ext4.h revision 02f310fc
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	int 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	spinlock_t of_lock;
1501	int of_blocks;			/* Number of orphan blocks in a file */
1502	__u32 of_csum_seed;		/* Checksum seed for orphan file */
1503	struct ext4_orphan_block *of_binfo;	/* Array with info about orphan
1504						 * file blocks */
1505};
1506
1507/*
1508 * fourth extended-fs super-block data in memory
1509 */
1510struct ext4_sb_info {
1511	unsigned long s_desc_size;	/* Size of a group descriptor in bytes */
1512	unsigned long s_inodes_per_block;/* Number of inodes per block */
1513	unsigned long s_blocks_per_group;/* Number of blocks in a group */
1514	unsigned long s_clusters_per_group; /* Number of clusters in a group */
1515	unsigned long s_inodes_per_group;/* Number of inodes in a group */
1516	unsigned long s_itb_per_group;	/* Number of inode table blocks per group */
1517	unsigned long s_gdb_count;	/* Number of group descriptor blocks */
1518	unsigned long s_desc_per_block;	/* Number of group descriptors per block */
1519	ext4_group_t s_groups_count;	/* Number of groups in the fs */
1520	ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1521	unsigned long s_overhead;  /* # of fs overhead clusters */
1522	unsigned int s_cluster_ratio;	/* Number of blocks per cluster */
1523	unsigned int s_cluster_bits;	/* log2 of s_cluster_ratio */
1524	loff_t s_bitmap_maxbytes;	/* max bytes for bitmap files */
1525	struct buffer_head * s_sbh;	/* Buffer containing the super block */
1526	struct ext4_super_block *s_es;	/* Pointer to the super block in the buffer */
1527	struct buffer_head * __rcu *s_group_desc;
1528	unsigned int s_mount_opt;
1529	unsigned int s_mount_opt2;
1530	unsigned long s_mount_flags;
1531	unsigned int s_def_mount_opt;
1532	ext4_fsblk_t s_sb_block;
1533	atomic64_t s_resv_clusters;
1534	kuid_t s_resuid;
1535	kgid_t s_resgid;
1536	unsigned short s_mount_state;
1537	unsigned short s_pad;
1538	int s_addr_per_block_bits;
1539	int s_desc_per_block_bits;
1540	int s_inode_size;
1541	int s_first_ino;
1542	unsigned int s_inode_readahead_blks;
1543	unsigned int s_inode_goal;
1544	u32 s_hash_seed[4];
1545	int s_def_hash_version;
1546	int s_hash_unsigned;	/* 3 if hash should be unsigned, 0 if not */
1547	struct percpu_counter s_freeclusters_counter;
1548	struct percpu_counter s_freeinodes_counter;
1549	struct percpu_counter s_dirs_counter;
1550	struct percpu_counter s_dirtyclusters_counter;
1551	struct percpu_counter s_sra_exceeded_retry_limit;
1552	struct blockgroup_lock *s_blockgroup_lock;
1553	struct proc_dir_entry *s_proc;
1554	struct kobject s_kobj;
1555	struct completion s_kobj_unregister;
1556	struct super_block *s_sb;
1557	struct buffer_head *s_mmp_bh;
1558
1559	/* Journaling */
1560	struct journal_s *s_journal;
1561	unsigned long s_ext4_flags;		/* Ext4 superblock flags */
1562	struct mutex s_orphan_lock;	/* Protects on disk list changes */
1563	struct list_head s_orphan;	/* List of orphaned inodes in on disk
1564					   list */
1565	struct ext4_orphan_info s_orphan_info;
1566	unsigned long s_commit_interval;
1567	u32 s_max_batch_time;
1568	u32 s_min_batch_time;
1569	struct block_device *s_journal_bdev;
1570#ifdef CONFIG_QUOTA
1571	/* Names of quota files with journalled quota */
1572	char __rcu *s_qf_names[EXT4_MAXQUOTAS];
1573	int s_jquota_fmt;			/* Format of quota to use */
1574#endif
1575	unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1576	struct ext4_system_blocks __rcu *s_system_blks;
1577
1578#ifdef EXTENTS_STATS
1579	/* ext4 extents stats */
1580	unsigned long s_ext_min;
1581	unsigned long s_ext_max;
1582	unsigned long s_depth_max;
1583	spinlock_t s_ext_stats_lock;
1584	unsigned long s_ext_blocks;
1585	unsigned long s_ext_extents;
1586#endif
1587
1588	/* for buddy allocator */
1589	struct ext4_group_info ** __rcu *s_group_info;
1590	struct inode *s_buddy_cache;
1591	spinlock_t s_md_lock;
1592	unsigned short *s_mb_offsets;
1593	unsigned int *s_mb_maxs;
1594	unsigned int s_group_info_size;
1595	unsigned int s_mb_free_pending;
1596	struct list_head s_freed_data_list;	/* List of blocks to be freed
1597						   after commit completed */
1598	struct list_head s_discard_list;
1599	struct work_struct s_discard_work;
1600	atomic_t s_retry_alloc_pending;
1601	struct rb_root s_mb_avg_fragment_size_root;
1602	rwlock_t s_mb_rb_lock;
1603	struct list_head *s_mb_largest_free_orders;
1604	rwlock_t *s_mb_largest_free_orders_locks;
1605
1606	/* tunables */
1607	unsigned long s_stripe;
1608	unsigned int s_mb_max_linear_groups;
1609	unsigned int s_mb_stream_request;
1610	unsigned int s_mb_max_to_scan;
1611	unsigned int s_mb_min_to_scan;
1612	unsigned int s_mb_stats;
1613	unsigned int s_mb_order2_reqs;
1614	unsigned int s_mb_group_prealloc;
1615	unsigned int s_mb_max_inode_prealloc;
1616	unsigned int s_max_dir_size_kb;
1617	/* where last allocation was done - for stream allocation */
1618	unsigned long s_mb_last_group;
1619	unsigned long s_mb_last_start;
1620	unsigned int s_mb_prefetch;
1621	unsigned int s_mb_prefetch_limit;
1622
1623	/* stats for buddy allocator */
1624	atomic_t s_bal_reqs;	/* number of reqs with len > 1 */
1625	atomic_t s_bal_success;	/* we found long enough chunks */
1626	atomic_t s_bal_allocated;	/* in blocks */
1627	atomic_t s_bal_ex_scanned;	/* total extents scanned */
1628	atomic_t s_bal_groups_scanned;	/* number of groups scanned */
1629	atomic_t s_bal_goals;	/* goal hits */
1630	atomic_t s_bal_breaks;	/* too long searches */
1631	atomic_t s_bal_2orders;	/* 2^order hits */
1632	atomic_t s_bal_cr0_bad_suggestions;
1633	atomic_t s_bal_cr1_bad_suggestions;
1634	atomic64_t s_bal_cX_groups_considered[4];
1635	atomic64_t s_bal_cX_hits[4];
1636	atomic64_t s_bal_cX_failed[4];		/* cX loop didn't find blocks */
1637	atomic_t s_mb_buddies_generated;	/* number of buddies generated */
1638	atomic64_t s_mb_generation_time;
1639	atomic_t s_mb_lost_chunks;
1640	atomic_t s_mb_preallocated;
1641	atomic_t s_mb_discarded;
1642	atomic_t s_lock_busy;
1643
1644	/* locality groups */
1645	struct ext4_locality_group __percpu *s_locality_groups;
1646
1647	/* for write statistics */
1648	unsigned long s_sectors_written_start;
1649	u64 s_kbytes_written;
1650
1651	/* the size of zero-out chunk */
1652	unsigned int s_extent_max_zeroout_kb;
1653
1654	unsigned int s_log_groups_per_flex;
1655	struct flex_groups * __rcu *s_flex_groups;
1656	ext4_group_t s_flex_groups_allocated;
1657
1658	/* workqueue for reserved extent conversions (buffered io) */
1659	struct workqueue_struct *rsv_conversion_wq;
1660
1661	/* timer for periodic error stats printing */
1662	struct timer_list s_err_report;
1663
1664	/* Lazy inode table initialization info */
1665	struct ext4_li_request *s_li_request;
1666	/* Wait multiplier for lazy initialization thread */
1667	unsigned int s_li_wait_mult;
1668
1669	/* Kernel thread for multiple mount protection */
1670	struct task_struct *s_mmp_tsk;
1671
1672	/* record the last minlen when FITRIM is called. */
1673	atomic_t s_last_trim_minblks;
1674
1675	/* Reference to checksum algorithm driver via cryptoapi */
1676	struct crypto_shash *s_chksum_driver;
1677
1678	/* Precomputed FS UUID checksum for seeding other checksums */
1679	__u32 s_csum_seed;
1680
1681	/* Reclaim extents from extent status tree */
1682	struct shrinker s_es_shrinker;
1683	struct list_head s_es_list;	/* List of inodes with reclaimable extents */
1684	long s_es_nr_inode;
1685	struct ext4_es_stats s_es_stats;
1686	struct mb_cache *s_ea_block_cache;
1687	struct mb_cache *s_ea_inode_cache;
1688	spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1689
1690	/* Journal triggers for checksum computation */
1691	struct ext4_journal_trigger s_journal_triggers[EXT4_JOURNAL_TRIGGER_COUNT];
1692
1693	/* Ratelimit ext4 messages. */
1694	struct ratelimit_state s_err_ratelimit_state;
1695	struct ratelimit_state s_warning_ratelimit_state;
1696	struct ratelimit_state s_msg_ratelimit_state;
1697	atomic_t s_warning_count;
1698	atomic_t s_msg_count;
1699
1700	/* Encryption policy for '-o test_dummy_encryption' */
1701	struct fscrypt_dummy_policy s_dummy_enc_policy;
1702
1703	/*
1704	 * Barrier between writepages ops and changing any inode's JOURNAL_DATA
1705	 * or EXTENTS flag.
1706	 */
1707	struct percpu_rw_semaphore s_writepages_rwsem;
1708	struct dax_device *s_daxdev;
1709#ifdef CONFIG_EXT4_DEBUG
1710	unsigned long s_simulate_fail;
1711#endif
1712	/* Record the errseq of the backing block device */
1713	errseq_t s_bdev_wb_err;
1714	spinlock_t s_bdev_wb_lock;
1715
1716	/* Information about errors that happened during this mount */
1717	spinlock_t s_error_lock;
1718	int s_add_error_count;
1719	int s_first_error_code;
1720	__u32 s_first_error_line;
1721	__u32 s_first_error_ino;
1722	__u64 s_first_error_block;
1723	const char *s_first_error_func;
1724	time64_t s_first_error_time;
1725	int s_last_error_code;
1726	__u32 s_last_error_line;
1727	__u32 s_last_error_ino;
1728	__u64 s_last_error_block;
1729	const char *s_last_error_func;
1730	time64_t s_last_error_time;
1731	/*
1732	 * If we are in a context where we cannot update error information in
1733	 * the on-disk superblock, we queue this work to do it.
1734	 */
1735	struct work_struct s_error_work;
1736
1737	/* Ext4 fast commit stuff */
1738	atomic_t s_fc_subtid;
1739	atomic_t s_fc_ineligible_updates;
1740	/*
1741	 * After commit starts, the main queue gets locked, and the further
1742	 * updates get added in the staging queue.
1743	 */
1744#define FC_Q_MAIN	0
1745#define FC_Q_STAGING	1
1746	struct list_head s_fc_q[2];	/* Inodes staged for fast commit
1747					 * that have data changes in them.
1748					 */
1749	struct list_head s_fc_dentry_q[2];	/* directory entry updates */
1750	unsigned int s_fc_bytes;
1751	/*
1752	 * Main fast commit lock. This lock protects accesses to the
1753	 * following fields:
1754	 * ei->i_fc_list, s_fc_dentry_q, s_fc_q, s_fc_bytes, s_fc_bh.
