xref: /openbmc/linux/fs/ext4/ext4.h (revision 66e61a9e)
1 /*
2  *  ext4.h
3  *
4  * Copyright (C) 1992, 1993, 1994, 1995
5  * Remy Card (card@masi.ibp.fr)
6  * Laboratoire MASI - Institut Blaise Pascal
7  * Universite Pierre et Marie Curie (Paris VI)
8  *
9  *  from
10  *
11  *  linux/include/linux/minix_fs.h
12  *
13  *  Copyright (C) 1991, 1992  Linus Torvalds
14  */
15 
16 #ifndef _EXT4_H
17 #define _EXT4_H
18 
19 #include <linux/types.h>
20 #include <linux/blkdev.h>
21 #include <linux/magic.h>
22 #include <linux/jbd2.h>
23 #include <linux/quota.h>
24 #include <linux/rwsem.h>
25 #include <linux/rbtree.h>
26 #include <linux/seqlock.h>
27 #include <linux/mutex.h>
28 #include <linux/timer.h>
29 #include <linux/wait.h>
30 #include <linux/blockgroup_lock.h>
31 #include <linux/percpu_counter.h>
32 #ifdef __KERNEL__
33 #include <linux/compat.h>
34 #endif
35 
36 /*
37  * The fourth extended filesystem constants/structures
38  */
39 
40 /*
41  * Define EXT4FS_DEBUG to produce debug messages
42  */
43 #undef EXT4FS_DEBUG
44 
45 /*
46  * Debug code
47  */
48 #ifdef EXT4FS_DEBUG
49 #define ext4_debug(f, a...)						\
50 	do {								\
51 		printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:",	\
52 			__FILE__, __LINE__, __func__);			\
53 		printk(KERN_DEBUG f, ## a);				\
54 	} while (0)
55 #else
56 #define ext4_debug(f, a...)	do {} while (0)
57 #endif
58 
59 #define EXT4_ERROR_INODE(inode, fmt, a...) \
60 	ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
61 
62 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...)			\
63 	ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
64 
65 #define EXT4_ERROR_FILE(file, fmt, a...)	\
66 	ext4_error_file(__func__, __LINE__, (file), (fmt), ## a)
67 
68 /* data type for block offset of block group */
69 typedef int ext4_grpblk_t;
70 
71 /* data type for filesystem-wide blocks number */
72 typedef unsigned long long ext4_fsblk_t;
73 
74 /* data type for file logical block number */
75 typedef __u32 ext4_lblk_t;
76 
77 /* data type for block group number */
78 typedef unsigned int ext4_group_t;
79 
80 /*
81  * Flags used in mballoc's allocation_context flags field.
82  *
83  * Also used to show what's going on for debugging purposes when the
84  * flag field is exported via the traceport interface
85  */
86 
87 /* prefer goal again. length */
88 #define EXT4_MB_HINT_MERGE		0x0001
89 /* blocks already reserved */
90 #define EXT4_MB_HINT_RESERVED		0x0002
91 /* metadata is being allocated */
92 #define EXT4_MB_HINT_METADATA		0x0004
93 /* first blocks in the file */
94 #define EXT4_MB_HINT_FIRST		0x0008
95 /* search for the best chunk */
96 #define EXT4_MB_HINT_BEST		0x0010
97 /* data is being allocated */
98 #define EXT4_MB_HINT_DATA		0x0020
99 /* don't preallocate (for tails) */
100 #define EXT4_MB_HINT_NOPREALLOC		0x0040
101 /* allocate for locality group */
102 #define EXT4_MB_HINT_GROUP_ALLOC	0x0080
103 /* allocate goal blocks or none */
104 #define EXT4_MB_HINT_GOAL_ONLY		0x0100
105 /* goal is meaningful */
106 #define EXT4_MB_HINT_TRY_GOAL		0x0200
107 /* blocks already pre-reserved by delayed allocation */
108 #define EXT4_MB_DELALLOC_RESERVED	0x0400
109 /* We are doing stream allocation */
110 #define EXT4_MB_STREAM_ALLOC		0x0800
111 
112 
113 struct ext4_allocation_request {
114 	/* target inode for block we're allocating */
115 	struct inode *inode;
116 	/* how many blocks we want to allocate */
117 	unsigned int len;
118 	/* logical block in target inode */
119 	ext4_lblk_t logical;
120 	/* the closest logical allocated block to the left */
121 	ext4_lblk_t lleft;
122 	/* the closest logical allocated block to the right */
123 	ext4_lblk_t lright;
124 	/* phys. target (a hint) */
125 	ext4_fsblk_t goal;
126 	/* phys. block for the closest logical allocated block to the left */
127 	ext4_fsblk_t pleft;
128 	/* phys. block for the closest logical allocated block to the right */
129 	ext4_fsblk_t pright;
130 	/* flags. see above EXT4_MB_HINT_* */
131 	unsigned int flags;
132 };
133 
134 /*
135  * Logical to physical block mapping, used by ext4_map_blocks()
136  *
137  * This structure is used to pass requests into ext4_map_blocks() as
138  * well as to store the information returned by ext4_map_blocks().  It
139  * takes less room on the stack than a struct buffer_head.
140  */
141 #define EXT4_MAP_NEW		(1 << BH_New)
142 #define EXT4_MAP_MAPPED		(1 << BH_Mapped)
143 #define EXT4_MAP_UNWRITTEN	(1 << BH_Unwritten)
144 #define EXT4_MAP_BOUNDARY	(1 << BH_Boundary)
145 #define EXT4_MAP_UNINIT		(1 << BH_Uninit)
146 #define EXT4_MAP_FLAGS		(EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
147 				 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\
148 				 EXT4_MAP_UNINIT)
149 
150 struct ext4_map_blocks {
151 	ext4_fsblk_t m_pblk;
152 	ext4_lblk_t m_lblk;
153 	unsigned int m_len;
154 	unsigned int m_flags;
155 };
156 
157 /*
158  * For delayed allocation tracking
159  */
160 struct mpage_da_data {
161 	struct inode *inode;
162 	sector_t b_blocknr;		/* start block number of extent */
163 	size_t b_size;			/* size of extent */
164 	unsigned long b_state;		/* state of the extent */
165 	unsigned long first_page, next_page;	/* extent of pages */
166 	struct writeback_control *wbc;
167 	int io_done;
168 	int pages_written;
169 	int retval;
170 };
171 #define	EXT4_IO_UNWRITTEN	0x1
172 typedef struct ext4_io_end {
173 	struct list_head	list;		/* per-file finished AIO list */
174 	struct inode		*inode;		/* file being written to */
175 	unsigned int		flag;		/* unwritten or not */
176 	struct page		*page;		/* page struct for buffer write */
177 	loff_t			offset;		/* offset in the file */
178 	ssize_t			size;		/* size of the extent */
179 	struct work_struct	work;		/* data work queue */
180 } ext4_io_end_t;
181 
182 /*
183  * Special inodes numbers
184  */
185 #define	EXT4_BAD_INO		 1	/* Bad blocks inode */
186 #define EXT4_ROOT_INO		 2	/* Root inode */
187 #define EXT4_BOOT_LOADER_INO	 5	/* Boot loader inode */
188 #define EXT4_UNDEL_DIR_INO	 6	/* Undelete directory inode */
189 #define EXT4_RESIZE_INO		 7	/* Reserved group descriptors inode */
190 #define EXT4_JOURNAL_INO	 8	/* Journal inode */
191 
192 /* First non-reserved inode for old ext4 filesystems */
193 #define EXT4_GOOD_OLD_FIRST_INO	11
194 
195 /*
196  * Maximal count of links to a file
197  */
198 #define EXT4_LINK_MAX		65000
199 
200 /*
201  * Macro-instructions used to manage several block sizes
202  */
203 #define EXT4_MIN_BLOCK_SIZE		1024
204 #define	EXT4_MAX_BLOCK_SIZE		65536
205 #define EXT4_MIN_BLOCK_LOG_SIZE		10
206 #ifdef __KERNEL__
207 # define EXT4_BLOCK_SIZE(s)		((s)->s_blocksize)
208 #else
209 # define EXT4_BLOCK_SIZE(s)		(EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
210 #endif
211 #define	EXT4_ADDR_PER_BLOCK(s)		(EXT4_BLOCK_SIZE(s) / sizeof(__u32))
212 #ifdef __KERNEL__
213 # define EXT4_BLOCK_SIZE_BITS(s)	((s)->s_blocksize_bits)
214 #else
215 # define EXT4_BLOCK_SIZE_BITS(s)	((s)->s_log_block_size + 10)
216 #endif
217 #ifdef __KERNEL__
218 #define	EXT4_ADDR_PER_BLOCK_BITS(s)	(EXT4_SB(s)->s_addr_per_block_bits)
219 #define EXT4_INODE_SIZE(s)		(EXT4_SB(s)->s_inode_size)
220 #define EXT4_FIRST_INO(s)		(EXT4_SB(s)->s_first_ino)
221 #else
222 #define EXT4_INODE_SIZE(s)	(((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
223 				 EXT4_GOOD_OLD_INODE_SIZE : \
224 				 (s)->s_inode_size)
225 #define EXT4_FIRST_INO(s)	(((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
226 				 EXT4_GOOD_OLD_FIRST_INO : \
227 				 (s)->s_first_ino)
228 #endif
229 #define EXT4_BLOCK_ALIGN(size, blkbits)		ALIGN((size), (1 << (blkbits)))
230 
231 /*
232  * Structure of a blocks group descriptor
233  */
234 struct ext4_group_desc
235 {
236 	__le32	bg_block_bitmap_lo;	/* Blocks bitmap block */
237 	__le32	bg_inode_bitmap_lo;	/* Inodes bitmap block */
