh);
2034 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2035 
2036 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2037 						ext4_group_t block_group);
2038 extern int ext4_wait_block_bitmap(struct super_block *sb,
2039 				  ext4_group_t block_group,
2040 				  struct buffer_head *bh);
2041 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2042 						  ext4_group_t block_group);
2043 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2044 					      ext4_group_t block_group,
2045 					      struct ext4_group_desc *gdp);
2046 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2047 
2048 /* crypto_policy.c */
2049 int ext4_is_child_context_consistent_with_parent(struct inode *parent,
2050 						 struct inode *child);
2051 int ext4_inherit_context(struct inode *parent, struct inode *child);
2052 void ext4_to_hex(char *dst, char *src, size_t src_size);
2053 int ext4_process_policy(const struct ext4_encryption_policy *policy,
2054 			struct inode *inode);
2055 int ext4_get_policy(struct inode *inode,
2056 		    struct ext4_encryption_policy *policy);
2057 
2058 /* crypto.c */
2059 extern struct kmem_cache *ext4_crypt_info_cachep;
2060 bool ext4_valid_contents_enc_mode(uint32_t mode);
2061 uint32_t ext4_validate_encryption_key_size(uint32_t mode, uint32_t size);
2062 extern struct workqueue_struct *ext4_read_workqueue;
2063 struct ext4_crypto_ctx *ext4_get_crypto_ctx(struct inode *inode);
2064 void ext4_release_crypto_ctx(struct ext4_crypto_ctx *ctx);
2065 void ext4_restore_control_page(struct page *data_page);
2066 struct page *ext4_encrypt(struct inode *inode,
2067 			  struct page *plaintext_page);
2068 int ext4_decrypt(struct ext4_crypto_ctx *ctx, struct page *page);
2069 int ext4_decrypt_one(struct inode *inode, struct page *page);
2070 int ext4_encrypted_zeroout(struct inode *inode, struct ext4_extent *ex);
2071 
2072 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2073 int ext4_init_crypto(void);
2074 void ext4_exit_crypto(void);
ext4_sb_has_crypto(struct super_block * sb)2075 static inline int ext4_sb_has_crypto(struct super_block *sb)
2076 {
2077 	return EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_ENCRYPT);
2078 }
2079 #else
ext4_init_crypto(void)2080 static inline int ext4_init_crypto(void) { return 0; }
ext4_exit_crypto(void)2081 static inline void ext4_exit_crypto(void) { }
ext4_sb_has_crypto(struct super_block * sb)2082 static inline int ext4_sb_has_crypto(struct super_block *sb)
2083 {
2084 	return 0;
2085 }
2086 #endif
2087 
2088 /* crypto_fname.c */
2089 bool ext4_valid_filenames_enc_mode(uint32_t mode);
2090 u32 ext4_fname_crypto_round_up(u32 size, u32 blksize);
2091 unsigned ext4_fname_encrypted_size(struct inode *inode, u32 ilen);
2092 int ext4_fname_crypto_alloc_buffer(struct inode *inode,
2093 				   u32 ilen, struct ext4_str *crypto_str);
2094 int _ext4_fname_disk_to_usr(struct inode *inode,
2095 			    struct dx_hash_info *hinfo,
2096 			    const struct ext4_str *iname,
2097 			    struct ext4_str *oname);
2098 int ext4_fname_disk_to_usr(struct inode *inode,
2099 			   struct dx_hash_info *hinfo,
2100 			   const struct ext4_dir_entry_2 *de,
2101 			   struct ext4_str *oname);
2102 int ext4_fname_usr_to_disk(struct inode *inode,
2103 			   const struct qstr *iname,
2104 			   struct ext4_str *oname);
2105 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2106 void ext4_fname_crypto_free_buffer(struct ext4_str *crypto_str);
2107 int ext4_fname_setup_filename(struct inode *dir, const struct qstr *iname,
2108 			      int lookup, struct ext4_filename *fname);
2109 void ext4_fname_free_filename(struct ext4_filename *fname);
2110 #else
2111 static inline
ext4_setup_fname_crypto(struct inode * inode)2112 int ext4_setup_fname_crypto(struct inode *inode)
2113 {
2114 	return 0;
2115 }
ext4_fname_crypto_free_buffer(struct ext4_str * p)2116 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)2117 static inline int ext4_fname_setup_filename(struct inode *dir,
2118 				     const struct qstr *iname,
2119 				     int lookup, struct ext4_filename *fname)
2120 {
2121 	fname->usr_fname = iname;
2122 	fname->disk_name.name = (unsigned char *) iname->name;
2123 	fname->disk_name.len = iname->len;
2124 	return 0;
2125 }
ext4_fname_free_filename(struct ext4_filename * fname)2126 static inline void ext4_fname_free_filename(struct ext4_filename *fname) { }
2127 #endif
2128 
2129 
2130 /* crypto_key.c */
2131 void ext4_free_crypt_info(struct ext4_crypt_info *ci);
2132 void ext4_free_encryption_info(struct inode *inode, struct ext4_crypt_info *ci);
2133 int _ext4_get_encryption_info(struct inode *inode);
2134 
2135 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2136 int ext4_has_encryption_key(struct inode *inode);
2137 
ext4_get_encryption_info(struct inode * inode)2138 static inline int ext4_get_encryption_info(struct inode *inode)
2139 {
2140 	struct ext4_crypt_info *ci = EXT4_I(inode)->i_crypt_info;
2141 
2142 	if (!ci ||
2143 	    (ci->ci_keyring_key &&
2144 	     (ci->ci_keyring_key->flags & ((1 << KEY_FLAG_INVALIDATED) |
2145 					   (1 << KEY_FLAG_REVOKED) |
2146 					   (1 << KEY_FLAG_DEAD)))))
2147 		return _ext4_get_encryption_info(inode);
