xref: /openbmc/linux/fs/ext4/ext4.h (revision 1dd24dae)
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 /*
61  * Turn on EXT_DEBUG to get lots of info about extents operations.
62  */
63 #define EXT_DEBUG__
64 #ifdef EXT_DEBUG
65 #define ext_debug(fmt, ...)	printk(fmt, ##__VA_ARGS__)
66 #else
67 #define ext_debug(fmt, ...)	no_printk(fmt, ##__VA_ARGS__)
68 #endif
69 
70 #define EXT4_ERROR_INODE(inode, fmt, a...) \
71 	ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
72 
73 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...)			\
74 	ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
75 
76 #define EXT4_ERROR_FILE(file, block, fmt, a...)				\
77 	ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
78 
79 /* data type for block offset of block group */
80 typedef int ext4_grpblk_t;
81 
82 /* data type for filesystem-wide blocks number */
83 typedef unsigned long long ext4_fsblk_t;
84 
85 /* data type for file logical block number */
86 typedef __u32 ext4_lblk_t;
87 
88 /* data type for block group number */
89 typedef unsigned int ext4_group_t;
90 
91 /*
92  * Flags used in mballoc's allocation_context flags field.
93  *
94  * Also used to show what's going on for debugging purposes when the
95  * flag field is exported via the traceport interface
96  */
97 
98 /* prefer goal again. length */
99 #define EXT4_MB_HINT_MERGE		0x0001
100 /* blocks already reserved */
101 #define EXT4_MB_HINT_RESERVED		0x0002
102 /* metadata is being allocated */
103 #define EXT4_MB_HINT_METADATA		0x0004
104 /* first blocks in the file */
105 #define EXT4_MB_HINT_FIRST		0x0008
106 /* search for the best chunk */
107 #define EXT4_MB_HINT_BEST		0x0010
108 /* data is being allocated */
109 #define EXT4_MB_HINT_DATA		0x0020
110 /* don't preallocate (for tails) */
111 #define EXT4_MB_HINT_NOPREALLOC		0x0040
112 /* allocate for locality group */
113 #define EXT4_MB_HINT_GROUP_ALLOC	0x0080
114 /* allocate goal blocks or none */
115 #define EXT4_MB_HINT_GOAL_ONLY		0x0100
116 /* goal is meaningful */
117 #define EXT4_MB_HINT_TRY_GOAL		0x0200
118 /* blocks already pre-reserved by delayed allocation */
119 #define EXT4_MB_DELALLOC_RESERVED	0x0400
120 /* We are doing stream allocation */
121 #define EXT4_MB_STREAM_ALLOC		0x0800
122 /* Use reserved root blocks if needed */
123 #define EXT4_MB_USE_ROOT_BLOCKS		0x1000
124 /* Use blocks from reserved pool */
125 #define EXT4_MB_USE_RESERVED		0x2000
126 
127 struct ext4_allocation_request {
128 	/* target inode for block we're allocating */
129 	struct inode *inode;
130 	/* how many blocks we want to allocate */
131 	unsigned int len;
132 	/* logical block in target inode */
133 	ext4_lblk_t logical;
134 	/* the closest logical allocated block to the left */
135 	ext4_lblk_t lleft;
136 	/* the closest logical allocated block to the right */
137 	ext4_lblk_t lright;
138 	/* phys. target (a hint) */
139 	ext4_fsblk_t goal;
140 	/* phys. block for the closest logical allocated block to the left */
141 	ext4_fsblk_t pleft;
142 	/* phys. block for the closest logical allocated block to the right */
143 	ext4_fsblk_t pright;
144 	/* flags. see above EXT4_MB_HINT_* */
145 	unsigned int flags;
146 };
147 
148 /*
149  * Logical to physical block mapping, used by ext4_map_blocks()
150  *
151  * This structure is used to pass requests into ext4_map_blocks() as
152  * well as to store the information returned by ext4_map_blocks().  It
153  * takes less room on the stack than a struct buffer_head.
154  */
155 #define EXT4_MAP_NEW		(1 << BH_New)
156 #define EXT4_MAP_MAPPED		(1 << BH_Mapped)
157 #define EXT4_MAP_UNWRITTEN	(1 << BH_Unwritten)
158 #define EXT4_MAP_BOUNDARY	(1 << BH_Boundary)
159 #define EXT4_MAP_UNINIT		(1 << BH_Uninit)
160 /* Sometimes (in the bigalloc case, from ext4_da_get_block_prep) the caller of
161  * ext4_map_blocks wants to know whether or not the underlying cluster has
162  * already been accounted for. EXT4_MAP_FROM_CLUSTER conveys to the caller that
163  * the requested mapping was from previously mapped (or delayed allocated)
164  * cluster. We use BH_AllocFromCluster only for this flag. BH_AllocFromCluster
165  * should never appear on buffer_head's state flags.
166  */
167 #define EXT4_MAP_FROM_CLUSTER	(1 << BH_AllocFromCluster)
168 #define EXT4_MAP_FLAGS		(EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
169 				 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\
170 				 EXT4_MAP_UNINIT | EXT4_MAP_FROM_CLUSTER)
171 
172 struct ext4_map_blocks {
173 	ext4_fsblk_t m_pblk;
174 	ext4_lblk_t m_lblk;
175 	unsigned int m_len;
176 	unsigned int m_flags;
177 };
178 
179 /*
180  * For delayed allocation tracking
181  */
182 struct mpage_da_data {
183 	struct inode *inode;
184 	sector_t b_blocknr;		/* start block number of extent */
185 	size_t b_size;			/* size of extent */
186 	unsigned long b_state;		/* state of the extent */
187 	unsigned long first_page, next_page;	/* extent of pages */
188 	struct writeback_control *wbc;
189 	int io_done;
190 	int pages_written;
191 	int retval;
192 };
193 
194 /*
195  * Flags for ext4_io_end->flags
196  */
197 #define	EXT4_IO_END_UNWRITTEN	0x0001
198 #define EXT4_IO_END_ERROR	0x0002
199 #define EXT4_IO_END_DIRECT	0x0004
200 
201 /*
202  * For converting uninitialized extents on a work queue.
203  */
204 typedef struct ext4_io_end {
205 	struct list_head	list;		/* per-file finished IO list */
206 	struct inode		*inode;		/* file being written to */
207 	unsigned int		flag;		/* unwritten or not */
208 	loff_t			offset;		/* offset in the file */
209 	ssize_t			size;		/* size of the extent */
210 	struct kiocb		*iocb;		/* iocb struct for AIO */
211 	int			result;		/* error value for AIO */
212 } ext4_io_end_t;
213 
214 struct ext4_io_submit {
215 	int			io_op;
216 	struct bio		*io_bio;
217 	ext4_io_end_t		*io_end;
218 	sector_t		io_next_block;
219 };
220 
221 /*
222  * Special inodes numbers
223  */
224 #define	EXT4_BAD_INO		 1	/* Bad blocks inode */
225 #define EXT4_ROOT_INO		 2	/* Root inode */
226 #define EXT4_USR_QUOTA_INO	 3	/* User quota inode */
227 #define EXT4_GRP_QUOTA_INO	 4	/* Group quota inode */
228 #define EXT4_BOOT_LOADER_INO	 5	/* Boot loader inode */
229 #define EXT4_UNDEL_DIR_INO	 6	/* Undelete directory inode */
230 #define EXT4_RESIZE_INO		 7	/* Reserved group descriptors inode */
231 #define EXT4_JOURNAL_INO	 8	/* Journal inode */
232 
233 /* First non-reserved inode for old ext4 filesystems */
234 #define EXT4_GOOD_OLD_FIRST_INO	11
235 
236 /*
237  * Maximal count of links to a file
238  */
239 #define EXT4_LINK_MAX		65000
240 
241 /*
242  * Macro-instructions used to manage several block sizes
243  */
244 #define EXT4_MIN_BLOCK_SIZE		1024
245 #define	EXT4_MAX_BLOCK_SIZE		65536
246 #define EXT4_MIN_BLOCK_LOG_SIZE		10
247 #define EXT4_MAX_BLOCK_LOG_SIZE		16
248 #ifdef __KERNEL__
249 # define EXT4_BLOCK_SIZE(s)		((s)->s_blocksize)
250 #else
251 # define EXT4_BLOCK_SIZE(s)		(EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
252 #endif
253 #define	EXT4_ADDR_PER_BLOCK(s)		(EXT4_BLOCK_SIZE(s) / sizeof(__u32))
254 #define EXT4_CLUSTER_SIZE(s)		(EXT4_BLOCK_SIZE(s) << \
255 					 EXT4_SB(s)->s_cluster_bits)
256 #ifdef __KERNEL__
257 # define EXT4_BLOCK_SIZE_BITS(s)	((s)->s_blocksize_bits)
258 # define EXT4_CLUSTER_BITS(s)		(EXT4_SB(s)->s_cluster_bits)
259 #else
260 # define EXT4_BLOCK_SIZE_BITS(s)	((s)->s_log_block_size + 10)
261 #endif
262 #ifdef __KERNEL__
263 #define	EXT4_ADDR_PER_BLOCK_BITS(s)	(EXT4_SB(s)->s_addr_per_block_bits)
264 #define EXT4_INODE_SIZE(s)		(EXT4_SB(s)->s_inode_size)
265 #define EXT4_FIRST_INO(s)		(EXT4_SB(s)->s_first_ino)
266 #else
267 #define EXT4_INODE_SIZE(s)	(((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
268 				 EXT4_GOOD_OLD_INODE_SIZE : \
269 				 (s)->s_inode_size)
270 #define EXT4_FIRST_INO(s)	(((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
271 				 EXT4_GOOD_OLD_FIRST_INO : \
272 				 (s)->s_first_ino)
273 #endif
274 #define EXT4_BLOCK_ALIGN(size, blkbits)		ALIGN((size), (1 << (blkbits)))
275 
276 /* Translate a block number to a cluster number */
277 #define EXT4_B2C(sbi, blk)	((blk) >> (sbi)->s_cluster_bits)
278 /* Translate a cluster number to a block number */
279 #define EXT4_C2B(sbi, cluster)	((cluster) << (sbi)->s_cluster_bits)
280 /* Translate # of blks to # of clusters */
281 #define EXT4_NUM_B2C(sbi, blks)	(((blks) + (sbi)->s_cluster_ratio - 1) >> \
282 				 (sbi)->s_cluster_bits)
283 
284 /*
285  * Structure of a blocks group descriptor
286  */
287 struct ext4_group_desc
288 {
289 	__le32	bg_block_bitmap_lo;	/* Blocks bitmap block */
290 	__le32	bg_inode_bitmap_lo;	/* Inodes bitmap block */
291 	__le32	bg_inode_table_lo;	/* Inodes table block */
292 	__le16	bg_free_blocks_count_lo;/* Free blocks count */
293 	__le16	bg_free_inodes_count_lo;/* Free inodes count */
294 	__le16	bg_used_dirs_count_lo;	/* Directories count */
295 	__le16	bg_flags;		/* EXT4_BG_flags (INODE_UNINIT, etc) */
296 	__le32  bg_exclude_bitmap_lo;   /* Exclude bitmap for snapshots */
297 	__le16  bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
298 	__le16  bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
299 	__le16  bg_itable_unused_lo;	/* Unused inodes count */
300 	__le16  bg_checksum;		/* crc16(sb_uuid+group+desc) */
301 	__le32	bg_block_bitmap_hi;	/* Blocks bitmap block MSB */
302 	__le32	bg_inode_bitmap_hi;	/* Inodes bitmap block MSB */
303 	__le32	bg_inode_table_hi;	/* Inodes table block MSB */
304 	__le16	bg_free_blocks_count_hi;/* Free blocks count MSB */
305 	__le16	bg_free_inodes_count_hi;/* Free inodes count MSB */
306 	__le16	bg_used_dirs_count_hi;	/* Directories count MSB */
307 	__le16  bg_itable_unused_hi;    /* Unused inodes count MSB */
308 	__le32  bg_exclude_bitmap_hi;   /* Exclude bitmap block MSB */
309 	__le16  bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
310 	__le16  bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
311 	__u32   bg_reserved;
312 };
313 
314 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END	\
315 	(offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
316 	 sizeof(__le16))
317 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END	\
318 	(offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
319 	 sizeof(__le16))
320 
321 /*
322  * Structure of a flex block group info
323  */
324 
325 struct flex_groups {
326 	atomic64_t	free_clusters;
327 	atomic_t	free_inodes;
328 	atomic_t	used_dirs;
329 };
330 
331 #define EXT4_BG_INODE_UNINIT	0x0001 /* Inode table/bitmap not in use */
332 #define EXT4_BG_BLOCK_UNINIT	0x0002 /* Block bitmap not in use */
333 #define EXT4_BG_INODE_ZEROED	0x0004 /* On-disk itable initialized to zero */
334 
335 /*
336  * Macro-instructions used to manage group descriptors
337  */
338 #define EXT4_MIN_DESC_SIZE		32
339 #define EXT4_MIN_DESC_SIZE_64BIT	64
340 #define	EXT4_MAX_DESC_SIZE		EXT4_MIN_BLOCK_SIZE
341 #define EXT4_DESC_SIZE(s)		(EXT4_SB(s)->s_desc_size)
342 #ifdef __KERNEL__
343 # define EXT4_BLOCKS_PER_GROUP(s)	(EXT4_SB(s)->s_blocks_per_group)
344 # define EXT4_CLUSTERS_PER_GROUP(s)	(EXT4_SB(s)->s_clusters_per_group)
345 # define EXT4_DESC_PER_BLOCK(s)		(EXT4_SB(s)->s_desc_per_block)
346 # define EXT4_INODES_PER_GROUP(s)	(EXT4_SB(s)->s_inodes_per_group)
347 # define EXT4_DESC_PER_BLOCK_BITS(s)	(EXT4_SB(s)->s_desc_per_block_bits)
348 #else
349 # define EXT4_BLOCKS_PER_GROUP(s)	((s)->s_blocks_per_group)
350 # define EXT4_DESC_PER_BLOCK(s)		(EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
351 # define EXT4_INODES_PER_GROUP(s)	((s)->s_inodes_per_group)
352 #endif
353 
354 /*
355  * Constants relative to the data blocks
356  */
357 #define	EXT4_NDIR_BLOCKS		12
358 #define	EXT4_IND_BLOCK			EXT4_NDIR_BLOCKS
359 #define	EXT4_DIND_BLOCK			(EXT4_IND_BLOCK + 1)
360 #define	EXT4_TIND_BLOCK			(EXT4_DIND_BLOCK + 1)
361 #define	EXT4_N_BLOCKS			(EXT4_TIND_BLOCK + 1)
362 
363 /*
364  * Inode flags
365  */
366 #define	EXT4_SECRM_FL			0x00000001 /* Secure deletion */
367 #define	EXT4_UNRM_FL			0x00000002 /* Undelete */
368 #define	EXT4_COMPR_FL			0x00000004 /* Compress file */
369 #define EXT4_SYNC_FL			0x00000008 /* Synchronous updates */
370 #define EXT4_IMMUTABLE_FL		0x00000010 /* Immutable file */
371 #define EXT4_APPEND_FL			0x00000020 /* writes to file may only append */
372 #define EXT4_NODUMP_FL			0x00000040 /* do not dump file */
373 #define EXT4_NOATIME_FL			0x00000080 /* do not update atime */
374 /* Reserved for compression usage... */
375 #define EXT4_DIRTY_FL			0x00000100
376 #define EXT4_COMPRBLK_FL		0x00000200 /* One or more compressed clusters */
377 #define EXT4_NOCOMPR_FL			0x00000400 /* Don't compress */
378 #define EXT4_ECOMPR_FL			0x00000800 /* Compression error */
379 /* End compression flags --- maybe not all used */
380 #define EXT4_INDEX_FL			0x00001000 /* hash-indexed directory */
381 #define EXT4_IMAGIC_FL			0x00002000 /* AFS directory */
382 #define EXT4_JOURNAL_DATA_FL		0x00004000 /* file data should be journaled */
383 #define EXT4_NOTAIL_FL			0x00008000 /* file tail should not be merged */
384 #define EXT4_DIRSYNC_FL			0x00010000 /* dirsync behaviour (directories only) */
385 #define EXT4_TOPDIR_FL			0x00020000 /* Top of directory hierarchies*/
386 #define EXT4_HUGE_FILE_FL               0x00040000 /* Set to each huge file */
387 #define EXT4_EXTENTS_FL			0x00080000 /* Inode uses extents */
388 #define EXT4_EA_INODE_FL	        0x00200000 /* Inode used for large EA */
