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/types.h>
21#include <linux/blkdev.h>
22#include <linux/magic.h>
23#include <linux/jbd2.h>
24#include <linux/quota.h>
25#include <linux/rwsem.h>
26#include <linux/rbtree.h>
27#include <linux/seqlock.h>
28#include <linux/mutex.h>
29#include <linux/timer.h>
30#include <linux/version.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 <crypto/hash.h>
37#include <linux/falloc.h>
38#include <linux/percpu-rwsem.h>
39#ifdef __KERNEL__
40#include <linux/compat.h>
41#endif
42
43#include <linux/fscrypt.h>
44#include <linux/fsverity.h>
45
46#include <linux/compiler.h>
47
48/*
49 * The fourth extended filesystem constants/structures
50 */
51
52/*
53 * with AGGRESSIVE_CHECK allocator runs consistency checks over
54 * structures. these checks slow things down a lot
55 */
56#define AGGRESSIVE_CHECK__
57
58/*
59 * with DOUBLE_CHECK defined mballoc creates persistent in-core
60 * bitmaps, maintains and uses them to check for double allocations
61 */
62#define DOUBLE_CHECK__
63
64/*
65 * Define EXT4FS_DEBUG to produce debug messages
66 */
67#undef EXT4FS_DEBUG
68
69/*
70 * Debug code
71 */
72#ifdef EXT4FS_DEBUG
73#define ext4_debug(f, a...) \
74 do { \
75 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
76 __FILE__, __LINE__, __func__); \
77 printk(KERN_DEBUG f, ## a); \
78 } while (0)
79#else
80#define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
81#endif
82
83/*
84 * Turn on EXT_DEBUG to get lots of info about extents operations.
85 */
86#define EXT_DEBUG__
87#ifdef EXT_DEBUG
88#define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__)
89#else
90#define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
91#endif
92
93/* data type for block offset of block group */
94typedef int ext4_grpblk_t;
95
96/* data type for filesystem-wide blocks number */
97typedef unsigned long long ext4_fsblk_t;
98
99/* data type for file logical block number */
100typedef __u32 ext4_lblk_t;
101
102/* data type for block group number */
103typedef unsigned int ext4_group_t;
104
105enum SHIFT_DIRECTION {
106 SHIFT_LEFT = 0,
107 SHIFT_RIGHT,
108};
109
110/*
111 * Flags used in mballoc's allocation_context flags field.
112 *
113 * Also used to show what's going on for debugging purposes when the
114 * flag field is exported via the traceport interface
115 */
116
117/* prefer goal again. length */
118#define EXT4_MB_HINT_MERGE 0x0001
119/* blocks already reserved */
120#define EXT4_MB_HINT_RESERVED 0x0002
121/* metadata is being allocated */
122#define EXT4_MB_HINT_METADATA 0x0004
123/* first blocks in the file */
124#define EXT4_MB_HINT_FIRST 0x0008
125/* search for the best chunk */
126#define EXT4_MB_HINT_BEST 0x0010
127/* data is being allocated */
128#define EXT4_MB_HINT_DATA 0x0020
129/* don't preallocate (for tails) */
130#define EXT4_MB_HINT_NOPREALLOC 0x0040
131/* allocate for locality group */
132#define EXT4_MB_HINT_GROUP_ALLOC 0x0080
133/* allocate goal blocks or none */
134#define EXT4_MB_HINT_GOAL_ONLY 0x0100
135/* goal is meaningful */
136#define EXT4_MB_HINT_TRY_GOAL 0x0200
137/* blocks already pre-reserved by delayed allocation */
138#define EXT4_MB_DELALLOC_RESERVED 0x0400
139/* We are doing stream allocation */
140#define EXT4_MB_STREAM_ALLOC 0x0800
141/* Use reserved root blocks if needed */
142#define EXT4_MB_USE_ROOT_BLOCKS 0x1000
143/* Use blocks from reserved pool */
144#define EXT4_MB_USE_RESERVED 0x2000
145
146struct ext4_allocation_request {
147 /* target inode for block we're allocating */
148 struct inode *inode;
149 /* how many blocks we want to allocate */
150 unsigned int len;
151 /* logical block in target inode */
152 ext4_lblk_t logical;
153 /* the closest logical allocated block to the left */
154 ext4_lblk_t lleft;
155 /* the closest logical allocated block to the right */
156 ext4_lblk_t lright;
157 /* phys. target (a hint) */
158 ext4_fsblk_t goal;
159 /* phys. block for the closest logical allocated block to the left */
160 ext4_fsblk_t pleft;
161 /* phys. block for the closest logical allocated block to the right */
162 ext4_fsblk_t pright;
163 /* flags. see above EXT4_MB_HINT_* */
164 unsigned int flags;
165};
166
167/*
168 * Logical to physical block mapping, used by ext4_map_blocks()
169 *
170 * This structure is used to pass requests into ext4_map_blocks() as
171 * well as to store the information returned by ext4_map_blocks(). It
172 * takes less room on the stack than a struct buffer_head.
173 */
174#define EXT4_MAP_NEW (1 << BH_New)
175#define EXT4_MAP_MAPPED (1 << BH_Mapped)
176#define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten)
177#define EXT4_MAP_BOUNDARY (1 << BH_Boundary)
178#define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
179 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
180
181struct ext4_map_blocks {
182 ext4_fsblk_t m_pblk;
183 ext4_lblk_t m_lblk;
184 unsigned int m_len;
185 unsigned int m_flags;
186};
187
188/*
189 * Block validity checking, system zone rbtree.
190 */
191struct ext4_system_blocks {
192 struct rb_root root;
193 struct rcu_head rcu;
194};
195
196/*
197 * Flags for ext4_io_end->flags
198 */
199#define EXT4_IO_END_UNWRITTEN 0x0001
200
201struct ext4_io_end_vec {
202 struct list_head list; /* list of io_end_vec */
203 loff_t offset; /* offset in the file */
204 ssize_t size; /* size of the extent */
205};
206
207/*
208 * For converting unwritten extents on a work queue. 'handle' is used for
209 * buffered writeback.
210 */
211typedef struct ext4_io_end {
212 struct list_head list; /* per-file finished IO list */
213 handle_t *handle; /* handle reserved for extent
214 * conversion */
215 struct inode *inode; /* file being written to */
216 struct bio *bio; /* Linked list of completed
217 * bios covering the extent */
218 unsigned int flag; /* unwritten or not */
219 atomic_t count; /* reference counter */
220 struct list_head list_vec; /* list of ext4_io_end_vec */
221} ext4_io_end_t;
222
223struct ext4_io_submit {
224 struct writeback_control *io_wbc;
225 struct bio *io_bio;
226 ext4_io_end_t *io_end;
227 sector_t io_next_block;
228};
229
230/*
231 * Special inodes numbers
232 */
233#define EXT4_BAD_INO 1 /* Bad blocks inode */
234#define EXT4_ROOT_INO 2 /* Root inode */
235#define EXT4_USR_QUOTA_INO 3 /* User quota inode */
236#define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
237#define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
238#define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
239#define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
240#define EXT4_JOURNAL_INO 8 /* Journal inode */
241
242/* First non-reserved inode for old ext4 filesystems */
243#define EXT4_GOOD_OLD_FIRST_INO 11
244
245/*
246 * Maximal count of links to a file
247 */
248#define EXT4_LINK_MAX 65000
249
250/*
251 * Macro-instructions used to manage several block sizes
252 */
253#define EXT4_MIN_BLOCK_SIZE 1024
254#define EXT4_MAX_BLOCK_SIZE 65536
255#define EXT4_MIN_BLOCK_LOG_SIZE 10
256#define EXT4_MAX_BLOCK_LOG_SIZE 16
257#define EXT4_MAX_CLUSTER_LOG_SIZE 30
258#ifdef __KERNEL__
259# define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
260#else
261# define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
262#endif
263#define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
264#define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
265 EXT4_SB(s)->s_cluster_bits)
266#ifdef __KERNEL__
267# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
268# define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
269#else
270# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
271#endif
272#ifdef __KERNEL__
273#define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
274#define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
275#define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
276#else
277#define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
278 EXT4_GOOD_OLD_INODE_SIZE : \
279 (s)->s_inode_size)
280#define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
281 EXT4_GOOD_OLD_FIRST_INO : \
282 (s)->s_first_ino)
283#endif
284#define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
285#define EXT4_MAX_BLOCKS(size, offset, blkbits) \
286 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \
287 blkbits))
288
289/* Translate a block number to a cluster number */
290#define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
291/* Translate a cluster number to a block number */
292#define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
293/* Translate # of blks to # of clusters */
294#define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
295 (sbi)->s_cluster_bits)
296/* Mask out the low bits to get the starting block of the cluster */
297#define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
298 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
299#define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
300 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
301/* Fill in the low bits to get the last block of the cluster */
302#define EXT4_LBLK_CFILL(sbi, lblk) ((lblk) | \
303 ((ext4_lblk_t) (sbi)->s_cluster_ratio - 1))
304/* Get the cluster offset */
305#define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
306 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
307#define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
308 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
309
310/*
311 * Structure of a blocks group descriptor
312 */
313struct ext4_group_desc
314{
315 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
316 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
317 __le32 bg_inode_table_lo; /* Inodes table block */
318 __le16 bg_free_blocks_count_lo;/* Free blocks count */
319 __le16 bg_free_inodes_count_lo;/* Free inodes count */
320 __le16 bg_used_dirs_count_lo; /* Directories count */
321 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
322 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
323 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
324 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
325 __le16 bg_itable_unused_lo; /* Unused inodes count */
326 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
327 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
328 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
329 __le32 bg_inode_table_hi; /* Inodes table block MSB */
330 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
331 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
332 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
333 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
334 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
335 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
336 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
337 __u32 bg_reserved;
338};
339
340#define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
341 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
342 sizeof(__le16))
343#define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
344 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
345 sizeof(__le16))
346
347/*
348 * Structure of a flex block group info
349 */
350
351struct flex_groups {
352 atomic64_t free_clusters;
353 atomic_t free_inodes;
354 atomic_t used_dirs;
355};
356
357#define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
358#define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
359#define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
360
361/*
362 * Macro-instructions used to manage group descriptors
363 */
364#define EXT4_MIN_DESC_SIZE 32
365#define EXT4_MIN_DESC_SIZE_64BIT 64
366#define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
367#define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
368#ifdef __KERNEL__
369# define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
370# define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
371# define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
372# define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
373# define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
374#else
375# define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
376# define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
377# define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
378#endif
379
380/*
381 * Constants relative to the data blocks
382 */
383#define EXT4_NDIR_BLOCKS 12
384#define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
385#define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
386#define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
387#define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
388
389/*
390 * Inode flags
391 */
392#define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
393#define EXT4_UNRM_FL 0x00000002 /* Undelete */
394#define EXT4_COMPR_FL 0x00000004 /* Compress file */
395#define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
396#define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
397#define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
398#define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
399#define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
400/* Reserved for compression usage... */
401#define EXT4_DIRTY_FL 0x00000100
402#define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
403#define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
404 /* nb: was previously EXT2_ECOMPR_FL */
405#define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */
406/* End compression flags --- maybe not all used */
407#define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
408#define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
409#define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
410#define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
411#define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
412#define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
413#define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
414#define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
415#define EXT4_VERITY_FL 0x00100000 /* Verity protected inode */
416#define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
417/* 0x00400000 was formerly EXT4_EOFBLOCKS_FL */
418#define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
419#define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
420#define EXT4_CASEFOLD_FL 0x40000000 /* Casefolded file */
421#define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
422
423#define EXT4_FL_USER_VISIBLE 0x705BDFFF /* User visible flags */
424#define EXT4_FL_USER_MODIFIABLE 0x604BC0FF /* User modifiable flags */
425
426/* Flags we can manipulate with through EXT4_IOC_FSSETXATTR */
427#define EXT4_FL_XFLAG_VISIBLE (EXT4_SYNC_FL | \
428 EXT4_IMMUTABLE_FL | \
429 EXT4_APPEND_FL | \
430 EXT4_NODUMP_FL | \
431 EXT4_NOATIME_FL | \
432 EXT4_PROJINHERIT_FL)
433
434/* Flags that should be inherited by new inodes from their parent. */
435#define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
436 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
437 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
438 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
439 EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL)
440
441/* Flags that are appropriate for regular files (all but dir-specific ones). */
442#define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL |\
443 EXT4_PROJINHERIT_FL))
444
445/* Flags that are appropriate for non-directories/regular files. */
446#define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
447
448/* The only flags that should be swapped */
449#define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL)
450
451/* Mask out flags that are inappropriate for the given type of inode. */
452static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
453{
454 if (S_ISDIR(mode))
455 return flags;
456 else if (S_ISREG(mode))
457 return flags & EXT4_REG_FLMASK;
458 else
459 return flags & EXT4_OTHER_FLMASK;
460}
461
462/*
463 * Inode flags used for atomic set/get
464 */
465enum {
466 EXT4_INODE_SECRM = 0, /* Secure deletion */
467 EXT4_INODE_UNRM = 1, /* Undelete */
468 EXT4_INODE_COMPR = 2, /* Compress file */
469 EXT4_INODE_SYNC = 3, /* Synchronous updates */
470 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
471 EXT4_INODE_APPEND = 5, /* writes to file may only append */
472 EXT4_INODE_NODUMP = 6, /* do not dump file */
473 EXT4_INODE_NOATIME = 7, /* do not update atime */
474/* Reserved for compression usage... */
475 EXT4_INODE_DIRTY = 8,
476 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
477 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
478 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */
479/* End compression flags --- maybe not all used */
480 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
481 EXT4_INODE_IMAGIC = 13, /* AFS directory */
482 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
483 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
484 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
485 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
486 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
487 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
488 EXT4_INODE_VERITY = 20, /* Verity protected inode */
489 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
490/* 22 was formerly EXT4_INODE_EOFBLOCKS */
491 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
492 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */
493 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
494};
495
496/*
497 * Since it's pretty easy to mix up bit numbers and hex values, we use a
498 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
499 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
500 * any extra space in the compiled kernel image, otherwise, the build will fail.
501 * It's important that these values are the same, since we are using
502 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
503 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
504 * values found in ext2, ext3 and ext4 filesystems, and of course the values
505 * defined in e2fsprogs.