1755	 */
1756	spinlock_t s_fc_lock;
1757	struct buffer_head *s_fc_bh;
1758	struct ext4_fc_stats s_fc_stats;
1759	u64 s_fc_avg_commit_time;
1760#ifdef CONFIG_EXT4_DEBUG
1761	int s_fc_debug_max_replay;
1762#endif
1763	struct ext4_fc_replay_state s_fc_replay_state;
1764};
1765
1766static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1767{
1768	return sb->s_fs_info;
1769}
1770static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1771{
1772	return container_of(inode, struct ext4_inode_info, vfs_inode);
1773}
1774
1775static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1776{
1777	return ino == EXT4_ROOT_INO ||
1778		(ino >= EXT4_FIRST_INO(sb) &&
1779		 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1780}
1781
1782/*
1783 * Returns: sbi->field[index]
1784 * Used to access an array element from the following sbi fields which require
1785 * rcu protection to avoid dereferencing an invalid pointer due to reassignment
1786 * - s_group_desc
1787 * - s_group_info
1788 * - s_flex_group
1789 */
1790#define sbi_array_rcu_deref(sbi, field, index)				   \
1791({									   \
1792	typeof(*((sbi)->field)) _v;					   \
1793	rcu_read_lock();						   \
1794	_v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index];	   \
1795	rcu_read_unlock();						   \
1796	_v;								   \
1797})
1798
1799/*
1800 * run-time mount flags
1801 */
1802enum {
1803	EXT4_MF_MNTDIR_SAMPLED,
1804	EXT4_MF_FS_ABORTED,	/* Fatal error detected */
1805	EXT4_MF_FC_INELIGIBLE,	/* Fast commit ineligible */
1806	EXT4_MF_FC_COMMITTING	/* File system underoing a fast
1807				 * commit.
1808				 */
1809};
1810
1811static inline void ext4_set_mount_flag(struct super_block *sb, int bit)
1812{
1813	set_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1814}
1815
1816static inline void ext4_clear_mount_flag(struct super_block *sb, int bit)
1817{
1818	clear_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1819}
1820
1821static inline int ext4_test_mount_flag(struct super_block *sb, int bit)
1822{
1823	return test_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1824}
1825
1826
1827/*
1828 * Simulate_fail codes
1829 */
1830#define EXT4_SIM_BBITMAP_EIO	1
1831#define EXT4_SIM_BBITMAP_CRC	2
1832#define EXT4_SIM_IBITMAP_EIO	3
1833#define EXT4_SIM_IBITMAP_CRC	4
1834#define EXT4_SIM_INODE_EIO	5
1835#define EXT4_SIM_INODE_CRC	6
1836#define EXT4_SIM_DIRBLOCK_EIO	7
1837#define EXT4_SIM_DIRBLOCK_CRC	8
1838
1839static inline bool ext4_simulate_fail(struct super_block *sb,
1840				     unsigned long code)
1841{
1842#ifdef CONFIG_EXT4_DEBUG
1843	struct ext4_sb_info *sbi = EXT4_SB(sb);
1844
1845	if (unlikely(sbi->s_simulate_fail == code)) {
1846		sbi->s_simulate_fail = 0;
1847		return true;
1848	}
1849#endif
1850	return false;
1851}
1852
1853static inline void ext4_simulate_fail_bh(struct super_block *sb,
1854					 struct buffer_head *bh,
1855					 unsigned long code)
1856{
1857	if (!IS_ERR(bh) && ext4_simulate_fail(sb, code))
1858		clear_buffer_uptodate(bh);
1859}
1860
1861/*
1862 * Error number codes for s_{first,last}_error_errno
1863 *
1864 * Linux errno numbers are architecture specific, so we need to translate
1865 * them into something which is architecture independent.   We don't define
1866 * codes for all errno's; just the ones which are most likely to be the cause
1867 * of an ext4_error() call.
1868 */
1869#define EXT4_ERR_UNKNOWN	 1
1870#define EXT4_ERR_EIO		 2
1871#define EXT4_ERR_ENOMEM		 3
1872#define EXT4_ERR_EFSBADCRC	 4
1873#define EXT4_ERR_EFSCORRUPTED	 5
1874#define EXT4_ERR_ENOSPC		 6
1875#define EXT4_ERR_ENOKEY		 7
1876#define EXT4_ERR_EROFS		 8
1877#define EXT4_ERR_EFBIG		 9
1878#define EXT4_ERR_EEXIST		10
1879#define EXT4_ERR_ERANGE		11
1880#define EXT4_ERR_EOVERFLOW	12
1881#define EXT4_ERR_EBUSY		13
1882#define EXT4_ERR_ENOTDIR	14
1883#define EXT4_ERR_ENOTEMPTY	15
1884#define EXT4_ERR_ESHUTDOWN	16
1885#define EXT4_ERR_EFAULT		17
1886
1887/*
1888 * Inode dynamic state flags
1889 */
1890enum {
1891	EXT4_STATE_JDATA,		/* journaled data exists */
1892	EXT4_STATE_NEW,			/* inode is newly created */
1893	EXT4_STATE_XATTR,		/* has in-inode xattrs */
1894	EXT4_STATE_NO_EXPAND,		/* No space for expansion */
1895	EXT4_STATE_DA_ALLOC_CLOSE,	/* Alloc DA blks on close */
1896	EXT4_STATE_EXT_MIGRATE,		/* Inode is migrating */
1897	EXT4_STATE_NEWENTRY,		/* File just added to dir */
1898	EXT4_STATE_MAY_INLINE_DATA,	/* may have in-inode data */
1899	EXT4_STATE_EXT_PRECACHED,	/* extents have been precached */
1900	EXT4_STATE_LUSTRE_EA_INODE,	/* Lustre-style ea_inode */
1901	EXT4_STATE_VERITY_IN_PROGRESS,	/* building fs-verity Merkle tree */
1902	EXT4_STATE_FC_COMMITTING,	/* Fast commit ongoing */
1903	EXT4_STATE_ORPHAN_FILE,		/* Inode orphaned in orphan file */
1904};
1905
1906#define EXT4_INODE_BIT_FNS(name, field, offset)				\
1907static inline int ext4_test_inode_##name(struct inode *inode, int bit)	\
1908{									\
1909	return test_bit(bit + (offset), &EXT4_I(inode)->i_##field);	\
1910}									\
1911static inline void ext4_set_inode_##name(struct inode *inode, int bit)	\
1912{									\
1913	set_bit(bit + (offset), &EXT4_I(inode)->i_##field);		\
1914}									\
1915static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1916{									\
1917	clear_bit(bit + (offset), &EXT4_I(inode)->i_##field);		\
1918}
1919
1920/* Add these declarations here only so that these functions can be
1921 * found by name.  Otherwise, they are very hard to locate. */
1922static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1923static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1924static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1925EXT4_INODE_BIT_FNS(flag, flags, 0)
1926
1927/* Add these declarations here only so that these functions can be
1928 * found by name.  Otherwise, they are very hard to locate. */
1929static inline int ext4_test_inode_state(struct inode *inode, int bit);
1930static inline void ext4_set_inode_state(struct inode *inode, int bit);
1931static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1932#if (BITS_PER_LONG < 64)
1933EXT4_INODE_BIT_FNS(state, state_flags, 0)
1934
1935static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1936{
1937	(ei)->i_state_flags = 0;
1938}
1939#else
1940EXT4_INODE_BIT_FNS(state, flags, 32)
1941
1942static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1943{
1944	/* We depend on the fact that callers will set i_flags */
1945}
1946#endif
1947#else
1948/* Assume that user mode programs are passing in an ext4fs superblock, not
1949 * a kernel struct super_block.  This will allow us to call the feature-test
1950 * macros from user land. */
1951#define EXT4_SB(sb)	(sb)
1952#endif
1953
1954static inline bool ext4_verity_in_progress(struct inode *inode)
1955{
1956	return IS_ENABLED(CONFIG_FS_VERITY) &&
1957	       ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS);
1958}
1959
1960#define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1961
1962/*
1963 * Codes for operating systems
1964 */
1965#define EXT4_OS_LINUX		0
1966#define EXT4_OS_HURD		1
1967#define EXT4_OS_MASIX		2
1968#define EXT4_OS_FREEBSD		3
1969#define EXT4_OS_LITES		4
1970
1971/*
1972 * Revision levels
1973 */
1974#define EXT4_GOOD_OLD_REV	0	/* The good old (original) format */
1975#define EXT4_DYNAMIC_REV	1	/* V2 format w/ dynamic inode sizes */
1976
1977#define EXT4_MAX_SUPP_REV	EXT4_DYNAMIC_REV
1978
1979#define EXT4_GOOD_OLD_INODE_SIZE 128
1980
1981#define EXT4_EXTRA_TIMESTAMP_MAX	(((s64)1 << 34) - 1  + S32_MIN)
1982#define EXT4_NON_EXTRA_TIMESTAMP_MAX	S32_MAX
1983#define EXT4_TIMESTAMP_MIN		S32_MIN
1984
1985/*
1986 * Feature set definitions
1987 */
1988
1989#define EXT4_FEATURE_COMPAT_DIR_PREALLOC	0x0001
1990#define EXT4_FEATURE_COMPAT_IMAGIC_INODES	0x0002
1991#define EXT4_FEATURE_COMPAT_HAS_JOURNAL		0x0004
1992#define EXT4_FEATURE_COMPAT_EXT_ATTR		0x0008
1993#define EXT4_FEATURE_COMPAT_RESIZE_INODE	0x0010
1994#define EXT4_FEATURE_COMPAT_DIR_INDEX		0x0020
1995#define EXT4_FEATURE_COMPAT_SPARSE_SUPER2	0x0200
1996/*
1997 * The reason why "FAST_COMMIT" is a compat feature is that, FS becomes
1998 * incompatible only if fast commit blocks are present in the FS. Since we
1999 * clear the journal (and thus the fast commit blocks), we don't mark FS as
2000 * incompatible. We also have a JBD2 incompat feature, which gets set when
2001 * there are fast commit blocks present in the journal.
2002 */
2003#define EXT4_FEATURE_COMPAT_FAST_COMMIT		0x0400
2004#define EXT4_FEATURE_COMPAT_STABLE_INODES	0x0800
2005#define EXT4_FEATURE_COMPAT_ORPHAN_FILE		0x1000	/* Orphan file exists */
2006
2007#define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER	0x0001
2008#define EXT4_FEATURE_RO_COMPAT_LARGE_FILE	0x0002
2009#define EXT4_FEATURE_RO_COMPAT_BTREE_DIR	0x0004
2010#define EXT4_FEATURE_RO_COMPAT_HUGE_FILE        0x0008
2011#define EXT4_FEATURE_RO_COMPAT_GDT_CSUM		0x0010
2012#define EXT4_FEATURE_RO_COMPAT_DIR_NLINK	0x0020
2013#define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE	0x0040
2014#define EXT4_FEATURE_RO_COMPAT_QUOTA		0x0100
2015#define EXT4_FEATURE_RO_COMPAT_BIGALLOC		0x0200
2016/*
2017 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM).  When
2018 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
2019 * all other data structures' checksums.  However, the METADATA_CSUM and
2020 * GDT_CSUM bits are mutually exclusive.