389 #define EXT4_EOFBLOCKS_FL		0x00400000 /* Blocks allocated beyond EOF */
390 #define EXT4_INLINE_DATA_FL		0x10000000 /* Inode has inline data. */
391 #define EXT4_RESERVED_FL		0x80000000 /* reserved for ext4 lib */
392 
393 #define EXT4_FL_USER_VISIBLE		0x004BDFFF /* User visible flags */
394 #define EXT4_FL_USER_MODIFIABLE		0x004380FF /* User modifiable flags */
395 
396 /* Flags that should be inherited by new inodes from their parent. */
397 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
398 			   EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
399 			   EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
400 			   EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
401 
402 /* Flags that are appropriate for regular files (all but dir-specific ones). */
403 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
404 
405 /* Flags that are appropriate for non-directories/regular files. */
406 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
407 
408 /* Mask out flags that are inappropriate for the given type of inode. */
409 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
410 {
411 	if (S_ISDIR(mode))
412 		return flags;
413 	else if (S_ISREG(mode))
414 		return flags & EXT4_REG_FLMASK;
415 	else
416 		return flags & EXT4_OTHER_FLMASK;
417 }
418 
419 /*
420  * Inode flags used for atomic set/get
421  */
422 enum {
423 	EXT4_INODE_SECRM	= 0,	/* Secure deletion */
424 	EXT4_INODE_UNRM		= 1,	/* Undelete */
425 	EXT4_INODE_COMPR	= 2,	/* Compress file */
426 	EXT4_INODE_SYNC		= 3,	/* Synchronous updates */
427 	EXT4_INODE_IMMUTABLE	= 4,	/* Immutable file */
428 	EXT4_INODE_APPEND	= 5,	/* writes to file may only append */
429 	EXT4_INODE_NODUMP	= 6,	/* do not dump file */
430 	EXT4_INODE_NOATIME	= 7,	/* do not update atime */
431 /* Reserved for compression usage... */
432 	EXT4_INODE_DIRTY	= 8,
433 	EXT4_INODE_COMPRBLK	= 9,	/* One or more compressed clusters */
434 	EXT4_INODE_NOCOMPR	= 10,	/* Don't compress */
435 	EXT4_INODE_ECOMPR	= 11,	/* Compression error */
436 /* End compression flags --- maybe not all used */
437 	EXT4_INODE_INDEX	= 12,	/* hash-indexed directory */
438 	EXT4_INODE_IMAGIC	= 13,	/* AFS directory */
439 	EXT4_INODE_JOURNAL_DATA	= 14,	/* file data should be journaled */
440 	EXT4_INODE_NOTAIL	= 15,	/* file tail should not be merged */
441 	EXT4_INODE_DIRSYNC	= 16,	/* dirsync behaviour (directories only) */
442 	EXT4_INODE_TOPDIR	= 17,	/* Top of directory hierarchies*/
443 	EXT4_INODE_HUGE_FILE	= 18,	/* Set to each huge file */
444 	EXT4_INODE_EXTENTS	= 19,	/* Inode uses extents */
445 	EXT4_INODE_EA_INODE	= 21,	/* Inode used for large EA */
446 	EXT4_INODE_EOFBLOCKS	= 22,	/* Blocks allocated beyond EOF */
447 	EXT4_INODE_INLINE_DATA	= 28,	/* Data in inode. */
448 	EXT4_INODE_RESERVED	= 31,	/* reserved for ext4 lib */
449 };
450 
451 /*
452  * Since it's pretty easy to mix up bit numbers and hex values, we use a
453  * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
454  * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
455  * any extra space in the compiled kernel image, otherwise, the build will fail.
456  * It's important that these values are the same, since we are using
457  * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
458  * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
459  * values found in ext2, ext3 and ext4 filesystems, and of course the values
460  * defined in e2fsprogs.
461  *
462  * It's not paranoia if the Murphy's Law really *is* out to get you.  :-)
463  */
464 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
465 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
466 
467 static inline void ext4_check_flag_values(void)
468 {
469 	CHECK_FLAG_VALUE(SECRM);
470 	CHECK_FLAG_VALUE(UNRM);
471 	CHECK_FLAG_VALUE(COMPR);
472 	CHECK_FLAG_VALUE(SYNC);
473 	CHECK_FLAG_VALUE(IMMUTABLE);
474 	CHECK_FLAG_VALUE(APPEND);
475 	CHECK_FLAG_VALUE(NODUMP);
476 	CHECK_FLAG_VALUE(NOATIME);
477 	CHECK_FLAG_VALUE(DIRTY);
478 	CHECK_FLAG_VALUE(COMPRBLK);
479 	CHECK_FLAG_VALUE(NOCOMPR);
480 	CHECK_FLAG_VALUE(ECOMPR);
481 	CHECK_FLAG_VALUE(INDEX);
482 	CHECK_FLAG_VALUE(IMAGIC);
483 	CHECK_FLAG_VALUE(JOURNAL_DATA);
484 	CHECK_FLAG_VALUE(NOTAIL);
485 	CHECK_FLAG_VALUE(DIRSYNC);
486 	CHECK_FLAG_VALUE(TOPDIR);
487 	CHECK_FLAG_VALUE(HUGE_FILE);
488 	CHECK_FLAG_VALUE(EXTENTS);
489 	CHECK_FLAG_VALUE(EA_INODE);
490 	CHECK_FLAG_VALUE(EOFBLOCKS);
491 	CHECK_FLAG_VALUE(INLINE_DATA);
492 	CHECK_FLAG_VALUE(RESERVED);
493 }
494 
495 /* Used to pass group descriptor data when online resize is done */
496 struct ext4_new_group_input {
497 	__u32 group;		/* Group number for this data */
498 	__u64 block_bitmap;	/* Absolute block number of block bitmap */
499 	__u64 inode_bitmap;	/* Absolute block number of inode bitmap */
500 	__u64 inode_table;	/* Absolute block number of inode table start */
501 	__u32 blocks_count;	/* Total number of blocks in this group */
502 	__u16 reserved_blocks;	/* Number of reserved blocks in this group */
503 	__u16 unused;
504 };
505 
506 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
507 struct compat_ext4_new_group_input {
508 	u32 group;
509 	compat_u64 block_bitmap;
510 	compat_u64 inode_bitmap;
511 	compat_u64 inode_table;
512 	u32 blocks_count;
513 	u16 reserved_blocks;
514 	u16 unused;
515 };
516 #endif
517 
518 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
519 struct ext4_new_group_data {
520 	__u32 group;
521 	__u64 block_bitmap;
522 	__u64 inode_bitmap;
523 	__u64 inode_table;
524 	__u32 blocks_count;
525 	__u16 reserved_blocks;
526 	__u16 unused;
527 	__u32 free_blocks_count;
528 };
529 
530 /* Indexes used to index group tables in ext4_new_group_data */
531 enum {
532 	BLOCK_BITMAP = 0,	/* block bitmap */
533 	INODE_BITMAP,		/* inode bitmap */
534 	INODE_TABLE,		/* inode tables */
535 	GROUP_TABLE_COUNT,
536 };
537 
538 /*
539  * Flags used by ext4_map_blocks()
540  */
541 	/* Allocate any needed blocks and/or convert an unitialized
542 	   extent to be an initialized ext4 */
543 #define EXT4_GET_BLOCKS_CREATE			0x0001
544 	/* Request the creation of an unitialized extent */
545 #define EXT4_GET_BLOCKS_UNINIT_EXT		0x0002
546 #define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT	(EXT4_GET_BLOCKS_UNINIT_EXT|\
547 						 EXT4_GET_BLOCKS_CREATE)
548 	/* Caller is from the delayed allocation writeout path
549 	 * finally doing the actual allocation of delayed blocks */
550 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE	0x0004
551 	/* caller is from the direct IO path, request to creation of an
552 	unitialized extents if not allocated, split the uninitialized
553 	extent if blocks has been preallocated already*/
554 #define EXT4_GET_BLOCKS_PRE_IO			0x0008
555 #define EXT4_GET_BLOCKS_CONVERT			0x0010
556 #define EXT4_GET_BLOCKS_IO_CREATE_EXT		(EXT4_GET_BLOCKS_PRE_IO|\
557 					 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
558 	/* Convert extent to initialized after IO complete */
559 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT		(EXT4_GET_BLOCKS_CONVERT|\
560 					 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
561 	/* Eventual metadata allocation (due to growing extent tree)
562 	 * should not fail, so try to use reserved blocks for that.*/
563 #define EXT4_GET_BLOCKS_METADATA_NOFAIL		0x0020
564 	/* Don't normalize allocation size (used for fallocate) */
565 #define EXT4_GET_BLOCKS_NO_NORMALIZE		0x0040
566 	/* Request will not result in inode size update (user for fallocate) */
567 #define EXT4_GET_BLOCKS_KEEP_SIZE		0x0080
568 	/* Do not take i_data_sem locking in ext4_map_blocks */
569 #define EXT4_GET_BLOCKS_NO_LOCK			0x0100
570 	/* Do not put hole in extent cache */
571 #define EXT4_GET_BLOCKS_NO_PUT_HOLE		0x0200
572 
573 /*
574  * Flags used by ext4_free_blocks
575  */
576 #define EXT4_FREE_BLOCKS_METADATA	0x0001
577 #define EXT4_FREE_BLOCKS_FORGET		0x0002
578 #define EXT4_FREE_BLOCKS_VALIDATED	0x0004
579 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE	0x0008
580 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER	0x0010
581 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER	0x0020
582 
583 /*
584  * Flags used by ext4_discard_partial_page_buffers
585  */
586 #define EXT4_DISCARD_PARTIAL_PG_ZERO_UNMAPPED	0x0001
587 
588 /*
589  * ioctl commands
590  */
591 #define	EXT4_IOC_GETFLAGS		FS_IOC_GETFLAGS
592 #define	EXT4_IOC_SETFLAGS		FS_IOC_SETFLAGS
593 #define	EXT4_IOC_GETVERSION		_IOR('f', 3, long)
594 #define	EXT4_IOC_SETVERSION		_IOW('f', 4, long)
595 #define	EXT4_IOC_GETVERSION_OLD		FS_IOC_GETVERSION
596 #define	EXT4_IOC_SETVERSION_OLD		FS_IOC_SETVERSION
597 #define EXT4_IOC_GETRSVSZ		_IOR('f', 5, long)
598 #define EXT4_IOC_SETRSVSZ		_IOW('f', 6, long)
599 #define EXT4_IOC_GROUP_EXTEND		_IOW('f', 7, unsigned long)
600 #define EXT4_IOC_GROUP_ADD		_IOW('f', 8, struct ext4_new_group_input)
601 #define EXT4_IOC_MIGRATE		_IO('f', 9)
602  /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
603  /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
604 #define EXT4_IOC_ALLOC_DA_BLKS		_IO('f', 12)
605 #define EXT4_IOC_MOVE_EXT		_IOWR('f', 15, struct move_extent)
606 #define EXT4_IOC_RESIZE_FS		_IOW('f', 16, __u64)
607 #define EXT4_IOC_SWAP_BOOT		_IO('f', 17)
608 
609 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
610 /*
611  * ioctl commands in 32 bit emulation
612  */
613 #define EXT4_IOC32_GETFLAGS		FS_IOC32_GETFLAGS
614 #define EXT4_IOC32_SETFLAGS		FS_IOC32_SETFLAGS
615 #define EXT4_IOC32_GETVERSION		_IOR('f', 3, int)
616 #define EXT4_IOC32_SETVERSION		_IOW('f', 4, int)
617 #define EXT4_IOC32_GETRSVSZ		_IOR('f', 5, int)
618 #define EXT4_IOC32_SETRSVSZ		_IOW('f', 6, int)
619 #define EXT4_IOC32_GROUP_EXTEND		_IOW('f', 7, unsigned int)
620 #define EXT4_IOC32_GROUP_ADD		_IOW('f', 8, struct compat_ext4_new_group_input)
621 #define EXT4_IOC32_GETVERSION_OLD	FS_IOC32_GETVERSION
622 #define EXT4_IOC32_SETVERSION_OLD	FS_IOC32_SETVERSION
623 #endif
624 
625 /* Max physical block we can address w/o extents */
626 #define EXT4_MAX_BLOCK_FILE_PHYS	0xFFFFFFFF
627 
628 /*
629  * Structure of an inode on the disk
630  */
631 struct ext4_inode {
632 	__le16	i_mode;		/* File mode */
633 	__le16	i_uid;		/* Low 16 bits of Owner Uid */
634 	__le32	i_size_lo;	/* Size in bytes */
635 	__le32	i_atime;	/* Access time */
636 	__le32	i_ctime;	/* Inode Change time */
637 	__le32	i_mtime;	/* Modification time */
638 	__le32	i_dtime;	/* Deletion Time */
639 	__le16	i_gid;		/* Low 16 bits of Group Id */
640 	__le16	i_links_count;	/* Links count */
641 	__le32	i_blocks_lo;	/* Blocks count */
642 	__le32	i_flags;	/* File flags */
643 	union {
644 		struct {
645 			__le32  l_i_version;
646 		} linux1;
647 		struct {
648 			__u32  h_i_translator;
649 		} hurd1;
650 		struct {
651 			__u32  m_i_reserved1;
652 		} masix1;
653 	} osd1;				/* OS dependent 1 */
654 	__le32	i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
655 	__le32	i_generation;	/* File version (for NFS) */
656 	__le32	i_file_acl_lo;	/* File ACL */
657 	__le32	i_size_high;
658 	__le32	i_obso_faddr;	/* Obsoleted fragment address */
659 	union {
660 		struct {
661 			__le16	l_i_blocks_high; /* were l_i_reserved1 */
662 			__le16	l_i_file_acl_high;
663 			__le16	l_i_uid_high;	/* these 2 fields */
664 			__le16	l_i_gid_high;	/* were reserved2[0] */
665 			__le16	l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
666 			__le16	l_i_reserved;
667 		} linux2;
668 		struct {
669 			__le16	h_i_reserved1;	/* Obsoleted fragment number/size which are removed in ext4 */
670 			__u16	h_i_mode_high;
671 			__u16	h_i_uid_high;
672 			__u16	h_i_gid_high;
673 			__u32	h_i_author;
674 		} hurd2;
675 		struct {
676 			__le16	h_i_reserved1;	/* Obsoleted fragment number/size which are removed in ext4 */
677 			__le16	m_i_file_acl_high;
678 			__u32	m_i_reserved2[2];
679 		} masix2;
680 	} osd2;				/* OS dependent 2 */
681 	__le16	i_extra_isize;
682 	__le16	i_checksum_hi;	/* crc32c(uuid+inum+inode) BE */
683 	__le32  i_ctime_extra;  /* extra Change time      (nsec << 2 | epoch) */
684 	__le32  i_mtime_extra;  /* extra Modification time(nsec << 2 | epoch) */
685 	__le32  i_atime_extra;  /* extra Access time      (nsec << 2 | epoch) */
686 	__le32  i_crtime;       /* File Creation time */
687 	__le32  i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
688 	__le32  i_version_hi;	/* high 32 bits for 64-bit version */
689 };
690 
691 struct move_extent {
692 	__u32 reserved;		/* should be zero */
693 	__u32 donor_fd;		/* donor file descriptor */
694 	__u64 orig_start;	/* logical start offset in block for orig */
695 	__u64 donor_start;	/* logical start offset in block for donor */
696 	__u64 len;		/* block length to be moved */
697 	__u64 moved_len;	/* moved block length */
698 };
699 
700 #define EXT4_EPOCH_BITS 2
701 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
702 #define EXT4_NSEC_MASK  (~0UL << EXT4_EPOCH_BITS)
703 
704 /*
705  * Extended fields will fit into an inode if the filesystem was formatted
706  * with large inodes (-I 256 or larger) and there are not currently any EAs
707  * consuming all of the available space. For new inodes we always reserve
708  * enough space for the kernel's known extended fields, but for inodes
709  * created with an old kernel this might not have been the case. None of
710  * the extended inode fields is critical for correct filesystem operation.
