xref: /openbmc/linux/fs/ext4/ext4.h (revision 424f0750edd5af866f80f5e65998e0610503cb5c)
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 #ifdef __KERNEL__
33 #include <linux/compat.h>
34 #endif
35 
36 /*
37  * The fourth extended filesystem constants/structures
38  */
39 
40 /*
41  * Define EXT4FS_DEBUG to produce debug messages
42  */
43 #undef EXT4FS_DEBUG
44 
45 /*
46  * Debug code
47  */
48 #ifdef EXT4FS_DEBUG
49 #define ext4_debug(f, a...)						\
50 	do {								\
51 		printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:",	\
52 			__FILE__, __LINE__, __func__);			\
53 		printk(KERN_DEBUG f, ## a);				\
54 	} while (0)
55 #else
56 #define ext4_debug(f, a...)	do {} while (0)
57 #endif
58 
59 #define EXT4_ERROR_INODE(inode, fmt, a...) \
60 	ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
61 
62 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...)			\
63 	ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
64 
65 #define EXT4_ERROR_FILE(file, block, fmt, a...)				\
66 	ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
67 
68 /* data type for block offset of block group */
69 typedef int ext4_grpblk_t;
70 
71 /* data type for filesystem-wide blocks number */
72 typedef unsigned long long ext4_fsblk_t;
73 
74 /* data type for file logical block number */
75 typedef __u32 ext4_lblk_t;
76 
77 /* data type for block group number */
78 typedef unsigned int ext4_group_t;
79 
80 /*
81  * Flags used in mballoc's allocation_context flags field.
82  *
83  * Also used to show what's going on for debugging purposes when the
84  * flag field is exported via the traceport interface
85  */
86 
87 /* prefer goal again. length */
88 #define EXT4_MB_HINT_MERGE		0x0001
89 /* blocks already reserved */
90 #define EXT4_MB_HINT_RESERVED		0x0002
91 /* metadata is being allocated */
92 #define EXT4_MB_HINT_METADATA		0x0004
93 /* first blocks in the file */
94 #define EXT4_MB_HINT_FIRST		0x0008
95 /* search for the best chunk */
96 #define EXT4_MB_HINT_BEST		0x0010
97 /* data is being allocated */
98 #define EXT4_MB_HINT_DATA		0x0020
99 /* don't preallocate (for tails) */
100 #define EXT4_MB_HINT_NOPREALLOC		0x0040
101 /* allocate for locality group */
102 #define EXT4_MB_HINT_GROUP_ALLOC	0x0080
103 /* allocate goal blocks or none */
104 #define EXT4_MB_HINT_GOAL_ONLY		0x0100
105 /* goal is meaningful */
106 #define EXT4_MB_HINT_TRY_GOAL		0x0200
107 /* blocks already pre-reserved by delayed allocation */
108 #define EXT4_MB_DELALLOC_RESERVED	0x0400
109 /* We are doing stream allocation */
110 #define EXT4_MB_STREAM_ALLOC		0x0800
111 /* Use reserved root blocks if needed */
112 #define EXT4_MB_USE_ROOT_BLOCKS		0x1000
113 
114 struct ext4_allocation_request {
115 	/* target inode for block we're allocating */
116 	struct inode *inode;
117 	/* how many blocks we want to allocate */
118 	unsigned int len;
119 	/* logical block in target inode */
120 	ext4_lblk_t logical;
121 	/* the closest logical allocated block to the left */
122 	ext4_lblk_t lleft;
123 	/* the closest logical allocated block to the right */
124 	ext4_lblk_t lright;
125 	/* phys. target (a hint) */
126 	ext4_fsblk_t goal;
127 	/* phys. block for the closest logical allocated block to the left */
128 	ext4_fsblk_t pleft;
129 	/* phys. block for the closest logical allocated block to the right */
130 	ext4_fsblk_t pright;
131 	/* flags. see above EXT4_MB_HINT_* */
132 	unsigned int flags;
133 };
134 
135 /*
136  * Logical to physical block mapping, used by ext4_map_blocks()
137  *
138  * This structure is used to pass requests into ext4_map_blocks() as
139  * well as to store the information returned by ext4_map_blocks().  It
140  * takes less room on the stack than a struct buffer_head.
141  */
142 #define EXT4_MAP_NEW		(1 << BH_New)
143 #define EXT4_MAP_MAPPED		(1 << BH_Mapped)
144 #define EXT4_MAP_UNWRITTEN	(1 << BH_Unwritten)
145 #define EXT4_MAP_BOUNDARY	(1 << BH_Boundary)
146 #define EXT4_MAP_UNINIT		(1 << BH_Uninit)
147 /* Sometimes (in the bigalloc case, from ext4_da_get_block_prep) the caller of
148  * ext4_map_blocks wants to know whether or not the underlying cluster has
149  * already been accounted for. EXT4_MAP_FROM_CLUSTER conveys to the caller that
150  * the requested mapping was from previously mapped (or delayed allocated)
151  * cluster. We use BH_AllocFromCluster only for this flag. BH_AllocFromCluster
152  * should never appear on buffer_head's state flags.
153  */
154 #define EXT4_MAP_FROM_CLUSTER	(1 << BH_AllocFromCluster)
155 #define EXT4_MAP_FLAGS		(EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
156 				 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\
157 				 EXT4_MAP_UNINIT | EXT4_MAP_FROM_CLUSTER)
158 
159 struct ext4_map_blocks {
160 	ext4_fsblk_t m_pblk;
161 	ext4_lblk_t m_lblk;
162 	unsigned int m_len;
163 	unsigned int m_flags;
164 };
165 
166 /*
167  * For delayed allocation tracking
168  */
169 struct mpage_da_data {
170 	struct inode *inode;
171 	sector_t b_blocknr;		/* start block number of extent */
172 	size_t b_size;			/* size of extent */
173 	unsigned long b_state;		/* state of the extent */
174 	unsigned long first_page, next_page;	/* extent of pages */
175 	struct writeback_control *wbc;
176 	int io_done;
177 	int pages_written;
178 	int retval;
179 };
180 
181 /*
182  * Flags for ext4_io_end->flags
183  */
184 #define	EXT4_IO_END_UNWRITTEN	0x0001
185 #define EXT4_IO_END_ERROR	0x0002
186 #define EXT4_IO_END_QUEUED	0x0004
187 
188 struct ext4_io_page {
189 	struct page	*p_page;
190 	atomic_t	p_count;
191 };
192 
193 #define MAX_IO_PAGES 128
194 
195 typedef struct ext4_io_end {
196 	struct list_head	list;		/* per-file finished IO list */
197 	struct inode		*inode;		/* file being written to */
198 	unsigned int		flag;		/* unwritten or not */
199 	struct page		*page;		/* page struct for buffer write */
200 	loff_t			offset;		/* offset in the file */
201 	ssize_t			size;		/* size of the extent */
202 	struct work_struct	work;		/* data work queue */
203 	struct kiocb		*iocb;		/* iocb struct for AIO */
204 	int			result;		/* error value for AIO */
205 	int			num_io_pages;
206 	struct ext4_io_page	*pages[MAX_IO_PAGES];
207 } ext4_io_end_t;
208 
209 struct ext4_io_submit {
210 	int			io_op;
211 	struct bio		*io_bio;
212 	ext4_io_end_t		*io_end;
213 	struct ext4_io_page	*io_page;
214 	sector_t		io_next_block;
215 };
216 
217 /*
218  * Special inodes numbers
219  */
220 #define	EXT4_BAD_INO		 1	/* Bad blocks inode */
221 #define EXT4_ROOT_INO		 2	/* Root inode */
222 #define EXT4_USR_QUOTA_INO	 3	/* User quota inode */
223 #define EXT4_GRP_QUOTA_INO	 4	/* Group quota inode */
224 #define EXT4_BOOT_LOADER_INO	 5	/* Boot loader inode */
225 #define EXT4_UNDEL_DIR_INO	 6	/* Undelete directory inode */
226 #define EXT4_RESIZE_INO		 7	/* Reserved group descriptors inode */
227 #define EXT4_JOURNAL_INO	 8	/* Journal inode */
228 
229 /* First non-reserved inode for old ext4 filesystems */
230 #define EXT4_GOOD_OLD_FIRST_INO	11
231 
232 /*
233  * Maximal count of links to a file
234  */
235 #define EXT4_LINK_MAX		65000
236 
237 /*
238  * Macro-instructions used to manage several block sizes
239  */
240 #define EXT4_MIN_BLOCK_SIZE		1024
241 #define	EXT4_MAX_BLOCK_SIZE		65536
242 #define EXT4_MIN_BLOCK_LOG_SIZE		10
243 #define EXT4_MAX_BLOCK_LOG_SIZE		16
244 #ifdef __KERNEL__
245 # define EXT4_BLOCK_SIZE(s)		((s)->s_blocksize)
246 #else
247 # define EXT4_BLOCK_SIZE(s)		(EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
248 #endif
249 #define	EXT4_ADDR_PER_BLOCK(s)		(EXT4_BLOCK_SIZE(s) / sizeof(__u32))
250 #define EXT4_CLUSTER_SIZE(s)		(EXT4_BLOCK_SIZE(s) << \
251 					 EXT4_SB(s)->s_cluster_bits)
252 #ifdef __KERNEL__
253 # define EXT4_BLOCK_SIZE_BITS(s)	((s)->s_blocksize_bits)
254 # define EXT4_CLUSTER_BITS(s)		(EXT4_SB(s)->s_cluster_bits)
255 #else
256 # define EXT4_BLOCK_SIZE_BITS(s)	((s)->s_log_block_size + 10)
257 #endif
258 #ifdef __KERNEL__
259 #define	EXT4_ADDR_PER_BLOCK_BITS(s)	(EXT4_SB(s)->s_addr_per_block_bits)
260 #define EXT4_INODE_SIZE(s)		(EXT4_SB(s)->s_inode_size)
261 #define EXT4_FIRST_INO(s)		(EXT4_SB(s)->s_first_ino)
262 #else
263 #define EXT4_INODE_SIZE(s)	(((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
264 				 EXT4_GOOD_OLD_INODE_SIZE : \
265 				 (s)->s_inode_size)
266 #define EXT4_FIRST_INO(s)	(((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
267 				 EXT4_GOOD_OLD_FIRST_INO : \
268 				 (s)->s_first_ino)
269 #endif
270 #define EXT4_BLOCK_ALIGN(size, blkbits)		ALIGN((size), (1 << (blkbits)))
271 
272 /* Translate a block number to a cluster number */
273 #define EXT4_B2C(sbi, blk)	((blk) >> (sbi)->s_cluster_bits)
274 /* Translate a cluster number to a block number */
275 #define EXT4_C2B(sbi, cluster)	((cluster) << (sbi)->s_cluster_bits)
276 /* Translate # of blks to # of clusters */
277 #define EXT4_NUM_B2C(sbi, blks)	(((blks) + (sbi)->s_cluster_ratio - 1) >> \
278 				 (sbi)->s_cluster_bits)
279 
280 /*
281  * Structure of a blocks group descriptor
282  */
283 struct ext4_group_desc
284 {
285 	__le32	bg_block_bitmap_lo;	/* Blocks bitmap block */
286 	__le32	bg_inode_bitmap_lo;	/* Inodes bitmap block */
287 	__le32	bg_inode_table_lo;	/* Inodes table block */
288 	__le16	bg_free_blocks_count_lo;/* Free blocks count */
289 	__le16	bg_free_inodes_count_lo;/* Free inodes count */
290 	__le16	bg_used_dirs_count_lo;	/* Directories count */
291 	__le16	bg_flags;		/* EXT4_BG_flags (INODE_UNINIT, etc) */
292 	__u32	bg_reserved[2];		/* Likely block/inode bitmap checksum */
293 	__le16  bg_itable_unused_lo;	/* Unused inodes count */
294 	__le16  bg_checksum;		/* crc16(sb_uuid+group+desc) */
295 	__le32	bg_block_bitmap_hi;	/* Blocks bitmap block MSB */
296 	__le32	bg_inode_bitmap_hi;	/* Inodes bitmap block MSB */
297 	__le32	bg_inode_table_hi;	/* Inodes table block MSB */
298 	__le16	bg_free_blocks_count_hi;/* Free blocks count MSB */
299 	__le16	bg_free_inodes_count_hi;/* Free inodes count MSB */
300 	__le16	bg_used_dirs_count_hi;	/* Directories count MSB */
301 	__le16  bg_itable_unused_hi;    /* Unused inodes count MSB */
302 	__u32	bg_reserved2[3];
303 };
304 
305 /*
306  * Structure of a flex block group info
307  */
308 
309 struct flex_groups {
310 	atomic_t free_inodes;
311 	atomic_t free_clusters;
312 	atomic_t used_dirs;
313 };
314 
315 #define EXT4_BG_INODE_UNINIT	0x0001 /* Inode table/bitmap not in use */
316 #define EXT4_BG_BLOCK_UNINIT	0x0002 /* Block bitmap not in use */
317 #define EXT4_BG_INODE_ZEROED	0x0004 /* On-disk itable initialized to zero */
318 
319 /*
320  * Macro-instructions used to manage group descriptors
321  */
322 #define EXT4_MIN_DESC_SIZE		32
323 #define EXT4_MIN_DESC_SIZE_64BIT	64
324 #define	EXT4_MAX_DESC_SIZE		EXT4_MIN_BLOCK_SIZE
325 #define EXT4_DESC_SIZE(s)		(EXT4_SB(s)->s_desc_size)
326 #ifdef __KERNEL__
327 # define EXT4_BLOCKS_PER_GROUP(s)	(EXT4_SB(s)->s_blocks_per_group)
