xref: /openbmc/linux/fs/ext4/ext4.h (revision e8e0929d)
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 
33 /*
34  * The fourth extended filesystem constants/structures
35  */
36 
37 /*
38  * Define EXT4FS_DEBUG to produce debug messages
39  */
40 #undef EXT4FS_DEBUG
41 
42 /*
43  * Debug code
44  */
45 #ifdef EXT4FS_DEBUG
46 #define ext4_debug(f, a...)						\
47 	do {								\
48 		printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:",	\
49 			__FILE__, __LINE__, __func__);			\
50 		printk(KERN_DEBUG f, ## a);				\
51 	} while (0)
52 #else
53 #define ext4_debug(f, a...)	do {} while (0)
54 #endif
55 
56 /* data type for block offset of block group */
57 typedef int ext4_grpblk_t;
58 
59 /* data type for filesystem-wide blocks number */
60 typedef unsigned long long ext4_fsblk_t;
61 
62 /* data type for file logical block number */
63 typedef __u32 ext4_lblk_t;
64 
65 /* data type for block group number */
66 typedef unsigned int ext4_group_t;
67 
68 /*
69  * Flags used in mballoc's allocation_context flags field.
70  *
71  * Also used to show what's going on for debugging purposes when the
72  * flag field is exported via the traceport interface
73  */
74 
75 /* prefer goal again. length */
76 #define EXT4_MB_HINT_MERGE		0x0001
77 /* blocks already reserved */
78 #define EXT4_MB_HINT_RESERVED		0x0002
79 /* metadata is being allocated */
80 #define EXT4_MB_HINT_METADATA		0x0004
81 /* first blocks in the file */
82 #define EXT4_MB_HINT_FIRST		0x0008
83 /* search for the best chunk */
84 #define EXT4_MB_HINT_BEST		0x0010
85 /* data is being allocated */
86 #define EXT4_MB_HINT_DATA		0x0020
87 /* don't preallocate (for tails) */
88 #define EXT4_MB_HINT_NOPREALLOC		0x0040
89 /* allocate for locality group */
90 #define EXT4_MB_HINT_GROUP_ALLOC	0x0080
91 /* allocate goal blocks or none */
92 #define EXT4_MB_HINT_GOAL_ONLY		0x0100
93 /* goal is meaningful */
94 #define EXT4_MB_HINT_TRY_GOAL		0x0200
95 /* blocks already pre-reserved by delayed allocation */
96 #define EXT4_MB_DELALLOC_RESERVED	0x0400
97 /* We are doing stream allocation */
98 #define EXT4_MB_STREAM_ALLOC		0x0800
99 
100 
101 struct ext4_allocation_request {
102 	/* target inode for block we're allocating */
103 	struct inode *inode;
104 	/* how many blocks we want to allocate */
105 	unsigned int len;
106 	/* logical block in target inode */
107 	ext4_lblk_t logical;
108 	/* the closest logical allocated block to the left */
109 	ext4_lblk_t lleft;
110 	/* the closest logical allocated block to the right */
111 	ext4_lblk_t lright;
112 	/* phys. target (a hint) */
113 	ext4_fsblk_t goal;
114 	/* phys. block for the closest logical allocated block to the left */
115 	ext4_fsblk_t pleft;
116 	/* phys. block for the closest logical allocated block to the right */
117 	ext4_fsblk_t pright;
118 	/* flags. see above EXT4_MB_HINT_* */
119 	unsigned int flags;
120 };
121 
122 /*
123  * For delayed allocation tracking
124  */
125 struct mpage_da_data {
126 	struct inode *inode;
127 	sector_t b_blocknr;		/* start block number of extent */
128 	size_t b_size;			/* size of extent */
129 	unsigned long b_state;		/* state of the extent */
130 	unsigned long first_page, next_page;	/* extent of pages */
131 	struct writeback_control *wbc;
132 	int io_done;
133 	int pages_written;
134 	int retval;
135 };
136 #define	DIO_AIO_UNWRITTEN	0x1
137 typedef struct ext4_io_end {
138 	struct list_head	list;		/* per-file finished AIO list */
139 	struct inode		*inode;		/* file being written to */
140 	unsigned int		flag;		/* unwritten or not */
141 	int			error;		/* I/O error code */
142 	ext4_lblk_t		offset;		/* offset in the file */
143 	size_t			size;		/* size of the extent */
144 	struct work_struct	work;		/* data work queue */
145 } ext4_io_end_t;
146 
147 /*
148  * Special inodes numbers
149  */
150 #define	EXT4_BAD_INO		 1	/* Bad blocks inode */
151 #define EXT4_ROOT_INO		 2	/* Root inode */
152 #define EXT4_BOOT_LOADER_INO	 5	/* Boot loader inode */
153 #define EXT4_UNDEL_DIR_INO	 6	/* Undelete directory inode */
154 #define EXT4_RESIZE_INO		 7	/* Reserved group descriptors inode */
155 #define EXT4_JOURNAL_INO	 8	/* Journal inode */
156 
157 /* First non-reserved inode for old ext4 filesystems */
158 #define EXT4_GOOD_OLD_FIRST_INO	11
159 
160 /*
161  * Maximal count of links to a file
162  */
163 #define EXT4_LINK_MAX		65000
164 
165 /*
166  * Macro-instructions used to manage several block sizes
167  */
168 #define EXT4_MIN_BLOCK_SIZE		1024
169 #define	EXT4_MAX_BLOCK_SIZE		65536
170 #define EXT4_MIN_BLOCK_LOG_SIZE		10
171 #ifdef __KERNEL__
172 # define EXT4_BLOCK_SIZE(s)		((s)->s_blocksize)
173 #else
174 # define EXT4_BLOCK_SIZE(s)		(EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
175 #endif
176 #define	EXT4_ADDR_PER_BLOCK(s)		(EXT4_BLOCK_SIZE(s) / sizeof(__u32))
177 #ifdef __KERNEL__
178 # define EXT4_BLOCK_SIZE_BITS(s)	((s)->s_blocksize_bits)
179 #else
180 # define EXT4_BLOCK_SIZE_BITS(s)	((s)->s_log_block_size + 10)
181 #endif
182 #ifdef __KERNEL__
183 #define	EXT4_ADDR_PER_BLOCK_BITS(s)	(EXT4_SB(s)->s_addr_per_block_bits)
184 #define EXT4_INODE_SIZE(s)		(EXT4_SB(s)->s_inode_size)
185 #define EXT4_FIRST_INO(s)		(EXT4_SB(s)->s_first_ino)
186 #else
187 #define EXT4_INODE_SIZE(s)	(((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
188 				 EXT4_GOOD_OLD_INODE_SIZE : \
189 				 (s)->s_inode_size)
190 #define EXT4_FIRST_INO(s)	(((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
191 				 EXT4_GOOD_OLD_FIRST_INO : \
192 				 (s)->s_first_ino)
193 #endif
194 #define EXT4_BLOCK_ALIGN(size, blkbits)		ALIGN((size), (1 << (blkbits)))
195 
196 /*
197  * Structure of a blocks group descriptor
198  */
199 struct ext4_group_desc
200 {
201 	__le32	bg_block_bitmap_lo;	/* Blocks bitmap block */
202 	__le32	bg_inode_bitmap_lo;	/* Inodes bitmap block */
203 	__le32	bg_inode_table_lo;	/* Inodes table block */
204 	__le16	bg_free_blocks_count_lo;/* Free blocks count */
205 	__le16	bg_free_inodes_count_lo;/* Free inodes count */
206 	__le16	bg_used_dirs_count_lo;	/* Directories count */
207 	__le16	bg_flags;		/* EXT4_BG_flags (INODE_UNINIT, etc) */
208 	__u32	bg_reserved[2];		/* Likely block/inode bitmap checksum */
209 	__le16  bg_itable_unused_lo;	/* Unused inodes count */
210 	__le16  bg_checksum;		/* crc16(sb_uuid+group+desc) */
211 	__le32	bg_block_bitmap_hi;	/* Blocks bitmap block MSB */
212 	__le32	bg_inode_bitmap_hi;	/* Inodes bitmap block MSB */
213 	__le32	bg_inode_table_hi;	/* Inodes table block MSB */
214 	__le16	bg_free_blocks_count_hi;/* Free blocks count MSB */
215 	__le16	bg_free_inodes_count_hi;/* Free inodes count MSB */
216 	__le16	bg_used_dirs_count_hi;	/* Directories count MSB */
217 	__le16  bg_itable_unused_hi;    /* Unused inodes count MSB */
218 	__u32	bg_reserved2[3];
219 };
220 
221 /*
222  * Structure of a flex block group info
223  */
224 
225 struct flex_groups {
226 	atomic_t free_inodes;
227 	atomic_t free_blocks;
228 	atomic_t used_dirs;
229 };
230 
231 #define EXT4_BG_INODE_UNINIT	0x0001 /* Inode table/bitmap not in use */
232 #define EXT4_BG_BLOCK_UNINIT	0x0002 /* Block bitmap not in use */
233 #define EXT4_BG_INODE_ZEROED	0x0004 /* On-disk itable initialized to zero */
234 
235 /*
236  * Macro-instructions used to manage group descriptors
237  */
238 #define EXT4_MIN_DESC_SIZE		32
239 #define EXT4_MIN_DESC_SIZE_64BIT	64
240 #define	EXT4_MAX_DESC_SIZE		EXT4_MIN_BLOCK_SIZE
241 #define EXT4_DESC_SIZE(s)		(EXT4_SB(s)->s_desc_size)
242 #ifdef __KERNEL__
243 # define EXT4_BLOCKS_PER_GROUP(s)	(EXT4_SB(s)->s_blocks_per_group)
244 # define EXT4_DESC_PER_BLOCK(s)		(EXT4_SB(s)->s_desc_per_block)
245 # define EXT4_INODES_PER_GROUP(s)	(EXT4_SB(s)->s_inodes_per_group)
246 # define EXT4_DESC_PER_BLOCK_BITS(s)	(EXT4_SB(s)->s_desc_per_block_bits)
247 #else
248 # define EXT4_BLOCKS_PER_GROUP(s)	((s)->s_blocks_per_group)
249 # define EXT4_DESC_PER_BLOCK(s)		(EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
250 # define EXT4_INODES_PER_GROUP(s)	((s)->s_inodes_per_group)
251 #endif
252 
253 /*
254  * Constants relative to the data blocks
255  */
