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 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/* first blocks in the file */
189#define EXT4_MB_HINT_FIRST 0x0008
190/* data is being allocated */
191#define EXT4_MB_HINT_DATA 0x0020
192/* don't preallocate (for tails) */
193#define EXT4_MB_HINT_NOPREALLOC 0x0040
194/* allocate for locality group */
195#define EXT4_MB_HINT_GROUP_ALLOC 0x0080
196/* allocate goal blocks or none */
197#define EXT4_MB_HINT_GOAL_ONLY 0x0100
198/* goal is meaningful */
199#define EXT4_MB_HINT_TRY_GOAL 0x0200
200/* blocks already pre-reserved by delayed allocation */
201#define EXT4_MB_DELALLOC_RESERVED 0x0400
202/* We are doing stream allocation */
203#define EXT4_MB_STREAM_ALLOC 0x0800
204/* Use reserved root blocks if needed */
205#define EXT4_MB_USE_ROOT_BLOCKS 0x1000
206/* Use blocks from reserved pool */
207#define EXT4_MB_USE_RESERVED 0x2000
208/* Do strict check for free blocks while retrying block allocation */
209#define EXT4_MB_STRICT_CHECK 0x4000
210
211struct ext4_allocation_request {
212 /* target inode for block we're allocating */
213 struct inode *inode;
214 /* how many blocks we want to allocate */
215 unsigned int len;
216 /* logical block in target inode */
217 ext4_lblk_t logical;
218 /* the closest logical allocated block to the left */
219 ext4_lblk_t lleft;
220 /* the closest logical allocated block to the right */
221 ext4_lblk_t lright;
222 /* phys. target (a hint) */
223 ext4_fsblk_t goal;
224 /* phys. block for the closest logical allocated block to the left */
225 ext4_fsblk_t pleft;
226 /* phys. block for the closest logical allocated block to the right */
227 ext4_fsblk_t pright;
228 /* flags. see above EXT4_MB_HINT_* */
229 unsigned int flags;
230};
231
232/*
233 * Logical to physical block mapping, used by ext4_map_blocks()
234 *
235 * This structure is used to pass requests into ext4_map_blocks() as
236 * well as to store the information returned by ext4_map_blocks(). It
237 * takes less room on the stack than a struct buffer_head.
238 */
239#define EXT4_MAP_NEW BIT(BH_New)
240#define EXT4_MAP_MAPPED BIT(BH_Mapped)
241#define EXT4_MAP_UNWRITTEN BIT(BH_Unwritten)
242#define EXT4_MAP_BOUNDARY BIT(BH_Boundary)
243#define EXT4_MAP_DELAYED BIT(BH_Delay)
244/*
245 * This is for use in ext4_map_query_blocks() for a special case where we can
246 * have a physically and logically contiguous blocks split across two leaf
247 * nodes instead of a single extent. This is required in case of atomic writes
248 * to know whether the returned extent is last in leaf. If yes, then lookup for
249 * next in leaf block in ext4_map_query_blocks_next_in_leaf().
250 * - This is never going to be added to any buffer head state.
251 * - We use the next available bit after BH_BITMAP_UPTODATE.
252 */
253#define EXT4_MAP_QUERY_LAST_IN_LEAF BIT(BH_BITMAP_UPTODATE + 1)
254#define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
255 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\
256 EXT4_MAP_DELAYED | EXT4_MAP_QUERY_LAST_IN_LEAF)
257
258struct ext4_map_blocks {
259 ext4_fsblk_t m_pblk;
260 ext4_lblk_t m_lblk;
261 unsigned int m_len;
262 unsigned int m_flags;
263 u64 m_seq;
264};
265
266/*
267 * Block validity checking, system zone rbtree.
268 */
269struct ext4_system_blocks {
270 struct rb_root root;
271 struct rcu_head rcu;
272};
273
274/*
275 * Flags for ext4_io_end->flags
276 */
277#define EXT4_IO_END_UNWRITTEN 0x0001
278#define EXT4_IO_END_FAILED 0x0002
279
280#define EXT4_IO_END_DEFER_COMPLETION (EXT4_IO_END_UNWRITTEN | EXT4_IO_END_FAILED)
281
282struct ext4_io_end_vec {
283 struct list_head list; /* list of io_end_vec */
284 loff_t offset; /* offset in the file */
285 ssize_t size; /* size of the extent */
286};
287
288/*
289 * For converting unwritten extents on a work queue. 'handle' is used for
290 * buffered writeback.
291 */
292typedef struct ext4_io_end {
293 struct list_head list; /* per-file finished IO list */
294 handle_t *handle; /* handle reserved for extent
295 * conversion */
296 struct inode *inode; /* file being written to */
297 struct bio *bio; /* Linked list of completed
298 * bios covering the extent */
299 unsigned int flag; /* unwritten or not */
300 refcount_t count; /* reference counter */
301 struct list_head list_vec; /* list of ext4_io_end_vec */
302} ext4_io_end_t;
303
304struct ext4_io_submit {
305 struct writeback_control *io_wbc;
306 struct bio *io_bio;
307 ext4_io_end_t *io_end;
308 sector_t io_next_block;
309};
310
311/*
312 * Special inodes numbers
313 */
314#define EXT4_BAD_INO 1 /* Bad blocks inode */
315#define EXT4_ROOT_INO 2 /* Root inode */
316#define EXT4_USR_QUOTA_INO 3 /* User quota inode */
317#define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
318#define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
319#define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
320#define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
321#define EXT4_JOURNAL_INO 8 /* Journal inode */
322
323/* First non-reserved inode for old ext4 filesystems */
324#define EXT4_GOOD_OLD_FIRST_INO 11
325
326/*
327 * Maximal count of links to a file
328 */
329#define EXT4_LINK_MAX 65000
330
331/*
332 * Macro-instructions used to manage several block sizes
333 */
334#define EXT4_MIN_BLOCK_SIZE 1024
335#define EXT4_MAX_BLOCK_SIZE 65536
336#define EXT4_MIN_BLOCK_LOG_SIZE 10
337#define EXT4_MAX_BLOCK_LOG_SIZE 16
338#define EXT4_MAX_CLUSTER_LOG_SIZE 30
339#ifdef __KERNEL__
340# define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
341#else
342# define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
343#endif
344#define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
345#define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
346 EXT4_SB(s)->s_cluster_bits)
347#ifdef __KERNEL__
348# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
349# define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
350#else
351# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
352#endif
353#ifdef __KERNEL__
354#define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
355#define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
356#define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
357#else
358#define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
359 EXT4_GOOD_OLD_INODE_SIZE : \
360 (s)->s_inode_size)
361#define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
362 EXT4_GOOD_OLD_FIRST_INO : \
363 (s)->s_first_ino)
364#endif
365#define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
366#define EXT4_MAX_BLOCKS(size, offset, blkbits) \
367 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \
368 blkbits))
369#define EXT4_B_TO_LBLK(inode, offset) \
370 (round_up((offset), i_blocksize(inode)) >> (inode)->i_blkbits)
371#define EXT4_LBLK_TO_B(inode, lblk) ((loff_t)(lblk) << (inode)->i_blkbits)
372
373/* Translate a block number to a page index */
374#define EXT4_LBLK_TO_PG(inode, lblk) (EXT4_LBLK_TO_B((inode), (lblk)) >> \
375 PAGE_SHIFT)
376/* Translate a page index to a block number */
377#define EXT4_PG_TO_LBLK(inode, pnum) (((loff_t)(pnum) << PAGE_SHIFT) >> \
378 (inode)->i_blkbits)
379/* Translate a block number to a cluster number */
380#define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
381/* Translate a cluster number to a block number */
382#define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
383/* Translate # of blks to # of clusters */
384#define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
385 (sbi)->s_cluster_bits)
386/* Mask out the low bits to get the starting block of the cluster */
387#define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
388 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
389#define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
390 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
391/* Fill in the low bits to get the last block of the cluster */
392#define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) | \
393 ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1))
394/* Get the cluster offset */
395#define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
396 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
397#define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
398 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
399
400/*
401 * Structure of a blocks group descriptor
402 */
403struct ext4_group_desc
404{
405 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
406 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
407 __le32 bg_inode_table_lo; /* Inodes table block */
408 __le16 bg_free_blocks_count_lo;/* Free blocks count */
409 __le16 bg_free_inodes_count_lo;/* Free inodes count */
410 __le16 bg_used_dirs_count_lo; /* Directories count */
411 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
412 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
413 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
414 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
415 __le16 bg_itable_unused_lo; /* Unused inodes count */
416 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
417 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
418 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
419 __le32 bg_inode_table_hi; /* Inodes table block MSB */
420 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
421 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
422 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
423 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
424 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
425 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
426 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
427 __u32 bg_reserved;
428};
429
430#define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
431 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
432 sizeof(__le16))
433#define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
434 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
435 sizeof(__le16))
436
437/*
438 * Structure of a flex block group info
439 */
440
441struct flex_groups {
442 atomic64_t free_clusters;
443 atomic_t free_inodes;
444 atomic_t used_dirs;
445};
446
447#define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
448#define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
449#define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
450
451/*
452 * Macro-instructions used to manage group descriptors
453 */
454#define EXT4_MIN_DESC_SIZE 32
455#define EXT4_MIN_DESC_SIZE_64BIT 64
456#define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
457#define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
458#ifdef __KERNEL__
459# define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
460# define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
461# define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
462# define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
463# define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
464#else
465# define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
466# define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
467# define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
468#endif
469
470/*
471 * Constants relative to the data blocks
472 */
473#define EXT4_NDIR_BLOCKS 12
474#define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
475#define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
476#define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
477#define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
478
479/*
480 * Inode flags
481 */
482#define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
483#define EXT4_UNRM_FL 0x00000002 /* Undelete */
484#define EXT4_COMPR_FL 0x00000004 /* Compress file */
485#define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
486#define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
487#define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
488#define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
489#define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
490/* Reserved for compression usage... */
491#define EXT4_DIRTY_FL 0x00000100
492#define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
493#define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
494 /* nb: was previously EXT2_ECOMPR_FL */
495#define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */
496/* End compression flags --- maybe not all used */
497#define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
498#define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
499#define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
500#define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
501#define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
502#define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
503#define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
504#define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
505#define EXT4_VERITY_FL 0x00100000 /* Verity protected inode */
506#define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
507/* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */
508
509#define EXT4_DAX_FL 0x02000000 /* Inode is DAX */
510
511#define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
512#define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
513#define EXT4_CASEFOLD_FL 0x40000000 /* Casefolded directory */
514#define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
515
516/* User modifiable flags */
517#define EXT4_FL_USER_MODIFIABLE (EXT4_SECRM_FL | \
518 EXT4_UNRM_FL | \
519 EXT4_COMPR_FL | \
520 EXT4_SYNC_FL | \
521 EXT4_IMMUTABLE_FL | \
522 EXT4_APPEND_FL | \
523 EXT4_NODUMP_FL | \
524 EXT4_NOATIME_FL | \
525 EXT4_JOURNAL_DATA_FL | \
526 EXT4_NOTAIL_FL | \
527 EXT4_DIRSYNC_FL | \
528 EXT4_TOPDIR_FL | \
529 EXT4_EXTENTS_FL | \
530 0x00400000 /* EXT4_EOFBLOCKS_FL */ | \
531 EXT4_DAX_FL | \
532 EXT4_PROJINHERIT_FL | \
533 EXT4_CASEFOLD_FL)
534
535/* User visible flags */
536#define EXT4_FL_USER_VISIBLE (EXT4_FL_USER_MODIFIABLE | \
537 EXT4_DIRTY_FL | \
538 EXT4_COMPRBLK_FL | \
539 EXT4_NOCOMPR_FL | \
540 EXT4_ENCRYPT_FL | \
541 EXT4_INDEX_FL | \
542 EXT4_VERITY_FL | \
543 EXT4_INLINE_DATA_FL)
544
545/* Flags that should be inherited by new inodes from their parent. */
546#define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
547 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
548 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
549 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
550 EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL |\
551 EXT4_DAX_FL)
552
553/* Flags that are appropriate for regular files (all but dir-specific ones). */
554#define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\
555 EXT4_PROJINHERIT_FL))
556
557/* Flags that are appropriate for non-directories/regular files. */
558#define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
559
560/* The only flags that should be swapped */
561#define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL)
562
563/* Flags which are mutually exclusive to DAX */
564#define EXT4_DAX_MUT_EXCL (EXT4_VERITY_FL | EXT4_ENCRYPT_FL |\
565 EXT4_JOURNAL_DATA_FL | EXT4_INLINE_DATA_FL)
566
567/* Mask out flags that are inappropriate for the given type of inode. */
568static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
569{
570 if (S_ISDIR(mode))
571 return flags;
572 else if (S_ISREG(mode))
573 return flags & EXT4_REG_FLMASK;
574 else
575 return flags & EXT4_OTHER_FLMASK;
576}
577
578/*
579 * Inode flags used for atomic set/get
580 */
581enum {
582 EXT4_INODE_SECRM = 0, /* Secure deletion */
583 EXT4_INODE_UNRM = 1, /* Undelete */
584 EXT4_INODE_COMPR = 2, /* Compress file */
585 EXT4_INODE_SYNC = 3, /* Synchronous updates */
586 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
587 EXT4_INODE_APPEND = 5, /* writes to file may only append */
588 EXT4_INODE_NODUMP = 6, /* do not dump file */
589 EXT4_INODE_NOATIME = 7, /* do not update atime */
590/* Reserved for compression usage... */
591 EXT4_INODE_DIRTY = 8,
592 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
593 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
594 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */
595/* End compression flags --- maybe not all used */
596 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
597 EXT4_INODE_IMAGIC = 13, /* AFS directory */
598 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
599 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
600 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
601 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
602 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
603 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
604 EXT4_INODE_VERITY = 20, /* Verity protected inode */
605 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
606/* 22 was formerly EXT4_INODE_EOFBLOCKS */
607 EXT4_INODE_DAX = 25, /* Inode is DAX */
608 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
609 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */
610 EXT4_INODE_CASEFOLD = 30, /* Casefolded directory */
611 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
612};
613
614/*
615 * Since it's pretty easy to mix up bit numbers and hex values, we use a
616 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
617 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
618 * any extra space in the compiled kernel image, otherwise, the build will fail.
619 * It's important that these values are the same, since we are using
620 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
621 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
622 * values found in ext2, ext3 and ext4 filesystems, and of course the values
623 * defined in e2fsprogs.
624 *
625 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
626 */
627#define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1U << EXT4_INODE_##FLAG))
628#define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
629
630static inline void ext4_check_flag_values(void)
631{
632 CHECK_FLAG_VALUE(SECRM);
633 CHECK_FLAG_VALUE(UNRM);
634 CHECK_FLAG_VALUE(COMPR);
635 CHECK_FLAG_VALUE(SYNC);
636 CHECK_FLAG_VALUE(IMMUTABLE);
637 CHECK_FLAG_VALUE(APPEND);
638 CHECK_FLAG_VALUE(NODUMP);
639 CHECK_FLAG_VALUE(NOATIME);
640 CHECK_FLAG_VALUE(DIRTY);
641 CHECK_FLAG_VALUE(COMPRBLK);
642 CHECK_FLAG_VALUE(NOCOMPR);
643 CHECK_FLAG_VALUE(ENCRYPT);
644 CHECK_FLAG_VALUE(INDEX);
645 CHECK_FLAG_VALUE(IMAGIC);
646 CHECK_FLAG_VALUE(JOURNAL_DATA);
647 CHECK_FLAG_VALUE(NOTAIL);
648 CHECK_FLAG_VALUE(DIRSYNC);
649 CHECK_FLAG_VALUE(TOPDIR);
650 CHECK_FLAG_VALUE(HUGE_FILE);
651 CHECK_FLAG_VALUE(EXTENTS);
652 CHECK_FLAG_VALUE(VERITY);
653 CHECK_FLAG_VALUE(EA_INODE);
654 CHECK_FLAG_VALUE(INLINE_DATA);
655 CHECK_FLAG_VALUE(PROJINHERIT);
656 CHECK_FLAG_VALUE(CASEFOLD);
657 CHECK_FLAG_VALUE(RESERVED);
658}
659
660#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
661struct compat_ext4_new_group_input {
662 u32 group;
663 compat_u64 block_bitmap;
664 compat_u64 inode_bitmap;
665 compat_u64 inode_table;
666 u32 blocks_count;
667 u16 reserved_blocks;
668 u16 unused;
669};
670#endif
671
672/* The struct ext4_new_group_input in kernel space, with free_blocks_count */
673struct ext4_new_group_data {
674 __u32 group;
675 __u64 block_bitmap;
676 __u64 inode_bitmap;
677 __u64 inode_table;
678 __u32 blocks_count;
679 __u16 reserved_blocks;
680 __u16 mdata_blocks;
681 __u32 free_clusters_count;
682};
683
684/* Indexes used to index group tables in ext4_new_group_data */
685enum {
686 BLOCK_BITMAP = 0, /* block bitmap */
687 INODE_BITMAP, /* inode bitmap */
688 INODE_TABLE, /* inode tables */
689 GROUP_TABLE_COUNT,
690};
691
692/*
693 * Flags used by ext4_map_blocks()
694 */
695 /* Allocate any needed blocks and/or convert an unwritten
696 extent to be an initialized ext4 */
697#define EXT4_GET_BLOCKS_CREATE 0x0001
698 /* Request the creation of an unwritten extent */
699#define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002
700#define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\
701 EXT4_GET_BLOCKS_CREATE)
702 /* Caller is from the delayed allocation writeout path
703 * finally doing the actual allocation of delayed blocks */
704#define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
705 /*
706 * This means that we cannot merge newly allocated extents, and if we
707 * found an unwritten extent, we need to split it.
