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 * Flags for ext4_io_end->flags
199 */
200 #define EXT4_IO_END_UNWRITTEN 0x0001
201
202 /*
203 * For converting unwritten extents on a work queue. 'handle' is used for
204 * buffered writeback.
205 */
206 typedef struct ext4_io_end {
207 struct list_head list; /* per-file finished IO list */
208 handle_t *handle; /* handle reserved for extent
209 * conversion */
210 struct inode *inode; /* file being written to */
211 struct bio *bio; /* Linked list of completed
212 * bios covering the extent */
213 unsigned int flag; /* unwritten or not */
214 atomic_t count; /* reference counter */
215 loff_t offset; /* offset in the file */
216 ssize_t size; /* size of the extent */
217 } ext4_io_end_t;
218
219 struct ext4_io_submit {
220 struct writeback_control *io_wbc;
221 struct bio *io_bio;
222 ext4_io_end_t *io_end;
223 sector_t io_next_block;
224 };
225
226 /*
227 * Special inodes numbers
228 */
229 #define EXT4_BAD_INO 1 /* Bad blocks inode */
230 #define EXT4_ROOT_INO 2 /* Root inode */
231 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */
232 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
233 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
234 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
235 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
236 #define EXT4_JOURNAL_INO 8 /* Journal inode */
237
238 /* First non-reserved inode for old ext4 filesystems */
239 #define EXT4_GOOD_OLD_FIRST_INO 11
240
241 /*
242 * Maximal count of links to a file
243 */
244 #define EXT4_LINK_MAX 65000
245
246 /*
247 * Macro-instructions used to manage several block sizes
248 */
249 #define EXT4_MIN_BLOCK_SIZE 1024
250 #define EXT4_MAX_BLOCK_SIZE 65536
251 #define EXT4_MIN_BLOCK_LOG_SIZE 10
252 #define EXT4_MAX_BLOCK_LOG_SIZE 16
253 #define EXT4_MAX_CLUSTER_LOG_SIZE 30
254 #ifdef __KERNEL__
255 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
256 #else
257 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
258 #endif
259 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
260 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
261 EXT4_SB(s)->s_cluster_bits)
262 #ifdef __KERNEL__
263 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
264 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
265 #else
266 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
267 #endif
268 #ifdef __KERNEL__
269 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
270 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
271 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
272 #else
273 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
274 EXT4_GOOD_OLD_INODE_SIZE : \
275 (s)->s_inode_size)
276 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
277 EXT4_GOOD_OLD_FIRST_INO : \
278 (s)->s_first_ino)
279 #endif
280 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
281 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \
282 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \
283 blkbits))
284
285 /* Translate a block number to a cluster number */
286 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
287 /* Translate a cluster number to a block number */
288 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
289 /* Translate # of blks to # of clusters */
290 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
291 (sbi)->s_cluster_bits)
292 /* Mask out the low bits to get the starting block of the cluster */
293 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
294 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
295 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
296 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
297 /* Get the cluster offset */
298 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
299 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
300 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
301 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
302
303 /*
304 * Structure of a blocks group descriptor
305 */
306 struct ext4_group_desc
307 {
308 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
309 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
310 __le32 bg_inode_table_lo; /* Inodes table block */
311 __le16 bg_free_blocks_count_lo;/* Free blocks count */
312 __le16 bg_free_inodes_count_lo;/* Free inodes count */
313 __le16 bg_used_dirs_count_lo; /* Directories count */
314 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
315 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
316 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
317 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
318 __le16 bg_itable_unused_lo; /* Unused inodes count */
319 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
320 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
321 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
322 __le32 bg_inode_table_hi; /* Inodes table block MSB */
323 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
324 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
325 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
326 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
327 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
328 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
329 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
330 __u32 bg_reserved;
331 };
332
333 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
334 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
335 sizeof(__le16))
336 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
337 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
338 sizeof(__le16))
339
340 /*
341 * Structure of a flex block group info
342 */
343
344 struct flex_groups {
345 atomic64_t free_clusters;
346 atomic_t free_inodes;
347 atomic_t used_dirs;
348 };
349
350 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
351 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
352 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
353
354 /*
355 * Macro-instructions used to manage group descriptors
356 */
357 #define EXT4_MIN_DESC_SIZE 32
358 #define EXT4_MIN_DESC_SIZE_64BIT 64
359 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
360 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
361 #ifdef __KERNEL__
362 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
363 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
364 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
365 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
366 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
367 #else
368 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
369 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
370 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
371 #endif
372
373 /*
374 * Constants relative to the data blocks
375 */
376 #define EXT4_NDIR_BLOCKS 12
377 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
378 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
379 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
380 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
381
382 /*
383 * Inode flags
384 */
385 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
386 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
387 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
388 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
389 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
390 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
391 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
392 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
393 /* Reserved for compression usage... */
394 #define EXT4_DIRTY_FL 0x00000100
395 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
396 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
397 /* nb: was previously EXT2_ECOMPR_FL */
398 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */
399 /* End compression flags --- maybe not all used */
400 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
401 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
402 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
403 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
404 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
405 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
406 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
407 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
408 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
409 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
410 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
411 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
412 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
413
414 #define EXT4_FL_USER_VISIBLE 0x304BDFFF /* User visible flags */
415 #define EXT4_FL_USER_MODIFIABLE 0x204BC0FF /* User modifiable flags */
416
417 /* Flags we can manipulate with through EXT4_IOC_FSSETXATTR */
418 #define EXT4_FL_XFLAG_VISIBLE (EXT4_SYNC_FL | \
419 EXT4_IMMUTABLE_FL | \
420 EXT4_APPEND_FL | \
421 EXT4_NODUMP_FL | \
422 EXT4_NOATIME_FL | \
423 EXT4_PROJINHERIT_FL)
424
425 /* Flags that should be inherited by new inodes from their parent. */
426 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
427 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
428 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
429 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
430 EXT4_PROJINHERIT_FL)
431
432 /* Flags that are appropriate for regular files (all but dir-specific ones). */
433 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
434
435 /* Flags that are appropriate for non-directories/regular files. */
436 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
437
438 /* Mask out flags that are inappropriate for the given type of inode. */
ext4_mask_flags(umode_t mode,__u32 flags)439 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
440 {
441 if (S_ISDIR(mode))
442 return flags;
443 else if (S_ISREG(mode))
444 return flags & EXT4_REG_FLMASK;
445 else
446 return flags & EXT4_OTHER_FLMASK;
447 }
448
449 /*
450 * Inode flags used for atomic set/get
451 */
452 enum {
453 EXT4_INODE_SECRM = 0, /* Secure deletion */
454 EXT4_INODE_UNRM = 1, /* Undelete */
455 EXT4_INODE_COMPR = 2, /* Compress file */
456 EXT4_INODE_SYNC = 3, /* Synchronous updates */
457 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
458 EXT4_INODE_APPEND = 5, /* writes to file may only append */
459 EXT4_INODE_NODUMP = 6, /* do not dump file */
460 EXT4_INODE_NOATIME = 7, /* do not update atime */
461 /* Reserved for compression usage... */
462 EXT4_INODE_DIRTY = 8,
463 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
464 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
465 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */
466 /* End compression flags --- maybe not all used */
467 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
468 EXT4_INODE_IMAGIC = 13, /* AFS directory */
469 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
470 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
471 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
472 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
473 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
474 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
475 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
476 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
477 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
478 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */
479 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
480 };
481
482 /*
483 * Since it's pretty easy to mix up bit numbers and hex values, we use a
484 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
485 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
486 * any extra space in the compiled kernel image, otherwise, the build will fail.
487 * It's important that these values are the same, since we are using
488 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
489 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
490 * values found in ext2, ext3 and ext4 filesystems, and of course the values
491 * defined in e2fsprogs.