256 #define	EXT4_NDIR_BLOCKS		12
257 #define	EXT4_IND_BLOCK			EXT4_NDIR_BLOCKS
258 #define	EXT4_DIND_BLOCK			(EXT4_IND_BLOCK + 1)
259 #define	EXT4_TIND_BLOCK			(EXT4_DIND_BLOCK + 1)
260 #define	EXT4_N_BLOCKS			(EXT4_TIND_BLOCK + 1)
261 
262 /*
263  * Inode flags
264  */
265 #define	EXT4_SECRM_FL			0x00000001 /* Secure deletion */
266 #define	EXT4_UNRM_FL			0x00000002 /* Undelete */
267 #define	EXT4_COMPR_FL			0x00000004 /* Compress file */
268 #define EXT4_SYNC_FL			0x00000008 /* Synchronous updates */
269 #define EXT4_IMMUTABLE_FL		0x00000010 /* Immutable file */
270 #define EXT4_APPEND_FL			0x00000020 /* writes to file may only append */
271 #define EXT4_NODUMP_FL			0x00000040 /* do not dump file */
272 #define EXT4_NOATIME_FL			0x00000080 /* do not update atime */
273 /* Reserved for compression usage... */
274 #define EXT4_DIRTY_FL			0x00000100
275 #define EXT4_COMPRBLK_FL		0x00000200 /* One or more compressed clusters */
276 #define EXT4_NOCOMPR_FL			0x00000400 /* Don't compress */
277 #define EXT4_ECOMPR_FL			0x00000800 /* Compression error */
278 /* End compression flags --- maybe not all used */
279 #define EXT4_INDEX_FL			0x00001000 /* hash-indexed directory */
280 #define EXT4_IMAGIC_FL			0x00002000 /* AFS directory */
281 #define EXT4_JOURNAL_DATA_FL		0x00004000 /* file data should be journaled */
282 #define EXT4_NOTAIL_FL			0x00008000 /* file tail should not be merged */
283 #define EXT4_DIRSYNC_FL			0x00010000 /* dirsync behaviour (directories only) */
284 #define EXT4_TOPDIR_FL			0x00020000 /* Top of directory hierarchies*/
285 #define EXT4_HUGE_FILE_FL               0x00040000 /* Set to each huge file */
286 #define EXT4_EXTENTS_FL			0x00080000 /* Inode uses extents */
287 #define EXT4_RESERVED_FL		0x80000000 /* reserved for ext4 lib */
288 
289 #define EXT4_FL_USER_VISIBLE		0x000BDFFF /* User visible flags */
290 #define EXT4_FL_USER_MODIFIABLE		0x000B80FF /* User modifiable flags */
291 
292 /* Flags that should be inherited by new inodes from their parent. */
293 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
294 			   EXT4_SYNC_FL | EXT4_IMMUTABLE_FL | EXT4_APPEND_FL |\
295 			   EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
296 			   EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
297 			   EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
298 
299 /* Flags that are appropriate for regular files (all but dir-specific ones). */
300 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
301 
302 /* Flags that are appropriate for non-directories/regular files. */
303 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
304 
305 /* Mask out flags that are inappropriate for the given type of inode. */
306 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
307 {
308 	if (S_ISDIR(mode))
309 		return flags;
310 	else if (S_ISREG(mode))
311 		return flags & EXT4_REG_FLMASK;
312 	else
313 		return flags & EXT4_OTHER_FLMASK;
314 }
315 
316 /*
317  * Inode dynamic state flags
318  */
319 #define EXT4_STATE_JDATA		0x00000001 /* journaled data exists */
320 #define EXT4_STATE_NEW			0x00000002 /* inode is newly created */
321 #define EXT4_STATE_XATTR		0x00000004 /* has in-inode xattrs */
322 #define EXT4_STATE_NO_EXPAND		0x00000008 /* No space for expansion */
323 #define EXT4_STATE_DA_ALLOC_CLOSE	0x00000010 /* Alloc DA blks on close */
324 #define EXT4_STATE_EXT_MIGRATE		0x00000020 /* Inode is migrating */
325 
326 /* Used to pass group descriptor data when online resize is done */
327 struct ext4_new_group_input {
328 	__u32 group;		/* Group number for this data */
329 	__u64 block_bitmap;	/* Absolute block number of block bitmap */
330 	__u64 inode_bitmap;	/* Absolute block number of inode bitmap */
331 	__u64 inode_table;	/* Absolute block number of inode table start */
332 	__u32 blocks_count;	/* Total number of blocks in this group */
333 	__u16 reserved_blocks;	/* Number of reserved blocks in this group */
334 	__u16 unused;
335 };
336 
337 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
338 struct ext4_new_group_data {
339 	__u32 group;
340 	__u64 block_bitmap;
341 	__u64 inode_bitmap;
342 	__u64 inode_table;
343 	__u32 blocks_count;
344 	__u16 reserved_blocks;
345 	__u16 unused;
346 	__u32 free_blocks_count;
347 };
348 
349 /*
350  * Flags used by ext4_get_blocks()
351  */
352 	/* Allocate any needed blocks and/or convert an unitialized
353 	   extent to be an initialized ext4 */
354 #define EXT4_GET_BLOCKS_CREATE			0x0001
355 	/* Request the creation of an unitialized extent */
356 #define EXT4_GET_BLOCKS_UNINIT_EXT		0x0002
357 #define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT	(EXT4_GET_BLOCKS_UNINIT_EXT|\
358 						 EXT4_GET_BLOCKS_CREATE)
359 	/* Caller is from the delayed allocation writeout path,
360 	   so set the magic i_delalloc_reserve_flag after taking the
361 	   inode allocation semaphore for */
362 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE	0x0004
363 	/* Call ext4_da_update_reserve_space() after successfully
364 	   allocating the blocks */
365 #define EXT4_GET_BLOCKS_UPDATE_RESERVE_SPACE	0x0008
366 	/* caller is from the direct IO path, request to creation of an
367 	unitialized extents if not allocated, split the uninitialized
368 	extent if blocks has been preallocated already*/
369 #define EXT4_GET_BLOCKS_DIO			0x0010
370 #define EXT4_GET_BLOCKS_CONVERT			0x0020
371 #define EXT4_GET_BLOCKS_DIO_CREATE_EXT		(EXT4_GET_BLOCKS_DIO|\
372 					 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
373 	/* Convert extent to initialized after direct IO complete */
374 #define EXT4_GET_BLOCKS_DIO_CONVERT_EXT		(EXT4_GET_BLOCKS_CONVERT|\
375 					 EXT4_GET_BLOCKS_DIO_CREATE_EXT)
376 
377 /*
378  * ioctl commands
379  */
380 #define	EXT4_IOC_GETFLAGS		FS_IOC_GETFLAGS
381 #define	EXT4_IOC_SETFLAGS		FS_IOC_SETFLAGS
382 #define	EXT4_IOC_GETVERSION		_IOR('f', 3, long)
383 #define	EXT4_IOC_SETVERSION		_IOW('f', 4, long)
384 #define	EXT4_IOC_GETVERSION_OLD		FS_IOC_GETVERSION
385 #define	EXT4_IOC_SETVERSION_OLD		FS_IOC_SETVERSION
386 #ifdef CONFIG_JBD2_DEBUG
387 #define EXT4_IOC_WAIT_FOR_READONLY	_IOR('f', 99, long)
388 #endif
389 #define EXT4_IOC_GETRSVSZ		_IOR('f', 5, long)
390 #define EXT4_IOC_SETRSVSZ		_IOW('f', 6, long)
391 #define EXT4_IOC_GROUP_EXTEND		_IOW('f', 7, unsigned long)
392 #define EXT4_IOC_GROUP_ADD		_IOW('f', 8, struct ext4_new_group_input)
393 #define EXT4_IOC_MIGRATE		_IO('f', 9)
394  /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
395  /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
396 #define EXT4_IOC_ALLOC_DA_BLKS		_IO('f', 12)
397 #define EXT4_IOC_MOVE_EXT		_IOWR('f', 15, struct move_extent)
398 
399 /*
400  * ioctl commands in 32 bit emulation
401  */
402 #define EXT4_IOC32_GETFLAGS		FS_IOC32_GETFLAGS
403 #define EXT4_IOC32_SETFLAGS		FS_IOC32_SETFLAGS
404 #define EXT4_IOC32_GETVERSION		_IOR('f', 3, int)
405 #define EXT4_IOC32_SETVERSION		_IOW('f', 4, int)
406 #define EXT4_IOC32_GETRSVSZ		_IOR('f', 5, int)
407 #define EXT4_IOC32_SETRSVSZ		_IOW('f', 6, int)
408 #define EXT4_IOC32_GROUP_EXTEND		_IOW('f', 7, unsigned int)
409 #ifdef CONFIG_JBD2_DEBUG
410 #define EXT4_IOC32_WAIT_FOR_READONLY	_IOR('f', 99, int)
411 #endif
412 #define EXT4_IOC32_GETVERSION_OLD	FS_IOC32_GETVERSION
413 #define EXT4_IOC32_SETVERSION_OLD	FS_IOC32_SETVERSION
414 
415 
416 /*
417  *  Mount options
418  */
419 struct ext4_mount_options {
420 	unsigned long s_mount_opt;
421 	uid_t s_resuid;
422 	gid_t s_resgid;
423 	unsigned long s_commit_interval;
424 	u32 s_min_batch_time, s_max_batch_time;
425 #ifdef CONFIG_QUOTA
426 	int s_jquota_fmt;
427 	char *s_qf_names[MAXQUOTAS];
428 #endif
429 };
430 
431 /* Max physical block we can addres w/o extents */
432 #define EXT4_MAX_BLOCK_FILE_PHYS	0xFFFFFFFF
433 
434 /*
435  * Structure of an inode on the disk
436  */
437 struct ext4_inode {
438 	__le16	i_mode;		/* File mode */
439 	__le16	i_uid;		/* Low 16 bits of Owner Uid */
440 	__le32	i_size_lo;	/* Size in bytes */
441 	__le32	i_atime;	/* Access time */
442 	__le32	i_ctime;	/* Inode Change time */
443 	__le32	i_mtime;	/* Modification time */
444 	__le32	i_dtime;	/* Deletion Time */
