xref: /openbmc/linux/fs/ext4/ext4.h (revision ca0faba0)
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		1
71 /* blocks already reserved */
72 #define EXT4_MB_HINT_RESERVED		2
73 /* metadata is being allocated */
74 #define EXT4_MB_HINT_METADATA		4
75 /* first blocks in the file */
76 #define EXT4_MB_HINT_FIRST		8
77 /* search for the best chunk */
78 #define EXT4_MB_HINT_BEST		16
79 /* data is being allocated */
80 #define EXT4_MB_HINT_DATA		32
81 /* don't preallocate (for tails) */
82 #define EXT4_MB_HINT_NOPREALLOC		64
83 /* allocate for locality group */
84 #define EXT4_MB_HINT_GROUP_ALLOC	128
85 /* allocate goal blocks or none */
86 #define EXT4_MB_HINT_GOAL_ONLY		256
87 /* goal is meaningful */
88 #define EXT4_MB_HINT_TRY_GOAL		512
89 /* blocks already pre-reserved by delayed allocation */
90 #define EXT4_MB_DELALLOC_RESERVED      1024
91 
92 
93 struct ext4_allocation_request {
94 	/* target inode for block we're allocating */
95 	struct inode *inode;
96 	/* logical block in target inode */
97 	ext4_lblk_t logical;
98 	/* phys. target (a hint) */
99 	ext4_fsblk_t goal;
100 	/* the closest logical allocated block to the left */
101 	ext4_lblk_t lleft;
102 	/* phys. block for ^^^ */
103 	ext4_fsblk_t pleft;
104 	/* the closest logical allocated block to the right */
105 	ext4_lblk_t lright;
106 	/* phys. block for ^^^ */
107 	ext4_fsblk_t pright;
108 	/* how many blocks we want to allocate */
109 	unsigned int len;
110 	/* flags. see above EXT4_MB_HINT_* */
111 	unsigned int flags;
112 };
113 
114 /*
115  * Special inodes numbers
116  */
117 #define	EXT4_BAD_INO		 1	/* Bad blocks inode */
118 #define EXT4_ROOT_INO		 2	/* Root inode */
119 #define EXT4_BOOT_LOADER_INO	 5	/* Boot loader inode */
120 #define EXT4_UNDEL_DIR_INO	 6	/* Undelete directory inode */
121 #define EXT4_RESIZE_INO		 7	/* Reserved group descriptors inode */
122 #define EXT4_JOURNAL_INO	 8	/* Journal inode */
123 
124 /* First non-reserved inode for old ext4 filesystems */
125 #define EXT4_GOOD_OLD_FIRST_INO	11
126 
127 /*
128  * Maximal count of links to a file
129  */
130 #define EXT4_LINK_MAX		65000
131 
132 /*
133  * Macro-instructions used to manage several block sizes
134  */
135 #define EXT4_MIN_BLOCK_SIZE		1024
136 #define	EXT4_MAX_BLOCK_SIZE		65536
137 #define EXT4_MIN_BLOCK_LOG_SIZE		10
138 #ifdef __KERNEL__
139 # define EXT4_BLOCK_SIZE(s)		((s)->s_blocksize)
140 #else
141 # define EXT4_BLOCK_SIZE(s)		(EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
142 #endif
143 #define	EXT4_ADDR_PER_BLOCK(s)		(EXT4_BLOCK_SIZE(s) / sizeof(__u32))
144 #ifdef __KERNEL__
145 # define EXT4_BLOCK_SIZE_BITS(s)	((s)->s_blocksize_bits)
146 #else
147 # define EXT4_BLOCK_SIZE_BITS(s)	((s)->s_log_block_size + 10)
148 #endif
149 #ifdef __KERNEL__
150 #define	EXT4_ADDR_PER_BLOCK_BITS(s)	(EXT4_SB(s)->s_addr_per_block_bits)
151 #define EXT4_INODE_SIZE(s)		(EXT4_SB(s)->s_inode_size)
152 #define EXT4_FIRST_INO(s)		(EXT4_SB(s)->s_first_ino)
153 #else
154 #define EXT4_INODE_SIZE(s)	(((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
155 				 EXT4_GOOD_OLD_INODE_SIZE : \
156 				 (s)->s_inode_size)
157 #define EXT4_FIRST_INO(s)	(((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
158 				 EXT4_GOOD_OLD_FIRST_INO : \
159 				 (s)->s_first_ino)
160 #endif
161 #define EXT4_BLOCK_ALIGN(size, blkbits)		ALIGN((size), (1 << (blkbits)))
162 
163 /*
164  * Structure of a blocks group descriptor
165  */
166 struct ext4_group_desc
167 {
168 	__le32	bg_block_bitmap_lo;	/* Blocks bitmap block */
169 	__le32	bg_inode_bitmap_lo;	/* Inodes bitmap block */
170 	__le32	bg_inode_table_lo;	/* Inodes table block */
171 	__le16	bg_free_blocks_count_lo;/* Free blocks count */
172 	__le16	bg_free_inodes_count_lo;/* Free inodes count */
173 	__le16	bg_used_dirs_count_lo;	/* Directories count */
174 	__le16	bg_flags;		/* EXT4_BG_flags (INODE_UNINIT, etc) */
175 	__u32	bg_reserved[2];		/* Likely block/inode bitmap checksum */
176 	__le16  bg_itable_unused_lo;	/* Unused inodes count */
177 	__le16  bg_checksum;		/* crc16(sb_uuid+group+desc) */
178 	__le32	bg_block_bitmap_hi;	/* Blocks bitmap block MSB */
179 	__le32	bg_inode_bitmap_hi;	/* Inodes bitmap block MSB */
180 	__le32	bg_inode_table_hi;	/* Inodes table block MSB */
181 	__le16	bg_free_blocks_count_hi;/* Free blocks count MSB */
182 	__le16	bg_free_inodes_count_hi;/* Free inodes count MSB */
183 	__le16	bg_used_dirs_count_hi;	/* Directories count MSB */
184 	__le16  bg_itable_unused_hi;    /* Unused inodes count MSB */
185 	__u32	bg_reserved2[3];
186 };
187 
188 /*
189  * Structure of a flex block group info
190  */
191 
192 struct flex_groups {
193 	atomic_t free_inodes;
194 	atomic_t free_blocks;
195 	atomic_t used_dirs;
196 };
197 
198 #define EXT4_BG_INODE_UNINIT	0x0001 /* Inode table/bitmap not in use */
199 #define EXT4_BG_BLOCK_UNINIT	0x0002 /* Block bitmap not in use */
200 #define EXT4_BG_INODE_ZEROED	0x0004 /* On-disk itable initialized to zero */
201 
202 /*
203  * Macro-instructions used to manage group descriptors
204  */
205 #define EXT4_MIN_DESC_SIZE		32
206 #define EXT4_MIN_DESC_SIZE_64BIT	64
207 #define	EXT4_MAX_DESC_SIZE		EXT4_MIN_BLOCK_SIZE
208 #define EXT4_DESC_SIZE(s)		(EXT4_SB(s)->s_desc_size)
209 #ifdef __KERNEL__
210 # define EXT4_BLOCKS_PER_GROUP(s)	(EXT4_SB(s)->s_blocks_per_group)
211 # define EXT4_DESC_PER_BLOCK(s)		(EXT4_SB(s)->s_desc_per_block)
212 # define EXT4_INODES_PER_GROUP(s)	(EXT4_SB(s)->s_inodes_per_group)
213 # define EXT4_DESC_PER_BLOCK_BITS(s)	(EXT4_SB(s)->s_desc_per_block_bits)
214 #else
215 # define EXT4_BLOCKS_PER_GROUP(s)	((s)->s_blocks_per_group)
216 # define EXT4_DESC_PER_BLOCK(s)		(EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
217 # define EXT4_INODES_PER_GROUP(s)	((s)->s_inodes_per_group)
218 #endif
219 
220 /*
221  * Constants relative to the data blocks
222  */
223 #define	EXT4_NDIR_BLOCKS		12
224 #define	EXT4_IND_BLOCK			EXT4_NDIR_BLOCKS
225 #define	EXT4_DIND_BLOCK			(EXT4_IND_BLOCK + 1)
226 #define	EXT4_TIND_BLOCK			(EXT4_DIND_BLOCK + 1)
227 #define	EXT4_N_BLOCKS			(EXT4_TIND_BLOCK + 1)
228 
229 /*
230  * Inode flags
231  */
232 #define	EXT4_SECRM_FL			0x00000001 /* Secure deletion */
233 #define	EXT4_UNRM_FL			0x00000002 /* Undelete */
234 #define	EXT4_COMPR_FL			0x00000004 /* Compress file */
235 #define EXT4_SYNC_FL			0x00000008 /* Synchronous updates */
236 #define EXT4_IMMUTABLE_FL		0x00000010 /* Immutable file */
237 #define EXT4_APPEND_FL			0x00000020 /* writes to file may only append */
238 #define EXT4_NODUMP_FL			0x00000040 /* do not dump file */
239 #define EXT4_NOATIME_FL			0x00000080 /* do not update atime */
240 /* Reserved for compression usage... */
241 #define EXT4_DIRTY_FL			0x00000100
242 #define EXT4_COMPRBLK_FL		0x00000200 /* One or more compressed clusters */
243 #define EXT4_NOCOMPR_FL			0x00000400 /* Don't compress */
244 #define EXT4_ECOMPR_FL			0x00000800 /* Compression error */
245 /* End compression flags --- maybe not all used */
246 #define EXT4_INDEX_FL			0x00001000 /* hash-indexed directory */
247 #define EXT4_IMAGIC_FL			0x00002000 /* AFS directory */
248 #define EXT4_JOURNAL_DATA_FL		0x00004000 /* file data should be journaled */
249 #define EXT4_NOTAIL_FL			0x00008000 /* file tail should not be merged */
250 #define EXT4_DIRSYNC_FL			0x00010000 /* dirsync behaviour (directories only) */
251 #define EXT4_TOPDIR_FL			0x00020000 /* Top of directory hierarchies*/
252 #define EXT4_HUGE_FILE_FL               0x00040000 /* Set to each huge file */