708 */
709#define EXT4_GET_BLOCKS_SPLIT_NOMERGE 0x0008
710 /*
711 * Caller is from the dio or dioread_nolock buffered IO, reqest to
712 * create an unwritten extent if it does not exist or split the
713 * found unwritten extent. Also do not merge the newly created
714 * unwritten extent, io end will convert unwritten to written,
715 * and try to merge the written extent.
716 */
717#define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_SPLIT_NOMERGE|\
718 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
719 /* Convert unwritten extent to initialized. */
720#define EXT4_GET_BLOCKS_CONVERT 0x0010
721 /* Eventual metadata allocation (due to growing extent tree)
722 * should not fail, so try to use reserved blocks for that.*/
723#define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
724 /* Don't normalize allocation size (used for fallocate) */
725#define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
726 /* Convert written extents to unwritten */
727#define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100
728 /* Write zeros to newly created written extents */
729#define EXT4_GET_BLOCKS_ZERO 0x0200
730#define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\
731 EXT4_GET_BLOCKS_ZERO)
732 /* Caller is in the context of data submission, such as writeback,
733 * fsync, etc. Especially, in the generic writeback path, caller will
734 * submit data before dropping transaction handle. This allows jbd2
735 * to avoid submitting data before commit. */
736#define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400
737 /* Convert extent to initialized after IO complete */
738#define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT |\
739 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT |\
740 EXT4_GET_BLOCKS_IO_SUBMIT)
741 /* Caller is in the atomic contex, find extent if it has been cached */
742#define EXT4_GET_BLOCKS_CACHED_NOWAIT 0x0800
743/*
744 * Atomic write caller needs this to query in the slow path of mixed mapping
745 * case, when a contiguous extent can be split across two adjacent leaf nodes.
746 * Look EXT4_MAP_QUERY_LAST_IN_LEAF.
747 */
748#define EXT4_GET_BLOCKS_QUERY_LAST_IN_LEAF 0x1000
749
750/*
751 * The bit position of these flags must not overlap with any of the
752 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(),
753 * read_extent_tree_block(), ext4_split_extent_at(),
754 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
755 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
756 * caching the extents when reading from the extent tree while a
757 * truncate or punch hole operation is in progress.
758 */
759#define EXT4_EX_NOCACHE 0x40000000
760#define EXT4_EX_FORCE_CACHE 0x20000000
761#define EXT4_EX_NOFAIL 0x10000000
762/*
763 * ext4_map_query_blocks() uses this filter mask to filter the flags needed to
764 * pass while lookup/querying of on disk extent tree.
765 */
766#define EXT4_EX_QUERY_FILTER (EXT4_EX_NOCACHE | EXT4_EX_FORCE_CACHE |\
767 EXT4_EX_NOFAIL |\
768 EXT4_GET_BLOCKS_QUERY_LAST_IN_LEAF)
769
770/*
771 * Flags used by ext4_free_blocks
772 */
773#define EXT4_FREE_BLOCKS_METADATA 0x0001
774#define EXT4_FREE_BLOCKS_FORGET 0x0002
775#define EXT4_FREE_BLOCKS_VALIDATED 0x0004
776#define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
777#define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
778#define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
779#define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040
780
781#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
782/*
783 * ioctl commands in 32 bit emulation
784 */
785#define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
786#define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
787#define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
788#define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
789#define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
790#define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
791#define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
792#define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
793#endif
794
795/* Max physical block we can address w/o extents */
796#define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
797
798/* Max logical block we can support */
799#define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFE
800
801/*
802 * Structure of an inode on the disk
803 */
804struct ext4_inode {
805 __le16 i_mode; /* File mode */
806 __le16 i_uid; /* Low 16 bits of Owner Uid */
807 __le32 i_size_lo; /* Size in bytes */
808 __le32 i_atime; /* Access time */
809 __le32 i_ctime; /* Inode Change time */
810 __le32 i_mtime; /* Modification time */
811 __le32 i_dtime; /* Deletion Time */
812 __le16 i_gid; /* Low 16 bits of Group Id */
813 __le16 i_links_count; /* Links count */
814 __le32 i_blocks_lo; /* Blocks count */
815 __le32 i_flags; /* File flags */
816 union {
817 struct {
818 __le32 l_i_version;
819 } linux1;
820 struct {
821 __u32 h_i_translator;
822 } hurd1;
823 struct {
824 __u32 m_i_reserved1;
825 } masix1;
826 } osd1; /* OS dependent 1 */
827 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
828 __le32 i_generation; /* File version (for NFS) */
829 __le32 i_file_acl_lo; /* File ACL */
830 __le32 i_size_high;
831 __le32 i_obso_faddr; /* Obsoleted fragment address */
832 union {
833 struct {
834 __le16 l_i_blocks_high; /* were l_i_reserved1 */
835 __le16 l_i_file_acl_high;
836 __le16 l_i_uid_high; /* these 2 fields */
837 __le16 l_i_gid_high; /* were reserved2[0] */
838 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
839 __le16 l_i_reserved;
840 } linux2;
841 struct {
842 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
843 __u16 h_i_mode_high;
844 __u16 h_i_uid_high;
845 __u16 h_i_gid_high;
846 __u32 h_i_author;
847 } hurd2;
848 struct {
849 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
850 __le16 m_i_file_acl_high;
851 __u32 m_i_reserved2[2];
852 } masix2;
853 } osd2; /* OS dependent 2 */
854 __le16 i_extra_isize;
855 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
856 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
857 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
858 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
859 __le32 i_crtime; /* File Creation time */
860 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
861 __le32 i_version_hi; /* high 32 bits for 64-bit version */
862 __le32 i_projid; /* Project ID */
863};
864
865#define EXT4_EPOCH_BITS 2
866#define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
867#define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
868
869/*
870 * Extended fields will fit into an inode if the filesystem was formatted
871 * with large inodes (-I 256 or larger) and there are not currently any EAs
872 * consuming all of the available space. For new inodes we always reserve
873 * enough space for the kernel's known extended fields, but for inodes
874 * created with an old kernel this might not have been the case. None of
875 * the extended inode fields is critical for correct filesystem operation.
876 * This macro checks if a certain field fits in the inode. Note that
877 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
878 */
879#define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
880 ((offsetof(typeof(*ext4_inode), field) + \
881 sizeof((ext4_inode)->field)) \
882 <= (EXT4_GOOD_OLD_INODE_SIZE + \
883 (einode)->i_extra_isize)) \
884
885/*
886 * We use an encoding that preserves the times for extra epoch "00":
887 *
888 * extra msb of adjust for signed
889 * epoch 32-bit 32-bit tv_sec to
890 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
891 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
892 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
893 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
894 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
895 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
896 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
897 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
898 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
899 *
900 * Note that previous versions of the kernel on 64-bit systems would
901 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
902 * 1970. e2fsck will correct this, assuming that it is run on the
903 * affected filesystem before 2242.
904 */
905
906static inline __le32 ext4_encode_extra_time(struct timespec64 ts)
907{
908 u32 extra = ((ts.tv_sec - (s32)ts.tv_sec) >> 32) & EXT4_EPOCH_MASK;
909 return cpu_to_le32(extra | (ts.tv_nsec << EXT4_EPOCH_BITS));
910}
911
912static inline struct timespec64 ext4_decode_extra_time(__le32 base,
913 __le32 extra)
914{
915 struct timespec64 ts = { .tv_sec = (signed)le32_to_cpu(base) };
916
917 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK)))
918 ts.tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
919 ts.tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
920 return ts;
921}
922
923#define EXT4_INODE_SET_XTIME_VAL(xtime, inode, raw_inode, ts) \
924do { \
925 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \
926 (raw_inode)->xtime = cpu_to_le32((ts).tv_sec); \
927 (raw_inode)->xtime ## _extra = ext4_encode_extra_time(ts); \
928 } else \
929 (raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (ts).tv_sec, S32_MIN, S32_MAX)); \
930} while (0)
931
932#define EXT4_INODE_SET_ATIME(inode, raw_inode) \
933 EXT4_INODE_SET_XTIME_VAL(i_atime, inode, raw_inode, inode_get_atime(inode))
934
935#define EXT4_INODE_SET_MTIME(inode, raw_inode) \
936 EXT4_INODE_SET_XTIME_VAL(i_mtime, inode, raw_inode, inode_get_mtime(inode))
937
938#define EXT4_INODE_SET_CTIME(inode, raw_inode) \
939 EXT4_INODE_SET_XTIME_VAL(i_ctime, inode, raw_inode, inode_get_ctime(inode))
940
941#define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
942 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
943 EXT4_INODE_SET_XTIME_VAL(xtime, &((einode)->vfs_inode), \
944 raw_inode, (einode)->xtime)
945
946#define EXT4_INODE_GET_XTIME_VAL(xtime, inode, raw_inode) \
947 (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra) ? \
948 ext4_decode_extra_time((raw_inode)->xtime, \
949 (raw_inode)->xtime ## _extra) : \
950 (struct timespec64) { \
951 .tv_sec = (signed)le32_to_cpu((raw_inode)->xtime) \
952 })
953
954#define EXT4_INODE_GET_ATIME(inode, raw_inode) \
955do { \
956 inode_set_atime_to_ts(inode, \
957 EXT4_INODE_GET_XTIME_VAL(i_atime, inode, raw_inode)); \
958} while (0)
959
960#define EXT4_INODE_GET_MTIME(inode, raw_inode) \
961do { \
962 inode_set_mtime_to_ts(inode, \
963 EXT4_INODE_GET_XTIME_VAL(i_mtime, inode, raw_inode)); \
964} while (0)
965
966#define EXT4_INODE_GET_CTIME(inode, raw_inode) \
967do { \
968 inode_set_ctime_to_ts(inode, \
969 EXT4_INODE_GET_XTIME_VAL(i_ctime, inode, raw_inode)); \
970} while (0)
971
972#define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
973do { \
974 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
975 (einode)->xtime = \
976 EXT4_INODE_GET_XTIME_VAL(xtime, &(einode->vfs_inode), \
977 raw_inode); \
978 else \
979 (einode)->xtime = (struct timespec64){0, 0}; \
980} while (0)
981
982#define i_disk_version osd1.linux1.l_i_version
983
984#if defined(__KERNEL__) || defined(__linux__)
985#define i_reserved1 osd1.linux1.l_i_reserved1
986#define i_file_acl_high osd2.linux2.l_i_file_acl_high
987#define i_blocks_high osd2.linux2.l_i_blocks_high
988#define i_uid_low i_uid
989#define i_gid_low i_gid
990#define i_uid_high osd2.linux2.l_i_uid_high
991#define i_gid_high osd2.linux2.l_i_gid_high
992#define i_checksum_lo osd2.linux2.l_i_checksum_lo
993
994#elif defined(__GNU__)
995
996#define i_translator osd1.hurd1.h_i_translator
997#define i_uid_high osd2.hurd2.h_i_uid_high
998#define i_gid_high osd2.hurd2.h_i_gid_high
999#define i_author osd2.hurd2.h_i_author
1000
1001#elif defined(__masix__)
1002
1003#define i_reserved1 osd1.masix1.m_i_reserved1
1004#define i_file_acl_high osd2.masix2.m_i_file_acl_high
1005#define i_reserved2 osd2.masix2.m_i_reserved2
1006
1007#endif /* defined(__KERNEL__) || defined(__linux__) */
1008
1009#include "extents_status.h"
1010#include "fast_commit.h"
1011
1012/*
1013 * Lock subclasses for i_data_sem in the ext4_inode_info structure.
1014 *
1015 * These are needed to avoid lockdep false positives when we need to
1016 * allocate blocks to the quota inode during ext4_map_blocks(), while
1017 * holding i_data_sem for a normal (non-quota) inode. Since we don't
1018 * do quota tracking for the quota inode, this avoids deadlock (as
1019 * well as infinite recursion, since it isn't turtles all the way
1020 * down...)
1021 *
1022 * I_DATA_SEM_NORMAL - Used for most inodes
1023 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode
1024 * where the second inode has larger inode number
1025 * than the first
1026 * I_DATA_SEM_QUOTA - Used for quota inodes only
1027 * I_DATA_SEM_EA - Used for ea_inodes only
1028 */
1029enum {
1030 I_DATA_SEM_NORMAL = 0,
1031 I_DATA_SEM_OTHER,
1032 I_DATA_SEM_QUOTA,
1033 I_DATA_SEM_EA
1034};
1035
1036
1037/*
1038 * fourth extended file system inode data in memory
1039 */
1040struct ext4_inode_info {
1041 __le32 i_data[15]; /* unconverted */
1042 __u32 i_dtime;
1043 ext4_fsblk_t i_file_acl;
1044
1045 /*
1046 * i_block_group is the number of the block group which contains
1047 * this file's inode. Constant across the lifetime of the inode,
1048 * it is used for making block allocation decisions - we try to
1049 * place a file's data blocks near its inode block, and new inodes
1050 * near to their parent directory's inode.
1051 */
1052 ext4_group_t i_block_group;
1053 ext4_lblk_t i_dir_start_lookup;
1054#if (BITS_PER_LONG < 64)
1055 unsigned long i_state_flags; /* Dynamic state flags */
1056#endif
1057 unsigned long i_flags;
1058
1059 /*
1060 * Extended attributes can be read independently of the main file
1061 * data. Taking i_rwsem even when reading would cause contention
1062 * between readers of EAs and writers of regular file data, so
1063 * instead we synchronize on xattr_sem when reading or changing
1064 * EAs.
1065 */
1066 struct rw_semaphore xattr_sem;
1067
1068 /*
1069 * Inodes with EXT4_STATE_ORPHAN_FILE use i_orphan_idx. Otherwise
1070 * i_orphan is used.
1071 */
1072 union {
1073 struct list_head i_orphan; /* unlinked but open inodes */
1074 unsigned int i_orphan_idx; /* Index in orphan file */
1075 };
1076
1077 /* Fast commit related info */
1078
1079 /* For tracking dentry create updates */
1080 struct list_head i_fc_dilist;
1081 struct list_head i_fc_list; /*
1082 * inodes that need fast commit
1083 * protected by sbi->s_fc_lock.
1084 */
1085
1086 /* Start of lblk range that needs to be committed in this fast commit */
1087 ext4_lblk_t i_fc_lblk_start;
1088
1089 /* End of lblk range that needs to be committed in this fast commit */
1090 ext4_lblk_t i_fc_lblk_len;
1091
1092 spinlock_t i_raw_lock; /* protects updates to the raw inode */
1093
1094 /* Fast commit wait queue for this inode */
1095 wait_queue_head_t i_fc_wait;
1096
1097 /*
1098 * Protect concurrent accesses on i_fc_lblk_start, i_fc_lblk_len
1099 * and inode's EXT4_FC_STATE_COMMITTING state bit.
1100 */
1101 spinlock_t i_fc_lock;
1102
1103 /*
1104 * i_disksize keeps track of what the inode size is ON DISK, not
1105 * in memory. During truncate, i_size is set to the new size by
1106 * the VFS prior to calling ext4_truncate(), but the filesystem won't
1107 * set i_disksize to 0 until the truncate is actually under way.
1108 *
1109 * The intent is that i_disksize always represents the blocks which
1110 * are used by this file. This allows recovery to restart truncate
1111 * on orphans if we crash during truncate. We actually write i_disksize
1112 * into the on-disk inode when writing inodes out, instead of i_size.
1113 *
1114 * The only time when i_disksize and i_size may be different is when
1115 * a truncate is in progress. The only things which change i_disksize
1116 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
1117 */
1118 loff_t i_disksize;
1119
1120 /*
1121 * i_data_sem is for serialising ext4_truncate() against
1122 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
1123 * data tree are chopped off during truncate. We can't do that in
1124 * ext4 because whenever we perform intermediate commits during
1125 * truncate, the inode and all the metadata blocks *must* be in a
1126 * consistent state which allows truncation of the orphans to restart
1127 * during recovery. Hence we must fix the get_block-vs-truncate race
1128 * by other means, so we have i_data_sem.
1129 */
1130 struct rw_semaphore i_data_sem;
1131 struct inode vfs_inode;
1132 struct jbd2_inode *jinode;
1133
1134 /*
1135 * File creation time. Its function is same as that of
1136 * struct timespec64 i_{a,c,m}time in the generic inode.