445 	__le16	i_gid;		/* Low 16 bits of Group Id */
446 	__le16	i_links_count;	/* Links count */
447 	__le32	i_blocks_lo;	/* Blocks count */
448 	__le32	i_flags;	/* File flags */
449 	union {
450 		struct {
451 			__le32  l_i_version;
452 		} linux1;
453 		struct {
454 			__u32  h_i_translator;
455 		} hurd1;
456 		struct {
457 			__u32  m_i_reserved1;
458 		} masix1;
459 	} osd1;				/* OS dependent 1 */
460 	__le32	i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
461 	__le32	i_generation;	/* File version (for NFS) */
462 	__le32	i_file_acl_lo;	/* File ACL */
463 	__le32	i_size_high;
464 	__le32	i_obso_faddr;	/* Obsoleted fragment address */
465 	union {
466 		struct {
467 			__le16	l_i_blocks_high; /* were l_i_reserved1 */
468 			__le16	l_i_file_acl_high;
469 			__le16	l_i_uid_high;	/* these 2 fields */
470 			__le16	l_i_gid_high;	/* were reserved2[0] */
471 			__u32	l_i_reserved2;
472 		} linux2;
473 		struct {
474 			__le16	h_i_reserved1;	/* Obsoleted fragment number/size which are removed in ext4 */
475 			__u16	h_i_mode_high;
476 			__u16	h_i_uid_high;
477 			__u16	h_i_gid_high;
478 			__u32	h_i_author;
479 		} hurd2;
480 		struct {
481 			__le16	h_i_reserved1;	/* Obsoleted fragment number/size which are removed in ext4 */
482 			__le16	m_i_file_acl_high;
483 			__u32	m_i_reserved2[2];
484 		} masix2;
485 	} osd2;				/* OS dependent 2 */
486 	__le16	i_extra_isize;
487 	__le16	i_pad1;
488 	__le32  i_ctime_extra;  /* extra Change time      (nsec << 2 | epoch) */
489 	__le32  i_mtime_extra;  /* extra Modification time(nsec << 2 | epoch) */
490 	__le32  i_atime_extra;  /* extra Access time      (nsec << 2 | epoch) */
491 	__le32  i_crtime;       /* File Creation time */
492 	__le32  i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
493 	__le32  i_version_hi;	/* high 32 bits for 64-bit version */
494 };
495 
496 struct move_extent {
497 	__u32 reserved;		/* should be zero */
498 	__u32 donor_fd;		/* donor file descriptor */
499 	__u64 orig_start;	/* logical start offset in block for orig */
500 	__u64 donor_start;	/* logical start offset in block for donor */
501 	__u64 len;		/* block length to be moved */
502 	__u64 moved_len;	/* moved block length */
503 };
504 
505 #define EXT4_EPOCH_BITS 2
506 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
507 #define EXT4_NSEC_MASK  (~0UL << EXT4_EPOCH_BITS)
508 
509 /*
510  * Extended fields will fit into an inode if the filesystem was formatted
511  * with large inodes (-I 256 or larger) and there are not currently any EAs
512  * consuming all of the available space. For new inodes we always reserve
513  * enough space for the kernel's known extended fields, but for inodes
514  * created with an old kernel this might not have been the case. None of
515  * the extended inode fields is critical for correct filesystem operation.
516  * This macro checks if a certain field fits in the inode. Note that
517  * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
518  */
519 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field)	\
520 	((offsetof(typeof(*ext4_inode), field) +	\
521 	  sizeof((ext4_inode)->field))			\
522 	<= (EXT4_GOOD_OLD_INODE_SIZE +			\
523 	    (einode)->i_extra_isize))			\
524 
525 static inline __le32 ext4_encode_extra_time(struct timespec *time)
526 {
527        return cpu_to_le32((sizeof(time->tv_sec) > 4 ?
528 			   (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) |
529                           ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK));
530 }
531 
532 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
533 {
534        if (sizeof(time->tv_sec) > 4)
535 	       time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK)
536 			       << 32;
537        time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
538 }
539 
540 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode)			       \
541 do {									       \
542 	(raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec);	       \
543 	if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     \
544 		(raw_inode)->xtime ## _extra =				       \
545 				ext4_encode_extra_time(&(inode)->xtime);       \
546 } while (0)
547 
548 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode)			       \
549 do {									       \
550 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))		       \
551 		(raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec);      \
552 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))	       \
553 		(raw_inode)->xtime ## _extra =				       \
554 				ext4_encode_extra_time(&(einode)->xtime);      \
555 } while (0)
556 
557 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode)			       \
558 do {									       \
559 	(inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime);       \
560 	if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     \
561 		ext4_decode_extra_time(&(inode)->xtime,			       \
562 				       raw_inode->xtime ## _extra);	       \
563 } while (0)
564 
565 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode)			       \
566 do {									       \
567 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))		       \
568 		(einode)->xtime.tv_sec = 				       \
569 			(signed)le32_to_cpu((raw_inode)->xtime);	       \
570 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))	       \
571 		ext4_decode_extra_time(&(einode)->xtime,		       \
572 				       raw_inode->xtime ## _extra);	       \
573 } while (0)
574 
575 #define i_disk_version osd1.linux1.l_i_version
576 
577 #if defined(__KERNEL__) || defined(__linux__)
578 #define i_reserved1	osd1.linux1.l_i_reserved1
579 #define i_file_acl_high	osd2.linux2.l_i_file_acl_high
580 #define i_blocks_high	osd2.linux2.l_i_blocks_high
581 #define i_uid_low	i_uid
582 #define i_gid_low	i_gid
583 #define i_uid_high	osd2.linux2.l_i_uid_high
584 #define i_gid_high	osd2.linux2.l_i_gid_high
585 #define i_reserved2	osd2.linux2.l_i_reserved2
586 
587 #elif defined(__GNU__)
588 
589 #define i_translator	osd1.hurd1.h_i_translator
590 #define i_uid_high	osd2.hurd2.h_i_uid_high
591 #define i_gid_high	osd2.hurd2.h_i_gid_high
592 #define i_author	osd2.hurd2.h_i_author
593 
594 #elif defined(__masix__)
595 
596 #define i_reserved1	osd1.masix1.m_i_reserved1
597 #define i_file_acl_high	osd2.masix2.m_i_file_acl_high
598 #define i_reserved2	osd2.masix2.m_i_reserved2
599 
600 #endif /* defined(__KERNEL__) || defined(__linux__) */
601 
602 /*
603  * storage for cached extent
604  */
605 struct ext4_ext_cache {
606 	ext4_fsblk_t	ec_start;
607 	ext4_lblk_t	ec_block;
608 	__u32		ec_len; /* must be 32bit to return holes */
609 	__u32		ec_type;
610 };
611 
612 /*
613  * fourth extended file system inode data in memory
614  */
615 struct ext4_inode_info {
616 	__le32	i_data[15];	/* unconverted */
617 	__u32	i_flags;
618 	ext4_fsblk_t	i_file_acl;
619 	__u32	i_dtime;
620 
621 	/*
622 	 * i_block_group is the number of the block group which contains
623 	 * this file's inode.  Constant across the lifetime of the inode,
624 	 * it is ued for making block allocation decisions - we try to
625 	 * place a file's data blocks near its inode block, and new inodes
626 	 * near to their parent directory's inode.
627 	 */
628 	ext4_group_t	i_block_group;
629 	__u32	i_state;		/* Dynamic state flags for ext4 */
630 
631 	ext4_lblk_t		i_dir_start_lookup;
632 #ifdef CONFIG_EXT4_FS_XATTR
633 	/*
634 	 * Extended attributes can be read independently of the main file
635 	 * data. Taking i_mutex even when reading would cause contention
636 	 * between readers of EAs and writers of regular file data, so
637 	 * instead we synchronize on xattr_sem when reading or changing
638 	 * EAs.
639 	 */
640 	struct rw_semaphore xattr_sem;
641 #endif
642 
643 	struct list_head i_orphan;	/* unlinked but open inodes */
644 
645 	/*
646 	 * i_disksize keeps track of what the inode size is ON DISK, not
647 	 * in memory.  During truncate, i_size is set to the new size by
648 	 * the VFS prior to calling ext4_truncate(), but the filesystem won't
649 	 * set i_disksize to 0 until the truncate is actually under way.
