xref: /kernel/linux/linux-5.10/fs/ext4/ext4.h (revision 8c2ecf20)
1// SPDX-License-Identifier: GPL-2.0
2/*
3 *  ext4.h
4 *
5 * Copyright (C) 1992, 1993, 1994, 1995
6 * Remy Card (card@masi.ibp.fr)
7 * Laboratoire MASI - Institut Blaise Pascal
8 * Universite Pierre et Marie Curie (Paris VI)
9 *
10 *  from
11 *
12 *  linux/include/linux/minix_fs.h
13 *
14 *  Copyright (C) 1991, 1992  Linus Torvalds
15 */
16
17#ifndef _EXT4_H
18#define _EXT4_H
19
20#include <linux/types.h>
21#include <linux/blkdev.h>
22#include <linux/magic.h>
23#include <linux/jbd2.h>
24#include <linux/quota.h>
25#include <linux/rwsem.h>
26#include <linux/rbtree.h>
27#include <linux/seqlock.h>
28#include <linux/mutex.h>
29#include <linux/timer.h>
30#include <linux/wait.h>
31#include <linux/sched/signal.h>
32#include <linux/blockgroup_lock.h>
33#include <linux/percpu_counter.h>
34#include <linux/ratelimit.h>
35#include <crypto/hash.h>
36#include <linux/falloc.h>
37#include <linux/percpu-rwsem.h>
38#include <linux/fiemap.h>
39#ifdef __KERNEL__
40#include <linux/compat.h>
41#endif
42
43#include <linux/fscrypt.h>
44#include <linux/fsverity.h>
45
46#include <linux/compiler.h>
47
48/*
49 * The fourth extended filesystem constants/structures
50 */
51
52/*
53 * with AGGRESSIVE_CHECK allocator runs consistency checks over
54 * structures. these checks slow things down a lot
55 */
56#define AGGRESSIVE_CHECK__
57
58/*
59 * with DOUBLE_CHECK defined mballoc creates persistent in-core
60 * bitmaps, maintains and uses them to check for double allocations
61 */
62#define DOUBLE_CHECK__
63
64/*
65 * Define EXT4FS_DEBUG to produce debug messages
66 */
67#undef EXT4FS_DEBUG
68
69/*
70 * Debug code
71 */
72#ifdef EXT4FS_DEBUG
73#define ext4_debug(f, a...)						\
74	do {								\
75		printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:",	\
76			__FILE__, __LINE__, __func__);			\
77		printk(KERN_DEBUG f, ## a);				\
78	} while (0)
79#else
80#define ext4_debug(fmt, ...)	no_printk(fmt, ##__VA_ARGS__)
81#endif
82
83 /*
84  * Turn on EXT_DEBUG to enable ext4_ext_show_path/leaf/move in extents.c
85  */
86#define EXT_DEBUG__
87
88/*
89 * Dynamic printk for controlled extents debugging.
90 */
91#ifdef CONFIG_EXT4_DEBUG
92#define ext_debug(ino, fmt, ...)					\
93	pr_debug("[%s/%d] EXT4-fs (%s): ino %lu: (%s, %d): %s:" fmt,	\
94		 current->comm, task_pid_nr(current),			\
95		 ino->i_sb->s_id, ino->i_ino, __FILE__, __LINE__,	\
96		 __func__, ##__VA_ARGS__)
97#else
98#define ext_debug(ino, fmt, ...)	no_printk(fmt, ##__VA_ARGS__)
99#endif
100
101/* data type for block offset of block group */
102typedef int ext4_grpblk_t;
103
104/* data type for filesystem-wide blocks number */
105typedef unsigned long long ext4_fsblk_t;
106
107/* data type for file logical block number */
108typedef __u32 ext4_lblk_t;
109
110/* data type for block group number */
111typedef unsigned int ext4_group_t;
112
113enum SHIFT_DIRECTION {
114	SHIFT_LEFT = 0,
115	SHIFT_RIGHT,
116};
117
118/*
119 * Flags used in mballoc's allocation_context flags field.
120 *
121 * Also used to show what's going on for debugging purposes when the
122 * flag field is exported via the traceport interface
123 */
124
125/* prefer goal again. length */
126#define EXT4_MB_HINT_MERGE		0x0001
127/* blocks already reserved */
128#define EXT4_MB_HINT_RESERVED		0x0002
129/* metadata is being allocated */
130#define EXT4_MB_HINT_METADATA		0x0004
131/* first blocks in the file */
132#define EXT4_MB_HINT_FIRST		0x0008
133/* search for the best chunk */
134#define EXT4_MB_HINT_BEST		0x0010
135/* data is being allocated */
136#define EXT4_MB_HINT_DATA		0x0020
137/* don't preallocate (for tails) */
138#define EXT4_MB_HINT_NOPREALLOC		0x0040
139/* allocate for locality group */
140#define EXT4_MB_HINT_GROUP_ALLOC	0x0080
141/* allocate goal blocks or none */
142#define EXT4_MB_HINT_GOAL_ONLY		0x0100
143/* goal is meaningful */
144#define EXT4_MB_HINT_TRY_GOAL		0x0200
145/* blocks already pre-reserved by delayed allocation */
146#define EXT4_MB_DELALLOC_RESERVED	0x0400
147/* We are doing stream allocation */
148#define EXT4_MB_STREAM_ALLOC		0x0800
149/* Use reserved root blocks if needed */
150#define EXT4_MB_USE_ROOT_BLOCKS		0x1000
151/* Use blocks from reserved pool */
152#define EXT4_MB_USE_RESERVED		0x2000
153/* Do strict check for free blocks while retrying block allocation */
154#define EXT4_MB_STRICT_CHECK		0x4000
155
156struct ext4_allocation_request {
157	/* target inode for block we're allocating */
158	struct inode *inode;
159	/* how many blocks we want to allocate */
160	unsigned int len;
161	/* logical block in target inode */
162	ext4_lblk_t logical;
163	/* the closest logical allocated block to the left */
164	ext4_lblk_t lleft;
165	/* the closest logical allocated block to the right */
166	ext4_lblk_t lright;
167	/* phys. target (a hint) */
168	ext4_fsblk_t goal;
169	/* phys. block for the closest logical allocated block to the left */
170	ext4_fsblk_t pleft;
171	/* phys. block for the closest logical allocated block to the right */
172	ext4_fsblk_t pright;
173	/* flags. see above EXT4_MB_HINT_* */
174	unsigned int flags;
175};
176
177/*
178 * Logical to physical block mapping, used by ext4_map_blocks()
179 *
180 * This structure is used to pass requests into ext4_map_blocks() as
181 * well as to store the information returned by ext4_map_blocks().  It
182 * takes less room on the stack than a struct buffer_head.
183 */
184#define EXT4_MAP_NEW		BIT(BH_New)
185#define EXT4_MAP_MAPPED		BIT(BH_Mapped)
186#define EXT4_MAP_UNWRITTEN	BIT(BH_Unwritten)
187#define EXT4_MAP_BOUNDARY	BIT(BH_Boundary)
188#define EXT4_MAP_FLAGS		(EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
189				 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
190
191struct ext4_map_blocks {
192	ext4_fsblk_t m_pblk;
193	ext4_lblk_t m_lblk;
194	unsigned int m_len;
195	unsigned int m_flags;
196};
197
198/*
199 * Block validity checking, system zone rbtree.
200 */
201struct ext4_system_blocks {
202	struct rb_root root;
203	struct rcu_head rcu;
204};
205
206/*
207 * Flags for ext4_io_end->flags
208 */
209#define	EXT4_IO_END_UNWRITTEN	0x0001
210
211struct ext4_io_end_vec {
212	struct list_head list;		/* list of io_end_vec */
213	loff_t offset;			/* offset in the file */
214	ssize_t size;			/* size of the extent */
215};
216
217/*
218 * For converting unwritten extents on a work queue. 'handle' is used for
219 * buffered writeback.
220 */
221typedef struct ext4_io_end {
222	struct list_head	list;		/* per-file finished IO list */
223	handle_t		*handle;	/* handle reserved for extent
224						 * conversion */
225	struct inode		*inode;		/* file being written to */
226	struct bio		*bio;		/* Linked list of completed
227						 * bios covering the extent */
228	unsigned int		flag;		/* unwritten or not */
229	atomic_t		count;		/* reference counter */
230	struct list_head	list_vec;	/* list of ext4_io_end_vec */
231} ext4_io_end_t;
232
233struct ext4_io_submit {
234	struct writeback_control *io_wbc;
235	struct bio		*io_bio;
236	ext4_io_end_t		*io_end;
237	sector_t		io_next_block;
238};
239
240/*
241 * Special inodes numbers
242 */
243#define	EXT4_BAD_INO		 1	/* Bad blocks inode */
244#define EXT4_ROOT_INO		 2	/* Root inode */
245#define EXT4_USR_QUOTA_INO	 3	/* User quota inode */
246#define EXT4_GRP_QUOTA_INO	 4	/* Group quota inode */
247#define EXT4_BOOT_LOADER_INO	 5	/* Boot loader inode */
248#define EXT4_UNDEL_DIR_INO	 6	/* Undelete directory inode */
249#define EXT4_RESIZE_INO		 7	/* Reserved group descriptors inode */
250#define EXT4_JOURNAL_INO	 8	/* Journal inode */
251
252/* First non-reserved inode for old ext4 filesystems */
253#define EXT4_GOOD_OLD_FIRST_INO	11
254
255/*
256 * Maximal count of links to a file
257 */
258#define EXT4_LINK_MAX		65000
259
260/*
261 * Macro-instructions used to manage several block sizes
262 */
263#define EXT4_MIN_BLOCK_SIZE		1024
264#define	EXT4_MAX_BLOCK_SIZE		65536
265#define EXT4_MIN_BLOCK_LOG_SIZE		10
266#define EXT4_MAX_BLOCK_LOG_SIZE		16
267#define EXT4_MAX_CLUSTER_LOG_SIZE	30
268#ifdef __KERNEL__
269# define EXT4_BLOCK_SIZE(s)		((s)->s_blocksize)
270#else
271# define EXT4_BLOCK_SIZE(s)		(EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
272#endif
273#define	EXT4_ADDR_PER_BLOCK(s)		(EXT4_BLOCK_SIZE(s) / sizeof(__u32))
274#define EXT4_CLUSTER_SIZE(s)		(EXT4_BLOCK_SIZE(s) << \
275					 EXT4_SB(s)->s_cluster_bits)
276#ifdef __KERNEL__
277# define EXT4_BLOCK_SIZE_BITS(s)	((s)->s_blocksize_bits)
278# define EXT4_CLUSTER_BITS(s)		(EXT4_SB(s)->s_cluster_bits)
279#else
280# define EXT4_BLOCK_SIZE_BITS(s)	((s)->s_log_block_size + 10)
281#endif
282#ifdef __KERNEL__
283#define	EXT4_ADDR_PER_BLOCK_BITS(s)	(EXT4_SB(s)->s_addr_per_block_bits)
284#define EXT4_INODE_SIZE(s)		(EXT4_SB(s)->s_inode_size)
285#define EXT4_FIRST_INO(s)		(EXT4_SB(s)->s_first_ino)
286#else
287#define EXT4_INODE_SIZE(s)	(((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
288				 EXT4_GOOD_OLD_INODE_SIZE : \
289				 (s)->s_inode_size)
290#define EXT4_FIRST_INO(s)	(((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
291				 EXT4_GOOD_OLD_FIRST_INO : \
292				 (s)->s_first_ino)
293#endif
294#define EXT4_BLOCK_ALIGN(size, blkbits)		ALIGN((size), (1 << (blkbits)))
295#define EXT4_MAX_BLOCKS(size, offset, blkbits) \
296	((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \
297								  blkbits))
298
299/* Translate a block number to a cluster number */
300#define EXT4_B2C(sbi, blk)	((blk) >> (sbi)->s_cluster_bits)
301/* Translate a cluster number to a block number */
302#define EXT4_C2B(sbi, cluster)	((cluster) << (sbi)->s_cluster_bits)
303/* Translate # of blks to # of clusters */
304#define EXT4_NUM_B2C(sbi, blks)	(((blks) + (sbi)->s_cluster_ratio - 1) >> \
305				 (sbi)->s_cluster_bits)
306/* Mask out the low bits to get the starting block of the cluster */
307#define EXT4_PBLK_CMASK(s, pblk) ((pblk) &				\
308				  ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
309#define EXT4_LBLK_CMASK(s, lblk) ((lblk) &				\
310				  ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
311/* Fill in the low bits to get the last block of the cluster */
312#define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) |				\
313				    ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1))
314/* Get the cluster offset */
315#define EXT4_PBLK_COFF(s, pblk) ((pblk) &				\
316				 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
317#define EXT4_LBLK_COFF(s, lblk) ((lblk) &				\
318				 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
319
320/*
321 * Structure of a blocks group descriptor
322 */
323struct ext4_group_desc
324{
325	__le32	bg_block_bitmap_lo;	/* Blocks bitmap block */
326	__le32	bg_inode_bitmap_lo;	/* Inodes bitmap block */
327	__le32	bg_inode_table_lo;	/* Inodes table block */
328	__le16	bg_free_blocks_count_lo;/* Free blocks count */
329	__le16	bg_free_inodes_count_lo;/* Free inodes count */
330	__le16	bg_used_dirs_count_lo;	/* Directories count */
331	__le16	bg_flags;		/* EXT4_BG_flags (INODE_UNINIT, etc) */
332	__le32  bg_exclude_bitmap_lo;   /* Exclude bitmap for snapshots */
333	__le16  bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
334	__le16  bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
335	__le16  bg_itable_unused_lo;	/* Unused inodes count */
336	__le16  bg_checksum;		/* crc16(sb_uuid+group+desc) */
337	__le32	bg_block_bitmap_hi;	/* Blocks bitmap block MSB */
338	__le32	bg_inode_bitmap_hi;	/* Inodes bitmap block MSB */
339	__le32	bg_inode_table_hi;	/* Inodes table block MSB */
340	__le16	bg_free_blocks_count_hi;/* Free blocks count MSB */
341	__le16	bg_free_inodes_count_hi;/* Free inodes count MSB */
342	__le16	bg_used_dirs_count_hi;	/* Directories count MSB */
343	__le16  bg_itable_unused_hi;    /* Unused inodes count MSB */
344	__le32  bg_exclude_bitmap_hi;   /* Exclude bitmap block MSB */
345	__le16  bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
346	__le16  bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
347	__u32   bg_reserved;
348};
349
350#define EXT4_BG_INODE_BITMAP_CSUM_HI_END	\
351	(offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
352	 sizeof(__le16))
353#define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END	\
354	(offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
355	 sizeof(__le16))
356
357/*
358 * Structure of a flex block group info
359 */
360
361struct flex_groups {
362	atomic64_t	free_clusters;
363	atomic_t	free_inodes;
364	atomic_t	used_dirs;
365};
366
367#define EXT4_BG_INODE_UNINIT	0x0001 /* Inode table/bitmap not in use */
368#define EXT4_BG_BLOCK_UNINIT	0x0002 /* Block bitmap not in use */
369#define EXT4_BG_INODE_ZEROED	0x0004 /* On-disk itable initialized to zero */
370
371/*
372 * Macro-instructions used to manage group descriptors
373 */
374#define EXT4_MIN_DESC_SIZE		32
375#define EXT4_MIN_DESC_SIZE_64BIT	64
376#define	EXT4_MAX_DESC_SIZE		EXT4_MIN_BLOCK_SIZE
377#define EXT4_DESC_SIZE(s)		(EXT4_SB(s)->s_desc_size)
378#ifdef __KERNEL__
379# define EXT4_BLOCKS_PER_GROUP(s)	(EXT4_SB(s)->s_blocks_per_group)
380# define EXT4_CLUSTERS_PER_GROUP(s)	(EXT4_SB(s)->s_clusters_per_group)
381# define EXT4_DESC_PER_BLOCK(s)		(EXT4_SB(s)->s_desc_per_block)
382# define EXT4_INODES_PER_GROUP(s)	(EXT4_SB(s)->s_inodes_per_group)
383# define EXT4_DESC_PER_BLOCK_BITS(s)	(EXT4_SB(s)->s_desc_per_block_bits)
384#else
385# define EXT4_BLOCKS_PER_GROUP(s)	((s)->s_blocks_per_group)
386# define EXT4_DESC_PER_BLOCK(s)		(EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
387# define EXT4_INODES_PER_GROUP(s)	((s)->s_inodes_per_group)
388#endif
389
390/*
391 * Constants relative to the data blocks
392 */
393#define	EXT4_NDIR_BLOCKS		12
394#define	EXT4_IND_BLOCK			EXT4_NDIR_BLOCKS
395#define	EXT4_DIND_BLOCK			(EXT4_IND_BLOCK + 1)
396#define	EXT4_TIND_BLOCK			(EXT4_DIND_BLOCK + 1)
397#define	EXT4_N_BLOCKS			(EXT4_TIND_BLOCK + 1)
398
399/*
400 * Inode flags
401 */
402#define	EXT4_SECRM_FL			0x00000001 /* Secure deletion */
403#define	EXT4_UNRM_FL			0x00000002 /* Undelete */
404#define	EXT4_COMPR_FL			0x00000004 /* Compress file */
405#define EXT4_SYNC_FL			0x00000008 /* Synchronous updates */
406#define EXT4_IMMUTABLE_FL		0x00000010 /* Immutable file */
407#define EXT4_APPEND_FL			0x00000020 /* writes to file may only append */
408#define EXT4_NODUMP_FL			0x00000040 /* do not dump file */
409#define EXT4_NOATIME_FL			0x00000080 /* do not update atime */
410/* Reserved for compression usage... */
411#define EXT4_DIRTY_FL			0x00000100
412#define EXT4_COMPRBLK_FL		0x00000200 /* One or more compressed clusters */
413#define EXT4_NOCOMPR_FL			0x00000400 /* Don't compress */
414	/* nb: was previously EXT2_ECOMPR_FL */
415#define EXT4_ENCRYPT_FL			0x00000800 /* encrypted file */
416/* End compression flags --- maybe not all used */
417#define EXT4_INDEX_FL			0x00001000 /* hash-indexed directory */
418#define EXT4_IMAGIC_FL			0x00002000 /* AFS directory */
419#define EXT4_JOURNAL_DATA_FL		0x00004000 /* file data should be journaled */
420#define EXT4_NOTAIL_FL			0x00008000 /* file tail should not be merged */
421#define EXT4_DIRSYNC_FL			0x00010000 /* dirsync behaviour (directories only) */
422#define EXT4_TOPDIR_FL			0x00020000 /* Top of directory hierarchies*/
423#define EXT4_HUGE_FILE_FL               0x00040000 /* Set to each huge file */
424#define EXT4_EXTENTS_FL			0x00080000 /* Inode uses extents */
425#define EXT4_VERITY_FL			0x00100000 /* Verity protected inode */
426#define EXT4_EA_INODE_FL	        0x00200000 /* Inode used for large EA */
427/* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */
428
429#define EXT4_DAX_FL			0x02000000 /* Inode is DAX */
430
431#define EXT4_INLINE_DATA_FL		0x10000000 /* Inode has inline data. */
432#define EXT4_PROJINHERIT_FL		0x20000000 /* Create with parents projid */
433#define EXT4_CASEFOLD_FL		0x40000000 /* Casefolded directory */
434#define EXT4_RESERVED_FL		0x80000000 /* reserved for ext4 lib */
435
436/* User modifiable flags */
437#define EXT4_FL_USER_MODIFIABLE		(EXT4_SECRM_FL | \
438					 EXT4_UNRM_FL | \
439					 EXT4_COMPR_FL | \
440					 EXT4_SYNC_FL | \
441					 EXT4_IMMUTABLE_FL | \
442					 EXT4_APPEND_FL | \
443					 EXT4_NODUMP_FL | \
444					 EXT4_NOATIME_FL | \
445					 EXT4_JOURNAL_DATA_FL | \
446					 EXT4_NOTAIL_FL | \
447					 EXT4_DIRSYNC_FL | \
448					 EXT4_TOPDIR_FL | \
449					 EXT4_EXTENTS_FL | \
450					 0x00400000 /* EXT4_EOFBLOCKS_FL */ | \
451					 EXT4_DAX_FL | \
452					 EXT4_PROJINHERIT_FL | \
453					 EXT4_CASEFOLD_FL)
454
455/* User visible flags */
456#define EXT4_FL_USER_VISIBLE		(EXT4_FL_USER_MODIFIABLE | \
457					 EXT4_DIRTY_FL | \
458					 EXT4_COMPRBLK_FL | \
459					 EXT4_NOCOMPR_FL | \
460					 EXT4_ENCRYPT_FL | \
461					 EXT4_INDEX_FL | \
462					 EXT4_VERITY_FL | \
463					 EXT4_INLINE_DATA_FL)
464
465/* Flags we can manipulate with through FS_IOC_FSSETXATTR */
466#define EXT4_FL_XFLAG_VISIBLE		(EXT4_SYNC_FL | \
467					 EXT4_IMMUTABLE_FL | \
468					 EXT4_APPEND_FL | \
469					 EXT4_NODUMP_FL | \
470					 EXT4_NOATIME_FL | \
471					 EXT4_PROJINHERIT_FL | \
472					 EXT4_DAX_FL)
473
474/* Flags that should be inherited by new inodes from their parent. */
475#define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
476			   EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
477			   EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
478			   EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
479			   EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL |\
480			   EXT4_DAX_FL)
481
482/* Flags that are appropriate for regular files (all but dir-specific ones). */
483#define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\
484			   EXT4_PROJINHERIT_FL))
485
486/* Flags that are appropriate for non-directories/regular files. */
487#define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
488
489/* The only flags that should be swapped */
490#define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL)
491
492/* Flags which are mutually exclusive to DAX */
493#define EXT4_DAX_MUT_EXCL (EXT4_VERITY_FL | EXT4_ENCRYPT_FL |\
494			   EXT4_JOURNAL_DATA_FL | EXT4_INLINE_DATA_FL)
495
496/* Mask out flags that are inappropriate for the given type of inode. */
497static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
498{
499	if (S_ISDIR(mode))
500		return flags;
501	else if (S_ISREG(mode))
502		return flags & EXT4_REG_FLMASK;
503	else
504		return flags & EXT4_OTHER_FLMASK;
505}
506
507/*
508 * Inode flags used for atomic set/get
509 */
510enum {
511	EXT4_INODE_SECRM	= 0,	/* Secure deletion */
512	EXT4_INODE_UNRM		= 1,	/* Undelete */
513	EXT4_INODE_COMPR	= 2,	/* Compress file */
514	EXT4_INODE_SYNC		= 3,	/* Synchronous updates */
515	EXT4_INODE_IMMUTABLE	= 4,	/* Immutable file */
516	EXT4_INODE_APPEND	= 5,	/* writes to file may only append */
517	EXT4_INODE_NODUMP	= 6,	/* do not dump file */
518	EXT4_INODE_NOATIME	= 7,	/* do not update atime */
519/* Reserved for compression usage... */
520	EXT4_INODE_DIRTY	= 8,
521	EXT4_INODE_COMPRBLK	= 9,	/* One or more compressed clusters */
522	EXT4_INODE_NOCOMPR	= 10,	/* Don't compress */
523	EXT4_INODE_ENCRYPT	= 11,	/* Encrypted file */
524/* End compression flags --- maybe not all used */
525	EXT4_INODE_INDEX	= 12,	/* hash-indexed directory */
526	EXT4_INODE_IMAGIC	= 13,	/* AFS directory */
527	EXT4_INODE_JOURNAL_DATA	= 14,	/* file data should be journaled */
528	EXT4_INODE_NOTAIL	= 15,	/* file tail should not be merged */
529	EXT4_INODE_DIRSYNC	= 16,	/* dirsync behaviour (directories only) */
530	EXT4_INODE_TOPDIR	= 17,	/* Top of directory hierarchies*/
531	EXT4_INODE_HUGE_FILE	= 18,	/* Set to each huge file */
532	EXT4_INODE_EXTENTS	= 19,	/* Inode uses extents */
533	EXT4_INODE_VERITY	= 20,	/* Verity protected inode */
534	EXT4_INODE_EA_INODE	= 21,	/* Inode used for large EA */
535/* 22 was formerly EXT4_INODE_EOFBLOCKS */
536	EXT4_INODE_DAX		= 25,	/* Inode is DAX */
537	EXT4_INODE_INLINE_DATA	= 28,	/* Data in inode. */
538	EXT4_INODE_PROJINHERIT	= 29,	/* Create with parents projid */
539	EXT4_INODE_CASEFOLD	= 30,	/* Casefolded directory */
540	EXT4_INODE_RESERVED	= 31,	/* reserved for ext4 lib */
541};
542
543/*
544 * Since it's pretty easy to mix up bit numbers and hex values, we use a
545 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
546 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
547 * any extra space in the compiled kernel image, otherwise, the build will fail.