1137 */
1138 struct timespec64 i_crtime;
1139
1140 /* mballoc */
1141 atomic_t i_prealloc_active;
1142
1143 /* allocation reservation info for delalloc */
1144 /* In case of bigalloc, this refer to clusters rather than blocks */
1145 unsigned int i_reserved_data_blocks;
1146 struct rb_root i_prealloc_node;
1147 rwlock_t i_prealloc_lock;
1148
1149 /* extents status tree */
1150 struct ext4_es_tree i_es_tree;
1151 rwlock_t i_es_lock;
1152 struct list_head i_es_list;
1153 unsigned int i_es_all_nr; /* protected by i_es_lock */
1154 unsigned int i_es_shk_nr; /* protected by i_es_lock */
1155 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for
1156 extents to shrink. Protected by
1157 i_es_lock */
1158 u64 i_es_seq; /* Change counter for extents.
1159 Protected by i_es_lock */
1160
1161 /* ialloc */
1162 ext4_group_t i_last_alloc_group;
1163
1164 /* pending cluster reservations for bigalloc file systems */
1165 struct ext4_pending_tree i_pending_tree;
1166
1167 /* on-disk additional length */
1168 __u16 i_extra_isize;
1169
1170 /* Indicate the inline data space. */
1171 u16 i_inline_off;
1172 u16 i_inline_size;
1173
1174#ifdef CONFIG_QUOTA
1175 /* quota space reservation, managed internally by quota code */
1176 qsize_t i_reserved_quota;
1177#endif
1178 spinlock_t i_block_reservation_lock;
1179
1180 /* Lock protecting lists below */
1181 spinlock_t i_completed_io_lock;
1182 /*
1183 * Completed IOs that need unwritten extents handling and have
1184 * transaction reserved
1185 */
1186 struct list_head i_rsv_conversion_list;
1187 struct work_struct i_rsv_conversion_work;
1188
1189 /*
1190 * Transactions that contain inode's metadata needed to complete
1191 * fsync and fdatasync, respectively.
1192 */
1193 tid_t i_sync_tid;
1194 tid_t i_datasync_tid;
1195
1196#ifdef CONFIG_QUOTA
1197 struct dquot __rcu *i_dquot[MAXQUOTAS];
1198#endif
1199
1200 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1201 __u32 i_csum_seed;
1202
1203 kprojid_t i_projid;
1204
1205#ifdef CONFIG_FS_ENCRYPTION
1206 struct fscrypt_inode_info *i_crypt_info;
1207#endif
1208
1209#ifdef CONFIG_FS_VERITY
1210 struct fsverity_info *i_verity_info;
1211#endif
1212};
1213
1214/*
1215 * File system states
1216 */
1217#define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
1218#define EXT4_ERROR_FS 0x0002 /* Errors detected */
1219#define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
1220#define EXT4_FC_REPLAY 0x0020 /* Fast commit replay ongoing */
1221
1222/*
1223 * Misc. filesystem flags
1224 */
1225#define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
1226#define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
1227#define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
1228
1229/*
1230 * Mount flags set via mount options or defaults
1231 */
1232#define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */
1233#define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
1234#define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
1235#define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
1236#define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
1237#define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
1238#define EXT4_MOUNT_ERRORS_MASK 0x00070
1239#define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
1240#define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
1241#ifdef CONFIG_FS_DAX
1242#define EXT4_MOUNT_DAX_ALWAYS 0x00200 /* Direct Access */
1243#else
1244#define EXT4_MOUNT_DAX_ALWAYS 0
1245#endif
1246#define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
1247#define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
1248#define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
1249#define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
1250#define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
1251#define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
1252#define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
1253#define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
1254#define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
1255#define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
1256#define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */
1257#define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota,
1258 * enable enforcement for hidden
1259 * quota files */
1260#define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable
1261 * enforcement for hidden quota
1262 * files */
1263#define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota
1264 * enforcement */
1265#define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
1266#define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
1267#define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1268#define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */
1269#define EXT4_MOUNT_NO_PREFETCH_BLOCK_BITMAPS 0x4000000
1270#define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
1271#define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
1272#define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
1273#define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
1274#define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
1275
1276/*
1277 * Mount flags set either automatically (could not be set by mount option)
1278 * based on per file system feature or property or in special cases such as
1279 * distinguishing between explicit mount option definition and default.
1280 */
1281#define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1282 specified delalloc */
1283#define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1284 size of blocksize * 8
1285 blocks */
1286#define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1287 file systems */
1288#define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly
1289 specified journal checksum */
1290
1291#define EXT4_MOUNT2_JOURNAL_FAST_COMMIT 0x00000010 /* Journal fast commit */
1292#define EXT4_MOUNT2_DAX_NEVER 0x00000020 /* Do not allow Direct Access */
1293#define EXT4_MOUNT2_DAX_INODE 0x00000040 /* For printing options only */
1294#define EXT4_MOUNT2_MB_OPTIMIZE_SCAN 0x00000080 /* Optimize group
1295 * scanning in mballoc
1296 */
1297#define EXT4_MOUNT2_ABORT 0x00000100 /* Abort filesystem */
1298
1299#define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1300 ~EXT4_MOUNT_##opt
1301#define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1302 EXT4_MOUNT_##opt
1303#define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1304 EXT4_MOUNT_##opt)
1305
1306#define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1307 ~EXT4_MOUNT2_##opt
1308#define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1309 EXT4_MOUNT2_##opt
1310#define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1311 EXT4_MOUNT2_##opt)
1312
1313#define ext4_test_and_set_bit __test_and_set_bit_le
1314#define ext4_set_bit __set_bit_le
1315#define ext4_test_and_clear_bit __test_and_clear_bit_le
1316#define ext4_clear_bit __clear_bit_le
1317#define ext4_test_bit test_bit_le
1318#define ext4_find_next_zero_bit find_next_zero_bit_le
1319#define ext4_find_next_bit find_next_bit_le
1320
1321extern void mb_set_bits(void *bm, int cur, int len);
1322
1323/*
1324 * Maximal mount counts between two filesystem checks
1325 */
1326#define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1327#define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1328
1329/*
1330 * Behaviour when detecting errors
1331 */
1332#define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1333#define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1334#define EXT4_ERRORS_PANIC 3 /* Panic */
1335#define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1336
1337/* Metadata checksum algorithm codes */
1338#define EXT4_CRC32C_CHKSUM 1
1339
1340#define EXT4_LABEL_MAX 16
1341
1342/*
1343 * Structure of the super block
1344 */
1345struct ext4_super_block {
1346/*00*/ __le32 s_inodes_count; /* Inodes count */
1347 __le32 s_blocks_count_lo; /* Blocks count */
1348 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1349 __le32 s_free_blocks_count_lo; /* Free blocks count */
1350/*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1351 __le32 s_first_data_block; /* First Data Block */
1352 __le32 s_log_block_size; /* Block size */
1353 __le32 s_log_cluster_size; /* Allocation cluster size */
1354/*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1355 __le32 s_clusters_per_group; /* # Clusters per group */
1356 __le32 s_inodes_per_group; /* # Inodes per group */
1357 __le32 s_mtime; /* Mount time */
1358/*30*/ __le32 s_wtime; /* Write time */
1359 __le16 s_mnt_count; /* Mount count */
1360 __le16 s_max_mnt_count; /* Maximal mount count */
1361 __le16 s_magic; /* Magic signature */
1362 __le16 s_state; /* File system state */
1363 __le16 s_errors; /* Behaviour when detecting errors */
1364 __le16 s_minor_rev_level; /* minor revision level */
1365/*40*/ __le32 s_lastcheck; /* time of last check */
1366 __le32 s_checkinterval; /* max. time between checks */
1367 __le32 s_creator_os; /* OS */
1368 __le32 s_rev_level; /* Revision level */
1369/*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1370 __le16 s_def_resgid; /* Default gid for reserved blocks */
1371 /*
1372 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1373 *
1374 * Note: the difference between the compatible feature set and
1375 * the incompatible feature set is that if there is a bit set
1376 * in the incompatible feature set that the kernel doesn't
1377 * know about, it should refuse to mount the filesystem.
1378 *
1379 * e2fsck's requirements are more strict; if it doesn't know
1380 * about a feature in either the compatible or incompatible
1381 * feature set, it must abort and not try to meddle with
1382 * things it doesn't understand...
1383 */
1384 __le32 s_first_ino; /* First non-reserved inode */
1385 __le16 s_inode_size; /* size of inode structure */
1386 __le16 s_block_group_nr; /* block group # of this superblock */
1387 __le32 s_feature_compat; /* compatible feature set */
1388/*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1389 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1390/*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1391/*78*/ char s_volume_name[EXT4_LABEL_MAX] __nonstring; /* volume name */
1392/*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */
1393/*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1394 /*
1395 * Performance hints. Directory preallocation should only
1396 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1397 */
1398 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1399 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1400 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1401 /*
1402 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1403 */
1404/*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1405/*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1406 __le32 s_journal_dev; /* device number of journal file */
1407 __le32 s_last_orphan; /* start of list of inodes to delete */
1408 __le32 s_hash_seed[4]; /* HTREE hash seed */
1409 __u8 s_def_hash_version; /* Default hash version to use */
1410 __u8 s_jnl_backup_type;
1411 __le16 s_desc_size; /* size of group descriptor */
1412/*100*/ __le32 s_default_mount_opts;
1413 __le32 s_first_meta_bg; /* First metablock block group */
1414 __le32 s_mkfs_time; /* When the filesystem was created */
1415 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1416 /* 64bit support valid if EXT4_FEATURE_INCOMPAT_64BIT */
1417/*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1418 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1419 __le32 s_free_blocks_count_hi; /* Free blocks count */
1420 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1421 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1422 __le32 s_flags; /* Miscellaneous flags */
1423 __le16 s_raid_stride; /* RAID stride */
1424 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1425 __le64 s_mmp_block; /* Block for multi-mount protection */
1426 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1427 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1428 __u8 s_checksum_type; /* metadata checksum algorithm used */
1429 __u8 s_encryption_level; /* versioning level for encryption */
1430 __u8 s_reserved_pad; /* Padding to next 32bits */
1431 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1432 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1433 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1434 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1435 snapshot's future use */
1436 __le32 s_snapshot_list; /* inode number of the head of the
1437 on-disk snapshot list */
1438#define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1439 __le32 s_error_count; /* number of fs errors */
1440 __le32 s_first_error_time; /* first time an error happened */
1441 __le32 s_first_error_ino; /* inode involved in first error */
1442 __le64 s_first_error_block; /* block involved of first error */
1443 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */
1444 __le32 s_first_error_line; /* line number where error happened */
1445 __le32 s_last_error_time; /* most recent time of an error */
1446 __le32 s_last_error_ino; /* inode involved in last error */
1447 __le32 s_last_error_line; /* line number where error happened */
1448 __le64 s_last_error_block; /* block involved of last error */
1449 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */
1450#define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1451 __u8 s_mount_opts[64];
1452 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1453 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1454 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1455 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
1456 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */
1457 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */
1458 __le32 s_lpf_ino; /* Location of the lost+found inode */
1459 __le32 s_prj_quota_inum; /* inode for tracking project quota */
1460 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */
1461 __u8 s_wtime_hi;
1462 __u8 s_mtime_hi;
1463 __u8 s_mkfs_time_hi;
1464 __u8 s_lastcheck_hi;
1465 __u8 s_first_error_time_hi;
1466 __u8 s_last_error_time_hi;
1467 __u8 s_first_error_errcode;
1468 __u8 s_last_error_errcode;
1469 __le16 s_encoding; /* Filename charset encoding */
1470 __le16 s_encoding_flags; /* Filename charset encoding flags */
1471 __le32 s_orphan_file_inum; /* Inode for tracking orphan inodes */
1472 __le16 s_def_resuid_hi;
1473 __le16 s_def_resgid_hi;
1474 __le32 s_reserved[93]; /* Padding to the end of the block */
1475 __le32 s_checksum; /* crc32c(superblock) */
1476};
1477
1478#define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1479
1480#ifdef __KERNEL__
1481
1482/* Number of quota types we support */
1483#define EXT4_MAXQUOTAS 3
1484
1485#define EXT4_ENC_UTF8_12_1 1
1486
1487/* Types of ext4 journal triggers */
1488enum ext4_journal_trigger_type {
1489 EXT4_JTR_ORPHAN_FILE,
1490 EXT4_JTR_NONE /* This must be the last entry for indexing to work! */
1491};
1492
1493#define EXT4_JOURNAL_TRIGGER_COUNT EXT4_JTR_NONE
1494
1495struct ext4_journal_trigger {
1496 struct jbd2_buffer_trigger_type tr_triggers;
1497 struct super_block *sb;
1498};
1499
1500static inline struct ext4_journal_trigger *EXT4_TRIGGER(
1501 struct jbd2_buffer_trigger_type *trigger)
1502{
1503 return container_of(trigger, struct ext4_journal_trigger, tr_triggers);
1504}
1505
1506#define EXT4_ORPHAN_BLOCK_MAGIC 0x0b10ca04
1507
1508/* Structure at the tail of orphan block */
1509struct ext4_orphan_block_tail {
1510 __le32 ob_magic;
1511 __le32 ob_checksum;
1512};
1513
1514static inline int ext4_inodes_per_orphan_block(struct super_block *sb)
1515{
1516 return (sb->s_blocksize - sizeof(struct ext4_orphan_block_tail)) /
1517 sizeof(u32);
1518}
1519
1520struct ext4_orphan_block {
1521 atomic_t ob_free_entries; /* Number of free orphan entries in block */
1522 struct buffer_head *ob_bh; /* Buffer for orphan block */
1523};
1524
1525/*
1526 * Info about orphan file.
