ext4.h revision 4a79a98c
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#define ASSERT(assert)						\
102do {									\
103	if (unlikely(!(assert))) {					\
104		printk(KERN_EMERG					\
105		       "Assertion failure in %s() at %s:%d: '%s'\n",	\
106		       __func__, __FILE__, __LINE__, #assert);		\
107		BUG();							\
108	}								\
109} while (0)
110
111/* data type for block offset of block group */
112typedef int ext4_grpblk_t;
113
114/* data type for filesystem-wide blocks number */
115typedef unsigned long long ext4_fsblk_t;
116
117/* data type for file logical block number */
118typedef __u32 ext4_lblk_t;
119
120/* data type for block group number */
121typedef unsigned int ext4_group_t;
122
123enum SHIFT_DIRECTION {
124	SHIFT_LEFT = 0,
125	SHIFT_RIGHT,
126};
127
128/*
129 * Flags used in mballoc's allocation_context flags field.
130 *
131 * Also used to show what's going on for debugging purposes when the
132 * flag field is exported via the traceport interface
133 */
134
135/* prefer goal again. length */
136#define EXT4_MB_HINT_MERGE		0x0001
137/* blocks already reserved */
138#define EXT4_MB_HINT_RESERVED		0x0002
139/* metadata is being allocated */
140#define EXT4_MB_HINT_METADATA		0x0004
141/* first blocks in the file */
142#define EXT4_MB_HINT_FIRST		0x0008
143/* search for the best chunk */
144#define EXT4_MB_HINT_BEST		0x0010
145/* data is being allocated */
146#define EXT4_MB_HINT_DATA		0x0020
147/* don't preallocate (for tails) */
148#define EXT4_MB_HINT_NOPREALLOC		0x0040
149/* allocate for locality group */
150#define EXT4_MB_HINT_GROUP_ALLOC	0x0080
151/* allocate goal blocks or none */
152#define EXT4_MB_HINT_GOAL_ONLY		0x0100
153/* goal is meaningful */
154#define EXT4_MB_HINT_TRY_GOAL		0x0200
155/* blocks already pre-reserved by delayed allocation */
156#define EXT4_MB_DELALLOC_RESERVED	0x0400
157/* We are doing stream allocation */
158#define EXT4_MB_STREAM_ALLOC		0x0800
159/* Use reserved root blocks if needed */
160#define EXT4_MB_USE_ROOT_BLOCKS		0x1000
161/* Use blocks from reserved pool */
162#define EXT4_MB_USE_RESERVED		0x2000
163/* Do strict check for free blocks while retrying block allocation */
164#define EXT4_MB_STRICT_CHECK		0x4000
165/* Large fragment size list lookup succeeded at least once for cr = 0 */
166#define EXT4_MB_CR0_OPTIMIZED		0x8000
167/* Avg fragment size rb tree lookup succeeded at least once for cr = 1 */
168#define EXT4_MB_CR1_OPTIMIZED		0x00010000
169/* Perform linear traversal for one group */
170#define EXT4_MB_SEARCH_NEXT_LINEAR	0x00020000
171struct ext4_allocation_request {
172	/* target inode for block we're allocating */
173	struct inode *inode;
174	/* how many blocks we want to allocate */
175	unsigned int len;
176	/* logical block in target inode */
177	ext4_lblk_t logical;
178	/* the closest logical allocated block to the left */
179	ext4_lblk_t lleft;
180	/* the closest logical allocated block to the right */
181	ext4_lblk_t lright;
182	/* phys. target (a hint) */
183	ext4_fsblk_t goal;
184	/* phys. block for the closest logical allocated block to the left */
185	ext4_fsblk_t pleft;
186	/* phys. block for the closest logical allocated block to the right */
187	ext4_fsblk_t pright;
188	/* flags. see above EXT4_MB_HINT_* */
189	unsigned int flags;
190};
191
192/*
193 * Logical to physical block mapping, used by ext4_map_blocks()
194 *
195 * This structure is used to pass requests into ext4_map_blocks() as
196 * well as to store the information returned by ext4_map_blocks().  It
197 * takes less room on the stack than a struct buffer_head.
198 */
199#define EXT4_MAP_NEW		BIT(BH_New)
200#define EXT4_MAP_MAPPED		BIT(BH_Mapped)
201#define EXT4_MAP_UNWRITTEN	BIT(BH_Unwritten)
202#define EXT4_MAP_BOUNDARY	BIT(BH_Boundary)
203#define EXT4_MAP_FLAGS		(EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
204				 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
205
206struct ext4_map_blocks {
207	ext4_fsblk_t m_pblk;
208	ext4_lblk_t m_lblk;
209	unsigned int m_len;
210	unsigned int m_flags;
211};
212
213/*
214 * Block validity checking, system zone rbtree.
215 */
216struct ext4_system_blocks {
217	struct rb_root root;
218	struct rcu_head rcu;
219};
220
221/*
222 * Flags for ext4_io_end->flags
223 */
224#define	EXT4_IO_END_UNWRITTEN	0x0001
225
226struct ext4_io_end_vec {
227	struct list_head list;		/* list of io_end_vec */
228	loff_t offset;			/* offset in the file */
229	ssize_t size;			/* size of the extent */
230};
231
232/*
233 * For converting unwritten extents on a work queue. 'handle' is used for
234 * buffered writeback.
235 */
236typedef struct ext4_io_end {
237	struct list_head	list;		/* per-file finished IO list */
238	handle_t		*handle;	/* handle reserved for extent
239						 * conversion */
240	struct inode		*inode;		/* file being written to */
241	struct bio		*bio;		/* Linked list of completed
242						 * bios covering the extent */
243	unsigned int		flag;		/* unwritten or not */
244	atomic_t		count;		/* reference counter */
245	struct list_head	list_vec;	/* list of ext4_io_end_vec */
246} ext4_io_end_t;
247
248struct ext4_io_submit {
249	struct writeback_control *io_wbc;
250	struct bio		*io_bio;
251	ext4_io_end_t		*io_end;
252	sector_t		io_next_block;
253};
254
255/*
256 * Special inodes numbers
257 */
258#define	EXT4_BAD_INO		 1	/* Bad blocks inode */
259#define EXT4_ROOT_INO		 2	/* Root inode */
260#define EXT4_USR_QUOTA_INO	 3	/* User quota inode */
261#define EXT4_GRP_QUOTA_INO	 4	/* Group quota inode */
262#define EXT4_BOOT_LOADER_INO	 5	/* Boot loader inode */
263#define EXT4_UNDEL_DIR_INO	 6	/* Undelete directory inode */
264#define EXT4_RESIZE_INO		 7	/* Reserved group descriptors inode */
265#define EXT4_JOURNAL_INO	 8	/* Journal inode */
266
267/* First non-reserved inode for old ext4 filesystems */
268#define EXT4_GOOD_OLD_FIRST_INO	11
269
270/*
271 * Maximal count of links to a file
272 */
273#define EXT4_LINK_MAX		65000
274
275/*
276 * Macro-instructions used to manage several block sizes
277 */
278#define EXT4_MIN_BLOCK_SIZE		1024
279#define	EXT4_MAX_BLOCK_SIZE		65536
280#define EXT4_MIN_BLOCK_LOG_SIZE		10
281#define EXT4_MAX_BLOCK_LOG_SIZE		16
282#define EXT4_MAX_CLUSTER_LOG_SIZE	30
283#ifdef __KERNEL__
284# define EXT4_BLOCK_SIZE(s)		((s)->s_blocksize)
285#else
286# define EXT4_BLOCK_SIZE(s)		(EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
287#endif
288#define	EXT4_ADDR_PER_BLOCK(s)		(EXT4_BLOCK_SIZE(s) / sizeof(__u32))
289#define EXT4_CLUSTER_SIZE(s)		(EXT4_BLOCK_SIZE(s) << \
290					 EXT4_SB(s)->s_cluster_bits)
291#ifdef __KERNEL__
292# define EXT4_BLOCK_SIZE_BITS(s)	((s)->s_blocksize_bits)
293# define EXT4_CLUSTER_BITS(s)		(EXT4_SB(s)->s_cluster_bits)
294#else
295# define EXT4_BLOCK_SIZE_BITS(s)	((s)->s_log_block_size + 10)
296#endif
297#ifdef __KERNEL__
298#define	EXT4_ADDR_PER_BLOCK_BITS(s)	(EXT4_SB(s)->s_addr_per_block_bits)
299#define EXT4_INODE_SIZE(s)		(EXT4_SB(s)->s_inode_size)
300#define EXT4_FIRST_INO(s)		(EXT4_SB(s)->s_first_ino)
301#else
302#define EXT4_INODE_SIZE(s)	(((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
303				 EXT4_GOOD_OLD_INODE_SIZE : \
304				 (s)->s_inode_size)
305#define EXT4_FIRST_INO(s)	(((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
306				 EXT4_GOOD_OLD_FIRST_INO : \
307				 (s)->s_first_ino)
308#endif
309#define EXT4_BLOCK_ALIGN(size, blkbits)		ALIGN((size), (1 << (blkbits)))
310#define EXT4_MAX_BLOCKS(size, offset, blkbits) \
311	((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \
312								  blkbits))
313
314/* Translate a block number to a cluster number */
315#define EXT4_B2C(sbi, blk)	((blk) >> (sbi)->s_cluster_bits)
316/* Translate a cluster number to a block number */
317#define EXT4_C2B(sbi, cluster)	((cluster) << (sbi)->s_cluster_bits)
318/* Translate # of blks to # of clusters */
319#define EXT4_NUM_B2C(sbi, blks)	(((blks) + (sbi)->s_cluster_ratio - 1) >> \
320				 (sbi)->s_cluster_bits)
321/* Mask out the low bits to get the starting block of the cluster */
322#define EXT4_PBLK_CMASK(s, pblk) ((pblk) &				\
323				  ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
324#define EXT4_LBLK_CMASK(s, lblk) ((lblk) &				\
325				  ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
326/* Fill in the low bits to get the last block of the cluster */
327#define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) |				\
328				    ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1))
329/* Get the cluster offset */
330#define EXT4_PBLK_COFF(s, pblk) ((pblk) &				\
331				 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
332#define EXT4_LBLK_COFF(s, lblk) ((lblk) &				\
333				 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
334
335/*
336 * Structure of a blocks group descriptor
337 */
338struct ext4_group_desc
339{
340	__le32	bg_block_bitmap_lo;	/* Blocks bitmap block */
341	__le32	bg_inode_bitmap_lo;	/* Inodes bitmap block */
342	__le32	bg_inode_table_lo;	/* Inodes table block */
343	__le16	bg_free_blocks_count_lo;/* Free blocks count */
344	__le16	bg_free_inodes_count_lo;/* Free inodes count */
345	__le16	bg_used_dirs_count_lo;	/* Directories count */
346	__le16	bg_flags;		/* EXT4_BG_flags (INODE_UNINIT, etc) */
347	__le32  bg_exclude_bitmap_lo;   /* Exclude bitmap for snapshots */
348	__le16  bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
349	__le16  bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
350	__le16  bg_itable_unused_lo;	/* Unused inodes count */
351	__le16  bg_checksum;		/* crc16(sb_uuid+group+desc) */
352	__le32	bg_block_bitmap_hi;	/* Blocks bitmap block MSB */
353	__le32	bg_inode_bitmap_hi;	/* Inodes bitmap block MSB */
354	__le32	bg_inode_table_hi;	/* Inodes table block MSB */
355	__le16	bg_free_blocks_count_hi;/* Free blocks count MSB */
356	__le16	bg_free_inodes_count_hi;/* Free inodes count MSB */
357	__le16	bg_used_dirs_count_hi;	/* Directories count MSB */
358	__le16  bg_itable_unused_hi;    /* Unused inodes count MSB */
359	__le32  bg_exclude_bitmap_hi;   /* Exclude bitmap block MSB */
360	__le16  bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
361	__le16  bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
362	__u32   bg_reserved;
363};
364
365#define EXT4_BG_INODE_BITMAP_CSUM_HI_END	\
366	(offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
367	 sizeof(__le16))
368#define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END	\
369	(offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
370	 sizeof(__le16))
371
372/*
373 * Structure of a flex block group info
374 */
375
376struct flex_groups {
377	atomic64_t	free_clusters;
378	atomic_t	free_inodes;
379	atomic_t	used_dirs;
380};
381
382#define EXT4_BG_INODE_UNINIT	0x0001 /* Inode table/bitmap not in use */
383#define EXT4_BG_BLOCK_UNINIT	0x0002 /* Block bitmap not in use */
384#define EXT4_BG_INODE_ZEROED	0x0004 /* On-disk itable initialized to zero */
385
386/*
387 * Macro-instructions used to manage group descriptors
388 */
389#define EXT4_MIN_DESC_SIZE		32
390#define EXT4_MIN_DESC_SIZE_64BIT	64
391#define	EXT4_MAX_DESC_SIZE		EXT4_MIN_BLOCK_SIZE
392#define EXT4_DESC_SIZE(s)		(EXT4_SB(s)->s_desc_size)
393#ifdef __KERNEL__
394# define EXT4_BLOCKS_PER_GROUP(s)	(EXT4_SB(s)->s_blocks_per_group)
395# define EXT4_CLUSTERS_PER_GROUP(s)	(EXT4_SB(s)->s_clusters_per_group)
396# define EXT4_DESC_PER_BLOCK(s)		(EXT4_SB(s)->s_desc_per_block)
397# define EXT4_INODES_PER_GROUP(s)	(EXT4_SB(s)->s_inodes_per_group)
398# define EXT4_DESC_PER_BLOCK_BITS(s)	(EXT4_SB(s)->s_desc_per_block_bits)
399#else
400# define EXT4_BLOCKS_PER_GROUP(s)	((s)->s_blocks_per_group)
401# define EXT4_DESC_PER_BLOCK(s)		(EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
402# define EXT4_INODES_PER_GROUP(s)	((s)->s_inodes_per_group)
403#endif
404
405/*
406 * Constants relative to the data blocks
407 */
408#define	EXT4_NDIR_BLOCKS		12
409#define	EXT4_IND_BLOCK			EXT4_NDIR_BLOCKS
410#define	EXT4_DIND_BLOCK			(EXT4_IND_BLOCK + 1)
411#define	EXT4_TIND_BLOCK			(EXT4_DIND_BLOCK + 1)
412#define	EXT4_N_BLOCKS			(EXT4_TIND_BLOCK + 1)
413
414/*
415 * Inode flags
416 */
417#define	EXT4_SECRM_FL			0x00000001 /* Secure deletion */
418#define	EXT4_UNRM_FL			0x00000002 /* Undelete */
419#define	EXT4_COMPR_FL			0x00000004 /* Compress file */
420#define EXT4_SYNC_FL			0x00000008 /* Synchronous updates */
421#define EXT4_IMMUTABLE_FL		0x00000010 /* Immutable file */
422#define EXT4_APPEND_FL			0x00000020 /* writes to file may only append */
423#define EXT4_NODUMP_FL			0x00000040 /* do not dump file */
424#define EXT4_NOATIME_FL			0x00000080 /* do not update atime */
425/* Reserved for compression usage... */
426#define EXT4_DIRTY_FL			0x00000100
427#define EXT4_COMPRBLK_FL		0x00000200 /* One or more compressed clusters */
428#define EXT4_NOCOMPR_FL			0x00000400 /* Don't compress */
429	/* nb: was previously EXT2_ECOMPR_FL */
430#define EXT4_ENCRYPT_FL			0x00000800 /* encrypted file */
431/* End compression flags --- maybe not all used */
432#define EXT4_INDEX_FL			0x00001000 /* hash-indexed directory */
433#define EXT4_IMAGIC_FL			0x00002000 /* AFS directory */
434#define EXT4_JOURNAL_DATA_FL		0x00004000 /* file data should be journaled */
435#define EXT4_NOTAIL_FL			0x00008000 /* file tail should not be merged */
436#define EXT4_DIRSYNC_FL			0x00010000 /* dirsync behaviour (directories only) */
437#define EXT4_TOPDIR_FL			0x00020000 /* Top of directory hierarchies*/
438#define EXT4_HUGE_FILE_FL               0x00040000 /* Set to each huge file */
439#define EXT4_EXTENTS_FL			0x00080000 /* Inode uses extents */
440#define EXT4_VERITY_FL			0x00100000 /* Verity protected inode */
441#define EXT4_EA_INODE_FL	        0x00200000 /* Inode used for large EA */
442/* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */
443
444#define EXT4_DAX_FL			0x02000000 /* Inode is DAX */
445
446#define EXT4_INLINE_DATA_FL		0x10000000 /* Inode has inline data. */
447#define EXT4_PROJINHERIT_FL		0x20000000 /* Create with parents projid */
448#define EXT4_CASEFOLD_FL		0x40000000 /* Casefolded directory */
449#define EXT4_RESERVED_FL		0x80000000 /* reserved for ext4 lib */
450
451/* User modifiable flags */
452#define EXT4_FL_USER_MODIFIABLE		(EXT4_SECRM_FL | \
453					 EXT4_UNRM_FL | \
454					 EXT4_COMPR_FL | \
455					 EXT4_SYNC_FL | \
456					 EXT4_IMMUTABLE_FL | \
457					 EXT4_APPEND_FL | \
458					 EXT4_NODUMP_FL | \
459					 EXT4_NOATIME_FL | \
460					 EXT4_JOURNAL_DATA_FL | \
461					 EXT4_NOTAIL_FL | \
462					 EXT4_DIRSYNC_FL | \
463					 EXT4_TOPDIR_FL | \
464					 EXT4_EXTENTS_FL | \
465					 0x00400000 /* EXT4_EOFBLOCKS_FL */ | \
466					 EXT4_DAX_FL | \
467					 EXT4_PROJINHERIT_FL | \
468					 EXT4_CASEFOLD_FL)
469
470/* User visible flags */
471#define EXT4_FL_USER_VISIBLE		(EXT4_FL_USER_MODIFIABLE | \
472					 EXT4_DIRTY_FL | \
473					 EXT4_COMPRBLK_FL | \
474					 EXT4_NOCOMPR_FL | \
475					 EXT4_ENCRYPT_FL | \
476					 EXT4_INDEX_FL | \
477					 EXT4_VERITY_FL | \
478					 EXT4_INLINE_DATA_FL)
479
480/* Flags that should be inherited by new inodes from their parent. */
481#define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
482			   EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
483			   EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
484			   EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
485			   EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL |\
486			   EXT4_DAX_FL)
487
488/* Flags that are appropriate for regular files (all but dir-specific ones). */
489#define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\
490			   EXT4_PROJINHERIT_FL))
491
492/* Flags that are appropriate for non-directories/regular files. */
493#define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
494
495/* The only flags that should be swapped */
496#define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL)
497
498/* Flags which are mutually exclusive to DAX */
499#define EXT4_DAX_MUT_EXCL (EXT4_VERITY_FL | EXT4_ENCRYPT_FL |\
500			   EXT4_JOURNAL_DATA_FL | EXT4_INLINE_DATA_FL)
501
502/* Mask out flags that are inappropriate for the given type of inode. */
503static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
504{
505	if (S_ISDIR(mode))
506		return flags;
507	else if (S_ISREG(mode))
508		return flags & EXT4_REG_FLMASK;
509	else
510		return flags & EXT4_OTHER_FLMASK;
511}
512
513/*
514 * Inode flags used for atomic set/get
515 */
516enum {
517	EXT4_INODE_SECRM	= 0,	/* Secure deletion */
518	EXT4_INODE_UNRM		= 1,	/* Undelete */
519	EXT4_INODE_COMPR	= 2,	/* Compress file */
520	EXT4_INODE_SYNC		= 3,	/* Synchronous updates */
521	EXT4_INODE_IMMUTABLE	= 4,	/* Immutable file */
522	EXT4_INODE_APPEND	= 5,	/* writes to file may only append */
523	EXT4_INODE_NODUMP	= 6,	/* do not dump file */
524	EXT4_INODE_NOATIME	= 7,	/* do not update atime */
525/* Reserved for compression usage... */
526	EXT4_INODE_DIRTY	= 8,
527	EXT4_INODE_COMPRBLK	= 9,	/* One or more compressed clusters */
528	EXT4_INODE_NOCOMPR	= 10,	/* Don't compress */
529	EXT4_INODE_ENCRYPT	= 11,	/* Encrypted file */
530/* End compression flags --- maybe not all used */
531	EXT4_INODE_INDEX	= 12,	/* hash-indexed directory */
532	EXT4_INODE_IMAGIC	= 13,	/* AFS directory */
533	EXT4_INODE_JOURNAL_DATA	= 14,	/* file data should be journaled */
534	EXT4_INODE_NOTAIL	= 15,	/* file tail should not be merged */
535	EXT4_INODE_DIRSYNC	= 16,	/* dirsync behaviour (directories only) */
536	EXT4_INODE_TOPDIR	= 17,	/* Top of directory hierarchies*/
537	EXT4_INODE_HUGE_FILE	= 18,	/* Set to each huge file */
538	EXT4_INODE_EXTENTS	= 19,	/* Inode uses extents */
539	EXT4_INODE_VERITY	= 20,	/* Verity protected inode */
540	EXT4_INODE_EA_INODE	= 21,	/* Inode used for large EA */
541/* 22 was formerly EXT4_INODE_EOFBLOCKS */
542	EXT4_INODE_DAX		= 25,	/* Inode is DAX */
543	EXT4_INODE_INLINE_DATA	= 28,	/* Data in inode. */
544	EXT4_INODE_PROJINHERIT	= 29,	/* Create with parents projid */
545	EXT4_INODE_CASEFOLD	= 30,	/* Casefolded directory */
546	EXT4_INODE_RESERVED	= 31,	/* reserved for ext4 lib */
547};
548
549/*
550 * Since it's pretty easy to mix up bit numbers and hex values, we use a
551 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
552 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
553 * any extra space in the compiled kernel image, otherwise, the build will fail.
