1 // SPDX-License-Identifier: GPL-2.0
2 /*
3 * ext4.h
4 *
5 * Copyright (C) 1992, 1993, 1994, 1995
6 * Remy Card (card@masi.ibp.fr)
7 * Laboratoire MASI - Institut Blaise Pascal
8 * Universite Pierre et Marie Curie (Paris VI)
9 *
10 * from
11 *
12 * linux/include/linux/minix_fs.h
13 *
14 * Copyright (C) 1991, 1992 Linus Torvalds
15 */
16
17 #ifndef _EXT4_H
18 #define _EXT4_H
19
20 #include <linux/types.h>
21 #include <linux/blkdev.h>
22 #include <linux/magic.h>
23 #include <linux/jbd2.h>
24 #include <linux/quota.h>
25 #include <linux/rwsem.h>
26 #include <linux/rbtree.h>
27 #include <linux/seqlock.h>
28 #include <linux/mutex.h>
29 #include <linux/timer.h>
30 #include <linux/version.h>
31 #include <linux/wait.h>
32 #include <linux/sched/signal.h>
33 #include <linux/blockgroup_lock.h>
34 #include <linux/percpu_counter.h>
35 #include <linux/ratelimit.h>
36 #include <crypto/hash.h>
37 #include <linux/falloc.h>
38 #include <linux/percpu-rwsem.h>
39 #ifdef __KERNEL__
40 #include <linux/compat.h>
41 #endif
42
43 #define __FS_HAS_ENCRYPTION IS_ENABLED(CONFIG_EXT4_FS_ENCRYPTION)
44 #include <linux/fscrypt.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 */
103 typedef int ext4_grpblk_t;
104
105 /* data type for filesystem-wide blocks number */
106 typedef unsigned long long ext4_fsblk_t;
107
108 /* data type for file logical block number */
109 typedef __u32 ext4_lblk_t;
110
111 /* data type for block group number */
112 typedef unsigned int ext4_group_t;
113
114 enum 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
155 struct 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
190 struct 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 */
200 struct 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 */
214 typedef 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
227 struct 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 */
314 struct 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
352 struct 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_EA_INODE_FL 0x00200000 /* Inode used for large EA */
417 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
418 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
419 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
420 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
421
422 #define EXT4_FL_USER_VISIBLE 0x304BDFFF /* User visible flags */
423 #define EXT4_FL_USER_MODIFIABLE 0x204BC0FF /* User modifiable flags */
424
425 /* Flags we can manipulate with through EXT4_IOC_FSSETXATTR */
426 #define EXT4_FL_XFLAG_VISIBLE (EXT4_SYNC_FL | \
427 EXT4_IMMUTABLE_FL | \
428 EXT4_APPEND_FL | \
429 EXT4_NODUMP_FL | \
430 EXT4_NOATIME_FL | \
431 EXT4_PROJINHERIT_FL)
432
433 /* Flags that should be inherited by new inodes from their parent. */
434 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
435 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
436 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
437 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
438 EXT4_PROJINHERIT_FL)
439
440 /* Flags that are appropriate for regular files (all but dir-specific ones). */
441 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
442
443 /* Flags that are appropriate for non-directories/regular files. */
444 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
445
446 /* The only flags that should be swapped */
447 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL)
448
449 /* Mask out flags that are inappropriate for the given type of inode. */
ext4_mask_flags(umode_t mode,__u32 flags)450 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
451 {
452 if (S_ISDIR(mode))
453 return flags;
454 else if (S_ISREG(mode))
455 return flags & EXT4_REG_FLMASK;
456 else
457 return flags & EXT4_OTHER_FLMASK;
458 }
459
460 /*
461 * Inode flags used for atomic set/get
462 */
463 enum {
464 EXT4_INODE_SECRM = 0, /* Secure deletion */
465 EXT4_INODE_UNRM = 1, /* Undelete */
466 EXT4_INODE_COMPR = 2, /* Compress file */
467 EXT4_INODE_SYNC = 3, /* Synchronous updates */
468 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
469 EXT4_INODE_APPEND = 5, /* writes to file may only append */
470 EXT4_INODE_NODUMP = 6, /* do not dump file */
471 EXT4_INODE_NOATIME = 7, /* do not update atime */
472 /* Reserved for compression usage... */
473 EXT4_INODE_DIRTY = 8,
474 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
475 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
476 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */
477 /* End compression flags --- maybe not all used */
478 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
479 EXT4_INODE_IMAGIC = 13, /* AFS directory */
480 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
481 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
482 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
483 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
484 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
485 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
486 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
487 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
488 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
489 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */
490 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
491 };
492
493 /*
494 * Since it's pretty easy to mix up bit numbers and hex values, we use a
495 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
496 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
497 * any extra space in the compiled kernel image, otherwise, the build will fail.
498 * It's important that these values are the same, since we are using
499 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
500 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
501 * values found in ext2, ext3 and ext4 filesystems, and of course the values
502 * defined in e2fsprogs.
503 *
504 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
505 */
506 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1U << EXT4_INODE_##FLAG))
507 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
508
ext4_check_flag_values(void)509 static inline void ext4_check_flag_values(void)
510 {
511 CHECK_FLAG_VALUE(SECRM);
512 CHECK_FLAG_VALUE(UNRM);
513 CHECK_FLAG_VALUE(COMPR);
514 CHECK_FLAG_VALUE(SYNC);
515 CHECK_FLAG_VALUE(IMMUTABLE);
516 CHECK_FLAG_VALUE(APPEND);
517 CHECK_FLAG_VALUE(NODUMP);
518 CHECK_FLAG_VALUE(NOATIME);
519 CHECK_FLAG_VALUE(DIRTY);
520 CHECK_FLAG_VALUE(COMPRBLK);
521 CHECK_FLAG_VALUE(NOCOMPR);
522 CHECK_FLAG_VALUE(ENCRYPT);
523 CHECK_FLAG_VALUE(INDEX);
524 CHECK_FLAG_VALUE(IMAGIC);
525 CHECK_FLAG_VALUE(JOURNAL_DATA);
526 CHECK_FLAG_VALUE(NOTAIL);
527 CHECK_FLAG_VALUE(DIRSYNC);
528 CHECK_FLAG_VALUE(TOPDIR);
529 CHECK_FLAG_VALUE(HUGE_FILE);
530 CHECK_FLAG_VALUE(EXTENTS);
531 CHECK_FLAG_VALUE(EA_INODE);
532 CHECK_FLAG_VALUE(EOFBLOCKS);
533 CHECK_FLAG_VALUE(INLINE_DATA);
534 CHECK_FLAG_VALUE(PROJINHERIT);
535 CHECK_FLAG_VALUE(RESERVED);
536 }
537
538 /* Used to pass group descriptor data when online resize is done */
539 struct ext4_new_group_input {
540 __u32 group; /* Group number for this data */
541 __u64 block_bitmap; /* Absolute block number of block bitmap */
542 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
543 __u64 inode_table; /* Absolute block number of inode table start */
544 __u32 blocks_count; /* Total number of blocks in this group */
545 __u16 reserved_blocks; /* Number of reserved blocks in this group */
546 __u16 unused;
547 };
548
549 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
550 struct compat_ext4_new_group_input {
551 u32 group;
552 compat_u64 block_bitmap;
553 compat_u64 inode_bitmap;
554 compat_u64 inode_table;
555 u32 blocks_count;
556 u16 reserved_blocks;
557 u16 unused;
558 };
559 #endif
560
561 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
562 struct ext4_new_group_data {
563 __u32 group;
564 __u64 block_bitmap;
565 __u64 inode_bitmap;
566 __u64 inode_table;
567 __u32 blocks_count;
568 __u16 reserved_blocks;
569 __u16 mdata_blocks;
570 __u32 free_clusters_count;
571 };
572
573 /* Indexes used to index group tables in ext4_new_group_data */
574 enum {
575 BLOCK_BITMAP = 0, /* block bitmap */
576 INODE_BITMAP, /* inode bitmap */
577 INODE_TABLE, /* inode tables */
578 GROUP_TABLE_COUNT,
579 };
580
581 /*
582 * Flags used by ext4_map_blocks()
583 */
584 /* Allocate any needed blocks and/or convert an unwritten
585 extent to be an initialized ext4 */
586 #define EXT4_GET_BLOCKS_CREATE 0x0001
587 /* Request the creation of an unwritten extent */
588 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002
589 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\
590 EXT4_GET_BLOCKS_CREATE)
591 /* Caller is from the delayed allocation writeout path
592 * finally doing the actual allocation of delayed blocks */
593 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
594 /* caller is from the direct IO path, request to creation of an
595 unwritten extents if not allocated, split the unwritten
596 extent if blocks has been preallocated already*/
597 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
598 #define EXT4_GET_BLOCKS_CONVERT 0x0010
599 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
600 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
601 /* Convert extent to initialized after IO complete */
602 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
603 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
604 /* Eventual metadata allocation (due to growing extent tree)
605 * should not fail, so try to use reserved blocks for that.*/
606 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
607 /* Don't normalize allocation size (used for fallocate) */
608 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
609 /* Request will not result in inode size update (user for fallocate) */
610 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080
611 /* Convert written extents to unwritten */
612 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100
613 /* Write zeros to newly created written extents */
614 #define EXT4_GET_BLOCKS_ZERO 0x0200
615 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\
616 EXT4_GET_BLOCKS_ZERO)
617 /* Caller will submit data before dropping transaction handle. This
618 * allows jbd2 to avoid submitting data before commit. */
619 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400
620
621 /*
622 * The bit position of these flags must not overlap with any of the
623 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(),
624 * read_extent_tree_block(), ext4_split_extent_at(),
625 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
626 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
627 * caching the extents when reading from the extent tree while a
628 * truncate or punch hole operation is in progress.
629 */
630 #define EXT4_EX_NOCACHE 0x40000000
631 #define EXT4_EX_FORCE_CACHE 0x20000000
632
633 /*
634 * Flags used by ext4_free_blocks
635 */
636 #define EXT4_FREE_BLOCKS_METADATA 0x0001
637 #define EXT4_FREE_BLOCKS_FORGET 0x0002
638 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
639 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
640 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
641 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
642
643 /*
644 * ioctl commands
645 */
646 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
647 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
648 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
649 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
650 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
651 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
652 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
653 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
654 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
655 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
656 #define EXT4_IOC_MIGRATE _IO('f', 9)
657 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
658 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
659 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
660 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
661 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
662 #define EXT4_IOC_SWAP_BOOT _IO('f', 17)
663 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18)
664 #define EXT4_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY
665 #define EXT4_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT
666 #define EXT4_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY
667
668 #define EXT4_IOC_FSGETXATTR FS_IOC_FSGETXATTR
669 #define EXT4_IOC_FSSETXATTR FS_IOC_FSSETXATTR
670
671 #define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32)
672
673 /*
674 * Flags for going down operation
675 */
676 #define EXT4_GOING_FLAGS_DEFAULT 0x0 /* going down */
677 #define EXT4_GOING_FLAGS_LOGFLUSH 0x1 /* flush log but not data */
678 #define EXT4_GOING_FLAGS_NOLOGFLUSH 0x2 /* don't flush log nor data */
679
680
681 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
682 /*
683 * ioctl commands in 32 bit emulation
684 */
685 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
686 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
687 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
688 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
689 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
690 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
691 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
692 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
693 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
694 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
695 #endif
696
697 /* Max physical block we can address w/o extents */
698 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
699
700 /* Max logical block we can support */
701 #define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFF
702
703 /*
704 * Structure of an inode on the disk
705 */
706 struct ext4_inode {
707 __le16 i_mode; /* File mode */
708 __le16 i_uid; /* Low 16 bits of Owner Uid */
709 __le32 i_size_lo; /* Size in bytes */
710 __le32 i_atime; /* Access time */
711 __le32 i_ctime; /* Inode Change time */
712 __le32 i_mtime; /* Modification time */
713 __le32 i_dtime; /* Deletion Time */
714 __le16 i_gid; /* Low 16 bits of Group Id */
715 __le16 i_links_count; /* Links count */
716 __le32 i_blocks_lo; /* Blocks count */
717 __le32 i_flags; /* File flags */
718 union {
719 struct {
720 __le32 l_i_version;
721 } linux1;
722 struct {
723 __u32 h_i_translator;
724 } hurd1;
725 struct {
726 __u32 m_i_reserved1;
727 } masix1;
728 } osd1; /* OS dependent 1 */
729 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
730 __le32 i_generation; /* File version (for NFS) */
731 __le32 i_file_acl_lo; /* File ACL */
732 __le32 i_size_high;
733 __le32 i_obso_faddr; /* Obsoleted fragment address */
734 union {
735 struct {
736 __le16 l_i_blocks_high; /* were l_i_reserved1 */
737 __le16 l_i_file_acl_high;
738 __le16 l_i_uid_high; /* these 2 fields */
739 __le16 l_i_gid_high; /* were reserved2[0] */
740 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
741 __le16 l_i_reserved;
742 } linux2;
743 struct {
744 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
745 __u16 h_i_mode_high;
746 __u16 h_i_uid_high;
747 __u16 h_i_gid_high;
748 __u32 h_i_author;
749 } hurd2;
750 struct {
751 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
752 __le16 m_i_file_acl_high;
753 __u32 m_i_reserved2[2];
754 } masix2;
755 } osd2; /* OS dependent 2 */
756 __le16 i_extra_isize;
757 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
758 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
759 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
760 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
761 __le32 i_crtime; /* File Creation time */
762 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
763 __le32 i_version_hi; /* high 32 bits for 64-bit version */
764 __le32 i_projid; /* Project ID */
765 };
766
767 struct move_extent {
768 __u32 reserved; /* should be zero */
769 __u32 donor_fd; /* donor file descriptor */
770 __u64 orig_start; /* logical start offset in block for orig */
771 __u64 donor_start; /* logical start offset in block for donor */
772 __u64 len; /* block length to be moved */
773 __u64 moved_len; /* moved block length */
774 };
775
776 #define EXT4_EPOCH_BITS 2
777 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
778 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
779
780 /*
781 * Extended fields will fit into an inode if the filesystem was formatted
782 * with large inodes (-I 256 or larger) and there are not currently any EAs
783 * consuming all of the available space. For new inodes we always reserve
784 * enough space for the kernel's known extended fields, but for inodes
785 * created with an old kernel this might not have been the case. None of
786 * the extended inode fields is critical for correct filesystem operation.
