blob: 56f9de2997afa5d96886c21e59c217316defb3dc [file] [log] [blame]
xjb04a4022021-11-25 15:01:52 +08001// 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/* Until this gets included into linux/compiler-gcc.h */
49#ifndef __nonstring
50#if defined(GCC_VERSION) && (GCC_VERSION >= 80000)
51#define __nonstring __attribute__((nonstring))
52#else
53#define __nonstring
54#endif
55#endif
56
57/*
58 * The fourth extended filesystem constants/structures
59 */
60
61/*
62 * with AGGRESSIVE_CHECK allocator runs consistency checks over
63 * structures. these checks slow things down a lot
64 */
65#define AGGRESSIVE_CHECK__
66
67/*
68 * with DOUBLE_CHECK defined mballoc creates persistent in-core
69 * bitmaps, maintains and uses them to check for double allocations
70 */
71#define DOUBLE_CHECK__
72
73/*
74 * Define EXT4FS_DEBUG to produce debug messages
75 */
76#undef EXT4FS_DEBUG
77
78/*
79 * Debug code
80 */
81#ifdef EXT4FS_DEBUG
82#define ext4_debug(f, a...) \
83 do { \
84 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
85 __FILE__, __LINE__, __func__); \
86 printk(KERN_DEBUG f, ## a); \
87 } while (0)
88#else
89#define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
90#endif
91
92/*
93 * Turn on EXT_DEBUG to get lots of info about extents operations.
94 */
95#define EXT_DEBUG__
96#ifdef EXT_DEBUG
97#define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__)
98#else
99#define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
100#endif
101
102/* data type for block offset of block group */
103typedef int ext4_grpblk_t;
104
105/* data type for filesystem-wide blocks number */
106typedef unsigned long long ext4_fsblk_t;
107
108/* data type for file logical block number */
109typedef __u32 ext4_lblk_t;
110
111/* data type for block group number */
112typedef unsigned int ext4_group_t;
113
114enum SHIFT_DIRECTION {
115 SHIFT_LEFT = 0,
116 SHIFT_RIGHT,
117};
118
119/*
120 * Flags used in mballoc's allocation_context flags field.
121 *
122 * Also used to show what's going on for debugging purposes when the
123 * flag field is exported via the traceport interface
124 */
125
126/* prefer goal again. length */
127#define EXT4_MB_HINT_MERGE 0x0001
128/* blocks already reserved */
129#define EXT4_MB_HINT_RESERVED 0x0002
130/* metadata is being allocated */
131#define EXT4_MB_HINT_METADATA 0x0004
132/* first blocks in the file */
133#define EXT4_MB_HINT_FIRST 0x0008
134/* search for the best chunk */
135#define EXT4_MB_HINT_BEST 0x0010
136/* data is being allocated */
137#define EXT4_MB_HINT_DATA 0x0020
138/* don't preallocate (for tails) */
139#define EXT4_MB_HINT_NOPREALLOC 0x0040
140/* allocate for locality group */
141#define EXT4_MB_HINT_GROUP_ALLOC 0x0080
142/* allocate goal blocks or none */
143#define EXT4_MB_HINT_GOAL_ONLY 0x0100
144/* goal is meaningful */
145#define EXT4_MB_HINT_TRY_GOAL 0x0200
146/* blocks already pre-reserved by delayed allocation */
147#define EXT4_MB_DELALLOC_RESERVED 0x0400
148/* We are doing stream allocation */
149#define EXT4_MB_STREAM_ALLOC 0x0800
150/* Use reserved root blocks if needed */
151#define EXT4_MB_USE_ROOT_BLOCKS 0x1000
152/* Use blocks from reserved pool */
153#define EXT4_MB_USE_RESERVED 0x2000
154
155struct ext4_allocation_request {
156 /* target inode for block we're allocating */
157 struct inode *inode;
158 /* how many blocks we want to allocate */
159 unsigned int len;
160 /* logical block in target inode */
161 ext4_lblk_t logical;
162 /* the closest logical allocated block to the left */
163 ext4_lblk_t lleft;
164 /* the closest logical allocated block to the right */
165 ext4_lblk_t lright;
166 /* phys. target (a hint) */
167 ext4_fsblk_t goal;
168 /* phys. block for the closest logical allocated block to the left */
169 ext4_fsblk_t pleft;
170 /* phys. block for the closest logical allocated block to the right */
171 ext4_fsblk_t pright;
172 /* flags. see above EXT4_MB_HINT_* */
173 unsigned int flags;
174};
175
176/*
177 * Logical to physical block mapping, used by ext4_map_blocks()
178 *
179 * This structure is used to pass requests into ext4_map_blocks() as
180 * well as to store the information returned by ext4_map_blocks(). It
181 * takes less room on the stack than a struct buffer_head.
182 */
183#define EXT4_MAP_NEW (1 << BH_New)
184#define EXT4_MAP_MAPPED (1 << BH_Mapped)
185#define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten)
186#define EXT4_MAP_BOUNDARY (1 << BH_Boundary)
187#define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
188 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
189
190struct ext4_map_blocks {
191 ext4_fsblk_t m_pblk;
192 ext4_lblk_t m_lblk;
193 unsigned int m_len;
194 unsigned int m_flags;
195};
196
197/*
198 * Block validity checking, system zone rbtree.
199 */
200struct ext4_system_blocks {
201 struct rb_root root;
202 struct rcu_head rcu;
203};
204
205/*
206 * Flags for ext4_io_end->flags
207 */
208#define EXT4_IO_END_UNWRITTEN 0x0001
209
210/*
211 * For converting unwritten extents on a work queue. 'handle' is used for
212 * buffered writeback.
213 */
214typedef struct ext4_io_end {
215 struct list_head list; /* per-file finished IO list */
216 handle_t *handle; /* handle reserved for extent
217 * conversion */
218 struct inode *inode; /* file being written to */
219 struct bio *bio; /* Linked list of completed
220 * bios covering the extent */
221 unsigned int flag; /* unwritten or not */
222 atomic_t count; /* reference counter */
223 loff_t offset; /* offset in the file */
224 ssize_t size; /* size of the extent */
225} ext4_io_end_t;
226
227struct ext4_io_submit {
228 struct writeback_control *io_wbc;
229 struct bio *io_bio;
230 ext4_io_end_t *io_end;
231 sector_t io_next_block;
232};
233
234/*
235 * Special inodes numbers
236 */
237#define EXT4_BAD_INO 1 /* Bad blocks inode */
238#define EXT4_ROOT_INO 2 /* Root inode */
239#define EXT4_USR_QUOTA_INO 3 /* User quota inode */
240#define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
241#define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
242#define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
243#define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
244#define EXT4_JOURNAL_INO 8 /* Journal inode */
245
246/* First non-reserved inode for old ext4 filesystems */
247#define EXT4_GOOD_OLD_FIRST_INO 11
248
249/*
250 * Maximal count of links to a file
251 */
252#define EXT4_LINK_MAX 65000
253
254/*
255 * Macro-instructions used to manage several block sizes
256 */
257#define EXT4_MIN_BLOCK_SIZE 1024
258#define EXT4_MAX_BLOCK_SIZE 65536
259#define EXT4_MIN_BLOCK_LOG_SIZE 10
260#define EXT4_MAX_BLOCK_LOG_SIZE 16
261#define EXT4_MAX_CLUSTER_LOG_SIZE 30
262#ifdef __KERNEL__
263# define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
264#else
265# define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
266#endif
267#define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
268#define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
269 EXT4_SB(s)->s_cluster_bits)
270#ifdef __KERNEL__
271# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
272# define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
273#else
274# define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
275#endif
276#ifdef __KERNEL__
277#define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
278#define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
279#define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
280#else
281#define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
282 EXT4_GOOD_OLD_INODE_SIZE : \
283 (s)->s_inode_size)
284#define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
285 EXT4_GOOD_OLD_FIRST_INO : \
286 (s)->s_first_ino)
287#endif
288#define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
289#define EXT4_MAX_BLOCKS(size, offset, blkbits) \
290 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \
291 blkbits))
292
293/* Translate a block number to a cluster number */
294#define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
295/* Translate a cluster number to a block number */
296#define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
297/* Translate # of blks to # of clusters */
298#define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
299 (sbi)->s_cluster_bits)
300/* Mask out the low bits to get the starting block of the cluster */
301#define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
302 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
303#define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
304 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
305/* Get the cluster offset */
306#define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
307 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
308#define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
309 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
310
311/*
312 * Structure of a blocks group descriptor
313 */
314struct ext4_group_desc
315{
316 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
317 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
318 __le32 bg_inode_table_lo; /* Inodes table block */
319 __le16 bg_free_blocks_count_lo;/* Free blocks count */
320 __le16 bg_free_inodes_count_lo;/* Free inodes count */
321 __le16 bg_used_dirs_count_lo; /* Directories count */
322 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
323 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
324 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
325 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
326 __le16 bg_itable_unused_lo; /* Unused inodes count */
327 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
328 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
329 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
330 __le32 bg_inode_table_hi; /* Inodes table block MSB */
331 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
332 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
333 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
334 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
335 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
336 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
337 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
338 __u32 bg_reserved;
339};
340
341#define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
342 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
343 sizeof(__le16))
344#define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
345 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
346 sizeof(__le16))
347
348/*
349 * Structure of a flex block group info
350 */
351
352struct flex_groups {
353 atomic64_t free_clusters;
354 atomic_t free_inodes;
355 atomic_t used_dirs;
356};
357
358#define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
359#define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
360#define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
361
362/*
363 * Macro-instructions used to manage group descriptors
364 */
365#define EXT4_MIN_DESC_SIZE 32
366#define EXT4_MIN_DESC_SIZE_64BIT 64
367#define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
368#define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
369#ifdef __KERNEL__
370# define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
371# define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
372# define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
373# define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
374# define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
375#else
376# define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
377# define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
378# define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
379#endif
380
381/*
382 * Constants relative to the data blocks
383 */
384#define EXT4_NDIR_BLOCKS 12
385#define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
386#define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
387#define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
388#define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
389
390/*
391 * Inode flags
392 */
393#define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
394#define EXT4_UNRM_FL 0x00000002 /* Undelete */
395#define EXT4_COMPR_FL 0x00000004 /* Compress file */
396#define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
397#define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
398#define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
399#define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
400#define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
401/* Reserved for compression usage... */
402#define EXT4_DIRTY_FL 0x00000100
403#define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
404#define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
405 /* nb: was previously EXT2_ECOMPR_FL */
406#define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */
407/* End compression flags --- maybe not all used */
408#define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
409#define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
410#define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
411#define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
412#define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
413#define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
414#define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
415#define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
416#define EXT4_VERITY_FL 0x00100000 /* Verity protected inode */
417#define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
418#define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
419#define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
420#define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
421#define EXT4_CASEFOLD_FL 0x40000000 /* Casefolded file */
422#define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
423
424#define EXT4_FL_USER_VISIBLE 0x705BDFFF /* User visible flags */
425#define EXT4_FL_USER_MODIFIABLE 0x604BC0FF /* User modifiable flags */
426
427/* Flags we can manipulate with through EXT4_IOC_FSSETXATTR */
428#define EXT4_FL_XFLAG_VISIBLE (EXT4_SYNC_FL | \
429 EXT4_IMMUTABLE_FL | \
430 EXT4_APPEND_FL | \
431 EXT4_NODUMP_FL | \
432 EXT4_NOATIME_FL | \
433 EXT4_PROJINHERIT_FL)
434
435/* Flags that should be inherited by new inodes from their parent. */
436#define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
437 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
438 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
439 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
440 EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL)
441
442/* Flags that are appropriate for regular files (all but dir-specific ones). */
443#define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_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 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
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(EOFBLOCKS);
537 CHECK_FLAG_VALUE(INLINE_DATA);
538 CHECK_FLAG_VALUE(PROJINHERIT);
539 CHECK_FLAG_VALUE(RESERVED);
540}
541
542/* Used to pass group descriptor data when online resize is done */
543struct ext4_new_group_input {
544 __u32 group; /* Group number for this data */
545 __u64 block_bitmap; /* Absolute block number of block bitmap */
546 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
547 __u64 inode_table; /* Absolute block number of inode table start */
548 __u32 blocks_count; /* Total number of blocks in this group */
549 __u16 reserved_blocks; /* Number of reserved blocks in this group */
550 __u16 unused;
551};
552
553#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
554struct compat_ext4_new_group_input {
555 u32 group;
556 compat_u64 block_bitmap;
557 compat_u64 inode_bitmap;
558 compat_u64 inode_table;
559 u32 blocks_count;
560 u16 reserved_blocks;
561 u16 unused;
562};
563#endif
564
565/* The struct ext4_new_group_input in kernel space, with free_blocks_count */
566struct ext4_new_group_data {
567 __u32 group;
568 __u64 block_bitmap;
569 __u64 inode_bitmap;
570 __u64 inode_table;
571 __u32 blocks_count;
572 __u16 reserved_blocks;
573 __u16 mdata_blocks;
574 __u32 free_clusters_count;
575};
576
577/* Indexes used to index group tables in ext4_new_group_data */
578enum {
579 BLOCK_BITMAP = 0, /* block bitmap */
580 INODE_BITMAP, /* inode bitmap */
581 INODE_TABLE, /* inode tables */
582 GROUP_TABLE_COUNT,
583};
584
585/*
586 * Flags used by ext4_map_blocks()
587 */
588 /* Allocate any needed blocks and/or convert an unwritten
589 extent to be an initialized ext4 */
590#define EXT4_GET_BLOCKS_CREATE 0x0001
591 /* Request the creation of an unwritten extent */
592#define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002
593#define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\
594 EXT4_GET_BLOCKS_CREATE)
595 /* Caller is from the delayed allocation writeout path
596 * finally doing the actual allocation of delayed blocks */
597#define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
598 /* caller is from the direct IO path, request to creation of an
599 unwritten extents if not allocated, split the unwritten
600 extent if blocks has been preallocated already*/
601#define EXT4_GET_BLOCKS_PRE_IO 0x0008
602#define EXT4_GET_BLOCKS_CONVERT 0x0010
603#define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
604 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
605 /* Convert extent to initialized after IO complete */
606#define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
607 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
608 /* Eventual metadata allocation (due to growing extent tree)
609 * should not fail, so try to use reserved blocks for that.*/
610#define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
611 /* Don't normalize allocation size (used for fallocate) */
612#define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
613 /* Request will not result in inode size update (user for fallocate) */
614#define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080
615 /* Convert written extents to unwritten */
616#define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100
617 /* Write zeros to newly created written extents */
618#define EXT4_GET_BLOCKS_ZERO 0x0200
619#define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\
620 EXT4_GET_BLOCKS_ZERO)
621 /* Caller will submit data before dropping transaction handle. This
622 * allows jbd2 to avoid submitting data before commit. */
623#define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400
624
625/*
626 * The bit position of these flags must not overlap with any of the
627 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(),
628 * read_extent_tree_block(), ext4_split_extent_at(),
629 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
630 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
631 * caching the extents when reading from the extent tree while a
632 * truncate or punch hole operation is in progress.
