xref: /openbmc/linux/fs/ext4/ext4.h (revision fd589a8f)
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 /* prefer goal again. length */
70 #define EXT4_MB_HINT_MERGE		0x0001
71 /* blocks already reserved */
72 #define EXT4_MB_HINT_RESERVED		0x0002
73 /* metadata is being allocated */
74 #define EXT4_MB_HINT_METADATA		0x0004
75 /* first blocks in the file */
76 #define EXT4_MB_HINT_FIRST		0x0008
77 /* search for the best chunk */
78 #define EXT4_MB_HINT_BEST		0x0010
79 /* data is being allocated */
80 #define EXT4_MB_HINT_DATA		0x0020
81 /* don't preallocate (for tails) */
82 #define EXT4_MB_HINT_NOPREALLOC		0x0040
83 /* allocate for locality group */
84 #define EXT4_MB_HINT_GROUP_ALLOC	0x0080
85 /* allocate goal blocks or none */
86 #define EXT4_MB_HINT_GOAL_ONLY		0x0100
87 /* goal is meaningful */
88 #define EXT4_MB_HINT_TRY_GOAL		0x0200
89 /* blocks already pre-reserved by delayed allocation */
90 #define EXT4_MB_DELALLOC_RESERVED	0x0400
91 /* We are doing stream allocation */
92 #define EXT4_MB_STREAM_ALLOC		0x0800
93 
94 
95 struct ext4_allocation_request {
96 	/* target inode for block we're allocating */
97 	struct inode *inode;
98 	/* how many blocks we want to allocate */
99 	unsigned int len;
100 	/* logical block in target inode */
101 	ext4_lblk_t logical;
102 	/* the closest logical allocated block to the left */
103 	ext4_lblk_t lleft;
104 	/* the closest logical allocated block to the right */
105 	ext4_lblk_t lright;
106 	/* phys. target (a hint) */
107 	ext4_fsblk_t goal;
108 	/* phys. block for the closest logical allocated block to the left */
109 	ext4_fsblk_t pleft;
110 	/* phys. block for the closest logical allocated block to the right */
111 	ext4_fsblk_t pright;
112 	/* flags. see above EXT4_MB_HINT_* */
113 	unsigned int flags;
114 };
115 
116 /*
117  * For delayed allocation tracking
118  */
119 struct mpage_da_data {
120 	struct inode *inode;
121 	sector_t b_blocknr;		/* start block number of extent */
122 	size_t b_size;			/* size of extent */
123 	unsigned long b_state;		/* state of the extent */
124 	unsigned long first_page, next_page;	/* extent of pages */
125 	struct writeback_control *wbc;
126 	int io_done;
127 	int pages_written;
128 	int retval;
129 };
130 
131 /*
132  * Special inodes numbers
133  */
134 #define	EXT4_BAD_INO		 1	/* Bad blocks inode */
135 #define EXT4_ROOT_INO		 2	/* Root inode */
136 #define EXT4_BOOT_LOADER_INO	 5	/* Boot loader inode */
137 #define EXT4_UNDEL_DIR_INO	 6	/* Undelete directory inode */
138 #define EXT4_RESIZE_INO		 7	/* Reserved group descriptors inode */
139 #define EXT4_JOURNAL_INO	 8	/* Journal inode */
140 
141 /* First non-reserved inode for old ext4 filesystems */
142 #define EXT4_GOOD_OLD_FIRST_INO	11
143 
144 /*
145  * Maximal count of links to a file
146  */
147 #define EXT4_LINK_MAX		65000
148 
149 /*
150  * Macro-instructions used to manage several block sizes
151  */
152 #define EXT4_MIN_BLOCK_SIZE		1024
153 #define	EXT4_MAX_BLOCK_SIZE		65536
154 #define EXT4_MIN_BLOCK_LOG_SIZE		10
155 #ifdef __KERNEL__
156 # define EXT4_BLOCK_SIZE(s)		((s)->s_blocksize)
157 #else
158 # define EXT4_BLOCK_SIZE(s)		(EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
159 #endif
160 #define	EXT4_ADDR_PER_BLOCK(s)		(EXT4_BLOCK_SIZE(s) / sizeof(__u32))
161 #ifdef __KERNEL__
162 # define EXT4_BLOCK_SIZE_BITS(s)	((s)->s_blocksize_bits)
163 #else
164 # define EXT4_BLOCK_SIZE_BITS(s)	((s)->s_log_block_size + 10)
165 #endif
166 #ifdef __KERNEL__
167 #define	EXT4_ADDR_PER_BLOCK_BITS(s)	(EXT4_SB(s)->s_addr_per_block_bits)
168 #define EXT4_INODE_SIZE(s)		(EXT4_SB(s)->s_inode_size)
169 #define EXT4_FIRST_INO(s)		(EXT4_SB(s)->s_first_ino)
170 #else
171 #define EXT4_INODE_SIZE(s)	(((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
172 				 EXT4_GOOD_OLD_INODE_SIZE : \
173 				 (s)->s_inode_size)
174 #define EXT4_FIRST_INO(s)	(((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
175 				 EXT4_GOOD_OLD_FIRST_INO : \
176 				 (s)->s_first_ino)
177 #endif
178 #define EXT4_BLOCK_ALIGN(size, blkbits)		ALIGN((size), (1 << (blkbits)))
179 
180 /*
181  * Structure of a blocks group descriptor
182  */
183 struct ext4_group_desc
184 {
185 	__le32	bg_block_bitmap_lo;	/* Blocks bitmap block */
186 	__le32	bg_inode_bitmap_lo;	/* Inodes bitmap block */
187 	__le32	bg_inode_table_lo;	/* Inodes table block */
188 	__le16	bg_free_blocks_count_lo;/* Free blocks count */
189 	__le16	bg_free_inodes_count_lo;/* Free inodes count */
190 	__le16	bg_used_dirs_count_lo;	/* Directories count */
191 	__le16	bg_flags;		/* EXT4_BG_flags (INODE_UNINIT, etc) */
192 	__u32	bg_reserved[2];		/* Likely block/inode bitmap checksum */
193 	__le16  bg_itable_unused_lo;	/* Unused inodes count */
194 	__le16  bg_checksum;		/* crc16(sb_uuid+group+desc) */
195 	__le32	bg_block_bitmap_hi;	/* Blocks bitmap block MSB */
196 	__le32	bg_inode_bitmap_hi;	/* Inodes bitmap block MSB */
197 	__le32	bg_inode_table_hi;	/* Inodes table block MSB */
198 	__le16	bg_free_blocks_count_hi;/* Free blocks count MSB */
199 	__le16	bg_free_inodes_count_hi;/* Free inodes count MSB */
200 	__le16	bg_used_dirs_count_hi;	/* Directories count MSB */
201 	__le16  bg_itable_unused_hi;    /* Unused inodes count MSB */
202 	__u32	bg_reserved2[3];
203 };
204 
205 /*
206  * Structure of a flex block group info
207  */
208 
209 struct flex_groups {
210 	atomic_t free_inodes;
211 	atomic_t free_blocks;
212 	atomic_t used_dirs;
213 };
214 
215 #define EXT4_BG_INODE_UNINIT	0x0001 /* Inode table/bitmap not in use */
216 #define EXT4_BG_BLOCK_UNINIT	0x0002 /* Block bitmap not in use */
217 #define EXT4_BG_INODE_ZEROED	0x0004 /* On-disk itable initialized to zero */
218 
219 /*
220  * Macro-instructions used to manage group descriptors
221  */
222 #define EXT4_MIN_DESC_SIZE		32
223 #define EXT4_MIN_DESC_SIZE_64BIT	64
224 #define	EXT4_MAX_DESC_SIZE		EXT4_MIN_BLOCK_SIZE
225 #define EXT4_DESC_SIZE(s)		(EXT4_SB(s)->s_desc_size)
226 #ifdef __KERNEL__
227 # define EXT4_BLOCKS_PER_GROUP(s)	(EXT4_SB(s)->s_blocks_per_group)
228 # define EXT4_DESC_PER_BLOCK(s)		(EXT4_SB(s)->s_desc_per_block)
229 # define EXT4_INODES_PER_GROUP(s)	(EXT4_SB(s)->s_inodes_per_group)
230 # define EXT4_DESC_PER_BLOCK_BITS(s)	(EXT4_SB(s)->s_desc_per_block_bits)
231 #else
232 # define EXT4_BLOCKS_PER_GROUP(s)	((s)->s_blocks_per_group)
233 # define EXT4_DESC_PER_BLOCK(s)		(EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
234 # define EXT4_INODES_PER_GROUP(s)	((s)->s_inodes_per_group)
235 #endif
236 
237 /*
238  * Constants relative to the data blocks
239  */
240 #define	EXT4_NDIR_BLOCKS		12
241 #define	EXT4_IND_BLOCK			EXT4_NDIR_BLOCKS
242 #define	EXT4_DIND_BLOCK			(EXT4_IND_BLOCK + 1)
243 #define	EXT4_TIND_BLOCK			(EXT4_DIND_BLOCK + 1)
244 #define	EXT4_N_BLOCKS			(EXT4_TIND_BLOCK + 1)
245 
246 /*
247  * Inode flags
248  */
249 #define	EXT4_SECRM_FL			0x00000001 /* Secure deletion */
250 #define	EXT4_UNRM_FL			0x00000002 /* Undelete */
251 #define	EXT4_COMPR_FL			0x00000004 /* Compress file */
252 #define EXT4_SYNC_FL			0x00000008 /* Synchronous updates */
253 #define EXT4_IMMUTABLE_FL		0x00000010 /* Immutable file */
254 #define EXT4_APPEND_FL			0x00000020 /* writes to file may only append */
255 #define EXT4_NODUMP_FL			0x00000040 /* do not dump file */
256 #define EXT4_NOATIME_FL			0x00000080 /* do not update atime */
257 /* Reserved for compression usage... */
258 #define EXT4_DIRTY_FL			0x00000100
259 #define EXT4_COMPRBLK_FL		0x00000200 /* One or more compressed clusters */
260 #define EXT4_NOCOMPR_FL			0x00000400 /* Don't compress */
261 #define EXT4_ECOMPR_FL			0x00000800 /* Compression error */
262 /* End compression flags --- maybe not all used */
263 #define EXT4_INDEX_FL			0x00001000 /* hash-indexed directory */
264 #define EXT4_IMAGIC_FL			0x00002000 /* AFS directory */
265 #define EXT4_JOURNAL_DATA_FL		0x00004000 /* file data should be journaled */
266 #define EXT4_NOTAIL_FL			0x00008000 /* file tail should not be merged */
267 #define EXT4_DIRSYNC_FL			0x00010000 /* dirsync behaviour (directories only) */
268 #define EXT4_TOPDIR_FL			0x00020000 /* Top of directory hierarchies*/
269 #define EXT4_HUGE_FILE_FL               0x00040000 /* Set to each huge file */
270 #define EXT4_EXTENTS_FL			0x00080000 /* Inode uses extents */
271 #define EXT4_RESERVED_FL		0x80000000 /* reserved for ext4 lib */
272 
273 #define EXT4_FL_USER_VISIBLE		0x000BDFFF /* User visible flags */
274 #define EXT4_FL_USER_MODIFIABLE		0x000B80FF /* User modifiable flags */