328 # define EXT4_CLUSTERS_PER_GROUP(s)	(EXT4_SB(s)->s_clusters_per_group)
329 # define EXT4_DESC_PER_BLOCK(s)		(EXT4_SB(s)->s_desc_per_block)
330 # define EXT4_INODES_PER_GROUP(s)	(EXT4_SB(s)->s_inodes_per_group)
331 # define EXT4_DESC_PER_BLOCK_BITS(s)	(EXT4_SB(s)->s_desc_per_block_bits)
332 #else
333 # define EXT4_BLOCKS_PER_GROUP(s)	((s)->s_blocks_per_group)
334 # define EXT4_DESC_PER_BLOCK(s)		(EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
335 # define EXT4_INODES_PER_GROUP(s)	((s)->s_inodes_per_group)
336 #endif
337 
338 /*
339  * Constants relative to the data blocks
340  */
341 #define	EXT4_NDIR_BLOCKS		12
342 #define	EXT4_IND_BLOCK			EXT4_NDIR_BLOCKS
343 #define	EXT4_DIND_BLOCK			(EXT4_IND_BLOCK + 1)
344 #define	EXT4_TIND_BLOCK			(EXT4_DIND_BLOCK + 1)
345 #define	EXT4_N_BLOCKS			(EXT4_TIND_BLOCK + 1)
346 
347 /*
348  * Inode flags
349  */
350 #define	EXT4_SECRM_FL			0x00000001 /* Secure deletion */
351 #define	EXT4_UNRM_FL			0x00000002 /* Undelete */
352 #define	EXT4_COMPR_FL			0x00000004 /* Compress file */
353 #define EXT4_SYNC_FL			0x00000008 /* Synchronous updates */
354 #define EXT4_IMMUTABLE_FL		0x00000010 /* Immutable file */
355 #define EXT4_APPEND_FL			0x00000020 /* writes to file may only append */
356 #define EXT4_NODUMP_FL			0x00000040 /* do not dump file */
357 #define EXT4_NOATIME_FL			0x00000080 /* do not update atime */
358 /* Reserved for compression usage... */
359 #define EXT4_DIRTY_FL			0x00000100
360 #define EXT4_COMPRBLK_FL		0x00000200 /* One or more compressed clusters */
361 #define EXT4_NOCOMPR_FL			0x00000400 /* Don't compress */
362 #define EXT4_ECOMPR_FL			0x00000800 /* Compression error */
363 /* End compression flags --- maybe not all used */
364 #define EXT4_INDEX_FL			0x00001000 /* hash-indexed directory */
365 #define EXT4_IMAGIC_FL			0x00002000 /* AFS directory */
366 #define EXT4_JOURNAL_DATA_FL		0x00004000 /* file data should be journaled */
367 #define EXT4_NOTAIL_FL			0x00008000 /* file tail should not be merged */
368 #define EXT4_DIRSYNC_FL			0x00010000 /* dirsync behaviour (directories only) */
369 #define EXT4_TOPDIR_FL			0x00020000 /* Top of directory hierarchies*/
370 #define EXT4_HUGE_FILE_FL               0x00040000 /* Set to each huge file */
371 #define EXT4_EXTENTS_FL			0x00080000 /* Inode uses extents */
372 #define EXT4_EA_INODE_FL	        0x00200000 /* Inode used for large EA */
373 #define EXT4_EOFBLOCKS_FL		0x00400000 /* Blocks allocated beyond EOF */
374 #define EXT4_RESERVED_FL		0x80000000 /* reserved for ext4 lib */
375 
376 #define EXT4_FL_USER_VISIBLE		0x004BDFFF /* User visible flags */
377 #define EXT4_FL_USER_MODIFIABLE		0x004B80FF /* User modifiable flags */
378 
379 /* Flags that should be inherited by new inodes from their parent. */
380 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
381 			   EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
382 			   EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
383 			   EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
384 
385 /* Flags that are appropriate for regular files (all but dir-specific ones). */
386 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
387 
388 /* Flags that are appropriate for non-directories/regular files. */
389 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
390 
391 /* Mask out flags that are inappropriate for the given type of inode. */
392 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
393 {
394 	if (S_ISDIR(mode))
395 		return flags;
396 	else if (S_ISREG(mode))
397 		return flags & EXT4_REG_FLMASK;
398 	else
399 		return flags & EXT4_OTHER_FLMASK;
400 }
401 
402 /*
403  * Inode flags used for atomic set/get
404  */
405 enum {
406 	EXT4_INODE_SECRM	= 0,	/* Secure deletion */
407 	EXT4_INODE_UNRM		= 1,	/* Undelete */
408 	EXT4_INODE_COMPR	= 2,	/* Compress file */
409 	EXT4_INODE_SYNC		= 3,	/* Synchronous updates */
410 	EXT4_INODE_IMMUTABLE	= 4,	/* Immutable file */
411 	EXT4_INODE_APPEND	= 5,	/* writes to file may only append */
412 	EXT4_INODE_NODUMP	= 6,	/* do not dump file */
413 	EXT4_INODE_NOATIME	= 7,	/* do not update atime */
414 /* Reserved for compression usage... */
415 	EXT4_INODE_DIRTY	= 8,
416 	EXT4_INODE_COMPRBLK	= 9,	/* One or more compressed clusters */
417 	EXT4_INODE_NOCOMPR	= 10,	/* Don't compress */
418 	EXT4_INODE_ECOMPR	= 11,	/* Compression error */
419 /* End compression flags --- maybe not all used */
420 	EXT4_INODE_INDEX	= 12,	/* hash-indexed directory */
421 	EXT4_INODE_IMAGIC	= 13,	/* AFS directory */
422 	EXT4_INODE_JOURNAL_DATA	= 14,	/* file data should be journaled */
423 	EXT4_INODE_NOTAIL	= 15,	/* file tail should not be merged */
424 	EXT4_INODE_DIRSYNC	= 16,	/* dirsync behaviour (directories only) */
425 	EXT4_INODE_TOPDIR	= 17,	/* Top of directory hierarchies*/
426 	EXT4_INODE_HUGE_FILE	= 18,	/* Set to each huge file */
427 	EXT4_INODE_EXTENTS	= 19,	/* Inode uses extents */
428 	EXT4_INODE_EA_INODE	= 21,	/* Inode used for large EA */
429 	EXT4_INODE_EOFBLOCKS	= 22,	/* Blocks allocated beyond EOF */
430 	EXT4_INODE_RESERVED	= 31,	/* reserved for ext4 lib */
431 };
432 
433 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
434 #define CHECK_FLAG_VALUE(FLAG) if (!TEST_FLAG_VALUE(FLAG)) { \
435 	printk(KERN_EMERG "EXT4 flag fail: " #FLAG ": %d %d\n", \
436 		EXT4_##FLAG##_FL, EXT4_INODE_##FLAG); BUG_ON(1); }
437 
438 /*
439  * Since it's pretty easy to mix up bit numbers and hex values, and we
440  * can't do a compile-time test for ENUM values, we use a run-time
441  * test to make sure that EXT4_XXX_FL is consistent with respect to
442  * EXT4_INODE_XXX.  If all is well the printk and BUG_ON will all drop
443  * out so it won't cost any extra space in the compiled kernel image.
444  * But it's important that these values are the same, since we are
445  * using EXT4_INODE_XXX to test for the flag values, but EXT4_XX_FL
446  * must be consistent with the values of FS_XXX_FL defined in
447  * include/linux/fs.h and the on-disk values found in ext2, ext3, and
448  * ext4 filesystems, and of course the values defined in e2fsprogs.
449  *
450  * It's not paranoia if the Murphy's Law really *is* out to get you.  :-)
451  */
452 static inline void ext4_check_flag_values(void)
453 {
454 	CHECK_FLAG_VALUE(SECRM);
455 	CHECK_FLAG_VALUE(UNRM);
456 	CHECK_FLAG_VALUE(COMPR);
457 	CHECK_FLAG_VALUE(SYNC);
458 	CHECK_FLAG_VALUE(IMMUTABLE);
459 	CHECK_FLAG_VALUE(APPEND);
460 	CHECK_FLAG_VALUE(NODUMP);
461 	CHECK_FLAG_VALUE(NOATIME);
462 	CHECK_FLAG_VALUE(DIRTY);
463 	CHECK_FLAG_VALUE(COMPRBLK);
464 	CHECK_FLAG_VALUE(NOCOMPR);
465 	CHECK_FLAG_VALUE(ECOMPR);
466 	CHECK_FLAG_VALUE(INDEX);
467 	CHECK_FLAG_VALUE(IMAGIC);
468 	CHECK_FLAG_VALUE(JOURNAL_DATA);
469 	CHECK_FLAG_VALUE(NOTAIL);
470 	CHECK_FLAG_VALUE(DIRSYNC);
471 	CHECK_FLAG_VALUE(TOPDIR);
472 	CHECK_FLAG_VALUE(HUGE_FILE);
473 	CHECK_FLAG_VALUE(EXTENTS);
474 	CHECK_FLAG_VALUE(EA_INODE);
475 	CHECK_FLAG_VALUE(EOFBLOCKS);
476 	CHECK_FLAG_VALUE(RESERVED);
477 }
478 
479 /* Used to pass group descriptor data when online resize is done */
480 struct ext4_new_group_input {
481 	__u32 group;		/* Group number for this data */
482 	__u64 block_bitmap;	/* Absolute block number of block bitmap */
483 	__u64 inode_bitmap;	/* Absolute block number of inode bitmap */
484 	__u64 inode_table;	/* Absolute block number of inode table start */
485 	__u32 blocks_count;	/* Total number of blocks in this group */
486 	__u16 reserved_blocks;	/* Number of reserved blocks in this group */
487 	__u16 unused;
488 };
489 
490 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
491 struct compat_ext4_new_group_input {
492 	u32 group;
493 	compat_u64 block_bitmap;
494 	compat_u64 inode_bitmap;
495 	compat_u64 inode_table;
496 	u32 blocks_count;
497 	u16 reserved_blocks;
498 	u16 unused;
499 };
500 #endif
501 
502 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
503 struct ext4_new_group_data {
504 	__u32 group;
505 	__u64 block_bitmap;
506 	__u64 inode_bitmap;
507 	__u64 inode_table;
508 	__u32 blocks_count;
509 	__u16 reserved_blocks;
510 	__u16 unused;
511 	__u32 free_blocks_count;
512 };
513 
514 /*
515  * Flags used by ext4_map_blocks()
516  */
517 	/* Allocate any needed blocks and/or convert an unitialized
518 	   extent to be an initialized ext4 */
519 #define EXT4_GET_BLOCKS_CREATE			0x0001
520 	/* Request the creation of an unitialized extent */
521 #define EXT4_GET_BLOCKS_UNINIT_EXT		0x0002
522 #define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT	(EXT4_GET_BLOCKS_UNINIT_EXT|\
523 						 EXT4_GET_BLOCKS_CREATE)
524 	/* Caller is from the delayed allocation writeout path,
525 	   so set the magic i_delalloc_reserve_flag after taking the
526 	   inode allocation semaphore for */
527 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE	0x0004
528 	/* caller is from the direct IO path, request to creation of an
529 	unitialized extents if not allocated, split the uninitialized
530 	extent if blocks has been preallocated already*/
531 #define EXT4_GET_BLOCKS_PRE_IO			0x0008
532 #define EXT4_GET_BLOCKS_CONVERT			0x0010
533 #define EXT4_GET_BLOCKS_IO_CREATE_EXT		(EXT4_GET_BLOCKS_PRE_IO|\
534 					 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
535 	/* Convert extent to initialized after IO complete */
536 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT		(EXT4_GET_BLOCKS_CONVERT|\
537 					 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
538 	/* Punch out blocks of an extent */
539 #define EXT4_GET_BLOCKS_PUNCH_OUT_EXT		0x0020
540 	/* Don't normalize allocation size (used for fallocate) */
541 #define EXT4_GET_BLOCKS_NO_NORMALIZE		0x0040
542 	/* Request will not result in inode size update (user for fallocate) */
543 #define EXT4_GET_BLOCKS_KEEP_SIZE		0x0080
544 
545 /*
546  * Flags used by ext4_free_blocks
547  */
548 #define EXT4_FREE_BLOCKS_METADATA	0x0001
549 #define EXT4_FREE_BLOCKS_FORGET		0x0002
550 #define EXT4_FREE_BLOCKS_VALIDATED	0x0004
551 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE	0x0008
552 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER	0x0010
553 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER	0x0020
554 
555 /*
556  * Flags used by ext4_discard_partial_page_buffers
557  */
558 #define EXT4_DISCARD_PARTIAL_PG_ZERO_UNMAPPED	0x0001
559 
560 /*
561  * ioctl commands
562  */
563 #define	EXT4_IOC_GETFLAGS		FS_IOC_GETFLAGS
564 #define	EXT4_IOC_SETFLAGS		FS_IOC_SETFLAGS
565 #define	EXT4_IOC_GETVERSION		_IOR('f', 3, long)
566 #define	EXT4_IOC_SETVERSION		_IOW('f', 4, long)
567 #define	EXT4_IOC_GETVERSION_OLD		FS_IOC_GETVERSION
568 #define	EXT4_IOC_SETVERSION_OLD		FS_IOC_SETVERSION
569 #define EXT4_IOC_GETRSVSZ		_IOR('f', 5, long)
570 #define EXT4_IOC_SETRSVSZ		_IOW('f', 6, long)
571 #define EXT4_IOC_GROUP_EXTEND		_IOW('f', 7, unsigned long)
572 #define EXT4_IOC_GROUP_ADD		_IOW('f', 8, struct ext4_new_group_input)
573 #define EXT4_IOC_MIGRATE		_IO('f', 9)
574  /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