548 * It's important that these values are the same, since we are using
549 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
550 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
551 * values found in ext2, ext3 and ext4 filesystems, and of course the values
552 * defined in e2fsprogs.
553 *
554 * It's not paranoia if the Murphy's Law really *is* out to get you.  :-)
555 */
556#define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1U << EXT4_INODE_##FLAG))
557#define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
558
559static inline void ext4_check_flag_values(void)
560{
561	CHECK_FLAG_VALUE(SECRM);
562	CHECK_FLAG_VALUE(UNRM);
563	CHECK_FLAG_VALUE(COMPR);
564	CHECK_FLAG_VALUE(SYNC);
565	CHECK_FLAG_VALUE(IMMUTABLE);
566	CHECK_FLAG_VALUE(APPEND);
567	CHECK_FLAG_VALUE(NODUMP);
568	CHECK_FLAG_VALUE(NOATIME);
569	CHECK_FLAG_VALUE(DIRTY);
570	CHECK_FLAG_VALUE(COMPRBLK);
571	CHECK_FLAG_VALUE(NOCOMPR);
572	CHECK_FLAG_VALUE(ENCRYPT);
573	CHECK_FLAG_VALUE(INDEX);
574	CHECK_FLAG_VALUE(IMAGIC);
575	CHECK_FLAG_VALUE(JOURNAL_DATA);
576	CHECK_FLAG_VALUE(NOTAIL);
577	CHECK_FLAG_VALUE(DIRSYNC);
578	CHECK_FLAG_VALUE(TOPDIR);
579	CHECK_FLAG_VALUE(HUGE_FILE);
580	CHECK_FLAG_VALUE(EXTENTS);
581	CHECK_FLAG_VALUE(VERITY);
582	CHECK_FLAG_VALUE(EA_INODE);
583	CHECK_FLAG_VALUE(INLINE_DATA);
584	CHECK_FLAG_VALUE(PROJINHERIT);
585	CHECK_FLAG_VALUE(CASEFOLD);
586	CHECK_FLAG_VALUE(RESERVED);
587}
588
589/* Used to pass group descriptor data when online resize is done */
590struct ext4_new_group_input {
591	__u32 group;		/* Group number for this data */
592	__u64 block_bitmap;	/* Absolute block number of block bitmap */
593	__u64 inode_bitmap;	/* Absolute block number of inode bitmap */
594	__u64 inode_table;	/* Absolute block number of inode table start */
595	__u32 blocks_count;	/* Total number of blocks in this group */
596	__u16 reserved_blocks;	/* Number of reserved blocks in this group */
597	__u16 unused;
598};
599
600#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
601struct compat_ext4_new_group_input {
602	u32 group;
603	compat_u64 block_bitmap;
604	compat_u64 inode_bitmap;
605	compat_u64 inode_table;
606	u32 blocks_count;
607	u16 reserved_blocks;
608	u16 unused;
609};
610#endif
611
612/* The struct ext4_new_group_input in kernel space, with free_blocks_count */
613struct ext4_new_group_data {
614	__u32 group;
615	__u64 block_bitmap;
616	__u64 inode_bitmap;
617	__u64 inode_table;
618	__u32 blocks_count;
619	__u16 reserved_blocks;
620	__u16 mdata_blocks;
621	__u32 free_clusters_count;
622};
623
624/* Indexes used to index group tables in ext4_new_group_data */
625enum {
626	BLOCK_BITMAP = 0,	/* block bitmap */
627	INODE_BITMAP,		/* inode bitmap */
628	INODE_TABLE,		/* inode tables */
629	GROUP_TABLE_COUNT,
630};
631
632/*
633 * Flags used by ext4_map_blocks()
634 */
635	/* Allocate any needed blocks and/or convert an unwritten
636	   extent to be an initialized ext4 */
637#define EXT4_GET_BLOCKS_CREATE			0x0001
638	/* Request the creation of an unwritten extent */
639#define EXT4_GET_BLOCKS_UNWRIT_EXT		0x0002
640#define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT	(EXT4_GET_BLOCKS_UNWRIT_EXT|\
641						 EXT4_GET_BLOCKS_CREATE)
642	/* Caller is from the delayed allocation writeout path
643	 * finally doing the actual allocation of delayed blocks */
644#define EXT4_GET_BLOCKS_DELALLOC_RESERVE	0x0004
645	/* caller is from the direct IO path, request to creation of an
646	unwritten extents if not allocated, split the unwritten
647	extent if blocks has been preallocated already*/
648#define EXT4_GET_BLOCKS_PRE_IO			0x0008
649#define EXT4_GET_BLOCKS_CONVERT			0x0010
650#define EXT4_GET_BLOCKS_IO_CREATE_EXT		(EXT4_GET_BLOCKS_PRE_IO|\
651					 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
652	/* Convert extent to initialized after IO complete */
653#define EXT4_GET_BLOCKS_IO_CONVERT_EXT		(EXT4_GET_BLOCKS_CONVERT|\
654					 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
655	/* Eventual metadata allocation (due to growing extent tree)
656	 * should not fail, so try to use reserved blocks for that.*/
657#define EXT4_GET_BLOCKS_METADATA_NOFAIL		0x0020
658	/* Don't normalize allocation size (used for fallocate) */
659#define EXT4_GET_BLOCKS_NO_NORMALIZE		0x0040
660	/* Convert written extents to unwritten */
661#define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN	0x0100
662	/* Write zeros to newly created written extents */
663#define EXT4_GET_BLOCKS_ZERO			0x0200
664#define EXT4_GET_BLOCKS_CREATE_ZERO		(EXT4_GET_BLOCKS_CREATE |\
665					EXT4_GET_BLOCKS_ZERO)
666	/* Caller will submit data before dropping transaction handle. This
667	 * allows jbd2 to avoid submitting data before commit. */
668#define EXT4_GET_BLOCKS_IO_SUBMIT		0x0400
669
670/*
671 * The bit position of these flags must not overlap with any of the
672 * EXT4_GET_BLOCKS_*.  They are used by ext4_find_extent(),
673 * read_extent_tree_block(), ext4_split_extent_at(),
674 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
675 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
676 * caching the extents when reading from the extent tree while a
677 * truncate or punch hole operation is in progress.
678 */
679#define EXT4_EX_NOCACHE				0x40000000
680#define EXT4_EX_FORCE_CACHE			0x20000000
681#define EXT4_EX_NOFAIL				0x10000000
682
683/*
684 * Flags used by ext4_free_blocks
685 */
686#define EXT4_FREE_BLOCKS_METADATA		0x0001
687#define EXT4_FREE_BLOCKS_FORGET			0x0002
688#define EXT4_FREE_BLOCKS_VALIDATED		0x0004
689#define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE		0x0008
690#define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER	0x0010
691#define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER	0x0020
692#define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER      0x0040
693
694/*
695 * ioctl commands
696 */
697#define	EXT4_IOC_GETVERSION		_IOR('f', 3, long)
698#define	EXT4_IOC_SETVERSION		_IOW('f', 4, long)
699#define	EXT4_IOC_GETVERSION_OLD		FS_IOC_GETVERSION
700#define	EXT4_IOC_SETVERSION_OLD		FS_IOC_SETVERSION
701#define EXT4_IOC_GETRSVSZ		_IOR('f', 5, long)
702#define EXT4_IOC_SETRSVSZ		_IOW('f', 6, long)
703#define EXT4_IOC_GROUP_EXTEND		_IOW('f', 7, unsigned long)
704#define EXT4_IOC_GROUP_ADD		_IOW('f', 8, struct ext4_new_group_input)
705#define EXT4_IOC_MIGRATE		_IO('f', 9)
706 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
707 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
708#define EXT4_IOC_ALLOC_DA_BLKS		_IO('f', 12)
709#define EXT4_IOC_MOVE_EXT		_IOWR('f', 15, struct move_extent)
710#define EXT4_IOC_RESIZE_FS		_IOW('f', 16, __u64)
711#define EXT4_IOC_SWAP_BOOT		_IO('f', 17)
712#define EXT4_IOC_PRECACHE_EXTENTS	_IO('f', 18)
713/* ioctl codes 19--39 are reserved for fscrypt */
714#define EXT4_IOC_CLEAR_ES_CACHE		_IO('f', 40)
715#define EXT4_IOC_GETSTATE		_IOW('f', 41, __u32)
716#define EXT4_IOC_GET_ES_CACHE		_IOWR('f', 42, struct fiemap)
717
718#define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32)
719
720/*
721 * Flags for going down operation
722 */
723#define EXT4_GOING_FLAGS_DEFAULT		0x0	/* going down */
724#define EXT4_GOING_FLAGS_LOGFLUSH		0x1	/* flush log but not data */
725#define EXT4_GOING_FLAGS_NOLOGFLUSH		0x2	/* don't flush log nor data */
726
727/*
728 * Flags returned by EXT4_IOC_GETSTATE
729 *
730 * We only expose to userspace a subset of the state flags in
731 * i_state_flags
732 */
733#define EXT4_STATE_FLAG_EXT_PRECACHED	0x00000001
734#define EXT4_STATE_FLAG_NEW		0x00000002
735#define EXT4_STATE_FLAG_NEWENTRY	0x00000004
736#define EXT4_STATE_FLAG_DA_ALLOC_CLOSE	0x00000008
737
738#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
739/*
740 * ioctl commands in 32 bit emulation
741 */
742#define EXT4_IOC32_GETVERSION		_IOR('f', 3, int)
743#define EXT4_IOC32_SETVERSION		_IOW('f', 4, int)
744#define EXT4_IOC32_GETRSVSZ		_IOR('f', 5, int)
745#define EXT4_IOC32_SETRSVSZ		_IOW('f', 6, int)
746#define EXT4_IOC32_GROUP_EXTEND		_IOW('f', 7, unsigned int)
747#define EXT4_IOC32_GROUP_ADD		_IOW('f', 8, struct compat_ext4_new_group_input)
748#define EXT4_IOC32_GETVERSION_OLD	FS_IOC32_GETVERSION
749#define EXT4_IOC32_SETVERSION_OLD	FS_IOC32_SETVERSION
750#endif
751
752/*
753 * Returned by EXT4_IOC_GET_ES_CACHE as an additional possible flag.
754 * It indicates that the entry in extent status cache is for a hole.
755 */
756#define EXT4_FIEMAP_EXTENT_HOLE		0x08000000
757
758/* Max physical block we can address w/o extents */
759#define EXT4_MAX_BLOCK_FILE_PHYS	0xFFFFFFFF
760
761/* Max logical block we can support */
762#define EXT4_MAX_LOGICAL_BLOCK		0xFFFFFFFE
763
764/*
765 * Structure of an inode on the disk
766 */
767struct ext4_inode {
768	__le16	i_mode;		/* File mode */
769	__le16	i_uid;		/* Low 16 bits of Owner Uid */
770	__le32	i_size_lo;	/* Size in bytes */
771	__le32	i_atime;	/* Access time */
772	__le32	i_ctime;	/* Inode Change time */
773	__le32	i_mtime;	/* Modification time */
774	__le32	i_dtime;	/* Deletion Time */
775	__le16	i_gid;		/* Low 16 bits of Group Id */
776	__le16	i_links_count;	/* Links count */
777	__le32	i_blocks_lo;	/* Blocks count */
778	__le32	i_flags;	/* File flags */
779	union {
780		struct {
781			__le32  l_i_version;
782		} linux1;
783		struct {
784			__u32  h_i_translator;
785		} hurd1;
786		struct {
787			__u32  m_i_reserved1;
788		} masix1;
789	} osd1;				/* OS dependent 1 */
790	__le32	i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
791	__le32	i_generation;	/* File version (for NFS) */
792	__le32	i_file_acl_lo;	/* File ACL */
793	__le32	i_size_high;
794	__le32	i_obso_faddr;	/* Obsoleted fragment address */
795	union {
796		struct {
797			__le16	l_i_blocks_high; /* were l_i_reserved1 */
798			__le16	l_i_file_acl_high;
799			__le16	l_i_uid_high;	/* these 2 fields */
800			__le16	l_i_gid_high;	/* were reserved2[0] */
801			__le16	l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
802			__le16	l_i_reserved;
803		} linux2;
804		struct {
805			__le16	h_i_reserved1;	/* Obsoleted fragment number/size which are removed in ext4 */
806			__u16	h_i_mode_high;
807			__u16	h_i_uid_high;
808			__u16	h_i_gid_high;
809			__u32	h_i_author;
810		} hurd2;
811		struct {
812			__le16	h_i_reserved1;	/* Obsoleted fragment number/size which are removed in ext4 */
813			__le16	m_i_file_acl_high;
814			__u32	m_i_reserved2[2];
815		} masix2;
816	} osd2;				/* OS dependent 2 */
817	__le16	i_extra_isize;
818	__le16	i_checksum_hi;	/* crc32c(uuid+inum+inode) BE */
819	__le32  i_ctime_extra;  /* extra Change time      (nsec << 2 | epoch) */
820	__le32  i_mtime_extra;  /* extra Modification time(nsec << 2 | epoch) */
821	__le32  i_atime_extra;  /* extra Access time      (nsec << 2 | epoch) */
822	__le32  i_crtime;       /* File Creation time */
823	__le32  i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
824	__le32  i_version_hi;	/* high 32 bits for 64-bit version */
825	__le32	i_projid;	/* Project ID */
826};
827
828struct move_extent {
829	__u32 reserved;		/* should be zero */
830	__u32 donor_fd;		/* donor file descriptor */
831	__u64 orig_start;	/* logical start offset in block for orig */
832	__u64 donor_start;	/* logical start offset in block for donor */
833	__u64 len;		/* block length to be moved */
834	__u64 moved_len;	/* moved block length */
835};
836
837#define EXT4_EPOCH_BITS 2
838#define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
839#define EXT4_NSEC_MASK  (~0UL << EXT4_EPOCH_BITS)
840
841/*
842 * Extended fields will fit into an inode if the filesystem was formatted
843 * with large inodes (-I 256 or larger) and there are not currently any EAs
844 * consuming all of the available space. For new inodes we always reserve
845 * enough space for the kernel's known extended fields, but for inodes
846 * created with an old kernel this might not have been the case. None of
847 * the extended inode fields is critical for correct filesystem operation.