633 */
634#define EXT4_EX_NOCACHE 0x40000000
635#define EXT4_EX_FORCE_CACHE 0x20000000
636
637/*
638 * Flags used by ext4_free_blocks
639 */
640#define EXT4_FREE_BLOCKS_METADATA 0x0001
641#define EXT4_FREE_BLOCKS_FORGET 0x0002
642#define EXT4_FREE_BLOCKS_VALIDATED 0x0004
643#define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
644#define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
645#define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
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
672#define EXT4_IOC_FSGETXATTR FS_IOC_FSGETXATTR
673#define EXT4_IOC_FSSETXATTR FS_IOC_FSSETXATTR
674
675#define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32)
676
677/*
678 * Flags for going down operation
679 */
680#define EXT4_GOING_FLAGS_DEFAULT 0x0 /* going down */
681#define EXT4_GOING_FLAGS_LOGFLUSH 0x1 /* flush log but not data */
682#define EXT4_GOING_FLAGS_NOLOGFLUSH 0x2 /* don't flush log nor data */
683
684
685#if defined(__KERNEL__) && defined(CONFIG_COMPAT)
686/*
687 * ioctl commands in 32 bit emulation
688 */
689#define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
690#define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
691#define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
692#define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
693#define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
694#define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
695#define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
696#define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
697#define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
698#define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
699#endif
700
701/* Max physical block we can address w/o extents */
702#define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
703
704/* Max logical block we can support */
705#define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFF
706
707/*
708 * Structure of an inode on the disk
709 */
710struct ext4_inode {
711 __le16 i_mode; /* File mode */
712 __le16 i_uid; /* Low 16 bits of Owner Uid */
713 __le32 i_size_lo; /* Size in bytes */
714 __le32 i_atime; /* Access time */
715 __le32 i_ctime; /* Inode Change time */
716 __le32 i_mtime; /* Modification time */
717 __le32 i_dtime; /* Deletion Time */
718 __le16 i_gid; /* Low 16 bits of Group Id */
719 __le16 i_links_count; /* Links count */
720 __le32 i_blocks_lo; /* Blocks count */
721 __le32 i_flags; /* File flags */
722 union {
723 struct {
724 __le32 l_i_version;
725 } linux1;
726 struct {
727 __u32 h_i_translator;
728 } hurd1;
729 struct {
730 __u32 m_i_reserved1;
731 } masix1;
732 } osd1; /* OS dependent 1 */
733 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
734 __le32 i_generation; /* File version (for NFS) */
735 __le32 i_file_acl_lo; /* File ACL */
736 __le32 i_size_high;
737 __le32 i_obso_faddr; /* Obsoleted fragment address */
738 union {
739 struct {
740 __le16 l_i_blocks_high; /* were l_i_reserved1 */
741 __le16 l_i_file_acl_high;
742 __le16 l_i_uid_high; /* these 2 fields */
743 __le16 l_i_gid_high; /* were reserved2[0] */
744 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
745 __le16 l_i_reserved;
746 } linux2;
747 struct {
748 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
749 __u16 h_i_mode_high;
750 __u16 h_i_uid_high;
751 __u16 h_i_gid_high;
752 __u32 h_i_author;
753 } hurd2;
754 struct {
755 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
756 __le16 m_i_file_acl_high;
757 __u32 m_i_reserved2[2];
758 } masix2;
759 } osd2; /* OS dependent 2 */
760 __le16 i_extra_isize;
761 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
762 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
763 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
764 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
765 __le32 i_crtime; /* File Creation time */
766 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
767 __le32 i_version_hi; /* high 32 bits for 64-bit version */
768 __le32 i_projid; /* Project ID */
769};
770
771struct move_extent {
772 __u32 reserved; /* should be zero */
773 __u32 donor_fd; /* donor file descriptor */
774 __u64 orig_start; /* logical start offset in block for orig */
775 __u64 donor_start; /* logical start offset in block for donor */
776 __u64 len; /* block length to be moved */
777 __u64 moved_len; /* moved block length */
778};
779
780#define EXT4_EPOCH_BITS 2
781#define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
782#define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
783
784/*
785 * Extended fields will fit into an inode if the filesystem was formatted
786 * with large inodes (-I 256 or larger) and there are not currently any EAs
787 * consuming all of the available space. For new inodes we always reserve
788 * enough space for the kernel's known extended fields, but for inodes
789 * created with an old kernel this might not have been the case. None of
790 * the extended inode fields is critical for correct filesystem operation.
791 * This macro checks if a certain field fits in the inode. Note that
792 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
793 */
794#define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
795 ((offsetof(typeof(*ext4_inode), field) + \
796 sizeof((ext4_inode)->field)) \
797 <= (EXT4_GOOD_OLD_INODE_SIZE + \
798 (einode)->i_extra_isize)) \
799
800/*
801 * We use an encoding that preserves the times for extra epoch "00":
802 *
803 * extra msb of adjust for signed
804 * epoch 32-bit 32-bit tv_sec to
805 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
806 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
807 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
808 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
809 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
810 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
811 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
812 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
813 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
814 *
815 * Note that previous versions of the kernel on 64-bit systems would
816 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
817 * 1970. e2fsck will correct this, assuming that it is run on the
818 * affected filesystem before 2242.
819 */
820
821static inline __le32 ext4_encode_extra_time(struct timespec64 *time)
822{
823 u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK;
824 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
825}
826
827static inline void ext4_decode_extra_time(struct timespec64 *time,
828 __le32 extra)
829{
830 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK))) {
831
832#if 1
833 /* Handle legacy encoding of pre-1970 dates with epoch
834 * bits 1,1. (This backwards compatibility may be removed
835 * at the discretion of the ext4 developers.)
836 */
837 u64 extra_bits = le32_to_cpu(extra) & EXT4_EPOCH_MASK;
838 if (extra_bits == 3 && ((time->tv_sec) & 0x80000000) != 0)
839 extra_bits = 0;
840 time->tv_sec += extra_bits << 32;
841#else
842 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
843#endif
844 }
845 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
846}
847
848#define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
849do { \
850 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
851 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {\
852 (raw_inode)->xtime ## _extra = \
853 ext4_encode_extra_time(&(inode)->xtime); \
854 } \
855} while (0)
856
857#define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
858do { \
859 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
860 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
861 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
862 (raw_inode)->xtime ## _extra = \
863 ext4_encode_extra_time(&(einode)->xtime); \
864} while (0)
865
866#define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
867do { \
868 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
869 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \
870 ext4_decode_extra_time(&(inode)->xtime, \
871 raw_inode->xtime ## _extra); \
872 } \
873 else \
874 (inode)->xtime.tv_nsec = 0; \
875} while (0)
876
877
878#define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
879do { \
880 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
881 (einode)->xtime.tv_sec = \
882 (signed)le32_to_cpu((raw_inode)->xtime); \
883 else \
884 (einode)->xtime.tv_sec = 0; \
885 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
886 ext4_decode_extra_time(&(einode)->xtime, \
887 raw_inode->xtime ## _extra); \
888 else \
889 (einode)->xtime.tv_nsec = 0; \
890} while (0)
891
892#define i_disk_version osd1.linux1.l_i_version
893
894#if defined(__KERNEL__) || defined(__linux__)
895#define i_reserved1 osd1.linux1.l_i_reserved1
896#define i_file_acl_high osd2.linux2.l_i_file_acl_high
897#define i_blocks_high osd2.linux2.l_i_blocks_high
898#define i_uid_low i_uid
899#define i_gid_low i_gid
900#define i_uid_high osd2.linux2.l_i_uid_high
901#define i_gid_high osd2.linux2.l_i_gid_high
902#define i_checksum_lo osd2.linux2.l_i_checksum_lo
903
904#elif defined(__GNU__)
905
906#define i_translator osd1.hurd1.h_i_translator
907#define i_uid_high osd2.hurd2.h_i_uid_high
908#define i_gid_high osd2.hurd2.h_i_gid_high
909#define i_author osd2.hurd2.h_i_author
910
911#elif defined(__masix__)
912
913#define i_reserved1 osd1.masix1.m_i_reserved1
914#define i_file_acl_high osd2.masix2.m_i_file_acl_high
915#define i_reserved2 osd2.masix2.m_i_reserved2
916
917#endif /* defined(__KERNEL__) || defined(__linux__) */
918
919#include "extents_status.h"
920
921/*
922 * Lock subclasses for i_data_sem in the ext4_inode_info structure.
923 *
924 * These are needed to avoid lockdep false positives when we need to
925 * allocate blocks to the quota inode during ext4_map_blocks(), while
926 * holding i_data_sem for a normal (non-quota) inode. Since we don't
927 * do quota tracking for the quota inode, this avoids deadlock (as
928 * well as infinite recursion, since it isn't turtles all the way
929 * down...)
930 *
931 * I_DATA_SEM_NORMAL - Used for most inodes
932 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode
933 * where the second inode has larger inode number
934 * than the first
935 * I_DATA_SEM_QUOTA - Used for quota inodes only
936 */
937enum {
938 I_DATA_SEM_NORMAL = 0,
939 I_DATA_SEM_OTHER,
940 I_DATA_SEM_QUOTA,
941};
942
943
944/*
945 * fourth extended file system inode data in memory
946 */
947struct ext4_inode_info {
948 __le32 i_data[15]; /* unconverted */
949 __u32 i_dtime;
950 ext4_fsblk_t i_file_acl;
951
952 /*
953 * i_block_group is the number of the block group which contains
954 * this file's inode. Constant across the lifetime of the inode,
955 * it is used for making block allocation decisions - we try to
956 * place a file's data blocks near its inode block, and new inodes
957 * near to their parent directory's inode.
958 */
959 ext4_group_t i_block_group;
960 ext4_lblk_t i_dir_start_lookup;
961#if (BITS_PER_LONG < 64)
962 unsigned long i_state_flags; /* Dynamic state flags */
963#endif
964 unsigned long i_flags;
965
966 /*
967 * Extended attributes can be read independently of the main file
968 * data. Taking i_mutex even when reading would cause contention
969 * between readers of EAs and writers of regular file data, so
970 * instead we synchronize on xattr_sem when reading or changing
971 * EAs.
972 */
973 struct rw_semaphore xattr_sem;
974
975 struct list_head i_orphan; /* unlinked but open inodes */
976
977 /*
978 * i_disksize keeps track of what the inode size is ON DISK, not
979 * in memory. During truncate, i_size is set to the new size by
980 * the VFS prior to calling ext4_truncate(), but the filesystem won't
981 * set i_disksize to 0 until the truncate is actually under way.