2148 	return 0;
2149 }
2150 
ext4_encryption_info(struct inode * inode)2151 static inline struct ext4_crypt_info *ext4_encryption_info(struct inode *inode)
2152 {
2153 	return EXT4_I(inode)->i_crypt_info;
2154 }
2155 
2156 #else
ext4_has_encryption_key(struct inode * inode)2157 static inline int ext4_has_encryption_key(struct inode *inode)
2158 {
2159 	return 0;
2160 }
ext4_get_encryption_info(struct inode * inode)2161 static inline int ext4_get_encryption_info(struct inode *inode)
2162 {
2163 	return 0;
2164 }
ext4_encryption_info(struct inode * inode)2165 static inline struct ext4_crypt_info *ext4_encryption_info(struct inode *inode)
2166 {
2167 	return NULL;
2168 }
2169 #endif
2170 
2171 
2172 /* dir.c */
2173 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2174 				  struct file *,
2175 				  struct ext4_dir_entry_2 *,
2176 				  struct buffer_head *, char *, int,
2177 				  unsigned int);
2178 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset)	\
2179 	unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2180 					(de), (bh), (buf), (size), (offset)))
2181 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2182 				__u32 minor_hash,
2183 				struct ext4_dir_entry_2 *dirent,
2184 				struct ext4_str *ent_name);
2185 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2186 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2187 			     struct buffer_head *bh,
2188 			     void *buf, int buf_size,
2189 			     struct ext4_filename *fname,
2190 			     struct ext4_dir_entry_2 **dest_de);
2191 int ext4_insert_dentry(struct inode *dir,
2192 		       struct inode *inode,
2193 		       struct ext4_dir_entry_2 *de,
2194 		       int buf_size,
2195 		       struct ext4_filename *fname);
ext4_update_dx_flag(struct inode * inode)2196 static inline void ext4_update_dx_flag(struct inode *inode)
2197 {
2198 	if (!EXT4_HAS_COMPAT_FEATURE(inode->i_sb,
2199 				     EXT4_FEATURE_COMPAT_DIR_INDEX))
2200 		ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2201 }
2202 static unsigned char ext4_filetype_table[] = {
2203 	DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2204 };
2205 
get_dtype(struct super_block * sb,int filetype)2206 static inline  unsigned char get_dtype(struct super_block *sb, int filetype)
2207 {
2208 	if (!EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE) ||
2209 	    (filetype >= EXT4_FT_MAX))
2210 		return DT_UNKNOWN;
2211 
2212 	return ext4_filetype_table[filetype];
2213 }
2214 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2215 			     void *buf, int buf_size);
2216 
2217 /* fsync.c */
2218 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2219 
2220 /* hash.c */
2221 extern int ext4fs_dirhash(const char *name, int len, struct
2222 			  dx_hash_info *hinfo);
2223 
2224 /* ialloc.c */
2225 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2226 				      const struct qstr *qstr, __u32 goal,
2227 				      uid_t *owner, int handle_type,
2228 				      unsigned int line_no, int nblocks);
2229 
2230 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner) \
2231 	__ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2232 			 0, 0, 0)
2233 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2234 				    type, nblocks)		    \
2235 	__ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2236 			 (type), __LINE__, (nblocks))
2237 
2238 
2239 extern void ext4_free_inode(handle_t *, struct inode *);
2240 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2241 extern unsigned long ext4_count_free_inodes(struct super_block *);
2242 extern unsigned long ext4_count_dirs(struct super_block *);
2243 extern void ext4_check_inodes_bitmap(struct super_block *);
2244 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2245 extern int ext4_init_inode_table(struct super_block *sb,
2246 				 ext4_group_t group, int barrier);
2247 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2248 
2249 /* mballoc.c */
2250 extern long ext4_mb_stats;
2251 extern long ext4_mb_max_to_scan;
2252 extern int ext4_mb_init(struct super_block *);
2253 extern int ext4_mb_release(struct super_block *);
2254 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2255 				struct ext4_allocation_request *, int *);
2256 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2257 extern void ext4_discard_preallocations(struct inode *);
2258 extern int __init ext4_init_mballoc(void);
2259 extern void ext4_exit_mballoc(void);
2260 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2261 			     struct buffer_head *bh, ext4_fsblk_t block,
2262 			     unsigned long count, int flags);
2263 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2264 				   ext4_group_t ngroups);
2265 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2266 		ext4_group_t i, struct ext4_group_desc *desc);
2267 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2268 				ext4_fsblk_t block, unsigned long count);
2269 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *,
2270 				unsigned long blkdev_flags);
2271 
2272 /* inode.c */
2273 int ext4_inode_is_fast_symlink(struct inode *inode);
2274 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2275 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2276 int ext4_get_block_write(struct inode *inode, sector_t iblock,