238 	__le32	bg_inode_table_lo;	/* Inodes table block */
239 	__le16	bg_free_blocks_count_lo;/* Free blocks count */
240 	__le16	bg_free_inodes_count_lo;/* Free inodes count */
241 	__le16	bg_used_dirs_count_lo;	/* Directories count */
242 	__le16	bg_flags;		/* EXT4_BG_flags (INODE_UNINIT, etc) */
243 	__u32	bg_reserved[2];		/* Likely block/inode bitmap checksum */
244 	__le16  bg_itable_unused_lo;	/* Unused inodes count */
245 	__le16  bg_checksum;		/* crc16(sb_uuid+group+desc) */
246 	__le32	bg_block_bitmap_hi;	/* Blocks bitmap block MSB */
247 	__le32	bg_inode_bitmap_hi;	/* Inodes bitmap block MSB */
248 	__le32	bg_inode_table_hi;	/* Inodes table block MSB */
249 	__le16	bg_free_blocks_count_hi;/* Free blocks count MSB */
250 	__le16	bg_free_inodes_count_hi;/* Free inodes count MSB */
251 	__le16	bg_used_dirs_count_hi;	/* Directories count MSB */
252 	__le16  bg_itable_unused_hi;    /* Unused inodes count MSB */
253 	__u32	bg_reserved2[3];
254 };
255 
256 /*
257  * Structure of a flex block group info
258  */
259 
260 struct flex_groups {
261 	atomic_t free_inodes;
262 	atomic_t free_blocks;
263 	atomic_t used_dirs;
264 };
265 
266 #define EXT4_BG_INODE_UNINIT	0x0001 /* Inode table/bitmap not in use */
267 #define EXT4_BG_BLOCK_UNINIT	0x0002 /* Block bitmap not in use */
268 #define EXT4_BG_INODE_ZEROED	0x0004 /* On-disk itable initialized to zero */
269 
270 /*
271  * Macro-instructions used to manage group descriptors
272  */
273 #define EXT4_MIN_DESC_SIZE		32
274 #define EXT4_MIN_DESC_SIZE_64BIT	64
275 #define	EXT4_MAX_DESC_SIZE		EXT4_MIN_BLOCK_SIZE
276 #define EXT4_DESC_SIZE(s)		(EXT4_SB(s)->s_desc_size)
277 #ifdef __KERNEL__
278 # define EXT4_BLOCKS_PER_GROUP(s)	(EXT4_SB(s)->s_blocks_per_group)
279 # define EXT4_DESC_PER_BLOCK(s)		(EXT4_SB(s)->s_desc_per_block)
280 # define EXT4_INODES_PER_GROUP(s)	(EXT4_SB(s)->s_inodes_per_group)
281 # define EXT4_DESC_PER_BLOCK_BITS(s)	(EXT4_SB(s)->s_desc_per_block_bits)
282 #else
283 # define EXT4_BLOCKS_PER_GROUP(s)	((s)->s_blocks_per_group)
284 # define EXT4_DESC_PER_BLOCK(s)		(EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
285 # define EXT4_INODES_PER_GROUP(s)	((s)->s_inodes_per_group)
286 #endif
287 
288 /*
289  * Constants relative to the data blocks
290  */
291 #define	EXT4_NDIR_BLOCKS		12
292 #define	EXT4_IND_BLOCK			EXT4_NDIR_BLOCKS
293 #define	EXT4_DIND_BLOCK			(EXT4_IND_BLOCK + 1)
294 #define	EXT4_TIND_BLOCK			(EXT4_DIND_BLOCK + 1)
295 #define	EXT4_N_BLOCKS			(EXT4_TIND_BLOCK + 1)
296 
297 /*
298  * Inode flags
299  */
300 #define	EXT4_SECRM_FL			0x00000001 /* Secure deletion */
301 #define	EXT4_UNRM_FL			0x00000002 /* Undelete */
302 #define	EXT4_COMPR_FL			0x00000004 /* Compress file */
303 #define EXT4_SYNC_FL			0x00000008 /* Synchronous updates */
304 #define EXT4_IMMUTABLE_FL		0x00000010 /* Immutable file */
305 #define EXT4_APPEND_FL			0x00000020 /* writes to file may only append */
306 #define EXT4_NODUMP_FL			0x00000040 /* do not dump file */
307 #define EXT4_NOATIME_FL			0x00000080 /* do not update atime */
308 /* Reserved for compression usage... */
309 #define EXT4_DIRTY_FL			0x00000100
310 #define EXT4_COMPRBLK_FL		0x00000200 /* One or more compressed clusters */
311 #define EXT4_NOCOMPR_FL			0x00000400 /* Don't compress */
312 #define EXT4_ECOMPR_FL			0x00000800 /* Compression error */
313 /* End compression flags --- maybe not all used */
314 #define EXT4_INDEX_FL			0x00001000 /* hash-indexed directory */
315 #define EXT4_IMAGIC_FL			0x00002000 /* AFS directory */
316 #define EXT4_JOURNAL_DATA_FL		0x00004000 /* file data should be journaled */
317 #define EXT4_NOTAIL_FL			0x00008000 /* file tail should not be merged */
318 #define EXT4_DIRSYNC_FL			0x00010000 /* dirsync behaviour (directories only) */
319 #define EXT4_TOPDIR_FL			0x00020000 /* Top of directory hierarchies*/
320 #define EXT4_HUGE_FILE_FL               0x00040000 /* Set to each huge file */
321 #define EXT4_EXTENTS_FL			0x00080000 /* Inode uses extents */
322 #define EXT4_EA_INODE_FL	        0x00200000 /* Inode used for large EA */
323 #define EXT4_EOFBLOCKS_FL		0x00400000 /* Blocks allocated beyond EOF */
324 #define EXT4_RESERVED_FL		0x80000000 /* reserved for ext4 lib */
325 
326 #define EXT4_FL_USER_VISIBLE		0x004BDFFF /* User visible flags */
327 #define EXT4_FL_USER_MODIFIABLE		0x004B80FF /* User modifiable flags */
328 
329 /* Flags that should be inherited by new inodes from their parent. */
330 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
331 			   EXT4_SYNC_FL | EXT4_IMMUTABLE_FL | EXT4_APPEND_FL |\
332 			   EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
333 			   EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
334 			   EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
335 
336 /* Flags that are appropriate for regular files (all but dir-specific ones). */
337 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
338 
339 /* Flags that are appropriate for non-directories/regular files. */
340 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
341 
342 /* Mask out flags that are inappropriate for the given type of inode. */
343 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
344 {
345 	if (S_ISDIR(mode))
346 		return flags;
347 	else if (S_ISREG(mode))
348 		return flags & EXT4_REG_FLMASK;
349 	else
350 		return flags & EXT4_OTHER_FLMASK;
351 }
352 
353 /*
354  * Inode flags used for atomic set/get
355  */
356 enum {
357 	EXT4_INODE_SECRM	= 0,	/* Secure deletion */
358 	EXT4_INODE_UNRM		= 1,	/* Undelete */
359 	EXT4_INODE_COMPR	= 2,	/* Compress file */
360 	EXT4_INODE_SYNC		= 3,	/* Synchronous updates */
361 	EXT4_INODE_IMMUTABLE	= 4,	/* Immutable file */
362 	EXT4_INODE_APPEND	= 5,	/* writes to file may only append */
363 	EXT4_INODE_NODUMP	= 6,	/* do not dump file */
364 	EXT4_INODE_NOATIME	= 7,	/* do not update atime */
365 /* Reserved for compression usage... */
366 	EXT4_INODE_DIRTY	= 8,
367 	EXT4_INODE_COMPRBLK	= 9,	/* One or more compressed clusters */
368 	EXT4_INODE_NOCOMPR	= 10,	/* Don't compress */
369 	EXT4_INODE_ECOMPR	= 11,	/* Compression error */
370 /* End compression flags --- maybe not all used */
371 	EXT4_INODE_INDEX	= 12,	/* hash-indexed directory */
372 	EXT4_INODE_IMAGIC	= 13,	/* AFS directory */
373 	EXT4_INODE_JOURNAL_DATA	= 14,	/* file data should be journaled */
374 	EXT4_INODE_NOTAIL	= 15,	/* file tail should not be merged */
375 	EXT4_INODE_DIRSYNC	= 16,	/* dirsync behaviour (directories only) */
376 	EXT4_INODE_TOPDIR	= 17,	/* Top of directory hierarchies*/
377 	EXT4_INODE_HUGE_FILE	= 18,	/* Set to each huge file */
378 	EXT4_INODE_EXTENTS	= 19,	/* Inode uses extents */
379 	EXT4_INODE_EA_INODE	= 21,	/* Inode used for large EA */
380 	EXT4_INODE_EOFBLOCKS	= 22,	/* Blocks allocated beyond EOF */
381 	EXT4_INODE_RESERVED	= 31,	/* reserved for ext4 lib */
382 };
383 
384 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
385 #define CHECK_FLAG_VALUE(FLAG) if (!TEST_FLAG_VALUE(FLAG)) { \
386 	printk(KERN_EMERG "EXT4 flag fail: " #FLAG ": %d %d\n", \
387 		EXT4_##FLAG##_FL, EXT4_INODE_##FLAG); BUG_ON(1); }
388 
389 /*
390  * Since it's pretty easy to mix up bit numbers and hex values, and we
391  * can't do a compile-time test for ENUM values, we use a run-time
392  * test to make sure that EXT4_XXX_FL is consistent with respect to
393  * EXT4_INODE_XXX.  If all is well the printk and BUG_ON will all drop
394  * out so it won't cost any extra space in the compiled kernel image.
395  * But it's important that these values are the same, since we are
396  * using EXT4_INODE_XXX to test for the flag values, but EXT4_XX_FL
397  * must be consistent with the values of FS_XXX_FL defined in
398  * include/linux/fs.h and the on-disk values found in ext2, ext3, and
399  * ext4 filesystems, and of course the values defined in e2fsprogs.