711  * This macro checks if a certain field fits in the inode. Note that
712  * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
713  */
714 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field)	\
715 	((offsetof(typeof(*ext4_inode), field) +	\
716 	  sizeof((ext4_inode)->field))			\
717 	<= (EXT4_GOOD_OLD_INODE_SIZE +			\
718 	    (einode)->i_extra_isize))			\
719 
720 static inline __le32 ext4_encode_extra_time(struct timespec *time)
721 {
722        return cpu_to_le32((sizeof(time->tv_sec) > 4 ?
723 			   (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) |
724                           ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK));
725 }
726 
727 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
728 {
729        if (sizeof(time->tv_sec) > 4)
730 	       time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK)
731 			       << 32;
732        time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
733 }
734 
735 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode)			       \
736 do {									       \
737 	(raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec);	       \
738 	if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     \
739 		(raw_inode)->xtime ## _extra =				       \
740 				ext4_encode_extra_time(&(inode)->xtime);       \
741 } while (0)
742 
743 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode)			       \
744 do {									       \
745 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))		       \
746 		(raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec);      \
747 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))	       \
748 		(raw_inode)->xtime ## _extra =				       \
749 				ext4_encode_extra_time(&(einode)->xtime);      \
750 } while (0)
751 
752 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode)			       \
753 do {									       \
754 	(inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime);       \
755 	if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     \
756 		ext4_decode_extra_time(&(inode)->xtime,			       \
757 				       raw_inode->xtime ## _extra);	       \
758 	else								       \
759 		(inode)->xtime.tv_nsec = 0;				       \
760 } while (0)
761 
762 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode)			       \
763 do {									       \
764 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))		       \
765 		(einode)->xtime.tv_sec = 				       \
766 			(signed)le32_to_cpu((raw_inode)->xtime);	       \
767 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))	       \
768 		ext4_decode_extra_time(&(einode)->xtime,		       \
769 				       raw_inode->xtime ## _extra);	       \
770 	else								       \
771 		(einode)->xtime.tv_nsec = 0;				       \
772 } while (0)
773 
774 #define i_disk_version osd1.linux1.l_i_version
775 
776 #if defined(__KERNEL__) || defined(__linux__)
777 #define i_reserved1	osd1.linux1.l_i_reserved1
778 #define i_file_acl_high	osd2.linux2.l_i_file_acl_high
779 #define i_blocks_high	osd2.linux2.l_i_blocks_high
780 #define i_uid_low	i_uid
781 #define i_gid_low	i_gid
782 #define i_uid_high	osd2.linux2.l_i_uid_high
783 #define i_gid_high	osd2.linux2.l_i_gid_high
784 #define i_checksum_lo	osd2.linux2.l_i_checksum_lo
785 
786 #elif defined(__GNU__)
787 
788 #define i_translator	osd1.hurd1.h_i_translator
789 #define i_uid_high	osd2.hurd2.h_i_uid_high
790 #define i_gid_high	osd2.hurd2.h_i_gid_high
791 #define i_author	osd2.hurd2.h_i_author
792 
793 #elif defined(__masix__)
794 
795 #define i_reserved1	osd1.masix1.m_i_reserved1
796 #define i_file_acl_high	osd2.masix2.m_i_file_acl_high
797 #define i_reserved2	osd2.masix2.m_i_reserved2
798 
799 #endif /* defined(__KERNEL__) || defined(__linux__) */
800 
801 #include "extents_status.h"
802 
803 /*
804  * fourth extended file system inode data in memory
805  */
806 struct ext4_inode_info {
807 	__le32	i_data[15];	/* unconverted */
808 	__u32	i_dtime;
809 	ext4_fsblk_t	i_file_acl;
810 
811 	/*
812 	 * i_block_group is the number of the block group which contains
813 	 * this file's inode.  Constant across the lifetime of the inode,
814 	 * it is ued for making block allocation decisions - we try to
815 	 * place a file's data blocks near its inode block, and new inodes
816 	 * near to their parent directory's inode.
817 	 */
818 	ext4_group_t	i_block_group;
819 	ext4_lblk_t	i_dir_start_lookup;
820 #if (BITS_PER_LONG < 64)
821 	unsigned long	i_state_flags;		/* Dynamic state flags */
822 #endif
823 	unsigned long	i_flags;
824 
825 	/*
826 	 * Extended attributes can be read independently of the main file
827 	 * data. Taking i_mutex even when reading would cause contention
828 	 * between readers of EAs and writers of regular file data, so
829 	 * instead we synchronize on xattr_sem when reading or changing
830 	 * EAs.
831 	 */
832 	struct rw_semaphore xattr_sem;
833 
834 	struct list_head i_orphan;	/* unlinked but open inodes */
835 
836 	/*
837 	 * i_disksize keeps track of what the inode size is ON DISK, not
838 	 * in memory.  During truncate, i_size is set to the new size by
839 	 * the VFS prior to calling ext4_truncate(), but the filesystem won't
840 	 * set i_disksize to 0 until the truncate is actually under way.
841 	 *
842 	 * The intent is that i_disksize always represents the blocks which
843 	 * are used by this file.  This allows recovery to restart truncate
844 	 * on orphans if we crash during truncate.  We actually write i_disksize
845 	 * into the on-disk inode when writing inodes out, instead of i_size.
846 	 *
847 	 * The only time when i_disksize and i_size may be different is when
848 	 * a truncate is in progress.  The only things which change i_disksize
849 	 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
850 	 */
851 	loff_t	i_disksize;
852 
853 	/*
854 	 * i_data_sem is for serialising ext4_truncate() against
855 	 * ext4_getblock().  In the 2.4 ext2 design, great chunks of inode's
856 	 * data tree are chopped off during truncate. We can't do that in
857 	 * ext4 because whenever we perform intermediate commits during
858 	 * truncate, the inode and all the metadata blocks *must* be in a
859 	 * consistent state which allows truncation of the orphans to restart
860 	 * during recovery.  Hence we must fix the get_block-vs-truncate race
861 	 * by other means, so we have i_data_sem.
862 	 */
863 	struct rw_semaphore i_data_sem;
864 	struct inode vfs_inode;
865 	struct jbd2_inode *jinode;
866 
867 	/*
868 	 * File creation time. Its function is same as that of
869 	 * struct timespec i_{a,c,m}time in the generic inode.
870 	 */
871 	struct timespec i_crtime;
872 
873 	/* mballoc */
874 	struct list_head i_prealloc_list;
875 	spinlock_t i_prealloc_lock;
876 
877 	/* extents status tree */
878 	struct ext4_es_tree i_es_tree;
879 	rwlock_t i_es_lock;
880 	struct list_head i_es_lru;
881 	unsigned int i_es_lru_nr;	/* protected by i_es_lock */
882 
883 	/* ialloc */
884 	ext4_group_t	i_last_alloc_group;
885 
886 	/* allocation reservation info for delalloc */
887 	/* In case of bigalloc, these refer to clusters rather than blocks */
888 	unsigned int i_reserved_data_blocks;
889 	unsigned int i_reserved_meta_blocks;
890 	unsigned int i_allocated_meta_blocks;
891 	ext4_lblk_t i_da_metadata_calc_last_lblock;
892 	int i_da_metadata_calc_len;
893 
894 	/* on-disk additional length */
895 	__u16 i_extra_isize;
896 
897 	/* Indicate the inline data space. */
898 	u16 i_inline_off;
899 	u16 i_inline_size;
900 
901 #ifdef CONFIG_QUOTA
902 	/* quota space reservation, managed internally by quota code */
903 	qsize_t i_reserved_quota;
904 #endif
905 
906 	/* completed IOs that might need unwritten extents handling */
907 	struct list_head i_completed_io_list;
908 	spinlock_t i_completed_io_lock;
909 	atomic_t i_ioend_count;	/* Number of outstanding io_end structs */
910 	atomic_t i_unwritten; /* Nr. of inflight conversions pending */
911 	struct work_struct i_unwritten_work;	/* deferred extent conversion */
912 
913 	spinlock_t i_block_reservation_lock;
914 
915 	/*
916 	 * Transactions that contain inode's metadata needed to complete
917 	 * fsync and fdatasync, respectively.
918 	 */
919 	tid_t i_sync_tid;
920 	tid_t i_datasync_tid;
921 
922 	/* Precomputed uuid+inum+igen checksum for seeding inode checksums */
923 	__u32 i_csum_seed;
924 };
925 
926 /*
927  * File system states
928  */
929 #define	EXT4_VALID_FS			0x0001	/* Unmounted cleanly */
930 #define	EXT4_ERROR_FS			0x0002	/* Errors detected */
931 #define	EXT4_ORPHAN_FS			0x0004	/* Orphans being recovered */
932 
933 /*
934  * Misc. filesystem flags
935  */
936 #define EXT2_FLAGS_SIGNED_HASH		0x0001  /* Signed dirhash in use */
937 #define EXT2_FLAGS_UNSIGNED_HASH	0x0002  /* Unsigned dirhash in use */
938 #define EXT2_FLAGS_TEST_FILESYS		0x0004	/* to test development code */
939 
940 /*
941  * Mount flags set via mount options or defaults
942  */
943 #define EXT4_MOUNT_GRPID		0x00004	/* Create files with directory's group */
944 #define EXT4_MOUNT_DEBUG		0x00008	/* Some debugging messages */
945 #define EXT4_MOUNT_ERRORS_CONT		0x00010	/* Continue on errors */
946 #define EXT4_MOUNT_ERRORS_RO		0x00020	/* Remount fs ro on errors */
947 #define EXT4_MOUNT_ERRORS_PANIC		0x00040	/* Panic on errors */
948 #define EXT4_MOUNT_ERRORS_MASK		0x00070
949 #define EXT4_MOUNT_MINIX_DF		0x00080	/* Mimics the Minix statfs */
950 #define EXT4_MOUNT_NOLOAD		0x00100	/* Don't use existing journal*/
951 #define EXT4_MOUNT_DATA_FLAGS		0x00C00	/* Mode for data writes: */
952 #define EXT4_MOUNT_JOURNAL_DATA		0x00400	/* Write data to journal */
953 #define EXT4_MOUNT_ORDERED_DATA		0x00800	/* Flush data before commit */
954 #define EXT4_MOUNT_WRITEBACK_DATA	0x00C00	/* No data ordering */
955 #define EXT4_MOUNT_UPDATE_JOURNAL	0x01000	/* Update the journal format */
956 #define EXT4_MOUNT_NO_UID32		0x02000  /* Disable 32-bit UIDs */
957 #define EXT4_MOUNT_XATTR_USER		0x04000	/* Extended user attributes */
958 #define EXT4_MOUNT_POSIX_ACL		0x08000	/* POSIX Access Control Lists */
959 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC	0x10000	/* No auto delalloc mapping */
960 #define EXT4_MOUNT_BARRIER		0x20000 /* Use block barriers */
961 #define EXT4_MOUNT_QUOTA		0x80000 /* Some quota option set */
962 #define EXT4_MOUNT_USRQUOTA		0x100000 /* "old" user quota */
963 #define EXT4_MOUNT_GRPQUOTA		0x200000 /* "old" group quota */
964 #define EXT4_MOUNT_DIOREAD_NOLOCK	0x400000 /* Enable support for dio read nolocking */
965 #define EXT4_MOUNT_JOURNAL_CHECKSUM	0x800000 /* Journal checksums */
966 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT	0x1000000 /* Journal Async Commit */
967 #define EXT4_MOUNT_DELALLOC		0x8000000 /* Delalloc support */
968 #define EXT4_MOUNT_DATA_ERR_ABORT	0x10000000 /* Abort on file data write */
969 #define EXT4_MOUNT_BLOCK_VALIDITY	0x20000000 /* Block validity checking */
970 #define EXT4_MOUNT_DISCARD		0x40000000 /* Issue DISCARD requests */
971 #define EXT4_MOUNT_INIT_INODE_TABLE	0x80000000 /* Initialize uninitialized itables */
972 
973 /*
974  * Mount flags set either automatically (could not be set by mount option)
975  * based on per file system feature or property or in special cases such as
976  * distinguishing between explicit mount option definition and default.