2277 			 struct buffer_head *bh_result, int create);
2278 int ext4_get_block(struct inode *inode, sector_t iblock,
2279 				struct buffer_head *bh_result, int create);
2280 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2281 			   struct buffer_head *bh, int create);
2282 int ext4_walk_page_buffers(handle_t *handle,
2283 			   struct buffer_head *head,
2284 			   unsigned from,
2285 			   unsigned to,
2286 			   int *partial,
2287 			   int (*fn)(handle_t *handle,
2288 				     struct buffer_head *bh));
2289 int do_journal_get_write_access(handle_t *handle,
2290 				struct buffer_head *bh);
2291 #define FALL_BACK_TO_NONDELALLOC 1
2292 #define CONVERT_INLINE_DATA	 2
2293 
2294 extern struct inode *ext4_iget(struct super_block *, unsigned long);
2295 extern struct inode *ext4_iget_normal(struct super_block *, unsigned long);
2296 extern int  ext4_write_inode(struct inode *, struct writeback_control *);
2297 extern int  ext4_setattr(struct dentry *, struct iattr *);
2298 extern int  ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
2299 				struct kstat *stat);
2300 extern void ext4_evict_inode(struct inode *);
2301 extern void ext4_clear_inode(struct inode *);
2302 extern int  ext4_sync_inode(handle_t *, struct inode *);
2303 extern void ext4_dirty_inode(struct inode *, int);
2304 extern int ext4_change_inode_journal_flag(struct inode *, int);
2305 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2306 extern int ext4_inode_attach_jinode(struct inode *inode);
2307 extern int ext4_can_truncate(struct inode *inode);
2308 extern void ext4_truncate(struct inode *);
2309 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2310 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
2311 extern void ext4_set_inode_flags(struct inode *);
2312 extern void ext4_get_inode_flags(struct ext4_inode_info *);
2313 extern int ext4_alloc_da_blocks(struct inode *inode);
2314 extern void ext4_set_aops(struct inode *inode);
2315 extern int ext4_writepage_trans_blocks(struct inode *);
2316 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2317 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2318 			     loff_t lstart, loff_t lend);
2319 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
2320 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2321 extern void ext4_da_update_reserve_space(struct inode *inode,
2322 					int used, int quota_claim);
2323 
2324 /* indirect.c */
2325 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2326 				struct ext4_map_blocks *map, int flags);
2327 extern ssize_t ext4_ind_direct_IO(int rw, struct kiocb *iocb,
2328 				  const struct iovec *iov, loff_t offset,
2329 				  unsigned long nr_segs);
2330 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2331 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2332 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2333 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2334 				 ext4_lblk_t start, ext4_lblk_t end);
2335 
2336 /* ioctl.c */
2337 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2338 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2339 
2340 /* migrate.c */
2341 extern int ext4_ext_migrate(struct inode *);
2342 extern int ext4_ind_migrate(struct inode *inode);
2343 
2344 /* namei.c */
2345 extern int ext4_dirent_csum_verify(struct inode *inode,
2346 				   struct ext4_dir_entry *dirent);
2347 extern int ext4_orphan_add(handle_t *, struct inode *);
2348 extern int ext4_orphan_del(handle_t *, struct inode *);
2349 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2350 				__u32 start_minor_hash, __u32 *next_hash);
2351 extern int ext4_search_dir(struct buffer_head *bh,
2352 			   char *search_buf,
2353 			   int buf_size,
2354 			   struct inode *dir,
2355 			   struct ext4_filename *fname,
2356 			   const struct qstr *d_name,
2357 			   unsigned int offset,
2358 			   struct ext4_dir_entry_2 **res_dir);
2359 extern int ext4_generic_delete_entry(handle_t *handle,
2360 				     struct inode *dir,
2361 				     struct ext4_dir_entry_2 *de_del,
2362 				     struct buffer_head *bh,
2363 				     void *entry_buf,
2364 				     int buf_size,
2365 				     int csum_size);
2366 extern int ext4_empty_dir(struct inode *inode);
2367 
2368 /* resize.c */
2369 extern int ext4_group_add(struct super_block *sb,
2370 				struct ext4_new_group_data *input);
2371 extern int ext4_group_extend(struct super_block *sb,
2372 				struct ext4_super_block *es,
2373 				ext4_fsblk_t n_blocks_count);
2374 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2375 
2376 /* super.c */
2377 extern int ext4_calculate_overhead(struct super_block *sb);
2378 extern void ext4_superblock_csum_set(struct super_block *sb);
2379 extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2380 extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2381 extern void ext4_kvfree(void *ptr);
2382 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2383 				    ext4_group_t ngroup);
2384 extern const char *ext4_decode_error(struct super_block *sb, int errno,
2385 				     char nbuf[16]);
2386 
2387 extern __printf(4, 5)
2388 void __ext4_error(struct super_block *, const char *, unsigned int,