492 *
493 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
494 */
495 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
496 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
497
ext4_check_flag_values(void)498 static inline void ext4_check_flag_values(void)
499 {
500 CHECK_FLAG_VALUE(SECRM);
501 CHECK_FLAG_VALUE(UNRM);
502 CHECK_FLAG_VALUE(COMPR);
503 CHECK_FLAG_VALUE(SYNC);
504 CHECK_FLAG_VALUE(IMMUTABLE);
505 CHECK_FLAG_VALUE(APPEND);
506 CHECK_FLAG_VALUE(NODUMP);
507 CHECK_FLAG_VALUE(NOATIME);
508 CHECK_FLAG_VALUE(DIRTY);
509 CHECK_FLAG_VALUE(COMPRBLK);
510 CHECK_FLAG_VALUE(NOCOMPR);
511 CHECK_FLAG_VALUE(ENCRYPT);
512 CHECK_FLAG_VALUE(INDEX);
513 CHECK_FLAG_VALUE(IMAGIC);
514 CHECK_FLAG_VALUE(JOURNAL_DATA);
515 CHECK_FLAG_VALUE(NOTAIL);
516 CHECK_FLAG_VALUE(DIRSYNC);
517 CHECK_FLAG_VALUE(TOPDIR);
518 CHECK_FLAG_VALUE(HUGE_FILE);
519 CHECK_FLAG_VALUE(EXTENTS);
520 CHECK_FLAG_VALUE(EA_INODE);
521 CHECK_FLAG_VALUE(EOFBLOCKS);
522 CHECK_FLAG_VALUE(INLINE_DATA);
523 CHECK_FLAG_VALUE(PROJINHERIT);
524 CHECK_FLAG_VALUE(RESERVED);
525 }
526
527 /* Used to pass group descriptor data when online resize is done */
528 struct ext4_new_group_input {
529 __u32 group; /* Group number for this data */
530 __u64 block_bitmap; /* Absolute block number of block bitmap */
531 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
532 __u64 inode_table; /* Absolute block number of inode table start */
533 __u32 blocks_count; /* Total number of blocks in this group */
534 __u16 reserved_blocks; /* Number of reserved blocks in this group */
535 __u16 unused;
536 };
537
538 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
539 struct compat_ext4_new_group_input {
540 u32 group;
541 compat_u64 block_bitmap;
542 compat_u64 inode_bitmap;
543 compat_u64 inode_table;
544 u32 blocks_count;
545 u16 reserved_blocks;
546 u16 unused;
547 };
548 #endif
549
550 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
551 struct ext4_new_group_data {
552 __u32 group;
553 __u64 block_bitmap;
554 __u64 inode_bitmap;
555 __u64 inode_table;
556 __u32 blocks_count;
557 __u16 reserved_blocks;
558 __u16 mdata_blocks;
559 __u32 free_clusters_count;
560 };
561
562 /* Indexes used to index group tables in ext4_new_group_data */
563 enum {
564 BLOCK_BITMAP = 0, /* block bitmap */
565 INODE_BITMAP, /* inode bitmap */
566 INODE_TABLE, /* inode tables */
567 GROUP_TABLE_COUNT,
568 };
569
570 /*
571 * Flags used by ext4_map_blocks()
572 */
573 /* Allocate any needed blocks and/or convert an unwritten
574 extent to be an initialized ext4 */
575 #define EXT4_GET_BLOCKS_CREATE 0x0001
576 /* Request the creation of an unwritten extent */
577 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002
578 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\
579 EXT4_GET_BLOCKS_CREATE)
580 /* Caller is from the delayed allocation writeout path
581 * finally doing the actual allocation of delayed blocks */
582 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
583 /* caller is from the direct IO path, request to creation of an
584 unwritten extents if not allocated, split the unwritten
585 extent if blocks has been preallocated already*/
586 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
587 #define EXT4_GET_BLOCKS_CONVERT 0x0010
588 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
589 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
590 /* Convert extent to initialized after IO complete */
591 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
592 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
593 /* Eventual metadata allocation (due to growing extent tree)
594 * should not fail, so try to use reserved blocks for that.*/
595 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
596 /* Don't normalize allocation size (used for fallocate) */
597 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
598 /* Request will not result in inode size update (user for fallocate) */
599 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080
600 /* Convert written extents to unwritten */
601 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100
602 /* Write zeros to newly created written extents */
603 #define EXT4_GET_BLOCKS_ZERO 0x0200
604 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\
605 EXT4_GET_BLOCKS_ZERO)
606 /* Caller will submit data before dropping transaction handle. This
607 * allows jbd2 to avoid submitting data before commit. */
608 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400
609
610 /*
611 * The bit position of these flags must not overlap with any of the
612 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(),
613 * read_extent_tree_block(), ext4_split_extent_at(),
614 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
615 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
616 * caching the extents when reading from the extent tree while a
617 * truncate or punch hole operation is in progress.
618 */
619 #define EXT4_EX_NOCACHE 0x40000000
620 #define EXT4_EX_FORCE_CACHE 0x20000000
621
622 /*
623 * Flags used by ext4_free_blocks
624 */
625 #define EXT4_FREE_BLOCKS_METADATA 0x0001
626 #define EXT4_FREE_BLOCKS_FORGET 0x0002
627 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
628 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
629 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
630 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
631
632 /*
633 * ioctl commands
634 */
635 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
636 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
637 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
638 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
639 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
640 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
641 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
642 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
643 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
644 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
645 #define EXT4_IOC_MIGRATE _IO('f', 9)
646 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
647 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
648 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
649 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
650 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
651 #define EXT4_IOC_SWAP_BOOT _IO('f', 17)
652 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18)
653 #define EXT4_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY
654 #define EXT4_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT
655 #define EXT4_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY
656
657 #define EXT4_IOC_FSGETXATTR FS_IOC_FSGETXATTR
658 #define EXT4_IOC_FSSETXATTR FS_IOC_FSSETXATTR
659
660 #define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32)
661
662 /*
663 * Flags for going down operation
664 */
665 #define EXT4_GOING_FLAGS_DEFAULT 0x0 /* going down */
666 #define EXT4_GOING_FLAGS_LOGFLUSH 0x1 /* flush log but not data */
667 #define EXT4_GOING_FLAGS_NOLOGFLUSH 0x2 /* don't flush log nor data */
668
669
670 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
671 /*
672 * ioctl commands in 32 bit emulation
673 */
674 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
675 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
676 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
677 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
678 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
679 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
680 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
681 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
682 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
683 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
684 #endif
685
686 /* Max physical block we can address w/o extents */
687 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
688
689 /* Max logical block we can support */
690 #define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFF
691
692 /*
693 * Structure of an inode on the disk
694 */
695 struct ext4_inode {
696 __le16 i_mode; /* File mode */
697 __le16 i_uid; /* Low 16 bits of Owner Uid */
698 __le32 i_size_lo; /* Size in bytes */
699 __le32 i_atime; /* Access time */
700 __le32 i_ctime; /* Inode Change time */
701 __le32 i_mtime; /* Modification time */
702 __le32 i_dtime; /* Deletion Time */
703 __le16 i_gid; /* Low 16 bits of Group Id */
704 __le16 i_links_count; /* Links count */
705 __le32 i_blocks_lo; /* Blocks count */
706 __le32 i_flags; /* File flags */
707 union {
708 struct {
709 __le32 l_i_version;
710 } linux1;
711 struct {
712 __u32 h_i_translator;
713 } hurd1;
714 struct {
715 __u32 m_i_reserved1;
716 } masix1;
717 } osd1; /* OS dependent 1 */
718 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
719 __le32 i_generation; /* File version (for NFS) */
720 __le32 i_file_acl_lo; /* File ACL */
721 __le32 i_size_high;
722 __le32 i_obso_faddr; /* Obsoleted fragment address */
723 union {
724 struct {
725 __le16 l_i_blocks_high; /* were l_i_reserved1 */
726 __le16 l_i_file_acl_high;
727 __le16 l_i_uid_high; /* these 2 fields */
728 __le16 l_i_gid_high; /* were reserved2[0] */
729 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
730 __le16 l_i_reserved;
731 } linux2;
732 struct {
733 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
734 __u16 h_i_mode_high;
735 __u16 h_i_uid_high;
736 __u16 h_i_gid_high;
737 __u32 h_i_author;
738 } hurd2;
739 struct {
740 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
741 __le16 m_i_file_acl_high;
742 __u32 m_i_reserved2[2];
743 } masix2;
744 } osd2; /* OS dependent 2 */
745 __le16 i_extra_isize;
746 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
747 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
748 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
749 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
750 __le32 i_crtime; /* File Creation time */
751 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
752 __le32 i_version_hi; /* high 32 bits for 64-bit version */
753 __le32 i_projid; /* Project ID */
754 };
755
756 struct move_extent {
757 __u32 reserved; /* should be zero */
758 __u32 donor_fd; /* donor file descriptor */
759 __u64 orig_start; /* logical start offset in block for orig */
760 __u64 donor_start; /* logical start offset in block for donor */
761 __u64 len; /* block length to be moved */
762 __u64 moved_len; /* moved block length */
763 };
764
765 #define EXT4_EPOCH_BITS 2
766 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
767 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
768
769 /*
770 * Extended fields will fit into an inode if the filesystem was formatted
771 * with large inodes (-I 256 or larger) and there are not currently any EAs
772 * consuming all of the available space. For new inodes we always reserve
773 * enough space for the kernel's known extended fields, but for inodes
774 * created with an old kernel this might not have been the case. None of
775 * the extended inode fields is critical for correct filesystem operation.
776 * This macro checks if a certain field fits in the inode. Note that
777 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
778 */
779 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
780 ((offsetof(typeof(*ext4_inode), field) + \
781 sizeof((ext4_inode)->field)) \
782 <= (EXT4_GOOD_OLD_INODE_SIZE + \
783 (einode)->i_extra_isize)) \
784
785 /*
786 * We use an encoding that preserves the times for extra epoch "00":
787 *
788 * extra msb of adjust for signed
789 * epoch 32-bit 32-bit tv_sec to
790 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
791 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
792 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
793 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
794 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
795 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
796 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
797 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
798 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
799 *
800 * Note that previous versions of the kernel on 64-bit systems would
801 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
802 * 1970. e2fsck will correct this, assuming that it is run on the
803 * affected filesystem before 2242.
804 */
805
ext4_encode_extra_time(struct timespec64 * time)806 static inline __le32 ext4_encode_extra_time(struct timespec64 *time)
807 {
808 u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK;
809 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
810 }
811
ext4_decode_extra_time(struct timespec64 * time,__le32 extra)812 static inline void ext4_decode_extra_time(struct timespec64 *time,
813 __le32 extra)
814 {
815 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK))) {
816
817 #if 1
818 /* Handle legacy encoding of pre-1970 dates with epoch
819 * bits 1,1. (This backwards compatibility may be removed
820 * at the discretion of the ext4 developers.)
