xref: /openbmc/linux/fs/ext4/ext4.h (revision feb0ab32)
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 <crypto/hash.h>
33 #ifdef __KERNEL__
34 #include <linux/compat.h>
35 #endif
36 
37 /*
38  * The fourth extended filesystem constants/structures
39  */
40 
41 /*
42  * Define EXT4FS_DEBUG to produce debug messages
43  */
44 #undef EXT4FS_DEBUG
45 
46 /*
47  * Debug code
48  */
49 #ifdef EXT4FS_DEBUG
50 #define ext4_debug(f, a...)						\
51 	do {								\
52 		printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:",	\
53 			__FILE__, __LINE__, __func__);			\
54 		printk(KERN_DEBUG f, ## a);				\
55 	} while (0)
56 #else
57 #define ext4_debug(fmt, ...)	no_printk(fmt, ##__VA_ARGS__)
58 #endif
59 
60 #define EXT4_ERROR_INODE(inode, fmt, a...) \
61 	ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
62 
63 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...)			\
64 	ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
65 
66 #define EXT4_ERROR_FILE(file, block, fmt, a...)				\
67 	ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
68 
69 /* data type for block offset of block group */
70 typedef int ext4_grpblk_t;
71 
72 /* data type for filesystem-wide blocks number */
73 typedef unsigned long long ext4_fsblk_t;
74 
75 /* data type for file logical block number */
76 typedef __u32 ext4_lblk_t;
77 
78 /* data type for block group number */
79 typedef unsigned int ext4_group_t;
80 
81 /*
82  * Flags used in mballoc's allocation_context flags field.
83  *
84  * Also used to show what's going on for debugging purposes when the
85  * flag field is exported via the traceport interface
86  */
87 
88 /* prefer goal again. length */
89 #define EXT4_MB_HINT_MERGE		0x0001
90 /* blocks already reserved */
91 #define EXT4_MB_HINT_RESERVED		0x0002
92 /* metadata is being allocated */
93 #define EXT4_MB_HINT_METADATA		0x0004
94 /* first blocks in the file */
95 #define EXT4_MB_HINT_FIRST		0x0008
96 /* search for the best chunk */
97 #define EXT4_MB_HINT_BEST		0x0010
98 /* data is being allocated */
99 #define EXT4_MB_HINT_DATA		0x0020
100 /* don't preallocate (for tails) */
101 #define EXT4_MB_HINT_NOPREALLOC		0x0040
102 /* allocate for locality group */
103 #define EXT4_MB_HINT_GROUP_ALLOC	0x0080
104 /* allocate goal blocks or none */
105 #define EXT4_MB_HINT_GOAL_ONLY		0x0100
106 /* goal is meaningful */
107 #define EXT4_MB_HINT_TRY_GOAL		0x0200
108 /* blocks already pre-reserved by delayed allocation */
109 #define EXT4_MB_DELALLOC_RESERVED	0x0400
110 /* We are doing stream allocation */
111 #define EXT4_MB_STREAM_ALLOC		0x0800
112 /* Use reserved root blocks if needed */
113 #define EXT4_MB_USE_ROOT_BLOCKS		0x1000
114 
115 struct ext4_allocation_request {
116 	/* target inode for block we're allocating */
117 	struct inode *inode;
118 	/* how many blocks we want to allocate */
119 	unsigned int len;
120 	/* logical block in target inode */
121 	ext4_lblk_t logical;
122 	/* the closest logical allocated block to the left */
123 	ext4_lblk_t lleft;
124 	/* the closest logical allocated block to the right */
125 	ext4_lblk_t lright;
126 	/* phys. target (a hint) */
127 	ext4_fsblk_t goal;
128 	/* phys. block for the closest logical allocated block to the left */
129 	ext4_fsblk_t pleft;
130 	/* phys. block for the closest logical allocated block to the right */
131 	ext4_fsblk_t pright;
132 	/* flags. see above EXT4_MB_HINT_* */
133 	unsigned int flags;
134 };
135 
136 /*
137  * Logical to physical block mapping, used by ext4_map_blocks()
138  *
139  * This structure is used to pass requests into ext4_map_blocks() as
140  * well as to store the information returned by ext4_map_blocks().  It
141  * takes less room on the stack than a struct buffer_head.
142  */
143 #define EXT4_MAP_NEW		(1 << BH_New)
144 #define EXT4_MAP_MAPPED		(1 << BH_Mapped)
145 #define EXT4_MAP_UNWRITTEN	(1 << BH_Unwritten)
146 #define EXT4_MAP_BOUNDARY	(1 << BH_Boundary)
147 #define EXT4_MAP_UNINIT		(1 << BH_Uninit)
148 /* Sometimes (in the bigalloc case, from ext4_da_get_block_prep) the caller of
149  * ext4_map_blocks wants to know whether or not the underlying cluster has
150  * already been accounted for. EXT4_MAP_FROM_CLUSTER conveys to the caller that
151  * the requested mapping was from previously mapped (or delayed allocated)
152  * cluster. We use BH_AllocFromCluster only for this flag. BH_AllocFromCluster
153  * should never appear on buffer_head's state flags.
154  */
155 #define EXT4_MAP_FROM_CLUSTER	(1 << BH_AllocFromCluster)
156 #define EXT4_MAP_FLAGS		(EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
157 				 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\
158 				 EXT4_MAP_UNINIT | EXT4_MAP_FROM_CLUSTER)
159 
160 struct ext4_map_blocks {
161 	ext4_fsblk_t m_pblk;
162 	ext4_lblk_t m_lblk;
163 	unsigned int m_len;
164 	unsigned int m_flags;
165 };
166 
167 /*
168  * For delayed allocation tracking
169  */
170 struct mpage_da_data {
171 	struct inode *inode;
172 	sector_t b_blocknr;		/* start block number of extent */
173 	size_t b_size;			/* size of extent */
174 	unsigned long b_state;		/* state of the extent */
175 	unsigned long first_page, next_page;	/* extent of pages */
176 	struct writeback_control *wbc;
177 	int io_done;
178 	int pages_written;
179 	int retval;
180 };
181 
182 /*
183  * Flags for ext4_io_end->flags
184  */
185 #define	EXT4_IO_END_UNWRITTEN	0x0001
186 #define EXT4_IO_END_ERROR	0x0002
187 #define EXT4_IO_END_QUEUED	0x0004
188 #define EXT4_IO_END_DIRECT	0x0008
189 #define EXT4_IO_END_IN_FSYNC	0x0010
190 
191 struct ext4_io_page {
192 	struct page	*p_page;
193 	atomic_t	p_count;
194 };
195 
196 #define MAX_IO_PAGES 128
197 
198 /*
199  * For converting uninitialized extents on a work queue.
200  *
201  * 'page' is only used from the writepage() path; 'pages' is only used for
202  * buffered writes; they are used to keep page references until conversion
203  * takes place.  For AIO/DIO, neither field is filled in.
204  */
205 typedef struct ext4_io_end {
206 	struct list_head	list;		/* per-file finished IO list */
207 	struct inode		*inode;		/* file being written to */
208 	unsigned int		flag;		/* unwritten or not */
209 	struct page		*page;		/* for writepage() path */
210 	loff_t			offset;		/* offset in the file */
211 	ssize_t			size;		/* size of the extent */
212 	struct work_struct	work;		/* data work queue */
213 	struct kiocb		*iocb;		/* iocb struct for AIO */
214 	int			result;		/* error value for AIO */
215 	int			num_io_pages;   /* for writepages() */
216 	struct ext4_io_page	*pages[MAX_IO_PAGES]; /* for writepages() */
217 } ext4_io_end_t;
218 
219 struct ext4_io_submit {
220 	int			io_op;
221 	struct bio		*io_bio;
222 	ext4_io_end_t		*io_end;
223 	struct ext4_io_page	*io_page;
224 	sector_t		io_next_block;
225 };
226 
227 /*
228  * Special inodes numbers
229  */
230 #define	EXT4_BAD_INO		 1	/* Bad blocks inode */
231 #define EXT4_ROOT_INO		 2	/* Root inode */
232 #define EXT4_USR_QUOTA_INO	 3	/* User quota inode */
233 #define EXT4_GRP_QUOTA_INO	 4	/* Group quota inode */
234 #define EXT4_BOOT_LOADER_INO	 5	/* Boot loader inode */
235 #define EXT4_UNDEL_DIR_INO	 6	/* Undelete directory inode */
236 #define EXT4_RESIZE_INO		 7	/* Reserved group descriptors inode */
237 #define EXT4_JOURNAL_INO	 8	/* Journal inode */
238 
239 /* First non-reserved inode for old ext4 filesystems */
240 #define EXT4_GOOD_OLD_FIRST_INO	11
241 
242 /*
243  * Maximal count of links to a file
244  */
245 #define EXT4_LINK_MAX		65000
246 
247 /*
248  * Macro-instructions used to manage several block sizes
249  */
250 #define EXT4_MIN_BLOCK_SIZE		1024
251 #define	EXT4_MAX_BLOCK_SIZE		65536
252 #define EXT4_MIN_BLOCK_LOG_SIZE		10
253 #define EXT4_MAX_BLOCK_LOG_SIZE		16
254 #ifdef __KERNEL__
255 # define EXT4_BLOCK_SIZE(s)		((s)->s_blocksize)
256 #else
257 # define EXT4_BLOCK_SIZE(s)		(EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
258 #endif
259 #define	EXT4_ADDR_PER_BLOCK(s)		(EXT4_BLOCK_SIZE(s) / sizeof(__u32))
260 #define EXT4_CLUSTER_SIZE(s)		(EXT4_BLOCK_SIZE(s) << \
261 					 EXT4_SB(s)->s_cluster_bits)
262 #ifdef __KERNEL__
263 # define EXT4_BLOCK_SIZE_BITS(s)	((s)->s_blocksize_bits)
264 # define EXT4_CLUSTER_BITS(s)		(EXT4_SB(s)->s_cluster_bits)
265 #else
266 # define EXT4_BLOCK_SIZE_BITS(s)	((s)->s_log_block_size + 10)
267 #endif
268 #ifdef __KERNEL__
269 #define	EXT4_ADDR_PER_BLOCK_BITS(s)	(EXT4_SB(s)->s_addr_per_block_bits)
270 #define EXT4_INODE_SIZE(s)		(EXT4_SB(s)->s_inode_size)
271 #define EXT4_FIRST_INO(s)		(EXT4_SB(s)->s_first_ino)
272 #else
273 #define EXT4_INODE_SIZE(s)	(((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
274 				 EXT4_GOOD_OLD_INODE_SIZE : \
275 				 (s)->s_inode_size)
276 #define EXT4_FIRST_INO(s)	(((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
277 				 EXT4_GOOD_OLD_FIRST_INO : \
278 				 (s)->s_first_ino)
279 #endif
280 #define EXT4_BLOCK_ALIGN(size, blkbits)		ALIGN((size), (1 << (blkbits)))
281 
282 /* Translate a block number to a cluster number */
283 #define EXT4_B2C(sbi, blk)	((blk) >> (sbi)->s_cluster_bits)
284 /* Translate a cluster number to a block number */
285 #define EXT4_C2B(sbi, cluster)	((cluster) << (sbi)->s_cluster_bits)
286 /* Translate # of blks to # of clusters */
287 #define EXT4_NUM_B2C(sbi, blks)	(((blks) + (sbi)->s_cluster_ratio - 1) >> \
288 				 (sbi)->s_cluster_bits)
289 
290 /*
291  * Structure of a blocks group descriptor
292  */
293 struct ext4_group_desc
294 {
295 	__le32	bg_block_bitmap_lo;	/* Blocks bitmap block */
296 	__le32	bg_inode_bitmap_lo;	/* Inodes bitmap block */
297 	__le32	bg_inode_table_lo;	/* Inodes table block */
298 	__le16	bg_free_blocks_count_lo;/* Free blocks count */
299 	__le16	bg_free_inodes_count_lo;/* Free inodes count */
300 	__le16	bg_used_dirs_count_lo;	/* Directories count */
301 	__le16	bg_flags;		/* EXT4_BG_flags (INODE_UNINIT, etc) */
302 	__le32  bg_exclude_bitmap_lo;   /* Exclude bitmap for snapshots */
303 	__le16  bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
304 	__le16  bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
305 	__le16  bg_itable_unused_lo;	/* Unused inodes count */
306 	__le16  bg_checksum;		/* crc16(sb_uuid+group+desc) */
307 	__le32	bg_block_bitmap_hi;	/* Blocks bitmap block MSB */
308 	__le32	bg_inode_bitmap_hi;	/* Inodes bitmap block MSB */
309 	__le32	bg_inode_table_hi;	/* Inodes table block MSB */
310 	__le16	bg_free_blocks_count_hi;/* Free blocks count MSB */
311 	__le16	bg_free_inodes_count_hi;/* Free inodes count MSB */
312 	__le16	bg_used_dirs_count_hi;	/* Directories count MSB */
313 	__le16  bg_itable_unused_hi;    /* Unused inodes count MSB */
314 	__le32  bg_exclude_bitmap_hi;   /* Exclude bitmap block MSB */
315 	__le16  bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
316 	__le16  bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
317 	__u32   bg_reserved;
318 };
319 
320 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END	\
321 	(offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
322 	 sizeof(__le16))
323 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END	\
324 	(offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
325 	 sizeof(__le16))
326 
327 /*
328  * Structure of a flex block group info
329  */
330 
331 struct flex_groups {
332 	atomic_t free_inodes;
333 	atomic_t free_clusters;
334 	atomic_t used_dirs;
335 };
336 
337 #define EXT4_BG_INODE_UNINIT	0x0001 /* Inode table/bitmap not in use */
338 #define EXT4_BG_BLOCK_UNINIT	0x0002 /* Block bitmap not in use */
