Linux kernel mirror (for testing)
git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel
os
linux
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/refcount.h>
21#include <linux/types.h>
22#include <linux/blkdev.h>
23#include <linux/magic.h>
24#include <linux/jbd2.h>
25#include <linux/quota.h>
26#include <linux/rwsem.h>
27#include <linux/rbtree.h>
28#include <linux/seqlock.h>
29#include <linux/mutex.h>
30#include <linux/timer.h>
31#include <linux/wait.h>
32#include <linux/sched/signal.h>
33#include <linux/blockgroup_lock.h>
34#include <linux/percpu_counter.h>
35#include <linux/ratelimit.h>
36#include <linux/crc32c.h>
37#include <linux/falloc.h>
38#include <linux/percpu-rwsem.h>
39#include <linux/fiemap.h>
40#ifdef __KERNEL__
41#include <linux/compat.h>
42#endif
43#include <uapi/linux/ext4.h>
44
45#include <linux/fscrypt.h>
46#include <linux/fsverity.h>
47
48#include <linux/compiler.h>
49
50/*
51 * The fourth extended filesystem constants/structures
52 */
53
54/*
55 * with AGGRESSIVE_CHECK allocator runs consistency checks over
56 * structures. these checks slow things down a lot
57 */
58#define AGGRESSIVE_CHECK__
59
60/*
61 * with DOUBLE_CHECK defined mballoc creates persistent in-core
62 * bitmaps, maintains and uses them to check for double allocations
63 */
64#define DOUBLE_CHECK__
65
66/*
67 * Define EXT4FS_DEBUG to produce debug messages
68 */
69#undef EXT4FS_DEBUG
70
71/*
72 * Debug code
73 */
74#ifdef EXT4FS_DEBUG
75#define ext4_debug(f, a...) \
76 do { \
77 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
78 __FILE__, __LINE__, __func__); \
79 printk(KERN_DEBUG f, ## a); \
80 } while (0)
81#else
82#define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
83#endif
84
85 /*
86 * Turn on EXT_DEBUG to enable ext4_ext_show_path/leaf/move in extents.c
87 */
88#define EXT_DEBUG__
89
90/*
91 * Dynamic printk for controlled extents debugging.
92 */
93#ifdef CONFIG_EXT4_DEBUG
94#define ext_debug(ino, fmt, ...) \
95 pr_debug("[%s/%d] EXT4-fs (%s): ino %lu: (%s, %d): %s:" fmt, \
96 current->comm, task_pid_nr(current), \
97 ino->i_sb->s_id, ino->i_ino, __FILE__, __LINE__, \
98 __func__, ##__VA_ARGS__)
99#else
100#define ext_debug(ino, fmt, ...) no_printk(fmt, ##__VA_ARGS__)
101#endif
102
103#define ASSERT(assert) \
104do { \
105 if (unlikely(!(assert))) { \
106 printk(KERN_EMERG \
107 "Assertion failure in %s() at %s:%d: '%s'\n", \
108 __func__, __FILE__, __LINE__, #assert); \
109 BUG(); \
110 } \
111} while (0)
112
113/* data type for block offset of block group */
114typedef int ext4_grpblk_t;
115
116/* data type for filesystem-wide blocks number */
117typedef unsigned long long ext4_fsblk_t;
118
119/* data type for file logical block number */
120typedef __u32 ext4_lblk_t;
121
122/* data type for block group number */
123typedef unsigned int ext4_group_t;
124
125enum SHIFT_DIRECTION {
126 SHIFT_LEFT = 0,
127 SHIFT_RIGHT,
128};
129
130/*
131 * For each criteria, mballoc has slightly different way of finding
132 * the required blocks nad usually, higher the criteria the slower the
133 * allocation. We start at lower criterias and keep falling back to
134 * higher ones if we are not able to find any blocks. Lower (earlier)
135 * criteria are faster.
136 */
137enum criteria {
138 /*
139 * Used when number of blocks needed is a power of 2. This
140 * doesn't trigger any disk IO except prefetch and is the
141 * fastest criteria.
142 */
143 CR_POWER2_ALIGNED,
144
145 /*
146 * Tries to lookup in-memory data structures to find the most
147 * suitable group that satisfies goal request. No disk IO
148 * except block prefetch.
149 */
150 CR_GOAL_LEN_FAST,
151
152 /*
153 * Same as CR_GOAL_LEN_FAST but is allowed to reduce the goal
154 * length to the best available length for faster allocation.
155 */
156 CR_BEST_AVAIL_LEN,
157
158 /*
159 * Reads each block group sequentially, performing disk IO if
160 * necessary, to find find_suitable block group. Tries to
161 * allocate goal length but might trim the request if nothing
162 * is found after enough tries.
163 */
164 CR_GOAL_LEN_SLOW,
165
166 /*
167 * Finds the first free set of blocks and allocates
168 * those. This is only used in rare cases when
169 * CR_GOAL_LEN_SLOW also fails to allocate anything.
170 */
171 CR_ANY_FREE,
172
173 /*
174 * Number of criterias defined.
175 */
176 EXT4_MB_NUM_CRS
177};
178
179/*
180 * Flags used in mballoc's allocation_context flags field.
181 *
182 * Also used to show what's going on for debugging purposes when the
183 * flag field is exported via the traceport interface
184 */
185
186/* prefer goal again. length */
187#define EXT4_MB_HINT_MERGE 0x0001
188/* blocks already reserved */
189#define EXT4_MB_HINT_RESERVED 0x0002
190/* metadata is being allocated */
191#define EXT4_MB_HINT_METADATA 0x0004
192/* first blocks in the file */
193#define EXT4_MB_HINT_FIRST 0x0008
194/* search for the best chunk */
195#define EXT4_MB_HINT_BEST 0x0010
196/* data is being allocated */
197#define EXT4_MB_HINT_DATA 0x0020
198/* don't preallocate (for tails) */
199#define EXT4_MB_HINT_NOPREALLOC 0x0040
200/* allocate for locality group */
201#define EXT4_MB_HINT_GROUP_ALLOC 0x0080
202/* allocate goal blocks or none */
203#define EXT4_MB_HINT_GOAL_ONLY 0x0100
204/* goal is meaningful */
205#define EXT4_MB_HINT_TRY_GOAL 0x0200
206/* blocks already pre-reserved by delayed allocation */
207#define EXT4_MB_DELALLOC_RESERVED 0x0400
208/* We are doing stream allocation */
209#define EXT4_MB_STREAM_ALLOC 0x0800
210/* Use reserved root blocks if needed */
211#define EXT4_MB_USE_ROOT_BLOCKS 0x1000
212/* Use blocks from reserved pool */
213#define EXT4_MB_USE_RESERVED 0x2000
214/* Do strict check for free blocks while retrying block allocation */
215#define EXT4_MB_STRICT_CHECK 0x4000
216/* Large fragment size list lookup succeeded at least once for
217 * CR_POWER2_ALIGNED */
218#define EXT4_MB_CR_POWER2_ALIGNED_OPTIMIZED 0x8000
219/* Avg fragment size rb tree lookup succeeded at least once for
220 * CR_GOAL_LEN_FAST */
221#define EXT4_MB_CR_GOAL_LEN_FAST_OPTIMIZED 0x00010000
222/* Avg fragment size rb tree lookup succeeded at least once for
223 * CR_BEST_AVAIL_LEN */
224#define EXT4_MB_CR_BEST_AVAIL_LEN_OPTIMIZED 0x00020000
225
226struct ext4_allocation_request {
227 /* target inode for block we're allocating */
228 struct inode *inode;
229 /* how many blocks we want to allocate */
230 unsigned int len;
231 /* logical block in target inode */
232 ext4_lblk_t logical;
233 /* the closest logical allocated block to the left */
234 ext4_lblk_t lleft;
235 /* the closest logical allocated block to the right */
236 ext4_lblk_t lright;
237 /* phys. target (a hint) */
238 ext4_fsblk_t goal;
239 /* phys. block for the closest logical allocated block to the left */
240 ext4_fsblk_t pleft;
241 /* phys. block for the closest logical allocated block to the right */
242 ext4_fsblk_t pright;
243 /* flags. see above EXT4_MB_HINT_* */
244 unsigned int flags;
245};
246
247/*
248 * Logical to physical block mapping, used by ext4_map_blocks()
249 *
250 * This structure is used to pass requests into ext4_map_blocks() as
251 * well as to store the information returned by ext4_map_blocks(). It
252 * takes less room on the stack than a struct buffer_head.
253 */
254#define EXT4_MAP_NEW BIT(BH_New)
255#define EXT4_MAP_MAPPED BIT(BH_Mapped)
256#define EXT4_MAP_UNWRITTEN BIT(BH_Unwritten)
257#define EXT4_MAP_BOUNDARY BIT(BH_Boundary)
258#define EXT4_MAP_DELAYED BIT(BH_Delay)
259#define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
260 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\
261 EXT4_MAP_DELAYED)
262
263struct ext4_map_blocks {
264 ext4_fsblk_t m_pblk;
265 ext4_lblk_t m_lblk;
266 unsigned int m_len;
267 unsigned int m_flags;
268};
269
270/*
271 * Block validity checking, system zone rbtree.
272 */
273struct ext4_system_blocks {
274 struct rb_root root;
275 struct rcu_head rcu;
276};
277
278/*
279 * Flags for ext4_io_end->flags
280 */
281#define EXT4_IO_END_UNWRITTEN 0x0001
282
283struct ext4_io_end_vec {
284 struct list_head list; /* list of io_end_vec */
285 loff_t offset; /* offset in the file */
286 ssize_t size; /* size of the extent */
287};
288
289/*
290 * For converting unwritten extents on a work queue. 'handle' is used for
291 * buffered writeback.
292 */
293typedef struct ext4_io_end {
294 struct list_head list; /* per-file finished IO list */
295 handle_t *handle; /* handle reserved for extent
296 * conversion */
297 struct inode *inode; /* file being written to */
298 struct bio *bio; /* Linked list of completed
299 * bios covering the extent */
300 unsigned int flag; /* unwritten or not */
301 refcount_t count; /* reference counter */
302 struct list_head list_vec; /* list of ext4_io_end_vec */
303} ext4_io_end_t;
304
305struct ext4_io_submit {
306 struct writeback_control *io_wbc;
307 struct bio *io_bio;
308 ext4_io_end_t *io_end;
309 sector_t io_next_block;
310};
311
312/*
313 * Special inodes numbers
314 */
315#define EXT4_BAD_INO 1 /* Bad blocks inode */
316#define EXT4_ROOT_INO 2 /* Root inode */
317#define EXT4_USR_QUOTA_INO 3 /* User quota inode */
318#define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
319#define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
320#define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
321#define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
322#define EXT4_JOURNAL_INO 8 /* Journal inode */
323
324/* First non-reserved inode for old ext4 filesystems */
325#define EXT4_GOOD_OLD_FIRST_INO 11
326
327/*
328 * Maximal count of links to a file
329 */
330#define EXT4_LINK_MAX 65000
331
332/*
333 * Macro-instructions used to manage several block sizes
334 */
335#define EXT4_MIN_BLOCK_SIZE 1024
336#define EXT4_MAX_BLOCK_SIZE 65536
337#define EXT4_MIN_BLOCK_LOG_SIZE 10
338#define EXT4_MAX_BLOCK_LOG_SIZE 16
339#define EXT4_MAX_CLUSTER_LOG_SIZE 30
340#ifdef __KERNEL__
341# define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
342#else
343# define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
344#endif
345#define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
346#define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
347 EXT4_SB(s)->s_cluster_bits)
348#ifdef __KERNEL__
349# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
350# define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
351#else
352# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
353#endif
354#ifdef __KERNEL__
355#define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
356#define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
357#define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
358#else
359#define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
360 EXT4_GOOD_OLD_INODE_SIZE : \
361 (s)->s_inode_size)
362#define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
363 EXT4_GOOD_OLD_FIRST_INO : \
364 (s)->s_first_ino)
365#endif
366#define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
367#define EXT4_MAX_BLOCKS(size, offset, blkbits) \
368 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \
369 blkbits))
370
371/* Translate a block number to a cluster number */
372#define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
373/* Translate a cluster number to a block number */
374#define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
375/* Translate # of blks to # of clusters */
376#define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
377 (sbi)->s_cluster_bits)
378/* Mask out the low bits to get the starting block of the cluster */
379#define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
380 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
381#define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
382 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
383/* Fill in the low bits to get the last block of the cluster */
384#define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) | \
385 ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1))
386/* Get the cluster offset */
387#define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
388 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
389#define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
390 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
391
392/*
393 * Structure of a blocks group descriptor
394 */
395struct ext4_group_desc
396{
397 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
398 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
399 __le32 bg_inode_table_lo; /* Inodes table block */
400 __le16 bg_free_blocks_count_lo;/* Free blocks count */
401 __le16 bg_free_inodes_count_lo;/* Free inodes count */
402 __le16 bg_used_dirs_count_lo; /* Directories count */
403 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
404 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
405 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
406 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
407 __le16 bg_itable_unused_lo; /* Unused inodes count */
408 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
409 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
410 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
411 __le32 bg_inode_table_hi; /* Inodes table block MSB */
412 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
413 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
414 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
415 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
416 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
417 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
418 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
419 __u32 bg_reserved;
420};
421
422#define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
423 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
424 sizeof(__le16))
425#define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
426 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
427 sizeof(__le16))
428
429/*
430 * Structure of a flex block group info
431 */
432
433struct flex_groups {
434 atomic64_t free_clusters;
435 atomic_t free_inodes;
436 atomic_t used_dirs;
437};
438
439#define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
440#define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
441#define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
442
443/*
444 * Macro-instructions used to manage group descriptors
445 */
446#define EXT4_MIN_DESC_SIZE 32
447#define EXT4_MIN_DESC_SIZE_64BIT 64
448#define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
449#define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
450#ifdef __KERNEL__
451# define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
452# define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
453# define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
454# define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
455# define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
456#else
457# define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
458# define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
459# define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
460#endif
461
462/*
463 * Constants relative to the data blocks
464 */
465#define EXT4_NDIR_BLOCKS 12
466#define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
467#define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
468#define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
469#define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
470
471/*
472 * Inode flags
473 */
474#define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
475#define EXT4_UNRM_FL 0x00000002 /* Undelete */
476#define EXT4_COMPR_FL 0x00000004 /* Compress file */
477#define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
478#define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
479#define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
480#define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
481#define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
482/* Reserved for compression usage... */
483#define EXT4_DIRTY_FL 0x00000100
484#define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
485#define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
486 /* nb: was previously EXT2_ECOMPR_FL */
487#define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */
488/* End compression flags --- maybe not all used */
489#define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
490#define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
491#define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
492#define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
493#define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
494#define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
495#define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
496#define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
497#define EXT4_VERITY_FL 0x00100000 /* Verity protected inode */
498#define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
499/* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */
500
501#define EXT4_DAX_FL 0x02000000 /* Inode is DAX */
502
503#define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
504#define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
505#define EXT4_CASEFOLD_FL 0x40000000 /* Casefolded directory */
506#define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
507
508/* User modifiable flags */
509#define EXT4_FL_USER_MODIFIABLE (EXT4_SECRM_FL | \
510 EXT4_UNRM_FL | \
511 EXT4_COMPR_FL | \
512 EXT4_SYNC_FL | \
513 EXT4_IMMUTABLE_FL | \
514 EXT4_APPEND_FL | \
515 EXT4_NODUMP_FL | \
516 EXT4_NOATIME_FL | \
517 EXT4_JOURNAL_DATA_FL | \
518 EXT4_NOTAIL_FL | \
519 EXT4_DIRSYNC_FL | \
520 EXT4_TOPDIR_FL | \
521 EXT4_EXTENTS_FL | \
522 0x00400000 /* EXT4_EOFBLOCKS_FL */ | \
523 EXT4_DAX_FL | \
524 EXT4_PROJINHERIT_FL | \
525 EXT4_CASEFOLD_FL)
526
527/* User visible flags */
528#define EXT4_FL_USER_VISIBLE (EXT4_FL_USER_MODIFIABLE | \
529 EXT4_DIRTY_FL | \
530 EXT4_COMPRBLK_FL | \
531 EXT4_NOCOMPR_FL | \
532 EXT4_ENCRYPT_FL | \
533 EXT4_INDEX_FL | \
534 EXT4_VERITY_FL | \
535 EXT4_INLINE_DATA_FL)
536
537/* Flags that should be inherited by new inodes from their parent. */
538#define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
539 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
540 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
541 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
542 EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL |\
543 EXT4_DAX_FL)
544
545/* Flags that are appropriate for regular files (all but dir-specific ones). */
546#define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\
547 EXT4_PROJINHERIT_FL))
548
549/* Flags that are appropriate for non-directories/regular files. */
550#define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
551
552/* The only flags that should be swapped */
553#define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL)
554
555/* Flags which are mutually exclusive to DAX */
556#define EXT4_DAX_MUT_EXCL (EXT4_VERITY_FL | EXT4_ENCRYPT_FL |\
557 EXT4_JOURNAL_DATA_FL | EXT4_INLINE_DATA_FL)
558
559/* Mask out flags that are inappropriate for the given type of inode. */
560static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
561{
562 if (S_ISDIR(mode))
563 return flags;
564 else if (S_ISREG(mode))
565 return flags & EXT4_REG_FLMASK;
566 else
567 return flags & EXT4_OTHER_FLMASK;
568}
569
570/*
571 * Inode flags used for atomic set/get
572 */
573enum {
574 EXT4_INODE_SECRM = 0, /* Secure deletion */
575 EXT4_INODE_UNRM = 1, /* Undelete */
576 EXT4_INODE_COMPR = 2, /* Compress file */
577 EXT4_INODE_SYNC = 3, /* Synchronous updates */
578 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
579 EXT4_INODE_APPEND = 5, /* writes to file may only append */
580 EXT4_INODE_NODUMP = 6, /* do not dump file */
581 EXT4_INODE_NOATIME = 7, /* do not update atime */
582/* Reserved for compression usage... */
583 EXT4_INODE_DIRTY = 8,
584 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
585 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
586 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */
587/* End compression flags --- maybe not all used */
588 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
589 EXT4_INODE_IMAGIC = 13, /* AFS directory */
590 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
591 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
592 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
593 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
594 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
595 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
596 EXT4_INODE_VERITY = 20, /* Verity protected inode */
597 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
598/* 22 was formerly EXT4_INODE_EOFBLOCKS */
599 EXT4_INODE_DAX = 25, /* Inode is DAX */
600 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
601 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */
602 EXT4_INODE_CASEFOLD = 30, /* Casefolded directory */
603 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
604};
605
606/*
607 * Since it's pretty easy to mix up bit numbers and hex values, we use a
608 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
609 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
610 * any extra space in the compiled kernel image, otherwise, the build will fail.
