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#include <linux/ratelimit.h>
33#include <crypto/hash.h>
34#include <linux/falloc.h>
35#ifdef __KERNEL__
36#include <linux/compat.h>
37#endif
38
39/*
40 * The fourth extended filesystem constants/structures
41 */
42
43/*
44 * Define EXT4FS_DEBUG to produce debug messages
45 */
46#undef EXT4FS_DEBUG
47
48/*
49 * Debug code
50 */
51#ifdef EXT4FS_DEBUG
52#define ext4_debug(f, a...)						\
53	do {								\
54		printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:",	\
55			__FILE__, __LINE__, __func__);			\
56		printk(KERN_DEBUG f, ## a);				\
57	} while (0)
58#else
59#define ext4_debug(fmt, ...)	no_printk(fmt, ##__VA_ARGS__)
60#endif
61
62/*
63 * Turn on EXT_DEBUG to get lots of info about extents operations.
64 */
65#define EXT_DEBUG__
66#ifdef EXT_DEBUG
67#define ext_debug(fmt, ...)	printk(fmt, ##__VA_ARGS__)
68#else
69#define ext_debug(fmt, ...)	no_printk(fmt, ##__VA_ARGS__)
70#endif
71
72#define EXT4_ERROR_INODE(inode, fmt, a...) \
73	ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
74
75#define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...)			\
76	ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
77
78#define EXT4_ERROR_FILE(file, block, fmt, a...)				\
79	ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
80
81/* data type for block offset of block group */
82typedef int ext4_grpblk_t;
83
84/* data type for filesystem-wide blocks number */
85typedef unsigned long long ext4_fsblk_t;
86
87/* data type for file logical block number */
88typedef __u32 ext4_lblk_t;
89
90/* data type for block group number */
91typedef unsigned int ext4_group_t;
92
93/*
94 * Flags used in mballoc's allocation_context flags field.
95 *
96 * Also used to show what's going on for debugging purposes when the
97 * flag field is exported via the traceport interface
98 */
99
100/* prefer goal again. length */
101#define EXT4_MB_HINT_MERGE		0x0001
102/* blocks already reserved */
103#define EXT4_MB_HINT_RESERVED		0x0002
104/* metadata is being allocated */
105#define EXT4_MB_HINT_METADATA		0x0004
106/* first blocks in the file */
107#define EXT4_MB_HINT_FIRST		0x0008
108/* search for the best chunk */
109#define EXT4_MB_HINT_BEST		0x0010
110/* data is being allocated */
111#define EXT4_MB_HINT_DATA		0x0020
112/* don't preallocate (for tails) */
113#define EXT4_MB_HINT_NOPREALLOC		0x0040
114/* allocate for locality group */
115#define EXT4_MB_HINT_GROUP_ALLOC	0x0080
116/* allocate goal blocks or none */
117#define EXT4_MB_HINT_GOAL_ONLY		0x0100
118/* goal is meaningful */
119#define EXT4_MB_HINT_TRY_GOAL		0x0200
120/* blocks already pre-reserved by delayed allocation */
121#define EXT4_MB_DELALLOC_RESERVED	0x0400
122/* We are doing stream allocation */
123#define EXT4_MB_STREAM_ALLOC		0x0800
124/* Use reserved root blocks if needed */
125#define EXT4_MB_USE_ROOT_BLOCKS		0x1000
126/* Use blocks from reserved pool */
127#define EXT4_MB_USE_RESERVED		0x2000
128
129struct ext4_allocation_request {
130	/* target inode for block we're allocating */
131	struct inode *inode;
132	/* how many blocks we want to allocate */
133	unsigned int len;
134	/* logical block in target inode */
135	ext4_lblk_t logical;
136	/* the closest logical allocated block to the left */
137	ext4_lblk_t lleft;
138	/* the closest logical allocated block to the right */
139	ext4_lblk_t lright;
140	/* phys. target (a hint) */
141	ext4_fsblk_t goal;
142	/* phys. block for the closest logical allocated block to the left */
143	ext4_fsblk_t pleft;
144	/* phys. block for the closest logical allocated block to the right */
145	ext4_fsblk_t pright;
146	/* flags. see above EXT4_MB_HINT_* */
147	unsigned int flags;
148};
149
150/*
151 * Logical to physical block mapping, used by ext4_map_blocks()
152 *
153 * This structure is used to pass requests into ext4_map_blocks() as
154 * well as to store the information returned by ext4_map_blocks().  It
155 * takes less room on the stack than a struct buffer_head.
156 */
157#define EXT4_MAP_NEW		(1 << BH_New)
158#define EXT4_MAP_MAPPED		(1 << BH_Mapped)
159#define EXT4_MAP_UNWRITTEN	(1 << BH_Unwritten)
160#define EXT4_MAP_BOUNDARY	(1 << BH_Boundary)
161#define EXT4_MAP_FLAGS		(EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
162				 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
163
164struct ext4_map_blocks {
165	ext4_fsblk_t m_pblk;
166	ext4_lblk_t m_lblk;
167	unsigned int m_len;
168	unsigned int m_flags;
169};
170
171/*
172 * Flags for ext4_io_end->flags
173 */
174#define	EXT4_IO_END_UNWRITTEN	0x0001
175
176/*
177 * For converting unwritten extents on a work queue. 'handle' is used for
178 * buffered writeback.
179 */
180typedef struct ext4_io_end {
181	struct list_head	list;		/* per-file finished IO list */
182	handle_t		*handle;	/* handle reserved for extent
183						 * conversion */
184	struct inode		*inode;		/* file being written to */
185	struct bio		*bio;		/* Linked list of completed
186						 * bios covering the extent */
187	unsigned int		flag;		/* unwritten or not */
188	loff_t			offset;		/* offset in the file */
189	ssize_t			size;		/* size of the extent */
190	atomic_t		count;		/* reference counter */
191} ext4_io_end_t;
192
193struct ext4_io_submit {
194	int			io_op;
195	struct bio		*io_bio;
196	ext4_io_end_t		*io_end;
197	sector_t		io_next_block;
198};
199
200/*
201 * Special inodes numbers
202 */
203#define	EXT4_BAD_INO		 1	/* Bad blocks inode */
204#define EXT4_ROOT_INO		 2	/* Root inode */
205#define EXT4_USR_QUOTA_INO	 3	/* User quota inode */
206#define EXT4_GRP_QUOTA_INO	 4	/* Group quota inode */
207#define EXT4_BOOT_LOADER_INO	 5	/* Boot loader inode */
208#define EXT4_UNDEL_DIR_INO	 6	/* Undelete directory inode */
209#define EXT4_RESIZE_INO		 7	/* Reserved group descriptors inode */
210#define EXT4_JOURNAL_INO	 8	/* Journal inode */
211
212/* First non-reserved inode for old ext4 filesystems */
213#define EXT4_GOOD_OLD_FIRST_INO	11
214
215/*
216 * Maximal count of links to a file
217 */
218#define EXT4_LINK_MAX		65000
219
220/*
221 * Macro-instructions used to manage several block sizes
222 */
223#define EXT4_MIN_BLOCK_SIZE		1024
224#define	EXT4_MAX_BLOCK_SIZE		65536
225#define EXT4_MIN_BLOCK_LOG_SIZE		10
226#define EXT4_MAX_BLOCK_LOG_SIZE		16
227#ifdef __KERNEL__
228# define EXT4_BLOCK_SIZE(s)		((s)->s_blocksize)
229#else
230# define EXT4_BLOCK_SIZE(s)		(EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
231#endif
232#define	EXT4_ADDR_PER_BLOCK(s)		(EXT4_BLOCK_SIZE(s) / sizeof(__u32))
233#define EXT4_CLUSTER_SIZE(s)		(EXT4_BLOCK_SIZE(s) << \
234					 EXT4_SB(s)->s_cluster_bits)
235#ifdef __KERNEL__
236# define EXT4_BLOCK_SIZE_BITS(s)	((s)->s_blocksize_bits)
237# define EXT4_CLUSTER_BITS(s)		(EXT4_SB(s)->s_cluster_bits)
238#else
239# define EXT4_BLOCK_SIZE_BITS(s)	((s)->s_log_block_size + 10)
240#endif
241#ifdef __KERNEL__
242#define	EXT4_ADDR_PER_BLOCK_BITS(s)	(EXT4_SB(s)->s_addr_per_block_bits)
243#define EXT4_INODE_SIZE(s)		(EXT4_SB(s)->s_inode_size)
244#define EXT4_FIRST_INO(s)		(EXT4_SB(s)->s_first_ino)
245#else
246#define EXT4_INODE_SIZE(s)	(((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
247				 EXT4_GOOD_OLD_INODE_SIZE : \
248				 (s)->s_inode_size)
249#define EXT4_FIRST_INO(s)	(((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
250				 EXT4_GOOD_OLD_FIRST_INO : \
251				 (s)->s_first_ino)
252#endif
253#define EXT4_BLOCK_ALIGN(size, blkbits)		ALIGN((size), (1 << (blkbits)))
254
255/* Translate a block number to a cluster number */
256#define EXT4_B2C(sbi, blk)	((blk) >> (sbi)->s_cluster_bits)
257/* Translate a cluster number to a block number */
258#define EXT4_C2B(sbi, cluster)	((cluster) << (sbi)->s_cluster_bits)
259/* Translate # of blks to # of clusters */
260#define EXT4_NUM_B2C(sbi, blks)	(((blks) + (sbi)->s_cluster_ratio - 1) >> \
261				 (sbi)->s_cluster_bits)
262/* Mask out the low bits to get the starting block of the cluster */
263#define EXT4_PBLK_CMASK(s, pblk) ((pblk) &				\
264				  ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
265#define EXT4_LBLK_CMASK(s, lblk) ((lblk) &				\
266				  ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
267/* Get the cluster offset */
268#define EXT4_PBLK_COFF(s, pblk) ((pblk) &				\
269				 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
270#define EXT4_LBLK_COFF(s, lblk) ((lblk) &				\
271				 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
272
273/*
274 * Structure of a blocks group descriptor
275 */
276struct ext4_group_desc
277{
278	__le32	bg_block_bitmap_lo;	/* Blocks bitmap block */
279	__le32	bg_inode_bitmap_lo;	/* Inodes bitmap block */
280	__le32	bg_inode_table_lo;	/* Inodes table block */
281	__le16	bg_free_blocks_count_lo;/* Free blocks count */
282	__le16	bg_free_inodes_count_lo;/* Free inodes count */
283	__le16	bg_used_dirs_count_lo;	/* Directories count */
284	__le16	bg_flags;		/* EXT4_BG_flags (INODE_UNINIT, etc) */
285	__le32  bg_exclude_bitmap_lo;   /* Exclude bitmap for snapshots */
286	__le16  bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
287	__le16  bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
288	__le16  bg_itable_unused_lo;	/* Unused inodes count */
289	__le16  bg_checksum;		/* crc16(sb_uuid+group+desc) */
290	__le32	bg_block_bitmap_hi;	/* Blocks bitmap block MSB */
291	__le32	bg_inode_bitmap_hi;	/* Inodes bitmap block MSB */
292	__le32	bg_inode_table_hi;	/* Inodes table block MSB */
293	__le16	bg_free_blocks_count_hi;/* Free blocks count MSB */
294	__le16	bg_free_inodes_count_hi;/* Free inodes count MSB */
295	__le16	bg_used_dirs_count_hi;	/* Directories count MSB */
296	__le16  bg_itable_unused_hi;    /* Unused inodes count MSB */
297	__le32  bg_exclude_bitmap_hi;   /* Exclude bitmap block MSB */
298	__le16  bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
299	__le16  bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
300	__u32   bg_reserved;
301};
302
303#define EXT4_BG_INODE_BITMAP_CSUM_HI_END	\
304	(offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
305	 sizeof(__le16))
306#define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END	\
307	(offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
308	 sizeof(__le16))
309
310/*
311 * Structure of a flex block group info
312 */
313
314struct flex_groups {
315	atomic64_t	free_clusters;
316	atomic_t	free_inodes;
317	atomic_t	used_dirs;
318};
319
320#define EXT4_BG_INODE_UNINIT	0x0001 /* Inode table/bitmap not in use */
321#define EXT4_BG_BLOCK_UNINIT	0x0002 /* Block bitmap not in use */
322#define EXT4_BG_INODE_ZEROED	0x0004 /* On-disk itable initialized to zero */
323
324/*
325 * Macro-instructions used to manage group descriptors
326 */
327#define EXT4_MIN_DESC_SIZE		32
328#define EXT4_MIN_DESC_SIZE_64BIT	64
329#define	EXT4_MAX_DESC_SIZE		EXT4_MIN_BLOCK_SIZE
330#define EXT4_DESC_SIZE(s)		(EXT4_SB(s)->s_desc_size)
331#ifdef __KERNEL__
332# define EXT4_BLOCKS_PER_GROUP(s)	(EXT4_SB(s)->s_blocks_per_group)
333# define EXT4_CLUSTERS_PER_GROUP(s)	(EXT4_SB(s)->s_clusters_per_group)
334# define EXT4_DESC_PER_BLOCK(s)		(EXT4_SB(s)->s_desc_per_block)
335# define EXT4_INODES_PER_GROUP(s)	(EXT4_SB(s)->s_inodes_per_group)
336# define EXT4_DESC_PER_BLOCK_BITS(s)	(EXT4_SB(s)->s_desc_per_block_bits)
337#else
338# define EXT4_BLOCKS_PER_GROUP(s)	((s)->s_blocks_per_group)
339# define EXT4_DESC_PER_BLOCK(s)		(EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
340# define EXT4_INODES_PER_GROUP(s)	((s)->s_inodes_per_group)
341#endif
342
343/*
344 * Constants relative to the data blocks
345 */
346#define	EXT4_NDIR_BLOCKS		12
347#define	EXT4_IND_BLOCK			EXT4_NDIR_BLOCKS
348#define	EXT4_DIND_BLOCK			(EXT4_IND_BLOCK + 1)
349#define	EXT4_TIND_BLOCK			(EXT4_DIND_BLOCK + 1)
350#define	EXT4_N_BLOCKS			(EXT4_TIND_BLOCK + 1)
351
352/*
353 * Inode flags
354 */
355#define	EXT4_SECRM_FL			0x00000001 /* Secure deletion */
356#define	EXT4_UNRM_FL			0x00000002 /* Undelete */
357#define	EXT4_COMPR_FL			0x00000004 /* Compress file */
358#define EXT4_SYNC_FL			0x00000008 /* Synchronous updates */
359#define EXT4_IMMUTABLE_FL		0x00000010 /* Immutable file */
360#define EXT4_APPEND_FL			0x00000020 /* writes to file may only append */
361#define EXT4_NODUMP_FL			0x00000040 /* do not dump file */
362#define EXT4_NOATIME_FL			0x00000080 /* do not update atime */
363/* Reserved for compression usage... */
364#define EXT4_DIRTY_FL			0x00000100
365#define EXT4_COMPRBLK_FL		0x00000200 /* One or more compressed clusters */
366#define EXT4_NOCOMPR_FL			0x00000400 /* Don't compress */
367	/* nb: was previously EXT2_ECOMPR_FL */
368#define EXT4_ENCRYPT_FL			0x00000800 /* encrypted file */
369/* End compression flags --- maybe not all used */
370#define EXT4_INDEX_FL			0x00001000 /* hash-indexed directory */
371#define EXT4_IMAGIC_FL			0x00002000 /* AFS directory */
372#define EXT4_JOURNAL_DATA_FL		0x00004000 /* file data should be journaled */
373#define EXT4_NOTAIL_FL			0x00008000 /* file tail should not be merged */
374#define EXT4_DIRSYNC_FL			0x00010000 /* dirsync behaviour (directories only) */
375#define EXT4_TOPDIR_FL			0x00020000 /* Top of directory hierarchies*/
376#define EXT4_HUGE_FILE_FL               0x00040000 /* Set to each huge file */
377#define EXT4_EXTENTS_FL			0x00080000 /* Inode uses extents */
378#define EXT4_EA_INODE_FL	        0x00200000 /* Inode used for large EA */
379#define EXT4_EOFBLOCKS_FL		0x00400000 /* Blocks allocated beyond EOF */
380#define EXT4_INLINE_DATA_FL		0x10000000 /* Inode has inline data. */
381#define EXT4_RESERVED_FL		0x80000000 /* reserved for ext4 lib */
382
383#define EXT4_FL_USER_VISIBLE		0x004BDFFF /* User visible flags */
384#define EXT4_FL_USER_MODIFIABLE		0x004380FF /* User modifiable flags */
385
386/* Flags that should be inherited by new inodes from their parent. */
387#define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
388			   EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
389			   EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
390			   EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
391
392/* Flags that are appropriate for regular files (all but dir-specific ones). */
393#define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
394
395/* Flags that are appropriate for non-directories/regular files. */
396#define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
397
398/* Mask out flags that are inappropriate for the given type of inode. */
399static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
400{
401	if (S_ISDIR(mode))
402		return flags;
403	else if (S_ISREG(mode))
404		return flags & EXT4_REG_FLMASK;
405	else
406		return flags & EXT4_OTHER_FLMASK;
407}
408
409/*
410 * Inode flags used for atomic set/get
411 */
412enum {
413	EXT4_INODE_SECRM	= 0,	/* Secure deletion */
414	EXT4_INODE_UNRM		= 1,	/* Undelete */
415	EXT4_INODE_COMPR	= 2,	/* Compress file */
416	EXT4_INODE_SYNC		= 3,	/* Synchronous updates */
417	EXT4_INODE_IMMUTABLE	= 4,	/* Immutable file */
418	EXT4_INODE_APPEND	= 5,	/* writes to file may only append */
419	EXT4_INODE_NODUMP	= 6,	/* do not dump file */
420	EXT4_INODE_NOATIME	= 7,	/* do not update atime */
421/* Reserved for compression usage... */
422	EXT4_INODE_DIRTY	= 8,
423	EXT4_INODE_COMPRBLK	= 9,	/* One or more compressed clusters */
424	EXT4_INODE_NOCOMPR	= 10,	/* Don't compress */
425	EXT4_INODE_ENCRYPT	= 11,	/* Encrypted file */
426/* End compression flags --- maybe not all used */
427	EXT4_INODE_INDEX	= 12,	/* hash-indexed directory */
428	EXT4_INODE_IMAGIC	= 13,	/* AFS directory */
429	EXT4_INODE_JOURNAL_DATA	= 14,	/* file data should be journaled */
430	EXT4_INODE_NOTAIL	= 15,	/* file tail should not be merged */
431	EXT4_INODE_DIRSYNC	= 16,	/* dirsync behaviour (directories only) */
432	EXT4_INODE_TOPDIR	= 17,	/* Top of directory hierarchies*/
433	EXT4_INODE_HUGE_FILE	= 18,	/* Set to each huge file */
434	EXT4_INODE_EXTENTS	= 19,	/* Inode uses extents */
435	EXT4_INODE_EA_INODE	= 21,	/* Inode used for large EA */
436	EXT4_INODE_EOFBLOCKS	= 22,	/* Blocks allocated beyond EOF */
437	EXT4_INODE_INLINE_DATA	= 28,	/* Data in inode. */
438	EXT4_INODE_RESERVED	= 31,	/* reserved for ext4 lib */
439};
440
441/*
442 * Since it's pretty easy to mix up bit numbers and hex values, we use a
443 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
444 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
445 * any extra space in the compiled kernel image, otherwise, the build will fail.