400  *
401  * It's not paranoia if the Murphy's Law really *is* out to get you.  :-)
402  */
403 static inline void ext4_check_flag_values(void)
404 {
405 	CHECK_FLAG_VALUE(SECRM);
406 	CHECK_FLAG_VALUE(UNRM);
407 	CHECK_FLAG_VALUE(COMPR);
408 	CHECK_FLAG_VALUE(SYNC);
409 	CHECK_FLAG_VALUE(IMMUTABLE);
410 	CHECK_FLAG_VALUE(APPEND);
411 	CHECK_FLAG_VALUE(NODUMP);
412 	CHECK_FLAG_VALUE(NOATIME);
413 	CHECK_FLAG_VALUE(DIRTY);
414 	CHECK_FLAG_VALUE(COMPRBLK);
415 	CHECK_FLAG_VALUE(NOCOMPR);
416 	CHECK_FLAG_VALUE(ECOMPR);
417 	CHECK_FLAG_VALUE(INDEX);
418 	CHECK_FLAG_VALUE(IMAGIC);
419 	CHECK_FLAG_VALUE(JOURNAL_DATA);
420 	CHECK_FLAG_VALUE(NOTAIL);
421 	CHECK_FLAG_VALUE(DIRSYNC);
422 	CHECK_FLAG_VALUE(TOPDIR);
423 	CHECK_FLAG_VALUE(HUGE_FILE);
424 	CHECK_FLAG_VALUE(EXTENTS);
425 	CHECK_FLAG_VALUE(EA_INODE);
426 	CHECK_FLAG_VALUE(EOFBLOCKS);
427 	CHECK_FLAG_VALUE(RESERVED);
428 }
429 
430 /* Used to pass group descriptor data when online resize is done */
431 struct ext4_new_group_input {
432 	__u32 group;		/* Group number for this data */
433 	__u64 block_bitmap;	/* Absolute block number of block bitmap */
434 	__u64 inode_bitmap;	/* Absolute block number of inode bitmap */
435 	__u64 inode_table;	/* Absolute block number of inode table start */
436 	__u32 blocks_count;	/* Total number of blocks in this group */
437 	__u16 reserved_blocks;	/* Number of reserved blocks in this group */
438 	__u16 unused;
439 };
440 
441 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
442 struct compat_ext4_new_group_input {
443 	u32 group;
444 	compat_u64 block_bitmap;
445 	compat_u64 inode_bitmap;
446 	compat_u64 inode_table;
447 	u32 blocks_count;
448 	u16 reserved_blocks;
449 	u16 unused;
450 };
451 #endif
452 
453 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
454 struct ext4_new_group_data {
455 	__u32 group;
456 	__u64 block_bitmap;
457 	__u64 inode_bitmap;
458 	__u64 inode_table;
459 	__u32 blocks_count;
460 	__u16 reserved_blocks;
461 	__u16 unused;
462 	__u32 free_blocks_count;
463 };
464 
465 /*
466  * Flags used by ext4_get_blocks()
467  */
468 	/* Allocate any needed blocks and/or convert an unitialized
469 	   extent to be an initialized ext4 */
470 #define EXT4_GET_BLOCKS_CREATE			0x0001
471 	/* Request the creation of an unitialized extent */
472 #define EXT4_GET_BLOCKS_UNINIT_EXT		0x0002
473 #define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT	(EXT4_GET_BLOCKS_UNINIT_EXT|\
474 						 EXT4_GET_BLOCKS_CREATE)
475 	/* Caller is from the delayed allocation writeout path,
476 	   so set the magic i_delalloc_reserve_flag after taking the
477 	   inode allocation semaphore for */
478 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE	0x0004
479 	/* caller is from the direct IO path, request to creation of an
480 	unitialized extents if not allocated, split the uninitialized
481 	extent if blocks has been preallocated already*/
482 #define EXT4_GET_BLOCKS_PRE_IO			0x0008
483 #define EXT4_GET_BLOCKS_CONVERT			0x0010
484 #define EXT4_GET_BLOCKS_IO_CREATE_EXT		(EXT4_GET_BLOCKS_PRE_IO|\
485 					 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
486 	/* Convert extent to initialized after IO complete */
487 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT		(EXT4_GET_BLOCKS_CONVERT|\
488 					 EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
489 
490 /*
491  * Flags used by ext4_free_blocks
492  */
493 #define EXT4_FREE_BLOCKS_METADATA	0x0001
494 #define EXT4_FREE_BLOCKS_FORGET		0x0002
495 #define EXT4_FREE_BLOCKS_VALIDATED	0x0004
496 
497 /*
498  * ioctl commands
499  */
500 #define	EXT4_IOC_GETFLAGS		FS_IOC_GETFLAGS
501 #define	EXT4_IOC_SETFLAGS		FS_IOC_SETFLAGS
502 #define	EXT4_IOC_GETVERSION		_IOR('f', 3, long)
503 #define	EXT4_IOC_SETVERSION		_IOW('f', 4, long)
504 #define	EXT4_IOC_GETVERSION_OLD		FS_IOC_GETVERSION
505 #define	EXT4_IOC_SETVERSION_OLD		FS_IOC_SETVERSION
506 #ifdef CONFIG_JBD2_DEBUG
507 #define EXT4_IOC_WAIT_FOR_READONLY	_IOR('f', 99, long)
508 #endif
509 #define EXT4_IOC_GETRSVSZ		_IOR('f', 5, long)
510 #define EXT4_IOC_SETRSVSZ		_IOW('f', 6, long)
511 #define EXT4_IOC_GROUP_EXTEND		_IOW('f', 7, unsigned long)
512 #define EXT4_IOC_GROUP_ADD		_IOW('f', 8, struct ext4_new_group_input)
513 #define EXT4_IOC_MIGRATE		_IO('f', 9)
514  /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
515  /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
516 #define EXT4_IOC_ALLOC_DA_BLKS		_IO('f', 12)
517 #define EXT4_IOC_MOVE_EXT		_IOWR('f', 15, struct move_extent)
518 
519 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
520 /*
521  * ioctl commands in 32 bit emulation
522  */
523 #define EXT4_IOC32_GETFLAGS		FS_IOC32_GETFLAGS
524 #define EXT4_IOC32_SETFLAGS		FS_IOC32_SETFLAGS
525 #define EXT4_IOC32_GETVERSION		_IOR('f', 3, int)
526 #define EXT4_IOC32_SETVERSION		_IOW('f', 4, int)
527 #define EXT4_IOC32_GETRSVSZ		_IOR('f', 5, int)
528 #define EXT4_IOC32_SETRSVSZ		_IOW('f', 6, int)
529 #define EXT4_IOC32_GROUP_EXTEND		_IOW('f', 7, unsigned int)
530 #define EXT4_IOC32_GROUP_ADD		_IOW('f', 8, struct compat_ext4_new_group_input)
531 #ifdef CONFIG_JBD2_DEBUG
532 #define EXT4_IOC32_WAIT_FOR_READONLY	_IOR('f', 99, int)
533 #endif
534 #define EXT4_IOC32_GETVERSION_OLD	FS_IOC32_GETVERSION
535 #define EXT4_IOC32_SETVERSION_OLD	FS_IOC32_SETVERSION
536 #endif
537 
538 
539 /*
540  *  Mount options
541  */
542 struct ext4_mount_options {
543 	unsigned long s_mount_opt;
544 	uid_t s_resuid;
545 	gid_t s_resgid;
546 	unsigned long s_commit_interval;
547 	u32 s_min_batch_time, s_max_batch_time;
548 #ifdef CONFIG_QUOTA
549 	int s_jquota_fmt;
550 	char *s_qf_names[MAXQUOTAS];
551 #endif
552 };
553 
554 /* Max physical block we can addres w/o extents */
555 #define EXT4_MAX_BLOCK_FILE_PHYS	0xFFFFFFFF
556 
557 /*
558  * Structure of an inode on the disk
559  */
560 struct ext4_inode {
561 	__le16	i_mode;		/* File mode */
562 	__le16	i_uid;		/* Low 16 bits of Owner Uid */
563 	__le32	i_size_lo;	/* Size in bytes */
564 	__le32	i_atime;	/* Access time */
565 	__le32	i_ctime;	/* Inode Change time */
566 	__le32	i_mtime;	/* Modification time */
567 	__le32	i_dtime;	/* Deletion Time */
568 	__le16	i_gid;		/* Low 16 bits of Group Id */
569 	__le16	i_links_count;	/* Links count */
570 	__le32	i_blocks_lo;	/* Blocks count */
571 	__le32	i_flags;	/* File flags */
572 	union {
573 		struct {
574 			__le32  l_i_version;
575 		} linux1;
576 		struct {
577 			__u32  h_i_translator;
578 		} hurd1;
579 		struct {
580 			__u32  m_i_reserved1;
581 		} masix1;
582 	} osd1;				/* OS dependent 1 */
583 	__le32	i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
584 	__le32	i_generation;	/* File version (for NFS) */
585 	__le32	i_file_acl_lo;	/* File ACL */
586 	__le32	i_size_high;
587 	__le32	i_obso_faddr;	/* Obsoleted fragment address */
588 	union {
589 		struct {
590 			__le16	l_i_blocks_high; /* were l_i_reserved1 */
591 			__le16	l_i_file_acl_high;
592 			__le16	l_i_uid_high;	/* these 2 fields */
593 			__le16	l_i_gid_high;	/* were reserved2[0] */
594 			__u32	l_i_reserved2;
595 		} linux2;
596 		struct {
597 			__le16	h_i_reserved1;	/* Obsoleted fragment number/size which are removed in ext4 */
598 			__u16	h_i_mode_high;
599 			__u16	h_i_uid_high;
600 			__u16	h_i_gid_high;
601 			__u32	h_i_author;
602 		} hurd2;
603 		struct {
604 			__le16	h_i_reserved1;	/* Obsoleted fragment number/size which are removed in ext4 */
605 			__le16	m_i_file_acl_high;
606 			__u32	m_i_reserved2[2];
607 		} masix2;
608 	} osd2;				/* OS dependent 2 */
609 	__le16	i_extra_isize;
610 	__le16	i_pad1;
611 	__le32  i_ctime_extra;  /* extra Change time      (nsec << 2 | epoch) */
612 	__le32  i_mtime_extra;  /* extra Modification time(nsec << 2 | epoch) */
613 	__le32  i_atime_extra;  /* extra Access time      (nsec << 2 | epoch) */
614 	__le32  i_crtime;       /* File Creation time */
615 	__le32  i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
616 	__le32  i_version_hi;	/* high 32 bits for 64-bit version */
617 };
618 
619 struct move_extent {
620 	__u32 reserved;		/* should be zero */
621 	__u32 donor_fd;		/* donor file descriptor */
622 	__u64 orig_start;	/* logical start offset in block for orig */
623 	__u64 donor_start;	/* logical start offset in block for donor */
624 	__u64 len;		/* block length to be moved */
625 	__u64 moved_len;	/* moved block length */
626 };
627 
628 #define EXT4_EPOCH_BITS 2
629 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
630 #define EXT4_NSEC_MASK  (~0UL << EXT4_EPOCH_BITS)
631 
632 /*
633  * Extended fields will fit into an inode if the filesystem was formatted
634  * with large inodes (-I 256 or larger) and there are not currently any EAs
635  * consuming all of the available space. For new inodes we always reserve
636  * enough space for the kernel's known extended fields, but for inodes
637  * created with an old kernel this might not have been the case. None of
638  * the extended inode fields is critical for correct filesystem operation.
639  * This macro checks if a certain field fits in the inode. Note that
640  * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
641  */
642 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field)	\
643 	((offsetof(typeof(*ext4_inode), field) +	\
644 	  sizeof((ext4_inode)->field))			\
645 	<= (EXT4_GOOD_OLD_INODE_SIZE +			\
646 	    (einode)->i_extra_isize))			\
647 
648 static inline __le32 ext4_encode_extra_time(struct timespec *time)
649 {
650        return cpu_to_le32((sizeof(time->tv_sec) > 4 ?
651 			   (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) |
652                           ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK));
653 }
654 
655 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
656 {
657        if (sizeof(time->tv_sec) > 4)
658 	       time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK)
659 			       << 32;
660        time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
661 }
662 
663 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode)			       \
664 do {									       \
665 	(raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec);	       \
666 	if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     \
667 		(raw_inode)->xtime ## _extra =				       \
668 				ext4_encode_extra_time(&(inode)->xtime);       \
669 } while (0)
670 
671 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode)			       \
672 do {									       \
673 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))		       \
674 		(raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec);      \
675 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))	       \
676 		(raw_inode)->xtime ## _extra =				       \
677 				ext4_encode_extra_time(&(einode)->xtime);      \
678 } while (0)
679 
680 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode)			       \
681 do {									       \
682 	(inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime);       \
683 	if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     \
684 		ext4_decode_extra_time(&(inode)->xtime,			       \
685 				       raw_inode->xtime ## _extra);	       \
686 } while (0)
687 
688 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode)			       \
689 do {									       \
690 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))		       \
691 		(einode)->xtime.tv_sec = 				       \
692 			(signed)le32_to_cpu((raw_inode)->xtime);	       \
693 	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))	       \
694 		ext4_decode_extra_time(&(einode)->xtime,		       \
695 				       raw_inode->xtime ## _extra);	       \
696 } while (0)
697 
698 #define i_disk_version osd1.linux1.l_i_version
699 
700 #if defined(__KERNEL__) || defined(__linux__)
701 #define i_reserved1	osd1.linux1.l_i_reserved1
702 #define i_file_acl_high	osd2.linux2.l_i_file_acl_high
703 #define i_blocks_high	osd2.linux2.l_i_blocks_high
704 #define i_uid_low	i_uid
705 #define i_gid_low	i_gid
706 #define i_uid_high	osd2.linux2.l_i_uid_high
707 #define i_gid_high	osd2.linux2.l_i_gid_high
708 #define i_reserved2	osd2.linux2.l_i_reserved2
709 
710 #elif defined(__GNU__)
711 
712 #define i_translator	osd1.hurd1.h_i_translator
713 #define i_uid_high	osd2.hurd2.h_i_uid_high
714 #define i_gid_high	osd2.hurd2.h_i_gid_high
715 #define i_author	osd2.hurd2.h_i_author
716 
717 #elif defined(__masix__)
718 
719 #define i_reserved1	osd1.masix1.m_i_reserved1
720 #define i_file_acl_high	osd2.masix2.m_i_file_acl_high
721 #define i_reserved2	osd2.masix2.m_i_reserved2
722 
723 #endif /* defined(__KERNEL__) || defined(__linux__) */
724 
725 /*
726  * storage for cached extent
727  */
728 struct ext4_ext_cache {
729 	ext4_fsblk_t	ec_start;
730 	ext4_lblk_t	ec_block;
731 	__u32		ec_len; /* must be 32bit to return holes */
732 	__u32		ec_type;
733 };
734 
735 /*
736  * fourth extended file system inode data in memory
737  */
738 struct ext4_inode_info {
739 	__le32	i_data[15];	/* unconverted */
740 	__u32	i_dtime;
741 	ext4_fsblk_t	i_file_acl;
742 
743 	/*
744 	 * i_block_group is the number of the block group which contains
745 	 * this file's inode.  Constant across the lifetime of the inode,
746 	 * it is ued for making block allocation decisions - we try to
747 	 * place a file's data blocks near its inode block, and new inodes
748 	 * near to their parent directory's inode.