611 * It's important that these values are the same, since we are using
612 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
613 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
614 * values found in ext2, ext3 and ext4 filesystems, and of course the values
615 * defined in e2fsprogs.
616 *
617 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
618 */
619#define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1U << EXT4_INODE_##FLAG))
620#define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
621
622static inline void ext4_check_flag_values(void)
623{
624 CHECK_FLAG_VALUE(SECRM);
625 CHECK_FLAG_VALUE(UNRM);
626 CHECK_FLAG_VALUE(COMPR);
627 CHECK_FLAG_VALUE(SYNC);
628 CHECK_FLAG_VALUE(IMMUTABLE);
629 CHECK_FLAG_VALUE(APPEND);
630 CHECK_FLAG_VALUE(NODUMP);
631 CHECK_FLAG_VALUE(NOATIME);
632 CHECK_FLAG_VALUE(DIRTY);
633 CHECK_FLAG_VALUE(COMPRBLK);
634 CHECK_FLAG_VALUE(NOCOMPR);
635 CHECK_FLAG_VALUE(ENCRYPT);
636 CHECK_FLAG_VALUE(INDEX);
637 CHECK_FLAG_VALUE(IMAGIC);
638 CHECK_FLAG_VALUE(JOURNAL_DATA);
639 CHECK_FLAG_VALUE(NOTAIL);
640 CHECK_FLAG_VALUE(DIRSYNC);
641 CHECK_FLAG_VALUE(TOPDIR);
642 CHECK_FLAG_VALUE(HUGE_FILE);
643 CHECK_FLAG_VALUE(EXTENTS);
644 CHECK_FLAG_VALUE(VERITY);
645 CHECK_FLAG_VALUE(EA_INODE);
646 CHECK_FLAG_VALUE(INLINE_DATA);
647 CHECK_FLAG_VALUE(PROJINHERIT);
648 CHECK_FLAG_VALUE(CASEFOLD);
649 CHECK_FLAG_VALUE(RESERVED);
650}
651
652#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
653struct compat_ext4_new_group_input {
654 u32 group;
655 compat_u64 block_bitmap;
656 compat_u64 inode_bitmap;
657 compat_u64 inode_table;
658 u32 blocks_count;
659 u16 reserved_blocks;
660 u16 unused;
661};
662#endif
663
664/* The struct ext4_new_group_input in kernel space, with free_blocks_count */
665struct ext4_new_group_data {
666 __u32 group;
667 __u64 block_bitmap;
668 __u64 inode_bitmap;
669 __u64 inode_table;
670 __u32 blocks_count;
671 __u16 reserved_blocks;
672 __u16 mdata_blocks;
673 __u32 free_clusters_count;
674};
675
676/* Indexes used to index group tables in ext4_new_group_data */
677enum {
678 BLOCK_BITMAP = 0, /* block bitmap */
679 INODE_BITMAP, /* inode bitmap */
680 INODE_TABLE, /* inode tables */
681 GROUP_TABLE_COUNT,
682};
683
684/*
685 * Flags used by ext4_map_blocks()
686 */
687 /* Allocate any needed blocks and/or convert an unwritten
688 extent to be an initialized ext4 */
689#define EXT4_GET_BLOCKS_CREATE 0x0001
690 /* Request the creation of an unwritten extent */
691#define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002
692#define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\
693 EXT4_GET_BLOCKS_CREATE)
694 /* Caller is from the delayed allocation writeout path
695 * finally doing the actual allocation of delayed blocks */
696#define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
697 /* caller is from the direct IO path, request to creation of an
698 unwritten extents if not allocated, split the unwritten
699 extent if blocks has been preallocated already*/
700#define EXT4_GET_BLOCKS_PRE_IO 0x0008
701#define EXT4_GET_BLOCKS_CONVERT 0x0010
702#define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
703 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
704 /* Convert extent to initialized after IO complete */
705#define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
706 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
707 /* Eventual metadata allocation (due to growing extent tree)
708 * should not fail, so try to use reserved blocks for that.*/
709#define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
710 /* Don't normalize allocation size (used for fallocate) */
711#define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
712 /* Convert written extents to unwritten */
713#define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100
714 /* Write zeros to newly created written extents */
715#define EXT4_GET_BLOCKS_ZERO 0x0200
716#define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\
717 EXT4_GET_BLOCKS_ZERO)
718 /* Caller will submit data before dropping transaction handle. This
719 * allows jbd2 to avoid submitting data before commit. */
720#define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400
721 /* Caller is in the atomic contex, find extent if it has been cached */
722#define EXT4_GET_BLOCKS_CACHED_NOWAIT 0x0800
723
724/*
725 * The bit position of these flags must not overlap with any of the
726 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(),
727 * read_extent_tree_block(), ext4_split_extent_at(),
728 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
729 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
730 * caching the extents when reading from the extent tree while a
731 * truncate or punch hole operation is in progress.
732 */
733#define EXT4_EX_NOCACHE 0x40000000
734#define EXT4_EX_FORCE_CACHE 0x20000000
735#define EXT4_EX_NOFAIL 0x10000000
736
737/*
738 * Flags used by ext4_free_blocks
739 */
740#define EXT4_FREE_BLOCKS_METADATA 0x0001
741#define EXT4_FREE_BLOCKS_FORGET 0x0002
742#define EXT4_FREE_BLOCKS_VALIDATED 0x0004
743#define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
744#define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
745#define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
746#define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040
747
748#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
749/*
750 * ioctl commands in 32 bit emulation
751 */
752#define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
753#define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
754#define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
755#define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
756#define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
757#define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
758#define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
759#define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
760#endif
761
762/* Max physical block we can address w/o extents */
763#define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
764
765/* Max logical block we can support */
766#define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFE
767
768/*
769 * Structure of an inode on the disk
770 */
771struct ext4_inode {
772 __le16 i_mode; /* File mode */
773 __le16 i_uid; /* Low 16 bits of Owner Uid */
774 __le32 i_size_lo; /* Size in bytes */
775 __le32 i_atime; /* Access time */
776 __le32 i_ctime; /* Inode Change time */
777 __le32 i_mtime; /* Modification time */
778 __le32 i_dtime; /* Deletion Time */
779 __le16 i_gid; /* Low 16 bits of Group Id */
780 __le16 i_links_count; /* Links count */
781 __le32 i_blocks_lo; /* Blocks count */
782 __le32 i_flags; /* File flags */
783 union {
784 struct {
785 __le32 l_i_version;
786 } linux1;
787 struct {
788 __u32 h_i_translator;
789 } hurd1;
790 struct {
791 __u32 m_i_reserved1;
792 } masix1;
793 } osd1; /* OS dependent 1 */
794 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
795 __le32 i_generation; /* File version (for NFS) */
796 __le32 i_file_acl_lo; /* File ACL */
797 __le32 i_size_high;
798 __le32 i_obso_faddr; /* Obsoleted fragment address */
799 union {
800 struct {
801 __le16 l_i_blocks_high; /* were l_i_reserved1 */
802 __le16 l_i_file_acl_high;
803 __le16 l_i_uid_high; /* these 2 fields */
804 __le16 l_i_gid_high; /* were reserved2[0] */
805 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
806 __le16 l_i_reserved;
807 } linux2;
808 struct {
809 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
810 __u16 h_i_mode_high;
811 __u16 h_i_uid_high;
812 __u16 h_i_gid_high;
813 __u32 h_i_author;
814 } hurd2;
815 struct {
816 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
817 __le16 m_i_file_acl_high;
818 __u32 m_i_reserved2[2];
819 } masix2;
820 } osd2; /* OS dependent 2 */
821 __le16 i_extra_isize;
822 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
823 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
824 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
825 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
826 __le32 i_crtime; /* File Creation time */
827 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
828 __le32 i_version_hi; /* high 32 bits for 64-bit version */
829 __le32 i_projid; /* Project ID */
830};
831
832#define EXT4_EPOCH_BITS 2
833#define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
834#define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
835
836/*
837 * Extended fields will fit into an inode if the filesystem was formatted
838 * with large inodes (-I 256 or larger) and there are not currently any EAs
839 * consuming all of the available space. For new inodes we always reserve
840 * enough space for the kernel's known extended fields, but for inodes
841 * created with an old kernel this might not have been the case. None of
842 * the extended inode fields is critical for correct filesystem operation.
843 * This macro checks if a certain field fits in the inode. Note that
844 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
845 */
846#define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
847 ((offsetof(typeof(*ext4_inode), field) + \
848 sizeof((ext4_inode)->field)) \
849 <= (EXT4_GOOD_OLD_INODE_SIZE + \
850 (einode)->i_extra_isize)) \
851
852/*
853 * We use an encoding that preserves the times for extra epoch "00":
854 *
855 * extra msb of adjust for signed
856 * epoch 32-bit 32-bit tv_sec to
857 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
858 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
859 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
860 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
861 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
862 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
863 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
864 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
865 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
866 *
867 * Note that previous versions of the kernel on 64-bit systems would
868 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
869 * 1970. e2fsck will correct this, assuming that it is run on the
870 * affected filesystem before 2242.
871 */
872
873static inline __le32 ext4_encode_extra_time(struct timespec64 ts)
874{
875 u32 extra = ((ts.tv_sec - (s32)ts.tv_sec) >> 32) & EXT4_EPOCH_MASK;
876 return cpu_to_le32(extra | (ts.tv_nsec << EXT4_EPOCH_BITS));
877}
878
879static inline struct timespec64 ext4_decode_extra_time(__le32 base,
880 __le32 extra)
881{
882 struct timespec64 ts = { .tv_sec = (signed)le32_to_cpu(base) };
883
884 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK)))
885 ts.tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
886 ts.tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
887 return ts;
888}
889
890#define EXT4_INODE_SET_XTIME_VAL(xtime, inode, raw_inode, ts) \
891do { \
892 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \
893 (raw_inode)->xtime = cpu_to_le32((ts).tv_sec); \
894 (raw_inode)->xtime ## _extra = ext4_encode_extra_time(ts); \
895 } else \
896 (raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (ts).tv_sec, S32_MIN, S32_MAX)); \
897} while (0)
898
899#define EXT4_INODE_SET_ATIME(inode, raw_inode) \
900 EXT4_INODE_SET_XTIME_VAL(i_atime, inode, raw_inode, inode_get_atime(inode))
901
902#define EXT4_INODE_SET_MTIME(inode, raw_inode) \
903 EXT4_INODE_SET_XTIME_VAL(i_mtime, inode, raw_inode, inode_get_mtime(inode))
904
905#define EXT4_INODE_SET_CTIME(inode, raw_inode) \
906 EXT4_INODE_SET_XTIME_VAL(i_ctime, inode, raw_inode, inode_get_ctime(inode))
907
908#define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
909 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
910 EXT4_INODE_SET_XTIME_VAL(xtime, &((einode)->vfs_inode), \
911 raw_inode, (einode)->xtime)
912
913#define EXT4_INODE_GET_XTIME_VAL(xtime, inode, raw_inode) \
914 (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra) ? \
915 ext4_decode_extra_time((raw_inode)->xtime, \
916 (raw_inode)->xtime ## _extra) : \
917 (struct timespec64) { \
918 .tv_sec = (signed)le32_to_cpu((raw_inode)->xtime) \
919 })
920
921#define EXT4_INODE_GET_ATIME(inode, raw_inode) \
922do { \
923 inode_set_atime_to_ts(inode, \
924 EXT4_INODE_GET_XTIME_VAL(i_atime, inode, raw_inode)); \
925} while (0)
926
927#define EXT4_INODE_GET_MTIME(inode, raw_inode) \
928do { \
929 inode_set_mtime_to_ts(inode, \
930 EXT4_INODE_GET_XTIME_VAL(i_mtime, inode, raw_inode)); \
931} while (0)
932
933#define EXT4_INODE_GET_CTIME(inode, raw_inode) \
934do { \
935 inode_set_ctime_to_ts(inode, \
936 EXT4_INODE_GET_XTIME_VAL(i_ctime, inode, raw_inode)); \
937} while (0)
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 = \
943 EXT4_INODE_GET_XTIME_VAL(xtime, &(einode->vfs_inode), \
944 raw_inode); \
945 else \
946 (einode)->xtime = (struct timespec64){0, 0}; \
947} while (0)
948
949#define i_disk_version osd1.linux1.l_i_version
950
951#if defined(__KERNEL__) || defined(__linux__)
952#define i_reserved1 osd1.linux1.l_i_reserved1
953#define i_file_acl_high osd2.linux2.l_i_file_acl_high
954#define i_blocks_high osd2.linux2.l_i_blocks_high
955#define i_uid_low i_uid
956#define i_gid_low i_gid
957#define i_uid_high osd2.linux2.l_i_uid_high
958#define i_gid_high osd2.linux2.l_i_gid_high
959#define i_checksum_lo osd2.linux2.l_i_checksum_lo
960
961#elif defined(__GNU__)
962
963#define i_translator osd1.hurd1.h_i_translator
964#define i_uid_high osd2.hurd2.h_i_uid_high
965#define i_gid_high osd2.hurd2.h_i_gid_high
966#define i_author osd2.hurd2.h_i_author
967
968#elif defined(__masix__)
969
970#define i_reserved1 osd1.masix1.m_i_reserved1
971#define i_file_acl_high osd2.masix2.m_i_file_acl_high
972#define i_reserved2 osd2.masix2.m_i_reserved2
973
974#endif /* defined(__KERNEL__) || defined(__linux__) */
975
976#include "extents_status.h"
977#include "fast_commit.h"
978
979/*
980 * Lock subclasses for i_data_sem in the ext4_inode_info structure.
981 *
982 * These are needed to avoid lockdep false positives when we need to
983 * allocate blocks to the quota inode during ext4_map_blocks(), while
984 * holding i_data_sem for a normal (non-quota) inode. Since we don't
985 * do quota tracking for the quota inode, this avoids deadlock (as
986 * well as infinite recursion, since it isn't turtles all the way
987 * down...)
988 *
989 * I_DATA_SEM_NORMAL - Used for most inodes
990 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode
991 * where the second inode has larger inode number
992 * than the first
993 * I_DATA_SEM_QUOTA - Used for quota inodes only
994 * I_DATA_SEM_EA - Used for ea_inodes only
995 */
996enum {
997 I_DATA_SEM_NORMAL = 0,
998 I_DATA_SEM_OTHER,
999 I_DATA_SEM_QUOTA,
1000 I_DATA_SEM_EA
1001};
1002
1003
1004/*
1005 * fourth extended file system inode data in memory
1006 */
1007struct ext4_inode_info {
1008 __le32 i_data[15]; /* unconverted */
1009 __u32 i_dtime;
1010 ext4_fsblk_t i_file_acl;
1011
1012 /*
1013 * i_block_group is the number of the block group which contains
1014 * this file's inode. Constant across the lifetime of the inode,
1015 * it is used for making block allocation decisions - we try to
1016 * place a file's data blocks near its inode block, and new inodes
1017 * near to their parent directory's inode.