506 *
507 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
508 */
509#define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
510#define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
511
512static inline void ext4_check_flag_values(void)
513{
514 CHECK_FLAG_VALUE(SECRM);
515 CHECK_FLAG_VALUE(UNRM);
516 CHECK_FLAG_VALUE(COMPR);
517 CHECK_FLAG_VALUE(SYNC);
518 CHECK_FLAG_VALUE(IMMUTABLE);
519 CHECK_FLAG_VALUE(APPEND);
520 CHECK_FLAG_VALUE(NODUMP);
521 CHECK_FLAG_VALUE(NOATIME);
522 CHECK_FLAG_VALUE(DIRTY);
523 CHECK_FLAG_VALUE(COMPRBLK);
524 CHECK_FLAG_VALUE(NOCOMPR);
525 CHECK_FLAG_VALUE(ENCRYPT);
526 CHECK_FLAG_VALUE(INDEX);
527 CHECK_FLAG_VALUE(IMAGIC);
528 CHECK_FLAG_VALUE(JOURNAL_DATA);
529 CHECK_FLAG_VALUE(NOTAIL);
530 CHECK_FLAG_VALUE(DIRSYNC);
531 CHECK_FLAG_VALUE(TOPDIR);
532 CHECK_FLAG_VALUE(HUGE_FILE);
533 CHECK_FLAG_VALUE(EXTENTS);
534 CHECK_FLAG_VALUE(VERITY);
535 CHECK_FLAG_VALUE(EA_INODE);
536 CHECK_FLAG_VALUE(INLINE_DATA);
537 CHECK_FLAG_VALUE(PROJINHERIT);
538 CHECK_FLAG_VALUE(RESERVED);
539}
540
541/* Used to pass group descriptor data when online resize is done */
542struct ext4_new_group_input {
543 __u32 group; /* Group number for this data */
544 __u64 block_bitmap; /* Absolute block number of block bitmap */
545 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
546 __u64 inode_table; /* Absolute block number of inode table start */
547 __u32 blocks_count; /* Total number of blocks in this group */
548 __u16 reserved_blocks; /* Number of reserved blocks in this group */
549 __u16 unused;
550};
551
552#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
553struct compat_ext4_new_group_input {
554 u32 group;
555 compat_u64 block_bitmap;
556 compat_u64 inode_bitmap;
557 compat_u64 inode_table;
558 u32 blocks_count;
559 u16 reserved_blocks;
560 u16 unused;
561};
562#endif
563
564/* The struct ext4_new_group_input in kernel space, with free_blocks_count */
565struct ext4_new_group_data {
566 __u32 group;
567 __u64 block_bitmap;
568 __u64 inode_bitmap;
569 __u64 inode_table;
570 __u32 blocks_count;
571 __u16 reserved_blocks;
572 __u16 mdata_blocks;
573 __u32 free_clusters_count;
574};
575
576/* Indexes used to index group tables in ext4_new_group_data */
577enum {
578 BLOCK_BITMAP = 0, /* block bitmap */
579 INODE_BITMAP, /* inode bitmap */
580 INODE_TABLE, /* inode tables */
581 GROUP_TABLE_COUNT,
582};
583
584/*
585 * Flags used by ext4_map_blocks()
586 */
587 /* Allocate any needed blocks and/or convert an unwritten
588 extent to be an initialized ext4 */
589#define EXT4_GET_BLOCKS_CREATE 0x0001
590 /* Request the creation of an unwritten extent */
591#define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002
592#define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\
593 EXT4_GET_BLOCKS_CREATE)
594 /* Caller is from the delayed allocation writeout path
595 * finally doing the actual allocation of delayed blocks */
596#define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
597 /* caller is from the direct IO path, request to creation of an
598 unwritten extents if not allocated, split the unwritten
599 extent if blocks has been preallocated already*/
600#define EXT4_GET_BLOCKS_PRE_IO 0x0008
601#define EXT4_GET_BLOCKS_CONVERT 0x0010
602#define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
603 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
604 /* Convert extent to initialized after IO complete */
605#define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
606 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
607 /* Eventual metadata allocation (due to growing extent tree)
608 * should not fail, so try to use reserved blocks for that.*/
609#define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
610 /* Don't normalize allocation size (used for fallocate) */
611#define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
612 /* Request will not result in inode size update (user for fallocate) */
613#define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080
614 /* Convert written extents to unwritten */
615#define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100
616 /* Write zeros to newly created written extents */
617#define EXT4_GET_BLOCKS_ZERO 0x0200
618#define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\
619 EXT4_GET_BLOCKS_ZERO)
620 /* Caller will submit data before dropping transaction handle. This
621 * allows jbd2 to avoid submitting data before commit. */
622#define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400
623
624/*
625 * The bit position of these flags must not overlap with any of the
626 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(),
627 * read_extent_tree_block(), ext4_split_extent_at(),
628 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
629 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
630 * caching the extents when reading from the extent tree while a
631 * truncate or punch hole operation is in progress.
632 */
633#define EXT4_EX_NOCACHE 0x40000000
634#define EXT4_EX_FORCE_CACHE 0x20000000
635
636/*
637 * Flags used by ext4_free_blocks
638 */
639#define EXT4_FREE_BLOCKS_METADATA 0x0001
640#define EXT4_FREE_BLOCKS_FORGET 0x0002
641#define EXT4_FREE_BLOCKS_VALIDATED 0x0004
642#define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
643#define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
644#define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
645#define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040
646
647/*
648 * ioctl commands
649 */
650#define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
651#define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
652#define EXT4_IOC_GETVERSION _IOR('f', 3, long)
653#define EXT4_IOC_SETVERSION _IOW('f', 4, long)
654#define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
655#define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
656#define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
657#define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
658#define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
659#define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
660#define EXT4_IOC_MIGRATE _IO('f', 9)
661 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
662 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
663#define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
664#define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
665#define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
666#define EXT4_IOC_SWAP_BOOT _IO('f', 17)
667#define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18)
668#define EXT4_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY
669#define EXT4_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT
670#define EXT4_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY
671/* ioctl codes 19--39 are reserved for fscrypt */
672#define EXT4_IOC_CLEAR_ES_CACHE _IO('f', 40)
673#define EXT4_IOC_GETSTATE _IOW('f', 41, __u32)
674#define EXT4_IOC_GET_ES_CACHE _IOWR('f', 42, struct fiemap)
675
676#define EXT4_IOC_FSGETXATTR FS_IOC_FSGETXATTR
677#define EXT4_IOC_FSSETXATTR FS_IOC_FSSETXATTR
678
679#define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32)
680
681/*
682 * Flags for going down operation
683 */
684#define EXT4_GOING_FLAGS_DEFAULT 0x0 /* going down */
685#define EXT4_GOING_FLAGS_LOGFLUSH 0x1 /* flush log but not data */
686#define EXT4_GOING_FLAGS_NOLOGFLUSH 0x2 /* don't flush log nor data */
687
688/*
689 * Flags returned by EXT4_IOC_GETSTATE
690 *
691 * We only expose to userspace a subset of the state flags in
692 * i_state_flags
693 */
694#define EXT4_STATE_FLAG_EXT_PRECACHED 0x00000001
695#define EXT4_STATE_FLAG_NEW 0x00000002
696#define EXT4_STATE_FLAG_NEWENTRY 0x00000004
697#define EXT4_STATE_FLAG_DA_ALLOC_CLOSE 0x00000008
698
699#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
700/*
701 * ioctl commands in 32 bit emulation
702 */
703#define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
704#define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
705#define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
706#define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
707#define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
708#define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
709#define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
710#define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
711#define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
712#define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
713#endif
714
715/*
716 * Returned by EXT4_IOC_GET_ES_CACHE as an additional possible flag.
717 * It indicates that the entry in extent status cache is for a hole.
718 */
719#define EXT4_FIEMAP_EXTENT_HOLE 0x08000000
720
721/* Max physical block we can address w/o extents */
722#define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
723
724/* Max logical block we can support */
725#define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFF
726
727/*
728 * Structure of an inode on the disk
729 */
730struct ext4_inode {
731 __le16 i_mode; /* File mode */
732 __le16 i_uid; /* Low 16 bits of Owner Uid */
733 __le32 i_size_lo; /* Size in bytes */
734 __le32 i_atime; /* Access time */
735 __le32 i_ctime; /* Inode Change time */
736 __le32 i_mtime; /* Modification time */
737 __le32 i_dtime; /* Deletion Time */
738 __le16 i_gid; /* Low 16 bits of Group Id */
739 __le16 i_links_count; /* Links count */
740 __le32 i_blocks_lo; /* Blocks count */
741 __le32 i_flags; /* File flags */
742 union {
743 struct {
744 __le32 l_i_version;
745 } linux1;
746 struct {
747 __u32 h_i_translator;
748 } hurd1;
749 struct {
750 __u32 m_i_reserved1;
751 } masix1;
752 } osd1; /* OS dependent 1 */
753 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
754 __le32 i_generation; /* File version (for NFS) */
755 __le32 i_file_acl_lo; /* File ACL */
756 __le32 i_size_high;
757 __le32 i_obso_faddr; /* Obsoleted fragment address */
758 union {
759 struct {
760 __le16 l_i_blocks_high; /* were l_i_reserved1 */
761 __le16 l_i_file_acl_high;
762 __le16 l_i_uid_high; /* these 2 fields */
763 __le16 l_i_gid_high; /* were reserved2[0] */
764 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
765 __le16 l_i_reserved;
766 } linux2;
767 struct {
768 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
769 __u16 h_i_mode_high;
770 __u16 h_i_uid_high;
771 __u16 h_i_gid_high;
772 __u32 h_i_author;
773 } hurd2;
774 struct {
775 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
776 __le16 m_i_file_acl_high;
777 __u32 m_i_reserved2[2];
778 } masix2;
779 } osd2; /* OS dependent 2 */
780 __le16 i_extra_isize;
781 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
782 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
783 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
784 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
785 __le32 i_crtime; /* File Creation time */
786 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
787 __le32 i_version_hi; /* high 32 bits for 64-bit version */
788 __le32 i_projid; /* Project ID */
789};
790
791struct move_extent {
792 __u32 reserved; /* should be zero */
793 __u32 donor_fd; /* donor file descriptor */
794 __u64 orig_start; /* logical start offset in block for orig */
795 __u64 donor_start; /* logical start offset in block for donor */
796 __u64 len; /* block length to be moved */
797 __u64 moved_len; /* moved block length */
798};
799
800#define EXT4_EPOCH_BITS 2
801#define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
802#define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
803
804/*
805 * Extended fields will fit into an inode if the filesystem was formatted
806 * with large inodes (-I 256 or larger) and there are not currently any EAs
807 * consuming all of the available space. For new inodes we always reserve
808 * enough space for the kernel's known extended fields, but for inodes
809 * created with an old kernel this might not have been the case. None of
810 * the extended inode fields is critical for correct filesystem operation.
811 * This macro checks if a certain field fits in the inode. Note that
812 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
813 */
814#define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
815 ((offsetof(typeof(*ext4_inode), field) + \
816 sizeof((ext4_inode)->field)) \
817 <= (EXT4_GOOD_OLD_INODE_SIZE + \
818 (einode)->i_extra_isize)) \
819
820/*
821 * We use an encoding that preserves the times for extra epoch "00":
822 *
823 * extra msb of adjust for signed
824 * epoch 32-bit 32-bit tv_sec to
825 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
826 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
827 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
828 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
829 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
830 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
831 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
832 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
833 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
834 *
835 * Note that previous versions of the kernel on 64-bit systems would
836 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
837 * 1970. e2fsck will correct this, assuming that it is run on the
838 * affected filesystem before 2242.
839 */
840
841static inline __le32 ext4_encode_extra_time(struct timespec64 *time)
842{
843 u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK;
844 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
845}
846
847static inline void ext4_decode_extra_time(struct timespec64 *time,
848 __le32 extra)
849{
850 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK)))
851 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
852 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
853}
854
855#define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
856do { \
857 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {\
858 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
859 (raw_inode)->xtime ## _extra = \
860 ext4_encode_extra_time(&(inode)->xtime); \
861 } \
862 else \
863 (raw_inode)->xtime = cpu_to_le32(clamp_t(int32_t, (inode)->xtime.tv_sec, S32_MIN, S32_MAX)); \
864} while (0)
865
866#define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
867do { \
868 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
869 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
870 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
871 (raw_inode)->xtime ## _extra = \
872 ext4_encode_extra_time(&(einode)->xtime); \
873} while (0)
874
875#define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
876do { \
877 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
878 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \
879 ext4_decode_extra_time(&(inode)->xtime, \
880 raw_inode->xtime ## _extra); \
881 } \
882 else \
883 (inode)->xtime.tv_nsec = 0; \
884} while (0)
885
886
887#define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
888do { \
889 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
890 (einode)->xtime.tv_sec = \
891 (signed)le32_to_cpu((raw_inode)->xtime); \
892 else \
893 (einode)->xtime.tv_sec = 0; \
894 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
895 ext4_decode_extra_time(&(einode)->xtime, \
896 raw_inode->xtime ## _extra); \
897 else \
898 (einode)->xtime.tv_nsec = 0; \
899} while (0)
900
901#define i_disk_version osd1.linux1.l_i_version
902
903#if defined(__KERNEL__) || defined(__linux__)
904#define i_reserved1 osd1.linux1.l_i_reserved1
905#define i_file_acl_high osd2.linux2.l_i_file_acl_high
906#define i_blocks_high osd2.linux2.l_i_blocks_high
907#define i_uid_low i_uid
908#define i_gid_low i_gid
909#define i_uid_high osd2.linux2.l_i_uid_high
910#define i_gid_high osd2.linux2.l_i_gid_high
911#define i_checksum_lo osd2.linux2.l_i_checksum_lo
912
913#elif defined(__GNU__)
914
915#define i_translator osd1.hurd1.h_i_translator
916#define i_uid_high osd2.hurd2.h_i_uid_high
917#define i_gid_high osd2.hurd2.h_i_gid_high
918#define i_author osd2.hurd2.h_i_author
919
920#elif defined(__masix__)
921
922#define i_reserved1 osd1.masix1.m_i_reserved1
923#define i_file_acl_high osd2.masix2.m_i_file_acl_high
924#define i_reserved2 osd2.masix2.m_i_reserved2
925
926#endif /* defined(__KERNEL__) || defined(__linux__) */
927
928#include "extents_status.h"
929
930/*
931 * Lock subclasses for i_data_sem in the ext4_inode_info structure.