446 * It's important that these values are the same, since we are using
447 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
448 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
449 * values found in ext2, ext3 and ext4 filesystems, and of course the values
450 * defined in e2fsprogs.
451 *
452 * It's not paranoia if the Murphy's Law really *is* out to get you.  :-)
453 */
454#define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
455#define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
456
457static inline void ext4_check_flag_values(void)
458{
459	CHECK_FLAG_VALUE(SECRM);
460	CHECK_FLAG_VALUE(UNRM);
461	CHECK_FLAG_VALUE(COMPR);
462	CHECK_FLAG_VALUE(SYNC);
463	CHECK_FLAG_VALUE(IMMUTABLE);
464	CHECK_FLAG_VALUE(APPEND);
465	CHECK_FLAG_VALUE(NODUMP);
466	CHECK_FLAG_VALUE(NOATIME);
467	CHECK_FLAG_VALUE(DIRTY);
468	CHECK_FLAG_VALUE(COMPRBLK);
469	CHECK_FLAG_VALUE(NOCOMPR);
470	CHECK_FLAG_VALUE(ENCRYPT);
471	CHECK_FLAG_VALUE(INDEX);
472	CHECK_FLAG_VALUE(IMAGIC);
473	CHECK_FLAG_VALUE(JOURNAL_DATA);
474	CHECK_FLAG_VALUE(NOTAIL);
475	CHECK_FLAG_VALUE(DIRSYNC);
476	CHECK_FLAG_VALUE(TOPDIR);
477	CHECK_FLAG_VALUE(HUGE_FILE);
478	CHECK_FLAG_VALUE(EXTENTS);
479	CHECK_FLAG_VALUE(EA_INODE);
480	CHECK_FLAG_VALUE(EOFBLOCKS);
481	CHECK_FLAG_VALUE(INLINE_DATA);
482	CHECK_FLAG_VALUE(RESERVED);
483}
484
485/* Used to pass group descriptor data when online resize is done */
486struct ext4_new_group_input {
487	__u32 group;		/* Group number for this data */
488	__u64 block_bitmap;	/* Absolute block number of block bitmap */
489	__u64 inode_bitmap;	/* Absolute block number of inode bitmap */
490	__u64 inode_table;	/* Absolute block number of inode table start */
491	__u32 blocks_count;	/* Total number of blocks in this group */
492	__u16 reserved_blocks;	/* Number of reserved blocks in this group */
493	__u16 unused;
494};
495
496#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
497struct compat_ext4_new_group_input {
498	u32 group;
499	compat_u64 block_bitmap;
500	compat_u64 inode_bitmap;
501	compat_u64 inode_table;
502	u32 blocks_count;
503	u16 reserved_blocks;
504	u16 unused;
505};
506#endif
507
508/* The struct ext4_new_group_input in kernel space, with free_blocks_count */
509struct ext4_new_group_data {
510	__u32 group;
511	__u64 block_bitmap;
512	__u64 inode_bitmap;
513	__u64 inode_table;
514	__u32 blocks_count;
515	__u16 reserved_blocks;
516	__u16 unused;
517	__u32 free_blocks_count;
518};
519
520/* Indexes used to index group tables in ext4_new_group_data */
521enum {
522	BLOCK_BITMAP = 0,	/* block bitmap */
523	INODE_BITMAP,		/* inode bitmap */
524	INODE_TABLE,		/* inode tables */
525	GROUP_TABLE_COUNT,
526};
527
528/*
529 * Flags used by ext4_map_blocks()
530 */
531	/* Allocate any needed blocks and/or convert an unwritten
532	   extent to be an initialized ext4 */
533#define EXT4_GET_BLOCKS_CREATE			0x0001
534	/* Request the creation of an unwritten extent */
535#define EXT4_GET_BLOCKS_UNWRIT_EXT		0x0002
536#define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT	(EXT4_GET_BLOCKS_UNWRIT_EXT|\
537						 EXT4_GET_BLOCKS_CREATE)
538	/* Caller is from the delayed allocation writeout path
539	 * finally doing the actual allocation of delayed blocks */
540#define EXT4_GET_BLOCKS_DELALLOC_RESERVE	0x0004
541	/* caller is from the direct IO path, request to creation of an
542	unwritten extents if not allocated, split the unwritten
543	extent if blocks has been preallocated already*/
544#define EXT4_GET_BLOCKS_PRE_IO			0x0008
545#define EXT4_GET_BLOCKS_CONVERT			0x0010
546#define EXT4_GET_BLOCKS_IO_CREATE_EXT		(EXT4_GET_BLOCKS_PRE_IO|\
547					 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
548	/* Convert extent to initialized after IO complete */
549#define EXT4_GET_BLOCKS_IO_CONVERT_EXT		(EXT4_GET_BLOCKS_CONVERT|\
550					 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
551	/* Eventual metadata allocation (due to growing extent tree)
552	 * should not fail, so try to use reserved blocks for that.*/
553#define EXT4_GET_BLOCKS_METADATA_NOFAIL		0x0020
554	/* Don't normalize allocation size (used for fallocate) */
555#define EXT4_GET_BLOCKS_NO_NORMALIZE		0x0040
556	/* Request will not result in inode size update (user for fallocate) */
557#define EXT4_GET_BLOCKS_KEEP_SIZE		0x0080
558	/* Do not take i_data_sem locking in ext4_map_blocks */
559#define EXT4_GET_BLOCKS_NO_LOCK			0x0100
560	/* Convert written extents to unwritten */
561#define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN	0x0200
562
563/*
564 * The bit position of these flags must not overlap with any of the
565 * EXT4_GET_BLOCKS_*.  They are used by ext4_find_extent(),
566 * read_extent_tree_block(), ext4_split_extent_at(),
567 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
568 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
569 * caching the extents when reading from the extent tree while a
570 * truncate or punch hole operation is in progress.
571 */
572#define EXT4_EX_NOCACHE				0x40000000
573#define EXT4_EX_FORCE_CACHE			0x20000000
574
575/*
576 * Flags used by ext4_free_blocks
577 */
578#define EXT4_FREE_BLOCKS_METADATA	0x0001
579#define EXT4_FREE_BLOCKS_FORGET		0x0002
580#define EXT4_FREE_BLOCKS_VALIDATED	0x0004
581#define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE	0x0008
582#define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER	0x0010
583#define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER	0x0020
584
585/* Encryption algorithms */
586#define EXT4_ENCRYPTION_MODE_INVALID		0
587#define EXT4_ENCRYPTION_MODE_AES_256_XTS	1
588#define EXT4_ENCRYPTION_MODE_AES_256_GCM	2
589#define EXT4_ENCRYPTION_MODE_AES_256_CBC	3
590#define EXT4_ENCRYPTION_MODE_AES_256_CTS	4
591
592#include "ext4_crypto.h"
593
594/*
595 * ioctl commands
596 */
597#define	EXT4_IOC_GETFLAGS		FS_IOC_GETFLAGS
598#define	EXT4_IOC_SETFLAGS		FS_IOC_SETFLAGS
599#define	EXT4_IOC_GETVERSION		_IOR('f', 3, long)
600#define	EXT4_IOC_SETVERSION		_IOW('f', 4, long)
601#define	EXT4_IOC_GETVERSION_OLD		FS_IOC_GETVERSION
602#define	EXT4_IOC_SETVERSION_OLD		FS_IOC_SETVERSION
603#define EXT4_IOC_GETRSVSZ		_IOR('f', 5, long)
604#define EXT4_IOC_SETRSVSZ		_IOW('f', 6, long)
605#define EXT4_IOC_GROUP_EXTEND		_IOW('f', 7, unsigned long)
606#define EXT4_IOC_GROUP_ADD		_IOW('f', 8, struct ext4_new_group_input)
607#define EXT4_IOC_MIGRATE		_IO('f', 9)
608 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
609 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
610#define EXT4_IOC_ALLOC_DA_BLKS		_IO('f', 12)
611#define EXT4_IOC_MOVE_EXT		_IOWR('f', 15, struct move_extent)
612#define EXT4_IOC_RESIZE_FS		_IOW('f', 16, __u64)
613#define EXT4_IOC_SWAP_BOOT		_IO('f', 17)
614#define EXT4_IOC_PRECACHE_EXTENTS	_IO('f', 18)
615#define EXT4_IOC_SET_ENCRYPTION_POLICY	_IOR('f', 19, struct ext4_encryption_policy)
616#define EXT4_IOC_GET_ENCRYPTION_PWSALT	_IOW('f', 20, __u8[16])
617#define EXT4_IOC_GET_ENCRYPTION_POLICY	_IOW('f', 21, struct ext4_encryption_policy)
618
619#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
620/*
621 * ioctl commands in 32 bit emulation
622 */
623#define EXT4_IOC32_GETFLAGS		FS_IOC32_GETFLAGS
624#define EXT4_IOC32_SETFLAGS		FS_IOC32_SETFLAGS
625#define EXT4_IOC32_GETVERSION		_IOR('f', 3, int)
626#define EXT4_IOC32_SETVERSION		_IOW('f', 4, int)
627#define EXT4_IOC32_GETRSVSZ		_IOR('f', 5, int)
628#define EXT4_IOC32_SETRSVSZ		_IOW('f', 6, int)
629#define EXT4_IOC32_GROUP_EXTEND		_IOW('f', 7, unsigned int)
630#define EXT4_IOC32_GROUP_ADD		_IOW('f', 8, struct compat_ext4_new_group_input)
631#define EXT4_IOC32_GETVERSION_OLD	FS_IOC32_GETVERSION
632#define EXT4_IOC32_SETVERSION_OLD	FS_IOC32_SETVERSION
633#endif
634
635/* Max physical block we can address w/o extents */
636#define EXT4_MAX_BLOCK_FILE_PHYS	0xFFFFFFFF
637
638/*
639 * Structure of an inode on the disk
640 */
641struct ext4_inode {
642	__le16	i_mode;		/* File mode */
643	__le16	i_uid;		/* Low 16 bits of Owner Uid */
644	__le32	i_size_lo;	/* Size in bytes */
645	__le32	i_atime;	/* Access time */
646	__le32	i_ctime;	/* Inode Change time */
647	__le32	i_mtime;	/* Modification time */
648	__le32	i_dtime;	/* Deletion Time */
649	__le16	i_gid;		/* Low 16 bits of Group Id */
650	__le16	i_links_count;	/* Links count */
651	__le32	i_blocks_lo;	/* Blocks count */
652	__le32	i_flags;	/* File flags */
653	union {
654		struct {
655			__le32  l_i_version;
656		} linux1;
657		struct {
658			__u32  h_i_translator;
659		} hurd1;
660		struct {
661			__u32  m_i_reserved1;
662		} masix1;
663	} osd1;				/* OS dependent 1 */
664	__le32	i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
665	__le32	i_generation;	/* File version (for NFS) */
666	__le32	i_file_acl_lo;	/* File ACL */
667	__le32	i_size_high;
668	__le32	i_obso_faddr;	/* Obsoleted fragment address */
669	union {
670		struct {
671			__le16	l_i_blocks_high; /* were l_i_reserved1 */
672			__le16	l_i_file_acl_high;
673			__le16	l_i_uid_high;	/* these 2 fields */
674			__le16	l_i_gid_high;	/* were reserved2[0] */
675			__le16	l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
676			__le16	l_i_reserved;
677		} linux2;
678		struct {
679			__le16	h_i_reserved1;	/* Obsoleted fragment number/size which are removed in ext4 */
680			__u16	h_i_mode_high;
681			__u16	h_i_uid_high;
682			__u16	h_i_gid_high;
683			__u32	h_i_author;
684		} hurd2;
685		struct {
686			__le16	h_i_reserved1;	/* Obsoleted fragment number/size which are removed in ext4 */
687			__le16	m_i_file_acl_high;
688			__u32	m_i_reserved2[2];
689		} masix2;
690	} osd2;				/* OS dependent 2 */
691	__le16	i_extra_isize;
692	__le16	i_checksum_hi;	/* crc32c(uuid+inum+inode) BE */
693	__le32  i_ctime_extra;  /* extra Change time      (nsec << 2 | epoch) */
694	__le32  i_mtime_extra;  /* extra Modification time(nsec << 2 | epoch) */
695	__le32  i_atime_extra;  /* extra Access time      (nsec << 2 | epoch) */
696	__le32  i_crtime;       /* File Creation time */
697	__le32  i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
698	__le32  i_version_hi;	/* high 32 bits for 64-bit version */
699};
700
701struct move_extent {
702	__u32 reserved;		/* should be zero */
703	__u32 donor_fd;		/* donor file descriptor */
704	__u64 orig_start;	/* logical start offset in block for orig */
705	__u64 donor_start;	/* logical start offset in block for donor */
706	__u64 len;		/* block length to be moved */
707	__u64 moved_len;	/* moved block length */
708};
709
710#define EXT4_EPOCH_BITS 2
711#define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
712#define EXT4_NSEC_MASK  (~0UL << EXT4_EPOCH_BITS)
713
714/*
715 * Extended fields will fit into an inode if the filesystem was formatted
716 * with large inodes (-I 256 or larger) and there are not currently any EAs
717 * consuming all of the available space. For new inodes we always reserve
718 * enough space for the kernel's known extended fields, but for inodes
719 * created with an old kernel this might not have been the case. None of
720 * the extended inode fields is critical for correct filesystem operation.