2021 */
2022#define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM	0x0400
2023#define EXT4_FEATURE_RO_COMPAT_READONLY		0x1000
2024#define EXT4_FEATURE_RO_COMPAT_PROJECT		0x2000
2025#define EXT4_FEATURE_RO_COMPAT_VERITY		0x8000
2026#define EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT	0x10000 /* Orphan file may be
2027							   non-empty */
2028
2029#define EXT4_FEATURE_INCOMPAT_COMPRESSION	0x0001
2030#define EXT4_FEATURE_INCOMPAT_FILETYPE		0x0002
2031#define EXT4_FEATURE_INCOMPAT_RECOVER		0x0004 /* Needs recovery */
2032#define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV	0x0008 /* Journal device */
2033#define EXT4_FEATURE_INCOMPAT_META_BG		0x0010
2034#define EXT4_FEATURE_INCOMPAT_EXTENTS		0x0040 /* extents support */
2035#define EXT4_FEATURE_INCOMPAT_64BIT		0x0080
2036#define EXT4_FEATURE_INCOMPAT_MMP               0x0100
2037#define EXT4_FEATURE_INCOMPAT_FLEX_BG		0x0200
2038#define EXT4_FEATURE_INCOMPAT_EA_INODE		0x0400 /* EA in inode */
2039#define EXT4_FEATURE_INCOMPAT_DIRDATA		0x1000 /* data in dirent */
2040#define EXT4_FEATURE_INCOMPAT_CSUM_SEED		0x2000
2041#define EXT4_FEATURE_INCOMPAT_LARGEDIR		0x4000 /* >2GB or 3-lvl htree */
2042#define EXT4_FEATURE_INCOMPAT_INLINE_DATA	0x8000 /* data in inode */
2043#define EXT4_FEATURE_INCOMPAT_ENCRYPT		0x10000
2044#define EXT4_FEATURE_INCOMPAT_CASEFOLD		0x20000
2045
2046extern void ext4_update_dynamic_rev(struct super_block *sb);
2047
2048#define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
2049static inline bool ext4_has_feature_##name(struct super_block *sb) \
2050{ \
2051	return ((EXT4_SB(sb)->s_es->s_feature_compat & \
2052		cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
2053} \
2054static inline void ext4_set_feature_##name(struct super_block *sb) \
2055{ \
2056	ext4_update_dynamic_rev(sb); \
2057	EXT4_SB(sb)->s_es->s_feature_compat |= \
2058		cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
2059} \
2060static inline void ext4_clear_feature_##name(struct super_block *sb) \
2061{ \
2062	EXT4_SB(sb)->s_es->s_feature_compat &= \
2063		~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
2064}
2065
2066#define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
2067static inline bool ext4_has_feature_##name(struct super_block *sb) \
2068{ \
2069	return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
2070		cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
2071} \
2072static inline void ext4_set_feature_##name(struct super_block *sb) \
2073{ \
2074	ext4_update_dynamic_rev(sb); \
2075	EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
2076		cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
2077} \
2078static inline void ext4_clear_feature_##name(struct super_block *sb) \
2079{ \
2080	EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
2081		~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
2082}
2083
2084#define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
2085static inline bool ext4_has_feature_##name(struct super_block *sb) \
2086{ \
2087	return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
2088		cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
2089} \
2090static inline void ext4_set_feature_##name(struct super_block *sb) \
2091{ \
2092	ext4_update_dynamic_rev(sb); \
2093	EXT4_SB(sb)->s_es->s_feature_incompat |= \
2094		cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
2095} \
2096static inline void ext4_clear_feature_##name(struct super_block *sb) \
2097{ \
2098	EXT4_SB(sb)->s_es->s_feature_incompat &= \
2099		~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
2100}
2101
2102EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc,		DIR_PREALLOC)
2103EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes,	IMAGIC_INODES)
2104EXT4_FEATURE_COMPAT_FUNCS(journal,		HAS_JOURNAL)
2105EXT4_FEATURE_COMPAT_FUNCS(xattr,		EXT_ATTR)
2106EXT4_FEATURE_COMPAT_FUNCS(resize_inode,		RESIZE_INODE)
2107EXT4_FEATURE_COMPAT_FUNCS(dir_index,		DIR_INDEX)
2108EXT4_FEATURE_COMPAT_FUNCS(sparse_super2,	SPARSE_SUPER2)
2109EXT4_FEATURE_COMPAT_FUNCS(fast_commit,		FAST_COMMIT)
2110EXT4_FEATURE_COMPAT_FUNCS(stable_inodes,	STABLE_INODES)
2111EXT4_FEATURE_COMPAT_FUNCS(orphan_file,		ORPHAN_FILE)
2112
2113EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super,	SPARSE_SUPER)
2114EXT4_FEATURE_RO_COMPAT_FUNCS(large_file,	LARGE_FILE)
2115EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir,		BTREE_DIR)
2116EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file,		HUGE_FILE)
2117EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum,		GDT_CSUM)
2118EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink,		DIR_NLINK)
2119EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize,	EXTRA_ISIZE)
2120EXT4_FEATURE_RO_COMPAT_FUNCS(quota,		QUOTA)
2121EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc,		BIGALLOC)
2122EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum,	METADATA_CSUM)
2123EXT4_FEATURE_RO_COMPAT_FUNCS(readonly,		READONLY)
2124EXT4_FEATURE_RO_COMPAT_FUNCS(project,		PROJECT)
2125EXT4_FEATURE_RO_COMPAT_FUNCS(verity,		VERITY)
2126EXT4_FEATURE_RO_COMPAT_FUNCS(orphan_present,	ORPHAN_PRESENT)
2127
2128EXT4_FEATURE_INCOMPAT_FUNCS(compression,	COMPRESSION)
2129EXT4_FEATURE_INCOMPAT_FUNCS(filetype,		FILETYPE)
2130EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery,	RECOVER)
2131EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev,	JOURNAL_DEV)
2132EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg,		META_BG)
2133EXT4_FEATURE_INCOMPAT_FUNCS(extents,		EXTENTS)
2134EXT4_FEATURE_INCOMPAT_FUNCS(64bit,		64BIT)
2135EXT4_FEATURE_INCOMPAT_FUNCS(mmp,		MMP)
2136EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg,		FLEX_BG)
2137EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode,		EA_INODE)
2138EXT4_FEATURE_INCOMPAT_FUNCS(dirdata,		DIRDATA)
2139EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed,		CSUM_SEED)
2140EXT4_FEATURE_INCOMPAT_FUNCS(largedir,		LARGEDIR)
2141EXT4_FEATURE_INCOMPAT_FUNCS(inline_data,	INLINE_DATA)
2142EXT4_FEATURE_INCOMPAT_FUNCS(encrypt,		ENCRYPT)
2143EXT4_FEATURE_INCOMPAT_FUNCS(casefold,		CASEFOLD)
2144
2145#define EXT2_FEATURE_COMPAT_SUPP	EXT4_FEATURE_COMPAT_EXT_ATTR
2146#define EXT2_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
2147					 EXT4_FEATURE_INCOMPAT_META_BG)
2148#define EXT2_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2149					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2150					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
2151
2152#define EXT3_FEATURE_COMPAT_SUPP	EXT4_FEATURE_COMPAT_EXT_ATTR
2153#define EXT3_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
2154					 EXT4_FEATURE_INCOMPAT_RECOVER| \
2155					 EXT4_FEATURE_INCOMPAT_META_BG)
2156#define EXT3_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2157					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2158					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
2159
2160#define EXT4_FEATURE_COMPAT_SUPP	(EXT4_FEATURE_COMPAT_EXT_ATTR| \
2161					 EXT4_FEATURE_COMPAT_ORPHAN_FILE)
2162#define EXT4_FEATURE_INCOMPAT_SUPP	(EXT4_FEATURE_INCOMPAT_FILETYPE| \
2163					 EXT4_FEATURE_INCOMPAT_RECOVER| \
2164					 EXT4_FEATURE_INCOMPAT_META_BG| \
2165					 EXT4_FEATURE_INCOMPAT_EXTENTS| \
2166					 EXT4_FEATURE_INCOMPAT_64BIT| \
2167					 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
2168					 EXT4_FEATURE_INCOMPAT_EA_INODE| \
2169					 EXT4_FEATURE_INCOMPAT_MMP | \
2170					 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
2171					 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
2172					 EXT4_FEATURE_INCOMPAT_CASEFOLD | \
2173					 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \
2174					 EXT4_FEATURE_INCOMPAT_LARGEDIR)
2175#define EXT4_FEATURE_RO_COMPAT_SUPP	(EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2176					 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2177					 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
2178					 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
2179					 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
2180					 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
2181					 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
2182					 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
2183					 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
2184					 EXT4_FEATURE_RO_COMPAT_QUOTA |\
2185					 EXT4_FEATURE_RO_COMPAT_PROJECT |\
2186					 EXT4_FEATURE_RO_COMPAT_VERITY |\
2187					 EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT)
2188
2189#define EXTN_FEATURE_FUNCS(ver) \
2190static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
2191{ \
2192	return ((EXT4_SB(sb)->s_es->s_feature_compat & \
2193		cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
2194} \
2195static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
2196{ \
2197	return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
2198		cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
2199} \
2200static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
2201{ \
2202	return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
2203		cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
2204}
2205
2206EXTN_FEATURE_FUNCS(2)
2207EXTN_FEATURE_FUNCS(3)
2208EXTN_FEATURE_FUNCS(4)
2209
2210static inline bool ext4_has_compat_features(struct super_block *sb)
2211{
2212	return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
2213}
2214static inline bool ext4_has_ro_compat_features(struct super_block *sb)
2215{
2216	return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
2217}
2218static inline bool ext4_has_incompat_features(struct super_block *sb)
2219{
2220	return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
2221}
2222
2223extern int ext4_feature_set_ok(struct super_block *sb, int readonly);
2224
2225/*
2226 * Superblock flags
2227 */
2228#define EXT4_FLAGS_RESIZING	0
2229#define EXT4_FLAGS_SHUTDOWN	1
2230#define EXT4_FLAGS_BDEV_IS_DAX	2
2231
2232static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi)
2233{
2234	return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags);
2235}
2236
2237/*
2238 * Default values for user and/or group using reserved blocks
2239 */
2240#define	EXT4_DEF_RESUID		0
2241#define	EXT4_DEF_RESGID		0
2242
2243/*
2244 * Default project ID
2245 */
2246#define	EXT4_DEF_PROJID		0
2247
2248#define EXT4_DEF_INODE_READAHEAD_BLKS	32
2249
2250/*
2251 * Default mount options
2252 */
2253#define EXT4_DEFM_DEBUG		0x0001
2254#define EXT4_DEFM_BSDGROUPS	0x0002
2255#define EXT4_DEFM_XATTR_USER	0x0004
2256#define EXT4_DEFM_ACL		0x0008
2257#define EXT4_DEFM_UID16		0x0010
2258#define EXT4_DEFM_JMODE		0x0060
2259#define EXT4_DEFM_JMODE_DATA	0x0020
2260#define EXT4_DEFM_JMODE_ORDERED	0x0040
2261#define EXT4_DEFM_JMODE_WBACK	0x0060
2262#define EXT4_DEFM_NOBARRIER	0x0100
2263#define EXT4_DEFM_BLOCK_VALIDITY 0x0200
2264#define EXT4_DEFM_DISCARD	0x0400
2265#define EXT4_DEFM_NODELALLOC	0x0800
2266
2267/*
2268 * Default journal batch times
2269 */
2270#define EXT4_DEF_MIN_BATCH_TIME	0
2271#define EXT4_DEF_MAX_BATCH_TIME	15000 /* 15ms */
2272
2273/*
2274 * Minimum number of groups in a flexgroup before we separate out
2275 * directories into the first block group of a flexgroup
2276 */
2277#define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME	4
2278
2279/*
2280 * Structure of a directory entry
2281 */
2282#define EXT4_NAME_LEN 255
2283
2284struct ext4_dir_entry {
2285	__le32	inode;			/* Inode number */
2286	__le16	rec_len;		/* Directory entry length */
2287	__le16	name_len;		/* Name length */
2288	char	name[EXT4_NAME_LEN];	/* File name */
2289};
2290
2291
2292/*
2293 * Encrypted Casefolded entries require saving the hash on disk. This structure
2294 * followed ext4_dir_entry_2's name[name_len] at the next 4 byte aligned
2295 * boundary.
2296 */
2297struct ext4_dir_entry_hash {
2298	__le32 hash;
2299	__le32 minor_hash;
2300};
2301
2302/*
2303 * The new version of the directory entry.  Since EXT4 structures are
2304 * stored in intel byte order, and the name_len field could never be
2305 * bigger than 255 chars, it's safe to reclaim the extra byte for the
2306 * file_type field.