253 #define EXT4_EXTENTS_FL			0x00080000 /* Inode uses extents */
254 #define EXT4_EXT_MIGRATE		0x00100000 /* Inode is migrating */
255 #define EXT4_RESERVED_FL		0x80000000 /* reserved for ext4 lib */
256 
257 #define EXT4_FL_USER_VISIBLE		0x000BDFFF /* User visible flags */
258 #define EXT4_FL_USER_MODIFIABLE		0x000B80FF /* User modifiable flags */
259 
260 /* Flags that should be inherited by new inodes from their parent. */
261 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
262 			   EXT4_SYNC_FL | EXT4_IMMUTABLE_FL | EXT4_APPEND_FL |\
263 			   EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
264 			   EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
265 			   EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
266 
267 /* Flags that are appropriate for regular files (all but dir-specific ones). */
268 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
269 
270 /* Flags that are appropriate for non-directories/regular files. */
271 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
272 
273 /* Mask out flags that are inappropriate for the given type of inode. */
274 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
275 {
276 	if (S_ISDIR(mode))
277 		return flags;
278 	else if (S_ISREG(mode))
279 		return flags & EXT4_REG_FLMASK;
280 	else
281 		return flags & EXT4_OTHER_FLMASK;
282 }
283 
284 /*
285  * Inode dynamic state flags
286  */
287 #define EXT4_STATE_JDATA		0x00000001 /* journaled data exists */
288 #define EXT4_STATE_NEW			0x00000002 /* inode is newly created */
289 #define EXT4_STATE_XATTR		0x00000004 /* has in-inode xattrs */
290 #define EXT4_STATE_NO_EXPAND		0x00000008 /* No space for expansion */
291 #define EXT4_STATE_DA_ALLOC_CLOSE	0x00000010 /* Alloc DA blks on close */
292 
293 /* Used to pass group descriptor data when online resize is done */
294 struct ext4_new_group_input {
295 	__u32 group;		/* Group number for this data */
296 	__u64 block_bitmap;	/* Absolute block number of block bitmap */
297 	__u64 inode_bitmap;	/* Absolute block number of inode bitmap */
298 	__u64 inode_table;	/* Absolute block number of inode table start */
299 	__u32 blocks_count;	/* Total number of blocks in this group */
300 	__u16 reserved_blocks;	/* Number of reserved blocks in this group */
301 	__u16 unused;
302 };
303 
304 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
305 struct ext4_new_group_data {
306 	__u32 group;
307 	__u64 block_bitmap;
308 	__u64 inode_bitmap;
309 	__u64 inode_table;
310 	__u32 blocks_count;
311 	__u16 reserved_blocks;
312 	__u16 unused;
313 	__u32 free_blocks_count;
314 };
315 
316 /*
317  * Following is used by preallocation code to tell get_blocks() that we
318  * want uninitialzed extents.
319  */
320 #define EXT4_CREATE_UNINITIALIZED_EXT		2
321 
322 /*
323  * ioctl commands
324  */
325 #define	EXT4_IOC_GETFLAGS		FS_IOC_GETFLAGS
326 #define	EXT4_IOC_SETFLAGS		FS_IOC_SETFLAGS
327 #define	EXT4_IOC_GETVERSION		_IOR('f', 3, long)
328 #define	EXT4_IOC_SETVERSION		_IOW('f', 4, long)
329 #define	EXT4_IOC_GETVERSION_OLD		FS_IOC_GETVERSION
330 #define	EXT4_IOC_SETVERSION_OLD		FS_IOC_SETVERSION
331 #ifdef CONFIG_JBD2_DEBUG
332 #define EXT4_IOC_WAIT_FOR_READONLY	_IOR('f', 99, long)
333 #endif
334 #define EXT4_IOC_GETRSVSZ		_IOR('f', 5, long)
335 #define EXT4_IOC_SETRSVSZ		_IOW('f', 6, long)
336 #define EXT4_IOC_GROUP_EXTEND		_IOW('f', 7, unsigned long)
337 #define EXT4_IOC_GROUP_ADD		_IOW('f', 8, struct ext4_new_group_input)
338 #define EXT4_IOC_MIGRATE		_IO('f', 9)
339  /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
340  /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
341 #define EXT4_IOC_ALLOC_DA_BLKS		_IO('f', 12)
342 
343 /*
344  * ioctl commands in 32 bit emulation
345  */
346 #define EXT4_IOC32_GETFLAGS		FS_IOC32_GETFLAGS
347 #define EXT4_IOC32_SETFLAGS		FS_IOC32_SETFLAGS
348 #define EXT4_IOC32_GETVERSION		_IOR('f', 3, int)
349 #define EXT4_IOC32_SETVERSION		_IOW('f', 4, int)
350 #define EXT4_IOC32_GETRSVSZ		_IOR('f', 5, int)
351 #define EXT4_IOC32_SETRSVSZ		_IOW('f', 6, int)
352 #define EXT4_IOC32_GROUP_EXTEND		_IOW('f', 7, unsigned int)
353 #ifdef CONFIG_JBD2_DEBUG
354 #define EXT4_IOC32_WAIT_FOR_READONLY	_IOR('f', 99, int)
355 #endif
356 #define EXT4_IOC32_GETVERSION_OLD	FS_IOC32_GETVERSION
357 #define EXT4_IOC32_SETVERSION_OLD	FS_IOC32_SETVERSION
358 
359 
360 /*
361  *  Mount options
362  */
363 struct ext4_mount_options {
364 	unsigned long s_mount_opt;
365 	uid_t s_resuid;
366 	gid_t s_resgid;
367 	unsigned long s_commit_interval;
368 	u32 s_min_batch_time, s_max_batch_time;
369 #ifdef CONFIG_QUOTA
370 	int s_jquota_fmt;
371 	char *s_qf_names[MAXQUOTAS];
372 #endif
373 };
374 
375 /*
376  * Structure of an inode on the disk
377  */
378 struct ext4_inode {
379 	__le16	i_mode;		/* File mode */
380 	__le16	i_uid;		/* Low 16 bits of Owner Uid */
381 	__le32	i_size_lo;	/* Size in bytes */
382 	__le32	i_atime;	/* Access time */
383 	__le32	i_ctime;	/* Inode Change time */
384 	__le32	i_mtime;	/* Modification time */
385 	__le32	i_dtime;	/* Deletion Time */
386 	__le16	i_gid;		/* Low 16 bits of Group Id */
387 	__le16	i_links_count;	/* Links count */
388 	__le32	i_blocks_lo;	/* Blocks count */
389 	__le32	i_flags;	/* File flags */
390 	union {
391 		struct {
392 			__le32  l_i_version;
393 		} linux1;
394 		struct {
395 			__u32  h_i_translator;
396 		} hurd1;
397 		struct {
398 			__u32  m_i_reserved1;
399 		} masix1;
400 	} osd1;				/* OS dependent 1 */
401 	__le32	i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
402 	__le32	i_generation;	/* File version (for NFS) */
403 	__le32	i_file_acl_lo;	/* File ACL */
404 	__le32	i_size_high;
405 	__le32	i_obso_faddr;	/* Obsoleted fragment address */
406 	union {
407 		struct {
408 			__le16	l_i_blocks_high; /* were l_i_reserved1 */
409 			__le16	l_i_file_acl_high;
410 			__le16	l_i_uid_high;	/* these 2 fields */
411 			__le16	l_i_gid_high;	/* were reserved2[0] */
412 			__u32	l_i_reserved2;
413 		} linux2;
414 		struct {
415 			__le16	h_i_reserved1;	/* Obsoleted fragment number/size which are removed in ext4 */
416 			__u16	h_i_mode_high;
417 			__u16	h_i_uid_high;
418 			__u16	h_i_gid_high;
419 			__u32	h_i_author;
420 		} hurd2;
421 		struct {
422 			__le16	h_i_reserved1;	/* Obsoleted fragment number/size which are removed in ext4 */
423 			__le16	m_i_file_acl_high;
424 			__u32	m_i_reserved2[2];
425 		} masix2;
426 	} osd2;				/* OS dependent 2 */
427 	__le16	i_extra_isize;
428 	__le16	i_pad1;
429 	__le32  i_ctime_extra;  /* extra Change time      (nsec << 2 | epoch) */
430 	__le32  i_mtime_extra;  /* extra Modification time(nsec << 2 | epoch) */
431 	__le32  i_atime_extra;  /* extra Access time      (nsec << 2 | epoch) */
432 	__le32  i_crtime;       /* File Creation time */
433 	__le32  i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
434 	__le32  i_version_hi;	/* high 32 bits for 64-bit version */
435 };
436 
437 
438 #define EXT4_EPOCH_BITS 2
439 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
440 #define EXT4_NSEC_MASK  (~0UL << EXT4_EPOCH_BITS)
441 
442 /*
443  * Extended fields will fit into an inode if the filesystem was formatted
444  * with large inodes (-I 256 or larger) and there are not currently any EAs
445  * consuming all of the available space. For new inodes we always reserve
446  * enough space for the kernel's known extended fields, but for inodes
447  * created with an old kernel this might not have been the case. None of
448  * the extended inode fields is critical for correct filesystem operation.