977  */
978 #define EXT4_MOUNT2_EXPLICIT_DELALLOC	0x00000001 /* User explicitly
979 						      specified delalloc */
980 #define EXT4_MOUNT2_STD_GROUP_SIZE	0x00000002 /* We have standard group
981 						      size of blocksize * 8
982 						      blocks */
983 
984 #define clear_opt(sb, opt)		EXT4_SB(sb)->s_mount_opt &= \
985 						~EXT4_MOUNT_##opt
986 #define set_opt(sb, opt)		EXT4_SB(sb)->s_mount_opt |= \
987 						EXT4_MOUNT_##opt
988 #define test_opt(sb, opt)		(EXT4_SB(sb)->s_mount_opt & \
989 					 EXT4_MOUNT_##opt)
990 
991 #define clear_opt2(sb, opt)		EXT4_SB(sb)->s_mount_opt2 &= \
992 						~EXT4_MOUNT2_##opt
993 #define set_opt2(sb, opt)		EXT4_SB(sb)->s_mount_opt2 |= \
994 						EXT4_MOUNT2_##opt
995 #define test_opt2(sb, opt)		(EXT4_SB(sb)->s_mount_opt2 & \
996 					 EXT4_MOUNT2_##opt)
997 
998 #define ext4_test_and_set_bit		__test_and_set_bit_le
999 #define ext4_set_bit			__set_bit_le
1000 #define ext4_set_bit_atomic		ext2_set_bit_atomic
1001 #define ext4_test_and_clear_bit		__test_and_clear_bit_le
1002 #define ext4_clear_bit			__clear_bit_le
1003 #define ext4_clear_bit_atomic		ext2_clear_bit_atomic
1004 #define ext4_test_bit			test_bit_le
1005 #define ext4_find_next_zero_bit		find_next_zero_bit_le
1006 #define ext4_find_next_bit		find_next_bit_le
1007 
1008 extern void ext4_set_bits(void *bm, int cur, int len);
1009 
1010 /*
1011  * Maximal mount counts between two filesystem checks
1012  */
1013 #define EXT4_DFL_MAX_MNT_COUNT		20	/* Allow 20 mounts */
1014 #define EXT4_DFL_CHECKINTERVAL		0	/* Don't use interval check */
1015 
1016 /*
1017  * Behaviour when detecting errors
1018  */
1019 #define EXT4_ERRORS_CONTINUE		1	/* Continue execution */
1020 #define EXT4_ERRORS_RO			2	/* Remount fs read-only */
1021 #define EXT4_ERRORS_PANIC		3	/* Panic */
1022 #define EXT4_ERRORS_DEFAULT		EXT4_ERRORS_CONTINUE
1023 
1024 /* Metadata checksum algorithm codes */
1025 #define EXT4_CRC32C_CHKSUM		1
1026 
1027 /*
1028  * Structure of the super block
1029  */
1030 struct ext4_super_block {
1031 /*00*/	__le32	s_inodes_count;		/* Inodes count */
1032 	__le32	s_blocks_count_lo;	/* Blocks count */
1033 	__le32	s_r_blocks_count_lo;	/* Reserved blocks count */
1034 	__le32	s_free_blocks_count_lo;	/* Free blocks count */
1035 /*10*/	__le32	s_free_inodes_count;	/* Free inodes count */
1036 	__le32	s_first_data_block;	/* First Data Block */
1037 	__le32	s_log_block_size;	/* Block size */
1038 	__le32	s_log_cluster_size;	/* Allocation cluster size */
1039 /*20*/	__le32	s_blocks_per_group;	/* # Blocks per group */
1040 	__le32	s_clusters_per_group;	/* # Clusters per group */
1041 	__le32	s_inodes_per_group;	/* # Inodes per group */
1042 	__le32	s_mtime;		/* Mount time */
1043 /*30*/	__le32	s_wtime;		/* Write time */
1044 	__le16	s_mnt_count;		/* Mount count */
1045 	__le16	s_max_mnt_count;	/* Maximal mount count */
1046 	__le16	s_magic;		/* Magic signature */
1047 	__le16	s_state;		/* File system state */
1048 	__le16	s_errors;		/* Behaviour when detecting errors */
1049 	__le16	s_minor_rev_level;	/* minor revision level */
1050 /*40*/	__le32	s_lastcheck;		/* time of last check */
1051 	__le32	s_checkinterval;	/* max. time between checks */
1052 	__le32	s_creator_os;		/* OS */
1053 	__le32	s_rev_level;		/* Revision level */
1054 /*50*/	__le16	s_def_resuid;		/* Default uid for reserved blocks */
1055 	__le16	s_def_resgid;		/* Default gid for reserved blocks */
1056 	/*
1057 	 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1058 	 *
1059 	 * Note: the difference between the compatible feature set and
1060 	 * the incompatible feature set is that if there is a bit set
1061 	 * in the incompatible feature set that the kernel doesn't
1062 	 * know about, it should refuse to mount the filesystem.
1063 	 *
1064 	 * e2fsck's requirements are more strict; if it doesn't know
1065 	 * about a feature in either the compatible or incompatible
1066 	 * feature set, it must abort and not try to meddle with
1067 	 * things it doesn't understand...
1068 	 */
1069 	__le32	s_first_ino;		/* First non-reserved inode */
1070 	__le16  s_inode_size;		/* size of inode structure */
1071 	__le16	s_block_group_nr;	/* block group # of this superblock */
1072 	__le32	s_feature_compat;	/* compatible feature set */
1073 /*60*/	__le32	s_feature_incompat;	/* incompatible feature set */
1074 	__le32	s_feature_ro_compat;	/* readonly-compatible feature set */
1075 /*68*/	__u8	s_uuid[16];		/* 128-bit uuid for volume */
1076 /*78*/	char	s_volume_name[16];	/* volume name */
1077 /*88*/	char	s_last_mounted[64];	/* directory where last mounted */
1078 /*C8*/	__le32	s_algorithm_usage_bitmap; /* For compression */
1079 	/*
1080 	 * Performance hints.  Directory preallocation should only
1081 	 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1082 	 */
1083 	__u8	s_prealloc_blocks;	/* Nr of blocks to try to preallocate*/
1084 	__u8	s_prealloc_dir_blocks;	/* Nr to preallocate for dirs */
1085 	__le16	s_reserved_gdt_blocks;	/* Per group desc for online growth */
1086 	/*
1087 	 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1088 	 */
1089 /*D0*/	__u8	s_journal_uuid[16];	/* uuid of journal superblock */
1090 /*E0*/	__le32	s_journal_inum;		/* inode number of journal file */
1091 	__le32	s_journal_dev;		/* device number of journal file */
1092 	__le32	s_last_orphan;		/* start of list of inodes to delete */
1093 	__le32	s_hash_seed[4];		/* HTREE hash seed */
1094 	__u8	s_def_hash_version;	/* Default hash version to use */
1095 	__u8	s_jnl_backup_type;
1096 	__le16  s_desc_size;		/* size of group descriptor */
1097 /*100*/	__le32	s_default_mount_opts;
1098 	__le32	s_first_meta_bg;	/* First metablock block group */
1099 	__le32	s_mkfs_time;		/* When the filesystem was created */
1100 	__le32	s_jnl_blocks[17];	/* Backup of the journal inode */
1101 	/* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1102 /*150*/	__le32	s_blocks_count_hi;	/* Blocks count */
1103 	__le32	s_r_blocks_count_hi;	/* Reserved blocks count */
1104 	__le32	s_free_blocks_count_hi;	/* Free blocks count */
1105 	__le16	s_min_extra_isize;	/* All inodes have at least # bytes */
1106 	__le16	s_want_extra_isize; 	/* New inodes should reserve # bytes */
1107 	__le32	s_flags;		/* Miscellaneous flags */
1108 	__le16  s_raid_stride;		/* RAID stride */
1109 	__le16  s_mmp_update_interval;  /* # seconds to wait in MMP checking */
1110 	__le64  s_mmp_block;            /* Block for multi-mount protection */
1111 	__le32  s_raid_stripe_width;    /* blocks on all data disks (N*stride)*/
1112 	__u8	s_log_groups_per_flex;  /* FLEX_BG group size */
1113 	__u8	s_checksum_type;	/* metadata checksum algorithm used */
1114 	__le16  s_reserved_pad;
1115 	__le64	s_kbytes_written;	/* nr of lifetime kilobytes written */
1116 	__le32	s_snapshot_inum;	/* Inode number of active snapshot */
1117 	__le32	s_snapshot_id;		/* sequential ID of active snapshot */
1118 	__le64	s_snapshot_r_blocks_count; /* reserved blocks for active
1119 					      snapshot's future use */
1120 	__le32	s_snapshot_list;	/* inode number of the head of the
1121 					   on-disk snapshot list */
1122 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1123 	__le32	s_error_count;		/* number of fs errors */
1124 	__le32	s_first_error_time;	/* first time an error happened */
1125 	__le32	s_first_error_ino;	/* inode involved in first error */
1126 	__le64	s_first_error_block;	/* block involved of first error */
1127 	__u8	s_first_error_func[32];	/* function where the error happened */
1128 	__le32	s_first_error_line;	/* line number where error happened */
1129 	__le32	s_last_error_time;	/* most recent time of an error */
1130 	__le32	s_last_error_ino;	/* inode involved in last error */
1131 	__le32	s_last_error_line;	/* line number where error happened */
1132 	__le64	s_last_error_block;	/* block involved of last error */
1133 	__u8	s_last_error_func[32];	/* function where the error happened */
1134 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1135 	__u8	s_mount_opts[64];
1136 	__le32	s_usr_quota_inum;	/* inode for tracking user quota */
1137 	__le32	s_grp_quota_inum;	/* inode for tracking group quota */
1138 	__le32	s_overhead_clusters;	/* overhead blocks/clusters in fs */
1139 	__le32	s_reserved[108];	/* Padding to the end of the block */
1140 	__le32	s_checksum;		/* crc32c(superblock) */
1141 };
1142 
1143 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1144 
1145 #ifdef __KERNEL__
1146 
1147 /*
1148  * run-time mount flags
1149  */
1150 #define EXT4_MF_MNTDIR_SAMPLED	0x0001
1151 #define EXT4_MF_FS_ABORTED	0x0002	/* Fatal error detected */
1152 
1153 /*
1154  * fourth extended-fs super-block data in memory
1155  */
1156 struct ext4_sb_info {
1157 	unsigned long s_desc_size;	/* Size of a group descriptor in bytes */
1158 	unsigned long s_inodes_per_block;/* Number of inodes per block */
1159 	unsigned long s_blocks_per_group;/* Number of blocks in a group */
1160 	unsigned long s_clusters_per_group; /* Number of clusters in a group */
1161 	unsigned long s_inodes_per_group;/* Number of inodes in a group */
1162 	unsigned long s_itb_per_group;	/* Number of inode table blocks per group */
1163 	unsigned long s_gdb_count;	/* Number of group descriptor blocks */
1164 	unsigned long s_desc_per_block;	/* Number of group descriptors per block */
1165 	ext4_group_t s_groups_count;	/* Number of groups in the fs */
1166 	ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1167 	unsigned long s_overhead;  /* # of fs overhead clusters */
1168 	unsigned int s_cluster_ratio;	/* Number of blocks per cluster */
1169 	unsigned int s_cluster_bits;	/* log2 of s_cluster_ratio */
1170 	loff_t s_bitmap_maxbytes;	/* max bytes for bitmap files */
1171 	struct buffer_head * s_sbh;	/* Buffer containing the super block */
1172 	struct ext4_super_block *s_es;	/* Pointer to the super block in the buffer */
1173 	struct buffer_head **s_group_desc;
1174 	unsigned int s_mount_opt;
1175 	unsigned int s_mount_opt2;
1176 	unsigned int s_mount_flags;
1177 	unsigned int s_def_mount_opt;
1178 	ext4_fsblk_t s_sb_block;
1179 	atomic64_t s_resv_clusters;
1180 	kuid_t s_resuid;
1181 	kgid_t s_resgid;
1182 	unsigned short s_mount_state;
1183 	unsigned short s_pad;
1184 	int s_addr_per_block_bits;
1185 	int s_desc_per_block_bits;
1186 	int s_inode_size;
1187 	int s_first_ino;
1188 	unsigned int s_inode_readahead_blks;
1189 	unsigned int s_inode_goal;
1190 	spinlock_t s_next_gen_lock;
1191 	u32 s_next_generation;
1192 	u32 s_hash_seed[4];
1193 	int s_def_hash_version;
1194 	int s_hash_unsigned;	/* 3 if hash should be signed, 0 if not */
1195 	struct percpu_counter s_freeclusters_counter;
1196 	struct percpu_counter s_freeinodes_counter;
1197 	struct percpu_counter s_dirs_counter;
1198 	struct percpu_counter s_dirtyclusters_counter;
1199 	struct blockgroup_lock *s_blockgroup_lock;
1200 	struct proc_dir_entry *s_proc;
1201 	struct kobject s_kobj;
1202 	struct completion s_kobj_unregister;
1203 	struct super_block *s_sb;
1204 
1205 	/* Journaling */
1206 	struct journal_s *s_journal;
1207 	struct list_head s_orphan;
1208 	struct mutex s_orphan_lock;
1209 	unsigned long s_resize_flags;		/* Flags indicating if there
1210 						   is a resizer */
1211 	unsigned long s_commit_interval;
1212 	u32 s_max_batch_time;
1213 	u32 s_min_batch_time;
1214 	struct block_device *journal_bdev;
1215 #ifdef CONFIG_QUOTA
1216 	char *s_qf_names[MAXQUOTAS];		/* Names of quota files with journalled quota */
1217 	int s_jquota_fmt;			/* Format of quota to use */
1218 #endif
1219 	unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1220 	struct rb_root system_blks;
1221 
1222 #ifdef EXTENTS_STATS
1223 	/* ext4 extents stats */
1224 	unsigned long s_ext_min;
1225 	unsigned long s_ext_max;
1226 	unsigned long s_depth_max;
1227 	spinlock_t s_ext_stats_lock;
1228 	unsigned long s_ext_blocks;
1229 	unsigned long s_ext_extents;
1230 #endif
1231 
1232 	/* for buddy allocator */
1233 	struct ext4_group_info ***s_group_info;
1234 	struct inode *s_buddy_cache;
1235 	spinlock_t s_md_lock;
1236 	unsigned short *s_mb_offsets;
1237 	unsigned int *s_mb_maxs;
1238 	unsigned int s_group_info_size;
1239 
1240 	/* tunables */
1241 	unsigned long s_stripe;
1242 	unsigned int s_mb_stream_request;
1243 	unsigned int s_mb_max_to_scan;
1244 	unsigned int s_mb_min_to_scan;
1245 	unsigned int s_mb_stats;
1246 	unsigned int s_mb_order2_reqs;
1247 	unsigned int s_mb_group_prealloc;
1248 	unsigned int s_max_writeback_mb_bump;
1249 	unsigned int s_max_dir_size_kb;
1250 	/* where last allocation was done - for stream allocation */
1251 	unsigned long s_mb_last_group;
1252 	unsigned long s_mb_last_start;
1253 
1254 	/* stats for buddy allocator */
1255 	atomic_t s_bal_reqs;	/* number of reqs with len > 1 */
1256 	atomic_t s_bal_success;	/* we found long enough chunks */
1257 	atomic_t s_bal_allocated;	/* in blocks */
1258 	atomic_t s_bal_ex_scanned;	/* total extents scanned */
1259 	atomic_t s_bal_goals;	/* goal hits */