1527 */
1528struct ext4_orphan_info {
1529 int of_blocks; /* Number of orphan blocks in a file */
1530 __u32 of_csum_seed; /* Checksum seed for orphan file */
1531 struct ext4_orphan_block *of_binfo; /* Array with info about orphan
1532 * file blocks */
1533};
1534
1535/*
1536 * fourth extended-fs super-block data in memory
1537 */
1538struct ext4_sb_info {
1539 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1540 unsigned long s_inodes_per_block;/* Number of inodes per block */
1541 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1542 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1543 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1544 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1545 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1546 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1547 ext4_group_t s_groups_count; /* Number of groups in the fs */
1548 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1549 unsigned long s_overhead; /* # of fs overhead clusters */
1550 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1551 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1552 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1553 struct buffer_head * s_sbh; /* Buffer containing the super block */
1554 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1555 /* Array of bh's for the block group descriptors */
1556 struct buffer_head * __rcu *s_group_desc;
1557 unsigned int s_mount_opt;
1558 unsigned int s_mount_opt2;
1559 unsigned long s_mount_flags;
1560 unsigned int s_def_mount_opt;
1561 unsigned int s_def_mount_opt2;
1562 ext4_fsblk_t s_sb_block;
1563 atomic64_t s_resv_clusters;
1564 kuid_t s_resuid;
1565 kgid_t s_resgid;
1566 unsigned short s_mount_state;
1567 unsigned short s_pad;
1568 int s_addr_per_block_bits;
1569 int s_desc_per_block_bits;
1570 int s_inode_size;
1571 int s_first_ino;
1572 unsigned int s_inode_readahead_blks;
1573 unsigned int s_inode_goal;
1574 u32 s_hash_seed[4];
1575 int s_def_hash_version;
1576 int s_hash_unsigned; /* 3 if hash should be unsigned, 0 if not */
1577 struct percpu_counter s_freeclusters_counter;
1578 struct percpu_counter s_freeinodes_counter;
1579 struct percpu_counter s_dirs_counter;
1580 struct percpu_counter s_dirtyclusters_counter;
1581 struct percpu_counter s_sra_exceeded_retry_limit;
1582 struct blockgroup_lock *s_blockgroup_lock;
1583 struct proc_dir_entry *s_proc;
1584 struct kobject s_kobj;
1585 struct completion s_kobj_unregister;
1586 struct super_block *s_sb;
1587 struct buffer_head *s_mmp_bh;
1588
1589 /* Journaling */
1590 struct journal_s *s_journal;
1591 unsigned long s_ext4_flags; /* Ext4 superblock flags */
1592 struct mutex s_orphan_lock; /* Protects on disk list changes */
1593 struct list_head s_orphan; /* List of orphaned inodes in on disk
1594 list */
1595 struct ext4_orphan_info s_orphan_info;
1596 unsigned long s_commit_interval;
1597 u32 s_max_batch_time;
1598 u32 s_min_batch_time;
1599 struct file *s_journal_bdev_file;
1600#ifdef CONFIG_QUOTA
1601 /* Names of quota files with journalled quota */
1602 char __rcu *s_qf_names[EXT4_MAXQUOTAS];
1603 int s_jquota_fmt; /* Format of quota to use */
1604#endif
1605 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1606 struct ext4_system_blocks __rcu *s_system_blks;
1607
1608#ifdef EXTENTS_STATS
1609 /* ext4 extents stats */
1610 unsigned long s_ext_min;
1611 unsigned long s_ext_max;
1612 unsigned long s_depth_max;
1613 spinlock_t s_ext_stats_lock;
1614 unsigned long s_ext_blocks;
1615 unsigned long s_ext_extents;
1616#endif
1617
1618 /* for buddy allocator */
1619 struct ext4_group_info ** __rcu *s_group_info;
1620 struct inode *s_buddy_cache;
1621 spinlock_t s_md_lock;
1622 unsigned short *s_mb_offsets;
1623 unsigned int *s_mb_maxs;
1624 unsigned int s_group_info_size;
1625 atomic_t s_mb_free_pending;
1626 struct list_head s_freed_data_list[2]; /* List of blocks to be freed
1627 after commit completed */
1628 struct list_head s_discard_list;
1629 struct work_struct s_discard_work;
1630 atomic_t s_retry_alloc_pending;
1631 struct xarray *s_mb_avg_fragment_size;
1632 struct xarray *s_mb_largest_free_orders;
1633
1634 /* tunables */
1635 unsigned long s_stripe;
1636 unsigned int s_mb_max_linear_groups;
1637 unsigned int s_mb_stream_request;
1638 unsigned int s_mb_max_to_scan;
1639 unsigned int s_mb_min_to_scan;
1640 unsigned int s_mb_stats;
1641 unsigned int s_mb_order2_reqs;
1642 unsigned int s_mb_group_prealloc;
1643 unsigned int s_max_dir_size_kb;
1644 unsigned int s_mb_prefetch;
1645 unsigned int s_mb_prefetch_limit;
1646 unsigned int s_mb_best_avail_max_trim_order;
1647 unsigned int s_sb_update_sec;
1648 unsigned int s_sb_update_kb;
1649
1650 /* where last allocation was done - for stream allocation */
1651 ext4_group_t *s_mb_last_groups;
1652 unsigned int s_mb_nr_global_goals;
1653
1654 /* stats for buddy allocator */
1655 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1656 atomic_t s_bal_success; /* we found long enough chunks */
1657 atomic_t s_bal_allocated; /* in blocks */
1658 atomic_t s_bal_ex_scanned; /* total extents scanned */
1659 atomic_t s_bal_cX_ex_scanned[EXT4_MB_NUM_CRS]; /* total extents scanned */
1660 atomic_t s_bal_groups_scanned; /* number of groups scanned */
1661 atomic_t s_bal_goals; /* goal hits */
1662 atomic_t s_bal_stream_goals; /* stream allocation global goal hits */
1663 atomic_t s_bal_len_goals; /* len goal hits */
1664 atomic_t s_bal_breaks; /* too long searches */
1665 atomic_t s_bal_2orders; /* 2^order hits */
1666 atomic64_t s_bal_cX_groups_considered[EXT4_MB_NUM_CRS];
1667 atomic64_t s_bal_cX_hits[EXT4_MB_NUM_CRS];
1668 atomic64_t s_bal_cX_failed[EXT4_MB_NUM_CRS]; /* cX loop didn't find blocks */
1669 atomic_t s_mb_buddies_generated; /* number of buddies generated */
1670 atomic64_t s_mb_generation_time;
1671 atomic_t s_mb_lost_chunks;
1672 atomic_t s_mb_preallocated;
1673 atomic_t s_mb_discarded;
1674 atomic_t s_lock_busy;
1675
1676 /* locality groups */
1677 struct ext4_locality_group __percpu *s_locality_groups;
1678
1679 /* for write statistics */
1680 unsigned long s_sectors_written_start;
1681 u64 s_kbytes_written;
1682
1683 /* the size of zero-out chunk */
1684 unsigned int s_extent_max_zeroout_kb;
1685
1686 unsigned int s_log_groups_per_flex;
1687 struct flex_groups * __rcu *s_flex_groups;
1688 ext4_group_t s_flex_groups_allocated;
1689
1690 /* workqueue for reserved extent conversions (buffered io) */
1691 struct workqueue_struct *rsv_conversion_wq;
1692
1693 /* timer for periodic error stats printing */
1694 struct timer_list s_err_report;
1695
1696 /* Lazy inode table initialization info */
1697 struct ext4_li_request *s_li_request;
1698 /* Wait multiplier for lazy initialization thread */
1699 unsigned int s_li_wait_mult;
1700
1701 /* Kernel thread for multiple mount protection */
1702 struct task_struct *s_mmp_tsk;
1703
1704 /* record the last minlen when FITRIM is called. */
1705 unsigned long s_last_trim_minblks;
1706
1707 /* minimum folio order of a page cache allocation */
1708 u16 s_min_folio_order;
1709 /* supported maximum folio order, 0 means not supported */
1710 u16 s_max_folio_order;
1711
1712 /* Precomputed FS UUID checksum for seeding other checksums */
1713 __u32 s_csum_seed;
1714
1715 /* Reclaim extents from extent status tree */
1716 struct shrinker *s_es_shrinker;
1717 struct list_head s_es_list; /* List of inodes with reclaimable extents */
1718 long s_es_nr_inode;
1719 struct ext4_es_stats s_es_stats;
1720 struct mb_cache *s_ea_block_cache;
1721 struct mb_cache *s_ea_inode_cache;
1722 spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1723
1724 /* Journal triggers for checksum computation */
1725 struct ext4_journal_trigger s_journal_triggers[EXT4_JOURNAL_TRIGGER_COUNT];
1726
1727 /* Ratelimit ext4 messages. */
1728 struct ratelimit_state s_err_ratelimit_state;
1729 struct ratelimit_state s_warning_ratelimit_state;
1730 struct ratelimit_state s_msg_ratelimit_state;
1731 atomic_t s_warning_count;
1732 atomic_t s_msg_count;
1733
1734 /* Encryption policy for '-o test_dummy_encryption' */
1735 struct fscrypt_dummy_policy s_dummy_enc_policy;
1736
1737 /*
1738 * Barrier between writepages ops and changing any inode's JOURNAL_DATA
1739 * or EXTENTS flag or between writepages ops and changing DELALLOC or
1740 * DIOREAD_NOLOCK mount options on remount.
1741 */
1742 struct percpu_rw_semaphore s_writepages_rwsem;
1743 struct dax_device *s_daxdev;
1744 u64 s_dax_part_off;
1745#ifdef CONFIG_EXT4_DEBUG
1746 unsigned long s_simulate_fail;
1747#endif
1748 /* Record the errseq of the backing block device */
1749 errseq_t s_bdev_wb_err;
1750 spinlock_t s_bdev_wb_lock;
1751
1752 /* Information about errors that happened during this mount */
1753 spinlock_t s_error_lock;
1754 int s_add_error_count;
1755 int s_first_error_code;
1756 __u32 s_first_error_line;
1757 __u32 s_first_error_ino;
1758 __u64 s_first_error_block;
1759 const char *s_first_error_func;
1760 time64_t s_first_error_time;
1761 int s_last_error_code;
1762 __u32 s_last_error_line;
1763 __u32 s_last_error_ino;
1764 __u64 s_last_error_block;
1765 const char *s_last_error_func;
1766 time64_t s_last_error_time;
1767 /*
1768 * If we are in a context where we cannot update the on-disk
1769 * superblock, we queue the work here. This is used to update
1770 * the error information in the superblock, and for periodic
1771 * updates of the superblock called from the commit callback
1772 * function.
1773 */
1774 struct work_struct s_sb_upd_work;
1775
1776 /* Atomic write unit values in bytes */
1777 unsigned int s_awu_min;
1778 unsigned int s_awu_max;
1779
1780 /* Ext4 fast commit sub transaction ID */
1781 atomic_t s_fc_subtid;
1782
1783 /*
1784 * After commit starts, the main queue gets locked, and the further
1785 * updates get added in the staging queue.
1786 */
1787#define FC_Q_MAIN 0
1788#define FC_Q_STAGING 1
1789 struct list_head s_fc_q[2]; /* Inodes staged for fast commit
1790 * that have data changes in them.
1791 */
1792 struct list_head s_fc_dentry_q[2]; /* directory entry updates */
1793 unsigned int s_fc_bytes;
1794 /*
1795 * Main fast commit lock. This lock protects accesses to the
1796 * following fields:
1797 * ei->i_fc_list, s_fc_dentry_q, s_fc_q, s_fc_bytes, s_fc_bh.
1798 */
1799 struct mutex s_fc_lock;
1800 struct buffer_head *s_fc_bh;
1801 struct ext4_fc_stats s_fc_stats;
1802 tid_t s_fc_ineligible_tid;
1803#ifdef CONFIG_EXT4_DEBUG
1804 int s_fc_debug_max_replay;
1805#endif
1806 struct ext4_fc_replay_state s_fc_replay_state;
1807};
1808
1809static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1810{
1811 return sb->s_fs_info;
1812}
1813static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1814{
1815 return container_of(inode, struct ext4_inode_info, vfs_inode);
1816}
1817
1818static inline int ext4_writepages_down_read(struct super_block *sb)
1819{
1820 percpu_down_read(&EXT4_SB(sb)->s_writepages_rwsem);
1821 return memalloc_nofs_save();
1822}
1823
1824static inline void ext4_writepages_up_read(struct super_block *sb, int ctx)
1825{
1826 memalloc_nofs_restore(ctx);
1827 percpu_up_read(&EXT4_SB(sb)->s_writepages_rwsem);
1828}
1829
1830static inline int ext4_writepages_down_write(struct super_block *sb)
1831{
1832 percpu_down_write(&EXT4_SB(sb)->s_writepages_rwsem);
1833 return memalloc_nofs_save();
1834}
1835
1836static inline void ext4_writepages_up_write(struct super_block *sb, int ctx)
1837{
1838 memalloc_nofs_restore(ctx);
1839 percpu_up_write(&EXT4_SB(sb)->s_writepages_rwsem);
1840}
1841
1842static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1843{
1844 return ino == EXT4_ROOT_INO ||
1845 (ino >= EXT4_FIRST_INO(sb) &&
1846 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1847}
1848
1849static inline int ext4_get_resuid(struct ext4_super_block *es)
1850{
1851 return le16_to_cpu(es->s_def_resuid) |
1852 le16_to_cpu(es->s_def_resuid_hi) << 16;
1853}
1854
1855static inline int ext4_get_resgid(struct ext4_super_block *es)
1856{
1857 return le16_to_cpu(es->s_def_resgid) |
1858 le16_to_cpu(es->s_def_resgid_hi) << 16;
1859}
1860
1861/*
1862 * Returns: sbi->field[index]
1863 * Used to access an array element from the following sbi fields which require
1864 * rcu protection to avoid dereferencing an invalid pointer due to reassignment
1865 * - s_group_desc
1866 * - s_group_info
1867 * - s_flex_group
1868 */
1869#define sbi_array_rcu_deref(sbi, field, index) \
1870({ \
1871 typeof(*((sbi)->field)) _v; \
1872 rcu_read_lock(); \
1873 _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index]; \
1874 rcu_read_unlock(); \
1875 _v; \
1876})
1877
1878/*
1879 * run-time mount flags
1880 */
1881enum {
1882 EXT4_MF_MNTDIR_SAMPLED,
1883 EXT4_MF_FC_INELIGIBLE, /* Fast commit ineligible */
1884 EXT4_MF_JOURNAL_DESTROY /* Journal is in process of destroying */
1885};
1886
1887static inline void ext4_set_mount_flag(struct super_block *sb, int bit)
1888{
1889 set_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1890}
1891
1892static inline void ext4_clear_mount_flag(struct super_block *sb, int bit)
1893{
1894 clear_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1895}
1896
1897static inline int ext4_test_mount_flag(struct super_block *sb, int bit)
1898{
1899 return test_bit(bit, &EXT4_SB(sb)->s_mount_flags);
1900}
1901
1902
1903/*
1904 * Simulate_fail codes
1905 */
1906#define EXT4_SIM_BBITMAP_EIO 1
1907#define EXT4_SIM_BBITMAP_CRC 2
1908#define EXT4_SIM_IBITMAP_EIO 3
1909#define EXT4_SIM_IBITMAP_CRC 4
1910#define EXT4_SIM_INODE_EIO 5
1911#define EXT4_SIM_INODE_CRC 6
1912#define EXT4_SIM_DIRBLOCK_EIO 7
1913#define EXT4_SIM_DIRBLOCK_CRC 8
1914
1915static inline bool ext4_simulate_fail(struct super_block *sb,
1916 unsigned long code)
1917{
1918#ifdef CONFIG_EXT4_DEBUG
1919 struct ext4_sb_info *sbi = EXT4_SB(sb);
1920
1921 if (unlikely(sbi->s_simulate_fail == code)) {
1922 sbi->s_simulate_fail = 0;
1923 return true;
1924 }
1925#endif
1926 return false;
1927}
1928
1929/*
1930 * Error number codes for s_{first,last}_error_errno
1931 *
1932 * Linux errno numbers are architecture specific, so we need to translate
1933 * them into something which is architecture independent. We don't define
1934 * codes for all errno's; just the ones which are most likely to be the cause
1935 * of an ext4_error() call.
1936 */
1937#define EXT4_ERR_UNKNOWN 1
1938#define EXT4_ERR_EIO 2
1939#define EXT4_ERR_ENOMEM 3
1940#define EXT4_ERR_EFSBADCRC 4
1941#define EXT4_ERR_EFSCORRUPTED 5
1942#define EXT4_ERR_ENOSPC 6
1943#define EXT4_ERR_ENOKEY 7
1944#define EXT4_ERR_EROFS 8
1945#define EXT4_ERR_EFBIG 9
1946#define EXT4_ERR_EEXIST 10
1947#define EXT4_ERR_ERANGE 11
1948#define EXT4_ERR_EOVERFLOW 12
1949#define EXT4_ERR_EBUSY 13
1950#define EXT4_ERR_ENOTDIR 14
1951#define EXT4_ERR_ENOTEMPTY 15
1952#define EXT4_ERR_ESHUTDOWN 16
1953#define EXT4_ERR_EFAULT 17
1954
1955/*
1956 * Inode dynamic state flags
1957 */
1958enum {
1959 EXT4_STATE_NEW, /* inode is newly created */
1960 EXT4_STATE_XATTR, /* has in-inode xattrs */
1961 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1962 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1963 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1964 EXT4_STATE_NEWENTRY, /* File just added to dir */
1965 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1966 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
1967 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */
1968 EXT4_STATE_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */
1969 EXT4_STATE_FC_COMMITTING, /* Fast commit ongoing */
1970 EXT4_STATE_FC_FLUSHING_DATA, /* Fast commit flushing data */
1971 EXT4_STATE_ORPHAN_FILE, /* Inode orphaned in orphan file */
1972};
1973
1974#define EXT4_INODE_BIT_FNS(name, field, offset) \
1975static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1976{ \
1977 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1978} \
1979static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1980{ \
1981 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1982} \
1983static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1984{ \
1985 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1986}
1987
1988/* Add these declarations here only so that these functions can be
1989 * found by name. Otherwise, they are very hard to locate. */
1990static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1991static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1992static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1993EXT4_INODE_BIT_FNS(flag, flags, 0)
1994
1995/* Add these declarations here only so that these functions can be
1996 * found by name. Otherwise, they are very hard to locate. */
1997static inline int ext4_test_inode_state(struct inode *inode, int bit);
1998static inline void ext4_set_inode_state(struct inode *inode, int bit);
1999static inline void ext4_clear_inode_state(struct inode *inode, int bit);
2000#if (BITS_PER_LONG < 64)
2001EXT4_INODE_BIT_FNS(state, state_flags, 0)
2002
2003static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
2004{
2005 (ei)->i_state_flags = 0;
2006}
2007#else
2008EXT4_INODE_BIT_FNS(state, flags, 32)
2009
2010static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
2011{
2012 /* We depend on the fact that callers will set i_flags */
2013}
2014#endif
2015#else
2016/* Assume that user mode programs are passing in an ext4fs superblock, not
2017 * a kernel struct super_block. This will allow us to call the feature-test
2018 * macros from user land. */
2019#define EXT4_SB(sb) (sb)
2020#endif
2021
2022static inline bool ext4_verity_in_progress(struct inode *inode)
2023{
2024 return IS_ENABLED(CONFIG_FS_VERITY) &&
2025 ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS);
2026}
2027
2028#define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
2029
2030/*
2031 * Check whether the inode is tracked as orphan (either in orphan file or
2032 * orphan list).
2033 */
2034static inline bool ext4_inode_orphan_tracked(struct inode *inode)
2035{
2036 return ext4_test_inode_state(inode, EXT4_STATE_ORPHAN_FILE) ||
2037 !list_empty(&EXT4_I(inode)->i_orphan);
2038}
2039
2040/*
2041 * Codes for operating systems
2042 */
2043#define EXT4_OS_LINUX 0
2044#define EXT4_OS_HURD 1
2045#define EXT4_OS_MASIX 2
2046#define EXT4_OS_FREEBSD 3
2047#define EXT4_OS_LITES 4
2048
2049/*
2050 * Revision levels
2051 */
2052#define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
2053#define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
2054
2055#define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
2056
2057#define EXT4_GOOD_OLD_INODE_SIZE 128
2058
2059#define EXT4_EXTRA_TIMESTAMP_MAX (((s64)1 << 34) - 1 + S32_MIN)
2060#define EXT4_NON_EXTRA_TIMESTAMP_MAX S32_MAX
2061#define EXT4_TIMESTAMP_MIN S32_MIN
2062
2063/*
2064 * Feature set definitions
2065 */
2066
2067#define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
2068#define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
2069#define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
2070#define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
2071#define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
2072#define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
2073#define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
2074/*
2075 * The reason why "FAST_COMMIT" is a compat feature is that, FS becomes
2076 * incompatible only if fast commit blocks are present in the FS. Since we
2077 * clear the journal (and thus the fast commit blocks), we don't mark FS as
2078 * incompatible. We also have a JBD2 incompat feature, which gets set when
2079 * there are fast commit blocks present in the journal.