848 * This macro checks if a certain field fits in the inode. Note that
849 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
850 */
851#define EXT4_FITS_IN_INODE(ext4_inode, einode, field)	\
852	((offsetof(typeof(*ext4_inode), field) +	\
853	  sizeof((ext4_inode)->field))			\
854	<= (EXT4_GOOD_OLD_INODE_SIZE +			\
855	    (einode)->i_extra_isize))			\
856
857/*
858 * We use an encoding that preserves the times for extra epoch "00":
859 *
860 * extra  msb of                         adjust for signed
861 * epoch  32-bit                         32-bit tv_sec to
862 * bits   time    decoded 64-bit tv_sec  64-bit tv_sec      valid time range
863 * 0 0    1    -0x80000000..-0x00000001  0x000000000 1901-12-13..1969-12-31
864 * 0 0    0    0x000000000..0x07fffffff  0x000000000 1970-01-01..2038-01-19
865 * 0 1    1    0x080000000..0x0ffffffff  0x100000000 2038-01-19..2106-02-07
866 * 0 1    0    0x100000000..0x17fffffff  0x100000000 2106-02-07..2174-02-25
867 * 1 0    1    0x180000000..0x1ffffffff  0x200000000 2174-02-25..2242-03-16
868 * 1 0    0    0x200000000..0x27fffffff  0x200000000 2242-03-16..2310-04-04
869 * 1 1    1    0x280000000..0x2ffffffff  0x300000000 2310-04-04..2378-04-22
870 * 1 1    0    0x300000000..0x37fffffff  0x300000000 2378-04-22..2446-05-10
871 *
872 * Note that previous versions of the kernel on 64-bit systems would
873 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
874 * 1970.  e2fsck will correct this, assuming that it is run on the
875 * affected filesystem before 2242.
876 */
877
878static inline __le32 ext4_encode_extra_time(struct timespec64 *time)
879{
880	u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK;
881	return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
882}
883
884static inline void ext4_decode_extra_time(struct timespec64 *time,
885					  __le32 extra)
886{
887	if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK)))
888		time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
889	time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
890}
891
892#define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode)				\
893do {										\
894	if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     {\
895		(raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec);	\
896		(raw_inode)->xtime ## _extra =					\
897				ext4_encode_extra_time(&(inode)->xtime);	\
898		}								\
899	else	\
900		(raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (inode)->xtime.tv_sec, S32_MIN, S32_MAX));	\
901} while (0)
902
903#define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode)			       \
904do {									       \
905	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))		       \
906		(raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec);      \
907	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))	       \
908		(raw_inode)->xtime ## _extra =				       \
909				ext4_encode_extra_time(&(einode)->xtime);      \
910} while (0)
911
912#define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode)				\
913do {										\
914	(inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime);	\
915	if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {	\
916		ext4_decode_extra_time(&(inode)->xtime,				\
917				       raw_inode->xtime ## _extra);		\
918		}								\
919	else									\
920		(inode)->xtime.tv_nsec = 0;					\
921} while (0)
922
923
924#define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode)			       \
925do {									       \
926	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))		       \
927		(einode)->xtime.tv_sec = 				       \
928			(signed)le32_to_cpu((raw_inode)->xtime);	       \
929	else								       \
930		(einode)->xtime.tv_sec = 0;				       \
931	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))	       \
932		ext4_decode_extra_time(&(einode)->xtime,		       \
933				       raw_inode->xtime ## _extra);	       \
934	else								       \
935		(einode)->xtime.tv_nsec = 0;				       \
936} while (0)
937
938#define i_disk_version osd1.linux1.l_i_version
939
940#if defined(__KERNEL__) || defined(__linux__)
941#define i_reserved1	osd1.linux1.l_i_reserved1
942#define i_file_acl_high	osd2.linux2.l_i_file_acl_high
943#define i_blocks_high	osd2.linux2.l_i_blocks_high
944#define i_uid_low	i_uid
945#define i_gid_low	i_gid
946#define i_uid_high	osd2.linux2.l_i_uid_high
947#define i_gid_high	osd2.linux2.l_i_gid_high
948#define i_checksum_lo	osd2.linux2.l_i_checksum_lo
949
950#elif defined(__GNU__)
951
952#define i_translator	osd1.hurd1.h_i_translator
953#define i_uid_high	osd2.hurd2.h_i_uid_high
954#define i_gid_high	osd2.hurd2.h_i_gid_high
955#define i_author	osd2.hurd2.h_i_author
956
957#elif defined(__masix__)
958
959#define i_reserved1	osd1.masix1.m_i_reserved1
960#define i_file_acl_high	osd2.masix2.m_i_file_acl_high
961#define i_reserved2	osd2.masix2.m_i_reserved2
962
963#endif /* defined(__KERNEL__) || defined(__linux__) */
964
965#include "extents_status.h"
966#include "fast_commit.h"
967
968/*
969 * Lock subclasses for i_data_sem in the ext4_inode_info structure.
970 *
971 * These are needed to avoid lockdep false positives when we need to
972 * allocate blocks to the quota inode during ext4_map_blocks(), while
973 * holding i_data_sem for a normal (non-quota) inode.  Since we don't
974 * do quota tracking for the quota inode, this avoids deadlock (as
975 * well as infinite recursion, since it isn't turtles all the way
976 * down...)
977 *
978 *  I_DATA_SEM_NORMAL - Used for most inodes
979 *  I_DATA_SEM_OTHER  - Used by move_inode.c for the second normal inode
980 *			  where the second inode has larger inode number
981 *			  than the first
982 *  I_DATA_SEM_QUOTA  - Used for quota inodes only
983 *  I_DATA_SEM_EA     - Used for ea_inodes only
984 */
985enum {
986	I_DATA_SEM_NORMAL = 0,
987	I_DATA_SEM_OTHER,
988	I_DATA_SEM_QUOTA,
989	I_DATA_SEM_EA
990};
991
992
993/*
994 * fourth extended file system inode data in memory
995 */
996struct ext4_inode_info {
997	__le32	i_data[15];	/* unconverted */
998	__u32	i_dtime;
999	ext4_fsblk_t	i_file_acl;
1000
1001	/*
1002	 * i_block_group is the number of the block group which contains
1003	 * this file's inode.  Constant across the lifetime of the inode,
1004	 * it is used for making block allocation decisions - we try to
1005	 * place a file's data blocks near its inode block, and new inodes
1006	 * near to their parent directory's inode.
1007	 */
1008	ext4_group_t	i_block_group;
1009	ext4_lblk_t	i_dir_start_lookup;
1010#if (BITS_PER_LONG < 64)
1011	unsigned long	i_state_flags;		/* Dynamic state flags */
1012#endif
1013	unsigned long	i_flags;
1014
1015	/*
1016	 * Extended attributes can be read independently of the main file
1017	 * data. Taking i_mutex even when reading would cause contention
1018	 * between readers of EAs and writers of regular file data, so
1019	 * instead we synchronize on xattr_sem when reading or changing
1020	 * EAs.
1021	 */
1022	struct rw_semaphore xattr_sem;
1023
1024	struct list_head i_orphan;	/* unlinked but open inodes */
1025
1026	/* Fast commit related info */
1027
1028	struct list_head i_fc_list;	/*
1029					 * inodes that need fast commit
1030					 * protected by sbi->s_fc_lock.
1031					 */
1032
1033	/* Start of lblk range that needs to be committed in this fast commit */
1034	ext4_lblk_t i_fc_lblk_start;
1035
1036	/* End of lblk range that needs to be committed in this fast commit */
1037	ext4_lblk_t i_fc_lblk_len;
1038
1039	/* Number of ongoing updates on this inode */
1040	atomic_t  i_fc_updates;
1041
1042	/* Fast commit wait queue for this inode */
1043	wait_queue_head_t i_fc_wait;
1044
1045	/* Protect concurrent accesses on i_fc_lblk_start, i_fc_lblk_len */
1046	struct mutex i_fc_lock;
1047
1048	/*
1049	 * i_disksize keeps track of what the inode size is ON DISK, not
1050	 * in memory.  During truncate, i_size is set to the new size by
1051	 * the VFS prior to calling ext4_truncate(), but the filesystem won't
1052	 * set i_disksize to 0 until the truncate is actually under way.
1053	 *
1054	 * The intent is that i_disksize always represents the blocks which
1055	 * are used by this file.  This allows recovery to restart truncate
1056	 * on orphans if we crash during truncate.  We actually write i_disksize
1057	 * into the on-disk inode when writing inodes out, instead of i_size.
1058	 *
1059	 * The only time when i_disksize and i_size may be different is when
1060	 * a truncate is in progress.  The only things which change i_disksize
1061	 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
1062	 */
1063	loff_t	i_disksize;
1064
1065	/*
1066	 * i_data_sem is for serialising ext4_truncate() against
1067	 * ext4_getblock().  In the 2.4 ext2 design, great chunks of inode's
1068	 * data tree are chopped off during truncate. We can't do that in
1069	 * ext4 because whenever we perform intermediate commits during
1070	 * truncate, the inode and all the metadata blocks *must* be in a
1071	 * consistent state which allows truncation of the orphans to restart
1072	 * during recovery.  Hence we must fix the get_block-vs-truncate race
1073	 * by other means, so we have i_data_sem.
1074	 */
1075	struct rw_semaphore i_data_sem;
1076	/*
1077	 * i_mmap_sem is for serializing page faults with truncate / punch hole
1078	 * operations. We have to make sure that new page cannot be faulted in
1079	 * a section of the inode that is being punched. We cannot easily use
1080	 * i_data_sem for this since we need protection for the whole punch
1081	 * operation and i_data_sem ranks below transaction start so we have
1082	 * to occasionally drop it.
1083	 */
1084	struct rw_semaphore i_mmap_sem;
1085	struct inode vfs_inode;
1086	struct jbd2_inode *jinode;
1087
1088	spinlock_t i_raw_lock;	/* protects updates to the raw inode */
1089
1090	/*
1091	 * File creation time. Its function is same as that of
1092	 * struct timespec64 i_{a,c,m}time in the generic inode.
1093	 */
1094	struct timespec64 i_crtime;
1095
1096	/* mballoc */
1097	atomic_t i_prealloc_active;
1098	struct list_head i_prealloc_list;
1099	spinlock_t i_prealloc_lock;
1100
1101	/* extents status tree */
1102	struct ext4_es_tree i_es_tree;
1103	rwlock_t i_es_lock;
1104	struct list_head i_es_list;
1105	unsigned int i_es_all_nr;	/* protected by i_es_lock */
1106	unsigned int i_es_shk_nr;	/* protected by i_es_lock */
1107	ext4_lblk_t i_es_shrink_lblk;	/* Offset where we start searching for
1108					   extents to shrink. Protected by
1109					   i_es_lock  */
1110
1111	/* ialloc */
1112	ext4_group_t	i_last_alloc_group;
1113
1114	/* allocation reservation info for delalloc */
1115	/* In case of bigalloc, this refer to clusters rather than blocks */
1116	unsigned int i_reserved_data_blocks;
1117
1118	/* pending cluster reservations for bigalloc file systems */
1119	struct ext4_pending_tree i_pending_tree;
1120
1121	/* on-disk additional length */
1122	__u16 i_extra_isize;
1123
1124	/* Indicate the inline data space. */
1125	u16 i_inline_off;
1126	u16 i_inline_size;
1127
1128#ifdef CONFIG_QUOTA
1129	/* quota space reservation, managed internally by quota code */
1130	qsize_t i_reserved_quota;
1131#endif
1132
1133	/* Lock protecting lists below */
1134	spinlock_t i_completed_io_lock;
1135	/*
1136	 * Completed IOs that need unwritten extents handling and have
1137	 * transaction reserved
1138	 */
1139	struct list_head i_rsv_conversion_list;
1140	struct work_struct i_rsv_conversion_work;
1141	atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1142
1143	spinlock_t i_block_reservation_lock;
1144
1145	/*
1146	 * Transactions that contain inode's metadata needed to complete
1147	 * fsync and fdatasync, respectively.
1148	 */
1149	tid_t i_sync_tid;
1150	tid_t i_datasync_tid;
1151
1152#ifdef CONFIG_QUOTA
1153	struct dquot *i_dquot[MAXQUOTAS];
1154#endif
1155
1156	/* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1157	__u32 i_csum_seed;
1158
1159	kprojid_t i_projid;
1160};
1161
1162/*
1163 * File system states
1164 */
1165#define	EXT4_VALID_FS			0x0001	/* Unmounted cleanly */
1166#define	EXT4_ERROR_FS			0x0002	/* Errors detected */
1167#define	EXT4_ORPHAN_FS			0x0004	/* Orphans being recovered */
1168#define EXT4_FC_REPLAY			0x0020	/* Fast commit replay ongoing */
1169
1170/*
1171 * Misc. filesystem flags
1172 */
1173#define EXT2_FLAGS_SIGNED_HASH		0x0001  /* Signed dirhash in use */
1174#define EXT2_FLAGS_UNSIGNED_HASH	0x0002  /* Unsigned dirhash in use */
1175#define EXT2_FLAGS_TEST_FILESYS		0x0004	/* to test development code */
1176
1177/*
1178 * Mount flags set via mount options or defaults
1179 */
1180#define EXT4_MOUNT_NO_MBCACHE		0x00001 /* Do not use mbcache */
1181#define EXT4_MOUNT_GRPID		0x00004	/* Create files with directory's group */
1182#define EXT4_MOUNT_DEBUG		0x00008	/* Some debugging messages */
1183#define EXT4_MOUNT_ERRORS_CONT		0x00010	/* Continue on errors */
1184#define EXT4_MOUNT_ERRORS_RO		0x00020	/* Remount fs ro on errors */
1185#define EXT4_MOUNT_ERRORS_PANIC		0x00040	/* Panic on errors */
1186#define EXT4_MOUNT_ERRORS_MASK		0x00070
1187#define EXT4_MOUNT_MINIX_DF		0x00080	/* Mimics the Minix statfs */
1188#define EXT4_MOUNT_NOLOAD		0x00100	/* Don't use existing journal*/
1189#ifdef CONFIG_FS_DAX
1190#define EXT4_MOUNT_DAX_ALWAYS		0x00200	/* Direct Access */
1191#else
1192#define EXT4_MOUNT_DAX_ALWAYS		0
1193#endif
1194#define EXT4_MOUNT_DATA_FLAGS		0x00C00	/* Mode for data writes: */
1195#define EXT4_MOUNT_JOURNAL_DATA		0x00400	/* Write data to journal */
1196#define EXT4_MOUNT_ORDERED_DATA		0x00800	/* Flush data before commit */
1197#define EXT4_MOUNT_WRITEBACK_DATA	0x00C00	/* No data ordering */
1198#define EXT4_MOUNT_UPDATE_JOURNAL	0x01000	/* Update the journal format */
1199#define EXT4_MOUNT_NO_UID32		0x02000  /* Disable 32-bit UIDs */
1200#define EXT4_MOUNT_XATTR_USER		0x04000	/* Extended user attributes */
1201#define EXT4_MOUNT_POSIX_ACL		0x08000	/* POSIX Access Control Lists */
1202#define EXT4_MOUNT_NO_AUTO_DA_ALLOC	0x10000	/* No auto delalloc mapping */
1203#define EXT4_MOUNT_BARRIER		0x20000 /* Use block barriers */
1204#define EXT4_MOUNT_QUOTA		0x40000 /* Some quota option set */
1205#define EXT4_MOUNT_USRQUOTA		0x80000 /* "old" user quota,
1206						 * enable enforcement for hidden
1207						 * quota files */
1208#define EXT4_MOUNT_GRPQUOTA		0x100000 /* "old" group quota, enable
1209						  * enforcement for hidden quota
1210						  * files */
1211#define EXT4_MOUNT_PRJQUOTA		0x200000 /* Enable project quota
1212						  * enforcement */
1213#define EXT4_MOUNT_DIOREAD_NOLOCK	0x400000 /* Enable support for dio read nolocking */
1214#define EXT4_MOUNT_JOURNAL_CHECKSUM	0x800000 /* Journal checksums */
1215#define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT	0x1000000 /* Journal Async Commit */
1216#define EXT4_MOUNT_WARN_ON_ERROR	0x2000000 /* Trigger WARN_ON on error */
1217#define EXT4_MOUNT_PREFETCH_BLOCK_BITMAPS 0x4000000
1218#define EXT4_MOUNT_DELALLOC		0x8000000 /* Delalloc support */
1219#define EXT4_MOUNT_DATA_ERR_ABORT	0x10000000 /* Abort on file data write */
1220#define EXT4_MOUNT_BLOCK_VALIDITY	0x20000000 /* Block validity checking */
1221#define EXT4_MOUNT_DISCARD		0x40000000 /* Issue DISCARD requests */
1222#define EXT4_MOUNT_INIT_INODE_TABLE	0x80000000 /* Initialize uninitialized itables */
1223
1224/*
1225 * Mount flags set either automatically (could not be set by mount option)
1226 * based on per file system feature or property or in special cases such as
1227 * distinguishing between explicit mount option definition and default.
1228 */
1229#define EXT4_MOUNT2_EXPLICIT_DELALLOC	0x00000001 /* User explicitly
1230						      specified delalloc */
1231#define EXT4_MOUNT2_STD_GROUP_SIZE	0x00000002 /* We have standard group
1232						      size of blocksize * 8
1233						      blocks */
1234#define EXT4_MOUNT2_HURD_COMPAT		0x00000004 /* Support HURD-castrated
1235						      file systems */
1236#define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM	0x00000008 /* User explicitly
1237						specified journal checksum */
1238
1239#define EXT4_MOUNT2_JOURNAL_FAST_COMMIT	0x00000010 /* Journal fast commit */
1240#define EXT4_MOUNT2_DAX_NEVER		0x00000020 /* Do not allow Direct Access */
1241#define EXT4_MOUNT2_DAX_INODE		0x00000040 /* For printing options only */
1242
1243
1244#define clear_opt(sb, opt)		EXT4_SB(sb)->s_mount_opt &= \
1245						~EXT4_MOUNT_##opt
1246#define set_opt(sb, opt)		EXT4_SB(sb)->s_mount_opt |= \
1247						EXT4_MOUNT_##opt
1248#define test_opt(sb, opt)		(EXT4_SB(sb)->s_mount_opt & \
1249					 EXT4_MOUNT_##opt)
1250
1251#define clear_opt2(sb, opt)		EXT4_SB(sb)->s_mount_opt2 &= \
1252						~EXT4_MOUNT2_##opt
1253#define set_opt2(sb, opt)		EXT4_SB(sb)->s_mount_opt2 |= \
1254						EXT4_MOUNT2_##opt
1255#define test_opt2(sb, opt)		(EXT4_SB(sb)->s_mount_opt2 & \
1256					 EXT4_MOUNT2_##opt)
1257
1258#define ext4_test_and_set_bit		__test_and_set_bit_le
1259#define ext4_set_bit			__set_bit_le
1260#define ext4_set_bit_atomic		ext2_set_bit_atomic
1261#define ext4_test_and_clear_bit		__test_and_clear_bit_le
1262#define ext4_clear_bit			__clear_bit_le
1263#define ext4_clear_bit_atomic		ext2_clear_bit_atomic
1264#define ext4_test_bit			test_bit_le
1265#define ext4_find_next_zero_bit		find_next_zero_bit_le
1266#define ext4_find_next_bit		find_next_bit_le
1267
1268extern void ext4_set_bits(void *bm, int cur, int len);
1269
1270/*
1271 * Maximal mount counts between two filesystem checks
1272 */
1273#define EXT4_DFL_MAX_MNT_COUNT		20	/* Allow 20 mounts */
1274#define EXT4_DFL_CHECKINTERVAL		0	/* Don't use interval check */
1275
1276/*
1277 * Behaviour when detecting errors
1278 */
1279#define EXT4_ERRORS_CONTINUE		1	/* Continue execution */
1280#define EXT4_ERRORS_RO			2	/* Remount fs read-only */
1281#define EXT4_ERRORS_PANIC		3	/* Panic */
1282#define EXT4_ERRORS_DEFAULT		EXT4_ERRORS_CONTINUE
1283
1284/* Metadata checksum algorithm codes */
1285#define EXT4_CRC32C_CHKSUM		1
1286
1287/*
1288 * Structure of the super block
1289 */
1290struct ext4_super_block {
1291/*00*/	__le32	s_inodes_count;		/* Inodes count */
1292	__le32	s_blocks_count_lo;	/* Blocks count */
1293	__le32	s_r_blocks_count_lo;	/* Reserved blocks count */
1294	__le32	s_free_blocks_count_lo;	/* Free blocks count */
1295/*10*/	__le32	s_free_inodes_count;	/* Free inodes count */
1296	__le32	s_first_data_block;	/* First Data Block */
1297	__le32	s_log_block_size;	/* Block size */
1298	__le32	s_log_cluster_size;	/* Allocation cluster size */
1299/*20*/	__le32	s_blocks_per_group;	/* # Blocks per group */
1300	__le32	s_clusters_per_group;	/* # Clusters per group */
1301	__le32	s_inodes_per_group;	/* # Inodes per group */
1302	__le32	s_mtime;		/* Mount time */
1303/*30*/	__le32	s_wtime;		/* Write time */
1304	__le16	s_mnt_count;		/* Mount count */
1305	__le16	s_max_mnt_count;	/* Maximal mount count */
1306	__le16	s_magic;		/* Magic signature */
1307	__le16	s_state;		/* File system state */
1308	__le16	s_errors;		/* Behaviour when detecting errors */
1309	__le16	s_minor_rev_level;	/* minor revision level */
1310/*40*/	__le32	s_lastcheck;		/* time of last check */
1311	__le32	s_checkinterval;	/* max. time between checks */
1312	__le32	s_creator_os;		/* OS */
1313	__le32	s_rev_level;		/* Revision level */
1314/*50*/	__le16	s_def_resuid;		/* Default uid for reserved blocks */
1315	__le16	s_def_resgid;		/* Default gid for reserved blocks */
1316	/*
1317	 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1318	 *
1319	 * Note: the difference between the compatible feature set and
1320	 * the incompatible feature set is that if there is a bit set
1321	 * in the incompatible feature set that the kernel doesn't
1322	 * know about, it should refuse to mount the filesystem.