821 */
822 u64 extra_bits = le32_to_cpu(extra) & EXT4_EPOCH_MASK;
823 if (extra_bits == 3 && ((time->tv_sec) & 0x80000000) != 0)
824 extra_bits = 0;
825 time->tv_sec += extra_bits << 32;
826 #else
827 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
828 #endif
829 }
830 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
831 }
832
833 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
834 do { \
835 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
836 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {\
837 (raw_inode)->xtime ## _extra = \
838 ext4_encode_extra_time(&(inode)->xtime); \
839 } \
840 } while (0)
841
842 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
843 do { \
844 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
845 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
846 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
847 (raw_inode)->xtime ## _extra = \
848 ext4_encode_extra_time(&(einode)->xtime); \
849 } while (0)
850
851 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
852 do { \
853 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
854 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \
855 ext4_decode_extra_time(&(inode)->xtime, \
856 raw_inode->xtime ## _extra); \
857 } \
858 else \
859 (inode)->xtime.tv_nsec = 0; \
860 } while (0)
861
862
863 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
864 do { \
865 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
866 (einode)->xtime.tv_sec = \
867 (signed)le32_to_cpu((raw_inode)->xtime); \
868 else \
869 (einode)->xtime.tv_sec = 0; \
870 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
871 ext4_decode_extra_time(&(einode)->xtime, \
872 raw_inode->xtime ## _extra); \
873 else \
874 (einode)->xtime.tv_nsec = 0; \
875 } while (0)
876
877 #define i_disk_version osd1.linux1.l_i_version
878
879 #if defined(__KERNEL__) || defined(__linux__)
880 #define i_reserved1 osd1.linux1.l_i_reserved1
881 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
882 #define i_blocks_high osd2.linux2.l_i_blocks_high
883 #define i_uid_low i_uid
884 #define i_gid_low i_gid
885 #define i_uid_high osd2.linux2.l_i_uid_high
886 #define i_gid_high osd2.linux2.l_i_gid_high
887 #define i_checksum_lo osd2.linux2.l_i_checksum_lo
888
889 #elif defined(__GNU__)
890
891 #define i_translator osd1.hurd1.h_i_translator
892 #define i_uid_high osd2.hurd2.h_i_uid_high
893 #define i_gid_high osd2.hurd2.h_i_gid_high
894 #define i_author osd2.hurd2.h_i_author
895
896 #elif defined(__masix__)
897
898 #define i_reserved1 osd1.masix1.m_i_reserved1
899 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
900 #define i_reserved2 osd2.masix2.m_i_reserved2
901
902 #endif /* defined(__KERNEL__) || defined(__linux__) */
903
904 #include "extents_status.h"
905
906 /*
907 * Lock subclasses for i_data_sem in the ext4_inode_info structure.
908 *
909 * These are needed to avoid lockdep false positives when we need to
910 * allocate blocks to the quota inode during ext4_map_blocks(), while
911 * holding i_data_sem for a normal (non-quota) inode. Since we don't
912 * do quota tracking for the quota inode, this avoids deadlock (as
913 * well as infinite recursion, since it isn't turtles all the way
914 * down...)
915 *
916 * I_DATA_SEM_NORMAL - Used for most inodes
917 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode
918 * where the second inode has larger inode number
919 * than the first
920 * I_DATA_SEM_QUOTA - Used for quota inodes only
921 */
922 enum {
923 I_DATA_SEM_NORMAL = 0,
924 I_DATA_SEM_OTHER,
925 I_DATA_SEM_QUOTA,
926 };
927
928
929 /*
930 * fourth extended file system inode data in memory
931 */
932 struct ext4_inode_info {
933 __le32 i_data[15]; /* unconverted */
934 __u32 i_dtime;
935 ext4_fsblk_t i_file_acl;
936
937 /*
938 * i_block_group is the number of the block group which contains
939 * this file's inode. Constant across the lifetime of the inode,
940 * it is used for making block allocation decisions - we try to
941 * place a file's data blocks near its inode block, and new inodes
942 * near to their parent directory's inode.
943 */
944 ext4_group_t i_block_group;
945 ext4_lblk_t i_dir_start_lookup;
946 #if (BITS_PER_LONG < 64)
947 unsigned long i_state_flags; /* Dynamic state flags */
948 #endif
949 unsigned long i_flags;
950
951 /*
952 * Extended attributes can be read independently of the main file
953 * data. Taking i_mutex even when reading would cause contention
954 * between readers of EAs and writers of regular file data, so
955 * instead we synchronize on xattr_sem when reading or changing
956 * EAs.
957 */
958 struct rw_semaphore xattr_sem;
959
960 struct list_head i_orphan; /* unlinked but open inodes */
961
962 /*
963 * i_disksize keeps track of what the inode size is ON DISK, not
964 * in memory. During truncate, i_size is set to the new size by
965 * the VFS prior to calling ext4_truncate(), but the filesystem won't
966 * set i_disksize to 0 until the truncate is actually under way.
967 *
968 * The intent is that i_disksize always represents the blocks which
969 * are used by this file. This allows recovery to restart truncate
970 * on orphans if we crash during truncate. We actually write i_disksize
971 * into the on-disk inode when writing inodes out, instead of i_size.
972 *
973 * The only time when i_disksize and i_size may be different is when
974 * a truncate is in progress. The only things which change i_disksize
975 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
976 */
977 loff_t i_disksize;
978
979 /*
980 * i_data_sem is for serialising ext4_truncate() against
981 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
982 * data tree are chopped off during truncate. We can't do that in
983 * ext4 because whenever we perform intermediate commits during
984 * truncate, the inode and all the metadata blocks *must* be in a
985 * consistent state which allows truncation of the orphans to restart
986 * during recovery. Hence we must fix the get_block-vs-truncate race
987 * by other means, so we have i_data_sem.
988 */
989 struct rw_semaphore i_data_sem;
990 /*
991 * i_mmap_sem is for serializing page faults with truncate / punch hole
992 * operations. We have to make sure that new page cannot be faulted in
993 * a section of the inode that is being punched. We cannot easily use
994 * i_data_sem for this since we need protection for the whole punch
995 * operation and i_data_sem ranks below transaction start so we have
996 * to occasionally drop it.
997 */
998 struct rw_semaphore i_mmap_sem;
999 struct inode vfs_inode;
1000 struct jbd2_inode *jinode;
1001
1002 spinlock_t i_raw_lock; /* protects updates to the raw inode */
1003
1004 /*
1005 * File creation time. Its function is same as that of
1006 * struct timespec64 i_{a,c,m}time in the generic inode.
1007 */
1008 struct timespec64 i_crtime;
1009
1010 /* mballoc */
1011 struct list_head i_prealloc_list;
1012 spinlock_t i_prealloc_lock;
1013
1014 /* extents status tree */
1015 struct ext4_es_tree i_es_tree;
1016 rwlock_t i_es_lock;
1017 struct list_head i_es_list;
1018 unsigned int i_es_all_nr; /* protected by i_es_lock */
1019 unsigned int i_es_shk_nr; /* protected by i_es_lock */
1020 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for
1021 extents to shrink. Protected by
1022 i_es_lock */
1023
1024 /* ialloc */
1025 ext4_group_t i_last_alloc_group;
1026
1027 /* allocation reservation info for delalloc */
1028 /* In case of bigalloc, this refer to clusters rather than blocks */
1029 unsigned int i_reserved_data_blocks;
1030 ext4_lblk_t i_da_metadata_calc_last_lblock;
1031 int i_da_metadata_calc_len;
1032
1033 /* on-disk additional length */
1034 __u16 i_extra_isize;
1035
1036 /* Indicate the inline data space. */
1037 u16 i_inline_off;
1038 u16 i_inline_size;
1039
1040 #ifdef CONFIG_QUOTA
1041 /* quota space reservation, managed internally by quota code */
1042 qsize_t i_reserved_quota;
1043 #endif
1044
1045 /* Lock protecting lists below */
1046 spinlock_t i_completed_io_lock;
1047 /*
1048 * Completed IOs that need unwritten extents handling and have
1049 * transaction reserved
1050 */
1051 struct list_head i_rsv_conversion_list;
1052 struct work_struct i_rsv_conversion_work;
1053 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1054
1055 spinlock_t i_block_reservation_lock;
1056
1057 /*
1058 * Transactions that contain inode's metadata needed to complete
1059 * fsync and fdatasync, respectively.
1060 */
1061 tid_t i_sync_tid;
1062 tid_t i_datasync_tid;
1063
1064 #ifdef CONFIG_QUOTA
1065 struct dquot *i_dquot[MAXQUOTAS];
1066 #endif
1067
1068 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1069 __u32 i_csum_seed;
1070
1071 kprojid_t i_projid;
1072 };
1073
1074 /*
1075 * File system states
1076 */
1077 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
1078 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
1079 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
1080
1081 /*
1082 * Misc. filesystem flags
1083 */
1084 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
1085 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
1086 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
1087
1088 /*
1089 * Mount flags set via mount options or defaults
1090 */
1091 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */
1092 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
1093 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
1094 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
1095 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
1096 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
1097 #define EXT4_MOUNT_ERRORS_MASK 0x00070
1098 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
1099 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
1100 #ifdef CONFIG_FS_DAX
1101 #define EXT4_MOUNT_DAX 0x00200 /* Direct Access */
1102 #else
1103 #define EXT4_MOUNT_DAX 0
1104 #endif
1105 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
1106 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
1107 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
1108 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
1109 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
1110 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
1111 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
1112 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
1113 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
1114 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
1115 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */
1116 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota,
1117 * enable enforcement for hidden
1118 * quota files */
1119 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable
1120 * enforcement for hidden quota
1121 * files */
1122 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota
1123 * enforcement */
1124 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
1125 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
1126 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1127 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */
1128 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
1129 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
1130 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
1131 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
1132 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
1133
1134 /*
1135 * Mount flags set either automatically (could not be set by mount option)
1136 * based on per file system feature or property or in special cases such as
1137 * distinguishing between explicit mount option definition and default.