275 
276 /* Flags that should be inherited by new inodes from their parent. */
277 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
278 			   EXT4_SYNC_FL | EXT4_IMMUTABLE_FL | EXT4_APPEND_FL |\
279 			   EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
280 			   EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
281 			   EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
282 
283 /* Flags that are appropriate for regular files (all but dir-specific ones). */
284 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
285 
286 /* Flags that are appropriate for non-directories/regular files. */
287 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
288 
289 /* Mask out flags that are inappropriate for the given type of inode. */
290 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
291 {
292 	if (S_ISDIR(mode))
293 		return flags;
294 	else if (S_ISREG(mode))
295 		return flags & EXT4_REG_FLMASK;
296 	else
297 		return flags & EXT4_OTHER_FLMASK;
298 }
299 
300 /*
301  * Inode dynamic state flags
302  */
303 #define EXT4_STATE_JDATA		0x00000001 /* journaled data exists */
304 #define EXT4_STATE_NEW			0x00000002 /* inode is newly created */
305 #define EXT4_STATE_XATTR		0x00000004 /* has in-inode xattrs */
306 #define EXT4_STATE_NO_EXPAND		0x00000008 /* No space for expansion */
307 #define EXT4_STATE_DA_ALLOC_CLOSE	0x00000010 /* Alloc DA blks on close */
308 #define EXT4_STATE_EXT_MIGRATE		0x00000020 /* Inode is migrating */
309 
310 /* Used to pass group descriptor data when online resize is done */
311 struct ext4_new_group_input {
312 	__u32 group;		/* Group number for this data */
313 	__u64 block_bitmap;	/* Absolute block number of block bitmap */
314 	__u64 inode_bitmap;	/* Absolute block number of inode bitmap */
315 	__u64 inode_table;	/* Absolute block number of inode table start */
316 	__u32 blocks_count;	/* Total number of blocks in this group */
317 	__u16 reserved_blocks;	/* Number of reserved blocks in this group */
318 	__u16 unused;
319 };
320 
321 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
322 struct ext4_new_group_data {
323 	__u32 group;
324 	__u64 block_bitmap;
325 	__u64 inode_bitmap;
326 	__u64 inode_table;
327 	__u32 blocks_count;
328 	__u16 reserved_blocks;
329 	__u16 unused;
330 	__u32 free_blocks_count;
331 };
332 
333 /*
334  * Flags used by ext4_get_blocks()
335  */
336 	/* Allocate any needed blocks and/or convert an unitialized
337 	   extent to be an initialized ext4 */
338 #define EXT4_GET_BLOCKS_CREATE			0x0001
339 	/* Request the creation of an unitialized extent */
340 #define EXT4_GET_BLOCKS_UNINIT_EXT		0x0002
341 #define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT	(EXT4_GET_BLOCKS_UNINIT_EXT|\
342 						 EXT4_GET_BLOCKS_CREATE)
343 	/* Caller is from the delayed allocation writeout path,
344 	   so set the magic i_delalloc_reserve_flag after taking the
345 	   inode allocation semaphore for */
346 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE	0x0004
347 	/* Call ext4_da_update_reserve_space() after successfully
348 	   allocating the blocks */
349 #define EXT4_GET_BLOCKS_UPDATE_RESERVE_SPACE	0x0008
350 
351 
352 /*
353  * ioctl commands
354  */
355 #define	EXT4_IOC_GETFLAGS		FS_IOC_GETFLAGS
356 #define	EXT4_IOC_SETFLAGS		FS_IOC_SETFLAGS
357 #define	EXT4_IOC_GETVERSION		_IOR('f', 3, long)
358 #define	EXT4_IOC_SETVERSION		_IOW('f', 4, long)
359 #define	EXT4_IOC_GETVERSION_OLD		FS_IOC_GETVERSION
360 #define	EXT4_IOC_SETVERSION_OLD		FS_IOC_SETVERSION
361 #ifdef CONFIG_JBD2_DEBUG
362 #define EXT4_IOC_WAIT_FOR_READONLY	_IOR('f', 99, long)
363 #endif
364 #define EXT4_IOC_GETRSVSZ		_IOR('f', 5, long)
365 #define EXT4_IOC_SETRSVSZ		_IOW('f', 6, long)
366 #define EXT4_IOC_GROUP_EXTEND		_IOW('f', 7, unsigned long)
367 #define EXT4_IOC_GROUP_ADD		_IOW('f', 8, struct ext4_new_group_input)
368 #define EXT4_IOC_MIGRATE		_IO('f', 9)
369  /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
370  /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
371 #define EXT4_IOC_ALLOC_DA_BLKS		_IO('f', 12)
372 #define EXT4_IOC_MOVE_EXT		_IOWR('f', 15, struct move_extent)
373 
374 /*
375  * ioctl commands in 32 bit emulation
376  */
377 #define EXT4_IOC32_GETFLAGS		FS_IOC32_GETFLAGS
378 #define EXT4_IOC32_SETFLAGS		FS_IOC32_SETFLAGS
379 #define EXT4_IOC32_GETVERSION		_IOR('f', 3, int)
380 #define EXT4_IOC32_SETVERSION		_IOW('f', 4, int)
381 #define EXT4_IOC32_GETRSVSZ		_IOR('f', 5, int)
382 #define EXT4_IOC32_SETRSVSZ		_IOW('f', 6, int)
383 #define EXT4_IOC32_GROUP_EXTEND		_IOW('f', 7, unsigned int)
384 #ifdef CONFIG_JBD2_DEBUG
385 #define EXT4_IOC32_WAIT_FOR_READONLY	_IOR('f', 99, int)
386 #endif
387 #define EXT4_IOC32_GETVERSION_OLD	FS_IOC32_GETVERSION
388 #define EXT4_IOC32_SETVERSION_OLD	FS_IOC32_SETVERSION
389 
390 
391 /*
392  *  Mount options
393  */
394 struct ext4_mount_options {
395 	unsigned long s_mount_opt;
396 	uid_t s_resuid;
397 	gid_t s_resgid;
398 	unsigned long s_commit_interval;
399 	u32 s_min_batch_time, s_max_batch_time;
400 #ifdef CONFIG_QUOTA
401 	int s_jquota_fmt;
402 	char *s_qf_names[MAXQUOTAS];
403 #endif
404 };
405 
406 /* Max physical block we can addres w/o extents */
407 #define EXT4_MAX_BLOCK_FILE_PHYS	0xFFFFFFFF
408 
409 /*
410  * Structure of an inode on the disk
411  */
412 struct ext4_inode {
413 	__le16	i_mode;		/* File mode */
414 	__le16	i_uid;		/* Low 16 bits of Owner Uid */
415 	__le32	i_size_lo;	/* Size in bytes */
416 	__le32	i_atime;	/* Access time */
417 	__le32	i_ctime;	/* Inode Change time */
418 	__le32	i_mtime;	/* Modification time */
419 	__le32	i_dtime;	/* Deletion Time */
420 	__le16	i_gid;		/* Low 16 bits of Group Id */
421 	__le16	i_links_count;	/* Links count */
422 	__le32	i_blocks_lo;	/* Blocks count */
423 	__le32	i_flags;	/* File flags */
424 	union {
425 		struct {
426 			__le32  l_i_version;
427 		} linux1;
428 		struct {
429 			__u32  h_i_translator;
430 		} hurd1;
431 		struct {
432 			__u32  m_i_reserved1;
433 		} masix1;
434 	} osd1;				/* OS dependent 1 */
435 	__le32	i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
436 	__le32	i_generation;	/* File version (for NFS) */
437 	__le32	i_file_acl_lo;	/* File ACL */
438 	__le32	i_size_high;
439 	__le32	i_obso_faddr;	/* Obsoleted fragment address */
440 	union {
441 		struct {
442 			__le16	l_i_blocks_high; /* were l_i_reserved1 */
443 			__le16	l_i_file_acl_high;
444 			__le16	l_i_uid_high;	/* these 2 fields */
445 			__le16	l_i_gid_high;	/* were reserved2[0] */
446 			__u32	l_i_reserved2;
447 		} linux2;
448 		struct {
449 			__le16	h_i_reserved1;	/* Obsoleted fragment number/size which are removed in ext4 */
450 			__u16	h_i_mode_high;
451 			__u16	h_i_uid_high;
452 			__u16	h_i_gid_high;
453 			__u32	h_i_author;
454 		} hurd2;
455 		struct {
456 			__le16	h_i_reserved1;	/* Obsoleted fragment number/size which are removed in ext4 */
457 			__le16	m_i_file_acl_high;
458 			__u32	m_i_reserved2[2];
459 		} masix2;
460 	} osd2;				/* OS dependent 2 */
461 	__le16	i_extra_isize;
462 	__le16	i_pad1;
463 	__le32  i_ctime_extra;  /* extra Change time      (nsec << 2 | epoch) */
464 	__le32  i_mtime_extra;  /* extra Modification time(nsec << 2 | epoch) */
465 	__le32  i_atime_extra;  /* extra Access time      (nsec << 2 | epoch) */
466 	__le32  i_crtime;       /* File Creation time */
467 	__le32  i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
468 	__le32  i_version_hi;	/* high 32 bits for 64-bit version */
469 };
470 
471 struct move_extent {
472 	__u32 reserved;		/* should be zero */
473 	__u32 donor_fd;		/* donor file descriptor */
474 	__u64 orig_start;	/* logical start offset in block for orig */
475 	__u64 donor_start;	/* logical start offset in block for donor */
476 	__u64 len;		/* block length to be moved */
477 	__u64 moved_len;	/* moved block length */
478 };
479 
480 #define EXT4_EPOCH_BITS 2
481 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
482 #define EXT4_NSEC_MASK  (~0UL << EXT4_EPOCH_BITS)
483 
484 /*
485  * Extended fields will fit into an inode if the filesystem was formatted
486  * with large inodes (-I 256 or larger) and there are not currently any EAs
487  * consuming all of the available space. For new inodes we always reserve
488  * enough space for the kernel's known extended fields, but for inodes
489  * created with an old kernel this might not have been the case. None of
490  * the extended inode fields is critical for correct filesystem operation.