2080 */
2081#define EXT4_FEATURE_COMPAT_FAST_COMMIT 0x0400
2082#define EXT4_FEATURE_COMPAT_STABLE_INODES 0x0800
2083#define EXT4_FEATURE_COMPAT_ORPHAN_FILE 0x1000 /* Orphan file exists */
2084
2085#define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
2086#define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
2087#define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
2088#define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
2089#define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
2090#define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
2091#define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
2092#define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
2093#define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
2094/*
2095 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
2096 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
2097 * all other data structures' checksums. However, the METADATA_CSUM and
2098 * GDT_CSUM bits are mutually exclusive.
2099 */
2100#define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
2101#define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000
2102#define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000
2103#define EXT4_FEATURE_RO_COMPAT_VERITY 0x8000
2104#define EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT 0x10000 /* Orphan file may be
2105 non-empty */
2106
2107#define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
2108#define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
2109#define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
2110#define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
2111#define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
2112#define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
2113#define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
2114#define EXT4_FEATURE_INCOMPAT_MMP 0x0100
2115#define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
2116#define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
2117#define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
2118#define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000
2119#define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
2120#define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
2121#define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000
2122#define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000
2123
2124extern void ext4_update_dynamic_rev(struct super_block *sb);
2125
2126#define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
2127static inline bool ext4_has_feature_##name(struct super_block *sb) \
2128{ \
2129 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
2130 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
2131} \
2132static inline void ext4_set_feature_##name(struct super_block *sb) \
2133{ \
2134 ext4_update_dynamic_rev(sb); \
2135 EXT4_SB(sb)->s_es->s_feature_compat |= \
2136 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
2137} \
2138static inline void ext4_clear_feature_##name(struct super_block *sb) \
2139{ \
2140 EXT4_SB(sb)->s_es->s_feature_compat &= \
2141 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
2142}
2143
2144#define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
2145static inline bool ext4_has_feature_##name(struct super_block *sb) \
2146{ \
2147 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
2148 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
2149} \
2150static inline void ext4_set_feature_##name(struct super_block *sb) \
2151{ \
2152 ext4_update_dynamic_rev(sb); \
2153 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
2154 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
2155} \
2156static inline void ext4_clear_feature_##name(struct super_block *sb) \
2157{ \
2158 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
2159 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
2160}
2161
2162#define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
2163static inline bool ext4_has_feature_##name(struct super_block *sb) \
2164{ \
2165 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
2166 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
2167} \
2168static inline void ext4_set_feature_##name(struct super_block *sb) \
2169{ \
2170 ext4_update_dynamic_rev(sb); \
2171 EXT4_SB(sb)->s_es->s_feature_incompat |= \
2172 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
2173} \
2174static inline void ext4_clear_feature_##name(struct super_block *sb) \
2175{ \
2176 EXT4_SB(sb)->s_es->s_feature_incompat &= \
2177 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
2178}
2179
2180EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC)
2181EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES)
2182EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL)
2183EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR)
2184EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE)
2185EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX)
2186EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2)
2187EXT4_FEATURE_COMPAT_FUNCS(fast_commit, FAST_COMMIT)
2188EXT4_FEATURE_COMPAT_FUNCS(stable_inodes, STABLE_INODES)
2189EXT4_FEATURE_COMPAT_FUNCS(orphan_file, ORPHAN_FILE)
2190
2191EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER)
2192EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE)
2193EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR)
2194EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE)
2195EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM)
2196EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK)
2197EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE)
2198EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA)
2199EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC)
2200EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM)
2201EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY)
2202EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT)
2203EXT4_FEATURE_RO_COMPAT_FUNCS(verity, VERITY)
2204EXT4_FEATURE_RO_COMPAT_FUNCS(orphan_present, ORPHAN_PRESENT)
2205
2206EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION)
2207EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE)
2208EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER)
2209EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV)
2210EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG)
2211EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS)
2212EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT)
2213EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP)
2214EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG)
2215EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE)
2216EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA)
2217EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED)
2218EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR)
2219EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA)
2220EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT)
2221EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD)
2222
2223#define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
2224#define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
2225 EXT4_FEATURE_INCOMPAT_META_BG)
2226#define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2227 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2228 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
2229
2230#define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
2231#define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
2232 EXT4_FEATURE_INCOMPAT_RECOVER| \
2233 EXT4_FEATURE_INCOMPAT_META_BG)
2234#define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2235 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2236 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
2237
2238#define EXT4_FEATURE_COMPAT_SUPP (EXT4_FEATURE_COMPAT_EXT_ATTR| \
2239 EXT4_FEATURE_COMPAT_ORPHAN_FILE)
2240#define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
2241 EXT4_FEATURE_INCOMPAT_RECOVER| \
2242 EXT4_FEATURE_INCOMPAT_META_BG| \
2243 EXT4_FEATURE_INCOMPAT_EXTENTS| \
2244 EXT4_FEATURE_INCOMPAT_64BIT| \
2245 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
2246 EXT4_FEATURE_INCOMPAT_EA_INODE| \
2247 EXT4_FEATURE_INCOMPAT_MMP | \
2248 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
2249 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
2250 EXT4_FEATURE_INCOMPAT_CASEFOLD | \
2251 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \
2252 EXT4_FEATURE_INCOMPAT_LARGEDIR)
2253#define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
2254 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
2255 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
2256 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
2257 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
2258 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
2259 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
2260 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
2261 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
2262 EXT4_FEATURE_RO_COMPAT_QUOTA |\
2263 EXT4_FEATURE_RO_COMPAT_PROJECT |\
2264 EXT4_FEATURE_RO_COMPAT_VERITY |\
2265 EXT4_FEATURE_RO_COMPAT_ORPHAN_PRESENT)
2266
2267#define EXTN_FEATURE_FUNCS(ver) \
2268static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
2269{ \
2270 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
2271 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
2272} \
2273static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
2274{ \
2275 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
2276 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
2277} \
2278static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
2279{ \
2280 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
2281 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
2282}
2283
2284EXTN_FEATURE_FUNCS(2)
2285EXTN_FEATURE_FUNCS(3)
2286EXTN_FEATURE_FUNCS(4)
2287
2288static inline bool ext4_has_compat_features(struct super_block *sb)
2289{
2290 return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
2291}
2292static inline bool ext4_has_ro_compat_features(struct super_block *sb)
2293{
2294 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
2295}
2296static inline bool ext4_has_incompat_features(struct super_block *sb)
2297{
2298 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
2299}
2300
2301extern int ext4_feature_set_ok(struct super_block *sb, int readonly);
2302
2303/*
2304 * Superblock flags
2305 */
2306enum {
2307 EXT4_FLAGS_RESIZING, /* Avoid superblock update and resize race */
2308 EXT4_FLAGS_SHUTDOWN, /* Prevent access to the file system */
2309 EXT4_FLAGS_BDEV_IS_DAX, /* Current block device support DAX */
2310 EXT4_FLAGS_EMERGENCY_RO,/* Emergency read-only due to fs errors */
2311};
2312
2313static inline int ext4_forced_shutdown(struct super_block *sb)
2314{
2315 return test_bit(EXT4_FLAGS_SHUTDOWN, &EXT4_SB(sb)->s_ext4_flags);
2316}
2317
2318static inline int ext4_emergency_ro(struct super_block *sb)
2319{
2320 return test_bit(EXT4_FLAGS_EMERGENCY_RO, &EXT4_SB(sb)->s_ext4_flags);
2321}
2322
2323static inline int ext4_emergency_state(struct super_block *sb)
2324{
2325 if (unlikely(ext4_forced_shutdown(sb)))
2326 return -EIO;
2327 if (unlikely(ext4_emergency_ro(sb)))
2328 return -EROFS;
2329 return 0;
2330}
2331
2332/*
2333 * Default values for user and/or group using reserved blocks
2334 */
2335#define EXT4_DEF_RESUID 0
2336#define EXT4_DEF_RESGID 0
2337
2338/*
2339 * Default project ID
2340 */
2341#define EXT4_DEF_PROJID 0
2342
2343#define EXT4_DEF_INODE_READAHEAD_BLKS 32
2344
2345/*
2346 * Default mount options
2347 */
2348#define EXT4_DEFM_DEBUG 0x0001
2349#define EXT4_DEFM_BSDGROUPS 0x0002
2350#define EXT4_DEFM_XATTR_USER 0x0004
2351#define EXT4_DEFM_ACL 0x0008
2352#define EXT4_DEFM_UID16 0x0010
2353#define EXT4_DEFM_JMODE 0x0060
2354#define EXT4_DEFM_JMODE_DATA 0x0020
2355#define EXT4_DEFM_JMODE_ORDERED 0x0040
2356#define EXT4_DEFM_JMODE_WBACK 0x0060
2357#define EXT4_DEFM_NOBARRIER 0x0100
2358#define EXT4_DEFM_BLOCK_VALIDITY 0x0200
2359#define EXT4_DEFM_DISCARD 0x0400
2360#define EXT4_DEFM_NODELALLOC 0x0800
2361
2362/*
2363 * Default journal batch times and ioprio.
2364 */
2365#define EXT4_DEF_MIN_BATCH_TIME 0
2366#define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
2367#define EXT4_DEF_JOURNAL_IOPRIO (IOPRIO_PRIO_VALUE(IOPRIO_CLASS_BE, 3))
2368
2369
2370/*
2371 * Default values for superblock update
2372 */
2373#define EXT4_DEF_SB_UPDATE_INTERVAL_SEC (3600) /* seconds (1 hour) */
2374#define EXT4_DEF_SB_UPDATE_INTERVAL_KB (16384) /* kilobytes (16MB) */
2375
2376
2377/*
2378 * Minimum number of groups in a flexgroup before we separate out
2379 * directories into the first block group of a flexgroup
2380 */
2381#define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
2382
2383/*
2384 * Structure of a directory entry
2385 */
2386#define EXT4_NAME_LEN 255
2387/*
2388 * Base length of the ext4 directory entry excluding the name length
2389 */
2390#define EXT4_BASE_DIR_LEN (sizeof(struct ext4_dir_entry_2) - EXT4_NAME_LEN)
2391
2392struct ext4_dir_entry {
2393 __le32 inode; /* Inode number */
2394 __le16 rec_len; /* Directory entry length */
2395 __le16 name_len; /* Name length */
2396 char name[EXT4_NAME_LEN]; /* File name */
2397};
2398
2399
2400/*
2401 * Encrypted Casefolded entries require saving the hash on disk. This structure
2402 * followed ext4_dir_entry_2's name[name_len] at the next 4 byte aligned
2403 * boundary.
2404 */
2405struct ext4_dir_entry_hash {
2406 __le32 hash;
2407 __le32 minor_hash;
2408};
2409
2410/*
2411 * The new version of the directory entry. Since EXT4 structures are
2412 * stored in intel byte order, and the name_len field could never be
2413 * bigger than 255 chars, it's safe to reclaim the extra byte for the
2414 * file_type field.
2415 */
2416struct ext4_dir_entry_2 {
2417 __le32 inode; /* Inode number */
2418 __le16 rec_len; /* Directory entry length */
2419 __u8 name_len; /* Name length */
2420 __u8 file_type; /* See file type macros EXT4_FT_* below */
2421 char name[EXT4_NAME_LEN]; /* File name */
2422};
2423
2424/*
2425 * Access the hashes at the end of ext4_dir_entry_2
2426 */
2427#define EXT4_DIRENT_HASHES(entry) \
2428 ((struct ext4_dir_entry_hash *) \
2429 (((void *)(entry)) + \
2430 ((8 + (entry)->name_len + EXT4_DIR_ROUND) & ~EXT4_DIR_ROUND)))
2431#define EXT4_DIRENT_HASH(entry) le32_to_cpu(EXT4_DIRENT_HASHES(entry)->hash)
2432#define EXT4_DIRENT_MINOR_HASH(entry) \
2433 le32_to_cpu(EXT4_DIRENT_HASHES(entry)->minor_hash)
2434
2435static inline bool ext4_hash_in_dirent(const struct inode *inode)
2436{
2437 return IS_CASEFOLDED(inode) && IS_ENCRYPTED(inode);
2438}
2439
2440/*
2441 * This is a bogus directory entry at the end of each leaf block that
2442 * records checksums.
2443 */
2444struct ext4_dir_entry_tail {
2445 __le32 det_reserved_zero1; /* Pretend to be unused */
2446 __le16 det_rec_len; /* 12 */
2447 __u8 det_reserved_zero2; /* Zero name length */
2448 __u8 det_reserved_ft; /* 0xDE, fake file type */
2449 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
2450};
2451
2452#define EXT4_DIRENT_TAIL(block, blocksize) \
2453 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
2454 ((blocksize) - \
2455 sizeof(struct ext4_dir_entry_tail))))
2456
2457/*
2458 * Ext4 directory file types. Only the low 3 bits are used. The
2459 * other bits are reserved for now.
2460 */
2461#define EXT4_FT_UNKNOWN 0
2462#define EXT4_FT_REG_FILE 1
2463#define EXT4_FT_DIR 2
2464#define EXT4_FT_CHRDEV 3
2465#define EXT4_FT_BLKDEV 4
2466#define EXT4_FT_FIFO 5
2467#define EXT4_FT_SOCK 6
2468#define EXT4_FT_SYMLINK 7
2469
2470#define EXT4_FT_MAX 8
2471
2472#define EXT4_FT_DIR_CSUM 0xDE
2473
2474/*
2475 * EXT4_DIR_PAD defines the directory entries boundaries
2476 *
2477 * NOTE: It must be a multiple of 4
2478 */
2479#define EXT4_DIR_PAD 4
2480#define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
2481#define EXT4_MAX_REC_LEN ((1<<16)-1)
2482
2483/*
2484 * The rec_len is dependent on the type of directory. Directories that are
2485 * casefolded and encrypted need to store the hash as well, so we add room for
2486 * ext4_extended_dir_entry_2. For all entries related to '.' or '..' you should
2487 * pass NULL for dir, as those entries do not use the extra fields.
2488 */
2489static inline unsigned int ext4_dir_rec_len(__u8 name_len,
2490 const struct inode *dir)
2491{
2492 int rec_len = (name_len + 8 + EXT4_DIR_ROUND);
2493
2494 if (dir && ext4_hash_in_dirent(dir))
2495 rec_len += sizeof(struct ext4_dir_entry_hash);
2496 return (rec_len & ~EXT4_DIR_ROUND);
2497}
2498
2499static inline unsigned int
2500ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
2501{
2502 unsigned len = le16_to_cpu(dlen);
2503
2504 if (len == EXT4_MAX_REC_LEN || len == 0)
2505 return blocksize;
2506 return (len & 65532) | ((len & 3) << 16);
2507}
2508
2509static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
2510{
2511 BUG_ON((len > blocksize) || (blocksize > (1 << 18)) || (len & 3));
2512 if (len < 65536)
2513 return cpu_to_le16(len);
2514 if (len == blocksize) {
2515 if (blocksize == 65536)
2516 return cpu_to_le16(EXT4_MAX_REC_LEN);
2517 else
2518 return cpu_to_le16(0);
2519 }
2520 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
2521}
2522
2523/*
2524 * Hash Tree Directory indexing
2525 * (c) Daniel Phillips, 2001
2526 */
2527
2528#define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
2529 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2530#define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \
2531 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir)))
2532#define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2533
2534/* Legal values for the dx_root hash_version field: */
2535
2536#define DX_HASH_LEGACY 0
2537#define DX_HASH_HALF_MD4 1
2538#define DX_HASH_TEA 2
2539#define DX_HASH_LEGACY_UNSIGNED 3
2540#define DX_HASH_HALF_MD4_UNSIGNED 4
2541#define DX_HASH_TEA_UNSIGNED 5
2542#define DX_HASH_SIPHASH 6
2543#define DX_HASH_LAST DX_HASH_SIPHASH
2544
2545static inline u32 ext4_chksum(u32 crc, const void *address, unsigned int length)
2546{
2547 return crc32c(crc, address, length);
2548}
2549
2550#ifdef __KERNEL__
2551
2552/* hash info structure used by the directory hash */
2553struct dx_hash_info
2554{
2555 u32 hash;
2556 u32 minor_hash;
2557 int hash_version;
2558 u32 *seed;
2559};
2560
2561
2562/* 32 and 64 bit signed EOF for dx directories */
2563#define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
2564#define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
2565
2566
2567/*
2568 * Control parameters used by ext4_htree_next_block
2569 */
2570#define HASH_NB_ALWAYS 1
2571
2572struct ext4_filename {
2573 const struct qstr *usr_fname;
2574 struct fscrypt_str disk_name;
2575 struct dx_hash_info hinfo;
2576#ifdef CONFIG_FS_ENCRYPTION
2577 struct fscrypt_str crypto_buf;
2578#endif
2579#if IS_ENABLED(CONFIG_UNICODE)
2580 struct qstr cf_name;
2581#endif
2582};
2583
2584#define fname_name(p) ((p)->disk_name.name)
2585#define fname_usr_name(p) ((p)->usr_fname->name)
2586#define fname_len(p) ((p)->disk_name.len)
2587
2588/*
2589 * Describe an inode's exact location on disk and in memory
2590 */
2591struct ext4_iloc
2592{
2593 struct buffer_head *bh;
2594 unsigned long offset;
2595 ext4_group_t block_group;
2596};
2597
2598static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2599{
2600 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2601}
2602
2603static inline bool ext4_is_quota_file(struct inode *inode)
2604{
2605 return IS_NOQUOTA(inode) &&
2606 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL);
2607}
2608
2609/*
2610 * This structure is stuffed into the struct file's private_data field
2611 * for directories. It is where we put information so that we can do
2612 * readdir operations in hash tree order.