932 *
933 * These are needed to avoid lockdep false positives when we need to
934 * allocate blocks to the quota inode during ext4_map_blocks(), while
935 * holding i_data_sem for a normal (non-quota) inode. Since we don't
936 * do quota tracking for the quota inode, this avoids deadlock (as
937 * well as infinite recursion, since it isn't turtles all the way
938 * down...)
939 *
940 * I_DATA_SEM_NORMAL - Used for most inodes
941 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode
942 * where the second inode has larger inode number
943 * than the first
944 * I_DATA_SEM_QUOTA - Used for quota inodes only
945 */
946enum {
947 I_DATA_SEM_NORMAL = 0,
948 I_DATA_SEM_OTHER,
949 I_DATA_SEM_QUOTA,
950};
951
952
953/*
954 * fourth extended file system inode data in memory
955 */
956struct ext4_inode_info {
957 __le32 i_data[15]; /* unconverted */
958 __u32 i_dtime;
959 ext4_fsblk_t i_file_acl;
960
961 /*
962 * i_block_group is the number of the block group which contains
963 * this file's inode. Constant across the lifetime of the inode,
964 * it is used for making block allocation decisions - we try to
965 * place a file's data blocks near its inode block, and new inodes
966 * near to their parent directory's inode.
967 */
968 ext4_group_t i_block_group;
969 ext4_lblk_t i_dir_start_lookup;
970#if (BITS_PER_LONG < 64)
971 unsigned long i_state_flags; /* Dynamic state flags */
972#endif
973 unsigned long i_flags;
974
975 /*
976 * Extended attributes can be read independently of the main file
977 * data. Taking i_mutex even when reading would cause contention
978 * between readers of EAs and writers of regular file data, so
979 * instead we synchronize on xattr_sem when reading or changing
980 * EAs.
981 */
982 struct rw_semaphore xattr_sem;
983
984 struct list_head i_orphan; /* unlinked but open inodes */
985
986 /*
987 * i_disksize keeps track of what the inode size is ON DISK, not
988 * in memory. During truncate, i_size is set to the new size by
989 * the VFS prior to calling ext4_truncate(), but the filesystem won't
990 * set i_disksize to 0 until the truncate is actually under way.
991 *
992 * The intent is that i_disksize always represents the blocks which
993 * are used by this file. This allows recovery to restart truncate
994 * on orphans if we crash during truncate. We actually write i_disksize
995 * into the on-disk inode when writing inodes out, instead of i_size.
996 *
997 * The only time when i_disksize and i_size may be different is when
998 * a truncate is in progress. The only things which change i_disksize
999 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
1000 */
1001 loff_t i_disksize;
1002
1003 /*
1004 * i_data_sem is for serialising ext4_truncate() against
1005 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
1006 * data tree are chopped off during truncate. We can't do that in
1007 * ext4 because whenever we perform intermediate commits during
1008 * truncate, the inode and all the metadata blocks *must* be in a
1009 * consistent state which allows truncation of the orphans to restart
1010 * during recovery. Hence we must fix the get_block-vs-truncate race
1011 * by other means, so we have i_data_sem.
1012 */
1013 struct rw_semaphore i_data_sem;
1014 /*
1015 * i_mmap_sem is for serializing page faults with truncate / punch hole
1016 * operations. We have to make sure that new page cannot be faulted in
1017 * a section of the inode that is being punched. We cannot easily use
1018 * i_data_sem for this since we need protection for the whole punch
1019 * operation and i_data_sem ranks below transaction start so we have
1020 * to occasionally drop it.
1021 */
1022 struct rw_semaphore i_mmap_sem;
1023 struct inode vfs_inode;
1024 struct jbd2_inode *jinode;
1025
1026 spinlock_t i_raw_lock; /* protects updates to the raw inode */
1027
1028 /*
1029 * File creation time. Its function is same as that of
1030 * struct timespec64 i_{a,c,m}time in the generic inode.
1031 */
1032 struct timespec64 i_crtime;
1033
1034 /* mballoc */
1035 struct list_head i_prealloc_list;
1036 spinlock_t i_prealloc_lock;
1037
1038 /* extents status tree */
1039 struct ext4_es_tree i_es_tree;
1040 rwlock_t i_es_lock;
1041 struct list_head i_es_list;
1042 unsigned int i_es_all_nr; /* protected by i_es_lock */
1043 unsigned int i_es_shk_nr; /* protected by i_es_lock */
1044 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for
1045 extents to shrink. Protected by
1046 i_es_lock */
1047
1048 /* ialloc */
1049 ext4_group_t i_last_alloc_group;
1050
1051 /* allocation reservation info for delalloc */
1052 /* In case of bigalloc, this refer to clusters rather than blocks */
1053 unsigned int i_reserved_data_blocks;
1054
1055 /* pending cluster reservations for bigalloc file systems */
1056 struct ext4_pending_tree i_pending_tree;
1057
1058 /* on-disk additional length */
1059 __u16 i_extra_isize;
1060
1061 /* Indicate the inline data space. */
1062 u16 i_inline_off;
1063 u16 i_inline_size;
1064
1065#ifdef CONFIG_QUOTA
1066 /* quota space reservation, managed internally by quota code */
1067 qsize_t i_reserved_quota;
1068#endif
1069
1070 /* Lock protecting lists below */
1071 spinlock_t i_completed_io_lock;
1072 /*
1073 * Completed IOs that need unwritten extents handling and have
1074 * transaction reserved
1075 */
1076 struct list_head i_rsv_conversion_list;
1077 struct work_struct i_rsv_conversion_work;
1078 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1079
1080 spinlock_t i_block_reservation_lock;
1081
1082 /*
1083 * Transactions that contain inode's metadata needed to complete
1084 * fsync and fdatasync, respectively.
1085 */
1086 tid_t i_sync_tid;
1087 tid_t i_datasync_tid;
1088
1089#ifdef CONFIG_QUOTA
1090 struct dquot *i_dquot[MAXQUOTAS];
1091#endif
1092
1093 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1094 __u32 i_csum_seed;
1095
1096 kprojid_t i_projid;
1097};
1098
1099/*
1100 * File system states
1101 */
1102#define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
1103#define EXT4_ERROR_FS 0x0002 /* Errors detected */
1104#define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
1105
1106/*
1107 * Misc. filesystem flags
1108 */
1109#define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
1110#define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
1111#define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
1112
1113/*
1114 * Mount flags set via mount options or defaults
1115 */
1116#define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */
1117#define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
1118#define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
1119#define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
1120#define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
1121#define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
1122#define EXT4_MOUNT_ERRORS_MASK 0x00070
1123#define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
1124#define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
1125#ifdef CONFIG_FS_DAX
1126#define EXT4_MOUNT_DAX 0x00200 /* Direct Access */
1127#else
1128#define EXT4_MOUNT_DAX 0
1129#endif
1130#define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
1131#define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
1132#define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
1133#define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
1134#define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
1135#define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
1136#define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
1137#define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
1138#define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
1139#define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
1140#define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */
1141#define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota,
1142 * enable enforcement for hidden
1143 * quota files */
1144#define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable
1145 * enforcement for hidden quota
1146 * files */
1147#define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota
1148 * enforcement */
1149#define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
1150#define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
1151#define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1152#define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */
1153#define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
1154#define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
1155#define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
1156#define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
1157#define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
1158
1159/*
1160 * Mount flags set either automatically (could not be set by mount option)
1161 * based on per file system feature or property or in special cases such as
1162 * distinguishing between explicit mount option definition and default.
1163 */
1164#define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1165 specified delalloc */
1166#define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1167 size of blocksize * 8
1168 blocks */
1169#define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1170 file systems */
1171
1172#define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly
1173 specified journal checksum */
1174
1175#define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1176 ~EXT4_MOUNT_##opt
1177#define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1178 EXT4_MOUNT_##opt
1179#define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1180 EXT4_MOUNT_##opt)
1181
1182#define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1183 ~EXT4_MOUNT2_##opt
1184#define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1185 EXT4_MOUNT2_##opt
1186#define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1187 EXT4_MOUNT2_##opt)
1188
1189#define ext4_test_and_set_bit __test_and_set_bit_le
1190#define ext4_set_bit __set_bit_le
1191#define ext4_set_bit_atomic ext2_set_bit_atomic
1192#define ext4_test_and_clear_bit __test_and_clear_bit_le
1193#define ext4_clear_bit __clear_bit_le
1194#define ext4_clear_bit_atomic ext2_clear_bit_atomic
1195#define ext4_test_bit test_bit_le
1196#define ext4_find_next_zero_bit find_next_zero_bit_le
1197#define ext4_find_next_bit find_next_bit_le
1198
1199extern void ext4_set_bits(void *bm, int cur, int len);
1200
1201/*
1202 * Maximal mount counts between two filesystem checks
1203 */
1204#define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1205#define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1206
1207/*
1208 * Behaviour when detecting errors
1209 */
1210#define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1211#define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1212#define EXT4_ERRORS_PANIC 3 /* Panic */
1213#define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1214
1215/* Metadata checksum algorithm codes */
1216#define EXT4_CRC32C_CHKSUM 1
1217
1218/*
1219 * Structure of the super block
1220 */
1221struct ext4_super_block {
1222/*00*/ __le32 s_inodes_count; /* Inodes count */
1223 __le32 s_blocks_count_lo; /* Blocks count */
1224 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1225 __le32 s_free_blocks_count_lo; /* Free blocks count */
1226/*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1227 __le32 s_first_data_block; /* First Data Block */
1228 __le32 s_log_block_size; /* Block size */
1229 __le32 s_log_cluster_size; /* Allocation cluster size */
1230/*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1231 __le32 s_clusters_per_group; /* # Clusters per group */
1232 __le32 s_inodes_per_group; /* # Inodes per group */
1233 __le32 s_mtime; /* Mount time */
1234/*30*/ __le32 s_wtime; /* Write time */
1235 __le16 s_mnt_count; /* Mount count */
1236 __le16 s_max_mnt_count; /* Maximal mount count */
1237 __le16 s_magic; /* Magic signature */
1238 __le16 s_state; /* File system state */
1239 __le16 s_errors; /* Behaviour when detecting errors */
1240 __le16 s_minor_rev_level; /* minor revision level */
1241/*40*/ __le32 s_lastcheck; /* time of last check */
1242 __le32 s_checkinterval; /* max. time between checks */
1243 __le32 s_creator_os; /* OS */
1244 __le32 s_rev_level; /* Revision level */
1245/*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1246 __le16 s_def_resgid; /* Default gid for reserved blocks */
1247 /*
1248 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1249 *
1250 * Note: the difference between the compatible feature set and
1251 * the incompatible feature set is that if there is a bit set
1252 * in the incompatible feature set that the kernel doesn't
1253 * know about, it should refuse to mount the filesystem.
1254 *
1255 * e2fsck's requirements are more strict; if it doesn't know
1256 * about a feature in either the compatible or incompatible
1257 * feature set, it must abort and not try to meddle with
1258 * things it doesn't understand...
1259 */
1260 __le32 s_first_ino; /* First non-reserved inode */
1261 __le16 s_inode_size; /* size of inode structure */
1262 __le16 s_block_group_nr; /* block group # of this superblock */
1263 __le32 s_feature_compat; /* compatible feature set */
1264/*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1265 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1266/*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1267/*78*/ char s_volume_name[16]; /* volume name */
1268/*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */
1269/*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1270 /*
1271 * Performance hints. Directory preallocation should only
1272 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1273 */
1274 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1275 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1276 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1277 /*
1278 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1279 */
1280/*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1281/*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1282 __le32 s_journal_dev; /* device number of journal file */
1283 __le32 s_last_orphan; /* start of list of inodes to delete */
1284 __le32 s_hash_seed[4]; /* HTREE hash seed */
1285 __u8 s_def_hash_version; /* Default hash version to use */
1286 __u8 s_jnl_backup_type;
1287 __le16 s_desc_size; /* size of group descriptor */
1288/*100*/ __le32 s_default_mount_opts;
1289 __le32 s_first_meta_bg; /* First metablock block group */
1290 __le32 s_mkfs_time; /* When the filesystem was created */
1291 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1292 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1293/*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1294 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1295 __le32 s_free_blocks_count_hi; /* Free blocks count */
1296 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1297 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1298 __le32 s_flags; /* Miscellaneous flags */
1299 __le16 s_raid_stride; /* RAID stride */
1300 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1301 __le64 s_mmp_block; /* Block for multi-mount protection */
1302 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1303 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1304 __u8 s_checksum_type; /* metadata checksum algorithm used */
1305 __u8 s_encryption_level; /* versioning level for encryption */
1306 __u8 s_reserved_pad; /* Padding to next 32bits */
1307 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1308 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1309 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1310 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1311 snapshot's future use */
1312 __le32 s_snapshot_list; /* inode number of the head of the
1313 on-disk snapshot list */
1314#define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1315 __le32 s_error_count; /* number of fs errors */
1316 __le32 s_first_error_time; /* first time an error happened */
1317 __le32 s_first_error_ino; /* inode involved in first error */
1318 __le64 s_first_error_block; /* block involved of first error */
1319 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */
1320 __le32 s_first_error_line; /* line number where error happened */
1321 __le32 s_last_error_time; /* most recent time of an error */
1322 __le32 s_last_error_ino; /* inode involved in last error */
1323 __le32 s_last_error_line; /* line number where error happened */
1324 __le64 s_last_error_block; /* block involved of last error */
1325 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */
1326#define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1327 __u8 s_mount_opts[64];
1328 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1329 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1330 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1331 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
1332 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */
1333 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */
1334 __le32 s_lpf_ino; /* Location of the lost+found inode */
1335 __le32 s_prj_quota_inum; /* inode for tracking project quota */
1336 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */
1337 __u8 s_wtime_hi;
1338 __u8 s_mtime_hi;
1339 __u8 s_mkfs_time_hi;
1340 __u8 s_lastcheck_hi;
1341 __u8 s_first_error_time_hi;
1342 __u8 s_last_error_time_hi;
1343 __u8 s_first_error_errcode;
1344 __u8 s_last_error_errcode;
1345 __le16 s_encoding; /* Filename charset encoding */
1346 __le16 s_encoding_flags; /* Filename charset encoding flags */
1347 __le32 s_reserved[95]; /* Padding to the end of the block */
1348 __le32 s_checksum; /* crc32c(superblock) */
1349};
1350
1351#define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1352
1353#ifdef __KERNEL__
1354
1355/*
1356 * run-time mount flags
1357 */
1358#define EXT4_MF_MNTDIR_SAMPLED 0x0001
1359#define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1360#define EXT4_MF_TEST_DUMMY_ENCRYPTION 0x0004
1361
1362#ifdef CONFIG_FS_ENCRYPTION
1363#define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \
1364 EXT4_MF_TEST_DUMMY_ENCRYPTION))
1365#else
1366#define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1367#endif
1368
1369/* Number of quota types we support */
1370#define EXT4_MAXQUOTAS 3
1371
1372#define EXT4_ENC_UTF8_12_1 1
1373
1374/*
1375 * Flags for ext4_sb_info.s_encoding_flags.