1323	 *
1324	 * e2fsck's requirements are more strict; if it doesn't know
1325	 * about a feature in either the compatible or incompatible
1326	 * feature set, it must abort and not try to meddle with
1327	 * things it doesn't understand...
1328	 */
1329	__le32	s_first_ino;		/* First non-reserved inode */
1330	__le16  s_inode_size;		/* size of inode structure */
1331	__le16	s_block_group_nr;	/* block group # of this superblock */
1332	__le32	s_feature_compat;	/* compatible feature set */
1333/*60*/	__le32	s_feature_incompat;	/* incompatible feature set */
1334	__le32	s_feature_ro_compat;	/* readonly-compatible feature set */
1335/*68*/	__u8	s_uuid[16];		/* 128-bit uuid for volume */
1336/*78*/	char	s_volume_name[16];	/* volume name */
1337/*88*/	char	s_last_mounted[64] __nonstring;	/* directory where last mounted */
1338/*C8*/	__le32	s_algorithm_usage_bitmap; /* For compression */
1339	/*
1340	 * Performance hints.  Directory preallocation should only
1341	 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1342	 */
1343	__u8	s_prealloc_blocks;	/* Nr of blocks to try to preallocate*/
1344	__u8	s_prealloc_dir_blocks;	/* Nr to preallocate for dirs */
1345	__le16	s_reserved_gdt_blocks;	/* Per group desc for online growth */
1346	/*
1347	 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1348	 */
1349/*D0*/	__u8	s_journal_uuid[16];	/* uuid of journal superblock */
1350/*E0*/	__le32	s_journal_inum;		/* inode number of journal file */
1351	__le32	s_journal_dev;		/* device number of journal file */
1352	__le32	s_last_orphan;		/* start of list of inodes to delete */
1353	__le32	s_hash_seed[4];		/* HTREE hash seed */
1354	__u8	s_def_hash_version;	/* Default hash version to use */
1355	__u8	s_jnl_backup_type;
1356	__le16  s_desc_size;		/* size of group descriptor */
1357/*100*/	__le32	s_default_mount_opts;
1358	__le32	s_first_meta_bg;	/* First metablock block group */
1359	__le32	s_mkfs_time;		/* When the filesystem was created */
1360	__le32	s_jnl_blocks[17];	/* Backup of the journal inode */
1361	/* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1362/*150*/	__le32	s_blocks_count_hi;	/* Blocks count */
1363	__le32	s_r_blocks_count_hi;	/* Reserved blocks count */
1364	__le32	s_free_blocks_count_hi;	/* Free blocks count */
1365	__le16	s_min_extra_isize;	/* All inodes have at least # bytes */
1366	__le16	s_want_extra_isize; 	/* New inodes should reserve # bytes */
1367	__le32	s_flags;		/* Miscellaneous flags */
1368	__le16  s_raid_stride;		/* RAID stride */
1369	__le16  s_mmp_update_interval;  /* # seconds to wait in MMP checking */
1370	__le64  s_mmp_block;            /* Block for multi-mount protection */
1371	__le32  s_raid_stripe_width;    /* blocks on all data disks (N*stride)*/
1372	__u8	s_log_groups_per_flex;  /* FLEX_BG group size */
1373	__u8	s_checksum_type;	/* metadata checksum algorithm used */
1374	__u8	s_encryption_level;	/* versioning level for encryption */
1375	__u8	s_reserved_pad;		/* Padding to next 32bits */
1376	__le64	s_kbytes_written;	/* nr of lifetime kilobytes written */
1377	__le32	s_snapshot_inum;	/* Inode number of active snapshot */
1378	__le32	s_snapshot_id;		/* sequential ID of active snapshot */
1379	__le64	s_snapshot_r_blocks_count; /* reserved blocks for active
1380					      snapshot's future use */
1381	__le32	s_snapshot_list;	/* inode number of the head of the
1382					   on-disk snapshot list */
1383#define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1384	__le32	s_error_count;		/* number of fs errors */
1385	__le32	s_first_error_time;	/* first time an error happened */
1386	__le32	s_first_error_ino;	/* inode involved in first error */
1387	__le64	s_first_error_block;	/* block involved of first error */
1388	__u8	s_first_error_func[32] __nonstring;	/* function where the error happened */
1389	__le32	s_first_error_line;	/* line number where error happened */
1390	__le32	s_last_error_time;	/* most recent time of an error */
1391	__le32	s_last_error_ino;	/* inode involved in last error */
1392	__le32	s_last_error_line;	/* line number where error happened */
1393	__le64	s_last_error_block;	/* block involved of last error */
1394	__u8	s_last_error_func[32] __nonstring;	/* function where the error happened */
1395#define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1396	__u8	s_mount_opts[64];
1397	__le32	s_usr_quota_inum;	/* inode for tracking user quota */
1398	__le32	s_grp_quota_inum;	/* inode for tracking group quota */
1399	__le32	s_overhead_clusters;	/* overhead blocks/clusters in fs */
1400	__le32	s_backup_bgs[2];	/* groups with sparse_super2 SBs */
1401	__u8	s_encrypt_algos[4];	/* Encryption algorithms in use  */
1402	__u8	s_encrypt_pw_salt[16];	/* Salt used for string2key algorithm */
1403	__le32	s_lpf_ino;		/* Location of the lost+found inode */
1404	__le32	s_prj_quota_inum;	/* inode for tracking project quota */
1405	__le32	s_checksum_seed;	/* crc32c(uuid) if csum_seed set */
1406	__u8	s_wtime_hi;
1407	__u8	s_mtime_hi;
1408	__u8	s_mkfs_time_hi;
1409	__u8	s_lastcheck_hi;
1410	__u8	s_first_error_time_hi;
1411	__u8	s_last_error_time_hi;
1412	__u8	s_first_error_errcode;
1413	__u8    s_last_error_errcode;
1414	__le16  s_encoding;		/* Filename charset encoding */
1415	__le16  s_encoding_flags;	/* Filename charset encoding flags */
1416	__le32	s_reserved[95];		/* Padding to the end of the block */
1417	__le32	s_checksum;		/* crc32c(superblock) */
1418};
1419
1420#define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1421
1422#ifdef __KERNEL__
1423
1424/* Number of quota types we support */
1425#define EXT4_MAXQUOTAS 3
1426
1427#define EXT4_ENC_UTF8_12_1	1
1428
1429/*
1430 * fourth extended-fs super-block data in memory
1431 */
1432struct ext4_sb_info {
1433	unsigned long s_desc_size;	/* Size of a group descriptor in bytes */
1434	unsigned long s_inodes_per_block;/* Number of inodes per block */
1435	unsigned long s_blocks_per_group;/* Number of blocks in a group */
1436	unsigned long s_clusters_per_group; /* Number of clusters in a group */
1437	unsigned long s_inodes_per_group;/* Number of inodes in a group */
1438	unsigned long s_itb_per_group;	/* Number of inode table blocks per group */
1439	unsigned long s_gdb_count;	/* Number of group descriptor blocks */
1440	unsigned long s_desc_per_block;	/* Number of group descriptors per block */
1441	ext4_group_t s_groups_count;	/* Number of groups in the fs */
1442	ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1443	unsigned long s_overhead;  /* # of fs overhead clusters */
1444	unsigned int s_cluster_ratio;	/* Number of blocks per cluster */
1445	unsigned int s_cluster_bits;	/* log2 of s_cluster_ratio */
1446	loff_t s_bitmap_maxbytes;	/* max bytes for bitmap files */
1447	struct buffer_head * s_sbh;	/* Buffer containing the super block */
1448	struct ext4_super_block *s_es;	/* Pointer to the super block in the buffer */
1449	struct buffer_head * __rcu *s_group_desc;
1450	unsigned int s_mount_opt;
1451	unsigned int s_mount_opt2;
1452	unsigned long s_mount_flags;
1453	unsigned int s_def_mount_opt;
1454	ext4_fsblk_t s_sb_block;
1455	atomic64_t s_resv_clusters;
1456	kuid_t s_resuid;
1457	kgid_t s_resgid;
1458	unsigned short s_mount_state;
1459	unsigned short s_pad;
1460	int s_addr_per_block_bits;
1461	int s_desc_per_block_bits;
1462	int s_inode_size;
1463	int s_first_ino;
1464	unsigned int s_inode_readahead_blks;
1465	unsigned int s_inode_goal;
1466	u32 s_hash_seed[4];
1467	int s_def_hash_version;
1468	int s_hash_unsigned;	/* 3 if hash should be signed, 0 if not */
1469	struct percpu_counter s_freeclusters_counter;
1470	struct percpu_counter s_freeinodes_counter;
1471	struct percpu_counter s_dirs_counter;
1472	struct percpu_counter s_dirtyclusters_counter;
1473	struct percpu_counter s_sra_exceeded_retry_limit;
1474	struct blockgroup_lock *s_blockgroup_lock;
1475	struct proc_dir_entry *s_proc;
1476	struct kobject s_kobj;
1477	struct completion s_kobj_unregister;
1478	struct super_block *s_sb;
1479	struct buffer_head *s_mmp_bh;
1480
1481	/* Journaling */
1482	struct journal_s *s_journal;
1483	struct list_head s_orphan;
1484	struct mutex s_orphan_lock;
1485	unsigned long s_ext4_flags;		/* Ext4 superblock flags */
1486	unsigned long s_commit_interval;
1487	u32 s_max_batch_time;
1488	u32 s_min_batch_time;
1489	struct block_device *s_journal_bdev;
1490#ifdef CONFIG_QUOTA
1491	/* Names of quota files with journalled quota */
1492	char __rcu *s_qf_names[EXT4_MAXQUOTAS];
1493	int s_jquota_fmt;			/* Format of quota to use */
1494#endif
1495	unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1496	struct ext4_system_blocks __rcu *s_system_blks;
1497
1498#ifdef EXTENTS_STATS
1499	/* ext4 extents stats */
1500	unsigned long s_ext_min;
1501	unsigned long s_ext_max;
1502	unsigned long s_depth_max;
1503	spinlock_t s_ext_stats_lock;
1504	unsigned long s_ext_blocks;
1505	unsigned long s_ext_extents;
1506#endif
1507
1508	/* for buddy allocator */
1509	struct ext4_group_info ** __rcu *s_group_info;
1510	struct inode *s_buddy_cache;
1511	spinlock_t s_md_lock;
1512	unsigned short *s_mb_offsets;
1513	unsigned int *s_mb_maxs;
1514	unsigned int s_group_info_size;
1515	unsigned int s_mb_free_pending;
1516	struct list_head s_freed_data_list;	/* List of blocks to be freed
1517						   after commit completed */
1518
1519	/* tunables */
1520	unsigned long s_stripe;
1521	unsigned int s_mb_stream_request;
1522	unsigned int s_mb_max_to_scan;
1523	unsigned int s_mb_min_to_scan;
1524	unsigned int s_mb_stats;
1525	unsigned int s_mb_order2_reqs;
1526	unsigned int s_mb_group_prealloc;
1527	unsigned int s_mb_max_inode_prealloc;
1528	unsigned int s_max_dir_size_kb;
1529	/* where last allocation was done - for stream allocation */
1530	unsigned long s_mb_last_group;
1531	unsigned long s_mb_last_start;
1532	unsigned int s_mb_prefetch;
1533	unsigned int s_mb_prefetch_limit;
1534
1535	/* stats for buddy allocator */
1536	atomic_t s_bal_reqs;	/* number of reqs with len > 1 */
1537	atomic_t s_bal_success;	/* we found long enough chunks */
1538	atomic_t s_bal_allocated;	/* in blocks */
1539	atomic_t s_bal_ex_scanned;	/* total extents scanned */
1540	atomic_t s_bal_groups_scanned;	/* number of groups scanned */
1541	atomic_t s_bal_goals;	/* goal hits */
1542	atomic_t s_bal_breaks;	/* too long searches */
1543	atomic_t s_bal_2orders;	/* 2^order hits */
1544	atomic64_t s_bal_cX_groups_considered[4];
1545	atomic64_t s_bal_cX_hits[4];
1546	atomic64_t s_bal_cX_failed[4];		/* cX loop didn't find blocks */
1547	atomic_t s_mb_buddies_generated;	/* number of buddies generated */
1548	atomic64_t s_mb_generation_time;
1549	atomic_t s_mb_lost_chunks;
1550	atomic_t s_mb_preallocated;
1551	atomic_t s_mb_discarded;
1552	atomic_t s_lock_busy;
1553
1554	/* locality groups */
1555	struct ext4_locality_group __percpu *s_locality_groups;
1556
1557	/* for write statistics */
1558	unsigned long s_sectors_written_start;
1559	u64 s_kbytes_written;
1560
1561	/* the size of zero-out chunk */
1562	unsigned int s_extent_max_zeroout_kb;
1563
1564	unsigned int s_log_groups_per_flex;
1565	struct flex_groups * __rcu *s_flex_groups;
1566	ext4_group_t s_flex_groups_allocated;
1567
1568	/* workqueue for reserved extent conversions (buffered io) */
1569	struct workqueue_struct *rsv_conversion_wq;
1570
1571	/* timer for periodic error stats printing */
1572	struct timer_list s_err_report;
1573
1574	/* Lazy inode table initialization info */
1575	struct ext4_li_request *s_li_request;
1576	/* Wait multiplier for lazy initialization thread */
1577	unsigned int s_li_wait_mult;
1578
1579	/* Kernel thread for multiple mount protection */
1580	struct task_struct *s_mmp_tsk;
1581
1582	/* record the last minlen when FITRIM is called. */
1583	unsigned long s_last_trim_minblks;
1584
1585	/* Reference to checksum algorithm driver via cryptoapi */
1586	struct crypto_shash *s_chksum_driver;
1587
1588	/* Precomputed FS UUID checksum for seeding other checksums */
1589	__u32 s_csum_seed;
1590
1591	/* Reclaim extents from extent status tree */
1592	struct shrinker s_es_shrinker;
1593	struct list_head s_es_list;	/* List of inodes with reclaimable extents */
1594	long s_es_nr_inode;
1595	struct ext4_es_stats s_es_stats;
1596	struct mb_cache *s_ea_block_cache;
1597	struct mb_cache *s_ea_inode_cache;
1598	spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1599
1600	/* Ratelimit ext4 messages. */
1601	struct ratelimit_state s_err_ratelimit_state;
1602	struct ratelimit_state s_warning_ratelimit_state;
1603	struct ratelimit_state s_msg_ratelimit_state;
1604	atomic_t s_warning_count;
1605	atomic_t s_msg_count;
1606
1607	/* Encryption policy for '-o test_dummy_encryption' */
1608	struct fscrypt_dummy_policy s_dummy_enc_policy;
1609
1610	/*
1611	 * Barrier between writepages ops and changing any inode's JOURNAL_DATA
1612	 * or EXTENTS flag.
1613	 */
1614	struct percpu_rw_semaphore s_writepages_rwsem;
1615	struct dax_device *s_daxdev;
1616#ifdef CONFIG_EXT4_DEBUG
1617	unsigned long s_simulate_fail;
1618#endif
1619	/* Record the errseq of the backing block device */
1620	errseq_t s_bdev_wb_err;
1621	spinlock_t s_bdev_wb_lock;
1622
1623	/* Information about errors that happened during this mount */
1624	spinlock_t s_error_lock;
1625	int s_add_error_count;
1626	int s_first_error_code;
1627	__u32 s_first_error_line;
1628	__u32 s_first_error_ino;
1629	__u64 s_first_error_block;
1630	const char *s_first_error_func;
1631	time64_t s_first_error_time;
1632	int s_last_error_code;
1633	__u32 s_last_error_line;
1634	__u32 s_last_error_ino;
1635	__u64 s_last_error_block;
1636	const char *s_last_error_func;
1637	time64_t s_last_error_time;
1638	/*
1639	 * If we are in a context where we cannot update error information in
1640	 * the on-disk superblock, we queue this work to do it.
1641	 */
1642	struct work_struct s_error_work;
1643
1644	/* Ext4 fast commit stuff */
1645	atomic_t s_fc_subtid;
1646	atomic_t s_fc_ineligible_updates;
1647	/*
1648	 * After commit starts, the main queue gets locked, and the further
1649	 * updates get added in the staging queue.
1650	 */
1651#define FC_Q_MAIN	0
1652#define FC_Q_STAGING	1
1653	struct list_head s_fc_q[2];	/* Inodes staged for fast commit
1654					 * that have data changes in them.
1655					 */
1656	struct list_head s_fc_dentry_q[2];	/* directory entry updates */
1657	unsigned int s_fc_bytes;
1658	/*
1659	 * Main fast commit lock. This lock protects accesses to the
1660	 * following fields:
1661	 * ei->i_fc_list, s_fc_dentry_q, s_fc_q, s_fc_bytes, s_fc_bh.