2307 */
2308struct ext4_dir_entry_2 {
2309	__le32	inode;			/* Inode number */
2310	__le16	rec_len;		/* Directory entry length */
2311	__u8	name_len;		/* Name length */
2312	__u8	file_type;		/* See file type macros EXT4_FT_* below */
2313	char	name[EXT4_NAME_LEN];	/* File name */
2314};
2315
2316/*
2317 * Access the hashes at the end of ext4_dir_entry_2
2318 */
2319#define EXT4_DIRENT_HASHES(entry) \
2320	((struct ext4_dir_entry_hash *) \
2321		(((void *)(entry)) + \
2322		((8 + (entry)->name_len + EXT4_DIR_ROUND) & ~EXT4_DIR_ROUND)))
2323#define EXT4_DIRENT_HASH(entry) le32_to_cpu(EXT4_DIRENT_HASHES(de)->hash)
2324#define EXT4_DIRENT_MINOR_HASH(entry) \
2325		le32_to_cpu(EXT4_DIRENT_HASHES(de)->minor_hash)
2326
2327static inline bool ext4_hash_in_dirent(const struct inode *inode)
2328{
2329	return IS_CASEFOLDED(inode) && IS_ENCRYPTED(inode);
2330}
2331
2332/*
2333 * This is a bogus directory entry at the end of each leaf block that
2334 * records checksums.
2335 */
2336struct ext4_dir_entry_tail {
2337	__le32	det_reserved_zero1;	/* Pretend to be unused */
2338	__le16	det_rec_len;		/* 12 */
2339	__u8	det_reserved_zero2;	/* Zero name length */
2340	__u8	det_reserved_ft;	/* 0xDE, fake file type */
2341	__le32	det_checksum;		/* crc32c(uuid+inum+dirblock) */
2342};
2343
2344#define EXT4_DIRENT_TAIL(block, blocksize) \
2345	((struct ext4_dir_entry_tail *)(((void *)(block)) + \
2346					((blocksize) - \
2347					 sizeof(struct ext4_dir_entry_tail))))
2348
2349/*
2350 * Ext4 directory file types.  Only the low 3 bits are used.  The
2351 * other bits are reserved for now.
2352 */
2353#define EXT4_FT_UNKNOWN		0
2354#define EXT4_FT_REG_FILE	1
2355#define EXT4_FT_DIR		2
2356#define EXT4_FT_CHRDEV		3
2357#define EXT4_FT_BLKDEV		4
2358#define EXT4_FT_FIFO		5
2359#define EXT4_FT_SOCK		6
2360#define EXT4_FT_SYMLINK		7
2361
2362#define EXT4_FT_MAX		8
2363
2364#define EXT4_FT_DIR_CSUM	0xDE
2365
2366/*
2367 * EXT4_DIR_PAD defines the directory entries boundaries
2368 *
2369 * NOTE: It must be a multiple of 4
2370 */
2371#define EXT4_DIR_PAD			4
2372#define EXT4_DIR_ROUND			(EXT4_DIR_PAD - 1)
2373#define EXT4_MAX_REC_LEN		((1<<16)-1)
2374
2375/*
2376 * The rec_len is dependent on the type of directory. Directories that are
2377 * casefolded and encrypted need to store the hash as well, so we add room for
2378 * ext4_extended_dir_entry_2. For all entries related to '.' or '..' you should
2379 * pass NULL for dir, as those entries do not use the extra fields.
2380 */
2381static inline unsigned int ext4_dir_rec_len(__u8 name_len,
2382						const struct inode *dir)
2383{
2384	int rec_len = (name_len + 8 + EXT4_DIR_ROUND);
2385
2386	if (dir && ext4_hash_in_dirent(dir))
2387		rec_len += sizeof(struct ext4_dir_entry_hash);
2388	return (rec_len & ~EXT4_DIR_ROUND);
2389}
2390
2391/*
2392 * If we ever get support for fs block sizes > page_size, we'll need
2393 * to remove the #if statements in the next two functions...
2394 */
2395static inline unsigned int
2396ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
2397{
2398	unsigned len = le16_to_cpu(dlen);
2399
2400#if (PAGE_SIZE >= 65536)
2401	if (len == EXT4_MAX_REC_LEN || len == 0)
2402		return blocksize;
2403	return (len & 65532) | ((len & 3) << 16);
2404#else
2405	return len;
2406#endif
2407}
2408
2409static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
2410{
2411	if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
2412		BUG();
2413#if (PAGE_SIZE >= 65536)
2414	if (len < 65536)
2415		return cpu_to_le16(len);
2416	if (len == blocksize) {
2417		if (blocksize == 65536)
2418			return cpu_to_le16(EXT4_MAX_REC_LEN);
2419		else
2420			return cpu_to_le16(0);
2421	}
2422	return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
2423#else
2424	return cpu_to_le16(len);
2425#endif
2426}
2427
2428/*
2429 * Hash Tree Directory indexing
2430 * (c) Daniel Phillips, 2001
2431 */
2432
2433#define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
2434		    ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2435#define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \
2436		    !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir)))
2437#define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2438
2439/* Legal values for the dx_root hash_version field: */
2440
2441#define DX_HASH_LEGACY			0
2442#define DX_HASH_HALF_MD4		1
2443#define DX_HASH_TEA			2
2444#define DX_HASH_LEGACY_UNSIGNED		3
2445#define DX_HASH_HALF_MD4_UNSIGNED	4
2446#define DX_HASH_TEA_UNSIGNED		5
2447#define DX_HASH_SIPHASH			6
2448
2449static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2450			      const void *address, unsigned int length)
2451{
2452	struct {
2453		struct shash_desc shash;
2454		char ctx[4];
2455	} desc;
2456
2457	BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2458
2459	desc.shash.tfm = sbi->s_chksum_driver;
2460	*(u32 *)desc.ctx = crc;
2461
2462	BUG_ON(crypto_shash_update(&desc.shash, address, length));
2463
2464	return *(u32 *)desc.ctx;
2465}
2466
2467#ifdef __KERNEL__
2468
2469/* hash info structure used by the directory hash */
2470struct dx_hash_info
2471{
2472	u32		hash;
2473	u32		minor_hash;
2474	int		hash_version;
2475	u32		*seed;
2476};
2477
2478
2479/* 32 and 64 bit signed EOF for dx directories */
2480#define EXT4_HTREE_EOF_32BIT   ((1UL  << (32 - 1)) - 1)
2481#define EXT4_HTREE_EOF_64BIT   ((1ULL << (64 - 1)) - 1)
2482
2483
2484/*
2485 * Control parameters used by ext4_htree_next_block
2486 */
2487#define HASH_NB_ALWAYS		1
2488
2489struct ext4_filename {
2490	const struct qstr *usr_fname;
2491	struct fscrypt_str disk_name;
2492	struct dx_hash_info hinfo;
2493#ifdef CONFIG_FS_ENCRYPTION
2494	struct fscrypt_str crypto_buf;
2495#endif
2496#ifdef CONFIG_UNICODE
2497	struct fscrypt_str cf_name;
2498#endif
2499};
2500
2501#define fname_name(p) ((p)->disk_name.name)
2502#define fname_usr_name(p) ((p)->usr_fname->name)
2503#define fname_len(p)  ((p)->disk_name.len)
2504
2505/*
2506 * Describe an inode's exact location on disk and in memory
2507 */
2508struct ext4_iloc
2509{
2510	struct buffer_head *bh;
2511	unsigned long offset;
2512	ext4_group_t block_group;
2513};
2514
2515static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2516{
2517	return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2518}
2519
2520static inline bool ext4_is_quota_file(struct inode *inode)
2521{
2522	return IS_NOQUOTA(inode) &&
2523	       !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL);
2524}
2525
2526/*
2527 * This structure is stuffed into the struct file's private_data field
2528 * for directories.  It is where we put information so that we can do
2529 * readdir operations in hash tree order.
2530 */
2531struct dir_private_info {
2532	struct rb_root	root;
2533	struct rb_node	*curr_node;
2534	struct fname	*extra_fname;
2535	loff_t		last_pos;
2536	__u32		curr_hash;
2537	__u32		curr_minor_hash;
2538	__u32		next_hash;
2539};
2540
2541/* calculate the first block number of the group */
2542static inline ext4_fsblk_t
2543ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2544{
2545	return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2546		le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2547}
2548
2549/*
2550 * Special error return code only used by dx_probe() and its callers.
2551 */
2552#define ERR_BAD_DX_DIR	(-(MAX_ERRNO - 1))
2553
2554/* htree levels for ext4 */
2555#define	EXT4_HTREE_LEVEL_COMPAT	2
2556#define	EXT4_HTREE_LEVEL	3
2557
2558static inline int ext4_dir_htree_level(struct super_block *sb)
2559{
2560	return ext4_has_feature_largedir(sb) ?
2561		EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT;
2562}
2563
2564/*
2565 * Timeout and state flag for lazy initialization inode thread.
2566 */
2567#define EXT4_DEF_LI_WAIT_MULT			10
2568#define EXT4_DEF_LI_MAX_START_DELAY		5
2569#define EXT4_LAZYINIT_QUIT			0x0001
2570#define EXT4_LAZYINIT_RUNNING			0x0002
2571
2572/*
2573 * Lazy inode table initialization info
2574 */
2575struct ext4_lazy_init {
2576	unsigned long		li_state;
2577	struct list_head	li_request_list;
2578	struct mutex		li_list_mtx;
2579};
2580
2581enum ext4_li_mode {
2582	EXT4_LI_MODE_PREFETCH_BBITMAP,
2583	EXT4_LI_MODE_ITABLE,
2584};
2585
2586struct ext4_li_request {
2587	struct super_block	*lr_super;
2588	enum ext4_li_mode	lr_mode;
2589	ext4_group_t		lr_first_not_zeroed;
2590	ext4_group_t		lr_next_group;
2591	struct list_head	lr_request;
2592	unsigned long		lr_next_sched;
2593	unsigned long		lr_timeout;
2594};
2595
2596struct ext4_features {
2597	struct kobject f_kobj;
2598	struct completion f_kobj_unregister;
2599};
2600
2601/*
2602 * This structure will be used for multiple mount protection. It will be
2603 * written into the block number saved in the s_mmp_block field in the
2604 * superblock. Programs that check MMP should assume that if
2605 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2606 * to use the filesystem, regardless of how old the timestamp is.
2607 */
2608#define EXT4_MMP_MAGIC     0x004D4D50U /* ASCII for MMP */
2609#define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2610#define EXT4_MMP_SEQ_FSCK  0xE24D4D50U /* mmp_seq value when being fscked */
2611#define EXT4_MMP_SEQ_MAX   0xE24D4D4FU /* maximum valid mmp_seq value */
2612
2613struct mmp_struct {
2614	__le32	mmp_magic;		/* Magic number for MMP */
2615	__le32	mmp_seq;		/* Sequence no. updated periodically */
2616
2617	/*
2618	 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2619	 * purposes and do not affect the correctness of the algorithm
2620	 */
2621	__le64	mmp_time;		/* Time last updated */
2622	char	mmp_nodename[64];	/* Node which last updated MMP block */
2623	char	mmp_bdevname[32];	/* Bdev which last updated MMP block */
2624
2625	/*
2626	 * mmp_check_interval is used to verify if the MMP block has been
2627	 * updated on the block device. The value is updated based on the
2628	 * maximum time to write the MMP block during an update cycle.
2629	 */
2630	__le16	mmp_check_interval;
2631
2632	__le16	mmp_pad1;
2633	__le32	mmp_pad2[226];
2634	__le32	mmp_checksum;		/* crc32c(uuid+mmp_block) */
2635};
2636
2637/* arguments passed to the mmp thread */
2638struct mmpd_data {
2639	struct buffer_head *bh; /* bh from initial read_mmp_block() */
2640	struct super_block *sb;  /* super block of the fs */
2641};
2642
2643/*
2644 * Check interval multiplier
2645 * The MMP block is written every update interval and initially checked every
2646 * update interval x the multiplier (the value is then adapted based on the
2647 * write latency). The reason is that writes can be delayed under load and we
2648 * don't want readers to incorrectly assume that the filesystem is no longer
2649 * in use.
2650 */
2651#define EXT4_MMP_CHECK_MULT		2UL
2652
2653/*
2654 * Minimum interval for MMP checking in seconds.
2655 */
2656#define EXT4_MMP_MIN_CHECK_INTERVAL	5UL
2657
2658/*
2659 * Maximum interval for MMP checking in seconds.