449  * This macro checks if a certain field fits in the inode. Note that
450  * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
451  */
452 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field)	\
453 	((offsetof(typeof(*ext4_inode), field) +	\
454 	  sizeof((ext4_inode)->field))			\
455 	<= (EXT4_GOOD_OLD_INODE_SIZE +			\
456 	    (einode)->i_extra_isize))			\
457 
458 static inline __le32 ext4_encode_extra_time(struct timespec *time)
459 {
460        return cpu_to_le32((sizeof(time->tv_sec) > 4 ?
461 			   time->tv_sec >> 32 : 0) |
462 			   ((time->tv_nsec << 2) & EXT4_NSEC_MASK));
463 }
464 
465 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
466 {
467        if (sizeof(time->tv_sec) > 4)
468 	       time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK)
469 			       << 32;
470        time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> 2;
471 }
472 
473 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode)			       \
474 do {									       \
475 	(raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec);	       \
476 	if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     \
477 		(raw_inode)->xtime ## _extra =				       \
478 				ext4_encode_extra_time(&(inode)->xtime);       \
479 } while (0)
480 
481 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode)			       \
482 do {									       \
483 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))		       \
484 		(raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec);      \
485 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))	       \
486 		(raw_inode)->xtime ## _extra =				       \
487 				ext4_encode_extra_time(&(einode)->xtime);      \
488 } while (0)
489 
490 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode)			       \
491 do {									       \
492 	(inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime);       \
493 	if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     \
494 		ext4_decode_extra_time(&(inode)->xtime,			       \
495 				       raw_inode->xtime ## _extra);	       \
496 } while (0)
497 
498 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode)			       \
499 do {									       \
500 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))		       \
501 		(einode)->xtime.tv_sec = 				       \
502 			(signed)le32_to_cpu((raw_inode)->xtime);	       \
503 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))	       \
504 		ext4_decode_extra_time(&(einode)->xtime,		       \
505 				       raw_inode->xtime ## _extra);	       \
506 } while (0)
507 
508 #define i_disk_version osd1.linux1.l_i_version
509 
510 #if defined(__KERNEL__) || defined(__linux__)
511 #define i_reserved1	osd1.linux1.l_i_reserved1
512 #define i_file_acl_high	osd2.linux2.l_i_file_acl_high
513 #define i_blocks_high	osd2.linux2.l_i_blocks_high
514 #define i_uid_low	i_uid
515 #define i_gid_low	i_gid
516 #define i_uid_high	osd2.linux2.l_i_uid_high
517 #define i_gid_high	osd2.linux2.l_i_gid_high
518 #define i_reserved2	osd2.linux2.l_i_reserved2
519 
520 #elif defined(__GNU__)
521 
522 #define i_translator	osd1.hurd1.h_i_translator
523 #define i_uid_high	osd2.hurd2.h_i_uid_high
524 #define i_gid_high	osd2.hurd2.h_i_gid_high
525 #define i_author	osd2.hurd2.h_i_author
526 
527 #elif defined(__masix__)
528 
529 #define i_reserved1	osd1.masix1.m_i_reserved1
530 #define i_file_acl_high	osd2.masix2.m_i_file_acl_high
531 #define i_reserved2	osd2.masix2.m_i_reserved2
532 
533 #endif /* defined(__KERNEL__) || defined(__linux__) */
534 
535 /*
536  * storage for cached extent
537  */
538 struct ext4_ext_cache {
539 	ext4_fsblk_t	ec_start;
540 	ext4_lblk_t	ec_block;
541 	__u32		ec_len; /* must be 32bit to return holes */
542 	__u32		ec_type;
543 };
544 
545 /*
546  * fourth extended file system inode data in memory
547  */
548 struct ext4_inode_info {
549 	__le32	i_data[15];	/* unconverted */
550 	__u32	i_flags;
551 	ext4_fsblk_t	i_file_acl;
552 	__u32	i_dtime;
553 
554 	/*
555 	 * i_block_group is the number of the block group which contains
556 	 * this file's inode.  Constant across the lifetime of the inode,
557 	 * it is ued for making block allocation decisions - we try to
558 	 * place a file's data blocks near its inode block, and new inodes
559 	 * near to their parent directory's inode.
560 	 */
561 	ext4_group_t	i_block_group;
562 	__u32	i_state;		/* Dynamic state flags for ext4 */
563 
564 	ext4_lblk_t		i_dir_start_lookup;
565 #ifdef CONFIG_EXT4_FS_XATTR
566 	/*
567 	 * Extended attributes can be read independently of the main file
568 	 * data. Taking i_mutex even when reading would cause contention
569 	 * between readers of EAs and writers of regular file data, so
570 	 * instead we synchronize on xattr_sem when reading or changing
571 	 * EAs.
572 	 */
573 	struct rw_semaphore xattr_sem;
574 #endif
575 #ifdef CONFIG_EXT4_FS_POSIX_ACL
576 	struct posix_acl	*i_acl;
577 	struct posix_acl	*i_default_acl;
578 #endif
579 
580 	struct list_head i_orphan;	/* unlinked but open inodes */
581 
582 	/*
583 	 * i_disksize keeps track of what the inode size is ON DISK, not
584 	 * in memory.  During truncate, i_size is set to the new size by
585 	 * the VFS prior to calling ext4_truncate(), but the filesystem won't
586 	 * set i_disksize to 0 until the truncate is actually under way.
587 	 *
588 	 * The intent is that i_disksize always represents the blocks which
589 	 * are used by this file.  This allows recovery to restart truncate
590 	 * on orphans if we crash during truncate.  We actually write i_disksize
591 	 * into the on-disk inode when writing inodes out, instead of i_size.
592 	 *
593 	 * The only time when i_disksize and i_size may be different is when
594 	 * a truncate is in progress.  The only things which change i_disksize
595 	 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
596 	 */
597 	loff_t	i_disksize;
598 
599 	/*
600 	 * i_data_sem is for serialising ext4_truncate() against
601 	 * ext4_getblock().  In the 2.4 ext2 design, great chunks of inode's
602 	 * data tree are chopped off during truncate. We can't do that in
603 	 * ext4 because whenever we perform intermediate commits during
604 	 * truncate, the inode and all the metadata blocks *must* be in a
605 	 * consistent state which allows truncation of the orphans to restart
606 	 * during recovery.  Hence we must fix the get_block-vs-truncate race
607 	 * by other means, so we have i_data_sem.
608 	 */
609 	struct rw_semaphore i_data_sem;
610 	struct inode vfs_inode;
611 	struct jbd2_inode jinode;
612 
613 	struct ext4_ext_cache i_cached_extent;
614 	/*
615 	 * File creation time. Its function is same as that of
616 	 * struct timespec i_{a,c,m}time in the generic inode.