787 * This macro checks if a certain field fits in the inode. Note that
788 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
789 */
790 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
791 ((offsetof(typeof(*ext4_inode), field) + \
792 sizeof((ext4_inode)->field)) \
793 <= (EXT4_GOOD_OLD_INODE_SIZE + \
794 (einode)->i_extra_isize)) \
795
796 /*
797 * We use an encoding that preserves the times for extra epoch "00":
798 *
799 * extra msb of adjust for signed
800 * epoch 32-bit 32-bit tv_sec to
801 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
802 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
803 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
804 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
805 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
806 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
807 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
808 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
809 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
810 *
811 * Note that previous versions of the kernel on 64-bit systems would
812 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
813 * 1970. e2fsck will correct this, assuming that it is run on the
814 * affected filesystem before 2242.
815 */
816
ext4_encode_extra_time(struct timespec64 * time)817 static inline __le32 ext4_encode_extra_time(struct timespec64 *time)
818 {
819 u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK;
820 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
821 }
822
ext4_decode_extra_time(struct timespec64 * time,__le32 extra)823 static inline void ext4_decode_extra_time(struct timespec64 *time,
824 __le32 extra)
825 {
826 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK))) {
827
828 #if 1
829 /* Handle legacy encoding of pre-1970 dates with epoch
830 * bits 1,1. (This backwards compatibility may be removed
831 * at the discretion of the ext4 developers.)
832 */
833 u64 extra_bits = le32_to_cpu(extra) & EXT4_EPOCH_MASK;
834 if (extra_bits == 3 && ((time->tv_sec) & 0x80000000) != 0)
835 extra_bits = 0;
836 time->tv_sec += extra_bits << 32;
837 #else
838 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
839 #endif
840 }
841 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
842 }
843
844 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
845 do { \
846 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
847 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {\
848 (raw_inode)->xtime ## _extra = \
849 ext4_encode_extra_time(&(inode)->xtime); \
850 } \
851 } while (0)
852
853 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
854 do { \
855 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
856 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
857 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
858 (raw_inode)->xtime ## _extra = \
859 ext4_encode_extra_time(&(einode)->xtime); \
860 } while (0)
861
862 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
863 do { \
864 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
865 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \
866 ext4_decode_extra_time(&(inode)->xtime, \
867 raw_inode->xtime ## _extra); \
868 } \
869 else \
870 (inode)->xtime.tv_nsec = 0; \
871 } while (0)
872
873
874 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
875 do { \
876 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
877 (einode)->xtime.tv_sec = \
878 (signed)le32_to_cpu((raw_inode)->xtime); \
879 else \
880 (einode)->xtime.tv_sec = 0; \
881 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
882 ext4_decode_extra_time(&(einode)->xtime, \
883 raw_inode->xtime ## _extra); \
884 else \
885 (einode)->xtime.tv_nsec = 0; \
886 } while (0)
887
888 #define i_disk_version osd1.linux1.l_i_version
889
890 #if defined(__KERNEL__) || defined(__linux__)
891 #define i_reserved1 osd1.linux1.l_i_reserved1
892 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
893 #define i_blocks_high osd2.linux2.l_i_blocks_high
894 #define i_uid_low i_uid
895 #define i_gid_low i_gid
896 #define i_uid_high osd2.linux2.l_i_uid_high
897 #define i_gid_high osd2.linux2.l_i_gid_high
898 #define i_checksum_lo osd2.linux2.l_i_checksum_lo
899
900 #elif defined(__GNU__)
901
902 #define i_translator osd1.hurd1.h_i_translator
903 #define i_uid_high osd2.hurd2.h_i_uid_high
904 #define i_gid_high osd2.hurd2.h_i_gid_high
905 #define i_author osd2.hurd2.h_i_author
906
907 #elif defined(__masix__)
908
909 #define i_reserved1 osd1.masix1.m_i_reserved1
910 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
911 #define i_reserved2 osd2.masix2.m_i_reserved2
912
913 #endif /* defined(__KERNEL__) || defined(__linux__) */
914
915 #include "extents_status.h"
916
917 /*
918 * Lock subclasses for i_data_sem in the ext4_inode_info structure.
919 *
920 * These are needed to avoid lockdep false positives when we need to
921 * allocate blocks to the quota inode during ext4_map_blocks(), while
922 * holding i_data_sem for a normal (non-quota) inode. Since we don't
923 * do quota tracking for the quota inode, this avoids deadlock (as
924 * well as infinite recursion, since it isn't turtles all the way
925 * down...)
926 *
927 * I_DATA_SEM_NORMAL - Used for most inodes
928 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode
929 * where the second inode has larger inode number
930 * than the first
931 * I_DATA_SEM_QUOTA - Used for quota inodes only
932 * I_DATA_SEM_EA - Used for ea_inodes only
933 */
934 enum {
935 I_DATA_SEM_NORMAL = 0,
936 I_DATA_SEM_OTHER,
937 I_DATA_SEM_QUOTA,
938 I_DATA_SEM_EA
939 };
940
941
942 /*
943 * fourth extended file system inode data in memory
944 */
945 struct ext4_inode_info {
946 __le32 i_data[15]; /* unconverted */
947 __u32 i_dtime;
948 ext4_fsblk_t i_file_acl;
949
950 /*
951 * i_block_group is the number of the block group which contains
952 * this file's inode. Constant across the lifetime of the inode,
953 * it is used for making block allocation decisions - we try to
954 * place a file's data blocks near its inode block, and new inodes
955 * near to their parent directory's inode.
956 */
957 ext4_group_t i_block_group;
958 ext4_lblk_t i_dir_start_lookup;
959 #if (BITS_PER_LONG < 64)
960 unsigned long i_state_flags; /* Dynamic state flags */
961 #endif
962 unsigned long i_flags;
963
964 /*
965 * Extended attributes can be read independently of the main file
966 * data. Taking i_mutex even when reading would cause contention
967 * between readers of EAs and writers of regular file data, so
968 * instead we synchronize on xattr_sem when reading or changing
969 * EAs.
970 */
971 struct rw_semaphore xattr_sem;
972
973 struct list_head i_orphan; /* unlinked but open inodes */
974
975 /*
976 * i_disksize keeps track of what the inode size is ON DISK, not
977 * in memory. During truncate, i_size is set to the new size by
978 * the VFS prior to calling ext4_truncate(), but the filesystem won't
979 * set i_disksize to 0 until the truncate is actually under way.
980 *
981 * The intent is that i_disksize always represents the blocks which
982 * are used by this file. This allows recovery to restart truncate
983 * on orphans if we crash during truncate. We actually write i_disksize
984 * into the on-disk inode when writing inodes out, instead of i_size.
985 *
986 * The only time when i_disksize and i_size may be different is when
987 * a truncate is in progress. The only things which change i_disksize
988 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
989 */
990 loff_t i_disksize;
991
992 /*
993 * i_data_sem is for serialising ext4_truncate() against
994 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
995 * data tree are chopped off during truncate. We can't do that in
996 * ext4 because whenever we perform intermediate commits during
997 * truncate, the inode and all the metadata blocks *must* be in a
998 * consistent state which allows truncation of the orphans to restart
999 * during recovery. Hence we must fix the get_block-vs-truncate race
1000 * by other means, so we have i_data_sem.
1001 */
1002 struct rw_semaphore i_data_sem;
1003 /*
1004 * i_mmap_sem is for serializing page faults with truncate / punch hole
1005 * operations. We have to make sure that new page cannot be faulted in
1006 * a section of the inode that is being punched. We cannot easily use
1007 * i_data_sem for this since we need protection for the whole punch
1008 * operation and i_data_sem ranks below transaction start so we have
1009 * to occasionally drop it.
1010 */
1011 struct rw_semaphore i_mmap_sem;
1012 struct inode vfs_inode;
1013 struct jbd2_inode *jinode;
1014
1015 spinlock_t i_raw_lock; /* protects updates to the raw inode */
1016
1017 /*
1018 * File creation time. Its function is same as that of
1019 * struct timespec64 i_{a,c,m}time in the generic inode.
1020 */
1021 struct timespec64 i_crtime;
1022
1023 /* mballoc */
1024 struct list_head i_prealloc_list;
1025 spinlock_t i_prealloc_lock;
1026
1027 /* extents status tree */
1028 struct ext4_es_tree i_es_tree;
1029 rwlock_t i_es_lock;
1030 struct list_head i_es_list;
1031 unsigned int i_es_all_nr; /* protected by i_es_lock */
1032 unsigned int i_es_shk_nr; /* protected by i_es_lock */
1033 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for
1034 extents to shrink. Protected by
1035 i_es_lock */
1036
1037 /* ialloc */
1038 ext4_group_t i_last_alloc_group;
1039
1040 /* allocation reservation info for delalloc */
1041 /* In case of bigalloc, this refer to clusters rather than blocks */
1042 unsigned int i_reserved_data_blocks;
1043 ext4_lblk_t i_da_metadata_calc_last_lblock;
1044 int i_da_metadata_calc_len;
1045
1046 /* on-disk additional length */
1047 __u16 i_extra_isize;
1048
1049 /* Indicate the inline data space. */
1050 u16 i_inline_off;
1051 u16 i_inline_size;
1052
1053 #ifdef CONFIG_QUOTA
1054 /* quota space reservation, managed internally by quota code */
1055 qsize_t i_reserved_quota;
1056 #endif
1057
1058 /* Lock protecting lists below */
1059 spinlock_t i_completed_io_lock;
1060 /*
1061 * Completed IOs that need unwritten extents handling and have
1062 * transaction reserved
1063 */
1064 struct list_head i_rsv_conversion_list;
1065 struct work_struct i_rsv_conversion_work;
1066 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1067
1068 spinlock_t i_block_reservation_lock;
1069
1070 /*
1071 * Transactions that contain inode's metadata needed to complete
1072 * fsync and fdatasync, respectively.