721 * This macro checks if a certain field fits in the inode. Note that
722 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
723 */
724#define EXT4_FITS_IN_INODE(ext4_inode, einode, field)	\
725	((offsetof(typeof(*ext4_inode), field) +	\
726	  sizeof((ext4_inode)->field))			\
727	<= (EXT4_GOOD_OLD_INODE_SIZE +			\
728	    (einode)->i_extra_isize))			\
729
730static inline __le32 ext4_encode_extra_time(struct timespec *time)
731{
732       return cpu_to_le32((sizeof(time->tv_sec) > 4 ?
733			   (time->tv_sec >> 32) & EXT4_EPOCH_MASK : 0) |
734                          ((time->tv_nsec << EXT4_EPOCH_BITS) & EXT4_NSEC_MASK));
735}
736
737static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
738{
739       if (sizeof(time->tv_sec) > 4)
740	       time->tv_sec |= (__u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK)
741			       << 32;
742       time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
743}
744
745#define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode)			       \
746do {									       \
747	(raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec);	       \
748	if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     \
749		(raw_inode)->xtime ## _extra =				       \
750				ext4_encode_extra_time(&(inode)->xtime);       \
751} while (0)
752
753#define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode)			       \
754do {									       \
755	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))		       \
756		(raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec);      \
757	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))	       \
758		(raw_inode)->xtime ## _extra =				       \
759				ext4_encode_extra_time(&(einode)->xtime);      \
760} while (0)
761
762#define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode)			       \
763do {									       \
764	(inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime);       \
765	if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     \
766		ext4_decode_extra_time(&(inode)->xtime,			       \
767				       raw_inode->xtime ## _extra);	       \
768	else								       \
769		(inode)->xtime.tv_nsec = 0;				       \
770} while (0)
771
772#define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode)			       \
773do {									       \
774	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))		       \
775		(einode)->xtime.tv_sec = 				       \
776			(signed)le32_to_cpu((raw_inode)->xtime);	       \
777	else								       \
778		(einode)->xtime.tv_sec = 0;				       \
779	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))	       \
780		ext4_decode_extra_time(&(einode)->xtime,		       \
781				       raw_inode->xtime ## _extra);	       \
782	else								       \
783		(einode)->xtime.tv_nsec = 0;				       \
784} while (0)
785
786#define i_disk_version osd1.linux1.l_i_version
787
788#if defined(__KERNEL__) || defined(__linux__)
789#define i_reserved1	osd1.linux1.l_i_reserved1
790#define i_file_acl_high	osd2.linux2.l_i_file_acl_high
791#define i_blocks_high	osd2.linux2.l_i_blocks_high
792#define i_uid_low	i_uid
793#define i_gid_low	i_gid
794#define i_uid_high	osd2.linux2.l_i_uid_high
795#define i_gid_high	osd2.linux2.l_i_gid_high
796#define i_checksum_lo	osd2.linux2.l_i_checksum_lo
797
798#elif defined(__GNU__)
799
800#define i_translator	osd1.hurd1.h_i_translator
801#define i_uid_high	osd2.hurd2.h_i_uid_high
802#define i_gid_high	osd2.hurd2.h_i_gid_high
803#define i_author	osd2.hurd2.h_i_author
804
805#elif defined(__masix__)
806
807#define i_reserved1	osd1.masix1.m_i_reserved1
808#define i_file_acl_high	osd2.masix2.m_i_file_acl_high
809#define i_reserved2	osd2.masix2.m_i_reserved2
810
811#endif /* defined(__KERNEL__) || defined(__linux__) */
812
813#include "extents_status.h"
814
815/*
816 * Lock subclasses for i_data_sem in the ext4_inode_info structure.
817 *
818 * These are needed to avoid lockdep false positives when we need to
819 * allocate blocks to the quota inode during ext4_map_blocks(), while
820 * holding i_data_sem for a normal (non-quota) inode.  Since we don't
821 * do quota tracking for the quota inode, this avoids deadlock (as
822 * well as infinite recursion, since it isn't turtles all the way
823 * down...)
824 *
825 *  I_DATA_SEM_NORMAL - Used for most inodes
826 *  I_DATA_SEM_OTHER  - Used by move_inode.c for the second normal inode
827 *			  where the second inode has larger inode number
828 *			  than the first
829 *  I_DATA_SEM_QUOTA  - Used for quota inodes only
830 */
831enum {
832	I_DATA_SEM_NORMAL = 0,
833	I_DATA_SEM_OTHER,
834	I_DATA_SEM_QUOTA,
835};
836
837
838/*
839 * fourth extended file system inode data in memory
840 */
841struct ext4_inode_info {
842	__le32	i_data[15];	/* unconverted */
843	__u32	i_dtime;
844	ext4_fsblk_t	i_file_acl;
845
846	/*
847	 * i_block_group is the number of the block group which contains
848	 * this file's inode.  Constant across the lifetime of the inode,
849	 * it is ued for making block allocation decisions - we try to
850	 * place a file's data blocks near its inode block, and new inodes
851	 * near to their parent directory's inode.
852	 */
853	ext4_group_t	i_block_group;
854	ext4_lblk_t	i_dir_start_lookup;
855#if (BITS_PER_LONG < 64)
856	unsigned long	i_state_flags;		/* Dynamic state flags */
857#endif
858	unsigned long	i_flags;
859
860	/*
861	 * Extended attributes can be read independently of the main file
862	 * data. Taking i_mutex even when reading would cause contention
863	 * between readers of EAs and writers of regular file data, so
864	 * instead we synchronize on xattr_sem when reading or changing
865	 * EAs.
866	 */
867	struct rw_semaphore xattr_sem;
868
869	struct list_head i_orphan;	/* unlinked but open inodes */
870
871	/*
872	 * i_disksize keeps track of what the inode size is ON DISK, not
873	 * in memory.  During truncate, i_size is set to the new size by
874	 * the VFS prior to calling ext4_truncate(), but the filesystem won't
875	 * set i_disksize to 0 until the truncate is actually under way.
876	 *
877	 * The intent is that i_disksize always represents the blocks which
878	 * are used by this file.  This allows recovery to restart truncate
879	 * on orphans if we crash during truncate.  We actually write i_disksize
880	 * into the on-disk inode when writing inodes out, instead of i_size.
881	 *
882	 * The only time when i_disksize and i_size may be different is when
883	 * a truncate is in progress.  The only things which change i_disksize
884	 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
885	 */
886	loff_t	i_disksize;
887
888	/*
889	 * i_data_sem is for serialising ext4_truncate() against
890	 * ext4_getblock().  In the 2.4 ext2 design, great chunks of inode's
891	 * data tree are chopped off during truncate. We can't do that in
892	 * ext4 because whenever we perform intermediate commits during
893	 * truncate, the inode and all the metadata blocks *must* be in a
894	 * consistent state which allows truncation of the orphans to restart
895	 * during recovery.  Hence we must fix the get_block-vs-truncate race
896	 * by other means, so we have i_data_sem.
897	 */
898	struct rw_semaphore i_data_sem;
899	/*
900	 * i_mmap_sem is for serializing page faults with truncate / punch hole
901	 * operations. We have to make sure that new page cannot be faulted in
902	 * a section of the inode that is being punched. We cannot easily use
903	 * i_data_sem for this since we need protection for the whole punch
904	 * operation and i_data_sem ranks below transaction start so we have
905	 * to occasionally drop it.
906	 */
907	struct rw_semaphore i_mmap_sem;
908	struct inode vfs_inode;
909	struct jbd2_inode *jinode;
910
911	spinlock_t i_raw_lock;	/* protects updates to the raw inode */
912
913	/*
914	 * File creation time. Its function is same as that of
915	 * struct timespec i_{a,c,m}time in the generic inode.
916	 */
917	struct timespec i_crtime;
918
919	/* mballoc */
920	struct list_head i_prealloc_list;
921	spinlock_t i_prealloc_lock;
922
923	/* extents status tree */
924	struct ext4_es_tree i_es_tree;
925	rwlock_t i_es_lock;
926	struct list_head i_es_list;
927	unsigned int i_es_all_nr;	/* protected by i_es_lock */
928	unsigned int i_es_shk_nr;	/* protected by i_es_lock */
929	ext4_lblk_t i_es_shrink_lblk;	/* Offset where we start searching for
930					   extents to shrink. Protected by
931					   i_es_lock  */
932
933	/* ialloc */
934	ext4_group_t	i_last_alloc_group;
935
936	/* allocation reservation info for delalloc */
937	/* In case of bigalloc, these refer to clusters rather than blocks */
938	unsigned int i_reserved_data_blocks;
939	unsigned int i_reserved_meta_blocks;
940	unsigned int i_allocated_meta_blocks;
941	ext4_lblk_t i_da_metadata_calc_last_lblock;
942	int i_da_metadata_calc_len;
943
944	/* on-disk additional length */
945	__u16 i_extra_isize;
946	char i_crypt_policy_flags;
947
948	/* Indicate the inline data space. */
949	u16 i_inline_off;
950	u16 i_inline_size;
951
952#ifdef CONFIG_QUOTA
953	/* quota space reservation, managed internally by quota code */
954	qsize_t i_reserved_quota;
955#endif
956
957	/* Lock protecting lists below */
958	spinlock_t i_completed_io_lock;
959	/*
960	 * Completed IOs that need unwritten extents handling and have
961	 * transaction reserved
962	 */
963	struct list_head i_rsv_conversion_list;
964	/*
965	 * Completed IOs that need unwritten extents handling and don't have
966	 * transaction reserved
967	 */
968	atomic_t i_ioend_count;	/* Number of outstanding io_end structs */
969	atomic_t i_unwritten; /* Nr. of inflight conversions pending */
970	struct work_struct i_rsv_conversion_work;
971
972	spinlock_t i_block_reservation_lock;
973
974	/*
975	 * Transactions that contain inode's metadata needed to complete
976	 * fsync and fdatasync, respectively.
977	 */
978	tid_t i_sync_tid;
979	tid_t i_datasync_tid;
980
981#ifdef CONFIG_QUOTA
982	struct dquot *i_dquot[MAXQUOTAS];
983#endif
984
985	/* Precomputed uuid+inum+igen checksum for seeding inode checksums */
986	__u32 i_csum_seed;
987
988#ifdef CONFIG_EXT4_FS_ENCRYPTION
989	/* Encryption params */
990	struct ext4_encryption_key i_encryption_key;
991#endif
992};
993
994/*
995 * File system states
996 */
997#define	EXT4_VALID_FS			0x0001	/* Unmounted cleanly */
998#define	EXT4_ERROR_FS			0x0002	/* Errors detected */
999#define	EXT4_ORPHAN_FS			0x0004	/* Orphans being recovered */
1000
1001/*
1002 * Misc. filesystem flags
1003 */
1004#define EXT2_FLAGS_SIGNED_HASH		0x0001  /* Signed dirhash in use */
1005#define EXT2_FLAGS_UNSIGNED_HASH	0x0002  /* Unsigned dirhash in use */
1006#define EXT2_FLAGS_TEST_FILESYS		0x0004	/* to test development code */
1007
1008/*
1009 * Mount flags set via mount options or defaults
1010 */
1011#define EXT4_MOUNT_GRPID		0x00004	/* Create files with directory's group */
1012#define EXT4_MOUNT_DEBUG		0x00008	/* Some debugging messages */
1013#define EXT4_MOUNT_ERRORS_CONT		0x00010	/* Continue on errors */
1014#define EXT4_MOUNT_ERRORS_RO		0x00020	/* Remount fs ro on errors */
1015#define EXT4_MOUNT_ERRORS_PANIC		0x00040	/* Panic on errors */
1016#define EXT4_MOUNT_ERRORS_MASK		0x00070
1017#define EXT4_MOUNT_MINIX_DF		0x00080	/* Mimics the Minix statfs */
1018#define EXT4_MOUNT_NOLOAD		0x00100	/* Don't use existing journal*/
1019#ifdef CONFIG_FS_DAX
1020#define EXT4_MOUNT_DAX			0x00200	/* Direct Access */
1021#else
1022#define EXT4_MOUNT_DAX			0
1023#endif
1024#define EXT4_MOUNT_DATA_FLAGS		0x00C00	/* Mode for data writes: */
1025#define EXT4_MOUNT_JOURNAL_DATA		0x00400	/* Write data to journal */
1026#define EXT4_MOUNT_ORDERED_DATA		0x00800	/* Flush data before commit */
1027#define EXT4_MOUNT_WRITEBACK_DATA	0x00C00	/* No data ordering */
1028#define EXT4_MOUNT_UPDATE_JOURNAL	0x01000	/* Update the journal format */
1029#define EXT4_MOUNT_NO_UID32		0x02000  /* Disable 32-bit UIDs */
1030#define EXT4_MOUNT_XATTR_USER		0x04000	/* Extended user attributes */
1031#define EXT4_MOUNT_POSIX_ACL		0x08000	/* POSIX Access Control Lists */
1032#define EXT4_MOUNT_NO_AUTO_DA_ALLOC	0x10000	/* No auto delalloc mapping */
1033#define EXT4_MOUNT_BARRIER		0x20000 /* Use block barriers */
1034#define EXT4_MOUNT_QUOTA		0x80000 /* Some quota option set */
1035#define EXT4_MOUNT_USRQUOTA		0x100000 /* "old" user quota */
1036#define EXT4_MOUNT_GRPQUOTA		0x200000 /* "old" group quota */
1037#define EXT4_MOUNT_DIOREAD_NOLOCK	0x400000 /* Enable support for dio read nolocking */
1038#define EXT4_MOUNT_JOURNAL_CHECKSUM	0x800000 /* Journal checksums */
1039#define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT	0x1000000 /* Journal Async Commit */
1040#define EXT4_MOUNT_DELALLOC		0x8000000 /* Delalloc support */
1041#define EXT4_MOUNT_DATA_ERR_ABORT	0x10000000 /* Abort on file data write */
1042#define EXT4_MOUNT_BLOCK_VALIDITY	0x20000000 /* Block validity checking */
1043#define EXT4_MOUNT_DISCARD		0x40000000 /* Issue DISCARD requests */
1044#define EXT4_MOUNT_INIT_INODE_TABLE	0x80000000 /* Initialize uninitialized itables */
1045
1046/*
1047 * Mount flags set either automatically (could not be set by mount option)
1048 * based on per file system feature or property or in special cases such as
1049 * distinguishing between explicit mount option definition and default.