554 * It's important that these values are the same, since we are using
555 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
556 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
557 * values found in ext2, ext3 and ext4 filesystems, and of course the values
558 * defined in e2fsprogs.
559 *
560 * It's not paranoia if the Murphy's Law really *is* out to get you.  :-)
561 */
562#define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
563#define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
564
565static inline void ext4_check_flag_values(void)
566{
567	CHECK_FLAG_VALUE(SECRM);
568	CHECK_FLAG_VALUE(UNRM);
569	CHECK_FLAG_VALUE(COMPR);
570	CHECK_FLAG_VALUE(SYNC);
571	CHECK_FLAG_VALUE(IMMUTABLE);
572	CHECK_FLAG_VALUE(APPEND);
573	CHECK_FLAG_VALUE(NODUMP);
574	CHECK_FLAG_VALUE(NOATIME);
575	CHECK_FLAG_VALUE(DIRTY);
576	CHECK_FLAG_VALUE(COMPRBLK);
577	CHECK_FLAG_VALUE(NOCOMPR);
578	CHECK_FLAG_VALUE(ENCRYPT);
579	CHECK_FLAG_VALUE(INDEX);
580	CHECK_FLAG_VALUE(IMAGIC);
581	CHECK_FLAG_VALUE(JOURNAL_DATA);
582	CHECK_FLAG_VALUE(NOTAIL);
583	CHECK_FLAG_VALUE(DIRSYNC);
584	CHECK_FLAG_VALUE(TOPDIR);
585	CHECK_FLAG_VALUE(HUGE_FILE);
586	CHECK_FLAG_VALUE(EXTENTS);
587	CHECK_FLAG_VALUE(VERITY);
588	CHECK_FLAG_VALUE(EA_INODE);
589	CHECK_FLAG_VALUE(INLINE_DATA);
590	CHECK_FLAG_VALUE(PROJINHERIT);
591	CHECK_FLAG_VALUE(CASEFOLD);
592	CHECK_FLAG_VALUE(RESERVED);
593}
594
595/* Used to pass group descriptor data when online resize is done */
596struct ext4_new_group_input {
597	__u32 group;		/* Group number for this data */
598	__u64 block_bitmap;	/* Absolute block number of block bitmap */
599	__u64 inode_bitmap;	/* Absolute block number of inode bitmap */
600	__u64 inode_table;	/* Absolute block number of inode table start */
601	__u32 blocks_count;	/* Total number of blocks in this group */
602	__u16 reserved_blocks;	/* Number of reserved blocks in this group */
603	__u16 unused;
604};
605
606#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
607struct compat_ext4_new_group_input {
608	u32 group;
609	compat_u64 block_bitmap;
610	compat_u64 inode_bitmap;
611	compat_u64 inode_table;
612	u32 blocks_count;
613	u16 reserved_blocks;
614	u16 unused;
615};
616#endif
617
618/* The struct ext4_new_group_input in kernel space, with free_blocks_count */
619struct ext4_new_group_data {
620	__u32 group;
621	__u64 block_bitmap;
622	__u64 inode_bitmap;
623	__u64 inode_table;
624	__u32 blocks_count;
625	__u16 reserved_blocks;
626	__u16 mdata_blocks;
627	__u32 free_clusters_count;
628};
629
630/* Indexes used to index group tables in ext4_new_group_data */
631enum {
632	BLOCK_BITMAP = 0,	/* block bitmap */
633	INODE_BITMAP,		/* inode bitmap */
634	INODE_TABLE,		/* inode tables */
635	GROUP_TABLE_COUNT,
636};
637
638/*
639 * Flags used by ext4_map_blocks()
640 */
641	/* Allocate any needed blocks and/or convert an unwritten
642	   extent to be an initialized ext4 */
643#define EXT4_GET_BLOCKS_CREATE			0x0001
644	/* Request the creation of an unwritten extent */
645#define EXT4_GET_BLOCKS_UNWRIT_EXT		0x0002
646#define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT	(EXT4_GET_BLOCKS_UNWRIT_EXT|\
647						 EXT4_GET_BLOCKS_CREATE)
648	/* Caller is from the delayed allocation writeout path
649	 * finally doing the actual allocation of delayed blocks */
650#define EXT4_GET_BLOCKS_DELALLOC_RESERVE	0x0004
651	/* caller is from the direct IO path, request to creation of an
652	unwritten extents if not allocated, split the unwritten
653	extent if blocks has been preallocated already*/
654#define EXT4_GET_BLOCKS_PRE_IO			0x0008
655#define EXT4_GET_BLOCKS_CONVERT			0x0010
656#define EXT4_GET_BLOCKS_IO_CREATE_EXT		(EXT4_GET_BLOCKS_PRE_IO|\
657					 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
658	/* Convert extent to initialized after IO complete */
659#define EXT4_GET_BLOCKS_IO_CONVERT_EXT		(EXT4_GET_BLOCKS_CONVERT|\
660					 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
661	/* Eventual metadata allocation (due to growing extent tree)
662	 * should not fail, so try to use reserved blocks for that.*/
663#define EXT4_GET_BLOCKS_METADATA_NOFAIL		0x0020
664	/* Don't normalize allocation size (used for fallocate) */
665#define EXT4_GET_BLOCKS_NO_NORMALIZE		0x0040
666	/* Convert written extents to unwritten */
667#define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN	0x0100
668	/* Write zeros to newly created written extents */
669#define EXT4_GET_BLOCKS_ZERO			0x0200
670#define EXT4_GET_BLOCKS_CREATE_ZERO		(EXT4_GET_BLOCKS_CREATE |\
671					EXT4_GET_BLOCKS_ZERO)
672	/* Caller will submit data before dropping transaction handle. This
673	 * allows jbd2 to avoid submitting data before commit. */
674#define EXT4_GET_BLOCKS_IO_SUBMIT		0x0400
675
676/*
677 * The bit position of these flags must not overlap with any of the
678 * EXT4_GET_BLOCKS_*.  They are used by ext4_find_extent(),
679 * read_extent_tree_block(), ext4_split_extent_at(),
680 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
681 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
682 * caching the extents when reading from the extent tree while a
683 * truncate or punch hole operation is in progress.
684 */
685#define EXT4_EX_NOCACHE				0x40000000
686#define EXT4_EX_FORCE_CACHE			0x20000000
687#define EXT4_EX_NOFAIL				0x10000000
688
689/*
690 * Flags used by ext4_free_blocks
691 */
692#define EXT4_FREE_BLOCKS_METADATA		0x0001
693#define EXT4_FREE_BLOCKS_FORGET			0x0002
694#define EXT4_FREE_BLOCKS_VALIDATED		0x0004
695#define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE		0x0008
696#define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER	0x0010
697#define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER	0x0020
698#define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER      0x0040
699
700/*
701 * ioctl commands
702 */
703#define	EXT4_IOC_GETVERSION		_IOR('f', 3, long)
704#define	EXT4_IOC_SETVERSION		_IOW('f', 4, long)
705#define	EXT4_IOC_GETVERSION_OLD		FS_IOC_GETVERSION
706#define	EXT4_IOC_SETVERSION_OLD		FS_IOC_SETVERSION
707#define EXT4_IOC_GETRSVSZ		_IOR('f', 5, long)
708#define EXT4_IOC_SETRSVSZ		_IOW('f', 6, long)
709#define EXT4_IOC_GROUP_EXTEND		_IOW('f', 7, unsigned long)
710#define EXT4_IOC_GROUP_ADD		_IOW('f', 8, struct ext4_new_group_input)
711#define EXT4_IOC_MIGRATE		_IO('f', 9)
712 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
713 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
714#define EXT4_IOC_ALLOC_DA_BLKS		_IO('f', 12)
715#define EXT4_IOC_MOVE_EXT		_IOWR('f', 15, struct move_extent)
716#define EXT4_IOC_RESIZE_FS		_IOW('f', 16, __u64)
717#define EXT4_IOC_SWAP_BOOT		_IO('f', 17)
718#define EXT4_IOC_PRECACHE_EXTENTS	_IO('f', 18)
719/* ioctl codes 19--39 are reserved for fscrypt */
720#define EXT4_IOC_CLEAR_ES_CACHE		_IO('f', 40)
721#define EXT4_IOC_GETSTATE		_IOW('f', 41, __u32)
722#define EXT4_IOC_GET_ES_CACHE		_IOWR('f', 42, struct fiemap)
723#define EXT4_IOC_CHECKPOINT		_IOW('f', 43, __u32)
724
725#define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32)
726
727/*
728 * Flags for going down operation
729 */
730#define EXT4_GOING_FLAGS_DEFAULT		0x0	/* going down */
731#define EXT4_GOING_FLAGS_LOGFLUSH		0x1	/* flush log but not data */
732#define EXT4_GOING_FLAGS_NOLOGFLUSH		0x2	/* don't flush log nor data */
733
734/*
735 * Flags returned by EXT4_IOC_GETSTATE
736 *
737 * We only expose to userspace a subset of the state flags in
738 * i_state_flags
739 */
740#define EXT4_STATE_FLAG_EXT_PRECACHED	0x00000001
741#define EXT4_STATE_FLAG_NEW		0x00000002
742#define EXT4_STATE_FLAG_NEWENTRY	0x00000004
743#define EXT4_STATE_FLAG_DA_ALLOC_CLOSE	0x00000008
744
745/* flags for ioctl EXT4_IOC_CHECKPOINT */
746#define EXT4_IOC_CHECKPOINT_FLAG_DISCARD	0x1
747#define EXT4_IOC_CHECKPOINT_FLAG_ZEROOUT	0x2
748#define EXT4_IOC_CHECKPOINT_FLAG_DRY_RUN	0x4
749#define EXT4_IOC_CHECKPOINT_FLAG_VALID		(EXT4_IOC_CHECKPOINT_FLAG_DISCARD | \
750						EXT4_IOC_CHECKPOINT_FLAG_ZEROOUT | \
751						EXT4_IOC_CHECKPOINT_FLAG_DRY_RUN)
752
753#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
754/*
755 * ioctl commands in 32 bit emulation
756 */
757#define EXT4_IOC32_GETVERSION		_IOR('f', 3, int)
758#define EXT4_IOC32_SETVERSION		_IOW('f', 4, int)
759#define EXT4_IOC32_GETRSVSZ		_IOR('f', 5, int)
760#define EXT4_IOC32_SETRSVSZ		_IOW('f', 6, int)
761#define EXT4_IOC32_GROUP_EXTEND		_IOW('f', 7, unsigned int)
762#define EXT4_IOC32_GROUP_ADD		_IOW('f', 8, struct compat_ext4_new_group_input)
763#define EXT4_IOC32_GETVERSION_OLD	FS_IOC32_GETVERSION
764#define EXT4_IOC32_SETVERSION_OLD	FS_IOC32_SETVERSION
765#endif
766
767/*
768 * Returned by EXT4_IOC_GET_ES_CACHE as an additional possible flag.
769 * It indicates that the entry in extent status cache is for a hole.
770 */
771#define EXT4_FIEMAP_EXTENT_HOLE		0x08000000
772
773/* Max physical block we can address w/o extents */
774#define EXT4_MAX_BLOCK_FILE_PHYS	0xFFFFFFFF
775
776/* Max logical block we can support */
777#define EXT4_MAX_LOGICAL_BLOCK		0xFFFFFFFE
778
779/*
780 * Structure of an inode on the disk
781 */
782struct ext4_inode {
783	__le16	i_mode;		/* File mode */
784	__le16	i_uid;		/* Low 16 bits of Owner Uid */
785	__le32	i_size_lo;	/* Size in bytes */
786	__le32	i_atime;	/* Access time */
787	__le32	i_ctime;	/* Inode Change time */
788	__le32	i_mtime;	/* Modification time */
789	__le32	i_dtime;	/* Deletion Time */
790	__le16	i_gid;		/* Low 16 bits of Group Id */
791	__le16	i_links_count;	/* Links count */
792	__le32	i_blocks_lo;	/* Blocks count */
793	__le32	i_flags;	/* File flags */
794	union {
795		struct {
796			__le32  l_i_version;
797		} linux1;
798		struct {
799			__u32  h_i_translator;
800		} hurd1;
801		struct {
802			__u32  m_i_reserved1;
803		} masix1;
804	} osd1;				/* OS dependent 1 */
805	__le32	i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
806	__le32	i_generation;	/* File version (for NFS) */
807	__le32	i_file_acl_lo;	/* File ACL */
808	__le32	i_size_high;
809	__le32	i_obso_faddr;	/* Obsoleted fragment address */
810	union {
811		struct {
812			__le16	l_i_blocks_high; /* were l_i_reserved1 */
813			__le16	l_i_file_acl_high;
814			__le16	l_i_uid_high;	/* these 2 fields */
815			__le16	l_i_gid_high;	/* were reserved2[0] */
816			__le16	l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
817			__le16	l_i_reserved;
818		} linux2;
819		struct {
820			__le16	h_i_reserved1;	/* Obsoleted fragment number/size which are removed in ext4 */
821			__u16	h_i_mode_high;
822			__u16	h_i_uid_high;
823			__u16	h_i_gid_high;
824			__u32	h_i_author;
825		} hurd2;
826		struct {
827			__le16	h_i_reserved1;	/* Obsoleted fragment number/size which are removed in ext4 */
828			__le16	m_i_file_acl_high;
829			__u32	m_i_reserved2[2];
830		} masix2;
831	} osd2;				/* OS dependent 2 */
832	__le16	i_extra_isize;
833	__le16	i_checksum_hi;	/* crc32c(uuid+inum+inode) BE */
834	__le32  i_ctime_extra;  /* extra Change time      (nsec << 2 | epoch) */
835	__le32  i_mtime_extra;  /* extra Modification time(nsec << 2 | epoch) */
836	__le32  i_atime_extra;  /* extra Access time      (nsec << 2 | epoch) */
837	__le32  i_crtime;       /* File Creation time */
838	__le32  i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
839	__le32  i_version_hi;	/* high 32 bits for 64-bit version */
840	__le32	i_projid;	/* Project ID */
841};
842
843struct move_extent {
844	__u32 reserved;		/* should be zero */
845	__u32 donor_fd;		/* donor file descriptor */
846	__u64 orig_start;	/* logical start offset in block for orig */
847	__u64 donor_start;	/* logical start offset in block for donor */
848	__u64 len;		/* block length to be moved */
849	__u64 moved_len;	/* moved block length */
850};
851
852#define EXT4_EPOCH_BITS 2
853#define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
854#define EXT4_NSEC_MASK  (~0UL << EXT4_EPOCH_BITS)
855
856/*
857 * Extended fields will fit into an inode if the filesystem was formatted
858 * with large inodes (-I 256 or larger) and there are not currently any EAs
859 * consuming all of the available space. For new inodes we always reserve
860 * enough space for the kernel's known extended fields, but for inodes
861 * created with an old kernel this might not have been the case. None of
862 * the extended inode fields is critical for correct filesystem operation.
863 * This macro checks if a certain field fits in the inode. Note that
864 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
865 */
866#define EXT4_FITS_IN_INODE(ext4_inode, einode, field)	\
867	((offsetof(typeof(*ext4_inode), field) +	\
868	  sizeof((ext4_inode)->field))			\
869	<= (EXT4_GOOD_OLD_INODE_SIZE +			\
870	    (einode)->i_extra_isize))			\
871
872/*
873 * We use an encoding that preserves the times for extra epoch "00":
874 *
875 * extra  msb of                         adjust for signed
876 * epoch  32-bit                         32-bit tv_sec to
877 * bits   time    decoded 64-bit tv_sec  64-bit tv_sec      valid time range
878 * 0 0    1    -0x80000000..-0x00000001  0x000000000 1901-12-13..1969-12-31
879 * 0 0    0    0x000000000..0x07fffffff  0x000000000 1970-01-01..2038-01-19
880 * 0 1    1    0x080000000..0x0ffffffff  0x100000000 2038-01-19..2106-02-07
881 * 0 1    0    0x100000000..0x17fffffff  0x100000000 2106-02-07..2174-02-25
882 * 1 0    1    0x180000000..0x1ffffffff  0x200000000 2174-02-25..2242-03-16
883 * 1 0    0    0x200000000..0x27fffffff  0x200000000 2242-03-16..2310-04-04
884 * 1 1    1    0x280000000..0x2ffffffff  0x300000000 2310-04-04..2378-04-22
885 * 1 1    0    0x300000000..0x37fffffff  0x300000000 2378-04-22..2446-05-10
886 *
887 * Note that previous versions of the kernel on 64-bit systems would
888 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
889 * 1970.  e2fsck will correct this, assuming that it is run on the
890 * affected filesystem before 2242.
891 */
892
893static inline __le32 ext4_encode_extra_time(struct timespec64 *time)
894{
895	u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK;
896	return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
897}
898
899static inline void ext4_decode_extra_time(struct timespec64 *time,
900					  __le32 extra)
901{
902	if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK)))
903		time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
904	time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
905}
906
907#define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode)				\
908do {										\
909	if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra))     {\
910		(raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec);	\
911		(raw_inode)->xtime ## _extra =					\
912				ext4_encode_extra_time(&(inode)->xtime);	\
913		}								\
914	else	\
915		(raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (inode)->xtime.tv_sec, S32_MIN, S32_MAX));	\
916} while (0)
917
918#define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode)			       \
919do {									       \
920	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))		       \
921		(raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec);      \
922	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))	       \
923		(raw_inode)->xtime ## _extra =				       \
924				ext4_encode_extra_time(&(einode)->xtime);      \
925} while (0)
926
927#define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode)				\
928do {										\
929	(inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime);	\
930	if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {	\
931		ext4_decode_extra_time(&(inode)->xtime,				\
932				       raw_inode->xtime ## _extra);		\
933		}								\
934	else									\
935		(inode)->xtime.tv_nsec = 0;					\
936} while (0)
937
938
939#define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode)			       \
940do {									       \
941	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime))		       \
942		(einode)->xtime.tv_sec = 				       \
943			(signed)le32_to_cpu((raw_inode)->xtime);	       \
944	else								       \
945		(einode)->xtime.tv_sec = 0;				       \
946	if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra))	       \
947		ext4_decode_extra_time(&(einode)->xtime,		       \
948				       raw_inode->xtime ## _extra);	       \
949	else								       \
950		(einode)->xtime.tv_nsec = 0;				       \
951} while (0)
952
953#define i_disk_version osd1.linux1.l_i_version
954
955#if defined(__KERNEL__) || defined(__linux__)
956#define i_reserved1	osd1.linux1.l_i_reserved1
957#define i_file_acl_high	osd2.linux2.l_i_file_acl_high
958#define i_blocks_high	osd2.linux2.l_i_blocks_high
959#define i_uid_low	i_uid
960#define i_gid_low	i_gid
961#define i_uid_high	osd2.linux2.l_i_uid_high
962#define i_gid_high	osd2.linux2.l_i_gid_high
963#define i_checksum_lo	osd2.linux2.l_i_checksum_lo
964
965#elif defined(__GNU__)
966
967#define i_translator	osd1.hurd1.h_i_translator
968#define i_uid_high	osd2.hurd2.h_i_uid_high
969#define i_gid_high	osd2.hurd2.h_i_gid_high
970#define i_author	osd2.hurd2.h_i_author
971
972#elif defined(__masix__)
973
974#define i_reserved1	osd1.masix1.m_i_reserved1
975#define i_file_acl_high	osd2.masix2.m_i_file_acl_high
976#define i_reserved2	osd2.masix2.m_i_reserved2
977
978#endif /* defined(__KERNEL__) || defined(__linux__) */
979
980#include "extents_status.h"
981#include "fast_commit.h"
982
983/*
984 * Lock subclasses for i_data_sem in the ext4_inode_info structure.