575  /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
576 #define EXT4_IOC_ALLOC_DA_BLKS		_IO('f', 12)
577 #define EXT4_IOC_MOVE_EXT		_IOWR('f', 15, struct move_extent)
578 
579 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
580 /*
581  * ioctl commands in 32 bit emulation
582  */
583 #define EXT4_IOC32_GETFLAGS		FS_IOC32_GETFLAGS
584 #define EXT4_IOC32_SETFLAGS		FS_IOC32_SETFLAGS
585 #define EXT4_IOC32_GETVERSION		_IOR('f', 3, int)
586 #define EXT4_IOC32_SETVERSION		_IOW('f', 4, int)
587 #define EXT4_IOC32_GETRSVSZ		_IOR('f', 5, int)
588 #define EXT4_IOC32_SETRSVSZ		_IOW('f', 6, int)
589 #define EXT4_IOC32_GROUP_EXTEND		_IOW('f', 7, unsigned int)
590 #define EXT4_IOC32_GROUP_ADD		_IOW('f', 8, struct compat_ext4_new_group_input)
591 #define EXT4_IOC32_GETVERSION_OLD	FS_IOC32_GETVERSION
592 #define EXT4_IOC32_SETVERSION_OLD	FS_IOC32_SETVERSION
593 #endif
594 
595 /* Max physical block we can address w/o extents */
596 #define EXT4_MAX_BLOCK_FILE_PHYS	0xFFFFFFFF
597 
598 /*
599  * Structure of an inode on the disk
600  */
601 struct ext4_inode {
602 	__le16	i_mode;		/* File mode */
603 	__le16	i_uid;		/* Low 16 bits of Owner Uid */
604 	__le32	i_size_lo;	/* Size in bytes */
605 	__le32	i_atime;	/* Access time */
606 	__le32	i_ctime;	/* Inode Change time */
607 	__le32	i_mtime;	/* Modification time */
608 	__le32	i_dtime;	/* Deletion Time */
609 	__le16	i_gid;		/* Low 16 bits of Group Id */
610 	__le16	i_links_count;	/* Links count */
611 	__le32	i_blocks_lo;	/* Blocks count */
612 	__le32	i_flags;	/* File flags */
613 	union {
614 		struct {
615 			__le32  l_i_version;
616 		} linux1;
617 		struct {
618 			__u32  h_i_translator;
619 		} hurd1;
620 		struct {
621 			__u32  m_i_reserved1;
622 		} masix1;
623 	} osd1;				/* OS dependent 1 */
624 	__le32	i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
625 	__le32	i_generation;	/* File version (for NFS) */
626 	__le32	i_file_acl_lo;	/* File ACL */
627 	__le32	i_size_high;
628 	__le32	i_obso_faddr;	/* Obsoleted fragment address */
629 	union {
630 		struct {
631 			__le16	l_i_blocks_high; /* were l_i_reserved1 */
632 			__le16	l_i_file_acl_high;
633 			__le16	l_i_uid_high;	/* these 2 fields */
634 			__le16	l_i_gid_high;	/* were reserved2[0] */
635 			__u32	l_i_reserved2;
636 		} linux2;
637 		struct {
638 			__le16	h_i_reserved1;	/* Obsoleted fragment number/size which are removed in ext4 */
639 			__u16	h_i_mode_high;
640 			__u16	h_i_uid_high;
641 			__u16	h_i_gid_high;
642 			__u32	h_i_author;
643 		} hurd2;
644 		struct {
645 			__le16	h_i_reserved1;	/* Obsoleted fragment number/size which are removed in ext4 */
646 			__le16	m_i_file_acl_high;
647 			__u32	m_i_reserved2[2];
648 		} masix2;
649 	} osd2;				/* OS dependent 2 */
650 	__le16	i_extra_isize;
651 	__le16	i_pad1;
652 	__le32  i_ctime_extra;  /* extra Change time      (nsec << 2 | epoch) */
653 	__le32  i_mtime_extra;  /* extra Modification time(nsec << 2 | epoch) */
654 	__le32  i_atime_extra;  /* extra Access time      (nsec << 2 | epoch) */
655 	__le32  i_crtime;       /* File Creation time */
656 	__le32  i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
657 	__le32  i_version_hi;	/* high 32 bits for 64-bit version */
658 };
659 
660 struct move_extent {
661 	__u32 reserved;		/* should be zero */
662 	__u32 donor_fd;		/* donor file descriptor */
663 	__u64 orig_start;	/* logical start offset in block for orig */
664 	__u64 donor_start;	/* logical start offset in block for donor */
665 	__u64 len;		/* block length to be moved */
666 	__u64 moved_len;	/* moved block length */
667 };
668 
669 #define EXT4_EPOCH_BITS 2
670 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
671 #define EXT4_NSEC_MASK  (~0UL << EXT4_EPOCH_BITS)
672 
673 /*
674  * Extended fields will fit into an inode if the filesystem was formatted
675  * with large inodes (-I 256 or larger) and there are not currently any EAs
676  * consuming all of the available space. For new inodes we always reserve
677  * enough space for the kernel's known extended fields, but for inodes
678  * created with an old kernel this might not have been the case. None of
679  * the extended inode fields is critical for correct filesystem operation.
680  * This macro checks if a certain field fits in the inode. Note that
681  * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
682  */
683 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field)	\
684 	((offsetof(typeof(*ext4_inode), field) +	\
685 	  sizeof((ext4_inode)->field))			\
686 	<= (EXT4_GOOD_OLD_INODE_SIZE +			\
687 	    (einode)->i_extra_isize))			\
688 
689 static inline __le32 ext4_encode_extra_time(struct timespec *time)
690 {
691        return cpu_to_le32((sizeof(time->tv_sec) > 4 ?
692 			   (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) |
693                           ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK));
694 }
695 
696 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
697 {
698        if (sizeof(time->tv_sec) > 4)
699 	       time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK)
700 			       << 32;
701        time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
702 }
703 
704 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode)			       \
705 do {									       \
706 	(raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec);	       \
707 	if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     \
708 		(raw_inode)->xtime ## _extra =				       \
709 				ext4_encode_extra_time(&(inode)->xtime);       \
710 } while (0)
711 
712 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode)			       \
713 do {									       \
714 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))		       \
715 		(raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec);      \
716 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))	       \
717 		(raw_inode)->xtime ## _extra =				       \
718 				ext4_encode_extra_time(&(einode)->xtime);      \
719 } while (0)
720 
721 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode)			       \
722 do {									       \
723 	(inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime);       \
724 	if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     \
725 		ext4_decode_extra_time(&(inode)->xtime,			       \
726 				       raw_inode->xtime ## _extra);	       \
727 	else								       \
728 		(inode)->xtime.tv_nsec = 0;				       \
729 } while (0)
730 
731 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode)			       \
732 do {									       \
733 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))		       \
734 		(einode)->xtime.tv_sec = 				       \
735 			(signed)le32_to_cpu((raw_inode)->xtime);	       \
736 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))	       \
737 		ext4_decode_extra_time(&(einode)->xtime,		       \
738 				       raw_inode->xtime ## _extra);	       \
739 	else								       \
740 		(einode)->xtime.tv_nsec = 0;				       \
741 } while (0)
742 
743 #define i_disk_version osd1.linux1.l_i_version
744 
745 #if defined(__KERNEL__) || defined(__linux__)
746 #define i_reserved1	osd1.linux1.l_i_reserved1
747 #define i_file_acl_high	osd2.linux2.l_i_file_acl_high
748 #define i_blocks_high	osd2.linux2.l_i_blocks_high
749 #define i_uid_low	i_uid
750 #define i_gid_low	i_gid
751 #define i_uid_high	osd2.linux2.l_i_uid_high
752 #define i_gid_high	osd2.linux2.l_i_gid_high
753 #define i_reserved2	osd2.linux2.l_i_reserved2
754 
755 #elif defined(__GNU__)
756 
757 #define i_translator	osd1.hurd1.h_i_translator
758 #define i_uid_high	osd2.hurd2.h_i_uid_high
759 #define i_gid_high	osd2.hurd2.h_i_gid_high
760 #define i_author	osd2.hurd2.h_i_author
761 
762 #elif defined(__masix__)
763 
764 #define i_reserved1	osd1.masix1.m_i_reserved1
765 #define i_file_acl_high	osd2.masix2.m_i_file_acl_high
766 #define i_reserved2	osd2.masix2.m_i_reserved2
767 
768 #endif /* defined(__KERNEL__) || defined(__linux__) */
769 
770 /*
771  * storage for cached extent
772  * If ec_len == 0, then the cache is invalid.
773  * If ec_start == 0, then the cache represents a gap (null mapping)
774  */
775 struct ext4_ext_cache {
776 	ext4_fsblk_t	ec_start;
777 	ext4_lblk_t	ec_block;
778 	__u32		ec_len; /* must be 32bit to return holes */
779 };
780 
781 /*
782  * fourth extended file system inode data in memory
783  */
784 struct ext4_inode_info {
785 	__le32	i_data[15];	/* unconverted */
786 	__u32	i_dtime;
787 	ext4_fsblk_t	i_file_acl;
788 
789 	/*
790 	 * i_block_group is the number of the block group which contains
791 	 * this file's inode.  Constant across the lifetime of the inode,
792 	 * it is ued for making block allocation decisions - we try to
793 	 * place a file's data blocks near its inode block, and new inodes
794 	 * near to their parent directory's inode.
795 	 */
796 	ext4_group_t	i_block_group;
797 	ext4_lblk_t	i_dir_start_lookup;
798 #if (BITS_PER_LONG < 64)
799 	unsigned long	i_state_flags;		/* Dynamic state flags */
800 #endif
801 	unsigned long	i_flags;
802 
803 #ifdef CONFIG_EXT4_FS_XATTR
804 	/*
805 	 * Extended attributes can be read independently of the main file
806 	 * data. Taking i_mutex even when reading would cause contention
807 	 * between readers of EAs and writers of regular file data, so
808 	 * instead we synchronize on xattr_sem when reading or changing
809 	 * EAs.
810 	 */
811 	struct rw_semaphore xattr_sem;
812 #endif
813 
814 	struct list_head i_orphan;	/* unlinked but open inodes */
815 
816 	/*
817 	 * i_disksize keeps track of what the inode size is ON DISK, not
818 	 * in memory.  During truncate, i_size is set to the new size by
819 	 * the VFS prior to calling ext4_truncate(), but the filesystem won't
820 	 * set i_disksize to 0 until the truncate is actually under way.
821 	 *
822 	 * The intent is that i_disksize always represents the blocks which
823 	 * are used by this file.  This allows recovery to restart truncate
824 	 * on orphans if we crash during truncate.  We actually write i_disksize
825 	 * into the on-disk inode when writing inodes out, instead of i_size.
826 	 *
827 	 * The only time when i_disksize and i_size may be different is when
828 	 * a truncate is in progress.  The only things which change i_disksize
829 	 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
830 	 */
831 	loff_t	i_disksize;
832 
833 	/*
834 	 * i_data_sem is for serialising ext4_truncate() against
835 	 * ext4_getblock().  In the 2.4 ext2 design, great chunks of inode's
836 	 * data tree are chopped off during truncate. We can't do that in
837 	 * ext4 because whenever we perform intermediate commits during
838 	 * truncate, the inode and all the metadata blocks *must* be in a
839 	 * consistent state which allows truncation of the orphans to restart
840 	 * during recovery.  Hence we must fix the get_block-vs-truncate race
841 	 * by other means, so we have i_data_sem.