1138 */
1139 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1140 specified delalloc */
1141 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1142 size of blocksize * 8
1143 blocks */
1144 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1145 file systems */
1146
1147 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly
1148 specified journal checksum */
1149
1150 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1151 ~EXT4_MOUNT_##opt
1152 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1153 EXT4_MOUNT_##opt
1154 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1155 EXT4_MOUNT_##opt)
1156
1157 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1158 ~EXT4_MOUNT2_##opt
1159 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1160 EXT4_MOUNT2_##opt
1161 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1162 EXT4_MOUNT2_##opt)
1163
1164 #define ext4_test_and_set_bit __test_and_set_bit_le
1165 #define ext4_set_bit __set_bit_le
1166 #define ext4_set_bit_atomic ext2_set_bit_atomic
1167 #define ext4_test_and_clear_bit __test_and_clear_bit_le
1168 #define ext4_clear_bit __clear_bit_le
1169 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
1170 #define ext4_test_bit test_bit_le
1171 #define ext4_find_next_zero_bit find_next_zero_bit_le
1172 #define ext4_find_next_bit find_next_bit_le
1173
1174 extern void ext4_set_bits(void *bm, int cur, int len);
1175
1176 /*
1177 * Maximal mount counts between two filesystem checks
1178 */
1179 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1180 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1181
1182 /*
1183 * Behaviour when detecting errors
1184 */
1185 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1186 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1187 #define EXT4_ERRORS_PANIC 3 /* Panic */
1188 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1189
1190 /* Metadata checksum algorithm codes */
1191 #define EXT4_CRC32C_CHKSUM 1
1192
1193 /*
1194 * Structure of the super block
1195 */
1196 struct ext4_super_block {
1197 /*00*/ __le32 s_inodes_count; /* Inodes count */
1198 __le32 s_blocks_count_lo; /* Blocks count */
1199 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1200 __le32 s_free_blocks_count_lo; /* Free blocks count */
1201 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1202 __le32 s_first_data_block; /* First Data Block */
1203 __le32 s_log_block_size; /* Block size */
1204 __le32 s_log_cluster_size; /* Allocation cluster size */
1205 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1206 __le32 s_clusters_per_group; /* # Clusters per group */
1207 __le32 s_inodes_per_group; /* # Inodes per group */
1208 __le32 s_mtime; /* Mount time */
1209 /*30*/ __le32 s_wtime; /* Write time */
1210 __le16 s_mnt_count; /* Mount count */
1211 __le16 s_max_mnt_count; /* Maximal mount count */
1212 __le16 s_magic; /* Magic signature */
1213 __le16 s_state; /* File system state */
1214 __le16 s_errors; /* Behaviour when detecting errors */
1215 __le16 s_minor_rev_level; /* minor revision level */
1216 /*40*/ __le32 s_lastcheck; /* time of last check */
1217 __le32 s_checkinterval; /* max. time between checks */
1218 __le32 s_creator_os; /* OS */
1219 __le32 s_rev_level; /* Revision level */
1220 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1221 __le16 s_def_resgid; /* Default gid for reserved blocks */
1222 /*
1223 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1224 *
1225 * Note: the difference between the compatible feature set and
1226 * the incompatible feature set is that if there is a bit set
1227 * in the incompatible feature set that the kernel doesn't
1228 * know about, it should refuse to mount the filesystem.
1229 *
1230 * e2fsck's requirements are more strict; if it doesn't know
1231 * about a feature in either the compatible or incompatible
1232 * feature set, it must abort and not try to meddle with
1233 * things it doesn't understand...
1234 */
1235 __le32 s_first_ino; /* First non-reserved inode */
1236 __le16 s_inode_size; /* size of inode structure */
1237 __le16 s_block_group_nr; /* block group # of this superblock */
1238 __le32 s_feature_compat; /* compatible feature set */
1239 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1240 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1241 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1242 /*78*/ char s_volume_name[16]; /* volume name */
1243 /*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */
1244 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1245 /*
1246 * Performance hints. Directory preallocation should only
1247 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1248 */
1249 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1250 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1251 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1252 /*
1253 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1254 */
1255 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1256 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1257 __le32 s_journal_dev; /* device number of journal file */
1258 __le32 s_last_orphan; /* start of list of inodes to delete */
1259 __le32 s_hash_seed[4]; /* HTREE hash seed */
1260 __u8 s_def_hash_version; /* Default hash version to use */
1261 __u8 s_jnl_backup_type;
1262 __le16 s_desc_size; /* size of group descriptor */
1263 /*100*/ __le32 s_default_mount_opts;
1264 __le32 s_first_meta_bg; /* First metablock block group */
1265 __le32 s_mkfs_time; /* When the filesystem was created */
1266 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1267 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1268 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1269 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1270 __le32 s_free_blocks_count_hi; /* Free blocks count */
1271 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1272 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1273 __le32 s_flags; /* Miscellaneous flags */
1274 __le16 s_raid_stride; /* RAID stride */
1275 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1276 __le64 s_mmp_block; /* Block for multi-mount protection */
1277 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1278 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1279 __u8 s_checksum_type; /* metadata checksum algorithm used */
1280 __u8 s_encryption_level; /* versioning level for encryption */
1281 __u8 s_reserved_pad; /* Padding to next 32bits */
1282 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1283 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1284 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1285 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1286 snapshot's future use */
1287 __le32 s_snapshot_list; /* inode number of the head of the
1288 on-disk snapshot list */
1289 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1290 __le32 s_error_count; /* number of fs errors */
1291 __le32 s_first_error_time; /* first time an error happened */
1292 __le32 s_first_error_ino; /* inode involved in first error */
1293 __le64 s_first_error_block; /* block involved of first error */
1294 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */
1295 __le32 s_first_error_line; /* line number where error happened */
1296 __le32 s_last_error_time; /* most recent time of an error */
1297 __le32 s_last_error_ino; /* inode involved in last error */
1298 __le32 s_last_error_line; /* line number where error happened */
1299 __le64 s_last_error_block; /* block involved of last error */
1300 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */
1301 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1302 __u8 s_mount_opts[64];
1303 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1304 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1305 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1306 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
1307 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */
1308 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */
1309 __le32 s_lpf_ino; /* Location of the lost+found inode */
1310 __le32 s_prj_quota_inum; /* inode for tracking project quota */
1311 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */
1312 __u8 s_wtime_hi;
1313 __u8 s_mtime_hi;
1314 __u8 s_mkfs_time_hi;
1315 __u8 s_lastcheck_hi;
1316 __u8 s_first_error_time_hi;
1317 __u8 s_last_error_time_hi;
1318 __u8 s_pad[2];
1319 __le32 s_reserved[96]; /* Padding to the end of the block */
1320 __le32 s_checksum; /* crc32c(superblock) */
1321 };
1322
1323 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1324
1325 #ifdef __KERNEL__
1326
1327 /*
1328 * run-time mount flags
1329 */
1330 #define EXT4_MF_MNTDIR_SAMPLED 0x0001
1331 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1332 #define EXT4_MF_TEST_DUMMY_ENCRYPTION 0x0004
1333
1334 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1335 #define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \
1336 EXT4_MF_TEST_DUMMY_ENCRYPTION))
1337 #else
1338 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1339 #endif
1340
1341 /* Number of quota types we support */
1342 #define EXT4_MAXQUOTAS 3
1343
1344 /*
1345 * fourth extended-fs super-block data in memory
1346 */
1347 struct ext4_sb_info {
1348 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1349 unsigned long s_inodes_per_block;/* Number of inodes per block */
1350 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1351 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1352 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1353 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1354 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1355 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1356 ext4_group_t s_groups_count; /* Number of groups in the fs */
1357 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1358 unsigned long s_overhead; /* # of fs overhead clusters */
1359 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1360 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1361 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1362 struct buffer_head * s_sbh; /* Buffer containing the super block */
1363 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1364 struct buffer_head **s_group_desc;
1365 unsigned int s_mount_opt;
1366 unsigned int s_mount_opt2;