650 	 *
651 	 * The intent is that i_disksize always represents the blocks which
652 	 * are used by this file.  This allows recovery to restart truncate
653 	 * on orphans if we crash during truncate.  We actually write i_disksize
654 	 * into the on-disk inode when writing inodes out, instead of i_size.
655 	 *
656 	 * The only time when i_disksize and i_size may be different is when
657 	 * a truncate is in progress.  The only things which change i_disksize
658 	 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
659 	 */
660 	loff_t	i_disksize;
661 
662 	/*
663 	 * i_data_sem is for serialising ext4_truncate() against
664 	 * ext4_getblock().  In the 2.4 ext2 design, great chunks of inode's
665 	 * data tree are chopped off during truncate. We can't do that in
666 	 * ext4 because whenever we perform intermediate commits during
667 	 * truncate, the inode and all the metadata blocks *must* be in a
668 	 * consistent state which allows truncation of the orphans to restart
669 	 * during recovery.  Hence we must fix the get_block-vs-truncate race
670 	 * by other means, so we have i_data_sem.
671 	 */
672 	struct rw_semaphore i_data_sem;
673 	struct inode vfs_inode;
674 	struct jbd2_inode jinode;
675 
676 	struct ext4_ext_cache i_cached_extent;
677 	/*
678 	 * File creation time. Its function is same as that of
679 	 * struct timespec i_{a,c,m}time in the generic inode.
680 	 */
681 	struct timespec i_crtime;
682 
683 	/* mballoc */
684 	struct list_head i_prealloc_list;
685 	spinlock_t i_prealloc_lock;
686 
687 	/* ialloc */
688 	ext4_group_t	i_last_alloc_group;
689 
690 	/* allocation reservation info for delalloc */
691 	unsigned int i_reserved_data_blocks;
692 	unsigned int i_reserved_meta_blocks;
693 	unsigned int i_allocated_meta_blocks;
694 	unsigned short i_delalloc_reserved_flag;
695 
696 	/* on-disk additional length */
697 	__u16 i_extra_isize;
698 
699 	spinlock_t i_block_reservation_lock;
700 
701 	/* completed async DIOs that might need unwritten extents handling */
702 	struct list_head i_aio_dio_complete_list;
703 	/* current io_end structure for async DIO write*/
704 	ext4_io_end_t *cur_aio_dio;
705 };
706 
707 /*
708  * File system states
709  */
710 #define	EXT4_VALID_FS			0x0001	/* Unmounted cleanly */
711 #define	EXT4_ERROR_FS			0x0002	/* Errors detected */
712 #define	EXT4_ORPHAN_FS			0x0004	/* Orphans being recovered */
713 
714 /*
715  * Misc. filesystem flags
716  */
717 #define EXT2_FLAGS_SIGNED_HASH		0x0001  /* Signed dirhash in use */
718 #define EXT2_FLAGS_UNSIGNED_HASH	0x0002  /* Unsigned dirhash in use */
719 #define EXT2_FLAGS_TEST_FILESYS		0x0004	/* to test development code */
720 
721 /*
722  * Mount flags
723  */
724 #define EXT4_MOUNT_OLDALLOC		0x00002  /* Don't use the new Orlov allocator */
725 #define EXT4_MOUNT_GRPID		0x00004	/* Create files with directory's group */
726 #define EXT4_MOUNT_DEBUG		0x00008	/* Some debugging messages */
727 #define EXT4_MOUNT_ERRORS_CONT		0x00010	/* Continue on errors */
728 #define EXT4_MOUNT_ERRORS_RO		0x00020	/* Remount fs ro on errors */
729 #define EXT4_MOUNT_ERRORS_PANIC		0x00040	/* Panic on errors */
730 #define EXT4_MOUNT_MINIX_DF		0x00080	/* Mimics the Minix statfs */
731 #define EXT4_MOUNT_NOLOAD		0x00100	/* Don't use existing journal*/
732 #define EXT4_MOUNT_DATA_FLAGS		0x00C00	/* Mode for data writes: */
733 #define EXT4_MOUNT_JOURNAL_DATA		0x00400	/* Write data to journal */
734 #define EXT4_MOUNT_ORDERED_DATA		0x00800	/* Flush data before commit */
735 #define EXT4_MOUNT_WRITEBACK_DATA	0x00C00	/* No data ordering */
736 #define EXT4_MOUNT_UPDATE_JOURNAL	0x01000	/* Update the journal format */
737 #define EXT4_MOUNT_NO_UID32		0x02000  /* Disable 32-bit UIDs */
738 #define EXT4_MOUNT_XATTR_USER		0x04000	/* Extended user attributes */
739 #define EXT4_MOUNT_POSIX_ACL		0x08000	/* POSIX Access Control Lists */
740 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC	0x10000	/* No auto delalloc mapping */
741 #define EXT4_MOUNT_BARRIER		0x20000 /* Use block barriers */
742 #define EXT4_MOUNT_NOBH			0x40000 /* No bufferheads */
743 #define EXT4_MOUNT_QUOTA		0x80000 /* Some quota option set */
744 #define EXT4_MOUNT_USRQUOTA		0x100000 /* "old" user quota */
745 #define EXT4_MOUNT_GRPQUOTA		0x200000 /* "old" group quota */
746 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT	0x1000000 /* Journal Async Commit */
747 #define EXT4_MOUNT_I_VERSION            0x2000000 /* i_version support */
748 #define EXT4_MOUNT_DELALLOC		0x8000000 /* Delalloc support */
749 #define EXT4_MOUNT_DATA_ERR_ABORT	0x10000000 /* Abort on file data write */
750 #define EXT4_MOUNT_BLOCK_VALIDITY	0x20000000 /* Block validity checking */
751 
752 #define clear_opt(o, opt)		o &= ~EXT4_MOUNT_##opt
753 #define set_opt(o, opt)			o |= EXT4_MOUNT_##opt
754 #define test_opt(sb, opt)		(EXT4_SB(sb)->s_mount_opt & \
755 					 EXT4_MOUNT_##opt)
756 
757 #define ext4_set_bit			ext2_set_bit
758 #define ext4_set_bit_atomic		ext2_set_bit_atomic
759 #define ext4_clear_bit			ext2_clear_bit
760 #define ext4_clear_bit_atomic		ext2_clear_bit_atomic
761 #define ext4_test_bit			ext2_test_bit
762 #define ext4_find_first_zero_bit	ext2_find_first_zero_bit
763 #define ext4_find_next_zero_bit		ext2_find_next_zero_bit
764 #define ext4_find_next_bit		ext2_find_next_bit
765 
766 /*
767  * Maximal mount counts between two filesystem checks
768  */
769 #define EXT4_DFL_MAX_MNT_COUNT		20	/* Allow 20 mounts */
770 #define EXT4_DFL_CHECKINTERVAL		0	/* Don't use interval check */
771 
772 /*
773  * Behaviour when detecting errors
774  */
775 #define EXT4_ERRORS_CONTINUE		1	/* Continue execution */
776 #define EXT4_ERRORS_RO			2	/* Remount fs read-only */
777 #define EXT4_ERRORS_PANIC		3	/* Panic */
778 #define EXT4_ERRORS_DEFAULT		EXT4_ERRORS_CONTINUE
779 
780 /*
781  * Structure of the super block
782  */
783 struct ext4_super_block {
784 /*00*/	__le32	s_inodes_count;		/* Inodes count */
785 	__le32	s_blocks_count_lo;	/* Blocks count */
786 	__le32	s_r_blocks_count_lo;	/* Reserved blocks count */
787 	__le32	s_free_blocks_count_lo;	/* Free blocks count */
788 /*10*/	__le32	s_free_inodes_count;	/* Free inodes count */
789 	__le32	s_first_data_block;	/* First Data Block */
790 	__le32	s_log_block_size;	/* Block size */
791 	__le32	s_obso_log_frag_size;	/* Obsoleted fragment size */
792 /*20*/	__le32	s_blocks_per_group;	/* # Blocks per group */
793 	__le32	s_obso_frags_per_group;	/* Obsoleted fragments per group */
794 	__le32	s_inodes_per_group;	/* # Inodes per group */
795 	__le32	s_mtime;		/* Mount time */
796 /*30*/	__le32	s_wtime;		/* Write time */
797 	__le16	s_mnt_count;		/* Mount count */
798 	__le16	s_max_mnt_count;	/* Maximal mount count */
799 	__le16	s_magic;		/* Magic signature */
800 	__le16	s_state;		/* File system state */
801 	__le16	s_errors;		/* Behaviour when detecting errors */
802 	__le16	s_minor_rev_level;	/* minor revision level */
803 /*40*/	__le32	s_lastcheck;		/* time of last check */
804 	__le32	s_checkinterval;	/* max. time between checks */
805 	__le32	s_creator_os;		/* OS */
806 	__le32	s_rev_level;		/* Revision level */
807 /*50*/	__le16	s_def_resuid;		/* Default uid for reserved blocks */
808 	__le16	s_def_resgid;		/* Default gid for reserved blocks */
809 	/*
810 	 * These fields are for EXT4_DYNAMIC_REV superblocks only.
811 	 *
812 	 * Note: the difference between the compatible feature set and
813 	 * the incompatible feature set is that if there is a bit set
814 	 * in the incompatible feature set that the kernel doesn't
815 	 * know about, it should refuse to mount the filesystem.
816 	 *
817 	 * e2fsck's requirements are more strict; if it doesn't know
818 	 * about a feature in either the compatible or incompatible
819 	 * feature set, it must abort and not try to meddle with
820 	 * things it doesn't understand...