1073 */
1074 tid_t i_sync_tid;
1075 tid_t i_datasync_tid;
1076
1077 #ifdef CONFIG_QUOTA
1078 struct dquot *i_dquot[MAXQUOTAS];
1079 #endif
1080
1081 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1082 __u32 i_csum_seed;
1083
1084 kprojid_t i_projid;
1085 };
1086
1087 /*
1088 * File system states
1089 */
1090 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
1091 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
1092 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
1093
1094 /*
1095 * Misc. filesystem flags
1096 */
1097 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
1098 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
1099 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
1100
1101 /*
1102 * Mount flags set via mount options or defaults
1103 */
1104 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */
1105 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
1106 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
1107 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
1108 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
1109 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
1110 #define EXT4_MOUNT_ERRORS_MASK 0x00070
1111 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
1112 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
1113 #ifdef CONFIG_FS_DAX
1114 #define EXT4_MOUNT_DAX 0x00200 /* Direct Access */
1115 #else
1116 #define EXT4_MOUNT_DAX 0
1117 #endif
1118 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
1119 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
1120 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
1121 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
1122 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
1123 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
1124 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
1125 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
1126 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
1127 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
1128 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */
1129 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota,
1130 * enable enforcement for hidden
1131 * quota files */
1132 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable
1133 * enforcement for hidden quota
1134 * files */
1135 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota
1136 * enforcement */
1137 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
1138 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
1139 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1140 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */
1141 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
1142 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
1143 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
1144 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
1145 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
1146
1147 /*
1148 * Mount flags set either automatically (could not be set by mount option)
1149 * based on per file system feature or property or in special cases such as
1150 * distinguishing between explicit mount option definition and default.
1151 */
1152 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1153 specified delalloc */
1154 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1155 size of blocksize * 8
1156 blocks */
1157 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1158 file systems */
1159
1160 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly
1161 specified journal checksum */
1162
1163 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1164 ~EXT4_MOUNT_##opt
1165 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1166 EXT4_MOUNT_##opt
1167 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1168 EXT4_MOUNT_##opt)
1169
1170 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1171 ~EXT4_MOUNT2_##opt
1172 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1173 EXT4_MOUNT2_##opt
1174 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1175 EXT4_MOUNT2_##opt)
1176
1177 #define ext4_test_and_set_bit __test_and_set_bit_le
1178 #define ext4_set_bit __set_bit_le
1179 #define ext4_set_bit_atomic ext2_set_bit_atomic
1180 #define ext4_test_and_clear_bit __test_and_clear_bit_le
1181 #define ext4_clear_bit __clear_bit_le
1182 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
1183 #define ext4_test_bit test_bit_le
1184 #define ext4_find_next_zero_bit find_next_zero_bit_le
1185 #define ext4_find_next_bit find_next_bit_le
1186
1187 extern void ext4_set_bits(void *bm, int cur, int len);
1188
1189 /*
1190 * Maximal mount counts between two filesystem checks
1191 */
1192 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1193 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1194
1195 /*
1196 * Behaviour when detecting errors
1197 */
1198 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1199 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1200 #define EXT4_ERRORS_PANIC 3 /* Panic */
1201 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1202
1203 /* Metadata checksum algorithm codes */
1204 #define EXT4_CRC32C_CHKSUM 1
1205
1206 /*
1207 * Structure of the super block
1208 */
1209 struct ext4_super_block {
1210 /*00*/ __le32 s_inodes_count; /* Inodes count */
1211 __le32 s_blocks_count_lo; /* Blocks count */
1212 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1213 __le32 s_free_blocks_count_lo; /* Free blocks count */
1214 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1215 __le32 s_first_data_block; /* First Data Block */
1216 __le32 s_log_block_size; /* Block size */
1217 __le32 s_log_cluster_size; /* Allocation cluster size */
1218 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1219 __le32 s_clusters_per_group; /* # Clusters per group */
1220 __le32 s_inodes_per_group; /* # Inodes per group */
1221 __le32 s_mtime; /* Mount time */
1222 /*30*/ __le32 s_wtime; /* Write time */
1223 __le16 s_mnt_count; /* Mount count */
1224 __le16 s_max_mnt_count; /* Maximal mount count */
1225 __le16 s_magic; /* Magic signature */
1226 __le16 s_state; /* File system state */
1227 __le16 s_errors; /* Behaviour when detecting errors */
1228 __le16 s_minor_rev_level; /* minor revision level */
1229 /*40*/ __le32 s_lastcheck; /* time of last check */
1230 __le32 s_checkinterval; /* max. time between checks */
1231 __le32 s_creator_os; /* OS */
1232 __le32 s_rev_level; /* Revision level */
1233 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1234 __le16 s_def_resgid; /* Default gid for reserved blocks */
1235 /*
1236 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1237 *
1238 * Note: the difference between the compatible feature set and
1239 * the incompatible feature set is that if there is a bit set
1240 * in the incompatible feature set that the kernel doesn't
1241 * know about, it should refuse to mount the filesystem.
1242 *
1243 * e2fsck's requirements are more strict; if it doesn't know
1244 * about a feature in either the compatible or incompatible
1245 * feature set, it must abort and not try to meddle with
1246 * things it doesn't understand...
1247 */
1248 __le32 s_first_ino; /* First non-reserved inode */
1249 __le16 s_inode_size; /* size of inode structure */
1250 __le16 s_block_group_nr; /* block group # of this superblock */
1251 __le32 s_feature_compat; /* compatible feature set */
1252 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1253 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1254 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1255 /*78*/ char s_volume_name[16]; /* volume name */
1256 /*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */
1257 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1258 /*
1259 * Performance hints. Directory preallocation should only
1260 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1261 */
1262 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1263 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1264 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1265 /*
1266 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1267 */
1268 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1269 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1270 __le32 s_journal_dev; /* device number of journal file */
1271 __le32 s_last_orphan; /* start of list of inodes to delete */
1272 __le32 s_hash_seed[4]; /* HTREE hash seed */
1273 __u8 s_def_hash_version; /* Default hash version to use */
1274 __u8 s_jnl_backup_type;
1275 __le16 s_desc_size; /* size of group descriptor */
1276 /*100*/ __le32 s_default_mount_opts;
1277 __le32 s_first_meta_bg; /* First metablock block group */
1278 __le32 s_mkfs_time; /* When the filesystem was created */
1279 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1280 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1281 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1282 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1283 __le32 s_free_blocks_count_hi; /* Free blocks count */
1284 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1285 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1286 __le32 s_flags; /* Miscellaneous flags */
1287 __le16 s_raid_stride; /* RAID stride */
1288 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1289 __le64 s_mmp_block; /* Block for multi-mount protection */
1290 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1291 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1292 __u8 s_checksum_type; /* metadata checksum algorithm used */
1293 __u8 s_encryption_level; /* versioning level for encryption */
1294 __u8 s_reserved_pad; /* Padding to next 32bits */
1295 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1296 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1297 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1298 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1299 snapshot's future use */
1300 __le32 s_snapshot_list; /* inode number of the head of the
1301 on-disk snapshot list */
1302 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1303 __le32 s_error_count; /* number of fs errors */
1304 __le32 s_first_error_time; /* first time an error happened */
1305 __le32 s_first_error_ino; /* inode involved in first error */
1306 __le64 s_first_error_block; /* block involved of first error */
1307 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */
1308 __le32 s_first_error_line; /* line number where error happened */
1309 __le32 s_last_error_time; /* most recent time of an error */
1310 __le32 s_last_error_ino; /* inode involved in last error */
1311 __le32 s_last_error_line; /* line number where error happened */
1312 __le64 s_last_error_block; /* block involved of last error */
1313 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */
1314 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1315 __u8 s_mount_opts[64];
1316 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1317 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1318 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1319 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
1320 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */
1321 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */
1322 __le32 s_lpf_ino; /* Location of the lost+found inode */
1323 __le32 s_prj_quota_inum; /* inode for tracking project quota */
1324 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */
1325 __u8 s_wtime_hi;
1326 __u8 s_mtime_hi;
1327 __u8 s_mkfs_time_hi;
1328 __u8 s_lastcheck_hi;
1329 __u8 s_first_error_time_hi;
1330 __u8 s_last_error_time_hi;
1331 __u8 s_pad[2];
1332 __le32 s_reserved[96]; /* Padding to the end of the block */
1333 __le32 s_checksum; /* crc32c(superblock) */
1334 };
1335
1336 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1337
1338 #ifdef __KERNEL__
1339
1340 /*
1341 * run-time mount flags
1342 */
1343 #define EXT4_MF_MNTDIR_SAMPLED 0x0001
1344 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1345 #define EXT4_MF_TEST_DUMMY_ENCRYPTION 0x0004
1346
1347 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1348 #define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \
1349 EXT4_MF_TEST_DUMMY_ENCRYPTION))
1350 #else
1351 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1352 #endif
1353
1354 /* Number of quota types we support */
1355 #define EXT4_MAXQUOTAS 3
1356
1357 /*
1358 * fourth extended-fs super-block data in memory
1359 */
1360 struct ext4_sb_info {
1361 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1362 unsigned long s_inodes_per_block;/* Number of inodes per block */
1363 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1364 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1365 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1366 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1367 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1368 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1369 ext4_group_t s_groups_count; /* Number of groups in the fs */
1370 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1371 unsigned long s_overhead; /* # of fs overhead clusters */
1372 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1373 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1374 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1375 struct buffer_head * s_sbh; /* Buffer containing the super block */
1376 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1377 struct buffer_head * __rcu *s_group_desc;
1378 unsigned int s_mount_opt;
1379 unsigned int s_mount_opt2;
1380 unsigned int s_mount_flags;
1381 unsigned int s_def_mount_opt;
1382 ext4_fsblk_t s_sb_block;
1383 atomic64_t s_resv_clusters;
1384 kuid_t s_resuid;
1385 kgid_t s_resgid;
1386 unsigned short s_mount_state;
1387 unsigned short s_pad;
1388 int s_addr_per_block_bits;
1389 int s_desc_per_block_bits;
1390 int s_inode_size;
1391 int s_first_ino;
1392 unsigned int s_inode_readahead_blks;
1393 unsigned int s_inode_goal;
1394 u32 s_hash_seed[4];
1395 int s_def_hash_version;
1396 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1397 struct percpu_counter s_freeclusters_counter;
1398 struct percpu_counter s_freeinodes_counter;
1399 struct percpu_counter s_dirs_counter;
1400 struct percpu_counter s_dirtyclusters_counter;
1401 struct blockgroup_lock *s_blockgroup_lock;
1402 struct proc_dir_entry *s_proc;
1403 struct kobject s_kobj;
1404 struct completion s_kobj_unregister;
1405 struct super_block *s_sb;
1406
1407 /* Journaling */
1408 struct journal_s *s_journal;
1409 struct list_head s_orphan;
1410 struct mutex s_orphan_lock;
1411 unsigned long s_ext4_flags; /* Ext4 superblock flags */
1412 unsigned long s_commit_interval;
1413 u32 s_max_batch_time;
1414 u32 s_min_batch_time;
1415 struct block_device *s_journal_bdev;
1416 #ifdef CONFIG_QUOTA
1417 /* Names of quota files with journalled quota */
1418 char __rcu *s_qf_names[EXT4_MAXQUOTAS];
1419 int s_jquota_fmt; /* Format of quota to use */
1420 #endif
1421 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1422 struct ext4_system_blocks __rcu *system_blks;
1423
1424 #ifdef EXTENTS_STATS
1425 /* ext4 extents stats */
1426 unsigned long s_ext_min;
1427 unsigned long s_ext_max;
1428 unsigned long s_depth_max;
1429 spinlock_t s_ext_stats_lock;
1430 unsigned long s_ext_blocks;
1431 unsigned long s_ext_extents;
1432 #endif
1433
1434 /* for buddy allocator */
1435 struct ext4_group_info ** __rcu *s_group_info;
1436 struct inode *s_buddy_cache;
1437 spinlock_t s_md_lock;
1438 unsigned short *s_mb_offsets;
1439 unsigned int *s_mb_maxs;
1440 unsigned int s_group_info_size;
1441 unsigned int s_mb_free_pending;
1442 struct list_head s_freed_data_list; /* List of blocks to be freed
1443 after commit completed */
1444
1445 /* tunables */
1446 unsigned long s_stripe;
1447 unsigned int s_mb_stream_request;
1448 unsigned int s_mb_max_to_scan;
1449 unsigned int s_mb_min_to_scan;
1450 unsigned int s_mb_stats;
1451 unsigned int s_mb_order2_reqs;
1452 unsigned int s_mb_group_prealloc;
1453 unsigned int s_max_dir_size_kb;
1454 /* where last allocation was done - for stream allocation */
1455 unsigned long s_mb_last_group;
1456 unsigned long s_mb_last_start;
1457
1458 /* stats for buddy allocator */
1459 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1460 atomic_t s_bal_success; /* we found long enough chunks */
1461 atomic_t s_bal_allocated; /* in blocks */
1462 atomic_t s_bal_ex_scanned; /* total extents scanned */
1463 atomic_t s_bal_goals; /* goal hits */
1464 atomic_t s_bal_breaks; /* too long searches */
1465 atomic_t s_bal_2orders; /* 2^order hits */
1466 spinlock_t s_bal_lock;
1467 unsigned long s_mb_buddies_generated;
1468 unsigned long long s_mb_generation_time;
1469 atomic_t s_mb_lost_chunks;
1470 atomic_t s_mb_preallocated;
1471 atomic_t s_mb_discarded;
1472 atomic_t s_lock_busy;
1473
1474 /* locality groups */
1475 struct ext4_locality_group __percpu *s_locality_groups;
1476
1477 /* for write statistics */
1478 unsigned long s_sectors_written_start;
1479 u64 s_kbytes_written;
1480
1481 /* the size of zero-out chunk */
1482 unsigned int s_extent_max_zeroout_kb;
1483
1484 unsigned int s_log_groups_per_flex;
1485 struct flex_groups * __rcu *s_flex_groups;
1486 ext4_group_t s_flex_groups_allocated;
1487
1488 /* workqueue for reserved extent conversions (buffered io) */
1489 struct workqueue_struct *rsv_conversion_wq;
1490
1491 /* timer for periodic error stats printing */
1492 struct timer_list s_err_report;
1493
1494 /* Lazy inode table initialization info */
1495 struct ext4_li_request *s_li_request;
1496 /* Wait multiplier for lazy initialization thread */
1497 unsigned int s_li_wait_mult;
1498
1499 /* Kernel thread for multiple mount protection */
1500 struct task_struct *s_mmp_tsk;
1501
1502 /* record the last minlen when FITRIM is called. */
1503 unsigned long s_last_trim_minblks;
1504
1505 /* Reference to checksum algorithm driver via cryptoapi */
1506 struct crypto_shash *s_chksum_driver;
1507
1508 /* Precomputed FS UUID checksum for seeding other checksums */
1509 __u32 s_csum_seed;
1510
1511 /* Reclaim extents from extent status tree */
1512 struct shrinker s_es_shrinker;
1513 struct list_head s_es_list; /* List of inodes with reclaimable extents */
1514 long s_es_nr_inode;
1515 struct ext4_es_stats s_es_stats;
1516 struct mb_cache *s_ea_block_cache;
1517 struct mb_cache *s_ea_inode_cache;
1518 spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1519
1520 /* Ratelimit ext4 messages. */
1521 struct ratelimit_state s_err_ratelimit_state;
1522 struct ratelimit_state s_warning_ratelimit_state;
1523 struct ratelimit_state s_msg_ratelimit_state;
1524
1525 /*
1526 * Barrier between writepages ops and changing any inode's JOURNAL_DATA
1527 * or EXTENTS flag.