1260 	atomic_t s_bal_breaks;	/* too long searches */
1261 	atomic_t s_bal_2orders;	/* 2^order hits */
1262 	spinlock_t s_bal_lock;
1263 	unsigned long s_mb_buddies_generated;
1264 	unsigned long long s_mb_generation_time;
1265 	atomic_t s_mb_lost_chunks;
1266 	atomic_t s_mb_preallocated;
1267 	atomic_t s_mb_discarded;
1268 	atomic_t s_lock_busy;
1269 
1270 	/* locality groups */
1271 	struct ext4_locality_group __percpu *s_locality_groups;
1272 
1273 	/* for write statistics */
1274 	unsigned long s_sectors_written_start;
1275 	u64 s_kbytes_written;
1276 
1277 	/* the size of zero-out chunk */
1278 	unsigned int s_extent_max_zeroout_kb;
1279 
1280 	unsigned int s_log_groups_per_flex;
1281 	struct flex_groups *s_flex_groups;
1282 	ext4_group_t s_flex_groups_allocated;
1283 
1284 	/* workqueue for dio unwritten */
1285 	struct workqueue_struct *dio_unwritten_wq;
1286 
1287 	/* timer for periodic error stats printing */
1288 	struct timer_list s_err_report;
1289 
1290 	/* Lazy inode table initialization info */
1291 	struct ext4_li_request *s_li_request;
1292 	/* Wait multiplier for lazy initialization thread */
1293 	unsigned int s_li_wait_mult;
1294 
1295 	/* Kernel thread for multiple mount protection */
1296 	struct task_struct *s_mmp_tsk;
1297 
1298 	/* record the last minlen when FITRIM is called. */
1299 	atomic_t s_last_trim_minblks;
1300 
1301 	/* Reference to checksum algorithm driver via cryptoapi */
1302 	struct crypto_shash *s_chksum_driver;
1303 
1304 	/* Precomputed FS UUID checksum for seeding other checksums */
1305 	__u32 s_csum_seed;
1306 
1307 	/* Reclaim extents from extent status tree */
1308 	struct shrinker s_es_shrinker;
1309 	struct list_head s_es_lru;
1310 	struct percpu_counter s_extent_cache_cnt;
1311 	spinlock_t s_es_lru_lock ____cacheline_aligned_in_smp;
1312 };
1313 
1314 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1315 {
1316 	return sb->s_fs_info;
1317 }
1318 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1319 {
1320 	return container_of(inode, struct ext4_inode_info, vfs_inode);
1321 }
1322 
1323 static inline struct timespec ext4_current_time(struct inode *inode)
1324 {
1325 	return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1326 		current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1327 }
1328 
1329 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1330 {
1331 	return ino == EXT4_ROOT_INO ||
1332 		ino == EXT4_USR_QUOTA_INO ||
1333 		ino == EXT4_GRP_QUOTA_INO ||
1334 		ino == EXT4_BOOT_LOADER_INO ||
1335 		ino == EXT4_JOURNAL_INO ||
1336 		ino == EXT4_RESIZE_INO ||
1337 		(ino >= EXT4_FIRST_INO(sb) &&
1338 		 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1339 }
1340 
1341 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
1342 					      struct ext4_io_end *io_end)
1343 {
1344 	if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
1345 		io_end->flag |= EXT4_IO_END_UNWRITTEN;
1346 		atomic_inc(&EXT4_I(inode)->i_unwritten);
1347 	}
1348 }
1349 
1350 static inline ext4_io_end_t *ext4_inode_aio(struct inode *inode)
1351 {
1352 	return inode->i_private;
1353 }
1354 
1355 static inline void ext4_inode_aio_set(struct inode *inode, ext4_io_end_t *io)
1356 {
1357 	inode->i_private = io;
1358 }
1359 
1360 /*
1361  * Inode dynamic state flags
1362  */
1363 enum {
1364 	EXT4_STATE_JDATA,		/* journaled data exists */
1365 	EXT4_STATE_NEW,			/* inode is newly created */
1366 	EXT4_STATE_XATTR,		/* has in-inode xattrs */
1367 	EXT4_STATE_NO_EXPAND,		/* No space for expansion */
1368 	EXT4_STATE_DA_ALLOC_CLOSE,	/* Alloc DA blks on close */
1369 	EXT4_STATE_EXT_MIGRATE,		/* Inode is migrating */
1370 	EXT4_STATE_DIO_UNWRITTEN,	/* need convert on dio done*/
1371 	EXT4_STATE_NEWENTRY,		/* File just added to dir */
1372 	EXT4_STATE_DELALLOC_RESERVED,	/* blks already reserved for delalloc */
1373 	EXT4_STATE_DIOREAD_LOCK,	/* Disable support for dio read
1374 					   nolocking */
1375 	EXT4_STATE_MAY_INLINE_DATA,	/* may have in-inode data */
1376 	EXT4_STATE_ORDERED_MODE,	/* data=ordered mode */
1377 };
1378 
1379 #define EXT4_INODE_BIT_FNS(name, field, offset)				\
1380 static inline int ext4_test_inode_##name(struct inode *inode, int bit)	\
1381 {									\
1382 	return test_bit(bit + (offset), &EXT4_I(inode)->i_##field);	\
1383 }									\
1384 static inline void ext4_set_inode_##name(struct inode *inode, int bit)	\
1385 {									\
1386 	set_bit(bit + (offset), &EXT4_I(inode)->i_##field);		\
1387 }									\
1388 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1389 {									\
1390 	clear_bit(bit + (offset), &EXT4_I(inode)->i_##field);		\
1391 }
1392 
1393 EXT4_INODE_BIT_FNS(flag, flags, 0)
1394 #if (BITS_PER_LONG < 64)
1395 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1396 
1397 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1398 {
1399 	(ei)->i_state_flags = 0;
1400 }
1401 #else
1402 EXT4_INODE_BIT_FNS(state, flags, 32)
1403 
1404 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1405 {
1406 	/* We depend on the fact that callers will set i_flags */
1407 }
1408 #endif
1409 #else
1410 /* Assume that user mode programs are passing in an ext4fs superblock, not
1411  * a kernel struct super_block.  This will allow us to call the feature-test
1412  * macros from user land. */
1413 #define EXT4_SB(sb)	(sb)
1414 #endif
1415 
1416 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1417 
1418 /*
1419  * Codes for operating systems
1420  */
1421 #define EXT4_OS_LINUX		0
1422 #define EXT4_OS_HURD		1
1423 #define EXT4_OS_MASIX		2
1424 #define EXT4_OS_FREEBSD		3
1425 #define EXT4_OS_LITES		4
1426 
1427 /*
1428  * Revision levels
1429  */
1430 #define EXT4_GOOD_OLD_REV	0	/* The good old (original) format */
1431 #define EXT4_DYNAMIC_REV	1	/* V2 format w/ dynamic inode sizes */
1432 
1433 #define EXT4_CURRENT_REV	EXT4_GOOD_OLD_REV
1434 #define EXT4_MAX_SUPP_REV	EXT4_DYNAMIC_REV
1435 
1436 #define EXT4_GOOD_OLD_INODE_SIZE 128
1437 
1438 /*
1439  * Feature set definitions
1440  */
1441 
1442 #define EXT4_HAS_COMPAT_FEATURE(sb,mask)			\
1443 	((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1444 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask)			\
1445 	((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1446 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask)			\
1447 	((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1448 #define EXT4_SET_COMPAT_FEATURE(sb,mask)			\
1449 	EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1450 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask)			\
1451 	EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1452 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask)			\
1453 	EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1454 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask)			\
1455 	EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1456 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask)			\
1457 	EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1458 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask)			\
1459 	EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1460 
1461 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC	0x0001
1462 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES	0x0002
1463 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL		0x0004
1464 #define EXT4_FEATURE_COMPAT_EXT_ATTR		0x0008
1465 #define EXT4_FEATURE_COMPAT_RESIZE_INODE	0x0010
1466 #define EXT4_FEATURE_COMPAT_DIR_INDEX		0x0020
1467 
1468 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER	0x0001
1469 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE	0x0002
1470 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR	0x0004
1471 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE        0x0008
1472 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM		0x0010
1473 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK	0x0020
1474 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE	0x0040
1475 #define EXT4_FEATURE_RO_COMPAT_QUOTA		0x0100
1476 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC		0x0200
1477 /*
1478  * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM).  When
1479  * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1480  * all other data structures' checksums.  However, the METADATA_CSUM and
1481  * GDT_CSUM bits are mutually exclusive.
1482  */
1483 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM	0x0400
1484 
1485 #define EXT4_FEATURE_INCOMPAT_COMPRESSION	0x0001
1486 #define EXT4_FEATURE_INCOMPAT_FILETYPE		0x0002
1487 #define EXT4_FEATURE_INCOMPAT_RECOVER		0x0004 /* Needs recovery */
1488 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV	0x0008 /* Journal device */
1489 #define EXT4_FEATURE_INCOMPAT_META_BG		0x0010
1490 #define EXT4_FEATURE_INCOMPAT_EXTENTS		0x0040 /* extents support */
1491 #define EXT4_FEATURE_INCOMPAT_64BIT		0x0080
1492 #define EXT4_FEATURE_INCOMPAT_MMP               0x0100
1493 #define EXT4_FEATURE_INCOMPAT_FLEX_BG		0x0200
1494 #define EXT4_FEATURE_INCOMPAT_EA_INODE		0x0400 /* EA in inode */
1495 #define EXT4_FEATURE_INCOMPAT_DIRDATA		0x1000 /* data in dirent */
1496 #define EXT4_FEATURE_INCOMPAT_BG_USE_META_CSUM	0x2000 /* use crc32c for bg */
1497 #define EXT4_FEATURE_INCOMPAT_LARGEDIR		0x4000 /* >2GB or 3-lvl htree */
1498 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA	0x8000 /* data in inode */
1499 
1500 #define EXT2_FEATURE_COMPAT_SUPP	EXT4_FEATURE_COMPAT_EXT_ATTR
1501 #define EXT2_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
1502 					 EXT4_FEATURE_INCOMPAT_META_BG)
1503 #define EXT2_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1504 					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1505 					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1506 
1507 #define EXT3_FEATURE_COMPAT_SUPP	EXT4_FEATURE_COMPAT_EXT_ATTR
1508 #define EXT3_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
1509 					 EXT4_FEATURE_INCOMPAT_RECOVER| \
1510 					 EXT4_FEATURE_INCOMPAT_META_BG)
1511 #define EXT3_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1512 					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1513 					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1514 
1515 #define EXT4_FEATURE_COMPAT_SUPP	EXT2_FEATURE_COMPAT_EXT_ATTR
1516 #define EXT4_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
1517 					 EXT4_FEATURE_INCOMPAT_RECOVER| \
1518 					 EXT4_FEATURE_INCOMPAT_META_BG| \
1519 					 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1520 					 EXT4_FEATURE_INCOMPAT_64BIT| \
1521 					 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1522 					 EXT4_FEATURE_INCOMPAT_MMP |	\
1523 					 EXT4_FEATURE_INCOMPAT_INLINE_DATA)
1524 #define EXT4_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1525 					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1526 					 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1527 					 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1528 					 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1529 					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1530 					 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1531 					 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1532 					 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1533 					 EXT4_FEATURE_RO_COMPAT_QUOTA)
1534 
1535 /*
1536  * Default values for user and/or group using reserved blocks
1537  */
1538 #define	EXT4_DEF_RESUID		0
1539 #define	EXT4_DEF_RESGID		0
1540 
1541 #define EXT4_DEF_INODE_READAHEAD_BLKS	32
1542 
1543 /*
1544  * Default mount options
1545  */
1546 #define EXT4_DEFM_DEBUG		0x0001
1547 #define EXT4_DEFM_BSDGROUPS	0x0002
1548 #define EXT4_DEFM_XATTR_USER	0x0004
1549 #define EXT4_DEFM_ACL		0x0008
1550 #define EXT4_DEFM_UID16		0x0010
1551 #define EXT4_DEFM_JMODE		0x0060
1552 #define EXT4_DEFM_JMODE_DATA	0x0020
1553 #define EXT4_DEFM_JMODE_ORDERED	0x0040
1554 #define EXT4_DEFM_JMODE_WBACK	0x0060
1555 #define EXT4_DEFM_NOBARRIER	0x0100
1556 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1557 #define EXT4_DEFM_DISCARD	0x0400
1558 #define EXT4_DEFM_NODELALLOC	0x0800
1559 
1560 /*
1561  * Default journal batch times
1562  */
1563 #define EXT4_DEF_MIN_BATCH_TIME	0
1564 #define EXT4_DEF_MAX_BATCH_TIME	15000 /* 15ms */
1565 
1566 /*
1567  * Minimum number of groups in a flexgroup before we separate out
1568  * directories into the first block group of a flexgroup
1569  */
1570 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME	4
1571 
1572 /*
1573  * Structure of a directory entry
1574  */
1575 #define EXT4_NAME_LEN 255
1576 
1577 struct ext4_dir_entry {
1578 	__le32	inode;			/* Inode number */
1579 	__le16	rec_len;		/* Directory entry length */
1580 	__le16	name_len;		/* Name length */
1581 	char	name[EXT4_NAME_LEN];	/* File name */
1582 };
1583 
1584 /*
1585  * The new version of the directory entry.  Since EXT4 structures are
1586  * stored in intel byte order, and the name_len field could never be
1587  * bigger than 255 chars, it's safe to reclaim the extra byte for the
1588  * file_type field.