2389 		  const char *, ...);
2390 extern __printf(5, 6)
2391 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2392 		      const char *, ...);
2393 extern __printf(5, 6)
2394 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2395 		     const char *, ...);
2396 extern void __ext4_std_error(struct super_block *, const char *,
2397 			     unsigned int, int);
2398 extern __printf(4, 5)
2399 void __ext4_abort(struct super_block *, const char *, unsigned int,
2400 		  const char *, ...);
2401 extern __printf(4, 5)
2402 void __ext4_warning(struct super_block *, const char *, unsigned int,
2403 		    const char *, ...);
2404 extern __printf(3, 4)
2405 void __ext4_msg(struct super_block *, const char *, const char *, ...);
2406 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2407 			   const char *, unsigned int, const char *);
2408 extern __printf(7, 8)
2409 void __ext4_grp_locked_error(const char *, unsigned int,
2410 			     struct super_block *, ext4_group_t,
2411 			     unsigned long, ext4_fsblk_t,
2412 			     const char *, ...);
2413 
2414 #ifdef CONFIG_PRINTK
2415 
2416 #define ext4_error_inode(inode, func, line, block, fmt, ...)		\
2417 	__ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__)
2418 #define ext4_error_file(file, func, line, block, fmt, ...)		\
2419 	__ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2420 #define ext4_error(sb, fmt, ...)					\
2421 	__ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2422 #define ext4_abort(sb, fmt, ...)					\
2423 	__ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2424 #define ext4_warning(sb, fmt, ...)					\
2425 	__ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2426 #define ext4_msg(sb, level, fmt, ...)				\
2427 	__ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2428 #define dump_mmp_msg(sb, mmp, msg)					\
2429 	__dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2430 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...)		\
2431 	__ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2432 				fmt, ##__VA_ARGS__)
2433 
2434 #else
2435 
2436 #define ext4_error_inode(inode, func, line, block, fmt, ...)		\
2437 do {									\
2438 	no_printk(fmt, ##__VA_ARGS__);					\
2439 	__ext4_error_inode(inode, "", 0, block, " ");			\
2440 } while (0)
2441 #define ext4_error_file(file, func, line, block, fmt, ...)		\
2442 do {									\
2443 	no_printk(fmt, ##__VA_ARGS__);					\
2444 	__ext4_error_file(file, "", 0, block, " ");			\
2445 } while (0)
2446 #define ext4_error(sb, fmt, ...)					\
2447 do {									\
2448 	no_printk(fmt, ##__VA_ARGS__);					\
2449 	__ext4_error(sb, "", 0, " ");					\
2450 } while (0)
2451 #define ext4_abort(sb, fmt, ...)					\
2452 do {									\
2453 	no_printk(fmt, ##__VA_ARGS__);					\
2454 	__ext4_abort(sb, "", 0, " ");					\
2455 } while (0)
2456 #define ext4_warning(sb, fmt, ...)					\
2457 do {									\
2458 	no_printk(fmt, ##__VA_ARGS__);					\
2459 	__ext4_warning(sb, "", 0, " ");					\
2460 } while (0)
2461 #define ext4_msg(sb, level, fmt, ...)					\
2462 do {									\
2463 	no_printk(fmt, ##__VA_ARGS__);					\
2464 	__ext4_msg(sb, "", " ");					\
2465 } while (0)
2466 #define dump_mmp_msg(sb, mmp, msg)					\
2467 	__dump_mmp_msg(sb, mmp, "", 0, "")
2468 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...)		\
2469 do {									\
2470 	no_printk(fmt, ##__VA_ARGS__);				\
2471 	__ext4_grp_locked_error("", 0, sb, grp, ino, block, " ");	\
2472 } while (0)
2473 
2474 #endif
2475 
2476 extern void ext4_update_dynamic_rev(struct super_block *sb);
2477 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2478 					__u32 compat);
2479 extern int ext4_update_rocompat_feature(handle_t *handle,
2480 					struct super_block *sb,	__u32 rocompat);
2481 extern int ext4_update_incompat_feature(handle_t *handle,
2482 					struct super_block *sb,	__u32 incompat);
2483 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2484 				      struct ext4_group_desc *bg);
2485 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2486 				      struct ext4_group_desc *bg);
2487 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2488 				     struct ext4_group_desc *bg);
2489 extern __u32 ext4_free_group_clusters(struct super_block *sb,
2490 				      struct ext4_group_desc *bg);
2491 extern __u32 ext4_free_inodes_count(struct super_block *sb,
2492 				 struct ext4_group_desc *bg);
2493 extern __u32 ext4_used_dirs_count(struct super_block *sb,
2494 				struct ext4_group_desc *bg);
2495 extern __u32 ext4_itable_unused_count(struct super_block *sb,
2496 				   struct ext4_group_desc *bg);
2497 extern void ext4_block_bitmap_set(struct super_block *sb,
2498 				  struct ext4_group_desc *bg, ext4_fsblk_t blk);
2499 extern void ext4_inode_bitmap_set(struct super_block *sb,
2500 				  struct ext4_group_desc *bg, ext4_fsblk_t blk);
2501 extern void ext4_inode_table_set(struct super_block *sb,
2502 				 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2503 extern void ext4_free_group_clusters_set(struct super_block *sb,
2504 					 struct ext4_group_desc *bg,
2505 					 __u32 count);
2506 extern void ext4_free_inodes_set(struct super_block *sb,