617 	 */
618 	struct timespec i_crtime;
619 
620 	/* mballoc */
621 	struct list_head i_prealloc_list;
622 	spinlock_t i_prealloc_lock;
623 
624 	/* ialloc */
625 	ext4_group_t	i_last_alloc_group;
626 
627 	/* allocation reservation info for delalloc */
628 	unsigned int i_reserved_data_blocks;
629 	unsigned int i_reserved_meta_blocks;
630 	unsigned int i_allocated_meta_blocks;
631 	unsigned short i_delalloc_reserved_flag;
632 
633 	/* on-disk additional length */
634 	__u16 i_extra_isize;
635 
636 	spinlock_t i_block_reservation_lock;
637 };
638 
639 /*
640  * File system states
641  */
642 #define	EXT4_VALID_FS			0x0001	/* Unmounted cleanly */
643 #define	EXT4_ERROR_FS			0x0002	/* Errors detected */
644 #define	EXT4_ORPHAN_FS			0x0004	/* Orphans being recovered */
645 
646 /*
647  * Misc. filesystem flags
648  */
649 #define EXT2_FLAGS_SIGNED_HASH		0x0001  /* Signed dirhash in use */
650 #define EXT2_FLAGS_UNSIGNED_HASH	0x0002  /* Unsigned dirhash in use */
651 #define EXT2_FLAGS_TEST_FILESYS		0x0004	/* to test development code */
652 
653 /*
654  * Mount flags
655  */
656 #define EXT4_MOUNT_OLDALLOC		0x00002  /* Don't use the new Orlov allocator */
657 #define EXT4_MOUNT_GRPID		0x00004	/* Create files with directory's group */
658 #define EXT4_MOUNT_DEBUG		0x00008	/* Some debugging messages */
659 #define EXT4_MOUNT_ERRORS_CONT		0x00010	/* Continue on errors */
660 #define EXT4_MOUNT_ERRORS_RO		0x00020	/* Remount fs ro on errors */
661 #define EXT4_MOUNT_ERRORS_PANIC		0x00040	/* Panic on errors */
662 #define EXT4_MOUNT_MINIX_DF		0x00080	/* Mimics the Minix statfs */
663 #define EXT4_MOUNT_NOLOAD		0x00100	/* Don't use existing journal*/
664 #define EXT4_MOUNT_ABORT		0x00200	/* Fatal error detected */
665 #define EXT4_MOUNT_DATA_FLAGS		0x00C00	/* Mode for data writes: */
666 #define EXT4_MOUNT_JOURNAL_DATA		0x00400	/* Write data to journal */
667 #define EXT4_MOUNT_ORDERED_DATA		0x00800	/* Flush data before commit */
668 #define EXT4_MOUNT_WRITEBACK_DATA	0x00C00	/* No data ordering */
669 #define EXT4_MOUNT_UPDATE_JOURNAL	0x01000	/* Update the journal format */
670 #define EXT4_MOUNT_NO_UID32		0x02000  /* Disable 32-bit UIDs */
671 #define EXT4_MOUNT_XATTR_USER		0x04000	/* Extended user attributes */
672 #define EXT4_MOUNT_POSIX_ACL		0x08000	/* POSIX Access Control Lists */
673 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC	0x10000	/* No auto delalloc mapping */
674 #define EXT4_MOUNT_BARRIER		0x20000 /* Use block barriers */
675 #define EXT4_MOUNT_NOBH			0x40000 /* No bufferheads */
676 #define EXT4_MOUNT_QUOTA		0x80000 /* Some quota option set */
677 #define EXT4_MOUNT_USRQUOTA		0x100000 /* "old" user quota */
678 #define EXT4_MOUNT_GRPQUOTA		0x200000 /* "old" group quota */
679 #define EXT4_MOUNT_JOURNAL_CHECKSUM	0x800000 /* Journal checksums */
680 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT	0x1000000 /* Journal Async Commit */
681 #define EXT4_MOUNT_I_VERSION            0x2000000 /* i_version support */
682 #define EXT4_MOUNT_DELALLOC		0x8000000 /* Delalloc support */
683 #define EXT4_MOUNT_DATA_ERR_ABORT	0x10000000 /* Abort on file data write */
684 
685 /* Compatibility, for having both ext2_fs.h and ext4_fs.h included at once */
686 #ifndef _LINUX_EXT2_FS_H
687 #define clear_opt(o, opt)		o &= ~EXT4_MOUNT_##opt
688 #define set_opt(o, opt)			o |= EXT4_MOUNT_##opt
689 #define test_opt(sb, opt)		(EXT4_SB(sb)->s_mount_opt & \
690 					 EXT4_MOUNT_##opt)
691 #else
692 #define EXT2_MOUNT_NOLOAD		EXT4_MOUNT_NOLOAD
693 #define EXT2_MOUNT_ABORT		EXT4_MOUNT_ABORT
694 #define EXT2_MOUNT_DATA_FLAGS		EXT4_MOUNT_DATA_FLAGS
695 #endif
696 
697 #define ext4_set_bit			ext2_set_bit
698 #define ext4_set_bit_atomic		ext2_set_bit_atomic
699 #define ext4_clear_bit			ext2_clear_bit
700 #define ext4_clear_bit_atomic		ext2_clear_bit_atomic
701 #define ext4_test_bit			ext2_test_bit
702 #define ext4_find_first_zero_bit	ext2_find_first_zero_bit
703 #define ext4_find_next_zero_bit		ext2_find_next_zero_bit
704 #define ext4_find_next_bit		ext2_find_next_bit
705 
706 /*
707  * Maximal mount counts between two filesystem checks
708  */
709 #define EXT4_DFL_MAX_MNT_COUNT		20	/* Allow 20 mounts */
710 #define EXT4_DFL_CHECKINTERVAL		0	/* Don't use interval check */
711 
712 /*
713  * Behaviour when detecting errors
714  */
715 #define EXT4_ERRORS_CONTINUE		1	/* Continue execution */
716 #define EXT4_ERRORS_RO			2	/* Remount fs read-only */
717 #define EXT4_ERRORS_PANIC		3	/* Panic */
718 #define EXT4_ERRORS_DEFAULT		EXT4_ERRORS_CONTINUE
719 
720 /*
721  * Structure of the super block
722  */
723 struct ext4_super_block {
724 /*00*/	__le32	s_inodes_count;		/* Inodes count */
725 	__le32	s_blocks_count_lo;	/* Blocks count */
726 	__le32	s_r_blocks_count_lo;	/* Reserved blocks count */
727 	__le32	s_free_blocks_count_lo;	/* Free blocks count */
728 /*10*/	__le32	s_free_inodes_count;	/* Free inodes count */
729 	__le32	s_first_data_block;	/* First Data Block */
730 	__le32	s_log_block_size;	/* Block size */
731 	__le32	s_obso_log_frag_size;	/* Obsoleted fragment size */
732 /*20*/	__le32	s_blocks_per_group;	/* # Blocks per group */
733 	__le32	s_obso_frags_per_group;	/* Obsoleted fragments per group */
734 	__le32	s_inodes_per_group;	/* # Inodes per group */
735 	__le32	s_mtime;		/* Mount time */
736 /*30*/	__le32	s_wtime;		/* Write time */
737 	__le16	s_mnt_count;		/* Mount count */
738 	__le16	s_max_mnt_count;	/* Maximal mount count */
739 	__le16	s_magic;		/* Magic signature */
740 	__le16	s_state;		/* File system state */
741 	__le16	s_errors;		/* Behaviour when detecting errors */
742 	__le16	s_minor_rev_level;	/* minor revision level */
743 /*40*/	__le32	s_lastcheck;		/* time of last check */
744 	__le32	s_checkinterval;	/* max. time between checks */
745 	__le32	s_creator_os;		/* OS */
746 	__le32	s_rev_level;		/* Revision level */
747 /*50*/	__le16	s_def_resuid;		/* Default uid for reserved blocks */
748 	__le16	s_def_resgid;		/* Default gid for reserved blocks */
749 	/*
750 	 * These fields are for EXT4_DYNAMIC_REV superblocks only.
751 	 *
752 	 * Note: the difference between the compatible feature set and
753 	 * the incompatible feature set is that if there is a bit set
754 	 * in the incompatible feature set that the kernel doesn't
755 	 * know about, it should refuse to mount the filesystem.
756 	 *
757 	 * e2fsck's requirements are more strict; if it doesn't know
758 	 * about a feature in either the compatible or incompatible
759 	 * feature set, it must abort and not try to meddle with
760 	 * things it doesn't understand...