842 	 */
843 	struct rw_semaphore i_data_sem;
844 	struct inode vfs_inode;
845 	struct jbd2_inode *jinode;
846 
847 	struct ext4_ext_cache i_cached_extent;
848 	/*
849 	 * File creation time. Its function is same as that of
850 	 * struct timespec i_{a,c,m}time in the generic inode.
851 	 */
852 	struct timespec i_crtime;
853 
854 	/* mballoc */
855 	struct list_head i_prealloc_list;
856 	spinlock_t i_prealloc_lock;
857 
858 	/* ialloc */
859 	ext4_group_t	i_last_alloc_group;
860 
861 	/* allocation reservation info for delalloc */
862 	/* In case of bigalloc, these refer to clusters rather than blocks */
863 	unsigned int i_reserved_data_blocks;
864 	unsigned int i_reserved_meta_blocks;
865 	unsigned int i_allocated_meta_blocks;
866 	ext4_lblk_t i_da_metadata_calc_last_lblock;
867 	int i_da_metadata_calc_len;
868 
869 	/* on-disk additional length */
870 	__u16 i_extra_isize;
871 
872 #ifdef CONFIG_QUOTA
873 	/* quota space reservation, managed internally by quota code */
874 	qsize_t i_reserved_quota;
875 #endif
876 
877 	/* completed IOs that might need unwritten extents handling */
878 	struct list_head i_completed_io_list;
879 	spinlock_t i_completed_io_lock;
880 	atomic_t i_ioend_count;	/* Number of outstanding io_end structs */
881 	/* current io_end structure for async DIO write*/
882 	ext4_io_end_t *cur_aio_dio;
883 	atomic_t i_aiodio_unwritten; /* Nr. of inflight conversions pending */
884 
885 	spinlock_t i_block_reservation_lock;
886 
887 	/*
888 	 * Transactions that contain inode's metadata needed to complete
889 	 * fsync and fdatasync, respectively.
890 	 */
891 	tid_t i_sync_tid;
892 	tid_t i_datasync_tid;
893 };
894 
895 /*
896  * File system states
897  */
898 #define	EXT4_VALID_FS			0x0001	/* Unmounted cleanly */
899 #define	EXT4_ERROR_FS			0x0002	/* Errors detected */
900 #define	EXT4_ORPHAN_FS			0x0004	/* Orphans being recovered */
901 
902 /*
903  * Misc. filesystem flags
904  */
905 #define EXT2_FLAGS_SIGNED_HASH		0x0001  /* Signed dirhash in use */
906 #define EXT2_FLAGS_UNSIGNED_HASH	0x0002  /* Unsigned dirhash in use */
907 #define EXT2_FLAGS_TEST_FILESYS		0x0004	/* to test development code */
908 
909 /*
910  * Mount flags
911  */
912 #define EXT4_MOUNT_GRPID		0x00004	/* Create files with directory's group */
913 #define EXT4_MOUNT_DEBUG		0x00008	/* Some debugging messages */
914 #define EXT4_MOUNT_ERRORS_CONT		0x00010	/* Continue on errors */
915 #define EXT4_MOUNT_ERRORS_RO		0x00020	/* Remount fs ro on errors */
916 #define EXT4_MOUNT_ERRORS_PANIC		0x00040	/* Panic on errors */
917 #define EXT4_MOUNT_MINIX_DF		0x00080	/* Mimics the Minix statfs */
918 #define EXT4_MOUNT_NOLOAD		0x00100	/* Don't use existing journal*/
919 #define EXT4_MOUNT_DATA_FLAGS		0x00C00	/* Mode for data writes: */
920 #define EXT4_MOUNT_JOURNAL_DATA		0x00400	/* Write data to journal */
921 #define EXT4_MOUNT_ORDERED_DATA		0x00800	/* Flush data before commit */
922 #define EXT4_MOUNT_WRITEBACK_DATA	0x00C00	/* No data ordering */
923 #define EXT4_MOUNT_UPDATE_JOURNAL	0x01000	/* Update the journal format */
924 #define EXT4_MOUNT_NO_UID32		0x02000  /* Disable 32-bit UIDs */
925 #define EXT4_MOUNT_XATTR_USER		0x04000	/* Extended user attributes */
926 #define EXT4_MOUNT_POSIX_ACL		0x08000	/* POSIX Access Control Lists */
927 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC	0x10000	/* No auto delalloc mapping */
928 #define EXT4_MOUNT_BARRIER		0x20000 /* Use block barriers */
929 #define EXT4_MOUNT_QUOTA		0x80000 /* Some quota option set */
930 #define EXT4_MOUNT_USRQUOTA		0x100000 /* "old" user quota */
931 #define EXT4_MOUNT_GRPQUOTA		0x200000 /* "old" group quota */
932 #define EXT4_MOUNT_DIOREAD_NOLOCK	0x400000 /* Enable support for dio read nolocking */
933 #define EXT4_MOUNT_JOURNAL_CHECKSUM	0x800000 /* Journal checksums */
934 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT	0x1000000 /* Journal Async Commit */
935 #define EXT4_MOUNT_I_VERSION            0x2000000 /* i_version support */
936 #define EXT4_MOUNT_MBLK_IO_SUBMIT	0x4000000 /* multi-block io submits */
937 #define EXT4_MOUNT_DELALLOC		0x8000000 /* Delalloc support */
938 #define EXT4_MOUNT_DATA_ERR_ABORT	0x10000000 /* Abort on file data write */
939 #define EXT4_MOUNT_BLOCK_VALIDITY	0x20000000 /* Block validity checking */
940 #define EXT4_MOUNT_DISCARD		0x40000000 /* Issue DISCARD requests */
941 #define EXT4_MOUNT_INIT_INODE_TABLE	0x80000000 /* Initialize uninitialized itables */
942 
943 #define EXT4_MOUNT2_EXPLICIT_DELALLOC	0x00000001 /* User explicitly
944 						      specified delalloc */
945 
946 #define clear_opt(sb, opt)		EXT4_SB(sb)->s_mount_opt &= \
947 						~EXT4_MOUNT_##opt
948 #define set_opt(sb, opt)		EXT4_SB(sb)->s_mount_opt |= \
949 						EXT4_MOUNT_##opt
950 #define test_opt(sb, opt)		(EXT4_SB(sb)->s_mount_opt & \
951 					 EXT4_MOUNT_##opt)
952 
953 #define clear_opt2(sb, opt)		EXT4_SB(sb)->s_mount_opt2 &= \
954 						~EXT4_MOUNT2_##opt
955 #define set_opt2(sb, opt)		EXT4_SB(sb)->s_mount_opt2 |= \
956 						EXT4_MOUNT2_##opt
957 #define test_opt2(sb, opt)		(EXT4_SB(sb)->s_mount_opt2 & \
958 					 EXT4_MOUNT2_##opt)
959 
960 #define ext4_set_bit			__test_and_set_bit_le
961 #define ext4_set_bit_atomic		ext2_set_bit_atomic
962 #define ext4_clear_bit			__test_and_clear_bit_le
963 #define ext4_clear_bit_atomic		ext2_clear_bit_atomic
964 #define ext4_test_bit			test_bit_le
965 #define ext4_find_first_zero_bit	find_first_zero_bit_le
966 #define ext4_find_next_zero_bit		find_next_zero_bit_le
967 #define ext4_find_next_bit		find_next_bit_le
968 
969 extern void ext4_set_bits(void *bm, int cur, int len);
970 
971 /*
972  * Maximal mount counts between two filesystem checks
973  */
974 #define EXT4_DFL_MAX_MNT_COUNT		20	/* Allow 20 mounts */
975 #define EXT4_DFL_CHECKINTERVAL		0	/* Don't use interval check */
976 
977 /*
978  * Behaviour when detecting errors
979  */
980 #define EXT4_ERRORS_CONTINUE		1	/* Continue execution */
981 #define EXT4_ERRORS_RO			2	/* Remount fs read-only */
982 #define EXT4_ERRORS_PANIC		3	/* Panic */
983 #define EXT4_ERRORS_DEFAULT		EXT4_ERRORS_CONTINUE
984 
985 /*
986  * Structure of the super block
987  */
988 struct ext4_super_block {
989 /*00*/	__le32	s_inodes_count;		/* Inodes count */
990 	__le32	s_blocks_count_lo;	/* Blocks count */
991 	__le32	s_r_blocks_count_lo;	/* Reserved blocks count */
992 	__le32	s_free_blocks_count_lo;	/* Free blocks count */
993 /*10*/	__le32	s_free_inodes_count;	/* Free inodes count */
994 	__le32	s_first_data_block;	/* First Data Block */
995 	__le32	s_log_block_size;	/* Block size */
996 	__le32	s_log_cluster_size;	/* Allocation cluster size */
997 /*20*/	__le32	s_blocks_per_group;	/* # Blocks per group */
998 	__le32	s_clusters_per_group;	/* # Clusters per group */
999 	__le32	s_inodes_per_group;	/* # Inodes per group */
1000 	__le32	s_mtime;		/* Mount time */
1001 /*30*/	__le32	s_wtime;		/* Write time */
1002 	__le16	s_mnt_count;		/* Mount count */
1003 	__le16	s_max_mnt_count;	/* Maximal mount count */
1004 	__le16	s_magic;		/* Magic signature */
1005 	__le16	s_state;		/* File system state */
1006 	__le16	s_errors;		/* Behaviour when detecting errors */
1007 	__le16	s_minor_rev_level;	/* minor revision level */
1008 /*40*/	__le32	s_lastcheck;		/* time of last check */
1009 	__le32	s_checkinterval;	/* max. time between checks */
1010 	__le32	s_creator_os;		/* OS */
1011 	__le32	s_rev_level;		/* Revision level */
1012 /*50*/	__le16	s_def_resuid;		/* Default uid for reserved blocks */
1013 	__le16	s_def_resgid;		/* Default gid for reserved blocks */
1014 	/*
1015 	 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1016 	 *
1017 	 * Note: the difference between the compatible feature set and
1018 	 * the incompatible feature set is that if there is a bit set
1019 	 * in the incompatible feature set that the kernel doesn't
1020 	 * know about, it should refuse to mount the filesystem.
1021 	 *
1022 	 * e2fsck's requirements are more strict; if it doesn't know
1023 	 * about a feature in either the compatible or incompatible
1024 	 * feature set, it must abort and not try to meddle with
1025 	 * things it doesn't understand...