1589  */
1590 struct ext4_dir_entry_2 {
1591 	__le32	inode;			/* Inode number */
1592 	__le16	rec_len;		/* Directory entry length */
1593 	__u8	name_len;		/* Name length */
1594 	__u8	file_type;
1595 	char	name[EXT4_NAME_LEN];	/* File name */
1596 };
1597 
1598 /*
1599  * This is a bogus directory entry at the end of each leaf block that
1600  * records checksums.
1601  */
1602 struct ext4_dir_entry_tail {
1603 	__le32	det_reserved_zero1;	/* Pretend to be unused */
1604 	__le16	det_rec_len;		/* 12 */
1605 	__u8	det_reserved_zero2;	/* Zero name length */
1606 	__u8	det_reserved_ft;	/* 0xDE, fake file type */
1607 	__le32	det_checksum;		/* crc32c(uuid+inum+dirblock) */
1608 };
1609 
1610 #define EXT4_DIRENT_TAIL(block, blocksize) \
1611 	((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1612 					((blocksize) - \
1613 					 sizeof(struct ext4_dir_entry_tail))))
1614 
1615 /*
1616  * Ext4 directory file types.  Only the low 3 bits are used.  The
1617  * other bits are reserved for now.
1618  */
1619 #define EXT4_FT_UNKNOWN		0
1620 #define EXT4_FT_REG_FILE	1
1621 #define EXT4_FT_DIR		2
1622 #define EXT4_FT_CHRDEV		3
1623 #define EXT4_FT_BLKDEV		4
1624 #define EXT4_FT_FIFO		5
1625 #define EXT4_FT_SOCK		6
1626 #define EXT4_FT_SYMLINK		7
1627 
1628 #define EXT4_FT_MAX		8
1629 
1630 #define EXT4_FT_DIR_CSUM	0xDE
1631 
1632 /*
1633  * EXT4_DIR_PAD defines the directory entries boundaries
1634  *
1635  * NOTE: It must be a multiple of 4
1636  */
1637 #define EXT4_DIR_PAD			4
1638 #define EXT4_DIR_ROUND			(EXT4_DIR_PAD - 1)
1639 #define EXT4_DIR_REC_LEN(name_len)	(((name_len) + 8 + EXT4_DIR_ROUND) & \
1640 					 ~EXT4_DIR_ROUND)
1641 #define EXT4_MAX_REC_LEN		((1<<16)-1)
1642 
1643 /*
1644  * If we ever get support for fs block sizes > page_size, we'll need
1645  * to remove the #if statements in the next two functions...
1646  */
1647 static inline unsigned int
1648 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1649 {
1650 	unsigned len = le16_to_cpu(dlen);
1651 
1652 #if (PAGE_CACHE_SIZE >= 65536)
1653 	if (len == EXT4_MAX_REC_LEN || len == 0)
1654 		return blocksize;
1655 	return (len & 65532) | ((len & 3) << 16);
1656 #else
1657 	return len;
1658 #endif
1659 }
1660 
1661 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1662 {
1663 	if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1664 		BUG();
1665 #if (PAGE_CACHE_SIZE >= 65536)
1666 	if (len < 65536)
1667 		return cpu_to_le16(len);
1668 	if (len == blocksize) {
1669 		if (blocksize == 65536)
1670 			return cpu_to_le16(EXT4_MAX_REC_LEN);
1671 		else
1672 			return cpu_to_le16(0);
1673 	}
1674 	return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1675 #else
1676 	return cpu_to_le16(len);
1677 #endif
1678 }
1679 
1680 /*
1681  * Hash Tree Directory indexing
1682  * (c) Daniel Phillips, 2001
1683  */
1684 
1685 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1686 				      EXT4_FEATURE_COMPAT_DIR_INDEX) && \
1687 		    ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
1688 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1689 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1690 
1691 /* Legal values for the dx_root hash_version field: */
1692 
1693 #define DX_HASH_LEGACY		0
1694 #define DX_HASH_HALF_MD4	1
1695 #define DX_HASH_TEA		2
1696 #define DX_HASH_LEGACY_UNSIGNED	3
1697 #define DX_HASH_HALF_MD4_UNSIGNED	4
1698 #define DX_HASH_TEA_UNSIGNED		5
1699 
1700 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
1701 			      const void *address, unsigned int length)
1702 {
1703 	struct {
1704 		struct shash_desc shash;
1705 		char ctx[4];
1706 	} desc;
1707 	int err;
1708 
1709 	BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
1710 
1711 	desc.shash.tfm = sbi->s_chksum_driver;
1712 	desc.shash.flags = 0;
1713 	*(u32 *)desc.ctx = crc;
1714 
1715 	err = crypto_shash_update(&desc.shash, address, length);
1716 	BUG_ON(err);
1717 
1718 	return *(u32 *)desc.ctx;
1719 }
1720 
1721 #ifdef __KERNEL__
1722 
1723 /* hash info structure used by the directory hash */
1724 struct dx_hash_info
1725 {
1726 	u32		hash;
1727 	u32		minor_hash;
1728 	int		hash_version;
1729 	u32		*seed;
1730 };
1731 
1732 
1733 /* 32 and 64 bit signed EOF for dx directories */
1734 #define EXT4_HTREE_EOF_32BIT   ((1UL  << (32 - 1)) - 1)
1735 #define EXT4_HTREE_EOF_64BIT   ((1ULL << (64 - 1)) - 1)
1736 
1737 
1738 /*
1739  * Control parameters used by ext4_htree_next_block
1740  */
1741 #define HASH_NB_ALWAYS		1
1742 
1743 
1744 /*
1745  * Describe an inode's exact location on disk and in memory
1746  */
1747 struct ext4_iloc
1748 {
1749 	struct buffer_head *bh;
1750 	unsigned long offset;
1751 	ext4_group_t block_group;
1752 };
1753 
1754 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
1755 {
1756 	return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
1757 }
1758 
1759 /*
1760  * This structure is stuffed into the struct file's private_data field
1761  * for directories.  It is where we put information so that we can do
1762  * readdir operations in hash tree order.
1763  */
1764 struct dir_private_info {
1765 	struct rb_root	root;
1766 	struct rb_node	*curr_node;
1767 	struct fname	*extra_fname;
1768 	loff_t		last_pos;
1769 	__u32		curr_hash;
1770 	__u32		curr_minor_hash;
1771 	__u32		next_hash;
1772 };
1773 
1774 /* calculate the first block number of the group */
1775 static inline ext4_fsblk_t
1776 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
1777 {
1778 	return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
1779 		le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
1780 }
1781 
1782 /*
1783  * Special error return code only used by dx_probe() and its callers.
1784  */
1785 #define ERR_BAD_DX_DIR	-75000
1786 
1787 /*
1788  * Timeout and state flag for lazy initialization inode thread.
1789  */
1790 #define EXT4_DEF_LI_WAIT_MULT			10
1791 #define EXT4_DEF_LI_MAX_START_DELAY		5
1792 #define EXT4_LAZYINIT_QUIT			0x0001
1793 #define EXT4_LAZYINIT_RUNNING			0x0002
1794 
1795 /*
1796  * Lazy inode table initialization info
1797  */
1798 struct ext4_lazy_init {
1799 	unsigned long		li_state;
1800 	struct list_head	li_request_list;
1801 	struct mutex		li_list_mtx;
1802 };
1803 
1804 struct ext4_li_request {
1805 	struct super_block	*lr_super;
1806 	struct ext4_sb_info	*lr_sbi;
1807 	ext4_group_t		lr_next_group;
1808 	struct list_head	lr_request;
1809 	unsigned long		lr_next_sched;
1810 	unsigned long		lr_timeout;
1811 };
1812 
1813 struct ext4_features {
1814 	struct kobject f_kobj;
1815 	struct completion f_kobj_unregister;
1816 };
1817 
1818 /*
1819  * This structure will be used for multiple mount protection. It will be
1820  * written into the block number saved in the s_mmp_block field in the
1821  * superblock. Programs that check MMP should assume that if
1822  * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
1823  * to use the filesystem, regardless of how old the timestamp is.
1824  */
1825 #define EXT4_MMP_MAGIC     0x004D4D50U /* ASCII for MMP */
1826 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
1827 #define EXT4_MMP_SEQ_FSCK  0xE24D4D50U /* mmp_seq value when being fscked */
1828 #define EXT4_MMP_SEQ_MAX   0xE24D4D4FU /* maximum valid mmp_seq value */
1829 
1830 struct mmp_struct {
1831 	__le32	mmp_magic;		/* Magic number for MMP */
1832 	__le32	mmp_seq;		/* Sequence no. updated periodically */
1833 
1834 	/*
1835 	 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
1836 	 * purposes and do not affect the correctness of the algorithm
1837 	 */
1838 	__le64	mmp_time;		/* Time last updated */
1839 	char	mmp_nodename[64];	/* Node which last updated MMP block */
1840 	char	mmp_bdevname[32];	/* Bdev which last updated MMP block */
1841 
1842 	/*
1843 	 * mmp_check_interval is used to verify if the MMP block has been
1844 	 * updated on the block device. The value is updated based on the
1845 	 * maximum time to write the MMP block during an update cycle.
1846 	 */
1847 	__le16	mmp_check_interval;
1848 
1849 	__le16	mmp_pad1;
1850 	__le32	mmp_pad2[226];
1851 	__le32	mmp_checksum;		/* crc32c(uuid+mmp_block) */
1852 };
1853 
1854 /* arguments passed to the mmp thread */
1855 struct mmpd_data {
1856 	struct buffer_head *bh; /* bh from initial read_mmp_block() */
1857 	struct super_block *sb;  /* super block of the fs */
1858 };
1859 
1860 /*
1861  * Check interval multiplier
1862  * The MMP block is written every update interval and initially checked every
1863  * update interval x the multiplier (the value is then adapted based on the
1864  * write latency). The reason is that writes can be delayed under load and we
1865  * don't want readers to incorrectly assume that the filesystem is no longer
1866  * in use.
1867  */
1868 #define EXT4_MMP_CHECK_MULT		2UL
1869 
1870 /*
1871  * Minimum interval for MMP checking in seconds.
1872  */
1873 #define EXT4_MMP_MIN_CHECK_INTERVAL	5UL
1874 
1875 /*
1876  * Maximum interval for MMP checking in seconds.
1877  */
1878 #define EXT4_MMP_MAX_CHECK_INTERVAL	300UL
1879 
1880 /*
1881  * Function prototypes
1882  */
1883 
1884 /*
1885  * Ok, these declarations are also in <linux/kernel.h> but none of the
1886  * ext4 source programs needs to include it so they are duplicated here.