1018 */
1019 ext4_group_t i_block_group;
1020 ext4_lblk_t i_dir_start_lookup;
1021#if (BITS_PER_LONG < 64)
1022 unsigned long i_state_flags; /* Dynamic state flags */
1023#endif
1024 unsigned long i_flags;
1025
1026 /*
1027 * Extended attributes can be read independently of the main file
1028 * data. Taking i_rwsem even when reading would cause contention
1029 * between readers of EAs and writers of regular file data, so
1030 * instead we synchronize on xattr_sem when reading or changing
1031 * EAs.
1032 */
1033 struct rw_semaphore xattr_sem;
1034
1035 /*
1036 * Inodes with EXT4_STATE_ORPHAN_FILE use i_orphan_idx. Otherwise
1037 * i_orphan is used.
1038 */
1039 union {
1040 struct list_head i_orphan; /* unlinked but open inodes */
1041 unsigned int i_orphan_idx; /* Index in orphan file */
1042 };
1043
1044 /* Fast commit related info */
1045
1046 /* For tracking dentry create updates */
1047 struct list_head i_fc_dilist;
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 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1062
1063 /* Fast commit wait queue for this inode */
1064 wait_queue_head_t i_fc_wait;
1065
1066 /* Protect concurrent accesses on i_fc_lblk_start, i_fc_lblk_len */
1067 struct mutex i_fc_lock;
1068
1069 /*
1070 * i_disksize keeps track of what the inode size is ON DISK, not
1071 * in memory. During truncate, i_size is set to the new size by
1072 * the VFS prior to calling ext4_truncate(), but the filesystem won't
1073 * set i_disksize to 0 until the truncate is actually under way.
1074 *
1075 * The intent is that i_disksize always represents the blocks which
1076 * are used by this file. This allows recovery to restart truncate
1077 * on orphans if we crash during truncate. We actually write i_disksize
1078 * into the on-disk inode when writing inodes out, instead of i_size.
1079 *
1080 * The only time when i_disksize and i_size may be different is when
1081 * a truncate is in progress. The only things which change i_disksize
1082 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
1083 */
1084 loff_t i_disksize;
1085
1086 /*
1087 * i_data_sem is for serialising ext4_truncate() against
1088 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
1089 * data tree are chopped off during truncate. We can't do that in
1090 * ext4 because whenever we perform intermediate commits during
1091 * truncate, the inode and all the metadata blocks *must* be in a
1092 * consistent state which allows truncation of the orphans to restart
1093 * during recovery. Hence we must fix the get_block-vs-truncate race
1094 * by other means, so we have i_data_sem.
1095 */
1096 struct rw_semaphore i_data_sem;
1097 struct inode vfs_inode;
1098 struct jbd2_inode *jinode;
1099
1100 spinlock_t i_raw_lock; /* protects updates to the raw inode */
1101
1102 /*
1103 * File creation time. Its function is same as that of
1104 * struct timespec64 i_{a,c,m}time in the generic inode.
1105 */
1106 struct timespec64 i_crtime;
1107
1108 /* mballoc */
1109 atomic_t i_prealloc_active;
1110
1111 /* allocation reservation info for delalloc */
1112 /* In case of bigalloc, this refer to clusters rather than blocks */
1113 unsigned int i_reserved_data_blocks;
1114 struct rb_root i_prealloc_node;
1115 rwlock_t i_prealloc_lock;
1116
1117 /* extents status tree */
1118 struct ext4_es_tree i_es_tree;
1119 rwlock_t i_es_lock;
1120 struct list_head i_es_list;
1121 unsigned int i_es_all_nr; /* protected by i_es_lock */
1122 unsigned int i_es_shk_nr; /* protected by i_es_lock */
1123 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for
1124 extents to shrink. Protected by
1125 i_es_lock */
1126
1127 /* ialloc */
1128 ext4_group_t i_last_alloc_group;
1129
1130 /* pending cluster reservations for bigalloc file systems */
1131 struct ext4_pending_tree i_pending_tree;
1132
1133 /* on-disk additional length */
1134 __u16 i_extra_isize;
1135
1136 /* Indicate the inline data space. */
1137 u16 i_inline_off;
1138 u16 i_inline_size;
1139
1140#ifdef CONFIG_QUOTA
1141 /* quota space reservation, managed internally by quota code */
1142 qsize_t i_reserved_quota;
1143#endif
1144
1145 /* Lock protecting lists below */
1146 spinlock_t i_completed_io_lock;
1147 /*
1148 * Completed IOs that need unwritten extents handling and have
1149 * transaction reserved
1150 */
1151 struct list_head i_rsv_conversion_list;
1152 struct work_struct i_rsv_conversion_work;
1153
1154 spinlock_t i_block_reservation_lock;
1155
1156 /*
1157 * Transactions that contain inode's metadata needed to complete
1158 * fsync and fdatasync, respectively.
1159 */
1160 tid_t i_sync_tid;
1161 tid_t i_datasync_tid;
1162
1163#ifdef CONFIG_QUOTA
1164 struct dquot __rcu *i_dquot[MAXQUOTAS];
1165#endif
1166
1167 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1168 __u32 i_csum_seed;
1169
1170 kprojid_t i_projid;
1171};
1172
1173/*
1174 * File system states
1175 */
1176#define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
1177#define EXT4_ERROR_FS 0x0002 /* Errors detected */
1178#define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
1179#define EXT4_FC_REPLAY 0x0020 /* Fast commit replay ongoing */
1180
1181/*
1182 * Misc. filesystem flags
1183 */
1184#define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
1185#define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
1186#define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
1187
1188/*
1189 * Mount flags set via mount options or defaults
1190 */
1191#define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */
1192#define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
1193#define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
1194#define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
1195#define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
1196#define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
1197#define EXT4_MOUNT_ERRORS_MASK 0x00070
1198#define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
1199#define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
1200#ifdef CONFIG_FS_DAX
1201#define EXT4_MOUNT_DAX_ALWAYS 0x00200 /* Direct Access */
1202#else
1203#define EXT4_MOUNT_DAX_ALWAYS 0
1204#endif
1205#define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
1206#define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
1207#define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
1208#define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
1209#define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
1210#define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
1211#define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
1212#define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
1213#define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
1214#define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
1215#define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */
1216#define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota,
1217 * enable enforcement for hidden
1218 * quota files */
1219#define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable
1220 * enforcement for hidden quota
1221 * files */
1222#define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota
1223 * enforcement */
1224#define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
1225#define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
1226#define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1227#define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */
1228#define EXT4_MOUNT_NO_PREFETCH_BLOCK_BITMAPS 0x4000000
1229#define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
1230#define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
1231#define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
1232#define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
1233#define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
1234
1235/*
1236 * Mount flags set either automatically (could not be set by mount option)
1237 * based on per file system feature or property or in special cases such as
1238 * distinguishing between explicit mount option definition and default.
1239 */
1240#define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1241 specified delalloc */
1242#define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1243 size of blocksize * 8
1244 blocks */
1245#define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1246 file systems */
1247#define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly
1248 specified journal checksum */
1249
1250#define EXT4_MOUNT2_JOURNAL_FAST_COMMIT 0x00000010 /* Journal fast commit */
1251#define EXT4_MOUNT2_DAX_NEVER 0x00000020 /* Do not allow Direct Access */
1252#define EXT4_MOUNT2_DAX_INODE 0x00000040 /* For printing options only */
1253#define EXT4_MOUNT2_MB_OPTIMIZE_SCAN 0x00000080 /* Optimize group
1254 * scanning in mballoc
1255 */
1256#define EXT4_MOUNT2_ABORT 0x00000100 /* Abort filesystem */
1257
1258#define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1259 ~EXT4_MOUNT_##opt
1260#define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1261 EXT4_MOUNT_##opt
1262#define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1263 EXT4_MOUNT_##opt)
1264
1265#define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1266 ~EXT4_MOUNT2_##opt
1267#define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1268 EXT4_MOUNT2_##opt
1269#define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1270 EXT4_MOUNT2_##opt)
1271
1272#define ext4_test_and_set_bit __test_and_set_bit_le
1273#define ext4_set_bit __set_bit_le
1274#define ext4_test_and_clear_bit __test_and_clear_bit_le
1275#define ext4_clear_bit __clear_bit_le
1276#define ext4_test_bit test_bit_le
1277#define ext4_find_next_zero_bit find_next_zero_bit_le
1278#define ext4_find_next_bit find_next_bit_le
1279
1280extern void mb_set_bits(void *bm, int cur, int len);
1281
1282/*
1283 * Maximal mount counts between two filesystem checks
1284 */
1285#define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1286#define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1287
1288/*
1289 * Behaviour when detecting errors
1290 */
1291#define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1292#define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1293#define EXT4_ERRORS_PANIC 3 /* Panic */
1294#define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1295
1296/* Metadata checksum algorithm codes */
1297#define EXT4_CRC32C_CHKSUM 1
1298
1299#define EXT4_LABEL_MAX 16
1300
1301/*
1302 * Structure of the super block
1303 */
1304struct ext4_super_block {
1305/*00*/ __le32 s_inodes_count; /* Inodes count */
1306 __le32 s_blocks_count_lo; /* Blocks count */
1307 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1308 __le32 s_free_blocks_count_lo; /* Free blocks count */
1309/*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1310 __le32 s_first_data_block; /* First Data Block */
1311 __le32 s_log_block_size; /* Block size */
1312 __le32 s_log_cluster_size; /* Allocation cluster size */
1313/*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1314 __le32 s_clusters_per_group; /* # Clusters per group */
1315 __le32 s_inodes_per_group; /* # Inodes per group */
1316 __le32 s_mtime; /* Mount time */
1317/*30*/ __le32 s_wtime; /* Write time */
1318 __le16 s_mnt_count; /* Mount count */
1319 __le16 s_max_mnt_count; /* Maximal mount count */
1320 __le16 s_magic; /* Magic signature */
1321 __le16 s_state; /* File system state */
1322 __le16 s_errors; /* Behaviour when detecting errors */
1323 __le16 s_minor_rev_level; /* minor revision level */
1324/*40*/ __le32 s_lastcheck; /* time of last check */
1325 __le32 s_checkinterval; /* max. time between checks */
1326 __le32 s_creator_os; /* OS */
1327 __le32 s_rev_level; /* Revision level */
1328/*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1329 __le16 s_def_resgid; /* Default gid for reserved blocks */
1330 /*
1331 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1332 *
1333 * Note: the difference between the compatible feature set and
1334 * the incompatible feature set is that if there is a bit set
1335 * in the incompatible feature set that the kernel doesn't
1336 * know about, it should refuse to mount the filesystem.
1337 *
1338 * e2fsck's requirements are more strict; if it doesn't know
1339 * about a feature in either the compatible or incompatible
1340 * feature set, it must abort and not try to meddle with
1341 * things it doesn't understand...
1342 */
1343 __le32 s_first_ino; /* First non-reserved inode */
1344 __le16 s_inode_size; /* size of inode structure */
1345 __le16 s_block_group_nr; /* block group # of this superblock */
1346 __le32 s_feature_compat; /* compatible feature set */
1347/*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1348 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1349/*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1350/*78*/ char s_volume_name[EXT4_LABEL_MAX] __nonstring; /* volume name */
1351/*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */
1352/*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1353 /*
1354 * Performance hints. Directory preallocation should only
1355 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1356 */
1357 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1358 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1359 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1360 /*
1361 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1362 */
1363/*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1364/*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1365 __le32 s_journal_dev; /* device number of journal file */
1366 __le32 s_last_orphan; /* start of list of inodes to delete */
1367 __le32 s_hash_seed[4]; /* HTREE hash seed */
1368 __u8 s_def_hash_version; /* Default hash version to use */
1369 __u8 s_jnl_backup_type;
1370 __le16 s_desc_size; /* size of group descriptor */
1371/*100*/ __le32 s_default_mount_opts;
1372 __le32 s_first_meta_bg; /* First metablock block group */
1373 __le32 s_mkfs_time; /* When the filesystem was created */
1374 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1375 /* 64bit support valid if EXT4_FEATURE_INCOMPAT_64BIT */
1376/*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1377 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1378 __le32 s_free_blocks_count_hi; /* Free blocks count */
1379 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1380 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1381 __le32 s_flags; /* Miscellaneous flags */
1382 __le16 s_raid_stride; /* RAID stride */
1383 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1384 __le64 s_mmp_block; /* Block for multi-mount protection */
1385 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1386 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1387 __u8 s_checksum_type; /* metadata checksum algorithm used */
1388 __u8 s_encryption_level; /* versioning level for encryption */
1389 __u8 s_reserved_pad; /* Padding to next 32bits */
1390 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1391 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1392 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1393 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1394 snapshot's future use */
1395 __le32 s_snapshot_list; /* inode number of the head of the
1396 on-disk snapshot list */
1397#define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1398 __le32 s_error_count; /* number of fs errors */
1399 __le32 s_first_error_time; /* first time an error happened */
1400 __le32 s_first_error_ino; /* inode involved in first error */
1401 __le64 s_first_error_block; /* block involved of first error */
1402 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */
1403 __le32 s_first_error_line; /* line number where error happened */
1404 __le32 s_last_error_time; /* most recent time of an error */
1405 __le32 s_last_error_ino; /* inode involved in last error */
1406 __le32 s_last_error_line; /* line number where error happened */
1407 __le64 s_last_error_block; /* block involved of last error */
1408 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */
1409#define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1410 __u8 s_mount_opts[64];
1411 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1412 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1413 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1414 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
1415 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */
1416 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */
1417 __le32 s_lpf_ino; /* Location of the lost+found inode */
1418 __le32 s_prj_quota_inum; /* inode for tracking project quota */
1419 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */
1420 __u8 s_wtime_hi;
1421 __u8 s_mtime_hi;
1422 __u8 s_mkfs_time_hi;
1423 __u8 s_lastcheck_hi;
1424 __u8 s_first_error_time_hi;
1425 __u8 s_last_error_time_hi;
1426 __u8 s_first_error_errcode;
1427 __u8 s_last_error_errcode;
1428 __le16 s_encoding; /* Filename charset encoding */
1429 __le16 s_encoding_flags; /* Filename charset encoding flags */
1430 __le32 s_orphan_file_inum; /* Inode for tracking orphan inodes */
1431 __le32 s_reserved[94]; /* Padding to the end of the block */
1432 __le32 s_checksum; /* crc32c(superblock) */
1433};
1434
1435#define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1436
1437#ifdef __KERNEL__
1438
1439/* Number of quota types we support */
1440#define EXT4_MAXQUOTAS 3
1441
1442#define EXT4_ENC_UTF8_12_1 1
1443
1444/* Types of ext4 journal triggers */
1445enum ext4_journal_trigger_type {
1446 EXT4_JTR_ORPHAN_FILE,
1447 EXT4_JTR_NONE /* This must be the last entry for indexing to work! */
1448};
1449
1450#define EXT4_JOURNAL_TRIGGER_COUNT EXT4_JTR_NONE
1451
1452struct ext4_journal_trigger {
1453 struct jbd2_buffer_trigger_type tr_triggers;
1454 struct super_block *sb;
1455};
1456
1457static inline struct ext4_journal_trigger *EXT4_TRIGGER(
1458 struct jbd2_buffer_trigger_type *trigger)
1459{
1460 return container_of(trigger, struct ext4_journal_trigger, tr_triggers);
1461}
1462
1463#define EXT4_ORPHAN_BLOCK_MAGIC 0x0b10ca04
1464
1465/* Structure at the tail of orphan block */
1466struct ext4_orphan_block_tail {
1467 __le32 ob_magic;
1468 __le32 ob_checksum;
1469};
1470
1471static inline int ext4_inodes_per_orphan_block(struct super_block *sb)
1472{
1473 return (sb->s_blocksize - sizeof(struct ext4_orphan_block_tail)) /
1474 sizeof(u32);
1475}
1476
1477struct ext4_orphan_block {
1478 atomic_t ob_free_entries; /* Number of free orphan entries in block */
1479 struct buffer_head *ob_bh; /* Buffer for orphan block */
1480};
1481
1482/*
1483 * Info about orphan file.