1050 */
1051#define EXT4_MOUNT2_EXPLICIT_DELALLOC	0x00000001 /* User explicitly
1052						      specified delalloc */
1053#define EXT4_MOUNT2_STD_GROUP_SIZE	0x00000002 /* We have standard group
1054						      size of blocksize * 8
1055						      blocks */
1056#define EXT4_MOUNT2_HURD_COMPAT		0x00000004 /* Support HURD-castrated
1057						      file systems */
1058
1059#define clear_opt(sb, opt)		EXT4_SB(sb)->s_mount_opt &= \
1060						~EXT4_MOUNT_##opt
1061#define set_opt(sb, opt)		EXT4_SB(sb)->s_mount_opt |= \
1062						EXT4_MOUNT_##opt
1063#define test_opt(sb, opt)		(EXT4_SB(sb)->s_mount_opt & \
1064					 EXT4_MOUNT_##opt)
1065
1066#define clear_opt2(sb, opt)		EXT4_SB(sb)->s_mount_opt2 &= \
1067						~EXT4_MOUNT2_##opt
1068#define set_opt2(sb, opt)		EXT4_SB(sb)->s_mount_opt2 |= \
1069						EXT4_MOUNT2_##opt
1070#define test_opt2(sb, opt)		(EXT4_SB(sb)->s_mount_opt2 & \
1071					 EXT4_MOUNT2_##opt)
1072
1073#define ext4_test_and_set_bit		__test_and_set_bit_le
1074#define ext4_set_bit			__set_bit_le
1075#define ext4_set_bit_atomic		ext2_set_bit_atomic
1076#define ext4_test_and_clear_bit		__test_and_clear_bit_le
1077#define ext4_clear_bit			__clear_bit_le
1078#define ext4_clear_bit_atomic		ext2_clear_bit_atomic
1079#define ext4_test_bit			test_bit_le
1080#define ext4_find_next_zero_bit		find_next_zero_bit_le
1081#define ext4_find_next_bit		find_next_bit_le
1082
1083extern void ext4_set_bits(void *bm, int cur, int len);
1084
1085/*
1086 * Maximal mount counts between two filesystem checks
1087 */
1088#define EXT4_DFL_MAX_MNT_COUNT		20	/* Allow 20 mounts */
1089#define EXT4_DFL_CHECKINTERVAL		0	/* Don't use interval check */
1090
1091/*
1092 * Behaviour when detecting errors
1093 */
1094#define EXT4_ERRORS_CONTINUE		1	/* Continue execution */
1095#define EXT4_ERRORS_RO			2	/* Remount fs read-only */
1096#define EXT4_ERRORS_PANIC		3	/* Panic */
1097#define EXT4_ERRORS_DEFAULT		EXT4_ERRORS_CONTINUE
1098
1099/* Metadata checksum algorithm codes */
1100#define EXT4_CRC32C_CHKSUM		1
1101
1102/*
1103 * Structure of the super block
1104 */
1105struct ext4_super_block {
1106/*00*/	__le32	s_inodes_count;		/* Inodes count */
1107	__le32	s_blocks_count_lo;	/* Blocks count */
1108	__le32	s_r_blocks_count_lo;	/* Reserved blocks count */
1109	__le32	s_free_blocks_count_lo;	/* Free blocks count */
1110/*10*/	__le32	s_free_inodes_count;	/* Free inodes count */
1111	__le32	s_first_data_block;	/* First Data Block */
1112	__le32	s_log_block_size;	/* Block size */
1113	__le32	s_log_cluster_size;	/* Allocation cluster size */
1114/*20*/	__le32	s_blocks_per_group;	/* # Blocks per group */
1115	__le32	s_clusters_per_group;	/* # Clusters per group */
1116	__le32	s_inodes_per_group;	/* # Inodes per group */
1117	__le32	s_mtime;		/* Mount time */
1118/*30*/	__le32	s_wtime;		/* Write time */
1119	__le16	s_mnt_count;		/* Mount count */
1120	__le16	s_max_mnt_count;	/* Maximal mount count */
1121	__le16	s_magic;		/* Magic signature */
1122	__le16	s_state;		/* File system state */
1123	__le16	s_errors;		/* Behaviour when detecting errors */
1124	__le16	s_minor_rev_level;	/* minor revision level */
1125/*40*/	__le32	s_lastcheck;		/* time of last check */
1126	__le32	s_checkinterval;	/* max. time between checks */
1127	__le32	s_creator_os;		/* OS */
1128	__le32	s_rev_level;		/* Revision level */
1129/*50*/	__le16	s_def_resuid;		/* Default uid for reserved blocks */
1130	__le16	s_def_resgid;		/* Default gid for reserved blocks */
1131	/*
1132	 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1133	 *
1134	 * Note: the difference between the compatible feature set and
1135	 * the incompatible feature set is that if there is a bit set
1136	 * in the incompatible feature set that the kernel doesn't
1137	 * know about, it should refuse to mount the filesystem.
1138	 *
1139	 * e2fsck's requirements are more strict; if it doesn't know
1140	 * about a feature in either the compatible or incompatible
1141	 * feature set, it must abort and not try to meddle with
1142	 * things it doesn't understand...
1143	 */
1144	__le32	s_first_ino;		/* First non-reserved inode */
1145	__le16  s_inode_size;		/* size of inode structure */
1146	__le16	s_block_group_nr;	/* block group # of this superblock */
1147	__le32	s_feature_compat;	/* compatible feature set */
1148/*60*/	__le32	s_feature_incompat;	/* incompatible feature set */
1149	__le32	s_feature_ro_compat;	/* readonly-compatible feature set */
1150/*68*/	__u8	s_uuid[16];		/* 128-bit uuid for volume */
1151/*78*/	char	s_volume_name[16];	/* volume name */
1152/*88*/	char	s_last_mounted[64];	/* directory where last mounted */
1153/*C8*/	__le32	s_algorithm_usage_bitmap; /* For compression */
1154	/*
1155	 * Performance hints.  Directory preallocation should only
1156	 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1157	 */
1158	__u8	s_prealloc_blocks;	/* Nr of blocks to try to preallocate*/
1159	__u8	s_prealloc_dir_blocks;	/* Nr to preallocate for dirs */
1160	__le16	s_reserved_gdt_blocks;	/* Per group desc for online growth */
1161	/*
1162	 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1163	 */
1164/*D0*/	__u8	s_journal_uuid[16];	/* uuid of journal superblock */
1165/*E0*/	__le32	s_journal_inum;		/* inode number of journal file */
1166	__le32	s_journal_dev;		/* device number of journal file */
1167	__le32	s_last_orphan;		/* start of list of inodes to delete */
1168	__le32	s_hash_seed[4];		/* HTREE hash seed */
1169	__u8	s_def_hash_version;	/* Default hash version to use */
1170	__u8	s_jnl_backup_type;
1171	__le16  s_desc_size;		/* size of group descriptor */
1172/*100*/	__le32	s_default_mount_opts;
1173	__le32	s_first_meta_bg;	/* First metablock block group */
1174	__le32	s_mkfs_time;		/* When the filesystem was created */
1175	__le32	s_jnl_blocks[17];	/* Backup of the journal inode */
1176	/* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1177/*150*/	__le32	s_blocks_count_hi;	/* Blocks count */
1178	__le32	s_r_blocks_count_hi;	/* Reserved blocks count */
1179	__le32	s_free_blocks_count_hi;	/* Free blocks count */
1180	__le16	s_min_extra_isize;	/* All inodes have at least # bytes */
1181	__le16	s_want_extra_isize; 	/* New inodes should reserve # bytes */
1182	__le32	s_flags;		/* Miscellaneous flags */
1183	__le16  s_raid_stride;		/* RAID stride */
1184	__le16  s_mmp_update_interval;  /* # seconds to wait in MMP checking */
1185	__le64  s_mmp_block;            /* Block for multi-mount protection */
1186	__le32  s_raid_stripe_width;    /* blocks on all data disks (N*stride)*/
1187	__u8	s_log_groups_per_flex;  /* FLEX_BG group size */
1188	__u8	s_checksum_type;	/* metadata checksum algorithm used */
1189	__u8	s_encryption_level;	/* versioning level for encryption */
1190	__u8	s_reserved_pad;		/* Padding to next 32bits */
1191	__le64	s_kbytes_written;	/* nr of lifetime kilobytes written */
1192	__le32	s_snapshot_inum;	/* Inode number of active snapshot */
1193	__le32	s_snapshot_id;		/* sequential ID of active snapshot */
1194	__le64	s_snapshot_r_blocks_count; /* reserved blocks for active
1195					      snapshot's future use */
1196	__le32	s_snapshot_list;	/* inode number of the head of the
1197					   on-disk snapshot list */
1198#define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1199	__le32	s_error_count;		/* number of fs errors */
1200	__le32	s_first_error_time;	/* first time an error happened */
1201	__le32	s_first_error_ino;	/* inode involved in first error */
1202	__le64	s_first_error_block;	/* block involved of first error */
1203	__u8	s_first_error_func[32];	/* function where the error happened */
1204	__le32	s_first_error_line;	/* line number where error happened */
1205	__le32	s_last_error_time;	/* most recent time of an error */
1206	__le32	s_last_error_ino;	/* inode involved in last error */
1207	__le32	s_last_error_line;	/* line number where error happened */
1208	__le64	s_last_error_block;	/* block involved of last error */
1209	__u8	s_last_error_func[32];	/* function where the error happened */
1210#define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1211	__u8	s_mount_opts[64];
1212	__le32	s_usr_quota_inum;	/* inode for tracking user quota */
1213	__le32	s_grp_quota_inum;	/* inode for tracking group quota */
1214	__le32	s_overhead_clusters;	/* overhead blocks/clusters in fs */
1215	__le32	s_backup_bgs[2];	/* groups with sparse_super2 SBs */
1216	__u8	s_encrypt_algos[4];	/* Encryption algorithms in use  */
1217	__u8	s_encrypt_pw_salt[16];	/* Salt used for string2key algorithm */
1218	__le32	s_lpf_ino;		/* Location of the lost+found inode */
1219	__le32	s_reserved[100];	/* Padding to the end of the block */
1220	__le32	s_checksum;		/* crc32c(superblock) */
1221};
1222
1223#define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1224
1225#ifdef __KERNEL__
1226
1227/*
1228 * run-time mount flags
1229 */
1230#define EXT4_MF_MNTDIR_SAMPLED		0x0001
1231#define EXT4_MF_FS_ABORTED		0x0002	/* Fatal error detected */
1232#define EXT4_MF_TEST_DUMMY_ENCRYPTION	0x0004
1233
1234#ifdef CONFIG_EXT4_FS_ENCRYPTION
1235#define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \
1236						EXT4_MF_TEST_DUMMY_ENCRYPTION))
1237#else
1238#define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1239#endif
1240
1241/* Number of quota types we support */
1242#define EXT4_MAXQUOTAS 2
1243
1244/*
1245 * fourth extended-fs super-block data in memory
1246 */
1247struct ext4_sb_info {
1248	unsigned long s_desc_size;	/* Size of a group descriptor in bytes */
1249	unsigned long s_inodes_per_block;/* Number of inodes per block */
1250	unsigned long s_blocks_per_group;/* Number of blocks in a group */
1251	unsigned long s_clusters_per_group; /* Number of clusters in a group */
1252	unsigned long s_inodes_per_group;/* Number of inodes in a group */
1253	unsigned long s_itb_per_group;	/* Number of inode table blocks per group */
1254	unsigned long s_gdb_count;	/* Number of group descriptor blocks */
1255	unsigned long s_desc_per_block;	/* Number of group descriptors per block */
1256	ext4_group_t s_groups_count;	/* Number of groups in the fs */
1257	ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1258	unsigned long s_overhead;  /* # of fs overhead clusters */
1259	unsigned int s_cluster_ratio;	/* Number of blocks per cluster */
1260	unsigned int s_cluster_bits;	/* log2 of s_cluster_ratio */
1261	loff_t s_bitmap_maxbytes;	/* max bytes for bitmap files */
1262	struct buffer_head * s_sbh;	/* Buffer containing the super block */
1263	struct ext4_super_block *s_es;	/* Pointer to the super block in the buffer */
1264	struct buffer_head **s_group_desc;
1265	unsigned int s_mount_opt;
1266	unsigned int s_mount_opt2;
1267	unsigned int s_mount_flags;
1268	unsigned int s_def_mount_opt;
1269	ext4_fsblk_t s_sb_block;
1270	atomic64_t s_resv_clusters;
1271	kuid_t s_resuid;
1272	kgid_t s_resgid;
1273	unsigned short s_mount_state;
1274	unsigned short s_pad;
1275	int s_addr_per_block_bits;
1276	int s_desc_per_block_bits;
1277	int s_inode_size;
1278	int s_first_ino;
1279	unsigned int s_inode_readahead_blks;
1280	unsigned int s_inode_goal;
1281	spinlock_t s_next_gen_lock;
1282	u32 s_next_generation;
1283	u32 s_hash_seed[4];
1284	int s_def_hash_version;
1285	int s_hash_unsigned;	/* 3 if hash should be signed, 0 if not */
1286	struct percpu_counter s_freeclusters_counter;
1287	struct percpu_counter s_freeinodes_counter;
1288	struct percpu_counter s_dirs_counter;
1289	struct percpu_counter s_dirtyclusters_counter;
1290	struct blockgroup_lock *s_blockgroup_lock;
1291	struct proc_dir_entry *s_proc;
1292	struct kobject s_kobj;
1293	struct completion s_kobj_unregister;
1294	struct super_block *s_sb;
1295
1296	/* Journaling */
1297	struct journal_s *s_journal;
1298	struct list_head s_orphan;
1299	struct mutex s_orphan_lock;
1300	unsigned long s_resize_flags;		/* Flags indicating if there
1301						   is a resizer */
1302	unsigned long s_commit_interval;
1303	u32 s_max_batch_time;
1304	u32 s_min_batch_time;
1305	struct block_device *journal_bdev;
1306#ifdef CONFIG_QUOTA
1307	char *s_qf_names[EXT4_MAXQUOTAS];	/* Names of quota files with journalled quota */
1308	int s_jquota_fmt;			/* Format of quota to use */
1309#endif
1310	unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1311	struct rb_root system_blks;
1312
1313#ifdef EXTENTS_STATS
1314	/* ext4 extents stats */
1315	unsigned long s_ext_min;
1316	unsigned long s_ext_max;
1317	unsigned long s_depth_max;
1318	spinlock_t s_ext_stats_lock;
1319	unsigned long s_ext_blocks;
1320	unsigned long s_ext_extents;
1321#endif
1322
1323	/* for buddy allocator */
1324	struct ext4_group_info ***s_group_info;
1325	struct inode *s_buddy_cache;
1326	spinlock_t s_md_lock;
1327	unsigned short *s_mb_offsets;
1328	unsigned int *s_mb_maxs;
1329	unsigned int s_group_info_size;
1330
1331	/* tunables */
1332	unsigned long s_stripe;
1333	unsigned int s_mb_stream_request;
1334	unsigned int s_mb_max_to_scan;
1335	unsigned int s_mb_min_to_scan;
1336	unsigned int s_mb_stats;
1337	unsigned int s_mb_order2_reqs;
1338	unsigned int s_mb_group_prealloc;
1339	unsigned int s_max_dir_size_kb;
1340	/* where last allocation was done - for stream allocation */
1341	unsigned long s_mb_last_group;
1342	unsigned long s_mb_last_start;
1343
1344	/* stats for buddy allocator */
1345	atomic_t s_bal_reqs;	/* number of reqs with len > 1 */
1346	atomic_t s_bal_success;	/* we found long enough chunks */
1347	atomic_t s_bal_allocated;	/* in blocks */
1348	atomic_t s_bal_ex_scanned;	/* total extents scanned */
1349	atomic_t s_bal_goals;	/* goal hits */
1350	atomic_t s_bal_breaks;	/* too long searches */
1351	atomic_t s_bal_2orders;	/* 2^order hits */
1352	spinlock_t s_bal_lock;
1353	unsigned long s_mb_buddies_generated;
1354	unsigned long long s_mb_generation_time;
1355	atomic_t s_mb_lost_chunks;
1356	atomic_t s_mb_preallocated;
1357	atomic_t s_mb_discarded;
1358	atomic_t s_lock_busy;
1359
1360	/* locality groups */
1361	struct ext4_locality_group __percpu *s_locality_groups;
1362
1363	/* for write statistics */
1364	unsigned long s_sectors_written_start;
1365	u64 s_kbytes_written;
1366
1367	/* the size of zero-out chunk */
1368	unsigned int s_extent_max_zeroout_kb;
1369
1370	unsigned int s_log_groups_per_flex;
1371	struct flex_groups *s_flex_groups;
1372	ext4_group_t s_flex_groups_allocated;
1373
1374	/* workqueue for reserved extent conversions (buffered io) */