761 	 */
762 	__le32	s_first_ino;		/* First non-reserved inode */
763 	__le16  s_inode_size;		/* size of inode structure */
764 	__le16	s_block_group_nr;	/* block group # of this superblock */
765 	__le32	s_feature_compat;	/* compatible feature set */
766 /*60*/	__le32	s_feature_incompat;	/* incompatible feature set */
767 	__le32	s_feature_ro_compat;	/* readonly-compatible feature set */
768 /*68*/	__u8	s_uuid[16];		/* 128-bit uuid for volume */
769 /*78*/	char	s_volume_name[16];	/* volume name */
770 /*88*/	char	s_last_mounted[64];	/* directory where last mounted */
771 /*C8*/	__le32	s_algorithm_usage_bitmap; /* For compression */
772 	/*
773 	 * Performance hints.  Directory preallocation should only
774 	 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
775 	 */
776 	__u8	s_prealloc_blocks;	/* Nr of blocks to try to preallocate*/
777 	__u8	s_prealloc_dir_blocks;	/* Nr to preallocate for dirs */
778 	__le16	s_reserved_gdt_blocks;	/* Per group desc for online growth */
779 	/*
780 	 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
781 	 */
782 /*D0*/	__u8	s_journal_uuid[16];	/* uuid of journal superblock */
783 /*E0*/	__le32	s_journal_inum;		/* inode number of journal file */
784 	__le32	s_journal_dev;		/* device number of journal file */
785 	__le32	s_last_orphan;		/* start of list of inodes to delete */
786 	__le32	s_hash_seed[4];		/* HTREE hash seed */
787 	__u8	s_def_hash_version;	/* Default hash version to use */
788 	__u8	s_reserved_char_pad;
789 	__le16  s_desc_size;		/* size of group descriptor */
790 /*100*/	__le32	s_default_mount_opts;
791 	__le32	s_first_meta_bg;	/* First metablock block group */
792 	__le32	s_mkfs_time;		/* When the filesystem was created */
793 	__le32	s_jnl_blocks[17];	/* Backup of the journal inode */
794 	/* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
795 /*150*/	__le32	s_blocks_count_hi;	/* Blocks count */
796 	__le32	s_r_blocks_count_hi;	/* Reserved blocks count */
797 	__le32	s_free_blocks_count_hi;	/* Free blocks count */
798 	__le16	s_min_extra_isize;	/* All inodes have at least # bytes */
799 	__le16	s_want_extra_isize; 	/* New inodes should reserve # bytes */
800 	__le32	s_flags;		/* Miscellaneous flags */
801 	__le16  s_raid_stride;		/* RAID stride */
802 	__le16  s_mmp_interval;         /* # seconds to wait in MMP checking */
803 	__le64  s_mmp_block;            /* Block for multi-mount protection */
804 	__le32  s_raid_stripe_width;    /* blocks on all data disks (N*stride)*/
805 	__u8	s_log_groups_per_flex;  /* FLEX_BG group size */
806 	__u8	s_reserved_char_pad2;
807 	__le16  s_reserved_pad;
808 	__le64	s_kbytes_written;	/* nr of lifetime kilobytes written */
809 	__u32   s_reserved[160];        /* Padding to the end of the block */
810 };
811 
812 #ifdef __KERNEL__
813 /*
814  * fourth extended-fs super-block data in memory
815  */
816 struct ext4_sb_info {
817 	unsigned long s_desc_size;	/* Size of a group descriptor in bytes */
818 	unsigned long s_inodes_per_block;/* Number of inodes per block */
819 	unsigned long s_blocks_per_group;/* Number of blocks in a group */
820 	unsigned long s_inodes_per_group;/* Number of inodes in a group */
821 	unsigned long s_itb_per_group;	/* Number of inode table blocks per group */
822 	unsigned long s_gdb_count;	/* Number of group descriptor blocks */
823 	unsigned long s_desc_per_block;	/* Number of group descriptors per block */
824 	ext4_group_t s_groups_count;	/* Number of groups in the fs */
825 	unsigned long s_overhead_last;  /* Last calculated overhead */
826 	unsigned long s_blocks_last;    /* Last seen block count */
827 	loff_t s_bitmap_maxbytes;	/* max bytes for bitmap files */
828 	struct buffer_head * s_sbh;	/* Buffer containing the super block */
829 	struct ext4_super_block *s_es;	/* Pointer to the super block in the buffer */
830 	struct buffer_head **s_group_desc;
831 	unsigned long  s_mount_opt;
832 	ext4_fsblk_t s_sb_block;
833 	uid_t s_resuid;
834 	gid_t s_resgid;
835 	unsigned short s_mount_state;
836 	unsigned short s_pad;
837 	int s_addr_per_block_bits;
838 	int s_desc_per_block_bits;
839 	int s_inode_size;
840 	int s_first_ino;
841 	unsigned int s_inode_readahead_blks;
842 	spinlock_t s_next_gen_lock;
843 	u32 s_next_generation;
844 	u32 s_hash_seed[4];
845 	int s_def_hash_version;
846 	int s_hash_unsigned;	/* 3 if hash should be signed, 0 if not */
847 	struct percpu_counter s_freeblocks_counter;
848 	struct percpu_counter s_freeinodes_counter;
849 	struct percpu_counter s_dirs_counter;
850 	struct percpu_counter s_dirtyblocks_counter;
851 	struct blockgroup_lock *s_blockgroup_lock;
852 	struct proc_dir_entry *s_proc;
853 	struct kobject s_kobj;
854 	struct completion s_kobj_unregister;
855 
856 	/* Journaling */
857 	struct inode *s_journal_inode;
858 	struct journal_s *s_journal;
859 	struct list_head s_orphan;
860 	struct mutex s_orphan_lock;
861 	struct mutex s_resize_lock;
862 	unsigned long s_commit_interval;
863 	u32 s_max_batch_time;
864 	u32 s_min_batch_time;
865 	struct block_device *journal_bdev;
866 #ifdef CONFIG_JBD2_DEBUG
867 	struct timer_list turn_ro_timer;	/* For turning read-only (crash simulation) */
868 	wait_queue_head_t ro_wait_queue;	/* For people waiting for the fs to go read-only */
869 #endif
870 #ifdef CONFIG_QUOTA
871 	char *s_qf_names[MAXQUOTAS];		/* Names of quota files with journalled quota */
872 	int s_jquota_fmt;			/* Format of quota to use */
873 #endif
874 	unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
875 
876 #ifdef EXTENTS_STATS
877 	/* ext4 extents stats */
878 	unsigned long s_ext_min;
879 	unsigned long s_ext_max;
880 	unsigned long s_depth_max;
881 	spinlock_t s_ext_stats_lock;
882 	unsigned long s_ext_blocks;
883 	unsigned long s_ext_extents;
884 #endif
885 
886 	/* for buddy allocator */
887 	struct ext4_group_info ***s_group_info;
888 	struct inode *s_buddy_cache;
889 	long s_blocks_reserved;
890 	spinlock_t s_reserve_lock;
891 	spinlock_t s_md_lock;
892 	tid_t s_last_transaction;
893 	unsigned short *s_mb_offsets;
894 	unsigned int *s_mb_maxs;
895 
896 	/* tunables */
897 	unsigned long s_stripe;
898 	unsigned int s_mb_stream_request;
899 	unsigned int s_mb_max_to_scan;
900 	unsigned int s_mb_min_to_scan;
901 	unsigned int s_mb_stats;
902 	unsigned int s_mb_order2_reqs;
903 	unsigned int s_mb_group_prealloc;
904 	/* where last allocation was done - for stream allocation */
905 	unsigned long s_mb_last_group;
906 	unsigned long s_mb_last_start;
907 
908 	/* history to debug policy */
909 	struct ext4_mb_history *s_mb_history;
910 	int s_mb_history_cur;
911 	int s_mb_history_max;
912 	int s_mb_history_num;
913 	spinlock_t s_mb_history_lock;
914 	int s_mb_history_filter;
915 
916 	/* stats for buddy allocator */
917 	spinlock_t s_mb_pa_lock;
918 	atomic_t s_bal_reqs;	/* number of reqs with len > 1 */
919 	atomic_t s_bal_success;	/* we found long enough chunks */
920 	atomic_t s_bal_allocated;	/* in blocks */
921 	atomic_t s_bal_ex_scanned;	/* total extents scanned */
922 	atomic_t s_bal_goals;	/* goal hits */
923 	atomic_t s_bal_breaks;	/* too long searches */
924 	atomic_t s_bal_2orders;	/* 2^order hits */
925 	spinlock_t s_bal_lock;
926 	unsigned long s_mb_buddies_generated;
927 	unsigned long long s_mb_generation_time;
928 	atomic_t s_mb_lost_chunks;
929 	atomic_t s_mb_preallocated;
930 	atomic_t s_mb_discarded;
931 
932 	/* locality groups */
933 	struct ext4_locality_group *s_locality_groups;
934 
935 	/* for write statistics */
936 	unsigned long s_sectors_written_start;
937 	u64 s_kbytes_written;
938 
939 	unsigned int s_log_groups_per_flex;
940 	struct flex_groups *s_flex_groups;
941 };
942 
943 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
944 {
945 	return sb->s_fs_info;
946 }
947 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
948 {
949 	return container_of(inode, struct ext4_inode_info, vfs_inode);
950 }
951 
952 static inline struct timespec ext4_current_time(struct inode *inode)
953 {
954 	return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
955 		current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
956 }
957 
958 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
959 {
960 	return ino == EXT4_ROOT_INO ||
961 		ino == EXT4_JOURNAL_INO ||
962 		ino == EXT4_RESIZE_INO ||
963 		(ino >= EXT4_FIRST_INO(sb) &&
964 		 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
965 }
966 
967 static inline spinlock_t *
968 sb_bgl_lock(struct ext4_sb_info *sbi, unsigned int block_group)
969 {