339 #define EXT4_BG_INODE_ZEROED	0x0004 /* On-disk itable initialized to zero */
340 
341 /*
342  * Macro-instructions used to manage group descriptors
343  */
344 #define EXT4_MIN_DESC_SIZE		32
345 #define EXT4_MIN_DESC_SIZE_64BIT	64
346 #define	EXT4_MAX_DESC_SIZE		EXT4_MIN_BLOCK_SIZE
347 #define EXT4_DESC_SIZE(s)		(EXT4_SB(s)->s_desc_size)
348 #ifdef __KERNEL__
349 # define EXT4_BLOCKS_PER_GROUP(s)	(EXT4_SB(s)->s_blocks_per_group)
350 # define EXT4_CLUSTERS_PER_GROUP(s)	(EXT4_SB(s)->s_clusters_per_group)
351 # define EXT4_DESC_PER_BLOCK(s)		(EXT4_SB(s)->s_desc_per_block)
352 # define EXT4_INODES_PER_GROUP(s)	(EXT4_SB(s)->s_inodes_per_group)
353 # define EXT4_DESC_PER_BLOCK_BITS(s)	(EXT4_SB(s)->s_desc_per_block_bits)
354 #else
355 # define EXT4_BLOCKS_PER_GROUP(s)	((s)->s_blocks_per_group)
356 # define EXT4_DESC_PER_BLOCK(s)		(EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
357 # define EXT4_INODES_PER_GROUP(s)	((s)->s_inodes_per_group)
358 #endif
359 
360 /*
361  * Constants relative to the data blocks
362  */
363 #define	EXT4_NDIR_BLOCKS		12
364 #define	EXT4_IND_BLOCK			EXT4_NDIR_BLOCKS
365 #define	EXT4_DIND_BLOCK			(EXT4_IND_BLOCK + 1)
366 #define	EXT4_TIND_BLOCK			(EXT4_DIND_BLOCK + 1)
367 #define	EXT4_N_BLOCKS			(EXT4_TIND_BLOCK + 1)
368 
369 /*
370  * Inode flags
371  */
372 #define	EXT4_SECRM_FL			0x00000001 /* Secure deletion */
373 #define	EXT4_UNRM_FL			0x00000002 /* Undelete */
374 #define	EXT4_COMPR_FL			0x00000004 /* Compress file */
375 #define EXT4_SYNC_FL			0x00000008 /* Synchronous updates */
376 #define EXT4_IMMUTABLE_FL		0x00000010 /* Immutable file */
377 #define EXT4_APPEND_FL			0x00000020 /* writes to file may only append */
378 #define EXT4_NODUMP_FL			0x00000040 /* do not dump file */
379 #define EXT4_NOATIME_FL			0x00000080 /* do not update atime */
380 /* Reserved for compression usage... */
381 #define EXT4_DIRTY_FL			0x00000100
382 #define EXT4_COMPRBLK_FL		0x00000200 /* One or more compressed clusters */
383 #define EXT4_NOCOMPR_FL			0x00000400 /* Don't compress */
384 #define EXT4_ECOMPR_FL			0x00000800 /* Compression error */
385 /* End compression flags --- maybe not all used */
386 #define EXT4_INDEX_FL			0x00001000 /* hash-indexed directory */
387 #define EXT4_IMAGIC_FL			0x00002000 /* AFS directory */
388 #define EXT4_JOURNAL_DATA_FL		0x00004000 /* file data should be journaled */
389 #define EXT4_NOTAIL_FL			0x00008000 /* file tail should not be merged */
390 #define EXT4_DIRSYNC_FL			0x00010000 /* dirsync behaviour (directories only) */
391 #define EXT4_TOPDIR_FL			0x00020000 /* Top of directory hierarchies*/
392 #define EXT4_HUGE_FILE_FL               0x00040000 /* Set to each huge file */
393 #define EXT4_EXTENTS_FL			0x00080000 /* Inode uses extents */
394 #define EXT4_EA_INODE_FL	        0x00200000 /* Inode used for large EA */
395 #define EXT4_EOFBLOCKS_FL		0x00400000 /* Blocks allocated beyond EOF */
396 #define EXT4_RESERVED_FL		0x80000000 /* reserved for ext4 lib */
397 
398 #define EXT4_FL_USER_VISIBLE		0x004BDFFF /* User visible flags */
399 #define EXT4_FL_USER_MODIFIABLE		0x004B80FF /* User modifiable flags */
400 
401 /* Flags that should be inherited by new inodes from their parent. */
402 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
403 			   EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
404 			   EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
405 			   EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
406 
407 /* Flags that are appropriate for regular files (all but dir-specific ones). */
408 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
409 
410 /* Flags that are appropriate for non-directories/regular files. */
411 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
412 
413 /* Mask out flags that are inappropriate for the given type of inode. */
414 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
415 {
416 	if (S_ISDIR(mode))
417 		return flags;
418 	else if (S_ISREG(mode))
419 		return flags & EXT4_REG_FLMASK;
420 	else
421 		return flags & EXT4_OTHER_FLMASK;
422 }
423 
424 /*
425  * Inode flags used for atomic set/get
426  */
427 enum {
428 	EXT4_INODE_SECRM	= 0,	/* Secure deletion */
429 	EXT4_INODE_UNRM		= 1,	/* Undelete */
430 	EXT4_INODE_COMPR	= 2,	/* Compress file */
431 	EXT4_INODE_SYNC		= 3,	/* Synchronous updates */
432 	EXT4_INODE_IMMUTABLE	= 4,	/* Immutable file */
433 	EXT4_INODE_APPEND	= 5,	/* writes to file may only append */
434 	EXT4_INODE_NODUMP	= 6,	/* do not dump file */
435 	EXT4_INODE_NOATIME	= 7,	/* do not update atime */
436 /* Reserved for compression usage... */
437 	EXT4_INODE_DIRTY	= 8,
438 	EXT4_INODE_COMPRBLK	= 9,	/* One or more compressed clusters */
439 	EXT4_INODE_NOCOMPR	= 10,	/* Don't compress */
440 	EXT4_INODE_ECOMPR	= 11,	/* Compression error */
441 /* End compression flags --- maybe not all used */
442 	EXT4_INODE_INDEX	= 12,	/* hash-indexed directory */
443 	EXT4_INODE_IMAGIC	= 13,	/* AFS directory */
444 	EXT4_INODE_JOURNAL_DATA	= 14,	/* file data should be journaled */
445 	EXT4_INODE_NOTAIL	= 15,	/* file tail should not be merged */
446 	EXT4_INODE_DIRSYNC	= 16,	/* dirsync behaviour (directories only) */
447 	EXT4_INODE_TOPDIR	= 17,	/* Top of directory hierarchies*/
448 	EXT4_INODE_HUGE_FILE	= 18,	/* Set to each huge file */
449 	EXT4_INODE_EXTENTS	= 19,	/* Inode uses extents */
450 	EXT4_INODE_EA_INODE	= 21,	/* Inode used for large EA */
451 	EXT4_INODE_EOFBLOCKS	= 22,	/* Blocks allocated beyond EOF */
452 	EXT4_INODE_RESERVED	= 31,	/* reserved for ext4 lib */
453 };
454 
455 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
456 #define CHECK_FLAG_VALUE(FLAG) if (!TEST_FLAG_VALUE(FLAG)) { \
457 	printk(KERN_EMERG "EXT4 flag fail: " #FLAG ": %d %d\n", \
458 		EXT4_##FLAG##_FL, EXT4_INODE_##FLAG); BUG_ON(1); }
459 
460 /*
461  * Since it's pretty easy to mix up bit numbers and hex values, and we
462  * can't do a compile-time test for ENUM values, we use a run-time
463  * test to make sure that EXT4_XXX_FL is consistent with respect to
464  * EXT4_INODE_XXX.  If all is well the printk and BUG_ON will all drop
465  * out so it won't cost any extra space in the compiled kernel image.
466  * But it's important that these values are the same, since we are
467  * using EXT4_INODE_XXX to test for the flag values, but EXT4_XX_FL
468  * must be consistent with the values of FS_XXX_FL defined in
469  * include/linux/fs.h and the on-disk values found in ext2, ext3, and
470  * ext4 filesystems, and of course the values defined in e2fsprogs.
471  *
472  * It's not paranoia if the Murphy's Law really *is* out to get you.  :-)
473  */
474 static inline void ext4_check_flag_values(void)
475 {
476 	CHECK_FLAG_VALUE(SECRM);
477 	CHECK_FLAG_VALUE(UNRM);
478 	CHECK_FLAG_VALUE(COMPR);
479 	CHECK_FLAG_VALUE(SYNC);
480 	CHECK_FLAG_VALUE(IMMUTABLE);
481 	CHECK_FLAG_VALUE(APPEND);
482 	CHECK_FLAG_VALUE(NODUMP);
483 	CHECK_FLAG_VALUE(NOATIME);
484 	CHECK_FLAG_VALUE(DIRTY);
485 	CHECK_FLAG_VALUE(COMPRBLK);
486 	CHECK_FLAG_VALUE(NOCOMPR);
487 	CHECK_FLAG_VALUE(ECOMPR);
488 	CHECK_FLAG_VALUE(INDEX);
489 	CHECK_FLAG_VALUE(IMAGIC);
490 	CHECK_FLAG_VALUE(JOURNAL_DATA);
491 	CHECK_FLAG_VALUE(NOTAIL);
492 	CHECK_FLAG_VALUE(DIRSYNC);
493 	CHECK_FLAG_VALUE(TOPDIR);
494 	CHECK_FLAG_VALUE(HUGE_FILE);
495 	CHECK_FLAG_VALUE(EXTENTS);
496 	CHECK_FLAG_VALUE(EA_INODE);
497 	CHECK_FLAG_VALUE(EOFBLOCKS);
498 	CHECK_FLAG_VALUE(RESERVED);
499 }
500 
501 /* Used to pass group descriptor data when online resize is done */
502 struct ext4_new_group_input {
503 	__u32 group;		/* Group number for this data */
504 	__u64 block_bitmap;	/* Absolute block number of block bitmap */
505 	__u64 inode_bitmap;	/* Absolute block number of inode bitmap */
506 	__u64 inode_table;	/* Absolute block number of inode table start */
507 	__u32 blocks_count;	/* Total number of blocks in this group */
508 	__u16 reserved_blocks;	/* Number of reserved blocks in this group */
509 	__u16 unused;
510 };
511 
512 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
513 struct compat_ext4_new_group_input {
514 	u32 group;
515 	compat_u64 block_bitmap;
516 	compat_u64 inode_bitmap;
517 	compat_u64 inode_table;
518 	u32 blocks_count;
519 	u16 reserved_blocks;
520 	u16 unused;
521 };
522 #endif
523 
524 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
525 struct ext4_new_group_data {
526 	__u32 group;
527 	__u64 block_bitmap;
528 	__u64 inode_bitmap;
529 	__u64 inode_table;
530 	__u32 blocks_count;
531 	__u16 reserved_blocks;
532 	__u16 unused;
533 	__u32 free_blocks_count;
534 };
535 
536 /* Indexes used to index group tables in ext4_new_group_data */
537 enum {
538 	BLOCK_BITMAP = 0,	/* block bitmap */
539 	INODE_BITMAP,		/* inode bitmap */
540 	INODE_TABLE,		/* inode tables */
541 	GROUP_TABLE_COUNT,
542 };
543 
544 /*
545  * Flags used by ext4_map_blocks()
546  */
547 	/* Allocate any needed blocks and/or convert an unitialized
548 	   extent to be an initialized ext4 */
549 #define EXT4_GET_BLOCKS_CREATE			0x0001
550 	/* Request the creation of an unitialized extent */
551 #define EXT4_GET_BLOCKS_UNINIT_EXT		0x0002
552 #define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT	(EXT4_GET_BLOCKS_UNINIT_EXT|\
553 						 EXT4_GET_BLOCKS_CREATE)
554 	/* Caller is from the delayed allocation writeout path,
555 	   so set the magic i_delalloc_reserve_flag after taking the
556 	   inode allocation semaphore for */
557 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE	0x0004
558 	/* caller is from the direct IO path, request to creation of an
559 	unitialized extents if not allocated, split the uninitialized
560 	extent if blocks has been preallocated already*/
561 #define EXT4_GET_BLOCKS_PRE_IO			0x0008
562 #define EXT4_GET_BLOCKS_CONVERT			0x0010
563 #define EXT4_GET_BLOCKS_IO_CREATE_EXT		(EXT4_GET_BLOCKS_PRE_IO|\
564 					 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
565 	/* Convert extent to initialized after IO complete */
566 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT		(EXT4_GET_BLOCKS_CONVERT|\
567 					 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
568 	/* Punch out blocks of an extent */
569 #define EXT4_GET_BLOCKS_PUNCH_OUT_EXT		0x0020
570 	/* Don't normalize allocation size (used for fallocate) */
571 #define EXT4_GET_BLOCKS_NO_NORMALIZE		0x0040
572 	/* Request will not result in inode size update (user for fallocate) */
573 #define EXT4_GET_BLOCKS_KEEP_SIZE		0x0080
574 
575 /*
576  * Flags used by ext4_free_blocks
577  */
578 #define EXT4_FREE_BLOCKS_METADATA	0x0001
579 #define EXT4_FREE_BLOCKS_FORGET		0x0002
580 #define EXT4_FREE_BLOCKS_VALIDATED	0x0004
581 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE	0x0008
582 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER	0x0010
583 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER	0x0020
584 
585 /*
586  * Flags used by ext4_discard_partial_page_buffers
587  */
588 #define EXT4_DISCARD_PARTIAL_PG_ZERO_UNMAPPED	0x0001
589 
590 /*
591  * ioctl commands
592  */
593 #define	EXT4_IOC_GETFLAGS		FS_IOC_GETFLAGS
594 #define	EXT4_IOC_SETFLAGS		FS_IOC_SETFLAGS
595 #define	EXT4_IOC_GETVERSION		_IOR('f', 3, long)
596 #define	EXT4_IOC_SETVERSION		_IOW('f', 4, long)
597 #define	EXT4_IOC_GETVERSION_OLD		FS_IOC_GETVERSION
598 #define	EXT4_IOC_SETVERSION_OLD		FS_IOC_SETVERSION
599 #define EXT4_IOC_GETRSVSZ		_IOR('f', 5, long)
600 #define EXT4_IOC_SETRSVSZ		_IOW('f', 6, long)
601 #define EXT4_IOC_GROUP_EXTEND		_IOW('f', 7, unsigned long)