985 *
986 * These are needed to avoid lockdep false positives when we need to
987 * allocate blocks to the quota inode during ext4_map_blocks(), while
988 * holding i_data_sem for a normal (non-quota) inode.  Since we don't
989 * do quota tracking for the quota inode, this avoids deadlock (as
990 * well as infinite recursion, since it isn't turtles all the way
991 * down...)
992 *
993 *  I_DATA_SEM_NORMAL - Used for most inodes
994 *  I_DATA_SEM_OTHER  - Used by move_inode.c for the second normal inode
995 *			  where the second inode has larger inode number
996 *			  than the first
997 *  I_DATA_SEM_QUOTA  - Used for quota inodes only
998 */
999enum {
1000	I_DATA_SEM_NORMAL = 0,
1001	I_DATA_SEM_OTHER,
1002	I_DATA_SEM_QUOTA,
1003};
1004
1005
1006/*
1007 * fourth extended file system inode data in memory
1008 */
1009struct ext4_inode_info {
1010	__le32	i_data[15];	/* unconverted */
1011	__u32	i_dtime;
1012	ext4_fsblk_t	i_file_acl;
1013
1014	/*
1015	 * i_block_group is the number of the block group which contains
1016	 * this file's inode.  Constant across the lifetime of the inode,
1017	 * it is used for making block allocation decisions - we try to
1018	 * place a file's data blocks near its inode block, and new inodes
1019	 * near to their parent directory's inode.
1020	 */
1021	ext4_group_t	i_block_group;
1022	ext4_lblk_t	i_dir_start_lookup;
1023#if (BITS_PER_LONG < 64)
1024	unsigned long	i_state_flags;		/* Dynamic state flags */
1025#endif
1026	unsigned long	i_flags;
1027
1028	/*
1029	 * Extended attributes can be read independently of the main file
1030	 * data. Taking i_mutex even when reading would cause contention
1031	 * between readers of EAs and writers of regular file data, so
1032	 * instead we synchronize on xattr_sem when reading or changing
1033	 * EAs.
1034	 */
1035	struct rw_semaphore xattr_sem;
1036
1037	/*
1038	 * Inodes with EXT4_STATE_ORPHAN_FILE use i_orphan_idx. Otherwise
1039	 * i_orphan is used.
1040	 */
1041	union {
1042		struct list_head i_orphan;	/* unlinked but open inodes */
1043		unsigned int i_orphan_idx;	/* Index in orphan file */
1044	};
1045
1046	/* Fast commit related info */
1047
1048	struct list_head i_fc_list;	/*
1049					 * inodes that need fast commit
1050					 * protected by sbi->s_fc_lock.
1051					 */
1052
1053	/* Start of lblk range that needs to be committed in this fast commit */
1054	ext4_lblk_t i_fc_lblk_start;
1055
1056	/* End of lblk range that needs to be committed in this fast commit */
1057	ext4_lblk_t i_fc_lblk_len;
1058
1059	/* Number of ongoing updates on this inode */
1060	atomic_t  i_fc_updates;
1061
1062	/* Fast commit wait queue for this inode */
1063	wait_queue_head_t i_fc_wait;
1064
1065	/* Protect concurrent accesses on i_fc_lblk_start, i_fc_lblk_len */
1066	struct mutex i_fc_lock;
1067
1068	/*
1069	 * i_disksize keeps track of what the inode size is ON DISK, not
1070	 * in memory.  During truncate, i_size is set to the new size by
1071	 * the VFS prior to calling ext4_truncate(), but the filesystem won't
1072	 * set i_disksize to 0 until the truncate is actually under way.
1073	 *
1074	 * The intent is that i_disksize always represents the blocks which
1075	 * are used by this file.  This allows recovery to restart truncate
1076	 * on orphans if we crash during truncate.  We actually write i_disksize
1077	 * into the on-disk inode when writing inodes out, instead of i_size.
1078	 *
1079	 * The only time when i_disksize and i_size may be different is when
1080	 * a truncate is in progress.  The only things which change i_disksize
1081	 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
1082	 */
1083	loff_t	i_disksize;
1084
1085	/*
1086	 * i_data_sem is for serialising ext4_truncate() against
1087	 * ext4_getblock().  In the 2.4 ext2 design, great chunks of inode's
1088	 * data tree are chopped off during truncate. We can't do that in
1089	 * ext4 because whenever we perform intermediate commits during
1090	 * truncate, the inode and all the metadata blocks *must* be in a
1091	 * consistent state which allows truncation of the orphans to restart
1092	 * during recovery.  Hence we must fix the get_block-vs-truncate race
1093	 * by other means, so we have i_data_sem.
1094	 */
1095	struct rw_semaphore i_data_sem;
1096	/*
1097	 * i_mmap_sem is for serializing page faults with truncate / punch hole
1098	 * operations. We have to make sure that new page cannot be faulted in
1099	 * a section of the inode that is being punched. We cannot easily use
1100	 * i_data_sem for this since we need protection for the whole punch
1101	 * operation and i_data_sem ranks below transaction start so we have
1102	 * to occasionally drop it.
1103	 */
1104	struct rw_semaphore i_mmap_sem;
1105	struct inode vfs_inode;
1106	struct jbd2_inode *jinode;
1107
1108	spinlock_t i_raw_lock;	/* protects updates to the raw inode */
1109
1110	/*
1111	 * File creation time. Its function is same as that of
1112	 * struct timespec64 i_{a,c,m}time in the generic inode.
1113	 */
1114	struct timespec64 i_crtime;
1115
1116	/* mballoc */
1117	atomic_t i_prealloc_active;
1118	struct list_head i_prealloc_list;
1119	spinlock_t i_prealloc_lock;
1120
1121	/* extents status tree */
1122	struct ext4_es_tree i_es_tree;
1123	rwlock_t i_es_lock;
1124	struct list_head i_es_list;
1125	unsigned int i_es_all_nr;	/* protected by i_es_lock */
1126	unsigned int i_es_shk_nr;	/* protected by i_es_lock */
1127	ext4_lblk_t i_es_shrink_lblk;	/* Offset where we start searching for
1128					   extents to shrink. Protected by
1129					   i_es_lock  */
1130
1131	/* ialloc */
1132	ext4_group_t	i_last_alloc_group;
1133
1134	/* allocation reservation info for delalloc */
1135	/* In case of bigalloc, this refer to clusters rather than blocks */
1136	unsigned int i_reserved_data_blocks;
1137
1138	/* pending cluster reservations for bigalloc file systems */
1139	struct ext4_pending_tree i_pending_tree;
1140
1141	/* on-disk additional length */
1142	__u16 i_extra_isize;
1143
1144	/* Indicate the inline data space. */
1145	u16 i_inline_off;
1146	u16 i_inline_size;
1147
1148#ifdef CONFIG_QUOTA
1149	/* quota space reservation, managed internally by quota code */
1150	qsize_t i_reserved_quota;
1151#endif
1152
1153	/* Lock protecting lists below */
1154	spinlock_t i_completed_io_lock;
1155	/*
1156	 * Completed IOs that need unwritten extents handling and have
1157	 * transaction reserved
1158	 */
1159	struct list_head i_rsv_conversion_list;
1160	struct work_struct i_rsv_conversion_work;
1161	atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1162
1163	spinlock_t i_block_reservation_lock;
1164
1165	/*
1166	 * Transactions that contain inode's metadata needed to complete
1167	 * fsync and fdatasync, respectively.
1168	 */
1169	tid_t i_sync_tid;
1170	tid_t i_datasync_tid;
1171
1172#ifdef CONFIG_QUOTA
1173	struct dquot *i_dquot[MAXQUOTAS];
1174#endif
1175
1176	/* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1177	__u32 i_csum_seed;
1178
1179	kprojid_t i_projid;
1180};
1181
1182/*
1183 * File system states
1184 */
1185#define	EXT4_VALID_FS			0x0001	/* Unmounted cleanly */
1186#define	EXT4_ERROR_FS			0x0002	/* Errors detected */
1187#define	EXT4_ORPHAN_FS			0x0004	/* Orphans being recovered */
1188#define EXT4_FC_REPLAY			0x0020	/* Fast commit replay ongoing */
1189
1190/*
1191 * Misc. filesystem flags
1192 */
1193#define EXT2_FLAGS_SIGNED_HASH		0x0001  /* Signed dirhash in use */
1194#define EXT2_FLAGS_UNSIGNED_HASH	0x0002  /* Unsigned dirhash in use */
1195#define EXT2_FLAGS_TEST_FILESYS		0x0004	/* to test development code */
1196
1197/*
1198 * Mount flags set via mount options or defaults
1199 */
1200#define EXT4_MOUNT_NO_MBCACHE		0x00001 /* Do not use mbcache */
1201#define EXT4_MOUNT_GRPID		0x00004	/* Create files with directory's group */
1202#define EXT4_MOUNT_DEBUG		0x00008	/* Some debugging messages */
1203#define EXT4_MOUNT_ERRORS_CONT		0x00010	/* Continue on errors */
1204#define EXT4_MOUNT_ERRORS_RO		0x00020	/* Remount fs ro on errors */
1205#define EXT4_MOUNT_ERRORS_PANIC		0x00040	/* Panic on errors */
1206#define EXT4_MOUNT_ERRORS_MASK		0x00070
1207#define EXT4_MOUNT_MINIX_DF		0x00080	/* Mimics the Minix statfs */
1208#define EXT4_MOUNT_NOLOAD		0x00100	/* Don't use existing journal*/
1209#ifdef CONFIG_FS_DAX
1210#define EXT4_MOUNT_DAX_ALWAYS		0x00200	/* Direct Access */
1211#else
1212#define EXT4_MOUNT_DAX_ALWAYS		0
1213#endif
1214#define EXT4_MOUNT_DATA_FLAGS		0x00C00	/* Mode for data writes: */
1215#define EXT4_MOUNT_JOURNAL_DATA		0x00400	/* Write data to journal */
1216#define EXT4_MOUNT_ORDERED_DATA		0x00800	/* Flush data before commit */
1217#define EXT4_MOUNT_WRITEBACK_DATA	0x00C00	/* No data ordering */
1218#define EXT4_MOUNT_UPDATE_JOURNAL	0x01000	/* Update the journal format */
1219#define EXT4_MOUNT_NO_UID32		0x02000  /* Disable 32-bit UIDs */
1220#define EXT4_MOUNT_XATTR_USER		0x04000	/* Extended user attributes */
1221#define EXT4_MOUNT_POSIX_ACL		0x08000	/* POSIX Access Control Lists */
1222#define EXT4_MOUNT_NO_AUTO_DA_ALLOC	0x10000	/* No auto delalloc mapping */
1223#define EXT4_MOUNT_BARRIER		0x20000 /* Use block barriers */
1224#define EXT4_MOUNT_QUOTA		0x40000 /* Some quota option set */
1225#define EXT4_MOUNT_USRQUOTA		0x80000 /* "old" user quota,
1226						 * enable enforcement for hidden
1227						 * quota files */
1228#define EXT4_MOUNT_GRPQUOTA		0x100000 /* "old" group quota, enable
1229						  * enforcement for hidden quota
1230						  * files */
1231#define EXT4_MOUNT_PRJQUOTA		0x200000 /* Enable project quota
1232						  * enforcement */
1233#define EXT4_MOUNT_DIOREAD_NOLOCK	0x400000 /* Enable support for dio read nolocking */
1234#define EXT4_MOUNT_JOURNAL_CHECKSUM	0x800000 /* Journal checksums */
1235#define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT	0x1000000 /* Journal Async Commit */
1236#define EXT4_MOUNT_WARN_ON_ERROR	0x2000000 /* Trigger WARN_ON on error */
1237#define EXT4_MOUNT_NO_PREFETCH_BLOCK_BITMAPS 0x4000000
1238#define EXT4_MOUNT_DELALLOC		0x8000000 /* Delalloc support */
1239#define EXT4_MOUNT_DATA_ERR_ABORT	0x10000000 /* Abort on file data write */
1240#define EXT4_MOUNT_BLOCK_VALIDITY	0x20000000 /* Block validity checking */
1241#define EXT4_MOUNT_DISCARD		0x40000000 /* Issue DISCARD requests */
1242#define EXT4_MOUNT_INIT_INODE_TABLE	0x80000000 /* Initialize uninitialized itables */
1243
1244/*
1245 * Mount flags set either automatically (could not be set by mount option)
1246 * based on per file system feature or property or in special cases such as
1247 * distinguishing between explicit mount option definition and default.
1248 */
1249#define EXT4_MOUNT2_EXPLICIT_DELALLOC	0x00000001 /* User explicitly
1250						      specified delalloc */
1251#define EXT4_MOUNT2_STD_GROUP_SIZE	0x00000002 /* We have standard group
1252						      size of blocksize * 8
1253						      blocks */
1254#define EXT4_MOUNT2_HURD_COMPAT		0x00000004 /* Support HURD-castrated
1255						      file systems */
1256#define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM	0x00000008 /* User explicitly
1257						specified journal checksum */
1258
1259#define EXT4_MOUNT2_JOURNAL_FAST_COMMIT	0x00000010 /* Journal fast commit */
1260#define EXT4_MOUNT2_DAX_NEVER		0x00000020 /* Do not allow Direct Access */
1261#define EXT4_MOUNT2_DAX_INODE		0x00000040 /* For printing options only */
1262#define EXT4_MOUNT2_MB_OPTIMIZE_SCAN	0x00000080 /* Optimize group
1263						    * scanning in mballoc
1264						    */
1265
1266#define clear_opt(sb, opt)		EXT4_SB(sb)->s_mount_opt &= \
1267						~EXT4_MOUNT_##opt
1268#define set_opt(sb, opt)		EXT4_SB(sb)->s_mount_opt |= \
1269						EXT4_MOUNT_##opt
1270#define test_opt(sb, opt)		(EXT4_SB(sb)->s_mount_opt & \
1271					 EXT4_MOUNT_##opt)
1272
1273#define clear_opt2(sb, opt)		EXT4_SB(sb)->s_mount_opt2 &= \
1274						~EXT4_MOUNT2_##opt
1275#define set_opt2(sb, opt)		EXT4_SB(sb)->s_mount_opt2 |= \
1276						EXT4_MOUNT2_##opt
1277#define test_opt2(sb, opt)		(EXT4_SB(sb)->s_mount_opt2 & \
1278					 EXT4_MOUNT2_##opt)
1279
1280#define ext4_test_and_set_bit		__test_and_set_bit_le
1281#define ext4_set_bit			__set_bit_le
1282#define ext4_set_bit_atomic		ext2_set_bit_atomic
1283#define ext4_test_and_clear_bit		__test_and_clear_bit_le
1284#define ext4_clear_bit			__clear_bit_le
1285#define ext4_clear_bit_atomic		ext2_clear_bit_atomic
1286#define ext4_test_bit			test_bit_le
1287#define ext4_find_next_zero_bit		find_next_zero_bit_le
1288#define ext4_find_next_bit		find_next_bit_le
1289
1290extern void ext4_set_bits(void *bm, int cur, int len);
1291
1292/*
1293 * Maximal mount counts between two filesystem checks
1294 */
1295#define EXT4_DFL_MAX_MNT_COUNT		20	/* Allow 20 mounts */
1296#define EXT4_DFL_CHECKINTERVAL		0	/* Don't use interval check */
1297
1298/*
1299 * Behaviour when detecting errors
1300 */
1301#define EXT4_ERRORS_CONTINUE		1	/* Continue execution */
1302#define EXT4_ERRORS_RO			2	/* Remount fs read-only */
1303#define EXT4_ERRORS_PANIC		3	/* Panic */
1304#define EXT4_ERRORS_DEFAULT		EXT4_ERRORS_CONTINUE
1305
1306/* Metadata checksum algorithm codes */
1307#define EXT4_CRC32C_CHKSUM		1
1308
1309/*
1310 * Structure of the super block
1311 */
1312struct ext4_super_block {
1313/*00*/	__le32	s_inodes_count;		/* Inodes count */
1314	__le32	s_blocks_count_lo;	/* Blocks count */
1315	__le32	s_r_blocks_count_lo;	/* Reserved blocks count */
1316	__le32	s_free_blocks_count_lo;	/* Free blocks count */
1317/*10*/	__le32	s_free_inodes_count;	/* Free inodes count */
1318	__le32	s_first_data_block;	/* First Data Block */
1319	__le32	s_log_block_size;	/* Block size */
1320	__le32	s_log_cluster_size;	/* Allocation cluster size */
1321/*20*/	__le32	s_blocks_per_group;	/* # Blocks per group */
1322	__le32	s_clusters_per_group;	/* # Clusters per group */
1323	__le32	s_inodes_per_group;	/* # Inodes per group */
1324	__le32	s_mtime;		/* Mount time */
1325/*30*/	__le32	s_wtime;		/* Write time */
1326	__le16	s_mnt_count;		/* Mount count */
1327	__le16	s_max_mnt_count;	/* Maximal mount count */
1328	__le16	s_magic;		/* Magic signature */
1329	__le16	s_state;		/* File system state */
1330	__le16	s_errors;		/* Behaviour when detecting errors */
1331	__le16	s_minor_rev_level;	/* minor revision level */
1332/*40*/	__le32	s_lastcheck;		/* time of last check */
1333	__le32	s_checkinterval;	/* max. time between checks */
1334	__le32	s_creator_os;		/* OS */
1335	__le32	s_rev_level;		/* Revision level */
1336/*50*/	__le16	s_def_resuid;		/* Default uid for reserved blocks */
1337	__le16	s_def_resgid;		/* Default gid for reserved blocks */
1338	/*
1339	 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1340	 *
1341	 * Note: the difference between the compatible feature set and
1342	 * the incompatible feature set is that if there is a bit set
1343	 * in the incompatible feature set that the kernel doesn't
1344	 * know about, it should refuse to mount the filesystem.
1345	 *
1346	 * e2fsck's requirements are more strict; if it doesn't know
1347	 * about a feature in either the compatible or incompatible
1348	 * feature set, it must abort and not try to meddle with
1349	 * things it doesn't understand...