1375	struct workqueue_struct *rsv_conversion_wq;
1376
1377	/* timer for periodic error stats printing */
1378	struct timer_list s_err_report;
1379
1380	/* Lazy inode table initialization info */
1381	struct ext4_li_request *s_li_request;
1382	/* Wait multiplier for lazy initialization thread */
1383	unsigned int s_li_wait_mult;
1384
1385	/* Kernel thread for multiple mount protection */
1386	struct task_struct *s_mmp_tsk;
1387
1388	/* record the last minlen when FITRIM is called. */
1389	atomic_t s_last_trim_minblks;
1390
1391	/* Reference to checksum algorithm driver via cryptoapi */
1392	struct crypto_shash *s_chksum_driver;
1393
1394	/* Precomputed FS UUID checksum for seeding other checksums */
1395	__u32 s_csum_seed;
1396
1397	/* Reclaim extents from extent status tree */
1398	struct shrinker s_es_shrinker;
1399	struct list_head s_es_list;	/* List of inodes with reclaimable extents */
1400	long s_es_nr_inode;
1401	struct ext4_es_stats s_es_stats;
1402	struct mb_cache *s_mb_cache;
1403	spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1404
1405	/* Ratelimit ext4 messages. */
1406	struct ratelimit_state s_err_ratelimit_state;
1407	struct ratelimit_state s_warning_ratelimit_state;
1408	struct ratelimit_state s_msg_ratelimit_state;
1409
1410#ifdef CONFIG_EXT4_FS_ENCRYPTION
1411	/* Encryption */
1412	uint32_t s_file_encryption_mode;
1413	uint32_t s_dir_encryption_mode;
1414#endif
1415};
1416
1417static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1418{
1419	return sb->s_fs_info;
1420}
1421static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1422{
1423	return container_of(inode, struct ext4_inode_info, vfs_inode);
1424}
1425
1426static inline struct timespec ext4_current_time(struct inode *inode)
1427{
1428	return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1429		current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1430}
1431
1432static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1433{
1434	return ino == EXT4_ROOT_INO ||
1435		ino == EXT4_USR_QUOTA_INO ||
1436		ino == EXT4_GRP_QUOTA_INO ||
1437		ino == EXT4_BOOT_LOADER_INO ||
1438		ino == EXT4_JOURNAL_INO ||
1439		ino == EXT4_RESIZE_INO ||
1440		(ino >= EXT4_FIRST_INO(sb) &&
1441		 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1442}
1443
1444static inline void ext4_set_io_unwritten_flag(struct inode *inode,
1445					      struct ext4_io_end *io_end)
1446{
1447	if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
1448		io_end->flag |= EXT4_IO_END_UNWRITTEN;
1449		atomic_inc(&EXT4_I(inode)->i_unwritten);
1450	}
1451}
1452
1453static inline ext4_io_end_t *ext4_inode_aio(struct inode *inode)
1454{
1455	return inode->i_private;
1456}
1457
1458static inline void ext4_inode_aio_set(struct inode *inode, ext4_io_end_t *io)
1459{
1460	inode->i_private = io;
1461}
1462
1463/*
1464 * Inode dynamic state flags
1465 */
1466enum {
1467	EXT4_STATE_JDATA,		/* journaled data exists */
1468	EXT4_STATE_NEW,			/* inode is newly created */
1469	EXT4_STATE_XATTR,		/* has in-inode xattrs */
1470	EXT4_STATE_NO_EXPAND,		/* No space for expansion */
1471	EXT4_STATE_DA_ALLOC_CLOSE,	/* Alloc DA blks on close */
1472	EXT4_STATE_EXT_MIGRATE,		/* Inode is migrating */
1473	EXT4_STATE_DIO_UNWRITTEN,	/* need convert on dio done*/
1474	EXT4_STATE_NEWENTRY,		/* File just added to dir */
1475	EXT4_STATE_DIOREAD_LOCK,	/* Disable support for dio read
1476					   nolocking */
1477	EXT4_STATE_MAY_INLINE_DATA,	/* may have in-inode data */
1478	EXT4_STATE_ORDERED_MODE,	/* data=ordered mode */
1479	EXT4_STATE_EXT_PRECACHED,	/* extents have been precached */
1480};
1481
1482#define EXT4_INODE_BIT_FNS(name, field, offset)				\
1483static inline int ext4_test_inode_##name(struct inode *inode, int bit)	\
1484{									\
1485	return test_bit(bit + (offset), &EXT4_I(inode)->i_##field);	\
1486}									\
1487static inline void ext4_set_inode_##name(struct inode *inode, int bit)	\
1488{									\
1489	set_bit(bit + (offset), &EXT4_I(inode)->i_##field);		\
1490}									\
1491static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1492{									\
1493	clear_bit(bit + (offset), &EXT4_I(inode)->i_##field);		\
1494}
1495
1496/* Add these declarations here only so that these functions can be
1497 * found by name.  Otherwise, they are very hard to locate. */
1498static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1499static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1500static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1501EXT4_INODE_BIT_FNS(flag, flags, 0)
1502
1503/* Add these declarations here only so that these functions can be
1504 * found by name.  Otherwise, they are very hard to locate. */
1505static inline int ext4_test_inode_state(struct inode *inode, int bit);
1506static inline void ext4_set_inode_state(struct inode *inode, int bit);
1507static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1508#if (BITS_PER_LONG < 64)
1509EXT4_INODE_BIT_FNS(state, state_flags, 0)
1510
1511static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1512{
1513	(ei)->i_state_flags = 0;
1514}
1515#else
1516EXT4_INODE_BIT_FNS(state, flags, 32)
1517
1518static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1519{
1520	/* We depend on the fact that callers will set i_flags */
1521}
1522#endif
1523#else
1524/* Assume that user mode programs are passing in an ext4fs superblock, not
1525 * a kernel struct super_block.  This will allow us to call the feature-test
1526 * macros from user land. */
1527#define EXT4_SB(sb)	(sb)
1528#endif
1529
1530/*
1531 * Returns true if the inode is inode is encrypted
1532 */
1533static inline int ext4_encrypted_inode(struct inode *inode)
1534{
1535#ifdef CONFIG_EXT4_FS_ENCRYPTION
1536	return ext4_test_inode_flag(inode, EXT4_INODE_ENCRYPT);
1537#else
1538	return 0;
1539#endif
1540}
1541
1542#define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1543
1544/*
1545 * Codes for operating systems
1546 */
1547#define EXT4_OS_LINUX		0
1548#define EXT4_OS_HURD		1
1549#define EXT4_OS_MASIX		2
1550#define EXT4_OS_FREEBSD		3
1551#define EXT4_OS_LITES		4
1552
1553/*
1554 * Revision levels
1555 */
1556#define EXT4_GOOD_OLD_REV	0	/* The good old (original) format */
1557#define EXT4_DYNAMIC_REV	1	/* V2 format w/ dynamic inode sizes */
1558
1559#define EXT4_CURRENT_REV	EXT4_GOOD_OLD_REV
1560#define EXT4_MAX_SUPP_REV	EXT4_DYNAMIC_REV
1561
1562#define EXT4_GOOD_OLD_INODE_SIZE 128
1563
1564/*
1565 * Feature set definitions
1566 */
1567
1568#define EXT4_HAS_COMPAT_FEATURE(sb,mask)			\
1569	((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1570#define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask)			\
1571	((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1572#define EXT4_HAS_INCOMPAT_FEATURE(sb,mask)			\
1573	((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1574#define EXT4_SET_COMPAT_FEATURE(sb,mask)			\
1575	EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1576#define EXT4_SET_RO_COMPAT_FEATURE(sb,mask)			\
1577	EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1578#define EXT4_SET_INCOMPAT_FEATURE(sb,mask)			\
1579	EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1580#define EXT4_CLEAR_COMPAT_FEATURE(sb,mask)			\
1581	EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1582#define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask)			\
1583	EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1584#define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask)			\
1585	EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1586
1587#define EXT4_FEATURE_COMPAT_DIR_PREALLOC	0x0001
1588#define EXT4_FEATURE_COMPAT_IMAGIC_INODES	0x0002
1589#define EXT4_FEATURE_COMPAT_HAS_JOURNAL		0x0004
1590#define EXT4_FEATURE_COMPAT_EXT_ATTR		0x0008
1591#define EXT4_FEATURE_COMPAT_RESIZE_INODE	0x0010
1592#define EXT4_FEATURE_COMPAT_DIR_INDEX		0x0020
1593#define EXT4_FEATURE_COMPAT_SPARSE_SUPER2	0x0200
1594
1595#define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER	0x0001
1596#define EXT4_FEATURE_RO_COMPAT_LARGE_FILE	0x0002
1597#define EXT4_FEATURE_RO_COMPAT_BTREE_DIR	0x0004
1598#define EXT4_FEATURE_RO_COMPAT_HUGE_FILE        0x0008
1599#define EXT4_FEATURE_RO_COMPAT_GDT_CSUM		0x0010
1600#define EXT4_FEATURE_RO_COMPAT_DIR_NLINK	0x0020
1601#define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE	0x0040
1602#define EXT4_FEATURE_RO_COMPAT_QUOTA		0x0100
1603#define EXT4_FEATURE_RO_COMPAT_BIGALLOC		0x0200
1604/*
1605 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM).  When
1606 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1607 * all other data structures' checksums.  However, the METADATA_CSUM and
1608 * GDT_CSUM bits are mutually exclusive.
1609 */
1610#define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM	0x0400
1611#define EXT4_FEATURE_RO_COMPAT_READONLY		0x1000
1612
1613#define EXT4_FEATURE_INCOMPAT_COMPRESSION	0x0001
1614#define EXT4_FEATURE_INCOMPAT_FILETYPE		0x0002
1615#define EXT4_FEATURE_INCOMPAT_RECOVER		0x0004 /* Needs recovery */
1616#define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV	0x0008 /* Journal device */
1617#define EXT4_FEATURE_INCOMPAT_META_BG		0x0010
1618#define EXT4_FEATURE_INCOMPAT_EXTENTS		0x0040 /* extents support */
1619#define EXT4_FEATURE_INCOMPAT_64BIT		0x0080
1620#define EXT4_FEATURE_INCOMPAT_MMP               0x0100
1621#define EXT4_FEATURE_INCOMPAT_FLEX_BG		0x0200
1622#define EXT4_FEATURE_INCOMPAT_EA_INODE		0x0400 /* EA in inode */
1623#define EXT4_FEATURE_INCOMPAT_DIRDATA		0x1000 /* data in dirent */
1624#define EXT4_FEATURE_INCOMPAT_BG_USE_META_CSUM	0x2000 /* use crc32c for bg */
1625#define EXT4_FEATURE_INCOMPAT_LARGEDIR		0x4000 /* >2GB or 3-lvl htree */
1626#define EXT4_FEATURE_INCOMPAT_INLINE_DATA	0x8000 /* data in inode */
1627#define EXT4_FEATURE_INCOMPAT_ENCRYPT		0x10000
1628
1629#define EXT2_FEATURE_COMPAT_SUPP	EXT4_FEATURE_COMPAT_EXT_ATTR
1630#define EXT2_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
1631					 EXT4_FEATURE_INCOMPAT_META_BG)
1632#define EXT2_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1633					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1634					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1635
1636#define EXT3_FEATURE_COMPAT_SUPP	EXT4_FEATURE_COMPAT_EXT_ATTR
1637#define EXT3_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
1638					 EXT4_FEATURE_INCOMPAT_RECOVER| \
1639					 EXT4_FEATURE_INCOMPAT_META_BG)
1640#define EXT3_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1641					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1642					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1643
1644#define EXT4_FEATURE_COMPAT_SUPP	EXT2_FEATURE_COMPAT_EXT_ATTR
1645#define EXT4_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
1646					 EXT4_FEATURE_INCOMPAT_RECOVER| \
1647					 EXT4_FEATURE_INCOMPAT_META_BG| \
1648					 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1649					 EXT4_FEATURE_INCOMPAT_64BIT| \
1650					 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1651					 EXT4_FEATURE_INCOMPAT_MMP | \
1652					 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
1653					 EXT4_FEATURE_INCOMPAT_ENCRYPT)
1654#define EXT4_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1655					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1656					 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1657					 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1658					 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1659					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1660					 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1661					 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1662					 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1663					 EXT4_FEATURE_RO_COMPAT_QUOTA)
1664
1665/*
1666 * Default values for user and/or group using reserved blocks
1667 */
1668#define	EXT4_DEF_RESUID		0
1669#define	EXT4_DEF_RESGID		0
1670
1671#define EXT4_DEF_INODE_READAHEAD_BLKS	32
1672
1673/*
1674 * Default mount options
1675 */
1676#define EXT4_DEFM_DEBUG		0x0001
1677#define EXT4_DEFM_BSDGROUPS	0x0002
1678#define EXT4_DEFM_XATTR_USER	0x0004
1679#define EXT4_DEFM_ACL		0x0008
1680#define EXT4_DEFM_UID16		0x0010
1681#define EXT4_DEFM_JMODE		0x0060
1682#define EXT4_DEFM_JMODE_DATA	0x0020
1683#define EXT4_DEFM_JMODE_ORDERED	0x0040
1684#define EXT4_DEFM_JMODE_WBACK	0x0060
1685#define EXT4_DEFM_NOBARRIER	0x0100
1686#define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1687#define EXT4_DEFM_DISCARD	0x0400
1688#define EXT4_DEFM_NODELALLOC	0x0800
1689
1690/*
1691 * Default journal batch times
1692 */
1693#define EXT4_DEF_MIN_BATCH_TIME	0
1694#define EXT4_DEF_MAX_BATCH_TIME	15000 /* 15ms */
1695
1696/*
1697 * Minimum number of groups in a flexgroup before we separate out
1698 * directories into the first block group of a flexgroup
1699 */
1700#define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME	4
1701
1702/*
1703 * Structure of a directory entry
1704 */
1705#define EXT4_NAME_LEN 255
1706
1707struct ext4_dir_entry {
1708	__le32	inode;			/* Inode number */
1709	__le16	rec_len;		/* Directory entry length */
1710	__le16	name_len;		/* Name length */
1711	char	name[EXT4_NAME_LEN];	/* File name */
1712};
1713
1714/*
1715 * The new version of the directory entry.  Since EXT4 structures are
1716 * stored in intel byte order, and the name_len field could never be
1717 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1718 * file_type field.
1719 */
1720struct ext4_dir_entry_2 {
1721	__le32	inode;			/* Inode number */
1722	__le16	rec_len;		/* Directory entry length */
1723	__u8	name_len;		/* Name length */
1724	__u8	file_type;
1725	char	name[EXT4_NAME_LEN];	/* File name */
1726};
1727
1728/*
1729 * This is a bogus directory entry at the end of each leaf block that
1730 * records checksums.
1731 */
1732struct ext4_dir_entry_tail {
1733	__le32	det_reserved_zero1;	/* Pretend to be unused */
1734	__le16	det_rec_len;		/* 12 */
1735	__u8	det_reserved_zero2;	/* Zero name length */
1736	__u8	det_reserved_ft;	/* 0xDE, fake file type */
1737	__le32	det_checksum;		/* crc32c(uuid+inum+dirblock) */
1738};
1739
1740#define EXT4_DIRENT_TAIL(block, blocksize) \
1741	((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1742					((blocksize) - \
1743					 sizeof(struct ext4_dir_entry_tail))))
1744
1745/*
1746 * Ext4 directory file types.  Only the low 3 bits are used.  The
1747 * other bits are reserved for now.