1376 */
1377#define EXT4_ENC_STRICT_MODE_FL (1 << 0)
1378
1379#define ext4_has_strict_mode(sbi) \
1380 (sbi->s_encoding_flags & EXT4_ENC_STRICT_MODE_FL)
1381
1382/*
1383 * fourth extended-fs super-block data in memory
1384 */
1385struct ext4_sb_info {
1386 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1387 unsigned long s_inodes_per_block;/* Number of inodes per block */
1388 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1389 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1390 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1391 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1392 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1393 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1394 ext4_group_t s_groups_count; /* Number of groups in the fs */
1395 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1396 unsigned long s_overhead; /* # of fs overhead clusters */
1397 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1398 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1399 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1400 struct buffer_head * s_sbh; /* Buffer containing the super block */
1401 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1402 struct buffer_head * __rcu *s_group_desc;
1403 unsigned int s_mount_opt;
1404 unsigned int s_mount_opt2;
1405 unsigned int s_mount_flags;
1406 unsigned int s_def_mount_opt;
1407 ext4_fsblk_t s_sb_block;
1408 atomic64_t s_resv_clusters;
1409 kuid_t s_resuid;
1410 kgid_t s_resgid;
1411 unsigned short s_mount_state;
1412 unsigned short s_pad;
1413 int s_addr_per_block_bits;
1414 int s_desc_per_block_bits;
1415 int s_inode_size;
1416 int s_first_ino;
1417 unsigned int s_inode_readahead_blks;
1418 unsigned int s_inode_goal;
1419 u32 s_hash_seed[4];
1420 int s_def_hash_version;
1421 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1422 struct percpu_counter s_freeclusters_counter;
1423 struct percpu_counter s_freeinodes_counter;
1424 struct percpu_counter s_dirs_counter;
1425 struct percpu_counter s_dirtyclusters_counter;
1426 struct blockgroup_lock *s_blockgroup_lock;
1427 struct proc_dir_entry *s_proc;
1428 struct kobject s_kobj;
1429 struct completion s_kobj_unregister;
1430 struct super_block *s_sb;
1431#ifdef CONFIG_UNICODE
1432 struct unicode_map *s_encoding;
1433 __u16 s_encoding_flags;
1434#endif
1435
1436 /* Journaling */
1437 struct journal_s *s_journal;
1438 struct list_head s_orphan;
1439 struct mutex s_orphan_lock;
1440 unsigned long s_ext4_flags; /* Ext4 superblock flags */
1441 unsigned long s_commit_interval;
1442 u32 s_max_batch_time;
1443 u32 s_min_batch_time;
1444 struct block_device *journal_bdev;
1445#ifdef CONFIG_QUOTA
1446 /* Names of quota files with journalled quota */
1447 char __rcu *s_qf_names[EXT4_MAXQUOTAS];
1448 int s_jquota_fmt; /* Format of quota to use */
1449#endif
1450 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1451 struct ext4_system_blocks __rcu *system_blks;
1452
1453#ifdef EXTENTS_STATS
1454 /* ext4 extents stats */
1455 unsigned long s_ext_min;
1456 unsigned long s_ext_max;
1457 unsigned long s_depth_max;
1458 spinlock_t s_ext_stats_lock;
1459 unsigned long s_ext_blocks;
1460 unsigned long s_ext_extents;
1461#endif
1462
1463 /* for buddy allocator */
1464 struct ext4_group_info ** __rcu *s_group_info;
1465 struct inode *s_buddy_cache;
1466 spinlock_t s_md_lock;
1467 unsigned short *s_mb_offsets;
1468 unsigned int *s_mb_maxs;
1469 unsigned int s_group_info_size;
1470 unsigned int s_mb_free_pending;
1471 struct list_head s_freed_data_list; /* List of blocks to be freed
1472 after commit completed */
1473
1474 /* tunables */
1475 unsigned long s_stripe;
1476 unsigned int s_mb_stream_request;
1477 unsigned int s_mb_max_to_scan;
1478 unsigned int s_mb_min_to_scan;
1479 unsigned int s_mb_stats;
1480 unsigned int s_mb_order2_reqs;
1481 unsigned int s_mb_group_prealloc;
1482 unsigned int s_max_dir_size_kb;
1483 /* where last allocation was done - for stream allocation */
1484 unsigned long s_mb_last_group;
1485 unsigned long s_mb_last_start;
1486
1487 /* stats for buddy allocator */
1488 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1489 atomic_t s_bal_success; /* we found long enough chunks */
1490 atomic_t s_bal_allocated; /* in blocks */
1491 atomic_t s_bal_ex_scanned; /* total extents scanned */
1492 atomic_t s_bal_goals; /* goal hits */
1493 atomic_t s_bal_breaks; /* too long searches */
1494 atomic_t s_bal_2orders; /* 2^order hits */
1495 spinlock_t s_bal_lock;
1496 unsigned long s_mb_buddies_generated;
1497 unsigned long long s_mb_generation_time;
1498 atomic_t s_mb_lost_chunks;
1499 atomic_t s_mb_preallocated;
1500 atomic_t s_mb_discarded;
1501 atomic_t s_lock_busy;
1502
1503 /* locality groups */
1504 struct ext4_locality_group __percpu *s_locality_groups;
1505
1506 /* for write statistics */
1507 unsigned long s_sectors_written_start;
1508 u64 s_kbytes_written;
1509
1510 /* the size of zero-out chunk */
1511 unsigned int s_extent_max_zeroout_kb;
1512
1513 unsigned int s_log_groups_per_flex;
1514 struct flex_groups * __rcu *s_flex_groups;
1515 ext4_group_t s_flex_groups_allocated;
1516
1517 /* workqueue for reserved extent conversions (buffered io) */
1518 struct workqueue_struct *rsv_conversion_wq;
1519
1520 /* timer for periodic error stats printing */
1521 struct timer_list s_err_report;
1522
1523 /* Lazy inode table initialization info */
1524 struct ext4_li_request *s_li_request;
1525 /* Wait multiplier for lazy initialization thread */
1526 unsigned int s_li_wait_mult;
1527
1528 /* Kernel thread for multiple mount protection */
1529 struct task_struct *s_mmp_tsk;
1530
1531 /* record the last minlen when FITRIM is called. */
1532 atomic_t s_last_trim_minblks;
1533
1534 /* Reference to checksum algorithm driver via cryptoapi */
1535 struct crypto_shash *s_chksum_driver;
1536
1537 /* Precomputed FS UUID checksum for seeding other checksums */
1538 __u32 s_csum_seed;
1539
1540 /* Reclaim extents from extent status tree */
1541 struct shrinker s_es_shrinker;
1542 struct list_head s_es_list; /* List of inodes with reclaimable extents */
1543 long s_es_nr_inode;
1544 struct ext4_es_stats s_es_stats;
1545 struct mb_cache *s_ea_block_cache;
1546 struct mb_cache *s_ea_inode_cache;
1547 spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1548
1549 /* Ratelimit ext4 messages. */
1550 struct ratelimit_state s_err_ratelimit_state;
1551 struct ratelimit_state s_warning_ratelimit_state;
1552 struct ratelimit_state s_msg_ratelimit_state;
1553
1554 /*
1555 * Barrier between writepages ops and changing any inode's JOURNAL_DATA
1556 * or EXTENTS flag.
1557 */
1558 struct percpu_rw_semaphore s_writepages_rwsem;
1559 struct dax_device *s_daxdev;
1560#ifdef CONFIG_EXT4_DEBUG
1561 unsigned long s_simulate_fail;
1562#endif
1563};
1564
1565static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1566{
1567 return sb->s_fs_info;
1568}
1569static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1570{
1571 return container_of(inode, struct ext4_inode_info, vfs_inode);
1572}
1573
1574static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1575{
1576 return ino == EXT4_ROOT_INO ||
1577 (ino >= EXT4_FIRST_INO(sb) &&
1578 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1579}
1580
1581/*
1582 * Returns: sbi->field[index]
1583 * Used to access an array element from the following sbi fields which require
1584 * rcu protection to avoid dereferencing an invalid pointer due to reassignment
1585 * - s_group_desc
1586 * - s_group_info
1587 * - s_flex_group
1588 */
1589#define sbi_array_rcu_deref(sbi, field, index) \
1590({ \
1591 typeof(*((sbi)->field)) _v; \
1592 rcu_read_lock(); \
1593 _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index]; \
1594 rcu_read_unlock(); \
1595 _v; \
1596})
1597
1598/*
1599 * Simulate_fail codes
1600 */
1601#define EXT4_SIM_BBITMAP_EIO 1
1602#define EXT4_SIM_BBITMAP_CRC 2
1603#define EXT4_SIM_IBITMAP_EIO 3
1604#define EXT4_SIM_IBITMAP_CRC 4
1605#define EXT4_SIM_INODE_EIO 5
1606#define EXT4_SIM_INODE_CRC 6
1607#define EXT4_SIM_DIRBLOCK_EIO 7
1608#define EXT4_SIM_DIRBLOCK_CRC 8
1609
1610static inline bool ext4_simulate_fail(struct super_block *sb,
1611 unsigned long code)
1612{
1613#ifdef CONFIG_EXT4_DEBUG
1614 struct ext4_sb_info *sbi = EXT4_SB(sb);
1615
1616 if (unlikely(sbi->s_simulate_fail == code)) {
1617 sbi->s_simulate_fail = 0;
1618 return true;
1619 }
1620#endif
1621 return false;
1622}
1623
1624static inline void ext4_simulate_fail_bh(struct super_block *sb,
1625 struct buffer_head *bh,
1626 unsigned long code)
1627{
1628 if (!IS_ERR(bh) && ext4_simulate_fail(sb, code))
1629 clear_buffer_uptodate(bh);
1630}
1631
1632/*
1633 * Error number codes for s_{first,last}_error_errno
1634 *
1635 * Linux errno numbers are architecture specific, so we need to translate
1636 * them into something which is architecture independent. We don't define
1637 * codes for all errno's; just the ones which are most likely to be the cause
1638 * of an ext4_error() call.
1639 */
1640#define EXT4_ERR_UNKNOWN 1
1641#define EXT4_ERR_EIO 2
1642#define EXT4_ERR_ENOMEM 3
1643#define EXT4_ERR_EFSBADCRC 4
1644#define EXT4_ERR_EFSCORRUPTED 5
1645#define EXT4_ERR_ENOSPC 6
1646#define EXT4_ERR_ENOKEY 7
1647#define EXT4_ERR_EROFS 8
1648#define EXT4_ERR_EFBIG 9
1649#define EXT4_ERR_EEXIST 10
1650#define EXT4_ERR_ERANGE 11
1651#define EXT4_ERR_EOVERFLOW 12
1652#define EXT4_ERR_EBUSY 13
1653#define EXT4_ERR_ENOTDIR 14
1654#define EXT4_ERR_ENOTEMPTY 15
1655#define EXT4_ERR_ESHUTDOWN 16
1656#define EXT4_ERR_EFAULT 17
1657
1658/*
1659 * Inode dynamic state flags
1660 */
1661enum {
1662 EXT4_STATE_JDATA, /* journaled data exists */
1663 EXT4_STATE_NEW, /* inode is newly created */
1664 EXT4_STATE_XATTR, /* has in-inode xattrs */
1665 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1666 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1667 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1668 EXT4_STATE_NEWENTRY, /* File just added to dir */
1669 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1670 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
1671 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */
1672 EXT4_STATE_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */
1673};
1674
1675#define EXT4_INODE_BIT_FNS(name, field, offset) \
1676static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1677{ \
1678 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1679} \
1680static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1681{ \
1682 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1683} \
1684static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1685{ \
1686 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1687}
1688
1689/* Add these declarations here only so that these functions can be
1690 * found by name. Otherwise, they are very hard to locate. */
1691static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1692static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1693static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1694EXT4_INODE_BIT_FNS(flag, flags, 0)
1695
1696/* Add these declarations here only so that these functions can be
1697 * found by name. Otherwise, they are very hard to locate. */
1698static inline int ext4_test_inode_state(struct inode *inode, int bit);
1699static inline void ext4_set_inode_state(struct inode *inode, int bit);
1700static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1701#if (BITS_PER_LONG < 64)
1702EXT4_INODE_BIT_FNS(state, state_flags, 0)
1703
1704static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1705{
1706 (ei)->i_state_flags = 0;
1707}
1708#else
1709EXT4_INODE_BIT_FNS(state, flags, 32)
1710
1711static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1712{
1713 /* We depend on the fact that callers will set i_flags */
1714}
1715#endif
1716#else
1717/* Assume that user mode programs are passing in an ext4fs superblock, not
1718 * a kernel struct super_block. This will allow us to call the feature-test
1719 * macros from user land. */
1720#define EXT4_SB(sb) (sb)
1721#endif
1722
1723static inline bool ext4_verity_in_progress(struct inode *inode)
1724{
1725 return IS_ENABLED(CONFIG_FS_VERITY) &&
1726 ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS);
1727}
1728
1729#define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1730
1731/*
1732 * Codes for operating systems
1733 */
1734#define EXT4_OS_LINUX 0
1735#define EXT4_OS_HURD 1
1736#define EXT4_OS_MASIX 2
1737#define EXT4_OS_FREEBSD 3
1738#define EXT4_OS_LITES 4
1739
1740/*
1741 * Revision levels
1742 */
1743#define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1744#define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1745
1746#define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1747#define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1748
1749#define EXT4_GOOD_OLD_INODE_SIZE 128
1750
1751#define EXT4_EXTRA_TIMESTAMP_MAX (((s64)1 << 34) - 1 + S32_MIN)
1752#define EXT4_NON_EXTRA_TIMESTAMP_MAX S32_MAX
1753#define EXT4_TIMESTAMP_MIN S32_MIN
1754
1755/*
1756 * Feature set definitions
1757 */
1758
1759#define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1760#define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1761#define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1762#define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1763#define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1764#define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1765#define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
1766#define EXT4_FEATURE_COMPAT_STABLE_INODES 0x0800
1767
1768#define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1769#define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1770#define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1771#define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1772#define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1773#define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1774#define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1775#define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1776#define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1777/*
1778 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1779 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1780 * all other data structures' checksums. However, the METADATA_CSUM and
1781 * GDT_CSUM bits are mutually exclusive.