602 #define EXT4_IOC_GROUP_ADD		_IOW('f', 8, struct ext4_new_group_input)
603 #define EXT4_IOC_MIGRATE		_IO('f', 9)
604  /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
605  /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
606 #define EXT4_IOC_ALLOC_DA_BLKS		_IO('f', 12)
607 #define EXT4_IOC_MOVE_EXT		_IOWR('f', 15, struct move_extent)
608 #define EXT4_IOC_RESIZE_FS		_IOW('f', 16, __u64)
609 
610 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
611 /*
612  * ioctl commands in 32 bit emulation
613  */
614 #define EXT4_IOC32_GETFLAGS		FS_IOC32_GETFLAGS
615 #define EXT4_IOC32_SETFLAGS		FS_IOC32_SETFLAGS
616 #define EXT4_IOC32_GETVERSION		_IOR('f', 3, int)
617 #define EXT4_IOC32_SETVERSION		_IOW('f', 4, int)
618 #define EXT4_IOC32_GETRSVSZ		_IOR('f', 5, int)
619 #define EXT4_IOC32_SETRSVSZ		_IOW('f', 6, int)
620 #define EXT4_IOC32_GROUP_EXTEND		_IOW('f', 7, unsigned int)
621 #define EXT4_IOC32_GROUP_ADD		_IOW('f', 8, struct compat_ext4_new_group_input)
622 #define EXT4_IOC32_GETVERSION_OLD	FS_IOC32_GETVERSION
623 #define EXT4_IOC32_SETVERSION_OLD	FS_IOC32_SETVERSION
624 #endif
625 
626 /* Max physical block we can address w/o extents */
627 #define EXT4_MAX_BLOCK_FILE_PHYS	0xFFFFFFFF
628 
629 /*
630  * Structure of an inode on the disk
631  */
632 struct ext4_inode {
633 	__le16	i_mode;		/* File mode */
634 	__le16	i_uid;		/* Low 16 bits of Owner Uid */
635 	__le32	i_size_lo;	/* Size in bytes */
636 	__le32	i_atime;	/* Access time */
637 	__le32	i_ctime;	/* Inode Change time */
638 	__le32	i_mtime;	/* Modification time */
639 	__le32	i_dtime;	/* Deletion Time */
640 	__le16	i_gid;		/* Low 16 bits of Group Id */
641 	__le16	i_links_count;	/* Links count */
642 	__le32	i_blocks_lo;	/* Blocks count */
643 	__le32	i_flags;	/* File flags */
644 	union {
645 		struct {
646 			__le32  l_i_version;
647 		} linux1;
648 		struct {
649 			__u32  h_i_translator;
650 		} hurd1;
651 		struct {
652 			__u32  m_i_reserved1;
653 		} masix1;
654 	} osd1;				/* OS dependent 1 */
655 	__le32	i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
656 	__le32	i_generation;	/* File version (for NFS) */
657 	__le32	i_file_acl_lo;	/* File ACL */
658 	__le32	i_size_high;
659 	__le32	i_obso_faddr;	/* Obsoleted fragment address */
660 	union {
661 		struct {
662 			__le16	l_i_blocks_high; /* were l_i_reserved1 */
663 			__le16	l_i_file_acl_high;
664 			__le16	l_i_uid_high;	/* these 2 fields */
665 			__le16	l_i_gid_high;	/* were reserved2[0] */
666 			__le16	l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
667 			__le16	l_i_reserved;
668 		} linux2;
669 		struct {
670 			__le16	h_i_reserved1;	/* Obsoleted fragment number/size which are removed in ext4 */
671 			__u16	h_i_mode_high;
672 			__u16	h_i_uid_high;
673 			__u16	h_i_gid_high;
674 			__u32	h_i_author;
675 		} hurd2;
676 		struct {
677 			__le16	h_i_reserved1;	/* Obsoleted fragment number/size which are removed in ext4 */
678 			__le16	m_i_file_acl_high;
679 			__u32	m_i_reserved2[2];
680 		} masix2;
681 	} osd2;				/* OS dependent 2 */
682 	__le16	i_extra_isize;
683 	__le16	i_checksum_hi;	/* crc32c(uuid+inum+inode) BE */
684 	__le32  i_ctime_extra;  /* extra Change time      (nsec << 2 | epoch) */
685 	__le32  i_mtime_extra;  /* extra Modification time(nsec << 2 | epoch) */
686 	__le32  i_atime_extra;  /* extra Access time      (nsec << 2 | epoch) */
687 	__le32  i_crtime;       /* File Creation time */
688 	__le32  i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
689 	__le32  i_version_hi;	/* high 32 bits for 64-bit version */
690 };
691 
692 struct move_extent {
693 	__u32 reserved;		/* should be zero */
694 	__u32 donor_fd;		/* donor file descriptor */
695 	__u64 orig_start;	/* logical start offset in block for orig */
696 	__u64 donor_start;	/* logical start offset in block for donor */
697 	__u64 len;		/* block length to be moved */
698 	__u64 moved_len;	/* moved block length */
699 };
700 
701 #define EXT4_EPOCH_BITS 2
702 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
703 #define EXT4_NSEC_MASK  (~0UL << EXT4_EPOCH_BITS)
704 
705 /*
706  * Extended fields will fit into an inode if the filesystem was formatted
707  * with large inodes (-I 256 or larger) and there are not currently any EAs
708  * consuming all of the available space. For new inodes we always reserve
709  * enough space for the kernel's known extended fields, but for inodes
710  * created with an old kernel this might not have been the case. None of
711  * the extended inode fields is critical for correct filesystem operation.
712  * This macro checks if a certain field fits in the inode. Note that
713  * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
714  */
715 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field)	\
716 	((offsetof(typeof(*ext4_inode), field) +	\
717 	  sizeof((ext4_inode)->field))			\
718 	<= (EXT4_GOOD_OLD_INODE_SIZE +			\
719 	    (einode)->i_extra_isize))			\
720 
721 static inline __le32 ext4_encode_extra_time(struct timespec *time)
722 {
723        return cpu_to_le32((sizeof(time->tv_sec) > 4 ?
724 			   (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) |
725                           ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK));
726 }
727 
728 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
729 {
730        if (sizeof(time->tv_sec) > 4)
731 	       time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK)
732 			       << 32;
733        time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
734 }
735 
736 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode)			       \
737 do {									       \
738 	(raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec);	       \
739 	if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     \
740 		(raw_inode)->xtime ## _extra =				       \
741 				ext4_encode_extra_time(&(inode)->xtime);       \
742 } while (0)
743 
744 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode)			       \
745 do {									       \
746 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))		       \
747 		(raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec);      \
748 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))	       \
749 		(raw_inode)->xtime ## _extra =				       \
750 				ext4_encode_extra_time(&(einode)->xtime);      \
751 } while (0)
752 
753 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode)			       \
754 do {									       \
755 	(inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime);       \
756 	if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     \
757 		ext4_decode_extra_time(&(inode)->xtime,			       \
758 				       raw_inode->xtime ## _extra);	       \
759 	else								       \
760 		(inode)->xtime.tv_nsec = 0;				       \
761 } while (0)
762 
763 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode)			       \
764 do {									       \
765 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))		       \
766 		(einode)->xtime.tv_sec = 				       \
767 			(signed)le32_to_cpu((raw_inode)->xtime);	       \
768 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))	       \
769 		ext4_decode_extra_time(&(einode)->xtime,		       \
770 				       raw_inode->xtime ## _extra);	       \
771 	else								       \
772 		(einode)->xtime.tv_nsec = 0;				       \
773 } while (0)
774 
775 #define i_disk_version osd1.linux1.l_i_version
776 
777 #if defined(__KERNEL__) || defined(__linux__)
778 #define i_reserved1	osd1.linux1.l_i_reserved1
779 #define i_file_acl_high	osd2.linux2.l_i_file_acl_high
780 #define i_blocks_high	osd2.linux2.l_i_blocks_high
781 #define i_uid_low	i_uid
782 #define i_gid_low	i_gid
783 #define i_uid_high	osd2.linux2.l_i_uid_high
784 #define i_gid_high	osd2.linux2.l_i_gid_high
785 #define i_checksum_lo	osd2.linux2.l_i_checksum_lo
786 
787 #elif defined(__GNU__)
788 
789 #define i_translator	osd1.hurd1.h_i_translator
790 #define i_uid_high	osd2.hurd2.h_i_uid_high
791 #define i_gid_high	osd2.hurd2.h_i_gid_high
792 #define i_author	osd2.hurd2.h_i_author
793 
794 #elif defined(__masix__)
795 
796 #define i_reserved1	osd1.masix1.m_i_reserved1
797 #define i_file_acl_high	osd2.masix2.m_i_file_acl_high
798 #define i_reserved2	osd2.masix2.m_i_reserved2
799 
800 #endif /* defined(__KERNEL__) || defined(__linux__) */
801 
802 /*
803  * storage for cached extent
804  * If ec_len == 0, then the cache is invalid.
805  * If ec_start == 0, then the cache represents a gap (null mapping)
806  */
807 struct ext4_ext_cache {
808 	ext4_fsblk_t	ec_start;
809 	ext4_lblk_t	ec_block;
810 	__u32		ec_len; /* must be 32bit to return holes */
811 };
812 
813 /*
814  * fourth extended file system inode data in memory
815  */
816 struct ext4_inode_info {
817 	__le32	i_data[15];	/* unconverted */
818 	__u32	i_dtime;
819 	ext4_fsblk_t	i_file_acl;
820 
821 	/*
822 	 * i_block_group is the number of the block group which contains
823 	 * this file's inode.  Constant across the lifetime of the inode,
824 	 * it is ued for making block allocation decisions - we try to
825 	 * place a file's data blocks near its inode block, and new inodes
826 	 * near to their parent directory's inode.
827 	 */
828 	ext4_group_t	i_block_group;
829 	ext4_lblk_t	i_dir_start_lookup;
830 #if (BITS_PER_LONG < 64)
831 	unsigned long	i_state_flags;		/* Dynamic state flags */
832 #endif
833 	unsigned long	i_flags;
834 
835 #ifdef CONFIG_EXT4_FS_XATTR
836 	/*
837 	 * Extended attributes can be read independently of the main file
838 	 * data. Taking i_mutex even when reading would cause contention
839 	 * between readers of EAs and writers of regular file data, so
840 	 * instead we synchronize on xattr_sem when reading or changing
841 	 * EAs.
842 	 */
843 	struct rw_semaphore xattr_sem;
844 #endif
845 
846 	struct list_head i_orphan;	/* unlinked but open inodes */
847 
848 	/*
849 	 * i_disksize keeps track of what the inode size is ON DISK, not
850 	 * in memory.  During truncate, i_size is set to the new size by
851 	 * the VFS prior to calling ext4_truncate(), but the filesystem won't
852 	 * set i_disksize to 0 until the truncate is actually under way.
853 	 *
854 	 * The intent is that i_disksize always represents the blocks which
855 	 * are used by this file.  This allows recovery to restart truncate
856 	 * on orphans if we crash during truncate.  We actually write i_disksize
857 	 * into the on-disk inode when writing inodes out, instead of i_size.
858 	 *
859 	 * The only time when i_disksize and i_size may be different is when
860 	 * a truncate is in progress.  The only things which change i_disksize
861 	 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
862 	 */
863 	loff_t	i_disksize;
864 
865 	/*
866 	 * i_data_sem is for serialising ext4_truncate() against
867 	 * ext4_getblock().  In the 2.4 ext2 design, great chunks of inode's
868 	 * data tree are chopped off during truncate. We can't do that in
869 	 * ext4 because whenever we perform intermediate commits during
870 	 * truncate, the inode and all the metadata blocks *must* be in a
871 	 * consistent state which allows truncation of the orphans to restart
872 	 * during recovery.  Hence we must fix the get_block-vs-truncate race
873 	 * by other means, so we have i_data_sem.
874 	 */
875 	struct rw_semaphore i_data_sem;
876 	struct inode vfs_inode;
877 	struct jbd2_inode *jinode;
878 
879 	struct ext4_ext_cache i_cached_extent;
880 	/*
881 	 * File creation time. Its function is same as that of
882 	 * struct timespec i_{a,c,m}time in the generic inode.