749 	 */
750 	ext4_group_t	i_block_group;
751 	unsigned long	i_state_flags;		/* Dynamic state flags */
752 	unsigned long	i_flags;
753 
754 	ext4_lblk_t		i_dir_start_lookup;
755 #ifdef CONFIG_EXT4_FS_XATTR
756 	/*
757 	 * Extended attributes can be read independently of the main file
758 	 * data. Taking i_mutex even when reading would cause contention
759 	 * between readers of EAs and writers of regular file data, so
760 	 * instead we synchronize on xattr_sem when reading or changing
761 	 * EAs.
762 	 */
763 	struct rw_semaphore xattr_sem;
764 #endif
765 
766 	struct list_head i_orphan;	/* unlinked but open inodes */
767 
768 	/*
769 	 * i_disksize keeps track of what the inode size is ON DISK, not
770 	 * in memory.  During truncate, i_size is set to the new size by
771 	 * the VFS prior to calling ext4_truncate(), but the filesystem won't
772 	 * set i_disksize to 0 until the truncate is actually under way.
773 	 *
774 	 * The intent is that i_disksize always represents the blocks which
775 	 * are used by this file.  This allows recovery to restart truncate
776 	 * on orphans if we crash during truncate.  We actually write i_disksize
777 	 * into the on-disk inode when writing inodes out, instead of i_size.
778 	 *
779 	 * The only time when i_disksize and i_size may be different is when
780 	 * a truncate is in progress.  The only things which change i_disksize
781 	 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
782 	 */
783 	loff_t	i_disksize;
784 
785 	/*
786 	 * i_data_sem is for serialising ext4_truncate() against
787 	 * ext4_getblock().  In the 2.4 ext2 design, great chunks of inode's
788 	 * data tree are chopped off during truncate. We can't do that in
789 	 * ext4 because whenever we perform intermediate commits during
790 	 * truncate, the inode and all the metadata blocks *must* be in a
791 	 * consistent state which allows truncation of the orphans to restart
792 	 * during recovery.  Hence we must fix the get_block-vs-truncate race
793 	 * by other means, so we have i_data_sem.
794 	 */
795 	struct rw_semaphore i_data_sem;
796 	struct inode vfs_inode;
797 	struct jbd2_inode jinode;
798 
799 	struct ext4_ext_cache i_cached_extent;
800 	/*
801 	 * File creation time. Its function is same as that of
802 	 * struct timespec i_{a,c,m}time in the generic inode.
803 	 */
804 	struct timespec i_crtime;
805 
806 	/* mballoc */
807 	struct list_head i_prealloc_list;
808 	spinlock_t i_prealloc_lock;
809 
810 	/* ialloc */
811 	ext4_group_t	i_last_alloc_group;
812 
813 	/* allocation reservation info for delalloc */
814 	unsigned int i_reserved_data_blocks;
815 	unsigned int i_reserved_meta_blocks;
816 	unsigned int i_allocated_meta_blocks;
817 	unsigned short i_delalloc_reserved_flag;
818 	sector_t i_da_metadata_calc_last_lblock;
819 	int i_da_metadata_calc_len;
820 
821 	/* on-disk additional length */
822 	__u16 i_extra_isize;
823 
824 	spinlock_t i_block_reservation_lock;
825 #ifdef CONFIG_QUOTA
826 	/* quota space reservation, managed internally by quota code */
827 	qsize_t i_reserved_quota;
828 #endif
829 
830 	/* completed IOs that might need unwritten extents handling */
831 	struct list_head i_completed_io_list;
832 	spinlock_t i_completed_io_lock;
833 	/* current io_end structure for async DIO write*/
834 	ext4_io_end_t *cur_aio_dio;
835 
836 	/*
837 	 * Transactions that contain inode's metadata needed to complete
838 	 * fsync and fdatasync, respectively.
839 	 */
840 	tid_t i_sync_tid;
841 	tid_t i_datasync_tid;
842 };
843 
844 /*
845  * File system states
846  */
847 #define	EXT4_VALID_FS			0x0001	/* Unmounted cleanly */
848 #define	EXT4_ERROR_FS			0x0002	/* Errors detected */
849 #define	EXT4_ORPHAN_FS			0x0004	/* Orphans being recovered */
850 
851 /*
852  * Misc. filesystem flags
853  */
854 #define EXT2_FLAGS_SIGNED_HASH		0x0001  /* Signed dirhash in use */
855 #define EXT2_FLAGS_UNSIGNED_HASH	0x0002  /* Unsigned dirhash in use */
856 #define EXT2_FLAGS_TEST_FILESYS		0x0004	/* to test development code */
857 
858 /*
859  * Mount flags
860  */
861 #define EXT4_MOUNT_OLDALLOC		0x00002  /* Don't use the new Orlov allocator */
862 #define EXT4_MOUNT_GRPID		0x00004	/* Create files with directory's group */
863 #define EXT4_MOUNT_DEBUG		0x00008	/* Some debugging messages */
864 #define EXT4_MOUNT_ERRORS_CONT		0x00010	/* Continue on errors */
865 #define EXT4_MOUNT_ERRORS_RO		0x00020	/* Remount fs ro on errors */
866 #define EXT4_MOUNT_ERRORS_PANIC		0x00040	/* Panic on errors */
867 #define EXT4_MOUNT_MINIX_DF		0x00080	/* Mimics the Minix statfs */
868 #define EXT4_MOUNT_NOLOAD		0x00100	/* Don't use existing journal*/
869 #define EXT4_MOUNT_DATA_FLAGS		0x00C00	/* Mode for data writes: */
870 #define EXT4_MOUNT_JOURNAL_DATA		0x00400	/* Write data to journal */
871 #define EXT4_MOUNT_ORDERED_DATA		0x00800	/* Flush data before commit */
872 #define EXT4_MOUNT_WRITEBACK_DATA	0x00C00	/* No data ordering */
873 #define EXT4_MOUNT_UPDATE_JOURNAL	0x01000	/* Update the journal format */
874 #define EXT4_MOUNT_NO_UID32		0x02000  /* Disable 32-bit UIDs */
875 #define EXT4_MOUNT_XATTR_USER		0x04000	/* Extended user attributes */
876 #define EXT4_MOUNT_POSIX_ACL		0x08000	/* POSIX Access Control Lists */
877 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC	0x10000	/* No auto delalloc mapping */
878 #define EXT4_MOUNT_BARRIER		0x20000 /* Use block barriers */
879 #define EXT4_MOUNT_QUOTA		0x80000 /* Some quota option set */
880 #define EXT4_MOUNT_USRQUOTA		0x100000 /* "old" user quota */
881 #define EXT4_MOUNT_GRPQUOTA		0x200000 /* "old" group quota */
882 #define EXT4_MOUNT_DIOREAD_NOLOCK	0x400000 /* Enable support for dio read nolocking */
883 #define EXT4_MOUNT_JOURNAL_CHECKSUM	0x800000 /* Journal checksums */
884 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT	0x1000000 /* Journal Async Commit */
885 #define EXT4_MOUNT_I_VERSION            0x2000000 /* i_version support */
886 #define EXT4_MOUNT_DELALLOC		0x8000000 /* Delalloc support */
887 #define EXT4_MOUNT_DATA_ERR_ABORT	0x10000000 /* Abort on file data write */
888 #define EXT4_MOUNT_BLOCK_VALIDITY	0x20000000 /* Block validity checking */
889 #define EXT4_MOUNT_DISCARD		0x40000000 /* Issue DISCARD requests */
890 
891 #define clear_opt(o, opt)		o &= ~EXT4_MOUNT_##opt
892 #define set_opt(o, opt)			o |= EXT4_MOUNT_##opt
893 #define test_opt(sb, opt)		(EXT4_SB(sb)->s_mount_opt & \
894 					 EXT4_MOUNT_##opt)
895 
896 #define ext4_set_bit			ext2_set_bit
897 #define ext4_set_bit_atomic		ext2_set_bit_atomic
898 #define ext4_clear_bit			ext2_clear_bit
899 #define ext4_clear_bit_atomic		ext2_clear_bit_atomic
900 #define ext4_test_bit			ext2_test_bit
901 #define ext4_find_first_zero_bit	ext2_find_first_zero_bit
902 #define ext4_find_next_zero_bit		ext2_find_next_zero_bit
903 #define ext4_find_next_bit		ext2_find_next_bit
904 
905 /*
906  * Maximal mount counts between two filesystem checks
907  */
908 #define EXT4_DFL_MAX_MNT_COUNT		20	/* Allow 20 mounts */
909 #define EXT4_DFL_CHECKINTERVAL		0	/* Don't use interval check */
910 
911 /*
912  * Behaviour when detecting errors
913  */
914 #define EXT4_ERRORS_CONTINUE		1	/* Continue execution */
915 #define EXT4_ERRORS_RO			2	/* Remount fs read-only */
916 #define EXT4_ERRORS_PANIC		3	/* Panic */
917 #define EXT4_ERRORS_DEFAULT		EXT4_ERRORS_CONTINUE
918 
919 /*
920  * Structure of the super block
921  */
922 struct ext4_super_block {
923 /*00*/	__le32	s_inodes_count;		/* Inodes count */
924 	__le32	s_blocks_count_lo;	/* Blocks count */
925 	__le32	s_r_blocks_count_lo;	/* Reserved blocks count */
926 	__le32	s_free_blocks_count_lo;	/* Free blocks count */
927 /*10*/	__le32	s_free_inodes_count;	/* Free inodes count */
928 	__le32	s_first_data_block;	/* First Data Block */
929 	__le32	s_log_block_size;	/* Block size */
930 	__le32	s_obso_log_frag_size;	/* Obsoleted fragment size */
931 /*20*/	__le32	s_blocks_per_group;	/* # Blocks per group */
932 	__le32	s_obso_frags_per_group;	/* Obsoleted fragments per group */
933 	__le32	s_inodes_per_group;	/* # Inodes per group */
934 	__le32	s_mtime;		/* Mount time */
935 /*30*/	__le32	s_wtime;		/* Write time */
936 	__le16	s_mnt_count;		/* Mount count */
937 	__le16	s_max_mnt_count;	/* Maximal mount count */
938 	__le16	s_magic;		/* Magic signature */
939 	__le16	s_state;		/* File system state */
940 	__le16	s_errors;		/* Behaviour when detecting errors */
941 	__le16	s_minor_rev_level;	/* minor revision level */
942 /*40*/	__le32	s_lastcheck;		/* time of last check */
943 	__le32	s_checkinterval;	/* max. time between checks */
944 	__le32	s_creator_os;		/* OS */
945 	__le32	s_rev_level;		/* Revision level */
946 /*50*/	__le16	s_def_resuid;		/* Default uid for reserved blocks */
947 	__le16	s_def_resgid;		/* Default gid for reserved blocks */
948 	/*
949 	 * These fields are for EXT4_DYNAMIC_REV superblocks only.