1528 */
1529 struct percpu_rw_semaphore s_writepages_rwsem;
1530 struct dax_device *s_daxdev;
1531 };
1532
EXT4_SB(struct super_block * sb)1533 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1534 {
1535 return sb->s_fs_info;
1536 }
EXT4_I(struct inode * inode)1537 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1538 {
1539 return container_of(inode, struct ext4_inode_info, vfs_inode);
1540 }
1541
ext4_valid_inum(struct super_block * sb,unsigned long ino)1542 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1543 {
1544 return ino == EXT4_ROOT_INO ||
1545 (ino >= EXT4_FIRST_INO(sb) &&
1546 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1547 }
1548
1549 /*
1550 * Returns: sbi->field[index]
1551 * Used to access an array element from the following sbi fields which require
1552 * rcu protection to avoid dereferencing an invalid pointer due to reassignment
1553 * - s_group_desc
1554 * - s_group_info
1555 * - s_flex_group
1556 */
1557 #define sbi_array_rcu_deref(sbi, field, index) \
1558 ({ \
1559 typeof(*((sbi)->field)) _v; \
1560 rcu_read_lock(); \
1561 _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index]; \
1562 rcu_read_unlock(); \
1563 _v; \
1564 })
1565
1566 /*
1567 * Inode dynamic state flags
1568 */
1569 enum {
1570 EXT4_STATE_JDATA, /* journaled data exists */
1571 EXT4_STATE_NEW, /* inode is newly created */
1572 EXT4_STATE_XATTR, /* has in-inode xattrs */
1573 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1574 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1575 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1576 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
1577 EXT4_STATE_NEWENTRY, /* File just added to dir */
1578 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1579 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
1580 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */
1581 };
1582
1583 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1584 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1585 { \
1586 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1587 } \
1588 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1589 { \
1590 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1591 } \
1592 static 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. */
1599 static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1600 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1601 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1602 EXT4_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. */
1606 static inline int ext4_test_inode_state(struct inode *inode, int bit);
1607 static inline void ext4_set_inode_state(struct inode *inode, int bit);
1608 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1609 #if (BITS_PER_LONG < 64)
1610 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1611
ext4_clear_state_flags(struct ext4_inode_info * ei)1612 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1613 {
1614 (ei)->i_state_flags = 0;
1615 }
1616 #else
1617 EXT4_INODE_BIT_FNS(state, flags, 32)
1618
ext4_clear_state_flags(struct ext4_inode_info * ei)1619 static 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
1631 /*
1632 * Returns true if the inode is inode is encrypted
1633 */
1634 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1635
1636 /*
1637 * Codes for operating systems
1638 */
1639 #define EXT4_OS_LINUX 0
1640 #define EXT4_OS_HURD 1
1641 #define EXT4_OS_MASIX 2
1642 #define EXT4_OS_FREEBSD 3
1643 #define EXT4_OS_LITES 4
1644
1645 /*
1646 * Revision levels
1647 */
1648 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1649 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1650
1651 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1652 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1653
1654 #define EXT4_GOOD_OLD_INODE_SIZE 128
1655
1656 /*
1657 * Feature set definitions
1658 */
1659
1660 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1661 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1662 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1663 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1664 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1665 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1666 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
1667
1668 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1669 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1670 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1671 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1672 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1673 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1674 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1675 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1676 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1677 /*
1678 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1679 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1680 * all other data structures' checksums. However, the METADATA_CSUM and
1681 * GDT_CSUM bits are mutually exclusive.
1682 */
1683 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
1684 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000
1685 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000
1686
1687 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1688 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1689 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1690 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1691 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1692 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1693 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1694 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1695 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1696 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1697 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1698 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000
1699 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
1700 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
1701 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000
1702
1703 extern void ext4_update_dynamic_rev(struct super_block *sb);
1704
1705 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
1706 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1707 { \
1708 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1709 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
1710 } \
1711 static inline void ext4_set_feature_##name(struct super_block *sb) \
1712 { \
1713 ext4_update_dynamic_rev(sb); \
1714 EXT4_SB(sb)->s_es->s_feature_compat |= \
1715 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1716 } \
1717 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1718 { \
1719 EXT4_SB(sb)->s_es->s_feature_compat &= \
1720 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1721 }
1722
1723 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
1724 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1725 { \
1726 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1727 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
1728 } \
1729 static inline void ext4_set_feature_##name(struct super_block *sb) \
1730 { \
1731 ext4_update_dynamic_rev(sb); \
1732 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
1733 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1734 } \
1735 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1736 { \
1737 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
1738 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1739 }
1740
1741 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
1742 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1743 { \
1744 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1745 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
1746 } \
1747 static inline void ext4_set_feature_##name(struct super_block *sb) \
1748 { \
1749 ext4_update_dynamic_rev(sb); \
1750 EXT4_SB(sb)->s_es->s_feature_incompat |= \
1751 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1752 } \
1753 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1754 { \
1755 EXT4_SB(sb)->s_es->s_feature_incompat &= \
1756 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1757 }
1758
EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc,DIR_PREALLOC)1759 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC)
1760 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES)
1761 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL)
1762 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR)
1763 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE)
1764 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX)
1765 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2)
1766
1767 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER)
1768 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE)
1769 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR)
1770 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE)
1771 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM)
1772 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK)
1773 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE)
1774 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA)
1775 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC)
1776 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM)
1777 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY)
1778 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT)
1779
1780 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION)
1781 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE)
1782 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER)
1783 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV)
1784 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG)
1785 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS)
1786 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT)
1787 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP)
1788 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG)
1789 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE)
1790 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA)
1791 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED)
1792 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR)
1793 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA)
1794 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT)
1795
1796 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1797 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1798 EXT4_FEATURE_INCOMPAT_META_BG)
1799 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1800 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1801 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1802
1803 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1804 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1805 EXT4_FEATURE_INCOMPAT_RECOVER| \
1806 EXT4_FEATURE_INCOMPAT_META_BG)
1807 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1808 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1809 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1810
1811 #define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1812 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1813 EXT4_FEATURE_INCOMPAT_RECOVER| \
1814 EXT4_FEATURE_INCOMPAT_META_BG| \
1815 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1816 EXT4_FEATURE_INCOMPAT_64BIT| \
1817 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1818 EXT4_FEATURE_INCOMPAT_EA_INODE| \
1819 EXT4_FEATURE_INCOMPAT_MMP | \
1820 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
1821 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
1822 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \
1823 EXT4_FEATURE_INCOMPAT_LARGEDIR)
1824 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1825 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1826 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1827 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1828 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1829 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1830 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1831 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1832 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1833 EXT4_FEATURE_RO_COMPAT_QUOTA |\
1834 EXT4_FEATURE_RO_COMPAT_PROJECT)
1835
1836 #define EXTN_FEATURE_FUNCS(ver) \
1837 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
1838 { \
1839 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1840 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
1841 } \
1842 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
1843 { \
1844 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1845 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
1846 } \
1847 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
1848 { \
1849 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1850 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
1851 }
1852
1853 EXTN_FEATURE_FUNCS(2)
1854 EXTN_FEATURE_FUNCS(3)
1855 EXTN_FEATURE_FUNCS(4)
1856
1857 static inline bool ext4_has_compat_features(struct super_block *sb)
1858 {
1859 return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
1860 }
ext4_has_ro_compat_features(struct super_block * sb)1861 static inline bool ext4_has_ro_compat_features(struct super_block *sb)
1862 {
1863 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
1864 }
ext4_has_incompat_features(struct super_block * sb)1865 static inline bool ext4_has_incompat_features(struct super_block *sb)
1866 {
1867 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
1868 }
1869
1870 /*
1871 * Superblock flags
1872 */
1873 #define EXT4_FLAGS_RESIZING 0
1874 #define EXT4_FLAGS_SHUTDOWN 1
1875
ext4_forced_shutdown(struct ext4_sb_info * sbi)1876 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi)
1877 {
1878 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags);
1879 }
1880
1881
1882 /*
1883 * Default values for user and/or group using reserved blocks
1884 */
1885 #define EXT4_DEF_RESUID 0
1886 #define EXT4_DEF_RESGID 0
1887
1888 /*
1889 * Default project ID
1890 */
1891 #define EXT4_DEF_PROJID 0
1892
1893 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
1894
1895 /*
1896 * Default mount options
1897 */
1898 #define EXT4_DEFM_DEBUG 0x0001
1899 #define EXT4_DEFM_BSDGROUPS 0x0002
1900 #define EXT4_DEFM_XATTR_USER 0x0004
1901 #define EXT4_DEFM_ACL 0x0008
1902 #define EXT4_DEFM_UID16 0x0010
1903 #define EXT4_DEFM_JMODE 0x0060
1904 #define EXT4_DEFM_JMODE_DATA 0x0020
1905 #define EXT4_DEFM_JMODE_ORDERED 0x0040
1906 #define EXT4_DEFM_JMODE_WBACK 0x0060
1907 #define EXT4_DEFM_NOBARRIER 0x0100
1908 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1909 #define EXT4_DEFM_DISCARD 0x0400
1910 #define EXT4_DEFM_NODELALLOC 0x0800
1911
1912 /*
1913 * Default journal batch times
1914 */
1915 #define EXT4_DEF_MIN_BATCH_TIME 0
1916 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1917
1918 /*
1919 * Minimum number of groups in a flexgroup before we separate out
1920 * directories into the first block group of a flexgroup
1921 */
1922 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1923
1924 /*
1925 * Structure of a directory entry
1926 */
1927 #define EXT4_NAME_LEN 255
1928
1929 struct ext4_dir_entry {
1930 __le32 inode; /* Inode number */
1931 __le16 rec_len; /* Directory entry length */
1932 __le16 name_len; /* Name length */
1933 char name[EXT4_NAME_LEN]; /* File name */
1934 };
1935
1936 /*
1937 * The new version of the directory entry. Since EXT4 structures are
1938 * stored in intel byte order, and the name_len field could never be
1939 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1940 * file_type field.