1484 */
1485struct ext4_orphan_info {
1486 int of_blocks; /* Number of orphan blocks in a file */
1487 __u32 of_csum_seed; /* Checksum seed for orphan file */
1488 struct ext4_orphan_block *of_binfo; /* Array with info about orphan
1489 * file blocks */
1490};
1491
1492/*
1493 * fourth extended-fs super-block data in memory
1494 */
1495struct ext4_sb_info {
1496 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1497 unsigned long s_inodes_per_block;/* Number of inodes per block */
1498 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1499 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1500 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1501 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1502 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1503 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1504 ext4_group_t s_groups_count; /* Number of groups in the fs */
1505 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1506 unsigned long s_overhead; /* # of fs overhead clusters */
1507 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1508 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1509 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1510 struct buffer_head * s_sbh; /* Buffer containing the super block */
1511 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1512 /* Array of bh's for the block group descriptors */
1513 struct buffer_head * __rcu *s_group_desc;
1514 unsigned int s_mount_opt;
1515 unsigned int s_mount_opt2;
1516 unsigned long s_mount_flags;
1517 unsigned int s_def_mount_opt;
1518 unsigned int s_def_mount_opt2;
1519 ext4_fsblk_t s_sb_block;
1520 atomic64_t s_resv_clusters;
1521 kuid_t s_resuid;
1522 kgid_t s_resgid;
1523 unsigned short s_mount_state;
1524 unsigned short s_pad;
1525 int s_addr_per_block_bits;
1526 int s_desc_per_block_bits;
1527 int s_inode_size;
1528 int s_first_ino;
1529 unsigned int s_inode_readahead_blks;
1530 unsigned int s_inode_goal;
1531 u32 s_hash_seed[4];
1532 int s_def_hash_version;
1533 int s_hash_unsigned; /* 3 if hash should be unsigned, 0 if not */
1534 struct percpu_counter s_freeclusters_counter;
1535 struct percpu_counter s_freeinodes_counter;
1536 struct percpu_counter s_dirs_counter;
1537 struct percpu_counter s_dirtyclusters_counter;
1538 struct percpu_counter s_sra_exceeded_retry_limit;
1539 struct blockgroup_lock *s_blockgroup_lock;
1540 struct proc_dir_entry *s_proc;
1541 struct kobject s_kobj;
1542 struct completion s_kobj_unregister;
1543 struct super_block *s_sb;
1544 struct buffer_head *s_mmp_bh;
1545
1546 /* Journaling */
1547 struct journal_s *s_journal;
1548 unsigned long s_ext4_flags; /* Ext4 superblock flags */
1549 struct mutex s_orphan_lock; /* Protects on disk list changes */
1550 struct list_head s_orphan; /* List of orphaned inodes in on disk
1551 list */
1552 struct ext4_orphan_info s_orphan_info;
1553 unsigned long s_commit_interval;
1554 u32 s_max_batch_time;
1555 u32 s_min_batch_time;
1556 struct file *s_journal_bdev_file;
1557#ifdef CONFIG_QUOTA
1558 /* Names of quota files with journalled quota */
1559 char __rcu *s_qf_names[EXT4_MAXQUOTAS];
1560 int s_jquota_fmt; /* Format of quota to use */
1561#endif
1562 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1563 struct ext4_system_blocks __rcu *s_system_blks;
1564
1565#ifdef EXTENTS_STATS
1566 /* ext4 extents stats */
1567 unsigned long s_ext_min;
1568 unsigned long s_ext_max;
1569 unsigned long s_depth_max;
1570 spinlock_t s_ext_stats_lock;
1571 unsigned long s_ext_blocks;
1572 unsigned long s_ext_extents;
1573#endif
1574
1575 /* for buddy allocator */
1576 struct ext4_group_info ** __rcu *s_group_info;
1577 struct inode *s_buddy_cache;
1578 spinlock_t s_md_lock;
1579 unsigned short *s_mb_offsets;
1580 unsigned int *s_mb_maxs;
1581 unsigned int s_group_info_size;
1582 unsigned int s_mb_free_pending;
1583 struct list_head s_freed_data_list[2]; /* List of blocks to be freed
1584 after commit completed */
1585 struct list_head s_discard_list;
1586 struct work_struct s_discard_work;
1587 atomic_t s_retry_alloc_pending;
1588 struct list_head *s_mb_avg_fragment_size;
1589 rwlock_t *s_mb_avg_fragment_size_locks;
1590 struct list_head *s_mb_largest_free_orders;
1591 rwlock_t *s_mb_largest_free_orders_locks;
1592
1593 /* tunables */
1594 unsigned long s_stripe;
1595 unsigned int s_mb_max_linear_groups;
1596 unsigned int s_mb_stream_request;
1597 unsigned int s_mb_max_to_scan;
1598 unsigned int s_mb_min_to_scan;
1599 unsigned int s_mb_stats;
1600 unsigned int s_mb_order2_reqs;
1601 unsigned int s_mb_group_prealloc;
1602 unsigned int s_max_dir_size_kb;
1603 /* where last allocation was done - for stream allocation */
1604 unsigned long s_mb_last_group;
1605 unsigned long s_mb_last_start;
1606 unsigned int s_mb_prefetch;
1607 unsigned int s_mb_prefetch_limit;
1608 unsigned int s_mb_best_avail_max_trim_order;
1609
1610 /* stats for buddy allocator */
1611 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1612 atomic_t s_bal_success; /* we found long enough chunks */
1613 atomic_t s_bal_allocated; /* in blocks */
1614 atomic_t s_bal_ex_scanned; /* total extents scanned */
1615 atomic_t s_bal_cX_ex_scanned[EXT4_MB_NUM_CRS]; /* total extents scanned */
1616 atomic_t s_bal_groups_scanned; /* number of groups scanned */
1617 atomic_t s_bal_goals; /* goal hits */
1618 atomic_t s_bal_len_goals; /* len goal hits */
1619 atomic_t s_bal_breaks; /* too long searches */
1620 atomic_t s_bal_2orders; /* 2^order hits */
1621 atomic_t s_bal_p2_aligned_bad_suggestions;
1622 atomic_t s_bal_goal_fast_bad_suggestions;
1623 atomic_t s_bal_best_avail_bad_suggestions;
1624 atomic64_t s_bal_cX_groups_considered[EXT4_MB_NUM_CRS];
1625 atomic64_t s_bal_cX_hits[EXT4_MB_NUM_CRS];
1626 atomic64_t s_bal_cX_failed[EXT4_MB_NUM_CRS]; /* cX loop didn't find blocks */
1627 atomic_t s_mb_buddies_generated; /* number of buddies generated */
1628 atomic64_t s_mb_generation_time;
1629 atomic_t s_mb_lost_chunks;
1630 atomic_t s_mb_preallocated;
1631 atomic_t s_mb_discarded;
1632 atomic_t s_lock_busy;
1633
1634 /* locality groups */
1635 struct ext4_locality_group __percpu *s_locality_groups;
1636
1637 /* for write statistics */
1638 unsigned long s_sectors_written_start;
1639 u64 s_kbytes_written;
1640
1641 /* the size of zero-out chunk */
1642 unsigned int s_extent_max_zeroout_kb;
1643
1644 unsigned int s_log_groups_per_flex;
1645 struct flex_groups * __rcu *s_flex_groups;
1646 ext4_group_t s_flex_groups_allocated;
1647
1648 /* workqueue for reserved extent conversions (buffered io) */
1649 struct workqueue_struct *rsv_conversion_wq;
1650
1651 /* timer for periodic error stats printing */
1652 struct timer_list s_err_report;
1653
1654 /* Lazy inode table initialization info */
1655 struct ext4_li_request *s_li_request;
1656 /* Wait multiplier for lazy initialization thread */
1657 unsigned int s_li_wait_mult;
1658
1659 /* Kernel thread for multiple mount protection */
1660 struct task_struct *s_mmp_tsk;
1661
1662 /* record the last minlen when FITRIM is called. */
1663 unsigned long s_last_trim_minblks;
1664
1665 /* Precomputed FS UUID checksum for seeding other checksums */
1666 __u32 s_csum_seed;
1667
1668 /* Reclaim extents from extent status tree */
1669 struct shrinker *s_es_shrinker;
1670 struct list_head s_es_list; /* List of inodes with reclaimable extents */
1671 long s_es_nr_inode;
1672 struct ext4_es_stats s_es_stats;
1673 struct mb_cache *s_ea_block_cache;
1674 struct mb_cache *s_ea_inode_cache;
1675 spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1676
1677 /* Journal triggers for checksum computation */
1678 struct ext4_journal_trigger s_journal_triggers[EXT4_JOURNAL_TRIGGER_COUNT];
1679
1680 /* Ratelimit ext4 messages. */
1681 struct ratelimit_state s_err_ratelimit_state;
1682 struct ratelimit_state s_warning_ratelimit_state;
1683 struct ratelimit_state s_msg_ratelimit_state;
1684 atomic_t s_warning_count;
1685 atomic_t s_msg_count;
1686
1687 /* Encryption policy for '-o test_dummy_encryption' */
1688 struct fscrypt_dummy_policy s_dummy_enc_policy;
1689
1690 /*
1691 * Barrier between writepages ops and changing any inode's JOURNAL_DATA
1692 * or EXTENTS flag or between writepages ops and changing DELALLOC or
1693 * DIOREAD_NOLOCK mount options on remount.
1694 */
1695 struct percpu_rw_semaphore s_writepages_rwsem;
1696 struct dax_device *s_daxdev;
1697 u64 s_dax_part_off;
1698#ifdef CONFIG_EXT4_DEBUG
1699 unsigned long s_simulate_fail;
1700#endif
1701 /* Record the errseq of the backing block device */
1702 errseq_t s_bdev_wb_err;
1703 spinlock_t s_bdev_wb_lock;
1704
1705 /* Information about errors that happened during this mount */
1706 spinlock_t s_error_lock;
1707 int s_add_error_count;
1708 int s_first_error_code;
1709 __u32 s_first_error_line;
1710 __u32 s_first_error_ino;
1711 __u64 s_first_error_block;
1712 const char *s_first_error_func;
1713 time64_t s_first_error_time;
1714 int s_last_error_code;
1715 __u32 s_last_error_line;
1716 __u32 s_last_error_ino;
1717 __u64 s_last_error_block;
1718 const char *s_last_error_func;
1719 time64_t s_last_error_time;
1720 /*
1721 * If we are in a context where we cannot update the on-disk
1722 * superblock, we queue the work here. This is used to update
1723 * the error information in the superblock, and for periodic
1724 * updates of the superblock called from the commit callback
1725 * function.
1726 */
1727 struct work_struct s_sb_upd_work;
1728
1729 /* Atomic write unit values in bytes */
1730 unsigned int s_awu_min;
1731 unsigned int s_awu_max;
1732
1733 /* Ext4 fast commit sub transaction ID */
1734 atomic_t s_fc_subtid;
1735
1736 /*
1737 * After commit starts, the main queue gets locked, and the further
1738 * updates get added in the staging queue.
1739 */
1740#define FC_Q_MAIN 0
1741#define FC_Q_STAGING 1
1742 struct list_head s_fc_q[2]; /* Inodes staged for fast commit
1743 * that have data changes in them.
1744 */
1745 struct list_head s_fc_dentry_q[2]; /* directory entry updates */
1746 unsigned int s_fc_bytes;
1747 /*
1748 * Main fast commit lock. This lock protects accesses to the
1749 * following fields:
1750 * ei->i_fc_list, s_fc_dentry_q, s_fc_q, s_fc_bytes, s_fc_bh.
1751 */
1752 spinlock_t s_fc_lock;
1753 struct buffer_head *s_fc_bh;
1754 struct ext4_fc_stats s_fc_stats;
1755 tid_t s_fc_ineligible_tid;
1756#ifdef CONFIG_EXT4_DEBUG
1757 int s_fc_debug_max_replay;
1758#endif
1759 struct ext4_fc_replay_state s_fc_replay_state;
1760};
1761
1762static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1763{
1764 return sb->s_fs_info;
1765}
1766static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1767{
1768 return container_of(inode, struct ext4_inode_info, vfs_inode);
1769}
1770
1771static inline int ext4_writepages_down_read(struct super_block *sb)
1772{
1773 percpu_down_read(&EXT4_SB(sb)->s_writepages_rwsem);
1774 return memalloc_nofs_save();
1775}
1776
1777static inline void ext4_writepages_up_read(struct super_block *sb, int ctx)
1778{
1779 memalloc_nofs_restore(ctx);
1780 percpu_up_read(&EXT4_SB(sb)->s_writepages_rwsem);
1781}
1782
1783static inline int ext4_writepages_down_write(struct super_block *sb)
1784{
1785 percpu_down_write(&EXT4_SB(sb)->s_writepages_rwsem);
1786 return memalloc_nofs_save();
1787}
1788
1789static inline void ext4_writepages_up_write(struct super_block *sb, int ctx)
1790{
1791 memalloc_nofs_restore(ctx);
1792 percpu_up_write(&EXT4_SB(sb)->s_writepages_rwsem);
1793}
1794
1795static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1796{
1797 return ino == EXT4_ROOT_INO ||
1798 (ino >= EXT4_FIRST_INO(sb) &&
1799 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1800}
1801
1802/*
1803 * Returns: sbi->field[index]
1804 * Used to access an array element from the following sbi fields which require
1805 * rcu protection to avoid dereferencing an invalid pointer due to reassignment
1806 * - s_group_desc
1807 * - s_group_info
1808 * - s_flex_group
1809 */
1810#define sbi_array_rcu_deref(sbi, field, index) \
1811({ \
1812 typeof(*((sbi)->field)) _v; \
1813 rcu_read_lock(); \
1814 _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index]; \
1815 rcu_read_unlock(); \
1816 _v; \
1817})
1818
1819/*
1820 * run-time mount flags
1821 */
1822enum {
1823 EXT4_MF_MNTDIR_SAMPLED,
1824 EXT4_MF_FC_INELIGIBLE /* Fast commit ineligible */
1825};
1826
1827static inline void ext4_set_mount_flag(struct super_block *sb, int bit)
1828{
1829 set_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1830}
1831
1832static inline void ext4_clear_mount_flag(struct super_block *sb, int bit)
1833{
1834 clear_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1835}
1836
1837static inline int ext4_test_mount_flag(struct super_block *sb, int bit)
1838{
1839 return test_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1840}
1841
1842
1843/*
1844 * Simulate_fail codes
1845 */
1846#define EXT4_SIM_BBITMAP_EIO 1
1847#define EXT4_SIM_BBITMAP_CRC 2
1848#define EXT4_SIM_IBITMAP_EIO 3
1849#define EXT4_SIM_IBITMAP_CRC 4
1850#define EXT4_SIM_INODE_EIO 5
1851#define EXT4_SIM_INODE_CRC 6
1852#define EXT4_SIM_DIRBLOCK_EIO 7
1853#define EXT4_SIM_DIRBLOCK_CRC 8
1854
1855static inline bool ext4_simulate_fail(struct super_block *sb,
1856 unsigned long code)
1857{
1858#ifdef CONFIG_EXT4_DEBUG
1859 struct ext4_sb_info *sbi = EXT4_SB(sb);
1860
1861 if (unlikely(sbi->s_simulate_fail == code)) {
1862 sbi->s_simulate_fail = 0;
1863 return true;
1864 }
1865#endif
1866 return false;
1867}
1868
1869/*
1870 * Error number codes for s_{first,last}_error_errno
1871 *
1872 * Linux errno numbers are architecture specific, so we need to translate
1873 * them into something which is architecture independent. We don't define
1874 * codes for all errno's; just the ones which are most likely to be the cause
1875 * of an ext4_error() call.