2660 */
2661#define EXT4_MMP_MAX_CHECK_INTERVAL	300UL
2662
2663/*
2664 * Function prototypes
2665 */
2666
2667/*
2668 * Ok, these declarations are also in <linux/kernel.h> but none of the
2669 * ext4 source programs needs to include it so they are duplicated here.
2670 */
2671# define NORET_TYPE	/**/
2672# define ATTRIB_NORET	__attribute__((noreturn))
2673# define NORET_AND	noreturn,
2674
2675/* bitmap.c */
2676extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2677void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2678				struct ext4_group_desc *gdp,
2679				struct buffer_head *bh, int sz);
2680int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2681				  struct ext4_group_desc *gdp,
2682				  struct buffer_head *bh, int sz);
2683void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2684				struct ext4_group_desc *gdp,
2685				struct buffer_head *bh);
2686int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2687				  struct ext4_group_desc *gdp,
2688				  struct buffer_head *bh);
2689
2690/* balloc.c */
2691extern void ext4_get_group_no_and_offset(struct super_block *sb,
2692					 ext4_fsblk_t blocknr,
2693					 ext4_group_t *blockgrpp,
2694					 ext4_grpblk_t *offsetp);
2695extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2696					  ext4_fsblk_t block);
2697
2698extern unsigned int ext4_block_group(struct super_block *sb,
2699			ext4_fsblk_t blocknr);
2700extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2701			ext4_fsblk_t blocknr);
2702extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2703extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2704			ext4_group_t group);
2705extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2706					 ext4_fsblk_t goal,
2707					 unsigned int flags,
2708					 unsigned long *count,
2709					 int *errp);
2710extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2711				    s64 nclusters, unsigned int flags);
2712extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2713extern void ext4_check_blocks_bitmap(struct super_block *);
2714extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2715						    ext4_group_t block_group,
2716						    struct buffer_head ** bh);
2717extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2718
2719extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2720						ext4_group_t block_group,
2721						bool ignore_locked);
2722extern int ext4_wait_block_bitmap(struct super_block *sb,
2723				  ext4_group_t block_group,
2724				  struct buffer_head *bh);
2725extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2726						  ext4_group_t block_group);
2727extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2728					      ext4_group_t block_group,
2729					      struct ext4_group_desc *gdp);
2730ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2731
2732#ifdef CONFIG_UNICODE
2733extern int ext4_fname_setup_ci_filename(struct inode *dir,
2734					 const struct qstr *iname,
2735					 struct ext4_filename *fname);
2736#endif
2737
2738#ifdef CONFIG_FS_ENCRYPTION
2739static inline void ext4_fname_from_fscrypt_name(struct ext4_filename *dst,
2740						const struct fscrypt_name *src)
2741{
2742	memset(dst, 0, sizeof(*dst));
2743
2744	dst->usr_fname = src->usr_fname;
2745	dst->disk_name = src->disk_name;
2746	dst->hinfo.hash = src->hash;
2747	dst->hinfo.minor_hash = src->minor_hash;
2748	dst->crypto_buf = src->crypto_buf;
2749}
2750
2751static inline int ext4_fname_setup_filename(struct inode *dir,
2752					    const struct qstr *iname,
2753					    int lookup,
2754					    struct ext4_filename *fname)
2755{
2756	struct fscrypt_name name;
2757	int err;
2758
2759	err = fscrypt_setup_filename(dir, iname, lookup, &name);
2760	if (err)
2761		return err;
2762
2763	ext4_fname_from_fscrypt_name(fname, &name);
2764
2765#ifdef CONFIG_UNICODE
2766	err = ext4_fname_setup_ci_filename(dir, iname, fname);
2767#endif
2768	return err;
2769}
2770
2771static inline int ext4_fname_prepare_lookup(struct inode *dir,
2772					    struct dentry *dentry,
2773					    struct ext4_filename *fname)
2774{
2775	struct fscrypt_name name;
2776	int err;
2777
2778	err = fscrypt_prepare_lookup(dir, dentry, &name);
2779	if (err)
2780		return err;
2781
2782	ext4_fname_from_fscrypt_name(fname, &name);
2783
2784#ifdef CONFIG_UNICODE
2785	err = ext4_fname_setup_ci_filename(dir, &dentry->d_name, fname);
2786#endif
2787	return err;
2788}
2789
2790static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2791{
2792	struct fscrypt_name name;
2793
2794	name.crypto_buf = fname->crypto_buf;
2795	fscrypt_free_filename(&name);
2796
2797	fname->crypto_buf.name = NULL;
2798	fname->usr_fname = NULL;
2799	fname->disk_name.name = NULL;
2800
2801#ifdef CONFIG_UNICODE
2802	kfree(fname->cf_name.name);
2803	fname->cf_name.name = NULL;
2804#endif
2805}
2806#else /* !CONFIG_FS_ENCRYPTION */
2807static inline int ext4_fname_setup_filename(struct inode *dir,
2808					    const struct qstr *iname,
2809					    int lookup,
2810					    struct ext4_filename *fname)
2811{
2812	int err = 0;
2813	fname->usr_fname = iname;
2814	fname->disk_name.name = (unsigned char *) iname->name;
2815	fname->disk_name.len = iname->len;
2816
2817#ifdef CONFIG_UNICODE
2818	err = ext4_fname_setup_ci_filename(dir, iname, fname);
2819#endif
2820
2821	return err;
2822}
2823
2824static inline int ext4_fname_prepare_lookup(struct inode *dir,
2825					    struct dentry *dentry,
2826					    struct ext4_filename *fname)
2827{
2828	return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname);
2829}
2830
2831static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2832{
2833#ifdef CONFIG_UNICODE
2834	kfree(fname->cf_name.name);
2835	fname->cf_name.name = NULL;
2836#endif
2837}
2838#endif /* !CONFIG_FS_ENCRYPTION */
2839
2840/* dir.c */
2841extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2842				  struct file *,
2843				  struct ext4_dir_entry_2 *,
2844				  struct buffer_head *, char *, int,
2845				  unsigned int);
2846#define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2847	unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2848				(de), (bh), (buf), (size), (offset)))
2849extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2850				__u32 minor_hash,
2851				struct ext4_dir_entry_2 *dirent,
2852				struct fscrypt_str *ent_name);
2853extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2854extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2855			     struct buffer_head *bh,
2856			     void *buf, int buf_size,
2857			     struct ext4_filename *fname,
2858			     struct ext4_dir_entry_2 **dest_de);
2859void ext4_insert_dentry(struct inode *dir, struct inode *inode,
2860			struct ext4_dir_entry_2 *de,
2861			int buf_size,
2862			struct ext4_filename *fname);
2863static inline void ext4_update_dx_flag(struct inode *inode)
2864{
2865	if (!ext4_has_feature_dir_index(inode->i_sb) &&
2866	    ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) {
2867		/* ext4_iget() should have caught this... */
2868		WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb));
2869		ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2870	}
2871}
2872static const unsigned char ext4_filetype_table[] = {
2873	DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2874};
2875
2876static inline  unsigned char get_dtype(struct super_block *sb, int filetype)
2877{
2878	if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2879		return DT_UNKNOWN;
2880
2881	return ext4_filetype_table[filetype];
2882}
2883extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2884			     void *buf, int buf_size);
2885
2886/* fsync.c */
2887extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2888
2889/* hash.c */
2890extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len,
2891			  struct dx_hash_info *hinfo);
2892
2893/* ialloc.c */
2894extern int ext4_mark_inode_used(struct super_block *sb, int ino);
2895extern struct inode *__ext4_new_inode(struct user_namespace *, handle_t *,
2896				      struct inode *, umode_t,
2897				      const struct qstr *qstr, __u32 goal,
2898				      uid_t *owner, __u32 i_flags,
2899				      int handle_type, unsigned int line_no,
2900				      int nblocks);
2901
2902#define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags)          \
2903	__ext4_new_inode(&init_user_ns, (handle), (dir), (mode), (qstr),       \
2904			 (goal), (owner), i_flags, 0, 0, 0)
2905#define ext4_new_inode_start_handle(mnt_userns, dir, mode, qstr, goal, owner, \
2906				    type, nblocks)		    \
2907	__ext4_new_inode((mnt_userns), NULL, (dir), (mode), (qstr), (goal), (owner), \
2908			 0, (type), __LINE__, (nblocks))
2909
2910
2911extern void ext4_free_inode(handle_t *, struct inode *);
2912extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2913extern unsigned long ext4_count_free_inodes(struct super_block *);
2914extern unsigned long ext4_count_dirs(struct super_block *);
2915extern void ext4_check_inodes_bitmap(struct super_block *);
2916extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2917extern int ext4_init_inode_table(struct super_block *sb,
2918				 ext4_group_t group, int barrier);
2919extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2920
2921/* fast_commit.c */
2922int ext4_fc_info_show(struct seq_file *seq, void *v);
2923void ext4_fc_init(struct super_block *sb, journal_t *journal);
2924void ext4_fc_init_inode(struct inode *inode);
2925void ext4_fc_track_range(handle_t *handle, struct inode *inode, ext4_lblk_t start,
2926			 ext4_lblk_t end);
2927void __ext4_fc_track_unlink(handle_t *handle, struct inode *inode,
2928	struct dentry *dentry);
2929void __ext4_fc_track_link(handle_t *handle, struct inode *inode,
2930	struct dentry *dentry);
2931void ext4_fc_track_unlink(handle_t *handle, struct dentry *dentry);
2932void ext4_fc_track_link(handle_t *handle, struct dentry *dentry);
2933void __ext4_fc_track_create(handle_t *handle, struct inode *inode,
2934			    struct dentry *dentry);
2935void ext4_fc_track_create(handle_t *handle, struct dentry *dentry);
2936void ext4_fc_track_inode(handle_t *handle, struct inode *inode);
2937void ext4_fc_mark_ineligible(struct super_block *sb, int reason);
2938void ext4_fc_start_ineligible(struct super_block *sb, int reason);
2939void ext4_fc_stop_ineligible(struct super_block *sb);
2940void ext4_fc_start_update(struct inode *inode);
2941void ext4_fc_stop_update(struct inode *inode);
2942void ext4_fc_del(struct inode *inode);
2943bool ext4_fc_replay_check_excluded(struct super_block *sb, ext4_fsblk_t block);
2944void ext4_fc_replay_cleanup(struct super_block *sb);
2945int ext4_fc_commit(journal_t *journal, tid_t commit_tid);
2946int __init ext4_fc_init_dentry_cache(void);
2947
2948/* mballoc.c */
2949extern const struct seq_operations ext4_mb_seq_groups_ops;
2950extern const struct seq_operations ext4_mb_seq_structs_summary_ops;
2951extern long ext4_mb_stats;
2952extern long ext4_mb_max_to_scan;
2953extern int ext4_seq_mb_stats_show(struct seq_file *seq, void *offset);
2954extern int ext4_mb_init(struct super_block *);
2955extern int ext4_mb_release(struct super_block *);
2956extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2957				struct ext4_allocation_request *, int *);
2958extern int ext4_mb_reserve_blocks(struct super_block *, int);
2959extern void ext4_discard_preallocations(struct inode *, unsigned int);
2960extern int __init ext4_init_mballoc(void);
2961extern void ext4_exit_mballoc(void);
2962extern ext4_group_t ext4_mb_prefetch(struct super_block *sb,
2963				     ext4_group_t group,