1026 	 */
1027 	__le32	s_first_ino;		/* First non-reserved inode */
1028 	__le16  s_inode_size;		/* size of inode structure */
1029 	__le16	s_block_group_nr;	/* block group # of this superblock */
1030 	__le32	s_feature_compat;	/* compatible feature set */
1031 /*60*/	__le32	s_feature_incompat;	/* incompatible feature set */
1032 	__le32	s_feature_ro_compat;	/* readonly-compatible feature set */
1033 /*68*/	__u8	s_uuid[16];		/* 128-bit uuid for volume */
1034 /*78*/	char	s_volume_name[16];	/* volume name */
1035 /*88*/	char	s_last_mounted[64];	/* directory where last mounted */
1036 /*C8*/	__le32	s_algorithm_usage_bitmap; /* For compression */
1037 	/*
1038 	 * Performance hints.  Directory preallocation should only
1039 	 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1040 	 */
1041 	__u8	s_prealloc_blocks;	/* Nr of blocks to try to preallocate*/
1042 	__u8	s_prealloc_dir_blocks;	/* Nr to preallocate for dirs */
1043 	__le16	s_reserved_gdt_blocks;	/* Per group desc for online growth */
1044 	/*
1045 	 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1046 	 */
1047 /*D0*/	__u8	s_journal_uuid[16];	/* uuid of journal superblock */
1048 /*E0*/	__le32	s_journal_inum;		/* inode number of journal file */
1049 	__le32	s_journal_dev;		/* device number of journal file */
1050 	__le32	s_last_orphan;		/* start of list of inodes to delete */
1051 	__le32	s_hash_seed[4];		/* HTREE hash seed */
1052 	__u8	s_def_hash_version;	/* Default hash version to use */
1053 	__u8	s_jnl_backup_type;
1054 	__le16  s_desc_size;		/* size of group descriptor */
1055 /*100*/	__le32	s_default_mount_opts;
1056 	__le32	s_first_meta_bg;	/* First metablock block group */
1057 	__le32	s_mkfs_time;		/* When the filesystem was created */
1058 	__le32	s_jnl_blocks[17];	/* Backup of the journal inode */
1059 	/* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1060 /*150*/	__le32	s_blocks_count_hi;	/* Blocks count */
1061 	__le32	s_r_blocks_count_hi;	/* Reserved blocks count */
1062 	__le32	s_free_blocks_count_hi;	/* Free blocks count */
1063 	__le16	s_min_extra_isize;	/* All inodes have at least # bytes */
1064 	__le16	s_want_extra_isize; 	/* New inodes should reserve # bytes */
1065 	__le32	s_flags;		/* Miscellaneous flags */
1066 	__le16  s_raid_stride;		/* RAID stride */
1067 	__le16  s_mmp_update_interval;  /* # seconds to wait in MMP checking */
1068 	__le64  s_mmp_block;            /* Block for multi-mount protection */
1069 	__le32  s_raid_stripe_width;    /* blocks on all data disks (N*stride)*/
1070 	__u8	s_log_groups_per_flex;  /* FLEX_BG group size */
1071 	__u8	s_reserved_char_pad;
1072 	__le16  s_reserved_pad;
1073 	__le64	s_kbytes_written;	/* nr of lifetime kilobytes written */
1074 	__le32	s_snapshot_inum;	/* Inode number of active snapshot */
1075 	__le32	s_snapshot_id;		/* sequential ID of active snapshot */
1076 	__le64	s_snapshot_r_blocks_count; /* reserved blocks for active
1077 					      snapshot's future use */
1078 	__le32	s_snapshot_list;	/* inode number of the head of the
1079 					   on-disk snapshot list */
1080 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1081 	__le32	s_error_count;		/* number of fs errors */
1082 	__le32	s_first_error_time;	/* first time an error happened */
1083 	__le32	s_first_error_ino;	/* inode involved in first error */
1084 	__le64	s_first_error_block;	/* block involved of first error */
1085 	__u8	s_first_error_func[32];	/* function where the error happened */
1086 	__le32	s_first_error_line;	/* line number where error happened */
1087 	__le32	s_last_error_time;	/* most recent time of an error */
1088 	__le32	s_last_error_ino;	/* inode involved in last error */
1089 	__le32	s_last_error_line;	/* line number where error happened */
1090 	__le64	s_last_error_block;	/* block involved of last error */
1091 	__u8	s_last_error_func[32];	/* function where the error happened */
1092 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1093 	__u8	s_mount_opts[64];
1094 	__le32	s_usr_quota_inum;	/* inode for tracking user quota */
1095 	__le32	s_grp_quota_inum;	/* inode for tracking group quota */
1096 	__le32	s_overhead_clusters;	/* overhead blocks/clusters in fs */
1097 	__le32  s_reserved[109];        /* Padding to the end of the block */
1098 };
1099 
1100 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1101 
1102 #ifdef __KERNEL__
1103 
1104 /*
1105  * run-time mount flags
1106  */
1107 #define EXT4_MF_MNTDIR_SAMPLED	0x0001
1108 #define EXT4_MF_FS_ABORTED	0x0002	/* Fatal error detected */
1109 
1110 /*
1111  * fourth extended-fs super-block data in memory
1112  */
1113 struct ext4_sb_info {
1114 	unsigned long s_desc_size;	/* Size of a group descriptor in bytes */
1115 	unsigned long s_inodes_per_block;/* Number of inodes per block */
1116 	unsigned long s_blocks_per_group;/* Number of blocks in a group */
1117 	unsigned long s_clusters_per_group; /* Number of clusters in a group */
1118 	unsigned long s_inodes_per_group;/* Number of inodes in a group */
1119 	unsigned long s_itb_per_group;	/* Number of inode table blocks per group */
1120 	unsigned long s_gdb_count;	/* Number of group descriptor blocks */
1121 	unsigned long s_desc_per_block;	/* Number of group descriptors per block */
1122 	ext4_group_t s_groups_count;	/* Number of groups in the fs */
1123 	ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1124 	unsigned long s_overhead_last;  /* Last calculated overhead */
1125 	unsigned long s_blocks_last;    /* Last seen block count */
1126 	unsigned int s_cluster_ratio;	/* Number of blocks per cluster */
1127 	unsigned int s_cluster_bits;	/* log2 of s_cluster_ratio */
1128 	loff_t s_bitmap_maxbytes;	/* max bytes for bitmap files */
1129 	struct buffer_head * s_sbh;	/* Buffer containing the super block */
1130 	struct ext4_super_block *s_es;	/* Pointer to the super block in the buffer */
1131 	struct buffer_head **s_group_desc;
1132 	unsigned int s_mount_opt;
1133 	unsigned int s_mount_opt2;
1134 	unsigned int s_mount_flags;
1135 	ext4_fsblk_t s_sb_block;
1136 	uid_t s_resuid;
1137 	gid_t s_resgid;
1138 	unsigned short s_mount_state;
1139 	unsigned short s_pad;
1140 	int s_addr_per_block_bits;
1141 	int s_desc_per_block_bits;
1142 	int s_inode_size;
1143 	int s_first_ino;
1144 	unsigned int s_inode_readahead_blks;
1145 	unsigned int s_inode_goal;
1146 	spinlock_t s_next_gen_lock;
1147 	u32 s_next_generation;
1148 	u32 s_hash_seed[4];
1149 	int s_def_hash_version;
1150 	int s_hash_unsigned;	/* 3 if hash should be signed, 0 if not */
1151 	struct percpu_counter s_freeclusters_counter;
1152 	struct percpu_counter s_freeinodes_counter;
1153 	struct percpu_counter s_dirs_counter;
1154 	struct percpu_counter s_dirtyclusters_counter;
1155 	struct blockgroup_lock *s_blockgroup_lock;
1156 	struct proc_dir_entry *s_proc;
1157 	struct kobject s_kobj;
1158 	struct completion s_kobj_unregister;
1159 
1160 	/* Journaling */
1161 	struct journal_s *s_journal;
1162 	struct list_head s_orphan;
1163 	struct mutex s_orphan_lock;
1164 	unsigned long s_resize_flags;		/* Flags indicating if there
1165 						   is a resizer */
1166 	unsigned long s_commit_interval;
1167 	u32 s_max_batch_time;
1168 	u32 s_min_batch_time;
1169 	struct block_device *journal_bdev;
1170 #ifdef CONFIG_QUOTA
1171 	char *s_qf_names[MAXQUOTAS];		/* Names of quota files with journalled quota */
1172 	int s_jquota_fmt;			/* Format of quota to use */
1173 #endif
1174 	unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1175 	struct rb_root system_blks;
1176 
1177 #ifdef EXTENTS_STATS
1178 	/* ext4 extents stats */
1179 	unsigned long s_ext_min;
1180 	unsigned long s_ext_max;
1181 	unsigned long s_depth_max;
1182 	spinlock_t s_ext_stats_lock;
1183 	unsigned long s_ext_blocks;
1184 	unsigned long s_ext_extents;
1185 #endif
1186 	/* ext4 extent cache stats */
1187 	unsigned long extent_cache_hits;
1188 	unsigned long extent_cache_misses;
1189 
1190 	/* for buddy allocator */
1191 	struct ext4_group_info ***s_group_info;
1192 	struct inode *s_buddy_cache;
1193 	spinlock_t s_md_lock;
1194 	unsigned short *s_mb_offsets;
1195 	unsigned int *s_mb_maxs;
1196 
1197 	/* tunables */
1198 	unsigned long s_stripe;
1199 	unsigned int s_mb_stream_request;
1200 	unsigned int s_mb_max_to_scan;
1201 	unsigned int s_mb_min_to_scan;
1202 	unsigned int s_mb_stats;
1203 	unsigned int s_mb_order2_reqs;
1204 	unsigned int s_mb_group_prealloc;
1205 	unsigned int s_max_writeback_mb_bump;
1206 	/* where last allocation was done - for stream allocation */
1207 	unsigned long s_mb_last_group;
1208 	unsigned long s_mb_last_start;
1209 
1210 	/* stats for buddy allocator */
1211 	atomic_t s_bal_reqs;	/* number of reqs with len > 1 */
1212 	atomic_t s_bal_success;	/* we found long enough chunks */
1213 	atomic_t s_bal_allocated;	/* in blocks */
1214 	atomic_t s_bal_ex_scanned;	/* total extents scanned */
1215 	atomic_t s_bal_goals;	/* goal hits */
1216 	atomic_t s_bal_breaks;	/* too long searches */
1217 	atomic_t s_bal_2orders;	/* 2^order hits */
1218 	spinlock_t s_bal_lock;
1219 	unsigned long s_mb_buddies_generated;
1220 	unsigned long long s_mb_generation_time;
1221 	atomic_t s_mb_lost_chunks;
1222 	atomic_t s_mb_preallocated;
1223 	atomic_t s_mb_discarded;
1224 	atomic_t s_lock_busy;
1225 
1226 	/* locality groups */
1227 	struct ext4_locality_group __percpu *s_locality_groups;
1228 
1229 	/* for write statistics */
1230 	unsigned long s_sectors_written_start;
1231 	u64 s_kbytes_written;
1232 
1233 	unsigned int s_log_groups_per_flex;
1234 	struct flex_groups *s_flex_groups;
1235 
1236 	/* workqueue for dio unwritten */
1237 	struct workqueue_struct *dio_unwritten_wq;
1238 
1239 	/* timer for periodic error stats printing */
1240 	struct timer_list s_err_report;
1241 
1242 	/* Lazy inode table initialization info */
1243 	struct ext4_li_request *s_li_request;
1244 	/* Wait multiplier for lazy initialization thread */
1245 	unsigned int s_li_wait_mult;
1246 
1247 	/* Kernel thread for multiple mount protection */
1248 	struct task_struct *s_mmp_tsk;
1249 
1250 	/* record the last minlen when FITRIM is called. */
1251 	atomic_t s_last_trim_minblks;
1252 };
1253 
1254 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1255 {
1256 	return sb->s_fs_info;
1257 }
1258 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1259 {
1260 	return container_of(inode, struct ext4_inode_info, vfs_inode);
1261 }
1262 
1263 static inline struct timespec ext4_current_time(struct inode *inode)
1264 {
1265 	return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1266 		current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1267 }
1268 
1269 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1270 {
1271 	return ino == EXT4_ROOT_INO ||
1272 		ino == EXT4_JOURNAL_INO ||
1273 		ino == EXT4_RESIZE_INO ||
1274 		(ino >= EXT4_FIRST_INO(sb) &&
1275 		 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1276 }
1277 
1278 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
1279 					      struct ext4_io_end *io_end)
1280 {
1281 	if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
1282 		io_end->flag |= EXT4_IO_END_UNWRITTEN;
1283 		atomic_inc(&EXT4_I(inode)->i_aiodio_unwritten);
1284 	}
1285 }
1286 
1287 /*
1288  * Inode dynamic state flags
1289  */
1290 enum {
1291 	EXT4_STATE_JDATA,		/* journaled data exists */
1292 	EXT4_STATE_NEW,			/* inode is newly created */
1293 	EXT4_STATE_XATTR,		/* has in-inode xattrs */
1294 	EXT4_STATE_NO_EXPAND,		/* No space for expansion */
1295 	EXT4_STATE_DA_ALLOC_CLOSE,	/* Alloc DA blks on close */
1296 	EXT4_STATE_EXT_MIGRATE,		/* Inode is migrating */
1297 	EXT4_STATE_DIO_UNWRITTEN,	/* need convert on dio done*/
1298 	EXT4_STATE_NEWENTRY,		/* File just added to dir */
1299 	EXT4_STATE_DELALLOC_RESERVED,	/* blks already reserved for delalloc */
1300 };
1301 
1302 #define EXT4_INODE_BIT_FNS(name, field, offset)				\
1303 static inline int ext4_test_inode_##name(struct inode *inode, int bit)	\
1304 {									\
1305 	return test_bit(bit + (offset), &EXT4_I(inode)->i_##field);	\
1306 }									\
1307 static inline void ext4_set_inode_##name(struct inode *inode, int bit)	\
1308 {									\
1309 	set_bit(bit + (offset), &EXT4_I(inode)->i_##field);		\
1310 }									\
1311 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1312 {									\
1313 	clear_bit(bit + (offset), &EXT4_I(inode)->i_##field);		\
1314 }
1315 
1316 EXT4_INODE_BIT_FNS(flag, flags, 0)
1317 #if (BITS_PER_LONG < 64)
1318 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1319 
1320 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1321 {
1322 	(ei)->i_state_flags = 0;
1323 }
1324 #else
1325 EXT4_INODE_BIT_FNS(state, flags, 32)
1326 
1327 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1328 {
1329 	/* We depend on the fact that callers will set i_flags */
1330 }
1331 #endif
1332 #else
1333 /* Assume that user mode programs are passing in an ext4fs superblock, not