1876 */
1877#define EXT4_ERR_UNKNOWN 1
1878#define EXT4_ERR_EIO 2
1879#define EXT4_ERR_ENOMEM 3
1880#define EXT4_ERR_EFSBADCRC 4
1881#define EXT4_ERR_EFSCORRUPTED 5
1882#define EXT4_ERR_ENOSPC 6
1883#define EXT4_ERR_ENOKEY 7
1884#define EXT4_ERR_EROFS 8
1885#define EXT4_ERR_EFBIG 9
1886#define EXT4_ERR_EEXIST 10
1887#define EXT4_ERR_ERANGE 11
1888#define EXT4_ERR_EOVERFLOW 12
1889#define EXT4_ERR_EBUSY 13
1890#define EXT4_ERR_ENOTDIR 14
1891#define EXT4_ERR_ENOTEMPTY 15
1892#define EXT4_ERR_ESHUTDOWN 16
1893#define EXT4_ERR_EFAULT 17
1894
1895/*
1896 * Inode dynamic state flags
1897 */
1898enum {
1899 EXT4_STATE_NEW, /* inode is newly created */
1900 EXT4_STATE_XATTR, /* has in-inode xattrs */
1901 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1902 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1903 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1904 EXT4_STATE_NEWENTRY, /* File just added to dir */
1905 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1906 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
1907 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */
1908 EXT4_STATE_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */
1909 EXT4_STATE_FC_COMMITTING, /* Fast commit ongoing */
1910 EXT4_STATE_ORPHAN_FILE, /* Inode orphaned in orphan file */
1911};
1912
1913#define EXT4_INODE_BIT_FNS(name, field, offset) \
1914static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1915{ \
1916 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1917} \
1918static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1919{ \
1920 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1921} \
1922static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1923{ \
1924 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1925}
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_flag(struct inode *inode, int bit);
1930static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1931static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1932EXT4_INODE_BIT_FNS(flag, flags, 0)
1933
1934/* Add these declarations here only so that these functions can be
1935 * found by name. Otherwise, they are very hard to locate. */
1936static inline int ext4_test_inode_state(struct inode *inode, int bit);
1937static inline void ext4_set_inode_state(struct inode *inode, int bit);
1938static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1939#if (BITS_PER_LONG < 64)
1940EXT4_INODE_BIT_FNS(state, state_flags, 0)
1941
1942static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1943{
1944 (ei)->i_state_flags = 0;
1945}
1946#else
1947EXT4_INODE_BIT_FNS(state, flags, 32)
1948
1949static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1950{
1951 /* We depend on the fact that callers will set i_flags */
1952}
1953#endif
1954#else
1955/* Assume that user mode programs are passing in an ext4fs superblock, not
1956 * a kernel struct super_block. This will allow us to call the feature-test
1957 * macros from user land. */
1958#define EXT4_SB(sb) (sb)
1959#endif
1960
1961static inline bool ext4_verity_in_progress(struct inode *inode)
1962{
1963 return IS_ENABLED(CONFIG_FS_VERITY) &&
1964 ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS);
1965}
1966
1967#define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1968
1969/*
1970 * Codes for operating systems
1971 */
1972#define EXT4_OS_LINUX 0
1973#define EXT4_OS_HURD 1
1974#define EXT4_OS_MASIX 2
1975#define EXT4_OS_FREEBSD 3
1976#define EXT4_OS_LITES 4
1977
1978/*
1979 * Revision levels
1980 */
1981#define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1982#define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1983
1984#define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1985
1986#define EXT4_GOOD_OLD_INODE_SIZE 128
1987
1988#define EXT4_EXTRA_TIMESTAMP_MAX (((s64)1 << 34) - 1 + S32_MIN)
1989#define EXT4_NON_EXTRA_TIMESTAMP_MAX S32_MAX
1990#define EXT4_TIMESTAMP_MIN S32_MIN
1991
1992/*
1993 * Feature set definitions
1994 */
1995
1996#define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1997#define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1998#define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1999#define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
2000#define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
2001#define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
2002#define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
2003/*
2004 * The reason why "FAST_COMMIT" is a compat feature is that, FS becomes
2005 * incompatible only if fast commit blocks are present in the FS. Since we
2006 * clear the journal (and thus the fast commit blocks), we don't mark FS as
2007 * incompatible. We also have a JBD2 incompat feature, which gets set when
2008 * there are fast commit blocks present in the journal.
2009 */
2010#define EXT4_FEATURE_COMPAT_FAST_COMMIT 0x0400
2011#define EXT4_FEATURE_COMPAT_STABLE_INODES 0x0800
2012#define EXT4_FEATURE_COMPAT_ORPHAN_FILE 0x1000 /* Orphan file exists */
2013
2014#define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
2015#define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
2016#define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
2017#define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
2018#define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
2019#define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
2020#define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
2021#define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
2022#define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
2023/*
2024 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
2025 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
2026 * all other data structures' checksums. However, the METADATA_CSUM and
2027 * GDT_CSUM bits are mutually exclusive.
2028 */
2029#define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
2030#define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000
2031#define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000
2032#define EXT4_FEATURE_RO_COMPAT_VERITY 0x8000
2033#define EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT 0x10000 /* Orphan file may be
2034 non-empty */
2035
2036#define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
2037#define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
2038#define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
2039#define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
2040#define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
2041#define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
2042#define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
2043#define EXT4_FEATURE_INCOMPAT_MMP 0x0100
2044#define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
2045#define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
2046#define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
2047#define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000
2048#define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
2049#define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
2050#define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000
2051#define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000
2052
2053extern void ext4_update_dynamic_rev(struct super_block *sb);
2054
2055#define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
2056static inline bool ext4_has_feature_##name(struct super_block *sb) \
2057{ \
2058 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
2059 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
2060} \
2061static inline void ext4_set_feature_##name(struct super_block *sb) \
2062{ \
2063 ext4_update_dynamic_rev(sb); \
2064 EXT4_SB(sb)->s_es->s_feature_compat |= \
2065 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
2066} \
2067static inline void ext4_clear_feature_##name(struct super_block *sb) \
2068{ \
2069 EXT4_SB(sb)->s_es->s_feature_compat &= \
2070 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
2071}
2072
2073#define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
2074static inline bool ext4_has_feature_##name(struct super_block *sb) \
2075{ \
2076 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
2077 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
2078} \
2079static inline void ext4_set_feature_##name(struct super_block *sb) \
2080{ \
2081 ext4_update_dynamic_rev(sb); \
2082 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
2083 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
2084} \
2085static inline void ext4_clear_feature_##name(struct super_block *sb) \
2086{ \
2087 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
2088 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
2089}
2090
2091#define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
2092static inline bool ext4_has_feature_##name(struct super_block *sb) \
2093{ \
2094 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
2095 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
2096} \
2097static inline void ext4_set_feature_##name(struct super_block *sb) \
2098{ \
2099 ext4_update_dynamic_rev(sb); \
2100 EXT4_SB(sb)->s_es->s_feature_incompat |= \
2101 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
2102} \
2103static inline void ext4_clear_feature_##name(struct super_block *sb) \
2104{ \
2105 EXT4_SB(sb)->s_es->s_feature_incompat &= \
2106 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
2107}
2108
2109EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC)
2110EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES)
2111EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL)
2112EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR)
2113EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE)
2114EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX)
2115EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2)
2116EXT4_FEATURE_COMPAT_FUNCS(fast_commit, FAST_COMMIT)
2117EXT4_FEATURE_COMPAT_FUNCS(stable_inodes, STABLE_INODES)
2118EXT4_FEATURE_COMPAT_FUNCS(orphan_file, ORPHAN_FILE)
2119
2120EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER)
2121EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE)
2122EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR)
2123EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE)
2124EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM)
2125EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK)
2126EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE)
2127EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA)
2128EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC)
2129EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM)
2130EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY)
2131EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT)
2132EXT4_FEATURE_RO_COMPAT_FUNCS(verity, VERITY)
2133EXT4_FEATURE_RO_COMPAT_FUNCS(orphan_present, ORPHAN_PRESENT)
2134
2135EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION)
2136EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE)
2137EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER)
2138EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV)
2139EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG)
2140EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS)
2141EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT)
2142EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP)
2143EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG)
2144EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE)
2145EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA)
2146EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED)
2147EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR)
2148EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA)
2149EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT)
2150EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD)
2151
2152#define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
2153#define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
2154 EXT4_FEATURE_INCOMPAT_META_BG)
2155#define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2156 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2157 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
2158
2159#define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
2160#define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
2161 EXT4_FEATURE_INCOMPAT_RECOVER| \
2162 EXT4_FEATURE_INCOMPAT_META_BG)
2163#define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2164 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2165 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
2166
2167#define EXT4_FEATURE_COMPAT_SUPP (EXT4_FEATURE_COMPAT_EXT_ATTR| \
2168 EXT4_FEATURE_COMPAT_ORPHAN_FILE)
2169#define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
2170 EXT4_FEATURE_INCOMPAT_RECOVER| \
2171 EXT4_FEATURE_INCOMPAT_META_BG| \
2172 EXT4_FEATURE_INCOMPAT_EXTENTS| \
2173 EXT4_FEATURE_INCOMPAT_64BIT| \
2174 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
2175 EXT4_FEATURE_INCOMPAT_EA_INODE| \
2176 EXT4_FEATURE_INCOMPAT_MMP | \
2177 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
2178 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
2179 EXT4_FEATURE_INCOMPAT_CASEFOLD | \
2180 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \
2181 EXT4_FEATURE_INCOMPAT_LARGEDIR)
2182#define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2183 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2184 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
2185 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
2186 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
2187 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
2188 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
2189 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
2190 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
2191 EXT4_FEATURE_RO_COMPAT_QUOTA |\
2192 EXT4_FEATURE_RO_COMPAT_PROJECT |\
2193 EXT4_FEATURE_RO_COMPAT_VERITY |\
2194 EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT)
2195
2196#define EXTN_FEATURE_FUNCS(ver) \
2197static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
2198{ \
2199 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
2200 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
2201} \
2202static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
2203{ \
2204 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
2205 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
2206} \
2207static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
2208{ \
2209 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
2210 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
2211}
2212
2213EXTN_FEATURE_FUNCS(2)
2214EXTN_FEATURE_FUNCS(3)
2215EXTN_FEATURE_FUNCS(4)
2216
2217static inline bool ext4_has_compat_features(struct super_block *sb)
2218{
2219 return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
2220}
2221static inline bool ext4_has_ro_compat_features(struct super_block *sb)
2222{
2223 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
2224}
2225static inline bool ext4_has_incompat_features(struct super_block *sb)
2226{
2227 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
2228}
2229
2230extern int ext4_feature_set_ok(struct super_block *sb, int readonly);
2231
2232/*
2233 * Superblock flags
2234 */
2235#define EXT4_FLAGS_RESIZING 0
2236#define EXT4_FLAGS_SHUTDOWN 1
2237#define EXT4_FLAGS_BDEV_IS_DAX 2
2238
2239static inline int ext4_forced_shutdown(struct super_block *sb)
2240{
2241 return test_bit(EXT4_FLAGS_SHUTDOWN, &EXT4_SB(sb)->s_ext4_flags);
2242}
2243
2244/*
2245 * Default values for user and/or group using reserved blocks
2246 */
2247#define EXT4_DEF_RESUID 0
2248#define EXT4_DEF_RESGID 0
2249
2250/*
2251 * Default project ID
2252 */
2253#define EXT4_DEF_PROJID 0
2254
2255#define EXT4_DEF_INODE_READAHEAD_BLKS 32
2256
2257/*
2258 * Default mount options
2259 */
2260#define EXT4_DEFM_DEBUG 0x0001
2261#define EXT4_DEFM_BSDGROUPS 0x0002
2262#define EXT4_DEFM_XATTR_USER 0x0004
2263#define EXT4_DEFM_ACL 0x0008
2264#define EXT4_DEFM_UID16 0x0010
2265#define EXT4_DEFM_JMODE 0x0060
2266#define EXT4_DEFM_JMODE_DATA 0x0020
2267#define EXT4_DEFM_JMODE_ORDERED 0x0040
2268#define EXT4_DEFM_JMODE_WBACK 0x0060
2269#define EXT4_DEFM_NOBARRIER 0x0100
2270#define EXT4_DEFM_BLOCK_VALIDITY 0x0200
2271#define EXT4_DEFM_DISCARD 0x0400
2272#define EXT4_DEFM_NODELALLOC 0x0800
2273
2274/*
2275 * Default journal batch times
2276 */
2277#define EXT4_DEF_MIN_BATCH_TIME 0
2278#define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
2279
2280/*
2281 * Minimum number of groups in a flexgroup before we separate out
2282 * directories into the first block group of a flexgroup
2283 */
2284#define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
2285
2286/*
2287 * Structure of a directory entry
2288 */
2289#define EXT4_NAME_LEN 255
2290/*
2291 * Base length of the ext4 directory entry excluding the name length
2292 */
2293#define EXT4_BASE_DIR_LEN (sizeof(struct ext4_dir_entry_2) - EXT4_NAME_LEN)
2294
2295struct ext4_dir_entry {
2296 __le32 inode; /* Inode number */
2297 __le16 rec_len; /* Directory entry length */
2298 __le16 name_len; /* Name length */
2299 char name[EXT4_NAME_LEN]; /* File name */
2300};
2301
2302
2303/*
2304 * Encrypted Casefolded entries require saving the hash on disk. This structure
2305 * followed ext4_dir_entry_2's name[name_len] at the next 4 byte aligned
2306 * boundary.
2307 */
2308struct ext4_dir_entry_hash {
2309 __le32 hash;
2310 __le32 minor_hash;
2311};
2312
2313/*
2314 * The new version of the directory entry. Since EXT4 structures are
2315 * stored in intel byte order, and the name_len field could never be
2316 * bigger than 255 chars, it's safe to reclaim the extra byte for the
2317 * file_type field.
2318 */
2319struct ext4_dir_entry_2 {
2320 __le32 inode; /* Inode number */
2321 __le16 rec_len; /* Directory entry length */
2322 __u8 name_len; /* Name length */
2323 __u8 file_type; /* See file type macros EXT4_FT_* below */
2324 char name[EXT4_NAME_LEN]; /* File name */
2325};
2326
2327/*
2328 * Access the hashes at the end of ext4_dir_entry_2
2329 */
2330#define EXT4_DIRENT_HASHES(entry) \
2331 ((struct ext4_dir_entry_hash *) \
2332 (((void *)(entry)) + \
2333 ((8 + (entry)->name_len + EXT4_DIR_ROUND) & ~EXT4_DIR_ROUND)))
2334#define EXT4_DIRENT_HASH(entry) le32_to_cpu(EXT4_DIRENT_HASHES(entry)->hash)
2335#define EXT4_DIRENT_MINOR_HASH(entry) \
2336 le32_to_cpu(EXT4_DIRENT_HASHES(entry)->minor_hash)
2337
2338static inline bool ext4_hash_in_dirent(const struct inode *inode)
2339{
2340 return IS_CASEFOLDED(inode) && IS_ENCRYPTED(inode);
2341}
2342
2343/*
2344 * This is a bogus directory entry at the end of each leaf block that
2345 * records checksums.
2346 */
2347struct ext4_dir_entry_tail {
2348 __le32 det_reserved_zero1; /* Pretend to be unused */
2349 __le16 det_rec_len; /* 12 */
2350 __u8 det_reserved_zero2; /* Zero name length */
2351 __u8 det_reserved_ft; /* 0xDE, fake file type */
2352 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
2353};
2354
2355#define EXT4_DIRENT_TAIL(block, blocksize) \
2356 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
2357 ((blocksize) - \
2358 sizeof(struct ext4_dir_entry_tail))))
2359
2360/*
2361 * Ext4 directory file types. Only the low 3 bits are used. The
2362 * other bits are reserved for now.
2363 */
2364#define EXT4_FT_UNKNOWN 0
2365#define EXT4_FT_REG_FILE 1
2366#define EXT4_FT_DIR 2
2367#define EXT4_FT_CHRDEV 3
2368#define EXT4_FT_BLKDEV 4
2369#define EXT4_FT_FIFO 5
2370#define EXT4_FT_SOCK 6
2371#define EXT4_FT_SYMLINK 7
2372
2373#define EXT4_FT_MAX 8
2374
2375#define EXT4_FT_DIR_CSUM 0xDE
2376
2377/*
2378 * EXT4_DIR_PAD defines the directory entries boundaries
2379 *
2380 * NOTE: It must be a multiple of 4
2381 */
2382#define EXT4_DIR_PAD 4
2383#define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
2384#define EXT4_MAX_REC_LEN ((1<<16)-1)
2385
2386/*
2387 * The rec_len is dependent on the type of directory. Directories that are
2388 * casefolded and encrypted need to store the hash as well, so we add room for
2389 * ext4_extended_dir_entry_2. For all entries related to '.' or '..' you should
2390 * pass NULL for dir, as those entries do not use the extra fields.
2391 */
2392static inline unsigned int ext4_dir_rec_len(__u8 name_len,
2393 const struct inode *dir)
2394{
2395 int rec_len = (name_len + 8 + EXT4_DIR_ROUND);
2396
2397 if (dir && ext4_hash_in_dirent(dir))
2398 rec_len += sizeof(struct ext4_dir_entry_hash);
2399 return (rec_len & ~EXT4_DIR_ROUND);
2400}
2401
2402/*
2403 * If we ever get support for fs block sizes > page_size, we'll need
2404 * to remove the #if statements in the next two functions...