2964				     unsigned int nr, int *cnt);
2965extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group,
2966				  unsigned int nr);
2967
2968extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2969			     struct buffer_head *bh, ext4_fsblk_t block,
2970			     unsigned long count, int flags);
2971extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2972				   ext4_group_t ngroups);
2973extern int ext4_mb_add_groupinfo(struct super_block *sb,
2974		ext4_group_t i, struct ext4_group_desc *desc);
2975extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2976				ext4_fsblk_t block, unsigned long count);
2977extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2978extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid);
2979extern void ext4_mb_mark_bb(struct super_block *sb, ext4_fsblk_t block,
2980		       int len, int state);
2981
2982/* inode.c */
2983void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw,
2984			 struct ext4_inode_info *ei);
2985int ext4_inode_is_fast_symlink(struct inode *inode);
2986struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2987struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2988int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count,
2989		     bool wait, struct buffer_head **bhs);
2990int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2991			     struct buffer_head *bh_result, int create);
2992int ext4_get_block(struct inode *inode, sector_t iblock,
2993		   struct buffer_head *bh_result, int create);
2994int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2995			   struct buffer_head *bh, int create);
2996int ext4_walk_page_buffers(handle_t *handle,
2997			   struct inode *inode,
2998			   struct buffer_head *head,
2999			   unsigned from,
3000			   unsigned to,
3001			   int *partial,
3002			   int (*fn)(handle_t *handle, struct inode *inode,
3003				     struct buffer_head *bh));
3004int do_journal_get_write_access(handle_t *handle, struct inode *inode,
3005				struct buffer_head *bh);
3006#define FALL_BACK_TO_NONDELALLOC 1
3007#define CONVERT_INLINE_DATA	 2
3008
3009typedef enum {
3010	EXT4_IGET_NORMAL =	0,
3011	EXT4_IGET_SPECIAL =	0x0001, /* OK to iget a system inode */
3012	EXT4_IGET_HANDLE = 	0x0002	/* Inode # is from a handle */
3013} ext4_iget_flags;
3014
3015extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
3016				 ext4_iget_flags flags, const char *function,
3017				 unsigned int line);
3018
3019#define ext4_iget(sb, ino, flags) \
3020	__ext4_iget((sb), (ino), (flags), __func__, __LINE__)
3021
3022extern int  ext4_write_inode(struct inode *, struct writeback_control *);
3023extern int  ext4_setattr(struct user_namespace *, struct dentry *,
3024			 struct iattr *);
3025extern int  ext4_getattr(struct user_namespace *, const struct path *,
3026			 struct kstat *, u32, unsigned int);
3027extern void ext4_evict_inode(struct inode *);
3028extern void ext4_clear_inode(struct inode *);
3029extern int  ext4_file_getattr(struct user_namespace *, const struct path *,
3030			      struct kstat *, u32, unsigned int);
3031extern int  ext4_sync_inode(handle_t *, struct inode *);
3032extern void ext4_dirty_inode(struct inode *, int);
3033extern int ext4_change_inode_journal_flag(struct inode *, int);
3034extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
3035extern int ext4_get_fc_inode_loc(struct super_block *sb, unsigned long ino,
3036			  struct ext4_iloc *iloc);
3037extern int ext4_inode_attach_jinode(struct inode *inode);
3038extern int ext4_can_truncate(struct inode *inode);
3039extern int ext4_truncate(struct inode *);
3040extern int ext4_break_layouts(struct inode *);
3041extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
3042extern void ext4_set_inode_flags(struct inode *, bool init);
3043extern int ext4_alloc_da_blocks(struct inode *inode);
3044extern void ext4_set_aops(struct inode *inode);
3045extern int ext4_writepage_trans_blocks(struct inode *);
3046extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
3047extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
3048			     loff_t lstart, loff_t lend);
3049extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf);
3050extern vm_fault_t ext4_filemap_fault(struct vm_fault *vmf);
3051extern qsize_t *ext4_get_reserved_space(struct inode *inode);
3052extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
3053extern void ext4_da_release_space(struct inode *inode, int to_free);
3054extern void ext4_da_update_reserve_space(struct inode *inode,
3055					int used, int quota_claim);
3056extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
3057			      ext4_fsblk_t pblk, ext4_lblk_t len);
3058
3059/* indirect.c */
3060extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
3061				struct ext4_map_blocks *map, int flags);
3062extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
3063extern void ext4_ind_truncate(handle_t *, struct inode *inode);
3064extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
3065				 ext4_lblk_t start, ext4_lblk_t end);
3066
3067/* ioctl.c */
3068extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
3069extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
3070int ext4_fileattr_set(struct user_namespace *mnt_userns,
3071		      struct dentry *dentry, struct fileattr *fa);
3072int ext4_fileattr_get(struct dentry *dentry, struct fileattr *fa);
3073extern void ext4_reset_inode_seed(struct inode *inode);
3074
3075/* migrate.c */
3076extern int ext4_ext_migrate(struct inode *);
3077extern int ext4_ind_migrate(struct inode *inode);
3078
3079/* namei.c */
3080extern int ext4_init_new_dir(handle_t *handle, struct inode *dir,
3081			     struct inode *inode);
3082extern int ext4_dirblock_csum_verify(struct inode *inode,
3083				     struct buffer_head *bh);
3084extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
3085				__u32 start_minor_hash, __u32 *next_hash);
3086extern int ext4_search_dir(struct buffer_head *bh,
3087			   char *search_buf,
3088			   int buf_size,
3089			   struct inode *dir,
3090			   struct ext4_filename *fname,
3091			   unsigned int offset,
3092			   struct ext4_dir_entry_2 **res_dir);
3093extern int ext4_generic_delete_entry(struct inode *dir,
3094				     struct ext4_dir_entry_2 *de_del,
3095				     struct buffer_head *bh,
3096				     void *entry_buf,
3097				     int buf_size,
3098				     int csum_size);
3099extern bool ext4_empty_dir(struct inode *inode);
3100
3101/* resize.c */
3102extern void ext4_kvfree_array_rcu(void *to_free);
3103extern int ext4_group_add(struct super_block *sb,
3104				struct ext4_new_group_data *input);
3105extern int ext4_group_extend(struct super_block *sb,
3106				struct ext4_super_block *es,
3107				ext4_fsblk_t n_blocks_count);
3108extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
3109
3110/* super.c */
3111extern struct buffer_head *ext4_sb_bread(struct super_block *sb,
3112					 sector_t block, int op_flags);
3113extern struct buffer_head *ext4_sb_bread_unmovable(struct super_block *sb,
3114						   sector_t block);
3115extern void ext4_read_bh_nowait(struct buffer_head *bh, int op_flags,
3116				bh_end_io_t *end_io);
3117extern int ext4_read_bh(struct buffer_head *bh, int op_flags,
3118			bh_end_io_t *end_io);
3119extern int ext4_read_bh_lock(struct buffer_head *bh, int op_flags, bool wait);
3120extern void ext4_sb_breadahead_unmovable(struct super_block *sb, sector_t block);
3121extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
3122extern int ext4_calculate_overhead(struct super_block *sb);
3123extern void ext4_superblock_csum_set(struct super_block *sb);
3124extern int ext4_alloc_flex_bg_array(struct super_block *sb,
3125				    ext4_group_t ngroup);
3126extern const char *ext4_decode_error(struct super_block *sb, int errno,
3127				     char nbuf[16]);
3128extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb,
3129					     ext4_group_t block_group,
3130					     unsigned int flags);
3131
3132extern __printf(7, 8)
3133void __ext4_error(struct super_block *, const char *, unsigned int, bool,
3134		  int, __u64, const char *, ...);
3135extern __printf(6, 7)
3136void __ext4_error_inode(struct inode *, const char *, unsigned int,
3137			ext4_fsblk_t, int, const char *, ...);
3138extern __printf(5, 6)
3139void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
3140		     const char *, ...);
3141extern void __ext4_std_error(struct super_block *, const char *,
3142			     unsigned int, int);
3143extern __printf(4, 5)
3144void __ext4_warning(struct super_block *, const char *, unsigned int,
3145		    const char *, ...);
3146extern __printf(4, 5)
3147void __ext4_warning_inode(const struct inode *inode, const char *function,
3148			  unsigned int line, const char *fmt, ...);
3149extern __printf(3, 4)
3150void __ext4_msg(struct super_block *, const char *, const char *, ...);
3151extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
3152			   const char *, unsigned int, const char *);
3153extern __printf(7, 8)
3154void __ext4_grp_locked_error(const char *, unsigned int,
3155			     struct super_block *, ext4_group_t,
3156			     unsigned long, ext4_fsblk_t,
3157			     const char *, ...);
3158
3159#define EXT4_ERROR_INODE(inode, fmt, a...) \
3160	ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
3161
3162#define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...)			\
3163	__ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a)
3164
3165#define ext4_error_inode_block(inode, block, err, fmt, a...)		\
3166	__ext4_error_inode((inode), __func__, __LINE__, (block), (err),	\
3167			   (fmt), ## a)
3168
3169#define EXT4_ERROR_FILE(file, block, fmt, a...)				\
3170	ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
3171
3172#define ext4_abort(sb, err, fmt, a...)					\
3173	__ext4_error((sb), __func__, __LINE__, true, (err), 0, (fmt), ## a)
3174
3175#ifdef CONFIG_PRINTK
3176
3177#define ext4_error_inode(inode, func, line, block, fmt, ...)		\
3178	__ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__)
3179#define ext4_error_inode_err(inode, func, line, block, err, fmt, ...)	\
3180	__ext4_error_inode((inode), (func), (line), (block), 		\
3181			   (err), (fmt), ##__VA_ARGS__)
3182#define ext4_error_file(file, func, line, block, fmt, ...)		\
3183	__ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
3184#define ext4_error(sb, fmt, ...)					\
3185	__ext4_error((sb), __func__, __LINE__, false, 0, 0, (fmt),	\
3186		##__VA_ARGS__)
3187#define ext4_error_err(sb, err, fmt, ...)				\
3188	__ext4_error((sb), __func__, __LINE__, false, (err), 0, (fmt),	\
3189		##__VA_ARGS__)
3190#define ext4_warning(sb, fmt, ...)					\
3191	__ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
3192#define ext4_warning_inode(inode, fmt, ...)				\
3193	__ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
3194#define ext4_msg(sb, level, fmt, ...)				\
3195	__ext4_msg(sb, level, fmt, ##__VA_ARGS__)
3196#define dump_mmp_msg(sb, mmp, msg)					\
3197	__dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
3198#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...)		\
3199	__ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
3200				fmt, ##__VA_ARGS__)
3201
3202#else
3203
3204#define ext4_error_inode(inode, func, line, block, fmt, ...)		\
3205do {									\
3206	no_printk(fmt, ##__VA_ARGS__);					\
3207	__ext4_error_inode(inode, "", 0, block, 0, " ");		\
3208} while (0)
3209#define ext4_error_inode_err(inode, func, line, block, err, fmt, ...)	\
3210do {									\
3211	no_printk(fmt, ##__VA_ARGS__);					\
3212	__ext4_error_inode(inode, "", 0, block, err, " ");		\