2613 */
2614struct dir_private_info {
2615 struct rb_root root;
2616 struct rb_node *curr_node;
2617 struct fname *extra_fname;
2618 loff_t last_pos;
2619 __u32 curr_hash;
2620 __u32 curr_minor_hash;
2621 __u32 next_hash;
2622 u64 cookie;
2623 bool initialized;
2624};
2625
2626/* calculate the first block number of the group */
2627static inline ext4_fsblk_t
2628ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2629{
2630 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2631 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2632}
2633
2634/*
2635 * Special error return code only used by dx_probe() and its callers.
2636 */
2637#define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
2638
2639/* htree levels for ext4 */
2640#define EXT4_HTREE_LEVEL_COMPAT 2
2641#define EXT4_HTREE_LEVEL 3
2642
2643static inline int ext4_dir_htree_level(struct super_block *sb)
2644{
2645 return ext4_has_feature_largedir(sb) ?
2646 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT;
2647}
2648
2649/*
2650 * Timeout and state flag for lazy initialization inode thread.
2651 */
2652#define EXT4_DEF_LI_WAIT_MULT 10
2653#define EXT4_DEF_LI_MAX_START_DELAY 5
2654#define EXT4_LAZYINIT_QUIT 0x0001
2655#define EXT4_LAZYINIT_RUNNING 0x0002
2656
2657/*
2658 * Lazy inode table initialization info
2659 */
2660struct ext4_lazy_init {
2661 unsigned long li_state;
2662 struct list_head li_request_list;
2663 struct mutex li_list_mtx;
2664};
2665
2666enum ext4_li_mode {
2667 EXT4_LI_MODE_PREFETCH_BBITMAP,
2668 EXT4_LI_MODE_ITABLE,
2669};
2670
2671struct ext4_li_request {
2672 struct super_block *lr_super;
2673 enum ext4_li_mode lr_mode;
2674 ext4_group_t lr_first_not_zeroed;
2675 ext4_group_t lr_next_group;
2676 struct list_head lr_request;
2677 unsigned long lr_next_sched;
2678 unsigned long lr_timeout;
2679};
2680
2681struct ext4_features {
2682 struct kobject f_kobj;
2683 struct completion f_kobj_unregister;
2684};
2685
2686/*
2687 * This structure will be used for multiple mount protection. It will be
2688 * written into the block number saved in the s_mmp_block field in the
2689 * superblock. Programs that check MMP should assume that if
2690 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2691 * to use the filesystem, regardless of how old the timestamp is.
2692 */
2693#define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
2694#define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2695#define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
2696#define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
2697
2698struct mmp_struct {
2699 __le32 mmp_magic; /* Magic number for MMP */
2700 __le32 mmp_seq; /* Sequence no. updated periodically */
2701
2702 /*
2703 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2704 * purposes and do not affect the correctness of the algorithm
2705 */
2706 __le64 mmp_time; /* Time last updated */
2707 char mmp_nodename[64]; /* Node which last updated MMP block */
2708 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
2709
2710 /*
2711 * mmp_check_interval is used to verify if the MMP block has been
2712 * updated on the block device. The value is updated based on the
2713 * maximum time to write the MMP block during an update cycle.
2714 */
2715 __le16 mmp_check_interval;
2716
2717 __le16 mmp_pad1;
2718 __le32 mmp_pad2[226];
2719 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
2720};
2721
2722/* arguments passed to the mmp thread */
2723struct mmpd_data {
2724 struct buffer_head *bh; /* bh from initial read_mmp_block() */
2725 struct super_block *sb; /* super block of the fs */
2726};
2727
2728/*
2729 * Check interval multiplier
2730 * The MMP block is written every update interval and initially checked every
2731 * update interval x the multiplier (the value is then adapted based on the
2732 * write latency). The reason is that writes can be delayed under load and we
2733 * don't want readers to incorrectly assume that the filesystem is no longer
2734 * in use.
2735 */
2736#define EXT4_MMP_CHECK_MULT 2UL
2737
2738/*
2739 * Minimum interval for MMP checking in seconds.
2740 */
2741#define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
2742
2743/*
2744 * Maximum interval for MMP checking in seconds.
2745 */
2746#define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
2747
2748/*
2749 * Function prototypes
2750 */
2751
2752/*
2753 * Ok, these declarations are also in <linux/kernel.h> but none of the
2754 * ext4 source programs needs to include it so they are duplicated here.
2755 */
2756# define NORET_TYPE /**/
2757# define ATTRIB_NORET __attribute__((noreturn))
2758# define NORET_AND noreturn,
2759
2760/* bitmap.c */
2761extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2762void ext4_inode_bitmap_csum_set(struct super_block *sb,
2763 struct ext4_group_desc *gdp,
2764 struct buffer_head *bh);
2765int ext4_inode_bitmap_csum_verify(struct super_block *sb,
2766 struct ext4_group_desc *gdp,
2767 struct buffer_head *bh);
2768void ext4_block_bitmap_csum_set(struct super_block *sb,
2769 struct ext4_group_desc *gdp,
2770 struct buffer_head *bh);
2771int ext4_block_bitmap_csum_verify(struct super_block *sb,
2772 struct ext4_group_desc *gdp,
2773 struct buffer_head *bh);
2774
2775/* balloc.c */
2776extern void ext4_get_group_no_and_offset(struct super_block *sb,
2777 ext4_fsblk_t blocknr,
2778 ext4_group_t *blockgrpp,
2779 ext4_grpblk_t *offsetp);
2780extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2781 ext4_fsblk_t block);
2782
2783extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2784extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2785 ext4_group_t group);
2786extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2787 ext4_fsblk_t goal,
2788 unsigned int flags,
2789 unsigned long *count,
2790 int *errp);
2791extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2792 s64 nclusters, unsigned int flags);
2793extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2794extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2795 ext4_group_t block_group,
2796 struct buffer_head ** bh);
2797extern struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2798 ext4_group_t group);
2799extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2800
2801extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2802 ext4_group_t block_group,
2803 bool ignore_locked);
2804extern int ext4_wait_block_bitmap(struct super_block *sb,
2805 ext4_group_t block_group,
2806 struct buffer_head *bh);
2807extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2808 ext4_group_t block_group);
2809extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2810 ext4_group_t block_group,
2811 struct ext4_group_desc *gdp);
2812ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2813
2814#if IS_ENABLED(CONFIG_UNICODE)
2815extern int ext4_fname_setup_ci_filename(struct inode *dir,
2816 const struct qstr *iname,
2817 struct ext4_filename *fname);
2818
2819static inline void ext4_fname_free_ci_filename(struct ext4_filename *fname)
2820{
2821 kfree(fname->cf_name.name);
2822 fname->cf_name.name = NULL;
2823}
2824#else
2825static inline int ext4_fname_setup_ci_filename(struct inode *dir,
2826 const struct qstr *iname,
2827 struct ext4_filename *fname)
2828{
2829 return 0;
2830}
2831
2832static inline void ext4_fname_free_ci_filename(struct ext4_filename *fname)
2833{
2834}
2835#endif
2836
2837/* ext4 encryption related stuff goes here crypto.c */
2838#ifdef CONFIG_FS_ENCRYPTION
2839extern const struct fscrypt_operations ext4_cryptops;
2840
2841int ext4_fname_setup_filename(struct inode *dir, const struct qstr *iname,
2842 int lookup, struct ext4_filename *fname);
2843
2844int ext4_fname_prepare_lookup(struct inode *dir, struct dentry *dentry,
2845 struct ext4_filename *fname);
2846
2847void ext4_fname_free_filename(struct ext4_filename *fname);
2848
2849int ext4_ioctl_get_encryption_pwsalt(struct file *filp, void __user *arg);
2850
2851#else /* !CONFIG_FS_ENCRYPTION */
2852static inline int ext4_fname_setup_filename(struct inode *dir,
2853 const struct qstr *iname,
2854 int lookup,
2855 struct ext4_filename *fname)
2856{
2857 fname->usr_fname = iname;
2858 fname->disk_name.name = (unsigned char *) iname->name;
2859 fname->disk_name.len = iname->len;
2860
2861 return ext4_fname_setup_ci_filename(dir, iname, fname);
2862}
2863
2864static inline int ext4_fname_prepare_lookup(struct inode *dir,
2865 struct dentry *dentry,
2866 struct ext4_filename *fname)
2867{
2868 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname);
2869}
2870
2871static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2872{
2873 ext4_fname_free_ci_filename(fname);
2874}
2875
2876static inline int ext4_ioctl_get_encryption_pwsalt(struct file *filp,
2877 void __user *arg)
2878{
2879 return -EOPNOTSUPP;
2880}
2881#endif /* !CONFIG_FS_ENCRYPTION */
2882
2883/* dir.c */
2884extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2885 struct file *,
2886 struct ext4_dir_entry_2 *,
2887 struct buffer_head *, char *, int,
2888 unsigned int);
2889#define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2890 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2891 (de), (bh), (buf), (size), (offset)))
2892extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2893 __u32 minor_hash,
2894 struct ext4_dir_entry_2 *dirent,
2895 struct fscrypt_str *ent_name);
2896extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2897extern int ext4_find_dest_de(struct inode *dir, struct buffer_head *bh,
2898 void *buf, int buf_size,
2899 struct ext4_filename *fname,
2900 struct ext4_dir_entry_2 **dest_de);
2901void ext4_insert_dentry(struct inode *dir, struct inode *inode,
2902 struct ext4_dir_entry_2 *de,
2903 int buf_size,
2904 struct ext4_filename *fname);
2905static inline void ext4_update_dx_flag(struct inode *inode)
2906{
2907 if (!ext4_has_feature_dir_index(inode->i_sb) &&
2908 ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) {
2909 /* ext4_iget() should have caught this... */
2910 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb));
2911 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2912 }
2913}
2914static const unsigned char ext4_filetype_table[] = {
2915 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2916};
2917
2918static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2919{
2920 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2921 return DT_UNKNOWN;
2922
2923 return ext4_filetype_table[filetype];
2924}
2925extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2926 void *buf, int buf_size);
2927
2928/* fsync.c */
2929extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2930
2931/* hash.c */
2932extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len,
2933 struct dx_hash_info *hinfo);
2934
2935/* ialloc.c */
2936extern int ext4_mark_inode_used(struct super_block *sb, int ino);
2937extern struct inode *__ext4_new_inode(struct mnt_idmap *, handle_t *,
2938 struct inode *, umode_t,
2939 const struct qstr *qstr, __u32 goal,
2940 uid_t *owner, __u32 i_flags,
2941 int handle_type, unsigned int line_no,
2942 int nblocks);
2943
2944#define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \
2945 __ext4_new_inode(&nop_mnt_idmap, (handle), (dir), (mode), (qstr), \
2946 (goal), (owner), i_flags, 0, 0, 0)
2947#define ext4_new_inode_start_handle(idmap, dir, mode, qstr, goal, owner, \
2948 type, nblocks) \
2949 __ext4_new_inode((idmap), NULL, (dir), (mode), (qstr), (goal), (owner), \
2950 0, (type), __LINE__, (nblocks))
2951
2952
2953extern void ext4_free_inode(handle_t *, struct inode *);
2954extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2955extern unsigned long ext4_count_free_inodes(struct super_block *);
2956extern unsigned long ext4_count_dirs(struct super_block *);
2957extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2958extern int ext4_init_inode_table(struct super_block *sb,
2959 ext4_group_t group, int barrier);
2960extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2961
2962/* fast_commit.c */
2963int ext4_fc_info_show(struct seq_file *seq, void *v);
2964void ext4_fc_init(struct super_block *sb, journal_t *journal);
2965void ext4_fc_init_inode(struct inode *inode);
2966void ext4_fc_track_range(handle_t *handle, struct inode *inode, ext4_lblk_t start,
2967 ext4_lblk_t end);
2968void __ext4_fc_track_unlink(handle_t *handle, struct inode *inode,
2969 struct dentry *dentry);
2970void __ext4_fc_track_link(handle_t *handle, struct inode *inode,
2971 struct dentry *dentry);
2972void ext4_fc_track_unlink(handle_t *handle, struct dentry *dentry);
2973void ext4_fc_track_link(handle_t *handle, struct dentry *dentry);
2974void __ext4_fc_track_create(handle_t *handle, struct inode *inode,
2975 struct dentry *dentry);
2976void ext4_fc_track_create(handle_t *handle, struct dentry *dentry);
2977void ext4_fc_track_inode(handle_t *handle, struct inode *inode);
2978void ext4_fc_mark_ineligible(struct super_block *sb, int reason, handle_t *handle);
2979void ext4_fc_del(struct inode *inode);
2980bool ext4_fc_replay_check_excluded(struct super_block *sb, ext4_fsblk_t block);
2981void ext4_fc_replay_cleanup(struct super_block *sb);
2982int ext4_fc_commit(journal_t *journal, tid_t commit_tid);
2983int __init ext4_fc_init_dentry_cache(void);
2984void ext4_fc_destroy_dentry_cache(void);
2985int ext4_fc_record_regions(struct super_block *sb, int ino,
2986 ext4_lblk_t lblk, ext4_fsblk_t pblk,
2987 int len, int replay);
2988
2989/* mballoc.c */
2990extern const struct seq_operations ext4_mb_seq_groups_ops;
2991extern const struct seq_operations ext4_mb_seq_structs_summary_ops;
2992extern int ext4_seq_mb_stats_show(struct seq_file *seq, void *offset);
2993extern int ext4_mb_init(struct super_block *);
2994extern void ext4_mb_release(struct super_block *);
2995extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2996 struct ext4_allocation_request *, int *);
2997extern void ext4_discard_preallocations(struct inode *);
2998extern int __init ext4_init_mballoc(void);
2999extern void ext4_exit_mballoc(void);
3000extern ext4_group_t ext4_mb_prefetch(struct super_block *sb,
3001 ext4_group_t group,
3002 unsigned int nr, int *cnt);
3003extern void ext4_mb_prefetch_fini(struct super_block *sb, ext4_group_t group,
3004 unsigned int nr);
3005
3006extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
3007 struct buffer_head *bh, ext4_fsblk_t block,
3008 unsigned long count, int flags);
3009extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
3010 ext4_group_t ngroups);
3011extern int ext4_mb_add_groupinfo(struct super_block *sb,
3012 ext4_group_t i, struct ext4_group_desc *desc);
3013extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
3014 ext4_fsblk_t block, unsigned long count);
3015extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
3016extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid);
3017extern void ext4_mb_mark_bb(struct super_block *sb, ext4_fsblk_t block,
3018 int len, bool state);
3019static inline bool ext4_mb_cr_expensive(enum criteria cr)
3020{
3021 return cr >= CR_GOAL_LEN_SLOW;
3022}
3023
3024/* inode.c */
3025void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw,
3026 struct ext4_inode_info *ei);
3027int ext4_inode_is_fast_symlink(struct inode *inode);
3028void ext4_check_map_extents_env(struct inode *inode);
3029struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
3030struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
3031int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count,
3032 bool wait, struct buffer_head **bhs);
3033int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
3034 struct buffer_head *bh_result, int create);
3035int ext4_get_block(struct inode *inode, sector_t iblock,
3036 struct buffer_head *bh_result, int create);
3037int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
3038 struct buffer_head *bh, int create);
3039int ext4_walk_page_buffers(handle_t *handle,
3040 struct inode *inode,
3041 struct buffer_head *head,
3042 unsigned from,
3043 unsigned to,
3044 int *partial,
3045 int (*fn)(handle_t *handle, struct inode *inode,
3046 struct buffer_head *bh));
3047int do_journal_get_write_access(handle_t *handle, struct inode *inode,
3048 struct buffer_head *bh);
3049void ext4_set_inode_mapping_order(struct inode *inode);
3050#define FALL_BACK_TO_NONDELALLOC 1
3051#define CONVERT_INLINE_DATA 2
3052
3053typedef enum {
3054 EXT4_IGET_NORMAL = 0,
3055 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */
3056 EXT4_IGET_HANDLE = 0x0002, /* Inode # is from a handle */
3057 EXT4_IGET_BAD = 0x0004, /* Allow to iget a bad inode */
3058 EXT4_IGET_EA_INODE = 0x0008 /* Inode should contain an EA value */
3059} ext4_iget_flags;
3060
3061extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
3062 ext4_iget_flags flags, const char *function,
3063 unsigned int line);
3064
3065#define ext4_iget(sb, ino, flags) \
3066 __ext4_iget((sb), (ino), (flags), __func__, __LINE__)
3067
3068extern int ext4_write_inode(struct inode *, struct writeback_control *);
3069extern int ext4_setattr(struct mnt_idmap *, struct dentry *,
3070 struct iattr *);
3071extern u32 ext4_dio_alignment(struct inode *inode);
3072extern int ext4_getattr(struct mnt_idmap *, const struct path *,
3073 struct kstat *, u32, unsigned int);
3074extern void ext4_evict_inode(struct inode *);
3075extern void ext4_clear_inode(struct inode *);
3076extern int ext4_file_getattr(struct mnt_idmap *, const struct path *,
3077 struct kstat *, u32, unsigned int);
3078extern void ext4_dirty_inode(struct inode *, int);