2405 */
2406static inline unsigned int
2407ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
2408{
2409 unsigned len = le16_to_cpu(dlen);
2410
2411#if (PAGE_SIZE >= 65536)
2412 if (len == EXT4_MAX_REC_LEN || len == 0)
2413 return blocksize;
2414 return (len & 65532) | ((len & 3) << 16);
2415#else
2416 return len;
2417#endif
2418}
2419
2420static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
2421{
2422 BUG_ON((len > blocksize) || (blocksize > (1 << 18)) || (len & 3));
2423#if (PAGE_SIZE >= 65536)
2424 if (len < 65536)
2425 return cpu_to_le16(len);
2426 if (len == blocksize) {
2427 if (blocksize == 65536)
2428 return cpu_to_le16(EXT4_MAX_REC_LEN);
2429 else
2430 return cpu_to_le16(0);
2431 }
2432 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
2433#else
2434 return cpu_to_le16(len);
2435#endif
2436}
2437
2438/*
2439 * Hash Tree Directory indexing
2440 * (c) Daniel Phillips, 2001
2441 */
2442
2443#define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
2444 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2445#define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \
2446 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir)))
2447#define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2448
2449/* Legal values for the dx_root hash_version field: */
2450
2451#define DX_HASH_LEGACY 0
2452#define DX_HASH_HALF_MD4 1
2453#define DX_HASH_TEA 2
2454#define DX_HASH_LEGACY_UNSIGNED 3
2455#define DX_HASH_HALF_MD4_UNSIGNED 4
2456#define DX_HASH_TEA_UNSIGNED 5
2457#define DX_HASH_SIPHASH 6
2458#define DX_HASH_LAST DX_HASH_SIPHASH
2459
2460static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2461 const void *address, unsigned int length)
2462{
2463 return crc32c(crc, address, length);
2464}
2465
2466#ifdef __KERNEL__
2467
2468/* hash info structure used by the directory hash */
2469struct dx_hash_info
2470{
2471 u32 hash;
2472 u32 minor_hash;
2473 int hash_version;
2474 u32 *seed;
2475};
2476
2477
2478/* 32 and 64 bit signed EOF for dx directories */
2479#define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
2480#define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
2481
2482
2483/*
2484 * Control parameters used by ext4_htree_next_block
2485 */
2486#define HASH_NB_ALWAYS 1
2487
2488struct ext4_filename {
2489 const struct qstr *usr_fname;
2490 struct fscrypt_str disk_name;
2491 struct dx_hash_info hinfo;
2492#ifdef CONFIG_FS_ENCRYPTION
2493 struct fscrypt_str crypto_buf;
2494#endif
2495#if IS_ENABLED(CONFIG_UNICODE)
2496 struct qstr cf_name;
2497#endif
2498};
2499
2500#define fname_name(p) ((p)->disk_name.name)
2501#define fname_usr_name(p) ((p)->usr_fname->name)
2502#define fname_len(p) ((p)->disk_name.len)
2503
2504/*
2505 * Describe an inode's exact location on disk and in memory
2506 */
2507struct ext4_iloc
2508{
2509 struct buffer_head *bh;
2510 unsigned long offset;
2511 ext4_group_t block_group;
2512};
2513
2514static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2515{
2516 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2517}
2518
2519static inline bool ext4_is_quota_file(struct inode *inode)
2520{
2521 return IS_NOQUOTA(inode) &&
2522 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL);
2523}
2524
2525/*
2526 * This structure is stuffed into the struct file's private_data field
2527 * for directories. It is where we put information so that we can do
2528 * readdir operations in hash tree order.
2529 */
2530struct dir_private_info {
2531 struct rb_root root;
2532 struct rb_node *curr_node;
2533 struct fname *extra_fname;
2534 loff_t last_pos;
2535 __u32 curr_hash;
2536 __u32 curr_minor_hash;
2537 __u32 next_hash;
2538 u64 cookie;
2539 bool initialized;
2540};
2541
2542/* calculate the first block number of the group */
2543static inline ext4_fsblk_t
2544ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2545{
2546 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2547 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2548}
2549
2550/*
2551 * Special error return code only used by dx_probe() and its callers.
2552 */
2553#define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
2554
2555/* htree levels for ext4 */
2556#define EXT4_HTREE_LEVEL_COMPAT 2
2557#define EXT4_HTREE_LEVEL 3
2558
2559static inline int ext4_dir_htree_level(struct super_block *sb)
2560{
2561 return ext4_has_feature_largedir(sb) ?
2562 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT;
2563}
2564
2565/*
2566 * Timeout and state flag for lazy initialization inode thread.
2567 */
2568#define EXT4_DEF_LI_WAIT_MULT 10
2569#define EXT4_DEF_LI_MAX_START_DELAY 5
2570#define EXT4_LAZYINIT_QUIT 0x0001
2571#define EXT4_LAZYINIT_RUNNING 0x0002
2572
2573/*
2574 * Lazy inode table initialization info
2575 */
2576struct ext4_lazy_init {
2577 unsigned long li_state;
2578 struct list_head li_request_list;
2579 struct mutex li_list_mtx;
2580};
2581
2582enum ext4_li_mode {
2583 EXT4_LI_MODE_PREFETCH_BBITMAP,
2584 EXT4_LI_MODE_ITABLE,
2585};
2586
2587struct ext4_li_request {
2588 struct super_block *lr_super;
2589 enum ext4_li_mode lr_mode;
2590 ext4_group_t lr_first_not_zeroed;
2591 ext4_group_t lr_next_group;
2592 struct list_head lr_request;
2593 unsigned long lr_next_sched;
2594 unsigned long lr_timeout;
2595};
2596
2597struct ext4_features {
2598 struct kobject f_kobj;
2599 struct completion f_kobj_unregister;
2600};
2601
2602/*
2603 * This structure will be used for multiple mount protection. It will be
2604 * written into the block number saved in the s_mmp_block field in the
2605 * superblock. Programs that check MMP should assume that if
2606 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2607 * to use the filesystem, regardless of how old the timestamp is.
2608 */
2609#define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
2610#define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2611#define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
2612#define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
2613
2614struct mmp_struct {
2615 __le32 mmp_magic; /* Magic number for MMP */
2616 __le32 mmp_seq; /* Sequence no. updated periodically */
2617
2618 /*
2619 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2620 * purposes and do not affect the correctness of the algorithm
2621 */
2622 __le64 mmp_time; /* Time last updated */
2623 char mmp_nodename[64]; /* Node which last updated MMP block */
2624 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
2625
2626 /*
2627 * mmp_check_interval is used to verify if the MMP block has been
2628 * updated on the block device. The value is updated based on the
2629 * maximum time to write the MMP block during an update cycle.
2630 */
2631 __le16 mmp_check_interval;
2632
2633 __le16 mmp_pad1;
2634 __le32 mmp_pad2[226];
2635 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
2636};
2637
2638/* arguments passed to the mmp thread */
2639struct mmpd_data {
2640 struct buffer_head *bh; /* bh from initial read_mmp_block() */
2641 struct super_block *sb; /* super block of the fs */
2642};
2643
2644/*
2645 * Check interval multiplier
2646 * The MMP block is written every update interval and initially checked every
2647 * update interval x the multiplier (the value is then adapted based on the
2648 * write latency). The reason is that writes can be delayed under load and we
2649 * don't want readers to incorrectly assume that the filesystem is no longer
2650 * in use.
2651 */
2652#define EXT4_MMP_CHECK_MULT 2UL
2653
2654/*
2655 * Minimum interval for MMP checking in seconds.
2656 */
2657#define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
2658
2659/*
2660 * Maximum interval for MMP checking in seconds.
2661 */
2662#define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
2663
2664/*
2665 * Function prototypes
2666 */
2667
2668/*
2669 * Ok, these declarations are also in <linux/kernel.h> but none of the
2670 * ext4 source programs needs to include it so they are duplicated here.
2671 */
2672# define NORET_TYPE /**/
2673# define ATTRIB_NORET __attribute__((noreturn))
2674# define NORET_AND noreturn,
2675
2676/* bitmap.c */
2677extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2678void ext4_inode_bitmap_csum_set(struct super_block *sb,
2679 struct ext4_group_desc *gdp,
2680 struct buffer_head *bh);
2681int ext4_inode_bitmap_csum_verify(struct super_block *sb,
2682 struct ext4_group_desc *gdp,
2683 struct buffer_head *bh);
2684void ext4_block_bitmap_csum_set(struct super_block *sb,
2685 struct ext4_group_desc *gdp,
2686 struct buffer_head *bh);
2687int ext4_block_bitmap_csum_verify(struct super_block *sb,
2688 struct ext4_group_desc *gdp,
2689 struct buffer_head *bh);
2690
2691/* balloc.c */
2692extern void ext4_get_group_no_and_offset(struct super_block *sb,
2693 ext4_fsblk_t blocknr,
2694 ext4_group_t *blockgrpp,
2695 ext4_grpblk_t *offsetp);
2696extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2697 ext4_fsblk_t block);
2698
2699extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2700extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2701 ext4_group_t group);
2702extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2703 ext4_fsblk_t goal,
2704 unsigned int flags,
2705 unsigned long *count,
2706 int *errp);
2707extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2708 s64 nclusters, unsigned int flags);
2709extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2710extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2711 ext4_group_t block_group,
2712 struct buffer_head ** bh);
2713extern struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2714 ext4_group_t group);
2715extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2716
2717extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2718 ext4_group_t block_group,
2719 bool ignore_locked);
2720extern int ext4_wait_block_bitmap(struct super_block *sb,
2721 ext4_group_t block_group,
2722 struct buffer_head *bh);
2723extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2724 ext4_group_t block_group);
2725extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2726 ext4_group_t block_group,
2727 struct ext4_group_desc *gdp);
2728ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2729
2730#if IS_ENABLED(CONFIG_UNICODE)
2731extern int ext4_fname_setup_ci_filename(struct inode *dir,
2732 const struct qstr *iname,
2733 struct ext4_filename *fname);
2734
2735static inline void ext4_fname_free_ci_filename(struct ext4_filename *fname)
2736{
2737 kfree(fname->cf_name.name);
2738 fname->cf_name.name = NULL;
2739}
2740#else
2741static inline int ext4_fname_setup_ci_filename(struct inode *dir,
2742 const struct qstr *iname,
2743 struct ext4_filename *fname)
2744{
2745 return 0;
2746}
2747
2748static inline void ext4_fname_free_ci_filename(struct ext4_filename *fname)
2749{
2750}
2751#endif
2752
2753/* ext4 encryption related stuff goes here crypto.c */
2754#ifdef CONFIG_FS_ENCRYPTION
2755extern const struct fscrypt_operations ext4_cryptops;
2756
2757int ext4_fname_setup_filename(struct inode *dir, const struct qstr *iname,
2758 int lookup, struct ext4_filename *fname);
2759
2760int ext4_fname_prepare_lookup(struct inode *dir, struct dentry *dentry,
2761 struct ext4_filename *fname);
2762
2763void ext4_fname_free_filename(struct ext4_filename *fname);
2764
2765int ext4_ioctl_get_encryption_pwsalt(struct file *filp, void __user *arg);
2766
2767#else /* !CONFIG_FS_ENCRYPTION */
2768static inline int ext4_fname_setup_filename(struct inode *dir,
2769 const struct qstr *iname,
2770 int lookup,
2771 struct ext4_filename *fname)
2772{
2773 fname->usr_fname = iname;
2774 fname->disk_name.name = (unsigned char *) iname->name;
2775 fname->disk_name.len = iname->len;
2776
2777 return ext4_fname_setup_ci_filename(dir, iname, fname);
2778}
2779
2780static inline int ext4_fname_prepare_lookup(struct inode *dir,
2781 struct dentry *dentry,
2782 struct ext4_filename *fname)
2783{
2784 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname);
2785}
2786
2787static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2788{
2789 ext4_fname_free_ci_filename(fname);
2790}
2791
2792static inline int ext4_ioctl_get_encryption_pwsalt(struct file *filp,
2793 void __user *arg)
2794{
2795 return -EOPNOTSUPP;
2796}
2797#endif /* !CONFIG_FS_ENCRYPTION */
2798
2799/* dir.c */
2800extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2801 struct file *,
2802 struct ext4_dir_entry_2 *,
2803 struct buffer_head *, char *, int,
2804 unsigned int);
2805#define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2806 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2807 (de), (bh), (buf), (size), (offset)))
2808extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2809 __u32 minor_hash,
2810 struct ext4_dir_entry_2 *dirent,
2811 struct fscrypt_str *ent_name);
2812extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2813extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2814 struct buffer_head *bh,
2815 void *buf, int buf_size,
2816 struct ext4_filename *fname,
2817 struct ext4_dir_entry_2 **dest_de);
2818void ext4_insert_dentry(struct inode *dir, struct inode *inode,
2819 struct ext4_dir_entry_2 *de,
2820 int buf_size,
2821 struct ext4_filename *fname);
2822static inline void ext4_update_dx_flag(struct inode *inode)
2823{
2824 if (!ext4_has_feature_dir_index(inode->i_sb) &&
2825 ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) {
2826 /* ext4_iget() should have caught this... */
2827 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb));
2828 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2829 }
2830}
2831static const unsigned char ext4_filetype_table[] = {
2832 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2833};
2834
2835static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2836{
2837 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2838 return DT_UNKNOWN;
2839
2840 return ext4_filetype_table[filetype];
2841}
2842extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2843 void *buf, int buf_size);
2844
2845/* fsync.c */
2846extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2847
2848/* hash.c */
2849extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len,
2850 struct dx_hash_info *hinfo);
2851
2852/* ialloc.c */
2853extern int ext4_mark_inode_used(struct super_block *sb, int ino);
2854extern struct inode *__ext4_new_inode(struct mnt_idmap *, handle_t *,
2855 struct inode *, umode_t,
2856 const struct qstr *qstr, __u32 goal,
2857 uid_t *owner, __u32 i_flags,
2858 int handle_type, unsigned int line_no,
2859 int nblocks);
2860
2861#define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \
2862 __ext4_new_inode(&nop_mnt_idmap, (handle), (dir), (mode), (qstr), \
2863 (goal), (owner), i_flags, 0, 0, 0)
2864#define ext4_new_inode_start_handle(idmap, dir, mode, qstr, goal, owner, \
2865 type, nblocks) \
2866 __ext4_new_inode((idmap), NULL, (dir), (mode), (qstr), (goal), (owner), \
2867 0, (type), __LINE__, (nblocks))
2868
2869
2870extern void ext4_free_inode(handle_t *, struct inode *);
2871extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2872extern unsigned long ext4_count_free_inodes(struct super_block *);
2873extern unsigned long ext4_count_dirs(struct super_block *);
2874extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2875extern int ext4_init_inode_table(struct super_block *sb,
2876 ext4_group_t group, int barrier);
2877extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2878
2879/* fast_commit.c */
2880int ext4_fc_info_show(struct seq_file *seq, void *v);
2881void ext4_fc_init(struct super_block *sb, journal_t *journal);
2882void ext4_fc_init_inode(struct inode *inode);
2883void ext4_fc_track_range(handle_t *handle, struct inode *inode, ext4_lblk_t start,
2884 ext4_lblk_t end);
2885void __ext4_fc_track_unlink(handle_t *handle, struct inode *inode,
2886 struct dentry *dentry);
2887void __ext4_fc_track_link(handle_t *handle, struct inode *inode,
2888 struct dentry *dentry);
2889void ext4_fc_track_unlink(handle_t *handle, struct dentry *dentry);
2890void ext4_fc_track_link(handle_t *handle, struct dentry *dentry);
2891void __ext4_fc_track_create(handle_t *handle, struct inode *inode,
2892 struct dentry *dentry);
2893void ext4_fc_track_create(handle_t *handle, struct dentry *dentry);
2894void ext4_fc_track_inode(handle_t *handle, struct inode *inode);
2895void ext4_fc_mark_ineligible(struct super_block *sb, int reason, handle_t *handle);
2896void ext4_fc_start_update(struct inode *inode);
2897void ext4_fc_stop_update(struct inode *inode);
2898void ext4_fc_del(struct inode *inode);
2899bool ext4_fc_replay_check_excluded(struct super_block *sb, ext4_fsblk_t block);
2900void ext4_fc_replay_cleanup(struct super_block *sb);
2901int ext4_fc_commit(journal_t *journal, tid_t commit_tid);
2902int __init ext4_fc_init_dentry_cache(void);
2903void ext4_fc_destroy_dentry_cache(void);
2904int ext4_fc_record_regions(struct super_block *sb, int ino,
2905 ext4_lblk_t lblk, ext4_fsblk_t pblk,
2906 int len, int replay);
2907
2908/* mballoc.c */
2909extern const struct seq_operations ext4_mb_seq_groups_ops;
2910extern const struct seq_operations ext4_mb_seq_structs_summary_ops;
2911extern int ext4_seq_mb_stats_show(struct seq_file *seq, void *offset);
2912extern int ext4_mb_init(struct super_block *);
2913extern void ext4_mb_release(struct super_block *);
2914extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2915 struct ext4_allocation_request *, int *);
2916extern void ext4_discard_preallocations(struct inode *);
2917extern int __init ext4_init_mballoc(void);
2918extern void ext4_exit_mballoc(void);
2919extern ext4_group_t ext4_mb_prefetch(struct super_block *sb,
2920 ext4_group_t group,
2921 unsigned int nr, int *cnt);
2922extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group,
2923 unsigned int nr);
2924
2925extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2926 struct buffer_head *bh, ext4_fsblk_t block,
2927 unsigned long count, int flags);
2928extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2929 ext4_group_t ngroups);
2930extern int ext4_mb_add_groupinfo(struct super_block *sb,
2931 ext4_group_t i, struct ext4_group_desc *desc);
2932extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2933 ext4_fsblk_t block, unsigned long count);
2934extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2935extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid);
2936extern void ext4_mb_mark_bb(struct super_block *sb, ext4_fsblk_t block,
2937 int len, bool state);
2938static inline bool ext4_mb_cr_expensive(enum criteria cr)
2939{
2940 return cr >= CR_GOAL_LEN_SLOW;
2941}
2942
2943/* inode.c */
2944void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw,
2945 struct ext4_inode_info *ei);
2946int ext4_inode_is_fast_symlink(struct inode *inode);
2947struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2948struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2949int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count,