1350	 */
1351	__le32	s_first_ino;		/* First non-reserved inode */
1352	__le16  s_inode_size;		/* size of inode structure */
1353	__le16	s_block_group_nr;	/* block group # of this superblock */
1354	__le32	s_feature_compat;	/* compatible feature set */
1355/*60*/	__le32	s_feature_incompat;	/* incompatible feature set */
1356	__le32	s_feature_ro_compat;	/* readonly-compatible feature set */
1357/*68*/	__u8	s_uuid[16];		/* 128-bit uuid for volume */
1358/*78*/	char	s_volume_name[16];	/* volume name */
1359/*88*/	char	s_last_mounted[64] __nonstring;	/* directory where last mounted */
1360/*C8*/	__le32	s_algorithm_usage_bitmap; /* For compression */
1361	/*
1362	 * Performance hints.  Directory preallocation should only
1363	 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1364	 */
1365	__u8	s_prealloc_blocks;	/* Nr of blocks to try to preallocate*/
1366	__u8	s_prealloc_dir_blocks;	/* Nr to preallocate for dirs */
1367	__le16	s_reserved_gdt_blocks;	/* Per group desc for online growth */
1368	/*
1369	 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1370	 */
1371/*D0*/	__u8	s_journal_uuid[16];	/* uuid of journal superblock */
1372/*E0*/	__le32	s_journal_inum;		/* inode number of journal file */
1373	__le32	s_journal_dev;		/* device number of journal file */
1374	__le32	s_last_orphan;		/* start of list of inodes to delete */
1375	__le32	s_hash_seed[4];		/* HTREE hash seed */
1376	__u8	s_def_hash_version;	/* Default hash version to use */
1377	__u8	s_jnl_backup_type;
1378	__le16  s_desc_size;		/* size of group descriptor */
1379/*100*/	__le32	s_default_mount_opts;
1380	__le32	s_first_meta_bg;	/* First metablock block group */
1381	__le32	s_mkfs_time;		/* When the filesystem was created */
1382	__le32	s_jnl_blocks[17];	/* Backup of the journal inode */
1383	/* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1384/*150*/	__le32	s_blocks_count_hi;	/* Blocks count */
1385	__le32	s_r_blocks_count_hi;	/* Reserved blocks count */
1386	__le32	s_free_blocks_count_hi;	/* Free blocks count */
1387	__le16	s_min_extra_isize;	/* All inodes have at least # bytes */
1388	__le16	s_want_extra_isize; 	/* New inodes should reserve # bytes */
1389	__le32	s_flags;		/* Miscellaneous flags */
1390	__le16  s_raid_stride;		/* RAID stride */
1391	__le16  s_mmp_update_interval;  /* # seconds to wait in MMP checking */
1392	__le64  s_mmp_block;            /* Block for multi-mount protection */
1393	__le32  s_raid_stripe_width;    /* blocks on all data disks (N*stride)*/
1394	__u8	s_log_groups_per_flex;  /* FLEX_BG group size */
1395	__u8	s_checksum_type;	/* metadata checksum algorithm used */
1396	__u8	s_encryption_level;	/* versioning level for encryption */
1397	__u8	s_reserved_pad;		/* Padding to next 32bits */
1398	__le64	s_kbytes_written;	/* nr of lifetime kilobytes written */
1399	__le32	s_snapshot_inum;	/* Inode number of active snapshot */
1400	__le32	s_snapshot_id;		/* sequential ID of active snapshot */
1401	__le64	s_snapshot_r_blocks_count; /* reserved blocks for active
1402					      snapshot's future use */
1403	__le32	s_snapshot_list;	/* inode number of the head of the
1404					   on-disk snapshot list */
1405#define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1406	__le32	s_error_count;		/* number of fs errors */
1407	__le32	s_first_error_time;	/* first time an error happened */
1408	__le32	s_first_error_ino;	/* inode involved in first error */
1409	__le64	s_first_error_block;	/* block involved of first error */
1410	__u8	s_first_error_func[32] __nonstring;	/* function where the error happened */
1411	__le32	s_first_error_line;	/* line number where error happened */
1412	__le32	s_last_error_time;	/* most recent time of an error */
1413	__le32	s_last_error_ino;	/* inode involved in last error */
1414	__le32	s_last_error_line;	/* line number where error happened */
1415	__le64	s_last_error_block;	/* block involved of last error */
1416	__u8	s_last_error_func[32] __nonstring;	/* function where the error happened */
1417#define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1418	__u8	s_mount_opts[64];
1419	__le32	s_usr_quota_inum;	/* inode for tracking user quota */
1420	__le32	s_grp_quota_inum;	/* inode for tracking group quota */
1421	__le32	s_overhead_clusters;	/* overhead blocks/clusters in fs */
1422	__le32	s_backup_bgs[2];	/* groups with sparse_super2 SBs */
1423	__u8	s_encrypt_algos[4];	/* Encryption algorithms in use  */
1424	__u8	s_encrypt_pw_salt[16];	/* Salt used for string2key algorithm */
1425	__le32	s_lpf_ino;		/* Location of the lost+found inode */
1426	__le32	s_prj_quota_inum;	/* inode for tracking project quota */
1427	__le32	s_checksum_seed;	/* crc32c(uuid) if csum_seed set */
1428	__u8	s_wtime_hi;
1429	__u8	s_mtime_hi;
1430	__u8	s_mkfs_time_hi;
1431	__u8	s_lastcheck_hi;
1432	__u8	s_first_error_time_hi;
1433	__u8	s_last_error_time_hi;
1434	__u8	s_first_error_errcode;
1435	__u8    s_last_error_errcode;
1436	__le16  s_encoding;		/* Filename charset encoding */
1437	__le16  s_encoding_flags;	/* Filename charset encoding flags */
1438	__le32  s_orphan_file_inum;	/* Inode for tracking orphan inodes */
1439	__le32	s_reserved[94];		/* Padding to the end of the block */
1440	__le32	s_checksum;		/* crc32c(superblock) */
1441};
1442
1443#define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1444
1445#ifdef __KERNEL__
1446
1447#ifdef CONFIG_FS_ENCRYPTION
1448#define DUMMY_ENCRYPTION_ENABLED(sbi) ((sbi)->s_dummy_enc_policy.policy != NULL)
1449#else
1450#define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1451#endif
1452
1453/* Number of quota types we support */
1454#define EXT4_MAXQUOTAS 3
1455
1456#define EXT4_ENC_UTF8_12_1	1
1457
1458/* Types of ext4 journal triggers */
1459enum ext4_journal_trigger_type {
1460	EXT4_JTR_ORPHAN_FILE,
1461	EXT4_JTR_NONE	/* This must be the last entry for indexing to work! */
1462};
1463
1464#define EXT4_JOURNAL_TRIGGER_COUNT EXT4_JTR_NONE
1465
1466struct ext4_journal_trigger {
1467	struct jbd2_buffer_trigger_type tr_triggers;
1468	struct super_block *sb;
1469};
1470
1471static inline struct ext4_journal_trigger *EXT4_TRIGGER(
1472				struct jbd2_buffer_trigger_type *trigger)
1473{
1474	return container_of(trigger, struct ext4_journal_trigger, tr_triggers);
1475}
1476
1477#define EXT4_ORPHAN_BLOCK_MAGIC 0x0b10ca04
1478
1479/* Structure at the tail of orphan block */
1480struct ext4_orphan_block_tail {
1481	__le32 ob_magic;
1482	__le32 ob_checksum;
1483};
1484
1485static inline int ext4_inodes_per_orphan_block(struct super_block *sb)
1486{
1487	return (sb->s_blocksize - sizeof(struct ext4_orphan_block_tail)) /
1488			sizeof(u32);
1489}
1490
1491struct ext4_orphan_block {
1492	atomic_t ob_free_entries;	/* Number of free orphan entries in block */
1493	struct buffer_head *ob_bh;	/* Buffer for orphan block */
1494};
1495
1496/*
1497 * Info about orphan file.
1498 */
1499struct ext4_orphan_info {
1500	int of_blocks;			/* Number of orphan blocks in a file */
1501	__u32 of_csum_seed;		/* Checksum seed for orphan file */
1502	struct ext4_orphan_block *of_binfo;	/* Array with info about orphan
1503						 * file blocks */
1504};
1505
1506/*
1507 * fourth extended-fs super-block data in memory
1508 */
1509struct ext4_sb_info {
1510	unsigned long s_desc_size;	/* Size of a group descriptor in bytes */
1511	unsigned long s_inodes_per_block;/* Number of inodes per block */
1512	unsigned long s_blocks_per_group;/* Number of blocks in a group */
1513	unsigned long s_clusters_per_group; /* Number of clusters in a group */
1514	unsigned long s_inodes_per_group;/* Number of inodes in a group */
1515	unsigned long s_itb_per_group;	/* Number of inode table blocks per group */
1516	unsigned long s_gdb_count;	/* Number of group descriptor blocks */
1517	unsigned long s_desc_per_block;	/* Number of group descriptors per block */
1518	ext4_group_t s_groups_count;	/* Number of groups in the fs */
1519	ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1520	unsigned long s_overhead;  /* # of fs overhead clusters */
1521	unsigned int s_cluster_ratio;	/* Number of blocks per cluster */
1522	unsigned int s_cluster_bits;	/* log2 of s_cluster_ratio */
1523	loff_t s_bitmap_maxbytes;	/* max bytes for bitmap files */
1524	struct buffer_head * s_sbh;	/* Buffer containing the super block */
1525	struct ext4_super_block *s_es;	/* Pointer to the super block in the buffer */
1526	struct buffer_head * __rcu *s_group_desc;
1527	unsigned int s_mount_opt;
1528	unsigned int s_mount_opt2;
1529	unsigned long s_mount_flags;
1530	unsigned int s_def_mount_opt;
1531	ext4_fsblk_t s_sb_block;
1532	atomic64_t s_resv_clusters;
1533	kuid_t s_resuid;
1534	kgid_t s_resgid;
1535	unsigned short s_mount_state;
1536	unsigned short s_pad;
1537	int s_addr_per_block_bits;
1538	int s_desc_per_block_bits;
1539	int s_inode_size;
1540	int s_first_ino;
1541	unsigned int s_inode_readahead_blks;
1542	unsigned int s_inode_goal;
1543	u32 s_hash_seed[4];
1544	int s_def_hash_version;
1545	int s_hash_unsigned;	/* 3 if hash should be unsigned, 0 if not */
1546	struct percpu_counter s_freeclusters_counter;
1547	struct percpu_counter s_freeinodes_counter;
1548	struct percpu_counter s_dirs_counter;
1549	struct percpu_counter s_dirtyclusters_counter;
1550	struct percpu_counter s_sra_exceeded_retry_limit;
1551	struct blockgroup_lock *s_blockgroup_lock;
1552	struct proc_dir_entry *s_proc;
1553	struct kobject s_kobj;
1554	struct completion s_kobj_unregister;
1555	struct super_block *s_sb;
1556	struct buffer_head *s_mmp_bh;
1557
1558	/* Journaling */
1559	struct journal_s *s_journal;
1560	unsigned long s_ext4_flags;		/* Ext4 superblock flags */
1561	struct mutex s_orphan_lock;	/* Protects on disk list changes */
1562	struct list_head s_orphan;	/* List of orphaned inodes in on disk
1563					   list */
1564	struct ext4_orphan_info s_orphan_info;
1565	unsigned long s_commit_interval;
1566	u32 s_max_batch_time;
1567	u32 s_min_batch_time;
1568	struct block_device *s_journal_bdev;
1569#ifdef CONFIG_QUOTA
1570	/* Names of quota files with journalled quota */
1571	char __rcu *s_qf_names[EXT4_MAXQUOTAS];
1572	int s_jquota_fmt;			/* Format of quota to use */
1573#endif
1574	unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1575	struct ext4_system_blocks __rcu *s_system_blks;
1576
1577#ifdef EXTENTS_STATS
1578	/* ext4 extents stats */
1579	unsigned long s_ext_min;
1580	unsigned long s_ext_max;
1581	unsigned long s_depth_max;
1582	spinlock_t s_ext_stats_lock;
1583	unsigned long s_ext_blocks;
1584	unsigned long s_ext_extents;
1585#endif
1586
1587	/* for buddy allocator */
1588	struct ext4_group_info ** __rcu *s_group_info;
1589	struct inode *s_buddy_cache;
1590	spinlock_t s_md_lock;
1591	unsigned short *s_mb_offsets;
1592	unsigned int *s_mb_maxs;
1593	unsigned int s_group_info_size;
1594	unsigned int s_mb_free_pending;
1595	struct list_head s_freed_data_list;	/* List of blocks to be freed
1596						   after commit completed */
1597	struct list_head s_discard_list;
1598	struct work_struct s_discard_work;
1599	atomic_t s_retry_alloc_pending;
1600	struct rb_root s_mb_avg_fragment_size_root;
1601	rwlock_t s_mb_rb_lock;
1602	struct list_head *s_mb_largest_free_orders;
1603	rwlock_t *s_mb_largest_free_orders_locks;
1604
1605	/* tunables */
1606	unsigned long s_stripe;
1607	unsigned int s_mb_max_linear_groups;
1608	unsigned int s_mb_stream_request;
1609	unsigned int s_mb_max_to_scan;
1610	unsigned int s_mb_min_to_scan;
1611	unsigned int s_mb_stats;
1612	unsigned int s_mb_order2_reqs;
1613	unsigned int s_mb_group_prealloc;
1614	unsigned int s_mb_max_inode_prealloc;
1615	unsigned int s_max_dir_size_kb;
1616	/* where last allocation was done - for stream allocation */
1617	unsigned long s_mb_last_group;
1618	unsigned long s_mb_last_start;
1619	unsigned int s_mb_prefetch;
1620	unsigned int s_mb_prefetch_limit;
1621
1622	/* stats for buddy allocator */
1623	atomic_t s_bal_reqs;	/* number of reqs with len > 1 */
1624	atomic_t s_bal_success;	/* we found long enough chunks */
1625	atomic_t s_bal_allocated;	/* in blocks */
1626	atomic_t s_bal_ex_scanned;	/* total extents scanned */
1627	atomic_t s_bal_groups_scanned;	/* number of groups scanned */
1628	atomic_t s_bal_goals;	/* goal hits */
1629	atomic_t s_bal_breaks;	/* too long searches */
1630	atomic_t s_bal_2orders;	/* 2^order hits */
1631	atomic_t s_bal_cr0_bad_suggestions;
1632	atomic_t s_bal_cr1_bad_suggestions;
1633	atomic64_t s_bal_cX_groups_considered[4];
1634	atomic64_t s_bal_cX_hits[4];
1635	atomic64_t s_bal_cX_failed[4];		/* cX loop didn't find blocks */
1636	atomic_t s_mb_buddies_generated;	/* number of buddies generated */
1637	atomic64_t s_mb_generation_time;
1638	atomic_t s_mb_lost_chunks;
1639	atomic_t s_mb_preallocated;
1640	atomic_t s_mb_discarded;
1641	atomic_t s_lock_busy;
1642
1643	/* locality groups */
1644	struct ext4_locality_group __percpu *s_locality_groups;
1645
1646	/* for write statistics */
1647	unsigned long s_sectors_written_start;
1648	u64 s_kbytes_written;
1649
1650	/* the size of zero-out chunk */
1651	unsigned int s_extent_max_zeroout_kb;
1652
1653	unsigned int s_log_groups_per_flex;
1654	struct flex_groups * __rcu *s_flex_groups;
1655	ext4_group_t s_flex_groups_allocated;
1656
1657	/* workqueue for reserved extent conversions (buffered io) */
1658	struct workqueue_struct *rsv_conversion_wq;
1659
1660	/* timer for periodic error stats printing */
1661	struct timer_list s_err_report;
1662
1663	/* Lazy inode table initialization info */
1664	struct ext4_li_request *s_li_request;
1665	/* Wait multiplier for lazy initialization thread */
1666	unsigned int s_li_wait_mult;
1667
1668	/* Kernel thread for multiple mount protection */
1669	struct task_struct *s_mmp_tsk;
1670
1671	/* record the last minlen when FITRIM is called. */
1672	atomic_t s_last_trim_minblks;
1673
1674	/* Reference to checksum algorithm driver via cryptoapi */
1675	struct crypto_shash *s_chksum_driver;
1676
1677	/* Precomputed FS UUID checksum for seeding other checksums */
1678	__u32 s_csum_seed;
1679
1680	/* Reclaim extents from extent status tree */
1681	struct shrinker s_es_shrinker;
1682	struct list_head s_es_list;	/* List of inodes with reclaimable extents */
1683	long s_es_nr_inode;
1684	struct ext4_es_stats s_es_stats;
1685	struct mb_cache *s_ea_block_cache;
1686	struct mb_cache *s_ea_inode_cache;
1687	spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1688
1689	/* Journal triggers for checksum computation */
1690	struct ext4_journal_trigger s_journal_triggers[EXT4_JOURNAL_TRIGGER_COUNT];
1691
1692	/* Ratelimit ext4 messages. */
1693	struct ratelimit_state s_err_ratelimit_state;
1694	struct ratelimit_state s_warning_ratelimit_state;
1695	struct ratelimit_state s_msg_ratelimit_state;
1696	atomic_t s_warning_count;
1697	atomic_t s_msg_count;
1698
1699	/* Encryption policy for '-o test_dummy_encryption' */
1700	struct fscrypt_dummy_policy s_dummy_enc_policy;
1701
1702	/*
1703	 * Barrier between writepages ops and changing any inode's JOURNAL_DATA
1704	 * or EXTENTS flag.
1705	 */
1706	struct percpu_rw_semaphore s_writepages_rwsem;
1707	struct dax_device *s_daxdev;
1708#ifdef CONFIG_EXT4_DEBUG
1709	unsigned long s_simulate_fail;
1710#endif
1711	/* Record the errseq of the backing block device */
1712	errseq_t s_bdev_wb_err;
1713	spinlock_t s_bdev_wb_lock;
1714
1715	/* Information about errors that happened during this mount */
1716	spinlock_t s_error_lock;
1717	int s_add_error_count;
1718	int s_first_error_code;
1719	__u32 s_first_error_line;
1720	__u32 s_first_error_ino;
1721	__u64 s_first_error_block;
1722	const char *s_first_error_func;
1723	time64_t s_first_error_time;
1724	int s_last_error_code;
1725	__u32 s_last_error_line;
1726	__u32 s_last_error_ino;
1727	__u64 s_last_error_block;
1728	const char *s_last_error_func;
1729	time64_t s_last_error_time;
1730	/*
1731	 * If we are in a context where we cannot update error information in
1732	 * the on-disk superblock, we queue this work to do it.
1733	 */
1734	struct work_struct s_error_work;
1735
1736	/* Ext4 fast commit stuff */
1737	atomic_t s_fc_subtid;
1738	atomic_t s_fc_ineligible_updates;
1739	/*
1740	 * After commit starts, the main queue gets locked, and the further
1741	 * updates get added in the staging queue.
1742	 */
1743#define FC_Q_MAIN	0
1744#define FC_Q_STAGING	1
1745	struct list_head s_fc_q[2];	/* Inodes staged for fast commit
1746					 * that have data changes in them.
1747					 */
1748	struct list_head s_fc_dentry_q[2];	/* directory entry updates */
1749	unsigned int s_fc_bytes;
1750	/*
1751	 * Main fast commit lock. This lock protects accesses to the
1752	 * following fields:
1753	 * ei->i_fc_list, s_fc_dentry_q, s_fc_q, s_fc_bytes, s_fc_bh.
1754	 */
1755	spinlock_t s_fc_lock;
1756	struct buffer_head *s_fc_bh;
1757	struct ext4_fc_stats s_fc_stats;
1758	u64 s_fc_avg_commit_time;
1759#ifdef CONFIG_EXT4_DEBUG
1760	int s_fc_debug_max_replay;
1761#endif
1762	struct ext4_fc_replay_state s_fc_replay_state;
1763};
1764
1765static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1766{
1767	return sb->s_fs_info;
1768}
1769static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1770{
1771	return container_of(inode, struct ext4_inode_info, vfs_inode);
1772}
1773
1774static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1775{
1776	return ino == EXT4_ROOT_INO ||
1777		(ino >= EXT4_FIRST_INO(sb) &&
1778		 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1779}
1780
1781/*
1782 * Returns: sbi->field[index]
1783 * Used to access an array element from the following sbi fields which require
1784 * rcu protection to avoid dereferencing an invalid pointer due to reassignment
1785 * - s_group_desc
1786 * - s_group_info
1787 * - s_flex_group
1788 */
1789#define sbi_array_rcu_deref(sbi, field, index)				   \
1790({									   \
1791	typeof(*((sbi)->field)) _v;					   \
1792	rcu_read_lock();						   \
1793	_v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index];	   \
1794	rcu_read_unlock();						   \
1795	_v;								   \
1796})
1797
1798/*
1799 * run-time mount flags
1800 */
1801enum {
1802	EXT4_MF_MNTDIR_SAMPLED,
1803	EXT4_MF_FS_ABORTED,	/* Fatal error detected */
1804	EXT4_MF_FC_INELIGIBLE,	/* Fast commit ineligible */
1805	EXT4_MF_FC_COMMITTING	/* File system underoing a fast
1806				 * commit.