1367 unsigned int s_mount_flags;
1368 unsigned int s_def_mount_opt;
1369 ext4_fsblk_t s_sb_block;
1370 atomic64_t s_resv_clusters;
1371 kuid_t s_resuid;
1372 kgid_t s_resgid;
1373 unsigned short s_mount_state;
1374 unsigned short s_pad;
1375 int s_addr_per_block_bits;
1376 int s_desc_per_block_bits;
1377 int s_inode_size;
1378 int s_first_ino;
1379 unsigned int s_inode_readahead_blks;
1380 unsigned int s_inode_goal;
1381 u32 s_hash_seed[4];
1382 int s_def_hash_version;
1383 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1384 struct percpu_counter s_freeclusters_counter;
1385 struct percpu_counter s_freeinodes_counter;
1386 struct percpu_counter s_dirs_counter;
1387 struct percpu_counter s_dirtyclusters_counter;
1388 struct blockgroup_lock *s_blockgroup_lock;
1389 struct proc_dir_entry *s_proc;
1390 struct kobject s_kobj;
1391 struct completion s_kobj_unregister;
1392 struct super_block *s_sb;
1393
1394 /* Journaling */
1395 struct journal_s *s_journal;
1396 struct list_head s_orphan;
1397 struct mutex s_orphan_lock;
1398 unsigned long s_ext4_flags; /* Ext4 superblock flags */
1399 unsigned long s_commit_interval;
1400 u32 s_max_batch_time;
1401 u32 s_min_batch_time;
1402 struct block_device *journal_bdev;
1403 #ifdef CONFIG_QUOTA
1404 char *s_qf_names[EXT4_MAXQUOTAS]; /* Names of quota files with journalled quota */
1405 int s_jquota_fmt; /* Format of quota to use */
1406 #endif
1407 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1408 struct rb_root system_blks;
1409
1410 #ifdef EXTENTS_STATS
1411 /* ext4 extents stats */
1412 unsigned long s_ext_min;
1413 unsigned long s_ext_max;
1414 unsigned long s_depth_max;
1415 spinlock_t s_ext_stats_lock;
1416 unsigned long s_ext_blocks;
1417 unsigned long s_ext_extents;
1418 #endif
1419
1420 /* for buddy allocator */
1421 struct ext4_group_info ***s_group_info;
1422 struct inode *s_buddy_cache;
1423 spinlock_t s_md_lock;
1424 unsigned short *s_mb_offsets;
1425 unsigned int *s_mb_maxs;
1426 unsigned int s_group_info_size;
1427 unsigned int s_mb_free_pending;
1428 struct list_head s_freed_data_list; /* List of blocks to be freed
1429 after commit completed */
1430
1431 /* tunables */
1432 unsigned long s_stripe;
1433 unsigned int s_mb_stream_request;
1434 unsigned int s_mb_max_to_scan;
1435 unsigned int s_mb_min_to_scan;
1436 unsigned int s_mb_stats;
1437 unsigned int s_mb_order2_reqs;
1438 unsigned int s_mb_group_prealloc;
1439 unsigned int s_max_dir_size_kb;
1440 /* where last allocation was done - for stream allocation */
1441 unsigned long s_mb_last_group;
1442 unsigned long s_mb_last_start;
1443
1444 /* stats for buddy allocator */
1445 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1446 atomic_t s_bal_success; /* we found long enough chunks */
1447 atomic_t s_bal_allocated; /* in blocks */
1448 atomic_t s_bal_ex_scanned; /* total extents scanned */
1449 atomic_t s_bal_goals; /* goal hits */
1450 atomic_t s_bal_breaks; /* too long searches */
1451 atomic_t s_bal_2orders; /* 2^order hits */
1452 spinlock_t s_bal_lock;
1453 unsigned long s_mb_buddies_generated;
1454 unsigned long long s_mb_generation_time;
1455 atomic_t s_mb_lost_chunks;
1456 atomic_t s_mb_preallocated;
1457 atomic_t s_mb_discarded;
1458 atomic_t s_lock_busy;
1459
1460 /* locality groups */
1461 struct ext4_locality_group __percpu *s_locality_groups;
1462
1463 /* for write statistics */
1464 unsigned long s_sectors_written_start;
1465 u64 s_kbytes_written;
1466
1467 /* the size of zero-out chunk */
1468 unsigned int s_extent_max_zeroout_kb;
1469
1470 unsigned int s_log_groups_per_flex;
1471 struct flex_groups *s_flex_groups;
1472 ext4_group_t s_flex_groups_allocated;
1473
1474 /* workqueue for reserved extent conversions (buffered io) */
1475 struct workqueue_struct *rsv_conversion_wq;
1476
1477 /* timer for periodic error stats printing */
1478 struct timer_list s_err_report;
1479
1480 /* Lazy inode table initialization info */
1481 struct ext4_li_request *s_li_request;
1482 /* Wait multiplier for lazy initialization thread */
1483 unsigned int s_li_wait_mult;
1484
1485 /* Kernel thread for multiple mount protection */
1486 struct task_struct *s_mmp_tsk;
1487
1488 /* record the last minlen when FITRIM is called. */
1489 atomic_t s_last_trim_minblks;
1490
1491 /* Reference to checksum algorithm driver via cryptoapi */
1492 struct crypto_shash *s_chksum_driver;
1493
1494 /* Precomputed FS UUID checksum for seeding other checksums */
1495 __u32 s_csum_seed;
1496
1497 /* Reclaim extents from extent status tree */
1498 struct shrinker s_es_shrinker;
1499 struct list_head s_es_list; /* List of inodes with reclaimable extents */
1500 long s_es_nr_inode;
1501 struct ext4_es_stats s_es_stats;
1502 struct mb_cache *s_ea_block_cache;
1503 struct mb_cache *s_ea_inode_cache;
1504 spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1505
1506 /* Ratelimit ext4 messages. */
1507 struct ratelimit_state s_err_ratelimit_state;
1508 struct ratelimit_state s_warning_ratelimit_state;
1509 struct ratelimit_state s_msg_ratelimit_state;
1510
1511 /* Barrier between changing inodes' journal flags and writepages ops. */
1512 struct percpu_rw_semaphore s_journal_flag_rwsem;
1513 struct dax_device *s_daxdev;
1514 };
1515
EXT4_SB(struct super_block * sb)1516 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1517 {
1518 return sb->s_fs_info;
1519 }
EXT4_I(struct inode * inode)1520 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1521 {
1522 return container_of(inode, struct ext4_inode_info, vfs_inode);
1523 }
1524
ext4_valid_inum(struct super_block * sb,unsigned long ino)1525 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1526 {
1527 return ino == EXT4_ROOT_INO ||
1528 (ino >= EXT4_FIRST_INO(sb) &&
1529 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1530 }
1531
1532 /*
1533 * Inode dynamic state flags
1534 */
1535 enum {
1536 EXT4_STATE_JDATA, /* journaled data exists */
1537 EXT4_STATE_NEW, /* inode is newly created */
1538 EXT4_STATE_XATTR, /* has in-inode xattrs */
1539 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1540 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1541 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1542 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
1543 EXT4_STATE_NEWENTRY, /* File just added to dir */
1544 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1545 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
1546 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */
1547 };
1548
1549 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1550 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1551 { \
1552 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1553 } \
1554 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1555 { \
1556 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1557 } \
1558 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1559 { \
1560 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1561 }
1562
1563 /* Add these declarations here only so that these functions can be
1564 * found by name. Otherwise, they are very hard to locate. */
1565 static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1566 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1567 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1568 EXT4_INODE_BIT_FNS(flag, flags, 0)
1569
1570 /* Add these declarations here only so that these functions can be
1571 * found by name. Otherwise, they are very hard to locate. */
1572 static inline int ext4_test_inode_state(struct inode *inode, int bit);
1573 static inline void ext4_set_inode_state(struct inode *inode, int bit);
1574 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1575 #if (BITS_PER_LONG < 64)
1576 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1577
ext4_clear_state_flags(struct ext4_inode_info * ei)1578 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1579 {
1580 (ei)->i_state_flags = 0;
1581 }
1582 #else
1583 EXT4_INODE_BIT_FNS(state, flags, 32)
1584
ext4_clear_state_flags(struct ext4_inode_info * ei)1585 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1586 {
1587 /* We depend on the fact that callers will set i_flags */
1588 }
1589 #endif
1590 #else
1591 /* Assume that user mode programs are passing in an ext4fs superblock, not
1592 * a kernel struct super_block. This will allow us to call the feature-test
1593 * macros from user land. */
1594 #define EXT4_SB(sb) (sb)
1595 #endif
1596
1597 /*
1598 * Returns true if the inode is inode is encrypted
1599 */
1600 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1601
1602 /*
1603 * Codes for operating systems
1604 */
1605 #define EXT4_OS_LINUX 0
1606 #define EXT4_OS_HURD 1
1607 #define EXT4_OS_MASIX 2
1608 #define EXT4_OS_FREEBSD 3
1609 #define EXT4_OS_LITES 4
1610
1611 /*
1612 * Revision levels
1613 */
1614 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1615 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1616
1617 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1618 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1619
1620 #define EXT4_GOOD_OLD_INODE_SIZE 128
1621
1622 /*
1623 * Feature set definitions
1624 */
1625
1626 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1627 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1628 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1629 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1630 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1631 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1632 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
1633
1634 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1635 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1636 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1637 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1638 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1639 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1640 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1641 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1642 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1643 /*
1644 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1645 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1646 * all other data structures' checksums. However, the METADATA_CSUM and
1647 * GDT_CSUM bits are mutually exclusive.