1662	 */
1663	spinlock_t s_fc_lock;
1664	struct buffer_head *s_fc_bh;
1665	struct ext4_fc_stats s_fc_stats;
1666	u64 s_fc_avg_commit_time;
1667#ifdef CONFIG_EXT4_DEBUG
1668	int s_fc_debug_max_replay;
1669#endif
1670	struct ext4_fc_replay_state s_fc_replay_state;
1671};
1672
1673static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1674{
1675	return sb->s_fs_info;
1676}
1677static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1678{
1679	return container_of(inode, struct ext4_inode_info, vfs_inode);
1680}
1681
1682static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1683{
1684	return ino == EXT4_ROOT_INO ||
1685		(ino >= EXT4_FIRST_INO(sb) &&
1686		 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1687}
1688
1689/*
1690 * Returns: sbi->field[index]
1691 * Used to access an array element from the following sbi fields which require
1692 * rcu protection to avoid dereferencing an invalid pointer due to reassignment
1693 * - s_group_desc
1694 * - s_group_info
1695 * - s_flex_group
1696 */
1697#define sbi_array_rcu_deref(sbi, field, index)				   \
1698({									   \
1699	typeof(*((sbi)->field)) _v;					   \
1700	rcu_read_lock();						   \
1701	_v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index];	   \
1702	rcu_read_unlock();						   \
1703	_v;								   \
1704})
1705
1706/*
1707 * run-time mount flags
1708 */
1709enum {
1710	EXT4_MF_MNTDIR_SAMPLED,
1711	EXT4_MF_FS_ABORTED,	/* Fatal error detected */
1712	EXT4_MF_FC_INELIGIBLE,	/* Fast commit ineligible */
1713	EXT4_MF_FC_COMMITTING	/* File system underoing a fast
1714				 * commit.
1715				 */
1716};
1717
1718static inline void ext4_set_mount_flag(struct super_block *sb, int bit)
1719{
1720	set_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1721}
1722
1723static inline void ext4_clear_mount_flag(struct super_block *sb, int bit)
1724{
1725	clear_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1726}
1727
1728static inline int ext4_test_mount_flag(struct super_block *sb, int bit)
1729{
1730	return test_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1731}
1732
1733
1734/*
1735 * Simulate_fail codes
1736 */
1737#define EXT4_SIM_BBITMAP_EIO	1
1738#define EXT4_SIM_BBITMAP_CRC	2
1739#define EXT4_SIM_IBITMAP_EIO	3
1740#define EXT4_SIM_IBITMAP_CRC	4
1741#define EXT4_SIM_INODE_EIO	5
1742#define EXT4_SIM_INODE_CRC	6
1743#define EXT4_SIM_DIRBLOCK_EIO	7
1744#define EXT4_SIM_DIRBLOCK_CRC	8
1745
1746static inline bool ext4_simulate_fail(struct super_block *sb,
1747				     unsigned long code)
1748{
1749#ifdef CONFIG_EXT4_DEBUG
1750	struct ext4_sb_info *sbi = EXT4_SB(sb);
1751
1752	if (unlikely(sbi->s_simulate_fail == code)) {
1753		sbi->s_simulate_fail = 0;
1754		return true;
1755	}
1756#endif
1757	return false;
1758}
1759
1760static inline void ext4_simulate_fail_bh(struct super_block *sb,
1761					 struct buffer_head *bh,
1762					 unsigned long code)
1763{
1764	if (!IS_ERR(bh) && ext4_simulate_fail(sb, code))
1765		clear_buffer_uptodate(bh);
1766}
1767
1768/*
1769 * Error number codes for s_{first,last}_error_errno
1770 *
1771 * Linux errno numbers are architecture specific, so we need to translate
1772 * them into something which is architecture independent.   We don't define
1773 * codes for all errno's; just the ones which are most likely to be the cause
1774 * of an ext4_error() call.
1775 */
1776#define EXT4_ERR_UNKNOWN	 1
1777#define EXT4_ERR_EIO		 2
1778#define EXT4_ERR_ENOMEM		 3
1779#define EXT4_ERR_EFSBADCRC	 4
1780#define EXT4_ERR_EFSCORRUPTED	 5
1781#define EXT4_ERR_ENOSPC		 6
1782#define EXT4_ERR_ENOKEY		 7
1783#define EXT4_ERR_EROFS		 8
1784#define EXT4_ERR_EFBIG		 9
1785#define EXT4_ERR_EEXIST		10
1786#define EXT4_ERR_ERANGE		11
1787#define EXT4_ERR_EOVERFLOW	12
1788#define EXT4_ERR_EBUSY		13
1789#define EXT4_ERR_ENOTDIR	14
1790#define EXT4_ERR_ENOTEMPTY	15
1791#define EXT4_ERR_ESHUTDOWN	16
1792#define EXT4_ERR_EFAULT		17
1793
1794/*
1795 * Inode dynamic state flags
1796 */
1797enum {
1798	EXT4_STATE_JDATA,		/* journaled data exists */
1799	EXT4_STATE_NEW,			/* inode is newly created */
1800	EXT4_STATE_XATTR,		/* has in-inode xattrs */
1801	EXT4_STATE_NO_EXPAND,		/* No space for expansion */
1802	EXT4_STATE_DA_ALLOC_CLOSE,	/* Alloc DA blks on close */
1803	EXT4_STATE_EXT_MIGRATE,		/* Inode is migrating */
1804	EXT4_STATE_NEWENTRY,		/* File just added to dir */
1805	EXT4_STATE_MAY_INLINE_DATA,	/* may have in-inode data */
1806	EXT4_STATE_EXT_PRECACHED,	/* extents have been precached */
1807	EXT4_STATE_LUSTRE_EA_INODE,	/* Lustre-style ea_inode */
1808	EXT4_STATE_VERITY_IN_PROGRESS,	/* building fs-verity Merkle tree */
1809	EXT4_STATE_FC_COMMITTING,	/* Fast commit ongoing */
1810};
1811
1812#define EXT4_INODE_BIT_FNS(name, field, offset)				\
1813static inline int ext4_test_inode_##name(struct inode *inode, int bit)	\
1814{									\
1815	return test_bit(bit + (offset), &EXT4_I(inode)->i_##field);	\
1816}									\
1817static inline void ext4_set_inode_##name(struct inode *inode, int bit)	\
1818{									\
1819	set_bit(bit + (offset), &EXT4_I(inode)->i_##field);		\
1820}									\
1821static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1822{									\
1823	clear_bit(bit + (offset), &EXT4_I(inode)->i_##field);		\
1824}
1825
1826/* Add these declarations here only so that these functions can be
1827 * found by name.  Otherwise, they are very hard to locate. */
1828static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1829static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1830static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1831EXT4_INODE_BIT_FNS(flag, flags, 0)
1832
1833/* Add these declarations here only so that these functions can be
1834 * found by name.  Otherwise, they are very hard to locate. */
1835static inline int ext4_test_inode_state(struct inode *inode, int bit);
1836static inline void ext4_set_inode_state(struct inode *inode, int bit);
1837static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1838#if (BITS_PER_LONG < 64)
1839EXT4_INODE_BIT_FNS(state, state_flags, 0)
1840
1841static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1842{
1843	(ei)->i_state_flags = 0;
1844}
1845#else
1846EXT4_INODE_BIT_FNS(state, flags, 32)
1847
1848static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1849{
1850	/* We depend on the fact that callers will set i_flags */
1851}
1852#endif
1853#else
1854/* Assume that user mode programs are passing in an ext4fs superblock, not
1855 * a kernel struct super_block.  This will allow us to call the feature-test
1856 * macros from user land. */
1857#define EXT4_SB(sb)	(sb)
1858#endif
1859
1860static inline bool ext4_verity_in_progress(struct inode *inode)
1861{
1862	return IS_ENABLED(CONFIG_FS_VERITY) &&
1863	       ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS);
1864}
1865
1866#define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1867
1868/*
1869 * Codes for operating systems
1870 */
1871#define EXT4_OS_LINUX		0
1872#define EXT4_OS_HURD		1
1873#define EXT4_OS_MASIX		2
1874#define EXT4_OS_FREEBSD		3
1875#define EXT4_OS_LITES		4
1876
1877/*
1878 * Revision levels
1879 */
1880#define EXT4_GOOD_OLD_REV	0	/* The good old (original) format */
1881#define EXT4_DYNAMIC_REV	1	/* V2 format w/ dynamic inode sizes */
1882
1883#define EXT4_CURRENT_REV	EXT4_GOOD_OLD_REV
1884#define EXT4_MAX_SUPP_REV	EXT4_DYNAMIC_REV
1885
1886#define EXT4_GOOD_OLD_INODE_SIZE 128
1887
1888#define EXT4_EXTRA_TIMESTAMP_MAX	(((s64)1 << 34) - 1  + S32_MIN)
1889#define EXT4_NON_EXTRA_TIMESTAMP_MAX	S32_MAX
1890#define EXT4_TIMESTAMP_MIN		S32_MIN
1891
1892/*
1893 * Feature set definitions
1894 */
1895
1896#define EXT4_FEATURE_COMPAT_DIR_PREALLOC	0x0001
1897#define EXT4_FEATURE_COMPAT_IMAGIC_INODES	0x0002
1898#define EXT4_FEATURE_COMPAT_HAS_JOURNAL		0x0004
1899#define EXT4_FEATURE_COMPAT_EXT_ATTR		0x0008
1900#define EXT4_FEATURE_COMPAT_RESIZE_INODE	0x0010
1901#define EXT4_FEATURE_COMPAT_DIR_INDEX		0x0020
1902#define EXT4_FEATURE_COMPAT_SPARSE_SUPER2	0x0200
1903/*
1904 * The reason why "FAST_COMMIT" is a compat feature is that, FS becomes
1905 * incompatible only if fast commit blocks are present in the FS. Since we
1906 * clear the journal (and thus the fast commit blocks), we don't mark FS as
1907 * incompatible. We also have a JBD2 incompat feature, which gets set when
1908 * there are fast commit blocks present in the journal.
1909 */
1910#define EXT4_FEATURE_COMPAT_FAST_COMMIT		0x0400
1911#define EXT4_FEATURE_COMPAT_STABLE_INODES	0x0800
1912
1913#define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER	0x0001
1914#define EXT4_FEATURE_RO_COMPAT_LARGE_FILE	0x0002
1915#define EXT4_FEATURE_RO_COMPAT_BTREE_DIR	0x0004
1916#define EXT4_FEATURE_RO_COMPAT_HUGE_FILE        0x0008
1917#define EXT4_FEATURE_RO_COMPAT_GDT_CSUM		0x0010
1918#define EXT4_FEATURE_RO_COMPAT_DIR_NLINK	0x0020
1919#define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE	0x0040
1920#define EXT4_FEATURE_RO_COMPAT_QUOTA		0x0100
1921#define EXT4_FEATURE_RO_COMPAT_BIGALLOC		0x0200
1922/*
1923 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM).  When
1924 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1925 * all other data structures' checksums.  However, the METADATA_CSUM and
1926 * GDT_CSUM bits are mutually exclusive.
1927 */
1928#define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM	0x0400
1929#define EXT4_FEATURE_RO_COMPAT_READONLY		0x1000
1930#define EXT4_FEATURE_RO_COMPAT_PROJECT		0x2000
1931#define EXT4_FEATURE_RO_COMPAT_VERITY		0x8000
1932
1933#define EXT4_FEATURE_INCOMPAT_COMPRESSION	0x0001
1934#define EXT4_FEATURE_INCOMPAT_FILETYPE		0x0002
1935#define EXT4_FEATURE_INCOMPAT_RECOVER		0x0004 /* Needs recovery */
1936#define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV	0x0008 /* Journal device */
1937#define EXT4_FEATURE_INCOMPAT_META_BG		0x0010
1938#define EXT4_FEATURE_INCOMPAT_EXTENTS		0x0040 /* extents support */
1939#define EXT4_FEATURE_INCOMPAT_64BIT		0x0080
1940#define EXT4_FEATURE_INCOMPAT_MMP               0x0100
1941#define EXT4_FEATURE_INCOMPAT_FLEX_BG		0x0200
1942#define EXT4_FEATURE_INCOMPAT_EA_INODE		0x0400 /* EA in inode */
1943#define EXT4_FEATURE_INCOMPAT_DIRDATA		0x1000 /* data in dirent */
1944#define EXT4_FEATURE_INCOMPAT_CSUM_SEED		0x2000
1945#define EXT4_FEATURE_INCOMPAT_LARGEDIR		0x4000 /* >2GB or 3-lvl htree */
1946#define EXT4_FEATURE_INCOMPAT_INLINE_DATA	0x8000 /* data in inode */
1947#define EXT4_FEATURE_INCOMPAT_ENCRYPT		0x10000
1948#define EXT4_FEATURE_INCOMPAT_CASEFOLD		0x20000
1949
1950extern void ext4_update_dynamic_rev(struct super_block *sb);
1951
1952#define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
1953static inline bool ext4_has_feature_##name(struct super_block *sb) \
1954{ \
1955	return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1956		cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
1957} \
1958static inline void ext4_set_feature_##name(struct super_block *sb) \
1959{ \
1960	ext4_update_dynamic_rev(sb); \
1961	EXT4_SB(sb)->s_es->s_feature_compat |= \
1962		cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1963} \
1964static inline void ext4_clear_feature_##name(struct super_block *sb) \
1965{ \
1966	EXT4_SB(sb)->s_es->s_feature_compat &= \
1967		~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1968}
1969
1970#define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
1971static inline bool ext4_has_feature_##name(struct super_block *sb) \
1972{ \
1973	return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1974		cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
1975} \
1976static inline void ext4_set_feature_##name(struct super_block *sb) \
1977{ \
1978	ext4_update_dynamic_rev(sb); \
1979	EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
1980		cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1981} \
1982static inline void ext4_clear_feature_##name(struct super_block *sb) \
1983{ \
1984	EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
1985		~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1986}
1987
1988#define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
1989static inline bool ext4_has_feature_##name(struct super_block *sb) \
1990{ \
1991	return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1992		cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
1993} \
1994static inline void ext4_set_feature_##name(struct super_block *sb) \
1995{ \
1996	ext4_update_dynamic_rev(sb); \
1997	EXT4_SB(sb)->s_es->s_feature_incompat |= \
1998		cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1999} \
2000static inline void ext4_clear_feature_##name(struct super_block *sb) \
2001{ \
2002	EXT4_SB(sb)->s_es->s_feature_incompat &= \
2003		~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
2004}
2005
2006EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc,		DIR_PREALLOC)
2007EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes,	IMAGIC_INODES)
2008EXT4_FEATURE_COMPAT_FUNCS(journal,		HAS_JOURNAL)
2009EXT4_FEATURE_COMPAT_FUNCS(xattr,		EXT_ATTR)
2010EXT4_FEATURE_COMPAT_FUNCS(resize_inode,		RESIZE_INODE)
2011EXT4_FEATURE_COMPAT_FUNCS(dir_index,		DIR_INDEX)
2012EXT4_FEATURE_COMPAT_FUNCS(sparse_super2,	SPARSE_SUPER2)
2013EXT4_FEATURE_COMPAT_FUNCS(fast_commit,		FAST_COMMIT)
2014EXT4_FEATURE_COMPAT_FUNCS(stable_inodes,	STABLE_INODES)
2015
2016EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super,	SPARSE_SUPER)
2017EXT4_FEATURE_RO_COMPAT_FUNCS(large_file,	LARGE_FILE)
2018EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir,		BTREE_DIR)
2019EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file,		HUGE_FILE)
2020EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum,		GDT_CSUM)
2021EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink,		DIR_NLINK)
2022EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize,	EXTRA_ISIZE)
2023EXT4_FEATURE_RO_COMPAT_FUNCS(quota,		QUOTA)
2024EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc,		BIGALLOC)
2025EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum,	METADATA_CSUM)
2026EXT4_FEATURE_RO_COMPAT_FUNCS(readonly,		READONLY)
2027EXT4_FEATURE_RO_COMPAT_FUNCS(project,		PROJECT)
2028EXT4_FEATURE_RO_COMPAT_FUNCS(verity,		VERITY)
2029
2030EXT4_FEATURE_INCOMPAT_FUNCS(compression,	COMPRESSION)
2031EXT4_FEATURE_INCOMPAT_FUNCS(filetype,		FILETYPE)
2032EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery,	RECOVER)
2033EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev,	JOURNAL_DEV)
2034EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg,		META_BG)
2035EXT4_FEATURE_INCOMPAT_FUNCS(extents,		EXTENTS)
2036EXT4_FEATURE_INCOMPAT_FUNCS(64bit,		64BIT)
2037EXT4_FEATURE_INCOMPAT_FUNCS(mmp,		MMP)
2038EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg,		FLEX_BG)
2039EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode,		EA_INODE)
2040EXT4_FEATURE_INCOMPAT_FUNCS(dirdata,		DIRDATA)
2041EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed,		CSUM_SEED)
2042EXT4_FEATURE_INCOMPAT_FUNCS(largedir,		LARGEDIR)
2043EXT4_FEATURE_INCOMPAT_FUNCS(inline_data,	INLINE_DATA)
2044EXT4_FEATURE_INCOMPAT_FUNCS(encrypt,		ENCRYPT)
2045EXT4_FEATURE_INCOMPAT_FUNCS(casefold,		CASEFOLD)
2046
2047#define EXT2_FEATURE_COMPAT_SUPP	EXT4_FEATURE_COMPAT_EXT_ATTR
2048#define EXT2_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
2049					 EXT4_FEATURE_INCOMPAT_META_BG)
2050#define EXT2_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2051					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2052					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
2053
2054#define EXT3_FEATURE_COMPAT_SUPP	EXT4_FEATURE_COMPAT_EXT_ATTR
2055#define EXT3_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
2056					 EXT4_FEATURE_INCOMPAT_RECOVER| \
2057					 EXT4_FEATURE_INCOMPAT_META_BG)
2058#define EXT3_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2059					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2060					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
2061
2062#define EXT4_FEATURE_COMPAT_SUPP	EXT4_FEATURE_COMPAT_EXT_ATTR
2063#define EXT4_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
2064					 EXT4_FEATURE_INCOMPAT_RECOVER| \
2065					 EXT4_FEATURE_INCOMPAT_META_BG| \
2066					 EXT4_FEATURE_INCOMPAT_EXTENTS| \
2067					 EXT4_FEATURE_INCOMPAT_64BIT| \
2068					 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
2069					 EXT4_FEATURE_INCOMPAT_EA_INODE| \
2070					 EXT4_FEATURE_INCOMPAT_MMP | \
2071					 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
2072					 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
2073					 EXT4_FEATURE_INCOMPAT_CASEFOLD | \
2074					 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \
2075					 EXT4_FEATURE_INCOMPAT_LARGEDIR)
2076#define EXT4_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2077					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2078					 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
2079					 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
2080					 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
2081					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
2082					 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
2083					 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
2084					 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
2085					 EXT4_FEATURE_RO_COMPAT_QUOTA |\
2086					 EXT4_FEATURE_RO_COMPAT_PROJECT |\
2087					 EXT4_FEATURE_RO_COMPAT_VERITY)
2088
2089#define EXTN_FEATURE_FUNCS(ver) \
2090static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
2091{ \
2092	return ((EXT4_SB(sb)->s_es->s_feature_compat & \
2093		cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
2094} \
2095static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
2096{ \
2097	return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
2098		cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
2099} \
2100static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
2101{ \
2102	return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
2103		cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
2104}
2105
2106EXTN_FEATURE_FUNCS(2)
2107EXTN_FEATURE_FUNCS(3)
2108EXTN_FEATURE_FUNCS(4)
2109
2110static inline bool ext4_has_compat_features(struct super_block *sb)
2111{
2112	return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
2113}
2114static inline bool ext4_has_ro_compat_features(struct super_block *sb)
2115{
2116	return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
2117}
2118static inline bool ext4_has_incompat_features(struct super_block *sb)
2119{
2120	return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
2121}
2122
2123/*
2124 * Superblock flags
2125 */
2126#define EXT4_FLAGS_RESIZING	0
2127#define EXT4_FLAGS_SHUTDOWN	1
2128#define EXT4_FLAGS_BDEV_IS_DAX	2
2129
2130static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi)
2131{
2132	return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags);
2133}
2134
2135
2136/*
2137 * Default values for user and/or group using reserved blocks
2138 */
2139#define	EXT4_DEF_RESUID		0
2140#define	EXT4_DEF_RESGID		0
2141
2142/*
2143 * Default project ID
2144 */
2145#define	EXT4_DEF_PROJID		0
2146
2147#define EXT4_DEF_INODE_READAHEAD_BLKS	32
2148
2149/*
2150 * Default mount options
2151 */
2152#define EXT4_DEFM_DEBUG		0x0001
2153#define EXT4_DEFM_BSDGROUPS	0x0002
2154#define EXT4_DEFM_XATTR_USER	0x0004
2155#define EXT4_DEFM_ACL		0x0008
2156#define EXT4_DEFM_UID16		0x0010
2157#define EXT4_DEFM_JMODE		0x0060
2158#define EXT4_DEFM_JMODE_DATA	0x0020
2159#define EXT4_DEFM_JMODE_ORDERED	0x0040
2160#define EXT4_DEFM_JMODE_WBACK	0x0060
2161#define EXT4_DEFM_NOBARRIER	0x0100
2162#define EXT4_DEFM_BLOCK_VALIDITY 0x0200
2163#define EXT4_DEFM_DISCARD	0x0400
2164#define EXT4_DEFM_NODELALLOC	0x0800
2165
2166/*
2167 * Default journal batch times
2168 */
2169#define EXT4_DEF_MIN_BATCH_TIME	0
2170#define EXT4_DEF_MAX_BATCH_TIME	15000 /* 15ms */
2171
2172/*
2173 * Minimum number of groups in a flexgroup before we separate out
2174 * directories into the first block group of a flexgroup
2175 */
2176#define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME	4
2177
2178/*
2179 * Structure of a directory entry
2180 */
2181#define EXT4_NAME_LEN 255
2182/*
2183 * Base length of the ext4 directory entry excluding the name length
2184 */
2185#define EXT4_BASE_DIR_LEN (sizeof(struct ext4_dir_entry_2) - EXT4_NAME_LEN)
2186
2187struct ext4_dir_entry {
2188	__le32	inode;			/* Inode number */
2189	__le16	rec_len;		/* Directory entry length */
2190	__le16	name_len;		/* Name length */
2191	char	name[EXT4_NAME_LEN];	/* File name */
2192};
2193
2194/*
2195 * The new version of the directory entry.  Since EXT4 structures are
2196 * stored in intel byte order, and the name_len field could never be
2197 * bigger than 255 chars, it's safe to reclaim the extra byte for the
2198 * file_type field.