1941 */
1942 struct ext4_dir_entry_2 {
1943 __le32 inode; /* Inode number */
1944 __le16 rec_len; /* Directory entry length */
1945 __u8 name_len; /* Name length */
1946 __u8 file_type;
1947 char name[EXT4_NAME_LEN]; /* File name */
1948 };
1949
1950 /*
1951 * This is a bogus directory entry at the end of each leaf block that
1952 * records checksums.
1953 */
1954 struct ext4_dir_entry_tail {
1955 __le32 det_reserved_zero1; /* Pretend to be unused */
1956 __le16 det_rec_len; /* 12 */
1957 __u8 det_reserved_zero2; /* Zero name length */
1958 __u8 det_reserved_ft; /* 0xDE, fake file type */
1959 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
1960 };
1961
1962 #define EXT4_DIRENT_TAIL(block, blocksize) \
1963 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1964 ((blocksize) - \
1965 sizeof(struct ext4_dir_entry_tail))))
1966
1967 /*
1968 * Ext4 directory file types. Only the low 3 bits are used. The
1969 * other bits are reserved for now.
1970 */
1971 #define EXT4_FT_UNKNOWN 0
1972 #define EXT4_FT_REG_FILE 1
1973 #define EXT4_FT_DIR 2
1974 #define EXT4_FT_CHRDEV 3
1975 #define EXT4_FT_BLKDEV 4
1976 #define EXT4_FT_FIFO 5
1977 #define EXT4_FT_SOCK 6
1978 #define EXT4_FT_SYMLINK 7
1979
1980 #define EXT4_FT_MAX 8
1981
1982 #define EXT4_FT_DIR_CSUM 0xDE
1983
1984 /*
1985 * EXT4_DIR_PAD defines the directory entries boundaries
1986 *
1987 * NOTE: It must be a multiple of 4
1988 */
1989 #define EXT4_DIR_PAD 4
1990 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1991 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1992 ~EXT4_DIR_ROUND)
1993 #define EXT4_MAX_REC_LEN ((1<<16)-1)
1994
1995 /*
1996 * If we ever get support for fs block sizes > page_size, we'll need
1997 * to remove the #if statements in the next two functions...
1998 */
1999 static inline unsigned int
ext4_rec_len_from_disk(__le16 dlen,unsigned blocksize)2000 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
2001 {
2002 unsigned len = le16_to_cpu(dlen);
2003
2004 #if (PAGE_SIZE >= 65536)
2005 if (len == EXT4_MAX_REC_LEN || len == 0)
2006 return blocksize;
2007 return (len & 65532) | ((len & 3) << 16);
2008 #else
2009 return len;
2010 #endif
2011 }
2012
ext4_rec_len_to_disk(unsigned len,unsigned blocksize)2013 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
2014 {
2015 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
2016 BUG();
2017 #if (PAGE_SIZE >= 65536)
2018 if (len < 65536)
2019 return cpu_to_le16(len);
2020 if (len == blocksize) {
2021 if (blocksize == 65536)
2022 return cpu_to_le16(EXT4_MAX_REC_LEN);
2023 else
2024 return cpu_to_le16(0);
2025 }
2026 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
2027 #else
2028 return cpu_to_le16(len);
2029 #endif
2030 }
2031
2032 /*
2033 * Hash Tree Directory indexing
2034 * (c) Daniel Phillips, 2001
2035 */
2036
2037 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
2038 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2039 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \
2040 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir)))
2041 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2042
2043 /* Legal values for the dx_root hash_version field: */
2044
2045 #define DX_HASH_LEGACY 0
2046 #define DX_HASH_HALF_MD4 1
2047 #define DX_HASH_TEA 2
2048 #define DX_HASH_LEGACY_UNSIGNED 3
2049 #define DX_HASH_HALF_MD4_UNSIGNED 4
2050 #define DX_HASH_TEA_UNSIGNED 5
2051
ext4_chksum(struct ext4_sb_info * sbi,u32 crc,const void * address,unsigned int length)2052 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2053 const void *address, unsigned int length)
2054 {
2055 struct {
2056 struct shash_desc shash;
2057 char ctx[4];
2058 } desc;
2059
2060 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2061
2062 desc.shash.tfm = sbi->s_chksum_driver;
2063 desc.shash.flags = 0;
2064 *(u32 *)desc.ctx = crc;
2065
2066 BUG_ON(crypto_shash_update(&desc.shash, address, length));
2067
2068 return *(u32 *)desc.ctx;
2069 }
2070
2071 #ifdef __KERNEL__
2072
2073 /* hash info structure used by the directory hash */
2074 struct dx_hash_info
2075 {
2076 u32 hash;
2077 u32 minor_hash;
2078 int hash_version;
2079 u32 *seed;
2080 };
2081
2082
2083 /* 32 and 64 bit signed EOF for dx directories */
2084 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
2085 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
2086
2087
2088 /*
2089 * Control parameters used by ext4_htree_next_block
2090 */
2091 #define HASH_NB_ALWAYS 1
2092
2093 struct ext4_filename {
2094 const struct qstr *usr_fname;
2095 struct fscrypt_str disk_name;
2096 struct dx_hash_info hinfo;
2097 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2098 struct fscrypt_str crypto_buf;
2099 #endif
2100 };
2101
2102 #define fname_name(p) ((p)->disk_name.name)
2103 #define fname_len(p) ((p)->disk_name.len)
2104
2105 /*
2106 * Describe an inode's exact location on disk and in memory
2107 */
2108 struct ext4_iloc
2109 {
2110 struct buffer_head *bh;
2111 unsigned long offset;
2112 ext4_group_t block_group;
2113 };
2114
ext4_raw_inode(struct ext4_iloc * iloc)2115 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2116 {
2117 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2118 }
2119
ext4_is_quota_file(struct inode * inode)2120 static inline bool ext4_is_quota_file(struct inode *inode)
2121 {
2122 return IS_NOQUOTA(inode) &&
2123 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL);
2124 }
2125
2126 /*
2127 * This structure is stuffed into the struct file's private_data field
2128 * for directories. It is where we put information so that we can do
2129 * readdir operations in hash tree order.
2130 */
2131 struct dir_private_info {
2132 struct rb_root root;
2133 struct rb_node *curr_node;
2134 struct fname *extra_fname;
2135 loff_t last_pos;
2136 __u32 curr_hash;
2137 __u32 curr_minor_hash;
2138 __u32 next_hash;
2139 };
2140
2141 /* calculate the first block number of the group */
2142 static inline ext4_fsblk_t
ext4_group_first_block_no(struct super_block * sb,ext4_group_t group_no)2143 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2144 {
2145 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2146 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2147 }
2148
2149 /*
2150 * Special error return code only used by dx_probe() and its callers.
2151 */
2152 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
2153
2154 /* htree levels for ext4 */
2155 #define EXT4_HTREE_LEVEL_COMPAT 2
2156 #define EXT4_HTREE_LEVEL 3
2157
ext4_dir_htree_level(struct super_block * sb)2158 static inline int ext4_dir_htree_level(struct super_block *sb)
2159 {
2160 return ext4_has_feature_largedir(sb) ?
2161 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT;
2162 }
2163
2164 /*
2165 * Timeout and state flag for lazy initialization inode thread.
2166 */
2167 #define EXT4_DEF_LI_WAIT_MULT 10
2168 #define EXT4_DEF_LI_MAX_START_DELAY 5
2169 #define EXT4_LAZYINIT_QUIT 0x0001
2170 #define EXT4_LAZYINIT_RUNNING 0x0002
2171
2172 /*
2173 * Lazy inode table initialization info
2174 */
2175 struct ext4_lazy_init {
2176 unsigned long li_state;
2177 struct list_head li_request_list;
2178 struct mutex li_list_mtx;
2179 };
2180
2181 struct ext4_li_request {
2182 struct super_block *lr_super;
2183 struct ext4_sb_info *lr_sbi;
2184 ext4_group_t lr_next_group;
2185 struct list_head lr_request;
2186 unsigned long lr_next_sched;
2187 unsigned long lr_timeout;
2188 };
2189
2190 struct ext4_features {
2191 struct kobject f_kobj;
2192 struct completion f_kobj_unregister;
2193 };
2194
2195 /*
2196 * This structure will be used for multiple mount protection. It will be
2197 * written into the block number saved in the s_mmp_block field in the
2198 * superblock. Programs that check MMP should assume that if
2199 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2200 * to use the filesystem, regardless of how old the timestamp is.
2201 */
2202 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
2203 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2204 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
2205 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
2206
2207 struct mmp_struct {
2208 __le32 mmp_magic; /* Magic number for MMP */
2209 __le32 mmp_seq; /* Sequence no. updated periodically */
2210
2211 /*
2212 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2213 * purposes and do not affect the correctness of the algorithm
2214 */
2215 __le64 mmp_time; /* Time last updated */
2216 char mmp_nodename[64]; /* Node which last updated MMP block */
2217 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
2218
2219 /*
2220 * mmp_check_interval is used to verify if the MMP block has been
2221 * updated on the block device. The value is updated based on the
2222 * maximum time to write the MMP block during an update cycle.
2223 */
2224 __le16 mmp_check_interval;
2225
2226 __le16 mmp_pad1;
2227 __le32 mmp_pad2[226];
2228 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
2229 };
2230
2231 /* arguments passed to the mmp thread */
2232 struct mmpd_data {
2233 struct buffer_head *bh; /* bh from initial read_mmp_block() */
2234 struct super_block *sb; /* super block of the fs */
2235 };
2236
2237 /*
2238 * Check interval multiplier
2239 * The MMP block is written every update interval and initially checked every
2240 * update interval x the multiplier (the value is then adapted based on the
2241 * write latency). The reason is that writes can be delayed under load and we
2242 * don't want readers to incorrectly assume that the filesystem is no longer
2243 * in use.
2244 */
2245 #define EXT4_MMP_CHECK_MULT 2UL
2246
2247 /*
2248 * Minimum interval for MMP checking in seconds.
2249 */
2250 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
2251
2252 /*
2253 * Maximum interval for MMP checking in seconds.
2254 */
2255 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
2256
2257 /*
2258 * Function prototypes
2259 */
2260
2261 /*
2262 * Ok, these declarations are also in <linux/kernel.h> but none of the
2263 * ext4 source programs needs to include it so they are duplicated here.