1782 */
1783#define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
1784#define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000
1785#define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000
1786#define EXT4_FEATURE_RO_COMPAT_VERITY 0x8000
1787
1788#define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1789#define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1790#define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1791#define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1792#define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1793#define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1794#define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1795#define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1796#define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1797#define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1798#define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1799#define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000
1800#define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
1801#define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
1802#define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000
1803#define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000
1804
1805extern void ext4_update_dynamic_rev(struct super_block *sb);
1806
1807#define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
1808static inline bool ext4_has_feature_##name(struct super_block *sb) \
1809{ \
1810 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1811 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
1812} \
1813static inline void ext4_set_feature_##name(struct super_block *sb) \
1814{ \
1815 ext4_update_dynamic_rev(sb); \
1816 EXT4_SB(sb)->s_es->s_feature_compat |= \
1817 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1818} \
1819static inline void ext4_clear_feature_##name(struct super_block *sb) \
1820{ \
1821 EXT4_SB(sb)->s_es->s_feature_compat &= \
1822 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1823}
1824
1825#define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
1826static inline bool ext4_has_feature_##name(struct super_block *sb) \
1827{ \
1828 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1829 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
1830} \
1831static inline void ext4_set_feature_##name(struct super_block *sb) \
1832{ \
1833 ext4_update_dynamic_rev(sb); \
1834 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
1835 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1836} \
1837static inline void ext4_clear_feature_##name(struct super_block *sb) \
1838{ \
1839 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
1840 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1841}
1842
1843#define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
1844static inline bool ext4_has_feature_##name(struct super_block *sb) \
1845{ \
1846 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1847 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
1848} \
1849static inline void ext4_set_feature_##name(struct super_block *sb) \
1850{ \
1851 ext4_update_dynamic_rev(sb); \
1852 EXT4_SB(sb)->s_es->s_feature_incompat |= \
1853 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1854} \
1855static inline void ext4_clear_feature_##name(struct super_block *sb) \
1856{ \
1857 EXT4_SB(sb)->s_es->s_feature_incompat &= \
1858 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1859}
1860
1861EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC)
1862EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES)
1863EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL)
1864EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR)
1865EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE)
1866EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX)
1867EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2)
1868EXT4_FEATURE_COMPAT_FUNCS(stable_inodes, STABLE_INODES)
1869
1870EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER)
1871EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE)
1872EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR)
1873EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE)
1874EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM)
1875EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK)
1876EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE)
1877EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA)
1878EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC)
1879EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM)
1880EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY)
1881EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT)
1882EXT4_FEATURE_RO_COMPAT_FUNCS(verity, VERITY)
1883
1884EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION)
1885EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE)
1886EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER)
1887EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV)
1888EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG)
1889EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS)
1890EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT)
1891EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP)
1892EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG)
1893EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE)
1894EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA)
1895EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED)
1896EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR)
1897EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA)
1898EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT)
1899EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD)
1900
1901#define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1902#define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1903 EXT4_FEATURE_INCOMPAT_META_BG)
1904#define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1905 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1906 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1907
1908#define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1909#define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1910 EXT4_FEATURE_INCOMPAT_RECOVER| \
1911 EXT4_FEATURE_INCOMPAT_META_BG)
1912#define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1913 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1914 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1915
1916#define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1917#define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1918 EXT4_FEATURE_INCOMPAT_RECOVER| \
1919 EXT4_FEATURE_INCOMPAT_META_BG| \
1920 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1921 EXT4_FEATURE_INCOMPAT_64BIT| \
1922 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1923 EXT4_FEATURE_INCOMPAT_EA_INODE| \
1924 EXT4_FEATURE_INCOMPAT_MMP | \
1925 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
1926 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
1927 EXT4_FEATURE_INCOMPAT_CASEFOLD | \
1928 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \
1929 EXT4_FEATURE_INCOMPAT_LARGEDIR)
1930#define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1931 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1932 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1933 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1934 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1935 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1936 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1937 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1938 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1939 EXT4_FEATURE_RO_COMPAT_QUOTA |\
1940 EXT4_FEATURE_RO_COMPAT_PROJECT |\
1941 EXT4_FEATURE_RO_COMPAT_VERITY)
1942
1943#define EXTN_FEATURE_FUNCS(ver) \
1944static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
1945{ \
1946 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1947 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
1948} \
1949static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
1950{ \
1951 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1952 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
1953} \
1954static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
1955{ \
1956 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1957 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
1958}
1959
1960EXTN_FEATURE_FUNCS(2)
1961EXTN_FEATURE_FUNCS(3)
1962EXTN_FEATURE_FUNCS(4)
1963
1964static inline bool ext4_has_compat_features(struct super_block *sb)
1965{
1966 return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
1967}
1968static inline bool ext4_has_ro_compat_features(struct super_block *sb)
1969{
1970 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
1971}
1972static inline bool ext4_has_incompat_features(struct super_block *sb)
1973{
1974 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
1975}
1976
1977/*
1978 * Superblock flags
1979 */
1980#define EXT4_FLAGS_RESIZING 0
1981#define EXT4_FLAGS_SHUTDOWN 1
1982
1983static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi)
1984{
1985 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags);
1986}
1987
1988
1989/*
1990 * Default values for user and/or group using reserved blocks
1991 */
1992#define EXT4_DEF_RESUID 0
1993#define EXT4_DEF_RESGID 0
1994
1995/*
1996 * Default project ID
1997 */
1998#define EXT4_DEF_PROJID 0
1999
2000#define EXT4_DEF_INODE_READAHEAD_BLKS 32
2001
2002/*
2003 * Default mount options
2004 */
2005#define EXT4_DEFM_DEBUG 0x0001
2006#define EXT4_DEFM_BSDGROUPS 0x0002
2007#define EXT4_DEFM_XATTR_USER 0x0004
2008#define EXT4_DEFM_ACL 0x0008
2009#define EXT4_DEFM_UID16 0x0010
2010#define EXT4_DEFM_JMODE 0x0060
2011#define EXT4_DEFM_JMODE_DATA 0x0020
2012#define EXT4_DEFM_JMODE_ORDERED 0x0040
2013#define EXT4_DEFM_JMODE_WBACK 0x0060
2014#define EXT4_DEFM_NOBARRIER 0x0100
2015#define EXT4_DEFM_BLOCK_VALIDITY 0x0200
2016#define EXT4_DEFM_DISCARD 0x0400
2017#define EXT4_DEFM_NODELALLOC 0x0800
2018
2019/*
2020 * Default journal batch times
2021 */
2022#define EXT4_DEF_MIN_BATCH_TIME 0
2023#define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
2024
2025/*
2026 * Minimum number of groups in a flexgroup before we separate out
2027 * directories into the first block group of a flexgroup
2028 */
2029#define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
2030
2031/*
2032 * Structure of a directory entry
2033 */
2034#define EXT4_NAME_LEN 255
2035
2036struct ext4_dir_entry {
2037 __le32 inode; /* Inode number */
2038 __le16 rec_len; /* Directory entry length */
2039 __le16 name_len; /* Name length */
2040 char name[EXT4_NAME_LEN]; /* File name */
2041};
2042
2043/*
2044 * The new version of the directory entry. Since EXT4 structures are
2045 * stored in intel byte order, and the name_len field could never be
2046 * bigger than 255 chars, it's safe to reclaim the extra byte for the
2047 * file_type field.
2048 */
2049struct ext4_dir_entry_2 {
2050 __le32 inode; /* Inode number */
2051 __le16 rec_len; /* Directory entry length */
2052 __u8 name_len; /* Name length */
2053 __u8 file_type;
2054 char name[EXT4_NAME_LEN]; /* File name */
2055};
2056
2057/*
2058 * This is a bogus directory entry at the end of each leaf block that
2059 * records checksums.
2060 */
2061struct ext4_dir_entry_tail {
2062 __le32 det_reserved_zero1; /* Pretend to be unused */
2063 __le16 det_rec_len; /* 12 */
2064 __u8 det_reserved_zero2; /* Zero name length */
2065 __u8 det_reserved_ft; /* 0xDE, fake file type */
2066 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
2067};
2068
2069#define EXT4_DIRENT_TAIL(block, blocksize) \
2070 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
2071 ((blocksize) - \
2072 sizeof(struct ext4_dir_entry_tail))))
2073
2074/*
2075 * Ext4 directory file types. Only the low 3 bits are used. The
2076 * other bits are reserved for now.
2077 */
2078#define EXT4_FT_UNKNOWN 0
2079#define EXT4_FT_REG_FILE 1
2080#define EXT4_FT_DIR 2
2081#define EXT4_FT_CHRDEV 3
2082#define EXT4_FT_BLKDEV 4
2083#define EXT4_FT_FIFO 5
2084#define EXT4_FT_SOCK 6
2085#define EXT4_FT_SYMLINK 7
2086
2087#define EXT4_FT_MAX 8
2088
2089#define EXT4_FT_DIR_CSUM 0xDE
2090
2091/*
2092 * EXT4_DIR_PAD defines the directory entries boundaries
2093 *
2094 * NOTE: It must be a multiple of 4
2095 */
2096#define EXT4_DIR_PAD 4
2097#define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
2098#define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
2099 ~EXT4_DIR_ROUND)
2100#define EXT4_MAX_REC_LEN ((1<<16)-1)
2101
2102/*
2103 * If we ever get support for fs block sizes > page_size, we'll need
2104 * to remove the #if statements in the next two functions...
2105 */
2106static inline unsigned int
2107ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
2108{
2109 unsigned len = le16_to_cpu(dlen);
2110
2111#if (PAGE_SIZE >= 65536)
2112 if (len == EXT4_MAX_REC_LEN || len == 0)
2113 return blocksize;
2114 return (len & 65532) | ((len & 3) << 16);
2115#else
2116 return len;
2117#endif
2118}
2119
2120static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
2121{
2122 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
2123 BUG();
2124#if (PAGE_SIZE >= 65536)
2125 if (len < 65536)
2126 return cpu_to_le16(len);
2127 if (len == blocksize) {
2128 if (blocksize == 65536)
2129 return cpu_to_le16(EXT4_MAX_REC_LEN);
2130 else
2131 return cpu_to_le16(0);
2132 }
2133 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
2134#else
2135 return cpu_to_le16(len);
2136#endif
2137}
2138
2139/*
2140 * Hash Tree Directory indexing
2141 * (c) Daniel Phillips, 2001
2142 */
2143
2144#define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
2145 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2146#define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \
2147 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir)))
2148#define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2149
2150/* Legal values for the dx_root hash_version field: */
2151
2152#define DX_HASH_LEGACY 0
2153#define DX_HASH_HALF_MD4 1
2154#define DX_HASH_TEA 2
2155#define DX_HASH_LEGACY_UNSIGNED 3
2156#define DX_HASH_HALF_MD4_UNSIGNED 4
2157#define DX_HASH_TEA_UNSIGNED 5
2158
2159static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2160 const void *address, unsigned int length)
2161{
2162 struct {
2163 struct shash_desc shash;
2164 char ctx[4];
2165 } desc;
2166
2167 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2168
2169 desc.shash.tfm = sbi->s_chksum_driver;
2170 *(u32 *)desc.ctx = crc;
2171
2172 BUG_ON(crypto_shash_update(&desc.shash, address, length));
2173
2174 return *(u32 *)desc.ctx;
2175}
2176
2177#ifdef __KERNEL__
2178
2179/* hash info structure used by the directory hash */
2180struct dx_hash_info
2181{
2182 u32 hash;
2183 u32 minor_hash;
2184 int hash_version;
2185 u32 *seed;
2186};
2187
2188
2189/* 32 and 64 bit signed EOF for dx directories */
2190#define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
2191#define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
2192
2193
2194/*
2195 * Control parameters used by ext4_htree_next_block
2196 */
2197#define HASH_NB_ALWAYS 1
2198
2199struct ext4_filename {
2200 const struct qstr *usr_fname;
2201 struct fscrypt_str disk_name;
2202 struct dx_hash_info hinfo;
2203#ifdef CONFIG_FS_ENCRYPTION
2204 struct fscrypt_str crypto_buf;
2205#endif
2206#ifdef CONFIG_UNICODE
2207 struct fscrypt_str cf_name;
2208#endif
2209};
2210
2211#define fname_name(p) ((p)->disk_name.name)
2212#define fname_len(p) ((p)->disk_name.len)
2213
2214/*
2215 * Describe an inode's exact location on disk and in memory
2216 */
2217struct ext4_iloc
2218{
2219 struct buffer_head *bh;
2220 unsigned long offset;
2221 ext4_group_t block_group;
2222};
2223
2224static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2225{
2226 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2227}
2228
2229static inline bool ext4_is_quota_file(struct inode *inode)
2230{
2231 return IS_NOQUOTA(inode) &&
2232 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL);
2233}
2234
2235/*
2236 * This structure is stuffed into the struct file's private_data field
2237 * for directories. It is where we put information so that we can do
2238 * readdir operations in hash tree order.