2199 */
2200struct ext4_dir_entry_2 {
2201	__le32	inode;			/* Inode number */
2202	__le16	rec_len;		/* Directory entry length */
2203	__u8	name_len;		/* Name length */
2204	__u8	file_type;		/* See file type macros EXT4_FT_* below */
2205	char	name[EXT4_NAME_LEN];	/* File name */
2206};
2207
2208/*
2209 * This is a bogus directory entry at the end of each leaf block that
2210 * records checksums.
2211 */
2212struct ext4_dir_entry_tail {
2213	__le32	det_reserved_zero1;	/* Pretend to be unused */
2214	__le16	det_rec_len;		/* 12 */
2215	__u8	det_reserved_zero2;	/* Zero name length */
2216	__u8	det_reserved_ft;	/* 0xDE, fake file type */
2217	__le32	det_checksum;		/* crc32c(uuid+inum+dirblock) */
2218};
2219
2220#define EXT4_DIRENT_TAIL(block, blocksize) \
2221	((struct ext4_dir_entry_tail *)(((void *)(block)) + \
2222					((blocksize) - \
2223					 sizeof(struct ext4_dir_entry_tail))))
2224
2225/*
2226 * Ext4 directory file types.  Only the low 3 bits are used.  The
2227 * other bits are reserved for now.
2228 */
2229#define EXT4_FT_UNKNOWN		0
2230#define EXT4_FT_REG_FILE	1
2231#define EXT4_FT_DIR		2
2232#define EXT4_FT_CHRDEV		3
2233#define EXT4_FT_BLKDEV		4
2234#define EXT4_FT_FIFO		5
2235#define EXT4_FT_SOCK		6
2236#define EXT4_FT_SYMLINK		7
2237
2238#define EXT4_FT_MAX		8
2239
2240#define EXT4_FT_DIR_CSUM	0xDE
2241
2242/*
2243 * EXT4_DIR_PAD defines the directory entries boundaries
2244 *
2245 * NOTE: It must be a multiple of 4
2246 */
2247#define EXT4_DIR_PAD			4
2248#define EXT4_DIR_ROUND			(EXT4_DIR_PAD - 1)
2249#define EXT4_DIR_REC_LEN(name_len)	(((name_len) + 8 + EXT4_DIR_ROUND) & \
2250					 ~EXT4_DIR_ROUND)
2251#define EXT4_MAX_REC_LEN		((1<<16)-1)
2252
2253/*
2254 * If we ever get support for fs block sizes > page_size, we'll need
2255 * to remove the #if statements in the next two functions...
2256 */
2257static inline unsigned int
2258ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
2259{
2260	unsigned len = le16_to_cpu(dlen);
2261
2262#if (PAGE_SIZE >= 65536)
2263	if (len == EXT4_MAX_REC_LEN || len == 0)
2264		return blocksize;
2265	return (len & 65532) | ((len & 3) << 16);
2266#else
2267	return len;
2268#endif
2269}
2270
2271static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
2272{
2273	if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
2274		BUG();
2275#if (PAGE_SIZE >= 65536)
2276	if (len < 65536)
2277		return cpu_to_le16(len);
2278	if (len == blocksize) {
2279		if (blocksize == 65536)
2280			return cpu_to_le16(EXT4_MAX_REC_LEN);
2281		else
2282			return cpu_to_le16(0);
2283	}
2284	return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
2285#else
2286	return cpu_to_le16(len);
2287#endif
2288}
2289
2290/*
2291 * Hash Tree Directory indexing
2292 * (c) Daniel Phillips, 2001
2293 */
2294
2295#define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
2296		    ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2297#define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \
2298		    !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir)))
2299#define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2300
2301/* Legal values for the dx_root hash_version field: */
2302
2303#define DX_HASH_LEGACY			0
2304#define DX_HASH_HALF_MD4		1
2305#define DX_HASH_TEA			2
2306#define DX_HASH_LEGACY_UNSIGNED		3
2307#define DX_HASH_HALF_MD4_UNSIGNED	4
2308#define DX_HASH_TEA_UNSIGNED		5
2309
2310static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2311			      const void *address, unsigned int length)
2312{
2313	struct {
2314		struct shash_desc shash;
2315		char ctx[4];
2316	} desc;
2317
2318	BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2319
2320	desc.shash.tfm = sbi->s_chksum_driver;
2321	*(u32 *)desc.ctx = crc;
2322
2323	BUG_ON(crypto_shash_update(&desc.shash, address, length));
2324
2325	return *(u32 *)desc.ctx;
2326}
2327
2328#ifdef __KERNEL__
2329
2330/* hash info structure used by the directory hash */
2331struct dx_hash_info
2332{
2333	u32		hash;
2334	u32		minor_hash;
2335	int		hash_version;
2336	u32		*seed;
2337};
2338
2339
2340/* 32 and 64 bit signed EOF for dx directories */
2341#define EXT4_HTREE_EOF_32BIT   ((1UL  << (32 - 1)) - 1)
2342#define EXT4_HTREE_EOF_64BIT   ((1ULL << (64 - 1)) - 1)
2343
2344
2345/*
2346 * Control parameters used by ext4_htree_next_block
2347 */
2348#define HASH_NB_ALWAYS		1
2349
2350struct ext4_filename {
2351	const struct qstr *usr_fname;
2352	struct fscrypt_str disk_name;
2353	struct dx_hash_info hinfo;
2354#ifdef CONFIG_FS_ENCRYPTION
2355	struct fscrypt_str crypto_buf;
2356#endif
2357#ifdef CONFIG_UNICODE
2358	struct fscrypt_str cf_name;
2359#endif
2360};
2361
2362#define fname_name(p) ((p)->disk_name.name)
2363#define fname_len(p)  ((p)->disk_name.len)
2364
2365/*
2366 * Describe an inode's exact location on disk and in memory
2367 */
2368struct ext4_iloc
2369{
2370	struct buffer_head *bh;
2371	unsigned long offset;
2372	ext4_group_t block_group;
2373};
2374
2375static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2376{
2377	return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2378}
2379
2380static inline bool ext4_is_quota_file(struct inode *inode)
2381{
2382	return IS_NOQUOTA(inode) &&
2383	       !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL);
2384}
2385
2386/*
2387 * This structure is stuffed into the struct file's private_data field
2388 * for directories.  It is where we put information so that we can do
2389 * readdir operations in hash tree order.
2390 */
2391struct dir_private_info {
2392	struct rb_root	root;
2393	struct rb_node	*curr_node;
2394	struct fname	*extra_fname;
2395	loff_t		last_pos;
2396	__u32		curr_hash;
2397	__u32		curr_minor_hash;
2398	__u32		next_hash;
2399};
2400
2401/* calculate the first block number of the group */
2402static inline ext4_fsblk_t
2403ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2404{
2405	return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2406		le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2407}
2408
2409/*
2410 * Special error return code only used by dx_probe() and its callers.
2411 */
2412#define ERR_BAD_DX_DIR	(-(MAX_ERRNO - 1))
2413
2414/* htree levels for ext4 */
2415#define	EXT4_HTREE_LEVEL_COMPAT	2
2416#define	EXT4_HTREE_LEVEL	3
2417
2418static inline int ext4_dir_htree_level(struct super_block *sb)
2419{
2420	return ext4_has_feature_largedir(sb) ?
2421		EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT;
2422}
2423
2424/*
2425 * Timeout and state flag for lazy initialization inode thread.
2426 */
2427#define EXT4_DEF_LI_WAIT_MULT			10
2428#define EXT4_DEF_LI_MAX_START_DELAY		5
2429#define EXT4_LAZYINIT_QUIT			0x0001
2430#define EXT4_LAZYINIT_RUNNING			0x0002
2431
2432/*
2433 * Lazy inode table initialization info
2434 */
2435struct ext4_lazy_init {
2436	unsigned long		li_state;
2437	struct list_head	li_request_list;
2438	struct mutex		li_list_mtx;
2439};
2440
2441enum ext4_li_mode {
2442	EXT4_LI_MODE_PREFETCH_BBITMAP,
2443	EXT4_LI_MODE_ITABLE,
2444};
2445
2446struct ext4_li_request {
2447	struct super_block	*lr_super;
2448	enum ext4_li_mode	lr_mode;
2449	ext4_group_t		lr_first_not_zeroed;
2450	ext4_group_t		lr_next_group;
2451	struct list_head	lr_request;
2452	unsigned long		lr_next_sched;
2453	unsigned long		lr_timeout;
2454};
2455
2456struct ext4_features {
2457	struct kobject f_kobj;
2458	struct completion f_kobj_unregister;
2459};
2460
2461/*
2462 * This structure will be used for multiple mount protection. It will be
2463 * written into the block number saved in the s_mmp_block field in the
2464 * superblock. Programs that check MMP should assume that if
2465 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2466 * to use the filesystem, regardless of how old the timestamp is.
2467 */
2468#define EXT4_MMP_MAGIC     0x004D4D50U /* ASCII for MMP */
2469#define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2470#define EXT4_MMP_SEQ_FSCK  0xE24D4D50U /* mmp_seq value when being fscked */
2471#define EXT4_MMP_SEQ_MAX   0xE24D4D4FU /* maximum valid mmp_seq value */
2472
2473struct mmp_struct {
2474	__le32	mmp_magic;		/* Magic number for MMP */
2475	__le32	mmp_seq;		/* Sequence no. updated periodically */
2476
2477	/*
2478	 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2479	 * purposes and do not affect the correctness of the algorithm
2480	 */
2481	__le64	mmp_time;		/* Time last updated */
2482	char	mmp_nodename[64];	/* Node which last updated MMP block */
2483	char	mmp_bdevname[32];	/* Bdev which last updated MMP block */
2484
2485	/*
2486	 * mmp_check_interval is used to verify if the MMP block has been
2487	 * updated on the block device. The value is updated based on the
2488	 * maximum time to write the MMP block during an update cycle.
2489	 */
2490	__le16	mmp_check_interval;
2491
2492	__le16	mmp_pad1;
2493	__le32	mmp_pad2[226];
2494	__le32	mmp_checksum;		/* crc32c(uuid+mmp_block) */
2495};
2496
2497/* arguments passed to the mmp thread */
2498struct mmpd_data {
2499	struct buffer_head *bh; /* bh from initial read_mmp_block() */
2500	struct super_block *sb;  /* super block of the fs */
2501};
2502
2503/*
2504 * Check interval multiplier
2505 * The MMP block is written every update interval and initially checked every
2506 * update interval x the multiplier (the value is then adapted based on the
2507 * write latency). The reason is that writes can be delayed under load and we
2508 * don't want readers to incorrectly assume that the filesystem is no longer
2509 * in use.
2510 */
2511#define EXT4_MMP_CHECK_MULT		2UL
2512
2513/*
2514 * Minimum interval for MMP checking in seconds.
2515 */
2516#define EXT4_MMP_MIN_CHECK_INTERVAL	5UL
2517
2518/*
2519 * Maximum interval for MMP checking in seconds.
2520 */
2521#define EXT4_MMP_MAX_CHECK_INTERVAL	300UL
2522
2523/*
2524 * Function prototypes
2525 */
2526
2527/*
2528 * Ok, these declarations are also in <linux/kernel.h> but none of the
2529 * ext4 source programs needs to include it so they are duplicated here.
2530 */
2531# define NORET_TYPE	/**/
2532# define ATTRIB_NORET	__attribute__((noreturn))
2533# define NORET_AND	noreturn,
2534
2535/* bitmap.c */
2536extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2537void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2538				struct ext4_group_desc *gdp,
2539				struct buffer_head *bh, int sz);
2540int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2541				  struct ext4_group_desc *gdp,
2542				  struct buffer_head *bh, int sz);
2543void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2544				struct ext4_group_desc *gdp,
2545				struct buffer_head *bh);
2546int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2547				  struct ext4_group_desc *gdp,
2548				  struct buffer_head *bh);
2549
2550/* balloc.c */
2551extern void ext4_get_group_no_and_offset(struct super_block *sb,
2552					 ext4_fsblk_t blocknr,
2553					 ext4_group_t *blockgrpp,
2554					 ext4_grpblk_t *offsetp);
2555extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2556					  ext4_fsblk_t block);
2557
2558extern unsigned int ext4_block_group(struct super_block *sb,
2559			ext4_fsblk_t blocknr);
2560extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2561			ext4_fsblk_t blocknr);
2562extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2563extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2564			ext4_group_t group);
2565extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2566					 ext4_fsblk_t goal,
2567					 unsigned int flags,
2568					 unsigned long *count,
2569					 int *errp);
2570extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2571				    s64 nclusters, unsigned int flags);
2572extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2573extern void ext4_check_blocks_bitmap(struct super_block *);
2574extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2575						    ext4_group_t block_group,
2576						    struct buffer_head ** bh);
2577extern struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2578						   ext4_group_t group);
2579extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2580
2581extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2582						ext4_group_t block_group,
2583						bool ignore_locked);
2584extern int ext4_wait_block_bitmap(struct super_block *sb,
2585				  ext4_group_t block_group,
2586				  struct buffer_head *bh);
2587extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2588						  ext4_group_t block_group);
2589extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2590					      ext4_group_t block_group,
2591					      struct ext4_group_desc *gdp);
2592ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2593
2594#ifdef CONFIG_UNICODE
2595extern void ext4_fname_setup_ci_filename(struct inode *dir,
2596					 const struct qstr *iname,
2597					 struct fscrypt_str *fname);
2598#endif
2599
2600#ifdef CONFIG_FS_ENCRYPTION
2601static inline void ext4_fname_from_fscrypt_name(struct ext4_filename *dst,
2602						const struct fscrypt_name *src)
2603{
2604	memset(dst, 0, sizeof(*dst));
2605
2606	dst->usr_fname = src->usr_fname;
2607	dst->disk_name = src->disk_name;
2608	dst->hinfo.hash = src->hash;
2609	dst->hinfo.minor_hash = src->minor_hash;
2610	dst->crypto_buf = src->crypto_buf;
2611}
2612
2613static inline int ext4_fname_setup_filename(struct inode *dir,
2614					    const struct qstr *iname,
2615					    int lookup,
2616					    struct ext4_filename *fname)
2617{
2618	struct fscrypt_name name;
2619	int err;
2620
2621	err = fscrypt_setup_filename(dir, iname, lookup, &name);
2622	if (err)
2623		return err;
2624
2625	ext4_fname_from_fscrypt_name(fname, &name);
2626
2627#ifdef CONFIG_UNICODE
2628	ext4_fname_setup_ci_filename(dir, iname, &fname->cf_name);
2629#endif
2630	return 0;
2631}
2632
2633static inline int ext4_fname_prepare_lookup(struct inode *dir,
2634					    struct dentry *dentry,
2635					    struct ext4_filename *fname)
2636{
2637	struct fscrypt_name name;
2638	int err;
2639
2640	err = fscrypt_prepare_lookup(dir, dentry, &name);
2641	if (err)
2642		return err;
2643
2644	ext4_fname_from_fscrypt_name(fname, &name);
2645
2646#ifdef CONFIG_UNICODE
2647	ext4_fname_setup_ci_filename(dir, &dentry->d_name, &fname->cf_name);
2648#endif
2649	return 0;
2650}
2651
2652static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2653{
2654	struct fscrypt_name name;
2655
2656	name.crypto_buf = fname->crypto_buf;
2657	fscrypt_free_filename(&name);
2658
2659	fname->crypto_buf.name = NULL;
2660	fname->usr_fname = NULL;
2661	fname->disk_name.name = NULL;
2662
2663#ifdef CONFIG_UNICODE
2664	kfree(fname->cf_name.name);
2665	fname->cf_name.name = NULL;
2666#endif
2667}
2668#else /* !CONFIG_FS_ENCRYPTION */
2669static inline int ext4_fname_setup_filename(struct inode *dir,
2670					    const struct qstr *iname,
2671					    int lookup,
2672					    struct ext4_filename *fname)
2673{
2674	fname->usr_fname = iname;
2675	fname->disk_name.name = (unsigned char *) iname->name;
2676	fname->disk_name.len = iname->len;
2677
2678#ifdef CONFIG_UNICODE
2679	ext4_fname_setup_ci_filename(dir, iname, &fname->cf_name);
2680#endif
2681
2682	return 0;
2683}
2684
2685static inline int ext4_fname_prepare_lookup(struct inode *dir,
2686					    struct dentry *dentry,
2687					    struct ext4_filename *fname)
2688{
2689	return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname);
2690}
2691
2692static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2693{
2694#ifdef CONFIG_UNICODE
2695	kfree(fname->cf_name.name);
2696	fname->cf_name.name = NULL;
2697#endif
2698}
2699#endif /* !CONFIG_FS_ENCRYPTION */
2700
2701/* dir.c */
2702extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2703				  struct file *,
2704				  struct ext4_dir_entry_2 *,
2705				  struct buffer_head *, char *, int,
2706				  unsigned int);
2707#define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset)	\
2708	unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2709					(de), (bh), (buf), (size), (offset)))
2710extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2711				__u32 minor_hash,
2712				struct ext4_dir_entry_2 *dirent,
2713				struct fscrypt_str *ent_name);
2714extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2715extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2716			     struct buffer_head *bh,
2717			     void *buf, int buf_size,
2718			     struct ext4_filename *fname,
2719			     struct ext4_dir_entry_2 **dest_de);
2720void ext4_insert_dentry(struct inode *inode,
2721			struct ext4_dir_entry_2 *de,
2722			int buf_size,
2723			struct ext4_filename *fname);
2724static inline void ext4_update_dx_flag(struct inode *inode)
2725{
2726	if (!ext4_has_feature_dir_index(inode->i_sb) &&
2727	    ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) {
2728		/* ext4_iget() should have caught this... */
2729		WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb));
2730		ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2731	}
2732}
2733static const unsigned char ext4_filetype_table[] = {
2734	DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2735};
2736
2737static inline  unsigned char get_dtype(struct super_block *sb, int filetype)
2738{
2739	if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2740		return DT_UNKNOWN;
2741
2742	return ext4_filetype_table[filetype];
2743}
2744extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2745			     void *buf, int buf_size);
2746
2747/* fsync.c */
2748extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2749
2750/* hash.c */
2751extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len,
2752			  struct dx_hash_info *hinfo);
2753
2754/* ialloc.c */
2755extern int ext4_mark_inode_used(struct super_block *sb, int ino);
2756extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2757				      const struct qstr *qstr, __u32 goal,
2758				      uid_t *owner, __u32 i_flags,
2759				      int handle_type, unsigned int line_no,
2760				      int nblocks);
2761
2762#define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \
2763	__ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2764			 i_flags, 0, 0, 0)
2765#define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2766				    type, nblocks)		    \
2767	__ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2768			 0, (type), __LINE__, (nblocks))
2769
2770
2771extern void ext4_free_inode(handle_t *, struct inode *);
2772extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2773extern unsigned long ext4_count_free_inodes(struct super_block *);
2774extern unsigned long ext4_count_dirs(struct super_block *);
2775extern void ext4_check_inodes_bitmap(struct super_block *);
2776extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2777extern int ext4_init_inode_table(struct super_block *sb,
2778				 ext4_group_t group, int barrier);
2779extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2780
2781/* fast_commit.c */
2782int ext4_fc_info_show(struct seq_file *seq, void *v);
2783void ext4_fc_init(struct super_block *sb, journal_t *journal);
2784void ext4_fc_init_inode(struct inode *inode);
2785void ext4_fc_track_range(handle_t *handle, struct inode *inode, ext4_lblk_t start,
2786			 ext4_lblk_t end);
2787void __ext4_fc_track_unlink(handle_t *handle, struct inode *inode,
2788	struct dentry *dentry);
2789void __ext4_fc_track_link(handle_t *handle, struct inode *inode,
2790	struct dentry *dentry);
2791void ext4_fc_track_unlink(handle_t *handle, struct dentry *dentry);
2792void ext4_fc_track_link(handle_t *handle, struct dentry *dentry);
2793void __ext4_fc_track_create(handle_t *handle, struct inode *inode,
2794			    struct dentry *dentry);
2795void ext4_fc_track_create(handle_t *handle, struct dentry *dentry);
2796void ext4_fc_track_inode(handle_t *handle, struct inode *inode);
2797void ext4_fc_mark_ineligible(struct super_block *sb, int reason);