970 	return bgl_lock_ptr(sbi->s_blockgroup_lock, block_group);
971 }
972 #else
973 /* Assume that user mode programs are passing in an ext4fs superblock, not
974  * a kernel struct super_block.  This will allow us to call the feature-test
975  * macros from user land. */
976 #define EXT4_SB(sb)	(sb)
977 #endif
978 
979 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
980 
981 /*
982  * Codes for operating systems
983  */
984 #define EXT4_OS_LINUX		0
985 #define EXT4_OS_HURD		1
986 #define EXT4_OS_MASIX		2
987 #define EXT4_OS_FREEBSD		3
988 #define EXT4_OS_LITES		4
989 
990 /*
991  * Revision levels
992  */
993 #define EXT4_GOOD_OLD_REV	0	/* The good old (original) format */
994 #define EXT4_DYNAMIC_REV	1	/* V2 format w/ dynamic inode sizes */
995 
996 #define EXT4_CURRENT_REV	EXT4_GOOD_OLD_REV
997 #define EXT4_MAX_SUPP_REV	EXT4_DYNAMIC_REV
998 
999 #define EXT4_GOOD_OLD_INODE_SIZE 128
1000 
1001 /*
1002  * Feature set definitions
1003  */
1004 
1005 #define EXT4_HAS_COMPAT_FEATURE(sb,mask)			\
1006 	((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1007 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask)			\
1008 	((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1009 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask)			\
1010 	((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1011 #define EXT4_SET_COMPAT_FEATURE(sb,mask)			\
1012 	EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1013 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask)			\
1014 	EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1015 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask)			\
1016 	EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1017 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask)			\
1018 	EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1019 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask)			\
1020 	EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1021 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask)			\
1022 	EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1023 
1024 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC	0x0001
1025 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES	0x0002
1026 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL		0x0004
1027 #define EXT4_FEATURE_COMPAT_EXT_ATTR		0x0008
1028 #define EXT4_FEATURE_COMPAT_RESIZE_INODE	0x0010
1029 #define EXT4_FEATURE_COMPAT_DIR_INDEX		0x0020
1030 
1031 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER	0x0001
1032 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE	0x0002
1033 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR	0x0004
1034 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE        0x0008
1035 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM		0x0010
1036 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK	0x0020
1037 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE	0x0040
1038 
1039 #define EXT4_FEATURE_INCOMPAT_COMPRESSION	0x0001
1040 #define EXT4_FEATURE_INCOMPAT_FILETYPE		0x0002
1041 #define EXT4_FEATURE_INCOMPAT_RECOVER		0x0004 /* Needs recovery */
1042 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV	0x0008 /* Journal device */
1043 #define EXT4_FEATURE_INCOMPAT_META_BG		0x0010
1044 #define EXT4_FEATURE_INCOMPAT_EXTENTS		0x0040 /* extents support */
1045 #define EXT4_FEATURE_INCOMPAT_64BIT		0x0080
1046 #define EXT4_FEATURE_INCOMPAT_MMP               0x0100
1047 #define EXT4_FEATURE_INCOMPAT_FLEX_BG		0x0200
1048 
1049 #define EXT4_FEATURE_COMPAT_SUPP	EXT2_FEATURE_COMPAT_EXT_ATTR
1050 #define EXT4_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
1051 					 EXT4_FEATURE_INCOMPAT_RECOVER| \
1052 					 EXT4_FEATURE_INCOMPAT_META_BG| \
1053 					 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1054 					 EXT4_FEATURE_INCOMPAT_64BIT| \
1055 					 EXT4_FEATURE_INCOMPAT_FLEX_BG)
1056 #define EXT4_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1057 					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1058 					 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1059 					 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1060 					 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1061 					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1062 					 EXT4_FEATURE_RO_COMPAT_HUGE_FILE)
1063 
1064 /*
1065  * Default values for user and/or group using reserved blocks
1066  */
1067 #define	EXT4_DEF_RESUID		0
1068 #define	EXT4_DEF_RESGID		0
1069 
1070 #define EXT4_DEF_INODE_READAHEAD_BLKS	32
1071 
1072 /*
1073  * Default mount options
1074  */
1075 #define EXT4_DEFM_DEBUG		0x0001
1076 #define EXT4_DEFM_BSDGROUPS	0x0002
1077 #define EXT4_DEFM_XATTR_USER	0x0004
1078 #define EXT4_DEFM_ACL		0x0008
1079 #define EXT4_DEFM_UID16		0x0010
1080 #define EXT4_DEFM_JMODE		0x0060
1081 #define EXT4_DEFM_JMODE_DATA	0x0020
1082 #define EXT4_DEFM_JMODE_ORDERED	0x0040
1083 #define EXT4_DEFM_JMODE_WBACK	0x0060
1084 
1085 /*
1086  * Default journal batch times
1087  */
1088 #define EXT4_DEF_MIN_BATCH_TIME	0
1089 #define EXT4_DEF_MAX_BATCH_TIME	15000 /* 15ms */
1090 
1091 /*
1092  * Minimum number of groups in a flexgroup before we separate out
1093  * directories into the first block group of a flexgroup
1094  */
1095 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME	4
1096 
1097 /*
1098  * Structure of a directory entry
1099  */
1100 #define EXT4_NAME_LEN 255
1101 
1102 struct ext4_dir_entry {
1103 	__le32	inode;			/* Inode number */
1104 	__le16	rec_len;		/* Directory entry length */
1105 	__le16	name_len;		/* Name length */
1106 	char	name[EXT4_NAME_LEN];	/* File name */
1107 };
1108 
1109 /*
1110  * The new version of the directory entry.  Since EXT4 structures are
1111  * stored in intel byte order, and the name_len field could never be
1112  * bigger than 255 chars, it's safe to reclaim the extra byte for the
1113  * file_type field.
1114  */
1115 struct ext4_dir_entry_2 {
1116 	__le32	inode;			/* Inode number */
1117 	__le16	rec_len;		/* Directory entry length */
1118 	__u8	name_len;		/* Name length */
1119 	__u8	file_type;
1120 	char	name[EXT4_NAME_LEN];	/* File name */
1121 };
1122 
1123 /*
1124  * Ext4 directory file types.  Only the low 3 bits are used.  The
1125  * other bits are reserved for now.
1126  */
1127 #define EXT4_FT_UNKNOWN		0
1128 #define EXT4_FT_REG_FILE	1
1129 #define EXT4_FT_DIR		2
1130 #define EXT4_FT_CHRDEV		3
1131 #define EXT4_FT_BLKDEV		4
1132 #define EXT4_FT_FIFO		5
1133 #define EXT4_FT_SOCK		6
1134 #define EXT4_FT_SYMLINK		7
1135 
1136 #define EXT4_FT_MAX		8
1137 
1138 /*
1139  * EXT4_DIR_PAD defines the directory entries boundaries
1140  *
1141  * NOTE: It must be a multiple of 4
1142  */
1143 #define EXT4_DIR_PAD			4
1144 #define EXT4_DIR_ROUND			(EXT4_DIR_PAD - 1)
1145 #define EXT4_DIR_REC_LEN(name_len)	(((name_len) + 8 + EXT4_DIR_ROUND) & \
1146 					 ~EXT4_DIR_ROUND)
1147 #define EXT4_MAX_REC_LEN		((1<<16)-1)
1148 
1149 /*
1150  * Hash Tree Directory indexing
1151  * (c) Daniel Phillips, 2001
1152  */
1153 
1154 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1155 				      EXT4_FEATURE_COMPAT_DIR_INDEX) && \
1156 		      (EXT4_I(dir)->i_flags & EXT4_INDEX_FL))
1157 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1158 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1159 
1160 /* Legal values for the dx_root hash_version field: */
1161 
1162 #define DX_HASH_LEGACY		0
1163 #define DX_HASH_HALF_MD4	1
1164 #define DX_HASH_TEA		2
1165 #define DX_HASH_LEGACY_UNSIGNED	3
1166 #define DX_HASH_HALF_MD4_UNSIGNED	4
1167 #define DX_HASH_TEA_UNSIGNED		5
1168 
1169 #ifdef __KERNEL__
1170 
1171 /* hash info structure used by the directory hash */
1172 struct dx_hash_info
1173 {
1174 	u32		hash;
1175 	u32		minor_hash;
1176 	int		hash_version;
1177 	u32		*seed;
1178 };
1179 
1180 #define EXT4_HTREE_EOF	0x7fffffff
1181 
1182 /*
1183  * Control parameters used by ext4_htree_next_block
1184  */
1185 #define HASH_NB_ALWAYS		1
1186 
1187 
1188 /*
1189  * Describe an inode's exact location on disk and in memory
1190  */
1191 struct ext4_iloc
1192 {
1193 	struct buffer_head *bh;
1194 	unsigned long offset;
1195 	ext4_group_t block_group;
1196 };
1197 
1198 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
1199 {
1200 	return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
1201 }
1202 
1203 /*
1204  * This structure is stuffed into the struct file's private_data field
1205  * for directories.  It is where we put information so that we can do
1206  * readdir operations in hash tree order.