491  * This macro checks if a certain field fits in the inode. Note that
492  * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
493  */
494 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field)	\
495 	((offsetof(typeof(*ext4_inode), field) +	\
496 	  sizeof((ext4_inode)->field))			\
497 	<= (EXT4_GOOD_OLD_INODE_SIZE +			\
498 	    (einode)->i_extra_isize))			\
499 
500 static inline __le32 ext4_encode_extra_time(struct timespec *time)
501 {
502        return cpu_to_le32((sizeof(time->tv_sec) > 4 ?
503 			   time->tv_sec >> 32 : 0) |
504 			   ((time->tv_nsec << 2) & EXT4_NSEC_MASK));
505 }
506 
507 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
508 {
509        if (sizeof(time->tv_sec) > 4)
510 	       time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK)
511 			       << 32;
512        time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> 2;
513 }
514 
515 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode)			       \
516 do {									       \
517 	(raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec);	       \
518 	if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     \
519 		(raw_inode)->xtime ## _extra =				       \
520 				ext4_encode_extra_time(&(inode)->xtime);       \
521 } while (0)
522 
523 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode)			       \
524 do {									       \
525 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))		       \
526 		(raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec);      \
527 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))	       \
528 		(raw_inode)->xtime ## _extra =				       \
529 				ext4_encode_extra_time(&(einode)->xtime);      \
530 } while (0)
531 
532 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode)			       \
533 do {									       \
534 	(inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime);       \
535 	if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     \
536 		ext4_decode_extra_time(&(inode)->xtime,			       \
537 				       raw_inode->xtime ## _extra);	       \
538 } while (0)
539 
540 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode)			       \
541 do {									       \
542 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))		       \
543 		(einode)->xtime.tv_sec = 				       \
544 			(signed)le32_to_cpu((raw_inode)->xtime);	       \
545 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))	       \
546 		ext4_decode_extra_time(&(einode)->xtime,		       \
547 				       raw_inode->xtime ## _extra);	       \
548 } while (0)
549 
550 #define i_disk_version osd1.linux1.l_i_version
551 
552 #if defined(__KERNEL__) || defined(__linux__)
553 #define i_reserved1	osd1.linux1.l_i_reserved1
554 #define i_file_acl_high	osd2.linux2.l_i_file_acl_high
555 #define i_blocks_high	osd2.linux2.l_i_blocks_high
556 #define i_uid_low	i_uid
557 #define i_gid_low	i_gid
558 #define i_uid_high	osd2.linux2.l_i_uid_high
559 #define i_gid_high	osd2.linux2.l_i_gid_high
560 #define i_reserved2	osd2.linux2.l_i_reserved2
561 
562 #elif defined(__GNU__)
563 
564 #define i_translator	osd1.hurd1.h_i_translator
565 #define i_uid_high	osd2.hurd2.h_i_uid_high
566 #define i_gid_high	osd2.hurd2.h_i_gid_high
567 #define i_author	osd2.hurd2.h_i_author
568 
569 #elif defined(__masix__)
570 
571 #define i_reserved1	osd1.masix1.m_i_reserved1
572 #define i_file_acl_high	osd2.masix2.m_i_file_acl_high
573 #define i_reserved2	osd2.masix2.m_i_reserved2
574 
575 #endif /* defined(__KERNEL__) || defined(__linux__) */
576 
577 /*
578  * storage for cached extent
579  */
580 struct ext4_ext_cache {
581 	ext4_fsblk_t	ec_start;
582 	ext4_lblk_t	ec_block;
583 	__u32		ec_len; /* must be 32bit to return holes */
584 	__u32		ec_type;
585 };
586 
587 /*
588  * fourth extended file system inode data in memory
589  */
590 struct ext4_inode_info {
591 	__le32	i_data[15];	/* unconverted */
592 	__u32	i_flags;
593 	ext4_fsblk_t	i_file_acl;
594 	__u32	i_dtime;
595 
596 	/*
597 	 * i_block_group is the number of the block group which contains
598 	 * this file's inode.  Constant across the lifetime of the inode,
599 	 * it is ued for making block allocation decisions - we try to
600 	 * place a file's data blocks near its inode block, and new inodes
601 	 * near to their parent directory's inode.
602 	 */
603 	ext4_group_t	i_block_group;
604 	__u32	i_state;		/* Dynamic state flags for ext4 */
605 
606 	ext4_lblk_t		i_dir_start_lookup;
607 #ifdef CONFIG_EXT4_FS_XATTR
608 	/*
609 	 * Extended attributes can be read independently of the main file
610 	 * data. Taking i_mutex even when reading would cause contention
611 	 * between readers of EAs and writers of regular file data, so
612 	 * instead we synchronize on xattr_sem when reading or changing
613 	 * EAs.
614 	 */
615 	struct rw_semaphore xattr_sem;
616 #endif
617 
618 	struct list_head i_orphan;	/* unlinked but open inodes */
619 
620 	/*
621 	 * i_disksize keeps track of what the inode size is ON DISK, not
622 	 * in memory.  During truncate, i_size is set to the new size by
623 	 * the VFS prior to calling ext4_truncate(), but the filesystem won't
624 	 * set i_disksize to 0 until the truncate is actually under way.
625 	 *
626 	 * The intent is that i_disksize always represents the blocks which
627 	 * are used by this file.  This allows recovery to restart truncate
628 	 * on orphans if we crash during truncate.  We actually write i_disksize
629 	 * into the on-disk inode when writing inodes out, instead of i_size.
630 	 *
631 	 * The only time when i_disksize and i_size may be different is when
632 	 * a truncate is in progress.  The only things which change i_disksize
633 	 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
634 	 */
635 	loff_t	i_disksize;
636 
637 	/*
638 	 * i_data_sem is for serialising ext4_truncate() against
639 	 * ext4_getblock().  In the 2.4 ext2 design, great chunks of inode's
640 	 * data tree are chopped off during truncate. We can't do that in
641 	 * ext4 because whenever we perform intermediate commits during
642 	 * truncate, the inode and all the metadata blocks *must* be in a
643 	 * consistent state which allows truncation of the orphans to restart
644 	 * during recovery.  Hence we must fix the get_block-vs-truncate race
645 	 * by other means, so we have i_data_sem.
646 	 */
647 	struct rw_semaphore i_data_sem;
648 	struct inode vfs_inode;
649 	struct jbd2_inode jinode;
650 
651 	struct ext4_ext_cache i_cached_extent;
652 	/*
653 	 * File creation time. Its function is same as that of
654 	 * struct timespec i_{a,c,m}time in the generic inode.