982 *
983 * The intent is that i_disksize always represents the blocks which
984 * are used by this file. This allows recovery to restart truncate
985 * on orphans if we crash during truncate. We actually write i_disksize
986 * into the on-disk inode when writing inodes out, instead of i_size.
987 *
988 * The only time when i_disksize and i_size may be different is when
989 * a truncate is in progress. The only things which change i_disksize
990 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
991 */
992 loff_t i_disksize;
993
994 /*
995 * i_data_sem is for serialising ext4_truncate() against
996 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
997 * data tree are chopped off during truncate. We can't do that in
998 * ext4 because whenever we perform intermediate commits during
999 * truncate, the inode and all the metadata blocks *must* be in a
1000 * consistent state which allows truncation of the orphans to restart
1001 * during recovery. Hence we must fix the get_block-vs-truncate race
1002 * by other means, so we have i_data_sem.
1003 */
1004 struct rw_semaphore i_data_sem;
1005 /*
1006 * i_mmap_sem is for serializing page faults with truncate / punch hole
1007 * operations. We have to make sure that new page cannot be faulted in
1008 * a section of the inode that is being punched. We cannot easily use
1009 * i_data_sem for this since we need protection for the whole punch
1010 * operation and i_data_sem ranks below transaction start so we have
1011 * to occasionally drop it.
1012 */
1013 struct rw_semaphore i_mmap_sem;
1014 struct inode vfs_inode;
1015 struct jbd2_inode *jinode;
1016
1017 spinlock_t i_raw_lock; /* protects updates to the raw inode */
1018
1019 /*
1020 * File creation time. Its function is same as that of
1021 * struct timespec64 i_{a,c,m}time in the generic inode.
1022 */
1023 struct timespec64 i_crtime;
1024
1025 /* mballoc */
1026 struct list_head i_prealloc_list;
1027 spinlock_t i_prealloc_lock;
1028
1029 /* extents status tree */
1030 struct ext4_es_tree i_es_tree;
1031 rwlock_t i_es_lock;
1032 struct list_head i_es_list;
1033 unsigned int i_es_all_nr; /* protected by i_es_lock */
1034 unsigned int i_es_shk_nr; /* protected by i_es_lock */
1035 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for
1036 extents to shrink. Protected by
1037 i_es_lock */
1038
1039 /* ialloc */
1040 ext4_group_t i_last_alloc_group;
1041
1042 /* allocation reservation info for delalloc */
1043 /* In case of bigalloc, this refer to clusters rather than blocks */
1044 unsigned int i_reserved_data_blocks;
1045 ext4_lblk_t i_da_metadata_calc_last_lblock;
1046 int i_da_metadata_calc_len;
1047
1048 /* on-disk additional length */
1049 __u16 i_extra_isize;
1050
1051 /* Indicate the inline data space. */
1052 u16 i_inline_off;
1053 u16 i_inline_size;
1054
1055#ifdef CONFIG_QUOTA
1056 /* quota space reservation, managed internally by quota code */
1057 qsize_t i_reserved_quota;
1058#endif
1059
1060 /* Lock protecting lists below */
1061 spinlock_t i_completed_io_lock;
1062 /*
1063 * Completed IOs that need unwritten extents handling and have
1064 * transaction reserved
1065 */
1066 struct list_head i_rsv_conversion_list;
1067 struct work_struct i_rsv_conversion_work;
1068 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1069
1070 spinlock_t i_block_reservation_lock;
1071
1072 /*
1073 * Transactions that contain inode's metadata needed to complete
1074 * fsync and fdatasync, respectively.
1075 */
1076 tid_t i_sync_tid;
1077 tid_t i_datasync_tid;
1078
1079#ifdef CONFIG_QUOTA
1080 struct dquot *i_dquot[MAXQUOTAS];
1081#endif
1082
1083 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1084 __u32 i_csum_seed;
1085
1086 kprojid_t i_projid;
1087};
1088
1089/*
1090 * File system states
1091 */
1092#define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
1093#define EXT4_ERROR_FS 0x0002 /* Errors detected */
1094#define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
1095
1096/*
1097 * Misc. filesystem flags
1098 */
1099#define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
1100#define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
1101#define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
1102
1103/*
1104 * Mount flags set via mount options or defaults
1105 */
1106#define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */
1107#define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
1108#define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
1109#define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
1110#define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
1111#define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
1112#define EXT4_MOUNT_ERRORS_MASK 0x00070
1113#define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
1114#define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
1115#ifdef CONFIG_FS_DAX
1116#define EXT4_MOUNT_DAX 0x00200 /* Direct Access */
1117#else
1118#define EXT4_MOUNT_DAX 0
1119#endif
1120#define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
1121#define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
1122#define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
1123#define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
1124#define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
1125#define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
1126#define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
1127#define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
1128#define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
1129#define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
1130#define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */
1131#define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota,
1132 * enable enforcement for hidden
1133 * quota files */
1134#define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable
1135 * enforcement for hidden quota
1136 * files */
1137#define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota
1138 * enforcement */
1139#define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
1140#define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
1141#define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1142#define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */
1143#define EXT4_MOUNT_INLINECRYPT 0x4000000 /* Inline encryption support */
1144#define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
1145#define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
1146#define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
1147#define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
1148#define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
1149
1150/*
1151 * Mount flags set either automatically (could not be set by mount option)
1152 * based on per file system feature or property or in special cases such as
1153 * distinguishing between explicit mount option definition and default.
1154 */
1155#define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1156 specified delalloc */
1157#define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1158 size of blocksize * 8
1159 blocks */
1160#define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1161 file systems */
1162
1163#define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly
1164 specified journal checksum */
1165
1166#define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1167 ~EXT4_MOUNT_##opt
1168#define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1169 EXT4_MOUNT_##opt
1170#define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1171 EXT4_MOUNT_##opt)
1172
1173#define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1174 ~EXT4_MOUNT2_##opt
1175#define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1176 EXT4_MOUNT2_##opt
1177#define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1178 EXT4_MOUNT2_##opt)
1179
1180#define ext4_test_and_set_bit __test_and_set_bit_le
1181#define ext4_set_bit __set_bit_le
1182#define ext4_set_bit_atomic ext2_set_bit_atomic
1183#define ext4_test_and_clear_bit __test_and_clear_bit_le
1184#define ext4_clear_bit __clear_bit_le
1185#define ext4_clear_bit_atomic ext2_clear_bit_atomic
1186#define ext4_test_bit test_bit_le
1187#define ext4_find_next_zero_bit find_next_zero_bit_le
1188#define ext4_find_next_bit find_next_bit_le
1189
1190extern void ext4_set_bits(void *bm, int cur, int len);
1191
1192/*
1193 * Maximal mount counts between two filesystem checks
1194 */
1195#define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1196#define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1197
1198/*
1199 * Behaviour when detecting errors
1200 */
1201#define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1202#define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1203#define EXT4_ERRORS_PANIC 3 /* Panic */
1204#define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1205
1206/* Metadata checksum algorithm codes */
1207#define EXT4_CRC32C_CHKSUM 1
1208
1209/*
1210 * Structure of the super block
1211 */
1212struct ext4_super_block {
1213/*00*/ __le32 s_inodes_count; /* Inodes count */
1214 __le32 s_blocks_count_lo; /* Blocks count */
1215 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1216 __le32 s_free_blocks_count_lo; /* Free blocks count */
1217/*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1218 __le32 s_first_data_block; /* First Data Block */
1219 __le32 s_log_block_size; /* Block size */
1220 __le32 s_log_cluster_size; /* Allocation cluster size */
1221/*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1222 __le32 s_clusters_per_group; /* # Clusters per group */
1223 __le32 s_inodes_per_group; /* # Inodes per group */
1224 __le32 s_mtime; /* Mount time */
1225/*30*/ __le32 s_wtime; /* Write time */
1226 __le16 s_mnt_count; /* Mount count */
1227 __le16 s_max_mnt_count; /* Maximal mount count */
1228 __le16 s_magic; /* Magic signature */
1229 __le16 s_state; /* File system state */
1230 __le16 s_errors; /* Behaviour when detecting errors */
1231 __le16 s_minor_rev_level; /* minor revision level */
1232/*40*/ __le32 s_lastcheck; /* time of last check */
1233 __le32 s_checkinterval; /* max. time between checks */
1234 __le32 s_creator_os; /* OS */
1235 __le32 s_rev_level; /* Revision level */
1236/*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1237 __le16 s_def_resgid; /* Default gid for reserved blocks */
1238 /*
1239 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1240 *
1241 * Note: the difference between the compatible feature set and
1242 * the incompatible feature set is that if there is a bit set
1243 * in the incompatible feature set that the kernel doesn't
1244 * know about, it should refuse to mount the filesystem.
1245 *
1246 * e2fsck's requirements are more strict; if it doesn't know
1247 * about a feature in either the compatible or incompatible
1248 * feature set, it must abort and not try to meddle with
1249 * things it doesn't understand...
1250 */
1251 __le32 s_first_ino; /* First non-reserved inode */
1252 __le16 s_inode_size; /* size of inode structure */
1253 __le16 s_block_group_nr; /* block group # of this superblock */
1254 __le32 s_feature_compat; /* compatible feature set */
1255/*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1256 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1257/*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1258/*78*/ char s_volume_name[16]; /* volume name */
1259/*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */
1260/*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1261 /*
1262 * Performance hints. Directory preallocation should only
1263 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1264 */
1265 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1266 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1267 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1268 /*
1269 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1270 */
1271/*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1272/*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1273 __le32 s_journal_dev; /* device number of journal file */
1274 __le32 s_last_orphan; /* start of list of inodes to delete */
1275 __le32 s_hash_seed[4]; /* HTREE hash seed */
1276 __u8 s_def_hash_version; /* Default hash version to use */
1277 __u8 s_jnl_backup_type;
1278 __le16 s_desc_size; /* size of group descriptor */
1279/*100*/ __le32 s_default_mount_opts;
1280 __le32 s_first_meta_bg; /* First metablock block group */
1281 __le32 s_mkfs_time; /* When the filesystem was created */
1282 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1283 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1284/*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1285 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1286 __le32 s_free_blocks_count_hi; /* Free blocks count */
1287 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1288 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1289 __le32 s_flags; /* Miscellaneous flags */
1290 __le16 s_raid_stride; /* RAID stride */
1291 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1292 __le64 s_mmp_block; /* Block for multi-mount protection */
1293 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1294 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1295 __u8 s_checksum_type; /* metadata checksum algorithm used */
1296 __u8 s_encryption_level; /* versioning level for encryption */
1297 __u8 s_reserved_pad; /* Padding to next 32bits */
1298 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1299 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1300 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1301 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1302 snapshot's future use */
1303 __le32 s_snapshot_list; /* inode number of the head of the
1304 on-disk snapshot list */
1305#define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1306 __le32 s_error_count; /* number of fs errors */
1307 __le32 s_first_error_time; /* first time an error happened */
1308 __le32 s_first_error_ino; /* inode involved in first error */
1309 __le64 s_first_error_block; /* block involved of first error */
1310 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */
1311 __le32 s_first_error_line; /* line number where error happened */
1312 __le32 s_last_error_time; /* most recent time of an error */
1313 __le32 s_last_error_ino; /* inode involved in last error */
1314 __le32 s_last_error_line; /* line number where error happened */
1315 __le64 s_last_error_block; /* block involved of last error */
1316 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */
1317#define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1318 __u8 s_mount_opts[64];
1319 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1320 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1321 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1322 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
1323 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */
1324 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */
1325 __le32 s_lpf_ino; /* Location of the lost+found inode */
1326 __le32 s_prj_quota_inum; /* inode for tracking project quota */
1327 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */
1328 __u8 s_wtime_hi;
1329 __u8 s_mtime_hi;
1330 __u8 s_mkfs_time_hi;
1331 __u8 s_lastcheck_hi;
1332 __u8 s_first_error_time_hi;
1333 __u8 s_last_error_time_hi;
1334 __u8 s_pad[2];
1335 __le16 s_encoding; /* Filename charset encoding */
1336 __le16 s_encoding_flags; /* Filename charset encoding flags */
1337 __le32 s_reserved[95]; /* Padding to the end of the block */
1338 __le32 s_checksum; /* crc32c(superblock) */
1339};
1340
1341#define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1342
1343#ifdef __KERNEL__
1344
1345/*
1346 * run-time mount flags
1347 */
1348#define EXT4_MF_MNTDIR_SAMPLED 0x0001
1349#define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1350#define EXT4_MF_TEST_DUMMY_ENCRYPTION 0x0004
1351
1352#ifdef CONFIG_FS_ENCRYPTION
1353#define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \
1354 EXT4_MF_TEST_DUMMY_ENCRYPTION))
1355#else
1356#define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1357#endif
1358
1359/* Number of quota types we support */
1360#define EXT4_MAXQUOTAS 3
1361
1362#define EXT4_ENC_UTF8_12_1 1
1363
1364/*
1365 * Flags for ext4_sb_info.s_encoding_flags.