1207  */
1208 struct dir_private_info {
1209 	struct rb_root	root;
1210 	struct rb_node	*curr_node;
1211 	struct fname	*extra_fname;
1212 	loff_t		last_pos;
1213 	__u32		curr_hash;
1214 	__u32		curr_minor_hash;
1215 	__u32		next_hash;
1216 };
1217 
1218 /* calculate the first block number of the group */
1219 static inline ext4_fsblk_t
1220 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
1221 {
1222 	return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
1223 		le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
1224 }
1225 
1226 /*
1227  * Special error return code only used by dx_probe() and its callers.
1228  */
1229 #define ERR_BAD_DX_DIR	-75000
1230 
1231 void ext4_get_group_no_and_offset(struct super_block *sb, ext4_fsblk_t blocknr,
1232 			ext4_group_t *blockgrpp, ext4_grpblk_t *offsetp);
1233 
1234 extern struct proc_dir_entry *ext4_proc_root;
1235 
1236 /*
1237  * Function prototypes
1238  */
1239 
1240 /*
1241  * Ok, these declarations are also in <linux/kernel.h> but none of the
1242  * ext4 source programs needs to include it so they are duplicated here.
1243  */
1244 # define NORET_TYPE	/**/
1245 # define ATTRIB_NORET	__attribute__((noreturn))
1246 # define NORET_AND	noreturn,
1247 
1248 /* bitmap.c */
1249 extern unsigned int ext4_count_free(struct buffer_head *, unsigned);
1250 
1251 /* balloc.c */
1252 extern unsigned int ext4_block_group(struct super_block *sb,
1253 			ext4_fsblk_t blocknr);
1254 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
1255 			ext4_fsblk_t blocknr);
1256 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
1257 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
1258 			ext4_group_t group);
1259 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
1260 			ext4_fsblk_t goal, unsigned long *count, int *errp);
1261 extern int ext4_claim_free_blocks(struct ext4_sb_info *sbi, s64 nblocks);
1262 extern int ext4_has_free_blocks(struct ext4_sb_info *sbi, s64 nblocks);
1263 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
1264 			ext4_fsblk_t block, unsigned long count, int metadata);
1265 extern void ext4_add_groupblocks(handle_t *handle, struct super_block *sb,
1266 				ext4_fsblk_t block, unsigned long count);
1267 extern ext4_fsblk_t ext4_count_free_blocks(struct super_block *);
1268 extern void ext4_check_blocks_bitmap(struct super_block *);
1269 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
1270 						    ext4_group_t block_group,
1271 						    struct buffer_head ** bh);
1272 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
1273 
1274 /* dir.c */
1275 extern int ext4_check_dir_entry(const char *, struct inode *,
1276 				struct ext4_dir_entry_2 *,
1277 				struct buffer_head *, unsigned int);
1278 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
1279 				    __u32 minor_hash,
1280 				    struct ext4_dir_entry_2 *dirent);
1281 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
1282 
1283 /* fsync.c */
1284 extern int ext4_sync_file(struct file *, struct dentry *, int);
1285 
1286 /* hash.c */
1287 extern int ext4fs_dirhash(const char *name, int len, struct
1288 			  dx_hash_info *hinfo);
1289 
1290 /* ialloc.c */
1291 extern struct inode * ext4_new_inode(handle_t *, struct inode *, int);
1292 extern void ext4_free_inode(handle_t *, struct inode *);
1293 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
1294 extern unsigned long ext4_count_free_inodes(struct super_block *);
1295 extern unsigned long ext4_count_dirs(struct super_block *);
1296 extern void ext4_check_inodes_bitmap(struct super_block *);
1297 
1298 /* mballoc.c */
1299 extern long ext4_mb_stats;
1300 extern long ext4_mb_max_to_scan;
1301 extern int ext4_mb_init(struct super_block *, int);
1302 extern int ext4_mb_release(struct super_block *);
1303 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
1304 				struct ext4_allocation_request *, int *);
1305 extern int ext4_mb_reserve_blocks(struct super_block *, int);
1306 extern void ext4_discard_preallocations(struct inode *);
1307 extern int __init init_ext4_mballoc(void);
1308 extern void exit_ext4_mballoc(void);
1309 extern void ext4_mb_free_blocks(handle_t *, struct inode *,
1310 		unsigned long, unsigned long, int, unsigned long *);
1311 extern int ext4_mb_add_groupinfo(struct super_block *sb,
1312 		ext4_group_t i, struct ext4_group_desc *desc);
1313 extern void ext4_mb_update_group_info(struct ext4_group_info *grp,
1314 		ext4_grpblk_t add);
1315 extern int ext4_mb_get_buddy_cache_lock(struct super_block *, ext4_group_t);
1316 extern void ext4_mb_put_buddy_cache_lock(struct super_block *,
1317 						ext4_group_t, int);
1318 /* inode.c */
1319 int ext4_forget(handle_t *handle, int is_metadata, struct inode *inode,
1320 		struct buffer_head *bh, ext4_fsblk_t blocknr);
1321 struct buffer_head *ext4_getblk(handle_t *, struct inode *,
1322 						ext4_lblk_t, int, int *);
1323 struct buffer_head *ext4_bread(handle_t *, struct inode *,
1324 						ext4_lblk_t, int, int *);
1325 int ext4_get_block(struct inode *inode, sector_t iblock,
1326 				struct buffer_head *bh_result, int create);
1327 
1328 extern struct inode *ext4_iget(struct super_block *, unsigned long);
1329 extern int  ext4_write_inode(struct inode *, int);
1330 extern int  ext4_setattr(struct dentry *, struct iattr *);
1331 extern int  ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
1332 				struct kstat *stat);
1333 extern void ext4_delete_inode(struct inode *);
1334 extern int  ext4_sync_inode(handle_t *, struct inode *);
1335 extern void ext4_dirty_inode(struct inode *);
1336 extern int ext4_change_inode_journal_flag(struct inode *, int);
1337 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
1338 extern int ext4_can_truncate(struct inode *inode);
1339 extern void ext4_truncate(struct inode *);
1340 extern void ext4_set_inode_flags(struct inode *);
1341 extern void ext4_get_inode_flags(struct ext4_inode_info *);
1342 extern int ext4_alloc_da_blocks(struct inode *inode);
1343 extern void ext4_set_aops(struct inode *inode);
1344 extern int ext4_writepage_trans_blocks(struct inode *);
1345 extern int ext4_meta_trans_blocks(struct inode *, int nrblocks, int idxblocks);
1346 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
1347 extern int ext4_block_truncate_page(handle_t *handle,
1348 		struct address_space *mapping, loff_t from);
1349 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
1350 extern qsize_t ext4_get_reserved_space(struct inode *inode);
1351 
1352 /* ioctl.c */
1353 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
1354 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
1355 
1356 /* migrate.c */
1357 extern int ext4_ext_migrate(struct inode *);
1358 
1359 /* namei.c */
1360 extern unsigned int ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize);
1361 extern __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize);
1362 extern int ext4_orphan_add(handle_t *, struct inode *);
1363 extern int ext4_orphan_del(handle_t *, struct inode *);
1364 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
1365 				__u32 start_minor_hash, __u32 *next_hash);
1366 
1367 /* resize.c */
1368 extern int ext4_group_add(struct super_block *sb,
1369 				struct ext4_new_group_data *input);
1370 extern int ext4_group_extend(struct super_block *sb,
1371 				struct ext4_super_block *es,
1372 				ext4_fsblk_t n_blocks_count);
1373 
1374 /* super.c */
1375 extern void ext4_error(struct super_block *, const char *, const char *, ...)
1376 	__attribute__ ((format (printf, 3, 4)));
1377 extern void __ext4_std_error(struct super_block *, const char *, int);
1378 extern void ext4_abort(struct super_block *, const char *, const char *, ...)
1379 	__attribute__ ((format (printf, 3, 4)));
1380 extern void ext4_warning(struct super_block *, const char *, const char *, ...)
1381 	__attribute__ ((format (printf, 3, 4)));
1382 extern void ext4_grp_locked_error(struct super_block *, ext4_group_t,
1383 				const char *, const char *, ...)