1807				 */
1808};
1809
1810static inline void ext4_set_mount_flag(struct super_block *sb, int bit)
1811{
1812	set_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1813}
1814
1815static inline void ext4_clear_mount_flag(struct super_block *sb, int bit)
1816{
1817	clear_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1818}
1819
1820static inline int ext4_test_mount_flag(struct super_block *sb, int bit)
1821{
1822	return test_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1823}
1824
1825
1826/*
1827 * Simulate_fail codes
1828 */
1829#define EXT4_SIM_BBITMAP_EIO	1
1830#define EXT4_SIM_BBITMAP_CRC	2
1831#define EXT4_SIM_IBITMAP_EIO	3
1832#define EXT4_SIM_IBITMAP_CRC	4
1833#define EXT4_SIM_INODE_EIO	5
1834#define EXT4_SIM_INODE_CRC	6
1835#define EXT4_SIM_DIRBLOCK_EIO	7
1836#define EXT4_SIM_DIRBLOCK_CRC	8
1837
1838static inline bool ext4_simulate_fail(struct super_block *sb,
1839				     unsigned long code)
1840{
1841#ifdef CONFIG_EXT4_DEBUG
1842	struct ext4_sb_info *sbi = EXT4_SB(sb);
1843
1844	if (unlikely(sbi->s_simulate_fail == code)) {
1845		sbi->s_simulate_fail = 0;
1846		return true;
1847	}
1848#endif
1849	return false;
1850}
1851
1852static inline void ext4_simulate_fail_bh(struct super_block *sb,
1853					 struct buffer_head *bh,
1854					 unsigned long code)
1855{
1856	if (!IS_ERR(bh) && ext4_simulate_fail(sb, code))
1857		clear_buffer_uptodate(bh);
1858}
1859
1860/*
1861 * Error number codes for s_{first,last}_error_errno
1862 *
1863 * Linux errno numbers are architecture specific, so we need to translate
1864 * them into something which is architecture independent.   We don't define
1865 * codes for all errno's; just the ones which are most likely to be the cause
1866 * of an ext4_error() call.
1867 */
1868#define EXT4_ERR_UNKNOWN	 1
1869#define EXT4_ERR_EIO		 2
1870#define EXT4_ERR_ENOMEM		 3
1871#define EXT4_ERR_EFSBADCRC	 4
1872#define EXT4_ERR_EFSCORRUPTED	 5
1873#define EXT4_ERR_ENOSPC		 6
1874#define EXT4_ERR_ENOKEY		 7
1875#define EXT4_ERR_EROFS		 8
1876#define EXT4_ERR_EFBIG		 9
1877#define EXT4_ERR_EEXIST		10
1878#define EXT4_ERR_ERANGE		11
1879#define EXT4_ERR_EOVERFLOW	12
1880#define EXT4_ERR_EBUSY		13
1881#define EXT4_ERR_ENOTDIR	14
1882#define EXT4_ERR_ENOTEMPTY	15
1883#define EXT4_ERR_ESHUTDOWN	16
1884#define EXT4_ERR_EFAULT		17
1885
1886/*
1887 * Inode dynamic state flags
1888 */
1889enum {
1890	EXT4_STATE_JDATA,		/* journaled data exists */
1891	EXT4_STATE_NEW,			/* inode is newly created */
1892	EXT4_STATE_XATTR,		/* has in-inode xattrs */
1893	EXT4_STATE_NO_EXPAND,		/* No space for expansion */
1894	EXT4_STATE_DA_ALLOC_CLOSE,	/* Alloc DA blks on close */
1895	EXT4_STATE_EXT_MIGRATE,		/* Inode is migrating */
1896	EXT4_STATE_NEWENTRY,		/* File just added to dir */
1897	EXT4_STATE_MAY_INLINE_DATA,	/* may have in-inode data */
1898	EXT4_STATE_EXT_PRECACHED,	/* extents have been precached */
1899	EXT4_STATE_LUSTRE_EA_INODE,	/* Lustre-style ea_inode */
1900	EXT4_STATE_VERITY_IN_PROGRESS,	/* building fs-verity Merkle tree */
1901	EXT4_STATE_FC_COMMITTING,	/* Fast commit ongoing */
1902	EXT4_STATE_ORPHAN_FILE,		/* Inode orphaned in orphan file */
1903};
1904
1905#define EXT4_INODE_BIT_FNS(name, field, offset)				\
1906static inline int ext4_test_inode_##name(struct inode *inode, int bit)	\
1907{									\
1908	return test_bit(bit + (offset), &EXT4_I(inode)->i_##field);	\
1909}									\
1910static inline void ext4_set_inode_##name(struct inode *inode, int bit)	\
1911{									\
1912	set_bit(bit + (offset), &EXT4_I(inode)->i_##field);		\
1913}									\
1914static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1915{									\
1916	clear_bit(bit + (offset), &EXT4_I(inode)->i_##field);		\
1917}
1918
1919/* Add these declarations here only so that these functions can be
1920 * found by name.  Otherwise, they are very hard to locate. */
1921static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1922static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1923static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1924EXT4_INODE_BIT_FNS(flag, flags, 0)
1925
1926/* Add these declarations here only so that these functions can be
1927 * found by name.  Otherwise, they are very hard to locate. */
1928static inline int ext4_test_inode_state(struct inode *inode, int bit);
1929static inline void ext4_set_inode_state(struct inode *inode, int bit);
1930static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1931#if (BITS_PER_LONG < 64)
1932EXT4_INODE_BIT_FNS(state, state_flags, 0)
1933
1934static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1935{
1936	(ei)->i_state_flags = 0;
1937}
1938#else
1939EXT4_INODE_BIT_FNS(state, flags, 32)
1940
1941static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1942{
1943	/* We depend on the fact that callers will set i_flags */
1944}
1945#endif
1946#else
1947/* Assume that user mode programs are passing in an ext4fs superblock, not
1948 * a kernel struct super_block.  This will allow us to call the feature-test
1949 * macros from user land. */
1950#define EXT4_SB(sb)	(sb)
1951#endif
1952
1953static inline bool ext4_verity_in_progress(struct inode *inode)
1954{
1955	return IS_ENABLED(CONFIG_FS_VERITY) &&
1956	       ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS);
1957}
1958
1959#define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1960
1961/*
1962 * Codes for operating systems
1963 */
1964#define EXT4_OS_LINUX		0
1965#define EXT4_OS_HURD		1
1966#define EXT4_OS_MASIX		2
1967#define EXT4_OS_FREEBSD		3
1968#define EXT4_OS_LITES		4
1969
1970/*
1971 * Revision levels
1972 */
1973#define EXT4_GOOD_OLD_REV	0	/* The good old (original) format */
1974#define EXT4_DYNAMIC_REV	1	/* V2 format w/ dynamic inode sizes */
1975
1976#define EXT4_MAX_SUPP_REV	EXT4_DYNAMIC_REV
1977
1978#define EXT4_GOOD_OLD_INODE_SIZE 128
1979
1980#define EXT4_EXTRA_TIMESTAMP_MAX	(((s64)1 << 34) - 1  + S32_MIN)
1981#define EXT4_NON_EXTRA_TIMESTAMP_MAX	S32_MAX
1982#define EXT4_TIMESTAMP_MIN		S32_MIN
1983
1984/*
1985 * Feature set definitions
1986 */
1987
1988#define EXT4_FEATURE_COMPAT_DIR_PREALLOC	0x0001
1989#define EXT4_FEATURE_COMPAT_IMAGIC_INODES	0x0002
1990#define EXT4_FEATURE_COMPAT_HAS_JOURNAL		0x0004
1991#define EXT4_FEATURE_COMPAT_EXT_ATTR		0x0008
1992#define EXT4_FEATURE_COMPAT_RESIZE_INODE	0x0010
1993#define EXT4_FEATURE_COMPAT_DIR_INDEX		0x0020
1994#define EXT4_FEATURE_COMPAT_SPARSE_SUPER2	0x0200
1995/*
1996 * The reason why "FAST_COMMIT" is a compat feature is that, FS becomes
1997 * incompatible only if fast commit blocks are present in the FS. Since we
1998 * clear the journal (and thus the fast commit blocks), we don't mark FS as
1999 * incompatible. We also have a JBD2 incompat feature, which gets set when
2000 * there are fast commit blocks present in the journal.
2001 */
2002#define EXT4_FEATURE_COMPAT_FAST_COMMIT		0x0400
2003#define EXT4_FEATURE_COMPAT_STABLE_INODES	0x0800
2004#define EXT4_FEATURE_COMPAT_ORPHAN_FILE		0x1000	/* Orphan file exists */
2005
2006#define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER	0x0001
2007#define EXT4_FEATURE_RO_COMPAT_LARGE_FILE	0x0002
2008#define EXT4_FEATURE_RO_COMPAT_BTREE_DIR	0x0004
2009#define EXT4_FEATURE_RO_COMPAT_HUGE_FILE        0x0008
2010#define EXT4_FEATURE_RO_COMPAT_GDT_CSUM		0x0010
2011#define EXT4_FEATURE_RO_COMPAT_DIR_NLINK	0x0020
2012#define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE	0x0040
2013#define EXT4_FEATURE_RO_COMPAT_QUOTA		0x0100
2014#define EXT4_FEATURE_RO_COMPAT_BIGALLOC		0x0200
2015/*
2016 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM).  When
2017 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
2018 * all other data structures' checksums.  However, the METADATA_CSUM and
2019 * GDT_CSUM bits are mutually exclusive.
2020 */
2021#define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM	0x0400
2022#define EXT4_FEATURE_RO_COMPAT_READONLY		0x1000
2023#define EXT4_FEATURE_RO_COMPAT_PROJECT		0x2000
2024#define EXT4_FEATURE_RO_COMPAT_VERITY		0x8000
2025#define EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT	0x10000 /* Orphan file may be
2026							   non-empty */
2027
2028#define EXT4_FEATURE_INCOMPAT_COMPRESSION	0x0001
2029#define EXT4_FEATURE_INCOMPAT_FILETYPE		0x0002
2030#define EXT4_FEATURE_INCOMPAT_RECOVER		0x0004 /* Needs recovery */
2031#define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV	0x0008 /* Journal device */
2032#define EXT4_FEATURE_INCOMPAT_META_BG		0x0010
2033#define EXT4_FEATURE_INCOMPAT_EXTENTS		0x0040 /* extents support */
2034#define EXT4_FEATURE_INCOMPAT_64BIT		0x0080
2035#define EXT4_FEATURE_INCOMPAT_MMP               0x0100
2036#define EXT4_FEATURE_INCOMPAT_FLEX_BG		0x0200
2037#define EXT4_FEATURE_INCOMPAT_EA_INODE		0x0400 /* EA in inode */
2038#define EXT4_FEATURE_INCOMPAT_DIRDATA		0x1000 /* data in dirent */
2039#define EXT4_FEATURE_INCOMPAT_CSUM_SEED		0x2000
2040#define EXT4_FEATURE_INCOMPAT_LARGEDIR		0x4000 /* >2GB or 3-lvl htree */
2041#define EXT4_FEATURE_INCOMPAT_INLINE_DATA	0x8000 /* data in inode */
2042#define EXT4_FEATURE_INCOMPAT_ENCRYPT		0x10000
2043#define EXT4_FEATURE_INCOMPAT_CASEFOLD		0x20000
2044
2045extern void ext4_update_dynamic_rev(struct super_block *sb);
2046
2047#define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
2048static inline bool ext4_has_feature_##name(struct super_block *sb) \
2049{ \
2050	return ((EXT4_SB(sb)->s_es->s_feature_compat & \
2051		cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
2052} \
2053static inline void ext4_set_feature_##name(struct super_block *sb) \
2054{ \
2055	ext4_update_dynamic_rev(sb); \
2056	EXT4_SB(sb)->s_es->s_feature_compat |= \
2057		cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
2058} \
2059static inline void ext4_clear_feature_##name(struct super_block *sb) \
2060{ \
2061	EXT4_SB(sb)->s_es->s_feature_compat &= \
2062		~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
2063}
2064
2065#define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
2066static inline bool ext4_has_feature_##name(struct super_block *sb) \
2067{ \
2068	return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
2069		cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
2070} \
2071static inline void ext4_set_feature_##name(struct super_block *sb) \
2072{ \
2073	ext4_update_dynamic_rev(sb); \
2074	EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
2075		cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
2076} \
2077static inline void ext4_clear_feature_##name(struct super_block *sb) \
2078{ \
2079	EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
2080		~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
2081}
2082
2083#define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
2084static inline bool ext4_has_feature_##name(struct super_block *sb) \
2085{ \
2086	return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
2087		cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
2088} \
2089static inline void ext4_set_feature_##name(struct super_block *sb) \
2090{ \
2091	ext4_update_dynamic_rev(sb); \
2092	EXT4_SB(sb)->s_es->s_feature_incompat |= \
2093		cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
2094} \
2095static inline void ext4_clear_feature_##name(struct super_block *sb) \
2096{ \
2097	EXT4_SB(sb)->s_es->s_feature_incompat &= \
2098		~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
2099}
2100
2101EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc,		DIR_PREALLOC)
2102EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes,	IMAGIC_INODES)
2103EXT4_FEATURE_COMPAT_FUNCS(journal,		HAS_JOURNAL)
2104EXT4_FEATURE_COMPAT_FUNCS(xattr,		EXT_ATTR)
2105EXT4_FEATURE_COMPAT_FUNCS(resize_inode,		RESIZE_INODE)
2106EXT4_FEATURE_COMPAT_FUNCS(dir_index,		DIR_INDEX)
2107EXT4_FEATURE_COMPAT_FUNCS(sparse_super2,	SPARSE_SUPER2)
2108EXT4_FEATURE_COMPAT_FUNCS(fast_commit,		FAST_COMMIT)
2109EXT4_FEATURE_COMPAT_FUNCS(stable_inodes,	STABLE_INODES)
2110EXT4_FEATURE_COMPAT_FUNCS(orphan_file,		ORPHAN_FILE)
2111
2112EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super,	SPARSE_SUPER)
2113EXT4_FEATURE_RO_COMPAT_FUNCS(large_file,	LARGE_FILE)
2114EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir,		BTREE_DIR)
2115EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file,		HUGE_FILE)
2116EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum,		GDT_CSUM)
2117EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink,		DIR_NLINK)
2118EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize,	EXTRA_ISIZE)
2119EXT4_FEATURE_RO_COMPAT_FUNCS(quota,		QUOTA)
2120EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc,		BIGALLOC)
2121EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum,	METADATA_CSUM)
2122EXT4_FEATURE_RO_COMPAT_FUNCS(readonly,		READONLY)
2123EXT4_FEATURE_RO_COMPAT_FUNCS(project,		PROJECT)
2124EXT4_FEATURE_RO_COMPAT_FUNCS(verity,		VERITY)
2125EXT4_FEATURE_RO_COMPAT_FUNCS(orphan_present,	ORPHAN_PRESENT)
2126
2127EXT4_FEATURE_INCOMPAT_FUNCS(compression,	COMPRESSION)
2128EXT4_FEATURE_INCOMPAT_FUNCS(filetype,		FILETYPE)
2129EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery,	RECOVER)
2130EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev,	JOURNAL_DEV)
2131EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg,		META_BG)
2132EXT4_FEATURE_INCOMPAT_FUNCS(extents,		EXTENTS)
2133EXT4_FEATURE_INCOMPAT_FUNCS(64bit,		64BIT)
2134EXT4_FEATURE_INCOMPAT_FUNCS(mmp,		MMP)
2135EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg,		FLEX_BG)
2136EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode,		EA_INODE)
2137EXT4_FEATURE_INCOMPAT_FUNCS(dirdata,		DIRDATA)
2138EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed,		CSUM_SEED)
2139EXT4_FEATURE_INCOMPAT_FUNCS(largedir,		LARGEDIR)
2140EXT4_FEATURE_INCOMPAT_FUNCS(inline_data,	INLINE_DATA)
2141EXT4_FEATURE_INCOMPAT_FUNCS(encrypt,		ENCRYPT)
2142EXT4_FEATURE_INCOMPAT_FUNCS(casefold,		CASEFOLD)
2143
2144#define EXT2_FEATURE_COMPAT_SUPP	EXT4_FEATURE_COMPAT_EXT_ATTR
2145#define EXT2_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
2146					 EXT4_FEATURE_INCOMPAT_META_BG)
2147#define EXT2_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2148					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2149					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
2150
2151#define EXT3_FEATURE_COMPAT_SUPP	EXT4_FEATURE_COMPAT_EXT_ATTR
2152#define EXT3_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
2153					 EXT4_FEATURE_INCOMPAT_RECOVER| \
2154					 EXT4_FEATURE_INCOMPAT_META_BG)
2155#define EXT3_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2156					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2157					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
2158
2159#define EXT4_FEATURE_COMPAT_SUPP	(EXT4_FEATURE_COMPAT_EXT_ATTR| \
2160					 EXT4_FEATURE_COMPAT_ORPHAN_FILE)
2161#define EXT4_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
2162					 EXT4_FEATURE_INCOMPAT_RECOVER| \
2163					 EXT4_FEATURE_INCOMPAT_META_BG| \
2164					 EXT4_FEATURE_INCOMPAT_EXTENTS| \
2165					 EXT4_FEATURE_INCOMPAT_64BIT| \
2166					 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
2167					 EXT4_FEATURE_INCOMPAT_EA_INODE| \
2168					 EXT4_FEATURE_INCOMPAT_MMP | \
2169					 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
2170					 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
2171					 EXT4_FEATURE_INCOMPAT_CASEFOLD | \
2172					 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \
2173					 EXT4_FEATURE_INCOMPAT_LARGEDIR)
2174#define EXT4_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2175					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2176					 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
2177					 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
2178					 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
2179					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
2180					 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
2181					 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
2182					 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
2183					 EXT4_FEATURE_RO_COMPAT_QUOTA |\
2184					 EXT4_FEATURE_RO_COMPAT_PROJECT |\
2185					 EXT4_FEATURE_RO_COMPAT_VERITY |\
2186					 EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT)
2187
2188#define EXTN_FEATURE_FUNCS(ver) \
2189static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
2190{ \
2191	return ((EXT4_SB(sb)->s_es->s_feature_compat & \
2192		cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
2193} \
2194static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
2195{ \
2196	return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
2197		cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
2198} \
2199static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
2200{ \
2201	return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
2202		cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
2203}
2204
2205EXTN_FEATURE_FUNCS(2)
2206EXTN_FEATURE_FUNCS(3)
2207EXTN_FEATURE_FUNCS(4)
2208
2209static inline bool ext4_has_compat_features(struct super_block *sb)
2210{
2211	return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
2212}
2213static inline bool ext4_has_ro_compat_features(struct super_block *sb)
2214{
2215	return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
2216}
2217static inline bool ext4_has_incompat_features(struct super_block *sb)
2218{
2219	return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
2220}
2221
2222extern int ext4_feature_set_ok(struct super_block *sb, int readonly);
2223
2224/*
2225 * Superblock flags
2226 */
2227#define EXT4_FLAGS_RESIZING	0
2228#define EXT4_FLAGS_SHUTDOWN	1
2229#define EXT4_FLAGS_BDEV_IS_DAX	2
2230
2231static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi)
2232{
2233	return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags);
2234}
2235
2236/*
2237 * Default values for user and/or group using reserved blocks
2238 */
2239#define	EXT4_DEF_RESUID		0
2240#define	EXT4_DEF_RESGID		0
2241
2242/*
2243 * Default project ID
2244 */
2245#define	EXT4_DEF_PROJID		0
2246
2247#define EXT4_DEF_INODE_READAHEAD_BLKS	32
2248
2249/*
2250 * Default mount options
2251 */
2252#define EXT4_DEFM_DEBUG		0x0001
2253#define EXT4_DEFM_BSDGROUPS	0x0002
2254#define EXT4_DEFM_XATTR_USER	0x0004
2255#define EXT4_DEFM_ACL		0x0008
2256#define EXT4_DEFM_UID16		0x0010
2257#define EXT4_DEFM_JMODE		0x0060
2258#define EXT4_DEFM_JMODE_DATA	0x0020
2259#define EXT4_DEFM_JMODE_ORDERED	0x0040
2260#define EXT4_DEFM_JMODE_WBACK	0x0060
2261#define EXT4_DEFM_NOBARRIER	0x0100
2262#define EXT4_DEFM_BLOCK_VALIDITY 0x0200
2263#define EXT4_DEFM_DISCARD	0x0400
2264#define EXT4_DEFM_NODELALLOC	0x0800
2265
2266/*
2267 * Default journal batch times
2268 */
2269#define EXT4_DEF_MIN_BATCH_TIME	0
2270#define EXT4_DEF_MAX_BATCH_TIME	15000 /* 15ms */
2271
2272/*
2273 * Minimum number of groups in a flexgroup before we separate out
2274 * directories into the first block group of a flexgroup
2275 */
2276#define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME	4
2277
2278/*
2279 * Structure of a directory entry
2280 */
2281#define EXT4_NAME_LEN 255
2282
2283struct ext4_dir_entry {
2284	__le32	inode;			/* Inode number */
2285	__le16	rec_len;		/* Directory entry length */
2286	__le16	name_len;		/* Name length */
2287	char	name[EXT4_NAME_LEN];	/* File name */
2288};
2289
2290
2291/*
2292 * Encrypted Casefolded entries require saving the hash on disk. This structure
2293 * followed ext4_dir_entry_2's name[name_len] at the next 4 byte aligned
2294 * boundary.