1366 */
1367#define EXT4_ENC_STRICT_MODE_FL (1 << 0)
1368
1369#define ext4_has_strict_mode(sbi) \
1370 (sbi->s_encoding_flags & EXT4_ENC_STRICT_MODE_FL)
1371
1372/*
1373 * fourth extended-fs super-block data in memory
1374 */
1375struct ext4_sb_info {
1376 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1377 unsigned long s_inodes_per_block;/* Number of inodes per block */
1378 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1379 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1380 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1381 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1382 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1383 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1384 ext4_group_t s_groups_count; /* Number of groups in the fs */
1385 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1386 unsigned long s_overhead; /* # of fs overhead clusters */
1387 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1388 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1389 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1390 struct buffer_head * s_sbh; /* Buffer containing the super block */
1391 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1392 struct buffer_head **s_group_desc;
1393 unsigned int s_mount_opt;
1394 unsigned int s_mount_opt2;
1395 unsigned int s_mount_flags;
1396 unsigned int s_def_mount_opt;
1397 ext4_fsblk_t s_sb_block;
1398 atomic64_t s_resv_clusters;
1399 kuid_t s_resuid;
1400 kgid_t s_resgid;
1401 unsigned short s_mount_state;
1402 unsigned short s_pad;
1403 int s_addr_per_block_bits;
1404 int s_desc_per_block_bits;
1405 int s_inode_size;
1406 int s_first_ino;
1407 unsigned int s_inode_readahead_blks;
1408 unsigned int s_inode_goal;
1409 u32 s_hash_seed[4];
1410 int s_def_hash_version;
1411 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1412 struct percpu_counter s_freeclusters_counter;
1413 struct percpu_counter s_freeinodes_counter;
1414 struct percpu_counter s_dirs_counter;
1415 struct percpu_counter s_dirtyclusters_counter;
1416 struct blockgroup_lock *s_blockgroup_lock;
1417 struct proc_dir_entry *s_proc;
1418 struct kobject s_kobj;
1419 struct completion s_kobj_unregister;
1420 struct super_block *s_sb;
1421#ifdef CONFIG_UNICODE
1422 struct unicode_map *s_encoding;
1423 __u16 s_encoding_flags;
1424#endif
1425
1426 /* Journaling */
1427 struct journal_s *s_journal;
1428 struct list_head s_orphan;
1429 struct mutex s_orphan_lock;
1430 unsigned long s_ext4_flags; /* Ext4 superblock flags */
1431 unsigned long s_commit_interval;
1432 u32 s_max_batch_time;
1433 u32 s_min_batch_time;
1434 struct block_device *journal_bdev;
1435#ifdef CONFIG_QUOTA
1436 /* Names of quota files with journalled quota */
1437 char __rcu *s_qf_names[EXT4_MAXQUOTAS];
1438 int s_jquota_fmt; /* Format of quota to use */
1439#endif
1440 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1441 struct ext4_system_blocks __rcu *system_blks;
1442
1443#ifdef EXTENTS_STATS
1444 /* ext4 extents stats */
1445 unsigned long s_ext_min;
1446 unsigned long s_ext_max;
1447 unsigned long s_depth_max;
1448 spinlock_t s_ext_stats_lock;
1449 unsigned long s_ext_blocks;
1450 unsigned long s_ext_extents;
1451#endif
1452
1453 /* for buddy allocator */
1454 struct ext4_group_info ***s_group_info;
1455 struct inode *s_buddy_cache;
1456 spinlock_t s_md_lock;
1457 unsigned short *s_mb_offsets;
1458 unsigned int *s_mb_maxs;
1459 unsigned int s_group_info_size;
1460 unsigned int s_mb_free_pending;
1461 struct list_head s_freed_data_list; /* List of blocks to be freed
1462 after commit completed */
1463
1464 /* tunables */
1465 unsigned long s_stripe;
1466 unsigned int s_mb_stream_request;
1467 unsigned int s_mb_max_to_scan;
1468 unsigned int s_mb_min_to_scan;
1469 unsigned int s_mb_stats;
1470 unsigned int s_mb_order2_reqs;
1471 unsigned int s_mb_group_prealloc;
1472 unsigned int s_max_dir_size_kb;
1473 /* where last allocation was done - for stream allocation */
1474 unsigned long s_mb_last_group;
1475 unsigned long s_mb_last_start;
1476
1477 /* stats for buddy allocator */
1478 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1479 atomic_t s_bal_success; /* we found long enough chunks */
1480 atomic_t s_bal_allocated; /* in blocks */
1481 atomic_t s_bal_ex_scanned; /* total extents scanned */
1482 atomic_t s_bal_goals; /* goal hits */
1483 atomic_t s_bal_breaks; /* too long searches */
1484 atomic_t s_bal_2orders; /* 2^order hits */
1485 spinlock_t s_bal_lock;
1486 unsigned long s_mb_buddies_generated;
1487 unsigned long long s_mb_generation_time;
1488 atomic_t s_mb_lost_chunks;
1489 atomic_t s_mb_preallocated;
1490 atomic_t s_mb_discarded;
1491 atomic_t s_lock_busy;
1492
1493 /* locality groups */
1494 struct ext4_locality_group __percpu *s_locality_groups;
1495
1496 /* for write statistics */
1497 unsigned long s_sectors_written_start;
1498 u64 s_kbytes_written;
1499
1500 /* the size of zero-out chunk */
1501 unsigned int s_extent_max_zeroout_kb;
1502
1503 unsigned int s_log_groups_per_flex;
1504 struct flex_groups *s_flex_groups;
1505 ext4_group_t s_flex_groups_allocated;
1506
1507 /* workqueue for reserved extent conversions (buffered io) */
1508 struct workqueue_struct *rsv_conversion_wq;
1509
1510 /* timer for periodic error stats printing */
1511 struct timer_list s_err_report;
1512
1513 /* Lazy inode table initialization info */
1514 struct ext4_li_request *s_li_request;
1515 /* Wait multiplier for lazy initialization thread */
1516 unsigned int s_li_wait_mult;
1517
1518 /* Kernel thread for multiple mount protection */
1519 struct task_struct *s_mmp_tsk;
1520
1521 /* record the last minlen when FITRIM is called. */
1522 atomic_t s_last_trim_minblks;
1523
1524 /* Reference to checksum algorithm driver via cryptoapi */
1525 struct crypto_shash *s_chksum_driver;
1526
1527 /* Precomputed FS UUID checksum for seeding other checksums */
1528 __u32 s_csum_seed;
1529
1530 /* Reclaim extents from extent status tree */
1531 struct shrinker s_es_shrinker;
1532 struct list_head s_es_list; /* List of inodes with reclaimable extents */
1533 long s_es_nr_inode;
1534 struct ext4_es_stats s_es_stats;
1535 struct mb_cache *s_ea_block_cache;
1536 struct mb_cache *s_ea_inode_cache;
1537 spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1538
1539 /* Ratelimit ext4 messages. */
1540 struct ratelimit_state s_err_ratelimit_state;
1541 struct ratelimit_state s_warning_ratelimit_state;
1542 struct ratelimit_state s_msg_ratelimit_state;
1543
1544 /* Barrier between changing inodes' journal flags and writepages ops. */
1545 struct percpu_rw_semaphore s_journal_flag_rwsem;
1546 struct dax_device *s_daxdev;
1547};
1548
1549static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1550{
1551 return sb->s_fs_info;
1552}
1553static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1554{
1555 return container_of(inode, struct ext4_inode_info, vfs_inode);
1556}
1557
1558static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1559{
1560 return ino == EXT4_ROOT_INO ||
1561 (ino >= EXT4_FIRST_INO(sb) &&
1562 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1563}
1564
1565/*
1566 * Inode dynamic state flags
1567 */
1568enum {
1569 EXT4_STATE_JDATA, /* journaled data exists */
1570 EXT4_STATE_NEW, /* inode is newly created */
1571 EXT4_STATE_XATTR, /* has in-inode xattrs */
1572 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1573 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1574 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1575 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
1576 EXT4_STATE_NEWENTRY, /* File just added to dir */
1577 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1578 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
1579 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */
1580 EXT4_STATE_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */
1581};
1582
1583#define EXT4_INODE_BIT_FNS(name, field, offset) \
1584static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1585{ \
1586 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1587} \
1588static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1589{ \
1590 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1591} \
1592static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1593{ \
1594 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1595}
1596
1597/* Add these declarations here only so that these functions can be
1598 * found by name. Otherwise, they are very hard to locate. */
1599static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1600static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1601static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1602EXT4_INODE_BIT_FNS(flag, flags, 0)
1603
1604/* Add these declarations here only so that these functions can be
1605 * found by name. Otherwise, they are very hard to locate. */
1606static inline int ext4_test_inode_state(struct inode *inode, int bit);
1607static inline void ext4_set_inode_state(struct inode *inode, int bit);
1608static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1609#if (BITS_PER_LONG < 64)
1610EXT4_INODE_BIT_FNS(state, state_flags, 0)
1611
1612static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1613{
1614 (ei)->i_state_flags = 0;
1615}
1616#else
1617EXT4_INODE_BIT_FNS(state, flags, 32)
1618
1619static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1620{
1621 /* We depend on the fact that callers will set i_flags */
1622}
1623#endif
1624#else
1625/* Assume that user mode programs are passing in an ext4fs superblock, not
1626 * a kernel struct super_block. This will allow us to call the feature-test
1627 * macros from user land. */
1628#define EXT4_SB(sb) (sb)
1629#endif
1630
1631static inline bool ext4_verity_in_progress(struct inode *inode)
1632{
1633 return IS_ENABLED(CONFIG_FS_VERITY) &&
1634 ext4_test_inode_state(inode, EXT4_STATE_VERITY_IN_PROGRESS);
1635}
1636
1637#define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1638
1639/*
1640 * Codes for operating systems
1641 */
1642#define EXT4_OS_LINUX 0
1643#define EXT4_OS_HURD 1
1644#define EXT4_OS_MASIX 2
1645#define EXT4_OS_FREEBSD 3
1646#define EXT4_OS_LITES 4
1647
1648/*
1649 * Revision levels
1650 */
1651#define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1652#define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1653
1654#define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1655#define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1656
1657#define EXT4_GOOD_OLD_INODE_SIZE 128
1658
1659/*
1660 * Feature set definitions
1661 */
1662
1663#define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1664#define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1665#define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1666#define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1667#define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1668#define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1669#define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
1670#define EXT4_FEATURE_COMPAT_STABLE_INODES 0x0800
1671
1672#define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1673#define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1674#define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1675#define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1676#define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1677#define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1678#define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1679#define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1680#define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1681/*
1682 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1683 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1684 * all other data structures' checksums. However, the METADATA_CSUM and
1685 * GDT_CSUM bits are mutually exclusive.