655 	 */
656 	struct timespec i_crtime;
657 
658 	/* mballoc */
659 	struct list_head i_prealloc_list;
660 	spinlock_t i_prealloc_lock;
661 
662 	/* ialloc */
663 	ext4_group_t	i_last_alloc_group;
664 
665 	/* allocation reservation info for delalloc */
666 	unsigned int i_reserved_data_blocks;
667 	unsigned int i_reserved_meta_blocks;
668 	unsigned int i_allocated_meta_blocks;
669 	unsigned short i_delalloc_reserved_flag;
670 
671 	/* on-disk additional length */
672 	__u16 i_extra_isize;
673 
674 	spinlock_t i_block_reservation_lock;
675 };
676 
677 /*
678  * File system states
679  */
680 #define	EXT4_VALID_FS			0x0001	/* Unmounted cleanly */
681 #define	EXT4_ERROR_FS			0x0002	/* Errors detected */
682 #define	EXT4_ORPHAN_FS			0x0004	/* Orphans being recovered */
683 
684 /*
685  * Misc. filesystem flags
686  */
687 #define EXT2_FLAGS_SIGNED_HASH		0x0001  /* Signed dirhash in use */
688 #define EXT2_FLAGS_UNSIGNED_HASH	0x0002  /* Unsigned dirhash in use */
689 #define EXT2_FLAGS_TEST_FILESYS		0x0004	/* to test development code */
690 
691 /*
692  * Mount flags
693  */
694 #define EXT4_MOUNT_OLDALLOC		0x00002  /* Don't use the new Orlov allocator */
695 #define EXT4_MOUNT_GRPID		0x00004	/* Create files with directory's group */
696 #define EXT4_MOUNT_DEBUG		0x00008	/* Some debugging messages */
697 #define EXT4_MOUNT_ERRORS_CONT		0x00010	/* Continue on errors */
698 #define EXT4_MOUNT_ERRORS_RO		0x00020	/* Remount fs ro on errors */
699 #define EXT4_MOUNT_ERRORS_PANIC		0x00040	/* Panic on errors */
700 #define EXT4_MOUNT_MINIX_DF		0x00080	/* Mimics the Minix statfs */
701 #define EXT4_MOUNT_NOLOAD		0x00100	/* Don't use existing journal*/
702 #define EXT4_MOUNT_DATA_FLAGS		0x00C00	/* Mode for data writes: */
703 #define EXT4_MOUNT_JOURNAL_DATA		0x00400	/* Write data to journal */
704 #define EXT4_MOUNT_ORDERED_DATA		0x00800	/* Flush data before commit */
705 #define EXT4_MOUNT_WRITEBACK_DATA	0x00C00	/* No data ordering */
706 #define EXT4_MOUNT_UPDATE_JOURNAL	0x01000	/* Update the journal format */
707 #define EXT4_MOUNT_NO_UID32		0x02000  /* Disable 32-bit UIDs */
708 #define EXT4_MOUNT_XATTR_USER		0x04000	/* Extended user attributes */
709 #define EXT4_MOUNT_POSIX_ACL		0x08000	/* POSIX Access Control Lists */
710 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC	0x10000	/* No auto delalloc mapping */
711 #define EXT4_MOUNT_BARRIER		0x20000 /* Use block barriers */
712 #define EXT4_MOUNT_NOBH			0x40000 /* No bufferheads */
713 #define EXT4_MOUNT_QUOTA		0x80000 /* Some quota option set */
714 #define EXT4_MOUNT_USRQUOTA		0x100000 /* "old" user quota */
715 #define EXT4_MOUNT_GRPQUOTA		0x200000 /* "old" group quota */
716 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT	0x1000000 /* Journal Async Commit */
717 #define EXT4_MOUNT_I_VERSION            0x2000000 /* i_version support */
718 #define EXT4_MOUNT_DELALLOC		0x8000000 /* Delalloc support */
719 #define EXT4_MOUNT_DATA_ERR_ABORT	0x10000000 /* Abort on file data write */
720 #define EXT4_MOUNT_BLOCK_VALIDITY	0x20000000 /* Block validity checking */
721 
722 #define clear_opt(o, opt)		o &= ~EXT4_MOUNT_##opt
723 #define set_opt(o, opt)			o |= EXT4_MOUNT_##opt
724 #define test_opt(sb, opt)		(EXT4_SB(sb)->s_mount_opt & \
725 					 EXT4_MOUNT_##opt)
726 
727 #define ext4_set_bit			ext2_set_bit
728 #define ext4_set_bit_atomic		ext2_set_bit_atomic
729 #define ext4_clear_bit			ext2_clear_bit
730 #define ext4_clear_bit_atomic		ext2_clear_bit_atomic
731 #define ext4_test_bit			ext2_test_bit
732 #define ext4_find_first_zero_bit	ext2_find_first_zero_bit
733 #define ext4_find_next_zero_bit		ext2_find_next_zero_bit
734 #define ext4_find_next_bit		ext2_find_next_bit
735 
736 /*
737  * Maximal mount counts between two filesystem checks
738  */
739 #define EXT4_DFL_MAX_MNT_COUNT		20	/* Allow 20 mounts */
740 #define EXT4_DFL_CHECKINTERVAL		0	/* Don't use interval check */
741 
742 /*
743  * Behaviour when detecting errors
744  */
745 #define EXT4_ERRORS_CONTINUE		1	/* Continue execution */
746 #define EXT4_ERRORS_RO			2	/* Remount fs read-only */
747 #define EXT4_ERRORS_PANIC		3	/* Panic */
748 #define EXT4_ERRORS_DEFAULT		EXT4_ERRORS_CONTINUE
749 
750 /*
751  * Structure of the super block
752  */
753 struct ext4_super_block {
754 /*00*/	__le32	s_inodes_count;		/* Inodes count */
755 	__le32	s_blocks_count_lo;	/* Blocks count */
756 	__le32	s_r_blocks_count_lo;	/* Reserved blocks count */
757 	__le32	s_free_blocks_count_lo;	/* Free blocks count */
758 /*10*/	__le32	s_free_inodes_count;	/* Free inodes count */
759 	__le32	s_first_data_block;	/* First Data Block */
760 	__le32	s_log_block_size;	/* Block size */
761 	__le32	s_obso_log_frag_size;	/* Obsoleted fragment size */
762 /*20*/	__le32	s_blocks_per_group;	/* # Blocks per group */
763 	__le32	s_obso_frags_per_group;	/* Obsoleted fragments per group */
764 	__le32	s_inodes_per_group;	/* # Inodes per group */
765 	__le32	s_mtime;		/* Mount time */
766 /*30*/	__le32	s_wtime;		/* Write time */
767 	__le16	s_mnt_count;		/* Mount count */
768 	__le16	s_max_mnt_count;	/* Maximal mount count */
769 	__le16	s_magic;		/* Magic signature */
770 	__le16	s_state;		/* File system state */
771 	__le16	s_errors;		/* Behaviour when detecting errors */
772 	__le16	s_minor_rev_level;	/* minor revision level */
773 /*40*/	__le32	s_lastcheck;		/* time of last check */
774 	__le32	s_checkinterval;	/* max. time between checks */
775 	__le32	s_creator_os;		/* OS */
776 	__le32	s_rev_level;		/* Revision level */
777 /*50*/	__le16	s_def_resuid;		/* Default uid for reserved blocks */
778 	__le16	s_def_resgid;		/* Default gid for reserved blocks */
779 	/*
780 	 * These fields are for EXT4_DYNAMIC_REV superblocks only.
781 	 *
782 	 * Note: the difference between the compatible feature set and
783 	 * the incompatible feature set is that if there is a bit set
784 	 * in the incompatible feature set that the kernel doesn't
785 	 * know about, it should refuse to mount the filesystem.
786 	 *
787 	 * e2fsck's requirements are more strict; if it doesn't know
788 	 * about a feature in either the compatible or incompatible
789 	 * feature set, it must abort and not try to meddle with
790 	 * things it doesn't understand...