3213} while (0)
3214#define ext4_error_file(file, func, line, block, fmt, ...)		\
3215do {									\
3216	no_printk(fmt, ##__VA_ARGS__);					\
3217	__ext4_error_file(file, "", 0, block, " ");			\
3218} while (0)
3219#define ext4_error(sb, fmt, ...)					\
3220do {									\
3221	no_printk(fmt, ##__VA_ARGS__);					\
3222	__ext4_error(sb, "", 0, false, 0, 0, " ");			\
3223} while (0)
3224#define ext4_error_err(sb, err, fmt, ...)				\
3225do {									\
3226	no_printk(fmt, ##__VA_ARGS__);					\
3227	__ext4_error(sb, "", 0, false, err, 0, " ");			\
3228} while (0)
3229#define ext4_warning(sb, fmt, ...)					\
3230do {									\
3231	no_printk(fmt, ##__VA_ARGS__);					\
3232	__ext4_warning(sb, "", 0, " ");					\
3233} while (0)
3234#define ext4_warning_inode(inode, fmt, ...)				\
3235do {									\
3236	no_printk(fmt, ##__VA_ARGS__);					\
3237	__ext4_warning_inode(inode, "", 0, " ");			\
3238} while (0)
3239#define ext4_msg(sb, level, fmt, ...)					\
3240do {									\
3241	no_printk(fmt, ##__VA_ARGS__);					\
3242	__ext4_msg(sb, "", " ");					\
3243} while (0)
3244#define dump_mmp_msg(sb, mmp, msg)					\
3245	__dump_mmp_msg(sb, mmp, "", 0, "")
3246#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...)		\
3247do {									\
3248	no_printk(fmt, ##__VA_ARGS__);				\
3249	__ext4_grp_locked_error("", 0, sb, grp, ino, block, " ");	\
3250} while (0)
3251
3252#endif
3253
3254extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
3255				      struct ext4_group_desc *bg);
3256extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
3257				      struct ext4_group_desc *bg);
3258extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
3259				     struct ext4_group_desc *bg);
3260extern __u32 ext4_free_group_clusters(struct super_block *sb,
3261				      struct ext4_group_desc *bg);
3262extern __u32 ext4_free_inodes_count(struct super_block *sb,
3263				 struct ext4_group_desc *bg);
3264extern __u32 ext4_used_dirs_count(struct super_block *sb,
3265				struct ext4_group_desc *bg);
3266extern __u32 ext4_itable_unused_count(struct super_block *sb,
3267				   struct ext4_group_desc *bg);
3268extern void ext4_block_bitmap_set(struct super_block *sb,
3269				  struct ext4_group_desc *bg, ext4_fsblk_t blk);
3270extern void ext4_inode_bitmap_set(struct super_block *sb,
3271				  struct ext4_group_desc *bg, ext4_fsblk_t blk);
3272extern void ext4_inode_table_set(struct super_block *sb,
3273				 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3274extern void ext4_free_group_clusters_set(struct super_block *sb,
3275					 struct ext4_group_desc *bg,
3276					 __u32 count);
3277extern void ext4_free_inodes_set(struct super_block *sb,
3278				struct ext4_group_desc *bg, __u32 count);
3279extern void ext4_used_dirs_set(struct super_block *sb,
3280				struct ext4_group_desc *bg, __u32 count);
3281extern void ext4_itable_unused_set(struct super_block *sb,
3282				   struct ext4_group_desc *bg, __u32 count);
3283extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
3284				       struct ext4_group_desc *gdp);
3285extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
3286				     struct ext4_group_desc *gdp);
3287extern int ext4_register_li_request(struct super_block *sb,
3288				    ext4_group_t first_not_zeroed);
3289
3290static inline int ext4_has_metadata_csum(struct super_block *sb)
3291{
3292	WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
3293		     !EXT4_SB(sb)->s_chksum_driver);
3294
3295	return ext4_has_feature_metadata_csum(sb) &&
3296	       (EXT4_SB(sb)->s_chksum_driver != NULL);
3297}
3298
3299static inline int ext4_has_group_desc_csum(struct super_block *sb)
3300{
3301	return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb);
3302}
3303
3304#define ext4_read_incompat_64bit_val(es, name) \
3305	(((es)->s_feature_incompat & cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT) \
3306		? (ext4_fsblk_t)le32_to_cpu(es->name##_hi) << 32 : 0) | \
3307		le32_to_cpu(es->name##_lo))
3308
3309static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
3310{
3311	return ext4_read_incompat_64bit_val(es, s_blocks_count);
3312}
3313
3314static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
3315{
3316	return ext4_read_incompat_64bit_val(es, s_r_blocks_count);
3317}
3318
3319static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
3320{
3321	return ext4_read_incompat_64bit_val(es, s_free_blocks_count);
3322}
3323
3324static inline void ext4_blocks_count_set(struct ext4_super_block *es,
3325					 ext4_fsblk_t blk)
3326{
3327	es->s_blocks_count_lo = cpu_to_le32((u32)blk);
3328	es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
3329}
3330
3331static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
3332					      ext4_fsblk_t blk)
3333{
3334	es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
3335	es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
3336}
3337
3338static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
3339					   ext4_fsblk_t blk)
3340{
3341	es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
3342	es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
3343}
3344
3345static inline loff_t ext4_isize(struct super_block *sb,
3346				struct ext4_inode *raw_inode)
3347{
3348	if (ext4_has_feature_largedir(sb) ||
3349	    S_ISREG(le16_to_cpu(raw_inode->i_mode)))
3350		return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
3351			le32_to_cpu(raw_inode->i_size_lo);
3352
3353	return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
3354}
3355
3356static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
3357{
3358	raw_inode->i_size_lo = cpu_to_le32(i_size);
3359	raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
3360}
3361
3362static inline
3363struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
3364					    ext4_group_t group)
3365{
3366	 struct ext4_group_info **grp_info;
3367	 long indexv, indexh;
3368	 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
3369	 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
3370	 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
3371	 grp_info = sbi_array_rcu_deref(EXT4_SB(sb), s_group_info, indexv);
3372	 return grp_info[indexh];
3373}
3374
3375/*
3376 * Reading s_groups_count requires using smp_rmb() afterwards.  See
3377 * the locking protocol documented in the comments of ext4_group_add()
3378 * in resize.c
3379 */
3380static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
3381{
3382	ext4_group_t	ngroups = EXT4_SB(sb)->s_groups_count;
3383
3384	smp_rmb();
3385	return ngroups;
3386}
3387
3388static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
3389					     ext4_group_t block_group)
3390{
3391	return block_group >> sbi->s_log_groups_per_flex;
3392}
3393
3394static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
3395{
3396	return 1 << sbi->s_log_groups_per_flex;
3397}
3398
3399#define ext4_std_error(sb, errno)				\
3400do {								\
3401	if ((errno))						\
3402		__ext4_std_error((sb), __func__, __LINE__, (errno));	\
3403} while (0)
3404
3405#ifdef CONFIG_SMP
3406/* Each CPU can accumulate percpu_counter_batch clusters in their local
3407 * counters. So we need to make sure we have free clusters more
3408 * than percpu_counter_batch  * nr_cpu_ids. Also add a window of 4 times.
3409 */
3410#define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
3411#else
3412#define EXT4_FREECLUSTERS_WATERMARK 0
3413#endif
3414
3415/* Update i_disksize. Requires i_mutex to avoid races with truncate */
3416static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
3417{
3418	WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
3419		     !inode_is_locked(inode));
3420	down_write(&EXT4_I(inode)->i_data_sem);
3421	if (newsize > EXT4_I(inode)->i_disksize)
3422		WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize);
3423	up_write(&EXT4_I(inode)->i_data_sem);
3424}
3425
3426/* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
3427static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
3428{
3429	int changed = 0;
3430
3431	if (newsize > inode->i_size) {
3432		i_size_write(inode, newsize);
3433		changed = 1;
3434	}
3435	if (newsize > EXT4_I(inode)->i_disksize) {
3436		ext4_update_i_disksize(inode, newsize);
3437		changed |= 2;
3438	}
3439	return changed;
3440}
3441
3442int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
3443				      loff_t len);
3444
3445struct ext4_group_info {
3446	unsigned long   bb_state;
3447#ifdef AGGRESSIVE_CHECK
3448	unsigned long	bb_check_counter;
3449#endif
3450	struct rb_root  bb_free_root;
3451	ext4_grpblk_t	bb_first_free;	/* first free block */
3452	ext4_grpblk_t	bb_free;	/* total free blocks */
3453	ext4_grpblk_t	bb_fragments;	/* nr of freespace fragments */
3454	ext4_grpblk_t	bb_largest_free_order;/* order of largest frag in BG */
3455	ext4_group_t	bb_group;	/* Group number */
3456	struct          list_head bb_prealloc_list;
3457#ifdef DOUBLE_CHECK
3458	void            *bb_bitmap;
3459#endif
3460	struct rw_semaphore alloc_sem;
3461	struct rb_node	bb_avg_fragment_size_rb;
3462	struct list_head bb_largest_free_order_node;
3463	ext4_grpblk_t	bb_counters[];	/* Nr of free power-of-two-block
3464					 * regions, index is order.
3465					 * bb_counters[3] = 5 means
3466					 * 5 free 8-block regions. */
3467};
3468
3469#define EXT4_GROUP_INFO_NEED_INIT_BIT		0
3470#define EXT4_GROUP_INFO_WAS_TRIMMED_BIT		1
3471#define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT	2
3472#define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT	3
3473#define EXT4_GROUP_INFO_BBITMAP_CORRUPT		\
3474	(1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT)
3475#define EXT4_GROUP_INFO_IBITMAP_CORRUPT		\
3476	(1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT)
3477#define EXT4_GROUP_INFO_BBITMAP_READ_BIT	4
3478
3479#define EXT4_MB_GRP_NEED_INIT(grp)	\
3480	(test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
3481#define EXT4_MB_GRP_BBITMAP_CORRUPT(grp)	\
3482	(test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3483#define EXT4_MB_GRP_IBITMAP_CORRUPT(grp)	\
3484	(test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3485
3486#define EXT4_MB_GRP_WAS_TRIMMED(grp)	\
3487	(test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3488#define EXT4_MB_GRP_SET_TRIMMED(grp)	\
3489	(set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3490#define EXT4_MB_GRP_CLEAR_TRIMMED(grp)	\
3491	(clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3492#define EXT4_MB_GRP_TEST_AND_SET_READ(grp)	\
3493	(test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state)))
3494
3495#define EXT4_MAX_CONTENTION		8
3496#define EXT4_CONTENTION_THRESHOLD	2
3497
3498static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
3499					      ext4_group_t group)
3500{
3501	return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
3502}
3503
3504/*
3505 * Returns true if the filesystem is busy enough that attempts to
3506 * access the block group locks has run into contention.
3507 */
3508static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
3509{
3510	return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
3511}
3512
3513static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
3514{
3515	spinlock_t *lock = ext4_group_lock_ptr(sb, group);
3516	if (spin_trylock(lock))
3517		/*
3518		 * We're able to grab the lock right away, so drop the
3519		 * lock contention counter.
3520		 */
3521		atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
3522	else {
3523		/*
3524		 * The lock is busy, so bump the contention counter,
3525		 * and then wait on the spin lock.