1748 */
1749#define EXT4_FT_UNKNOWN		0
1750#define EXT4_FT_REG_FILE	1
1751#define EXT4_FT_DIR		2
1752#define EXT4_FT_CHRDEV		3
1753#define EXT4_FT_BLKDEV		4
1754#define EXT4_FT_FIFO		5
1755#define EXT4_FT_SOCK		6
1756#define EXT4_FT_SYMLINK		7
1757
1758#define EXT4_FT_MAX		8
1759
1760#define EXT4_FT_DIR_CSUM	0xDE
1761
1762/*
1763 * EXT4_DIR_PAD defines the directory entries boundaries
1764 *
1765 * NOTE: It must be a multiple of 4
1766 */
1767#define EXT4_DIR_PAD			4
1768#define EXT4_DIR_ROUND			(EXT4_DIR_PAD - 1)
1769#define EXT4_DIR_REC_LEN(name_len)	(((name_len) + 8 + EXT4_DIR_ROUND) & \
1770					 ~EXT4_DIR_ROUND)
1771#define EXT4_MAX_REC_LEN		((1<<16)-1)
1772
1773/*
1774 * If we ever get support for fs block sizes > page_size, we'll need
1775 * to remove the #if statements in the next two functions...
1776 */
1777static inline unsigned int
1778ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1779{
1780	unsigned len = le16_to_cpu(dlen);
1781
1782#if (PAGE_CACHE_SIZE >= 65536)
1783	if (len == EXT4_MAX_REC_LEN || len == 0)
1784		return blocksize;
1785	return (len & 65532) | ((len & 3) << 16);
1786#else
1787	return len;
1788#endif
1789}
1790
1791static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1792{
1793	if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1794		BUG();
1795#if (PAGE_CACHE_SIZE >= 65536)
1796	if (len < 65536)
1797		return cpu_to_le16(len);
1798	if (len == blocksize) {
1799		if (blocksize == 65536)
1800			return cpu_to_le16(EXT4_MAX_REC_LEN);
1801		else
1802			return cpu_to_le16(0);
1803	}
1804	return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1805#else
1806	return cpu_to_le16(len);
1807#endif
1808}
1809
1810/*
1811 * Hash Tree Directory indexing
1812 * (c) Daniel Phillips, 2001
1813 */
1814
1815#define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
1816				      EXT4_FEATURE_COMPAT_DIR_INDEX) && \
1817		    ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
1818#define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1819#define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1820
1821/* Legal values for the dx_root hash_version field: */
1822
1823#define DX_HASH_LEGACY		0
1824#define DX_HASH_HALF_MD4	1
1825#define DX_HASH_TEA		2
1826#define DX_HASH_LEGACY_UNSIGNED	3
1827#define DX_HASH_HALF_MD4_UNSIGNED	4
1828#define DX_HASH_TEA_UNSIGNED		5
1829
1830static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
1831			      const void *address, unsigned int length)
1832{
1833	struct {
1834		struct shash_desc shash;
1835		char ctx[4];
1836	} desc;
1837	int err;
1838
1839	BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
1840
1841	desc.shash.tfm = sbi->s_chksum_driver;
1842	desc.shash.flags = 0;
1843	*(u32 *)desc.ctx = crc;
1844
1845	err = crypto_shash_update(&desc.shash, address, length);
1846	BUG_ON(err);
1847
1848	return *(u32 *)desc.ctx;
1849}
1850
1851#ifdef __KERNEL__
1852
1853/* hash info structure used by the directory hash */
1854struct dx_hash_info
1855{
1856	u32		hash;
1857	u32		minor_hash;
1858	int		hash_version;
1859	u32		*seed;
1860};
1861
1862
1863/* 32 and 64 bit signed EOF for dx directories */
1864#define EXT4_HTREE_EOF_32BIT   ((1UL  << (32 - 1)) - 1)
1865#define EXT4_HTREE_EOF_64BIT   ((1ULL << (64 - 1)) - 1)
1866
1867
1868/*
1869 * Control parameters used by ext4_htree_next_block
1870 */
1871#define HASH_NB_ALWAYS		1
1872
1873
1874/*
1875 * Describe an inode's exact location on disk and in memory
1876 */
1877struct ext4_iloc
1878{
1879	struct buffer_head *bh;
1880	unsigned long offset;
1881	ext4_group_t block_group;
1882};
1883
1884static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
1885{
1886	return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
1887}
1888
1889/*
1890 * This structure is stuffed into the struct file's private_data field
1891 * for directories.  It is where we put information so that we can do
1892 * readdir operations in hash tree order.
1893 */
1894struct dir_private_info {
1895	struct rb_root	root;
1896	struct rb_node	*curr_node;
1897	struct fname	*extra_fname;
1898	loff_t		last_pos;
1899	__u32		curr_hash;
1900	__u32		curr_minor_hash;
1901	__u32		next_hash;
1902};
1903
1904/* calculate the first block number of the group */
1905static inline ext4_fsblk_t
1906ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
1907{
1908	return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
1909		le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
1910}
1911
1912/*
1913 * Special error return code only used by dx_probe() and its callers.
1914 */
1915#define ERR_BAD_DX_DIR	(-(MAX_ERRNO - 1))
1916
1917/*
1918 * Timeout and state flag for lazy initialization inode thread.
1919 */
1920#define EXT4_DEF_LI_WAIT_MULT			10
1921#define EXT4_DEF_LI_MAX_START_DELAY		5
1922#define EXT4_LAZYINIT_QUIT			0x0001
1923#define EXT4_LAZYINIT_RUNNING			0x0002
1924
1925/*
1926 * Lazy inode table initialization info
1927 */
1928struct ext4_lazy_init {
1929	unsigned long		li_state;
1930	struct list_head	li_request_list;
1931	struct mutex		li_list_mtx;
1932};
1933
1934struct ext4_li_request {
1935	struct super_block	*lr_super;
1936	struct ext4_sb_info	*lr_sbi;
1937	ext4_group_t		lr_next_group;
1938	struct list_head	lr_request;
1939	unsigned long		lr_next_sched;
1940	unsigned long		lr_timeout;
1941};
1942
1943struct ext4_features {
1944	struct kobject f_kobj;
1945	struct completion f_kobj_unregister;
1946};
1947
1948/*
1949 * This structure will be used for multiple mount protection. It will be
1950 * written into the block number saved in the s_mmp_block field in the
1951 * superblock. Programs that check MMP should assume that if
1952 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
1953 * to use the filesystem, regardless of how old the timestamp is.
1954 */
1955#define EXT4_MMP_MAGIC     0x004D4D50U /* ASCII for MMP */
1956#define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
1957#define EXT4_MMP_SEQ_FSCK  0xE24D4D50U /* mmp_seq value when being fscked */
1958#define EXT4_MMP_SEQ_MAX   0xE24D4D4FU /* maximum valid mmp_seq value */
1959
1960struct mmp_struct {
1961	__le32	mmp_magic;		/* Magic number for MMP */
1962	__le32	mmp_seq;		/* Sequence no. updated periodically */
1963
1964	/*
1965	 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
1966	 * purposes and do not affect the correctness of the algorithm
1967	 */
1968	__le64	mmp_time;		/* Time last updated */
1969	char	mmp_nodename[64];	/* Node which last updated MMP block */
1970	char	mmp_bdevname[32];	/* Bdev which last updated MMP block */
1971
1972	/*
1973	 * mmp_check_interval is used to verify if the MMP block has been
1974	 * updated on the block device. The value is updated based on the
1975	 * maximum time to write the MMP block during an update cycle.
1976	 */
1977	__le16	mmp_check_interval;
1978
1979	__le16	mmp_pad1;
1980	__le32	mmp_pad2[226];
1981	__le32	mmp_checksum;		/* crc32c(uuid+mmp_block) */
1982};
1983
1984/* arguments passed to the mmp thread */
1985struct mmpd_data {
1986	struct buffer_head *bh; /* bh from initial read_mmp_block() */
1987	struct super_block *sb;  /* super block of the fs */
1988};
1989
1990/*
1991 * Check interval multiplier
1992 * The MMP block is written every update interval and initially checked every
1993 * update interval x the multiplier (the value is then adapted based on the
1994 * write latency). The reason is that writes can be delayed under load and we
1995 * don't want readers to incorrectly assume that the filesystem is no longer
1996 * in use.
1997 */
1998#define EXT4_MMP_CHECK_MULT		2UL
1999
2000/*
2001 * Minimum interval for MMP checking in seconds.
2002 */
2003#define EXT4_MMP_MIN_CHECK_INTERVAL	5UL
2004
2005/*
2006 * Maximum interval for MMP checking in seconds.
2007 */
2008#define EXT4_MMP_MAX_CHECK_INTERVAL	300UL
2009
2010/*
2011 * Function prototypes
2012 */
2013
2014/*
2015 * Ok, these declarations are also in <linux/kernel.h> but none of the
2016 * ext4 source programs needs to include it so they are duplicated here.
2017 */
2018# define NORET_TYPE	/**/
2019# define ATTRIB_NORET	__attribute__((noreturn))
2020# define NORET_AND	noreturn,
2021
2022/* bitmap.c */
2023extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2024void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2025				struct ext4_group_desc *gdp,
2026				struct buffer_head *bh, int sz);
2027int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2028				  struct ext4_group_desc *gdp,
2029				  struct buffer_head *bh, int sz);
2030void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2031				struct ext4_group_desc *gdp,
2032				struct buffer_head *bh);
2033int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2034				  struct ext4_group_desc *gdp,
2035				  struct buffer_head *bh);
2036
2037/* balloc.c */
2038extern void ext4_get_group_no_and_offset(struct super_block *sb,
2039					 ext4_fsblk_t blocknr,
2040					 ext4_group_t *blockgrpp,
2041					 ext4_grpblk_t *offsetp);
2042extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2043					  ext4_fsblk_t block);
2044
2045extern unsigned int ext4_block_group(struct super_block *sb,
2046			ext4_fsblk_t blocknr);
2047extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2048			ext4_fsblk_t blocknr);
2049extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2050extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2051			ext4_group_t group);
2052extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2053					 ext4_fsblk_t goal,
2054					 unsigned int flags,
2055					 unsigned long *count,
2056					 int *errp);
2057extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2058				    s64 nclusters, unsigned int flags);
2059extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2060extern void ext4_check_blocks_bitmap(struct super_block *);
2061extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2062						    ext4_group_t block_group,
2063						    struct buffer_head ** bh);
2064extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2065
2066extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2067						ext4_group_t block_group);
2068extern int ext4_wait_block_bitmap(struct super_block *sb,
2069				  ext4_group_t block_group,
2070				  struct buffer_head *bh);
2071extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2072						  ext4_group_t block_group);
2073extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2074					      ext4_group_t block_group,
2075					      struct ext4_group_desc *gdp);
2076ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2077
2078/* crypto_policy.c */
2079int ext4_is_child_context_consistent_with_parent(struct inode *parent,
2080						 struct inode *child);
2081int ext4_inherit_context(struct inode *parent, struct inode *child);
2082void ext4_to_hex(char *dst, char *src, size_t src_size);
2083int ext4_process_policy(const struct ext4_encryption_policy *policy,
2084			struct inode *inode);
2085int ext4_get_policy(struct inode *inode,
2086		    struct ext4_encryption_policy *policy);
2087
2088/* crypto.c */
2089bool ext4_valid_contents_enc_mode(uint32_t mode);
2090uint32_t ext4_validate_encryption_key_size(uint32_t mode, uint32_t size);
2091extern struct workqueue_struct *ext4_read_workqueue;
2092struct ext4_crypto_ctx *ext4_get_crypto_ctx(struct inode *inode);
2093void ext4_release_crypto_ctx(struct ext4_crypto_ctx *ctx);
2094void ext4_restore_control_page(struct page *data_page);
2095struct page *ext4_encrypt(struct inode *inode,
2096			  struct page *plaintext_page);
2097int ext4_decrypt(struct ext4_crypto_ctx *ctx, struct page *page);
2098int ext4_decrypt_one(struct inode *inode, struct page *page);
2099int ext4_encrypted_zeroout(struct inode *inode, struct ext4_extent *ex);
2100
2101#ifdef CONFIG_EXT4_FS_ENCRYPTION
2102int ext4_init_crypto(void);
2103void ext4_exit_crypto(void);
2104static inline int ext4_sb_has_crypto(struct super_block *sb)
2105{
2106	return EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_ENCRYPT);
2107}
2108#else
2109static inline int ext4_init_crypto(void) { return 0; }
2110static inline void ext4_exit_crypto(void) { }
2111static inline int ext4_sb_has_crypto(struct super_block *sb)
2112{
2113	return 0;
2114}
2115#endif
2116
2117/* crypto_fname.c */
2118bool ext4_valid_filenames_enc_mode(uint32_t mode);
2119u32 ext4_fname_crypto_round_up(u32 size, u32 blksize);
2120int ext4_fname_crypto_alloc_buffer(struct ext4_fname_crypto_ctx *ctx,
2121				   u32 ilen, struct ext4_str *crypto_str);
2122int _ext4_fname_disk_to_usr(struct ext4_fname_crypto_ctx *ctx,
2123			    struct dx_hash_info *hinfo,
2124			    const struct ext4_str *iname,
2125			    struct ext4_str *oname);
2126int ext4_fname_disk_to_usr(struct ext4_fname_crypto_ctx *ctx,
2127			   struct dx_hash_info *hinfo,
2128			   const struct ext4_dir_entry_2 *de,
2129			   struct ext4_str *oname);
2130int ext4_fname_usr_to_disk(struct ext4_fname_crypto_ctx *ctx,
2131			   const struct qstr *iname,
2132			   struct ext4_str *oname);
2133int ext4_fname_usr_to_hash(struct ext4_fname_crypto_ctx *ctx,
2134			   const struct qstr *iname,
2135			   struct dx_hash_info *hinfo);
2136int ext4_fname_crypto_namelen_on_disk(struct ext4_fname_crypto_ctx *ctx,
2137				      u32 namelen);
2138int ext4_fname_match(struct ext4_fname_crypto_ctx *ctx, struct ext4_str *cstr,
2139		     int len, const char * const name,
2140		     struct ext4_dir_entry_2 *de);
2141
2142
2143#ifdef CONFIG_EXT4_FS_ENCRYPTION
2144void ext4_put_fname_crypto_ctx(struct ext4_fname_crypto_ctx **ctx);
2145struct ext4_fname_crypto_ctx *ext4_get_fname_crypto_ctx(struct inode *inode,
2146							u32 max_len);
2147void ext4_fname_crypto_free_buffer(struct ext4_str *crypto_str);
2148#else
2149static inline
2150void ext4_put_fname_crypto_ctx(struct ext4_fname_crypto_ctx **ctx) { }
2151static inline
2152struct ext4_fname_crypto_ctx *ext4_get_fname_crypto_ctx(struct inode *inode,
2153							u32 max_len)
2154{
2155	return NULL;
2156}
2157static inline void ext4_fname_crypto_free_buffer(struct ext4_str *p) { }
2158#endif
2159
2160
2161/* crypto_key.c */
2162int ext4_generate_encryption_key(struct inode *inode);
2163
2164#ifdef CONFIG_EXT4_FS_ENCRYPTION
2165int ext4_has_encryption_key(struct inode *inode);
2166#else
2167static inline int ext4_has_encryption_key(struct inode *inode)
2168{
2169	return 0;
2170}
2171#endif
2172
2173
2174/* dir.c */
2175extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2176				  struct file *,
2177				  struct ext4_dir_entry_2 *,
2178				  struct buffer_head *, char *, int,
2179				  unsigned int);
2180#define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset)	\
2181	unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2182					(de), (bh), (buf), (size), (offset)))
2183extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2184				__u32 minor_hash,
2185				struct ext4_dir_entry_2 *dirent,
2186				struct ext4_str *ent_name);
2187extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2188extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2189			     struct buffer_head *bh,
2190			     void *buf, int buf_size,
2191			     const char *name, int namelen,
2192			     struct ext4_dir_entry_2 **dest_de);
2193int ext4_insert_dentry(struct inode *dir,
2194			struct inode *inode,
2195			struct ext4_dir_entry_2 *de,
2196			int buf_size,
2197		       const struct qstr *iname,
2198			const char *name, int namelen);
2199static inline void ext4_update_dx_flag(struct inode *inode)
2200{
2201	if (!EXT4_HAS_COMPAT_FEATURE(inode->i_sb,
2202				     EXT4_FEATURE_COMPAT_DIR_INDEX))
2203		ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2204}
2205static unsigned char ext4_filetype_table[] = {
2206	DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2207};
2208
2209static inline  unsigned char get_dtype(struct super_block *sb, int filetype)
2210{
2211	if (!EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE) ||
2212	    (filetype >= EXT4_FT_MAX))
2213		return DT_UNKNOWN;
2214
2215	return ext4_filetype_table[filetype];
2216}
2217extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2218			     void *buf, int buf_size);
2219
2220/* fsync.c */
2221extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2222
2223/* hash.c */
2224extern int ext4fs_dirhash(const char *name, int len, struct
2225			  dx_hash_info *hinfo);
2226
2227/* ialloc.c */
2228extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2229				      const struct qstr *qstr, __u32 goal,
2230				      uid_t *owner, int handle_type,
2231				      unsigned int line_no, int nblocks);
2232
2233#define ext4_new_inode(handle, dir, mode, qstr, goal, owner) \
2234	__ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2235			 0, 0, 0)