1686 */
1687#define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
1688#define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000
1689#define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000
1690#define EXT4_FEATURE_RO_COMPAT_VERITY 0x8000
1691
1692#define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1693#define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1694#define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1695#define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1696#define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1697#define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1698#define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1699#define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1700#define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1701#define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1702#define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1703#define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000
1704#define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
1705#define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
1706#define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000
1707#define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000
1708
1709extern void ext4_update_dynamic_rev(struct super_block *sb);
1710
1711#define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
1712static inline bool ext4_has_feature_##name(struct super_block *sb) \
1713{ \
1714 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1715 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
1716} \
1717static inline void ext4_set_feature_##name(struct super_block *sb) \
1718{ \
1719 ext4_update_dynamic_rev(sb); \
1720 EXT4_SB(sb)->s_es->s_feature_compat |= \
1721 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1722} \
1723static inline void ext4_clear_feature_##name(struct super_block *sb) \
1724{ \
1725 EXT4_SB(sb)->s_es->s_feature_compat &= \
1726 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1727}
1728
1729#define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
1730static inline bool ext4_has_feature_##name(struct super_block *sb) \
1731{ \
1732 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1733 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
1734} \
1735static inline void ext4_set_feature_##name(struct super_block *sb) \
1736{ \
1737 ext4_update_dynamic_rev(sb); \
1738 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
1739 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1740} \
1741static inline void ext4_clear_feature_##name(struct super_block *sb) \
1742{ \
1743 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
1744 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1745}
1746
1747#define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
1748static inline bool ext4_has_feature_##name(struct super_block *sb) \
1749{ \
1750 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1751 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
1752} \
1753static inline void ext4_set_feature_##name(struct super_block *sb) \
1754{ \
1755 ext4_update_dynamic_rev(sb); \
1756 EXT4_SB(sb)->s_es->s_feature_incompat |= \
1757 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1758} \
1759static inline void ext4_clear_feature_##name(struct super_block *sb) \
1760{ \
1761 EXT4_SB(sb)->s_es->s_feature_incompat &= \
1762 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1763}
1764
1765EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC)
1766EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES)
1767EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL)
1768EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR)
1769EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE)
1770EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX)
1771EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2)
1772EXT4_FEATURE_COMPAT_FUNCS(stable_inodes, STABLE_INODES)
1773
1774EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER)
1775EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE)
1776EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR)
1777EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE)
1778EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM)
1779EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK)
1780EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE)
1781EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA)
1782EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC)
1783EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM)
1784EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY)
1785EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT)
1786EXT4_FEATURE_RO_COMPAT_FUNCS(verity, VERITY)
1787
1788EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION)
1789EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE)
1790EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER)
1791EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV)
1792EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG)
1793EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS)
1794EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT)
1795EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP)
1796EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG)
1797EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE)
1798EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA)
1799EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED)
1800EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR)
1801EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA)
1802EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT)
1803EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD)
1804
1805#define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1806#define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1807 EXT4_FEATURE_INCOMPAT_META_BG)
1808#define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1809 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1810 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1811
1812#define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1813#define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1814 EXT4_FEATURE_INCOMPAT_RECOVER| \
1815 EXT4_FEATURE_INCOMPAT_META_BG)
1816#define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1817 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1818 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1819
1820#define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1821#define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1822 EXT4_FEATURE_INCOMPAT_RECOVER| \
1823 EXT4_FEATURE_INCOMPAT_META_BG| \
1824 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1825 EXT4_FEATURE_INCOMPAT_64BIT| \
1826 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1827 EXT4_FEATURE_INCOMPAT_EA_INODE| \
1828 EXT4_FEATURE_INCOMPAT_MMP | \
1829 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
1830 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
1831 EXT4_FEATURE_INCOMPAT_CASEFOLD | \
1832 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \
1833 EXT4_FEATURE_INCOMPAT_LARGEDIR)
1834#define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1835 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1836 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1837 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1838 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1839 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1840 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1841 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1842 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1843 EXT4_FEATURE_RO_COMPAT_QUOTA |\
1844 EXT4_FEATURE_RO_COMPAT_PROJECT |\
1845 EXT4_FEATURE_RO_COMPAT_VERITY)
1846
1847#define EXTN_FEATURE_FUNCS(ver) \
1848static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
1849{ \
1850 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1851 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
1852} \
1853static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
1854{ \
1855 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1856 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
1857} \
1858static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
1859{ \
1860 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1861 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
1862}
1863
1864EXTN_FEATURE_FUNCS(2)
1865EXTN_FEATURE_FUNCS(3)
1866EXTN_FEATURE_FUNCS(4)
1867
1868static inline bool ext4_has_compat_features(struct super_block *sb)
1869{
1870 return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
1871}
1872static inline bool ext4_has_ro_compat_features(struct super_block *sb)
1873{
1874 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
1875}
1876static inline bool ext4_has_incompat_features(struct super_block *sb)
1877{
1878 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
1879}
1880
1881/*
1882 * Superblock flags
1883 */
1884#define EXT4_FLAGS_RESIZING 0
1885#define EXT4_FLAGS_SHUTDOWN 1
1886
1887static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi)
1888{
1889 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags);
1890}
1891
1892
1893/*
1894 * Default values for user and/or group using reserved blocks
1895 */
1896#define EXT4_DEF_RESUID 0
1897#define EXT4_DEF_RESGID 0
1898
1899/*
1900 * Default project ID
1901 */
1902#define EXT4_DEF_PROJID 0
1903
1904#define EXT4_DEF_INODE_READAHEAD_BLKS 32
1905
1906/*
1907 * Default mount options
1908 */
1909#define EXT4_DEFM_DEBUG 0x0001
1910#define EXT4_DEFM_BSDGROUPS 0x0002
1911#define EXT4_DEFM_XATTR_USER 0x0004
1912#define EXT4_DEFM_ACL 0x0008
1913#define EXT4_DEFM_UID16 0x0010
1914#define EXT4_DEFM_JMODE 0x0060
1915#define EXT4_DEFM_JMODE_DATA 0x0020
1916#define EXT4_DEFM_JMODE_ORDERED 0x0040
1917#define EXT4_DEFM_JMODE_WBACK 0x0060
1918#define EXT4_DEFM_NOBARRIER 0x0100
1919#define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1920#define EXT4_DEFM_DISCARD 0x0400
1921#define EXT4_DEFM_NODELALLOC 0x0800
1922
1923/*
1924 * Default journal batch times
1925 */
1926#define EXT4_DEF_MIN_BATCH_TIME 0
1927#define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1928
1929/*
1930 * Minimum number of groups in a flexgroup before we separate out
1931 * directories into the first block group of a flexgroup
1932 */
1933#define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1934
1935/*
1936 * Structure of a directory entry
1937 */
1938#define EXT4_NAME_LEN 255
1939
1940struct ext4_dir_entry {
1941 __le32 inode; /* Inode number */
1942 __le16 rec_len; /* Directory entry length */
1943 __le16 name_len; /* Name length */
1944 char name[EXT4_NAME_LEN]; /* File name */
1945};
1946
1947/*
1948 * The new version of the directory entry. Since EXT4 structures are
1949 * stored in intel byte order, and the name_len field could never be
1950 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1951 * file_type field.
1952 */
1953struct ext4_dir_entry_2 {
1954 __le32 inode; /* Inode number */
1955 __le16 rec_len; /* Directory entry length */
1956 __u8 name_len; /* Name length */
1957 __u8 file_type;
1958 char name[EXT4_NAME_LEN]; /* File name */
1959};
1960
1961/*
1962 * This is a bogus directory entry at the end of each leaf block that
1963 * records checksums.
1964 */
1965struct ext4_dir_entry_tail {
1966 __le32 det_reserved_zero1; /* Pretend to be unused */
1967 __le16 det_rec_len; /* 12 */
1968 __u8 det_reserved_zero2; /* Zero name length */
1969 __u8 det_reserved_ft; /* 0xDE, fake file type */
1970 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
1971};
1972
1973#define EXT4_DIRENT_TAIL(block, blocksize) \
1974 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1975 ((blocksize) - \
1976 sizeof(struct ext4_dir_entry_tail))))
1977
1978/*
1979 * Ext4 directory file types. Only the low 3 bits are used. The
1980 * other bits are reserved for now.
1981 */
1982#define EXT4_FT_UNKNOWN 0
1983#define EXT4_FT_REG_FILE 1
1984#define EXT4_FT_DIR 2
1985#define EXT4_FT_CHRDEV 3
1986#define EXT4_FT_BLKDEV 4
1987#define EXT4_FT_FIFO 5
1988#define EXT4_FT_SOCK 6
1989#define EXT4_FT_SYMLINK 7
1990
1991#define EXT4_FT_MAX 8
1992
1993#define EXT4_FT_DIR_CSUM 0xDE
1994
1995/*
1996 * EXT4_DIR_PAD defines the directory entries boundaries
1997 *
1998 * NOTE: It must be a multiple of 4
1999 */
2000#define EXT4_DIR_PAD 4
2001#define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
2002#define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
2003 ~EXT4_DIR_ROUND)
2004#define EXT4_MAX_REC_LEN ((1<<16)-1)
2005
2006/*
2007 * If we ever get support for fs block sizes > page_size, we'll need
2008 * to remove the #if statements in the next two functions...
2009 */
2010static inline unsigned int
2011ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
2012{
2013 unsigned len = le16_to_cpu(dlen);
2014
2015#if (PAGE_SIZE >= 65536)
2016 if (len == EXT4_MAX_REC_LEN || len == 0)
2017 return blocksize;
2018 return (len & 65532) | ((len & 3) << 16);
2019#else
2020 return len;
2021#endif
2022}
2023
2024static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
2025{
2026 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
2027 BUG();
2028#if (PAGE_SIZE >= 65536)
2029 if (len < 65536)
2030 return cpu_to_le16(len);
2031 if (len == blocksize) {
2032 if (blocksize == 65536)
2033 return cpu_to_le16(EXT4_MAX_REC_LEN);
2034 else
2035 return cpu_to_le16(0);
2036 }
2037 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
2038#else
2039 return cpu_to_le16(len);
2040#endif
2041}
2042
2043/*
2044 * Hash Tree Directory indexing
2045 * (c) Daniel Phillips, 2001
2046 */
2047
2048#define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
2049 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2050#define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \
2051 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir)))
2052#define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2053
2054/* Legal values for the dx_root hash_version field: */
2055
2056#define DX_HASH_LEGACY 0
2057#define DX_HASH_HALF_MD4 1
2058#define DX_HASH_TEA 2
2059#define DX_HASH_LEGACY_UNSIGNED 3
2060#define DX_HASH_HALF_MD4_UNSIGNED 4
2061#define DX_HASH_TEA_UNSIGNED 5
2062
2063static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2064 const void *address, unsigned int length)
2065{
2066 struct {
2067 struct shash_desc shash;
2068 char ctx[4];
2069 } desc;
2070
2071 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2072
2073 desc.shash.tfm = sbi->s_chksum_driver;
2074 desc.shash.flags = 0;
2075 *(u32 *)desc.ctx = crc;
2076
2077 BUG_ON(crypto_shash_update(&desc.shash, address, length));
2078
2079 return *(u32 *)desc.ctx;
2080}
2081
2082#ifdef __KERNEL__
2083
2084/* hash info structure used by the directory hash */
2085struct dx_hash_info
2086{
2087 u32 hash;
2088 u32 minor_hash;
2089 int hash_version;
2090 u32 *seed;
2091};
2092
2093
2094/* 32 and 64 bit signed EOF for dx directories */
2095#define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
2096#define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
2097
2098
2099/*
2100 * Control parameters used by ext4_htree_next_block
2101 */
2102#define HASH_NB_ALWAYS 1
2103
2104struct ext4_filename {
2105 const struct qstr *usr_fname;
2106 struct fscrypt_str disk_name;
2107 struct dx_hash_info hinfo;
2108#ifdef CONFIG_FS_ENCRYPTION
2109 struct fscrypt_str crypto_buf;
2110#endif
2111#ifdef CONFIG_UNICODE
2112 struct fscrypt_str cf_name;
2113#endif
2114};
2115
2116#define fname_name(p) ((p)->disk_name.name)
2117#define fname_len(p) ((p)->disk_name.len)
2118
2119/*
2120 * Describe an inode's exact location on disk and in memory
2121 */
2122struct ext4_iloc
2123{
2124 struct buffer_head *bh;
2125 unsigned long offset;
2126 ext4_group_t block_group;
2127};
2128
2129static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2130{
2131 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2132}
2133
2134static inline bool ext4_is_quota_file(struct inode *inode)
2135{
2136 return IS_NOQUOTA(inode) &&
2137 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL);
2138}
2139
2140/*
2141 * This structure is stuffed into the struct file's private_data field
2142 * for directories. It is where we put information so that we can do
2143 * readdir operations in hash tree order.
2144 */
2145struct dir_private_info {
2146 struct rb_root root;
2147 struct rb_node *curr_node;
2148 struct fname *extra_fname;
2149 loff_t last_pos;
2150 __u32 curr_hash;
2151 __u32 curr_minor_hash;
2152 __u32 next_hash;
2153};
2154
2155/* calculate the first block number of the group */
2156static inline ext4_fsblk_t
2157ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2158{
2159 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2160 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2161}
2162
2163/*
2164 * Special error return code only used by dx_probe() and its callers.
2165 */
2166#define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
2167
2168/* htree levels for ext4 */
2169#define EXT4_HTREE_LEVEL_COMPAT 2
2170#define EXT4_HTREE_LEVEL 3
2171
2172static inline int ext4_dir_htree_level(struct super_block *sb)
2173{
2174 return ext4_has_feature_largedir(sb) ?
2175 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT;
2176}
2177
2178/*
2179 * Timeout and state flag for lazy initialization inode thread.
2180 */
2181#define EXT4_DEF_LI_WAIT_MULT 10
2182#define EXT4_DEF_LI_MAX_START_DELAY 5
2183#define EXT4_LAZYINIT_QUIT 0x0001
2184#define EXT4_LAZYINIT_RUNNING 0x0002
2185
2186/*
2187 * Lazy inode table initialization info
2188 */
2189struct ext4_lazy_init {
2190 unsigned long li_state;
2191 struct list_head li_request_list;
2192 struct mutex li_list_mtx;
2193};
2194
2195struct ext4_li_request {
2196 struct super_block *lr_super;
2197 struct ext4_sb_info *lr_sbi;
2198 ext4_group_t lr_next_group;
2199 struct list_head lr_request;
2200 unsigned long lr_next_sched;
2201 unsigned long lr_timeout;
2202};
2203
2204struct ext4_features {
2205 struct kobject f_kobj;
2206 struct completion f_kobj_unregister;
2207};
2208
2209/*
2210 * This structure will be used for multiple mount protection. It will be
2211 * written into the block number saved in the s_mmp_block field in the
2212 * superblock. Programs that check MMP should assume that if
2213 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2214 * to use the filesystem, regardless of how old the timestamp is.