2295 */
2296struct ext4_dir_entry_hash {
2297	__le32 hash;
2298	__le32 minor_hash;
2299};
2300
2301/*
2302 * The new version of the directory entry.  Since EXT4 structures are
2303 * stored in intel byte order, and the name_len field could never be
2304 * bigger than 255 chars, it's safe to reclaim the extra byte for the
2305 * file_type field.
2306 */
2307struct ext4_dir_entry_2 {
2308	__le32	inode;			/* Inode number */
2309	__le16	rec_len;		/* Directory entry length */
2310	__u8	name_len;		/* Name length */
2311	__u8	file_type;		/* See file type macros EXT4_FT_* below */
2312	char	name[EXT4_NAME_LEN];	/* File name */
2313};
2314
2315/*
2316 * Access the hashes at the end of ext4_dir_entry_2
2317 */
2318#define EXT4_DIRENT_HASHES(entry) \
2319	((struct ext4_dir_entry_hash *) \
2320		(((void *)(entry)) + \
2321		((8 + (entry)->name_len + EXT4_DIR_ROUND) & ~EXT4_DIR_ROUND)))
2322#define EXT4_DIRENT_HASH(entry) le32_to_cpu(EXT4_DIRENT_HASHES(de)->hash)
2323#define EXT4_DIRENT_MINOR_HASH(entry) \
2324		le32_to_cpu(EXT4_DIRENT_HASHES(de)->minor_hash)
2325
2326static inline bool ext4_hash_in_dirent(const struct inode *inode)
2327{
2328	return IS_CASEFOLDED(inode) && IS_ENCRYPTED(inode);
2329}
2330
2331/*
2332 * This is a bogus directory entry at the end of each leaf block that
2333 * records checksums.
2334 */
2335struct ext4_dir_entry_tail {
2336	__le32	det_reserved_zero1;	/* Pretend to be unused */
2337	__le16	det_rec_len;		/* 12 */
2338	__u8	det_reserved_zero2;	/* Zero name length */
2339	__u8	det_reserved_ft;	/* 0xDE, fake file type */
2340	__le32	det_checksum;		/* crc32c(uuid+inum+dirblock) */
2341};
2342
2343#define EXT4_DIRENT_TAIL(block, blocksize) \
2344	((struct ext4_dir_entry_tail *)(((void *)(block)) + \
2345					((blocksize) - \
2346					 sizeof(struct ext4_dir_entry_tail))))
2347
2348/*
2349 * Ext4 directory file types.  Only the low 3 bits are used.  The
2350 * other bits are reserved for now.
2351 */
2352#define EXT4_FT_UNKNOWN		0
2353#define EXT4_FT_REG_FILE	1
2354#define EXT4_FT_DIR		2
2355#define EXT4_FT_CHRDEV		3
2356#define EXT4_FT_BLKDEV		4
2357#define EXT4_FT_FIFO		5
2358#define EXT4_FT_SOCK		6
2359#define EXT4_FT_SYMLINK		7
2360
2361#define EXT4_FT_MAX		8
2362
2363#define EXT4_FT_DIR_CSUM	0xDE
2364
2365/*
2366 * EXT4_DIR_PAD defines the directory entries boundaries
2367 *
2368 * NOTE: It must be a multiple of 4
2369 */
2370#define EXT4_DIR_PAD			4
2371#define EXT4_DIR_ROUND			(EXT4_DIR_PAD - 1)
2372#define EXT4_MAX_REC_LEN		((1<<16)-1)
2373
2374/*
2375 * The rec_len is dependent on the type of directory. Directories that are
2376 * casefolded and encrypted need to store the hash as well, so we add room for
2377 * ext4_extended_dir_entry_2. For all entries related to '.' or '..' you should
2378 * pass NULL for dir, as those entries do not use the extra fields.
2379 */
2380static inline unsigned int ext4_dir_rec_len(__u8 name_len,
2381						const struct inode *dir)
2382{
2383	int rec_len = (name_len + 8 + EXT4_DIR_ROUND);
2384
2385	if (dir && ext4_hash_in_dirent(dir))
2386		rec_len += sizeof(struct ext4_dir_entry_hash);
2387	return (rec_len & ~EXT4_DIR_ROUND);
2388}
2389
2390/*
2391 * If we ever get support for fs block sizes > page_size, we'll need
2392 * to remove the #if statements in the next two functions...
2393 */
2394static inline unsigned int
2395ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
2396{
2397	unsigned len = le16_to_cpu(dlen);
2398
2399#if (PAGE_SIZE >= 65536)
2400	if (len == EXT4_MAX_REC_LEN || len == 0)
2401		return blocksize;
2402	return (len & 65532) | ((len & 3) << 16);
2403#else
2404	return len;
2405#endif
2406}
2407
2408static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
2409{
2410	if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
2411		BUG();
2412#if (PAGE_SIZE >= 65536)
2413	if (len < 65536)
2414		return cpu_to_le16(len);
2415	if (len == blocksize) {
2416		if (blocksize == 65536)
2417			return cpu_to_le16(EXT4_MAX_REC_LEN);
2418		else
2419			return cpu_to_le16(0);
2420	}
2421	return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
2422#else
2423	return cpu_to_le16(len);
2424#endif
2425}
2426
2427/*
2428 * Hash Tree Directory indexing
2429 * (c) Daniel Phillips, 2001
2430 */
2431
2432#define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
2433		    ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2434#define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \
2435		    !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir)))
2436#define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2437
2438/* Legal values for the dx_root hash_version field: */
2439
2440#define DX_HASH_LEGACY			0
2441#define DX_HASH_HALF_MD4		1
2442#define DX_HASH_TEA			2
2443#define DX_HASH_LEGACY_UNSIGNED		3
2444#define DX_HASH_HALF_MD4_UNSIGNED	4
2445#define DX_HASH_TEA_UNSIGNED		5
2446#define DX_HASH_SIPHASH			6
2447
2448static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2449			      const void *address, unsigned int length)
2450{
2451	struct {
2452		struct shash_desc shash;
2453		char ctx[4];
2454	} desc;
2455
2456	BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2457
2458	desc.shash.tfm = sbi->s_chksum_driver;
2459	*(u32 *)desc.ctx = crc;
2460
2461	BUG_ON(crypto_shash_update(&desc.shash, address, length));
2462
2463	return *(u32 *)desc.ctx;
2464}
2465
2466#ifdef __KERNEL__
2467
2468/* hash info structure used by the directory hash */
2469struct dx_hash_info
2470{
2471	u32		hash;
2472	u32		minor_hash;
2473	int		hash_version;
2474	u32		*seed;
2475};
2476
2477
2478/* 32 and 64 bit signed EOF for dx directories */
2479#define EXT4_HTREE_EOF_32BIT   ((1UL  << (32 - 1)) - 1)
2480#define EXT4_HTREE_EOF_64BIT   ((1ULL << (64 - 1)) - 1)
2481
2482
2483/*
2484 * Control parameters used by ext4_htree_next_block
2485 */
2486#define HASH_NB_ALWAYS		1
2487
2488struct ext4_filename {
2489	const struct qstr *usr_fname;
2490	struct fscrypt_str disk_name;
2491	struct dx_hash_info hinfo;
2492#ifdef CONFIG_FS_ENCRYPTION
2493	struct fscrypt_str crypto_buf;
2494#endif
2495#ifdef CONFIG_UNICODE
2496	struct fscrypt_str cf_name;
2497#endif
2498};
2499
2500#define fname_name(p) ((p)->disk_name.name)
2501#define fname_usr_name(p) ((p)->usr_fname->name)
2502#define fname_len(p)  ((p)->disk_name.len)
2503
2504/*
2505 * Describe an inode's exact location on disk and in memory
2506 */
2507struct ext4_iloc
2508{
2509	struct buffer_head *bh;
2510	unsigned long offset;
2511	ext4_group_t block_group;
2512};
2513
2514static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2515{
2516	return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2517}
2518
2519static inline bool ext4_is_quota_file(struct inode *inode)
2520{
2521	return IS_NOQUOTA(inode) &&
2522	       !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL);
2523}
2524
2525/*
2526 * This structure is stuffed into the struct file's private_data field
2527 * for directories.  It is where we put information so that we can do
2528 * readdir operations in hash tree order.
2529 */
2530struct dir_private_info {
2531	struct rb_root	root;
2532	struct rb_node	*curr_node;
2533	struct fname	*extra_fname;
2534	loff_t		last_pos;
2535	__u32		curr_hash;
2536	__u32		curr_minor_hash;
2537	__u32		next_hash;
2538};
2539
2540/* calculate the first block number of the group */
2541static inline ext4_fsblk_t
2542ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2543{
2544	return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2545		le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2546}
2547
2548/*
2549 * Special error return code only used by dx_probe() and its callers.
2550 */
2551#define ERR_BAD_DX_DIR	(-(MAX_ERRNO - 1))
2552
2553/* htree levels for ext4 */
2554#define	EXT4_HTREE_LEVEL_COMPAT	2
2555#define	EXT4_HTREE_LEVEL	3
2556
2557static inline int ext4_dir_htree_level(struct super_block *sb)
2558{
2559	return ext4_has_feature_largedir(sb) ?
2560		EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT;
2561}
2562
2563/*
2564 * Timeout and state flag for lazy initialization inode thread.
2565 */
2566#define EXT4_DEF_LI_WAIT_MULT			10
2567#define EXT4_DEF_LI_MAX_START_DELAY		5
2568#define EXT4_LAZYINIT_QUIT			0x0001
2569#define EXT4_LAZYINIT_RUNNING			0x0002
2570
2571/*
2572 * Lazy inode table initialization info
2573 */
2574struct ext4_lazy_init {
2575	unsigned long		li_state;
2576	struct list_head	li_request_list;
2577	struct mutex		li_list_mtx;
2578};
2579
2580enum ext4_li_mode {
2581	EXT4_LI_MODE_PREFETCH_BBITMAP,
2582	EXT4_LI_MODE_ITABLE,
2583};
2584
2585struct ext4_li_request {
2586	struct super_block	*lr_super;
2587	enum ext4_li_mode	lr_mode;
2588	ext4_group_t		lr_first_not_zeroed;
2589	ext4_group_t		lr_next_group;
2590	struct list_head	lr_request;
2591	unsigned long		lr_next_sched;
2592	unsigned long		lr_timeout;
2593};
2594
2595struct ext4_features {
2596	struct kobject f_kobj;
2597	struct completion f_kobj_unregister;
2598};
2599
2600/*
2601 * This structure will be used for multiple mount protection. It will be
2602 * written into the block number saved in the s_mmp_block field in the
2603 * superblock. Programs that check MMP should assume that if
2604 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2605 * to use the filesystem, regardless of how old the timestamp is.
2606 */
2607#define EXT4_MMP_MAGIC     0x004D4D50U /* ASCII for MMP */
2608#define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2609#define EXT4_MMP_SEQ_FSCK  0xE24D4D50U /* mmp_seq value when being fscked */
2610#define EXT4_MMP_SEQ_MAX   0xE24D4D4FU /* maximum valid mmp_seq value */
2611
2612struct mmp_struct {
2613	__le32	mmp_magic;		/* Magic number for MMP */
2614	__le32	mmp_seq;		/* Sequence no. updated periodically */
2615
2616	/*
2617	 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2618	 * purposes and do not affect the correctness of the algorithm
2619	 */
2620	__le64	mmp_time;		/* Time last updated */
2621	char	mmp_nodename[64];	/* Node which last updated MMP block */
2622	char	mmp_bdevname[32];	/* Bdev which last updated MMP block */
2623
2624	/*
2625	 * mmp_check_interval is used to verify if the MMP block has been
2626	 * updated on the block device. The value is updated based on the
2627	 * maximum time to write the MMP block during an update cycle.
2628	 */
2629	__le16	mmp_check_interval;
2630
2631	__le16	mmp_pad1;
2632	__le32	mmp_pad2[226];
2633	__le32	mmp_checksum;		/* crc32c(uuid+mmp_block) */
2634};
2635
2636/* arguments passed to the mmp thread */
2637struct mmpd_data {
2638	struct buffer_head *bh; /* bh from initial read_mmp_block() */
2639	struct super_block *sb;  /* super block of the fs */
2640};
2641
2642/*
2643 * Check interval multiplier
2644 * The MMP block is written every update interval and initially checked every
2645 * update interval x the multiplier (the value is then adapted based on the
2646 * write latency). The reason is that writes can be delayed under load and we
2647 * don't want readers to incorrectly assume that the filesystem is no longer
2648 * in use.
2649 */
2650#define EXT4_MMP_CHECK_MULT		2UL
2651
2652/*
2653 * Minimum interval for MMP checking in seconds.
2654 */
2655#define EXT4_MMP_MIN_CHECK_INTERVAL	5UL
2656
2657/*
2658 * Maximum interval for MMP checking in seconds.
2659 */
2660#define EXT4_MMP_MAX_CHECK_INTERVAL	300UL
2661
2662/*
2663 * Function prototypes
2664 */
2665
2666/*
2667 * Ok, these declarations are also in <linux/kernel.h> but none of the
2668 * ext4 source programs needs to include it so they are duplicated here.
2669 */
2670# define NORET_TYPE	/**/
2671# define ATTRIB_NORET	__attribute__((noreturn))
2672# define NORET_AND	noreturn,
2673
2674/* bitmap.c */
2675extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2676void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2677				struct ext4_group_desc *gdp,
2678				struct buffer_head *bh, int sz);
2679int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2680				  struct ext4_group_desc *gdp,
2681				  struct buffer_head *bh, int sz);
2682void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2683				struct ext4_group_desc *gdp,
2684				struct buffer_head *bh);
2685int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2686				  struct ext4_group_desc *gdp,
2687				  struct buffer_head *bh);
2688
2689/* balloc.c */
2690extern void ext4_get_group_no_and_offset(struct super_block *sb,
2691					 ext4_fsblk_t blocknr,
2692					 ext4_group_t *blockgrpp,
2693					 ext4_grpblk_t *offsetp);
2694extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2695					  ext4_fsblk_t block);
2696
2697extern unsigned int ext4_block_group(struct super_block *sb,
2698			ext4_fsblk_t blocknr);
2699extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2700			ext4_fsblk_t blocknr);
2701extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2702extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2703			ext4_group_t group);
2704extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2705					 ext4_fsblk_t goal,
2706					 unsigned int flags,
2707					 unsigned long *count,
2708					 int *errp);
2709extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2710				    s64 nclusters, unsigned int flags);
2711extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2712extern void ext4_check_blocks_bitmap(struct super_block *);
2713extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2714						    ext4_group_t block_group,
2715						    struct buffer_head ** bh);
2716extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2717
2718extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2719						ext4_group_t block_group,
2720						bool ignore_locked);
2721extern int ext4_wait_block_bitmap(struct super_block *sb,
2722				  ext4_group_t block_group,
2723				  struct buffer_head *bh);
2724extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2725						  ext4_group_t block_group);
2726extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2727					      ext4_group_t block_group,
2728					      struct ext4_group_desc *gdp);
2729ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2730
2731#ifdef CONFIG_UNICODE
2732extern int ext4_fname_setup_ci_filename(struct inode *dir,
2733					 const struct qstr *iname,
2734					 struct ext4_filename *fname);
2735#endif
2736
2737#ifdef CONFIG_FS_ENCRYPTION
2738static inline void ext4_fname_from_fscrypt_name(struct ext4_filename *dst,
2739						const struct fscrypt_name *src)
2740{
2741	memset(dst, 0, sizeof(*dst));
2742
2743	dst->usr_fname = src->usr_fname;
2744	dst->disk_name = src->disk_name;
2745	dst->hinfo.hash = src->hash;
2746	dst->hinfo.minor_hash = src->minor_hash;
2747	dst->crypto_buf = src->crypto_buf;
2748}
2749
2750static inline int ext4_fname_setup_filename(struct inode *dir,
2751					    const struct qstr *iname,
2752					    int lookup,
2753					    struct ext4_filename *fname)
2754{
2755	struct fscrypt_name name;
2756	int err;
2757
2758	err = fscrypt_setup_filename(dir, iname, lookup, &name);
2759	if (err)
2760		return err;
2761
2762	ext4_fname_from_fscrypt_name(fname, &name);
2763
2764#ifdef CONFIG_UNICODE
2765	err = ext4_fname_setup_ci_filename(dir, iname, fname);
2766#endif
2767	return err;
2768}
2769
2770static inline int ext4_fname_prepare_lookup(struct inode *dir,
2771					    struct dentry *dentry,
2772					    struct ext4_filename *fname)
2773{
2774	struct fscrypt_name name;
2775	int err;
2776
2777	err = fscrypt_prepare_lookup(dir, dentry, &name);
2778	if (err)
2779		return err;
2780
2781	ext4_fname_from_fscrypt_name(fname, &name);
2782
2783#ifdef CONFIG_UNICODE
2784	err = ext4_fname_setup_ci_filename(dir, &dentry->d_name, fname);
2785#endif
2786	return err;
2787}
2788
2789static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2790{
2791	struct fscrypt_name name;
2792
2793	name.crypto_buf = fname->crypto_buf;
2794	fscrypt_free_filename(&name);
2795
2796	fname->crypto_buf.name = NULL;
2797	fname->usr_fname = NULL;
2798	fname->disk_name.name = NULL;
2799
2800#ifdef CONFIG_UNICODE
2801	kfree(fname->cf_name.name);
2802	fname->cf_name.name = NULL;
2803#endif
2804}
2805#else /* !CONFIG_FS_ENCRYPTION */
2806static inline int ext4_fname_setup_filename(struct inode *dir,
2807					    const struct qstr *iname,
2808					    int lookup,
2809					    struct ext4_filename *fname)
2810{
2811	int err = 0;
2812	fname->usr_fname = iname;
2813	fname->disk_name.name = (unsigned char *) iname->name;
2814	fname->disk_name.len = iname->len;
2815
2816#ifdef CONFIG_UNICODE
2817	err = ext4_fname_setup_ci_filename(dir, iname, fname);
2818#endif
2819
2820	return err;
2821}
2822
2823static inline int ext4_fname_prepare_lookup(struct inode *dir,
2824					    struct dentry *dentry,
2825					    struct ext4_filename *fname)
2826{
2827	return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname);
2828}
2829
2830static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2831{
2832#ifdef CONFIG_UNICODE
2833	kfree(fname->cf_name.name);
2834	fname->cf_name.name = NULL;
2835#endif
2836}
2837#endif /* !CONFIG_FS_ENCRYPTION */
2838
2839/* dir.c */
2840extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2841				  struct file *,
2842				  struct ext4_dir_entry_2 *,
2843				  struct buffer_head *, char *, int,
2844				  unsigned int);
2845#define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2846	unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2847				(de), (bh), (buf), (size), (offset)))
2848extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2849				__u32 minor_hash,
2850				struct ext4_dir_entry_2 *dirent,
2851				struct fscrypt_str *ent_name);
2852extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2853extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2854			     struct buffer_head *bh,
2855			     void *buf, int buf_size,
2856			     struct ext4_filename *fname,
2857			     struct ext4_dir_entry_2 **dest_de);
2858void ext4_insert_dentry(struct inode *dir, struct inode *inode,
2859			struct ext4_dir_entry_2 *de,
2860			int buf_size,
2861			struct ext4_filename *fname);
2862static inline void ext4_update_dx_flag(struct inode *inode)
2863{
2864	if (!ext4_has_feature_dir_index(inode->i_sb) &&
2865	    ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) {
2866		/* ext4_iget() should have caught this... */
2867		WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb));
2868		ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2869	}
2870}
2871static const unsigned char ext4_filetype_table[] = {
2872	DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2873};
2874
2875static inline  unsigned char get_dtype(struct super_block *sb, int filetype)
2876{
2877	if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2878		return DT_UNKNOWN;
2879
2880	return ext4_filetype_table[filetype];
2881}
2882extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2883			     void *buf, int buf_size);
2884
2885/* fsync.c */
2886extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2887
2888/* hash.c */
2889extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len,
2890			  struct dx_hash_info *hinfo);
2891
2892/* ialloc.c */
2893extern int ext4_mark_inode_used(struct super_block *sb, int ino);
2894extern struct inode *__ext4_new_inode(struct user_namespace *, handle_t *,
2895				      struct inode *, umode_t,
2896				      const struct qstr *qstr, __u32 goal,
2897				      uid_t *owner, __u32 i_flags,
2898				      int handle_type, unsigned int line_no,
2899				      int nblocks);
2900
2901#define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags)          \
2902	__ext4_new_inode(&init_user_ns, (handle), (dir), (mode), (qstr),       \
2903			 (goal), (owner), i_flags, 0, 0, 0)
2904#define ext4_new_inode_start_handle(mnt_userns, dir, mode, qstr, goal, owner, \
2905				    type, nblocks)		    \
2906	__ext4_new_inode((mnt_userns), NULL, (dir), (mode), (qstr), (goal), (owner), \
2907			 0, (type), __LINE__, (nblocks))
2908
2909
2910extern void ext4_free_inode(handle_t *, struct inode *);
2911extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2912extern unsigned long ext4_count_free_inodes(struct super_block *);
2913extern unsigned long ext4_count_dirs(struct super_block *);
2914extern void ext4_check_inodes_bitmap(struct super_block *);
2915extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2916extern int ext4_init_inode_table(struct super_block *sb,
2917				 ext4_group_t group, int barrier);
2918extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2919
2920/* fast_commit.c */
2921int ext4_fc_info_show(struct seq_file *seq, void *v);
2922void ext4_fc_init(struct super_block *sb, journal_t *journal);
2923void ext4_fc_init_inode(struct inode *inode);
2924void ext4_fc_track_range(handle_t *handle, struct inode *inode, ext4_lblk_t start,
2925			 ext4_lblk_t end);
2926void __ext4_fc_track_unlink(handle_t *handle, struct inode *inode,