3079extern int ext4_change_inode_journal_flag(struct inode *, int);
3080extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
3081extern int ext4_get_fc_inode_loc(struct super_block *sb, unsigned long ino,
3082 struct ext4_iloc *iloc);
3083extern int ext4_inode_attach_jinode(struct inode *inode);
3084extern int ext4_can_truncate(struct inode *inode);
3085extern int ext4_truncate(struct inode *);
3086extern int ext4_break_layouts(struct inode *);
3087extern int ext4_truncate_page_cache_block_range(struct inode *inode,
3088 loff_t start, loff_t end);
3089extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length);
3090extern void ext4_set_inode_flags(struct inode *, bool init);
3091extern int ext4_alloc_da_blocks(struct inode *inode);
3092extern void ext4_set_aops(struct inode *inode);
3093extern int ext4_normal_submit_inode_data_buffers(struct jbd2_inode *jinode);
3094extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
3095extern int ext4_chunk_trans_extent(struct inode *inode, int nrblocks);
3096extern int ext4_meta_trans_blocks(struct inode *inode, int lblocks,
3097 int pextents);
3098extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
3099 loff_t lstart, loff_t lend);
3100extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf);
3101extern qsize_t *ext4_get_reserved_space(struct inode *inode);
3102extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
3103extern void ext4_da_release_space(struct inode *inode, int to_free);
3104extern void ext4_da_update_reserve_space(struct inode *inode,
3105 int used, int quota_claim);
3106extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
3107 ext4_fsblk_t pblk, ext4_lblk_t len);
3108
3109static inline bool is_special_ino(struct super_block *sb, unsigned long ino)
3110{
3111 struct ext4_super_block *es = EXT4_SB(sb)->s_es;
3112
3113 return (ino < EXT4_FIRST_INO(sb) && ino != EXT4_ROOT_INO) ||
3114 ino == le32_to_cpu(es->s_usr_quota_inum) ||
3115 ino == le32_to_cpu(es->s_grp_quota_inum) ||
3116 ino == le32_to_cpu(es->s_prj_quota_inum) ||
3117 ino == le32_to_cpu(es->s_orphan_file_inum);
3118}
3119
3120/* indirect.c */
3121extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
3122 struct ext4_map_blocks *map, int flags);
3123extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
3124extern void ext4_ind_truncate(handle_t *, struct inode *inode);
3125extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
3126 ext4_lblk_t start, ext4_lblk_t end);
3127
3128/* ioctl.c */
3129extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
3130extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
3131int ext4_fileattr_set(struct mnt_idmap *idmap,
3132 struct dentry *dentry, struct file_kattr *fa);
3133int ext4_fileattr_get(struct dentry *dentry, struct file_kattr *fa);
3134extern void ext4_reset_inode_seed(struct inode *inode);
3135int ext4_update_overhead(struct super_block *sb, bool force);
3136int ext4_force_shutdown(struct super_block *sb, u32 flags);
3137
3138/* migrate.c */
3139extern int ext4_ext_migrate(struct inode *);
3140extern int ext4_ind_migrate(struct inode *inode);
3141
3142/* namei.c */
3143extern int ext4_init_new_dir(handle_t *handle, struct inode *dir,
3144 struct inode *inode);
3145extern int ext4_dirblock_csum_verify(struct inode *inode,
3146 struct buffer_head *bh);
3147extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
3148 __u32 start_minor_hash, __u32 *next_hash);
3149extern int ext4_search_dir(struct buffer_head *bh,
3150 char *search_buf,
3151 int buf_size,
3152 struct inode *dir,
3153 struct ext4_filename *fname,
3154 unsigned int offset,
3155 struct ext4_dir_entry_2 **res_dir);
3156extern int ext4_generic_delete_entry(struct inode *dir,
3157 struct ext4_dir_entry_2 *de_del,
3158 struct buffer_head *bh,
3159 void *entry_buf,
3160 int buf_size,
3161 int csum_size);
3162extern bool ext4_empty_dir(struct inode *inode);
3163
3164/* resize.c */
3165extern void ext4_kvfree_array_rcu(void *to_free);
3166extern int ext4_group_add(struct super_block *sb,
3167 struct ext4_new_group_data *input);
3168extern int ext4_group_extend(struct super_block *sb,
3169 struct ext4_super_block *es,
3170 ext4_fsblk_t n_blocks_count);
3171extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
3172extern unsigned int ext4_list_backups(struct super_block *sb,
3173 unsigned int *three, unsigned int *five,
3174 unsigned int *seven);
3175
3176/* super.c */
3177extern struct buffer_head *ext4_sb_bread(struct super_block *sb,
3178 sector_t block, blk_opf_t op_flags);
3179extern struct buffer_head *ext4_sb_bread_unmovable(struct super_block *sb,
3180 sector_t block);
3181extern struct buffer_head *ext4_sb_bread_nofail(struct super_block *sb,
3182 sector_t block);
3183extern void ext4_read_bh_nowait(struct buffer_head *bh, blk_opf_t op_flags,
3184 bh_end_io_t *end_io, bool simu_fail);
3185extern int ext4_read_bh(struct buffer_head *bh, blk_opf_t op_flags,
3186 bh_end_io_t *end_io, bool simu_fail);
3187extern int ext4_read_bh_lock(struct buffer_head *bh, blk_opf_t op_flags, bool wait);
3188extern void ext4_sb_breadahead_unmovable(struct super_block *sb, sector_t block);
3189extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
3190extern int ext4_calculate_overhead(struct super_block *sb);
3191extern __le32 ext4_superblock_csum(struct ext4_super_block *es);
3192extern void ext4_superblock_csum_set(struct super_block *sb);
3193extern int ext4_alloc_flex_bg_array(struct super_block *sb,
3194 ext4_group_t ngroup);
3195extern const char *ext4_decode_error(struct super_block *sb, int errno,
3196 char nbuf[16]);
3197extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb,
3198 ext4_group_t block_group,
3199 unsigned int flags);
3200extern unsigned int ext4_num_base_meta_blocks(struct super_block *sb,
3201 ext4_group_t block_group);
3202
3203extern __printf(7, 8)
3204void __ext4_error(struct super_block *, const char *, unsigned int, bool,
3205 int, __u64, const char *, ...);
3206extern __printf(6, 7)
3207void __ext4_error_inode(struct inode *, const char *, unsigned int,
3208 ext4_fsblk_t, int, const char *, ...);
3209extern __printf(5, 6)
3210void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
3211 const char *, ...);
3212extern void __ext4_std_error(struct super_block *, const char *,
3213 unsigned int, int);
3214extern __printf(4, 5)
3215void __ext4_warning(struct super_block *, const char *, unsigned int,
3216 const char *, ...);
3217extern __printf(4, 5)
3218void __ext4_warning_inode(const struct inode *inode, const char *function,
3219 unsigned int line, const char *fmt, ...);
3220extern __printf(3, 4)
3221void __ext4_msg(struct super_block *, const char *, const char *, ...);
3222extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
3223 const char *, unsigned int, const char *);
3224extern __printf(7, 8)
3225void __ext4_grp_locked_error(const char *, unsigned int,
3226 struct super_block *, ext4_group_t,
3227 unsigned long, ext4_fsblk_t,
3228 const char *, ...);
3229
3230#define EXT4_ERROR_INODE(inode, fmt, a...) \
3231 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
3232
3233#define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...) \
3234 __ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a)
3235
3236#define ext4_error_inode_block(inode, block, err, fmt, a...) \
3237 __ext4_error_inode((inode), __func__, __LINE__, (block), (err), \
3238 (fmt), ## a)
3239
3240#define EXT4_ERROR_FILE(file, block, fmt, a...) \
3241 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
3242
3243#define ext4_abort(sb, err, fmt, a...) \
3244 __ext4_error((sb), __func__, __LINE__, true, (err), 0, (fmt), ## a)
3245
3246#ifdef CONFIG_PRINTK
3247
3248#define ext4_error_inode(inode, func, line, block, fmt, ...) \
3249 __ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__)
3250#define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \
3251 __ext4_error_inode((inode), (func), (line), (block), \
3252 (err), (fmt), ##__VA_ARGS__)
3253#define ext4_error_file(file, func, line, block, fmt, ...) \
3254 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
3255#define ext4_error(sb, fmt, ...) \
3256 __ext4_error((sb), __func__, __LINE__, false, 0, 0, (fmt), \
3257 ##__VA_ARGS__)
3258#define ext4_error_err(sb, err, fmt, ...) \
3259 __ext4_error((sb), __func__, __LINE__, false, (err), 0, (fmt), \
3260 ##__VA_ARGS__)
3261#define ext4_warning(sb, fmt, ...) \
3262 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
3263#define ext4_warning_inode(inode, fmt, ...) \
3264 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
3265#define ext4_msg(sb, level, fmt, ...) \
3266 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
3267#define dump_mmp_msg(sb, mmp, msg) \
3268 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
3269#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
3270 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
3271 fmt, ##__VA_ARGS__)
3272
3273#else
3274
3275#define ext4_error_inode(inode, func, line, block, fmt, ...) \
3276do { \
3277 no_printk(fmt, ##__VA_ARGS__); \
3278 __ext4_error_inode(inode, "", 0, block, 0, " "); \
3279} while (0)
3280#define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \
3281do { \
3282 no_printk(fmt, ##__VA_ARGS__); \
3283 __ext4_error_inode(inode, "", 0, block, err, " "); \
3284} while (0)
3285#define ext4_error_file(file, func, line, block, fmt, ...) \
3286do { \
3287 no_printk(fmt, ##__VA_ARGS__); \
3288 __ext4_error_file(file, "", 0, block, " "); \
3289} while (0)
3290#define ext4_error(sb, fmt, ...) \
3291do { \
3292 no_printk(fmt, ##__VA_ARGS__); \
3293 __ext4_error(sb, "", 0, false, 0, 0, " "); \
3294} while (0)
3295#define ext4_error_err(sb, err, fmt, ...) \
3296do { \
3297 no_printk(fmt, ##__VA_ARGS__); \
3298 __ext4_error(sb, "", 0, false, err, 0, " "); \
3299} while (0)
3300#define ext4_warning(sb, fmt, ...) \
3301do { \
3302 no_printk(fmt, ##__VA_ARGS__); \
3303 __ext4_warning(sb, "", 0, " "); \
3304} while (0)
3305#define ext4_warning_inode(inode, fmt, ...) \
3306do { \
3307 no_printk(fmt, ##__VA_ARGS__); \
3308 __ext4_warning_inode(inode, "", 0, " "); \
3309} while (0)
3310#define ext4_msg(sb, level, fmt, ...) \
3311do { \
3312 no_printk(fmt, ##__VA_ARGS__); \
3313 __ext4_msg(sb, "", " "); \
3314} while (0)
3315#define dump_mmp_msg(sb, mmp, msg) \
3316 __dump_mmp_msg(sb, mmp, "", 0, "")
3317#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
3318do { \
3319 no_printk(fmt, ##__VA_ARGS__); \
3320 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
3321} while (0)
3322
3323#endif
3324
3325extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
3326 struct ext4_group_desc *bg);
3327extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
3328 struct ext4_group_desc *bg);
3329extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
3330 struct ext4_group_desc *bg);
3331extern __u32 ext4_free_group_clusters(struct super_block *sb,
3332 struct ext4_group_desc *bg);
3333extern __u32 ext4_free_inodes_count(struct super_block *sb,
3334 struct ext4_group_desc *bg);
3335extern __u32 ext4_used_dirs_count(struct super_block *sb,
3336 struct ext4_group_desc *bg);
3337extern __u32 ext4_itable_unused_count(struct super_block *sb,
3338 struct ext4_group_desc *bg);
3339extern void ext4_block_bitmap_set(struct super_block *sb,
3340 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3341extern void ext4_inode_bitmap_set(struct super_block *sb,
3342 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3343extern void ext4_inode_table_set(struct super_block *sb,
3344 struct ext4_group_desc *bg, ext4_fsblk_t blk);
3345extern void ext4_free_group_clusters_set(struct super_block *sb,
3346 struct ext4_group_desc *bg,
3347 __u32 count);
3348extern void ext4_free_inodes_set(struct super_block *sb,
3349 struct ext4_group_desc *bg, __u32 count);
3350extern void ext4_used_dirs_set(struct super_block *sb,
3351 struct ext4_group_desc *bg, __u32 count);
3352extern void ext4_itable_unused_set(struct super_block *sb,
3353 struct ext4_group_desc *bg, __u32 count);
3354extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
3355 struct ext4_group_desc *gdp);
3356extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
3357 struct ext4_group_desc *gdp);
3358extern int ext4_register_li_request(struct super_block *sb,
3359 ext4_group_t first_not_zeroed);
3360
3361static inline int ext4_has_group_desc_csum(struct super_block *sb)
3362{
3363 return ext4_has_feature_gdt_csum(sb) ||
3364 ext4_has_feature_metadata_csum(sb);
3365}
3366
3367#define ext4_read_incompat_64bit_val(es, name) \
3368 (((es)->s_feature_incompat & cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT) \
3369 ? (ext4_fsblk_t)le32_to_cpu(es->name##_hi) << 32 : 0) | \
3370 le32_to_cpu(es->name##_lo))
3371
3372static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
3373{
3374 return ext4_read_incompat_64bit_val(es, s_blocks_count);
3375}
3376
3377static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
3378{
3379 return ext4_read_incompat_64bit_val(es, s_r_blocks_count);
3380}
3381
3382static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
3383{
3384 return ext4_read_incompat_64bit_val(es, s_free_blocks_count);
3385}
3386
3387static inline void ext4_blocks_count_set(struct ext4_super_block *es,
3388 ext4_fsblk_t blk)
3389{
3390 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
3391 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
3392}
3393
3394static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
3395 ext4_fsblk_t blk)
3396{
3397 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
3398 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
3399}
3400
3401static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
3402 ext4_fsblk_t blk)
3403{
3404 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
3405 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
3406}
3407
3408static inline loff_t ext4_isize(struct super_block *sb,
3409 struct ext4_inode *raw_inode)
3410{
3411 if (ext4_has_feature_largedir(sb) ||
3412 S_ISREG(le16_to_cpu(raw_inode->i_mode)))
3413 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
3414 le32_to_cpu(raw_inode->i_size_lo);
3415
3416 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
3417}
3418
3419static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
3420{
3421 raw_inode->i_size_lo = cpu_to_le32(i_size);
3422 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
3423}
3424
3425/*
3426 * Reading s_groups_count requires using smp_rmb() afterwards. See
3427 * the locking protocol documented in the comments of ext4_group_add()
3428 * in resize.c
3429 */
3430static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
3431{
3432 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
3433
3434 smp_rmb();
3435 return ngroups;
3436}
3437
3438static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
3439 ext4_group_t block_group)
3440{
3441 return block_group >> sbi->s_log_groups_per_flex;
3442}
3443
3444static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
3445{
3446 return 1 << sbi->s_log_groups_per_flex;
3447}
3448
3449static inline loff_t ext4_get_maxbytes(struct inode *inode)
3450{
3451 if (ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS))
3452 return inode->i_sb->s_maxbytes;
3453 return EXT4_SB(inode->i_sb)->s_bitmap_maxbytes;
3454}
3455
3456#define ext4_std_error(sb, errno) \
3457do { \
3458 if ((errno)) \
3459 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
3460} while (0)
3461
3462#ifdef CONFIG_SMP
3463/* Each CPU can accumulate percpu_counter_batch clusters in their local
3464 * counters. So we need to make sure we have free clusters more
3465 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
3466 */
3467#define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
3468#else
3469#define EXT4_FREECLUSTERS_WATERMARK 0
3470#endif
3471
3472/* Update i_disksize. Requires i_rwsem to avoid races with truncate */
3473static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
3474{
3475 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
3476 !inode_is_locked(inode));
3477 down_write(&EXT4_I(inode)->i_data_sem);
3478 if (newsize > EXT4_I(inode)->i_disksize)
3479 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize);
3480 up_write(&EXT4_I(inode)->i_data_sem);
3481}
3482
3483/* Update i_size, i_disksize. Requires i_rwsem to avoid races with truncate */
3484static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
3485{
3486 int changed = 0;
3487
3488 if (newsize > inode->i_size) {
3489 i_size_write(inode, newsize);
3490 changed = 1;
3491 }
3492 if (newsize > EXT4_I(inode)->i_disksize) {
3493 ext4_update_i_disksize(inode, newsize);
3494 changed |= 2;
3495 }
3496 return changed;
3497}
3498
3499int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
3500 loff_t len);
3501
3502struct ext4_group_info {
3503 unsigned long bb_state;
3504#ifdef AGGRESSIVE_CHECK
3505 unsigned long bb_check_counter;
3506#endif
3507 struct rb_root bb_free_root;
3508 ext4_grpblk_t bb_first_free; /* first free block */
3509 ext4_grpblk_t bb_free; /* total free blocks */
3510 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
3511 int bb_avg_fragment_size_order; /* order of average
3512 fragment in BG */
3513 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
3514 ext4_group_t bb_group; /* Group number */
3515 struct list_head bb_prealloc_list;
3516#ifdef DOUBLE_CHECK
3517 void *bb_bitmap;
3518#endif
3519 struct rw_semaphore alloc_sem;
3520 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
3521 * regions, index is order.