1648 */
1649 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
1650 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000
1651 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000
1652
1653 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1654 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1655 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1656 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1657 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1658 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1659 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1660 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1661 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1662 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1663 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1664 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000
1665 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
1666 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
1667 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000
1668
1669 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
1670 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1671 { \
1672 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1673 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
1674 } \
1675 static inline void ext4_set_feature_##name(struct super_block *sb) \
1676 { \
1677 EXT4_SB(sb)->s_es->s_feature_compat |= \
1678 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1679 } \
1680 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1681 { \
1682 EXT4_SB(sb)->s_es->s_feature_compat &= \
1683 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1684 }
1685
1686 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
1687 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1688 { \
1689 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1690 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
1691 } \
1692 static inline void ext4_set_feature_##name(struct super_block *sb) \
1693 { \
1694 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
1695 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1696 } \
1697 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1698 { \
1699 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
1700 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1701 }
1702
1703 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
1704 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1705 { \
1706 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1707 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
1708 } \
1709 static inline void ext4_set_feature_##name(struct super_block *sb) \
1710 { \
1711 EXT4_SB(sb)->s_es->s_feature_incompat |= \
1712 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1713 } \
1714 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1715 { \
1716 EXT4_SB(sb)->s_es->s_feature_incompat &= \
1717 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1718 }
1719
EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc,DIR_PREALLOC)1720 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC)
1721 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES)
1722 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL)
1723 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR)
1724 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE)
1725 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX)
1726 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2)
1727
1728 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER)
1729 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE)
1730 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR)
1731 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE)
1732 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM)
1733 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK)
1734 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE)
1735 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA)
1736 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC)
1737 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM)
1738 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY)
1739 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT)
1740
1741 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION)
1742 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE)
1743 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER)
1744 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV)
1745 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG)
1746 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS)
1747 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT)
1748 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP)
1749 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG)
1750 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE)
1751 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA)
1752 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED)
1753 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR)
1754 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA)
1755 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT)
1756
1757 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1758 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1759 EXT4_FEATURE_INCOMPAT_META_BG)
1760 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1761 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1762 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1763
1764 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1765 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1766 EXT4_FEATURE_INCOMPAT_RECOVER| \
1767 EXT4_FEATURE_INCOMPAT_META_BG)
1768 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1769 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1770 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1771
1772 #define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1773 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1774 EXT4_FEATURE_INCOMPAT_RECOVER| \
1775 EXT4_FEATURE_INCOMPAT_META_BG| \
1776 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1777 EXT4_FEATURE_INCOMPAT_64BIT| \
1778 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1779 EXT4_FEATURE_INCOMPAT_EA_INODE| \
1780 EXT4_FEATURE_INCOMPAT_MMP | \
1781 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
1782 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
1783 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \
1784 EXT4_FEATURE_INCOMPAT_LARGEDIR)
1785 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1786 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1787 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1788 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1789 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1790 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1791 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1792 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1793 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1794 EXT4_FEATURE_RO_COMPAT_QUOTA |\
1795 EXT4_FEATURE_RO_COMPAT_PROJECT)
1796
1797 #define EXTN_FEATURE_FUNCS(ver) \
1798 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
1799 { \
1800 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1801 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
1802 } \
1803 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
1804 { \
1805 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1806 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
1807 } \
1808 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
1809 { \
1810 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1811 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
1812 }
1813
1814 EXTN_FEATURE_FUNCS(2)
1815 EXTN_FEATURE_FUNCS(3)
1816 EXTN_FEATURE_FUNCS(4)
1817
1818 static inline bool ext4_has_compat_features(struct super_block *sb)
1819 {
1820 return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
1821 }
ext4_has_ro_compat_features(struct super_block * sb)1822 static inline bool ext4_has_ro_compat_features(struct super_block *sb)
1823 {
1824 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
1825 }
ext4_has_incompat_features(struct super_block * sb)1826 static inline bool ext4_has_incompat_features(struct super_block *sb)
1827 {
1828 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
1829 }
1830
1831 /*
1832 * Superblock flags
1833 */
1834 #define EXT4_FLAGS_RESIZING 0
1835 #define EXT4_FLAGS_SHUTDOWN 1
1836
ext4_forced_shutdown(struct ext4_sb_info * sbi)1837 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi)
1838 {
1839 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags);
1840 }
1841
1842
1843 /*
1844 * Default values for user and/or group using reserved blocks
1845 */
1846 #define EXT4_DEF_RESUID 0
1847 #define EXT4_DEF_RESGID 0
1848
1849 /*
1850 * Default project ID
1851 */
1852 #define EXT4_DEF_PROJID 0
1853
1854 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
1855
1856 /*
1857 * Default mount options
1858 */
1859 #define EXT4_DEFM_DEBUG 0x0001
1860 #define EXT4_DEFM_BSDGROUPS 0x0002
1861 #define EXT4_DEFM_XATTR_USER 0x0004
1862 #define EXT4_DEFM_ACL 0x0008
1863 #define EXT4_DEFM_UID16 0x0010
1864 #define EXT4_DEFM_JMODE 0x0060
1865 #define EXT4_DEFM_JMODE_DATA 0x0020
1866 #define EXT4_DEFM_JMODE_ORDERED 0x0040
1867 #define EXT4_DEFM_JMODE_WBACK 0x0060
1868 #define EXT4_DEFM_NOBARRIER 0x0100
1869 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1870 #define EXT4_DEFM_DISCARD 0x0400
1871 #define EXT4_DEFM_NODELALLOC 0x0800
1872
1873 /*
1874 * Default journal batch times
1875 */
1876 #define EXT4_DEF_MIN_BATCH_TIME 0
1877 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1878
1879 /*
1880 * Minimum number of groups in a flexgroup before we separate out
1881 * directories into the first block group of a flexgroup
1882 */
1883 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1884
1885 /*
1886 * Structure of a directory entry
1887 */
1888 #define EXT4_NAME_LEN 255
1889
1890 struct ext4_dir_entry {
1891 __le32 inode; /* Inode number */
1892 __le16 rec_len; /* Directory entry length */
1893 __le16 name_len; /* Name length */
1894 char name[EXT4_NAME_LEN]; /* File name */
1895 };
1896
1897 /*
1898 * The new version of the directory entry. Since EXT4 structures are
1899 * stored in intel byte order, and the name_len field could never be
1900 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1901 * file_type field.
1902 */
1903 struct ext4_dir_entry_2 {
1904 __le32 inode; /* Inode number */
1905 __le16 rec_len; /* Directory entry length */
1906 __u8 name_len; /* Name length */
1907 __u8 file_type;
1908 char name[EXT4_NAME_LEN]; /* File name */
1909 };
1910
1911 /*
1912 * This is a bogus directory entry at the end of each leaf block that
1913 * records checksums.
1914 */
1915 struct ext4_dir_entry_tail {
1916 __le32 det_reserved_zero1; /* Pretend to be unused */
1917 __le16 det_rec_len; /* 12 */
1918 __u8 det_reserved_zero2; /* Zero name length */
1919 __u8 det_reserved_ft; /* 0xDE, fake file type */
1920 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
1921 };
1922
1923 #define EXT4_DIRENT_TAIL(block, blocksize) \
1924 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1925 ((blocksize) - \
1926 sizeof(struct ext4_dir_entry_tail))))
1927
1928 /*
1929 * Ext4 directory file types. Only the low 3 bits are used. The
1930 * other bits are reserved for now.
1931 */
1932 #define EXT4_FT_UNKNOWN 0
1933 #define EXT4_FT_REG_FILE 1
1934 #define EXT4_FT_DIR 2
1935 #define EXT4_FT_CHRDEV 3
1936 #define EXT4_FT_BLKDEV 4
1937 #define EXT4_FT_FIFO 5
1938 #define EXT4_FT_SOCK 6
1939 #define EXT4_FT_SYMLINK 7
1940
1941 #define EXT4_FT_MAX 8
1942
1943 #define EXT4_FT_DIR_CSUM 0xDE
1944
1945 /*
1946 * EXT4_DIR_PAD defines the directory entries boundaries
1947 *
1948 * NOTE: It must be a multiple of 4
1949 */
1950 #define EXT4_DIR_PAD 4
1951 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1952 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1953 ~EXT4_DIR_ROUND)
1954 #define EXT4_MAX_REC_LEN ((1<<16)-1)
1955
1956 /*
1957 * If we ever get support for fs block sizes > page_size, we'll need
1958 * to remove the #if statements in the next two functions...
1959 */
1960 static inline unsigned int
ext4_rec_len_from_disk(__le16 dlen,unsigned blocksize)1961 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1962 {
1963 unsigned len = le16_to_cpu(dlen);
1964
1965 #if (PAGE_SIZE >= 65536)
1966 if (len == EXT4_MAX_REC_LEN || len == 0)
1967 return blocksize;
1968 return (len & 65532) | ((len & 3) << 16);
1969 #else
1970 return len;
1971 #endif
1972 }
1973
ext4_rec_len_to_disk(unsigned len,unsigned blocksize)1974 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1975 {
1976 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1977 BUG();
1978 #if (PAGE_SIZE >= 65536)
1979 if (len < 65536)
1980 return cpu_to_le16(len);
1981 if (len == blocksize) {
1982 if (blocksize == 65536)
1983 return cpu_to_le16(EXT4_MAX_REC_LEN);
1984 else
1985 return cpu_to_le16(0);
1986 }
1987 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1988 #else
1989 return cpu_to_le16(len);
1990 #endif
1991 }
1992
1993 /*
1994 * Hash Tree Directory indexing
1995 * (c) Daniel Phillips, 2001
1996 */
1997
1998 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
1999 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2000 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \
2001 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir)))
2002 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2003
2004 /* Legal values for the dx_root hash_version field: */
2005
2006 #define DX_HASH_LEGACY 0
2007 #define DX_HASH_HALF_MD4 1
2008 #define DX_HASH_TEA 2
2009 #define DX_HASH_LEGACY_UNSIGNED 3
2010 #define DX_HASH_HALF_MD4_UNSIGNED 4
2011 #define DX_HASH_TEA_UNSIGNED 5
2012
ext4_chksum(struct ext4_sb_info * sbi,u32 crc,const void * address,unsigned int length)2013 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2014 const void *address, unsigned int length)
2015 {
2016 struct {
2017 struct shash_desc shash;
2018 char ctx[4];
2019 } desc;
2020
2021 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2022
2023 desc.shash.tfm = sbi->s_chksum_driver;
2024 desc.shash.flags = 0;
2025 *(u32 *)desc.ctx = crc;
2026
2027 BUG_ON(crypto_shash_update(&desc.shash, address, length));
2028
2029 return *(u32 *)desc.ctx;
2030 }
2031
2032 #ifdef __KERNEL__
2033
2034 /* hash info structure used by the directory hash */
2035 struct dx_hash_info
2036 {
2037 u32 hash;
2038 u32 minor_hash;
2039 int hash_version;
2040 u32 *seed;
2041 };
2042
2043
2044 /* 32 and 64 bit signed EOF for dx directories */
2045 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
2046 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
2047
2048
2049 /*
2050 * Control parameters used by ext4_htree_next_block
2051 */
2052 #define HASH_NB_ALWAYS 1
2053
2054 struct ext4_filename {
2055 const struct qstr *usr_fname;
2056 struct fscrypt_str disk_name;
2057 struct dx_hash_info hinfo;
2058 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2059 struct fscrypt_str crypto_buf;
2060 #endif
2061 };
2062
2063 #define fname_name(p) ((p)->disk_name.name)
2064 #define fname_len(p) ((p)->disk_name.len)
2065
2066 /*
2067 * Describe an inode's exact location on disk and in memory
2068 */
2069 struct ext4_iloc
2070 {
2071 struct buffer_head *bh;
2072 unsigned long offset;
2073 ext4_group_t block_group;
2074 };
2075
ext4_raw_inode(struct ext4_iloc * iloc)2076 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2077 {
2078 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2079 }
2080
ext4_is_quota_file(struct inode * inode)2081 static inline bool ext4_is_quota_file(struct inode *inode)
2082 {
2083 return IS_NOQUOTA(inode) &&
2084 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL);
2085 }
2086
2087 /*
2088 * This structure is stuffed into the struct file's private_data field
2089 * for directories. It is where we put information so that we can do
2090 * readdir operations in hash tree order.