2927	struct dentry *dentry);
2928void __ext4_fc_track_link(handle_t *handle, struct inode *inode,
2929	struct dentry *dentry);
2930void ext4_fc_track_unlink(handle_t *handle, struct dentry *dentry);
2931void ext4_fc_track_link(handle_t *handle, struct dentry *dentry);
2932void __ext4_fc_track_create(handle_t *handle, struct inode *inode,
2933			    struct dentry *dentry);
2934void ext4_fc_track_create(handle_t *handle, struct dentry *dentry);
2935void ext4_fc_track_inode(handle_t *handle, struct inode *inode);
2936void ext4_fc_mark_ineligible(struct super_block *sb, int reason);
2937void ext4_fc_start_ineligible(struct super_block *sb, int reason);
2938void ext4_fc_stop_ineligible(struct super_block *sb);
2939void ext4_fc_start_update(struct inode *inode);
2940void ext4_fc_stop_update(struct inode *inode);
2941void ext4_fc_del(struct inode *inode);
2942bool ext4_fc_replay_check_excluded(struct super_block *sb, ext4_fsblk_t block);
2943void ext4_fc_replay_cleanup(struct super_block *sb);
2944int ext4_fc_commit(journal_t *journal, tid_t commit_tid);
2945int __init ext4_fc_init_dentry_cache(void);
2946
2947/* mballoc.c */
2948extern const struct seq_operations ext4_mb_seq_groups_ops;
2949extern const struct seq_operations ext4_mb_seq_structs_summary_ops;
2950extern long ext4_mb_stats;
2951extern long ext4_mb_max_to_scan;
2952extern int ext4_seq_mb_stats_show(struct seq_file *seq, void *offset);
2953extern int ext4_mb_init(struct super_block *);
2954extern int ext4_mb_release(struct super_block *);
2955extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2956				struct ext4_allocation_request *, int *);
2957extern int ext4_mb_reserve_blocks(struct super_block *, int);
2958extern void ext4_discard_preallocations(struct inode *, unsigned int);
2959extern int __init ext4_init_mballoc(void);
2960extern void ext4_exit_mballoc(void);
2961extern ext4_group_t ext4_mb_prefetch(struct super_block *sb,
2962				     ext4_group_t group,
2963				     unsigned int nr, int *cnt);
2964extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group,
2965				  unsigned int nr);
2966
2967extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2968			     struct buffer_head *bh, ext4_fsblk_t block,
2969			     unsigned long count, int flags);
2970extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2971				   ext4_group_t ngroups);
2972extern int ext4_mb_add_groupinfo(struct super_block *sb,
2973		ext4_group_t i, struct ext4_group_desc *desc);
2974extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2975				ext4_fsblk_t block, unsigned long count);
2976extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2977extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid);
2978extern void ext4_mb_mark_bb(struct super_block *sb, ext4_fsblk_t block,
2979		       int len, int state);
2980
2981/* inode.c */
2982void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw,
2983			 struct ext4_inode_info *ei);
2984int ext4_inode_is_fast_symlink(struct inode *inode);
2985struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2986struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2987int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count,
2988		     bool wait, struct buffer_head **bhs);
2989int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2990			     struct buffer_head *bh_result, int create);
2991int ext4_get_block(struct inode *inode, sector_t iblock,
2992		   struct buffer_head *bh_result, int create);
2993int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2994			   struct buffer_head *bh, int create);
2995int ext4_walk_page_buffers(handle_t *handle,
2996			   struct inode *inode,
2997			   struct buffer_head *head,
2998			   unsigned from,
2999			   unsigned to,
3000			   int *partial,
3001			   int (*fn)(handle_t *handle, struct inode *inode,
3002				     struct buffer_head *bh));
3003int do_journal_get_write_access(handle_t *handle, struct inode *inode,
3004				struct buffer_head *bh);
3005#define FALL_BACK_TO_NONDELALLOC 1
3006#define CONVERT_INLINE_DATA	 2
3007
3008typedef enum {
3009	EXT4_IGET_NORMAL =	0,
3010	EXT4_IGET_SPECIAL =	0x0001, /* OK to iget a system inode */
3011	EXT4_IGET_HANDLE = 	0x0002	/* Inode # is from a handle */
3012} ext4_iget_flags;
3013
3014extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
3015				 ext4_iget_flags flags, const char *function,
3016				 unsigned int line);
3017
3018#define ext4_iget(sb, ino, flags) \
3019	__ext4_iget((sb), (ino), (flags), __func__, __LINE__)
3020
3021extern int  ext4_write_inode(struct inode *, struct writeback_control *);
3022extern int  ext4_setattr(struct user_namespace *, struct dentry *,
3023			 struct iattr *);
3024extern int  ext4_getattr(struct user_namespace *, const struct path *,
3025			 struct kstat *, u32, unsigned int);
3026extern void ext4_evict_inode(struct inode *);
3027extern void ext4_clear_inode(struct inode *);
3028extern int  ext4_file_getattr(struct user_namespace *, const struct path *,
3029			      struct kstat *, u32, unsigned int);
3030extern int  ext4_sync_inode(handle_t *, struct inode *);
3031extern void ext4_dirty_inode(struct inode *, int);
3032extern int ext4_change_inode_journal_flag(struct inode *, int);
3033extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
3034extern int ext4_get_fc_inode_loc(struct super_block *sb, unsigned long ino,
3035			  struct ext4_iloc *iloc);
3036extern int ext4_inode_attach_jinode(struct inode *inode);
3037extern int ext4_can_truncate(struct inode *inode);
3038extern int ext4_truncate(struct inode *);
3039extern int ext4_break_layouts(struct inode *);
3040extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
3041extern void ext4_set_inode_flags(struct inode *, bool init);
3042extern int ext4_alloc_da_blocks(struct inode *inode);
3043extern void ext4_set_aops(struct inode *inode);
3044extern int ext4_writepage_trans_blocks(struct inode *);
3045extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
3046extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
3047			     loff_t lstart, loff_t lend);
3048extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf);
3049extern vm_fault_t ext4_filemap_fault(struct vm_fault *vmf);
3050extern qsize_t *ext4_get_reserved_space(struct inode *inode);
3051extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
3052extern void ext4_da_release_space(struct inode *inode, int to_free);
3053extern void ext4_da_update_reserve_space(struct inode *inode,
3054					int used, int quota_claim);
3055extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
3056			      ext4_fsblk_t pblk, ext4_lblk_t len);
3057
3058/* indirect.c */
3059extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
3060				struct ext4_map_blocks *map, int flags);
3061extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
3062extern void ext4_ind_truncate(handle_t *, struct inode *inode);
3063extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
3064				 ext4_lblk_t start, ext4_lblk_t end);
3065
3066/* ioctl.c */
3067extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
3068extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
3069int ext4_fileattr_set(struct user_namespace *mnt_userns,
3070		      struct dentry *dentry, struct fileattr *fa);
3071int ext4_fileattr_get(struct dentry *dentry, struct fileattr *fa);
3072extern void ext4_reset_inode_seed(struct inode *inode);
3073
3074/* migrate.c */
3075extern int ext4_ext_migrate(struct inode *);
3076extern int ext4_ind_migrate(struct inode *inode);
3077
3078/* namei.c */
3079extern int ext4_init_new_dir(handle_t *handle, struct inode *dir,
3080			     struct inode *inode);
3081extern int ext4_dirblock_csum_verify(struct inode *inode,
3082				     struct buffer_head *bh);
3083extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
3084				__u32 start_minor_hash, __u32 *next_hash);
3085extern int ext4_search_dir(struct buffer_head *bh,
3086			   char *search_buf,
3087			   int buf_size,
3088			   struct inode *dir,
3089			   struct ext4_filename *fname,
3090			   unsigned int offset,
3091			   struct ext4_dir_entry_2 **res_dir);
3092extern int ext4_generic_delete_entry(struct inode *dir,
3093				     struct ext4_dir_entry_2 *de_del,
3094				     struct buffer_head *bh,
3095				     void *entry_buf,
3096				     int buf_size,
3097				     int csum_size);
3098extern bool ext4_empty_dir(struct inode *inode);
3099
3100/* resize.c */
3101extern void ext4_kvfree_array_rcu(void *to_free);
3102extern int ext4_group_add(struct super_block *sb,
3103				struct ext4_new_group_data *input);
3104extern int ext4_group_extend(struct super_block *sb,
3105				struct ext4_super_block *es,
3106				ext4_fsblk_t n_blocks_count);
3107extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
3108
3109/* super.c */
3110extern struct buffer_head *ext4_sb_bread(struct super_block *sb,
3111					 sector_t block, int op_flags);
3112extern struct buffer_head *ext4_sb_bread_unmovable(struct super_block *sb,
3113						   sector_t block);
3114extern void ext4_read_bh_nowait(struct buffer_head *bh, int op_flags,
3115				bh_end_io_t *end_io);
3116extern int ext4_read_bh(struct buffer_head *bh, int op_flags,
3117			bh_end_io_t *end_io);
3118extern int ext4_read_bh_lock(struct buffer_head *bh, int op_flags, bool wait);
3119extern void ext4_sb_breadahead_unmovable(struct super_block *sb, sector_t block);
3120extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
3121extern int ext4_calculate_overhead(struct super_block *sb);
3122extern void ext4_superblock_csum_set(struct super_block *sb);
3123extern int ext4_alloc_flex_bg_array(struct super_block *sb,
3124				    ext4_group_t ngroup);
3125extern const char *ext4_decode_error(struct super_block *sb, int errno,
3126				     char nbuf[16]);
3127extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb,
3128					     ext4_group_t block_group,
3129					     unsigned int flags);
3130
3131extern __printf(7, 8)
3132void __ext4_error(struct super_block *, const char *, unsigned int, bool,
3133		  int, __u64, const char *, ...);
3134extern __printf(6, 7)
3135void __ext4_error_inode(struct inode *, const char *, unsigned int,
3136			ext4_fsblk_t, int, const char *, ...);
3137extern __printf(5, 6)
3138void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
3139		     const char *, ...);
3140extern void __ext4_std_error(struct super_block *, const char *,
3141			     unsigned int, int);
3142extern __printf(4, 5)
3143void __ext4_warning(struct super_block *, const char *, unsigned int,
3144		    const char *, ...);
3145extern __printf(4, 5)
3146void __ext4_warning_inode(const struct inode *inode, const char *function,
3147			  unsigned int line, const char *fmt, ...);
3148extern __printf(3, 4)
3149void __ext4_msg(struct super_block *, const char *, const char *, ...);
3150extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
3151			   const char *, unsigned int, const char *);
3152extern __printf(7, 8)
3153void __ext4_grp_locked_error(const char *, unsigned int,
3154			     struct super_block *, ext4_group_t,
3155			     unsigned long, ext4_fsblk_t,
3156			     const char *, ...);
3157
3158#define EXT4_ERROR_INODE(inode, fmt, a...) \
3159	ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
3160
3161#define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...)			\
3162	__ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a)
3163
3164#define ext4_error_inode_block(inode, block, err, fmt, a...)		\
3165	__ext4_error_inode((inode), __func__, __LINE__, (block), (err),	\
3166			   (fmt), ## a)
3167
3168#define EXT4_ERROR_FILE(file, block, fmt, a...)				\
3169	ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
3170
3171#define ext4_abort(sb, err, fmt, a...)					\
3172	__ext4_error((sb), __func__, __LINE__, true, (err), 0, (fmt), ## a)
3173
3174#ifdef CONFIG_PRINTK
3175
3176#define ext4_error_inode(inode, func, line, block, fmt, ...)		\
3177	__ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__)
3178#define ext4_error_inode_err(inode, func, line, block, err, fmt, ...)	\
3179	__ext4_error_inode((inode), (func), (line), (block), 		\
3180			   (err), (fmt), ##__VA_ARGS__)
3181#define ext4_error_file(file, func, line, block, fmt, ...)		\
3182	__ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
3183#define ext4_error(sb, fmt, ...)					\
3184	__ext4_error((sb), __func__, __LINE__, false, 0, 0, (fmt),	\
3185		##__VA_ARGS__)
3186#define ext4_error_err(sb, err, fmt, ...)				\
3187	__ext4_error((sb), __func__, __LINE__, false, (err), 0, (fmt),	\
3188		##__VA_ARGS__)
3189#define ext4_warning(sb, fmt, ...)					\
3190	__ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
3191#define ext4_warning_inode(inode, fmt, ...)				\
3192	__ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
3193#define ext4_msg(sb, level, fmt, ...)				\
3194	__ext4_msg(sb, level, fmt, ##__VA_ARGS__)
3195#define dump_mmp_msg(sb, mmp, msg)					\
3196	__dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
3197#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...)		\
3198	__ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
3199				fmt, ##__VA_ARGS__)
3200
3201#else
3202
3203#define ext4_error_inode(inode, func, line, block, fmt, ...)		\
3204do {									\
3205	no_printk(fmt, ##__VA_ARGS__);					\
3206	__ext4_error_inode(inode, "", 0, block, 0, " ");		\
3207} while (0)
3208#define ext4_error_inode_err(inode, func, line, block, err, fmt, ...)	\
3209do {									\
3210	no_printk(fmt, ##__VA_ARGS__);					\
3211	__ext4_error_inode(inode, "", 0, block, err, " ");		\
3212} while (0)
3213#define ext4_error_file(file, func, line, block, fmt, ...)		\
3214do {									\
3215	no_printk(fmt, ##__VA_ARGS__);					\
3216	__ext4_error_file(file, "", 0, block, " ");			\
3217} while (0)
3218#define ext4_error(sb, fmt, ...)					\
3219do {									\
3220	no_printk(fmt, ##__VA_ARGS__);					\
3221	__ext4_error(sb, "", 0, false, 0, 0, " ");			\
3222} while (0)
3223#define ext4_error_err(sb, err, fmt, ...)				\
3224do {									\
3225	no_printk(fmt, ##__VA_ARGS__);					\
3226	__ext4_error(sb, "", 0, false, err, 0, " ");			\
3227} while (0)
3228#define ext4_warning(sb, fmt, ...)					\
3229do {									\
3230	no_printk(fmt, ##__VA_ARGS__);					\
3231	__ext4_warning(sb, "", 0, " ");					\
3232} while (0)
3233#define ext4_warning_inode(inode, fmt, ...)				\
3234do {									\
3235	no_printk(fmt, ##__VA_ARGS__);					\
3236	__ext4_warning_inode(inode, "", 0, " ");			\
3237} while (0)
3238#define ext4_msg(sb, level, fmt, ...)					\
3239do {									\
3240	no_printk(fmt, ##__VA_ARGS__);					\
3241	__ext4_msg(sb, "", " ");					\
3242} while (0)
3243#define dump_mmp_msg(sb, mmp, msg)					\
3244	__dump_mmp_msg(sb, mmp, "", 0, "")
3245#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...)		\
3246do {									\
3247	no_printk(fmt, ##__VA_ARGS__);				\
3248	__ext4_grp_locked_error("", 0, sb, grp, ino, block, " ");	\
3249} while (0)
3250
3251#endif
3252
3253extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
3254				      struct ext4_group_desc *bg);
3255extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
3256				      struct ext4_group_desc *bg);
3257extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
3258				     struct ext4_group_desc *bg);
3259extern __u32 ext4_free_group_clusters(struct super_block *sb,
3260				      struct ext4_group_desc *bg);
3261extern __u32 ext4_free_inodes_count(struct super_block *sb,
3262				 struct ext4_group_desc *bg);
3263extern __u32 ext4_used_dirs_count(struct super_block *sb,
3264				struct ext4_group_desc *bg);
3265extern __u32 ext4_itable_unused_count(struct super_block *sb,
3266				   struct ext4_group_desc *bg);
3267extern void ext4_block_bitmap_set(struct super_block *sb,
3268				  struct ext4_group_desc *bg, ext4_fsblk_t blk);
3269extern void ext4_inode_bitmap_set(struct super_block *sb,
3270				  struct ext4_group_desc *bg, ext4_fsblk_t blk);
3271extern void ext4_inode_table_set(struct super_block *sb,
3272				 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3273extern void ext4_free_group_clusters_set(struct super_block *sb,
3274					 struct ext4_group_desc *bg,
3275					 __u32 count);
3276extern void ext4_free_inodes_set(struct super_block *sb,
3277				struct ext4_group_desc *bg, __u32 count);
3278extern void ext4_used_dirs_set(struct super_block *sb,
3279				struct ext4_group_desc *bg, __u32 count);
3280extern void ext4_itable_unused_set(struct super_block *sb,
3281				   struct ext4_group_desc *bg, __u32 count);
3282extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
3283				       struct ext4_group_desc *gdp);
3284extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
3285				     struct ext4_group_desc *gdp);
3286extern int ext4_register_li_request(struct super_block *sb,
3287				    ext4_group_t first_not_zeroed);
3288
3289static inline int ext4_has_metadata_csum(struct super_block *sb)
3290{
3291	WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
3292		     !EXT4_SB(sb)->s_chksum_driver);
3293
3294	return ext4_has_feature_metadata_csum(sb) &&
3295	       (EXT4_SB(sb)->s_chksum_driver != NULL);
3296}
3297
3298static inline int ext4_has_group_desc_csum(struct super_block *sb)
3299{
3300	return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb);
3301}
3302
3303#define ext4_read_incompat_64bit_val(es, name) \
3304	(((es)->s_feature_incompat & cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT) \
3305		? (ext4_fsblk_t)le32_to_cpu(es->name##_hi) << 32 : 0) | \
3306		le32_to_cpu(es->name##_lo))
3307
3308static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
3309{
3310	return ext4_read_incompat_64bit_val(es, s_blocks_count);
3311}
3312
3313static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
3314{
3315	return ext4_read_incompat_64bit_val(es, s_r_blocks_count);
3316}
3317
3318static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
3319{
3320	return ext4_read_incompat_64bit_val(es, s_free_blocks_count);
3321}
3322
3323static inline void ext4_blocks_count_set(struct ext4_super_block *es,
3324					 ext4_fsblk_t blk)
3325{
3326	es->s_blocks_count_lo = cpu_to_le32((u32)blk);
3327	es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
3328}
3329
3330static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
3331					      ext4_fsblk_t blk)
3332{
3333	es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
3334	es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
3335}
3336
3337static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
3338					   ext4_fsblk_t blk)
3339{
3340	es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
3341	es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
3342}
3343
3344static inline loff_t ext4_isize(struct super_block *sb,
3345				struct ext4_inode *raw_inode)
3346{
3347	if (ext4_has_feature_largedir(sb) ||
3348	    S_ISREG(le16_to_cpu(raw_inode->i_mode)))
3349		return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
3350			le32_to_cpu(raw_inode->i_size_lo);
3351
3352	return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
3353}
3354
3355static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
3356{
3357	raw_inode->i_size_lo = cpu_to_le32(i_size);
3358	raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
3359}
3360
3361static inline
3362struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
3363					    ext4_group_t group)
3364{
3365	 struct ext4_group_info **grp_info;
3366	 long indexv, indexh;
3367	 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
3368	 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
3369	 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
3370	 grp_info = sbi_array_rcu_deref(EXT4_SB(sb), s_group_info, indexv);
3371	 return grp_info[indexh];
3372}
3373
3374/*
3375 * Reading s_groups_count requires using smp_rmb() afterwards.  See
3376 * the locking protocol documented in the comments of ext4_group_add()
3377 * in resize.c
3378 */
3379static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
3380{
3381	ext4_group_t	ngroups = EXT4_SB(sb)->s_groups_count;
3382
3383	smp_rmb();
3384	return ngroups;
3385}
3386
3387static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
3388					     ext4_group_t block_group)
3389{
3390	return block_group >> sbi->s_log_groups_per_flex;
3391}
3392
3393static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
3394{
3395	return 1 << sbi->s_log_groups_per_flex;
3396}
3397
3398#define ext4_std_error(sb, errno)				\
3399do {								\
3400	if ((errno))						\
3401		__ext4_std_error((sb), __func__, __LINE__, (errno));	\
3402} while (0)
3403
3404#ifdef CONFIG_SMP
3405/* Each CPU can accumulate percpu_counter_batch clusters in their local
3406 * counters. So we need to make sure we have free clusters more
3407 * than percpu_counter_batch  * nr_cpu_ids. Also add a window of 4 times.