2239 */
2240struct dir_private_info {
2241 struct rb_root root;
2242 struct rb_node *curr_node;
2243 struct fname *extra_fname;
2244 loff_t last_pos;
2245 __u32 curr_hash;
2246 __u32 curr_minor_hash;
2247 __u32 next_hash;
2248};
2249
2250/* calculate the first block number of the group */
2251static inline ext4_fsblk_t
2252ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2253{
2254 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2255 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2256}
2257
2258/*
2259 * Special error return code only used by dx_probe() and its callers.
2260 */
2261#define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
2262
2263/* htree levels for ext4 */
2264#define EXT4_HTREE_LEVEL_COMPAT 2
2265#define EXT4_HTREE_LEVEL 3
2266
2267static inline int ext4_dir_htree_level(struct super_block *sb)
2268{
2269 return ext4_has_feature_largedir(sb) ?
2270 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT;
2271}
2272
2273/*
2274 * Timeout and state flag for lazy initialization inode thread.
2275 */
2276#define EXT4_DEF_LI_WAIT_MULT 10
2277#define EXT4_DEF_LI_MAX_START_DELAY 5
2278#define EXT4_LAZYINIT_QUIT 0x0001
2279#define EXT4_LAZYINIT_RUNNING 0x0002
2280
2281/*
2282 * Lazy inode table initialization info
2283 */
2284struct ext4_lazy_init {
2285 unsigned long li_state;
2286 struct list_head li_request_list;
2287 struct mutex li_list_mtx;
2288};
2289
2290struct ext4_li_request {
2291 struct super_block *lr_super;
2292 struct ext4_sb_info *lr_sbi;
2293 ext4_group_t lr_next_group;
2294 struct list_head lr_request;
2295 unsigned long lr_next_sched;
2296 unsigned long lr_timeout;
2297};
2298
2299struct ext4_features {
2300 struct kobject f_kobj;
2301 struct completion f_kobj_unregister;
2302};
2303
2304/*
2305 * This structure will be used for multiple mount protection. It will be
2306 * written into the block number saved in the s_mmp_block field in the
2307 * superblock. Programs that check MMP should assume that if
2308 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2309 * to use the filesystem, regardless of how old the timestamp is.
2310 */
2311#define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
2312#define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2313#define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
2314#define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
2315
2316struct mmp_struct {
2317 __le32 mmp_magic; /* Magic number for MMP */
2318 __le32 mmp_seq; /* Sequence no. updated periodically */
2319
2320 /*
2321 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2322 * purposes and do not affect the correctness of the algorithm
2323 */
2324 __le64 mmp_time; /* Time last updated */
2325 char mmp_nodename[64]; /* Node which last updated MMP block */
2326 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
2327
2328 /*
2329 * mmp_check_interval is used to verify if the MMP block has been
2330 * updated on the block device. The value is updated based on the
2331 * maximum time to write the MMP block during an update cycle.
2332 */
2333 __le16 mmp_check_interval;
2334
2335 __le16 mmp_pad1;
2336 __le32 mmp_pad2[226];
2337 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
2338};
2339
2340/* arguments passed to the mmp thread */
2341struct mmpd_data {
2342 struct buffer_head *bh; /* bh from initial read_mmp_block() */
2343 struct super_block *sb; /* super block of the fs */
2344};
2345
2346/*
2347 * Check interval multiplier
2348 * The MMP block is written every update interval and initially checked every
2349 * update interval x the multiplier (the value is then adapted based on the
2350 * write latency). The reason is that writes can be delayed under load and we
2351 * don't want readers to incorrectly assume that the filesystem is no longer
2352 * in use.
2353 */
2354#define EXT4_MMP_CHECK_MULT 2UL
2355
2356/*
2357 * Minimum interval for MMP checking in seconds.
2358 */
2359#define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
2360
2361/*
2362 * Maximum interval for MMP checking in seconds.
2363 */
2364#define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
2365
2366/*
2367 * Function prototypes
2368 */
2369
2370/*
2371 * Ok, these declarations are also in <linux/kernel.h> but none of the
2372 * ext4 source programs needs to include it so they are duplicated here.
2373 */
2374# define NORET_TYPE /**/
2375# define ATTRIB_NORET __attribute__((noreturn))
2376# define NORET_AND noreturn,
2377
2378/* bitmap.c */
2379extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2380void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2381 struct ext4_group_desc *gdp,
2382 struct buffer_head *bh, int sz);
2383int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2384 struct ext4_group_desc *gdp,
2385 struct buffer_head *bh, int sz);
2386void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2387 struct ext4_group_desc *gdp,
2388 struct buffer_head *bh);
2389int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2390 struct ext4_group_desc *gdp,
2391 struct buffer_head *bh);
2392
2393/* balloc.c */
2394extern void ext4_get_group_no_and_offset(struct super_block *sb,
2395 ext4_fsblk_t blocknr,
2396 ext4_group_t *blockgrpp,
2397 ext4_grpblk_t *offsetp);
2398extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2399 ext4_fsblk_t block);
2400
2401extern unsigned int ext4_block_group(struct super_block *sb,
2402 ext4_fsblk_t blocknr);
2403extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2404 ext4_fsblk_t blocknr);
2405extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2406extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2407 ext4_group_t group);
2408extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2409 ext4_fsblk_t goal,
2410 unsigned int flags,
2411 unsigned long *count,
2412 int *errp);
2413extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2414 s64 nclusters, unsigned int flags);
2415extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2416extern void ext4_check_blocks_bitmap(struct super_block *);
2417extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2418 ext4_group_t block_group,
2419 struct buffer_head ** bh);
2420extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2421
2422extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2423 ext4_group_t block_group);
2424extern int ext4_wait_block_bitmap(struct super_block *sb,
2425 ext4_group_t block_group,
2426 struct buffer_head *bh);
2427extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2428 ext4_group_t block_group);
2429extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2430 ext4_group_t block_group,
2431 struct ext4_group_desc *gdp);
2432ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2433
2434#ifdef CONFIG_UNICODE
2435extern void ext4_fname_setup_ci_filename(struct inode *dir,
2436 const struct qstr *iname,
2437 struct fscrypt_str *fname);
2438#endif
2439
2440#ifdef CONFIG_FS_ENCRYPTION
2441static inline void ext4_fname_from_fscrypt_name(struct ext4_filename *dst,
2442 const struct fscrypt_name *src)
2443{
2444 memset(dst, 0, sizeof(*dst));
2445
2446 dst->usr_fname = src->usr_fname;
2447 dst->disk_name = src->disk_name;
2448 dst->hinfo.hash = src->hash;
2449 dst->hinfo.minor_hash = src->minor_hash;
2450 dst->crypto_buf = src->crypto_buf;
2451}
2452
2453static inline int ext4_fname_setup_filename(struct inode *dir,
2454 const struct qstr *iname,
2455 int lookup,
2456 struct ext4_filename *fname)
2457{
2458 struct fscrypt_name name;
2459 int err;
2460
2461 err = fscrypt_setup_filename(dir, iname, lookup, &name);
2462 if (err)
2463 return err;
2464
2465 ext4_fname_from_fscrypt_name(fname, &name);
2466
2467#ifdef CONFIG_UNICODE
2468 ext4_fname_setup_ci_filename(dir, iname, &fname->cf_name);
2469#endif
2470 return 0;
2471}
2472
2473static inline int ext4_fname_prepare_lookup(struct inode *dir,
2474 struct dentry *dentry,
2475 struct ext4_filename *fname)
2476{
2477 struct fscrypt_name name;
2478 int err;
2479
2480 err = fscrypt_prepare_lookup(dir, dentry, &name);
2481 if (err)
2482 return err;
2483
2484 ext4_fname_from_fscrypt_name(fname, &name);
2485
2486#ifdef CONFIG_UNICODE
2487 ext4_fname_setup_ci_filename(dir, &dentry->d_name, &fname->cf_name);
2488#endif
2489 return 0;
2490}
2491
2492static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2493{
2494 struct fscrypt_name name;
2495
2496 name.crypto_buf = fname->crypto_buf;
2497 fscrypt_free_filename(&name);
2498
2499 fname->crypto_buf.name = NULL;
2500 fname->usr_fname = NULL;
2501 fname->disk_name.name = NULL;
2502
2503#ifdef CONFIG_UNICODE
2504 kfree(fname->cf_name.name);
2505 fname->cf_name.name = NULL;
2506#endif
2507}
2508#else /* !CONFIG_FS_ENCRYPTION */
2509static inline int ext4_fname_setup_filename(struct inode *dir,
2510 const struct qstr *iname,
2511 int lookup,
2512 struct ext4_filename *fname)
2513{
2514 fname->usr_fname = iname;
2515 fname->disk_name.name = (unsigned char *) iname->name;
2516 fname->disk_name.len = iname->len;
2517
2518#ifdef CONFIG_UNICODE
2519 ext4_fname_setup_ci_filename(dir, iname, &fname->cf_name);
2520#endif
2521
2522 return 0;
2523}
2524
2525static inline int ext4_fname_prepare_lookup(struct inode *dir,
2526 struct dentry *dentry,
2527 struct ext4_filename *fname)
2528{
2529 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname);
2530}
2531
2532static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2533{
2534#ifdef CONFIG_UNICODE
2535 kfree(fname->cf_name.name);
2536 fname->cf_name.name = NULL;
2537#endif
2538}
2539#endif /* !CONFIG_FS_ENCRYPTION */
2540
2541/* dir.c */
2542extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2543 struct file *,
2544 struct ext4_dir_entry_2 *,
2545 struct buffer_head *, char *, int,
2546 unsigned int);
2547#define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2548 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2549 (de), (bh), (buf), (size), (offset)))
2550extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2551 __u32 minor_hash,
2552 struct ext4_dir_entry_2 *dirent,
2553 struct fscrypt_str *ent_name);
2554extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2555extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2556 struct buffer_head *bh,
2557 void *buf, int buf_size,
2558 struct ext4_filename *fname,
2559 struct ext4_dir_entry_2 **dest_de);
2560void ext4_insert_dentry(struct inode *inode,
2561 struct ext4_dir_entry_2 *de,
2562 int buf_size,
2563 struct ext4_filename *fname);
2564static inline void ext4_update_dx_flag(struct inode *inode)
2565{
2566 if (!ext4_has_feature_dir_index(inode->i_sb)) {
2567 /* ext4_iget() should have caught this... */
2568 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb));
2569 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2570 }
2571}
2572static const unsigned char ext4_filetype_table[] = {
2573 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2574};
2575
2576static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2577{
2578 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2579 return DT_UNKNOWN;
2580
2581 return ext4_filetype_table[filetype];
2582}
2583extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2584 void *buf, int buf_size);
2585
2586/* fsync.c */
2587extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2588
2589/* hash.c */
2590extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len,
2591 struct dx_hash_info *hinfo);
2592
2593/* ialloc.c */
2594extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2595 const struct qstr *qstr, __u32 goal,
2596 uid_t *owner, __u32 i_flags,
2597 int handle_type, unsigned int line_no,
2598 int nblocks);
2599
2600#define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \
2601 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2602 i_flags, 0, 0, 0)
2603#define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2604 type, nblocks) \
2605 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2606 0, (type), __LINE__, (nblocks))
2607
2608
2609extern void ext4_free_inode(handle_t *, struct inode *);
2610extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2611extern unsigned long ext4_count_free_inodes(struct super_block *);
2612extern unsigned long ext4_count_dirs(struct super_block *);
2613extern void ext4_check_inodes_bitmap(struct super_block *);
2614extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2615extern int ext4_init_inode_table(struct super_block *sb,
2616 ext4_group_t group, int barrier);
2617extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2618
2619/* mballoc.c */
2620extern const struct seq_operations ext4_mb_seq_groups_ops;
2621extern long ext4_mb_stats;
2622extern long ext4_mb_max_to_scan;
2623extern int ext4_mb_init(struct super_block *);
2624extern int ext4_mb_release(struct super_block *);
2625extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2626 struct ext4_allocation_request *, int *);
2627extern int ext4_mb_reserve_blocks(struct super_block *, int);
2628extern void ext4_discard_preallocations(struct inode *);
2629extern int __init ext4_init_mballoc(void);
2630extern void ext4_exit_mballoc(void);
2631extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2632 struct buffer_head *bh, ext4_fsblk_t block,
2633 unsigned long count, int flags);
2634extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2635 ext4_group_t ngroups);
2636extern int ext4_mb_add_groupinfo(struct super_block *sb,
2637 ext4_group_t i, struct ext4_group_desc *desc);
2638extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2639 ext4_fsblk_t block, unsigned long count);
2640extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2641extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid);
2642
2643/* inode.c */
2644int ext4_inode_is_fast_symlink(struct inode *inode);
2645struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2646struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2647int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count,
2648 bool wait, struct buffer_head **bhs);
2649int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2650 struct buffer_head *bh_result, int create);
2651int ext4_get_block(struct inode *inode, sector_t iblock,
2652 struct buffer_head *bh_result, int create);
2653int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2654 struct buffer_head *bh, int create);
2655int ext4_walk_page_buffers(handle_t *handle,
2656 struct buffer_head *head,
2657 unsigned from,
2658 unsigned to,
2659 int *partial,
2660 int (*fn)(handle_t *handle,
2661 struct buffer_head *bh));
2662int do_journal_get_write_access(handle_t *handle,
2663 struct buffer_head *bh);
2664#define FALL_BACK_TO_NONDELALLOC 1
2665#define CONVERT_INLINE_DATA 2
2666
2667typedef enum {
2668 EXT4_IGET_NORMAL = 0,
2669 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */
2670 EXT4_IGET_HANDLE = 0x0002 /* Inode # is from a handle */
2671} ext4_iget_flags;
2672
2673extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
2674 ext4_iget_flags flags, const char *function,
2675 unsigned int line);
2676
2677#define ext4_iget(sb, ino, flags) \
2678 __ext4_iget((sb), (ino), (flags), __func__, __LINE__)
2679
2680extern int ext4_write_inode(struct inode *, struct writeback_control *);
2681extern int ext4_setattr(struct dentry *, struct iattr *);
2682extern int ext4_getattr(const struct path *, struct kstat *, u32, unsigned int);
2683extern void ext4_evict_inode(struct inode *);
2684extern void ext4_clear_inode(struct inode *);
2685extern int ext4_file_getattr(const struct path *, struct kstat *, u32, unsigned int);
2686extern int ext4_sync_inode(handle_t *, struct inode *);
2687extern void ext4_dirty_inode(struct inode *, int);
2688extern int ext4_change_inode_journal_flag(struct inode *, int);
2689extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2690extern int ext4_inode_attach_jinode(struct inode *inode);
2691extern int ext4_can_truncate(struct inode *inode);