791 	 */
792 	__le32	s_first_ino;		/* First non-reserved inode */
793 	__le16  s_inode_size;		/* size of inode structure */
794 	__le16	s_block_group_nr;	/* block group # of this superblock */
795 	__le32	s_feature_compat;	/* compatible feature set */
796 /*60*/	__le32	s_feature_incompat;	/* incompatible feature set */
797 	__le32	s_feature_ro_compat;	/* readonly-compatible feature set */
798 /*68*/	__u8	s_uuid[16];		/* 128-bit uuid for volume */
799 /*78*/	char	s_volume_name[16];	/* volume name */
800 /*88*/	char	s_last_mounted[64];	/* directory where last mounted */
801 /*C8*/	__le32	s_algorithm_usage_bitmap; /* For compression */
802 	/*
803 	 * Performance hints.  Directory preallocation should only
804 	 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
805 	 */
806 	__u8	s_prealloc_blocks;	/* Nr of blocks to try to preallocate*/
807 	__u8	s_prealloc_dir_blocks;	/* Nr to preallocate for dirs */
808 	__le16	s_reserved_gdt_blocks;	/* Per group desc for online growth */
809 	/*
810 	 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
811 	 */
812 /*D0*/	__u8	s_journal_uuid[16];	/* uuid of journal superblock */
813 /*E0*/	__le32	s_journal_inum;		/* inode number of journal file */
814 	__le32	s_journal_dev;		/* device number of journal file */
815 	__le32	s_last_orphan;		/* start of list of inodes to delete */
816 	__le32	s_hash_seed[4];		/* HTREE hash seed */
817 	__u8	s_def_hash_version;	/* Default hash version to use */
818 	__u8	s_reserved_char_pad;
819 	__le16  s_desc_size;		/* size of group descriptor */
820 /*100*/	__le32	s_default_mount_opts;
821 	__le32	s_first_meta_bg;	/* First metablock block group */
822 	__le32	s_mkfs_time;		/* When the filesystem was created */
823 	__le32	s_jnl_blocks[17];	/* Backup of the journal inode */
824 	/* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
825 /*150*/	__le32	s_blocks_count_hi;	/* Blocks count */
826 	__le32	s_r_blocks_count_hi;	/* Reserved blocks count */
827 	__le32	s_free_blocks_count_hi;	/* Free blocks count */
828 	__le16	s_min_extra_isize;	/* All inodes have at least # bytes */
829 	__le16	s_want_extra_isize; 	/* New inodes should reserve # bytes */
830 	__le32	s_flags;		/* Miscellaneous flags */
831 	__le16  s_raid_stride;		/* RAID stride */
832 	__le16  s_mmp_interval;         /* # seconds to wait in MMP checking */
833 	__le64  s_mmp_block;            /* Block for multi-mount protection */
834 	__le32  s_raid_stripe_width;    /* blocks on all data disks (N*stride)*/
835 	__u8	s_log_groups_per_flex;  /* FLEX_BG group size */
836 	__u8	s_reserved_char_pad2;
837 	__le16  s_reserved_pad;
838 	__le64	s_kbytes_written;	/* nr of lifetime kilobytes written */
839 	__u32   s_reserved[160];        /* Padding to the end of the block */
840 };
841 
842 #ifdef __KERNEL__
843 
844 /*
845  * run-time mount flags
846  */
847 #define EXT4_MF_MNTDIR_SAMPLED	0x0001
848 #define EXT4_MF_FS_ABORTED	0x0002	/* Fatal error detected */
849 
850 /*
851  * fourth extended-fs super-block data in memory
852  */
853 struct ext4_sb_info {
854 	unsigned long s_desc_size;	/* Size of a group descriptor in bytes */
855 	unsigned long s_inodes_per_block;/* Number of inodes per block */
856 	unsigned long s_blocks_per_group;/* Number of blocks in a group */
857 	unsigned long s_inodes_per_group;/* Number of inodes in a group */
858 	unsigned long s_itb_per_group;	/* Number of inode table blocks per group */
859 	unsigned long s_gdb_count;	/* Number of group descriptor blocks */
860 	unsigned long s_desc_per_block;	/* Number of group descriptors per block */
861 	ext4_group_t s_groups_count;	/* Number of groups in the fs */
862 	ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
863 	unsigned long s_overhead_last;  /* Last calculated overhead */
864 	unsigned long s_blocks_last;    /* Last seen block count */
865 	loff_t s_bitmap_maxbytes;	/* max bytes for bitmap files */
866 	struct buffer_head * s_sbh;	/* Buffer containing the super block */
867 	struct ext4_super_block *s_es;	/* Pointer to the super block in the buffer */
868 	struct buffer_head **s_group_desc;
869 	unsigned int s_mount_opt;
870 	unsigned int s_mount_flags;
871 	ext4_fsblk_t s_sb_block;
872 	uid_t s_resuid;
873 	gid_t s_resgid;
874 	unsigned short s_mount_state;
875 	unsigned short s_pad;
876 	int s_addr_per_block_bits;
877 	int s_desc_per_block_bits;
878 	int s_inode_size;
879 	int s_first_ino;
880 	unsigned int s_inode_readahead_blks;
881 	unsigned int s_inode_goal;
882 	spinlock_t s_next_gen_lock;
883 	u32 s_next_generation;
884 	u32 s_hash_seed[4];
885 	int s_def_hash_version;
886 	int s_hash_unsigned;	/* 3 if hash should be signed, 0 if not */
887 	struct percpu_counter s_freeblocks_counter;
888 	struct percpu_counter s_freeinodes_counter;
889 	struct percpu_counter s_dirs_counter;
890 	struct percpu_counter s_dirtyblocks_counter;
891 	struct blockgroup_lock *s_blockgroup_lock;
892 	struct proc_dir_entry *s_proc;
893 	struct kobject s_kobj;
894 	struct completion s_kobj_unregister;
895 
896 	/* Journaling */
897 	struct inode *s_journal_inode;
898 	struct journal_s *s_journal;
899 	struct list_head s_orphan;
900 	struct mutex s_orphan_lock;
901 	struct mutex s_resize_lock;
902 	unsigned long s_commit_interval;
903 	u32 s_max_batch_time;
904 	u32 s_min_batch_time;
905 	struct block_device *journal_bdev;
906 #ifdef CONFIG_JBD2_DEBUG
907 	struct timer_list turn_ro_timer;	/* For turning read-only (crash simulation) */
908 	wait_queue_head_t ro_wait_queue;	/* For people waiting for the fs to go read-only */
909 #endif
910 #ifdef CONFIG_QUOTA
911 	char *s_qf_names[MAXQUOTAS];		/* Names of quota files with journalled quota */
912 	int s_jquota_fmt;			/* Format of quota to use */
913 #endif
914 	unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
915 	struct rb_root system_blks;
916 
917 #ifdef EXTENTS_STATS
918 	/* ext4 extents stats */
919 	unsigned long s_ext_min;
920 	unsigned long s_ext_max;
921 	unsigned long s_depth_max;
922 	spinlock_t s_ext_stats_lock;
923 	unsigned long s_ext_blocks;
924 	unsigned long s_ext_extents;
925 #endif
926 
927 	/* for buddy allocator */
928 	struct ext4_group_info ***s_group_info;
929 	struct inode *s_buddy_cache;
930 	long s_blocks_reserved;
931 	spinlock_t s_reserve_lock;
932 	spinlock_t s_md_lock;
933 	tid_t s_last_transaction;
934 	unsigned short *s_mb_offsets;
935 	unsigned int *s_mb_maxs;
936 
937 	/* tunables */
938 	unsigned long s_stripe;
939 	unsigned int s_mb_stream_request;
940 	unsigned int s_mb_max_to_scan;
941 	unsigned int s_mb_min_to_scan;
942 	unsigned int s_mb_stats;
943 	unsigned int s_mb_order2_reqs;
944 	unsigned int s_mb_group_prealloc;
945 	/* where last allocation was done - for stream allocation */
946 	unsigned long s_mb_last_group;
947 	unsigned long s_mb_last_start;
948 
949 	/* history to debug policy */
950 	struct ext4_mb_history *s_mb_history;
951 	int s_mb_history_cur;
952 	int s_mb_history_max;
953 	int s_mb_history_num;
954 	spinlock_t s_mb_history_lock;
955 	int s_mb_history_filter;
956 
957 	/* stats for buddy allocator */
958 	spinlock_t s_mb_pa_lock;
959 	atomic_t s_bal_reqs;	/* number of reqs with len > 1 */
960 	atomic_t s_bal_success;	/* we found long enough chunks */
961 	atomic_t s_bal_allocated;	/* in blocks */
962 	atomic_t s_bal_ex_scanned;	/* total extents scanned */
963 	atomic_t s_bal_goals;	/* goal hits */
964 	atomic_t s_bal_breaks;	/* too long searches */
965 	atomic_t s_bal_2orders;	/* 2^order hits */
966 	spinlock_t s_bal_lock;
967 	unsigned long s_mb_buddies_generated;
968 	unsigned long long s_mb_generation_time;
969 	atomic_t s_mb_lost_chunks;
970 	atomic_t s_mb_preallocated;
971 	atomic_t s_mb_discarded;
972 	atomic_t s_lock_busy;
973 
974 	/* locality groups */
975 	struct ext4_locality_group *s_locality_groups;
976 
977 	/* for write statistics */
978 	unsigned long s_sectors_written_start;
979 	u64 s_kbytes_written;
980 
981 	unsigned int s_log_groups_per_flex;
982 	struct flex_groups *s_flex_groups;
983 };
984 
985 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
986 {
987 	return sb->s_fs_info;
988 }
989 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
990 {
991 	return container_of(inode, struct ext4_inode_info, vfs_inode);
992 }
993 
994 static inline struct timespec ext4_current_time(struct inode *inode)
995 {
996 	return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
997 		current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
998 }
999 
1000 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1001 {
1002 	return ino == EXT4_ROOT_INO ||
1003 		ino == EXT4_JOURNAL_INO ||
1004 		ino == EXT4_RESIZE_INO ||
1005 		(ino >= EXT4_FIRST_INO(sb) &&
1006 		 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1007 }
1008 #else
1009 /* Assume that user mode programs are passing in an ext4fs superblock, not
1010  * a kernel struct super_block.  This will allow us to call the feature-test
1011  * macros from user land. */
1012 #define EXT4_SB(sb)	(sb)
1013 #endif
1014 
1015 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1016 
1017 /*
1018  * Codes for operating systems
1019  */
1020 #define EXT4_OS_LINUX		0
1021 #define EXT4_OS_HURD		1