2236#define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2237				    type, nblocks)		    \
2238	__ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2239			 (type), __LINE__, (nblocks))
2240
2241
2242extern void ext4_free_inode(handle_t *, struct inode *);
2243extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2244extern unsigned long ext4_count_free_inodes(struct super_block *);
2245extern unsigned long ext4_count_dirs(struct super_block *);
2246extern void ext4_check_inodes_bitmap(struct super_block *);
2247extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2248extern int ext4_init_inode_table(struct super_block *sb,
2249				 ext4_group_t group, int barrier);
2250extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2251
2252/* mballoc.c */
2253extern long ext4_mb_stats;
2254extern long ext4_mb_max_to_scan;
2255extern int ext4_mb_init(struct super_block *);
2256extern int ext4_mb_release(struct super_block *);
2257extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2258				struct ext4_allocation_request *, int *);
2259extern int ext4_mb_reserve_blocks(struct super_block *, int);
2260extern void ext4_discard_preallocations(struct inode *);
2261extern int __init ext4_init_mballoc(void);
2262extern void ext4_exit_mballoc(void);
2263extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2264			     struct buffer_head *bh, ext4_fsblk_t block,
2265			     unsigned long count, int flags);
2266extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2267				   ext4_group_t ngroups);
2268extern int ext4_mb_add_groupinfo(struct super_block *sb,
2269		ext4_group_t i, struct ext4_group_desc *desc);
2270extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2271				ext4_fsblk_t block, unsigned long count);
2272extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2273
2274/* inode.c */
2275int ext4_inode_is_fast_symlink(struct inode *inode);
2276struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2277struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2278int ext4_get_block_write(struct inode *inode, sector_t iblock,
2279			 struct buffer_head *bh_result, int create);
2280int ext4_get_block(struct inode *inode, sector_t iblock,
2281				struct buffer_head *bh_result, int create);
2282int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2283			   struct buffer_head *bh, int create);
2284int ext4_walk_page_buffers(handle_t *handle,
2285			   struct buffer_head *head,
2286			   unsigned from,
2287			   unsigned to,
2288			   int *partial,
2289			   int (*fn)(handle_t *handle,
2290				     struct buffer_head *bh));
2291int do_journal_get_write_access(handle_t *handle,
2292				struct buffer_head *bh);
2293#define FALL_BACK_TO_NONDELALLOC 1
2294#define CONVERT_INLINE_DATA	 2
2295
2296extern struct inode *ext4_iget(struct super_block *, unsigned long);
2297extern struct inode *ext4_iget_normal(struct super_block *, unsigned long);
2298extern int  ext4_write_inode(struct inode *, struct writeback_control *);
2299extern int  ext4_setattr(struct dentry *, struct iattr *);
2300extern int  ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
2301				struct kstat *stat);
2302extern void ext4_evict_inode(struct inode *);
2303extern void ext4_clear_inode(struct inode *);
2304extern int  ext4_sync_inode(handle_t *, struct inode *);
2305extern void ext4_dirty_inode(struct inode *, int);
2306extern int ext4_change_inode_journal_flag(struct inode *, int);
2307extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2308extern int ext4_inode_attach_jinode(struct inode *inode);
2309extern int ext4_can_truncate(struct inode *inode);
2310extern void ext4_truncate(struct inode *);
2311extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2312extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
2313extern void ext4_set_inode_flags(struct inode *);
2314extern void ext4_get_inode_flags(struct ext4_inode_info *);
2315extern int ext4_alloc_da_blocks(struct inode *inode);
2316extern void ext4_set_aops(struct inode *inode);
2317extern int ext4_writepage_trans_blocks(struct inode *);
2318extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2319extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2320			     loff_t lstart, loff_t lend);
2321extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
2322extern int ext4_filemap_fault(struct vm_area_struct *vma, struct vm_fault *vmf);
2323extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2324extern void ext4_da_update_reserve_space(struct inode *inode,
2325					int used, int quota_claim);
2326
2327/* indirect.c */
2328extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2329				struct ext4_map_blocks *map, int flags);
2330extern ssize_t ext4_ind_direct_IO(struct kiocb *iocb, struct iov_iter *iter,
2331				  loff_t offset);
2332extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2333extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2334extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2335extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2336				 ext4_lblk_t start, ext4_lblk_t end);
2337
2338/* ioctl.c */
2339extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2340extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2341
2342/* migrate.c */
2343extern int ext4_ext_migrate(struct inode *);
2344extern int ext4_ind_migrate(struct inode *inode);
2345
2346/* namei.c */
2347extern int ext4_dirent_csum_verify(struct inode *inode,
2348				   struct ext4_dir_entry *dirent);
2349extern int ext4_orphan_add(handle_t *, struct inode *);
2350extern int ext4_orphan_del(handle_t *, struct inode *);
2351extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2352				__u32 start_minor_hash, __u32 *next_hash);
2353extern int search_dir(struct buffer_head *bh,
2354		      char *search_buf,
2355		      int buf_size,
2356		      struct inode *dir,
2357		      const struct qstr *d_name,
2358		      unsigned int offset,
2359		      struct ext4_dir_entry_2 **res_dir);
2360extern int ext4_generic_delete_entry(handle_t *handle,
2361				     struct inode *dir,
2362				     struct ext4_dir_entry_2 *de_del,
2363				     struct buffer_head *bh,
2364				     void *entry_buf,
2365				     int buf_size,
2366				     int csum_size);
2367extern int ext4_empty_dir(struct inode *inode);
2368
2369/* resize.c */
2370extern int ext4_group_add(struct super_block *sb,
2371				struct ext4_new_group_data *input);
2372extern int ext4_group_extend(struct super_block *sb,
2373				struct ext4_super_block *es,
2374				ext4_fsblk_t n_blocks_count);
2375extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2376
2377/* super.c */
2378extern int ext4_calculate_overhead(struct super_block *sb);
2379extern void ext4_superblock_csum_set(struct super_block *sb);
2380extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2381extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2382extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2383				    ext4_group_t ngroup);
2384extern const char *ext4_decode_error(struct super_block *sb, int errno,
2385				     char nbuf[16]);
2386
2387extern __printf(4, 5)
2388void __ext4_error(struct super_block *, const char *, unsigned int,
2389		  const char *, ...);
2390extern __printf(5, 6)
2391void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2392		      const char *, ...);
2393extern __printf(5, 6)
2394void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2395		     const char *, ...);
2396extern void __ext4_std_error(struct super_block *, const char *,
2397			     unsigned int, int);
2398extern __printf(4, 5)
2399void __ext4_abort(struct super_block *, const char *, unsigned int,
2400		  const char *, ...);
2401extern __printf(4, 5)
2402void __ext4_warning(struct super_block *, const char *, unsigned int,
2403		    const char *, ...);
2404extern __printf(3, 4)
2405void __ext4_msg(struct super_block *, const char *, const char *, ...);
2406extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2407			   const char *, unsigned int, const char *);
2408extern __printf(7, 8)
2409void __ext4_grp_locked_error(const char *, unsigned int,
2410			     struct super_block *, ext4_group_t,
2411			     unsigned long, ext4_fsblk_t,
2412			     const char *, ...);
2413
2414#ifdef CONFIG_PRINTK
2415
2416#define ext4_error_inode(inode, func, line, block, fmt, ...)		\
2417	__ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__)
2418#define ext4_error_file(file, func, line, block, fmt, ...)		\
2419	__ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2420#define ext4_error(sb, fmt, ...)					\
2421	__ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2422#define ext4_abort(sb, fmt, ...)					\
2423	__ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2424#define ext4_warning(sb, fmt, ...)					\
2425	__ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2426#define ext4_msg(sb, level, fmt, ...)				\
2427	__ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2428#define dump_mmp_msg(sb, mmp, msg)					\
2429	__dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2430#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...)		\
2431	__ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2432				fmt, ##__VA_ARGS__)
2433
2434#else
2435
2436#define ext4_error_inode(inode, func, line, block, fmt, ...)		\
2437do {									\
2438	no_printk(fmt, ##__VA_ARGS__);					\
2439	__ext4_error_inode(inode, "", 0, block, " ");			\
2440} while (0)
2441#define ext4_error_file(file, func, line, block, fmt, ...)		\
2442do {									\
2443	no_printk(fmt, ##__VA_ARGS__);					\
2444	__ext4_error_file(file, "", 0, block, " ");			\
2445} while (0)
2446#define ext4_error(sb, fmt, ...)					\
2447do {									\
2448	no_printk(fmt, ##__VA_ARGS__);					\
2449	__ext4_error(sb, "", 0, " ");					\
2450} while (0)
2451#define ext4_abort(sb, fmt, ...)					\
2452do {									\
2453	no_printk(fmt, ##__VA_ARGS__);					\
2454	__ext4_abort(sb, "", 0, " ");					\
2455} while (0)
2456#define ext4_warning(sb, fmt, ...)					\
2457do {									\
2458	no_printk(fmt, ##__VA_ARGS__);					\
2459	__ext4_warning(sb, "", 0, " ");					\
2460} while (0)
2461#define ext4_msg(sb, level, fmt, ...)					\
2462do {									\
2463	no_printk(fmt, ##__VA_ARGS__);					\
2464	__ext4_msg(sb, "", " ");					\
2465} while (0)
2466#define dump_mmp_msg(sb, mmp, msg)					\
2467	__dump_mmp_msg(sb, mmp, "", 0, "")
2468#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...)		\
2469do {									\
2470	no_printk(fmt, ##__VA_ARGS__);				\
2471	__ext4_grp_locked_error("", 0, sb, grp, ino, block, " ");	\
2472} while (0)
2473
2474#endif
2475
2476extern void ext4_update_dynamic_rev(struct super_block *sb);
2477extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2478					__u32 compat);
2479extern int ext4_update_rocompat_feature(handle_t *handle,
2480					struct super_block *sb,	__u32 rocompat);
2481extern int ext4_update_incompat_feature(handle_t *handle,
2482					struct super_block *sb,	__u32 incompat);
2483extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2484				      struct ext4_group_desc *bg);
2485extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2486				      struct ext4_group_desc *bg);
2487extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2488				     struct ext4_group_desc *bg);
2489extern __u32 ext4_free_group_clusters(struct super_block *sb,
2490				      struct ext4_group_desc *bg);
2491extern __u32 ext4_free_inodes_count(struct super_block *sb,
2492				 struct ext4_group_desc *bg);
2493extern __u32 ext4_used_dirs_count(struct super_block *sb,
2494				struct ext4_group_desc *bg);
2495extern __u32 ext4_itable_unused_count(struct super_block *sb,
2496				   struct ext4_group_desc *bg);
2497extern void ext4_block_bitmap_set(struct super_block *sb,
2498				  struct ext4_group_desc *bg, ext4_fsblk_t blk);
2499extern void ext4_inode_bitmap_set(struct super_block *sb,
2500				  struct ext4_group_desc *bg, ext4_fsblk_t blk);
2501extern void ext4_inode_table_set(struct super_block *sb,
2502				 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2503extern void ext4_free_group_clusters_set(struct super_block *sb,
2504					 struct ext4_group_desc *bg,
2505					 __u32 count);
2506extern void ext4_free_inodes_set(struct super_block *sb,
2507				struct ext4_group_desc *bg, __u32 count);
2508extern void ext4_used_dirs_set(struct super_block *sb,
2509				struct ext4_group_desc *bg, __u32 count);
2510extern void ext4_itable_unused_set(struct super_block *sb,
2511				   struct ext4_group_desc *bg, __u32 count);
2512extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2513				       struct ext4_group_desc *gdp);
2514extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2515				     struct ext4_group_desc *gdp);
2516extern int ext4_register_li_request(struct super_block *sb,
2517				    ext4_group_t first_not_zeroed);
2518
2519static inline int ext4_has_group_desc_csum(struct super_block *sb)
2520{
2521	return EXT4_HAS_RO_COMPAT_FEATURE(sb,
2522					  EXT4_FEATURE_RO_COMPAT_GDT_CSUM) ||
2523	       (EXT4_SB(sb)->s_chksum_driver != NULL);
2524}
2525
2526static inline int ext4_has_metadata_csum(struct super_block *sb)
2527{
2528	WARN_ON_ONCE(EXT4_HAS_RO_COMPAT_FEATURE(sb,
2529			EXT4_FEATURE_RO_COMPAT_METADATA_CSUM) &&
2530		     !EXT4_SB(sb)->s_chksum_driver);
2531
2532	return (EXT4_SB(sb)->s_chksum_driver != NULL);
2533}
2534static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2535{
2536	return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2537		le32_to_cpu(es->s_blocks_count_lo);
2538}
2539
2540static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2541{
2542	return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2543		le32_to_cpu(es->s_r_blocks_count_lo);
2544}
2545
2546static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2547{
2548	return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2549		le32_to_cpu(es->s_free_blocks_count_lo);
2550}
2551
2552static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2553					 ext4_fsblk_t blk)
2554{
2555	es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2556	es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2557}
2558
2559static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2560					      ext4_fsblk_t blk)
2561{
2562	es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2563	es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2564}
2565
2566static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2567					   ext4_fsblk_t blk)
2568{
2569	es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2570	es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2571}
2572
2573static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
2574{
2575	if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2576		return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2577			le32_to_cpu(raw_inode->i_size_lo);
2578	else
2579		return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2580}
2581
2582static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2583{
2584	raw_inode->i_size_lo = cpu_to_le32(i_size);
2585	raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2586}
2587
2588static inline
2589struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2590					    ext4_group_t group)
2591{
2592	 struct ext4_group_info ***grp_info;
2593	 long indexv, indexh;
2594	 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
2595	 grp_info = EXT4_SB(sb)->s_group_info;
2596	 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2597	 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2598	 return grp_info[indexv][indexh];
2599}
2600
2601/*
2602 * Reading s_groups_count requires using smp_rmb() afterwards.  See
2603 * the locking protocol documented in the comments of ext4_group_add()
2604 * in resize.c
2605 */
2606static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2607{
2608	ext4_group_t	ngroups = EXT4_SB(sb)->s_groups_count;
2609
2610	smp_rmb();
2611	return ngroups;
2612}
2613
2614static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2615					     ext4_group_t block_group)
2616{
2617	return block_group >> sbi->s_log_groups_per_flex;
2618}
2619
2620static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2621{
2622	return 1 << sbi->s_log_groups_per_flex;
2623}
2624
2625#define ext4_std_error(sb, errno)				\
2626do {								\
2627	if ((errno))						\
2628		__ext4_std_error((sb), __func__, __LINE__, (errno));	\
2629} while (0)
2630
2631#ifdef CONFIG_SMP
2632/* Each CPU can accumulate percpu_counter_batch clusters in their local
2633 * counters. So we need to make sure we have free clusters more
2634 * than percpu_counter_batch  * nr_cpu_ids. Also add a window of 4 times.