1334  * a kernel struct super_block.  This will allow us to call the feature-test
1335  * macros from user land. */
1336 #define EXT4_SB(sb)	(sb)
1337 #endif
1338 
1339 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1340 
1341 /*
1342  * Codes for operating systems
1343  */
1344 #define EXT4_OS_LINUX		0
1345 #define EXT4_OS_HURD		1
1346 #define EXT4_OS_MASIX		2
1347 #define EXT4_OS_FREEBSD		3
1348 #define EXT4_OS_LITES		4
1349 
1350 /*
1351  * Revision levels
1352  */
1353 #define EXT4_GOOD_OLD_REV	0	/* The good old (original) format */
1354 #define EXT4_DYNAMIC_REV	1	/* V2 format w/ dynamic inode sizes */
1355 
1356 #define EXT4_CURRENT_REV	EXT4_GOOD_OLD_REV
1357 #define EXT4_MAX_SUPP_REV	EXT4_DYNAMIC_REV
1358 
1359 #define EXT4_GOOD_OLD_INODE_SIZE 128
1360 
1361 /*
1362  * Feature set definitions
1363  */
1364 
1365 #define EXT4_HAS_COMPAT_FEATURE(sb,mask)			\
1366 	((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1367 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask)			\
1368 	((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1369 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask)			\
1370 	((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1371 #define EXT4_SET_COMPAT_FEATURE(sb,mask)			\
1372 	EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1373 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask)			\
1374 	EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1375 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask)			\
1376 	EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1377 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask)			\
1378 	EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1379 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask)			\
1380 	EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1381 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask)			\
1382 	EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1383 
1384 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC	0x0001
1385 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES	0x0002
1386 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL		0x0004
1387 #define EXT4_FEATURE_COMPAT_EXT_ATTR		0x0008
1388 #define EXT4_FEATURE_COMPAT_RESIZE_INODE	0x0010
1389 #define EXT4_FEATURE_COMPAT_DIR_INDEX		0x0020
1390 
1391 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER	0x0001
1392 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE	0x0002
1393 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR	0x0004
1394 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE        0x0008
1395 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM		0x0010
1396 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK	0x0020
1397 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE	0x0040
1398 #define EXT4_FEATURE_RO_COMPAT_QUOTA		0x0100
1399 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC		0x0200
1400 
1401 #define EXT4_FEATURE_INCOMPAT_COMPRESSION	0x0001
1402 #define EXT4_FEATURE_INCOMPAT_FILETYPE		0x0002
1403 #define EXT4_FEATURE_INCOMPAT_RECOVER		0x0004 /* Needs recovery */
1404 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV	0x0008 /* Journal device */
1405 #define EXT4_FEATURE_INCOMPAT_META_BG		0x0010
1406 #define EXT4_FEATURE_INCOMPAT_EXTENTS		0x0040 /* extents support */
1407 #define EXT4_FEATURE_INCOMPAT_64BIT		0x0080
1408 #define EXT4_FEATURE_INCOMPAT_MMP               0x0100
1409 #define EXT4_FEATURE_INCOMPAT_FLEX_BG		0x0200
1410 #define EXT4_FEATURE_INCOMPAT_EA_INODE		0x0400 /* EA in inode */
1411 #define EXT4_FEATURE_INCOMPAT_DIRDATA		0x1000 /* data in dirent */
1412 
1413 #define EXT2_FEATURE_COMPAT_SUPP	EXT4_FEATURE_COMPAT_EXT_ATTR
1414 #define EXT2_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
1415 					 EXT4_FEATURE_INCOMPAT_META_BG)
1416 #define EXT2_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1417 					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1418 					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1419 
1420 #define EXT3_FEATURE_COMPAT_SUPP	EXT4_FEATURE_COMPAT_EXT_ATTR
1421 #define EXT3_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
1422 					 EXT4_FEATURE_INCOMPAT_RECOVER| \
1423 					 EXT4_FEATURE_INCOMPAT_META_BG)
1424 #define EXT3_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1425 					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1426 					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1427 
1428 #define EXT4_FEATURE_COMPAT_SUPP	EXT2_FEATURE_COMPAT_EXT_ATTR
1429 #define EXT4_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
1430 					 EXT4_FEATURE_INCOMPAT_RECOVER| \
1431 					 EXT4_FEATURE_INCOMPAT_META_BG| \
1432 					 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1433 					 EXT4_FEATURE_INCOMPAT_64BIT| \
1434 					 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1435 					 EXT4_FEATURE_INCOMPAT_MMP)
1436 #define EXT4_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1437 					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1438 					 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1439 					 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1440 					 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1441 					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1442 					 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1443 					 EXT4_FEATURE_RO_COMPAT_BIGALLOC)
1444 
1445 /*
1446  * Default values for user and/or group using reserved blocks
1447  */
1448 #define	EXT4_DEF_RESUID		0
1449 #define	EXT4_DEF_RESGID		0
1450 
1451 #define EXT4_DEF_INODE_READAHEAD_BLKS	32
1452 
1453 /*
1454  * Default mount options
1455  */
1456 #define EXT4_DEFM_DEBUG		0x0001
1457 #define EXT4_DEFM_BSDGROUPS	0x0002
1458 #define EXT4_DEFM_XATTR_USER	0x0004
1459 #define EXT4_DEFM_ACL		0x0008
1460 #define EXT4_DEFM_UID16		0x0010
1461 #define EXT4_DEFM_JMODE		0x0060
1462 #define EXT4_DEFM_JMODE_DATA	0x0020
1463 #define EXT4_DEFM_JMODE_ORDERED	0x0040
1464 #define EXT4_DEFM_JMODE_WBACK	0x0060
1465 #define EXT4_DEFM_NOBARRIER	0x0100
1466 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1467 #define EXT4_DEFM_DISCARD	0x0400
1468 #define EXT4_DEFM_NODELALLOC	0x0800
1469 
1470 /*
1471  * Default journal batch times
1472  */
1473 #define EXT4_DEF_MIN_BATCH_TIME	0
1474 #define EXT4_DEF_MAX_BATCH_TIME	15000 /* 15ms */
1475 
1476 /*
1477  * Minimum number of groups in a flexgroup before we separate out
1478  * directories into the first block group of a flexgroup
1479  */
1480 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME	4
1481 
1482 /*
1483  * Structure of a directory entry
1484  */
1485 #define EXT4_NAME_LEN 255
1486 
1487 struct ext4_dir_entry {
1488 	__le32	inode;			/* Inode number */
1489 	__le16	rec_len;		/* Directory entry length */
1490 	__le16	name_len;		/* Name length */
1491 	char	name[EXT4_NAME_LEN];	/* File name */
1492 };
1493 
1494 /*
1495  * The new version of the directory entry.  Since EXT4 structures are
1496  * stored in intel byte order, and the name_len field could never be
1497  * bigger than 255 chars, it's safe to reclaim the extra byte for the
1498  * file_type field.
1499  */
1500 struct ext4_dir_entry_2 {
1501 	__le32	inode;			/* Inode number */
1502 	__le16	rec_len;		/* Directory entry length */
1503 	__u8	name_len;		/* Name length */
1504 	__u8	file_type;
1505 	char	name[EXT4_NAME_LEN];	/* File name */
1506 };
1507 
1508 /*
1509  * Ext4 directory file types.  Only the low 3 bits are used.  The
1510  * other bits are reserved for now.
1511  */
1512 #define EXT4_FT_UNKNOWN		0
1513 #define EXT4_FT_REG_FILE	1
1514 #define EXT4_FT_DIR		2
1515 #define EXT4_FT_CHRDEV		3
1516 #define EXT4_FT_BLKDEV		4
1517 #define EXT4_FT_FIFO		5
1518 #define EXT4_FT_SOCK		6
1519 #define EXT4_FT_SYMLINK		7
1520 
1521 #define EXT4_FT_MAX		8
1522 
1523 /*
1524  * EXT4_DIR_PAD defines the directory entries boundaries
1525  *
1526  * NOTE: It must be a multiple of 4
1527  */
1528 #define EXT4_DIR_PAD			4
1529 #define EXT4_DIR_ROUND			(EXT4_DIR_PAD - 1)
1530 #define EXT4_DIR_REC_LEN(name_len)	(((name_len) + 8 + EXT4_DIR_ROUND) & \
1531 					 ~EXT4_DIR_ROUND)
1532 #define EXT4_MAX_REC_LEN		((1<<16)-1)
1533 
1534 /*
1535  * If we ever get support for fs block sizes > page_size, we'll need
1536  * to remove the #if statements in the next two functions...
1537  */
1538 static inline unsigned int
1539 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1540 {
1541 	unsigned len = le16_to_cpu(dlen);
1542 
1543 #if (PAGE_CACHE_SIZE >= 65536)
1544 	if (len == EXT4_MAX_REC_LEN || len == 0)
1545 		return blocksize;
1546 	return (len & 65532) | ((len & 3) << 16);
1547 #else
1548 	return len;
1549 #endif
1550 }
1551 
1552 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1553 {
1554 	if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1555 		BUG();
1556 #if (PAGE_CACHE_SIZE >= 65536)
1557 	if (len < 65536)
1558 		return cpu_to_le16(len);
1559 	if (len == blocksize) {
1560 		if (blocksize == 65536)
1561 			return cpu_to_le16(EXT4_MAX_REC_LEN);
1562 		else
1563 			return cpu_to_le16(0);
1564 	}
1565 	return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1566 #else
1567 	return cpu_to_le16(len);
1568 #endif
1569 }
1570 
1571 /*
1572  * Hash Tree Directory indexing
1573  * (c) Daniel Phillips, 2001
1574  */
1575 
1576 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1577 				      EXT4_FEATURE_COMPAT_DIR_INDEX) && \
1578 		    ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
1579 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1580 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1581 
1582 /* Legal values for the dx_root hash_version field: */
1583 
1584 #define DX_HASH_LEGACY		0
1585 #define DX_HASH_HALF_MD4	1
1586 #define DX_HASH_TEA		2
1587 #define DX_HASH_LEGACY_UNSIGNED	3
1588 #define DX_HASH_HALF_MD4_UNSIGNED	4
1589 #define DX_HASH_TEA_UNSIGNED		5
1590 
1591 #ifdef __KERNEL__
1592 
1593 /* hash info structure used by the directory hash */
1594 struct dx_hash_info
1595 {
1596 	u32		hash;
1597 	u32		minor_hash;
1598 	int		hash_version;
1599 	u32		*seed;
1600 };
1601 
1602 #define EXT4_HTREE_EOF	0x7fffffff
1603 
1604 /*
1605  * Control parameters used by ext4_htree_next_block
1606  */
1607 #define HASH_NB_ALWAYS		1
1608 
1609 
1610 /*
1611  * Describe an inode's exact location on disk and in memory
1612  */
1613 struct ext4_iloc
1614 {
1615 	struct buffer_head *bh;
1616 	unsigned long offset;
1617 	ext4_group_t block_group;
1618 };
1619 
1620 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
1621 {
1622 	return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
1623 }
1624 
1625 /*
1626  * This structure is stuffed into the struct file's private_data field
1627  * for directories.  It is where we put information so that we can do
1628  * readdir operations in hash tree order.
1629  */
1630 struct dir_private_info {
1631 	struct rb_root	root;
1632 	struct rb_node	*curr_node;
1633 	struct fname	*extra_fname;
1634 	loff_t		last_pos;
1635 	__u32		curr_hash;
1636 	__u32		curr_minor_hash;
1637 	__u32		next_hash;
1638 };
1639 
1640 /* calculate the first block number of the group */
1641 static inline ext4_fsblk_t
1642 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
1643 {
1644 	return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
1645 		le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
1646 }
1647 
1648 /*
1649  * Special error return code only used by dx_probe() and its callers.
1650  */
1651 #define ERR_BAD_DX_DIR	-75000
1652 
1653 void ext4_get_group_no_and_offset(struct super_block *sb, ext4_fsblk_t blocknr,
1654 			ext4_group_t *blockgrpp, ext4_grpblk_t *offsetp);
1655 
1656 /*
1657  * Timeout and state flag for lazy initialization inode thread.
1658  */
1659 #define EXT4_DEF_LI_WAIT_MULT			10
1660 #define EXT4_DEF_LI_MAX_START_DELAY		5
1661 #define EXT4_LAZYINIT_QUIT			0x0001
1662 #define EXT4_LAZYINIT_RUNNING			0x0002
1663 
1664 /*
1665  * Lazy inode table initialization info
1666  */
1667 struct ext4_lazy_init {
1668 	unsigned long		li_state;
1669 	struct list_head	li_request_list;
1670 	struct mutex		li_list_mtx;
1671 };
1672 
1673 struct ext4_li_request {
1674 	struct super_block	*lr_super;
1675 	struct ext4_sb_info	*lr_sbi;
1676 	ext4_group_t		lr_next_group;
1677 	struct list_head	lr_request;
1678 	unsigned long		lr_next_sched;
1679 	unsigned long		lr_timeout;
1680 };
1681 
1682 struct ext4_features {
1683 	struct kobject f_kobj;
1684 	struct completion f_kobj_unregister;
1685 };
1686 
1687 /*
1688  * This structure will be used for multiple mount protection. It will be
1689  * written into the block number saved in the s_mmp_block field in the
1690  * superblock. Programs that check MMP should assume that if
1691  * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
1692  * to use the filesystem, regardless of how old the timestamp is.