2507 				struct ext4_group_desc *bg, __u32 count);
2508 extern void ext4_used_dirs_set(struct super_block *sb,
2509 				struct ext4_group_desc *bg, __u32 count);
2510 extern void ext4_itable_unused_set(struct super_block *sb,
2511 				   struct ext4_group_desc *bg, __u32 count);
2512 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2513 				       struct ext4_group_desc *gdp);
2514 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2515 				     struct ext4_group_desc *gdp);
2516 extern int ext4_register_li_request(struct super_block *sb,
2517 				    ext4_group_t first_not_zeroed);
2518 
ext4_has_group_desc_csum(struct super_block * sb)2519 static inline int ext4_has_group_desc_csum(struct super_block *sb)
2520 {
2521 	return EXT4_HAS_RO_COMPAT_FEATURE(sb,
2522 					  EXT4_FEATURE_RO_COMPAT_GDT_CSUM) ||
2523 	       (EXT4_SB(sb)->s_chksum_driver != NULL);
2524 }
2525 
ext4_has_metadata_csum(struct super_block * sb)2526 static inline int ext4_has_metadata_csum(struct super_block *sb)
2527 {
2528 	WARN_ON_ONCE(EXT4_HAS_RO_COMPAT_FEATURE(sb,
2529 			EXT4_FEATURE_RO_COMPAT_METADATA_CSUM) &&
2530 		     !EXT4_SB(sb)->s_chksum_driver);
2531 
2532 	return (EXT4_SB(sb)->s_chksum_driver != NULL);
2533 }
ext4_blocks_count(struct ext4_super_block * es)2534 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2535 {
2536 	return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2537 		le32_to_cpu(es->s_blocks_count_lo);
2538 }
2539 
ext4_r_blocks_count(struct ext4_super_block * es)2540 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2541 {
2542 	return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2543 		le32_to_cpu(es->s_r_blocks_count_lo);
2544 }
2545 
ext4_free_blocks_count(struct ext4_super_block * es)2546 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2547 {
2548 	return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2549 		le32_to_cpu(es->s_free_blocks_count_lo);
2550 }
2551 
ext4_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)2552 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2553 					 ext4_fsblk_t blk)
2554 {
2555 	es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2556 	es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2557 }
2558 
ext4_free_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)2559 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2560 					      ext4_fsblk_t blk)
2561 {
2562 	es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2563 	es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2564 }
2565 
ext4_r_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)2566 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2567 					   ext4_fsblk_t blk)
2568 {
2569 	es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2570 	es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2571 }
2572 
ext4_isize(struct ext4_inode * raw_inode)2573 static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
2574 {
2575 	if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2576 		return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2577 			le32_to_cpu(raw_inode->i_size_lo);
2578 	else
2579 		return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2580 }
2581 
ext4_isize_set(struct ext4_inode * raw_inode,loff_t i_size)2582 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2583 {
2584 	raw_inode->i_size_lo = cpu_to_le32(i_size);
2585 	raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2586 }
2587 
2588 static inline
ext4_get_group_info(struct super_block * sb,ext4_group_t group)2589 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2590 					    ext4_group_t group)
2591 {
2592 	 struct ext4_group_info ***grp_info;
2593 	 long indexv, indexh;
2594 	 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
2595 	 grp_info = EXT4_SB(sb)->s_group_info;
2596 	 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2597 	 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2598 	 return grp_info[indexv][indexh];
2599 }
2600 
2601 /*
2602  * Reading s_groups_count requires using smp_rmb() afterwards.  See
2603  * the locking protocol documented in the comments of ext4_group_add()
2604  * in resize.c
2605  */
ext4_get_groups_count(struct super_block * sb)2606 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2607 {
2608 	ext4_group_t	ngroups = EXT4_SB(sb)->s_groups_count;
2609 
2610 	smp_rmb();
2611 	return ngroups;
2612 }
2613 
ext4_flex_group(struct ext4_sb_info * sbi,ext4_group_t block_group)2614 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2615 					     ext4_group_t block_group)
2616 {
2617 	return block_group >> sbi->s_log_groups_per_flex;
2618 }
2619 
ext4_flex_bg_size(struct ext4_sb_info * sbi)2620 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2621 {
2622 	return 1 << sbi->s_log_groups_per_flex;
2623 }
2624 
2625 #define ext4_std_error(sb, errno)				\
2626 do {								\
2627 	if ((errno))						\
2628 		__ext4_std_error((sb), __func__, __LINE__, (errno));	\
2629 } while (0)
2630 
2631 #ifdef CONFIG_SMP
2632 /* Each CPU can accumulate percpu_counter_batch clusters in their local
2633  * counters. So we need to make sure we have free clusters more
2634  * than percpu_counter_batch  * nr_cpu_ids. Also add a window of 4 times.