2264 */
2265 # define NORET_TYPE /**/
2266 # define ATTRIB_NORET __attribute__((noreturn))
2267 # define NORET_AND noreturn,
2268
2269 /* bitmap.c */
2270 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2271 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2272 struct ext4_group_desc *gdp,
2273 struct buffer_head *bh, int sz);
2274 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2275 struct ext4_group_desc *gdp,
2276 struct buffer_head *bh, int sz);
2277 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2278 struct ext4_group_desc *gdp,
2279 struct buffer_head *bh);
2280 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2281 struct ext4_group_desc *gdp,
2282 struct buffer_head *bh);
2283
2284 /* balloc.c */
2285 extern void ext4_get_group_no_and_offset(struct super_block *sb,
2286 ext4_fsblk_t blocknr,
2287 ext4_group_t *blockgrpp,
2288 ext4_grpblk_t *offsetp);
2289 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2290 ext4_fsblk_t block);
2291
2292 extern unsigned int ext4_block_group(struct super_block *sb,
2293 ext4_fsblk_t blocknr);
2294 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2295 ext4_fsblk_t blocknr);
2296 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2297 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2298 ext4_group_t group);
2299 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2300 ext4_fsblk_t goal,
2301 unsigned int flags,
2302 unsigned long *count,
2303 int *errp);
2304 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2305 s64 nclusters, unsigned int flags);
2306 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2307 extern void ext4_check_blocks_bitmap(struct super_block *);
2308 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2309 ext4_group_t block_group,
2310 struct buffer_head ** bh);
2311 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2312
2313 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2314 ext4_group_t block_group);
2315 extern int ext4_wait_block_bitmap(struct super_block *sb,
2316 ext4_group_t block_group,
2317 struct buffer_head *bh);
2318 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2319 ext4_group_t block_group);
2320 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2321 ext4_group_t block_group,
2322 struct ext4_group_desc *gdp);
2323 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2324
ext4_encrypted_inode(struct inode * inode)2325 static inline bool ext4_encrypted_inode(struct inode *inode)
2326 {
2327 return ext4_test_inode_flag(inode, EXT4_INODE_ENCRYPT);
2328 }
2329
2330 #ifdef CONFIG_EXT4_FS_ENCRYPTION
ext4_fname_from_fscrypt_name(struct ext4_filename * dst,const struct fscrypt_name * src)2331 static inline void ext4_fname_from_fscrypt_name(struct ext4_filename *dst,
2332 const struct fscrypt_name *src)
2333 {
2334 memset(dst, 0, sizeof(*dst));
2335
2336 dst->usr_fname = src->usr_fname;
2337 dst->disk_name = src->disk_name;
2338 dst->hinfo.hash = src->hash;
2339 dst->hinfo.minor_hash = src->minor_hash;
2340 dst->crypto_buf = src->crypto_buf;
2341 }
2342
ext4_fname_setup_filename(struct inode * dir,const struct qstr * iname,int lookup,struct ext4_filename * fname)2343 static inline int ext4_fname_setup_filename(struct inode *dir,
2344 const struct qstr *iname,
2345 int lookup,
2346 struct ext4_filename *fname)
2347 {
2348 struct fscrypt_name name;
2349 int err;
2350
2351 err = fscrypt_setup_filename(dir, iname, lookup, &name);
2352 if (err)
2353 return err;
2354
2355 ext4_fname_from_fscrypt_name(fname, &name);
2356 return 0;
2357 }
2358
ext4_fname_prepare_lookup(struct inode * dir,struct dentry * dentry,struct ext4_filename * fname)2359 static inline int ext4_fname_prepare_lookup(struct inode *dir,
2360 struct dentry *dentry,
2361 struct ext4_filename *fname)
2362 {
2363 struct fscrypt_name name;
2364 int err;
2365
2366 err = fscrypt_prepare_lookup(dir, dentry, &name);
2367 if (err)
2368 return err;
2369
2370 ext4_fname_from_fscrypt_name(fname, &name);
2371 return 0;
2372 }
2373
ext4_fname_free_filename(struct ext4_filename * fname)2374 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2375 {
2376 struct fscrypt_name name;
2377
2378 name.crypto_buf = fname->crypto_buf;
2379 fscrypt_free_filename(&name);
2380
2381 fname->crypto_buf.name = NULL;
2382 fname->usr_fname = NULL;
2383 fname->disk_name.name = NULL;
2384 }
2385 #else /* !CONFIG_EXT4_FS_ENCRYPTION */
ext4_fname_setup_filename(struct inode * dir,const struct qstr * iname,int lookup,struct ext4_filename * fname)2386 static inline int ext4_fname_setup_filename(struct inode *dir,
2387 const struct qstr *iname,
2388 int lookup,
2389 struct ext4_filename *fname)
2390 {
2391 fname->usr_fname = iname;
2392 fname->disk_name.name = (unsigned char *) iname->name;
2393 fname->disk_name.len = iname->len;
2394 return 0;
2395 }
2396
ext4_fname_prepare_lookup(struct inode * dir,struct dentry * dentry,struct ext4_filename * fname)2397 static inline int ext4_fname_prepare_lookup(struct inode *dir,
2398 struct dentry *dentry,
2399 struct ext4_filename *fname)
2400 {
2401 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname);
2402 }
2403
ext4_fname_free_filename(struct ext4_filename * fname)2404 static inline void ext4_fname_free_filename(struct ext4_filename *fname) { }
2405 #endif /* !CONFIG_EXT4_FS_ENCRYPTION */
2406
2407 /* dir.c */
2408 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2409 struct file *,
2410 struct ext4_dir_entry_2 *,
2411 struct buffer_head *, char *, int,
2412 unsigned int);
2413 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2414 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2415 (de), (bh), (buf), (size), (offset)))
2416 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2417 __u32 minor_hash,
2418 struct ext4_dir_entry_2 *dirent,
2419 struct fscrypt_str *ent_name);
2420 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2421 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2422 struct buffer_head *bh,
2423 void *buf, int buf_size,
2424 struct ext4_filename *fname,
2425 struct ext4_dir_entry_2 **dest_de);
2426 void ext4_insert_dentry(struct inode *inode,
2427 struct ext4_dir_entry_2 *de,
2428 int buf_size,
2429 struct ext4_filename *fname);
ext4_update_dx_flag(struct inode * inode)2430 static inline void ext4_update_dx_flag(struct inode *inode)
2431 {
2432 if (!ext4_has_feature_dir_index(inode->i_sb) &&
2433 ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) {
2434 /* ext4_iget() should have caught this... */
2435 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb));
2436 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2437 }
2438 }
2439 static const unsigned char ext4_filetype_table[] = {
2440 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2441 };
2442
get_dtype(struct super_block * sb,int filetype)2443 static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2444 {
2445 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2446 return DT_UNKNOWN;
2447
2448 return ext4_filetype_table[filetype];
2449 }
2450 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2451 void *buf, int buf_size);
2452
2453 /* fsync.c */
2454 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2455
2456 /* hash.c */
2457 extern int ext4fs_dirhash(const char *name, int len, struct
2458 dx_hash_info *hinfo);
2459
2460 /* ialloc.c */
2461 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2462 const struct qstr *qstr, __u32 goal,
2463 uid_t *owner, __u32 i_flags,
2464 int handle_type, unsigned int line_no,
2465 int nblocks);
2466
2467 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \
2468 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2469 i_flags, 0, 0, 0)
2470 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2471 type, nblocks) \
2472 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2473 0, (type), __LINE__, (nblocks))
2474
2475
2476 extern void ext4_free_inode(handle_t *, struct inode *);
2477 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2478 extern unsigned long ext4_count_free_inodes(struct super_block *);
2479 extern unsigned long ext4_count_dirs(struct super_block *);
2480 extern void ext4_check_inodes_bitmap(struct super_block *);
2481 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2482 extern int ext4_init_inode_table(struct super_block *sb,
2483 ext4_group_t group, int barrier);
2484 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2485
2486 /* mballoc.c */
2487 extern const struct seq_operations ext4_mb_seq_groups_ops;
2488 extern long ext4_mb_stats;
2489 extern long ext4_mb_max_to_scan;
2490 extern int ext4_mb_init(struct super_block *);
2491 extern int ext4_mb_release(struct super_block *);
2492 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2493 struct ext4_allocation_request *, int *);
2494 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2495 extern void ext4_discard_preallocations(struct inode *);
2496 extern int __init ext4_init_mballoc(void);
2497 extern void ext4_exit_mballoc(void);
2498 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2499 struct buffer_head *bh, ext4_fsblk_t block,
2500 unsigned long count, int flags);
2501 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2502 ext4_group_t ngroups);
2503 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2504 ext4_group_t i, struct ext4_group_desc *desc);
2505 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2506 ext4_fsblk_t block, unsigned long count);
2507 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2508 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid);
2509
2510 /* inode.c */
2511 int ext4_inode_is_fast_symlink(struct inode *inode);
2512 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2513 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2514 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count,
2515 bool wait, struct buffer_head **bhs);
2516 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2517 struct buffer_head *bh_result, int create);
2518 int ext4_get_block(struct inode *inode, sector_t iblock,
2519 struct buffer_head *bh_result, int create);
2520 int ext4_dio_get_block(struct inode *inode, sector_t iblock,
2521 struct buffer_head *bh_result, int create);
2522 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2523 struct buffer_head *bh, int create);
2524 int ext4_walk_page_buffers(handle_t *handle,
2525 struct buffer_head *head,
2526 unsigned from,
2527 unsigned to,
2528 int *partial,
2529 int (*fn)(handle_t *handle,
2530 struct buffer_head *bh));
2531 int do_journal_get_write_access(handle_t *handle,
2532 struct buffer_head *bh);
2533 #define FALL_BACK_TO_NONDELALLOC 1
2534 #define CONVERT_INLINE_DATA 2
2535
2536 typedef enum {
2537 EXT4_IGET_NORMAL = 0,
2538 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */
2539 EXT4_IGET_HANDLE = 0x0002 /* Inode # is from a handle */
2540 } ext4_iget_flags;
2541
2542 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
2543 ext4_iget_flags flags, const char *function,
2544 unsigned int line);
2545
2546 #define ext4_iget(sb, ino, flags) \
2547 __ext4_iget((sb), (ino), (flags), __func__, __LINE__)
2548
2549 extern int ext4_write_inode(struct inode *, struct writeback_control *);
2550 extern int ext4_setattr(struct dentry *, struct iattr *);
2551 extern int ext4_getattr(const struct path *, struct kstat *, u32, unsigned int);
2552 extern void ext4_evict_inode(struct inode *);
2553 extern void ext4_clear_inode(struct inode *);
2554 extern int ext4_file_getattr(const struct path *, struct kstat *, u32, unsigned int);
2555 extern int ext4_sync_inode(handle_t *, struct inode *);
2556 extern void ext4_dirty_inode(struct inode *, int);
2557 extern int ext4_change_inode_journal_flag(struct inode *, int);
2558 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2559 extern int ext4_inode_attach_jinode(struct inode *inode);
2560 extern int ext4_can_truncate(struct inode *inode);
2561 extern int ext4_truncate(struct inode *);
2562 extern int ext4_break_layouts(struct inode *);
2563 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2564 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
2565 extern void ext4_set_inode_flags(struct inode *);
2566 extern int ext4_alloc_da_blocks(struct inode *inode);
2567 extern void ext4_set_aops(struct inode *inode);
2568 extern int ext4_writepage_trans_blocks(struct inode *);
2569 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2570 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2571 loff_t lstart, loff_t lend);
2572 extern int ext4_page_mkwrite(struct vm_fault *vmf);
2573 extern int ext4_filemap_fault(struct vm_fault *vmf);
2574 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2575 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
2576 extern void ext4_da_update_reserve_space(struct inode *inode,
2577 int used, int quota_claim);
2578 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
2579 ext4_fsblk_t pblk, ext4_lblk_t len);
2580
2581 /* indirect.c */
2582 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2583 struct ext4_map_blocks *map, int flags);
2584 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2585 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2586 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2587 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2588 ext4_lblk_t start, ext4_lblk_t end);
2589
2590 /* ioctl.c */
2591 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2592 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2593
2594 /* migrate.c */
2595 extern int ext4_ext_migrate(struct inode *);
2596 extern int ext4_ind_migrate(struct inode *inode);
2597
2598 /* namei.c */
2599 extern int ext4_dirent_csum_verify(struct inode *inode,
2600 struct ext4_dir_entry *dirent);
2601 extern int ext4_orphan_add(handle_t *, struct inode *);
2602 extern int ext4_orphan_del(handle_t *, struct inode *);
2603 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2604 __u32 start_minor_hash, __u32 *next_hash);