1022 #define EXT4_OS_MASIX		2
1023 #define EXT4_OS_FREEBSD		3
1024 #define EXT4_OS_LITES		4
1025 
1026 /*
1027  * Revision levels
1028  */
1029 #define EXT4_GOOD_OLD_REV	0	/* The good old (original) format */
1030 #define EXT4_DYNAMIC_REV	1	/* V2 format w/ dynamic inode sizes */
1031 
1032 #define EXT4_CURRENT_REV	EXT4_GOOD_OLD_REV
1033 #define EXT4_MAX_SUPP_REV	EXT4_DYNAMIC_REV
1034 
1035 #define EXT4_GOOD_OLD_INODE_SIZE 128
1036 
1037 /*
1038  * Feature set definitions
1039  */
1040 
1041 #define EXT4_HAS_COMPAT_FEATURE(sb,mask)			\
1042 	((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1043 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask)			\
1044 	((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1045 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask)			\
1046 	((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1047 #define EXT4_SET_COMPAT_FEATURE(sb,mask)			\
1048 	EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1049 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask)			\
1050 	EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1051 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask)			\
1052 	EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1053 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask)			\
1054 	EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1055 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask)			\
1056 	EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1057 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask)			\
1058 	EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1059 
1060 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC	0x0001
1061 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES	0x0002
1062 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL		0x0004
1063 #define EXT4_FEATURE_COMPAT_EXT_ATTR		0x0008
1064 #define EXT4_FEATURE_COMPAT_RESIZE_INODE	0x0010
1065 #define EXT4_FEATURE_COMPAT_DIR_INDEX		0x0020
1066 
1067 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER	0x0001
1068 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE	0x0002
1069 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR	0x0004
1070 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE        0x0008
1071 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM		0x0010
1072 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK	0x0020
1073 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE	0x0040
1074 
1075 #define EXT4_FEATURE_INCOMPAT_COMPRESSION	0x0001
1076 #define EXT4_FEATURE_INCOMPAT_FILETYPE		0x0002
1077 #define EXT4_FEATURE_INCOMPAT_RECOVER		0x0004 /* Needs recovery */
1078 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV	0x0008 /* Journal device */
1079 #define EXT4_FEATURE_INCOMPAT_META_BG		0x0010
1080 #define EXT4_FEATURE_INCOMPAT_EXTENTS		0x0040 /* extents support */
1081 #define EXT4_FEATURE_INCOMPAT_64BIT		0x0080
1082 #define EXT4_FEATURE_INCOMPAT_MMP               0x0100
1083 #define EXT4_FEATURE_INCOMPAT_FLEX_BG		0x0200
1084 
1085 #define EXT4_FEATURE_COMPAT_SUPP	EXT2_FEATURE_COMPAT_EXT_ATTR
1086 #define EXT4_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
1087 					 EXT4_FEATURE_INCOMPAT_RECOVER| \
1088 					 EXT4_FEATURE_INCOMPAT_META_BG| \
1089 					 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1090 					 EXT4_FEATURE_INCOMPAT_64BIT| \
1091 					 EXT4_FEATURE_INCOMPAT_FLEX_BG)
1092 #define EXT4_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1093 					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1094 					 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1095 					 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1096 					 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1097 					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1098 					 EXT4_FEATURE_RO_COMPAT_HUGE_FILE)
1099 
1100 /*
1101  * Default values for user and/or group using reserved blocks
1102  */
1103 #define	EXT4_DEF_RESUID		0
1104 #define	EXT4_DEF_RESGID		0
1105 
1106 #define EXT4_DEF_INODE_READAHEAD_BLKS	32
1107 
1108 /*
1109  * Default mount options
1110  */
1111 #define EXT4_DEFM_DEBUG		0x0001
1112 #define EXT4_DEFM_BSDGROUPS	0x0002
1113 #define EXT4_DEFM_XATTR_USER	0x0004
1114 #define EXT4_DEFM_ACL		0x0008
1115 #define EXT4_DEFM_UID16		0x0010
1116 #define EXT4_DEFM_JMODE		0x0060
1117 #define EXT4_DEFM_JMODE_DATA	0x0020
1118 #define EXT4_DEFM_JMODE_ORDERED	0x0040
1119 #define EXT4_DEFM_JMODE_WBACK	0x0060
1120 
1121 /*
1122  * Default journal batch times
1123  */
1124 #define EXT4_DEF_MIN_BATCH_TIME	0
1125 #define EXT4_DEF_MAX_BATCH_TIME	15000 /* 15ms */
1126 
1127 /*
1128  * Minimum number of groups in a flexgroup before we separate out
1129  * directories into the first block group of a flexgroup
1130  */
1131 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME	4
1132 
1133 /*
1134  * Structure of a directory entry
1135  */
1136 #define EXT4_NAME_LEN 255
1137 
1138 struct ext4_dir_entry {
1139 	__le32	inode;			/* Inode number */
1140 	__le16	rec_len;		/* Directory entry length */
1141 	__le16	name_len;		/* Name length */
1142 	char	name[EXT4_NAME_LEN];	/* File name */
1143 };
1144 
1145 /*
1146  * The new version of the directory entry.  Since EXT4 structures are
1147  * stored in intel byte order, and the name_len field could never be
1148  * bigger than 255 chars, it's safe to reclaim the extra byte for the
1149  * file_type field.
1150  */
1151 struct ext4_dir_entry_2 {
1152 	__le32	inode;			/* Inode number */
1153 	__le16	rec_len;		/* Directory entry length */
1154 	__u8	name_len;		/* Name length */
1155 	__u8	file_type;
1156 	char	name[EXT4_NAME_LEN];	/* File name */
1157 };
1158 
1159 /*
1160  * Ext4 directory file types.  Only the low 3 bits are used.  The
1161  * other bits are reserved for now.
1162  */
1163 #define EXT4_FT_UNKNOWN		0
1164 #define EXT4_FT_REG_FILE	1
1165 #define EXT4_FT_DIR		2
1166 #define EXT4_FT_CHRDEV		3
1167 #define EXT4_FT_BLKDEV		4
1168 #define EXT4_FT_FIFO		5
1169 #define EXT4_FT_SOCK		6
1170 #define EXT4_FT_SYMLINK		7
1171 
1172 #define EXT4_FT_MAX		8
1173 
1174 /*
1175  * EXT4_DIR_PAD defines the directory entries boundaries
1176  *
1177  * NOTE: It must be a multiple of 4
1178  */
1179 #define EXT4_DIR_PAD			4
1180 #define EXT4_DIR_ROUND			(EXT4_DIR_PAD - 1)
1181 #define EXT4_DIR_REC_LEN(name_len)	(((name_len) + 8 + EXT4_DIR_ROUND) & \
1182 					 ~EXT4_DIR_ROUND)
1183 #define EXT4_MAX_REC_LEN		((1<<16)-1)
1184 
1185 /*
1186  * Hash Tree Directory indexing
1187  * (c) Daniel Phillips, 2001
1188  */
1189 
1190 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1191 				      EXT4_FEATURE_COMPAT_DIR_INDEX) && \
1192 		      (EXT4_I(dir)->i_flags & EXT4_INDEX_FL))
1193 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1194 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1195 
1196 /* Legal values for the dx_root hash_version field: */
1197 
1198 #define DX_HASH_LEGACY		0
1199 #define DX_HASH_HALF_MD4	1
1200 #define DX_HASH_TEA		2
1201 #define DX_HASH_LEGACY_UNSIGNED	3
1202 #define DX_HASH_HALF_MD4_UNSIGNED	4
1203 #define DX_HASH_TEA_UNSIGNED		5
1204 
1205 #ifdef __KERNEL__
1206 
1207 /* hash info structure used by the directory hash */
1208 struct dx_hash_info
1209 {
1210 	u32		hash;
1211 	u32		minor_hash;
1212 	int		hash_version;
1213 	u32		*seed;
1214 };
1215 
1216 #define EXT4_HTREE_EOF	0x7fffffff
1217 
1218 /*
1219  * Control parameters used by ext4_htree_next_block
1220  */
1221 #define HASH_NB_ALWAYS		1
1222 
1223 
1224 /*
1225  * Describe an inode's exact location on disk and in memory
1226  */
1227 struct ext4_iloc
1228 {
1229 	struct buffer_head *bh;
1230 	unsigned long offset;
1231 	ext4_group_t block_group;
1232 };
1233 
1234 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
1235 {
1236 	return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
1237 }
1238 
1239 /*
1240  * This structure is stuffed into the struct file's private_data field
1241  * for directories.  It is where we put information so that we can do
1242  * readdir operations in hash tree order.
1243  */
1244 struct dir_private_info {
1245 	struct rb_root	root;
1246 	struct rb_node	*curr_node;
1247 	struct fname	*extra_fname;
1248 	loff_t		last_pos;
1249 	__u32		curr_hash;
1250 	__u32		curr_minor_hash;
1251 	__u32		next_hash;
1252 };
1253 
1254 /* calculate the first block number of the group */
1255 static inline ext4_fsblk_t
1256 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
1257 {
1258 	return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
1259 		le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
1260 }
1261 
1262 /*
1263  * Special error return code only used by dx_probe() and its callers.
1264  */
1265 #define ERR_BAD_DX_DIR	-75000
1266 
1267 void ext4_get_group_no_and_offset(struct super_block *sb, ext4_fsblk_t blocknr,
1268 			ext4_group_t *blockgrpp, ext4_grpblk_t *offsetp);
1269 
1270 extern struct proc_dir_entry *ext4_proc_root;
1271 
1272 /*
1273  * Function prototypes
1274  */
1275 
1276 /*
1277  * Ok, these declarations are also in <linux/kernel.h> but none of the
1278  * ext4 source programs needs to include it so they are duplicated here.