1693  */
1694 #define EXT4_MMP_MAGIC     0x004D4D50U /* ASCII for MMP */
1695 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
1696 #define EXT4_MMP_SEQ_FSCK  0xE24D4D50U /* mmp_seq value when being fscked */
1697 #define EXT4_MMP_SEQ_MAX   0xE24D4D4FU /* maximum valid mmp_seq value */
1698 
1699 struct mmp_struct {
1700 	__le32	mmp_magic;		/* Magic number for MMP */
1701 	__le32	mmp_seq;		/* Sequence no. updated periodically */
1702 
1703 	/*
1704 	 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
1705 	 * purposes and do not affect the correctness of the algorithm
1706 	 */
1707 	__le64	mmp_time;		/* Time last updated */
1708 	char	mmp_nodename[64];	/* Node which last updated MMP block */
1709 	char	mmp_bdevname[32];	/* Bdev which last updated MMP block */
1710 
1711 	/*
1712 	 * mmp_check_interval is used to verify if the MMP block has been
1713 	 * updated on the block device. The value is updated based on the
1714 	 * maximum time to write the MMP block during an update cycle.
1715 	 */
1716 	__le16	mmp_check_interval;
1717 
1718 	__le16	mmp_pad1;
1719 	__le32	mmp_pad2[227];
1720 };
1721 
1722 /* arguments passed to the mmp thread */
1723 struct mmpd_data {
1724 	struct buffer_head *bh; /* bh from initial read_mmp_block() */
1725 	struct super_block *sb;  /* super block of the fs */
1726 };
1727 
1728 /*
1729  * Check interval multiplier
1730  * The MMP block is written every update interval and initially checked every
1731  * update interval x the multiplier (the value is then adapted based on the
1732  * write latency). The reason is that writes can be delayed under load and we
1733  * don't want readers to incorrectly assume that the filesystem is no longer
1734  * in use.
1735  */
1736 #define EXT4_MMP_CHECK_MULT		2UL
1737 
1738 /*
1739  * Minimum interval for MMP checking in seconds.
1740  */
1741 #define EXT4_MMP_MIN_CHECK_INTERVAL	5UL
1742 
1743 /*
1744  * Maximum interval for MMP checking in seconds.
1745  */
1746 #define EXT4_MMP_MAX_CHECK_INTERVAL	300UL
1747 
1748 /*
1749  * Function prototypes
1750  */
1751 
1752 /*
1753  * Ok, these declarations are also in <linux/kernel.h> but none of the
1754  * ext4 source programs needs to include it so they are duplicated here.
1755  */
1756 # define NORET_TYPE	/**/
1757 # define ATTRIB_NORET	__attribute__((noreturn))
1758 # define NORET_AND	noreturn,
1759 
1760 /* bitmap.c */
1761 extern unsigned int ext4_count_free(struct buffer_head *, unsigned);
1762 
1763 /* balloc.c */
1764 extern unsigned int ext4_block_group(struct super_block *sb,
1765 			ext4_fsblk_t blocknr);
1766 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
1767 			ext4_fsblk_t blocknr);
1768 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
1769 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
1770 			ext4_group_t group);
1771 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
1772 					 ext4_fsblk_t goal,
1773 					 unsigned int flags,
1774 					 unsigned long *count,
1775 					 int *errp);
1776 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
1777 				    s64 nclusters, unsigned int flags);
1778 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
1779 extern void ext4_check_blocks_bitmap(struct super_block *);
1780 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
1781 						    ext4_group_t block_group,
1782 						    struct buffer_head ** bh);
1783 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
1784 struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
1785 				      ext4_group_t block_group);
1786 extern void ext4_init_block_bitmap(struct super_block *sb,
1787 				   struct buffer_head *bh,
1788 				   ext4_group_t group,
1789 				   struct ext4_group_desc *desc);
1790 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
1791 					      ext4_group_t block_group,
1792 					      struct ext4_group_desc *gdp);
1793 extern unsigned ext4_num_base_meta_clusters(struct super_block *sb,
1794 					    ext4_group_t block_group);
1795 extern unsigned ext4_num_overhead_clusters(struct super_block *sb,
1796 					   ext4_group_t block_group,
1797 					   struct ext4_group_desc *gdp);
1798 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
1799 
1800 /* dir.c */
1801 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
1802 				  struct file *,
1803 				  struct ext4_dir_entry_2 *,
1804 				  struct buffer_head *, unsigned int);
1805 #define ext4_check_dir_entry(dir, filp, de, bh, offset)			\
1806 	unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
1807 					(de), (bh), (offset)))
1808 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
1809 				    __u32 minor_hash,
1810 				    struct ext4_dir_entry_2 *dirent);
1811 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
1812 
1813 /* fsync.c */
1814 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
1815 extern int ext4_flush_completed_IO(struct inode *);
1816 
1817 /* hash.c */
1818 extern int ext4fs_dirhash(const char *name, int len, struct
1819 			  dx_hash_info *hinfo);
1820 
1821 /* ialloc.c */
1822 extern struct inode *ext4_new_inode(handle_t *, struct inode *, int,
1823 				    const struct qstr *qstr, __u32 goal,
1824 				    uid_t *owner);
1825 extern void ext4_free_inode(handle_t *, struct inode *);
1826 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
1827 extern unsigned long ext4_count_free_inodes(struct super_block *);
1828 extern unsigned long ext4_count_dirs(struct super_block *);
1829 extern void ext4_check_inodes_bitmap(struct super_block *);
1830 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
1831 extern int ext4_init_inode_table(struct super_block *sb,
1832 				 ext4_group_t group, int barrier);
1833 
1834 /* mballoc.c */
1835 extern long ext4_mb_stats;
1836 extern long ext4_mb_max_to_scan;
1837 extern int ext4_mb_init(struct super_block *, int);
1838 extern int ext4_mb_release(struct super_block *);
1839 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
1840 				struct ext4_allocation_request *, int *);
1841 extern int ext4_mb_reserve_blocks(struct super_block *, int);
1842 extern void ext4_discard_preallocations(struct inode *);
1843 extern int __init ext4_init_mballoc(void);
1844 extern void ext4_exit_mballoc(void);
1845 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
1846 			     struct buffer_head *bh, ext4_fsblk_t block,
1847 			     unsigned long count, int flags);
1848 extern int ext4_mb_add_groupinfo(struct super_block *sb,
1849 		ext4_group_t i, struct ext4_group_desc *desc);
1850 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
1851 				ext4_fsblk_t block, unsigned long count);
1852 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
1853 
1854 /* inode.c */
1855 struct buffer_head *ext4_getblk(handle_t *, struct inode *,
1856 						ext4_lblk_t, int, int *);
1857 struct buffer_head *ext4_bread(handle_t *, struct inode *,
1858 						ext4_lblk_t, int, int *);
1859 int ext4_get_block(struct inode *inode, sector_t iblock,
1860 				struct buffer_head *bh_result, int create);
1861 
1862 extern struct inode *ext4_iget(struct super_block *, unsigned long);
1863 extern int  ext4_write_inode(struct inode *, struct writeback_control *);
1864 extern int  ext4_setattr(struct dentry *, struct iattr *);
1865 extern int  ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
1866 				struct kstat *stat);
1867 extern void ext4_evict_inode(struct inode *);
1868 extern void ext4_clear_inode(struct inode *);
1869 extern int  ext4_sync_inode(handle_t *, struct inode *);
1870 extern void ext4_dirty_inode(struct inode *, int);
1871 extern int ext4_change_inode_journal_flag(struct inode *, int);
1872 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
1873 extern int ext4_can_truncate(struct inode *inode);
1874 extern void ext4_truncate(struct inode *);
1875 extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length);
1876 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
1877 extern void ext4_set_inode_flags(struct inode *);
1878 extern void ext4_get_inode_flags(struct ext4_inode_info *);
1879 extern int ext4_alloc_da_blocks(struct inode *inode);
1880 extern void ext4_set_aops(struct inode *inode);
1881 extern int ext4_writepage_trans_blocks(struct inode *);
1882 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
1883 extern int ext4_block_truncate_page(handle_t *handle,
1884 		struct address_space *mapping, loff_t from);
1885 extern int ext4_block_zero_page_range(handle_t *handle,
1886 		struct address_space *mapping, loff_t from, loff_t length);
1887 extern int ext4_discard_partial_page_buffers(handle_t *handle,
1888 		struct address_space *mapping, loff_t from,
1889 		loff_t length, int flags);
1890 extern int ext4_discard_partial_page_buffers_no_lock(handle_t *handle,
1891 		struct inode *inode, struct page *page, loff_t from,
1892 		loff_t length, int flags);
1893 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
1894 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
1895 extern void ext4_da_update_reserve_space(struct inode *inode,
1896 					int used, int quota_claim);
1897 
1898 /* indirect.c */
1899 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
1900 				struct ext4_map_blocks *map, int flags);
1901 extern ssize_t ext4_ind_direct_IO(int rw, struct kiocb *iocb,
1902 				const struct iovec *iov, loff_t offset,
1903 				unsigned long nr_segs);
1904 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
1905 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks, int chunk);
1906 extern void ext4_ind_truncate(struct inode *inode);
1907 
1908 /* ioctl.c */
1909 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
1910 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
1911 
1912 /* migrate.c */
1913 extern int ext4_ext_migrate(struct inode *);
1914 
1915 /* namei.c */
1916 extern int ext4_orphan_add(handle_t *, struct inode *);
1917 extern int ext4_orphan_del(handle_t *, struct inode *);
1918 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
1919 				__u32 start_minor_hash, __u32 *next_hash);
1920 
1921 /* resize.c */
1922 extern int ext4_group_add(struct super_block *sb,
1923 				struct ext4_new_group_data *input);
1924 extern int ext4_group_extend(struct super_block *sb,
1925 				struct ext4_super_block *es,
1926 				ext4_fsblk_t n_blocks_count);
1927 
1928 /* super.c */
1929 extern void *ext4_kvmalloc(size_t size, gfp_t flags);
1930 extern void *ext4_kvzalloc(size_t size, gfp_t flags);
1931 extern void ext4_kvfree(void *ptr);
1932 extern __printf(4, 5)
1933 void __ext4_error(struct super_block *, const char *, unsigned int,
1934 		  const char *, ...);
1935 #define ext4_error(sb, message...)	__ext4_error(sb, __func__,	\
1936 						     __LINE__, ## message)
1937 extern __printf(5, 6)
1938 void ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
1939 		      const char *, ...);
1940 extern __printf(5, 6)
1941 void ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
1942 		     const char *, ...);
1943 extern void __ext4_std_error(struct super_block *, const char *,
1944 			     unsigned int, int);
1945 extern __printf(4, 5)
1946 void __ext4_abort(struct super_block *, const char *, unsigned int,
1947 		  const char *, ...);
1948 #define ext4_abort(sb, message...)	__ext4_abort(sb, __func__, \
1949 						       __LINE__, ## message)
1950 extern __printf(4, 5)
1951 void __ext4_warning(struct super_block *, const char *, unsigned int,
1952 		    const char *, ...);
1953 #define ext4_warning(sb, message...)	__ext4_warning(sb, __func__, \
1954 						       __LINE__, ## message)
1955 extern __printf(3, 4)
1956 void ext4_msg(struct super_block *, const char *, const char *, ...);
1957 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
1958 			   const char *, unsigned int, const char *);
1959 #define dump_mmp_msg(sb, mmp, msg)	__dump_mmp_msg(sb, mmp, __func__, \
1960 						       __LINE__, msg)
1961 extern __printf(7, 8)
1962 void __ext4_grp_locked_error(const char *, unsigned int,
1963 			     struct super_block *, ext4_group_t,
1964 			     unsigned long, ext4_fsblk_t,
1965 			     const char *, ...);
1966 #define ext4_grp_locked_error(sb, grp, message...) \
1967 	__ext4_grp_locked_error(__func__, __LINE__, (sb), (grp), ## message)
1968 extern void ext4_update_dynamic_rev(struct super_block *sb);
1969 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
1970 					__u32 compat);
1971 extern int ext4_update_rocompat_feature(handle_t *handle,
1972 					struct super_block *sb,	__u32 rocompat);
1973 extern int ext4_update_incompat_feature(handle_t *handle,
1974 					struct super_block *sb,	__u32 incompat);
1975 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
1976 				      struct ext4_group_desc *bg);
1977 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
1978 				      struct ext4_group_desc *bg);
1979 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
1980 				     struct ext4_group_desc *bg);
1981 extern __u32 ext4_free_group_clusters(struct super_block *sb,
1982 				      struct ext4_group_desc *bg);
1983 extern __u32 ext4_free_inodes_count(struct super_block *sb,
1984 				 struct ext4_group_desc *bg);
1985 extern __u32 ext4_used_dirs_count(struct super_block *sb,
1986 				struct ext4_group_desc *bg);
1987 extern __u32 ext4_itable_unused_count(struct super_block *sb,
1988 				   struct ext4_group_desc *bg);
1989 extern void ext4_block_bitmap_set(struct super_block *sb,
1990 				  struct ext4_group_desc *bg, ext4_fsblk_t blk);
1991 extern void ext4_inode_bitmap_set(struct super_block *sb,
1992 				  struct ext4_group_desc *bg, ext4_fsblk_t blk);
1993 extern void ext4_inode_table_set(struct super_block *sb,
1994 				 struct ext4_group_desc *bg, ext4_fsblk_t blk);
1995 extern void ext4_free_group_clusters_set(struct super_block *sb,
1996 					 struct ext4_group_desc *bg,
1997 					 __u32 count);
1998 extern void ext4_free_inodes_set(struct super_block *sb,
1999 				struct ext4_group_desc *bg, __u32 count);
2000 extern void ext4_used_dirs_set(struct super_block *sb,
2001 				struct ext4_group_desc *bg, __u32 count);
2002 extern void ext4_itable_unused_set(struct super_block *sb,
2003 				   struct ext4_group_desc *bg, __u32 count);
2004 extern __le16 ext4_group_desc_csum(struct ext4_sb_info *sbi, __u32 group,
2005 				   struct ext4_group_desc *gdp);
2006 extern int ext4_group_desc_csum_verify(struct ext4_sb_info *sbi, __u32 group,
2007 				       struct ext4_group_desc *gdp);
2008 
2009 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2010 {
2011 	return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2012 		le32_to_cpu(es->s_blocks_count_lo);
2013 }
2014 
2015 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2016 {
2017 	return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2018 		le32_to_cpu(es->s_r_blocks_count_lo);
2019 }
2020 
2021 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2022 {
2023 	return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2024 		le32_to_cpu(es->s_free_blocks_count_lo);
2025 }
2026 
2027 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2028 					 ext4_fsblk_t blk)
2029 {
2030 	es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2031 	es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2032 }
2033 
2034 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2035 					      ext4_fsblk_t blk)
2036 {
2037 	es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2038 	es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2039 }
2040 
2041 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2042 					   ext4_fsblk_t blk)
2043 {
2044 	es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2045 	es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2046 }
2047 
2048 static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
2049 {
2050 	if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2051 		return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2052 			le32_to_cpu(raw_inode->i_size_lo);
2053 	else
2054 		return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2055 }
2056 
2057 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2058 {
2059 	raw_inode->i_size_lo = cpu_to_le32(i_size);
2060 	raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2061 }
2062 
2063 static inline
2064 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2065 					    ext4_group_t group)
2066 {
2067 	 struct ext4_group_info ***grp_info;
2068 	 long indexv, indexh;
2069 	 grp_info = EXT4_SB(sb)->s_group_info;
2070 	 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2071 	 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2072 	 return grp_info[indexv][indexh];
2073 }
2074 
2075 /*
2076  * Reading s_groups_count requires using smp_rmb() afterwards.  See
2077  * the locking protocol documented in the comments of ext4_group_add()
2078  * in resize.c
2079  */
2080 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2081 {
2082 	ext4_group_t	ngroups = EXT4_SB(sb)->s_groups_count;
2083 
2084 	smp_rmb();
2085 	return ngroups;
2086 }
2087 
2088 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2089 					     ext4_group_t block_group)
2090 {
2091 	return block_group >> sbi->s_log_groups_per_flex;
2092 }
2093 
2094 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2095 {
2096 	return 1 << sbi->s_log_groups_per_flex;
2097 }
2098 
2099 #define ext4_std_error(sb, errno)				\
2100 do {								\
2101 	if ((errno))						\
2102 		__ext4_std_error((sb), __func__, __LINE__, (errno));	\
2103 } while (0)
2104 
2105 #ifdef CONFIG_SMP
2106 /* Each CPU can accumulate percpu_counter_batch clusters in their local
2107  * counters. So we need to make sure we have free clusters more
2108  * than percpu_counter_batch  * nr_cpu_ids. Also add a window of 4 times.