950 	 *
951 	 * Note: the difference between the compatible feature set and
952 	 * the incompatible feature set is that if there is a bit set
953 	 * in the incompatible feature set that the kernel doesn't
954 	 * know about, it should refuse to mount the filesystem.
955 	 *
956 	 * e2fsck's requirements are more strict; if it doesn't know
957 	 * about a feature in either the compatible or incompatible
958 	 * feature set, it must abort and not try to meddle with
959 	 * things it doesn't understand...
960 	 */
961 	__le32	s_first_ino;		/* First non-reserved inode */
962 	__le16  s_inode_size;		/* size of inode structure */
963 	__le16	s_block_group_nr;	/* block group # of this superblock */
964 	__le32	s_feature_compat;	/* compatible feature set */
965 /*60*/	__le32	s_feature_incompat;	/* incompatible feature set */
966 	__le32	s_feature_ro_compat;	/* readonly-compatible feature set */
967 /*68*/	__u8	s_uuid[16];		/* 128-bit uuid for volume */
968 /*78*/	char	s_volume_name[16];	/* volume name */
969 /*88*/	char	s_last_mounted[64];	/* directory where last mounted */
970 /*C8*/	__le32	s_algorithm_usage_bitmap; /* For compression */
971 	/*
972 	 * Performance hints.  Directory preallocation should only
973 	 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
974 	 */
975 	__u8	s_prealloc_blocks;	/* Nr of blocks to try to preallocate*/
976 	__u8	s_prealloc_dir_blocks;	/* Nr to preallocate for dirs */
977 	__le16	s_reserved_gdt_blocks;	/* Per group desc for online growth */
978 	/*
979 	 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
980 	 */
981 /*D0*/	__u8	s_journal_uuid[16];	/* uuid of journal superblock */
982 /*E0*/	__le32	s_journal_inum;		/* inode number of journal file */
983 	__le32	s_journal_dev;		/* device number of journal file */
984 	__le32	s_last_orphan;		/* start of list of inodes to delete */
985 	__le32	s_hash_seed[4];		/* HTREE hash seed */
986 	__u8	s_def_hash_version;	/* Default hash version to use */
987 	__u8	s_reserved_char_pad;
988 	__le16  s_desc_size;		/* size of group descriptor */
989 /*100*/	__le32	s_default_mount_opts;
990 	__le32	s_first_meta_bg;	/* First metablock block group */
991 	__le32	s_mkfs_time;		/* When the filesystem was created */
992 	__le32	s_jnl_blocks[17];	/* Backup of the journal inode */
993 	/* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
994 /*150*/	__le32	s_blocks_count_hi;	/* Blocks count */
995 	__le32	s_r_blocks_count_hi;	/* Reserved blocks count */
996 	__le32	s_free_blocks_count_hi;	/* Free blocks count */
997 	__le16	s_min_extra_isize;	/* All inodes have at least # bytes */
998 	__le16	s_want_extra_isize; 	/* New inodes should reserve # bytes */
999 	__le32	s_flags;		/* Miscellaneous flags */
1000 	__le16  s_raid_stride;		/* RAID stride */
1001 	__le16  s_mmp_interval;         /* # seconds to wait in MMP checking */
1002 	__le64  s_mmp_block;            /* Block for multi-mount protection */
1003 	__le32  s_raid_stripe_width;    /* blocks on all data disks (N*stride)*/
1004 	__u8	s_log_groups_per_flex;  /* FLEX_BG group size */
1005 	__u8	s_reserved_char_pad2;
1006 	__le16  s_reserved_pad;
1007 	__le64	s_kbytes_written;	/* nr of lifetime kilobytes written */
1008 	__le32	s_snapshot_inum;	/* Inode number of active snapshot */
1009 	__le32	s_snapshot_id;		/* sequential ID of active snapshot */
1010 	__le64	s_snapshot_r_blocks_count; /* reserved blocks for active
1011 					      snapshot's future use */
1012 	__le32	s_snapshot_list;	/* inode number of the head of the
1013 					   on-disk snapshot list */
1014 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1015 	__le32	s_error_count;		/* number of fs errors */
1016 	__le32	s_first_error_time;	/* first time an error happened */
1017 	__le32	s_first_error_ino;	/* inode involved in first error */
1018 	__le64	s_first_error_block;	/* block involved of first error */
1019 	__u8	s_first_error_func[32];	/* function where the error happened */
1020 	__le32	s_first_error_line;	/* line number where error happened */
1021 	__le32	s_last_error_time;	/* most recent time of an error */
1022 	__le32	s_last_error_ino;	/* inode involved in last error */
1023 	__le32	s_last_error_line;	/* line number where error happened */
1024 	__le64	s_last_error_block;	/* block involved of last error */
1025 	__u8	s_last_error_func[32];	/* function where the error happened */
1026 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_reserved)
1027 	__le32   s_reserved[128];        /* Padding to the end of the block */
1028 };
1029 
1030 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1031 
1032 #ifdef __KERNEL__
1033 
1034 /*
1035  * run-time mount flags
1036  */
1037 #define EXT4_MF_MNTDIR_SAMPLED	0x0001
1038 #define EXT4_MF_FS_ABORTED	0x0002	/* Fatal error detected */
1039 
1040 /*
1041  * fourth extended-fs super-block data in memory
1042  */
1043 struct ext4_sb_info {
1044 	unsigned long s_desc_size;	/* Size of a group descriptor in bytes */
1045 	unsigned long s_inodes_per_block;/* Number of inodes per block */
1046 	unsigned long s_blocks_per_group;/* Number of blocks in a group */
1047 	unsigned long s_inodes_per_group;/* Number of inodes in a group */
1048 	unsigned long s_itb_per_group;	/* Number of inode table blocks per group */
1049 	unsigned long s_gdb_count;	/* Number of group descriptor blocks */
1050 	unsigned long s_desc_per_block;	/* Number of group descriptors per block */
1051 	ext4_group_t s_groups_count;	/* Number of groups in the fs */
1052 	ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1053 	unsigned long s_overhead_last;  /* Last calculated overhead */
1054 	unsigned long s_blocks_last;    /* Last seen block count */
1055 	loff_t s_bitmap_maxbytes;	/* max bytes for bitmap files */
1056 	struct buffer_head * s_sbh;	/* Buffer containing the super block */
1057 	struct ext4_super_block *s_es;	/* Pointer to the super block in the buffer */
1058 	struct buffer_head **s_group_desc;
1059 	unsigned int s_mount_opt;
1060 	unsigned int s_mount_flags;
1061 	ext4_fsblk_t s_sb_block;
1062 	uid_t s_resuid;
1063 	gid_t s_resgid;
1064 	unsigned short s_mount_state;
1065 	unsigned short s_pad;
1066 	int s_addr_per_block_bits;
1067 	int s_desc_per_block_bits;
1068 	int s_inode_size;
1069 	int s_first_ino;
1070 	unsigned int s_inode_readahead_blks;
1071 	unsigned int s_inode_goal;
1072 	spinlock_t s_next_gen_lock;
1073 	u32 s_next_generation;
1074 	u32 s_hash_seed[4];
1075 	int s_def_hash_version;
1076 	int s_hash_unsigned;	/* 3 if hash should be signed, 0 if not */
1077 	struct percpu_counter s_freeblocks_counter;
1078 	struct percpu_counter s_freeinodes_counter;
1079 	struct percpu_counter s_dirs_counter;
1080 	struct percpu_counter s_dirtyblocks_counter;
1081 	struct blockgroup_lock *s_blockgroup_lock;
1082 	struct proc_dir_entry *s_proc;
1083 	struct kobject s_kobj;
1084 	struct completion s_kobj_unregister;
1085 
1086 	/* Journaling */
1087 	struct inode *s_journal_inode;
1088 	struct journal_s *s_journal;
1089 	struct list_head s_orphan;
1090 	struct mutex s_orphan_lock;
1091 	struct mutex s_resize_lock;
1092 	unsigned long s_commit_interval;
1093 	u32 s_max_batch_time;
1094 	u32 s_min_batch_time;
1095 	struct block_device *journal_bdev;
1096 #ifdef CONFIG_JBD2_DEBUG
1097 	struct timer_list turn_ro_timer;	/* For turning read-only (crash simulation) */
1098 	wait_queue_head_t ro_wait_queue;	/* For people waiting for the fs to go read-only */
1099 #endif
1100 #ifdef CONFIG_QUOTA
1101 	char *s_qf_names[MAXQUOTAS];		/* Names of quota files with journalled quota */
1102 	int s_jquota_fmt;			/* Format of quota to use */
1103 #endif
1104 	unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1105 	struct rb_root system_blks;
1106 
1107 #ifdef EXTENTS_STATS
1108 	/* ext4 extents stats */
1109 	unsigned long s_ext_min;
1110 	unsigned long s_ext_max;
1111 	unsigned long s_depth_max;
1112 	spinlock_t s_ext_stats_lock;
1113 	unsigned long s_ext_blocks;
1114 	unsigned long s_ext_extents;
1115 #endif
1116 
1117 	/* for buddy allocator */
1118 	struct ext4_group_info ***s_group_info;
1119 	struct inode *s_buddy_cache;
1120 	long s_blocks_reserved;
1121 	spinlock_t s_reserve_lock;
1122 	spinlock_t s_md_lock;
1123 	tid_t s_last_transaction;
1124 	unsigned short *s_mb_offsets;
1125 	unsigned int *s_mb_maxs;
1126 
1127 	/* tunables */
1128 	unsigned long s_stripe;
1129 	unsigned int s_mb_stream_request;
1130 	unsigned int s_mb_max_to_scan;
1131 	unsigned int s_mb_min_to_scan;
1132 	unsigned int s_mb_stats;
1133 	unsigned int s_mb_order2_reqs;
1134 	unsigned int s_mb_group_prealloc;
1135 	unsigned int s_max_writeback_mb_bump;
1136 	/* where last allocation was done - for stream allocation */
1137 	unsigned long s_mb_last_group;
1138 	unsigned long s_mb_last_start;
1139 
1140 	/* stats for buddy allocator */
1141 	spinlock_t s_mb_pa_lock;
1142 	atomic_t s_bal_reqs;	/* number of reqs with len > 1 */
1143 	atomic_t s_bal_success;	/* we found long enough chunks */
1144 	atomic_t s_bal_allocated;	/* in blocks */
1145 	atomic_t s_bal_ex_scanned;	/* total extents scanned */
1146 	atomic_t s_bal_goals;	/* goal hits */
1147 	atomic_t s_bal_breaks;	/* too long searches */
1148 	atomic_t s_bal_2orders;	/* 2^order hits */
1149 	spinlock_t s_bal_lock;
1150 	unsigned long s_mb_buddies_generated;
1151 	unsigned long long s_mb_generation_time;
1152 	atomic_t s_mb_lost_chunks;
1153 	atomic_t s_mb_preallocated;
1154 	atomic_t s_mb_discarded;
1155 	atomic_t s_lock_busy;
1156 
1157 	/* locality groups */
1158 	struct ext4_locality_group __percpu *s_locality_groups;
1159 
1160 	/* for write statistics */
1161 	unsigned long s_sectors_written_start;
1162 	u64 s_kbytes_written;
1163 
1164 	unsigned int s_log_groups_per_flex;
1165 	struct flex_groups *s_flex_groups;
1166 
1167 	/* workqueue for dio unwritten */
1168 	struct workqueue_struct *dio_unwritten_wq;
1169 
1170 	/* timer for periodic error stats printing */
1171 	struct timer_list s_err_report;
1172 };
1173 
1174 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1175 {
1176 	return sb->s_fs_info;
1177 }
1178 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1179 {
1180 	return container_of(inode, struct ext4_inode_info, vfs_inode);
1181 }
1182 