1279  */
1280 # define NORET_TYPE	/**/
1281 # define ATTRIB_NORET	__attribute__((noreturn))
1282 # define NORET_AND	noreturn,
1283 
1284 /* bitmap.c */
1285 extern unsigned int ext4_count_free(struct buffer_head *, unsigned);
1286 
1287 /* balloc.c */
1288 extern unsigned int ext4_block_group(struct super_block *sb,
1289 			ext4_fsblk_t blocknr);
1290 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
1291 			ext4_fsblk_t blocknr);
1292 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
1293 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
1294 			ext4_group_t group);
1295 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
1296 			ext4_fsblk_t goal, unsigned long *count, int *errp);
1297 extern int ext4_claim_free_blocks(struct ext4_sb_info *sbi, s64 nblocks);
1298 extern int ext4_has_free_blocks(struct ext4_sb_info *sbi, s64 nblocks);
1299 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
1300 			ext4_fsblk_t block, unsigned long count, int metadata);
1301 extern void ext4_add_groupblocks(handle_t *handle, struct super_block *sb,
1302 				ext4_fsblk_t block, unsigned long count);
1303 extern ext4_fsblk_t ext4_count_free_blocks(struct super_block *);
1304 extern void ext4_check_blocks_bitmap(struct super_block *);
1305 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
1306 						    ext4_group_t block_group,
1307 						    struct buffer_head ** bh);
1308 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
1309 struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
1310 				      ext4_group_t block_group);
1311 extern unsigned ext4_init_block_bitmap(struct super_block *sb,
1312 				       struct buffer_head *bh,
1313 				       ext4_group_t group,
1314 				       struct ext4_group_desc *desc);
1315 #define ext4_free_blocks_after_init(sb, group, desc)			\
1316 		ext4_init_block_bitmap(sb, NULL, group, desc)
1317 
1318 /* dir.c */
1319 extern int ext4_check_dir_entry(const char *, struct inode *,
1320 				struct ext4_dir_entry_2 *,
1321 				struct buffer_head *, unsigned int);
1322 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
1323 				    __u32 minor_hash,
1324 				    struct ext4_dir_entry_2 *dirent);
1325 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
1326 
1327 /* fsync.c */
1328 extern int ext4_sync_file(struct file *, struct dentry *, int);
1329 
1330 /* hash.c */
1331 extern int ext4fs_dirhash(const char *name, int len, struct
1332 			  dx_hash_info *hinfo);
1333 
1334 /* ialloc.c */
1335 extern struct inode *ext4_new_inode(handle_t *, struct inode *, int,
1336 				    const struct qstr *qstr, __u32 goal);
1337 extern void ext4_free_inode(handle_t *, struct inode *);
1338 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
1339 extern unsigned long ext4_count_free_inodes(struct super_block *);
1340 extern unsigned long ext4_count_dirs(struct super_block *);
1341 extern void ext4_check_inodes_bitmap(struct super_block *);
1342 extern unsigned ext4_init_inode_bitmap(struct super_block *sb,
1343 				       struct buffer_head *bh,
1344 				       ext4_group_t group,
1345 				       struct ext4_group_desc *desc);
1346 extern void mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
1347 
1348 /* mballoc.c */
1349 extern long ext4_mb_stats;
1350 extern long ext4_mb_max_to_scan;
1351 extern int ext4_mb_init(struct super_block *, int);
1352 extern int ext4_mb_release(struct super_block *);
1353 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
1354 				struct ext4_allocation_request *, int *);
1355 extern int ext4_mb_reserve_blocks(struct super_block *, int);
1356 extern void ext4_discard_preallocations(struct inode *);
1357 extern int __init init_ext4_mballoc(void);
1358 extern void exit_ext4_mballoc(void);
1359 extern void ext4_mb_free_blocks(handle_t *, struct inode *,
1360 		ext4_fsblk_t, unsigned long, int, unsigned long *);
1361 extern int ext4_mb_add_groupinfo(struct super_block *sb,
1362 		ext4_group_t i, struct ext4_group_desc *desc);
1363 extern int ext4_mb_get_buddy_cache_lock(struct super_block *, ext4_group_t);
1364 extern void ext4_mb_put_buddy_cache_lock(struct super_block *,
1365 						ext4_group_t, int);
1366 /* inode.c */
1367 int ext4_forget(handle_t *handle, int is_metadata, struct inode *inode,
1368 		struct buffer_head *bh, ext4_fsblk_t blocknr);
1369 struct buffer_head *ext4_getblk(handle_t *, struct inode *,
1370 						ext4_lblk_t, int, int *);
1371 struct buffer_head *ext4_bread(handle_t *, struct inode *,
1372 						ext4_lblk_t, int, int *);
1373 int ext4_get_block(struct inode *inode, sector_t iblock,
1374 				struct buffer_head *bh_result, int create);
1375 
1376 extern struct inode *ext4_iget(struct super_block *, unsigned long);
1377 extern int  ext4_write_inode(struct inode *, int);
1378 extern int  ext4_setattr(struct dentry *, struct iattr *);
1379 extern int  ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
1380 				struct kstat *stat);
1381 extern void ext4_delete_inode(struct inode *);
1382 extern int  ext4_sync_inode(handle_t *, struct inode *);
1383 extern void ext4_dirty_inode(struct inode *);
1384 extern int ext4_change_inode_journal_flag(struct inode *, int);
1385 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
1386 extern int ext4_can_truncate(struct inode *inode);
1387 extern void ext4_truncate(struct inode *);
1388 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
1389 extern void ext4_set_inode_flags(struct inode *);
1390 extern void ext4_get_inode_flags(struct ext4_inode_info *);
1391 extern int ext4_alloc_da_blocks(struct inode *inode);
1392 extern void ext4_set_aops(struct inode *inode);
1393 extern int ext4_writepage_trans_blocks(struct inode *);
1394 extern int ext4_meta_trans_blocks(struct inode *, int nrblocks, int idxblocks);
1395 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
1396 extern int ext4_block_truncate_page(handle_t *handle,
1397 		struct address_space *mapping, loff_t from);
1398 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
1399 extern qsize_t ext4_get_reserved_space(struct inode *inode);
1400 
1401 /* ioctl.c */
1402 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
1403 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
1404 
1405 /* migrate.c */
1406 extern int ext4_ext_migrate(struct inode *);
1407 
1408 /* namei.c */
1409 extern unsigned int ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize);
1410 extern __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize);
1411 extern int ext4_orphan_add(handle_t *, struct inode *);
1412 extern int ext4_orphan_del(handle_t *, struct inode *);
1413 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
1414 				__u32 start_minor_hash, __u32 *next_hash);
1415 
1416 /* resize.c */
1417 extern int ext4_group_add(struct super_block *sb,
1418 				struct ext4_new_group_data *input);
1419 extern int ext4_group_extend(struct super_block *sb,
1420 				struct ext4_super_block *es,
1421 				ext4_fsblk_t n_blocks_count);
1422 
1423 /* super.c */
1424 extern void ext4_error(struct super_block *, const char *, const char *, ...)
1425 	__attribute__ ((format (printf, 3, 4)));
1426 extern void __ext4_std_error(struct super_block *, const char *, int);
1427 extern void ext4_abort(struct super_block *, const char *, const char *, ...)
1428 	__attribute__ ((format (printf, 3, 4)));
1429 extern void ext4_warning(struct super_block *, const char *, const char *, ...)
1430 	__attribute__ ((format (printf, 3, 4)));
1431 extern void ext4_msg(struct super_block *, const char *, const char *, ...)
1432 	__attribute__ ((format (printf, 3, 4)));
1433 extern void ext4_grp_locked_error(struct super_block *, ext4_group_t,
1434 				const char *, const char *, ...)