2215 */
2216#define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
2217#define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2218#define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
2219#define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
2220
2221struct mmp_struct {
2222 __le32 mmp_magic; /* Magic number for MMP */
2223 __le32 mmp_seq; /* Sequence no. updated periodically */
2224
2225 /*
2226 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2227 * purposes and do not affect the correctness of the algorithm
2228 */
2229 __le64 mmp_time; /* Time last updated */
2230 char mmp_nodename[64]; /* Node which last updated MMP block */
2231 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
2232
2233 /*
2234 * mmp_check_interval is used to verify if the MMP block has been
2235 * updated on the block device. The value is updated based on the
2236 * maximum time to write the MMP block during an update cycle.
2237 */
2238 __le16 mmp_check_interval;
2239
2240 __le16 mmp_pad1;
2241 __le32 mmp_pad2[226];
2242 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
2243};
2244
2245/* arguments passed to the mmp thread */
2246struct mmpd_data {
2247 struct buffer_head *bh; /* bh from initial read_mmp_block() */
2248 struct super_block *sb; /* super block of the fs */
2249};
2250
2251/*
2252 * Check interval multiplier
2253 * The MMP block is written every update interval and initially checked every
2254 * update interval x the multiplier (the value is then adapted based on the
2255 * write latency). The reason is that writes can be delayed under load and we
2256 * don't want readers to incorrectly assume that the filesystem is no longer
2257 * in use.
2258 */
2259#define EXT4_MMP_CHECK_MULT 2UL
2260
2261/*
2262 * Minimum interval for MMP checking in seconds.
2263 */
2264#define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
2265
2266/*
2267 * Maximum interval for MMP checking in seconds.
2268 */
2269#define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
2270
2271/*
2272 * Function prototypes
2273 */
2274
2275/*
2276 * Ok, these declarations are also in <linux/kernel.h> but none of the
2277 * ext4 source programs needs to include it so they are duplicated here.
2278 */
2279# define NORET_TYPE /**/
2280# define ATTRIB_NORET __attribute__((noreturn))
2281# define NORET_AND noreturn,
2282
2283/* bitmap.c */
2284extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2285void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2286 struct ext4_group_desc *gdp,
2287 struct buffer_head *bh, int sz);
2288int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2289 struct ext4_group_desc *gdp,
2290 struct buffer_head *bh, int sz);
2291void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2292 struct ext4_group_desc *gdp,
2293 struct buffer_head *bh);
2294int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2295 struct ext4_group_desc *gdp,
2296 struct buffer_head *bh);
2297
2298/* balloc.c */
2299extern void ext4_get_group_no_and_offset(struct super_block *sb,
2300 ext4_fsblk_t blocknr,
2301 ext4_group_t *blockgrpp,
2302 ext4_grpblk_t *offsetp);
2303extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2304 ext4_fsblk_t block);
2305
2306extern unsigned int ext4_block_group(struct super_block *sb,
2307 ext4_fsblk_t blocknr);
2308extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2309 ext4_fsblk_t blocknr);
2310extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2311extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2312 ext4_group_t group);
2313extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2314 ext4_fsblk_t goal,
2315 unsigned int flags,
2316 unsigned long *count,
2317 int *errp);
2318extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2319 s64 nclusters, unsigned int flags);
2320extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2321extern void ext4_check_blocks_bitmap(struct super_block *);
2322extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2323 ext4_group_t block_group,
2324 struct buffer_head ** bh);
2325extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2326
2327extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2328 ext4_group_t block_group);
2329extern int ext4_wait_block_bitmap(struct super_block *sb,
2330 ext4_group_t block_group,
2331 struct buffer_head *bh);
2332extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2333 ext4_group_t block_group);
2334extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2335 ext4_group_t block_group,
2336 struct ext4_group_desc *gdp);
2337ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2338
2339#ifdef CONFIG_UNICODE
2340extern void ext4_fname_setup_ci_filename(struct inode *dir,
2341 const struct qstr *iname,
2342 struct fscrypt_str *fname);
2343#endif
2344
2345#ifdef CONFIG_FS_ENCRYPTION
2346static inline void ext4_fname_from_fscrypt_name(struct ext4_filename *dst,
2347 const struct fscrypt_name *src)
2348{
2349 memset(dst, 0, sizeof(*dst));
2350
2351 dst->usr_fname = src->usr_fname;
2352 dst->disk_name = src->disk_name;
2353 dst->hinfo.hash = src->hash;
2354 dst->hinfo.minor_hash = src->minor_hash;
2355 dst->crypto_buf = src->crypto_buf;
2356}
2357
2358static inline int ext4_fname_setup_filename(struct inode *dir,
2359 const struct qstr *iname,
2360 int lookup,
2361 struct ext4_filename *fname)
2362{
2363 struct fscrypt_name name;
2364 int err;
2365
2366 err = fscrypt_setup_filename(dir, iname, lookup, &name);
2367 if (err)
2368 return err;
2369
2370 ext4_fname_from_fscrypt_name(fname, &name);
2371
2372#ifdef CONFIG_UNICODE
2373 ext4_fname_setup_ci_filename(dir, iname, &fname->cf_name);
2374#endif
2375 return 0;
2376}
2377
2378static inline int ext4_fname_prepare_lookup(struct inode *dir,
2379 struct dentry *dentry,
2380 struct ext4_filename *fname)
2381{
2382 struct fscrypt_name name;
2383 int err;
2384
2385 err = fscrypt_prepare_lookup(dir, dentry, &name);
2386 if (err)
2387 return err;
2388
2389 ext4_fname_from_fscrypt_name(fname, &name);
2390
2391#ifdef CONFIG_UNICODE
2392 ext4_fname_setup_ci_filename(dir, &dentry->d_name, &fname->cf_name);
2393#endif
2394 return 0;
2395}
2396
2397static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2398{
2399 struct fscrypt_name name;
2400
2401 name.crypto_buf = fname->crypto_buf;
2402 fscrypt_free_filename(&name);
2403
2404 fname->crypto_buf.name = NULL;
2405 fname->usr_fname = NULL;
2406 fname->disk_name.name = NULL;
2407
2408#ifdef CONFIG_UNICODE
2409 kfree(fname->cf_name.name);
2410 fname->cf_name.name = NULL;
2411#endif
2412}
2413#else /* !CONFIG_FS_ENCRYPTION */
2414static inline int ext4_fname_setup_filename(struct inode *dir,
2415 const struct qstr *iname,
2416 int lookup,
2417 struct ext4_filename *fname)
2418{
2419 fname->usr_fname = iname;
2420 fname->disk_name.name = (unsigned char *) iname->name;
2421 fname->disk_name.len = iname->len;
2422
2423#ifdef CONFIG_UNICODE
2424 ext4_fname_setup_ci_filename(dir, iname, &fname->cf_name);
2425#endif
2426
2427 return 0;
2428}
2429
2430static inline int ext4_fname_prepare_lookup(struct inode *dir,
2431 struct dentry *dentry,
2432 struct ext4_filename *fname)
2433{
2434 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname);
2435}
2436
2437static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2438{
2439#ifdef CONFIG_UNICODE
2440 kfree(fname->cf_name.name);
2441 fname->cf_name.name = NULL;
2442#endif
2443}
2444#endif /* !CONFIG_FS_ENCRYPTION */
2445
2446/* dir.c */
2447extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2448 struct file *,
2449 struct ext4_dir_entry_2 *,
2450 struct buffer_head *, char *, int,
2451 unsigned int);
2452#define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2453 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2454 (de), (bh), (buf), (size), (offset)))
2455extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2456 __u32 minor_hash,
2457 struct ext4_dir_entry_2 *dirent,
2458 struct fscrypt_str *ent_name);
2459extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2460extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2461 struct buffer_head *bh,
2462 void *buf, int buf_size,
2463 struct ext4_filename *fname,
2464 struct ext4_dir_entry_2 **dest_de);
2465void ext4_insert_dentry(struct inode *inode,
2466 struct ext4_dir_entry_2 *de,
2467 int buf_size,
2468 struct ext4_filename *fname);
2469static inline void ext4_update_dx_flag(struct inode *inode)
2470{
2471 if (!ext4_has_feature_dir_index(inode->i_sb))
2472 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2473}
2474static const unsigned char ext4_filetype_table[] = {
2475 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2476};
2477
2478static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2479{
2480 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2481 return DT_UNKNOWN;
2482
2483 return ext4_filetype_table[filetype];
2484}
2485extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2486 void *buf, int buf_size);
2487
2488/* fsync.c */
2489extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2490
2491/* hash.c */
2492extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len,
2493 struct dx_hash_info *hinfo);
2494
2495/* ialloc.c */
2496extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2497 const struct qstr *qstr, __u32 goal,
2498 uid_t *owner, __u32 i_flags,
2499 int handle_type, unsigned int line_no,
2500 int nblocks);
2501
2502#define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \
2503 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2504 i_flags, 0, 0, 0)
2505#define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2506 type, nblocks) \
2507 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2508 0, (type), __LINE__, (nblocks))
2509
2510
2511extern void ext4_free_inode(handle_t *, struct inode *);
2512extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2513extern unsigned long ext4_count_free_inodes(struct super_block *);
2514extern unsigned long ext4_count_dirs(struct super_block *);
2515extern void ext4_check_inodes_bitmap(struct super_block *);
2516extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2517extern int ext4_init_inode_table(struct super_block *sb,
2518 ext4_group_t group, int barrier);
2519extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2520
2521/* mballoc.c */
2522extern const struct seq_operations ext4_mb_seq_groups_ops;
2523extern long ext4_mb_stats;
2524extern long ext4_mb_max_to_scan;
2525extern int ext4_mb_init(struct super_block *);
2526extern int ext4_mb_release(struct super_block *);
2527extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2528 struct ext4_allocation_request *, int *);
2529extern int ext4_mb_reserve_blocks(struct super_block *, int);
2530extern void ext4_discard_preallocations(struct inode *);
2531extern int __init ext4_init_mballoc(void);
2532extern void ext4_exit_mballoc(void);
2533extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2534 struct buffer_head *bh, ext4_fsblk_t block,
2535 unsigned long count, int flags);
2536extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2537 ext4_group_t ngroups);
2538extern int ext4_mb_add_groupinfo(struct super_block *sb,
2539 ext4_group_t i, struct ext4_group_desc *desc);
2540extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2541 ext4_fsblk_t block, unsigned long count);
2542extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2543extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid);
2544
2545/* inode.c */
2546int ext4_inode_is_fast_symlink(struct inode *inode);
2547struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2548struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2549int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count,
2550 bool wait, struct buffer_head **bhs);
2551int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2552 struct buffer_head *bh_result, int create);
2553int ext4_get_block(struct inode *inode, sector_t iblock,
2554 struct buffer_head *bh_result, int create);
2555int ext4_dio_get_block(struct inode *inode, sector_t iblock,
2556 struct buffer_head *bh_result, int create);
2557int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2558 struct buffer_head *bh, int create);
2559int ext4_walk_page_buffers(handle_t *handle,
2560 struct buffer_head *head,
2561 unsigned from,
2562 unsigned to,
2563 int *partial,
2564 int (*fn)(handle_t *handle,
2565 struct buffer_head *bh));
2566int do_journal_get_write_access(handle_t *handle,
2567 struct buffer_head *bh);
2568#define FALL_BACK_TO_NONDELALLOC 1
2569#define CONVERT_INLINE_DATA 2
2570
2571typedef enum {
2572 EXT4_IGET_NORMAL = 0,
2573 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */
2574 EXT4_IGET_HANDLE = 0x0002 /* Inode # is from a handle */
2575} ext4_iget_flags;
2576
2577extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
2578 ext4_iget_flags flags, const char *function,
2579 unsigned int line);
2580
2581#define ext4_iget(sb, ino, flags) \
2582 __ext4_iget((sb), (ino), (flags), __func__, __LINE__)
2583
2584extern int ext4_write_inode(struct inode *, struct writeback_control *);
2585extern int ext4_setattr(struct dentry *, struct iattr *);
2586extern int ext4_getattr(const struct path *, struct kstat *, u32, unsigned int);
2587extern void ext4_evict_inode(struct inode *);
2588extern void ext4_clear_inode(struct inode *);
2589extern int ext4_file_getattr(const struct path *, struct kstat *, u32, unsigned int);
2590extern int ext4_sync_inode(handle_t *, struct inode *);
2591extern void ext4_dirty_inode(struct inode *, int);
2592extern int ext4_change_inode_journal_flag(struct inode *, int);
2593extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2594extern int ext4_inode_attach_jinode(struct inode *inode);
2595extern int ext4_can_truncate(struct inode *inode);
2596extern int ext4_truncate(struct inode *);
2597extern int ext4_break_layouts(struct inode *);
2598extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2599extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
2600extern void ext4_set_inode_flags(struct inode *);
2601extern int ext4_alloc_da_blocks(struct inode *inode);
2602extern void ext4_set_aops(struct inode *inode);
2603extern int ext4_writepage_trans_blocks(struct inode *);
2604extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2605extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2606 loff_t lstart, loff_t lend);