883 	 */
884 	struct timespec i_crtime;
885 
886 	/* mballoc */
887 	struct list_head i_prealloc_list;
888 	spinlock_t i_prealloc_lock;
889 
890 	/* ialloc */
891 	ext4_group_t	i_last_alloc_group;
892 
893 	/* allocation reservation info for delalloc */
894 	/* In case of bigalloc, these refer to clusters rather than blocks */
895 	unsigned int i_reserved_data_blocks;
896 	unsigned int i_reserved_meta_blocks;
897 	unsigned int i_allocated_meta_blocks;
898 	ext4_lblk_t i_da_metadata_calc_last_lblock;
899 	int i_da_metadata_calc_len;
900 
901 	/* on-disk additional length */
902 	__u16 i_extra_isize;
903 
904 #ifdef CONFIG_QUOTA
905 	/* quota space reservation, managed internally by quota code */
906 	qsize_t i_reserved_quota;
907 #endif
908 
909 	/* completed IOs that might need unwritten extents handling */
910 	struct list_head i_completed_io_list;
911 	spinlock_t i_completed_io_lock;
912 	atomic_t i_ioend_count;	/* Number of outstanding io_end structs */
913 	/* current io_end structure for async DIO write*/
914 	ext4_io_end_t *cur_aio_dio;
915 	atomic_t i_aiodio_unwritten; /* Nr. of inflight conversions pending */
916 
917 	spinlock_t i_block_reservation_lock;
918 
919 	/*
920 	 * Transactions that contain inode's metadata needed to complete
921 	 * fsync and fdatasync, respectively.
922 	 */
923 	tid_t i_sync_tid;
924 	tid_t i_datasync_tid;
925 
926 	/* Precomputed uuid+inum+igen checksum for seeding inode checksums */
927 	__u32 i_csum_seed;
928 };
929 
930 /*
931  * File system states
932  */
933 #define	EXT4_VALID_FS			0x0001	/* Unmounted cleanly */
934 #define	EXT4_ERROR_FS			0x0002	/* Errors detected */
935 #define	EXT4_ORPHAN_FS			0x0004	/* Orphans being recovered */
936 
937 /*
938  * Misc. filesystem flags
939  */
940 #define EXT2_FLAGS_SIGNED_HASH		0x0001  /* Signed dirhash in use */
941 #define EXT2_FLAGS_UNSIGNED_HASH	0x0002  /* Unsigned dirhash in use */
942 #define EXT2_FLAGS_TEST_FILESYS		0x0004	/* to test development code */
943 
944 /*
945  * Mount flags
946  */
947 #define EXT4_MOUNT_GRPID		0x00004	/* Create files with directory's group */
948 #define EXT4_MOUNT_DEBUG		0x00008	/* Some debugging messages */
949 #define EXT4_MOUNT_ERRORS_CONT		0x00010	/* Continue on errors */
950 #define EXT4_MOUNT_ERRORS_RO		0x00020	/* Remount fs ro on errors */
951 #define EXT4_MOUNT_ERRORS_PANIC		0x00040	/* Panic on errors */
952 #define EXT4_MOUNT_ERRORS_MASK		0x00070
953 #define EXT4_MOUNT_MINIX_DF		0x00080	/* Mimics the Minix statfs */
954 #define EXT4_MOUNT_NOLOAD		0x00100	/* Don't use existing journal*/
955 #define EXT4_MOUNT_DATA_FLAGS		0x00C00	/* Mode for data writes: */
956 #define EXT4_MOUNT_JOURNAL_DATA		0x00400	/* Write data to journal */
957 #define EXT4_MOUNT_ORDERED_DATA		0x00800	/* Flush data before commit */
958 #define EXT4_MOUNT_WRITEBACK_DATA	0x00C00	/* No data ordering */
959 #define EXT4_MOUNT_UPDATE_JOURNAL	0x01000	/* Update the journal format */
960 #define EXT4_MOUNT_NO_UID32		0x02000  /* Disable 32-bit UIDs */
961 #define EXT4_MOUNT_XATTR_USER		0x04000	/* Extended user attributes */
962 #define EXT4_MOUNT_POSIX_ACL		0x08000	/* POSIX Access Control Lists */
963 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC	0x10000	/* No auto delalloc mapping */
964 #define EXT4_MOUNT_BARRIER		0x20000 /* Use block barriers */
965 #define EXT4_MOUNT_QUOTA		0x80000 /* Some quota option set */
966 #define EXT4_MOUNT_USRQUOTA		0x100000 /* "old" user quota */
967 #define EXT4_MOUNT_GRPQUOTA		0x200000 /* "old" group quota */
968 #define EXT4_MOUNT_DIOREAD_NOLOCK	0x400000 /* Enable support for dio read nolocking */
969 #define EXT4_MOUNT_JOURNAL_CHECKSUM	0x800000 /* Journal checksums */
970 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT	0x1000000 /* Journal Async Commit */
971 #define EXT4_MOUNT_MBLK_IO_SUBMIT	0x4000000 /* multi-block io submits */
972 #define EXT4_MOUNT_DELALLOC		0x8000000 /* Delalloc support */
973 #define EXT4_MOUNT_DATA_ERR_ABORT	0x10000000 /* Abort on file data write */
974 #define EXT4_MOUNT_BLOCK_VALIDITY	0x20000000 /* Block validity checking */
975 #define EXT4_MOUNT_DISCARD		0x40000000 /* Issue DISCARD requests */
976 #define EXT4_MOUNT_INIT_INODE_TABLE	0x80000000 /* Initialize uninitialized itables */
977 
978 #define EXT4_MOUNT2_EXPLICIT_DELALLOC	0x00000001 /* User explicitly
979 						      specified delalloc */
980 
981 #define clear_opt(sb, opt)		EXT4_SB(sb)->s_mount_opt &= \
982 						~EXT4_MOUNT_##opt
983 #define set_opt(sb, opt)		EXT4_SB(sb)->s_mount_opt |= \
984 						EXT4_MOUNT_##opt
985 #define test_opt(sb, opt)		(EXT4_SB(sb)->s_mount_opt & \
986 					 EXT4_MOUNT_##opt)
987 
988 #define clear_opt2(sb, opt)		EXT4_SB(sb)->s_mount_opt2 &= \
989 						~EXT4_MOUNT2_##opt
990 #define set_opt2(sb, opt)		EXT4_SB(sb)->s_mount_opt2 |= \
991 						EXT4_MOUNT2_##opt
992 #define test_opt2(sb, opt)		(EXT4_SB(sb)->s_mount_opt2 & \
993 					 EXT4_MOUNT2_##opt)
994 
995 #define ext4_test_and_set_bit		__test_and_set_bit_le
996 #define ext4_set_bit			__set_bit_le
997 #define ext4_set_bit_atomic		ext2_set_bit_atomic
998 #define ext4_test_and_clear_bit		__test_and_clear_bit_le
999 #define ext4_clear_bit			__clear_bit_le
1000 #define ext4_clear_bit_atomic		ext2_clear_bit_atomic
1001 #define ext4_test_bit			test_bit_le
1002 #define ext4_find_next_zero_bit		find_next_zero_bit_le
1003 #define ext4_find_next_bit		find_next_bit_le
1004 
1005 extern void ext4_set_bits(void *bm, int cur, int len);
1006 
1007 /*
1008  * Maximal mount counts between two filesystem checks
1009  */
1010 #define EXT4_DFL_MAX_MNT_COUNT		20	/* Allow 20 mounts */
1011 #define EXT4_DFL_CHECKINTERVAL		0	/* Don't use interval check */
1012 
1013 /*
1014  * Behaviour when detecting errors
1015  */
1016 #define EXT4_ERRORS_CONTINUE		1	/* Continue execution */
1017 #define EXT4_ERRORS_RO			2	/* Remount fs read-only */
1018 #define EXT4_ERRORS_PANIC		3	/* Panic */
1019 #define EXT4_ERRORS_DEFAULT		EXT4_ERRORS_CONTINUE
1020 
1021 /* Metadata checksum algorithm codes */
1022 #define EXT4_CRC32C_CHKSUM		1
1023 
1024 /*
1025  * Structure of the super block
1026  */
1027 struct ext4_super_block {
1028 /*00*/	__le32	s_inodes_count;		/* Inodes count */
1029 	__le32	s_blocks_count_lo;	/* Blocks count */
1030 	__le32	s_r_blocks_count_lo;	/* Reserved blocks count */
1031 	__le32	s_free_blocks_count_lo;	/* Free blocks count */
1032 /*10*/	__le32	s_free_inodes_count;	/* Free inodes count */
1033 	__le32	s_first_data_block;	/* First Data Block */
1034 	__le32	s_log_block_size;	/* Block size */
1035 	__le32	s_log_cluster_size;	/* Allocation cluster size */
1036 /*20*/	__le32	s_blocks_per_group;	/* # Blocks per group */
1037 	__le32	s_clusters_per_group;	/* # Clusters per group */
1038 	__le32	s_inodes_per_group;	/* # Inodes per group */
1039 	__le32	s_mtime;		/* Mount time */
1040 /*30*/	__le32	s_wtime;		/* Write time */
1041 	__le16	s_mnt_count;		/* Mount count */
1042 	__le16	s_max_mnt_count;	/* Maximal mount count */
1043 	__le16	s_magic;		/* Magic signature */
1044 	__le16	s_state;		/* File system state */
1045 	__le16	s_errors;		/* Behaviour when detecting errors */
1046 	__le16	s_minor_rev_level;	/* minor revision level */
1047 /*40*/	__le32	s_lastcheck;		/* time of last check */
1048 	__le32	s_checkinterval;	/* max. time between checks */
1049 	__le32	s_creator_os;		/* OS */
1050 	__le32	s_rev_level;		/* Revision level */
1051 /*50*/	__le16	s_def_resuid;		/* Default uid for reserved blocks */
1052 	__le16	s_def_resgid;		/* Default gid for reserved blocks */
1053 	/*
1054 	 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1055 	 *
1056 	 * Note: the difference between the compatible feature set and
1057 	 * the incompatible feature set is that if there is a bit set
1058 	 * in the incompatible feature set that the kernel doesn't
1059 	 * know about, it should refuse to mount the filesystem.
1060 	 *
1061 	 * e2fsck's requirements are more strict; if it doesn't know
1062 	 * about a feature in either the compatible or incompatible
1063 	 * feature set, it must abort and not try to meddle with
1064 	 * things it doesn't understand...