1384 	__attribute__ ((format (printf, 4, 5)));
1385 extern void ext4_update_dynamic_rev(struct super_block *sb);
1386 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
1387 					__u32 compat);
1388 extern int ext4_update_rocompat_feature(handle_t *handle,
1389 					struct super_block *sb,	__u32 rocompat);
1390 extern int ext4_update_incompat_feature(handle_t *handle,
1391 					struct super_block *sb,	__u32 incompat);
1392 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
1393 				      struct ext4_group_desc *bg);
1394 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
1395 				      struct ext4_group_desc *bg);
1396 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
1397 				     struct ext4_group_desc *bg);
1398 extern __u32 ext4_free_blks_count(struct super_block *sb,
1399 				struct ext4_group_desc *bg);
1400 extern __u32 ext4_free_inodes_count(struct super_block *sb,
1401 				 struct ext4_group_desc *bg);
1402 extern __u32 ext4_used_dirs_count(struct super_block *sb,
1403 				struct ext4_group_desc *bg);
1404 extern __u32 ext4_itable_unused_count(struct super_block *sb,
1405 				   struct ext4_group_desc *bg);
1406 extern void ext4_block_bitmap_set(struct super_block *sb,
1407 				  struct ext4_group_desc *bg, ext4_fsblk_t blk);
1408 extern void ext4_inode_bitmap_set(struct super_block *sb,
1409 				  struct ext4_group_desc *bg, ext4_fsblk_t blk);
1410 extern void ext4_inode_table_set(struct super_block *sb,
1411 				 struct ext4_group_desc *bg, ext4_fsblk_t blk);
1412 extern void ext4_free_blks_set(struct super_block *sb,
1413 			       struct ext4_group_desc *bg, __u32 count);
1414 extern void ext4_free_inodes_set(struct super_block *sb,
1415 				struct ext4_group_desc *bg, __u32 count);
1416 extern void ext4_used_dirs_set(struct super_block *sb,
1417 				struct ext4_group_desc *bg, __u32 count);
1418 extern void ext4_itable_unused_set(struct super_block *sb,
1419 				   struct ext4_group_desc *bg, __u32 count);
1420 
1421 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
1422 {
1423 	return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
1424 		le32_to_cpu(es->s_blocks_count_lo);
1425 }
1426 
1427 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
1428 {
1429 	return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
1430 		le32_to_cpu(es->s_r_blocks_count_lo);
1431 }
1432 
1433 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
1434 {
1435 	return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
1436 		le32_to_cpu(es->s_free_blocks_count_lo);
1437 }
1438 
1439 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
1440 					 ext4_fsblk_t blk)
1441 {
1442 	es->s_blocks_count_lo = cpu_to_le32((u32)blk);
1443 	es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
1444 }
1445 
1446 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
1447 					      ext4_fsblk_t blk)
1448 {
1449 	es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
1450 	es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
1451 }
1452 
1453 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
1454 					   ext4_fsblk_t blk)
1455 {
1456 	es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
1457 	es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
1458 }
1459 
1460 static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
1461 {
1462 	if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
1463 		return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
1464 			le32_to_cpu(raw_inode->i_size_lo);
1465 	else
1466 		return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
1467 }
1468 
1469 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
1470 {
1471 	raw_inode->i_size_lo = cpu_to_le32(i_size);
1472 	raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
1473 }
1474 
1475 static inline
1476 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
1477 					    ext4_group_t group)
1478 {
1479 	 struct ext4_group_info ***grp_info;
1480 	 long indexv, indexh;
1481 	 grp_info = EXT4_SB(sb)->s_group_info;
1482 	 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
1483 	 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
1484 	 return grp_info[indexv][indexh];
1485 }
1486 
1487 /*
1488  * Reading s_groups_count requires using smp_rmb() afterwards.  See
1489  * the locking protocol documented in the comments of ext4_group_add()
1490  * in resize.c
1491  */
1492 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
1493 {
1494 	ext4_group_t	ngroups = EXT4_SB(sb)->s_groups_count;
1495 
1496 	smp_rmb();
1497 	return ngroups;
1498 }
1499 
1500 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
1501 					     ext4_group_t block_group)
1502 {
1503 	return block_group >> sbi->s_log_groups_per_flex;
1504 }
1505 
1506 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
1507 {
1508 	return 1 << sbi->s_log_groups_per_flex;
1509 }
1510 
1511 #define ext4_std_error(sb, errno)				\
1512 do {								\
1513 	if ((errno))						\
1514 		__ext4_std_error((sb), __func__, (errno));	\
1515 } while (0)
1516 
1517 #ifdef CONFIG_SMP
1518 /* Each CPU can accumulate percpu_counter_batch blocks in their local
1519  * counters. So we need to make sure we have free blocks more
1520  * than percpu_counter_batch  * nr_cpu_ids. Also add a window of 4 times.
1521  */
1522 #define EXT4_FREEBLOCKS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
1523 #else
1524 #define EXT4_FREEBLOCKS_WATERMARK 0
1525 #endif
1526 
1527 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
1528 {
1529 	/*
1530 	 * XXX: replace with spinlock if seen contended -bzzz
1531 	 */
1532 	down_write(&EXT4_I(inode)->i_data_sem);
1533 	if (newsize > EXT4_I(inode)->i_disksize)
1534 		EXT4_I(inode)->i_disksize = newsize;
1535 	up_write(&EXT4_I(inode)->i_data_sem);
1536 	return ;
1537 }
1538 
1539 struct ext4_group_info {
1540 	unsigned long   bb_state;
1541 	struct rb_root  bb_free_root;
1542 	unsigned short  bb_first_free;
1543 	unsigned short  bb_free;
1544 	unsigned short  bb_fragments;
1545 	struct          list_head bb_prealloc_list;
1546 #ifdef DOUBLE_CHECK
1547 	void            *bb_bitmap;
1548 #endif
1549 	struct rw_semaphore alloc_sem;
1550 	unsigned short  bb_counters[];
1551 };
1552 
1553 #define EXT4_GROUP_INFO_NEED_INIT_BIT	0
1554 #define EXT4_GROUP_INFO_LOCKED_BIT	1
1555 
1556 #define EXT4_MB_GRP_NEED_INIT(grp)	\
1557 	(test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
1558 
1559 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
1560 {
1561 	struct ext4_group_info *grinfo = ext4_get_group_info(sb, group);
1562 
1563 	bit_spin_lock(EXT4_GROUP_INFO_LOCKED_BIT, &(grinfo->bb_state));
1564 }
1565 
1566 static inline void ext4_unlock_group(struct super_block *sb,
1567 					ext4_group_t group)
1568 {
1569 	struct ext4_group_info *grinfo = ext4_get_group_info(sb, group);
1570 
1571 	bit_spin_unlock(EXT4_GROUP_INFO_LOCKED_BIT, &(grinfo->bb_state));
1572 }
1573 
1574 static inline int ext4_is_group_locked(struct super_block *sb,
1575 					ext4_group_t group)
1576 {
1577 	struct ext4_group_info *grinfo = ext4_get_group_info(sb, group);
1578 
1579 	return bit_spin_is_locked(EXT4_GROUP_INFO_LOCKED_BIT,
1580 						&(grinfo->bb_state));
1581 }
1582 
1583 /*
1584  * Inodes and files operations
1585  */
1586 
1587 /* dir.c */
1588 extern const struct file_operations ext4_dir_operations;
1589 
1590 /* file.c */
1591 extern const struct inode_operations ext4_file_inode_operations;
1592 extern const struct file_operations ext4_file_operations;
1593 
1594 /* namei.c */
1595 extern const struct inode_operations ext4_dir_inode_operations;
1596 extern const struct inode_operations ext4_special_inode_operations;
1597 
1598 /* symlink.c */
1599 extern const struct inode_operations ext4_symlink_inode_operations;
1600 extern const struct inode_operations ext4_fast_symlink_inode_operations;
1601 
1602 /* extents.c */
1603 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
1604 extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
1605 extern int ext4_ext_index_trans_blocks(struct inode *inode, int nrblocks,
1606 				       int chunk);
1607 extern int ext4_ext_get_blocks(handle_t *handle, struct inode *inode,
1608 			       ext4_lblk_t iblock, unsigned int max_blocks,
1609 			       struct buffer_head *bh_result,
1610 			       int create, int extend_disksize);
1611 extern void ext4_ext_truncate(struct inode *);
1612 extern void ext4_ext_init(struct super_block *);
1613 extern void ext4_ext_release(struct super_block *);
1614 extern long ext4_fallocate(struct inode *inode, int mode, loff_t offset,
1615 			  loff_t len);
1616 extern int ext4_get_blocks_wrap(handle_t *handle, struct inode *inode,
1617 			sector_t block, unsigned int max_blocks,
1618 			struct buffer_head *bh, int create,
1619 			int extend_disksize, int flag);
1620 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
1621 			__u64 start, __u64 len);
1622 
1623 /*
1624  * Add new method to test wether block and inode bitmaps are properly
1625  * initialized. With uninit_bg reading the block from disk is not enough
1626  * to mark the bitmap uptodate. We need to also zero-out the bitmap
1627  */
1628 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
1629 
1630 static inline int bitmap_uptodate(struct buffer_head *bh)
1631 {
1632 	return (buffer_uptodate(bh) &&
1633 			test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
1634 }
1635 static inline void set_bitmap_uptodate(struct buffer_head *bh)
1636 {
1637 	set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
1638 }
1639 
1640 #endif	/* __KERNEL__ */
1641 
1642 #endif	/* _EXT4_H */
1643