2607extern int ext4_page_mkwrite(struct vm_fault *vmf);
2608extern int ext4_filemap_fault(struct vm_fault *vmf);
2609extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2610extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
2611extern void ext4_da_update_reserve_space(struct inode *inode,
2612 int used, int quota_claim);
2613extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
2614 ext4_fsblk_t pblk, ext4_lblk_t len);
2615
2616/* indirect.c */
2617extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2618 struct ext4_map_blocks *map, int flags);
2619extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2620extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2621extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2622extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2623 ext4_lblk_t start, ext4_lblk_t end);
2624
2625/* ioctl.c */
2626extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2627extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2628
2629/* migrate.c */
2630extern int ext4_ext_migrate(struct inode *);
2631extern int ext4_ind_migrate(struct inode *inode);
2632
2633/* namei.c */
2634extern int ext4_dirent_csum_verify(struct inode *inode,
2635 struct ext4_dir_entry *dirent);
2636extern int ext4_orphan_add(handle_t *, struct inode *);
2637extern int ext4_orphan_del(handle_t *, struct inode *);
2638extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2639 __u32 start_minor_hash, __u32 *next_hash);
2640extern int ext4_search_dir(struct buffer_head *bh,
2641 char *search_buf,
2642 int buf_size,
2643 struct inode *dir,
2644 struct ext4_filename *fname,
2645 unsigned int offset,
2646 struct ext4_dir_entry_2 **res_dir);
2647extern int ext4_generic_delete_entry(handle_t *handle,
2648 struct inode *dir,
2649 struct ext4_dir_entry_2 *de_del,
2650 struct buffer_head *bh,
2651 void *entry_buf,
2652 int buf_size,
2653 int csum_size);
2654extern bool ext4_empty_dir(struct inode *inode);
2655
2656/* resize.c */
2657extern int ext4_group_add(struct super_block *sb,
2658 struct ext4_new_group_data *input);
2659extern int ext4_group_extend(struct super_block *sb,
2660 struct ext4_super_block *es,
2661 ext4_fsblk_t n_blocks_count);
2662extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2663
2664/* super.c */
2665extern struct buffer_head *ext4_sb_bread(struct super_block *sb,
2666 sector_t block, int op_flags);
2667extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
2668extern int ext4_calculate_overhead(struct super_block *sb);
2669extern void ext4_superblock_csum_set(struct super_block *sb);
2670extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2671extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2672extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2673 ext4_group_t ngroup);
2674extern const char *ext4_decode_error(struct super_block *sb, int errno,
2675 char nbuf[16]);
2676extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb,
2677 ext4_group_t block_group,
2678 unsigned int flags);
2679
2680extern __printf(4, 5)
2681void __ext4_error(struct super_block *, const char *, unsigned int,
2682 const char *, ...);
2683extern __printf(5, 6)
2684void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2685 const char *, ...);
2686extern __printf(5, 6)
2687void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2688 const char *, ...);
2689extern void __ext4_std_error(struct super_block *, const char *,
2690 unsigned int, int);
2691extern __printf(4, 5)
2692void __ext4_abort(struct super_block *, const char *, unsigned int,
2693 const char *, ...);
2694extern __printf(4, 5)
2695void __ext4_warning(struct super_block *, const char *, unsigned int,
2696 const char *, ...);
2697extern __printf(4, 5)
2698void __ext4_warning_inode(const struct inode *inode, const char *function,
2699 unsigned int line, const char *fmt, ...);
2700extern __printf(3, 4)
2701void __ext4_msg(struct super_block *, const char *, const char *, ...);
2702extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2703 const char *, unsigned int, const char *);
2704extern __printf(7, 8)
2705void __ext4_grp_locked_error(const char *, unsigned int,
2706 struct super_block *, ext4_group_t,
2707 unsigned long, ext4_fsblk_t,
2708 const char *, ...);
2709
2710#define EXT4_ERROR_INODE(inode, fmt, a...) \
2711 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
2712
2713#define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
2714 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
2715
2716#define EXT4_ERROR_FILE(file, block, fmt, a...) \
2717 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
2718
2719#ifdef CONFIG_PRINTK
2720
2721#define ext4_error_inode(inode, func, line, block, fmt, ...) \
2722 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__)
2723#define ext4_error_file(file, func, line, block, fmt, ...) \
2724 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2725#define ext4_error(sb, fmt, ...) \
2726 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2727#define ext4_abort(sb, fmt, ...) \
2728 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2729#define ext4_warning(sb, fmt, ...) \
2730 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2731#define ext4_warning_inode(inode, fmt, ...) \
2732 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
2733#define ext4_msg(sb, level, fmt, ...) \
2734 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2735#define dump_mmp_msg(sb, mmp, msg) \
2736 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2737#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2738 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2739 fmt, ##__VA_ARGS__)
2740
2741#else
2742
2743#define ext4_error_inode(inode, func, line, block, fmt, ...) \
2744do { \
2745 no_printk(fmt, ##__VA_ARGS__); \
2746 __ext4_error_inode(inode, "", 0, block, " "); \
2747} while (0)
2748#define ext4_error_file(file, func, line, block, fmt, ...) \
2749do { \
2750 no_printk(fmt, ##__VA_ARGS__); \
2751 __ext4_error_file(file, "", 0, block, " "); \
2752} while (0)
2753#define ext4_error(sb, fmt, ...) \
2754do { \
2755 no_printk(fmt, ##__VA_ARGS__); \
2756 __ext4_error(sb, "", 0, " "); \
2757} while (0)
2758#define ext4_abort(sb, fmt, ...) \
2759do { \
2760 no_printk(fmt, ##__VA_ARGS__); \
2761 __ext4_abort(sb, "", 0, " "); \
2762} while (0)
2763#define ext4_warning(sb, fmt, ...) \
2764do { \
2765 no_printk(fmt, ##__VA_ARGS__); \
2766 __ext4_warning(sb, "", 0, " "); \
2767} while (0)
2768#define ext4_warning_inode(inode, fmt, ...) \
2769do { \
2770 no_printk(fmt, ##__VA_ARGS__); \
2771 __ext4_warning_inode(inode, "", 0, " "); \
2772} while (0)
2773#define ext4_msg(sb, level, fmt, ...) \
2774do { \
2775 no_printk(fmt, ##__VA_ARGS__); \
2776 __ext4_msg(sb, "", " "); \
2777} while (0)
2778#define dump_mmp_msg(sb, mmp, msg) \
2779 __dump_mmp_msg(sb, mmp, "", 0, "")
2780#define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2781do { \
2782 no_printk(fmt, ##__VA_ARGS__); \
2783 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
2784} while (0)
2785
2786#endif
2787
2788extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2789 __u32 compat);
2790extern int ext4_update_rocompat_feature(handle_t *handle,
2791 struct super_block *sb, __u32 rocompat);
2792extern int ext4_update_incompat_feature(handle_t *handle,
2793 struct super_block *sb, __u32 incompat);
2794extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2795 struct ext4_group_desc *bg);
2796extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2797 struct ext4_group_desc *bg);
2798extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2799 struct ext4_group_desc *bg);
2800extern __u32 ext4_free_group_clusters(struct super_block *sb,
2801 struct ext4_group_desc *bg);
2802extern __u32 ext4_free_inodes_count(struct super_block *sb,
2803 struct ext4_group_desc *bg);
2804extern __u32 ext4_used_dirs_count(struct super_block *sb,
2805 struct ext4_group_desc *bg);
2806extern __u32 ext4_itable_unused_count(struct super_block *sb,
2807 struct ext4_group_desc *bg);
2808extern void ext4_block_bitmap_set(struct super_block *sb,
2809 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2810extern void ext4_inode_bitmap_set(struct super_block *sb,
2811 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2812extern void ext4_inode_table_set(struct super_block *sb,
2813 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2814extern void ext4_free_group_clusters_set(struct super_block *sb,
2815 struct ext4_group_desc *bg,
2816 __u32 count);
2817extern void ext4_free_inodes_set(struct super_block *sb,
2818 struct ext4_group_desc *bg, __u32 count);
2819extern void ext4_used_dirs_set(struct super_block *sb,
2820 struct ext4_group_desc *bg, __u32 count);
2821extern void ext4_itable_unused_set(struct super_block *sb,
2822 struct ext4_group_desc *bg, __u32 count);
2823extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2824 struct ext4_group_desc *gdp);
2825extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2826 struct ext4_group_desc *gdp);
2827extern int ext4_register_li_request(struct super_block *sb,
2828 ext4_group_t first_not_zeroed);
2829
2830static inline int ext4_has_metadata_csum(struct super_block *sb)
2831{
2832 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
2833 !EXT4_SB(sb)->s_chksum_driver);
2834
2835 return ext4_has_feature_metadata_csum(sb) &&
2836 (EXT4_SB(sb)->s_chksum_driver != NULL);
2837}
2838
2839static inline int ext4_has_group_desc_csum(struct super_block *sb)
2840{
2841 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb);
2842}
2843
2844static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2845{
2846 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2847 le32_to_cpu(es->s_blocks_count_lo);
2848}
2849
2850static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2851{
2852 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2853 le32_to_cpu(es->s_r_blocks_count_lo);
2854}
2855
2856static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2857{
2858 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2859 le32_to_cpu(es->s_free_blocks_count_lo);
2860}
2861
2862static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2863 ext4_fsblk_t blk)
2864{
2865 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2866 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2867}
2868
2869static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2870 ext4_fsblk_t blk)
2871{
2872 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2873 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2874}
2875
2876static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2877 ext4_fsblk_t blk)
2878{
2879 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2880 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2881}
2882
2883static inline loff_t ext4_isize(struct super_block *sb,
2884 struct ext4_inode *raw_inode)
2885{
2886 if (ext4_has_feature_largedir(sb) ||
2887 S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2888 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2889 le32_to_cpu(raw_inode->i_size_lo);
2890
2891 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2892}
2893
2894static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2895{
2896 raw_inode->i_size_lo = cpu_to_le32(i_size);
2897 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2898}
2899
2900static inline
2901struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2902 ext4_group_t group)
2903{
2904 struct ext4_group_info ***grp_info;
2905 long indexv, indexh;
2906 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
2907 grp_info = EXT4_SB(sb)->s_group_info;
2908 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2909 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2910 return grp_info[indexv][indexh];
2911}
2912
2913/*
2914 * Reading s_groups_count requires using smp_rmb() afterwards. See
2915 * the locking protocol documented in the comments of ext4_group_add()
2916 * in resize.c
2917 */
2918static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2919{
2920 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
2921
2922 smp_rmb();
2923 return ngroups;
2924}
2925
2926static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2927 ext4_group_t block_group)
2928{
2929 return block_group >> sbi->s_log_groups_per_flex;
2930}
2931
2932static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2933{
2934 return 1 << sbi->s_log_groups_per_flex;
2935}
2936
2937#define ext4_std_error(sb, errno) \
2938do { \
2939 if ((errno)) \
2940 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
2941} while (0)
2942
2943#ifdef CONFIG_SMP
2944/* Each CPU can accumulate percpu_counter_batch clusters in their local
2945 * counters. So we need to make sure we have free clusters more
2946 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
2947 */
2948#define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2949#else
2950#define EXT4_FREECLUSTERS_WATERMARK 0
2951#endif
2952
2953/* Update i_disksize. Requires i_mutex to avoid races with truncate */
2954static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2955{
2956 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
2957 !inode_is_locked(inode));
2958 down_write(&EXT4_I(inode)->i_data_sem);
2959 if (newsize > EXT4_I(inode)->i_disksize)
2960 EXT4_I(inode)->i_disksize = newsize;
2961 up_write(&EXT4_I(inode)->i_data_sem);
2962}
2963
2964/* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
2965static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
2966{
2967 int changed = 0;
2968
2969 if (newsize > inode->i_size) {
2970 i_size_write(inode, newsize);
2971 changed = 1;
2972 }
2973 if (newsize > EXT4_I(inode)->i_disksize) {
2974 ext4_update_i_disksize(inode, newsize);
2975 changed |= 2;
2976 }
2977 return changed;
2978}
2979
2980int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
2981 loff_t len);
2982
2983struct ext4_group_info {
2984 unsigned long bb_state;
2985 struct rb_root bb_free_root;
2986 ext4_grpblk_t bb_first_free; /* first free block */
2987 ext4_grpblk_t bb_free; /* total free blocks */
2988 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
2989 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
2990 struct list_head bb_prealloc_list;
2991#ifdef DOUBLE_CHECK
2992 void *bb_bitmap;
2993#endif
2994 struct rw_semaphore alloc_sem;
2995 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
2996 * regions, index is order.