1065 	 */
1066 	__le32	s_first_ino;		/* First non-reserved inode */
1067 	__le16  s_inode_size;		/* size of inode structure */
1068 	__le16	s_block_group_nr;	/* block group # of this superblock */
1069 	__le32	s_feature_compat;	/* compatible feature set */
1070 /*60*/	__le32	s_feature_incompat;	/* incompatible feature set */
1071 	__le32	s_feature_ro_compat;	/* readonly-compatible feature set */
1072 /*68*/	__u8	s_uuid[16];		/* 128-bit uuid for volume */
1073 /*78*/	char	s_volume_name[16];	/* volume name */
1074 /*88*/	char	s_last_mounted[64];	/* directory where last mounted */
1075 /*C8*/	__le32	s_algorithm_usage_bitmap; /* For compression */
1076 	/*
1077 	 * Performance hints.  Directory preallocation should only
1078 	 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1079 	 */
1080 	__u8	s_prealloc_blocks;	/* Nr of blocks to try to preallocate*/
1081 	__u8	s_prealloc_dir_blocks;	/* Nr to preallocate for dirs */
1082 	__le16	s_reserved_gdt_blocks;	/* Per group desc for online growth */
1083 	/*
1084 	 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1085 	 */
1086 /*D0*/	__u8	s_journal_uuid[16];	/* uuid of journal superblock */
1087 /*E0*/	__le32	s_journal_inum;		/* inode number of journal file */
1088 	__le32	s_journal_dev;		/* device number of journal file */
1089 	__le32	s_last_orphan;		/* start of list of inodes to delete */
1090 	__le32	s_hash_seed[4];		/* HTREE hash seed */
1091 	__u8	s_def_hash_version;	/* Default hash version to use */
1092 	__u8	s_jnl_backup_type;
1093 	__le16  s_desc_size;		/* size of group descriptor */
1094 /*100*/	__le32	s_default_mount_opts;
1095 	__le32	s_first_meta_bg;	/* First metablock block group */
1096 	__le32	s_mkfs_time;		/* When the filesystem was created */
1097 	__le32	s_jnl_blocks[17];	/* Backup of the journal inode */
1098 	/* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1099 /*150*/	__le32	s_blocks_count_hi;	/* Blocks count */
1100 	__le32	s_r_blocks_count_hi;	/* Reserved blocks count */
1101 	__le32	s_free_blocks_count_hi;	/* Free blocks count */
1102 	__le16	s_min_extra_isize;	/* All inodes have at least # bytes */
1103 	__le16	s_want_extra_isize; 	/* New inodes should reserve # bytes */
1104 	__le32	s_flags;		/* Miscellaneous flags */
1105 	__le16  s_raid_stride;		/* RAID stride */
1106 	__le16  s_mmp_update_interval;  /* # seconds to wait in MMP checking */
1107 	__le64  s_mmp_block;            /* Block for multi-mount protection */
1108 	__le32  s_raid_stripe_width;    /* blocks on all data disks (N*stride)*/
1109 	__u8	s_log_groups_per_flex;  /* FLEX_BG group size */
1110 	__u8	s_checksum_type;	/* metadata checksum algorithm used */
1111 	__le16  s_reserved_pad;
1112 	__le64	s_kbytes_written;	/* nr of lifetime kilobytes written */
1113 	__le32	s_snapshot_inum;	/* Inode number of active snapshot */
1114 	__le32	s_snapshot_id;		/* sequential ID of active snapshot */
1115 	__le64	s_snapshot_r_blocks_count; /* reserved blocks for active
1116 					      snapshot's future use */
1117 	__le32	s_snapshot_list;	/* inode number of the head of the
1118 					   on-disk snapshot list */
1119 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1120 	__le32	s_error_count;		/* number of fs errors */
1121 	__le32	s_first_error_time;	/* first time an error happened */
1122 	__le32	s_first_error_ino;	/* inode involved in first error */
1123 	__le64	s_first_error_block;	/* block involved of first error */
1124 	__u8	s_first_error_func[32];	/* function where the error happened */
1125 	__le32	s_first_error_line;	/* line number where error happened */
1126 	__le32	s_last_error_time;	/* most recent time of an error */
1127 	__le32	s_last_error_ino;	/* inode involved in last error */
1128 	__le32	s_last_error_line;	/* line number where error happened */
1129 	__le64	s_last_error_block;	/* block involved of last error */
1130 	__u8	s_last_error_func[32];	/* function where the error happened */
1131 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1132 	__u8	s_mount_opts[64];
1133 	__le32	s_usr_quota_inum;	/* inode for tracking user quota */
1134 	__le32	s_grp_quota_inum;	/* inode for tracking group quota */
1135 	__le32	s_overhead_clusters;	/* overhead blocks/clusters in fs */
1136 	__le32	s_reserved[108];	/* Padding to the end of the block */
1137 	__le32	s_checksum;		/* crc32c(superblock) */
1138 };
1139 
1140 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1141 
1142 #ifdef __KERNEL__
1143 
1144 /*
1145  * run-time mount flags
1146  */
1147 #define EXT4_MF_MNTDIR_SAMPLED	0x0001
1148 #define EXT4_MF_FS_ABORTED	0x0002	/* Fatal error detected */
1149 
1150 /*
1151  * fourth extended-fs super-block data in memory
1152  */
1153 struct ext4_sb_info {
1154 	unsigned long s_desc_size;	/* Size of a group descriptor in bytes */
1155 	unsigned long s_inodes_per_block;/* Number of inodes per block */
1156 	unsigned long s_blocks_per_group;/* Number of blocks in a group */
1157 	unsigned long s_clusters_per_group; /* Number of clusters in a group */
1158 	unsigned long s_inodes_per_group;/* Number of inodes in a group */
1159 	unsigned long s_itb_per_group;	/* Number of inode table blocks per group */
1160 	unsigned long s_gdb_count;	/* Number of group descriptor blocks */
1161 	unsigned long s_desc_per_block;	/* Number of group descriptors per block */
1162 	ext4_group_t s_groups_count;	/* Number of groups in the fs */
1163 	ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1164 	unsigned long s_overhead_last;  /* Last calculated overhead */
1165 	unsigned long s_blocks_last;    /* Last seen block count */
1166 	unsigned int s_cluster_ratio;	/* Number of blocks per cluster */
1167 	unsigned int s_cluster_bits;	/* log2 of s_cluster_ratio */
1168 	loff_t s_bitmap_maxbytes;	/* max bytes for bitmap files */
1169 	struct buffer_head * s_sbh;	/* Buffer containing the super block */
1170 	struct ext4_super_block *s_es;	/* Pointer to the super block in the buffer */
1171 	struct buffer_head **s_group_desc;
1172 	unsigned int s_mount_opt;
1173 	unsigned int s_mount_opt2;
1174 	unsigned int s_mount_flags;
1175 	unsigned int s_def_mount_opt;
1176 	ext4_fsblk_t s_sb_block;
1177 	uid_t s_resuid;
1178 	gid_t s_resgid;
1179 	unsigned short s_mount_state;
1180 	unsigned short s_pad;
1181 	int s_addr_per_block_bits;
1182 	int s_desc_per_block_bits;
1183 	int s_inode_size;
1184 	int s_first_ino;
1185 	unsigned int s_inode_readahead_blks;
1186 	unsigned int s_inode_goal;
1187 	spinlock_t s_next_gen_lock;
1188 	u32 s_next_generation;
1189 	u32 s_hash_seed[4];
1190 	int s_def_hash_version;
1191 	int s_hash_unsigned;	/* 3 if hash should be signed, 0 if not */
1192 	struct percpu_counter s_freeclusters_counter;
1193 	struct percpu_counter s_freeinodes_counter;
1194 	struct percpu_counter s_dirs_counter;
1195 	struct percpu_counter s_dirtyclusters_counter;
1196 	struct blockgroup_lock *s_blockgroup_lock;
1197 	struct proc_dir_entry *s_proc;
1198 	struct kobject s_kobj;
1199 	struct completion s_kobj_unregister;
1200 
1201 	/* Journaling */
1202 	struct journal_s *s_journal;
1203 	struct list_head s_orphan;
1204 	struct mutex s_orphan_lock;
1205 	unsigned long s_resize_flags;		/* Flags indicating if there
1206 						   is a resizer */
1207 	unsigned long s_commit_interval;
1208 	u32 s_max_batch_time;
1209 	u32 s_min_batch_time;
1210 	struct block_device *journal_bdev;
1211 #ifdef CONFIG_QUOTA
1212 	char *s_qf_names[MAXQUOTAS];		/* Names of quota files with journalled quota */
1213 	int s_jquota_fmt;			/* Format of quota to use */
1214 #endif
1215 	unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1216 	struct rb_root system_blks;
1217 
1218 #ifdef EXTENTS_STATS
1219 	/* ext4 extents stats */
1220 	unsigned long s_ext_min;
1221 	unsigned long s_ext_max;
1222 	unsigned long s_depth_max;
1223 	spinlock_t s_ext_stats_lock;
1224 	unsigned long s_ext_blocks;
1225 	unsigned long s_ext_extents;
1226 #endif
1227 
1228 	/* for buddy allocator */
1229 	struct ext4_group_info ***s_group_info;
1230 	struct inode *s_buddy_cache;
1231 	spinlock_t s_md_lock;
1232 	unsigned short *s_mb_offsets;
1233 	unsigned int *s_mb_maxs;
1234 
1235 	/* tunables */
1236 	unsigned long s_stripe;
1237 	unsigned int s_mb_stream_request;
1238 	unsigned int s_mb_max_to_scan;
1239 	unsigned int s_mb_min_to_scan;
1240 	unsigned int s_mb_stats;
1241 	unsigned int s_mb_order2_reqs;
1242 	unsigned int s_mb_group_prealloc;
1243 	unsigned int s_max_writeback_mb_bump;
1244 	/* where last allocation was done - for stream allocation */
1245 	unsigned long s_mb_last_group;
1246 	unsigned long s_mb_last_start;
1247 
1248 	/* stats for buddy allocator */
1249 	atomic_t s_bal_reqs;	/* number of reqs with len > 1 */
1250 	atomic_t s_bal_success;	/* we found long enough chunks */
1251 	atomic_t s_bal_allocated;	/* in blocks */
1252 	atomic_t s_bal_ex_scanned;	/* total extents scanned */
1253 	atomic_t s_bal_goals;	/* goal hits */
1254 	atomic_t s_bal_breaks;	/* too long searches */
1255 	atomic_t s_bal_2orders;	/* 2^order hits */
1256 	spinlock_t s_bal_lock;
1257 	unsigned long s_mb_buddies_generated;
1258 	unsigned long long s_mb_generation_time;
1259 	atomic_t s_mb_lost_chunks;
1260 	atomic_t s_mb_preallocated;
1261 	atomic_t s_mb_discarded;
1262 	atomic_t s_lock_busy;
1263 
1264 	/* locality groups */
1265 	struct ext4_locality_group __percpu *s_locality_groups;
1266 
1267 	/* for write statistics */
1268 	unsigned long s_sectors_written_start;
1269 	u64 s_kbytes_written;
1270 
1271 	unsigned int s_log_groups_per_flex;
1272 	struct flex_groups *s_flex_groups;
1273 
1274 	/* workqueue for dio unwritten */
1275 	struct workqueue_struct *dio_unwritten_wq;
1276 
1277 	/* timer for periodic error stats printing */
1278 	struct timer_list s_err_report;
1279 
1280 	/* Lazy inode table initialization info */
1281 	struct ext4_li_request *s_li_request;
1282 	/* Wait multiplier for lazy initialization thread */
1283 	unsigned int s_li_wait_mult;
1284 
1285 	/* Kernel thread for multiple mount protection */
1286 	struct task_struct *s_mmp_tsk;
1287 
1288 	/* record the last minlen when FITRIM is called. */
1289 	atomic_t s_last_trim_minblks;
1290 
1291 	/* Reference to checksum algorithm driver via cryptoapi */
1292 	struct crypto_shash *s_chksum_driver;
1293 
1294 	/* Precomputed FS UUID checksum for seeding other checksums */
1295 	__u32 s_csum_seed;
1296 };
1297 
1298 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1299 {
1300 	return sb->s_fs_info;
1301 }
1302 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1303 {
1304 	return container_of(inode, struct ext4_inode_info, vfs_inode);
1305 }
1306 
1307 static inline struct timespec ext4_current_time(struct inode *inode)
1308 {
1309 	return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1310 		current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1311 }
1312 
1313 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1314 {
1315 	return ino == EXT4_ROOT_INO ||
1316 		ino == EXT4_JOURNAL_INO ||
1317 		ino == EXT4_RESIZE_INO ||
1318 		(ino >= EXT4_FIRST_INO(sb) &&
1319 		 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1320 }
1321 
1322 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
1323 					      struct ext4_io_end *io_end)
1324 {
1325 	if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
1326 		io_end->flag |= EXT4_IO_END_UNWRITTEN;
1327 		atomic_inc(&EXT4_I(inode)->i_aiodio_unwritten);
1328 	}
1329 }
1330 
1331 /*
1332  * Inode dynamic state flags
1333  */
1334 enum {
1335 	EXT4_STATE_JDATA,		/* journaled data exists */
1336 	EXT4_STATE_NEW,			/* inode is newly created */
1337 	EXT4_STATE_XATTR,		/* has in-inode xattrs */
1338 	EXT4_STATE_NO_EXPAND,		/* No space for expansion */
1339 	EXT4_STATE_DA_ALLOC_CLOSE,	/* Alloc DA blks on close */
1340 	EXT4_STATE_EXT_MIGRATE,		/* Inode is migrating */
1341 	EXT4_STATE_DIO_UNWRITTEN,	/* need convert on dio done*/
1342 	EXT4_STATE_NEWENTRY,		/* File just added to dir */
1343 	EXT4_STATE_DELALLOC_RESERVED,	/* blks already reserved for delalloc */
1344 };
1345 
1346 #define EXT4_INODE_BIT_FNS(name, field, offset)				\
1347 static inline int ext4_test_inode_##name(struct inode *inode, int bit)	\
1348 {									\
1349 	return test_bit(bit + (offset), &EXT4_I(inode)->i_##field);	\
1350 }									\
1351 static inline void ext4_set_inode_##name(struct inode *inode, int bit)	\
1352 {									\
1353 	set_bit(bit + (offset), &EXT4_I(inode)->i_##field);		\
1354 }									\
1355 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1356 {									\
1357 	clear_bit(bit + (offset), &EXT4_I(inode)->i_##field);		\
1358 }
1359 
1360 EXT4_INODE_BIT_FNS(flag, flags, 0)
1361 #if (BITS_PER_LONG < 64)
1362 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1363 
1364 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1365 {
1366 	(ei)->i_state_flags = 0;
1367 }
1368 #else
1369 EXT4_INODE_BIT_FNS(state, flags, 32)
1370 
1371 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1372 {
1373 	/* We depend on the fact that callers will set i_flags */
1374 }
1375 #endif
1376 #else
1377 /* Assume that user mode programs are passing in an ext4fs superblock, not
1378  * a kernel struct super_block.  This will allow us to call the feature-test
1379  * macros from user land. */
1380 #define EXT4_SB(sb)	(sb)
1381 #endif
1382 
1383 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1384 
1385 /*
1386  * Codes for operating systems
1387  */
1388 #define EXT4_OS_LINUX		0
1389 #define EXT4_OS_HURD		1
1390 #define EXT4_OS_MASIX		2
1391 #define EXT4_OS_FREEBSD		3
1392 #define EXT4_OS_LITES		4
1393 
1394 /*
1395  * Revision levels
1396  */
1397 #define EXT4_GOOD_OLD_REV	0	/* The good old (original) format */
1398 #define EXT4_DYNAMIC_REV	1	/* V2 format w/ dynamic inode sizes */
1399 
1400 #define EXT4_CURRENT_REV	EXT4_GOOD_OLD_REV
1401 #define EXT4_MAX_SUPP_REV	EXT4_DYNAMIC_REV
1402 
1403 #define EXT4_GOOD_OLD_INODE_SIZE 128
1404 
1405 /*
1406  * Feature set definitions
1407  */
1408 
1409 #define EXT4_HAS_COMPAT_FEATURE(sb,mask)			\
1410 	((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1411 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask)			\
1412 	((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1413 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask)			\
1414 	((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1415 #define EXT4_SET_COMPAT_FEATURE(sb,mask)			\
1416 	EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1417 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask)			\
1418 	EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1419 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask)			\
1420 	EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1421 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask)			\
1422 	EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1423 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask)			\
1424 	EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1425 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask)			\
1426 	EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1427 
1428 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC	0x0001
1429 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES	0x0002
1430 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL		0x0004
1431 #define EXT4_FEATURE_COMPAT_EXT_ATTR		0x0008
1432 #define EXT4_FEATURE_COMPAT_RESIZE_INODE	0x0010
1433 #define EXT4_FEATURE_COMPAT_DIR_INDEX		0x0020
1434 
1435 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER	0x0001
1436 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE	0x0002
1437 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR	0x0004
1438 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE        0x0008
1439 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM		0x0010
1440 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK	0x0020
1441 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE	0x0040
1442 #define EXT4_FEATURE_RO_COMPAT_QUOTA		0x0100
1443 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC		0x0200
1444 /*
1445  * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM).  When
1446  * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1447  * all other data structures' checksums.  However, the METADATA_CSUM and
1448  * GDT_CSUM bits are mutually exclusive.