821 	 */
822 	__le32	s_first_ino;		/* First non-reserved inode */
823 	__le16  s_inode_size;		/* size of inode structure */
824 	__le16	s_block_group_nr;	/* block group # of this superblock */
825 	__le32	s_feature_compat;	/* compatible feature set */
826 /*60*/	__le32	s_feature_incompat;	/* incompatible feature set */
827 	__le32	s_feature_ro_compat;	/* readonly-compatible feature set */
828 /*68*/	__u8	s_uuid[16];		/* 128-bit uuid for volume */
829 /*78*/	char	s_volume_name[16];	/* volume name */
830 /*88*/	char	s_last_mounted[64];	/* directory where last mounted */
831 /*C8*/	__le32	s_algorithm_usage_bitmap; /* For compression */
832 	/*
833 	 * Performance hints.  Directory preallocation should only
834 	 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
835 	 */
836 	__u8	s_prealloc_blocks;	/* Nr of blocks to try to preallocate*/
837 	__u8	s_prealloc_dir_blocks;	/* Nr to preallocate for dirs */
838 	__le16	s_reserved_gdt_blocks;	/* Per group desc for online growth */
839 	/*
840 	 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
841 	 */
842 /*D0*/	__u8	s_journal_uuid[16];	/* uuid of journal superblock */
843 /*E0*/	__le32	s_journal_inum;		/* inode number of journal file */
844 	__le32	s_journal_dev;		/* device number of journal file */
845 	__le32	s_last_orphan;		/* start of list of inodes to delete */
846 	__le32	s_hash_seed[4];		/* HTREE hash seed */
847 	__u8	s_def_hash_version;	/* Default hash version to use */
848 	__u8	s_reserved_char_pad;
849 	__le16  s_desc_size;		/* size of group descriptor */
850 /*100*/	__le32	s_default_mount_opts;
851 	__le32	s_first_meta_bg;	/* First metablock block group */
852 	__le32	s_mkfs_time;		/* When the filesystem was created */
853 	__le32	s_jnl_blocks[17];	/* Backup of the journal inode */
854 	/* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
855 /*150*/	__le32	s_blocks_count_hi;	/* Blocks count */
856 	__le32	s_r_blocks_count_hi;	/* Reserved blocks count */
857 	__le32	s_free_blocks_count_hi;	/* Free blocks count */
858 	__le16	s_min_extra_isize;	/* All inodes have at least # bytes */
859 	__le16	s_want_extra_isize; 	/* New inodes should reserve # bytes */
860 	__le32	s_flags;		/* Miscellaneous flags */
861 	__le16  s_raid_stride;		/* RAID stride */
862 	__le16  s_mmp_interval;         /* # seconds to wait in MMP checking */
863 	__le64  s_mmp_block;            /* Block for multi-mount protection */
864 	__le32  s_raid_stripe_width;    /* blocks on all data disks (N*stride)*/
865 	__u8	s_log_groups_per_flex;  /* FLEX_BG group size */
866 	__u8	s_reserved_char_pad2;
867 	__le16  s_reserved_pad;
868 	__le64	s_kbytes_written;	/* nr of lifetime kilobytes written */
869 	__u32   s_reserved[160];        /* Padding to the end of the block */
870 };
871 
872 #ifdef __KERNEL__
873 
874 /*
875  * run-time mount flags
876  */
877 #define EXT4_MF_MNTDIR_SAMPLED	0x0001
878 #define EXT4_MF_FS_ABORTED	0x0002	/* Fatal error detected */
879 
880 /*
881  * fourth extended-fs super-block data in memory
882  */
883 struct ext4_sb_info {
884 	unsigned long s_desc_size;	/* Size of a group descriptor in bytes */
885 	unsigned long s_inodes_per_block;/* Number of inodes per block */
886 	unsigned long s_blocks_per_group;/* Number of blocks in a group */
887 	unsigned long s_inodes_per_group;/* Number of inodes in a group */
888 	unsigned long s_itb_per_group;	/* Number of inode table blocks per group */
889 	unsigned long s_gdb_count;	/* Number of group descriptor blocks */
890 	unsigned long s_desc_per_block;	/* Number of group descriptors per block */
891 	ext4_group_t s_groups_count;	/* Number of groups in the fs */
892 	ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
893 	unsigned long s_overhead_last;  /* Last calculated overhead */
894 	unsigned long s_blocks_last;    /* Last seen block count */
895 	loff_t s_bitmap_maxbytes;	/* max bytes for bitmap files */
896 	struct buffer_head * s_sbh;	/* Buffer containing the super block */
897 	struct ext4_super_block *s_es;	/* Pointer to the super block in the buffer */
898 	struct buffer_head **s_group_desc;
899 	unsigned int s_mount_opt;
900 	unsigned int s_mount_flags;
901 	ext4_fsblk_t s_sb_block;
902 	uid_t s_resuid;
903 	gid_t s_resgid;
904 	unsigned short s_mount_state;
905 	unsigned short s_pad;
906 	int s_addr_per_block_bits;
907 	int s_desc_per_block_bits;
908 	int s_inode_size;
909 	int s_first_ino;
910 	unsigned int s_inode_readahead_blks;
911 	unsigned int s_inode_goal;
912 	spinlock_t s_next_gen_lock;
913 	u32 s_next_generation;
914 	u32 s_hash_seed[4];
915 	int s_def_hash_version;
916 	int s_hash_unsigned;	/* 3 if hash should be signed, 0 if not */
917 	struct percpu_counter s_freeblocks_counter;
918 	struct percpu_counter s_freeinodes_counter;
919 	struct percpu_counter s_dirs_counter;
920 	struct percpu_counter s_dirtyblocks_counter;
921 	struct blockgroup_lock *s_blockgroup_lock;
922 	struct proc_dir_entry *s_proc;
923 	struct kobject s_kobj;
924 	struct completion s_kobj_unregister;
925 
926 	/* Journaling */
927 	struct inode *s_journal_inode;
928 	struct journal_s *s_journal;
929 	struct list_head s_orphan;
930 	struct mutex s_orphan_lock;
931 	struct mutex s_resize_lock;
932 	unsigned long s_commit_interval;
933 	u32 s_max_batch_time;
934 	u32 s_min_batch_time;
935 	struct block_device *journal_bdev;
936 #ifdef CONFIG_JBD2_DEBUG
937 	struct timer_list turn_ro_timer;	/* For turning read-only (crash simulation) */
938 	wait_queue_head_t ro_wait_queue;	/* For people waiting for the fs to go read-only */
939 #endif
940 #ifdef CONFIG_QUOTA
941 	char *s_qf_names[MAXQUOTAS];		/* Names of quota files with journalled quota */
942 	int s_jquota_fmt;			/* Format of quota to use */
943 #endif
944 	unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
945 	struct rb_root system_blks;
946 
947 #ifdef EXTENTS_STATS
948 	/* ext4 extents stats */
949 	unsigned long s_ext_min;
950 	unsigned long s_ext_max;
951 	unsigned long s_depth_max;
952 	spinlock_t s_ext_stats_lock;
953 	unsigned long s_ext_blocks;
954 	unsigned long s_ext_extents;
955 #endif
956 
957 	/* for buddy allocator */
958 	struct ext4_group_info ***s_group_info;
959 	struct inode *s_buddy_cache;
960 	long s_blocks_reserved;
961 	spinlock_t s_reserve_lock;
962 	spinlock_t s_md_lock;
963 	tid_t s_last_transaction;
964 	unsigned short *s_mb_offsets;
965 	unsigned int *s_mb_maxs;
966 
967 	/* tunables */
968 	unsigned long s_stripe;
969 	unsigned int s_mb_stream_request;
970 	unsigned int s_mb_max_to_scan;
971 	unsigned int s_mb_min_to_scan;
972 	unsigned int s_mb_stats;
973 	unsigned int s_mb_order2_reqs;
974 	unsigned int s_mb_group_prealloc;
975 	unsigned int s_max_writeback_mb_bump;
976 	/* where last allocation was done - for stream allocation */
977 	unsigned long s_mb_last_group;
978 	unsigned long s_mb_last_start;
979 
980 	/* stats for buddy allocator */
981 	spinlock_t s_mb_pa_lock;
982 	atomic_t s_bal_reqs;	/* number of reqs with len > 1 */
983 	atomic_t s_bal_success;	/* we found long enough chunks */
984 	atomic_t s_bal_allocated;	/* in blocks */
985 	atomic_t s_bal_ex_scanned;	/* total extents scanned */
986 	atomic_t s_bal_goals;	/* goal hits */
987 	atomic_t s_bal_breaks;	/* too long searches */
988 	atomic_t s_bal_2orders;	/* 2^order hits */
989 	spinlock_t s_bal_lock;
990 	unsigned long s_mb_buddies_generated;
991 	unsigned long long s_mb_generation_time;
992 	atomic_t s_mb_lost_chunks;
993 	atomic_t s_mb_preallocated;
994 	atomic_t s_mb_discarded;
995 	atomic_t s_lock_busy;
996 
997 	/* locality groups */
998 	struct ext4_locality_group *s_locality_groups;
999 
1000 	/* for write statistics */
1001 	unsigned long s_sectors_written_start;
1002 	u64 s_kbytes_written;
1003 
1004 	unsigned int s_log_groups_per_flex;
1005 	struct flex_groups *s_flex_groups;
1006 
1007 	/* workqueue for dio unwritten */
1008 	struct workqueue_struct *dio_unwritten_wq;
1009 };
1010 
1011 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1012 {
1013 	return sb->s_fs_info;
1014 }
1015 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1016 {
1017 	return container_of(inode, struct ext4_inode_info, vfs_inode);
1018 }
1019 
1020 static inline struct timespec ext4_current_time(struct inode *inode)
1021 {
1022 	return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1023 		current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1024 }
1025 
1026 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1027 {
1028 	return ino == EXT4_ROOT_INO ||
1029 		ino == EXT4_JOURNAL_INO ||
1030 		ino == EXT4_RESIZE_INO ||
1031 		(ino >= EXT4_FIRST_INO(sb) &&
1032 		 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1033 }
1034 #else
1035 /* Assume that user mode programs are passing in an ext4fs superblock, not
1036  * a kernel struct super_block.  This will allow us to call the feature-test
1037  * macros from user land. */
1038 #define EXT4_SB(sb)	(sb)