2091 */
2092 struct dir_private_info {
2093 struct rb_root root;
2094 struct rb_node *curr_node;
2095 struct fname *extra_fname;
2096 loff_t last_pos;
2097 __u32 curr_hash;
2098 __u32 curr_minor_hash;
2099 __u32 next_hash;
2100 };
2101
2102 /* calculate the first block number of the group */
2103 static inline ext4_fsblk_t
ext4_group_first_block_no(struct super_block * sb,ext4_group_t group_no)2104 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2105 {
2106 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2107 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2108 }
2109
2110 /*
2111 * Special error return code only used by dx_probe() and its callers.
2112 */
2113 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
2114
2115 /* htree levels for ext4 */
2116 #define EXT4_HTREE_LEVEL_COMPAT 2
2117 #define EXT4_HTREE_LEVEL 3
2118
ext4_dir_htree_level(struct super_block * sb)2119 static inline int ext4_dir_htree_level(struct super_block *sb)
2120 {
2121 return ext4_has_feature_largedir(sb) ?
2122 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT;
2123 }
2124
2125 /*
2126 * Timeout and state flag for lazy initialization inode thread.
2127 */
2128 #define EXT4_DEF_LI_WAIT_MULT 10
2129 #define EXT4_DEF_LI_MAX_START_DELAY 5
2130 #define EXT4_LAZYINIT_QUIT 0x0001
2131 #define EXT4_LAZYINIT_RUNNING 0x0002
2132
2133 /*
2134 * Lazy inode table initialization info
2135 */
2136 struct ext4_lazy_init {
2137 unsigned long li_state;
2138 struct list_head li_request_list;
2139 struct mutex li_list_mtx;
2140 };
2141
2142 struct ext4_li_request {
2143 struct super_block *lr_super;
2144 struct ext4_sb_info *lr_sbi;
2145 ext4_group_t lr_next_group;
2146 struct list_head lr_request;
2147 unsigned long lr_next_sched;
2148 unsigned long lr_timeout;
2149 };
2150
2151 struct ext4_features {
2152 struct kobject f_kobj;
2153 struct completion f_kobj_unregister;
2154 };
2155
2156 /*
2157 * This structure will be used for multiple mount protection. It will be
2158 * written into the block number saved in the s_mmp_block field in the
2159 * superblock. Programs that check MMP should assume that if
2160 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2161 * to use the filesystem, regardless of how old the timestamp is.
2162 */
2163 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
2164 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2165 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
2166 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
2167
2168 struct mmp_struct {
2169 __le32 mmp_magic; /* Magic number for MMP */
2170 __le32 mmp_seq; /* Sequence no. updated periodically */
2171
2172 /*
2173 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2174 * purposes and do not affect the correctness of the algorithm
2175 */
2176 __le64 mmp_time; /* Time last updated */
2177 char mmp_nodename[64]; /* Node which last updated MMP block */
2178 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
2179
2180 /*
2181 * mmp_check_interval is used to verify if the MMP block has been
2182 * updated on the block device. The value is updated based on the
2183 * maximum time to write the MMP block during an update cycle.
2184 */
2185 __le16 mmp_check_interval;
2186
2187 __le16 mmp_pad1;
2188 __le32 mmp_pad2[226];
2189 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
2190 };
2191
2192 /* arguments passed to the mmp thread */
2193 struct mmpd_data {
2194 struct buffer_head *bh; /* bh from initial read_mmp_block() */
2195 struct super_block *sb; /* super block of the fs */
2196 };
2197
2198 /*
2199 * Check interval multiplier
2200 * The MMP block is written every update interval and initially checked every
2201 * update interval x the multiplier (the value is then adapted based on the
2202 * write latency). The reason is that writes can be delayed under load and we
2203 * don't want readers to incorrectly assume that the filesystem is no longer
2204 * in use.
2205 */
2206 #define EXT4_MMP_CHECK_MULT 2UL
2207
2208 /*
2209 * Minimum interval for MMP checking in seconds.
2210 */
2211 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
2212
2213 /*
2214 * Maximum interval for MMP checking in seconds.
2215 */
2216 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
2217
2218 /*
2219 * Function prototypes
2220 */
2221
2222 /*
2223 * Ok, these declarations are also in <linux/kernel.h> but none of the
2224 * ext4 source programs needs to include it so they are duplicated here.
2225 */
2226 # define NORET_TYPE /**/
2227 # define ATTRIB_NORET __attribute__((noreturn))
2228 # define NORET_AND noreturn,
2229
2230 /* bitmap.c */
2231 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2232 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2233 struct ext4_group_desc *gdp,
2234 struct buffer_head *bh, int sz);
2235 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2236 struct ext4_group_desc *gdp,
2237 struct buffer_head *bh, int sz);
2238 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2239 struct ext4_group_desc *gdp,
2240 struct buffer_head *bh);
2241 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2242 struct ext4_group_desc *gdp,
2243 struct buffer_head *bh);
2244
2245 /* balloc.c */
2246 extern void ext4_get_group_no_and_offset(struct super_block *sb,
2247 ext4_fsblk_t blocknr,
2248 ext4_group_t *blockgrpp,
2249 ext4_grpblk_t *offsetp);
2250 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2251 ext4_fsblk_t block);
2252
2253 extern unsigned int ext4_block_group(struct super_block *sb,
2254 ext4_fsblk_t blocknr);
2255 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2256 ext4_fsblk_t blocknr);
2257 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2258 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2259 ext4_group_t group);
2260 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2261 ext4_fsblk_t goal,
2262 unsigned int flags,
2263 unsigned long *count,
2264 int *errp);
2265 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2266 s64 nclusters, unsigned int flags);
2267 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2268 extern void ext4_check_blocks_bitmap(struct super_block *);
2269 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2270 ext4_group_t block_group,
2271 struct buffer_head ** bh);
2272 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2273
2274 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2275 ext4_group_t block_group);
2276 extern int ext4_wait_block_bitmap(struct super_block *sb,
2277 ext4_group_t block_group,
2278 struct buffer_head *bh);
2279 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2280 ext4_group_t block_group);
2281 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2282 ext4_group_t block_group,
2283 struct ext4_group_desc *gdp);
2284 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2285
ext4_encrypted_inode(struct inode * inode)2286 static inline bool ext4_encrypted_inode(struct inode *inode)
2287 {
2288 return ext4_test_inode_flag(inode, EXT4_INODE_ENCRYPT);
2289 }
2290
2291 #ifdef CONFIG_EXT4_FS_ENCRYPTION
ext4_fname_setup_filename(struct inode * dir,const struct qstr * iname,int lookup,struct ext4_filename * fname)2292 static inline int ext4_fname_setup_filename(struct inode *dir,
2293 const struct qstr *iname,
2294 int lookup, struct ext4_filename *fname)
2295 {
2296 struct fscrypt_name name;
2297 int err;
2298
2299 memset(fname, 0, sizeof(struct ext4_filename));
2300
2301 err = fscrypt_setup_filename(dir, iname, lookup, &name);
2302
2303 fname->usr_fname = name.usr_fname;
2304 fname->disk_name = name.disk_name;
2305 fname->hinfo.hash = name.hash;
2306 fname->hinfo.minor_hash = name.minor_hash;
2307 fname->crypto_buf = name.crypto_buf;
2308 return err;
2309 }
2310
ext4_fname_free_filename(struct ext4_filename * fname)2311 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2312 {
2313 struct fscrypt_name name;
2314
2315 name.crypto_buf = fname->crypto_buf;
2316 fscrypt_free_filename(&name);
2317
2318 fname->crypto_buf.name = NULL;
2319 fname->usr_fname = NULL;
2320 fname->disk_name.name = NULL;
2321 }
2322 #else
ext4_fname_setup_filename(struct inode * dir,const struct qstr * iname,int lookup,struct ext4_filename * fname)2323 static inline int ext4_fname_setup_filename(struct inode *dir,
2324 const struct qstr *iname,
2325 int lookup, struct ext4_filename *fname)
2326 {
2327 fname->usr_fname = iname;
2328 fname->disk_name.name = (unsigned char *) iname->name;
2329 fname->disk_name.len = iname->len;
2330 return 0;
2331 }
ext4_fname_free_filename(struct ext4_filename * fname)2332 static inline void ext4_fname_free_filename(struct ext4_filename *fname) { }
2333
2334 #endif
2335
2336 /* dir.c */
2337 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2338 struct file *,
2339 struct ext4_dir_entry_2 *,
2340 struct buffer_head *, char *, int,
2341 unsigned int);
2342 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2343 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2344 (de), (bh), (buf), (size), (offset)))
2345 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2346 __u32 minor_hash,
2347 struct ext4_dir_entry_2 *dirent,
2348 struct fscrypt_str *ent_name);
2349 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2350 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2351 struct buffer_head *bh,
2352 void *buf, int buf_size,
2353 struct ext4_filename *fname,
2354 struct ext4_dir_entry_2 **dest_de);
2355 void ext4_insert_dentry(struct inode *inode,
2356 struct ext4_dir_entry_2 *de,
2357 int buf_size,
2358 struct ext4_filename *fname);
ext4_update_dx_flag(struct inode * inode)2359 static inline void ext4_update_dx_flag(struct inode *inode)
2360 {
2361 if (!ext4_has_feature_dir_index(inode->i_sb))
2362 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2363 }
2364 static const unsigned char ext4_filetype_table[] = {
2365 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2366 };
2367
get_dtype(struct super_block * sb,int filetype)2368 static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2369 {
2370 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2371 return DT_UNKNOWN;
2372
2373 return ext4_filetype_table[filetype];
2374 }
2375 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2376 void *buf, int buf_size);
2377
2378 /* fsync.c */
2379 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2380
2381 /* hash.c */
2382 extern int ext4fs_dirhash(const char *name, int len, struct
2383 dx_hash_info *hinfo);
2384
2385 /* ialloc.c */
2386 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2387 const struct qstr *qstr, __u32 goal,
2388 uid_t *owner, __u32 i_flags,
2389 int handle_type, unsigned int line_no,
2390 int nblocks);
2391
2392 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \
2393 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2394 i_flags, 0, 0, 0)
2395 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2396 type, nblocks) \