1449  */
1450 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM	0x0400
1451 
1452 #define EXT4_FEATURE_INCOMPAT_COMPRESSION	0x0001
1453 #define EXT4_FEATURE_INCOMPAT_FILETYPE		0x0002
1454 #define EXT4_FEATURE_INCOMPAT_RECOVER		0x0004 /* Needs recovery */
1455 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV	0x0008 /* Journal device */
1456 #define EXT4_FEATURE_INCOMPAT_META_BG		0x0010
1457 #define EXT4_FEATURE_INCOMPAT_EXTENTS		0x0040 /* extents support */
1458 #define EXT4_FEATURE_INCOMPAT_64BIT		0x0080
1459 #define EXT4_FEATURE_INCOMPAT_MMP               0x0100
1460 #define EXT4_FEATURE_INCOMPAT_FLEX_BG		0x0200
1461 #define EXT4_FEATURE_INCOMPAT_EA_INODE		0x0400 /* EA in inode */
1462 #define EXT4_FEATURE_INCOMPAT_DIRDATA		0x1000 /* data in dirent */
1463 #define EXT4_FEATURE_INCOMPAT_BG_USE_META_CSUM	0x2000 /* use crc32c for bg */
1464 #define EXT4_FEATURE_INCOMPAT_LARGEDIR		0x4000 /* >2GB or 3-lvl htree */
1465 #define EXT4_FEATURE_INCOMPAT_INLINEDATA	0x8000 /* data in inode */
1466 
1467 #define EXT2_FEATURE_COMPAT_SUPP	EXT4_FEATURE_COMPAT_EXT_ATTR
1468 #define EXT2_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
1469 					 EXT4_FEATURE_INCOMPAT_META_BG)
1470 #define EXT2_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1471 					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1472 					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1473 
1474 #define EXT3_FEATURE_COMPAT_SUPP	EXT4_FEATURE_COMPAT_EXT_ATTR
1475 #define EXT3_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
1476 					 EXT4_FEATURE_INCOMPAT_RECOVER| \
1477 					 EXT4_FEATURE_INCOMPAT_META_BG)
1478 #define EXT3_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1479 					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1480 					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1481 
1482 #define EXT4_FEATURE_COMPAT_SUPP	EXT2_FEATURE_COMPAT_EXT_ATTR
1483 #define EXT4_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
1484 					 EXT4_FEATURE_INCOMPAT_RECOVER| \
1485 					 EXT4_FEATURE_INCOMPAT_META_BG| \
1486 					 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1487 					 EXT4_FEATURE_INCOMPAT_64BIT| \
1488 					 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1489 					 EXT4_FEATURE_INCOMPAT_MMP)
1490 #define EXT4_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1491 					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1492 					 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1493 					 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1494 					 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1495 					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1496 					 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1497 					 EXT4_FEATURE_RO_COMPAT_BIGALLOC)
1498 
1499 /*
1500  * Default values for user and/or group using reserved blocks
1501  */
1502 #define	EXT4_DEF_RESUID		0
1503 #define	EXT4_DEF_RESGID		0
1504 
1505 #define EXT4_DEF_INODE_READAHEAD_BLKS	32
1506 
1507 /*
1508  * Default mount options
1509  */
1510 #define EXT4_DEFM_DEBUG		0x0001
1511 #define EXT4_DEFM_BSDGROUPS	0x0002
1512 #define EXT4_DEFM_XATTR_USER	0x0004
1513 #define EXT4_DEFM_ACL		0x0008
1514 #define EXT4_DEFM_UID16		0x0010
1515 #define EXT4_DEFM_JMODE		0x0060
1516 #define EXT4_DEFM_JMODE_DATA	0x0020
1517 #define EXT4_DEFM_JMODE_ORDERED	0x0040
1518 #define EXT4_DEFM_JMODE_WBACK	0x0060
1519 #define EXT4_DEFM_NOBARRIER	0x0100
1520 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1521 #define EXT4_DEFM_DISCARD	0x0400
1522 #define EXT4_DEFM_NODELALLOC	0x0800
1523 
1524 /*
1525  * Default journal batch times
1526  */
1527 #define EXT4_DEF_MIN_BATCH_TIME	0
1528 #define EXT4_DEF_MAX_BATCH_TIME	15000 /* 15ms */
1529 
1530 /*
1531  * Minimum number of groups in a flexgroup before we separate out
1532  * directories into the first block group of a flexgroup
1533  */
1534 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME	4
1535 
1536 /*
1537  * Structure of a directory entry
1538  */
1539 #define EXT4_NAME_LEN 255
1540 
1541 struct ext4_dir_entry {
1542 	__le32	inode;			/* Inode number */
1543 	__le16	rec_len;		/* Directory entry length */
1544 	__le16	name_len;		/* Name length */
1545 	char	name[EXT4_NAME_LEN];	/* File name */
1546 };
1547 
1548 /*
1549  * The new version of the directory entry.  Since EXT4 structures are
1550  * stored in intel byte order, and the name_len field could never be
1551  * bigger than 255 chars, it's safe to reclaim the extra byte for the
1552  * file_type field.
1553  */
1554 struct ext4_dir_entry_2 {
1555 	__le32	inode;			/* Inode number */
1556 	__le16	rec_len;		/* Directory entry length */
1557 	__u8	name_len;		/* Name length */
1558 	__u8	file_type;
1559 	char	name[EXT4_NAME_LEN];	/* File name */
1560 };
1561 
1562 /*
1563  * This is a bogus directory entry at the end of each leaf block that
1564  * records checksums.
1565  */
1566 struct ext4_dir_entry_tail {
1567 	__le32	det_reserved_zero1;	/* Pretend to be unused */
1568 	__le16	det_rec_len;		/* 12 */
1569 	__u8	det_reserved_zero2;	/* Zero name length */
1570 	__u8	det_reserved_ft;	/* 0xDE, fake file type */
1571 	__le32	det_checksum;		/* crc32c(uuid+inum+dirblock) */
1572 };
1573 
1574 /*
1575  * Ext4 directory file types.  Only the low 3 bits are used.  The
1576  * other bits are reserved for now.
1577  */
1578 #define EXT4_FT_UNKNOWN		0
1579 #define EXT4_FT_REG_FILE	1
1580 #define EXT4_FT_DIR		2
1581 #define EXT4_FT_CHRDEV		3
1582 #define EXT4_FT_BLKDEV		4
1583 #define EXT4_FT_FIFO		5
1584 #define EXT4_FT_SOCK		6
1585 #define EXT4_FT_SYMLINK		7
1586 
1587 #define EXT4_FT_MAX		8
1588 
1589 #define EXT4_FT_DIR_CSUM	0xDE
1590 
1591 /*
1592  * EXT4_DIR_PAD defines the directory entries boundaries
1593  *
1594  * NOTE: It must be a multiple of 4
1595  */
1596 #define EXT4_DIR_PAD			4
1597 #define EXT4_DIR_ROUND			(EXT4_DIR_PAD - 1)
1598 #define EXT4_DIR_REC_LEN(name_len)	(((name_len) + 8 + EXT4_DIR_ROUND) & \
1599 					 ~EXT4_DIR_ROUND)
1600 #define EXT4_MAX_REC_LEN		((1<<16)-1)
1601 
1602 /*
1603  * If we ever get support for fs block sizes > page_size, we'll need
1604  * to remove the #if statements in the next two functions...
1605  */
1606 static inline unsigned int
1607 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1608 {
1609 	unsigned len = le16_to_cpu(dlen);
1610 
1611 #if (PAGE_CACHE_SIZE >= 65536)
1612 	if (len == EXT4_MAX_REC_LEN || len == 0)
1613 		return blocksize;
1614 	return (len & 65532) | ((len & 3) << 16);
1615 #else
1616 	return len;
1617 #endif
1618 }
1619 
1620 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1621 {
1622 	if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1623 		BUG();
1624 #if (PAGE_CACHE_SIZE >= 65536)
1625 	if (len < 65536)
1626 		return cpu_to_le16(len);
1627 	if (len == blocksize) {
1628 		if (blocksize == 65536)
1629 			return cpu_to_le16(EXT4_MAX_REC_LEN);
1630 		else
1631 			return cpu_to_le16(0);
1632 	}
1633 	return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1634 #else
1635 	return cpu_to_le16(len);
1636 #endif
1637 }
1638 
1639 /*
1640  * Hash Tree Directory indexing
1641  * (c) Daniel Phillips, 2001
1642  */
1643 
1644 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1645 				      EXT4_FEATURE_COMPAT_DIR_INDEX) && \
1646 		    ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
1647 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1648 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1649 
1650 /* Legal values for the dx_root hash_version field: */
1651 
1652 #define DX_HASH_LEGACY		0
1653 #define DX_HASH_HALF_MD4	1
1654 #define DX_HASH_TEA		2
1655 #define DX_HASH_LEGACY_UNSIGNED	3
1656 #define DX_HASH_HALF_MD4_UNSIGNED	4
1657 #define DX_HASH_TEA_UNSIGNED		5
1658 
1659 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
1660 			      const void *address, unsigned int length)
1661 {
1662 	struct {
1663 		struct shash_desc shash;
1664 		char ctx[crypto_shash_descsize(sbi->s_chksum_driver)];
1665 	} desc;
1666 	int err;
1667 
1668 	desc.shash.tfm = sbi->s_chksum_driver;
1669 	desc.shash.flags = 0;
1670 	*(u32 *)desc.ctx = crc;
1671 
1672 	err = crypto_shash_update(&desc.shash, address, length);
1673 	BUG_ON(err);
1674 
1675 	return *(u32 *)desc.ctx;
1676 }
1677 
1678 #ifdef __KERNEL__
1679 
1680 /* hash info structure used by the directory hash */
1681 struct dx_hash_info
1682 {
1683 	u32		hash;
1684 	u32		minor_hash;
1685 	int		hash_version;
1686 	u32		*seed;
1687 };
1688 
1689 
1690 /* 32 and 64 bit signed EOF for dx directories */
1691 #define EXT4_HTREE_EOF_32BIT   ((1UL  << (32 - 1)) - 1)
1692 #define EXT4_HTREE_EOF_64BIT   ((1ULL << (64 - 1)) - 1)
1693 
1694 
1695 /*
1696  * Control parameters used by ext4_htree_next_block
1697  */
1698 #define HASH_NB_ALWAYS		1
1699 
1700 
1701 /*
1702  * Describe an inode's exact location on disk and in memory
1703  */
1704 struct ext4_iloc
1705 {
1706 	struct buffer_head *bh;
1707 	unsigned long offset;
1708 	ext4_group_t block_group;
1709 };
1710 
1711 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
1712 {
1713 	return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
1714 }
1715 
1716 /*
1717  * This structure is stuffed into the struct file's private_data field
1718  * for directories.  It is where we put information so that we can do
1719  * readdir operations in hash tree order.
1720  */
1721 struct dir_private_info {
1722 	struct rb_root	root;
1723 	struct rb_node	*curr_node;
1724 	struct fname	*extra_fname;
1725 	loff_t		last_pos;
1726 	__u32		curr_hash;
1727 	__u32		curr_minor_hash;
1728 	__u32		next_hash;
1729 };
1730 
1731 /* calculate the first block number of the group */
1732 static inline ext4_fsblk_t
1733 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
1734 {
1735 	return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
1736 		le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
1737 }
1738 
1739 /*
1740  * Special error return code only used by dx_probe() and its callers.
1741  */
1742 #define ERR_BAD_DX_DIR	-75000
1743 
1744 void ext4_get_group_no_and_offset(struct super_block *sb, ext4_fsblk_t blocknr,
1745 			ext4_group_t *blockgrpp, ext4_grpblk_t *offsetp);
1746 
1747 /*
1748  * Timeout and state flag for lazy initialization inode thread.
1749  */
1750 #define EXT4_DEF_LI_WAIT_MULT			10
1751 #define EXT4_DEF_LI_MAX_START_DELAY		5
1752 #define EXT4_LAZYINIT_QUIT			0x0001
1753 #define EXT4_LAZYINIT_RUNNING			0x0002
1754 
1755 /*
1756  * Lazy inode table initialization info
1757  */
1758 struct ext4_lazy_init {
1759 	unsigned long		li_state;
1760 	struct list_head	li_request_list;
1761 	struct mutex		li_list_mtx;
1762 };
1763 
1764 struct ext4_li_request {
1765 	struct super_block	*lr_super;
1766 	struct ext4_sb_info	*lr_sbi;
1767 	ext4_group_t		lr_next_group;
1768 	struct list_head	lr_request;
1769 	unsigned long		lr_next_sched;
1770 	unsigned long		lr_timeout;
1771 };
1772 
1773 struct ext4_features {
1774 	struct kobject f_kobj;
1775 	struct completion f_kobj_unregister;
1776 };
1777 
1778 /*
1779  * This structure will be used for multiple mount protection. It will be
1780  * written into the block number saved in the s_mmp_block field in the
1781  * superblock. Programs that check MMP should assume that if
1782  * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
1783  * to use the filesystem, regardless of how old the timestamp is.