1039 #endif
1040 
1041 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1042 
1043 /*
1044  * Codes for operating systems
1045  */
1046 #define EXT4_OS_LINUX		0
1047 #define EXT4_OS_HURD		1
1048 #define EXT4_OS_MASIX		2
1049 #define EXT4_OS_FREEBSD		3
1050 #define EXT4_OS_LITES		4
1051 
1052 /*
1053  * Revision levels
1054  */
1055 #define EXT4_GOOD_OLD_REV	0	/* The good old (original) format */
1056 #define EXT4_DYNAMIC_REV	1	/* V2 format w/ dynamic inode sizes */
1057 
1058 #define EXT4_CURRENT_REV	EXT4_GOOD_OLD_REV
1059 #define EXT4_MAX_SUPP_REV	EXT4_DYNAMIC_REV
1060 
1061 #define EXT4_GOOD_OLD_INODE_SIZE 128
1062 
1063 /*
1064  * Feature set definitions
1065  */
1066 
1067 #define EXT4_HAS_COMPAT_FEATURE(sb,mask)			\
1068 	((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1069 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask)			\
1070 	((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1071 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask)			\
1072 	((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1073 #define EXT4_SET_COMPAT_FEATURE(sb,mask)			\
1074 	EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1075 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask)			\
1076 	EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1077 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask)			\
1078 	EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1079 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask)			\
1080 	EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1081 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask)			\
1082 	EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1083 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask)			\
1084 	EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1085 
1086 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC	0x0001
1087 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES	0x0002
1088 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL		0x0004
1089 #define EXT4_FEATURE_COMPAT_EXT_ATTR		0x0008
1090 #define EXT4_FEATURE_COMPAT_RESIZE_INODE	0x0010
1091 #define EXT4_FEATURE_COMPAT_DIR_INDEX		0x0020
1092 
1093 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER	0x0001
1094 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE	0x0002
1095 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR	0x0004
1096 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE        0x0008
1097 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM		0x0010
1098 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK	0x0020
1099 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE	0x0040
1100 
1101 #define EXT4_FEATURE_INCOMPAT_COMPRESSION	0x0001
1102 #define EXT4_FEATURE_INCOMPAT_FILETYPE		0x0002
1103 #define EXT4_FEATURE_INCOMPAT_RECOVER		0x0004 /* Needs recovery */
1104 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV	0x0008 /* Journal device */
1105 #define EXT4_FEATURE_INCOMPAT_META_BG		0x0010
1106 #define EXT4_FEATURE_INCOMPAT_EXTENTS		0x0040 /* extents support */
1107 #define EXT4_FEATURE_INCOMPAT_64BIT		0x0080
1108 #define EXT4_FEATURE_INCOMPAT_MMP               0x0100
1109 #define EXT4_FEATURE_INCOMPAT_FLEX_BG		0x0200
1110 
1111 #define EXT4_FEATURE_COMPAT_SUPP	EXT2_FEATURE_COMPAT_EXT_ATTR
1112 #define EXT4_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
1113 					 EXT4_FEATURE_INCOMPAT_RECOVER| \
1114 					 EXT4_FEATURE_INCOMPAT_META_BG| \
1115 					 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1116 					 EXT4_FEATURE_INCOMPAT_64BIT| \
1117 					 EXT4_FEATURE_INCOMPAT_FLEX_BG)
1118 #define EXT4_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1119 					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1120 					 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1121 					 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1122 					 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1123 					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1124 					 EXT4_FEATURE_RO_COMPAT_HUGE_FILE)
1125 
1126 /*
1127  * Default values for user and/or group using reserved blocks
1128  */
1129 #define	EXT4_DEF_RESUID		0
1130 #define	EXT4_DEF_RESGID		0
1131 
1132 #define EXT4_DEF_INODE_READAHEAD_BLKS	32
1133 
1134 /*
1135  * Default mount options
1136  */
1137 #define EXT4_DEFM_DEBUG		0x0001
1138 #define EXT4_DEFM_BSDGROUPS	0x0002
1139 #define EXT4_DEFM_XATTR_USER	0x0004
1140 #define EXT4_DEFM_ACL		0x0008
1141 #define EXT4_DEFM_UID16		0x0010
1142 #define EXT4_DEFM_JMODE		0x0060
1143 #define EXT4_DEFM_JMODE_DATA	0x0020
1144 #define EXT4_DEFM_JMODE_ORDERED	0x0040
1145 #define EXT4_DEFM_JMODE_WBACK	0x0060
1146 
1147 /*
1148  * Default journal batch times
1149  */
1150 #define EXT4_DEF_MIN_BATCH_TIME	0
1151 #define EXT4_DEF_MAX_BATCH_TIME	15000 /* 15ms */
1152 
1153 /*
1154  * Minimum number of groups in a flexgroup before we separate out
1155  * directories into the first block group of a flexgroup
1156  */
1157 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME	4
1158 
1159 /*
1160  * Structure of a directory entry
1161  */
1162 #define EXT4_NAME_LEN 255
1163 
1164 struct ext4_dir_entry {
1165 	__le32	inode;			/* Inode number */
1166 	__le16	rec_len;		/* Directory entry length */
1167 	__le16	name_len;		/* Name length */
1168 	char	name[EXT4_NAME_LEN];	/* File name */
1169 };
1170 
1171 /*
1172  * The new version of the directory entry.  Since EXT4 structures are
1173  * stored in intel byte order, and the name_len field could never be
1174  * bigger than 255 chars, it's safe to reclaim the extra byte for the
1175  * file_type field.
1176  */
1177 struct ext4_dir_entry_2 {
1178 	__le32	inode;			/* Inode number */
1179 	__le16	rec_len;		/* Directory entry length */
1180 	__u8	name_len;		/* Name length */
1181 	__u8	file_type;
1182 	char	name[EXT4_NAME_LEN];	/* File name */
1183 };
1184 
1185 /*
1186  * Ext4 directory file types.  Only the low 3 bits are used.  The
1187  * other bits are reserved for now.
1188  */
1189 #define EXT4_FT_UNKNOWN		0
1190 #define EXT4_FT_REG_FILE	1
1191 #define EXT4_FT_DIR		2
1192 #define EXT4_FT_CHRDEV		3
1193 #define EXT4_FT_BLKDEV		4
1194 #define EXT4_FT_FIFO		5
1195 #define EXT4_FT_SOCK		6
1196 #define EXT4_FT_SYMLINK		7
1197 
1198 #define EXT4_FT_MAX		8
1199 
1200 /*
1201  * EXT4_DIR_PAD defines the directory entries boundaries
1202  *
1203  * NOTE: It must be a multiple of 4
1204  */
1205 #define EXT4_DIR_PAD			4
1206 #define EXT4_DIR_ROUND			(EXT4_DIR_PAD - 1)
1207 #define EXT4_DIR_REC_LEN(name_len)	(((name_len) + 8 + EXT4_DIR_ROUND) & \
1208 					 ~EXT4_DIR_ROUND)
1209 #define EXT4_MAX_REC_LEN		((1<<16)-1)
1210 
1211 /*
1212  * Hash Tree Directory indexing
1213  * (c) Daniel Phillips, 2001
1214  */
1215 
1216 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1217 				      EXT4_FEATURE_COMPAT_DIR_INDEX) && \
1218 		      (EXT4_I(dir)->i_flags & EXT4_INDEX_FL))
1219 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1220 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1221 
1222 /* Legal values for the dx_root hash_version field: */
1223 
1224 #define DX_HASH_LEGACY		0
1225 #define DX_HASH_HALF_MD4	1
1226 #define DX_HASH_TEA		2
1227 #define DX_HASH_LEGACY_UNSIGNED	3
1228 #define DX_HASH_HALF_MD4_UNSIGNED	4
1229 #define DX_HASH_TEA_UNSIGNED		5
1230 
1231 #ifdef __KERNEL__
1232 
1233 /* hash info structure used by the directory hash */
1234 struct dx_hash_info
1235 {
1236 	u32		hash;
1237 	u32		minor_hash;
1238 	int		hash_version;
1239 	u32		*seed;
1240 };
1241 
1242 #define EXT4_HTREE_EOF	0x7fffffff
1243 
1244 /*
1245  * Control parameters used by ext4_htree_next_block
1246  */
1247 #define HASH_NB_ALWAYS		1
1248 
1249 
1250 /*
1251  * Describe an inode's exact location on disk and in memory
1252  */
1253 struct ext4_iloc
1254 {
1255 	struct buffer_head *bh;
1256 	unsigned long offset;
1257 	ext4_group_t block_group;
1258 };
1259 
1260 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
1261 {
1262 	return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
1263 }
1264 
1265 /*
1266  * This structure is stuffed into the struct file's private_data field
1267  * for directories.  It is where we put information so that we can do
1268  * readdir operations in hash tree order.
1269  */
1270 struct dir_private_info {
1271 	struct rb_root	root;
1272 	struct rb_node	*curr_node;
1273 	struct fname	*extra_fname;
1274 	loff_t		last_pos;
1275 	__u32		curr_hash;
1276 	__u32		curr_minor_hash;
1277 	__u32		next_hash;
1278 };
1279 
1280 /* calculate the first block number of the group */
1281 static inline ext4_fsblk_t
1282 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
1283 {
1284 	return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
1285 		le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
1286 }
1287 
1288 /*
1289  * Special error return code only used by dx_probe() and its callers.