2798void ext4_fc_start_ineligible(struct super_block *sb, int reason);
2799void ext4_fc_stop_ineligible(struct super_block *sb);
2800void ext4_fc_start_update(struct inode *inode);
2801void ext4_fc_stop_update(struct inode *inode);
2802void ext4_fc_del(struct inode *inode);
2803bool ext4_fc_replay_check_excluded(struct super_block *sb, ext4_fsblk_t block);
2804void ext4_fc_replay_cleanup(struct super_block *sb);
2805int ext4_fc_commit(journal_t *journal, tid_t commit_tid);
2806int __init ext4_fc_init_dentry_cache(void);
2807void ext4_fc_destroy_dentry_cache(void);
2808int ext4_fc_record_regions(struct super_block *sb, int ino,
2809			   ext4_lblk_t lblk, ext4_fsblk_t pblk,
2810			   int len, int replay);
2811
2812/* mballoc.c */
2813extern const struct seq_operations ext4_mb_seq_groups_ops;
2814extern long ext4_mb_stats;
2815extern long ext4_mb_max_to_scan;
2816extern int ext4_seq_mb_stats_show(struct seq_file *seq, void *offset);
2817extern int ext4_mb_init(struct super_block *);
2818extern int ext4_mb_release(struct super_block *);
2819extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2820				struct ext4_allocation_request *, int *);
2821extern int ext4_mb_reserve_blocks(struct super_block *, int);
2822extern void ext4_discard_preallocations(struct inode *, unsigned int);
2823extern int __init ext4_init_mballoc(void);
2824extern void ext4_exit_mballoc(void);
2825extern ext4_group_t ext4_mb_prefetch(struct super_block *sb,
2826				     ext4_group_t group,
2827				     unsigned int nr, int *cnt);
2828extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group,
2829				  unsigned int nr);
2830
2831extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2832			     struct buffer_head *bh, ext4_fsblk_t block,
2833			     unsigned long count, int flags);
2834extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2835				   ext4_group_t ngroups);
2836extern int ext4_mb_add_groupinfo(struct super_block *sb,
2837		ext4_group_t i, struct ext4_group_desc *desc);
2838extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2839				ext4_fsblk_t block, unsigned long count);
2840extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2841extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid);
2842extern void ext4_mb_mark_bb(struct super_block *sb, ext4_fsblk_t block,
2843		       int len, int state);
2844
2845/* inode.c */
2846void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw,
2847			 struct ext4_inode_info *ei);
2848int ext4_inode_is_fast_symlink(struct inode *inode);
2849struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2850struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2851int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count,
2852		     bool wait, struct buffer_head **bhs);
2853int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2854			     struct buffer_head *bh_result, int create);
2855int ext4_get_block(struct inode *inode, sector_t iblock,
2856		   struct buffer_head *bh_result, int create);
2857int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2858			   struct buffer_head *bh, int create);
2859int ext4_walk_page_buffers(handle_t *handle,
2860			   struct buffer_head *head,
2861			   unsigned from,
2862			   unsigned to,
2863			   int *partial,
2864			   int (*fn)(handle_t *handle,
2865				     struct buffer_head *bh));
2866int do_journal_get_write_access(handle_t *handle,
2867				struct buffer_head *bh);
2868#define FALL_BACK_TO_NONDELALLOC 1
2869#define CONVERT_INLINE_DATA	 2
2870
2871typedef enum {
2872	EXT4_IGET_NORMAL =	0,
2873	EXT4_IGET_SPECIAL =	0x0001, /* OK to iget a system inode */
2874	EXT4_IGET_HANDLE = 	0x0002,	/* Inode # is from a handle */
2875	EXT4_IGET_BAD =		0x0004, /* Allow to iget a bad inode */
2876	EXT4_IGET_EA_INODE =	0x0008	/* Inode should contain an EA value */
2877} ext4_iget_flags;
2878
2879extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
2880				 ext4_iget_flags flags, const char *function,
2881				 unsigned int line);
2882
2883#define ext4_iget(sb, ino, flags) \
2884	__ext4_iget((sb), (ino), (flags), __func__, __LINE__)
2885
2886extern int  ext4_write_inode(struct inode *, struct writeback_control *);
2887extern int  ext4_setattr(struct dentry *, struct iattr *);
2888extern int  ext4_getattr(const struct path *, struct kstat *, u32, unsigned int);
2889extern void ext4_evict_inode(struct inode *);
2890extern void ext4_clear_inode(struct inode *);
2891extern int  ext4_file_getattr(const struct path *, struct kstat *, u32, unsigned int);
2892extern int  ext4_sync_inode(handle_t *, struct inode *);
2893extern void ext4_dirty_inode(struct inode *, int);
2894extern int ext4_change_inode_journal_flag(struct inode *, int);
2895extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2896extern int ext4_get_fc_inode_loc(struct super_block *sb, unsigned long ino,
2897			  struct ext4_iloc *iloc);
2898extern int ext4_inode_attach_jinode(struct inode *inode);
2899extern int ext4_can_truncate(struct inode *inode);
2900extern int ext4_truncate(struct inode *);
2901extern int ext4_break_layouts(struct inode *);
2902extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length);
2903extern void ext4_set_inode_flags(struct inode *, bool init);
2904extern int ext4_alloc_da_blocks(struct inode *inode);
2905extern void ext4_set_aops(struct inode *inode);
2906extern int ext4_writepage_trans_blocks(struct inode *);
2907extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2908extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2909			     loff_t lstart, loff_t lend);
2910extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf);
2911extern vm_fault_t ext4_filemap_fault(struct vm_fault *vmf);
2912extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2913extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
2914extern void ext4_da_release_space(struct inode *inode, int to_free);
2915extern void ext4_da_update_reserve_space(struct inode *inode,
2916					int used, int quota_claim);
2917extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
2918			      ext4_fsblk_t pblk, ext4_lblk_t len);
2919
2920/* indirect.c */
2921extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2922				struct ext4_map_blocks *map, int flags);
2923extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2924extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2925extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2926				 ext4_lblk_t start, ext4_lblk_t end);
2927
2928/* ioctl.c */
2929extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2930extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2931extern void ext4_reset_inode_seed(struct inode *inode);
2932
2933/* migrate.c */
2934extern int ext4_ext_migrate(struct inode *);
2935extern int ext4_ind_migrate(struct inode *inode);
2936
2937/* namei.c */
2938extern int ext4_init_new_dir(handle_t *handle, struct inode *dir,
2939			     struct inode *inode);
2940extern int ext4_dirblock_csum_verify(struct inode *inode,
2941				     struct buffer_head *bh);
2942extern int ext4_orphan_add(handle_t *, struct inode *);
2943extern int ext4_orphan_del(handle_t *, struct inode *);
2944extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2945				__u32 start_minor_hash, __u32 *next_hash);
2946extern int ext4_search_dir(struct buffer_head *bh,
2947			   char *search_buf,
2948			   int buf_size,
2949			   struct inode *dir,
2950			   struct ext4_filename *fname,
2951			   unsigned int offset,
2952			   struct ext4_dir_entry_2 **res_dir);
2953extern int ext4_generic_delete_entry(struct inode *dir,
2954				     struct ext4_dir_entry_2 *de_del,
2955				     struct buffer_head *bh,
2956				     void *entry_buf,
2957				     int buf_size,
2958				     int csum_size);
2959extern bool ext4_empty_dir(struct inode *inode);
2960
2961/* resize.c */
2962extern void ext4_kvfree_array_rcu(void *to_free);
2963extern int ext4_group_add(struct super_block *sb,
2964				struct ext4_new_group_data *input);
2965extern int ext4_group_extend(struct super_block *sb,
2966				struct ext4_super_block *es,
2967				ext4_fsblk_t n_blocks_count);
2968extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2969
2970/* super.c */
2971extern struct buffer_head *ext4_sb_bread(struct super_block *sb,
2972					 sector_t block, int op_flags);
2973extern struct buffer_head *ext4_sb_bread_unmovable(struct super_block *sb,
2974						   sector_t block);
2975extern void ext4_read_bh_nowait(struct buffer_head *bh, int op_flags,
2976				bh_end_io_t *end_io);
2977extern int ext4_read_bh(struct buffer_head *bh, int op_flags,
2978			bh_end_io_t *end_io);
2979extern int ext4_read_bh_lock(struct buffer_head *bh, int op_flags, bool wait);
2980extern void ext4_sb_breadahead_unmovable(struct super_block *sb, sector_t block);
2981extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
2982extern int ext4_calculate_overhead(struct super_block *sb);
2983extern void ext4_superblock_csum_set(struct super_block *sb);
2984extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2985				    ext4_group_t ngroup);
2986extern const char *ext4_decode_error(struct super_block *sb, int errno,
2987				     char nbuf[16]);
2988extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb,
2989					     ext4_group_t block_group,
2990					     unsigned int flags);
2991extern unsigned int ext4_num_base_meta_blocks(struct super_block *sb,
2992					      ext4_group_t block_group);
2993
2994extern __printf(7, 8)
2995void __ext4_error(struct super_block *, const char *, unsigned int, bool,
2996		  int, __u64, const char *, ...);
2997extern __printf(6, 7)
2998void __ext4_error_inode(struct inode *, const char *, unsigned int,
2999			ext4_fsblk_t, int, const char *, ...);
3000extern __printf(5, 6)
3001void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
3002		     const char *, ...);
3003extern void __ext4_std_error(struct super_block *, const char *,
3004			     unsigned int, int);
3005extern __printf(4, 5)
3006void __ext4_warning(struct super_block *, const char *, unsigned int,
3007		    const char *, ...);
3008extern __printf(4, 5)
3009void __ext4_warning_inode(const struct inode *inode, const char *function,
3010			  unsigned int line, const char *fmt, ...);
3011extern __printf(3, 4)
3012void __ext4_msg(struct super_block *, const char *, const char *, ...);
3013extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
3014			   const char *, unsigned int, const char *);
3015extern __printf(7, 8)
3016void __ext4_grp_locked_error(const char *, unsigned int,
3017			     struct super_block *, ext4_group_t,
3018			     unsigned long, ext4_fsblk_t,
3019			     const char *, ...);
3020
3021#define EXT4_ERROR_INODE(inode, fmt, a...) \
3022	ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
3023
3024#define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...)			\
3025	__ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a)
3026
3027#define ext4_error_inode_block(inode, block, err, fmt, a...)		\
3028	__ext4_error_inode((inode), __func__, __LINE__, (block), (err),	\
3029			   (fmt), ## a)
3030
3031#define EXT4_ERROR_FILE(file, block, fmt, a...)				\
3032	ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
3033
3034#define ext4_abort(sb, err, fmt, a...)					\
3035	__ext4_error((sb), __func__, __LINE__, true, (err), 0, (fmt), ## a)
3036
3037#ifdef CONFIG_PRINTK
3038
3039#define ext4_error_inode(inode, func, line, block, fmt, ...)		\
3040	__ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__)
3041#define ext4_error_inode_err(inode, func, line, block, err, fmt, ...)	\
3042	__ext4_error_inode((inode), (func), (line), (block), 		\
3043			   (err), (fmt), ##__VA_ARGS__)
3044#define ext4_error_file(file, func, line, block, fmt, ...)		\
3045	__ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
3046#define ext4_error(sb, fmt, ...)					\
3047	__ext4_error((sb), __func__, __LINE__, false, 0, 0, (fmt),	\
3048		##__VA_ARGS__)
3049#define ext4_error_err(sb, err, fmt, ...)				\
3050	__ext4_error((sb), __func__, __LINE__, false, (err), 0, (fmt),	\
3051		##__VA_ARGS__)
3052#define ext4_warning(sb, fmt, ...)					\
3053	__ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
3054#define ext4_warning_inode(inode, fmt, ...)				\
3055	__ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
3056#define ext4_msg(sb, level, fmt, ...)				\
3057	__ext4_msg(sb, level, fmt, ##__VA_ARGS__)
3058#define dump_mmp_msg(sb, mmp, msg)					\
3059	__dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
3060#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...)		\
3061	__ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
3062				fmt, ##__VA_ARGS__)
3063
3064#else
3065
3066#define ext4_error_inode(inode, func, line, block, fmt, ...)		\
3067do {									\
3068	no_printk(fmt, ##__VA_ARGS__);					\
3069	__ext4_error_inode(inode, "", 0, block, 0, " ");		\
3070} while (0)
3071#define ext4_error_inode_err(inode, func, line, block, err, fmt, ...)	\
3072do {									\
3073	no_printk(fmt, ##__VA_ARGS__);					\
3074	__ext4_error_inode(inode, "", 0, block, err, " ");		\
3075} while (0)
3076#define ext4_error_file(file, func, line, block, fmt, ...)		\
3077do {									\
3078	no_printk(fmt, ##__VA_ARGS__);					\
3079	__ext4_error_file(file, "", 0, block, " ");			\
3080} while (0)
3081#define ext4_error(sb, fmt, ...)					\
3082do {									\
3083	no_printk(fmt, ##__VA_ARGS__);					\
3084	__ext4_error(sb, "", 0, false, 0, 0, " ");			\
3085} while (0)
3086#define ext4_error_err(sb, err, fmt, ...)				\
3087do {									\
3088	no_printk(fmt, ##__VA_ARGS__);					\
3089	__ext4_error(sb, "", 0, false, err, 0, " ");			\
3090} while (0)
3091#define ext4_warning(sb, fmt, ...)					\
3092do {									\
3093	no_printk(fmt, ##__VA_ARGS__);					\
3094	__ext4_warning(sb, "", 0, " ");					\
3095} while (0)
3096#define ext4_warning_inode(inode, fmt, ...)				\
3097do {									\
3098	no_printk(fmt, ##__VA_ARGS__);					\
3099	__ext4_warning_inode(inode, "", 0, " ");			\
3100} while (0)
3101#define ext4_msg(sb, level, fmt, ...)					\
3102do {									\
3103	no_printk(fmt, ##__VA_ARGS__);					\
3104	__ext4_msg(sb, "", " ");					\
3105} while (0)
3106#define dump_mmp_msg(sb, mmp, msg)					\
3107	__dump_mmp_msg(sb, mmp, "", 0, "")
3108#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...)		\
3109do {									\
3110	no_printk(fmt, ##__VA_ARGS__);				\
3111	__ext4_grp_locked_error("", 0, sb, grp, ino, block, " ");	\
3112} while (0)
3113
3114#endif
3115
3116extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
3117				      struct ext4_group_desc *bg);
3118extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
3119				      struct ext4_group_desc *bg);
3120extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
3121				     struct ext4_group_desc *bg);
3122extern __u32 ext4_free_group_clusters(struct super_block *sb,
3123				      struct ext4_group_desc *bg);
3124extern __u32 ext4_free_inodes_count(struct super_block *sb,
3125				 struct ext4_group_desc *bg);
3126extern __u32 ext4_used_dirs_count(struct super_block *sb,
3127				struct ext4_group_desc *bg);
3128extern __u32 ext4_itable_unused_count(struct super_block *sb,
3129				   struct ext4_group_desc *bg);
3130extern void ext4_block_bitmap_set(struct super_block *sb,
3131				  struct ext4_group_desc *bg, ext4_fsblk_t blk);
3132extern void ext4_inode_bitmap_set(struct super_block *sb,
3133				  struct ext4_group_desc *bg, ext4_fsblk_t blk);
3134extern void ext4_inode_table_set(struct super_block *sb,
3135				 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3136extern void ext4_free_group_clusters_set(struct super_block *sb,
3137					 struct ext4_group_desc *bg,
3138					 __u32 count);
3139extern void ext4_free_inodes_set(struct super_block *sb,
3140				struct ext4_group_desc *bg, __u32 count);
3141extern void ext4_used_dirs_set(struct super_block *sb,
3142				struct ext4_group_desc *bg, __u32 count);
3143extern void ext4_itable_unused_set(struct super_block *sb,
3144				   struct ext4_group_desc *bg, __u32 count);
3145extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
3146				       struct ext4_group_desc *gdp);
3147extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
3148				     struct ext4_group_desc *gdp);
3149extern int ext4_register_li_request(struct super_block *sb,
3150				    ext4_group_t first_not_zeroed);
3151
3152static inline int ext4_has_metadata_csum(struct super_block *sb)
3153{
3154	WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
3155		     !EXT4_SB(sb)->s_chksum_driver);
3156
3157	return ext4_has_feature_metadata_csum(sb) &&
3158	       (EXT4_SB(sb)->s_chksum_driver != NULL);
3159}
3160
3161static inline int ext4_has_group_desc_csum(struct super_block *sb)
3162{
3163	return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb);
3164}
3165
3166#define ext4_read_incompat_64bit_val(es, name) \
3167	(((es)->s_feature_incompat & cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT) \
3168		? (ext4_fsblk_t)le32_to_cpu(es->name##_hi) << 32 : 0) | \
3169		le32_to_cpu(es->name##_lo))
3170
3171static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
3172{
3173	return ext4_read_incompat_64bit_val(es, s_blocks_count);
3174}
3175
3176static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
3177{
3178	return ext4_read_incompat_64bit_val(es, s_r_blocks_count);
3179}
3180
3181static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
3182{
3183	return ext4_read_incompat_64bit_val(es, s_free_blocks_count);
3184}
3185
3186static inline void ext4_blocks_count_set(struct ext4_super_block *es,
3187					 ext4_fsblk_t blk)
3188{
3189	es->s_blocks_count_lo = cpu_to_le32((u32)blk);
3190	es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
3191}
3192
3193static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
3194					      ext4_fsblk_t blk)
3195{
3196	es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
3197	es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
3198}
3199
3200static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
3201					   ext4_fsblk_t blk)
3202{
3203	es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
3204	es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
3205}
3206
3207static inline loff_t ext4_isize(struct super_block *sb,
3208				struct ext4_inode *raw_inode)
3209{
3210	if (ext4_has_feature_largedir(sb) ||
3211	    S_ISREG(le16_to_cpu(raw_inode->i_mode)))
3212		return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
3213			le32_to_cpu(raw_inode->i_size_lo);
3214
3215	return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
3216}
3217
3218static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
3219{
3220	raw_inode->i_size_lo = cpu_to_le32(i_size);
3221	raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
3222}
3223
3224/*
3225 * Reading s_groups_count requires using smp_rmb() afterwards.  See
3226 * the locking protocol documented in the comments of ext4_group_add()
3227 * in resize.c
3228 */
3229static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
3230{
3231	ext4_group_t	ngroups = EXT4_SB(sb)->s_groups_count;
3232
3233	smp_rmb();
3234	return ngroups;
3235}
3236
3237static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
3238					     ext4_group_t block_group)
3239{
3240	return block_group >> sbi->s_log_groups_per_flex;
3241}
3242
3243static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
3244{
3245	return 1 << sbi->s_log_groups_per_flex;
3246}
3247
3248#define ext4_std_error(sb, errno)				\
3249do {								\
3250	if ((errno))						\
3251		__ext4_std_error((sb), __func__, __LINE__, (errno));	\
3252} while (0)
3253
3254#ifdef CONFIG_SMP
3255/* Each CPU can accumulate percpu_counter_batch clusters in their local
3256 * counters. So we need to make sure we have free clusters more
3257 * than percpu_counter_batch  * nr_cpu_ids. Also add a window of 4 times.