1183 static inline struct timespec ext4_current_time(struct inode *inode)
1184 {
1185 	return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1186 		current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1187 }
1188 
1189 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1190 {
1191 	return ino == EXT4_ROOT_INO ||
1192 		ino == EXT4_JOURNAL_INO ||
1193 		ino == EXT4_RESIZE_INO ||
1194 		(ino >= EXT4_FIRST_INO(sb) &&
1195 		 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1196 }
1197 
1198 /*
1199  * Inode dynamic state flags
1200  */
1201 enum {
1202 	EXT4_STATE_JDATA,		/* journaled data exists */
1203 	EXT4_STATE_NEW,			/* inode is newly created */
1204 	EXT4_STATE_XATTR,		/* has in-inode xattrs */
1205 	EXT4_STATE_NO_EXPAND,		/* No space for expansion */
1206 	EXT4_STATE_DA_ALLOC_CLOSE,	/* Alloc DA blks on close */
1207 	EXT4_STATE_EXT_MIGRATE,		/* Inode is migrating */
1208 	EXT4_STATE_DIO_UNWRITTEN,	/* need convert on dio done*/
1209 	EXT4_STATE_NEWENTRY,		/* File just added to dir */
1210 };
1211 
1212 #define EXT4_INODE_BIT_FNS(name, field)					\
1213 static inline int ext4_test_inode_##name(struct inode *inode, int bit)	\
1214 {									\
1215 	return test_bit(bit, &EXT4_I(inode)->i_##field);		\
1216 }									\
1217 static inline void ext4_set_inode_##name(struct inode *inode, int bit)	\
1218 {									\
1219 	set_bit(bit, &EXT4_I(inode)->i_##field);			\
1220 }									\
1221 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1222 {									\
1223 	clear_bit(bit, &EXT4_I(inode)->i_##field);			\
1224 }
1225 
1226 EXT4_INODE_BIT_FNS(flag, flags)
1227 EXT4_INODE_BIT_FNS(state, state_flags)
1228 #else
1229 /* Assume that user mode programs are passing in an ext4fs superblock, not
1230  * a kernel struct super_block.  This will allow us to call the feature-test
1231  * macros from user land. */
1232 #define EXT4_SB(sb)	(sb)
1233 #endif
1234 
1235 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1236 
1237 /*
1238  * Codes for operating systems
1239  */
1240 #define EXT4_OS_LINUX		0
1241 #define EXT4_OS_HURD		1
1242 #define EXT4_OS_MASIX		2
1243 #define EXT4_OS_FREEBSD		3
1244 #define EXT4_OS_LITES		4
1245 
1246 /*
1247  * Revision levels
1248  */
1249 #define EXT4_GOOD_OLD_REV	0	/* The good old (original) format */
1250 #define EXT4_DYNAMIC_REV	1	/* V2 format w/ dynamic inode sizes */
1251 
1252 #define EXT4_CURRENT_REV	EXT4_GOOD_OLD_REV
1253 #define EXT4_MAX_SUPP_REV	EXT4_DYNAMIC_REV
1254 
1255 #define EXT4_GOOD_OLD_INODE_SIZE 128
1256 
1257 /*
1258  * Feature set definitions
1259  */
1260 
1261 #define EXT4_HAS_COMPAT_FEATURE(sb,mask)			\
1262 	((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1263 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask)			\
1264 	((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1265 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask)			\
1266 	((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1267 #define EXT4_SET_COMPAT_FEATURE(sb,mask)			\
1268 	EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1269 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask)			\
1270 	EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1271 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask)			\
1272 	EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1273 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask)			\
1274 	EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1275 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask)			\
1276 	EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1277 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask)			\
1278 	EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1279 
1280 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC	0x0001
1281 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES	0x0002
1282 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL		0x0004
1283 #define EXT4_FEATURE_COMPAT_EXT_ATTR		0x0008
1284 #define EXT4_FEATURE_COMPAT_RESIZE_INODE	0x0010
1285 #define EXT4_FEATURE_COMPAT_DIR_INDEX		0x0020
1286 
1287 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER	0x0001
1288 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE	0x0002
1289 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR	0x0004
1290 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE        0x0008
1291 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM		0x0010
1292 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK	0x0020
1293 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE	0x0040
1294 
1295 #define EXT4_FEATURE_INCOMPAT_COMPRESSION	0x0001
1296 #define EXT4_FEATURE_INCOMPAT_FILETYPE		0x0002
1297 #define EXT4_FEATURE_INCOMPAT_RECOVER		0x0004 /* Needs recovery */
1298 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV	0x0008 /* Journal device */
1299 #define EXT4_FEATURE_INCOMPAT_META_BG		0x0010
1300 #define EXT4_FEATURE_INCOMPAT_EXTENTS		0x0040 /* extents support */
1301 #define EXT4_FEATURE_INCOMPAT_64BIT		0x0080
1302 #define EXT4_FEATURE_INCOMPAT_MMP               0x0100
1303 #define EXT4_FEATURE_INCOMPAT_FLEX_BG		0x0200
1304 #define EXT4_FEATURE_INCOMPAT_EA_INODE		0x0400 /* EA in inode */
1305 #define EXT4_FEATURE_INCOMPAT_DIRDATA		0x1000 /* data in dirent */
1306 
1307 #define EXT4_FEATURE_COMPAT_SUPP	EXT2_FEATURE_COMPAT_EXT_ATTR
1308 #define EXT4_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
1309 					 EXT4_FEATURE_INCOMPAT_RECOVER| \
1310 					 EXT4_FEATURE_INCOMPAT_META_BG| \
1311 					 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1312 					 EXT4_FEATURE_INCOMPAT_64BIT| \
1313 					 EXT4_FEATURE_INCOMPAT_FLEX_BG)
1314 #define EXT4_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1315 					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1316 					 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1317 					 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1318 					 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1319 					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1320 					 EXT4_FEATURE_RO_COMPAT_HUGE_FILE)
1321 
1322 /*
1323  * Default values for user and/or group using reserved blocks
1324  */
1325 #define	EXT4_DEF_RESUID		0
1326 #define	EXT4_DEF_RESGID		0
1327 
1328 #define EXT4_DEF_INODE_READAHEAD_BLKS	32
1329 
1330 /*
1331  * Default mount options
1332  */
1333 #define EXT4_DEFM_DEBUG		0x0001
1334 #define EXT4_DEFM_BSDGROUPS	0x0002
1335 #define EXT4_DEFM_XATTR_USER	0x0004
1336 #define EXT4_DEFM_ACL		0x0008
1337 #define EXT4_DEFM_UID16		0x0010
1338 #define EXT4_DEFM_JMODE		0x0060
1339 #define EXT4_DEFM_JMODE_DATA	0x0020
1340 #define EXT4_DEFM_JMODE_ORDERED	0x0040
1341 #define EXT4_DEFM_JMODE_WBACK	0x0060
1342 
1343 /*
1344  * Default journal batch times
1345  */
1346 #define EXT4_DEF_MIN_BATCH_TIME	0
1347 #define EXT4_DEF_MAX_BATCH_TIME	15000 /* 15ms */
1348 
1349 /*
1350  * Minimum number of groups in a flexgroup before we separate out
1351  * directories into the first block group of a flexgroup
1352  */
1353 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME	4
1354 
1355 /*
1356  * Structure of a directory entry
1357  */
1358 #define EXT4_NAME_LEN 255
1359 
1360 struct ext4_dir_entry {
1361 	__le32	inode;			/* Inode number */
1362 	__le16	rec_len;		/* Directory entry length */
1363 	__le16	name_len;		/* Name length */
1364 	char	name[EXT4_NAME_LEN];	/* File name */
1365 };
1366 
1367 /*
1368  * The new version of the directory entry.  Since EXT4 structures are
1369  * stored in intel byte order, and the name_len field could never be
1370  * bigger than 255 chars, it's safe to reclaim the extra byte for the
1371  * file_type field.
1372  */
1373 struct ext4_dir_entry_2 {
1374 	__le32	inode;			/* Inode number */
1375 	__le16	rec_len;		/* Directory entry length */
1376 	__u8	name_len;		/* Name length */
1377 	__u8	file_type;
1378 	char	name[EXT4_NAME_LEN];	/* File name */
1379 };
1380 
1381 /*
1382  * Ext4 directory file types.  Only the low 3 bits are used.  The
1383  * other bits are reserved for now.
1384  */
1385 #define EXT4_FT_UNKNOWN		0
1386 #define EXT4_FT_REG_FILE	1
1387 #define EXT4_FT_DIR		2
1388 #define EXT4_FT_CHRDEV		3
1389 #define EXT4_FT_BLKDEV		4
1390 #define EXT4_FT_FIFO		5
1391 #define EXT4_FT_SOCK		6
1392 #define EXT4_FT_SYMLINK		7
1393 
1394 #define EXT4_FT_MAX		8
1395 
1396 /*
1397  * EXT4_DIR_PAD defines the directory entries boundaries
1398  *
1399  * NOTE: It must be a multiple of 4
1400  */
1401 #define EXT4_DIR_PAD			4
1402 #define EXT4_DIR_ROUND			(EXT4_DIR_PAD - 1)
1403 #define EXT4_DIR_REC_LEN(name_len)	(((name_len) + 8 + EXT4_DIR_ROUND) & \
1404 					 ~EXT4_DIR_ROUND)
1405 #define EXT4_MAX_REC_LEN		((1<<16)-1)
1406 
1407 /*
1408  * Hash Tree Directory indexing
1409  * (c) Daniel Phillips, 2001
1410  */
1411 
1412 #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1413 				      EXT4_FEATURE_COMPAT_DIR_INDEX) && \
1414 		    ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
1415 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1416 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1417 
1418 /* Legal values for the dx_root hash_version field: */
1419 
1420 #define DX_HASH_LEGACY		0
1421 #define DX_HASH_HALF_MD4	1
1422 #define DX_HASH_TEA		2
1423 #define DX_HASH_LEGACY_UNSIGNED	3
1424 #define DX_HASH_HALF_MD4_UNSIGNED	4
1425 #define DX_HASH_TEA_UNSIGNED		5
1426 
1427 #ifdef __KERNEL__
1428 
1429 /* hash info structure used by the directory hash */
1430 struct dx_hash_info
1431 {
1432 	u32		hash;
1433 	u32		minor_hash;
1434 	int		hash_version;
1435 	u32		*seed;
1436 };
1437 
1438 #define EXT4_HTREE_EOF	0x7fffffff
1439 
1440 /*
1441  * Control parameters used by ext4_htree_next_block
1442  */
1443 #define HASH_NB_ALWAYS		1
1444 
1445 
1446 /*
1447  * Describe an inode's exact location on disk and in memory
1448  */
1449 struct ext4_iloc
1450 {
1451 	struct buffer_head *bh;
1452 	unsigned long offset;
1453 	ext4_group_t block_group;
1454 };
1455 
1456 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
1457 {
1458 	return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
1459 }
1460 
1461 /*
1462  * This structure is stuffed into the struct file's private_data field
1463  * for directories.  It is where we put information so that we can do
1464  * readdir operations in hash tree order.