2950 bool wait, struct buffer_head **bhs);
2951int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2952 struct buffer_head *bh_result, int create);
2953int ext4_get_block(struct inode *inode, sector_t iblock,
2954 struct buffer_head *bh_result, int create);
2955int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2956 struct buffer_head *bh, int create);
2957int ext4_walk_page_buffers(handle_t *handle,
2958 struct inode *inode,
2959 struct buffer_head *head,
2960 unsigned from,
2961 unsigned to,
2962 int *partial,
2963 int (*fn)(handle_t *handle, struct inode *inode,
2964 struct buffer_head *bh));
2965int do_journal_get_write_access(handle_t *handle, struct inode *inode,
2966 struct buffer_head *bh);
2967#define FALL_BACK_TO_NONDELALLOC 1
2968#define CONVERT_INLINE_DATA 2
2969
2970typedef enum {
2971 EXT4_IGET_NORMAL = 0,
2972 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */
2973 EXT4_IGET_HANDLE = 0x0002, /* Inode # is from a handle */
2974 EXT4_IGET_BAD = 0x0004, /* Allow to iget a bad inode */
2975 EXT4_IGET_EA_INODE = 0x0008 /* Inode should contain an EA value */
2976} ext4_iget_flags;
2977
2978extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
2979 ext4_iget_flags flags, const char *function,
2980 unsigned int line);
2981
2982#define ext4_iget(sb, ino, flags) \
2983 __ext4_iget((sb), (ino), (flags), __func__, __LINE__)
2984
2985extern int ext4_write_inode(struct inode *, struct writeback_control *);
2986extern int ext4_setattr(struct mnt_idmap *, struct dentry *,
2987 struct iattr *);
2988extern u32 ext4_dio_alignment(struct inode *inode);
2989extern int ext4_getattr(struct mnt_idmap *, const struct path *,
2990 struct kstat *, u32, unsigned int);
2991extern void ext4_evict_inode(struct inode *);
2992extern void ext4_clear_inode(struct inode *);
2993extern int ext4_file_getattr(struct mnt_idmap *, const struct path *,
2994 struct kstat *, u32, unsigned int);
2995extern void ext4_dirty_inode(struct inode *, int);
2996extern int ext4_change_inode_journal_flag(struct inode *, int);
2997extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2998extern int ext4_get_fc_inode_loc(struct super_block *sb, unsigned long ino,
2999 struct ext4_iloc *iloc);
3000extern int ext4_inode_attach_jinode(struct inode *inode);
3001extern int ext4_can_truncate(struct inode *inode);
3002extern int ext4_truncate(struct inode *);
3003extern int ext4_break_layouts(struct inode *);
3004extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length);
3005extern void ext4_set_inode_flags(struct inode *, bool init);
3006extern int ext4_alloc_da_blocks(struct inode *inode);
3007extern void ext4_set_aops(struct inode *inode);
3008extern int ext4_writepage_trans_blocks(struct inode *);
3009extern int ext4_normal_submit_inode_data_buffers(struct jbd2_inode *jinode);
3010extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
3011extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
3012 loff_t lstart, loff_t lend);
3013extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf);
3014extern qsize_t *ext4_get_reserved_space(struct inode *inode);
3015extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
3016extern void ext4_da_release_space(struct inode *inode, int to_free);
3017extern void ext4_da_update_reserve_space(struct inode *inode,
3018 int used, int quota_claim);
3019extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
3020 ext4_fsblk_t pblk, ext4_lblk_t len);
3021
3022/* indirect.c */
3023extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
3024 struct ext4_map_blocks *map, int flags);
3025extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
3026extern void ext4_ind_truncate(handle_t *, struct inode *inode);
3027extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
3028 ext4_lblk_t start, ext4_lblk_t end);
3029
3030/* ioctl.c */
3031extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
3032extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
3033int ext4_fileattr_set(struct mnt_idmap *idmap,
3034 struct dentry *dentry, struct fileattr *fa);
3035int ext4_fileattr_get(struct dentry *dentry, struct fileattr *fa);
3036extern void ext4_reset_inode_seed(struct inode *inode);
3037int ext4_update_overhead(struct super_block *sb, bool force);
3038int ext4_force_shutdown(struct super_block *sb, u32 flags);
3039
3040/* migrate.c */
3041extern int ext4_ext_migrate(struct inode *);
3042extern int ext4_ind_migrate(struct inode *inode);
3043
3044/* namei.c */
3045extern int ext4_init_new_dir(handle_t *handle, struct inode *dir,
3046 struct inode *inode);
3047extern int ext4_dirblock_csum_verify(struct inode *inode,
3048 struct buffer_head *bh);
3049extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
3050 __u32 start_minor_hash, __u32 *next_hash);
3051extern int ext4_search_dir(struct buffer_head *bh,
3052 char *search_buf,
3053 int buf_size,
3054 struct inode *dir,
3055 struct ext4_filename *fname,
3056 unsigned int offset,
3057 struct ext4_dir_entry_2 **res_dir);
3058extern int ext4_generic_delete_entry(struct inode *dir,
3059 struct ext4_dir_entry_2 *de_del,
3060 struct buffer_head *bh,
3061 void *entry_buf,
3062 int buf_size,
3063 int csum_size);
3064extern bool ext4_empty_dir(struct inode *inode);
3065
3066/* resize.c */
3067extern void ext4_kvfree_array_rcu(void *to_free);
3068extern int ext4_group_add(struct super_block *sb,
3069 struct ext4_new_group_data *input);
3070extern int ext4_group_extend(struct super_block *sb,
3071 struct ext4_super_block *es,
3072 ext4_fsblk_t n_blocks_count);
3073extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
3074extern unsigned int ext4_list_backups(struct super_block *sb,
3075 unsigned int *three, unsigned int *five,
3076 unsigned int *seven);
3077
3078/* super.c */
3079extern struct buffer_head *ext4_sb_bread(struct super_block *sb,
3080 sector_t block, blk_opf_t op_flags);
3081extern struct buffer_head *ext4_sb_bread_unmovable(struct super_block *sb,
3082 sector_t block);
3083extern void ext4_read_bh_nowait(struct buffer_head *bh, blk_opf_t op_flags,
3084 bh_end_io_t *end_io, bool simu_fail);
3085extern int ext4_read_bh(struct buffer_head *bh, blk_opf_t op_flags,
3086 bh_end_io_t *end_io, bool simu_fail);
3087extern int ext4_read_bh_lock(struct buffer_head *bh, blk_opf_t op_flags, bool wait);
3088extern void ext4_sb_breadahead_unmovable(struct super_block *sb, sector_t block);
3089extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
3090extern int ext4_calculate_overhead(struct super_block *sb);
3091extern __le32 ext4_superblock_csum(struct super_block *sb,
3092 struct ext4_super_block *es);
3093extern void ext4_superblock_csum_set(struct super_block *sb);
3094extern int ext4_alloc_flex_bg_array(struct super_block *sb,
3095 ext4_group_t ngroup);
3096extern const char *ext4_decode_error(struct super_block *sb, int errno,
3097 char nbuf[16]);
3098extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb,
3099 ext4_group_t block_group,
3100 unsigned int flags);
3101extern unsigned int ext4_num_base_meta_blocks(struct super_block *sb,
3102 ext4_group_t block_group);
3103
3104extern __printf(7, 8)
3105void __ext4_error(struct super_block *, const char *, unsigned int, bool,
3106 int, __u64, const char *, ...);
3107extern __printf(6, 7)
3108void __ext4_error_inode(struct inode *, const char *, unsigned int,
3109 ext4_fsblk_t, int, const char *, ...);
3110extern __printf(5, 6)
3111void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
3112 const char *, ...);
3113extern void __ext4_std_error(struct super_block *, const char *,
3114 unsigned int, int);
3115extern __printf(4, 5)
3116void __ext4_warning(struct super_block *, const char *, unsigned int,
3117 const char *, ...);
3118extern __printf(4, 5)
3119void __ext4_warning_inode(const struct inode *inode, const char *function,
3120 unsigned int line, const char *fmt, ...);
3121extern __printf(3, 4)
3122void __ext4_msg(struct super_block *, const char *, const char *, ...);
3123extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
3124 const char *, unsigned int, const char *);
3125extern __printf(7, 8)
3126void __ext4_grp_locked_error(const char *, unsigned int,
3127 struct super_block *, ext4_group_t,
3128 unsigned long, ext4_fsblk_t,
3129 const char *, ...);
3130
3131#define EXT4_ERROR_INODE(inode, fmt, a...) \
3132 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
3133
3134#define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...) \
3135 __ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a)
3136
3137#define ext4_error_inode_block(inode, block, err, fmt, a...) \
3138 __ext4_error_inode((inode), __func__, __LINE__, (block), (err), \
3139 (fmt), ## a)
3140
3141#define EXT4_ERROR_FILE(file, block, fmt, a...) \
3142 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
3143
3144#define ext4_abort(sb, err, fmt, a...) \
3145 __ext4_error((sb), __func__, __LINE__, true, (err), 0, (fmt), ## a)
3146
3147#ifdef CONFIG_PRINTK
3148
3149#define ext4_error_inode(inode, func, line, block, fmt, ...) \
3150 __ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__)
3151#define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \
3152 __ext4_error_inode((inode), (func), (line), (block), \
3153 (err), (fmt), ##__VA_ARGS__)
3154#define ext4_error_file(file, func, line, block, fmt, ...) \
3155 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
3156#define ext4_error(sb, fmt, ...) \
3157 __ext4_error((sb), __func__, __LINE__, false, 0, 0, (fmt), \
3158 ##__VA_ARGS__)
3159#define ext4_error_err(sb, err, fmt, ...) \
3160 __ext4_error((sb), __func__, __LINE__, false, (err), 0, (fmt), \
3161 ##__VA_ARGS__)
3162#define ext4_warning(sb, fmt, ...) \
3163 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
3164#define ext4_warning_inode(inode, fmt, ...) \
3165 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
3166#define ext4_msg(sb, level, fmt, ...) \
3167 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
3168#define dump_mmp_msg(sb, mmp, msg) \
3169 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
3170#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
3171 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
3172 fmt, ##__VA_ARGS__)
3173
3174#else
3175
3176#define ext4_error_inode(inode, func, line, block, fmt, ...) \
3177do { \
3178 no_printk(fmt, ##__VA_ARGS__); \
3179 __ext4_error_inode(inode, "", 0, block, 0, " "); \
3180} while (0)
3181#define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \
3182do { \
3183 no_printk(fmt, ##__VA_ARGS__); \
3184 __ext4_error_inode(inode, "", 0, block, err, " "); \
3185} while (0)
3186#define ext4_error_file(file, func, line, block, fmt, ...) \
3187do { \
3188 no_printk(fmt, ##__VA_ARGS__); \
3189 __ext4_error_file(file, "", 0, block, " "); \
3190} while (0)
3191#define ext4_error(sb, fmt, ...) \
3192do { \
3193 no_printk(fmt, ##__VA_ARGS__); \
3194 __ext4_error(sb, "", 0, false, 0, 0, " "); \
3195} while (0)
3196#define ext4_error_err(sb, err, fmt, ...) \
3197do { \
3198 no_printk(fmt, ##__VA_ARGS__); \
3199 __ext4_error(sb, "", 0, false, err, 0, " "); \
3200} while (0)
3201#define ext4_warning(sb, fmt, ...) \
3202do { \
3203 no_printk(fmt, ##__VA_ARGS__); \
3204 __ext4_warning(sb, "", 0, " "); \
3205} while (0)
3206#define ext4_warning_inode(inode, fmt, ...) \
3207do { \
3208 no_printk(fmt, ##__VA_ARGS__); \
3209 __ext4_warning_inode(inode, "", 0, " "); \
3210} while (0)
3211#define ext4_msg(sb, level, fmt, ...) \
3212do { \
3213 no_printk(fmt, ##__VA_ARGS__); \
3214 __ext4_msg(sb, "", " "); \
3215} while (0)
3216#define dump_mmp_msg(sb, mmp, msg) \
3217 __dump_mmp_msg(sb, mmp, "", 0, "")
3218#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
3219do { \
3220 no_printk(fmt, ##__VA_ARGS__); \
3221 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
3222} while (0)
3223
3224#endif
3225
3226extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
3227 struct ext4_group_desc *bg);
3228extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
3229 struct ext4_group_desc *bg);
3230extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
3231 struct ext4_group_desc *bg);
3232extern __u32 ext4_free_group_clusters(struct super_block *sb,
3233 struct ext4_group_desc *bg);
3234extern __u32 ext4_free_inodes_count(struct super_block *sb,
3235 struct ext4_group_desc *bg);
3236extern __u32 ext4_used_dirs_count(struct super_block *sb,
3237 struct ext4_group_desc *bg);
3238extern __u32 ext4_itable_unused_count(struct super_block *sb,
3239 struct ext4_group_desc *bg);
3240extern void ext4_block_bitmap_set(struct super_block *sb,
3241 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3242extern void ext4_inode_bitmap_set(struct super_block *sb,
3243 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3244extern void ext4_inode_table_set(struct super_block *sb,
3245 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3246extern void ext4_free_group_clusters_set(struct super_block *sb,
3247 struct ext4_group_desc *bg,
3248 __u32 count);
3249extern void ext4_free_inodes_set(struct super_block *sb,
3250 struct ext4_group_desc *bg, __u32 count);
3251extern void ext4_used_dirs_set(struct super_block *sb,
3252 struct ext4_group_desc *bg, __u32 count);
3253extern void ext4_itable_unused_set(struct super_block *sb,
3254 struct ext4_group_desc *bg, __u32 count);
3255extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
3256 struct ext4_group_desc *gdp);
3257extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
3258 struct ext4_group_desc *gdp);
3259extern int ext4_register_li_request(struct super_block *sb,
3260 ext4_group_t first_not_zeroed);
3261
3262static inline int ext4_has_metadata_csum(struct super_block *sb)
3263{
3264 return ext4_has_feature_metadata_csum(sb);
3265}
3266
3267static inline int ext4_has_group_desc_csum(struct super_block *sb)
3268{
3269 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb);
3270}
3271
3272#define ext4_read_incompat_64bit_val(es, name) \
3273 (((es)->s_feature_incompat & cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT) \
3274 ? (ext4_fsblk_t)le32_to_cpu(es->name##_hi) << 32 : 0) | \
3275 le32_to_cpu(es->name##_lo))
3276
3277static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
3278{
3279 return ext4_read_incompat_64bit_val(es, s_blocks_count);
3280}
3281
3282static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
3283{
3284 return ext4_read_incompat_64bit_val(es, s_r_blocks_count);
3285}
3286
3287static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
3288{
3289 return ext4_read_incompat_64bit_val(es, s_free_blocks_count);
3290}
3291
3292static inline void ext4_blocks_count_set(struct ext4_super_block *es,
3293 ext4_fsblk_t blk)
3294{
3295 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
3296 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
3297}
3298
3299static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
3300 ext4_fsblk_t blk)
3301{
3302 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
3303 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
3304}
3305
3306static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
3307 ext4_fsblk_t blk)
3308{
3309 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
3310 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
3311}
3312
3313static inline loff_t ext4_isize(struct super_block *sb,
3314 struct ext4_inode *raw_inode)
3315{
3316 if (ext4_has_feature_largedir(sb) ||
3317 S_ISREG(le16_to_cpu(raw_inode->i_mode)))
3318 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
3319 le32_to_cpu(raw_inode->i_size_lo);
3320
3321 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
3322}
3323
3324static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
3325{
3326 raw_inode->i_size_lo = cpu_to_le32(i_size);
3327 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
3328}
3329
3330/*
3331 * Reading s_groups_count requires using smp_rmb() afterwards. See
3332 * the locking protocol documented in the comments of ext4_group_add()
3333 * in resize.c
3334 */
3335static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
3336{
3337 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
3338
3339 smp_rmb();
3340 return ngroups;
3341}
3342
3343static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
3344 ext4_group_t block_group)
3345{
3346 return block_group >> sbi->s_log_groups_per_flex;
3347}
3348
3349static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
3350{
3351 return 1 << sbi->s_log_groups_per_flex;
3352}
3353
3354#define ext4_std_error(sb, errno) \
3355do { \
3356 if ((errno)) \
3357 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
3358} while (0)
3359
3360#ifdef CONFIG_SMP
3361/* Each CPU can accumulate percpu_counter_batch clusters in their local
3362 * counters. So we need to make sure we have free clusters more
3363 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
3364 */
3365#define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
3366#else
3367#define EXT4_FREECLUSTERS_WATERMARK 0
3368#endif
3369
3370/* Update i_disksize. Requires i_rwsem to avoid races with truncate */
3371static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
3372{
3373 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
3374 !inode_is_locked(inode));
3375 down_write(&EXT4_I(inode)->i_data_sem);
3376 if (newsize > EXT4_I(inode)->i_disksize)
3377 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize);
3378 up_write(&EXT4_I(inode)->i_data_sem);
3379}
3380
3381/* Update i_size, i_disksize. Requires i_rwsem to avoid races with truncate */
3382static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
3383{
3384 int changed = 0;
3385
3386 if (newsize > inode->i_size) {
3387 i_size_write(inode, newsize);
3388 changed = 1;
3389 }
3390 if (newsize > EXT4_I(inode)->i_disksize) {
3391 ext4_update_i_disksize(inode, newsize);
3392 changed |= 2;
3393 }
3394 return changed;
3395}
3396
3397int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
3398 loff_t len);
3399
3400struct ext4_group_info {
3401 unsigned long bb_state;
3402#ifdef AGGRESSIVE_CHECK
3403 unsigned long bb_check_counter;
3404#endif
3405 struct rb_root bb_free_root;
3406 ext4_grpblk_t bb_first_free; /* first free block */
3407 ext4_grpblk_t bb_free; /* total free blocks */
3408 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
3409 int bb_avg_fragment_size_order; /* order of average
3410 fragment in BG */
3411 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
3412 ext4_group_t bb_group; /* Group number */
3413 struct list_head bb_prealloc_list;
3414#ifdef DOUBLE_CHECK
3415 void *bb_bitmap;
3416#endif
3417 struct rw_semaphore alloc_sem;
3418 struct list_head bb_avg_fragment_size_node;
3419 struct list_head bb_largest_free_order_node;
3420 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
3421 * regions, index is order.