2635  */
2636 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2637 #else
2638 #define EXT4_FREECLUSTERS_WATERMARK 0
2639 #endif
2640 
2641 /* Update i_disksize. Requires i_mutex to avoid races with truncate */
ext4_update_i_disksize(struct inode * inode,loff_t newsize)2642 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2643 {
2644 	WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
2645 		     !mutex_is_locked(&inode->i_mutex));
2646 	down_write(&EXT4_I(inode)->i_data_sem);
2647 	if (newsize > EXT4_I(inode)->i_disksize)
2648 		EXT4_I(inode)->i_disksize = newsize;
2649 	up_write(&EXT4_I(inode)->i_data_sem);
2650 }
2651 
2652 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
ext4_update_inode_size(struct inode * inode,loff_t newsize)2653 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
2654 {
2655 	int changed = 0;
2656 
2657 	if (newsize > inode->i_size) {
2658 		i_size_write(inode, newsize);
2659 		changed = 1;
2660 	}
2661 	if (newsize > EXT4_I(inode)->i_disksize) {
2662 		ext4_update_i_disksize(inode, newsize);
2663 		changed |= 2;
2664 	}
2665 	return changed;
2666 }
2667 
2668 struct ext4_group_info {
2669 	unsigned long   bb_state;
2670 	struct rb_root  bb_free_root;
2671 	ext4_grpblk_t	bb_first_free;	/* first free block */
2672 	ext4_grpblk_t	bb_free;	/* total free blocks */
2673 	ext4_grpblk_t	bb_fragments;	/* nr of freespace fragments */
2674 	ext4_grpblk_t	bb_largest_free_order;/* order of largest frag in BG */
2675 	struct          list_head bb_prealloc_list;
2676 #ifdef DOUBLE_CHECK
2677 	void            *bb_bitmap;
2678 #endif
2679 	struct rw_semaphore alloc_sem;
2680 	ext4_grpblk_t	bb_counters[];	/* Nr of free power-of-two-block
2681 					 * regions, index is order.
2682 					 * bb_counters[3] = 5 means
2683 					 * 5 free 8-block regions. */
2684 };
2685 
2686 #define EXT4_GROUP_INFO_NEED_INIT_BIT		0
2687 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT		1
2688 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT	2
2689 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT	3
2690 
2691 #define EXT4_MB_GRP_NEED_INIT(grp)	\
2692 	(test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2693 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp)	\
2694 	(test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2695 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp)	\
2696 	(test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2697 
2698 #define EXT4_MB_GRP_WAS_TRIMMED(grp)	\
2699 	(test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2700 #define EXT4_MB_GRP_SET_TRIMMED(grp)	\
2701 	(set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2702 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp)	\
2703 	(clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2704 
2705 #define EXT4_MAX_CONTENTION		8
2706 #define EXT4_CONTENTION_THRESHOLD	2
2707 
ext4_group_lock_ptr(struct super_block * sb,ext4_group_t group)2708 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2709 					      ext4_group_t group)
2710 {
2711 	return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2712 }
2713 
2714 /*
2715  * Returns true if the filesystem is busy enough that attempts to
2716  * access the block group locks has run into contention.
2717  */
ext4_fs_is_busy(struct ext4_sb_info * sbi)2718 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2719 {
2720 	return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2721 }
2722 
ext4_lock_group(struct super_block * sb,ext4_group_t group)2723 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2724 {
2725 	spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2726 	if (spin_trylock(lock))
2727 		/*
2728 		 * We're able to grab the lock right away, so drop the
2729 		 * lock contention counter.
2730 		 */
2731 		atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2732 	else {
2733 		/*
2734 		 * The lock is busy, so bump the contention counter,
2735 		 * and then wait on the spin lock.
2736 		 */
2737 		atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2738 				  EXT4_MAX_CONTENTION);
2739 		spin_lock(lock);
2740 	}
2741 }
2742 
ext4_unlock_group(struct super_block * sb,ext4_group_t group)2743 static inline void ext4_unlock_group(struct super_block *sb,
2744 					ext4_group_t group)
2745 {
2746 	spin_unlock(ext4_group_lock_ptr(sb, group));
2747 }
2748 
2749 /*
2750  * Block validity checking
2751  */
2752 #define ext4_check_indirect_blockref(inode, bh)				\
2753 	ext4_check_blockref(__func__, __LINE__, inode,			\
2754 			    (__le32 *)(bh)->b_data,			\
2755 			    EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2756 
2757 #define ext4_ind_check_inode(inode)					\
2758 	ext4_check_blockref(__func__, __LINE__, inode,			\
2759 			    EXT4_I(inode)->i_data,			\
2760 			    EXT4_NDIR_BLOCKS)
2761 
2762 /*
2763  * Inodes and files operations
2764  */
2765 
2766 /* dir.c */
2767 extern const struct file_operations ext4_dir_operations;
2768 
2769 /* file.c */
2770 extern const struct inode_operations ext4_file_inode_operations;
2771 extern const struct file_operations ext4_file_operations;
2772 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
2773 
2774 /* inline.c */
2775 extern int ext4_get_max_inline_size(struct inode *inode);
2776 extern int ext4_find_inline_data_nolock(struct inode *inode);
2777 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
2778 				 unsigned int len);