1784  */
1785 #define EXT4_MMP_MAGIC     0x004D4D50U /* ASCII for MMP */
1786 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
1787 #define EXT4_MMP_SEQ_FSCK  0xE24D4D50U /* mmp_seq value when being fscked */
1788 #define EXT4_MMP_SEQ_MAX   0xE24D4D4FU /* maximum valid mmp_seq value */
1789 
1790 struct mmp_struct {
1791 	__le32	mmp_magic;		/* Magic number for MMP */
1792 	__le32	mmp_seq;		/* Sequence no. updated periodically */
1793 
1794 	/*
1795 	 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
1796 	 * purposes and do not affect the correctness of the algorithm
1797 	 */
1798 	__le64	mmp_time;		/* Time last updated */
1799 	char	mmp_nodename[64];	/* Node which last updated MMP block */
1800 	char	mmp_bdevname[32];	/* Bdev which last updated MMP block */
1801 
1802 	/*
1803 	 * mmp_check_interval is used to verify if the MMP block has been
1804 	 * updated on the block device. The value is updated based on the
1805 	 * maximum time to write the MMP block during an update cycle.
1806 	 */
1807 	__le16	mmp_check_interval;
1808 
1809 	__le16	mmp_pad1;
1810 	__le32	mmp_pad2[226];
1811 	__le32	mmp_checksum;		/* crc32c(uuid+mmp_block) */
1812 };
1813 
1814 /* arguments passed to the mmp thread */
1815 struct mmpd_data {
1816 	struct buffer_head *bh; /* bh from initial read_mmp_block() */
1817 	struct super_block *sb;  /* super block of the fs */
1818 };
1819 
1820 /*
1821  * Check interval multiplier
1822  * The MMP block is written every update interval and initially checked every
1823  * update interval x the multiplier (the value is then adapted based on the
1824  * write latency). The reason is that writes can be delayed under load and we
1825  * don't want readers to incorrectly assume that the filesystem is no longer
1826  * in use.
1827  */
1828 #define EXT4_MMP_CHECK_MULT		2UL
1829 
1830 /*
1831  * Minimum interval for MMP checking in seconds.
1832  */
1833 #define EXT4_MMP_MIN_CHECK_INTERVAL	5UL
1834 
1835 /*
1836  * Maximum interval for MMP checking in seconds.
1837  */
1838 #define EXT4_MMP_MAX_CHECK_INTERVAL	300UL
1839 
1840 /*
1841  * Function prototypes
1842  */
1843 
1844 /*
1845  * Ok, these declarations are also in <linux/kernel.h> but none of the
1846  * ext4 source programs needs to include it so they are duplicated here.
1847  */
1848 # define NORET_TYPE	/**/
1849 # define ATTRIB_NORET	__attribute__((noreturn))
1850 # define NORET_AND	noreturn,
1851 
1852 /* bitmap.c */
1853 extern unsigned int ext4_count_free(struct buffer_head *, unsigned);
1854 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
1855 				struct ext4_group_desc *gdp,
1856 				struct buffer_head *bh, int sz);
1857 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
1858 				  struct ext4_group_desc *gdp,
1859 				  struct buffer_head *bh, int sz);
1860 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
1861 				struct ext4_group_desc *gdp,
1862 				struct buffer_head *bh, int sz);
1863 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
1864 				  struct ext4_group_desc *gdp,
1865 				  struct buffer_head *bh, int sz);
1866 
1867 /* balloc.c */
1868 extern void ext4_validate_block_bitmap(struct super_block *sb,
1869 				       struct ext4_group_desc *desc,
1870 				       unsigned int block_group,
1871 				       struct buffer_head *bh);
1872 extern unsigned int ext4_block_group(struct super_block *sb,
1873 			ext4_fsblk_t blocknr);
1874 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
1875 			ext4_fsblk_t blocknr);
1876 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
1877 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
1878 			ext4_group_t group);
1879 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
1880 					 ext4_fsblk_t goal,
1881 					 unsigned int flags,
1882 					 unsigned long *count,
1883 					 int *errp);
1884 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
1885 				    s64 nclusters, unsigned int flags);
1886 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
1887 extern void ext4_check_blocks_bitmap(struct super_block *);
1888 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
1889 						    ext4_group_t block_group,
1890 						    struct buffer_head ** bh);
1891 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
1892 
1893 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
1894 						ext4_group_t block_group);
1895 extern int ext4_wait_block_bitmap(struct super_block *sb,
1896 				  ext4_group_t block_group,
1897 				  struct buffer_head *bh);
1898 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
1899 						  ext4_group_t block_group);
1900 extern void ext4_init_block_bitmap(struct super_block *sb,
1901 				   struct buffer_head *bh,
1902 				   ext4_group_t group,
1903 				   struct ext4_group_desc *desc);
1904 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
1905 					      ext4_group_t block_group,
1906 					      struct ext4_group_desc *gdp);
1907 extern unsigned ext4_num_overhead_clusters(struct super_block *sb,
1908 					   ext4_group_t block_group,
1909 					   struct ext4_group_desc *gdp);
1910 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
1911 
1912 /* dir.c */
1913 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
1914 				  struct file *,
1915 				  struct ext4_dir_entry_2 *,
1916 				  struct buffer_head *, unsigned int);
1917 #define ext4_check_dir_entry(dir, filp, de, bh, offset)			\
1918 	unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
1919 					(de), (bh), (offset)))
1920 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
1921 				    __u32 minor_hash,
1922 				    struct ext4_dir_entry_2 *dirent);
1923 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
1924 
1925 /* fsync.c */
1926 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
1927 extern int ext4_flush_completed_IO(struct inode *);
1928 
1929 /* hash.c */
1930 extern int ext4fs_dirhash(const char *name, int len, struct
1931 			  dx_hash_info *hinfo);
1932 
1933 /* ialloc.c */
1934 extern struct inode *ext4_new_inode(handle_t *, struct inode *, umode_t,
1935 				    const struct qstr *qstr, __u32 goal,
1936 				    uid_t *owner);
1937 extern void ext4_free_inode(handle_t *, struct inode *);
1938 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
1939 extern unsigned long ext4_count_free_inodes(struct super_block *);
1940 extern unsigned long ext4_count_dirs(struct super_block *);
1941 extern void ext4_check_inodes_bitmap(struct super_block *);
1942 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
1943 extern int ext4_init_inode_table(struct super_block *sb,
1944 				 ext4_group_t group, int barrier);
1945 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
1946 
1947 /* mballoc.c */
1948 extern long ext4_mb_stats;
1949 extern long ext4_mb_max_to_scan;
1950 extern int ext4_mb_init(struct super_block *, int);
1951 extern int ext4_mb_release(struct super_block *);
1952 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
1953 				struct ext4_allocation_request *, int *);
1954 extern int ext4_mb_reserve_blocks(struct super_block *, int);
1955 extern void ext4_discard_preallocations(struct inode *);
1956 extern int __init ext4_init_mballoc(void);
1957 extern void ext4_exit_mballoc(void);
1958 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
1959 			     struct buffer_head *bh, ext4_fsblk_t block,
1960 			     unsigned long count, int flags);
1961 extern int ext4_mb_add_groupinfo(struct super_block *sb,
1962 		ext4_group_t i, struct ext4_group_desc *desc);
1963 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
1964 				ext4_fsblk_t block, unsigned long count);
1965 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
1966 
1967 /* inode.c */
1968 struct buffer_head *ext4_getblk(handle_t *, struct inode *,
1969 						ext4_lblk_t, int, int *);
1970 struct buffer_head *ext4_bread(handle_t *, struct inode *,
1971 						ext4_lblk_t, int, int *);
1972 int ext4_get_block(struct inode *inode, sector_t iblock,
1973 				struct buffer_head *bh_result, int create);
1974 
1975 extern struct inode *ext4_iget(struct super_block *, unsigned long);
1976 extern int  ext4_write_inode(struct inode *, struct writeback_control *);
1977 extern int  ext4_setattr(struct dentry *, struct iattr *);
1978 extern int  ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
1979 				struct kstat *stat);
1980 extern void ext4_evict_inode(struct inode *);
1981 extern void ext4_clear_inode(struct inode *);
1982 extern int  ext4_sync_inode(handle_t *, struct inode *);
1983 extern void ext4_dirty_inode(struct inode *, int);
1984 extern int ext4_change_inode_journal_flag(struct inode *, int);
1985 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
1986 extern int ext4_can_truncate(struct inode *inode);
1987 extern void ext4_truncate(struct inode *);
1988 extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length);
1989 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
1990 extern void ext4_set_inode_flags(struct inode *);
1991 extern void ext4_get_inode_flags(struct ext4_inode_info *);
1992 extern int ext4_alloc_da_blocks(struct inode *inode);
1993 extern void ext4_set_aops(struct inode *inode);
1994 extern int ext4_writepage_trans_blocks(struct inode *);
1995 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
1996 extern int ext4_discard_partial_page_buffers(handle_t *handle,
1997 		struct address_space *mapping, loff_t from,
1998 		loff_t length, int flags);
1999 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
2000 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2001 extern void ext4_da_update_reserve_space(struct inode *inode,
2002 					int used, int quota_claim);
2003 
2004 /* indirect.c */
2005 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2006 				struct ext4_map_blocks *map, int flags);
2007 extern ssize_t ext4_ind_direct_IO(int rw, struct kiocb *iocb,
2008 				const struct iovec *iov, loff_t offset,
2009 				unsigned long nr_segs);
2010 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2011 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks, int chunk);
2012 extern void ext4_ind_truncate(struct inode *inode);
2013 
2014 /* ioctl.c */
2015 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2016 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2017 
2018 /* migrate.c */
2019 extern int ext4_ext_migrate(struct inode *);
2020 
2021 /* namei.c */
2022 extern int ext4_dirent_csum_verify(struct inode *inode,
2023 				   struct ext4_dir_entry *dirent);
2024 extern int ext4_orphan_add(handle_t *, struct inode *);
2025 extern int ext4_orphan_del(handle_t *, struct inode *);
2026 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2027 				__u32 start_minor_hash, __u32 *next_hash);
2028 
2029 /* resize.c */
2030 extern int ext4_group_add(struct super_block *sb,
2031 				struct ext4_new_group_data *input);
2032 extern int ext4_group_extend(struct super_block *sb,
2033 				struct ext4_super_block *es,
2034 				ext4_fsblk_t n_blocks_count);
2035 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2036 
2037 /* super.c */
2038 extern int ext4_superblock_csum_verify(struct super_block *sb,
2039 				       struct ext4_super_block *es);
2040 extern void ext4_superblock_csum_set(struct super_block *sb,
2041 				     struct ext4_super_block *es);
2042 extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2043 extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2044 extern void ext4_kvfree(void *ptr);
2045 extern __printf(4, 5)
2046 void __ext4_error(struct super_block *, const char *, unsigned int,
2047 		  const char *, ...);
2048 #define ext4_error(sb, message...)	__ext4_error(sb, __func__,	\
2049 						     __LINE__, ## message)
2050 extern __printf(5, 6)
2051 void ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2052 		      const char *, ...);
2053 extern __printf(5, 6)
2054 void ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2055 		     const char *, ...);
2056 extern void __ext4_std_error(struct super_block *, const char *,
2057 			     unsigned int, int);
2058 extern __printf(4, 5)
2059 void __ext4_abort(struct super_block *, const char *, unsigned int,
2060 		  const char *, ...);
2061 #define ext4_abort(sb, message...)	__ext4_abort(sb, __func__, \
2062 						       __LINE__, ## message)
2063 extern __printf(4, 5)
2064 void __ext4_warning(struct super_block *, const char *, unsigned int,
2065 		    const char *, ...);
2066 #define ext4_warning(sb, message...)	__ext4_warning(sb, __func__, \
2067 						       __LINE__, ## message)
2068 extern __printf(3, 4)
2069 void ext4_msg(struct super_block *, const char *, const char *, ...);
2070 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2071 			   const char *, unsigned int, const char *);
2072 #define dump_mmp_msg(sb, mmp, msg)	__dump_mmp_msg(sb, mmp, __func__, \
2073 						       __LINE__, msg)
2074 extern __printf(7, 8)
2075 void __ext4_grp_locked_error(const char *, unsigned int,
2076 			     struct super_block *, ext4_group_t,
2077 			     unsigned long, ext4_fsblk_t,
2078 			     const char *, ...);
2079 #define ext4_grp_locked_error(sb, grp, message...) \
2080 	__ext4_grp_locked_error(__func__, __LINE__, (sb), (grp), ## message)
2081 extern void ext4_update_dynamic_rev(struct super_block *sb);
2082 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2083 					__u32 compat);
2084 extern int ext4_update_rocompat_feature(handle_t *handle,
2085 					struct super_block *sb,	__u32 rocompat);
2086 extern int ext4_update_incompat_feature(handle_t *handle,
2087 					struct super_block *sb,	__u32 incompat);
2088 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2089 				      struct ext4_group_desc *bg);
2090 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2091 				      struct ext4_group_desc *bg);
2092 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2093 				     struct ext4_group_desc *bg);
2094 extern __u32 ext4_free_group_clusters(struct super_block *sb,
2095 				      struct ext4_group_desc *bg);
2096 extern __u32 ext4_free_inodes_count(struct super_block *sb,