3526		 */
3527		atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
3528				  EXT4_MAX_CONTENTION);
3529		spin_lock(lock);
3530	}
3531}
3532
3533static inline void ext4_unlock_group(struct super_block *sb,
3534					ext4_group_t group)
3535{
3536	spin_unlock(ext4_group_lock_ptr(sb, group));
3537}
3538
3539#ifdef CONFIG_QUOTA
3540static inline bool ext4_quota_capable(struct super_block *sb)
3541{
3542	return (test_opt(sb, QUOTA) || ext4_has_feature_quota(sb));
3543}
3544
3545static inline bool ext4_is_quota_journalled(struct super_block *sb)
3546{
3547	struct ext4_sb_info *sbi = EXT4_SB(sb);
3548
3549	return (ext4_has_feature_quota(sb) ||
3550		sbi->s_qf_names[USRQUOTA] || sbi->s_qf_names[GRPQUOTA]);
3551}
3552int ext4_enable_quotas(struct super_block *sb);
3553#endif
3554
3555/*
3556 * Block validity checking
3557 */
3558#define ext4_check_indirect_blockref(inode, bh)				\
3559	ext4_check_blockref(__func__, __LINE__, inode,			\
3560			    (__le32 *)(bh)->b_data,			\
3561			    EXT4_ADDR_PER_BLOCK((inode)->i_sb))
3562
3563#define ext4_ind_check_inode(inode)					\
3564	ext4_check_blockref(__func__, __LINE__, inode,			\
3565			    EXT4_I(inode)->i_data,			\
3566			    EXT4_NDIR_BLOCKS)
3567
3568/*
3569 * Inodes and files operations
3570 */
3571
3572/* dir.c */
3573extern const struct file_operations ext4_dir_operations;
3574
3575/* file.c */
3576extern const struct inode_operations ext4_file_inode_operations;
3577extern const struct file_operations ext4_file_operations;
3578extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
3579
3580/* inline.c */
3581extern int ext4_get_max_inline_size(struct inode *inode);
3582extern int ext4_find_inline_data_nolock(struct inode *inode);
3583extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
3584				 unsigned int len);
3585extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
3586
3587extern int ext4_readpage_inline(struct inode *inode, struct page *page);
3588extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3589					 struct inode *inode,
3590					 loff_t pos, unsigned len,
3591					 unsigned flags,
3592					 struct page **pagep);
3593extern int ext4_write_inline_data_end(struct inode *inode,
3594				      loff_t pos, unsigned len,
3595				      unsigned copied,
3596				      struct page *page);
3597extern struct buffer_head *
3598ext4_journalled_write_inline_data(struct inode *inode,
3599				  unsigned len,
3600				  struct page *page);
3601extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
3602					   struct inode *inode,
3603					   loff_t pos, unsigned len,
3604					   unsigned flags,
3605					   struct page **pagep,
3606					   void **fsdata);
3607extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
3608					 unsigned len, unsigned copied,
3609					 struct page *page);
3610extern int ext4_try_add_inline_entry(handle_t *handle,
3611				     struct ext4_filename *fname,
3612				     struct inode *dir, struct inode *inode);
3613extern int ext4_try_create_inline_dir(handle_t *handle,
3614				      struct inode *parent,
3615				      struct inode *inode);
3616extern int ext4_read_inline_dir(struct file *filp,
3617				struct dir_context *ctx,
3618				int *has_inline_data);
3619extern int ext4_inlinedir_to_tree(struct file *dir_file,
3620				  struct inode *dir, ext4_lblk_t block,
3621				  struct dx_hash_info *hinfo,
3622				  __u32 start_hash, __u32 start_minor_hash,
3623				  int *has_inline_data);
3624extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3625					struct ext4_filename *fname,
3626					struct ext4_dir_entry_2 **res_dir,
3627					int *has_inline_data);
3628extern int ext4_delete_inline_entry(handle_t *handle,
3629				    struct inode *dir,
3630				    struct ext4_dir_entry_2 *de_del,
3631				    struct buffer_head *bh,
3632				    int *has_inline_data);
3633extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3634extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3635					struct ext4_dir_entry_2 **parent_de,
3636					int *retval);
3637extern int ext4_inline_data_fiemap(struct inode *inode,
3638				   struct fiemap_extent_info *fieinfo,
3639				   int *has_inline, __u64 start, __u64 len);
3640
3641struct iomap;
3642extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap);
3643
3644extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3645
3646extern int ext4_convert_inline_data(struct inode *inode);
3647
3648static inline int ext4_has_inline_data(struct inode *inode)
3649{
3650	return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3651	       EXT4_I(inode)->i_inline_off;
3652}
3653
3654/* namei.c */
3655extern const struct inode_operations ext4_dir_inode_operations;
3656extern const struct inode_operations ext4_special_inode_operations;
3657extern struct dentry *ext4_get_parent(struct dentry *child);
3658extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3659				 struct ext4_dir_entry_2 *de,
3660				 int blocksize, int csum_size,
3661				 unsigned int parent_ino, int dotdot_real_len);
3662extern void ext4_initialize_dirent_tail(struct buffer_head *bh,
3663					unsigned int blocksize);
3664extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode,
3665				      struct buffer_head *bh);
3666extern int __ext4_unlink(handle_t *handle, struct inode *dir, const struct qstr *d_name,
3667			 struct inode *inode);
3668extern int __ext4_link(struct inode *dir, struct inode *inode,
3669		       struct dentry *dentry);
3670
3671#define S_SHIFT 12
3672static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = {
3673	[S_IFREG >> S_SHIFT]	= EXT4_FT_REG_FILE,
3674	[S_IFDIR >> S_SHIFT]	= EXT4_FT_DIR,
3675	[S_IFCHR >> S_SHIFT]	= EXT4_FT_CHRDEV,
3676	[S_IFBLK >> S_SHIFT]	= EXT4_FT_BLKDEV,
3677	[S_IFIFO >> S_SHIFT]	= EXT4_FT_FIFO,
3678	[S_IFSOCK >> S_SHIFT]	= EXT4_FT_SOCK,
3679	[S_IFLNK >> S_SHIFT]	= EXT4_FT_SYMLINK,
3680};
3681
3682static inline void ext4_set_de_type(struct super_block *sb,
3683				struct ext4_dir_entry_2 *de,
3684				umode_t mode) {
3685	if (ext4_has_feature_filetype(sb))
3686		de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3687}
3688
3689/* readpages.c */
3690extern int ext4_mpage_readpages(struct inode *inode,
3691		struct readahead_control *rac, struct page *page);
3692extern int __init ext4_init_post_read_processing(void);
3693extern void ext4_exit_post_read_processing(void);
3694
3695/* symlink.c */
3696extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3697extern const struct inode_operations ext4_symlink_inode_operations;
3698extern const struct inode_operations ext4_fast_symlink_inode_operations;
3699
3700/* sysfs.c */
3701extern void ext4_notify_error_sysfs(struct ext4_sb_info *sbi);
3702extern int ext4_register_sysfs(struct super_block *sb);
3703extern void ext4_unregister_sysfs(struct super_block *sb);
3704extern int __init ext4_init_sysfs(void);
3705extern void ext4_exit_sysfs(void);
3706
3707/* block_validity */
3708extern void ext4_release_system_zone(struct super_block *sb);
3709extern int ext4_setup_system_zone(struct super_block *sb);
3710extern int __init ext4_init_system_zone(void);
3711extern void ext4_exit_system_zone(void);
3712extern int ext4_inode_block_valid(struct inode *inode,
3713				  ext4_fsblk_t start_blk,
3714				  unsigned int count);
3715extern int ext4_check_blockref(const char *, unsigned int,
3716			       struct inode *, __le32 *, unsigned int);
3717
3718/* extents.c */
3719struct ext4_ext_path;
3720struct ext4_extent;
3721
3722/*
3723 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3724 * __le32.
3725 */
3726#define EXT_MAX_BLOCKS	0xffffffff
3727
3728extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode);
3729extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3730extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3731			       struct ext4_map_blocks *map, int flags);
3732extern int ext4_ext_truncate(handle_t *, struct inode *);
3733extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3734				 ext4_lblk_t end);
3735extern void ext4_ext_init(struct super_block *);
3736extern void ext4_ext_release(struct super_block *);
3737extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3738			  loff_t len);
3739extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3740					  loff_t offset, ssize_t len);
3741extern int ext4_convert_unwritten_io_end_vec(handle_t *handle,
3742					     ext4_io_end_t *io_end);
3743extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3744			   struct ext4_map_blocks *map, int flags);
3745extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3746						   int num,
3747						   struct ext4_ext_path *path);
3748extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3749				  struct ext4_ext_path **,
3750				  struct ext4_extent *, int);
3751extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3752					      struct ext4_ext_path **,
3753					      int flags);
3754extern void ext4_ext_drop_refs(struct ext4_ext_path *);
3755extern int ext4_ext_check_inode(struct inode *inode);
3756extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3757extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3758			__u64 start, __u64 len);
3759extern int ext4_get_es_cache(struct inode *inode,
3760			     struct fiemap_extent_info *fieinfo,
3761			     __u64 start, __u64 len);
3762extern int ext4_ext_precache(struct inode *inode);
3763extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3764				struct inode *inode2, ext4_lblk_t lblk1,
3765			     ext4_lblk_t lblk2,  ext4_lblk_t count,
3766			     int mark_unwritten,int *err);
3767extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu);
3768extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode,
3769				       int check_cred, int restart_cred,
3770				       int revoke_cred);
3771extern void ext4_ext_replay_shrink_inode(struct inode *inode, ext4_lblk_t end);
3772extern int ext4_ext_replay_set_iblocks(struct inode *inode);
3773extern int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start,
3774		int len, int unwritten, ext4_fsblk_t pblk);
3775extern int ext4_ext_clear_bb(struct inode *inode);
3776
3777
3778/* move_extent.c */
3779extern void ext4_double_down_write_data_sem(struct inode *first,
3780					    struct inode *second);
3781extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3782					  struct inode *donor_inode);
3783extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3784			     __u64 start_orig, __u64 start_donor,
3785			     __u64 len, __u64 *moved_len);
3786
3787/* page-io.c */
3788extern int __init ext4_init_pageio(void);
3789extern void ext4_exit_pageio(void);
3790extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3791extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3792extern int ext4_put_io_end(ext4_io_end_t *io_end);
3793extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3794extern void ext4_io_submit_init(struct ext4_io_submit *io,
3795				struct writeback_control *wbc);
3796extern void ext4_end_io_rsv_work(struct work_struct *work);
3797extern void ext4_io_submit(struct ext4_io_submit *io);
3798extern int ext4_bio_write_page(struct ext4_io_submit *io,
3799			       struct page *page,
3800			       int len,
3801			       bool keep_towrite);
3802extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end);
3803extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end);
3804
3805/* mmp.c */
3806extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3807
3808/* mmp.c */
3809extern void ext4_stop_mmpd(struct ext4_sb_info *sbi);
3810
3811/* verity.c */
3812extern const struct fsverity_operations ext4_verityops;
3813
3814/* orphan.c */
3815extern int ext4_orphan_add(handle_t *, struct inode *);
3816extern int ext4_orphan_del(handle_t *, struct inode *);
3817extern void ext4_orphan_cleanup(struct super_block *sb,
3818				struct ext4_super_block *es);
3819extern void ext4_release_orphan_info(struct super_block *sb);
3820extern int ext4_init_orphan_info(struct super_block *sb);
3821extern int ext4_orphan_file_empty(struct super_block *sb);
3822extern void ext4_orphan_file_block_trigger(
3823				struct jbd2_buffer_trigger_type *triggers,
3824				struct buffer_head *bh,
3825				void *data, size_t size);
3826
3827/*
3828 * Add new method to test whether block and inode bitmaps are properly
3829 * initialized. With uninit_bg reading the block from disk is not enough
3830 * to mark the bitmap uptodate. We need to also zero-out the bitmap
3831 */
3832#define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3833
3834static inline int bitmap_uptodate(struct buffer_head *bh)
3835{
3836	return (buffer_uptodate(bh) &&
3837			test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3838}
3839static inline void set_bitmap_uptodate(struct buffer_head *bh)
3840{
3841	set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3842}
3843
3844#define in_range(b, first, len)	((b) >= (first) && (b) <= (first) + (len) - 1)
3845
3846/* For ioend & aio unwritten conversion wait queues */
3847#define EXT4_WQ_HASH_SZ		37
3848#define ext4_ioend_wq(v)   (&ext4__ioend_wq[((unsigned long)(v)) %\
3849					    EXT4_WQ_HASH_SZ])
3850extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3851
3852extern int ext4_resize_begin(struct super_block *sb);
3853extern void ext4_resize_end(struct super_block *sb);
3854
3855static inline void ext4_set_io_unwritten_flag(struct inode *inode,
3856					      struct ext4_io_end *io_end)
3857{
3858	if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
3859		io_end->flag |= EXT4_IO_END_UNWRITTEN;
3860		atomic_inc(&EXT4_I(inode)->i_unwritten);
3861	}
3862}
3863
3864static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3865{
3866	struct inode *inode = io_end->inode;
3867
3868	if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
3869		io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3870		/* Wake up anyone waiting on unwritten extent conversion */
3871		if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
3872			wake_up_all(ext4_ioend_wq(inode));
3873	}
3874}
3875
3876extern const struct iomap_ops ext4_iomap_ops;
3877extern const struct iomap_ops ext4_iomap_overwrite_ops;
3878extern const struct iomap_ops ext4_iomap_report_ops;
3879
3880static inline int ext4_buffer_uptodate(struct buffer_head *bh)
3881{
3882	/*
3883	 * If the buffer has the write error flag, we have failed
3884	 * to write out data in the block.  In this  case, we don't
3885	 * have to read the block because we may read the old data
3886	 * successfully.
3887	 */
3888	if (buffer_write_io_error(bh))
3889		set_buffer_uptodate(bh);
3890	return buffer_uptodate(bh);
3891}
3892
3893#endif	/* __KERNEL__ */
3894
3895#define EFSBADCRC	EBADMSG		/* Bad CRC detected */
3896#define EFSCORRUPTED	EUCLEAN		/* Filesystem is corrupted */
3897
3898#endif	/* _EXT4_H */
3899