2109  */
2110 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2111 #else
2112 #define EXT4_FREECLUSTERS_WATERMARK 0
2113 #endif
2114 
2115 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2116 {
2117 	/*
2118 	 * XXX: replace with spinlock if seen contended -bzzz
2119 	 */
2120 	down_write(&EXT4_I(inode)->i_data_sem);
2121 	if (newsize > EXT4_I(inode)->i_disksize)
2122 		EXT4_I(inode)->i_disksize = newsize;
2123 	up_write(&EXT4_I(inode)->i_data_sem);
2124 	return ;
2125 }
2126 
2127 struct ext4_group_info {
2128 	unsigned long   bb_state;
2129 	struct rb_root  bb_free_root;
2130 	ext4_grpblk_t	bb_first_free;	/* first free block */
2131 	ext4_grpblk_t	bb_free;	/* total free blocks */
2132 	ext4_grpblk_t	bb_fragments;	/* nr of freespace fragments */
2133 	ext4_grpblk_t	bb_largest_free_order;/* order of largest frag in BG */
2134 	struct          list_head bb_prealloc_list;
2135 #ifdef DOUBLE_CHECK
2136 	void            *bb_bitmap;
2137 #endif
2138 	struct rw_semaphore alloc_sem;
2139 	ext4_grpblk_t	bb_counters[];	/* Nr of free power-of-two-block
2140 					 * regions, index is order.
2141 					 * bb_counters[3] = 5 means
2142 					 * 5 free 8-block regions. */
2143 };
2144 
2145 #define EXT4_GROUP_INFO_NEED_INIT_BIT		0
2146 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT		1
2147 
2148 #define EXT4_MB_GRP_NEED_INIT(grp)	\
2149 	(test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2150 
2151 #define EXT4_MB_GRP_WAS_TRIMMED(grp)	\
2152 	(test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2153 #define EXT4_MB_GRP_SET_TRIMMED(grp)	\
2154 	(set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2155 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp)	\
2156 	(clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2157 
2158 #define EXT4_MAX_CONTENTION		8
2159 #define EXT4_CONTENTION_THRESHOLD	2
2160 
2161 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2162 					      ext4_group_t group)
2163 {
2164 	return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2165 }
2166 
2167 /*
2168  * Returns true if the filesystem is busy enough that attempts to
2169  * access the block group locks has run into contention.
2170  */
2171 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2172 {
2173 	return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2174 }
2175 
2176 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2177 {
2178 	spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2179 	if (spin_trylock(lock))
2180 		/*
2181 		 * We're able to grab the lock right away, so drop the
2182 		 * lock contention counter.
2183 		 */
2184 		atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2185 	else {
2186 		/*
2187 		 * The lock is busy, so bump the contention counter,
2188 		 * and then wait on the spin lock.
2189 		 */
2190 		atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2191 				  EXT4_MAX_CONTENTION);
2192 		spin_lock(lock);
2193 	}
2194 }
2195 
2196 static inline void ext4_unlock_group(struct super_block *sb,
2197 					ext4_group_t group)
2198 {
2199 	spin_unlock(ext4_group_lock_ptr(sb, group));
2200 }
2201 
2202 static inline void ext4_mark_super_dirty(struct super_block *sb)
2203 {
2204 	if (EXT4_SB(sb)->s_journal == NULL)
2205 		sb->s_dirt =1;
2206 }
2207 
2208 /*
2209  * Block validity checking
2210  */
2211 #define ext4_check_indirect_blockref(inode, bh)				\
2212 	ext4_check_blockref(__func__, __LINE__, inode,			\
2213 			    (__le32 *)(bh)->b_data,			\
2214 			    EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2215 
2216 #define ext4_ind_check_inode(inode)					\
2217 	ext4_check_blockref(__func__, __LINE__, inode,			\
2218 			    EXT4_I(inode)->i_data,			\
2219 			    EXT4_NDIR_BLOCKS)
2220 
2221 /*
2222  * Inodes and files operations
2223  */
2224 
2225 /* dir.c */
2226 extern const struct file_operations ext4_dir_operations;
2227 
2228 /* file.c */
2229 extern const struct inode_operations ext4_file_inode_operations;
2230 extern const struct file_operations ext4_file_operations;
2231 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
2232 
2233 /* namei.c */
2234 extern const struct inode_operations ext4_dir_inode_operations;
2235 extern const struct inode_operations ext4_special_inode_operations;
2236 extern struct dentry *ext4_get_parent(struct dentry *child);
2237 
2238 /* symlink.c */
2239 extern const struct inode_operations ext4_symlink_inode_operations;
2240 extern const struct inode_operations ext4_fast_symlink_inode_operations;
2241 
2242 /* block_validity */
2243 extern void ext4_release_system_zone(struct super_block *sb);
2244 extern int ext4_setup_system_zone(struct super_block *sb);
2245 extern int __init ext4_init_system_zone(void);
2246 extern void ext4_exit_system_zone(void);
2247 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
2248 				 ext4_fsblk_t start_blk,
2249 				 unsigned int count);
2250 extern int ext4_check_blockref(const char *, unsigned int,
2251 			       struct inode *, __le32 *, unsigned int);
2252 
2253 /* extents.c */
2254 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
2255 extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
2256 extern int ext4_ext_index_trans_blocks(struct inode *inode, int nrblocks,
2257 				       int chunk);
2258 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
2259 			       struct ext4_map_blocks *map, int flags);
2260 extern void ext4_ext_truncate(struct inode *);
2261 extern int ext4_ext_punch_hole(struct file *file, loff_t offset,
2262 				loff_t length);
2263 extern void ext4_ext_init(struct super_block *);
2264 extern void ext4_ext_release(struct super_block *);
2265 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
2266 			  loff_t len);
2267 extern int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset,
2268 			  ssize_t len);
2269 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
2270 			   struct ext4_map_blocks *map, int flags);
2271 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
2272 			__u64 start, __u64 len);
2273 /* move_extent.c */
2274 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
2275 			     __u64 start_orig, __u64 start_donor,
2276 			     __u64 len, __u64 *moved_len);
2277 
2278 /* page-io.c */
2279 extern int __init ext4_init_pageio(void);
2280 extern void ext4_exit_pageio(void);
2281 extern void ext4_ioend_wait(struct inode *);
2282 extern void ext4_free_io_end(ext4_io_end_t *io);
2283 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
2284 extern int ext4_end_io_nolock(ext4_io_end_t *io);
2285 extern void ext4_io_submit(struct ext4_io_submit *io);
2286 extern int ext4_bio_write_page(struct ext4_io_submit *io,
2287 			       struct page *page,
2288 			       int len,
2289 			       struct writeback_control *wbc);
2290 
2291 /* mmp.c */
2292 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
2293 
2294 /* BH_Uninit flag: blocks are allocated but uninitialized on disk */
2295 enum ext4_state_bits {
2296 	BH_Uninit	/* blocks are allocated but uninitialized on disk */
2297 	  = BH_JBDPrivateStart,
2298 	BH_AllocFromCluster,	/* allocated blocks were part of already
2299 				 * allocated cluster. Note that this flag will
2300 				 * never, ever appear in a buffer_head's state
2301 				 * flag. See EXT4_MAP_FROM_CLUSTER to see where
2302 				 * this is used. */
2303 	BH_Da_Mapped,	/* Delayed allocated block that now has a mapping. This
2304 			 * flag is set when ext4_map_blocks is called on a
2305 			 * delayed allocated block to get its real mapping. */
2306 };
2307 
2308 BUFFER_FNS(Uninit, uninit)
2309 TAS_BUFFER_FNS(Uninit, uninit)
2310 BUFFER_FNS(Da_Mapped, da_mapped)
2311 
2312 /*
2313  * Add new method to test wether block and inode bitmaps are properly
2314  * initialized. With uninit_bg reading the block from disk is not enough
2315  * to mark the bitmap uptodate. We need to also zero-out the bitmap
2316  */
2317 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
2318 
2319 static inline int bitmap_uptodate(struct buffer_head *bh)
2320 {
2321 	return (buffer_uptodate(bh) &&
2322 			test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
2323 }
2324 static inline void set_bitmap_uptodate(struct buffer_head *bh)
2325 {
2326 	set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
2327 }
2328 
2329 #define in_range(b, first, len)	((b) >= (first) && (b) <= (first) + (len) - 1)
2330 
2331 /* For ioend & aio unwritten conversion wait queues */
2332 #define EXT4_WQ_HASH_SZ		37
2333 #define ext4_ioend_wq(v)   (&ext4__ioend_wq[((unsigned long)(v)) %\
2334 					    EXT4_WQ_HASH_SZ])
2335 #define ext4_aio_mutex(v)  (&ext4__aio_mutex[((unsigned long)(v)) %\
2336 					     EXT4_WQ_HASH_SZ])
2337 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
2338 extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ];
2339 
2340 #define EXT4_RESIZING	0
2341 extern int ext4_resize_begin(struct super_block *sb);
2342 extern void ext4_resize_end(struct super_block *sb);
2343 
2344 #endif	/* __KERNEL__ */
2345 
2346 #include "ext4_extents.h"
2347 
2348 #endif	/* _EXT4_H */
2349