2097 				 struct ext4_group_desc *bg);
2098 extern __u32 ext4_used_dirs_count(struct super_block *sb,
2099 				struct ext4_group_desc *bg);
2100 extern __u32 ext4_itable_unused_count(struct super_block *sb,
2101 				   struct ext4_group_desc *bg);
2102 extern void ext4_block_bitmap_set(struct super_block *sb,
2103 				  struct ext4_group_desc *bg, ext4_fsblk_t blk);
2104 extern void ext4_inode_bitmap_set(struct super_block *sb,
2105 				  struct ext4_group_desc *bg, ext4_fsblk_t blk);
2106 extern void ext4_inode_table_set(struct super_block *sb,
2107 				 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2108 extern void ext4_free_group_clusters_set(struct super_block *sb,
2109 					 struct ext4_group_desc *bg,
2110 					 __u32 count);
2111 extern void ext4_free_inodes_set(struct super_block *sb,
2112 				struct ext4_group_desc *bg, __u32 count);
2113 extern void ext4_used_dirs_set(struct super_block *sb,
2114 				struct ext4_group_desc *bg, __u32 count);
2115 extern void ext4_itable_unused_set(struct super_block *sb,
2116 				   struct ext4_group_desc *bg, __u32 count);
2117 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2118 				       struct ext4_group_desc *gdp);
2119 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2120 				     struct ext4_group_desc *gdp);
2121 
2122 static inline int ext4_has_group_desc_csum(struct super_block *sb)
2123 {
2124 	return EXT4_HAS_RO_COMPAT_FEATURE(sb,
2125 					  EXT4_FEATURE_RO_COMPAT_GDT_CSUM |
2126 					  EXT4_FEATURE_RO_COMPAT_METADATA_CSUM);
2127 }
2128 
2129 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2130 {
2131 	return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2132 		le32_to_cpu(es->s_blocks_count_lo);
2133 }
2134 
2135 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2136 {
2137 	return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2138 		le32_to_cpu(es->s_r_blocks_count_lo);
2139 }
2140 
2141 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2142 {
2143 	return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2144 		le32_to_cpu(es->s_free_blocks_count_lo);
2145 }
2146 
2147 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2148 					 ext4_fsblk_t blk)
2149 {
2150 	es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2151 	es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2152 }
2153 
2154 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2155 					      ext4_fsblk_t blk)
2156 {
2157 	es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2158 	es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2159 }
2160 
2161 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2162 					   ext4_fsblk_t blk)
2163 {
2164 	es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2165 	es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2166 }
2167 
2168 static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
2169 {
2170 	if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2171 		return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2172 			le32_to_cpu(raw_inode->i_size_lo);
2173 	else
2174 		return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2175 }
2176 
2177 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2178 {
2179 	raw_inode->i_size_lo = cpu_to_le32(i_size);
2180 	raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2181 }
2182 
2183 static inline
2184 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2185 					    ext4_group_t group)
2186 {
2187 	 struct ext4_group_info ***grp_info;
2188 	 long indexv, indexh;
2189 	 grp_info = EXT4_SB(sb)->s_group_info;
2190 	 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2191 	 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2192 	 return grp_info[indexv][indexh];
2193 }
2194 
2195 /*
2196  * Reading s_groups_count requires using smp_rmb() afterwards.  See
2197  * the locking protocol documented in the comments of ext4_group_add()
2198  * in resize.c
2199  */
2200 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2201 {
2202 	ext4_group_t	ngroups = EXT4_SB(sb)->s_groups_count;
2203 
2204 	smp_rmb();
2205 	return ngroups;
2206 }
2207 
2208 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2209 					     ext4_group_t block_group)
2210 {
2211 	return block_group >> sbi->s_log_groups_per_flex;
2212 }
2213 
2214 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2215 {
2216 	return 1 << sbi->s_log_groups_per_flex;
2217 }
2218 
2219 #define ext4_std_error(sb, errno)				\
2220 do {								\
2221 	if ((errno))						\
2222 		__ext4_std_error((sb), __func__, __LINE__, (errno));	\
2223 } while (0)
2224 
2225 #ifdef CONFIG_SMP
2226 /* Each CPU can accumulate percpu_counter_batch clusters in their local
2227  * counters. So we need to make sure we have free clusters more
2228  * than percpu_counter_batch  * nr_cpu_ids. Also add a window of 4 times.
2229  */
2230 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2231 #else
2232 #define EXT4_FREECLUSTERS_WATERMARK 0
2233 #endif
2234 
2235 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2236 {
2237 	/*
2238 	 * XXX: replace with spinlock if seen contended -bzzz
2239 	 */
2240 	down_write(&EXT4_I(inode)->i_data_sem);
2241 	if (newsize > EXT4_I(inode)->i_disksize)
2242 		EXT4_I(inode)->i_disksize = newsize;
2243 	up_write(&EXT4_I(inode)->i_data_sem);
2244 	return ;
2245 }
2246 
2247 struct ext4_group_info {
2248 	unsigned long   bb_state;
2249 	struct rb_root  bb_free_root;
2250 	ext4_grpblk_t	bb_first_free;	/* first free block */
2251 	ext4_grpblk_t	bb_free;	/* total free blocks */
2252 	ext4_grpblk_t	bb_fragments;	/* nr of freespace fragments */
2253 	ext4_grpblk_t	bb_largest_free_order;/* order of largest frag in BG */
2254 	struct          list_head bb_prealloc_list;
2255 #ifdef DOUBLE_CHECK
2256 	void            *bb_bitmap;
2257 #endif
2258 	struct rw_semaphore alloc_sem;
2259 	ext4_grpblk_t	bb_counters[];	/* Nr of free power-of-two-block
2260 					 * regions, index is order.
2261 					 * bb_counters[3] = 5 means
2262 					 * 5 free 8-block regions. */
2263 };
2264 
2265 #define EXT4_GROUP_INFO_NEED_INIT_BIT		0
2266 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT		1
2267 
2268 #define EXT4_MB_GRP_NEED_INIT(grp)	\
2269 	(test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2270 
2271 #define EXT4_MB_GRP_WAS_TRIMMED(grp)	\
2272 	(test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2273 #define EXT4_MB_GRP_SET_TRIMMED(grp)	\
2274 	(set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2275 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp)	\
2276 	(clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2277 
2278 #define EXT4_MAX_CONTENTION		8
2279 #define EXT4_CONTENTION_THRESHOLD	2
2280 
2281 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2282 					      ext4_group_t group)
2283 {
2284 	return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2285 }
2286 
2287 /*
2288  * Returns true if the filesystem is busy enough that attempts to
2289  * access the block group locks has run into contention.
2290  */
2291 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2292 {
2293 	return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2294 }
2295 
2296 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2297 {
2298 	spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2299 	if (spin_trylock(lock))
2300 		/*
2301 		 * We're able to grab the lock right away, so drop the
2302 		 * lock contention counter.
2303 		 */
2304 		atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2305 	else {
2306 		/*
2307 		 * The lock is busy, so bump the contention counter,
2308 		 * and then wait on the spin lock.
2309 		 */
2310 		atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2311 				  EXT4_MAX_CONTENTION);
2312 		spin_lock(lock);
2313 	}
2314 }
2315 
2316 static inline void ext4_unlock_group(struct super_block *sb,
2317 					ext4_group_t group)
2318 {
2319 	spin_unlock(ext4_group_lock_ptr(sb, group));
2320 }
2321 
2322 static inline void ext4_mark_super_dirty(struct super_block *sb)
2323 {
2324 	struct ext4_super_block *es = EXT4_SB(sb)->s_es;
2325 
2326 	ext4_superblock_csum_set(sb, es);
2327 	if (EXT4_SB(sb)->s_journal == NULL)
2328 		sb->s_dirt =1;
2329 }
2330 
2331 /*
2332  * Block validity checking
2333  */
2334 #define ext4_check_indirect_blockref(inode, bh)				\
2335 	ext4_check_blockref(__func__, __LINE__, inode,			\
2336 			    (__le32 *)(bh)->b_data,			\
2337 			    EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2338 
2339 #define ext4_ind_check_inode(inode)					\
2340 	ext4_check_blockref(__func__, __LINE__, inode,			\
2341 			    EXT4_I(inode)->i_data,			\
2342 			    EXT4_NDIR_BLOCKS)
2343 
2344 /*
2345  * Inodes and files operations
2346  */
2347 
2348 /* dir.c */
2349 extern const struct file_operations ext4_dir_operations;
2350 
2351 /* file.c */
2352 extern const struct inode_operations ext4_file_inode_operations;
2353 extern const struct file_operations ext4_file_operations;
2354 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
2355 
2356 /* namei.c */
2357 extern const struct inode_operations ext4_dir_inode_operations;
2358 extern const struct inode_operations ext4_special_inode_operations;
2359 extern struct dentry *ext4_get_parent(struct dentry *child);
2360 
2361 /* symlink.c */
2362 extern const struct inode_operations ext4_symlink_inode_operations;
2363 extern const struct inode_operations ext4_fast_symlink_inode_operations;
2364 
2365 /* block_validity */
2366 extern void ext4_release_system_zone(struct super_block *sb);
2367 extern int ext4_setup_system_zone(struct super_block *sb);
2368 extern int __init ext4_init_system_zone(void);
2369 extern void ext4_exit_system_zone(void);
2370 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
2371 				 ext4_fsblk_t start_blk,
2372 				 unsigned int count);
2373 extern int ext4_check_blockref(const char *, unsigned int,
2374 			       struct inode *, __le32 *, unsigned int);
2375 
2376 /* extents.c */
2377 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
2378 extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
2379 extern int ext4_ext_index_trans_blocks(struct inode *inode, int nrblocks,
2380 				       int chunk);
2381 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
2382 			       struct ext4_map_blocks *map, int flags);
2383 extern void ext4_ext_truncate(struct inode *);
2384 extern int ext4_ext_punch_hole(struct file *file, loff_t offset,
2385 				loff_t length);
2386 extern void ext4_ext_init(struct super_block *);
2387 extern void ext4_ext_release(struct super_block *);
2388 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
2389 			  loff_t len);
2390 extern int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset,
2391 			  ssize_t len);
2392 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
2393 			   struct ext4_map_blocks *map, int flags);
2394 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
2395 			__u64 start, __u64 len);
2396 /* move_extent.c */
2397 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
2398 			     __u64 start_orig, __u64 start_donor,
2399 			     __u64 len, __u64 *moved_len);
2400 
2401 /* page-io.c */
2402 extern int __init ext4_init_pageio(void);
2403 extern void ext4_exit_pageio(void);
2404 extern void ext4_ioend_wait(struct inode *);
2405 extern void ext4_free_io_end(ext4_io_end_t *io);
2406 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
2407 extern int ext4_end_io_nolock(ext4_io_end_t *io);
2408 extern void ext4_io_submit(struct ext4_io_submit *io);
2409 extern int ext4_bio_write_page(struct ext4_io_submit *io,
2410 			       struct page *page,
2411 			       int len,
2412 			       struct writeback_control *wbc);
2413 
2414 /* mmp.c */
2415 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
2416 
2417 /* BH_Uninit flag: blocks are allocated but uninitialized on disk */
2418 enum ext4_state_bits {
2419 	BH_Uninit	/* blocks are allocated but uninitialized on disk */
2420 	  = BH_JBDPrivateStart,
2421 	BH_AllocFromCluster,	/* allocated blocks were part of already
2422 				 * allocated cluster. Note that this flag will
2423 				 * never, ever appear in a buffer_head's state
2424 				 * flag. See EXT4_MAP_FROM_CLUSTER to see where
2425 				 * this is used. */
2426 	BH_Da_Mapped,	/* Delayed allocated block that now has a mapping. This
2427 			 * flag is set when ext4_map_blocks is called on a
2428 			 * delayed allocated block to get its real mapping. */
2429 };
2430 
2431 BUFFER_FNS(Uninit, uninit)
2432 TAS_BUFFER_FNS(Uninit, uninit)
2433 BUFFER_FNS(Da_Mapped, da_mapped)
2434 
2435 /*
2436  * Add new method to test wether block and inode bitmaps are properly
2437  * initialized. With uninit_bg reading the block from disk is not enough
2438  * to mark the bitmap uptodate. We need to also zero-out the bitmap
2439  */
2440 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
2441 
2442 static inline int bitmap_uptodate(struct buffer_head *bh)
2443 {
2444 	return (buffer_uptodate(bh) &&
2445 			test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
2446 }
2447 static inline void set_bitmap_uptodate(struct buffer_head *bh)
2448 {
2449 	set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
2450 }
2451 
2452 #define in_range(b, first, len)	((b) >= (first) && (b) <= (first) + (len) - 1)
2453 
2454 /* For ioend & aio unwritten conversion wait queues */
2455 #define EXT4_WQ_HASH_SZ		37
2456 #define ext4_ioend_wq(v)   (&ext4__ioend_wq[((unsigned long)(v)) %\
2457 					    EXT4_WQ_HASH_SZ])
2458 #define ext4_aio_mutex(v)  (&ext4__aio_mutex[((unsigned long)(v)) %\
2459 					     EXT4_WQ_HASH_SZ])
2460 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
2461 extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ];
2462 
2463 #define EXT4_RESIZING	0
2464 extern int ext4_resize_begin(struct super_block *sb);
2465 extern void ext4_resize_end(struct super_block *sb);
2466 
2467 #endif	/* __KERNEL__ */
2468 
2469 #include "ext4_extents.h"
2470 
2471 #endif	/* _EXT4_H */
2472