3422 * bb_counters[3] = 5 means
3423 * 5 free 8-block regions. */
3424};
3425
3426#define EXT4_GROUP_INFO_NEED_INIT_BIT 0
3427#define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
3428#define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
3429#define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
3430#define EXT4_GROUP_INFO_BBITMAP_CORRUPT \
3431 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT)
3432#define EXT4_GROUP_INFO_IBITMAP_CORRUPT \
3433 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT)
3434#define EXT4_GROUP_INFO_BBITMAP_READ_BIT 4
3435
3436#define EXT4_MB_GRP_NEED_INIT(grp) \
3437 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
3438#define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
3439 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3440#define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
3441 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3442
3443#define EXT4_MB_GRP_WAS_TRIMMED(grp) \
3444 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3445#define EXT4_MB_GRP_SET_TRIMMED(grp) \
3446 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3447#define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
3448 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3449#define EXT4_MB_GRP_TEST_AND_SET_READ(grp) \
3450 (test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state)))
3451
3452#define EXT4_MAX_CONTENTION 8
3453#define EXT4_CONTENTION_THRESHOLD 2
3454
3455static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
3456 ext4_group_t group)
3457{
3458 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
3459}
3460
3461/*
3462 * Returns true if the filesystem is busy enough that attempts to
3463 * access the block group locks has run into contention.
3464 */
3465static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
3466{
3467 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
3468}
3469
3470static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
3471{
3472 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
3473 if (spin_trylock(lock))
3474 /*
3475 * We're able to grab the lock right away, so drop the
3476 * lock contention counter.
3477 */
3478 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
3479 else {
3480 /*
3481 * The lock is busy, so bump the contention counter,
3482 * and then wait on the spin lock.
3483 */
3484 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
3485 EXT4_MAX_CONTENTION);
3486 spin_lock(lock);
3487 }
3488}
3489
3490static inline void ext4_unlock_group(struct super_block *sb,
3491 ext4_group_t group)
3492{
3493 spin_unlock(ext4_group_lock_ptr(sb, group));
3494}
3495
3496#ifdef CONFIG_QUOTA
3497static inline bool ext4_quota_capable(struct super_block *sb)
3498{
3499 return (test_opt(sb, QUOTA) || ext4_has_feature_quota(sb));
3500}
3501
3502static inline bool ext4_is_quota_journalled(struct super_block *sb)
3503{
3504 struct ext4_sb_info *sbi = EXT4_SB(sb);
3505
3506 return (ext4_has_feature_quota(sb) ||
3507 sbi->s_qf_names[USRQUOTA] || sbi->s_qf_names[GRPQUOTA]);
3508}
3509int ext4_enable_quotas(struct super_block *sb);
3510#endif
3511
3512/*
3513 * Block validity checking
3514 */
3515#define ext4_check_indirect_blockref(inode, bh) \
3516 ext4_check_blockref(__func__, __LINE__, inode, \
3517 (__le32 *)(bh)->b_data, \
3518 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
3519
3520#define ext4_ind_check_inode(inode) \
3521 ext4_check_blockref(__func__, __LINE__, inode, \
3522 EXT4_I(inode)->i_data, \
3523 EXT4_NDIR_BLOCKS)
3524
3525/*
3526 * Inodes and files operations
3527 */
3528
3529/* dir.c */
3530extern const struct file_operations ext4_dir_operations;
3531
3532/* file.c */
3533extern const struct inode_operations ext4_file_inode_operations;
3534extern const struct file_operations ext4_file_operations;
3535extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
3536
3537/* inline.c */
3538extern int ext4_get_max_inline_size(struct inode *inode);
3539extern int ext4_find_inline_data_nolock(struct inode *inode);
3540extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
3541
3542int ext4_readpage_inline(struct inode *inode, struct folio *folio);
3543extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3544 struct inode *inode,
3545 loff_t pos, unsigned len,
3546 struct folio **foliop);
3547int ext4_write_inline_data_end(struct inode *inode, loff_t pos, unsigned len,
3548 unsigned copied, struct folio *folio);
3549extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
3550 struct inode *inode,
3551 loff_t pos, unsigned len,
3552 struct folio **foliop,
3553 void **fsdata);
3554extern int ext4_try_add_inline_entry(handle_t *handle,
3555 struct ext4_filename *fname,
3556 struct inode *dir, struct inode *inode);
3557extern int ext4_try_create_inline_dir(handle_t *handle,
3558 struct inode *parent,
3559 struct inode *inode);
3560extern int ext4_read_inline_dir(struct file *filp,
3561 struct dir_context *ctx,
3562 int *has_inline_data);
3563extern int ext4_inlinedir_to_tree(struct file *dir_file,
3564 struct inode *dir, ext4_lblk_t block,
3565 struct dx_hash_info *hinfo,
3566 __u32 start_hash, __u32 start_minor_hash,
3567 int *has_inline_data);
3568extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3569 struct ext4_filename *fname,
3570 struct ext4_dir_entry_2 **res_dir,
3571 int *has_inline_data);
3572extern int ext4_delete_inline_entry(handle_t *handle,
3573 struct inode *dir,
3574 struct ext4_dir_entry_2 *de_del,
3575 struct buffer_head *bh,
3576 int *has_inline_data);
3577extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3578extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3579 struct ext4_dir_entry_2 **parent_de,
3580 int *retval);
3581extern void *ext4_read_inline_link(struct inode *inode);
3582
3583struct iomap;
3584extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap);
3585
3586extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3587
3588extern int ext4_convert_inline_data(struct inode *inode);
3589
3590static inline int ext4_has_inline_data(struct inode *inode)
3591{
3592 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3593 EXT4_I(inode)->i_inline_off;
3594}
3595
3596/* namei.c */
3597extern const struct inode_operations ext4_dir_inode_operations;
3598extern const struct inode_operations ext4_special_inode_operations;
3599extern struct dentry *ext4_get_parent(struct dentry *child);
3600extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3601 struct ext4_dir_entry_2 *de,
3602 int blocksize, int csum_size,
3603 unsigned int parent_ino, int dotdot_real_len);
3604extern void ext4_initialize_dirent_tail(struct buffer_head *bh,
3605 unsigned int blocksize);
3606extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode,
3607 struct buffer_head *bh);
3608extern int __ext4_unlink(struct inode *dir, const struct qstr *d_name,
3609 struct inode *inode, struct dentry *dentry);
3610extern int __ext4_link(struct inode *dir, struct inode *inode,
3611 struct dentry *dentry);
3612
3613#define S_SHIFT 12
3614static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = {
3615 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
3616 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
3617 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
3618 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
3619 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
3620 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
3621 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
3622};
3623
3624static inline void ext4_set_de_type(struct super_block *sb,
3625 struct ext4_dir_entry_2 *de,
3626 umode_t mode) {
3627 if (ext4_has_feature_filetype(sb))
3628 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3629}
3630
3631/* readpages.c */
3632extern int ext4_mpage_readpages(struct inode *inode,
3633 struct readahead_control *rac, struct folio *folio);
3634extern int __init ext4_init_post_read_processing(void);
3635extern void ext4_exit_post_read_processing(void);
3636
3637/* symlink.c */
3638extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3639extern const struct inode_operations ext4_symlink_inode_operations;
3640extern const struct inode_operations ext4_fast_symlink_inode_operations;
3641
3642/* sysfs.c */
3643extern void ext4_notify_error_sysfs(struct ext4_sb_info *sbi);
3644extern int ext4_register_sysfs(struct super_block *sb);
3645extern void ext4_unregister_sysfs(struct super_block *sb);
3646extern int __init ext4_init_sysfs(void);
3647extern void ext4_exit_sysfs(void);
3648
3649/* block_validity */
3650extern void ext4_release_system_zone(struct super_block *sb);
3651extern int ext4_setup_system_zone(struct super_block *sb);
3652extern int __init ext4_init_system_zone(void);
3653extern void ext4_exit_system_zone(void);
3654extern int ext4_inode_block_valid(struct inode *inode,
3655 ext4_fsblk_t start_blk,
3656 unsigned int count);
3657extern int ext4_check_blockref(const char *, unsigned int,
3658 struct inode *, __le32 *, unsigned int);
3659extern int ext4_sb_block_valid(struct super_block *sb, struct inode *inode,
3660 ext4_fsblk_t start_blk, unsigned int count);
3661
3662
3663/* extents.c */
3664struct ext4_ext_path;
3665struct ext4_extent;
3666
3667/*
3668 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3669 * __le32.
3670 */
3671#define EXT_MAX_BLOCKS 0xffffffff
3672
3673extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode);
3674extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3675extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3676 struct ext4_map_blocks *map, int flags);
3677extern int ext4_ext_truncate(handle_t *, struct inode *);
3678extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3679 ext4_lblk_t end);
3680extern void ext4_ext_init(struct super_block *);
3681extern void ext4_ext_release(struct super_block *);
3682extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3683 loff_t len);
3684extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3685 loff_t offset, ssize_t len);
3686extern int ext4_convert_unwritten_io_end_vec(handle_t *handle,
3687 ext4_io_end_t *io_end);
3688extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3689 struct ext4_map_blocks *map, int flags);
3690extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3691 int num,
3692 struct ext4_ext_path *path);
3693extern struct ext4_ext_path *ext4_ext_insert_extent(
3694 handle_t *handle, struct inode *inode,
3695 struct ext4_ext_path *path,
3696 struct ext4_extent *newext, int gb_flags);
3697extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3698 struct ext4_ext_path *,
3699 int flags);
3700extern void ext4_free_ext_path(struct ext4_ext_path *);
3701extern int ext4_ext_check_inode(struct inode *inode);
3702extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3703extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3704 __u64 start, __u64 len);
3705extern int ext4_get_es_cache(struct inode *inode,
3706 struct fiemap_extent_info *fieinfo,
3707 __u64 start, __u64 len);
3708extern int ext4_ext_precache(struct inode *inode);
3709extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3710 struct inode *inode2, ext4_lblk_t lblk1,
3711 ext4_lblk_t lblk2, ext4_lblk_t count,
3712 int mark_unwritten,int *err);
3713extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu);
3714extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode,
3715 int check_cred, int restart_cred,
3716 int revoke_cred);
3717extern void ext4_ext_replay_shrink_inode(struct inode *inode, ext4_lblk_t end);
3718extern int ext4_ext_replay_set_iblocks(struct inode *inode);
3719extern int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start,
3720 int len, int unwritten, ext4_fsblk_t pblk);
3721extern int ext4_ext_clear_bb(struct inode *inode);
3722
3723
3724/* move_extent.c */
3725extern void ext4_double_down_write_data_sem(struct inode *first,
3726 struct inode *second);
3727extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3728 struct inode *donor_inode);
3729extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3730 __u64 start_orig, __u64 start_donor,
3731 __u64 len, __u64 *moved_len);
3732
3733/* page-io.c */
3734extern int __init ext4_init_pageio(void);
3735extern void ext4_exit_pageio(void);
3736extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3737extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3738extern int ext4_put_io_end(ext4_io_end_t *io_end);
3739extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3740extern void ext4_io_submit_init(struct ext4_io_submit *io,
3741 struct writeback_control *wbc);
3742extern void ext4_end_io_rsv_work(struct work_struct *work);
3743extern void ext4_io_submit(struct ext4_io_submit *io);
3744int ext4_bio_write_folio(struct ext4_io_submit *io, struct folio *page,
3745 size_t len);
3746extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end);
3747extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end);
3748
3749/* mmp.c */
3750extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3751
3752/* mmp.c */
3753extern void ext4_stop_mmpd(struct ext4_sb_info *sbi);
3754
3755/* verity.c */
3756extern const struct fsverity_operations ext4_verityops;
3757
3758/* orphan.c */
3759extern int ext4_orphan_add(handle_t *, struct inode *);
3760extern int ext4_orphan_del(handle_t *, struct inode *);
3761extern void ext4_orphan_cleanup(struct super_block *sb,
3762 struct ext4_super_block *es);
3763extern void ext4_release_orphan_info(struct super_block *sb);
3764extern int ext4_init_orphan_info(struct super_block *sb);
3765extern int ext4_orphan_file_empty(struct super_block *sb);
3766extern void ext4_orphan_file_block_trigger(
3767 struct jbd2_buffer_trigger_type *triggers,
3768 struct buffer_head *bh,
3769 void *data, size_t size);
3770
3771/*
3772 * Add new method to test whether block and inode bitmaps are properly
3773 * initialized. With uninit_bg reading the block from disk is not enough
3774 * to mark the bitmap uptodate. We need to also zero-out the bitmap
3775 */
3776#define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3777
3778static inline int bitmap_uptodate(struct buffer_head *bh)
3779{
3780 return (buffer_uptodate(bh) &&
3781 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3782}
3783static inline void set_bitmap_uptodate(struct buffer_head *bh)
3784{
3785 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3786}
3787
3788/* For ioend & aio unwritten conversion wait queues */
3789#define EXT4_WQ_HASH_SZ 37
3790#define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
3791 EXT4_WQ_HASH_SZ])
3792extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3793
3794extern int ext4_resize_begin(struct super_block *sb);
3795extern int ext4_resize_end(struct super_block *sb, bool update_backups);
3796
3797static inline void ext4_set_io_unwritten_flag(struct inode *inode,
3798 struct ext4_io_end *io_end)
3799{
3800 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
3801 io_end->flag |= EXT4_IO_END_UNWRITTEN;
3802 atomic_inc(&EXT4_I(inode)->i_unwritten);
3803 }
3804}
3805
3806static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3807{
3808 struct inode *inode = io_end->inode;
3809
3810 if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
3811 io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3812 /* Wake up anyone waiting on unwritten extent conversion */
3813 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
3814 wake_up_all(ext4_ioend_wq(inode));
3815 }
3816}
3817
3818extern const struct iomap_ops ext4_iomap_ops;
3819extern const struct iomap_ops ext4_iomap_overwrite_ops;
3820extern const struct iomap_ops ext4_iomap_report_ops;
3821
3822static inline int ext4_buffer_uptodate(struct buffer_head *bh)
3823{
3824 /*
3825 * If the buffer has the write error flag, we have failed
3826 * to write out data in the block. In this case, we don't
3827 * have to read the block because we may read the old data
3828 * successfully.
3829 */
3830 if (buffer_write_io_error(bh))
3831 set_buffer_uptodate(bh);
3832 return buffer_uptodate(bh);
3833}
3834
3835static inline bool ext4_inode_can_atomic_write(struct inode *inode)
3836{
3837
3838 return S_ISREG(inode->i_mode) && EXT4_SB(inode->i_sb)->s_awu_min > 0;
3839}
3840
3841extern int ext4_block_write_begin(handle_t *handle, struct folio *folio,
3842 loff_t pos, unsigned len,
3843 get_block_t *get_block);
3844#endif /* __KERNEL__ */
3845
3846#define EFSBADCRC EBADMSG /* Bad CRC detected */
3847#define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
3848
3849#endif /* _EXT4_H */