3522 * bb_counters[3] = 5 means
3523 * 5 free 8-block regions. */
3524};
3525
3526#define EXT4_GROUP_INFO_NEED_INIT_BIT 0
3527#define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
3528#define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
3529#define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
3530#define EXT4_GROUP_INFO_BBITMAP_CORRUPT \
3531 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT)
3532#define EXT4_GROUP_INFO_IBITMAP_CORRUPT \
3533 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT)
3534#define EXT4_GROUP_INFO_BBITMAP_READ_BIT 4
3535
3536#define EXT4_MB_GRP_NEED_INIT(grp) \
3537 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
3538#define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
3539 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3540#define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
3541 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3542
3543#define EXT4_MB_GRP_WAS_TRIMMED(grp) \
3544 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3545#define EXT4_MB_GRP_SET_TRIMMED(grp) \
3546 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3547#define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
3548 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3549#define EXT4_MB_GRP_TEST_AND_SET_READ(grp) \
3550 (test_and_set_bit(EXT4_GROUP_INFO_BBITMAP_READ_BIT, &((grp)->bb_state)))
3551
3552#define EXT4_MAX_CONTENTION 8
3553#define EXT4_CONTENTION_THRESHOLD 2
3554
3555static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
3556 ext4_group_t group)
3557{
3558 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
3559}
3560
3561/*
3562 * Returns true if the filesystem is busy enough that attempts to
3563 * access the block group locks has run into contention.
3564 */
3565static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
3566{
3567 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
3568}
3569
3570static inline bool ext4_try_lock_group(struct super_block *sb, ext4_group_t group)
3571{
3572 if (!spin_trylock(ext4_group_lock_ptr(sb, group)))
3573 return false;
3574 /*
3575 * We're able to grab the lock right away, so drop the lock
3576 * contention counter.
3577 */
3578 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
3579 return true;
3580}
3581
3582static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
3583{
3584 if (!ext4_try_lock_group(sb, group)) {
3585 /*
3586 * The lock is busy, so bump the contention counter,
3587 * and then wait on the spin lock.
3588 */
3589 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
3590 EXT4_MAX_CONTENTION);
3591 spin_lock(ext4_group_lock_ptr(sb, group));
3592 }
3593}
3594
3595static inline void ext4_unlock_group(struct super_block *sb,
3596 ext4_group_t group)
3597{
3598 spin_unlock(ext4_group_lock_ptr(sb, group));
3599}
3600
3601#ifdef CONFIG_QUOTA
3602static inline bool ext4_quota_capable(struct super_block *sb)
3603{
3604 return (test_opt(sb, QUOTA) || ext4_has_feature_quota(sb));
3605}
3606
3607static inline bool ext4_is_quota_journalled(struct super_block *sb)
3608{
3609 struct ext4_sb_info *sbi = EXT4_SB(sb);
3610
3611 return (ext4_has_feature_quota(sb) ||
3612 sbi->s_qf_names[USRQUOTA] || sbi->s_qf_names[GRPQUOTA]);
3613}
3614int ext4_enable_quotas(struct super_block *sb);
3615#endif
3616
3617/*
3618 * Block validity checking
3619 */
3620#define ext4_check_indirect_blockref(inode, bh) \
3621 ext4_check_blockref(__func__, __LINE__, inode, \
3622 (__le32 *)(bh)->b_data, \
3623 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
3624
3625#define ext4_ind_check_inode(inode) \
3626 ext4_check_blockref(__func__, __LINE__, inode, \
3627 EXT4_I(inode)->i_data, \
3628 EXT4_NDIR_BLOCKS)
3629
3630/*
3631 * Inodes and files operations
3632 */
3633
3634/* dir.c */
3635extern const struct file_operations ext4_dir_operations;
3636
3637/* file.c */
3638extern const struct inode_operations ext4_file_inode_operations;
3639extern const struct file_operations ext4_file_operations;
3640extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
3641
3642/* inline.c */
3643extern int ext4_get_max_inline_size(struct inode *inode);
3644extern int ext4_find_inline_data_nolock(struct inode *inode);
3645extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
3646extern void ext4_update_final_de(void *de_buf, int old_size, int new_size);
3647
3648int ext4_readpage_inline(struct inode *inode, struct folio *folio);
3649extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3650 struct inode *inode,
3651 loff_t pos, unsigned len,
3652 struct folio **foliop);
3653int ext4_write_inline_data_end(struct inode *inode, loff_t pos, unsigned len,
3654 unsigned copied, struct folio *folio);
3655extern int ext4_generic_write_inline_data(struct address_space *mapping,
3656 struct inode *inode,
3657 loff_t pos, unsigned len,
3658 struct folio **foliop,
3659 void **fsdata, bool da);
3660extern int ext4_try_add_inline_entry(handle_t *handle,
3661 struct ext4_filename *fname,
3662 struct inode *dir, struct inode *inode);
3663extern int ext4_try_create_inline_dir(handle_t *handle,
3664 struct inode *parent,
3665 struct inode *inode);
3666extern int ext4_read_inline_dir(struct file *filp,
3667 struct dir_context *ctx,
3668 int *has_inline_data);
3669extern int ext4_inlinedir_to_tree(struct file *dir_file,
3670 struct inode *dir, ext4_lblk_t block,
3671 struct dx_hash_info *hinfo,
3672 __u32 start_hash, __u32 start_minor_hash,
3673 int *has_inline_data);
3674extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3675 struct ext4_filename *fname,
3676 struct ext4_dir_entry_2 **res_dir,
3677 int *has_inline_data);
3678extern int ext4_delete_inline_entry(handle_t *handle,
3679 struct inode *dir,
3680 struct ext4_dir_entry_2 *de_del,
3681 struct buffer_head *bh,
3682 int *has_inline_data);
3683extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3684extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3685 struct ext4_dir_entry_2 **parent_de,
3686 int *retval);
3687extern void *ext4_read_inline_link(struct inode *inode);
3688
3689struct iomap;
3690extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap);
3691
3692extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3693
3694extern int ext4_convert_inline_data(struct inode *inode);
3695
3696static inline int ext4_has_inline_data(struct inode *inode)
3697{
3698 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3699 EXT4_I(inode)->i_inline_off;
3700}
3701
3702/* namei.c */
3703extern const struct inode_operations ext4_dir_inode_operations;
3704extern const struct inode_operations ext4_special_inode_operations;
3705extern struct dentry *ext4_get_parent(struct dentry *child);
3706extern int ext4_init_dirblock(handle_t *handle, struct inode *inode,
3707 struct buffer_head *dir_block,
3708 unsigned int parent_ino, void *inline_buf,
3709 int inline_size);
3710extern void ext4_initialize_dirent_tail(struct buffer_head *bh,
3711 unsigned int blocksize);
3712extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode,
3713 struct buffer_head *bh);
3714extern int __ext4_unlink(struct inode *dir, const struct qstr *d_name,
3715 struct inode *inode, struct dentry *dentry);
3716extern int __ext4_link(struct inode *dir, struct inode *inode,
3717 struct dentry *dentry);
3718
3719#define S_SHIFT 12
3720static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = {
3721 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
3722 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
3723 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
3724 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
3725 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
3726 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
3727 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
3728};
3729
3730static inline void ext4_set_de_type(struct super_block *sb,
3731 struct ext4_dir_entry_2 *de,
3732 umode_t mode) {
3733 if (ext4_has_feature_filetype(sb))
3734 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3735}
3736
3737/* readpages.c */
3738extern int ext4_mpage_readpages(struct inode *inode,
3739 struct readahead_control *rac, struct folio *folio);
3740extern int __init ext4_init_post_read_processing(void);
3741extern void ext4_exit_post_read_processing(void);
3742
3743/* symlink.c */
3744extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3745extern const struct inode_operations ext4_symlink_inode_operations;
3746extern const struct inode_operations ext4_fast_symlink_inode_operations;
3747
3748/* sysfs.c */
3749extern void ext4_notify_error_sysfs(struct ext4_sb_info *sbi);
3750extern int ext4_register_sysfs(struct super_block *sb);
3751extern void ext4_unregister_sysfs(struct super_block *sb);
3752extern int __init ext4_init_sysfs(void);
3753extern void ext4_exit_sysfs(void);
3754
3755/* block_validity */
3756extern void ext4_release_system_zone(struct super_block *sb);
3757extern int ext4_setup_system_zone(struct super_block *sb);
3758extern int __init ext4_init_system_zone(void);
3759extern void ext4_exit_system_zone(void);
3760extern int ext4_inode_block_valid(struct inode *inode,
3761 ext4_fsblk_t start_blk,
3762 unsigned int count);
3763extern int ext4_check_blockref(const char *, unsigned int,
3764 struct inode *, __le32 *, unsigned int);
3765extern int ext4_sb_block_valid(struct super_block *sb, struct inode *inode,
3766 ext4_fsblk_t start_blk, unsigned int count);
3767
3768
3769/* extents.c */
3770struct ext4_ext_path;
3771struct ext4_extent;
3772
3773/*
3774 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3775 * __le32.
3776 */
3777#define EXT_MAX_BLOCKS 0xffffffff
3778
3779extern void ext4_ext_tree_init(handle_t *handle, struct inode *inode);
3780extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3781extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3782 struct ext4_map_blocks *map, int flags);
3783extern int ext4_ext_truncate(handle_t *, struct inode *);
3784extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3785 ext4_lblk_t end);
3786extern void ext4_ext_init(struct super_block *);
3787extern void ext4_ext_release(struct super_block *);
3788extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3789 loff_t len);
3790extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3791 loff_t offset, ssize_t len);
3792extern int ext4_convert_unwritten_extents_atomic(handle_t *handle,
3793 struct inode *inode, loff_t offset, ssize_t len);
3794extern int ext4_convert_unwritten_io_end_vec(handle_t *handle,
3795 ext4_io_end_t *io_end);
3796extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3797 struct ext4_map_blocks *map, int flags);
3798extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3799 int num,
3800 struct ext4_ext_path *path);
3801extern struct ext4_ext_path *ext4_ext_insert_extent(
3802 handle_t *handle, struct inode *inode,
3803 struct ext4_ext_path *path,
3804 struct ext4_extent *newext, int gb_flags);
3805extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3806 struct ext4_ext_path *,
3807 int flags);
3808extern void ext4_free_ext_path(struct ext4_ext_path *);
3809extern int ext4_ext_check_inode(struct inode *inode);
3810extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3811extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3812 __u64 start, __u64 len);
3813extern int ext4_get_es_cache(struct inode *inode,
3814 struct fiemap_extent_info *fieinfo,
3815 __u64 start, __u64 len);
3816extern int ext4_ext_precache(struct inode *inode);
3817extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3818 struct inode *inode2, ext4_lblk_t lblk1,
3819 ext4_lblk_t lblk2, ext4_lblk_t count,
3820 int mark_unwritten,int *err);
3821extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu);
3822extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode,
3823 int check_cred, int restart_cred,
3824 int revoke_cred);
3825extern void ext4_ext_replay_shrink_inode(struct inode *inode, ext4_lblk_t end);
3826extern int ext4_ext_replay_set_iblocks(struct inode *inode);
3827extern int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start,
3828 int len, int unwritten, ext4_fsblk_t pblk);
3829extern int ext4_ext_clear_bb(struct inode *inode);
3830
3831
3832/* move_extent.c */
3833extern void ext4_double_down_write_data_sem(struct inode *first,
3834 struct inode *second);
3835extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3836 struct inode *donor_inode);
3837extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3838 __u64 start_orig, __u64 start_donor,
3839 __u64 len, __u64 *moved_len);
3840
3841/* page-io.c */
3842extern int __init ext4_init_pageio(void);
3843extern void ext4_exit_pageio(void);
3844extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3845extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3846extern int ext4_put_io_end(ext4_io_end_t *io_end);
3847extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3848extern void ext4_io_submit_init(struct ext4_io_submit *io,
3849 struct writeback_control *wbc);
3850extern void ext4_end_io_rsv_work(struct work_struct *work);
3851extern void ext4_io_submit(struct ext4_io_submit *io);
3852int ext4_bio_write_folio(struct ext4_io_submit *io, struct folio *page,
3853 size_t len);
3854extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end);
3855extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end);
3856
3857/* mmp.c */
3858extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3859
3860/* mmp.c */
3861extern void ext4_stop_mmpd(struct ext4_sb_info *sbi);
3862
3863/* verity.c */
3864extern const struct fsverity_operations ext4_verityops;
3865
3866/* orphan.c */
3867extern int ext4_orphan_add(handle_t *, struct inode *);
3868extern int ext4_orphan_del(handle_t *, struct inode *);
3869extern void ext4_orphan_cleanup(struct super_block *sb,
3870 struct ext4_super_block *es);
3871extern void ext4_release_orphan_info(struct super_block *sb);
3872extern int ext4_init_orphan_info(struct super_block *sb);
3873extern int ext4_orphan_file_empty(struct super_block *sb);
3874extern void ext4_orphan_file_block_trigger(
3875 struct jbd2_buffer_trigger_type *triggers,
3876 struct buffer_head *bh,
3877 void *data, size_t size);
3878
3879/*
3880 * Add new method to test whether block and inode bitmaps are properly
3881 * initialized. With uninit_bg reading the block from disk is not enough
3882 * to mark the bitmap uptodate. We need to also zero-out the bitmap
3883 */
3884#define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3885
3886static inline int bitmap_uptodate(struct buffer_head *bh)
3887{
3888 return (buffer_uptodate(bh) &&
3889 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3890}
3891static inline void set_bitmap_uptodate(struct buffer_head *bh)
3892{
3893 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3894}
3895
3896extern int ext4_resize_begin(struct super_block *sb);
3897extern int ext4_resize_end(struct super_block *sb, bool update_backups);
3898
3899static inline void ext4_set_io_unwritten_flag(struct ext4_io_end *io_end)
3900{
3901 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN))
3902 io_end->flag |= EXT4_IO_END_UNWRITTEN;
3903}
3904
3905static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3906{
3907 if (io_end->flag & EXT4_IO_END_UNWRITTEN)
3908 io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3909}
3910
3911extern const struct iomap_ops ext4_iomap_ops;
3912extern const struct iomap_ops ext4_iomap_overwrite_ops;
3913extern const struct iomap_ops ext4_iomap_report_ops;
3914
3915static inline int ext4_buffer_uptodate(struct buffer_head *bh)
3916{
3917 /*
3918 * If the buffer has the write error flag, we have failed
3919 * to write out data in the block. In this case, we don't
3920 * have to read the block because we may read the old data
3921 * successfully.
3922 */
3923 if (buffer_write_io_error(bh))
3924 set_buffer_uptodate(bh);
3925 return buffer_uptodate(bh);
3926}
3927
3928static inline bool ext4_inode_can_atomic_write(struct inode *inode)
3929{
3930
3931 return S_ISREG(inode->i_mode) &&
3932 ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS) &&
3933 EXT4_SB(inode->i_sb)->s_awu_min > 0;
3934}
3935
3936extern int ext4_block_write_begin(handle_t *handle, struct folio *folio,
3937 loff_t pos, unsigned len,
3938 get_block_t *get_block);
3939#endif /* __KERNEL__ */
3940
3941#define EFSBADCRC EBADMSG /* Bad CRC detected */
3942#define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
3943
3944#endif /* _EXT4_H */