2605 extern int ext4_search_dir(struct buffer_head *bh,
2606 char *search_buf,
2607 int buf_size,
2608 struct inode *dir,
2609 struct ext4_filename *fname,
2610 unsigned int offset,
2611 struct ext4_dir_entry_2 **res_dir);
2612 extern int ext4_generic_delete_entry(handle_t *handle,
2613 struct inode *dir,
2614 struct ext4_dir_entry_2 *de_del,
2615 struct buffer_head *bh,
2616 void *entry_buf,
2617 int buf_size,
2618 int csum_size);
2619 extern bool ext4_empty_dir(struct inode *inode);
2620
2621 /* resize.c */
2622 extern void ext4_kvfree_array_rcu(void *to_free);
2623 extern int ext4_group_add(struct super_block *sb,
2624 struct ext4_new_group_data *input);
2625 extern int ext4_group_extend(struct super_block *sb,
2626 struct ext4_super_block *es,
2627 ext4_fsblk_t n_blocks_count);
2628 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2629
2630 /* super.c */
2631 extern struct buffer_head *ext4_sb_bread(struct super_block *sb,
2632 sector_t block, int op_flags);
2633 extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
2634 extern int ext4_calculate_overhead(struct super_block *sb);
2635 extern void ext4_superblock_csum_set(struct super_block *sb);
2636 extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2637 extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2638 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2639 ext4_group_t ngroup);
2640 extern const char *ext4_decode_error(struct super_block *sb, int errno,
2641 char nbuf[16]);
2642 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb,
2643 ext4_group_t block_group,
2644 unsigned int flags);
2645
2646 extern __printf(4, 5)
2647 void __ext4_error(struct super_block *, const char *, unsigned int,
2648 const char *, ...);
2649 extern __printf(5, 6)
2650 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2651 const char *, ...);
2652 extern __printf(5, 6)
2653 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2654 const char *, ...);
2655 extern void __ext4_std_error(struct super_block *, const char *,
2656 unsigned int, int);
2657 extern __printf(4, 5)
2658 void __ext4_abort(struct super_block *, const char *, unsigned int,
2659 const char *, ...);
2660 extern __printf(4, 5)
2661 void __ext4_warning(struct super_block *, const char *, unsigned int,
2662 const char *, ...);
2663 extern __printf(4, 5)
2664 void __ext4_warning_inode(const struct inode *inode, const char *function,
2665 unsigned int line, const char *fmt, ...);
2666 extern __printf(3, 4)
2667 void __ext4_msg(struct super_block *, const char *, const char *, ...);
2668 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2669 const char *, unsigned int, const char *);
2670 extern __printf(7, 8)
2671 void __ext4_grp_locked_error(const char *, unsigned int,
2672 struct super_block *, ext4_group_t,
2673 unsigned long, ext4_fsblk_t,
2674 const char *, ...);
2675
2676 #define EXT4_ERROR_INODE(inode, fmt, a...) \
2677 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
2678
2679 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
2680 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
2681
2682 #define EXT4_ERROR_FILE(file, block, fmt, a...) \
2683 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
2684
2685 #ifdef CONFIG_PRINTK
2686
2687 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2688 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__)
2689 #define ext4_error_file(file, func, line, block, fmt, ...) \
2690 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2691 #define ext4_error(sb, fmt, ...) \
2692 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2693 #define ext4_abort(sb, fmt, ...) \
2694 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2695 #define ext4_warning(sb, fmt, ...) \
2696 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2697 #define ext4_warning_inode(inode, fmt, ...) \
2698 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
2699 #define ext4_msg(sb, level, fmt, ...) \
2700 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2701 #define dump_mmp_msg(sb, mmp, msg) \
2702 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2703 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2704 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2705 fmt, ##__VA_ARGS__)
2706
2707 #else
2708
2709 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2710 do { \
2711 no_printk(fmt, ##__VA_ARGS__); \
2712 __ext4_error_inode(inode, "", 0, block, " "); \
2713 } while (0)
2714 #define ext4_error_file(file, func, line, block, fmt, ...) \
2715 do { \
2716 no_printk(fmt, ##__VA_ARGS__); \
2717 __ext4_error_file(file, "", 0, block, " "); \
2718 } while (0)
2719 #define ext4_error(sb, fmt, ...) \
2720 do { \
2721 no_printk(fmt, ##__VA_ARGS__); \
2722 __ext4_error(sb, "", 0, " "); \
2723 } while (0)
2724 #define ext4_abort(sb, fmt, ...) \
2725 do { \
2726 no_printk(fmt, ##__VA_ARGS__); \
2727 __ext4_abort(sb, "", 0, " "); \
2728 } while (0)
2729 #define ext4_warning(sb, fmt, ...) \
2730 do { \
2731 no_printk(fmt, ##__VA_ARGS__); \
2732 __ext4_warning(sb, "", 0, " "); \
2733 } while (0)
2734 #define ext4_warning_inode(inode, fmt, ...) \
2735 do { \
2736 no_printk(fmt, ##__VA_ARGS__); \
2737 __ext4_warning_inode(inode, "", 0, " "); \
2738 } while (0)
2739 #define ext4_msg(sb, level, fmt, ...) \
2740 do { \
2741 no_printk(fmt, ##__VA_ARGS__); \
2742 __ext4_msg(sb, "", " "); \
2743 } while (0)
2744 #define dump_mmp_msg(sb, mmp, msg) \
2745 __dump_mmp_msg(sb, mmp, "", 0, "")
2746 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2747 do { \
2748 no_printk(fmt, ##__VA_ARGS__); \
2749 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
2750 } while (0)
2751
2752 #endif
2753
2754 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2755 __u32 compat);
2756 extern int ext4_update_rocompat_feature(handle_t *handle,
2757 struct super_block *sb, __u32 rocompat);
2758 extern int ext4_update_incompat_feature(handle_t *handle,
2759 struct super_block *sb, __u32 incompat);
2760 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2761 struct ext4_group_desc *bg);
2762 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2763 struct ext4_group_desc *bg);
2764 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2765 struct ext4_group_desc *bg);
2766 extern __u32 ext4_free_group_clusters(struct super_block *sb,
2767 struct ext4_group_desc *bg);
2768 extern __u32 ext4_free_inodes_count(struct super_block *sb,
2769 struct ext4_group_desc *bg);
2770 extern __u32 ext4_used_dirs_count(struct super_block *sb,
2771 struct ext4_group_desc *bg);
2772 extern __u32 ext4_itable_unused_count(struct super_block *sb,
2773 struct ext4_group_desc *bg);
2774 extern void ext4_block_bitmap_set(struct super_block *sb,
2775 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2776 extern void ext4_inode_bitmap_set(struct super_block *sb,
2777 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2778 extern void ext4_inode_table_set(struct super_block *sb,
2779 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2780 extern void ext4_free_group_clusters_set(struct super_block *sb,
2781 struct ext4_group_desc *bg,
2782 __u32 count);
2783 extern void ext4_free_inodes_set(struct super_block *sb,
2784 struct ext4_group_desc *bg, __u32 count);
2785 extern void ext4_used_dirs_set(struct super_block *sb,
2786 struct ext4_group_desc *bg, __u32 count);
2787 extern void ext4_itable_unused_set(struct super_block *sb,
2788 struct ext4_group_desc *bg, __u32 count);
2789 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2790 struct ext4_group_desc *gdp);
2791 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2792 struct ext4_group_desc *gdp);
2793 extern int ext4_register_li_request(struct super_block *sb,
2794 ext4_group_t first_not_zeroed);
2795
ext4_has_metadata_csum(struct super_block * sb)2796 static inline int ext4_has_metadata_csum(struct super_block *sb)
2797 {
2798 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
2799 !EXT4_SB(sb)->s_chksum_driver);
2800
2801 return ext4_has_feature_metadata_csum(sb) &&
2802 (EXT4_SB(sb)->s_chksum_driver != NULL);
2803 }
2804
ext4_has_group_desc_csum(struct super_block * sb)2805 static inline int ext4_has_group_desc_csum(struct super_block *sb)
2806 {
2807 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb);
2808 }
2809
ext4_blocks_count(struct ext4_super_block * es)2810 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2811 {
2812 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2813 le32_to_cpu(es->s_blocks_count_lo);
2814 }
2815
ext4_r_blocks_count(struct ext4_super_block * es)2816 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2817 {
2818 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2819 le32_to_cpu(es->s_r_blocks_count_lo);
2820 }
2821
ext4_free_blocks_count(struct ext4_super_block * es)2822 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2823 {
2824 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2825 le32_to_cpu(es->s_free_blocks_count_lo);
2826 }
2827
ext4_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)2828 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2829 ext4_fsblk_t blk)
2830 {
2831 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2832 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2833 }
2834
ext4_free_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)2835 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2836 ext4_fsblk_t blk)
2837 {
2838 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2839 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2840 }
2841
ext4_r_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)2842 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2843 ext4_fsblk_t blk)
2844 {
2845 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2846 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2847 }
2848
ext4_isize(struct super_block * sb,struct ext4_inode * raw_inode)2849 static inline loff_t ext4_isize(struct super_block *sb,
2850 struct ext4_inode *raw_inode)
2851 {
2852 if (ext4_has_feature_largedir(sb) ||
2853 S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2854 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2855 le32_to_cpu(raw_inode->i_size_lo);
2856
2857 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2858 }
2859
ext4_isize_set(struct ext4_inode * raw_inode,loff_t i_size)2860 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2861 {
2862 raw_inode->i_size_lo = cpu_to_le32(i_size);
2863 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2864 }
2865
2866 static inline
ext4_get_group_info(struct super_block * sb,ext4_group_t group)2867 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2868 ext4_group_t group)
2869 {
2870 struct ext4_group_info **grp_info;
2871 long indexv, indexh;
2872 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
2873 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2874 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2875 grp_info = sbi_array_rcu_deref(EXT4_SB(sb), s_group_info, indexv);
2876 return grp_info[indexh];
2877 }
2878
2879 /*
2880 * Reading s_groups_count requires using smp_rmb() afterwards. See
2881 * the locking protocol documented in the comments of ext4_group_add()
2882 * in resize.c
2883 */
ext4_get_groups_count(struct super_block * sb)2884 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2885 {
2886 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
2887
2888 smp_rmb();
2889 return ngroups;
2890 }
2891
ext4_flex_group(struct ext4_sb_info * sbi,ext4_group_t block_group)2892 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2893 ext4_group_t block_group)
2894 {
2895 return block_group >> sbi->s_log_groups_per_flex;
2896 }
2897
ext4_flex_bg_size(struct ext4_sb_info * sbi)2898 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2899 {
2900 return 1 << sbi->s_log_groups_per_flex;
2901 }
2902
2903 #define ext4_std_error(sb, errno) \
2904 do { \
2905 if ((errno)) \
2906 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
2907 } while (0)
2908
2909 #ifdef CONFIG_SMP
2910 /* Each CPU can accumulate percpu_counter_batch clusters in their local
2911 * counters. So we need to make sure we have free clusters more
2912 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
2913 */
2914 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2915 #else
2916 #define EXT4_FREECLUSTERS_WATERMARK 0
2917 #endif
2918
2919 /* Update i_disksize. Requires i_mutex to avoid races with truncate */
ext4_update_i_disksize(struct inode * inode,loff_t newsize)2920 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2921 {
2922 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
2923 !inode_is_locked(inode));
2924 down_write(&EXT4_I(inode)->i_data_sem);
2925 if (newsize > EXT4_I(inode)->i_disksize)
2926 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize);
2927 up_write(&EXT4_I(inode)->i_data_sem);
2928 }
2929
2930 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
ext4_update_inode_size(struct inode * inode,loff_t newsize)2931 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
2932 {
2933 int changed = 0;
2934
2935 if (newsize > inode->i_size) {
2936 i_size_write(inode, newsize);
2937 changed = 1;
2938 }
2939 if (newsize > EXT4_I(inode)->i_disksize) {
2940 ext4_update_i_disksize(inode, newsize);
2941 changed |= 2;
2942 }
2943 return changed;
2944 }
2945
2946 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
2947 loff_t len);
2948
2949 struct ext4_group_info {
2950 unsigned long bb_state;
2951 struct rb_root bb_free_root;
2952 ext4_grpblk_t bb_first_free; /* first free block */
2953 ext4_grpblk_t bb_free; /* total free blocks */
2954 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
2955 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
2956 struct list_head bb_prealloc_list;
2957 #ifdef DOUBLE_CHECK
2958 void *bb_bitmap;
2959 #endif
2960 struct rw_semaphore alloc_sem;
2961 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
2962 * regions, index is order.