2997 * bb_counters[3] = 5 means
2998 * 5 free 8-block regions. */
2999};
3000
3001#define EXT4_GROUP_INFO_NEED_INIT_BIT 0
3002#define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
3003#define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
3004#define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
3005#define EXT4_GROUP_INFO_BBITMAP_CORRUPT \
3006 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT)
3007#define EXT4_GROUP_INFO_IBITMAP_CORRUPT \
3008 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT)
3009
3010#define EXT4_MB_GRP_NEED_INIT(grp) \
3011 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
3012#define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
3013 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3014#define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
3015 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
3016
3017#define EXT4_MB_GRP_WAS_TRIMMED(grp) \
3018 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3019#define EXT4_MB_GRP_SET_TRIMMED(grp) \
3020 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3021#define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
3022 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
3023
3024#define EXT4_MAX_CONTENTION 8
3025#define EXT4_CONTENTION_THRESHOLD 2
3026
3027static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
3028 ext4_group_t group)
3029{
3030 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
3031}
3032
3033/*
3034 * Returns true if the filesystem is busy enough that attempts to
3035 * access the block group locks has run into contention.
3036 */
3037static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
3038{
3039 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
3040}
3041
3042static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
3043{
3044 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
3045 if (spin_trylock(lock))
3046 /*
3047 * We're able to grab the lock right away, so drop the
3048 * lock contention counter.
3049 */
3050 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
3051 else {
3052 /*
3053 * The lock is busy, so bump the contention counter,
3054 * and then wait on the spin lock.
3055 */
3056 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
3057 EXT4_MAX_CONTENTION);
3058 spin_lock(lock);
3059 }
3060}
3061
3062static inline void ext4_unlock_group(struct super_block *sb,
3063 ext4_group_t group)
3064{
3065 spin_unlock(ext4_group_lock_ptr(sb, group));
3066}
3067
3068/*
3069 * Block validity checking
3070 */
3071#define ext4_check_indirect_blockref(inode, bh) \
3072 ext4_check_blockref(__func__, __LINE__, inode, \
3073 (__le32 *)(bh)->b_data, \
3074 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
3075
3076#define ext4_ind_check_inode(inode) \
3077 ext4_check_blockref(__func__, __LINE__, inode, \
3078 EXT4_I(inode)->i_data, \
3079 EXT4_NDIR_BLOCKS)
3080
3081/*
3082 * Inodes and files operations
3083 */
3084
3085/* dir.c */
3086extern const struct file_operations ext4_dir_operations;
3087
3088#ifdef CONFIG_UNICODE
3089extern const struct dentry_operations ext4_dentry_ops;
3090#endif
3091
3092/* file.c */
3093extern const struct inode_operations ext4_file_inode_operations;
3094extern const struct file_operations ext4_file_operations;
3095extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
3096
3097/* inline.c */
3098extern int ext4_get_max_inline_size(struct inode *inode);
3099extern int ext4_find_inline_data_nolock(struct inode *inode);
3100extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
3101 unsigned int len);
3102extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
3103
3104extern int ext4_readpage_inline(struct inode *inode, struct page *page);
3105extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3106 struct inode *inode,
3107 loff_t pos, unsigned len,
3108 unsigned flags,
3109 struct page **pagep);
3110extern int ext4_write_inline_data_end(struct inode *inode,
3111 loff_t pos, unsigned len,
3112 unsigned copied,
3113 struct page *page);
3114extern struct buffer_head *
3115ext4_journalled_write_inline_data(struct inode *inode,
3116 unsigned len,
3117 struct page *page);
3118extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
3119 struct inode *inode,
3120 loff_t pos, unsigned len,
3121 unsigned flags,
3122 struct page **pagep,
3123 void **fsdata);
3124extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
3125 unsigned len, unsigned copied,
3126 struct page *page);
3127extern int ext4_try_add_inline_entry(handle_t *handle,
3128 struct ext4_filename *fname,
3129 struct inode *dir, struct inode *inode);
3130extern int ext4_try_create_inline_dir(handle_t *handle,
3131 struct inode *parent,
3132 struct inode *inode);
3133extern int ext4_read_inline_dir(struct file *filp,
3134 struct dir_context *ctx,
3135 int *has_inline_data);
3136extern int htree_inlinedir_to_tree(struct file *dir_file,
3137 struct inode *dir, ext4_lblk_t block,
3138 struct dx_hash_info *hinfo,
3139 __u32 start_hash, __u32 start_minor_hash,
3140 int *has_inline_data);
3141extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3142 struct ext4_filename *fname,
3143 struct ext4_dir_entry_2 **res_dir,
3144 int *has_inline_data);
3145extern int ext4_delete_inline_entry(handle_t *handle,
3146 struct inode *dir,
3147 struct ext4_dir_entry_2 *de_del,
3148 struct buffer_head *bh,
3149 int *has_inline_data);
3150extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3151extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3152 struct ext4_dir_entry_2 **parent_de,
3153 int *retval);
3154extern int ext4_inline_data_fiemap(struct inode *inode,
3155 struct fiemap_extent_info *fieinfo,
3156 int *has_inline, __u64 start, __u64 len);
3157
3158struct iomap;
3159extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap);
3160
3161extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3162
3163extern int ext4_convert_inline_data(struct inode *inode);
3164
3165static inline int ext4_has_inline_data(struct inode *inode)
3166{
3167 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3168 EXT4_I(inode)->i_inline_off;
3169}
3170
3171/* namei.c */
3172extern const struct inode_operations ext4_dir_inode_operations;
3173extern const struct inode_operations ext4_special_inode_operations;
3174extern struct dentry *ext4_get_parent(struct dentry *child);
3175extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3176 struct ext4_dir_entry_2 *de,
3177 int blocksize, int csum_size,
3178 unsigned int parent_ino, int dotdot_real_len);
3179extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t,
3180 unsigned int blocksize);
3181extern int ext4_handle_dirty_dirent_node(handle_t *handle,
3182 struct inode *inode,
3183 struct buffer_head *bh);
3184extern int ext4_ci_compare(const struct inode *parent,
3185 const struct qstr *fname,
3186 const struct qstr *entry, bool quick);
3187
3188#define S_SHIFT 12
3189static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = {
3190 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
3191 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
3192 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
3193 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
3194 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
3195 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
3196 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
3197};
3198
3199static inline void ext4_set_de_type(struct super_block *sb,
3200 struct ext4_dir_entry_2 *de,
3201 umode_t mode) {
3202 if (ext4_has_feature_filetype(sb))
3203 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3204}
3205
3206/* readpages.c */
3207extern int ext4_mpage_readpages(struct address_space *mapping,
3208 struct list_head *pages, struct page *page,
3209 unsigned nr_pages, bool is_readahead);
3210extern int __init ext4_init_post_read_processing(void);
3211extern void ext4_exit_post_read_processing(void);
3212
3213/* symlink.c */
3214extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3215extern const struct inode_operations ext4_symlink_inode_operations;
3216extern const struct inode_operations ext4_fast_symlink_inode_operations;
3217
3218/* sysfs.c */
3219extern int ext4_register_sysfs(struct super_block *sb);
3220extern void ext4_unregister_sysfs(struct super_block *sb);
3221extern int __init ext4_init_sysfs(void);
3222extern void ext4_exit_sysfs(void);
3223
3224/* block_validity */
3225extern void ext4_release_system_zone(struct super_block *sb);
3226extern int ext4_setup_system_zone(struct super_block *sb);
3227extern int __init ext4_init_system_zone(void);
3228extern void ext4_exit_system_zone(void);
3229extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
3230 ext4_fsblk_t start_blk,
3231 unsigned int count);
3232extern int ext4_check_blockref(const char *, unsigned int,
3233 struct inode *, __le32 *, unsigned int);
3234
3235/* extents.c */
3236struct ext4_ext_path;
3237struct ext4_extent;
3238
3239/*
3240 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3241 * __le32.
3242 */
3243#define EXT_MAX_BLOCKS 0xffffffff
3244
3245extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
3246extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
3247extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3248extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3249 struct ext4_map_blocks *map, int flags);
3250extern int ext4_ext_truncate(handle_t *, struct inode *);
3251extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3252 ext4_lblk_t end);
3253extern void ext4_ext_init(struct super_block *);
3254extern void ext4_ext_release(struct super_block *);
3255extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3256 loff_t len);
3257extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3258 loff_t offset, ssize_t len);
3259extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3260 struct ext4_map_blocks *map, int flags);
3261extern int ext4_ext_calc_metadata_amount(struct inode *inode,
3262 ext4_lblk_t lblocks);
3263extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3264 int num,
3265 struct ext4_ext_path *path);
3266extern int ext4_can_extents_be_merged(struct inode *inode,
3267 struct ext4_extent *ex1,
3268 struct ext4_extent *ex2);
3269extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3270 struct ext4_ext_path **,
3271 struct ext4_extent *, int);
3272extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3273 struct ext4_ext_path **,
3274 int flags);
3275extern void ext4_ext_drop_refs(struct ext4_ext_path *);
3276extern int ext4_ext_check_inode(struct inode *inode);
3277extern int ext4_find_delalloc_range(struct inode *inode,
3278 ext4_lblk_t lblk_start,
3279 ext4_lblk_t lblk_end);
3280extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk);
3281extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3282extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3283 __u64 start, __u64 len);
3284extern int ext4_ext_precache(struct inode *inode);
3285extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len);
3286extern int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len);
3287extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3288 struct inode *inode2, ext4_lblk_t lblk1,
3289 ext4_lblk_t lblk2, ext4_lblk_t count,
3290 int mark_unwritten,int *err);
3291
3292/* move_extent.c */
3293extern void ext4_double_down_write_data_sem(struct inode *first,
3294 struct inode *second);
3295extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3296 struct inode *donor_inode);
3297extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3298 __u64 start_orig, __u64 start_donor,
3299 __u64 len, __u64 *moved_len);
3300
3301/* page-io.c */
3302extern int __init ext4_init_pageio(void);
3303extern void ext4_exit_pageio(void);
3304extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3305extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3306extern int ext4_put_io_end(ext4_io_end_t *io_end);
3307extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3308extern void ext4_io_submit_init(struct ext4_io_submit *io,
3309 struct writeback_control *wbc);
3310extern void ext4_end_io_rsv_work(struct work_struct *work);
3311extern void ext4_io_submit(struct ext4_io_submit *io);
3312extern int ext4_bio_write_page(struct ext4_io_submit *io,
3313 struct page *page,
3314 int len,
3315 struct writeback_control *wbc,
3316 bool keep_towrite);
3317
3318/* mmp.c */
3319extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3320
3321/* verity.c */
3322extern const struct fsverity_operations ext4_verityops;
3323
3324/*
3325 * Add new method to test whether block and inode bitmaps are properly
3326 * initialized. With uninit_bg reading the block from disk is not enough
3327 * to mark the bitmap uptodate. We need to also zero-out the bitmap
3328 */
3329#define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3330
3331static inline int bitmap_uptodate(struct buffer_head *bh)
3332{
3333 return (buffer_uptodate(bh) &&
3334 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3335}
3336static inline void set_bitmap_uptodate(struct buffer_head *bh)
3337{
3338 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3339}
3340
3341#define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
3342
3343/* For ioend & aio unwritten conversion wait queues */
3344#define EXT4_WQ_HASH_SZ 37
3345#define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
3346 EXT4_WQ_HASH_SZ])
3347extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3348
3349extern int ext4_resize_begin(struct super_block *sb);
3350extern void ext4_resize_end(struct super_block *sb);
3351
3352static inline void ext4_set_io_unwritten_flag(struct inode *inode,
3353 struct ext4_io_end *io_end)
3354{
3355 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
3356 io_end->flag |= EXT4_IO_END_UNWRITTEN;
3357 atomic_inc(&EXT4_I(inode)->i_unwritten);
3358 }
3359}
3360
3361static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3362{
3363 struct inode *inode = io_end->inode;
3364
3365 if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
3366 io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3367 /* Wake up anyone waiting on unwritten extent conversion */
3368 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
3369 wake_up_all(ext4_ioend_wq(inode));
3370 }
3371}
3372
3373extern const struct iomap_ops ext4_iomap_ops;
3374
3375#endif /* __KERNEL__ */
3376
3377#define EFSBADCRC EBADMSG /* Bad CRC detected */
3378#define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
3379
3380#endif /* _EXT4_H */