3258 */
3259#define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
3260#else
3261#define EXT4_FREECLUSTERS_WATERMARK 0
3262#endif
3263
3264/* Update i_disksize. Requires i_mutex to avoid races with truncate */
3265static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
3266{
3267	WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
3268		     !inode_is_locked(inode));
3269	down_write(&EXT4_I(inode)->i_data_sem);
3270	if (newsize > EXT4_I(inode)->i_disksize)
3271		WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize);
3272	up_write(&EXT4_I(inode)->i_data_sem);
3273}
3274
3275/* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
3276static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
3277{
3278	int changed = 0;
3279
3280	if (newsize > inode->i_size) {
3281		i_size_write(inode, newsize);
3282		changed = 1;
3283	}
3284	if (newsize > EXT4_I(inode)->i_disksize) {
3285		ext4_update_i_disksize(inode, newsize);
3286		changed |= 2;
3287	}
3288	return changed;
3289}
3290
3291int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
3292				      loff_t len);
3293
3294struct ext4_group_info {
3295	unsigned long   bb_state;
3296#ifdef AGGRESSIVE_CHECK
3297	unsigned long	bb_check_counter;
3298#endif
3299	struct rb_root  bb_free_root;
3300	ext4_grpblk_t	bb_first_free;	/* first free block */
3301	ext4_grpblk_t	bb_free;	/* total free blocks */
3302	ext4_grpblk_t	bb_fragments;	/* nr of freespace fragments */
3303	ext4_grpblk_t	bb_largest_free_order;/* order of largest frag in BG */
3304	struct          list_head bb_prealloc_list;
3305#ifdef DOUBLE_CHECK
3306	void            *bb_bitmap;
3307#endif
3308	struct rw_semaphore alloc_sem;
3309	ext4_grpblk_t	bb_counters[];	/* Nr of free power-of-two-block
3310					 * regions, index is order.
3311					 * bb_counters[3] = 5 means
3312					 * 5 free 8-block regions. */
3313};
3314
3315#define EXT4_GROUP_INFO_NEED_INIT_BIT		0
3316#define EXT4_GROUP_INFO_WAS_TRIMMED_BIT		1
3317#define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT	2
3318#define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT	3
3319#define EXT4_GROUP_INFO_BBITMAP_CORRUPT		\
3320	(1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT)
3321#define EXT4_GROUP_INFO_IBITMAP_CORRUPT		\
3322	(1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT)
3323#define EXT4_GROUP_INFO_BBITMAP_READ_BIT	4
3324
3325#define EXT4_MB_GRP_NEED_INIT(grp)	\
3326	(test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
3327#define EXT4_MB_GRP_BBITMAP_CORRUPT(grp)	\
3328	(test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3329#define EXT4_MB_GRP_IBITMAP_CORRUPT(grp)	\
3330	(test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3331
3332#define EXT4_MB_GRP_WAS_TRIMMED(grp)	\
3333	(test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3334#define EXT4_MB_GRP_SET_TRIMMED(grp)	\
3335	(set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3336#define EXT4_MB_GRP_CLEAR_TRIMMED(grp)	\
3337	(clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3338#define EXT4_MB_GRP_TEST_AND_SET_READ(grp)	\
3339	(test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state)))
3340
3341#define EXT4_MAX_CONTENTION		8
3342#define EXT4_CONTENTION_THRESHOLD	2
3343
3344static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
3345					      ext4_group_t group)
3346{
3347	return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
3348}
3349
3350/*
3351 * Returns true if the filesystem is busy enough that attempts to
3352 * access the block group locks has run into contention.
3353 */
3354static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
3355{
3356	return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
3357}
3358
3359static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
3360{
3361	spinlock_t *lock = ext4_group_lock_ptr(sb, group);
3362	if (spin_trylock(lock))
3363		/*
3364		 * We're able to grab the lock right away, so drop the
3365		 * lock contention counter.
3366		 */
3367		atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
3368	else {
3369		/*
3370		 * The lock is busy, so bump the contention counter,
3371		 * and then wait on the spin lock.
3372		 */
3373		atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
3374				  EXT4_MAX_CONTENTION);
3375		spin_lock(lock);
3376	}
3377}
3378
3379static inline void ext4_unlock_group(struct super_block *sb,
3380					ext4_group_t group)
3381{
3382	spin_unlock(ext4_group_lock_ptr(sb, group));
3383}
3384
3385/*
3386 * Block validity checking
3387 */
3388#define ext4_check_indirect_blockref(inode, bh)				\
3389	ext4_check_blockref(__func__, __LINE__, inode,			\
3390			    (__le32 *)(bh)->b_data,			\
3391			    EXT4_ADDR_PER_BLOCK((inode)->i_sb))
3392
3393#define ext4_ind_check_inode(inode)					\
3394	ext4_check_blockref(__func__, __LINE__, inode,			\
3395			    EXT4_I(inode)->i_data,			\
3396			    EXT4_NDIR_BLOCKS)
3397
3398/*
3399 * Inodes and files operations
3400 */
3401
3402/* dir.c */
3403extern const struct file_operations ext4_dir_operations;
3404
3405#ifdef CONFIG_UNICODE
3406extern const struct dentry_operations ext4_dentry_ops;
3407#endif
3408
3409/* file.c */
3410extern const struct inode_operations ext4_file_inode_operations;
3411extern const struct file_operations ext4_file_operations;
3412extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
3413
3414/* inline.c */
3415extern int ext4_get_max_inline_size(struct inode *inode);
3416extern int ext4_find_inline_data_nolock(struct inode *inode);
3417extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
3418				 unsigned int len);
3419extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
3420
3421extern int ext4_readpage_inline(struct inode *inode, struct page *page);
3422extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3423					 struct inode *inode,
3424					 loff_t pos, unsigned len,
3425					 unsigned flags,
3426					 struct page **pagep);
3427extern int ext4_write_inline_data_end(struct inode *inode,
3428				      loff_t pos, unsigned len,
3429				      unsigned copied,
3430				      struct page *page);
3431extern struct buffer_head *
3432ext4_journalled_write_inline_data(struct inode *inode,
3433				  unsigned len,
3434				  struct page *page);
3435extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
3436					   struct inode *inode,
3437					   loff_t pos, unsigned len,
3438					   unsigned flags,
3439					   struct page **pagep,
3440					   void **fsdata);
3441extern int ext4_try_add_inline_entry(handle_t *handle,
3442				     struct ext4_filename *fname,
3443				     struct inode *dir, struct inode *inode);
3444extern int ext4_try_create_inline_dir(handle_t *handle,
3445				      struct inode *parent,
3446				      struct inode *inode);
3447extern int ext4_read_inline_dir(struct file *filp,
3448				struct dir_context *ctx,
3449				int *has_inline_data);
3450extern int ext4_inlinedir_to_tree(struct file *dir_file,
3451				  struct inode *dir, ext4_lblk_t block,
3452				  struct dx_hash_info *hinfo,
3453				  __u32 start_hash, __u32 start_minor_hash,
3454				  int *has_inline_data);
3455extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3456					struct ext4_filename *fname,
3457					struct ext4_dir_entry_2 **res_dir,
3458					int *has_inline_data);
3459extern int ext4_delete_inline_entry(handle_t *handle,
3460				    struct inode *dir,
3461				    struct ext4_dir_entry_2 *de_del,
3462				    struct buffer_head *bh,
3463				    int *has_inline_data);
3464extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3465extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3466					struct ext4_dir_entry_2 **parent_de,
3467					int *retval);
3468extern int ext4_inline_data_fiemap(struct inode *inode,
3469				   struct fiemap_extent_info *fieinfo,
3470				   int *has_inline, __u64 start, __u64 len);
3471
3472struct iomap;
3473extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap);
3474
3475extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3476
3477extern int ext4_convert_inline_data(struct inode *inode);
3478
3479static inline int ext4_has_inline_data(struct inode *inode)
3480{
3481	return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3482	       EXT4_I(inode)->i_inline_off;
3483}
3484
3485/* namei.c */
3486extern const struct inode_operations ext4_dir_inode_operations;
3487extern const struct inode_operations ext4_special_inode_operations;
3488extern struct dentry *ext4_get_parent(struct dentry *child);
3489extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3490				 struct ext4_dir_entry_2 *de,
3491				 int blocksize, int csum_size,
3492				 unsigned int parent_ino, int dotdot_real_len);
3493extern void ext4_initialize_dirent_tail(struct buffer_head *bh,
3494					unsigned int blocksize);
3495extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode,
3496				      struct buffer_head *bh);
3497extern int ext4_ci_compare(const struct inode *parent,
3498			   const struct qstr *fname,
3499			   const struct qstr *entry, bool quick);
3500extern int __ext4_unlink(struct inode *dir, const struct qstr *d_name,
3501			 struct inode *inode, struct dentry *dentry);
3502extern int __ext4_link(struct inode *dir, struct inode *inode,
3503		       struct dentry *dentry);
3504
3505#define S_SHIFT 12
3506static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = {
3507	[S_IFREG >> S_SHIFT]	= EXT4_FT_REG_FILE,
3508	[S_IFDIR >> S_SHIFT]	= EXT4_FT_DIR,
3509	[S_IFCHR >> S_SHIFT]	= EXT4_FT_CHRDEV,
3510	[S_IFBLK >> S_SHIFT]	= EXT4_FT_BLKDEV,
3511	[S_IFIFO >> S_SHIFT]	= EXT4_FT_FIFO,
3512	[S_IFSOCK >> S_SHIFT]	= EXT4_FT_SOCK,
3513	[S_IFLNK >> S_SHIFT]	= EXT4_FT_SYMLINK,
3514};
3515
3516static inline void ext4_set_de_type(struct super_block *sb,
3517				struct ext4_dir_entry_2 *de,
3518				umode_t mode) {
3519	if (ext4_has_feature_filetype(sb))
3520		de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3521}
3522
3523/* readpages.c */
3524extern int ext4_mpage_readpages(struct inode *inode,
3525		struct readahead_control *rac, struct page *page);
3526extern int __init ext4_init_post_read_processing(void);
3527extern void ext4_exit_post_read_processing(void);
3528
3529/* symlink.c */
3530extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3531extern const struct inode_operations ext4_symlink_inode_operations;
3532extern const struct inode_operations ext4_fast_symlink_inode_operations;
3533
3534/* sysfs.c */
3535extern int ext4_register_sysfs(struct super_block *sb);
3536extern void ext4_unregister_sysfs(struct super_block *sb);
3537extern int __init ext4_init_sysfs(void);
3538extern void ext4_exit_sysfs(void);
3539
3540/* block_validity */
3541extern void ext4_release_system_zone(struct super_block *sb);
3542extern int ext4_setup_system_zone(struct super_block *sb);
3543extern int __init ext4_init_system_zone(void);
3544extern void ext4_exit_system_zone(void);
3545extern int ext4_inode_block_valid(struct inode *inode,
3546				  ext4_fsblk_t start_blk,
3547				  unsigned int count);
3548extern int ext4_check_blockref(const char *, unsigned int,
3549			       struct inode *, __le32 *, unsigned int);
3550extern int ext4_sb_block_valid(struct super_block *sb, struct inode *inode,
3551				ext4_fsblk_t start_blk, unsigned int count);
3552
3553
3554/* extents.c */
3555struct ext4_ext_path;
3556struct ext4_extent;
3557
3558/*
3559 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3560 * __le32.
3561 */
3562#define EXT_MAX_BLOCKS	0xffffffff
3563
3564extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode);
3565extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3566extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3567			       struct ext4_map_blocks *map, int flags);
3568extern int ext4_ext_truncate(handle_t *, struct inode *);
3569extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3570				 ext4_lblk_t end);
3571extern void ext4_ext_init(struct super_block *);
3572extern void ext4_ext_release(struct super_block *);
3573extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3574			  loff_t len);
3575extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3576					  loff_t offset, ssize_t len);
3577extern int ext4_convert_unwritten_io_end_vec(handle_t *handle,
3578					     ext4_io_end_t *io_end);
3579extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3580			   struct ext4_map_blocks *map, int flags);
3581extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3582						   int num,
3583						   struct ext4_ext_path *path);
3584extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3585				  struct ext4_ext_path **,
3586				  struct ext4_extent *, int);
3587extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3588					      struct ext4_ext_path **,
3589					      int flags);
3590extern void ext4_ext_drop_refs(struct ext4_ext_path *);
3591extern int ext4_ext_check_inode(struct inode *inode);
3592extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3593extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3594			__u64 start, __u64 len);
3595extern int ext4_get_es_cache(struct inode *inode,
3596			     struct fiemap_extent_info *fieinfo,
3597			     __u64 start, __u64 len);
3598extern int ext4_ext_precache(struct inode *inode);
3599extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3600				struct inode *inode2, ext4_lblk_t lblk1,
3601			     ext4_lblk_t lblk2,  ext4_lblk_t count,
3602			     int mark_unwritten,int *err);
3603extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu);
3604extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode,
3605				       int check_cred, int restart_cred,
3606				       int revoke_cred);
3607extern void ext4_ext_replay_shrink_inode(struct inode *inode, ext4_lblk_t end);
3608extern int ext4_ext_replay_set_iblocks(struct inode *inode);
3609extern int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start,
3610		int len, int unwritten, ext4_fsblk_t pblk);
3611extern int ext4_ext_clear_bb(struct inode *inode);
3612
3613
3614/* move_extent.c */
3615extern void ext4_double_down_write_data_sem(struct inode *first,
3616					    struct inode *second);
3617extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3618					  struct inode *donor_inode);
3619extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3620			     __u64 start_orig, __u64 start_donor,
3621			     __u64 len, __u64 *moved_len);
3622
3623/* page-io.c */
3624extern int __init ext4_init_pageio(void);
3625extern void ext4_exit_pageio(void);
3626extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3627extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3628extern int ext4_put_io_end(ext4_io_end_t *io_end);
3629extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3630extern void ext4_io_submit_init(struct ext4_io_submit *io,
3631				struct writeback_control *wbc);
3632extern void ext4_end_io_rsv_work(struct work_struct *work);
3633extern void ext4_io_submit(struct ext4_io_submit *io);
3634extern int ext4_bio_write_page(struct ext4_io_submit *io,
3635			       struct page *page,
3636			       int len,
3637			       struct writeback_control *wbc,
3638			       bool keep_towrite);
3639extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end);
3640extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end);
3641
3642/* mmp.c */
3643extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3644
3645/* mmp.c */
3646extern void ext4_stop_mmpd(struct ext4_sb_info *sbi);
3647
3648/* verity.c */
3649extern const struct fsverity_operations ext4_verityops;
3650
3651/*
3652 * Add new method to test whether block and inode bitmaps are properly
3653 * initialized. With uninit_bg reading the block from disk is not enough
3654 * to mark the bitmap uptodate. We need to also zero-out the bitmap
3655 */
3656#define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3657
3658static inline int bitmap_uptodate(struct buffer_head *bh)
3659{
3660	return (buffer_uptodate(bh) &&
3661			test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3662}
3663static inline void set_bitmap_uptodate(struct buffer_head *bh)
3664{
3665	set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3666}
3667
3668#define in_range(b, first, len)	((b) >= (first) && (b) <= (first) + (len) - 1)
3669
3670/* For ioend & aio unwritten conversion wait queues */
3671#define EXT4_WQ_HASH_SZ		37
3672#define ext4_ioend_wq(v)   (&ext4__ioend_wq[((unsigned long)(v)) %\
3673					    EXT4_WQ_HASH_SZ])
3674extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3675
3676extern int ext4_resize_begin(struct super_block *sb);
3677extern void ext4_resize_end(struct super_block *sb);
3678
3679static inline void ext4_set_io_unwritten_flag(struct inode *inode,
3680					      struct ext4_io_end *io_end)
3681{
3682	if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
3683		io_end->flag |= EXT4_IO_END_UNWRITTEN;
3684		atomic_inc(&EXT4_I(inode)->i_unwritten);
3685	}
3686}
3687
3688static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3689{
3690	struct inode *inode = io_end->inode;
3691
3692	if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
3693		io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3694		/* Wake up anyone waiting on unwritten extent conversion */
3695		if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
3696			wake_up_all(ext4_ioend_wq(inode));
3697	}
3698}
3699
3700extern const struct iomap_ops ext4_iomap_ops;
3701extern const struct iomap_ops ext4_iomap_overwrite_ops;
3702extern const struct iomap_ops ext4_iomap_report_ops;
3703
3704static inline int ext4_buffer_uptodate(struct buffer_head *bh)
3705{
3706	/*
3707	 * If the buffer has the write error flag, we have failed
3708	 * to write out data in the block.  In this  case, we don't
3709	 * have to read the block because we may read the old data
3710	 * successfully.
3711	 */
3712	if (!buffer_uptodate(bh) && buffer_write_io_error(bh))
3713		set_buffer_uptodate(bh);
3714	return buffer_uptodate(bh);
3715}
3716
3717#endif	/* __KERNEL__ */
3718
3719#define EFSBADCRC	EBADMSG		/* Bad CRC detected */
3720#define EFSCORRUPTED	EUCLEAN		/* Filesystem is corrupted */
3721
3722#endif	/* _EXT4_H */
3723