1887  */
1888 # define NORET_TYPE	/**/
1889 # define ATTRIB_NORET	__attribute__((noreturn))
1890 # define NORET_AND	noreturn,
1891 
1892 /* bitmap.c */
1893 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
1894 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
1895 				struct ext4_group_desc *gdp,
1896 				struct buffer_head *bh, int sz);
1897 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
1898 				  struct ext4_group_desc *gdp,
1899 				  struct buffer_head *bh, int sz);
1900 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
1901 				struct ext4_group_desc *gdp,
1902 				struct buffer_head *bh);
1903 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
1904 				  struct ext4_group_desc *gdp,
1905 				  struct buffer_head *bh);
1906 
1907 /* balloc.c */
1908 extern void ext4_get_group_no_and_offset(struct super_block *sb,
1909 					 ext4_fsblk_t blocknr,
1910 					 ext4_group_t *blockgrpp,
1911 					 ext4_grpblk_t *offsetp);
1912 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
1913 					  ext4_fsblk_t block);
1914 
1915 extern void ext4_validate_block_bitmap(struct super_block *sb,
1916 				       struct ext4_group_desc *desc,
1917 				       unsigned int block_group,
1918 				       struct buffer_head *bh);
1919 extern unsigned int ext4_block_group(struct super_block *sb,
1920 			ext4_fsblk_t blocknr);
1921 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
1922 			ext4_fsblk_t blocknr);
1923 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
1924 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
1925 			ext4_group_t group);
1926 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
1927 					 ext4_fsblk_t goal,
1928 					 unsigned int flags,
1929 					 unsigned long *count,
1930 					 int *errp);
1931 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
1932 				    s64 nclusters, unsigned int flags);
1933 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
1934 extern void ext4_check_blocks_bitmap(struct super_block *);
1935 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
1936 						    ext4_group_t block_group,
1937 						    struct buffer_head ** bh);
1938 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
1939 
1940 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
1941 						ext4_group_t block_group);
1942 extern int ext4_wait_block_bitmap(struct super_block *sb,
1943 				  ext4_group_t block_group,
1944 				  struct buffer_head *bh);
1945 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
1946 						  ext4_group_t block_group);
1947 extern void ext4_init_block_bitmap(struct super_block *sb,
1948 				   struct buffer_head *bh,
1949 				   ext4_group_t group,
1950 				   struct ext4_group_desc *desc);
1951 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
1952 					      ext4_group_t block_group,
1953 					      struct ext4_group_desc *gdp);
1954 extern unsigned ext4_num_overhead_clusters(struct super_block *sb,
1955 					   ext4_group_t block_group,
1956 					   struct ext4_group_desc *gdp);
1957 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
1958 
1959 /* dir.c */
1960 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
1961 				  struct file *,
1962 				  struct ext4_dir_entry_2 *,
1963 				  struct buffer_head *, char *, int,
1964 				  unsigned int);
1965 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset)	\
1966 	unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
1967 					(de), (bh), (buf), (size), (offset)))
1968 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
1969 				    __u32 minor_hash,
1970 				    struct ext4_dir_entry_2 *dirent);
1971 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
1972 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
1973 			     struct buffer_head *bh,
1974 			     void *buf, int buf_size,
1975 			     const char *name, int namelen,
1976 			     struct ext4_dir_entry_2 **dest_de);
1977 void ext4_insert_dentry(struct inode *inode,
1978 			struct ext4_dir_entry_2 *de,
1979 			int buf_size,
1980 			const char *name, int namelen);
1981 static inline void ext4_update_dx_flag(struct inode *inode)
1982 {
1983 	if (!EXT4_HAS_COMPAT_FEATURE(inode->i_sb,
1984 				     EXT4_FEATURE_COMPAT_DIR_INDEX))
1985 		ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
1986 }
1987 static unsigned char ext4_filetype_table[] = {
1988 	DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
1989 };
1990 
1991 static inline  unsigned char get_dtype(struct super_block *sb, int filetype)
1992 {
1993 	if (!EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE) ||
1994 	    (filetype >= EXT4_FT_MAX))
1995 		return DT_UNKNOWN;
1996 
1997 	return ext4_filetype_table[filetype];
1998 }
1999 
2000 /* fsync.c */
2001 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2002 extern int ext4_flush_unwritten_io(struct inode *);
2003 
2004 /* hash.c */
2005 extern int ext4fs_dirhash(const char *name, int len, struct
2006 			  dx_hash_info *hinfo);
2007 
2008 /* ialloc.c */
2009 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2010 				      const struct qstr *qstr, __u32 goal,
2011 				      uid_t *owner, int handle_type,
2012 				      unsigned int line_no, int nblocks);
2013 
2014 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner) \
2015 	__ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2016 			 0, 0, 0)
2017 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2018 				    type, nblocks)		    \
2019 	__ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2020 			 (type), __LINE__, (nblocks))
2021 
2022 
2023 extern void ext4_free_inode(handle_t *, struct inode *);
2024 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2025 extern unsigned long ext4_count_free_inodes(struct super_block *);
2026 extern unsigned long ext4_count_dirs(struct super_block *);
2027 extern void ext4_check_inodes_bitmap(struct super_block *);
2028 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2029 extern int ext4_init_inode_table(struct super_block *sb,
2030 				 ext4_group_t group, int barrier);
2031 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2032 
2033 /* mballoc.c */
2034 extern long ext4_mb_stats;
2035 extern long ext4_mb_max_to_scan;
2036 extern int ext4_mb_init(struct super_block *);
2037 extern int ext4_mb_release(struct super_block *);
2038 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2039 				struct ext4_allocation_request *, int *);
2040 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2041 extern void ext4_discard_preallocations(struct inode *);
2042 extern int __init ext4_init_mballoc(void);
2043 extern void ext4_exit_mballoc(void);
2044 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2045 			     struct buffer_head *bh, ext4_fsblk_t block,
2046 			     unsigned long count, int flags);
2047 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2048 				   ext4_group_t ngroups);
2049 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2050 		ext4_group_t i, struct ext4_group_desc *desc);
2051 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2052 				ext4_fsblk_t block, unsigned long count);
2053 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2054 
2055 /* inode.c */
2056 struct buffer_head *ext4_getblk(handle_t *, struct inode *,
2057 						ext4_lblk_t, int, int *);
2058 struct buffer_head *ext4_bread(handle_t *, struct inode *,
2059 						ext4_lblk_t, int, int *);
2060 int ext4_get_block_write(struct inode *inode, sector_t iblock,
2061 			 struct buffer_head *bh_result, int create);
2062 int ext4_get_block(struct inode *inode, sector_t iblock,
2063 				struct buffer_head *bh_result, int create);
2064 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2065 			   struct buffer_head *bh, int create);
2066 int ext4_walk_page_buffers(handle_t *handle,
2067 			   struct buffer_head *head,
2068 			   unsigned from,
2069 			   unsigned to,
2070 			   int *partial,
2071 			   int (*fn)(handle_t *handle,
2072 				     struct buffer_head *bh));
2073 int do_journal_get_write_access(handle_t *handle,
2074 				struct buffer_head *bh);
2075 #define FALL_BACK_TO_NONDELALLOC 1
2076 #define CONVERT_INLINE_DATA	 2
2077 
2078 extern struct inode *ext4_iget(struct super_block *, unsigned long);
2079 extern int  ext4_write_inode(struct inode *, struct writeback_control *);
2080 extern int  ext4_setattr(struct dentry *, struct iattr *);
2081 extern int  ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
2082 				struct kstat *stat);
2083 extern void ext4_evict_inode(struct inode *);
2084 extern void ext4_clear_inode(struct inode *);
2085 extern int  ext4_sync_inode(handle_t *, struct inode *);
2086 extern void ext4_dirty_inode(struct inode *, int);
2087 extern int ext4_change_inode_journal_flag(struct inode *, int);
2088 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2089 extern int ext4_can_truncate(struct inode *inode);
2090 extern void ext4_truncate(struct inode *);
2091 extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length);
2092 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
2093 extern void ext4_set_inode_flags(struct inode *);
2094 extern void ext4_get_inode_flags(struct ext4_inode_info *);
2095 extern int ext4_alloc_da_blocks(struct inode *inode);
2096 extern void ext4_set_aops(struct inode *inode);
2097 extern int ext4_writepage_trans_blocks(struct inode *);
2098 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2099 extern int ext4_discard_partial_page_buffers(handle_t *handle,
2100 		struct address_space *mapping, loff_t from,
2101 		loff_t length, int flags);
2102 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
2103 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2104 extern void ext4_da_update_reserve_space(struct inode *inode,
2105 					int used, int quota_claim);
2106 
2107 /* indirect.c */
2108 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2109 				struct ext4_map_blocks *map, int flags);
2110 extern ssize_t ext4_ind_direct_IO(int rw, struct kiocb *iocb,
2111 				const struct iovec *iov, loff_t offset,
2112 				unsigned long nr_segs);
2113 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2114 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks, int chunk);
2115 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2116 extern int ext4_free_hole_blocks(handle_t *handle, struct inode *inode,
2117 				 ext4_lblk_t first, ext4_lblk_t stop);
2118 
2119 /* ioctl.c */
2120 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2121 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2122 
2123 /* migrate.c */
2124 extern int ext4_ext_migrate(struct inode *);
2125 extern int ext4_ind_migrate(struct inode *inode);
2126 
2127 /* namei.c */
2128 extern int ext4_dirent_csum_verify(struct inode *inode,
2129 				   struct ext4_dir_entry *dirent);
2130 extern int ext4_orphan_add(handle_t *, struct inode *);
2131 extern int ext4_orphan_del(handle_t *, struct inode *);
2132 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2133 				__u32 start_minor_hash, __u32 *next_hash);
2134 extern int search_dir(struct buffer_head *bh,
2135 		      char *search_buf,
2136 		      int buf_size,
2137 		      struct inode *dir,
2138 		      const struct qstr *d_name,
2139 		      unsigned int offset,
2140 		      struct ext4_dir_entry_2 **res_dir);
2141 extern int ext4_generic_delete_entry(handle_t *handle,
2142 				     struct inode *dir,
2143 				     struct ext4_dir_entry_2 *de_del,
2144 				     struct buffer_head *bh,
2145 				     void *entry_buf,
2146 				     int buf_size,
2147 				     int csum_size);
2148 
2149 /* resize.c */
2150 extern int ext4_group_add(struct super_block *sb,
2151 				struct ext4_new_group_data *input);
2152 extern int ext4_group_extend(struct super_block *sb,
2153 				struct ext4_super_block *es,
2154 				ext4_fsblk_t n_blocks_count);
2155 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2156 
2157 /* super.c */
2158 extern int ext4_calculate_overhead(struct super_block *sb);
2159 extern int ext4_superblock_csum_verify(struct super_block *sb,
2160 				       struct ext4_super_block *es);
2161 extern void ext4_superblock_csum_set(struct super_block *sb);
2162 extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2163 extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2164 extern void ext4_kvfree(void *ptr);
2165 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2166 				    ext4_group_t ngroup);
2167 extern const char *ext4_decode_error(struct super_block *sb, int errno,
2168 				     char nbuf[16]);
2169 extern __printf(4, 5)
2170 void __ext4_error(struct super_block *, const char *, unsigned int,
2171 		  const char *, ...);
2172 #define ext4_error(sb, message...)	__ext4_error(sb, __func__,	\
2173 						     __LINE__, ## message)
2174 extern __printf(5, 6)
2175 void ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2176 		      const char *, ...);
2177 extern __printf(5, 6)
2178 void ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2179 		     const char *, ...);
2180 extern void __ext4_std_error(struct super_block *, const char *,
2181 			     unsigned int, int);
2182 extern __printf(4, 5)
2183 void __ext4_abort(struct super_block *, const char *, unsigned int,
2184 		  const char *, ...);
2185 #define ext4_abort(sb, message...)	__ext4_abort(sb, __func__, \
2186 						       __LINE__, ## message)
2187 extern __printf(4, 5)
2188 void __ext4_warning(struct super_block *, const char *, unsigned int,
2189 		    const char *, ...);
2190 #define ext4_warning(sb, message...)	__ext4_warning(sb, __func__, \
2191 						       __LINE__, ## message)
2192 extern __printf(3, 4)
2193 void ext4_msg(struct super_block *, const char *, const char *, ...);
2194 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2195 			   const char *, unsigned int, const char *);
2196 #define dump_mmp_msg(sb, mmp, msg)	__dump_mmp_msg(sb, mmp, __func__, \
2197 						       __LINE__, msg)
2198 extern __printf(7, 8)
2199 void __ext4_grp_locked_error(const char *, unsigned int,
2200 			     struct super_block *, ext4_group_t,
2201 			     unsigned long, ext4_fsblk_t,
2202 			     const char *, ...);
2203 #define ext4_grp_locked_error(sb, grp, message...) \
2204 	__ext4_grp_locked_error(__func__, __LINE__, (sb), (grp), ## message)
2205 extern void ext4_update_dynamic_rev(struct super_block *sb);
2206 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2207 					__u32 compat);
2208 extern int ext4_update_rocompat_feature(handle_t *handle,
2209 					struct super_block *sb,	__u32 rocompat);
2210 extern int ext4_update_incompat_feature(handle_t *handle,
2211 					struct super_block *sb,	__u32 incompat);
2212 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2213 				      struct ext4_group_desc *bg);
2214 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2215 				      struct ext4_group_desc *bg);
2216 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2217 				     struct ext4_group_desc *bg);
2218 extern __u32 ext4_free_group_clusters(struct super_block *sb,
2219 				      struct ext4_group_desc *bg);
2220 extern __u32 ext4_free_inodes_count(struct super_block *sb,
2221 				 struct ext4_group_desc *bg);
2222 extern __u32 ext4_used_dirs_count(struct super_block *sb,
2223 				struct ext4_group_desc *bg);
2224 extern __u32 ext4_itable_unused_count(struct super_block *sb,
2225 				   struct ext4_group_desc *bg);
2226 extern void ext4_block_bitmap_set(struct super_block *sb,
2227 				  struct ext4_group_desc *bg, ext4_fsblk_t blk);
2228 extern void ext4_inode_bitmap_set(struct super_block *sb,
2229 				  struct ext4_group_desc *bg, ext4_fsblk_t blk);
2230 extern void ext4_inode_table_set(struct super_block *sb,
2231 				 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2232 extern void ext4_free_group_clusters_set(struct super_block *sb,
2233 					 struct ext4_group_desc *bg,
2234 					 __u32 count);
2235 extern void ext4_free_inodes_set(struct super_block *sb,
2236 				struct ext4_group_desc *bg, __u32 count);
2237 extern void ext4_used_dirs_set(struct super_block *sb,
2238 				struct ext4_group_desc *bg, __u32 count);
2239 extern void ext4_itable_unused_set(struct super_block *sb,
2240 				   struct ext4_group_desc *bg, __u32 count);
2241 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2242 				       struct ext4_group_desc *gdp);
2243 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2244 				     struct ext4_group_desc *gdp);
2245 extern int ext4_register_li_request(struct super_block *sb,
2246 				    ext4_group_t first_not_zeroed);
2247 
2248 static inline int ext4_has_group_desc_csum(struct super_block *sb)
2249 {
2250 	return EXT4_HAS_RO_COMPAT_FEATURE(sb,
2251 					  EXT4_FEATURE_RO_COMPAT_GDT_CSUM |
2252 					  EXT4_FEATURE_RO_COMPAT_METADATA_CSUM);
2253 }
2254 
2255 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2256 {
2257 	return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2258 		le32_to_cpu(es->s_blocks_count_lo);
2259 }
2260 
2261 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2262 {
2263 	return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2264 		le32_to_cpu(es->s_r_blocks_count_lo);
2265 }
2266 
2267 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2268 {
2269 	return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2270 		le32_to_cpu(es->s_free_blocks_count_lo);
2271 }
2272 
2273 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2274 					 ext4_fsblk_t blk)
2275 {
2276 	es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2277 	es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2278 }
2279 
2280 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2281 					      ext4_fsblk_t blk)
2282 {
2283 	es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2284 	es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2285 }
2286 
2287 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2288 					   ext4_fsblk_t blk)
2289 {
2290 	es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2291 	es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2292 }
2293 
2294 static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
2295 {
2296 	if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2297 		return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2298 			le32_to_cpu(raw_inode->i_size_lo);
2299 	else
2300 		return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2301 }
2302 
2303 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2304 {
2305 	raw_inode->i_size_lo = cpu_to_le32(i_size);
2306 	raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2307 }
2308 
2309 static inline
2310 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2311 					    ext4_group_t group)
2312 {
2313 	 struct ext4_group_info ***grp_info;
2314 	 long indexv, indexh;
2315 	 grp_info = EXT4_SB(sb)->s_group_info;
2316 	 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2317 	 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2318 	 return grp_info[indexv][indexh];
2319 }
2320 
2321 /*
2322  * Reading s_groups_count requires using smp_rmb() afterwards.  See
2323  * the locking protocol documented in the comments of ext4_group_add()
2324  * in resize.c
2325  */
2326 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2327 {
2328 	ext4_group_t	ngroups = EXT4_SB(sb)->s_groups_count;
2329 
2330 	smp_rmb();
2331 	return ngroups;
2332 }
2333 
2334 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2335 					     ext4_group_t block_group)
2336 {
2337 	return block_group >> sbi->s_log_groups_per_flex;
2338 }
2339 
2340 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2341 {
2342 	return 1 << sbi->s_log_groups_per_flex;
2343 }
2344 
2345 #define ext4_std_error(sb, errno)				\
2346 do {								\
2347 	if ((errno))						\
2348 		__ext4_std_error((sb), __func__, __LINE__, (errno));	\
2349 } while (0)
2350 
2351 #ifdef CONFIG_SMP
2352 /* Each CPU can accumulate percpu_counter_batch clusters in their local
2353  * counters. So we need to make sure we have free clusters more
2354  * than percpu_counter_batch  * nr_cpu_ids. Also add a window of 4 times.
2355  */
2356 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2357 #else
2358 #define EXT4_FREECLUSTERS_WATERMARK 0
2359 #endif
2360 
2361 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2362 {
2363 	/*
2364 	 * XXX: replace with spinlock if seen contended -bzzz
2365 	 */
2366 	down_write(&EXT4_I(inode)->i_data_sem);
2367 	if (newsize > EXT4_I(inode)->i_disksize)
2368 		EXT4_I(inode)->i_disksize = newsize;
2369 	up_write(&EXT4_I(inode)->i_data_sem);
2370 	return ;
2371 }
2372 
2373 struct ext4_group_info {
2374 	unsigned long   bb_state;
2375 	struct rb_root  bb_free_root;
2376 	ext4_grpblk_t	bb_first_free;	/* first free block */
2377 	ext4_grpblk_t	bb_free;	/* total free blocks */
2378 	ext4_grpblk_t	bb_fragments;	/* nr of freespace fragments */
2379 	ext4_grpblk_t	bb_largest_free_order;/* order of largest frag in BG */
2380 	struct          list_head bb_prealloc_list;
2381 #ifdef DOUBLE_CHECK
2382 	void            *bb_bitmap;
2383 #endif
2384 	struct rw_semaphore alloc_sem;
2385 	ext4_grpblk_t	bb_counters[];	/* Nr of free power-of-two-block
2386 					 * regions, index is order.