2779 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
2780 
2781 extern int ext4_readpage_inline(struct inode *inode, struct page *page);
2782 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
2783 					 struct inode *inode,
2784 					 loff_t pos, unsigned len,
2785 					 unsigned flags,
2786 					 struct page **pagep);
2787 extern int ext4_write_inline_data_end(struct inode *inode,
2788 				      loff_t pos, unsigned len,
2789 				      unsigned copied,
2790 				      struct page *page);
2791 extern struct buffer_head *
2792 ext4_journalled_write_inline_data(struct inode *inode,
2793 				  unsigned len,
2794 				  struct page *page);
2795 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
2796 					   struct inode *inode,
2797 					   loff_t pos, unsigned len,
2798 					   unsigned flags,
2799 					   struct page **pagep,
2800 					   void **fsdata);
2801 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
2802 					 unsigned len, unsigned copied,
2803 					 struct page *page);
2804 extern int ext4_try_add_inline_entry(handle_t *handle,
2805 				     struct ext4_filename *fname,
2806 				     struct dentry *dentry,
2807 				     struct inode *inode);
2808 extern int ext4_try_create_inline_dir(handle_t *handle,
2809 				      struct inode *parent,
2810 				      struct inode *inode);
2811 extern int ext4_read_inline_dir(struct file *filp,
2812 				struct dir_context *ctx,
2813 				int *has_inline_data);
2814 extern int htree_inlinedir_to_tree(struct file *dir_file,
2815 				   struct inode *dir, ext4_lblk_t block,
2816 				   struct dx_hash_info *hinfo,
2817 				   __u32 start_hash, __u32 start_minor_hash,
2818 				   int *has_inline_data);
2819 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
2820 					struct ext4_filename *fname,
2821 					const struct qstr *d_name,
2822 					struct ext4_dir_entry_2 **res_dir,
2823 					int *has_inline_data);
2824 extern int ext4_delete_inline_entry(handle_t *handle,
2825 				    struct inode *dir,
2826 				    struct ext4_dir_entry_2 *de_del,
2827 				    struct buffer_head *bh,
2828 				    int *has_inline_data);
2829 extern int empty_inline_dir(struct inode *dir, int *has_inline_data);
2830 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
2831 					struct ext4_dir_entry_2 **parent_de,
2832 					int *retval);
2833 extern int ext4_inline_data_fiemap(struct inode *inode,
2834 				   struct fiemap_extent_info *fieinfo,
2835 				   int *has_inline);
2836 extern int ext4_try_to_evict_inline_data(handle_t *handle,
2837 					 struct inode *inode,
2838 					 int needed);
2839 extern void ext4_inline_data_truncate(struct inode *inode, int *has_inline);
2840 
2841 extern int ext4_convert_inline_data(struct inode *inode);
2842 
ext4_has_inline_data(struct inode * inode)2843 static inline int ext4_has_inline_data(struct inode *inode)
2844 {
2845 	return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
2846 	       EXT4_I(inode)->i_inline_off;
2847 }
2848 
2849 /* namei.c */
2850 extern const struct inode_operations ext4_dir_inode_operations;
2851 extern const struct inode_operations ext4_special_inode_operations;
2852 extern struct dentry *ext4_get_parent(struct dentry *child);
2853 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
2854 				 struct ext4_dir_entry_2 *de,
2855 				 int blocksize, int csum_size,
2856 				 unsigned int parent_ino, int dotdot_real_len);
2857 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t,
2858 				   unsigned int blocksize);
2859 extern int ext4_handle_dirty_dirent_node(handle_t *handle,
2860 					 struct inode *inode,
2861 					 struct buffer_head *bh);
2862 #define S_SHIFT 12
2863 static unsigned char ext4_type_by_mode[S_IFMT >> S_SHIFT] = {
2864 	[S_IFREG >> S_SHIFT]	= EXT4_FT_REG_FILE,
2865 	[S_IFDIR >> S_SHIFT]	= EXT4_FT_DIR,
2866 	[S_IFCHR >> S_SHIFT]	= EXT4_FT_CHRDEV,
2867 	[S_IFBLK >> S_SHIFT]	= EXT4_FT_BLKDEV,
2868 	[S_IFIFO >> S_SHIFT]	= EXT4_FT_FIFO,
2869 	[S_IFSOCK >> S_SHIFT]	= EXT4_FT_SOCK,
2870 	[S_IFLNK >> S_SHIFT]	= EXT4_FT_SYMLINK,
2871 };
2872 
ext4_set_de_type(struct super_block * sb,struct ext4_dir_entry_2 * de,umode_t mode)2873 static inline void ext4_set_de_type(struct super_block *sb,
2874 				struct ext4_dir_entry_2 *de,
2875 				umode_t mode) {
2876 	if (EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE))
2877 		de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
2878 }
2879 
2880 /* readpages.c */
2881 extern int ext4_mpage_readpages(struct address_space *mapping,
2882 				struct list_head *pages, struct page *page,
2883 				unsigned nr_pages);
2884 
2885 /* symlink.c */
2886 extern const struct inode_operations ext4_symlink_inode_operations;
2887 extern const struct inode_operations ext4_fast_symlink_inode_operations;
2888 
2889 /* block_validity */
2890 extern void ext4_release_system_zone(struct super_block *sb);
2891 extern int ext4_setup_system_zone(struct super_block *sb);
2892 extern int __init ext4_init_system_zone(void);
2893 extern void ext4_exit_system_zone(void);
2894 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
2895 				 ext4_fsblk_t start_blk,
2896 				 unsigned int count);
2897 extern int ext4_check_blockref(const char *, unsigned int,
2898 			       struct inode *, __le32 *, unsigned int);
2899 
2900 /* extents.c */
2901 struct ext4_ext_path;
2902 struct ext4_extent;
2903 
2904 /*
2905  * Maximum number of logical blocks in a file; ext4_extent's ee_block is
2906  * __le32.