1465  */
1466 struct dir_private_info {
1467 	struct rb_root	root;
1468 	struct rb_node	*curr_node;
1469 	struct fname	*extra_fname;
1470 	loff_t		last_pos;
1471 	__u32		curr_hash;
1472 	__u32		curr_minor_hash;
1473 	__u32		next_hash;
1474 };
1475 
1476 /* calculate the first block number of the group */
1477 static inline ext4_fsblk_t
1478 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
1479 {
1480 	return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
1481 		le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
1482 }
1483 
1484 /*
1485  * Special error return code only used by dx_probe() and its callers.
1486  */
1487 #define ERR_BAD_DX_DIR	-75000
1488 
1489 void ext4_get_group_no_and_offset(struct super_block *sb, ext4_fsblk_t blocknr,
1490 			ext4_group_t *blockgrpp, ext4_grpblk_t *offsetp);
1491 
1492 extern struct proc_dir_entry *ext4_proc_root;
1493 
1494 /*
1495  * Function prototypes
1496  */
1497 
1498 /*
1499  * Ok, these declarations are also in <linux/kernel.h> but none of the
1500  * ext4 source programs needs to include it so they are duplicated here.
1501  */
1502 # define NORET_TYPE	/**/
1503 # define ATTRIB_NORET	__attribute__((noreturn))
1504 # define NORET_AND	noreturn,
1505 
1506 /* bitmap.c */
1507 extern unsigned int ext4_count_free(struct buffer_head *, unsigned);
1508 
1509 /* balloc.c */
1510 extern unsigned int ext4_block_group(struct super_block *sb,
1511 			ext4_fsblk_t blocknr);
1512 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
1513 			ext4_fsblk_t blocknr);
1514 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
1515 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
1516 			ext4_group_t group);
1517 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
1518 			ext4_fsblk_t goal, unsigned long *count, int *errp);
1519 extern int ext4_claim_free_blocks(struct ext4_sb_info *sbi, s64 nblocks);
1520 extern int ext4_has_free_blocks(struct ext4_sb_info *sbi, s64 nblocks);
1521 extern void ext4_add_groupblocks(handle_t *handle, struct super_block *sb,
1522 				ext4_fsblk_t block, unsigned long count);
1523 extern ext4_fsblk_t ext4_count_free_blocks(struct super_block *);
1524 extern void ext4_check_blocks_bitmap(struct super_block *);
1525 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
1526 						    ext4_group_t block_group,
1527 						    struct buffer_head ** bh);
1528 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
1529 struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
1530 				      ext4_group_t block_group);
1531 extern unsigned ext4_init_block_bitmap(struct super_block *sb,
1532 				       struct buffer_head *bh,
1533 				       ext4_group_t group,
1534 				       struct ext4_group_desc *desc);
1535 #define ext4_free_blocks_after_init(sb, group, desc)			\
1536 		ext4_init_block_bitmap(sb, NULL, group, desc)
1537 
1538 /* dir.c */
1539 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
1540 				  struct ext4_dir_entry_2 *,
1541 				  struct buffer_head *, unsigned int);
1542 #define ext4_check_dir_entry(dir, de, bh, offset) \
1543 	__ext4_check_dir_entry(__func__, __LINE__, (dir), (de), (bh), (offset))
1544 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
1545 				    __u32 minor_hash,
1546 				    struct ext4_dir_entry_2 *dirent);
1547 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
1548 
1549 /* fsync.c */
1550 extern int ext4_sync_file(struct file *, int);
1551 
1552 /* hash.c */
1553 extern int ext4fs_dirhash(const char *name, int len, struct
1554 			  dx_hash_info *hinfo);
1555 
1556 /* ialloc.c */
1557 extern struct inode *ext4_new_inode(handle_t *, struct inode *, int,
1558 				    const struct qstr *qstr, __u32 goal);
1559 extern void ext4_free_inode(handle_t *, struct inode *);
1560 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
1561 extern unsigned long ext4_count_free_inodes(struct super_block *);
1562 extern unsigned long ext4_count_dirs(struct super_block *);
1563 extern void ext4_check_inodes_bitmap(struct super_block *);
1564 extern unsigned ext4_init_inode_bitmap(struct super_block *sb,
1565 				       struct buffer_head *bh,
1566 				       ext4_group_t group,
1567 				       struct ext4_group_desc *desc);
1568 extern void mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
1569 
1570 /* mballoc.c */
1571 extern long ext4_mb_stats;
1572 extern long ext4_mb_max_to_scan;
1573 extern int ext4_mb_init(struct super_block *, int);
1574 extern int ext4_mb_release(struct super_block *);
1575 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
1576 				struct ext4_allocation_request *, int *);
1577 extern int ext4_mb_reserve_blocks(struct super_block *, int);
1578 extern void ext4_discard_preallocations(struct inode *);
1579 extern int __init init_ext4_mballoc(void);
1580 extern void exit_ext4_mballoc(void);
1581 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
1582 			     struct buffer_head *bh, ext4_fsblk_t block,
1583 			     unsigned long count, int flags);
1584 extern int ext4_mb_add_groupinfo(struct super_block *sb,
1585 		ext4_group_t i, struct ext4_group_desc *desc);
1586 extern int ext4_mb_get_buddy_cache_lock(struct super_block *, ext4_group_t);
1587 extern void ext4_mb_put_buddy_cache_lock(struct super_block *,
1588 						ext4_group_t, int);
1589 /* inode.c */
1590 struct buffer_head *ext4_getblk(handle_t *, struct inode *,
1591 						ext4_lblk_t, int, int *);
1592 struct buffer_head *ext4_bread(handle_t *, struct inode *,
1593 						ext4_lblk_t, int, int *);
1594 int ext4_get_block(struct inode *inode, sector_t iblock,
1595 				struct buffer_head *bh_result, int create);
1596 
1597 extern struct inode *ext4_iget(struct super_block *, unsigned long);
1598 extern int  ext4_write_inode(struct inode *, struct writeback_control *);
1599 extern int  ext4_setattr(struct dentry *, struct iattr *);
1600 extern int  ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
1601 				struct kstat *stat);
1602 extern void ext4_delete_inode(struct inode *);
1603 extern int  ext4_sync_inode(handle_t *, struct inode *);
1604 extern void ext4_dirty_inode(struct inode *);
1605 extern int ext4_change_inode_journal_flag(struct inode *, int);
1606 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
1607 extern int ext4_can_truncate(struct inode *inode);
1608 extern void ext4_truncate(struct inode *);
1609 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
1610 extern void ext4_set_inode_flags(struct inode *);
1611 extern void ext4_get_inode_flags(struct ext4_inode_info *);
1612 extern int ext4_alloc_da_blocks(struct inode *inode);
1613 extern void ext4_set_aops(struct inode *inode);
1614 extern int ext4_writepage_trans_blocks(struct inode *);
1615 extern int ext4_meta_trans_blocks(struct inode *, int nrblocks, int idxblocks);
1616 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
1617 extern int ext4_block_truncate_page(handle_t *handle,
1618 		struct address_space *mapping, loff_t from);
1619 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
1620 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
1621 extern int flush_completed_IO(struct inode *inode);
1622 extern void ext4_da_update_reserve_space(struct inode *inode,
1623 					int used, int quota_claim);
1624 /* ioctl.c */
1625 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
1626 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
1627 
1628 /* migrate.c */
1629 extern int ext4_ext_migrate(struct inode *);
1630 
1631 /* namei.c */
1632 extern unsigned int ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize);
1633 extern __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize);
1634 extern int ext4_orphan_add(handle_t *, struct inode *);
1635 extern int ext4_orphan_del(handle_t *, struct inode *);
1636 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
1637 				__u32 start_minor_hash, __u32 *next_hash);
1638 
1639 /* resize.c */
1640 extern int ext4_group_add(struct super_block *sb,
1641 				struct ext4_new_group_data *input);
1642 extern int ext4_group_extend(struct super_block *sb,
1643 				struct ext4_super_block *es,
1644 				ext4_fsblk_t n_blocks_count);
1645 
1646 /* super.c */
1647 extern void __ext4_error(struct super_block *, const char *, unsigned int,
1648 			 const char *, ...)
1649 	__attribute__ ((format (printf, 4, 5)));
1650 #define ext4_error(sb, message...)	__ext4_error(sb, __func__,	\
1651 						     __LINE__, ## message)
1652 extern void ext4_error_inode(struct inode *, const char *, unsigned int,
1653 			     ext4_fsblk_t, const char *, ...)
1654 	__attribute__ ((format (printf, 5, 6)));
1655 extern void ext4_error_file(struct file *, const char *, unsigned int,
1656 			    const char *, ...)
1657 	__attribute__ ((format (printf, 4, 5)));
1658 extern void __ext4_std_error(struct super_block *, const char *,
1659 			     unsigned int, int);
1660 extern void __ext4_abort(struct super_block *, const char *, unsigned int,
1661 		       const char *, ...)
1662 	__attribute__ ((format (printf, 4, 5)));
1663 #define ext4_abort(sb, message...)	__ext4_abort(sb, __func__, \
1664 						       __LINE__, ## message)
1665 extern void __ext4_warning(struct super_block *, const char *, unsigned int,
1666 			  const char *, ...)
1667 	__attribute__ ((format (printf, 4, 5)));
1668 #define ext4_warning(sb, message...)	__ext4_warning(sb, __func__, \
1669 						       __LINE__, ## message)
1670 extern void ext4_msg(struct super_block *, const char *, const char *, ...)
1671 	__attribute__ ((format (printf, 3, 4)));
1672 extern void __ext4_grp_locked_error(const char *, unsigned int, \
1673 				    struct super_block *, ext4_group_t, \
1674 				    unsigned long, ext4_fsblk_t, \
1675 				    const char *, ...)