2387 					 * bb_counters[3] = 5 means
2388 					 * 5 free 8-block regions. */
2389 };
2390 
2391 #define EXT4_GROUP_INFO_NEED_INIT_BIT		0
2392 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT		1
2393 
2394 #define EXT4_MB_GRP_NEED_INIT(grp)	\
2395 	(test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2396 
2397 #define EXT4_MB_GRP_WAS_TRIMMED(grp)	\
2398 	(test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2399 #define EXT4_MB_GRP_SET_TRIMMED(grp)	\
2400 	(set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2401 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp)	\
2402 	(clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2403 
2404 #define EXT4_MAX_CONTENTION		8
2405 #define EXT4_CONTENTION_THRESHOLD	2
2406 
2407 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2408 					      ext4_group_t group)
2409 {
2410 	return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2411 }
2412 
2413 /*
2414  * Returns true if the filesystem is busy enough that attempts to
2415  * access the block group locks has run into contention.
2416  */
2417 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2418 {
2419 	return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2420 }
2421 
2422 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2423 {
2424 	spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2425 	if (spin_trylock(lock))
2426 		/*
2427 		 * We're able to grab the lock right away, so drop the
2428 		 * lock contention counter.
2429 		 */
2430 		atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2431 	else {
2432 		/*
2433 		 * The lock is busy, so bump the contention counter,
2434 		 * and then wait on the spin lock.
2435 		 */
2436 		atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2437 				  EXT4_MAX_CONTENTION);
2438 		spin_lock(lock);
2439 	}
2440 }
2441 
2442 static inline void ext4_unlock_group(struct super_block *sb,
2443 					ext4_group_t group)
2444 {
2445 	spin_unlock(ext4_group_lock_ptr(sb, group));
2446 }
2447 
2448 /*
2449  * Block validity checking
2450  */
2451 #define ext4_check_indirect_blockref(inode, bh)				\
2452 	ext4_check_blockref(__func__, __LINE__, inode,			\
2453 			    (__le32 *)(bh)->b_data,			\
2454 			    EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2455 
2456 #define ext4_ind_check_inode(inode)					\
2457 	ext4_check_blockref(__func__, __LINE__, inode,			\
2458 			    EXT4_I(inode)->i_data,			\
2459 			    EXT4_NDIR_BLOCKS)
2460 
2461 /*
2462  * Inodes and files operations
2463  */
2464 
2465 /* dir.c */
2466 extern const struct file_operations ext4_dir_operations;
2467 
2468 /* file.c */
2469 extern const struct inode_operations ext4_file_inode_operations;
2470 extern const struct file_operations ext4_file_operations;
2471 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
2472 extern void ext4_unwritten_wait(struct inode *inode);
2473 
2474 /* inline.c */
2475 extern int ext4_has_inline_data(struct inode *inode);
2476 extern int ext4_get_inline_size(struct inode *inode);
2477 extern int ext4_get_max_inline_size(struct inode *inode);
2478 extern int ext4_find_inline_data_nolock(struct inode *inode);
2479 extern void ext4_write_inline_data(struct inode *inode,
2480 				   struct ext4_iloc *iloc,
2481 				   void *buffer, loff_t pos,
2482 				   unsigned int len);
2483 extern int ext4_prepare_inline_data(handle_t *handle, struct inode *inode,
2484 				    unsigned int len);
2485 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
2486 				 unsigned int len);
2487 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
2488 
2489 extern int ext4_readpage_inline(struct inode *inode, struct page *page);
2490 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
2491 					 struct inode *inode,
2492 					 loff_t pos, unsigned len,
2493 					 unsigned flags,
2494 					 struct page **pagep);
2495 extern int ext4_write_inline_data_end(struct inode *inode,
2496 				      loff_t pos, unsigned len,
2497 				      unsigned copied,
2498 				      struct page *page);
2499 extern struct buffer_head *
2500 ext4_journalled_write_inline_data(struct inode *inode,
2501 				  unsigned len,
2502 				  struct page *page);
2503 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
2504 					   struct inode *inode,
2505 					   loff_t pos, unsigned len,
2506 					   unsigned flags,
2507 					   struct page **pagep,
2508 					   void **fsdata);
2509 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
2510 					 unsigned len, unsigned copied,
2511 					 struct page *page);
2512 extern int ext4_try_add_inline_entry(handle_t *handle, struct dentry *dentry,
2513 				     struct inode *inode);
2514 extern int ext4_try_create_inline_dir(handle_t *handle,
2515 				      struct inode *parent,
2516 				      struct inode *inode);
2517 extern int ext4_read_inline_dir(struct file *filp,
2518 				void *dirent, filldir_t filldir,
2519 				int *has_inline_data);
2520 extern int htree_inlinedir_to_tree(struct file *dir_file,
2521 				   struct inode *dir, ext4_lblk_t block,
2522 				   struct dx_hash_info *hinfo,
2523 				   __u32 start_hash, __u32 start_minor_hash,
2524 				   int *has_inline_data);
2525 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
2526 					const struct qstr *d_name,
2527 					struct ext4_dir_entry_2 **res_dir,
2528 					int *has_inline_data);
2529 extern int ext4_delete_inline_entry(handle_t *handle,
2530 				    struct inode *dir,
2531 				    struct ext4_dir_entry_2 *de_del,
2532 				    struct buffer_head *bh,
2533 				    int *has_inline_data);
2534 extern int empty_inline_dir(struct inode *dir, int *has_inline_data);
2535 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
2536 					struct ext4_dir_entry_2 **parent_de,
2537 					int *retval);
2538 extern int ext4_inline_data_fiemap(struct inode *inode,
2539 				   struct fiemap_extent_info *fieinfo,
2540 				   int *has_inline);
2541 extern int ext4_try_to_evict_inline_data(handle_t *handle,
2542 					 struct inode *inode,
2543 					 int needed);
2544 extern void ext4_inline_data_truncate(struct inode *inode, int *has_inline);
2545 
2546 extern int ext4_convert_inline_data(struct inode *inode);
2547 
2548 /* namei.c */
2549 extern const struct inode_operations ext4_dir_inode_operations;
2550 extern const struct inode_operations ext4_special_inode_operations;
2551 extern struct dentry *ext4_get_parent(struct dentry *child);
2552 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
2553 				 struct ext4_dir_entry_2 *de,
2554 				 int blocksize, int csum_size,
2555 				 unsigned int parent_ino, int dotdot_real_len);
2556 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t,
2557 				   unsigned int blocksize);
2558 extern int ext4_handle_dirty_dirent_node(handle_t *handle,
2559 					 struct inode *inode,
2560 					 struct buffer_head *bh);
2561 #define S_SHIFT 12
2562 static unsigned char ext4_type_by_mode[S_IFMT >> S_SHIFT] = {
2563 	[S_IFREG >> S_SHIFT]	= EXT4_FT_REG_FILE,
2564 	[S_IFDIR >> S_SHIFT]	= EXT4_FT_DIR,
2565 	[S_IFCHR >> S_SHIFT]	= EXT4_FT_CHRDEV,
2566 	[S_IFBLK >> S_SHIFT]	= EXT4_FT_BLKDEV,
2567 	[S_IFIFO >> S_SHIFT]	= EXT4_FT_FIFO,
2568 	[S_IFSOCK >> S_SHIFT]	= EXT4_FT_SOCK,
2569 	[S_IFLNK >> S_SHIFT]	= EXT4_FT_SYMLINK,
2570 };
2571 
2572 static inline void ext4_set_de_type(struct super_block *sb,
2573 				struct ext4_dir_entry_2 *de,
2574 				umode_t mode) {
2575 	if (EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE))
2576 		de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
2577 }
2578 
2579 
2580 /* symlink.c */
2581 extern const struct inode_operations ext4_symlink_inode_operations;
2582 extern const struct inode_operations ext4_fast_symlink_inode_operations;
2583 
2584 /* block_validity */
2585 extern void ext4_release_system_zone(struct super_block *sb);
2586 extern int ext4_setup_system_zone(struct super_block *sb);
2587 extern int __init ext4_init_system_zone(void);
2588 extern void ext4_exit_system_zone(void);
2589 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
2590 				 ext4_fsblk_t start_blk,
2591 				 unsigned int count);
2592 extern int ext4_check_blockref(const char *, unsigned int,
2593 			       struct inode *, __le32 *, unsigned int);
2594 
2595 /* extents.c */
2596 struct ext4_ext_path;
2597 struct ext4_extent;
2598 
2599 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
2600 extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
2601 extern int ext4_ext_index_trans_blocks(struct inode *inode, int nrblocks,
2602 				       int chunk);
2603 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
2604 			       struct ext4_map_blocks *map, int flags);
2605 extern void ext4_ext_truncate(handle_t *, struct inode *);
2606 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
2607 				 ext4_lblk_t end);
2608 extern void ext4_ext_init(struct super_block *);
2609 extern void ext4_ext_release(struct super_block *);
2610 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
2611 			  loff_t len);
2612 extern int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset,
2613 			  ssize_t len);
2614 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
2615 			   struct ext4_map_blocks *map, int flags);
2616 extern int ext4_ext_calc_metadata_amount(struct inode *inode,
2617 					 ext4_lblk_t lblocks);
2618 extern int ext4_extent_tree_init(handle_t *, struct inode *);
2619 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
2620 						   int num,
2621 						   struct ext4_ext_path *path);
2622 extern int ext4_can_extents_be_merged(struct inode *inode,
2623 				      struct ext4_extent *ex1,
2624 				      struct ext4_extent *ex2);
2625 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
2626 				  struct ext4_ext_path *,
2627 				  struct ext4_extent *, int);
2628 extern struct ext4_ext_path *ext4_ext_find_extent(struct inode *, ext4_lblk_t,
2629 						  struct ext4_ext_path *);
2630 extern void ext4_ext_drop_refs(struct ext4_ext_path *);
2631 extern int ext4_ext_check_inode(struct inode *inode);
2632 extern int ext4_find_delalloc_range(struct inode *inode,
2633 				    ext4_lblk_t lblk_start,
2634 				    ext4_lblk_t lblk_end);
2635 extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk);
2636 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
2637 			__u64 start, __u64 len);
2638 
2639 
2640 /* move_extent.c */
2641 extern void ext4_double_down_write_data_sem(struct inode *first,
2642 					    struct inode *second);
2643 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
2644 					  struct inode *donor_inode);
2645 void ext4_inode_double_lock(struct inode *inode1, struct inode *inode2);
2646 void ext4_inode_double_unlock(struct inode *inode1, struct inode *inode2);
2647 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
2648 			     __u64 start_orig, __u64 start_donor,
2649 			     __u64 len, __u64 *moved_len);
2650 
2651 /* page-io.c */
2652 extern int __init ext4_init_pageio(void);
2653 extern void ext4_add_complete_io(ext4_io_end_t *io_end);
2654 extern void ext4_exit_pageio(void);
2655 extern void ext4_ioend_shutdown(struct inode *);
2656 extern void ext4_free_io_end(ext4_io_end_t *io);
2657 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
2658 extern void ext4_end_io_work(struct work_struct *work);
2659 extern void ext4_io_submit(struct ext4_io_submit *io);
2660 extern int ext4_bio_write_page(struct ext4_io_submit *io,
2661 			       struct page *page,
2662 			       int len,
2663 			       struct writeback_control *wbc);
2664 
2665 /* mmp.c */
2666 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
2667 extern void ext4_mmp_csum_set(struct super_block *sb, struct mmp_struct *mmp);
2668 extern int ext4_mmp_csum_verify(struct super_block *sb,
2669 				struct mmp_struct *mmp);
2670 
2671 /* BH_Uninit flag: blocks are allocated but uninitialized on disk */
2672 enum ext4_state_bits {
2673 	BH_Uninit	/* blocks are allocated but uninitialized on disk */
2674 	  = BH_JBDPrivateStart,
2675 	BH_AllocFromCluster,	/* allocated blocks were part of already
2676 				 * allocated cluster. Note that this flag will
2677 				 * never, ever appear in a buffer_head's state
2678 				 * flag. See EXT4_MAP_FROM_CLUSTER to see where
2679 				 * this is used. */
2680 };
2681 
2682 BUFFER_FNS(Uninit, uninit)
2683 TAS_BUFFER_FNS(Uninit, uninit)
2684 
2685 /*
2686  * Add new method to test whether block and inode bitmaps are properly
2687  * initialized. With uninit_bg reading the block from disk is not enough
2688  * to mark the bitmap uptodate. We need to also zero-out the bitmap
2689  */
2690 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
2691 
2692 static inline int bitmap_uptodate(struct buffer_head *bh)
2693 {
2694 	return (buffer_uptodate(bh) &&
2695 			test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
2696 }
2697 static inline void set_bitmap_uptodate(struct buffer_head *bh)
2698 {
2699 	set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
2700 }
2701 
2702 /*
2703  * Disable DIO read nolock optimization, so new dioreaders will be forced
2704  * to grab i_mutex
2705  */
2706 static inline void ext4_inode_block_unlocked_dio(struct inode *inode)
2707 {
2708 	ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
2709 	smp_mb();
2710 }
2711 static inline void ext4_inode_resume_unlocked_dio(struct inode *inode)
2712 {
2713 	smp_mb();
2714 	ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
2715 }
2716 
2717 #define in_range(b, first, len)	((b) >= (first) && (b) <= (first) + (len) - 1)
2718 
2719 /* For ioend & aio unwritten conversion wait queues */
2720 #define EXT4_WQ_HASH_SZ		37
2721 #define ext4_ioend_wq(v)   (&ext4__ioend_wq[((unsigned long)(v)) %\
2722 					    EXT4_WQ_HASH_SZ])
2723 #define ext4_aio_mutex(v)  (&ext4__aio_mutex[((unsigned long)(v)) %\
2724 					     EXT4_WQ_HASH_SZ])
2725 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
2726 extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ];
2727 
2728 #define EXT4_RESIZING	0
2729 extern int ext4_resize_begin(struct super_block *sb);
2730 extern void ext4_resize_end(struct super_block *sb);
2731 
2732 #endif	/* __KERNEL__ */
2733 
2734 #endif	/* _EXT4_H */
2735