2907  */
2908 #define EXT_MAX_BLOCKS	0xffffffff
2909 
2910 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
2911 extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
2912 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
2913 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
2914 			       struct ext4_map_blocks *map, int flags);
2915 extern void ext4_ext_truncate(handle_t *, struct inode *);
2916 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
2917 				 ext4_lblk_t end);
2918 extern void ext4_ext_init(struct super_block *);
2919 extern void ext4_ext_release(struct super_block *);
2920 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
2921 			  loff_t len);
2922 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
2923 					  loff_t offset, ssize_t len);
2924 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
2925 			   struct ext4_map_blocks *map, int flags);
2926 extern int ext4_ext_calc_metadata_amount(struct inode *inode,
2927 					 ext4_lblk_t lblocks);
2928 extern int ext4_extent_tree_init(handle_t *, struct inode *);
2929 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
2930 						   int num,
2931 						   struct ext4_ext_path *path);
2932 extern int ext4_can_extents_be_merged(struct inode *inode,
2933 				      struct ext4_extent *ex1,
2934 				      struct ext4_extent *ex2);
2935 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
2936 				  struct ext4_ext_path **,
2937 				  struct ext4_extent *, int);
2938 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
2939 					      struct ext4_ext_path **,
2940 					      int flags);
2941 extern void ext4_ext_drop_refs(struct ext4_ext_path *);
2942 extern int ext4_ext_check_inode(struct inode *inode);
2943 extern int ext4_find_delalloc_range(struct inode *inode,
2944 				    ext4_lblk_t lblk_start,
2945 				    ext4_lblk_t lblk_end);
2946 extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk);
2947 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
2948 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
2949 			__u64 start, __u64 len);
2950 extern int ext4_ext_precache(struct inode *inode);
2951 extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len);
2952 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
2953 				struct inode *inode2, ext4_lblk_t lblk1,
2954 			     ext4_lblk_t lblk2,  ext4_lblk_t count,
2955 			     int mark_unwritten,int *err);
2956 
2957 /* move_extent.c */
2958 extern void ext4_double_down_write_data_sem(struct inode *first,
2959 					    struct inode *second);
2960 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
2961 					  struct inode *donor_inode);
2962 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
2963 			     __u64 start_orig, __u64 start_donor,
2964 			     __u64 len, __u64 *moved_len);
2965 
2966 /* page-io.c */
2967 extern int __init ext4_init_pageio(void);
2968 extern void ext4_exit_pageio(void);
2969 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
2970 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
2971 extern int ext4_put_io_end(ext4_io_end_t *io_end);
2972 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
2973 extern void ext4_io_submit_init(struct ext4_io_submit *io,
2974 				struct writeback_control *wbc);
2975 extern void ext4_end_io_rsv_work(struct work_struct *work);
2976 extern void ext4_io_submit(struct ext4_io_submit *io);
2977 extern int ext4_bio_write_page(struct ext4_io_submit *io,
2978 			       struct page *page,
2979 			       int len,
2980 			       struct writeback_control *wbc,
2981 			       bool keep_towrite);
2982 
2983 /* mmp.c */
2984 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
2985 
2986 /*
2987  * Note that these flags will never ever appear in a buffer_head's state flag.
2988  * See EXT4_MAP_... to see where this is used.
2989  */
2990 enum ext4_state_bits {
2991 	BH_AllocFromCluster	/* allocated blocks were part of already
2992 				 * allocated cluster. */
2993 	= BH_JBDPrivateStart
2994 };
2995 
2996 /*
2997  * Add new method to test whether block and inode bitmaps are properly
2998  * initialized. With uninit_bg reading the block from disk is not enough
2999  * to mark the bitmap uptodate. We need to also zero-out the bitmap
3000  */
3001 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3002 
bitmap_uptodate(struct buffer_head * bh)3003 static inline int bitmap_uptodate(struct buffer_head *bh)
3004 {
3005 	return (buffer_uptodate(bh) &&
3006 			test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3007 }
set_bitmap_uptodate(struct buffer_head * bh)3008 static inline void set_bitmap_uptodate(struct buffer_head *bh)
3009 {
3010 	set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3011 }
3012 
3013 /*
3014  * Disable DIO read nolock optimization, so new dioreaders will be forced
3015  * to grab i_mutex
3016  */
ext4_inode_block_unlocked_dio(struct inode * inode)3017 static inline void ext4_inode_block_unlocked_dio(struct inode *inode)
3018 {
3019 	ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
3020 	smp_mb();
3021 }
ext4_inode_resume_unlocked_dio(struct inode * inode)3022 static inline void ext4_inode_resume_unlocked_dio(struct inode *inode)
3023 {
3024 	smp_mb();
3025 	ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
3026 }
3027 
3028 #define in_range(b, first, len)	((b) >= (first) && (b) <= (first) + (len) - 1)
3029 
3030 /* For ioend & aio unwritten conversion wait queues */
3031 #define EXT4_WQ_HASH_SZ		37
3032 #define ext4_ioend_wq(v)   (&ext4__ioend_wq[((unsigned long)(v)) %\
3033 					    EXT4_WQ_HASH_SZ])
3034 #define ext4_aio_mutex(v)  (&ext4__aio_mutex[((unsigned long)(v)) %\
3035 					     EXT4_WQ_HASH_SZ])
3036 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3037 extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ];
3038 
3039 #define EXT4_RESIZING	0
3040 extern int ext4_resize_begin(struct super_block *sb);
3041 extern void ext4_resize_end(struct super_block *sb);
3042 
3043 #endif	/* __KERNEL__ */
3044 
3045 #endif	/* _EXT4_H */
3046