3408 */
3409#define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
3410#else
3411#define EXT4_FREECLUSTERS_WATERMARK 0
3412#endif
3413
3414/* Update i_disksize. Requires i_mutex to avoid races with truncate */
3415static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
3416{
3417	WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
3418		     !inode_is_locked(inode));
3419	down_write(&EXT4_I(inode)->i_data_sem);
3420	if (newsize > EXT4_I(inode)->i_disksize)
3421		WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize);
3422	up_write(&EXT4_I(inode)->i_data_sem);
3423}
3424
3425/* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
3426static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
3427{
3428	int changed = 0;
3429
3430	if (newsize > inode->i_size) {
3431		i_size_write(inode, newsize);
3432		changed = 1;
3433	}
3434	if (newsize > EXT4_I(inode)->i_disksize) {
3435		ext4_update_i_disksize(inode, newsize);
3436		changed |= 2;
3437	}
3438	return changed;
3439}
3440
3441int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
3442				      loff_t len);
3443
3444struct ext4_group_info {
3445	unsigned long   bb_state;
3446#ifdef AGGRESSIVE_CHECK
3447	unsigned long	bb_check_counter;
3448#endif
3449	struct rb_root  bb_free_root;
3450	ext4_grpblk_t	bb_first_free;	/* first free block */
3451	ext4_grpblk_t	bb_free;	/* total free blocks */
3452	ext4_grpblk_t	bb_fragments;	/* nr of freespace fragments */
3453	ext4_grpblk_t	bb_largest_free_order;/* order of largest frag in BG */
3454	ext4_group_t	bb_group;	/* Group number */
3455	struct          list_head bb_prealloc_list;
3456#ifdef DOUBLE_CHECK
3457	void            *bb_bitmap;
3458#endif
3459	struct rw_semaphore alloc_sem;
3460	struct rb_node	bb_avg_fragment_size_rb;
3461	struct list_head bb_largest_free_order_node;
3462	ext4_grpblk_t	bb_counters[];	/* Nr of free power-of-two-block
3463					 * regions, index is order.
3464					 * bb_counters[3] = 5 means
3465					 * 5 free 8-block regions. */
3466};
3467
3468#define EXT4_GROUP_INFO_NEED_INIT_BIT		0
3469#define EXT4_GROUP_INFO_WAS_TRIMMED_BIT		1
3470#define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT	2
3471#define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT	3
3472#define EXT4_GROUP_INFO_BBITMAP_CORRUPT		\
3473	(1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT)
3474#define EXT4_GROUP_INFO_IBITMAP_CORRUPT		\
3475	(1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT)
3476#define EXT4_GROUP_INFO_BBITMAP_READ_BIT	4
3477
3478#define EXT4_MB_GRP_NEED_INIT(grp)	\
3479	(test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
3480#define EXT4_MB_GRP_BBITMAP_CORRUPT(grp)	\
3481	(test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3482#define EXT4_MB_GRP_IBITMAP_CORRUPT(grp)	\
3483	(test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3484
3485#define EXT4_MB_GRP_WAS_TRIMMED(grp)	\
3486	(test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3487#define EXT4_MB_GRP_SET_TRIMMED(grp)	\
3488	(set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3489#define EXT4_MB_GRP_CLEAR_TRIMMED(grp)	\
3490	(clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3491#define EXT4_MB_GRP_TEST_AND_SET_READ(grp)	\
3492	(test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state)))
3493
3494#define EXT4_MAX_CONTENTION		8
3495#define EXT4_CONTENTION_THRESHOLD	2
3496
3497static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
3498					      ext4_group_t group)
3499{
3500	return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
3501}
3502
3503/*
3504 * Returns true if the filesystem is busy enough that attempts to
3505 * access the block group locks has run into contention.
3506 */
3507static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
3508{
3509	return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
3510}
3511
3512static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
3513{
3514	spinlock_t *lock = ext4_group_lock_ptr(sb, group);
3515	if (spin_trylock(lock))
3516		/*
3517		 * We're able to grab the lock right away, so drop the
3518		 * lock contention counter.
3519		 */
3520		atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
3521	else {
3522		/*
3523		 * The lock is busy, so bump the contention counter,
3524		 * and then wait on the spin lock.
3525		 */
3526		atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
3527				  EXT4_MAX_CONTENTION);
3528		spin_lock(lock);
3529	}
3530}
3531
3532static inline void ext4_unlock_group(struct super_block *sb,
3533					ext4_group_t group)
3534{
3535	spin_unlock(ext4_group_lock_ptr(sb, group));
3536}
3537
3538#ifdef CONFIG_QUOTA
3539static inline bool ext4_quota_capable(struct super_block *sb)
3540{
3541	return (test_opt(sb, QUOTA) || ext4_has_feature_quota(sb));
3542}
3543
3544static inline bool ext4_is_quota_journalled(struct super_block *sb)
3545{
3546	struct ext4_sb_info *sbi = EXT4_SB(sb);
3547
3548	return (ext4_has_feature_quota(sb) ||
3549		sbi->s_qf_names[USRQUOTA] || sbi->s_qf_names[GRPQUOTA]);
3550}
3551int ext4_enable_quotas(struct super_block *sb);
3552#endif
3553
3554/*
3555 * Block validity checking
3556 */
3557#define ext4_check_indirect_blockref(inode, bh)				\
3558	ext4_check_blockref(__func__, __LINE__, inode,			\
3559			    (__le32 *)(bh)->b_data,			\
3560			    EXT4_ADDR_PER_BLOCK((inode)->i_sb))
3561
3562#define ext4_ind_check_inode(inode)					\
3563	ext4_check_blockref(__func__, __LINE__, inode,			\
3564			    EXT4_I(inode)->i_data,			\
3565			    EXT4_NDIR_BLOCKS)
3566
3567/*
3568 * Inodes and files operations
3569 */
3570
3571/* dir.c */
3572extern const struct file_operations ext4_dir_operations;
3573
3574/* file.c */
3575extern const struct inode_operations ext4_file_inode_operations;
3576extern const struct file_operations ext4_file_operations;
3577extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
3578
3579/* inline.c */
3580extern int ext4_get_max_inline_size(struct inode *inode);
3581extern int ext4_find_inline_data_nolock(struct inode *inode);
3582extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
3583				 unsigned int len);
3584extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
3585
3586extern int ext4_readpage_inline(struct inode *inode, struct page *page);
3587extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3588					 struct inode *inode,
3589					 loff_t pos, unsigned len,
3590					 unsigned flags,
3591					 struct page **pagep);
3592extern int ext4_write_inline_data_end(struct inode *inode,
3593				      loff_t pos, unsigned len,
3594				      unsigned copied,
3595				      struct page *page);
3596extern struct buffer_head *
3597ext4_journalled_write_inline_data(struct inode *inode,
3598				  unsigned len,
3599				  struct page *page);
3600extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
3601					   struct inode *inode,
3602					   loff_t pos, unsigned len,
3603					   unsigned flags,
3604					   struct page **pagep,
3605					   void **fsdata);
3606extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
3607					 unsigned len, unsigned copied,
3608					 struct page *page);
3609extern int ext4_try_add_inline_entry(handle_t *handle,
3610				     struct ext4_filename *fname,
3611				     struct inode *dir, struct inode *inode);
3612extern int ext4_try_create_inline_dir(handle_t *handle,
3613				      struct inode *parent,
3614				      struct inode *inode);
3615extern int ext4_read_inline_dir(struct file *filp,
3616				struct dir_context *ctx,
3617				int *has_inline_data);
3618extern int ext4_inlinedir_to_tree(struct file *dir_file,
3619				  struct inode *dir, ext4_lblk_t block,
3620				  struct dx_hash_info *hinfo,
3621				  __u32 start_hash, __u32 start_minor_hash,
3622				  int *has_inline_data);
3623extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3624					struct ext4_filename *fname,
3625					struct ext4_dir_entry_2 **res_dir,
3626					int *has_inline_data);
3627extern int ext4_delete_inline_entry(handle_t *handle,
3628				    struct inode *dir,
3629				    struct ext4_dir_entry_2 *de_del,
3630				    struct buffer_head *bh,
3631				    int *has_inline_data);
3632extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3633extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3634					struct ext4_dir_entry_2 **parent_de,
3635					int *retval);
3636extern int ext4_inline_data_fiemap(struct inode *inode,
3637				   struct fiemap_extent_info *fieinfo,
3638				   int *has_inline, __u64 start, __u64 len);
3639
3640struct iomap;
3641extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap);
3642
3643extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3644
3645extern int ext4_convert_inline_data(struct inode *inode);
3646
3647static inline int ext4_has_inline_data(struct inode *inode)
3648{
3649	return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3650	       EXT4_I(inode)->i_inline_off;
3651}
3652
3653/* namei.c */
3654extern const struct inode_operations ext4_dir_inode_operations;
3655extern const struct inode_operations ext4_special_inode_operations;
3656extern struct dentry *ext4_get_parent(struct dentry *child);
3657extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3658				 struct ext4_dir_entry_2 *de,
3659				 int blocksize, int csum_size,
3660				 unsigned int parent_ino, int dotdot_real_len);
3661extern void ext4_initialize_dirent_tail(struct buffer_head *bh,
3662					unsigned int blocksize);
3663extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode,
3664				      struct buffer_head *bh);
3665extern int __ext4_unlink(handle_t *handle, struct inode *dir, const struct qstr *d_name,
3666			 struct inode *inode);
3667extern int __ext4_link(struct inode *dir, struct inode *inode,
3668		       struct dentry *dentry);
3669
3670#define S_SHIFT 12
3671static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = {
3672	[S_IFREG >> S_SHIFT]	= EXT4_FT_REG_FILE,
3673	[S_IFDIR >> S_SHIFT]	= EXT4_FT_DIR,
3674	[S_IFCHR >> S_SHIFT]	= EXT4_FT_CHRDEV,
3675	[S_IFBLK >> S_SHIFT]	= EXT4_FT_BLKDEV,
3676	[S_IFIFO >> S_SHIFT]	= EXT4_FT_FIFO,
3677	[S_IFSOCK >> S_SHIFT]	= EXT4_FT_SOCK,
3678	[S_IFLNK >> S_SHIFT]	= EXT4_FT_SYMLINK,
3679};
3680
3681static inline void ext4_set_de_type(struct super_block *sb,
3682				struct ext4_dir_entry_2 *de,
3683				umode_t mode) {
3684	if (ext4_has_feature_filetype(sb))
3685		de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3686}
3687
3688/* readpages.c */
3689extern int ext4_mpage_readpages(struct inode *inode,
3690		struct readahead_control *rac, struct page *page);
3691extern int __init ext4_init_post_read_processing(void);
3692extern void ext4_exit_post_read_processing(void);
3693
3694/* symlink.c */
3695extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3696extern const struct inode_operations ext4_symlink_inode_operations;
3697extern const struct inode_operations ext4_fast_symlink_inode_operations;
3698
3699/* sysfs.c */
3700extern void ext4_notify_error_sysfs(struct ext4_sb_info *sbi);
3701extern int ext4_register_sysfs(struct super_block *sb);
3702extern void ext4_unregister_sysfs(struct super_block *sb);
3703extern int __init ext4_init_sysfs(void);
3704extern void ext4_exit_sysfs(void);
3705
3706/* block_validity */
3707extern void ext4_release_system_zone(struct super_block *sb);
3708extern int ext4_setup_system_zone(struct super_block *sb);
3709extern int __init ext4_init_system_zone(void);
3710extern void ext4_exit_system_zone(void);
3711extern int ext4_inode_block_valid(struct inode *inode,
3712				  ext4_fsblk_t start_blk,
3713				  unsigned int count);
3714extern int ext4_check_blockref(const char *, unsigned int,
3715			       struct inode *, __le32 *, unsigned int);
3716
3717/* extents.c */
3718struct ext4_ext_path;
3719struct ext4_extent;
3720
3721/*
3722 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3723 * __le32.
3724 */
3725#define EXT_MAX_BLOCKS	0xffffffff
3726
3727extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode);
3728extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3729extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3730			       struct ext4_map_blocks *map, int flags);
3731extern int ext4_ext_truncate(handle_t *, struct inode *);
3732extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3733				 ext4_lblk_t end);
3734extern void ext4_ext_init(struct super_block *);
3735extern void ext4_ext_release(struct super_block *);
3736extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3737			  loff_t len);
3738extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3739					  loff_t offset, ssize_t len);
3740extern int ext4_convert_unwritten_io_end_vec(handle_t *handle,
3741					     ext4_io_end_t *io_end);
3742extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3743			   struct ext4_map_blocks *map, int flags);
3744extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3745						   int num,
3746						   struct ext4_ext_path *path);
3747extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3748				  struct ext4_ext_path **,
3749				  struct ext4_extent *, int);
3750extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3751					      struct ext4_ext_path **,
3752					      int flags);
3753extern void ext4_ext_drop_refs(struct ext4_ext_path *);
3754extern int ext4_ext_check_inode(struct inode *inode);
3755extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3756extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3757			__u64 start, __u64 len);
3758extern int ext4_get_es_cache(struct inode *inode,
3759			     struct fiemap_extent_info *fieinfo,
3760			     __u64 start, __u64 len);
3761extern int ext4_ext_precache(struct inode *inode);
3762extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3763				struct inode *inode2, ext4_lblk_t lblk1,
3764			     ext4_lblk_t lblk2,  ext4_lblk_t count,
3765			     int mark_unwritten,int *err);
3766extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu);
3767extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode,
3768				       int check_cred, int restart_cred,
3769				       int revoke_cred);
3770extern void ext4_ext_replay_shrink_inode(struct inode *inode, ext4_lblk_t end);
3771extern int ext4_ext_replay_set_iblocks(struct inode *inode);
3772extern int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start,
3773		int len, int unwritten, ext4_fsblk_t pblk);
3774extern int ext4_ext_clear_bb(struct inode *inode);
3775
3776
3777/* move_extent.c */
3778extern void ext4_double_down_write_data_sem(struct inode *first,
3779					    struct inode *second);
3780extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3781					  struct inode *donor_inode);
3782extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3783			     __u64 start_orig, __u64 start_donor,
3784			     __u64 len, __u64 *moved_len);
3785
3786/* page-io.c */
3787extern int __init ext4_init_pageio(void);
3788extern void ext4_exit_pageio(void);
3789extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3790extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3791extern int ext4_put_io_end(ext4_io_end_t *io_end);
3792extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3793extern void ext4_io_submit_init(struct ext4_io_submit *io,
3794				struct writeback_control *wbc);
3795extern void ext4_end_io_rsv_work(struct work_struct *work);
3796extern void ext4_io_submit(struct ext4_io_submit *io);
3797extern int ext4_bio_write_page(struct ext4_io_submit *io,
3798			       struct page *page,
3799			       int len,
3800			       bool keep_towrite);
3801extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end);
3802extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end);
3803
3804/* mmp.c */
3805extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3806
3807/* mmp.c */
3808extern void ext4_stop_mmpd(struct ext4_sb_info *sbi);
3809
3810/* verity.c */
3811extern const struct fsverity_operations ext4_verityops;
3812
3813/* orphan.c */
3814extern int ext4_orphan_add(handle_t *, struct inode *);
3815extern int ext4_orphan_del(handle_t *, struct inode *);
3816extern void ext4_orphan_cleanup(struct super_block *sb,
3817				struct ext4_super_block *es);
3818extern void ext4_release_orphan_info(struct super_block *sb);
3819extern int ext4_init_orphan_info(struct super_block *sb);
3820extern int ext4_orphan_file_empty(struct super_block *sb);
3821extern void ext4_orphan_file_block_trigger(
3822				struct jbd2_buffer_trigger_type *triggers,
3823				struct buffer_head *bh,
3824				void *data, size_t size);
3825
3826/*
3827 * Add new method to test whether block and inode bitmaps are properly
3828 * initialized. With uninit_bg reading the block from disk is not enough
3829 * to mark the bitmap uptodate. We need to also zero-out the bitmap
3830 */
3831#define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3832
3833static inline int bitmap_uptodate(struct buffer_head *bh)
3834{
3835	return (buffer_uptodate(bh) &&
3836			test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3837}
3838static inline void set_bitmap_uptodate(struct buffer_head *bh)
3839{
3840	set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3841}
3842
3843#define in_range(b, first, len)	((b) >= (first) && (b) <= (first) + (len) - 1)
3844
3845/* For ioend & aio unwritten conversion wait queues */
3846#define EXT4_WQ_HASH_SZ		37
3847#define ext4_ioend_wq(v)   (&ext4__ioend_wq[((unsigned long)(v)) %\
3848					    EXT4_WQ_HASH_SZ])
3849extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3850
3851extern int ext4_resize_begin(struct super_block *sb);
3852extern void ext4_resize_end(struct super_block *sb);
3853
3854static inline void ext4_set_io_unwritten_flag(struct inode *inode,
3855					      struct ext4_io_end *io_end)
3856{
3857	if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
3858		io_end->flag |= EXT4_IO_END_UNWRITTEN;
3859		atomic_inc(&EXT4_I(inode)->i_unwritten);
3860	}
3861}
3862
3863static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3864{
3865	struct inode *inode = io_end->inode;
3866
3867	if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
3868		io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3869		/* Wake up anyone waiting on unwritten extent conversion */
3870		if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
3871			wake_up_all(ext4_ioend_wq(inode));
3872	}
3873}
3874
3875extern const struct iomap_ops ext4_iomap_ops;
3876extern const struct iomap_ops ext4_iomap_overwrite_ops;
3877extern const struct iomap_ops ext4_iomap_report_ops;
3878
3879static inline int ext4_buffer_uptodate(struct buffer_head *bh)
3880{
3881	/*
3882	 * If the buffer has the write error flag, we have failed
3883	 * to write out data in the block.  In this  case, we don't
3884	 * have to read the block because we may read the old data
3885	 * successfully.
3886	 */
3887	if (buffer_write_io_error(bh))
3888		set_buffer_uptodate(bh);
3889	return buffer_uptodate(bh);
3890}
3891
3892#endif	/* __KERNEL__ */
3893
3894#define EFSBADCRC	EBADMSG		/* Bad CRC detected */
3895#define EFSCORRUPTED	EUCLEAN		/* Filesystem is corrupted */
3896
3897#endif	/* _EXT4_H */
3898