1290  */
1291 #define ERR_BAD_DX_DIR	-75000
1292 
1293 void ext4_get_group_no_and_offset(struct super_block *sb, ext4_fsblk_t blocknr,
1294 			ext4_group_t *blockgrpp, ext4_grpblk_t *offsetp);
1295 
1296 extern struct proc_dir_entry *ext4_proc_root;
1297 
1298 /*
1299  * Function prototypes
1300  */
1301 
1302 /*
1303  * Ok, these declarations are also in <linux/kernel.h> but none of the
1304  * ext4 source programs needs to include it so they are duplicated here.
1305  */
1306 # define NORET_TYPE	/**/
1307 # define ATTRIB_NORET	__attribute__((noreturn))
1308 # define NORET_AND	noreturn,
1309 
1310 /* bitmap.c */
1311 extern unsigned int ext4_count_free(struct buffer_head *, unsigned);
1312 
1313 /* balloc.c */
1314 extern unsigned int ext4_block_group(struct super_block *sb,
1315 			ext4_fsblk_t blocknr);
1316 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
1317 			ext4_fsblk_t blocknr);
1318 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
1319 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
1320 			ext4_group_t group);
1321 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
1322 			ext4_fsblk_t goal, unsigned long *count, int *errp);
1323 extern int ext4_claim_free_blocks(struct ext4_sb_info *sbi, s64 nblocks);
1324 extern int ext4_has_free_blocks(struct ext4_sb_info *sbi, s64 nblocks);
1325 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
1326 			ext4_fsblk_t block, unsigned long count, int metadata);
1327 extern void ext4_add_groupblocks(handle_t *handle, struct super_block *sb,
1328 				ext4_fsblk_t block, unsigned long count);
1329 extern ext4_fsblk_t ext4_count_free_blocks(struct super_block *);
1330 extern void ext4_check_blocks_bitmap(struct super_block *);
1331 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
1332 						    ext4_group_t block_group,
1333 						    struct buffer_head ** bh);
1334 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
1335 struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
1336 				      ext4_group_t block_group);
1337 extern unsigned ext4_init_block_bitmap(struct super_block *sb,
1338 				       struct buffer_head *bh,
1339 				       ext4_group_t group,
1340 				       struct ext4_group_desc *desc);
1341 #define ext4_free_blocks_after_init(sb, group, desc)			\
1342 		ext4_init_block_bitmap(sb, NULL, group, desc)
1343 
1344 /* dir.c */
1345 extern int ext4_check_dir_entry(const char *, struct inode *,
1346 				struct ext4_dir_entry_2 *,
1347 				struct buffer_head *, unsigned int);
1348 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
1349 				    __u32 minor_hash,
1350 				    struct ext4_dir_entry_2 *dirent);
1351 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
1352 
1353 /* fsync.c */
1354 extern int ext4_sync_file(struct file *, struct dentry *, int);
1355 
1356 /* hash.c */
1357 extern int ext4fs_dirhash(const char *name, int len, struct
1358 			  dx_hash_info *hinfo);
1359 
1360 /* ialloc.c */
1361 extern struct inode *ext4_new_inode(handle_t *, struct inode *, int,
1362 				    const struct qstr *qstr, __u32 goal);
1363 extern void ext4_free_inode(handle_t *, struct inode *);
1364 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
1365 extern unsigned long ext4_count_free_inodes(struct super_block *);
1366 extern unsigned long ext4_count_dirs(struct super_block *);
1367 extern void ext4_check_inodes_bitmap(struct super_block *);
1368 extern unsigned ext4_init_inode_bitmap(struct super_block *sb,
1369 				       struct buffer_head *bh,
1370 				       ext4_group_t group,
1371 				       struct ext4_group_desc *desc);
1372 extern void mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
1373 
1374 /* mballoc.c */
1375 extern long ext4_mb_stats;
1376 extern long ext4_mb_max_to_scan;
1377 extern int ext4_mb_init(struct super_block *, int);
1378 extern int ext4_mb_release(struct super_block *);
1379 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
1380 				struct ext4_allocation_request *, int *);
1381 extern int ext4_mb_reserve_blocks(struct super_block *, int);
1382 extern void ext4_discard_preallocations(struct inode *);
1383 extern int __init init_ext4_mballoc(void);
1384 extern void exit_ext4_mballoc(void);
1385 extern void ext4_mb_free_blocks(handle_t *, struct inode *,
1386 		ext4_fsblk_t, unsigned long, int, unsigned long *);
1387 extern int ext4_mb_add_groupinfo(struct super_block *sb,
1388 		ext4_group_t i, struct ext4_group_desc *desc);
1389 extern int ext4_mb_get_buddy_cache_lock(struct super_block *, ext4_group_t);
1390 extern void ext4_mb_put_buddy_cache_lock(struct super_block *,
1391 						ext4_group_t, int);
1392 /* inode.c */
1393 int ext4_forget(handle_t *handle, int is_metadata, struct inode *inode,
1394 		struct buffer_head *bh, ext4_fsblk_t blocknr);
1395 struct buffer_head *ext4_getblk(handle_t *, struct inode *,
1396 						ext4_lblk_t, int, int *);
1397 struct buffer_head *ext4_bread(handle_t *, struct inode *,
1398 						ext4_lblk_t, int, int *);
1399 int ext4_get_block(struct inode *inode, sector_t iblock,
1400 				struct buffer_head *bh_result, int create);
1401 
1402 extern struct inode *ext4_iget(struct super_block *, unsigned long);
1403 extern int  ext4_write_inode(struct inode *, int);
1404 extern int  ext4_setattr(struct dentry *, struct iattr *);
1405 extern int  ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
1406 				struct kstat *stat);
1407 extern void ext4_delete_inode(struct inode *);
1408 extern int  ext4_sync_inode(handle_t *, struct inode *);
1409 extern void ext4_dirty_inode(struct inode *);
1410 extern int ext4_change_inode_journal_flag(struct inode *, int);
1411 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
1412 extern int ext4_can_truncate(struct inode *inode);
1413 extern void ext4_truncate(struct inode *);
1414 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
1415 extern void ext4_set_inode_flags(struct inode *);
1416 extern void ext4_get_inode_flags(struct ext4_inode_info *);
1417 extern int ext4_alloc_da_blocks(struct inode *inode);
1418 extern void ext4_set_aops(struct inode *inode);
1419 extern int ext4_writepage_trans_blocks(struct inode *);
1420 extern int ext4_meta_trans_blocks(struct inode *, int nrblocks, int idxblocks);
1421 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
1422 extern int ext4_block_truncate_page(handle_t *handle,
1423 		struct address_space *mapping, loff_t from);
1424 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
1425 extern qsize_t ext4_get_reserved_space(struct inode *inode);
1426 extern int flush_aio_dio_completed_IO(struct inode *inode);
1427 /* ioctl.c */
1428 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
1429 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
1430 
1431 /* migrate.c */
1432 extern int ext4_ext_migrate(struct inode *);
1433 
1434 /* namei.c */
1435 extern unsigned int ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize);
1436 extern __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize);
1437 extern int ext4_orphan_add(handle_t *, struct inode *);
1438 extern int ext4_orphan_del(handle_t *, struct inode *);
1439 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
1440 				__u32 start_minor_hash, __u32 *next_hash);
1441 
1442 /* resize.c */
1443 extern int ext4_group_add(struct super_block *sb,
1444 				struct ext4_new_group_data *input);
1445 extern int ext4_group_extend(struct super_block *sb,
1446 				struct ext4_super_block *es,
1447 				ext4_fsblk_t n_blocks_count);
1448 
1449 /* super.c */
1450 extern void ext4_error(struct super_block *, const char *, const char *, ...)
1451 	__attribute__ ((format (printf, 3, 4)));
1452 extern void __ext4_std_error(struct super_block *, const char *, int);
1453 extern void ext4_abort(struct super_block *, const char *, const char *, ...)
1454 	__attribute__ ((format (printf, 3, 4)));
1455 extern void ext4_warning(struct super_block *, const char *, const char *, ...)
1456 	__attribute__ ((format (printf, 3, 4)));
1457 extern void ext4_msg(struct super_block *, const char *, const char *, ...)
1458 	__attribute__ ((format (printf, 3, 4)));
1459 extern void ext4_grp_locked_error(struct super_block *, ext4_group_t,
1460 				const char *, const char *, ...)