2692extern int ext4_truncate(struct inode *);
2693extern int ext4_break_layouts(struct inode *);
2694extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2695extern void ext4_set_inode_flags(struct inode *);
2696extern int ext4_alloc_da_blocks(struct inode *inode);
2697extern void ext4_set_aops(struct inode *inode);
2698extern int ext4_writepage_trans_blocks(struct inode *);
2699extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2700extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2701 loff_t lstart, loff_t lend);
2702extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf);
2703extern vm_fault_t ext4_filemap_fault(struct vm_fault *vmf);
2704extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2705extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
2706extern void ext4_da_release_space(struct inode *inode, int to_free);
2707extern void ext4_da_update_reserve_space(struct inode *inode,
2708 int used, int quota_claim);
2709extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
2710 ext4_fsblk_t pblk, ext4_lblk_t len);
2711
2712/* indirect.c */
2713extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2714 struct ext4_map_blocks *map, int flags);
2715extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2716extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2717extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2718 ext4_lblk_t start, ext4_lblk_t end);
2719
2720/* ioctl.c */
2721extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2722extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2723
2724/* migrate.c */
2725extern int ext4_ext_migrate(struct inode *);
2726extern int ext4_ind_migrate(struct inode *inode);
2727
2728/* namei.c */
2729extern int ext4_dirblock_csum_verify(struct inode *inode,
2730 struct buffer_head *bh);
2731extern int ext4_orphan_add(handle_t *, struct inode *);
2732extern int ext4_orphan_del(handle_t *, struct inode *);
2733extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2734 __u32 start_minor_hash, __u32 *next_hash);
2735extern int ext4_search_dir(struct buffer_head *bh,
2736 char *search_buf,
2737 int buf_size,
2738 struct inode *dir,
2739 struct ext4_filename *fname,
2740 unsigned int offset,
2741 struct ext4_dir_entry_2 **res_dir);
2742extern int ext4_generic_delete_entry(handle_t *handle,
2743 struct inode *dir,
2744 struct ext4_dir_entry_2 *de_del,
2745 struct buffer_head *bh,
2746 void *entry_buf,
2747 int buf_size,
2748 int csum_size);
2749extern bool ext4_empty_dir(struct inode *inode);
2750
2751/* resize.c */
2752extern void ext4_kvfree_array_rcu(void *to_free);
2753extern int ext4_group_add(struct super_block *sb,
2754 struct ext4_new_group_data *input);
2755extern int ext4_group_extend(struct super_block *sb,
2756 struct ext4_super_block *es,
2757 ext4_fsblk_t n_blocks_count);
2758extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2759
2760/* super.c */
2761extern struct buffer_head *ext4_sb_bread(struct super_block *sb,
2762 sector_t block, int op_flags);
2763extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
2764extern int ext4_calculate_overhead(struct super_block *sb);
2765extern void ext4_superblock_csum_set(struct super_block *sb);
2766extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2767 ext4_group_t ngroup);
2768extern const char *ext4_decode_error(struct super_block *sb, int errno,
2769 char nbuf[16]);
2770extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb,
2771 ext4_group_t block_group,
2772 unsigned int flags);
2773
2774extern __printf(6, 7)
2775void __ext4_error(struct super_block *, const char *, unsigned int, int, __u64,
2776 const char *, ...);
2777extern __printf(6, 7)
2778void __ext4_error_inode(struct inode *, const char *, unsigned int,
2779 ext4_fsblk_t, int, const char *, ...);
2780extern __printf(5, 6)
2781void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2782 const char *, ...);
2783extern void __ext4_std_error(struct super_block *, const char *,
2784 unsigned int, int);
2785extern __printf(5, 6)
2786void __ext4_abort(struct super_block *, const char *, unsigned int, int,
2787 const char *, ...);
2788extern __printf(4, 5)
2789void __ext4_warning(struct super_block *, const char *, unsigned int,
2790 const char *, ...);
2791extern __printf(4, 5)
2792void __ext4_warning_inode(const struct inode *inode, const char *function,
2793 unsigned int line, const char *fmt, ...);
2794extern __printf(3, 4)
2795void __ext4_msg(struct super_block *, const char *, const char *, ...);
2796extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2797 const char *, unsigned int, const char *);
2798extern __printf(7, 8)
2799void __ext4_grp_locked_error(const char *, unsigned int,
2800 struct super_block *, ext4_group_t,
2801 unsigned long, ext4_fsblk_t,
2802 const char *, ...);
2803
2804#define EXT4_ERROR_INODE(inode, fmt, a...) \
2805 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
2806
2807#define EXT4_ERROR_INODE_ERR(inode, err, fmt, a...) \
2808 __ext4_error_inode((inode), __func__, __LINE__, 0, (err), (fmt), ## a)
2809
2810#define ext4_error_inode_block(inode, block, err, fmt, a...) \
2811 __ext4_error_inode((inode), __func__, __LINE__, (block), (err), \
2812 (fmt), ## a)
2813
2814#define EXT4_ERROR_FILE(file, block, fmt, a...) \
2815 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
2816
2817#ifdef CONFIG_PRINTK
2818
2819#define ext4_error_inode(inode, func, line, block, fmt, ...) \
2820 __ext4_error_inode(inode, func, line, block, 0, fmt, ##__VA_ARGS__)
2821#define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \
2822 __ext4_error_inode((inode), (func), (line), (block), \
2823 (err), (fmt), ##__VA_ARGS__)
2824#define ext4_error_file(file, func, line, block, fmt, ...) \
2825 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2826#define ext4_error(sb, fmt, ...) \
2827 __ext4_error((sb), __func__, __LINE__, 0, 0, (fmt), ##__VA_ARGS__)
2828#define ext4_error_err(sb, err, fmt, ...) \
2829 __ext4_error((sb), __func__, __LINE__, (err), 0, (fmt), ##__VA_ARGS__)
2830#define ext4_abort(sb, err, fmt, ...) \
2831 __ext4_abort((sb), __func__, __LINE__, (err), (fmt), ##__VA_ARGS__)
2832#define ext4_warning(sb, fmt, ...) \
2833 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2834#define ext4_warning_inode(inode, fmt, ...) \
2835 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
2836#define ext4_msg(sb, level, fmt, ...) \
2837 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2838#define dump_mmp_msg(sb, mmp, msg) \
2839 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2840#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2841 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2842 fmt, ##__VA_ARGS__)
2843
2844#else
2845
2846#define ext4_error_inode(inode, func, line, block, fmt, ...) \
2847do { \
2848 no_printk(fmt, ##__VA_ARGS__); \
2849 __ext4_error_inode(inode, "", 0, block, 0, " "); \
2850} while (0)
2851#define ext4_error_inode_err(inode, func, line, block, err, fmt, ...) \
2852do { \
2853 no_printk(fmt, ##__VA_ARGS__); \
2854 __ext4_error_inode(inode, "", 0, block, err, " "); \
2855} while (0)
2856#define ext4_error_file(file, func, line, block, fmt, ...) \
2857do { \
2858 no_printk(fmt, ##__VA_ARGS__); \
2859 __ext4_error_file(file, "", 0, block, " "); \
2860} while (0)
2861#define ext4_error(sb, fmt, ...) \
2862do { \
2863 no_printk(fmt, ##__VA_ARGS__); \
2864 __ext4_error(sb, "", 0, 0, 0, " "); \
2865} while (0)
2866#define ext4_error_err(sb, err, fmt, ...) \
2867do { \
2868 no_printk(fmt, ##__VA_ARGS__); \
2869 __ext4_error(sb, "", 0, err, 0, " "); \
2870} while (0)
2871#define ext4_abort(sb, err, fmt, ...) \
2872do { \
2873 no_printk(fmt, ##__VA_ARGS__); \
2874 __ext4_abort(sb, "", 0, err, " "); \
2875} while (0)
2876#define ext4_warning(sb, fmt, ...) \
2877do { \
2878 no_printk(fmt, ##__VA_ARGS__); \
2879 __ext4_warning(sb, "", 0, " "); \
2880} while (0)
2881#define ext4_warning_inode(inode, fmt, ...) \
2882do { \
2883 no_printk(fmt, ##__VA_ARGS__); \
2884 __ext4_warning_inode(inode, "", 0, " "); \
2885} while (0)
2886#define ext4_msg(sb, level, fmt, ...) \
2887do { \
2888 no_printk(fmt, ##__VA_ARGS__); \
2889 __ext4_msg(sb, "", " "); \
2890} while (0)
2891#define dump_mmp_msg(sb, mmp, msg) \
2892 __dump_mmp_msg(sb, mmp, "", 0, "")
2893#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2894do { \
2895 no_printk(fmt, ##__VA_ARGS__); \
2896 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
2897} while (0)
2898
2899#endif
2900
2901extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2902 __u32 compat);
2903extern int ext4_update_rocompat_feature(handle_t *handle,
2904 struct super_block *sb, __u32 rocompat);
2905extern int ext4_update_incompat_feature(handle_t *handle,
2906 struct super_block *sb, __u32 incompat);
2907extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2908 struct ext4_group_desc *bg);
2909extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2910 struct ext4_group_desc *bg);
2911extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2912 struct ext4_group_desc *bg);
2913extern __u32 ext4_free_group_clusters(struct super_block *sb,
2914 struct ext4_group_desc *bg);
2915extern __u32 ext4_free_inodes_count(struct super_block *sb,
2916 struct ext4_group_desc *bg);
2917extern __u32 ext4_used_dirs_count(struct super_block *sb,
2918 struct ext4_group_desc *bg);
2919extern __u32 ext4_itable_unused_count(struct super_block *sb,
2920 struct ext4_group_desc *bg);
2921extern void ext4_block_bitmap_set(struct super_block *sb,
2922 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2923extern void ext4_inode_bitmap_set(struct super_block *sb,
2924 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2925extern void ext4_inode_table_set(struct super_block *sb,
2926 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2927extern void ext4_free_group_clusters_set(struct super_block *sb,
2928 struct ext4_group_desc *bg,
2929 __u32 count);
2930extern void ext4_free_inodes_set(struct super_block *sb,
2931 struct ext4_group_desc *bg, __u32 count);
2932extern void ext4_used_dirs_set(struct super_block *sb,
2933 struct ext4_group_desc *bg, __u32 count);
2934extern void ext4_itable_unused_set(struct super_block *sb,
2935 struct ext4_group_desc *bg, __u32 count);
2936extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2937 struct ext4_group_desc *gdp);
2938extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2939 struct ext4_group_desc *gdp);
2940extern int ext4_register_li_request(struct super_block *sb,
2941 ext4_group_t first_not_zeroed);
2942
2943static inline int ext4_has_metadata_csum(struct super_block *sb)
2944{
2945 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
2946 !EXT4_SB(sb)->s_chksum_driver);
2947
2948 return ext4_has_feature_metadata_csum(sb) &&
2949 (EXT4_SB(sb)->s_chksum_driver != NULL);
2950}
2951
2952static inline int ext4_has_group_desc_csum(struct super_block *sb)
2953{
2954 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb);
2955}
2956
2957static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2958{
2959 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2960 le32_to_cpu(es->s_blocks_count_lo);
2961}
2962
2963static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2964{
2965 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2966 le32_to_cpu(es->s_r_blocks_count_lo);
2967}
2968
2969static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2970{
2971 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2972 le32_to_cpu(es->s_free_blocks_count_lo);
2973}
2974
2975static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2976 ext4_fsblk_t blk)
2977{
2978 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2979 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2980}
2981
2982static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2983 ext4_fsblk_t blk)
2984{
2985 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2986 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2987}
2988
2989static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2990 ext4_fsblk_t blk)
2991{
2992 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2993 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2994}
2995
2996static inline loff_t ext4_isize(struct super_block *sb,
2997 struct ext4_inode *raw_inode)
2998{
2999 if (ext4_has_feature_largedir(sb) ||
3000 S_ISREG(le16_to_cpu(raw_inode->i_mode)))
3001 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
3002 le32_to_cpu(raw_inode->i_size_lo);
3003
3004 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
3005}
3006
3007static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
3008{
3009 raw_inode->i_size_lo = cpu_to_le32(i_size);
3010 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
3011}
3012
3013static inline
3014struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
3015 ext4_group_t group)
3016{
3017 struct ext4_group_info **grp_info;
3018 long indexv, indexh;
3019 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
3020 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
3021 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
3022 grp_info = sbi_array_rcu_deref(EXT4_SB(sb), s_group_info, indexv);
3023 return grp_info[indexh];
3024}
3025
3026/*
3027 * Reading s_groups_count requires using smp_rmb() afterwards. See
3028 * the locking protocol documented in the comments of ext4_group_add()
3029 * in resize.c
3030 */
3031static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
3032{
3033 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
3034
3035 smp_rmb();
3036 return ngroups;
3037}
3038
3039static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
3040 ext4_group_t block_group)
3041{
3042 return block_group >> sbi->s_log_groups_per_flex;
3043}
3044
3045static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
3046{
3047 return 1 << sbi->s_log_groups_per_flex;
3048}
3049
3050#define ext4_std_error(sb, errno) \
3051do { \
3052 if ((errno)) \
3053 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
3054} while (0)
3055
3056#ifdef CONFIG_SMP
3057/* Each CPU can accumulate percpu_counter_batch clusters in their local
3058 * counters. So we need to make sure we have free clusters more
3059 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
3060 */
3061#define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
3062#else
3063#define EXT4_FREECLUSTERS_WATERMARK 0
3064#endif
3065
3066/* Update i_disksize. Requires i_mutex to avoid races with truncate */
3067static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
3068{
3069 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
3070 !inode_is_locked(inode));
3071 down_write(&EXT4_I(inode)->i_data_sem);
3072 if (newsize > EXT4_I(inode)->i_disksize)
3073 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize);
3074 up_write(&EXT4_I(inode)->i_data_sem);
3075}
3076
3077/* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
3078static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
3079{
3080 int changed = 0;
3081
3082 if (newsize > inode->i_size) {
3083 i_size_write(inode, newsize);
3084 changed = 1;
3085 }
3086 if (newsize > EXT4_I(inode)->i_disksize) {
3087 ext4_update_i_disksize(inode, newsize);
3088 changed |= 2;
3089 }
3090 return changed;
3091}
3092
3093int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
3094 loff_t len);
3095
3096struct ext4_group_info {
3097 unsigned long bb_state;
3098 struct rb_root bb_free_root;
3099 ext4_grpblk_t bb_first_free; /* first free block */
3100 ext4_grpblk_t bb_free; /* total free blocks */
3101 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
3102 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
3103 struct list_head bb_prealloc_list;
3104#ifdef DOUBLE_CHECK
3105 void *bb_bitmap;
3106#endif
3107 struct rw_semaphore alloc_sem;
3108 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
3109 * regions, index is order.