2635 */
2636#define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2637#else
2638#define EXT4_FREECLUSTERS_WATERMARK 0
2639#endif
2640
2641/* Update i_disksize. Requires i_mutex to avoid races with truncate */
2642static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2643{
2644	WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
2645		     !mutex_is_locked(&inode->i_mutex));
2646	down_write(&EXT4_I(inode)->i_data_sem);
2647	if (newsize > EXT4_I(inode)->i_disksize)
2648		EXT4_I(inode)->i_disksize = newsize;
2649	up_write(&EXT4_I(inode)->i_data_sem);
2650}
2651
2652/* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
2653static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
2654{
2655	int changed = 0;
2656
2657	if (newsize > inode->i_size) {
2658		i_size_write(inode, newsize);
2659		changed = 1;
2660	}
2661	if (newsize > EXT4_I(inode)->i_disksize) {
2662		ext4_update_i_disksize(inode, newsize);
2663		changed |= 2;
2664	}
2665	return changed;
2666}
2667
2668int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
2669				      loff_t len);
2670
2671struct ext4_group_info {
2672	unsigned long   bb_state;
2673	struct rb_root  bb_free_root;
2674	ext4_grpblk_t	bb_first_free;	/* first free block */
2675	ext4_grpblk_t	bb_free;	/* total free blocks */
2676	ext4_grpblk_t	bb_fragments;	/* nr of freespace fragments */
2677	ext4_grpblk_t	bb_largest_free_order;/* order of largest frag in BG */
2678	struct          list_head bb_prealloc_list;
2679#ifdef DOUBLE_CHECK
2680	void            *bb_bitmap;
2681#endif
2682	struct rw_semaphore alloc_sem;
2683	ext4_grpblk_t	bb_counters[];	/* Nr of free power-of-two-block
2684					 * regions, index is order.
2685					 * bb_counters[3] = 5 means
2686					 * 5 free 8-block regions. */
2687};
2688
2689#define EXT4_GROUP_INFO_NEED_INIT_BIT		0
2690#define EXT4_GROUP_INFO_WAS_TRIMMED_BIT		1
2691#define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT	2
2692#define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT	3
2693
2694#define EXT4_MB_GRP_NEED_INIT(grp)	\
2695	(test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2696#define EXT4_MB_GRP_BBITMAP_CORRUPT(grp)	\
2697	(test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2698#define EXT4_MB_GRP_IBITMAP_CORRUPT(grp)	\
2699	(test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2700
2701#define EXT4_MB_GRP_WAS_TRIMMED(grp)	\
2702	(test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2703#define EXT4_MB_GRP_SET_TRIMMED(grp)	\
2704	(set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2705#define EXT4_MB_GRP_CLEAR_TRIMMED(grp)	\
2706	(clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2707
2708#define EXT4_MAX_CONTENTION		8
2709#define EXT4_CONTENTION_THRESHOLD	2
2710
2711static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2712					      ext4_group_t group)
2713{
2714	return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2715}
2716
2717/*
2718 * Returns true if the filesystem is busy enough that attempts to
2719 * access the block group locks has run into contention.
2720 */
2721static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2722{
2723	return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2724}
2725
2726static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2727{
2728	spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2729	if (spin_trylock(lock))
2730		/*
2731		 * We're able to grab the lock right away, so drop the
2732		 * lock contention counter.
2733		 */
2734		atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2735	else {
2736		/*
2737		 * The lock is busy, so bump the contention counter,
2738		 * and then wait on the spin lock.
2739		 */
2740		atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2741				  EXT4_MAX_CONTENTION);
2742		spin_lock(lock);
2743	}
2744}
2745
2746static inline void ext4_unlock_group(struct super_block *sb,
2747					ext4_group_t group)
2748{
2749	spin_unlock(ext4_group_lock_ptr(sb, group));
2750}
2751
2752/*
2753 * Block validity checking
2754 */
2755#define ext4_check_indirect_blockref(inode, bh)				\
2756	ext4_check_blockref(__func__, __LINE__, inode,			\
2757			    (__le32 *)(bh)->b_data,			\
2758			    EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2759
2760#define ext4_ind_check_inode(inode)					\
2761	ext4_check_blockref(__func__, __LINE__, inode,			\
2762			    EXT4_I(inode)->i_data,			\
2763			    EXT4_NDIR_BLOCKS)
2764
2765/*
2766 * Inodes and files operations
2767 */
2768
2769/* dir.c */
2770extern const struct file_operations ext4_dir_operations;
2771
2772/* file.c */
2773extern const struct inode_operations ext4_file_inode_operations;
2774extern const struct file_operations ext4_file_operations;
2775extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
2776
2777/* inline.c */
2778extern int ext4_get_max_inline_size(struct inode *inode);
2779extern int ext4_find_inline_data_nolock(struct inode *inode);
2780extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
2781				 unsigned int len);
2782extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
2783
2784extern int ext4_readpage_inline(struct inode *inode, struct page *page);
2785extern int ext4_try_to_write_inline_data(struct address_space *mapping,
2786					 struct inode *inode,
2787					 loff_t pos, unsigned len,
2788					 unsigned flags,
2789					 struct page **pagep);
2790extern int ext4_write_inline_data_end(struct inode *inode,
2791				      loff_t pos, unsigned len,
2792				      unsigned copied,
2793				      struct page *page);
2794extern struct buffer_head *
2795ext4_journalled_write_inline_data(struct inode *inode,
2796				  unsigned len,
2797				  struct page *page);
2798extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
2799					   struct inode *inode,
2800					   loff_t pos, unsigned len,
2801					   unsigned flags,
2802					   struct page **pagep,
2803					   void **fsdata);
2804extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
2805					 unsigned len, unsigned copied,
2806					 struct page *page);
2807extern int ext4_try_add_inline_entry(handle_t *handle, struct dentry *dentry,
2808				     struct inode *inode);
2809extern int ext4_try_create_inline_dir(handle_t *handle,
2810				      struct inode *parent,
2811				      struct inode *inode);
2812extern int ext4_read_inline_dir(struct file *filp,
2813				struct dir_context *ctx,
2814				int *has_inline_data);
2815extern int htree_inlinedir_to_tree(struct file *dir_file,
2816				   struct inode *dir, ext4_lblk_t block,
2817				   struct dx_hash_info *hinfo,
2818				   __u32 start_hash, __u32 start_minor_hash,
2819				   int *has_inline_data);
2820extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
2821					const struct qstr *d_name,
2822					struct ext4_dir_entry_2 **res_dir,
2823					int *has_inline_data);
2824extern int ext4_delete_inline_entry(handle_t *handle,
2825				    struct inode *dir,
2826				    struct ext4_dir_entry_2 *de_del,
2827				    struct buffer_head *bh,
2828				    int *has_inline_data);
2829extern int empty_inline_dir(struct inode *dir, int *has_inline_data);
2830extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
2831					struct ext4_dir_entry_2 **parent_de,
2832					int *retval);
2833extern int ext4_inline_data_fiemap(struct inode *inode,
2834				   struct fiemap_extent_info *fieinfo,
2835				   int *has_inline, __u64 start, __u64 len);
2836extern int ext4_try_to_evict_inline_data(handle_t *handle,
2837					 struct inode *inode,
2838					 int needed);
2839extern void ext4_inline_data_truncate(struct inode *inode, int *has_inline);
2840
2841extern int ext4_convert_inline_data(struct inode *inode);
2842
2843static inline int ext4_has_inline_data(struct inode *inode)
2844{
2845	return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
2846	       EXT4_I(inode)->i_inline_off;
2847}
2848
2849/* namei.c */
2850extern const struct inode_operations ext4_dir_inode_operations;
2851extern const struct inode_operations ext4_special_inode_operations;
2852extern struct dentry *ext4_get_parent(struct dentry *child);
2853extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
2854				 struct ext4_dir_entry_2 *de,
2855				 int blocksize, int csum_size,
2856				 unsigned int parent_ino, int dotdot_real_len);
2857extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t,
2858				   unsigned int blocksize);
2859extern int ext4_handle_dirty_dirent_node(handle_t *handle,
2860					 struct inode *inode,
2861					 struct buffer_head *bh);
2862#define S_SHIFT 12
2863static unsigned char ext4_type_by_mode[S_IFMT >> S_SHIFT] = {
2864	[S_IFREG >> S_SHIFT]	= EXT4_FT_REG_FILE,
2865	[S_IFDIR >> S_SHIFT]	= EXT4_FT_DIR,
2866	[S_IFCHR >> S_SHIFT]	= EXT4_FT_CHRDEV,
2867	[S_IFBLK >> S_SHIFT]	= EXT4_FT_BLKDEV,
2868	[S_IFIFO >> S_SHIFT]	= EXT4_FT_FIFO,
2869	[S_IFSOCK >> S_SHIFT]	= EXT4_FT_SOCK,
2870	[S_IFLNK >> S_SHIFT]	= EXT4_FT_SYMLINK,
2871};
2872
2873static inline void ext4_set_de_type(struct super_block *sb,
2874				struct ext4_dir_entry_2 *de,
2875				umode_t mode) {
2876	if (EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE))
2877		de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
2878}
2879
2880/* readpages.c */
2881extern int ext4_mpage_readpages(struct address_space *mapping,
2882				struct list_head *pages, struct page *page,
2883				unsigned nr_pages);
2884
2885/* symlink.c */
2886extern const struct inode_operations ext4_symlink_inode_operations;
2887extern const struct inode_operations ext4_fast_symlink_inode_operations;
2888
2889/* block_validity */
2890extern void ext4_release_system_zone(struct super_block *sb);
2891extern int ext4_setup_system_zone(struct super_block *sb);
2892extern int __init ext4_init_system_zone(void);
2893extern void ext4_exit_system_zone(void);
2894extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
2895				 ext4_fsblk_t start_blk,
2896				 unsigned int count);
2897extern int ext4_check_blockref(const char *, unsigned int,
2898			       struct inode *, __le32 *, unsigned int);
2899
2900/* extents.c */
2901struct ext4_ext_path;
2902struct ext4_extent;
2903
2904/*
2905 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
2906 * __le32.
2907 */
2908#define EXT_MAX_BLOCKS	0xffffffff
2909
2910extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
2911extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
2912extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
2913extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
2914			       struct ext4_map_blocks *map, int flags);
2915extern void ext4_ext_truncate(handle_t *, struct inode *);
2916extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
2917				 ext4_lblk_t end);
2918extern void ext4_ext_init(struct super_block *);
2919extern void ext4_ext_release(struct super_block *);
2920extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
2921			  loff_t len);
2922extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
2923					  loff_t offset, ssize_t len);
2924extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
2925			   struct ext4_map_blocks *map, int flags);
2926extern int ext4_ext_calc_metadata_amount(struct inode *inode,
2927					 ext4_lblk_t lblocks);
2928extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
2929						   int num,
2930						   struct ext4_ext_path *path);
2931extern int ext4_can_extents_be_merged(struct inode *inode,
2932				      struct ext4_extent *ex1,
2933				      struct ext4_extent *ex2);
2934extern int ext4_ext_insert_extent(handle_t *, struct inode *,
2935				  struct ext4_ext_path **,
2936				  struct ext4_extent *, int);
2937extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
2938					      struct ext4_ext_path **,
2939					      int flags);
2940extern void ext4_ext_drop_refs(struct ext4_ext_path *);
2941extern int ext4_ext_check_inode(struct inode *inode);
2942extern int ext4_find_delalloc_range(struct inode *inode,
2943				    ext4_lblk_t lblk_start,
2944				    ext4_lblk_t lblk_end);
2945extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk);
2946extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
2947extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
2948			__u64 start, __u64 len);
2949extern int ext4_ext_precache(struct inode *inode);
2950extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len);
2951extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
2952				struct inode *inode2, ext4_lblk_t lblk1,
2953			     ext4_lblk_t lblk2,  ext4_lblk_t count,
2954			     int mark_unwritten,int *err);
2955
2956/* move_extent.c */
2957extern void ext4_double_down_write_data_sem(struct inode *first,
2958					    struct inode *second);
2959extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
2960					  struct inode *donor_inode);
2961extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
2962			     __u64 start_orig, __u64 start_donor,
2963			     __u64 len, __u64 *moved_len);
2964
2965/* page-io.c */
2966extern int __init ext4_init_pageio(void);
2967extern void ext4_exit_pageio(void);
2968extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
2969extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
2970extern int ext4_put_io_end(ext4_io_end_t *io_end);
2971extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
2972extern void ext4_io_submit_init(struct ext4_io_submit *io,
2973				struct writeback_control *wbc);
2974extern void ext4_end_io_rsv_work(struct work_struct *work);
2975extern void ext4_io_submit(struct ext4_io_submit *io);
2976extern int ext4_bio_write_page(struct ext4_io_submit *io,
2977			       struct page *page,
2978			       int len,
2979			       struct writeback_control *wbc,
2980			       bool keep_towrite);
2981
2982/* mmp.c */
2983extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
2984
2985/*
2986 * Add new method to test whether block and inode bitmaps are properly
2987 * initialized. With uninit_bg reading the block from disk is not enough
2988 * to mark the bitmap uptodate. We need to also zero-out the bitmap
2989 */
2990#define BH_BITMAP_UPTODATE BH_JBDPrivateStart
2991
2992static inline int bitmap_uptodate(struct buffer_head *bh)
2993{
2994	return (buffer_uptodate(bh) &&
2995			test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
2996}
2997static inline void set_bitmap_uptodate(struct buffer_head *bh)
2998{
2999	set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3000}
3001
3002/*
3003 * Disable DIO read nolock optimization, so new dioreaders will be forced
3004 * to grab i_mutex
3005 */
3006static inline void ext4_inode_block_unlocked_dio(struct inode *inode)
3007{
3008	ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
3009	smp_mb();
3010}
3011static inline void ext4_inode_resume_unlocked_dio(struct inode *inode)
3012{
3013	smp_mb();
3014	ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
3015}
3016
3017#define in_range(b, first, len)	((b) >= (first) && (b) <= (first) + (len) - 1)
3018
3019/* For ioend & aio unwritten conversion wait queues */
3020#define EXT4_WQ_HASH_SZ		37
3021#define ext4_ioend_wq(v)   (&ext4__ioend_wq[((unsigned long)(v)) %\
3022					    EXT4_WQ_HASH_SZ])
3023#define ext4_aio_mutex(v)  (&ext4__aio_mutex[((unsigned long)(v)) %\
3024					     EXT4_WQ_HASH_SZ])
3025extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3026extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ];
3027
3028#define EXT4_RESIZING	0
3029extern int ext4_resize_begin(struct super_block *sb);
3030extern void ext4_resize_end(struct super_block *sb);
3031
3032#endif	/* __KERNEL__ */
3033
3034#endif	/* _EXT4_H */
3035