1676 	__attribute__ ((format (printf, 7, 8)));
1677 #define ext4_grp_locked_error(sb, grp, message...) \
1678 	__ext4_grp_locked_error(__func__, __LINE__, (sb), (grp), ## message)
1679 extern void ext4_update_dynamic_rev(struct super_block *sb);
1680 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
1681 					__u32 compat);
1682 extern int ext4_update_rocompat_feature(handle_t *handle,
1683 					struct super_block *sb,	__u32 rocompat);
1684 extern int ext4_update_incompat_feature(handle_t *handle,
1685 					struct super_block *sb,	__u32 incompat);
1686 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
1687 				      struct ext4_group_desc *bg);
1688 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
1689 				      struct ext4_group_desc *bg);
1690 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
1691 				     struct ext4_group_desc *bg);
1692 extern __u32 ext4_free_blks_count(struct super_block *sb,
1693 				struct ext4_group_desc *bg);
1694 extern __u32 ext4_free_inodes_count(struct super_block *sb,
1695 				 struct ext4_group_desc *bg);
1696 extern __u32 ext4_used_dirs_count(struct super_block *sb,
1697 				struct ext4_group_desc *bg);
1698 extern __u32 ext4_itable_unused_count(struct super_block *sb,
1699 				   struct ext4_group_desc *bg);
1700 extern void ext4_block_bitmap_set(struct super_block *sb,
1701 				  struct ext4_group_desc *bg, ext4_fsblk_t blk);
1702 extern void ext4_inode_bitmap_set(struct super_block *sb,
1703 				  struct ext4_group_desc *bg, ext4_fsblk_t blk);
1704 extern void ext4_inode_table_set(struct super_block *sb,
1705 				 struct ext4_group_desc *bg, ext4_fsblk_t blk);
1706 extern void ext4_free_blks_set(struct super_block *sb,
1707 			       struct ext4_group_desc *bg, __u32 count);
1708 extern void ext4_free_inodes_set(struct super_block *sb,
1709 				struct ext4_group_desc *bg, __u32 count);
1710 extern void ext4_used_dirs_set(struct super_block *sb,
1711 				struct ext4_group_desc *bg, __u32 count);
1712 extern void ext4_itable_unused_set(struct super_block *sb,
1713 				   struct ext4_group_desc *bg, __u32 count);
1714 extern __le16 ext4_group_desc_csum(struct ext4_sb_info *sbi, __u32 group,
1715 				   struct ext4_group_desc *gdp);
1716 extern int ext4_group_desc_csum_verify(struct ext4_sb_info *sbi, __u32 group,
1717 				       struct ext4_group_desc *gdp);
1718 
1719 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
1720 {
1721 	return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
1722 		le32_to_cpu(es->s_blocks_count_lo);
1723 }
1724 
1725 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
1726 {
1727 	return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
1728 		le32_to_cpu(es->s_r_blocks_count_lo);
1729 }
1730 
1731 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
1732 {
1733 	return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
1734 		le32_to_cpu(es->s_free_blocks_count_lo);
1735 }
1736 
1737 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
1738 					 ext4_fsblk_t blk)
1739 {
1740 	es->s_blocks_count_lo = cpu_to_le32((u32)blk);
1741 	es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
1742 }
1743 
1744 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
1745 					      ext4_fsblk_t blk)
1746 {
1747 	es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
1748 	es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
1749 }
1750 
1751 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
1752 					   ext4_fsblk_t blk)
1753 {
1754 	es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
1755 	es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
1756 }
1757 
1758 static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
1759 {
1760 	if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
1761 		return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
1762 			le32_to_cpu(raw_inode->i_size_lo);
1763 	else
1764 		return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
1765 }
1766 
1767 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
1768 {
1769 	raw_inode->i_size_lo = cpu_to_le32(i_size);
1770 	raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
1771 }
1772 
1773 static inline
1774 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
1775 					    ext4_group_t group)
1776 {
1777 	 struct ext4_group_info ***grp_info;
1778 	 long indexv, indexh;
1779 	 grp_info = EXT4_SB(sb)->s_group_info;
1780 	 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
1781 	 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
1782 	 return grp_info[indexv][indexh];
1783 }
1784 
1785 /*
1786  * Reading s_groups_count requires using smp_rmb() afterwards.  See
1787  * the locking protocol documented in the comments of ext4_group_add()
1788  * in resize.c
1789  */
1790 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
1791 {
1792 	ext4_group_t	ngroups = EXT4_SB(sb)->s_groups_count;
1793 
1794 	smp_rmb();
1795 	return ngroups;
1796 }
1797 
1798 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
1799 					     ext4_group_t block_group)
1800 {
1801 	return block_group >> sbi->s_log_groups_per_flex;
1802 }
1803 
1804 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
1805 {
1806 	return 1 << sbi->s_log_groups_per_flex;
1807 }
1808 
1809 #define ext4_std_error(sb, errno)				\
1810 do {								\
1811 	if ((errno))						\
1812 		__ext4_std_error((sb), __func__, __LINE__, (errno));	\
1813 } while (0)
1814 
1815 #ifdef CONFIG_SMP
1816 /* Each CPU can accumulate percpu_counter_batch blocks in their local
1817  * counters. So we need to make sure we have free blocks more
1818  * than percpu_counter_batch  * nr_cpu_ids. Also add a window of 4 times.
1819  */
1820 #define EXT4_FREEBLOCKS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
1821 #else
1822 #define EXT4_FREEBLOCKS_WATERMARK 0
1823 #endif
1824 
1825 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
1826 {
1827 	/*
1828 	 * XXX: replace with spinlock if seen contended -bzzz
1829 	 */
1830 	down_write(&EXT4_I(inode)->i_data_sem);
1831 	if (newsize > EXT4_I(inode)->i_disksize)
1832 		EXT4_I(inode)->i_disksize = newsize;
1833 	up_write(&EXT4_I(inode)->i_data_sem);
1834 	return ;
1835 }
1836 
1837 struct ext4_group_info {
1838 	unsigned long   bb_state;
1839 	struct rb_root  bb_free_root;
1840 	ext4_grpblk_t	bb_first_free;	/* first free block */
1841 	ext4_grpblk_t	bb_free;	/* total free blocks */
1842 	ext4_grpblk_t	bb_fragments;	/* nr of freespace fragments */
1843 	ext4_grpblk_t	bb_largest_free_order;/* order of largest frag in BG */
1844 	struct          list_head bb_prealloc_list;
1845 #ifdef DOUBLE_CHECK
1846 	void            *bb_bitmap;
1847 #endif
1848 	struct rw_semaphore alloc_sem;
1849 	ext4_grpblk_t	bb_counters[];	/* Nr of free power-of-two-block
1850 					 * regions, index is order.
1851 					 * bb_counters[3] = 5 means
1852 					 * 5 free 8-block regions. */
1853 };
1854 
1855 #define EXT4_GROUP_INFO_NEED_INIT_BIT	0
1856 
1857 #define EXT4_MB_GRP_NEED_INIT(grp)	\
1858 	(test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
1859 
1860 #define EXT4_MAX_CONTENTION		8
1861 #define EXT4_CONTENTION_THRESHOLD	2
1862 
1863 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
1864 					      ext4_group_t group)
1865 {
1866 	return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
1867 }
1868 
1869 /*
1870  * Returns true if the filesystem is busy enough that attempts to
1871  * access the block group locks has run into contention.
1872  */
1873 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
1874 {
1875 	return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
1876 }
1877 
1878 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
1879 {
1880 	spinlock_t *lock = ext4_group_lock_ptr(sb, group);
1881 	if (spin_trylock(lock))
1882 		/*
1883 		 * We're able to grab the lock right away, so drop the
1884 		 * lock contention counter.
1885 		 */
1886 		atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
1887 	else {
1888 		/*
1889 		 * The lock is busy, so bump the contention counter,
1890 		 * and then wait on the spin lock.
1891 		 */
1892 		atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
1893 				  EXT4_MAX_CONTENTION);
1894 		spin_lock(lock);
1895 	}
1896 }
1897 
1898 static inline void ext4_unlock_group(struct super_block *sb,
1899 					ext4_group_t group)
1900 {
1901 	spin_unlock(ext4_group_lock_ptr(sb, group));
1902 }
1903 
1904 static inline void ext4_mark_super_dirty(struct super_block *sb)
1905 {
1906 	if (EXT4_SB(sb)->s_journal == NULL)
1907 		sb->s_dirt =1;
1908 }
1909 
1910 /*
1911  * Inodes and files operations
1912  */
1913 
1914 /* dir.c */
1915 extern const struct file_operations ext4_dir_operations;
1916 
1917 /* file.c */
1918 extern const struct inode_operations ext4_file_inode_operations;
1919 extern const struct file_operations ext4_file_operations;
1920 
1921 /* namei.c */
1922 extern const struct inode_operations ext4_dir_inode_operations;
1923 extern const struct inode_operations ext4_special_inode_operations;
1924 extern struct dentry *ext4_get_parent(struct dentry *child);
1925 
1926 /* symlink.c */
1927 extern const struct inode_operations ext4_symlink_inode_operations;
1928 extern const struct inode_operations ext4_fast_symlink_inode_operations;
1929 
1930 /* block_validity */
1931 extern void ext4_release_system_zone(struct super_block *sb);
1932 extern int ext4_setup_system_zone(struct super_block *sb);
1933 extern int __init init_ext4_system_zone(void);
1934 extern void exit_ext4_system_zone(void);
1935 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
1936 				 ext4_fsblk_t start_blk,
1937 				 unsigned int count);
1938 
1939 /* extents.c */
1940 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
1941 extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
1942 extern int ext4_ext_index_trans_blocks(struct inode *inode, int nrblocks,
1943 				       int chunk);
1944 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
1945 			       struct ext4_map_blocks *map, int flags);
1946 extern void ext4_ext_truncate(struct inode *);
1947 extern void ext4_ext_init(struct super_block *);
1948 extern void ext4_ext_release(struct super_block *);
1949 extern long ext4_fallocate(struct inode *inode, int mode, loff_t offset,
1950 			  loff_t len);
1951 extern int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset,
1952 			  ssize_t len);
1953 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
1954 			   struct ext4_map_blocks *map, int flags);
1955 extern int ext4_get_blocks(handle_t *handle, struct inode *inode,
1956 			   sector_t block, unsigned int max_blocks,
1957 			   struct buffer_head *bh, int flags);
1958 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
1959 			__u64 start, __u64 len);
1960 /* move_extent.c */
1961 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
1962 			     __u64 start_orig, __u64 start_donor,
1963 			     __u64 len, __u64 *moved_len);
1964 
1965 
1966 /* BH_Uninit flag: blocks are allocated but uninitialized on disk */
1967 enum ext4_state_bits {
1968 	BH_Uninit	/* blocks are allocated but uninitialized on disk */
1969 	  = BH_JBDPrivateStart,
1970 };
1971 
1972 BUFFER_FNS(Uninit, uninit)
1973 TAS_BUFFER_FNS(Uninit, uninit)
1974 
1975 /*
1976  * Add new method to test wether block and inode bitmaps are properly
1977  * initialized. With uninit_bg reading the block from disk is not enough
1978  * to mark the bitmap uptodate. We need to also zero-out the bitmap
1979  */
1980 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
1981 
1982 static inline int bitmap_uptodate(struct buffer_head *bh)
1983 {
1984 	return (buffer_uptodate(bh) &&
1985 			test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
1986 }
1987 static inline void set_bitmap_uptodate(struct buffer_head *bh)
1988 {
1989 	set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
1990 }
1991 
1992 #define in_range(b, first, len)	((b) >= (first) && (b) <= (first) + (len) - 1)
1993 
1994 #endif	/* __KERNEL__ */
1995 
1996 #endif	/* _EXT4_H */
1997