3110 * bb_counters[3] = 5 means
3111 * 5 free 8-block regions. */
3112};
3113
3114#define EXT4_GROUP_INFO_NEED_INIT_BIT 0
3115#define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
3116#define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
3117#define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
3118#define EXT4_GROUP_INFO_BBITMAP_CORRUPT \
3119 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT)
3120#define EXT4_GROUP_INFO_IBITMAP_CORRUPT \
3121 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT)
3122
3123#define EXT4_MB_GRP_NEED_INIT(grp) \
3124 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
3125#define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
3126 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3127#define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
3128 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3129
3130#define EXT4_MB_GRP_WAS_TRIMMED(grp) \
3131 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3132#define EXT4_MB_GRP_SET_TRIMMED(grp) \
3133 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3134#define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
3135 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3136
3137#define EXT4_MAX_CONTENTION 8
3138#define EXT4_CONTENTION_THRESHOLD 2
3139
3140static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
3141 ext4_group_t group)
3142{
3143 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
3144}
3145
3146/*
3147 * Returns true if the filesystem is busy enough that attempts to
3148 * access the block group locks has run into contention.
3149 */
3150static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
3151{
3152 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
3153}
3154
3155static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
3156{
3157 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
3158 if (spin_trylock(lock))
3159 /*
3160 * We're able to grab the lock right away, so drop the
3161 * lock contention counter.
3162 */
3163 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
3164 else {
3165 /*
3166 * The lock is busy, so bump the contention counter,
3167 * and then wait on the spin lock.
3168 */
3169 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
3170 EXT4_MAX_CONTENTION);
3171 spin_lock(lock);
3172 }
3173}
3174
3175static inline void ext4_unlock_group(struct super_block *sb,
3176 ext4_group_t group)
3177{
3178 spin_unlock(ext4_group_lock_ptr(sb, group));
3179}
3180
3181/*
3182 * Block validity checking
3183 */
3184#define ext4_check_indirect_blockref(inode, bh) \
3185 ext4_check_blockref(__func__, __LINE__, inode, \
3186 (__le32 *)(bh)->b_data, \
3187 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
3188
3189#define ext4_ind_check_inode(inode) \
3190 ext4_check_blockref(__func__, __LINE__, inode, \
3191 EXT4_I(inode)->i_data, \
3192 EXT4_NDIR_BLOCKS)
3193
3194/*
3195 * Inodes and files operations
3196 */
3197
3198/* dir.c */
3199extern const struct file_operations ext4_dir_operations;
3200
3201#ifdef CONFIG_UNICODE
3202extern const struct dentry_operations ext4_dentry_ops;
3203#endif
3204
3205/* file.c */
3206extern const struct inode_operations ext4_file_inode_operations;
3207extern const struct file_operations ext4_file_operations;
3208extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
3209
3210/* inline.c */
3211extern int ext4_get_max_inline_size(struct inode *inode);
3212extern int ext4_find_inline_data_nolock(struct inode *inode);
3213extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
3214 unsigned int len);
3215extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
3216
3217extern int ext4_readpage_inline(struct inode *inode, struct page *page);
3218extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3219 struct inode *inode,
3220 loff_t pos, unsigned len,
3221 unsigned flags,
3222 struct page **pagep);
3223extern int ext4_write_inline_data_end(struct inode *inode,
3224 loff_t pos, unsigned len,
3225 unsigned copied,
3226 struct page *page);
3227extern struct buffer_head *
3228ext4_journalled_write_inline_data(struct inode *inode,
3229 unsigned len,
3230 struct page *page);
3231extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
3232 struct inode *inode,
3233 loff_t pos, unsigned len,
3234 unsigned flags,
3235 struct page **pagep,
3236 void **fsdata);
3237extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
3238 unsigned len, unsigned copied,
3239 struct page *page);
3240extern int ext4_try_add_inline_entry(handle_t *handle,
3241 struct ext4_filename *fname,
3242 struct inode *dir, struct inode *inode);
3243extern int ext4_try_create_inline_dir(handle_t *handle,
3244 struct inode *parent,
3245 struct inode *inode);
3246extern int ext4_read_inline_dir(struct file *filp,
3247 struct dir_context *ctx,
3248 int *has_inline_data);
3249extern int ext4_inlinedir_to_tree(struct file *dir_file,
3250 struct inode *dir, ext4_lblk_t block,
3251 struct dx_hash_info *hinfo,
3252 __u32 start_hash, __u32 start_minor_hash,
3253 int *has_inline_data);
3254extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3255 struct ext4_filename *fname,
3256 struct ext4_dir_entry_2 **res_dir,
3257 int *has_inline_data);
3258extern int ext4_delete_inline_entry(handle_t *handle,
3259 struct inode *dir,
3260 struct ext4_dir_entry_2 *de_del,
3261 struct buffer_head *bh,
3262 int *has_inline_data);
3263extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3264extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3265 struct ext4_dir_entry_2 **parent_de,
3266 int *retval);
3267extern int ext4_inline_data_fiemap(struct inode *inode,
3268 struct fiemap_extent_info *fieinfo,
3269 int *has_inline, __u64 start, __u64 len);
3270
3271struct iomap;
3272extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap);
3273
3274extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3275
3276extern int ext4_convert_inline_data(struct inode *inode);
3277
3278static inline int ext4_has_inline_data(struct inode *inode)
3279{
3280 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3281 EXT4_I(inode)->i_inline_off;
3282}
3283
3284/* namei.c */
3285extern const struct inode_operations ext4_dir_inode_operations;
3286extern const struct inode_operations ext4_special_inode_operations;
3287extern struct dentry *ext4_get_parent(struct dentry *child);
3288extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3289 struct ext4_dir_entry_2 *de,
3290 int blocksize, int csum_size,
3291 unsigned int parent_ino, int dotdot_real_len);
3292extern void ext4_initialize_dirent_tail(struct buffer_head *bh,
3293 unsigned int blocksize);
3294extern int ext4_handle_dirty_dirblock(handle_t *handle, struct inode *inode,
3295 struct buffer_head *bh);
3296extern int ext4_ci_compare(const struct inode *parent,
3297 const struct qstr *fname,
3298 const struct qstr *entry, bool quick);
3299
3300#define S_SHIFT 12
3301static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = {
3302 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
3303 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
3304 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
3305 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
3306 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
3307 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
3308 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
3309};
3310
3311static inline void ext4_set_de_type(struct super_block *sb,
3312 struct ext4_dir_entry_2 *de,
3313 umode_t mode) {
3314 if (ext4_has_feature_filetype(sb))
3315 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3316}
3317
3318/* readpages.c */
3319extern int ext4_mpage_readpages(struct address_space *mapping,
3320 struct list_head *pages, struct page *page,
3321 unsigned nr_pages, bool is_readahead);
3322extern int __init ext4_init_post_read_processing(void);
3323extern void ext4_exit_post_read_processing(void);
3324
3325/* symlink.c */
3326extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3327extern const struct inode_operations ext4_symlink_inode_operations;
3328extern const struct inode_operations ext4_fast_symlink_inode_operations;
3329
3330/* sysfs.c */
3331extern int ext4_register_sysfs(struct super_block *sb);
3332extern void ext4_unregister_sysfs(struct super_block *sb);
3333extern int __init ext4_init_sysfs(void);
3334extern void ext4_exit_sysfs(void);
3335
3336/* block_validity */
3337extern void ext4_release_system_zone(struct super_block *sb);
3338extern int ext4_setup_system_zone(struct super_block *sb);
3339extern int __init ext4_init_system_zone(void);
3340extern void ext4_exit_system_zone(void);
3341extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
3342 ext4_fsblk_t start_blk,
3343 unsigned int count);
3344extern int ext4_check_blockref(const char *, unsigned int,
3345 struct inode *, __le32 *, unsigned int);
3346
3347/* extents.c */
3348struct ext4_ext_path;
3349struct ext4_extent;
3350
3351/*
3352 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3353 * __le32.
3354 */
3355#define EXT_MAX_BLOCKS 0xffffffff
3356
3357extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
3358extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3359extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3360 struct ext4_map_blocks *map, int flags);
3361extern int ext4_ext_truncate(handle_t *, struct inode *);
3362extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3363 ext4_lblk_t end);
3364extern void ext4_ext_init(struct super_block *);
3365extern void ext4_ext_release(struct super_block *);
3366extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3367 loff_t len);
3368extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3369 loff_t offset, ssize_t len);
3370extern int ext4_convert_unwritten_io_end_vec(handle_t *handle,
3371 ext4_io_end_t *io_end);
3372extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3373 struct ext4_map_blocks *map, int flags);
3374extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3375 int num,
3376 struct ext4_ext_path *path);
3377extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3378 struct ext4_ext_path **,
3379 struct ext4_extent *, int);
3380extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3381 struct ext4_ext_path **,
3382 int flags);
3383extern void ext4_ext_drop_refs(struct ext4_ext_path *);
3384extern int ext4_ext_check_inode(struct inode *inode);
3385extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3386extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3387 __u64 start, __u64 len);
3388extern int ext4_get_es_cache(struct inode *inode,
3389 struct fiemap_extent_info *fieinfo,
3390 __u64 start, __u64 len);
3391extern int ext4_ext_precache(struct inode *inode);
3392extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3393 struct inode *inode2, ext4_lblk_t lblk1,
3394 ext4_lblk_t lblk2, ext4_lblk_t count,
3395 int mark_unwritten,int *err);
3396extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu);
3397extern int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode,
3398 int check_cred, int restart_cred,
3399 int revoke_cred);
3400
3401
3402/* move_extent.c */
3403extern void ext4_double_down_write_data_sem(struct inode *first,
3404 struct inode *second);
3405extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3406 struct inode *donor_inode);
3407extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3408 __u64 start_orig, __u64 start_donor,
3409 __u64 len, __u64 *moved_len);
3410
3411/* page-io.c */
3412extern int __init ext4_init_pageio(void);
3413extern void ext4_exit_pageio(void);
3414extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3415extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3416extern int ext4_put_io_end(ext4_io_end_t *io_end);
3417extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3418extern void ext4_io_submit_init(struct ext4_io_submit *io,
3419 struct writeback_control *wbc);
3420extern void ext4_end_io_rsv_work(struct work_struct *work);
3421extern void ext4_io_submit(struct ext4_io_submit *io);
3422extern int ext4_bio_write_page(struct ext4_io_submit *io,
3423 struct page *page,
3424 int len,
3425 struct writeback_control *wbc,
3426 bool keep_towrite);
3427extern struct ext4_io_end_vec *ext4_alloc_io_end_vec(ext4_io_end_t *io_end);
3428extern struct ext4_io_end_vec *ext4_last_io_end_vec(ext4_io_end_t *io_end);
3429
3430/* mmp.c */
3431extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3432
3433/* verity.c */
3434extern const struct fsverity_operations ext4_verityops;
3435
3436/*
3437 * Add new method to test whether block and inode bitmaps are properly
3438 * initialized. With uninit_bg reading the block from disk is not enough
3439 * to mark the bitmap uptodate. We need to also zero-out the bitmap
3440 */
3441#define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3442
3443static inline int bitmap_uptodate(struct buffer_head *bh)
3444{
3445 return (buffer_uptodate(bh) &&
3446 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3447}
3448static inline void set_bitmap_uptodate(struct buffer_head *bh)
3449{
3450 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3451}
3452
3453#define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
3454
3455/* For ioend & aio unwritten conversion wait queues */
3456#define EXT4_WQ_HASH_SZ 37
3457#define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
3458 EXT4_WQ_HASH_SZ])
3459extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3460
3461extern int ext4_resize_begin(struct super_block *sb);
3462extern void ext4_resize_end(struct super_block *sb);
3463
3464static inline void ext4_set_io_unwritten_flag(struct inode *inode,
3465 struct ext4_io_end *io_end)
3466{
3467 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
3468 io_end->flag |= EXT4_IO_END_UNWRITTEN;
3469 atomic_inc(&EXT4_I(inode)->i_unwritten);
3470 }
3471}
3472
3473static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3474{
3475 struct inode *inode = io_end->inode;
3476
3477 if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
3478 io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3479 /* Wake up anyone waiting on unwritten extent conversion */
3480 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
3481 wake_up_all(ext4_ioend_wq(inode));
3482 }
3483}
3484
3485extern const struct iomap_ops ext4_iomap_ops;
3486extern const struct iomap_ops ext4_iomap_overwrite_ops;
3487extern const struct iomap_ops ext4_iomap_report_ops;
3488
3489static inline int ext4_buffer_uptodate(struct buffer_head *bh)
3490{
3491 /*
3492 * If the buffer has the write error flag, we have failed
3493 * to write out data in the block. In this case, we don't
3494 * have to read the block because we may read the old data
3495 * successfully.
3496 */
3497 if (!buffer_uptodate(bh) && buffer_write_io_error(bh))
3498 set_buffer_uptodate(bh);
3499 return buffer_uptodate(bh);
3500}
3501
3502#endif /* __KERNEL__ */
3503
3504#define EFSBADCRC EBADMSG /* Bad CRC detected */
3505#define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
3506
3507#endif /* _EXT4_H */