1435 	__attribute__ ((format (printf, 4, 5)));
1436 extern void ext4_update_dynamic_rev(struct super_block *sb);
1437 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
1438 					__u32 compat);
1439 extern int ext4_update_rocompat_feature(handle_t *handle,
1440 					struct super_block *sb,	__u32 rocompat);
1441 extern int ext4_update_incompat_feature(handle_t *handle,
1442 					struct super_block *sb,	__u32 incompat);
1443 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
1444 				      struct ext4_group_desc *bg);
1445 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
1446 				      struct ext4_group_desc *bg);
1447 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
1448 				     struct ext4_group_desc *bg);
1449 extern __u32 ext4_free_blks_count(struct super_block *sb,
1450 				struct ext4_group_desc *bg);
1451 extern __u32 ext4_free_inodes_count(struct super_block *sb,
1452 				 struct ext4_group_desc *bg);
1453 extern __u32 ext4_used_dirs_count(struct super_block *sb,
1454 				struct ext4_group_desc *bg);
1455 extern __u32 ext4_itable_unused_count(struct super_block *sb,
1456 				   struct ext4_group_desc *bg);
1457 extern void ext4_block_bitmap_set(struct super_block *sb,
1458 				  struct ext4_group_desc *bg, ext4_fsblk_t blk);
1459 extern void ext4_inode_bitmap_set(struct super_block *sb,
1460 				  struct ext4_group_desc *bg, ext4_fsblk_t blk);
1461 extern void ext4_inode_table_set(struct super_block *sb,
1462 				 struct ext4_group_desc *bg, ext4_fsblk_t blk);
1463 extern void ext4_free_blks_set(struct super_block *sb,
1464 			       struct ext4_group_desc *bg, __u32 count);
1465 extern void ext4_free_inodes_set(struct super_block *sb,
1466 				struct ext4_group_desc *bg, __u32 count);
1467 extern void ext4_used_dirs_set(struct super_block *sb,
1468 				struct ext4_group_desc *bg, __u32 count);
1469 extern void ext4_itable_unused_set(struct super_block *sb,
1470 				   struct ext4_group_desc *bg, __u32 count);
1471 extern __le16 ext4_group_desc_csum(struct ext4_sb_info *sbi, __u32 group,
1472 				   struct ext4_group_desc *gdp);
1473 extern int ext4_group_desc_csum_verify(struct ext4_sb_info *sbi, __u32 group,
1474 				       struct ext4_group_desc *gdp);
1475 
1476 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
1477 {
1478 	return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
1479 		le32_to_cpu(es->s_blocks_count_lo);
1480 }
1481 
1482 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
1483 {
1484 	return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
1485 		le32_to_cpu(es->s_r_blocks_count_lo);
1486 }
1487 
1488 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
1489 {
1490 	return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
1491 		le32_to_cpu(es->s_free_blocks_count_lo);
1492 }
1493 
1494 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
1495 					 ext4_fsblk_t blk)
1496 {
1497 	es->s_blocks_count_lo = cpu_to_le32((u32)blk);
1498 	es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
1499 }
1500 
1501 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
1502 					      ext4_fsblk_t blk)
1503 {
1504 	es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
1505 	es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
1506 }
1507 
1508 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
1509 					   ext4_fsblk_t blk)
1510 {
1511 	es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
1512 	es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
1513 }
1514 
1515 static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
1516 {
1517 	if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
1518 		return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
1519 			le32_to_cpu(raw_inode->i_size_lo);
1520 	else
1521 		return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
1522 }
1523 
1524 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
1525 {
1526 	raw_inode->i_size_lo = cpu_to_le32(i_size);
1527 	raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
1528 }
1529 
1530 static inline
1531 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
1532 					    ext4_group_t group)
1533 {
1534 	 struct ext4_group_info ***grp_info;
1535 	 long indexv, indexh;
1536 	 grp_info = EXT4_SB(sb)->s_group_info;
1537 	 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
1538 	 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
1539 	 return grp_info[indexv][indexh];
1540 }
1541 
1542 /*
1543  * Reading s_groups_count requires using smp_rmb() afterwards.  See
1544  * the locking protocol documented in the comments of ext4_group_add()
1545  * in resize.c
1546  */
1547 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
1548 {
1549 	ext4_group_t	ngroups = EXT4_SB(sb)->s_groups_count;
1550 
1551 	smp_rmb();
1552 	return ngroups;
1553 }
1554 
1555 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
1556 					     ext4_group_t block_group)
1557 {
1558 	return block_group >> sbi->s_log_groups_per_flex;
1559 }
1560 
1561 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
1562 {
1563 	return 1 << sbi->s_log_groups_per_flex;
1564 }
1565 
1566 #define ext4_std_error(sb, errno)				\
1567 do {								\
1568 	if ((errno))						\
1569 		__ext4_std_error((sb), __func__, (errno));	\
1570 } while (0)
1571 
1572 #ifdef CONFIG_SMP
1573 /* Each CPU can accumulate percpu_counter_batch blocks in their local
1574  * counters. So we need to make sure we have free blocks more
1575  * than percpu_counter_batch  * nr_cpu_ids. Also add a window of 4 times.
1576  */
1577 #define EXT4_FREEBLOCKS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
1578 #else
1579 #define EXT4_FREEBLOCKS_WATERMARK 0
1580 #endif
1581 
1582 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
1583 {
1584 	/*
1585 	 * XXX: replace with spinlock if seen contended -bzzz
1586 	 */
1587 	down_write(&EXT4_I(inode)->i_data_sem);
1588 	if (newsize > EXT4_I(inode)->i_disksize)
1589 		EXT4_I(inode)->i_disksize = newsize;
1590 	up_write(&EXT4_I(inode)->i_data_sem);
1591 	return ;
1592 }
1593 
1594 struct ext4_group_info {
1595 	unsigned long   bb_state;
1596 	struct rb_root  bb_free_root;
1597 	ext4_grpblk_t	bb_first_free;	/* first free block */
1598 	ext4_grpblk_t	bb_free;	/* total free blocks */
1599 	ext4_grpblk_t	bb_fragments;	/* nr of freespace fragments */
1600 	struct          list_head bb_prealloc_list;
1601 #ifdef DOUBLE_CHECK
1602 	void            *bb_bitmap;
1603 #endif
1604 	struct rw_semaphore alloc_sem;
1605 	ext4_grpblk_t	bb_counters[];	/* Nr of free power-of-two-block
1606 					 * regions, index is order.
1607 					 * bb_counters[3] = 5 means
1608 					 * 5 free 8-block regions. */
1609 };
1610 
1611 #define EXT4_GROUP_INFO_NEED_INIT_BIT	0
1612 
1613 #define EXT4_MB_GRP_NEED_INIT(grp)	\
1614 	(test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
1615 
1616 #define EXT4_MAX_CONTENTION		8
1617 #define EXT4_CONTENTION_THRESHOLD	2
1618 
1619 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
1620 					      ext4_group_t group)
1621 {
1622 	return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
1623 }
1624 
1625 /*
1626  * Returns true if the filesystem is busy enough that attempts to
1627  * access the block group locks has run into contention.
1628  */
1629 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
1630 {
1631 	return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
1632 }
1633 
1634 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
1635 {
1636 	spinlock_t *lock = ext4_group_lock_ptr(sb, group);
1637 	if (spin_trylock(lock))
1638 		/*
1639 		 * We're able to grab the lock right away, so drop the
1640 		 * lock contention counter.
1641 		 */
1642 		atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
1643 	else {
1644 		/*
1645 		 * The lock is busy, so bump the contention counter,
1646 		 * and then wait on the spin lock.
1647 		 */
1648 		atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
1649 				  EXT4_MAX_CONTENTION);
1650 		spin_lock(lock);
1651 	}
1652 }
1653 
1654 static inline void ext4_unlock_group(struct super_block *sb,
1655 					ext4_group_t group)
1656 {
1657 	spin_unlock(ext4_group_lock_ptr(sb, group));
1658 }
1659 
1660 /*
1661  * Inodes and files operations
1662  */
1663 
1664 /* dir.c */
1665 extern const struct file_operations ext4_dir_operations;
1666 
1667 /* file.c */
1668 extern const struct inode_operations ext4_file_inode_operations;
1669 extern const struct file_operations ext4_file_operations;
1670 
1671 /* namei.c */
1672 extern const struct inode_operations ext4_dir_inode_operations;
1673 extern const struct inode_operations ext4_special_inode_operations;
1674 extern struct dentry *ext4_get_parent(struct dentry *child);
1675 
1676 /* symlink.c */
1677 extern const struct inode_operations ext4_symlink_inode_operations;
1678 extern const struct inode_operations ext4_fast_symlink_inode_operations;
1679 
1680 /* block_validity */
1681 extern void ext4_release_system_zone(struct super_block *sb);
1682 extern int ext4_setup_system_zone(struct super_block *sb);
1683 extern int __init init_ext4_system_zone(void);
1684 extern void exit_ext4_system_zone(void);
1685 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
1686 				 ext4_fsblk_t start_blk,
1687 				 unsigned int count);
1688 
1689 /* extents.c */
1690 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
1691 extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
1692 extern int ext4_ext_index_trans_blocks(struct inode *inode, int nrblocks,
1693 				       int chunk);
1694 extern int ext4_ext_get_blocks(handle_t *handle, struct inode *inode,
1695 			       ext4_lblk_t iblock, unsigned int max_blocks,
1696 			       struct buffer_head *bh_result, int flags);
1697 extern void ext4_ext_truncate(struct inode *);
1698 extern void ext4_ext_init(struct super_block *);
1699 extern void ext4_ext_release(struct super_block *);
1700 extern long ext4_fallocate(struct inode *inode, int mode, loff_t offset,
1701 			  loff_t len);
1702 extern int ext4_get_blocks(handle_t *handle, struct inode *inode,
1703 			   sector_t block, unsigned int max_blocks,
1704 			   struct buffer_head *bh, int flags);
1705 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
1706 			__u64 start, __u64 len);
1707 /* move_extent.c */
1708 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
1709 			     __u64 start_orig, __u64 start_donor,
1710 			     __u64 len, __u64 *moved_len);
1711 
1712 
1713 /*
1714  * Add new method to test wether block and inode bitmaps are properly
1715  * initialized. With uninit_bg reading the block from disk is not enough
1716  * to mark the bitmap uptodate. We need to also zero-out the bitmap
1717  */
1718 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
1719 
1720 static inline int bitmap_uptodate(struct buffer_head *bh)
1721 {
1722 	return (buffer_uptodate(bh) &&
1723 			test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
1724 }
1725 static inline void set_bitmap_uptodate(struct buffer_head *bh)
1726 {
1727 	set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
1728 }
1729 
1730 #endif	/* __KERNEL__ */
1731 
1732 #endif	/* _EXT4_H */
1733