1461 	__attribute__ ((format (printf, 4, 5)));
1462 extern void ext4_update_dynamic_rev(struct super_block *sb);
1463 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
1464 					__u32 compat);
1465 extern int ext4_update_rocompat_feature(handle_t *handle,
1466 					struct super_block *sb,	__u32 rocompat);
1467 extern int ext4_update_incompat_feature(handle_t *handle,
1468 					struct super_block *sb,	__u32 incompat);
1469 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
1470 				      struct ext4_group_desc *bg);
1471 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
1472 				      struct ext4_group_desc *bg);
1473 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
1474 				     struct ext4_group_desc *bg);
1475 extern __u32 ext4_free_blks_count(struct super_block *sb,
1476 				struct ext4_group_desc *bg);
1477 extern __u32 ext4_free_inodes_count(struct super_block *sb,
1478 				 struct ext4_group_desc *bg);
1479 extern __u32 ext4_used_dirs_count(struct super_block *sb,
1480 				struct ext4_group_desc *bg);
1481 extern __u32 ext4_itable_unused_count(struct super_block *sb,
1482 				   struct ext4_group_desc *bg);
1483 extern void ext4_block_bitmap_set(struct super_block *sb,
1484 				  struct ext4_group_desc *bg, ext4_fsblk_t blk);
1485 extern void ext4_inode_bitmap_set(struct super_block *sb,
1486 				  struct ext4_group_desc *bg, ext4_fsblk_t blk);
1487 extern void ext4_inode_table_set(struct super_block *sb,
1488 				 struct ext4_group_desc *bg, ext4_fsblk_t blk);
1489 extern void ext4_free_blks_set(struct super_block *sb,
1490 			       struct ext4_group_desc *bg, __u32 count);
1491 extern void ext4_free_inodes_set(struct super_block *sb,
1492 				struct ext4_group_desc *bg, __u32 count);
1493 extern void ext4_used_dirs_set(struct super_block *sb,
1494 				struct ext4_group_desc *bg, __u32 count);
1495 extern void ext4_itable_unused_set(struct super_block *sb,
1496 				   struct ext4_group_desc *bg, __u32 count);
1497 extern __le16 ext4_group_desc_csum(struct ext4_sb_info *sbi, __u32 group,
1498 				   struct ext4_group_desc *gdp);
1499 extern int ext4_group_desc_csum_verify(struct ext4_sb_info *sbi, __u32 group,
1500 				       struct ext4_group_desc *gdp);
1501 
1502 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
1503 {
1504 	return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
1505 		le32_to_cpu(es->s_blocks_count_lo);
1506 }
1507 
1508 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
1509 {
1510 	return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
1511 		le32_to_cpu(es->s_r_blocks_count_lo);
1512 }
1513 
1514 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
1515 {
1516 	return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
1517 		le32_to_cpu(es->s_free_blocks_count_lo);
1518 }
1519 
1520 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
1521 					 ext4_fsblk_t blk)
1522 {
1523 	es->s_blocks_count_lo = cpu_to_le32((u32)blk);
1524 	es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
1525 }
1526 
1527 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
1528 					      ext4_fsblk_t blk)
1529 {
1530 	es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
1531 	es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
1532 }
1533 
1534 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
1535 					   ext4_fsblk_t blk)
1536 {
1537 	es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
1538 	es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
1539 }
1540 
1541 static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
1542 {
1543 	if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
1544 		return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
1545 			le32_to_cpu(raw_inode->i_size_lo);
1546 	else
1547 		return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
1548 }
1549 
1550 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
1551 {
1552 	raw_inode->i_size_lo = cpu_to_le32(i_size);
1553 	raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
1554 }
1555 
1556 static inline
1557 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
1558 					    ext4_group_t group)
1559 {
1560 	 struct ext4_group_info ***grp_info;
1561 	 long indexv, indexh;
1562 	 grp_info = EXT4_SB(sb)->s_group_info;
1563 	 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
1564 	 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
1565 	 return grp_info[indexv][indexh];
1566 }
1567 
1568 /*
1569  * Reading s_groups_count requires using smp_rmb() afterwards.  See
1570  * the locking protocol documented in the comments of ext4_group_add()
1571  * in resize.c
1572  */
1573 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
1574 {
1575 	ext4_group_t	ngroups = EXT4_SB(sb)->s_groups_count;
1576 
1577 	smp_rmb();
1578 	return ngroups;
1579 }
1580 
1581 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
1582 					     ext4_group_t block_group)
1583 {
1584 	return block_group >> sbi->s_log_groups_per_flex;
1585 }
1586 
1587 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
1588 {
1589 	return 1 << sbi->s_log_groups_per_flex;
1590 }
1591 
1592 #define ext4_std_error(sb, errno)				\
1593 do {								\
1594 	if ((errno))						\
1595 		__ext4_std_error((sb), __func__, (errno));	\
1596 } while (0)
1597 
1598 #ifdef CONFIG_SMP
1599 /* Each CPU can accumulate percpu_counter_batch blocks in their local
1600  * counters. So we need to make sure we have free blocks more
1601  * than percpu_counter_batch  * nr_cpu_ids. Also add a window of 4 times.
1602  */
1603 #define EXT4_FREEBLOCKS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
1604 #else
1605 #define EXT4_FREEBLOCKS_WATERMARK 0
1606 #endif
1607 
1608 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
1609 {
1610 	/*
1611 	 * XXX: replace with spinlock if seen contended -bzzz
1612 	 */
1613 	down_write(&EXT4_I(inode)->i_data_sem);
1614 	if (newsize > EXT4_I(inode)->i_disksize)
1615 		EXT4_I(inode)->i_disksize = newsize;
1616 	up_write(&EXT4_I(inode)->i_data_sem);
1617 	return ;
1618 }
1619 
1620 struct ext4_group_info {
1621 	unsigned long   bb_state;
1622 	struct rb_root  bb_free_root;
1623 	ext4_grpblk_t	bb_first_free;	/* first free block */
1624 	ext4_grpblk_t	bb_free;	/* total free blocks */
1625 	ext4_grpblk_t	bb_fragments;	/* nr of freespace fragments */
1626 	struct          list_head bb_prealloc_list;
1627 #ifdef DOUBLE_CHECK
1628 	void            *bb_bitmap;
1629 #endif
1630 	struct rw_semaphore alloc_sem;
1631 	ext4_grpblk_t	bb_counters[];	/* Nr of free power-of-two-block
1632 					 * regions, index is order.
1633 					 * bb_counters[3] = 5 means
1634 					 * 5 free 8-block regions. */
1635 };
1636 
1637 #define EXT4_GROUP_INFO_NEED_INIT_BIT	0
1638 
1639 #define EXT4_MB_GRP_NEED_INIT(grp)	\
1640 	(test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
1641 
1642 #define EXT4_MAX_CONTENTION		8
1643 #define EXT4_CONTENTION_THRESHOLD	2
1644 
1645 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
1646 					      ext4_group_t group)
1647 {
1648 	return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
1649 }
1650 
1651 /*
1652  * Returns true if the filesystem is busy enough that attempts to
1653  * access the block group locks has run into contention.
1654  */
1655 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
1656 {
1657 	return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
1658 }
1659 
1660 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
1661 {
1662 	spinlock_t *lock = ext4_group_lock_ptr(sb, group);
1663 	if (spin_trylock(lock))
1664 		/*
1665 		 * We're able to grab the lock right away, so drop the
1666 		 * lock contention counter.
1667 		 */
1668 		atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
1669 	else {
1670 		/*
1671 		 * The lock is busy, so bump the contention counter,
1672 		 * and then wait on the spin lock.
1673 		 */
1674 		atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
1675 				  EXT4_MAX_CONTENTION);
1676 		spin_lock(lock);
1677 	}
1678 }
1679 
1680 static inline void ext4_unlock_group(struct super_block *sb,
1681 					ext4_group_t group)
1682 {
1683 	spin_unlock(ext4_group_lock_ptr(sb, group));
1684 }
1685 
1686 /*
1687  * Inodes and files operations
1688  */
1689 
1690 /* dir.c */
1691 extern const struct file_operations ext4_dir_operations;
1692 
1693 /* file.c */
1694 extern const struct inode_operations ext4_file_inode_operations;
1695 extern const struct file_operations ext4_file_operations;
1696 
1697 /* namei.c */
1698 extern const struct inode_operations ext4_dir_inode_operations;
1699 extern const struct inode_operations ext4_special_inode_operations;
1700 extern struct dentry *ext4_get_parent(struct dentry *child);
1701 
1702 /* symlink.c */
1703 extern const struct inode_operations ext4_symlink_inode_operations;
1704 extern const struct inode_operations ext4_fast_symlink_inode_operations;
1705 
1706 /* block_validity */
1707 extern void ext4_release_system_zone(struct super_block *sb);
1708 extern int ext4_setup_system_zone(struct super_block *sb);
1709 extern int __init init_ext4_system_zone(void);
1710 extern void exit_ext4_system_zone(void);
1711 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
1712 				 ext4_fsblk_t start_blk,
1713 				 unsigned int count);
1714 
1715 /* extents.c */
1716 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
1717 extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
1718 extern int ext4_ext_index_trans_blocks(struct inode *inode, int nrblocks,
1719 				       int chunk);
1720 extern int ext4_ext_get_blocks(handle_t *handle, struct inode *inode,
1721 			       ext4_lblk_t iblock, unsigned int max_blocks,
1722 			       struct buffer_head *bh_result, int flags);
1723 extern void ext4_ext_truncate(struct inode *);
1724 extern void ext4_ext_init(struct super_block *);
1725 extern void ext4_ext_release(struct super_block *);
1726 extern long ext4_fallocate(struct inode *inode, int mode, loff_t offset,
1727 			  loff_t len);
1728 extern int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset,
1729 			  loff_t len);
1730 extern int ext4_get_blocks(handle_t *handle, struct inode *inode,
1731 			   sector_t block, unsigned int max_blocks,
1732 			   struct buffer_head *bh, int flags);
1733 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
1734 			__u64 start, __u64 len);
1735 /* move_extent.c */
1736 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
1737 			     __u64 start_orig, __u64 start_donor,
1738 			     __u64 len, __u64 *moved_len);
1739 
1740 
1741 /*
1742  * Add new method to test wether block and inode bitmaps are properly
1743  * initialized. With uninit_bg reading the block from disk is not enough
1744  * to mark the bitmap uptodate. We need to also zero-out the bitmap
1745  */
1746 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
1747 
1748 static inline int bitmap_uptodate(struct buffer_head *bh)
1749 {
1750 	return (buffer_uptodate(bh) &&
1751 			test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
1752 }
1753 static inline void set_bitmap_uptodate(struct buffer_head *bh)
1754 {
1755 	set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
1756 }
1757 
1758 #endif	/* __KERNEL__ */
1759 
1760 #endif	/* _EXT4_H */
1761