2963 * bb_counters[3] = 5 means
2964 * 5 free 8-block regions. */
2965 };
2966
2967 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
2968 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
2969 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
2970 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
2971 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \
2972 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT)
2973 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \
2974 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT)
2975
2976 #define EXT4_MB_GRP_NEED_INIT(grp) \
2977 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2978 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
2979 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2980 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
2981 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2982
2983 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
2984 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2985 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
2986 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2987 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
2988 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2989
2990 #define EXT4_MAX_CONTENTION 8
2991 #define EXT4_CONTENTION_THRESHOLD 2
2992
ext4_group_lock_ptr(struct super_block * sb,ext4_group_t group)2993 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2994 ext4_group_t group)
2995 {
2996 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2997 }
2998
2999 /*
3000 * Returns true if the filesystem is busy enough that attempts to
3001 * access the block group locks has run into contention.
3002 */
ext4_fs_is_busy(struct ext4_sb_info * sbi)3003 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
3004 {
3005 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
3006 }
3007
ext4_lock_group(struct super_block * sb,ext4_group_t group)3008 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
3009 {
3010 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
3011 if (spin_trylock(lock))
3012 /*
3013 * We're able to grab the lock right away, so drop the
3014 * lock contention counter.
3015 */
3016 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
3017 else {
3018 /*
3019 * The lock is busy, so bump the contention counter,
3020 * and then wait on the spin lock.
3021 */
3022 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
3023 EXT4_MAX_CONTENTION);
3024 spin_lock(lock);
3025 }
3026 }
3027
ext4_unlock_group(struct super_block * sb,ext4_group_t group)3028 static inline void ext4_unlock_group(struct super_block *sb,
3029 ext4_group_t group)
3030 {
3031 spin_unlock(ext4_group_lock_ptr(sb, group));
3032 }
3033
3034 /*
3035 * Block validity checking
3036 */
3037 #define ext4_check_indirect_blockref(inode, bh) \
3038 ext4_check_blockref(__func__, __LINE__, inode, \
3039 (__le32 *)(bh)->b_data, \
3040 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
3041
3042 #define ext4_ind_check_inode(inode) \
3043 ext4_check_blockref(__func__, __LINE__, inode, \
3044 EXT4_I(inode)->i_data, \
3045 EXT4_NDIR_BLOCKS)
3046
3047 /*
3048 * Inodes and files operations
3049 */
3050
3051 /* dir.c */
3052 extern const struct file_operations ext4_dir_operations;
3053
3054 /* file.c */
3055 extern const struct inode_operations ext4_file_inode_operations;
3056 extern const struct file_operations ext4_file_operations;
3057 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
3058
3059 /* inline.c */
3060 extern int ext4_get_max_inline_size(struct inode *inode);
3061 extern int ext4_find_inline_data_nolock(struct inode *inode);
3062 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
3063 unsigned int len);
3064 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
3065
3066 extern int ext4_readpage_inline(struct inode *inode, struct page *page);
3067 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3068 struct inode *inode,
3069 loff_t pos, unsigned len,
3070 unsigned flags,
3071 struct page **pagep);
3072 extern int ext4_write_inline_data_end(struct inode *inode,
3073 loff_t pos, unsigned len,
3074 unsigned copied,
3075 struct page *page);
3076 extern struct buffer_head *
3077 ext4_journalled_write_inline_data(struct inode *inode,
3078 unsigned len,
3079 struct page *page);
3080 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
3081 struct inode *inode,
3082 loff_t pos, unsigned len,
3083 unsigned flags,
3084 struct page **pagep,
3085 void **fsdata);
3086 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
3087 unsigned len, unsigned copied,
3088 struct page *page);
3089 extern int ext4_try_add_inline_entry(handle_t *handle,
3090 struct ext4_filename *fname,
3091 struct inode *dir, struct inode *inode);
3092 extern int ext4_try_create_inline_dir(handle_t *handle,
3093 struct inode *parent,
3094 struct inode *inode);
3095 extern int ext4_read_inline_dir(struct file *filp,
3096 struct dir_context *ctx,
3097 int *has_inline_data);
3098 extern int htree_inlinedir_to_tree(struct file *dir_file,
3099 struct inode *dir, ext4_lblk_t block,
3100 struct dx_hash_info *hinfo,
3101 __u32 start_hash, __u32 start_minor_hash,
3102 int *has_inline_data);
3103 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3104 struct ext4_filename *fname,
3105 struct ext4_dir_entry_2 **res_dir,
3106 int *has_inline_data);
3107 extern int ext4_delete_inline_entry(handle_t *handle,
3108 struct inode *dir,
3109 struct ext4_dir_entry_2 *de_del,
3110 struct buffer_head *bh,
3111 int *has_inline_data);
3112 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3113 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3114 struct ext4_dir_entry_2 **parent_de,
3115 int *retval);
3116 extern int ext4_inline_data_fiemap(struct inode *inode,
3117 struct fiemap_extent_info *fieinfo,
3118 int *has_inline, __u64 start, __u64 len);
3119
3120 struct iomap;
3121 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap);
3122
3123 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3124
3125 extern int ext4_convert_inline_data(struct inode *inode);
3126
ext4_has_inline_data(struct inode * inode)3127 static inline int ext4_has_inline_data(struct inode *inode)
3128 {
3129 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3130 EXT4_I(inode)->i_inline_off;
3131 }
3132
3133 /* namei.c */
3134 extern const struct inode_operations ext4_dir_inode_operations;
3135 extern const struct inode_operations ext4_special_inode_operations;
3136 extern struct dentry *ext4_get_parent(struct dentry *child);
3137 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3138 struct ext4_dir_entry_2 *de,
3139 int blocksize, int csum_size,
3140 unsigned int parent_ino, int dotdot_real_len);
3141 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t,
3142 unsigned int blocksize);
3143 extern int ext4_handle_dirty_dirent_node(handle_t *handle,
3144 struct inode *inode,
3145 struct buffer_head *bh);
3146 #define S_SHIFT 12
3147 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = {
3148 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
3149 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
3150 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
3151 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
3152 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
3153 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
3154 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
3155 };
3156
ext4_set_de_type(struct super_block * sb,struct ext4_dir_entry_2 * de,umode_t mode)3157 static inline void ext4_set_de_type(struct super_block *sb,
3158 struct ext4_dir_entry_2 *de,
3159 umode_t mode) {
3160 if (ext4_has_feature_filetype(sb))
3161 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3162 }
3163
3164 /* readpages.c */
3165 extern int ext4_mpage_readpages(struct address_space *mapping,
3166 struct list_head *pages, struct page *page,
3167 unsigned nr_pages, bool is_readahead);
3168
3169 /* symlink.c */
3170 extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3171 extern const struct inode_operations ext4_symlink_inode_operations;
3172 extern const struct inode_operations ext4_fast_symlink_inode_operations;
3173
3174 /* sysfs.c */
3175 extern int ext4_register_sysfs(struct super_block *sb);
3176 extern void ext4_unregister_sysfs(struct super_block *sb);
3177 extern int __init ext4_init_sysfs(void);
3178 extern void ext4_exit_sysfs(void);
3179
3180 /* block_validity */
3181 extern void ext4_release_system_zone(struct super_block *sb);
3182 extern int ext4_setup_system_zone(struct super_block *sb);
3183 extern int __init ext4_init_system_zone(void);
3184 extern void ext4_exit_system_zone(void);
3185 extern int ext4_inode_block_valid(struct inode *inode,
3186 ext4_fsblk_t start_blk,
3187 unsigned int count);
3188 extern int ext4_check_blockref(const char *, unsigned int,
3189 struct inode *, __le32 *, unsigned int);
3190
3191 /* extents.c */
3192 struct ext4_ext_path;
3193 struct ext4_extent;
3194
3195 /*
3196 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3197 * __le32.
3198 */
3199 #define EXT_MAX_BLOCKS 0xffffffff
3200
3201 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
3202 extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
3203 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3204 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3205 struct ext4_map_blocks *map, int flags);
3206 extern int ext4_ext_truncate(handle_t *, struct inode *);
3207 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3208 ext4_lblk_t end);
3209 extern void ext4_ext_init(struct super_block *);
3210 extern void ext4_ext_release(struct super_block *);
3211 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3212 loff_t len);
3213 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3214 loff_t offset, ssize_t len);
3215 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3216 struct ext4_map_blocks *map, int flags);
3217 extern int ext4_ext_calc_metadata_amount(struct inode *inode,
3218 ext4_lblk_t lblocks);
3219 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3220 int num,
3221 struct ext4_ext_path *path);
3222 extern int ext4_can_extents_be_merged(struct inode *inode,
3223 struct ext4_extent *ex1,
3224 struct ext4_extent *ex2);
3225 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3226 struct ext4_ext_path **,
3227 struct ext4_extent *, int);
3228 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3229 struct ext4_ext_path **,
3230 int flags);
3231 extern void ext4_ext_drop_refs(struct ext4_ext_path *);
3232 extern int ext4_ext_check_inode(struct inode *inode);
3233 extern int ext4_find_delalloc_range(struct inode *inode,
3234 ext4_lblk_t lblk_start,
3235 ext4_lblk_t lblk_end);
3236 extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk);
3237 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3238 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3239 __u64 start, __u64 len);
3240 extern int ext4_ext_precache(struct inode *inode);
3241 extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len);
3242 extern int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len);
3243 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3244 struct inode *inode2, ext4_lblk_t lblk1,
3245 ext4_lblk_t lblk2, ext4_lblk_t count,
3246 int mark_unwritten,int *err);
3247
3248 /* move_extent.c */
3249 extern void ext4_double_down_write_data_sem(struct inode *first,
3250 struct inode *second);
3251 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3252 struct inode *donor_inode);
3253 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3254 __u64 start_orig, __u64 start_donor,
3255 __u64 len, __u64 *moved_len);
3256
3257 /* page-io.c */
3258 extern int __init ext4_init_pageio(void);
3259 extern void ext4_exit_pageio(void);
3260 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3261 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3262 extern int ext4_put_io_end(ext4_io_end_t *io_end);
3263 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3264 extern void ext4_io_submit_init(struct ext4_io_submit *io,
3265 struct writeback_control *wbc);
3266 extern void ext4_end_io_rsv_work(struct work_struct *work);
3267 extern void ext4_io_submit(struct ext4_io_submit *io);
3268 extern int ext4_bio_write_page(struct ext4_io_submit *io,
3269 struct page *page,
3270 int len,
3271 struct writeback_control *wbc,
3272 bool keep_towrite);
3273
3274 /* mmp.c */
3275 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3276
3277 /*
3278 * Add new method to test whether block and inode bitmaps are properly
3279 * initialized. With uninit_bg reading the block from disk is not enough
3280 * to mark the bitmap uptodate. We need to also zero-out the bitmap
3281 */
3282 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3283
bitmap_uptodate(struct buffer_head * bh)3284 static inline int bitmap_uptodate(struct buffer_head *bh)
3285 {
3286 return (buffer_uptodate(bh) &&
3287 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3288 }
set_bitmap_uptodate(struct buffer_head * bh)3289 static inline void set_bitmap_uptodate(struct buffer_head *bh)
3290 {
3291 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3292 }
3293
3294 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
3295
3296 /* For ioend & aio unwritten conversion wait queues */
3297 #define EXT4_WQ_HASH_SZ 37
3298 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
3299 EXT4_WQ_HASH_SZ])
3300 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3301
3302 extern int ext4_resize_begin(struct super_block *sb);
3303 extern void ext4_resize_end(struct super_block *sb);
3304
ext4_set_io_unwritten_flag(struct inode * inode,struct ext4_io_end * io_end)3305 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
3306 struct ext4_io_end *io_end)
3307 {
3308 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
3309 io_end->flag |= EXT4_IO_END_UNWRITTEN;
3310 atomic_inc(&EXT4_I(inode)->i_unwritten);
3311 }
3312 }
3313
ext4_clear_io_unwritten_flag(ext4_io_end_t * io_end)3314 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3315 {
3316 struct inode *inode = io_end->inode;
3317
3318 if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
3319 io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3320 /* Wake up anyone waiting on unwritten extent conversion */
3321 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
3322 wake_up_all(ext4_ioend_wq(inode));
3323 }
3324 }
3325
3326 extern const struct iomap_ops ext4_iomap_ops;
3327
3328 #endif /* __KERNEL__ */
3329
3330 #define EFSBADCRC EBADMSG /* Bad CRC detected */
3331 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
3332
3333 #endif /* _EXT4_H */
3334