2397 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2398 0, (type), __LINE__, (nblocks))
2399
2400
2401 extern void ext4_free_inode(handle_t *, struct inode *);
2402 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2403 extern unsigned long ext4_count_free_inodes(struct super_block *);
2404 extern unsigned long ext4_count_dirs(struct super_block *);
2405 extern void ext4_check_inodes_bitmap(struct super_block *);
2406 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2407 extern int ext4_init_inode_table(struct super_block *sb,
2408 ext4_group_t group, int barrier);
2409 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2410
2411 /* mballoc.c */
2412 extern const struct seq_operations ext4_mb_seq_groups_ops;
2413 extern long ext4_mb_stats;
2414 extern long ext4_mb_max_to_scan;
2415 extern int ext4_mb_init(struct super_block *);
2416 extern int ext4_mb_release(struct super_block *);
2417 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2418 struct ext4_allocation_request *, int *);
2419 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2420 extern void ext4_discard_preallocations(struct inode *);
2421 extern int __init ext4_init_mballoc(void);
2422 extern void ext4_exit_mballoc(void);
2423 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2424 struct buffer_head *bh, ext4_fsblk_t block,
2425 unsigned long count, int flags);
2426 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2427 ext4_group_t ngroups);
2428 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2429 ext4_group_t i, struct ext4_group_desc *desc);
2430 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2431 ext4_fsblk_t block, unsigned long count);
2432 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2433 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid);
2434
2435 /* inode.c */
2436 int ext4_inode_is_fast_symlink(struct inode *inode);
2437 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2438 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2439 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count,
2440 bool wait, struct buffer_head **bhs);
2441 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2442 struct buffer_head *bh_result, int create);
2443 int ext4_get_block(struct inode *inode, sector_t iblock,
2444 struct buffer_head *bh_result, int create);
2445 int ext4_dio_get_block(struct inode *inode, sector_t iblock,
2446 struct buffer_head *bh_result, int create);
2447 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2448 struct buffer_head *bh, int create);
2449 int ext4_walk_page_buffers(handle_t *handle,
2450 struct buffer_head *head,
2451 unsigned from,
2452 unsigned to,
2453 int *partial,
2454 int (*fn)(handle_t *handle,
2455 struct buffer_head *bh));
2456 int do_journal_get_write_access(handle_t *handle,
2457 struct buffer_head *bh);
2458 #define FALL_BACK_TO_NONDELALLOC 1
2459 #define CONVERT_INLINE_DATA 2
2460
2461 extern struct inode *ext4_iget(struct super_block *, unsigned long);
2462 extern struct inode *ext4_iget_normal(struct super_block *, unsigned long);
2463 extern int ext4_write_inode(struct inode *, struct writeback_control *);
2464 extern int ext4_setattr(struct dentry *, struct iattr *);
2465 extern int ext4_getattr(const struct path *, struct kstat *, u32, unsigned int);
2466 extern void ext4_evict_inode(struct inode *);
2467 extern void ext4_clear_inode(struct inode *);
2468 extern int ext4_file_getattr(const struct path *, struct kstat *, u32, unsigned int);
2469 extern int ext4_sync_inode(handle_t *, struct inode *);
2470 extern void ext4_dirty_inode(struct inode *, int);
2471 extern int ext4_change_inode_journal_flag(struct inode *, int);
2472 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2473 extern int ext4_inode_attach_jinode(struct inode *inode);
2474 extern int ext4_can_truncate(struct inode *inode);
2475 extern int ext4_truncate(struct inode *);
2476 extern int ext4_break_layouts(struct inode *);
2477 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2478 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
2479 extern void ext4_set_inode_flags(struct inode *);
2480 extern int ext4_alloc_da_blocks(struct inode *inode);
2481 extern void ext4_set_aops(struct inode *inode);
2482 extern int ext4_writepage_trans_blocks(struct inode *);
2483 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2484 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2485 loff_t lstart, loff_t lend);
2486 extern int ext4_page_mkwrite(struct vm_fault *vmf);
2487 extern int ext4_filemap_fault(struct vm_fault *vmf);
2488 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2489 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
2490 extern void ext4_da_update_reserve_space(struct inode *inode,
2491 int used, int quota_claim);
2492 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
2493 ext4_fsblk_t pblk, ext4_lblk_t len);
2494
2495 /* indirect.c */
2496 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2497 struct ext4_map_blocks *map, int flags);
2498 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2499 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2500 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2501 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2502 ext4_lblk_t start, ext4_lblk_t end);
2503
2504 /* ioctl.c */
2505 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2506 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2507
2508 /* migrate.c */
2509 extern int ext4_ext_migrate(struct inode *);
2510 extern int ext4_ind_migrate(struct inode *inode);
2511
2512 /* namei.c */
2513 extern int ext4_dirent_csum_verify(struct inode *inode,
2514 struct ext4_dir_entry *dirent);
2515 extern int ext4_orphan_add(handle_t *, struct inode *);
2516 extern int ext4_orphan_del(handle_t *, struct inode *);
2517 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2518 __u32 start_minor_hash, __u32 *next_hash);
2519 extern int ext4_search_dir(struct buffer_head *bh,
2520 char *search_buf,
2521 int buf_size,
2522 struct inode *dir,
2523 struct ext4_filename *fname,
2524 unsigned int offset,
2525 struct ext4_dir_entry_2 **res_dir);
2526 extern int ext4_generic_delete_entry(handle_t *handle,
2527 struct inode *dir,
2528 struct ext4_dir_entry_2 *de_del,
2529 struct buffer_head *bh,
2530 void *entry_buf,
2531 int buf_size,
2532 int csum_size);
2533 extern bool ext4_empty_dir(struct inode *inode);
2534
2535 /* resize.c */
2536 extern int ext4_group_add(struct super_block *sb,
2537 struct ext4_new_group_data *input);
2538 extern int ext4_group_extend(struct super_block *sb,
2539 struct ext4_super_block *es,
2540 ext4_fsblk_t n_blocks_count);
2541 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2542
2543 /* super.c */
2544 extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
2545 extern int ext4_calculate_overhead(struct super_block *sb);
2546 extern void ext4_superblock_csum_set(struct super_block *sb);
2547 extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2548 extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2549 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2550 ext4_group_t ngroup);
2551 extern const char *ext4_decode_error(struct super_block *sb, int errno,
2552 char nbuf[16]);
2553 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb,
2554 ext4_group_t block_group,
2555 unsigned int flags);
2556
2557 extern __printf(4, 5)
2558 void __ext4_error(struct super_block *, const char *, unsigned int,
2559 const char *, ...);
2560 extern __printf(5, 6)
2561 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2562 const char *, ...);
2563 extern __printf(5, 6)
2564 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2565 const char *, ...);
2566 extern void __ext4_std_error(struct super_block *, const char *,
2567 unsigned int, int);
2568 extern __printf(4, 5)
2569 void __ext4_abort(struct super_block *, const char *, unsigned int,
2570 const char *, ...);
2571 extern __printf(4, 5)
2572 void __ext4_warning(struct super_block *, const char *, unsigned int,
2573 const char *, ...);
2574 extern __printf(4, 5)
2575 void __ext4_warning_inode(const struct inode *inode, const char *function,
2576 unsigned int line, const char *fmt, ...);
2577 extern __printf(3, 4)
2578 void __ext4_msg(struct super_block *, const char *, const char *, ...);
2579 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2580 const char *, unsigned int, const char *);
2581 extern __printf(7, 8)
2582 void __ext4_grp_locked_error(const char *, unsigned int,
2583 struct super_block *, ext4_group_t,
2584 unsigned long, ext4_fsblk_t,
2585 const char *, ...);
2586
2587 #define EXT4_ERROR_INODE(inode, fmt, a...) \
2588 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
2589
2590 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
2591 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
2592
2593 #define EXT4_ERROR_FILE(file, block, fmt, a...) \
2594 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
2595
2596 #ifdef CONFIG_PRINTK
2597
2598 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2599 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__)
2600 #define ext4_error_file(file, func, line, block, fmt, ...) \
2601 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2602 #define ext4_error(sb, fmt, ...) \
2603 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2604 #define ext4_abort(sb, fmt, ...) \
2605 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2606 #define ext4_warning(sb, fmt, ...) \
2607 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2608 #define ext4_warning_inode(inode, fmt, ...) \
2609 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
2610 #define ext4_msg(sb, level, fmt, ...) \
2611 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2612 #define dump_mmp_msg(sb, mmp, msg) \
2613 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2614 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2615 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2616 fmt, ##__VA_ARGS__)
2617
2618 #else
2619
2620 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2621 do { \
2622 no_printk(fmt, ##__VA_ARGS__); \
2623 __ext4_error_inode(inode, "", 0, block, " "); \
2624 } while (0)
2625 #define ext4_error_file(file, func, line, block, fmt, ...) \
2626 do { \
2627 no_printk(fmt, ##__VA_ARGS__); \
2628 __ext4_error_file(file, "", 0, block, " "); \
2629 } while (0)
2630 #define ext4_error(sb, fmt, ...) \
2631 do { \
2632 no_printk(fmt, ##__VA_ARGS__); \
2633 __ext4_error(sb, "", 0, " "); \
2634 } while (0)
2635 #define ext4_abort(sb, fmt, ...) \
2636 do { \
2637 no_printk(fmt, ##__VA_ARGS__); \
2638 __ext4_abort(sb, "", 0, " "); \
2639 } while (0)
2640 #define ext4_warning(sb, fmt, ...) \
2641 do { \
2642 no_printk(fmt, ##__VA_ARGS__); \
2643 __ext4_warning(sb, "", 0, " "); \
2644 } while (0)
2645 #define ext4_warning_inode(inode, fmt, ...) \
2646 do { \
2647 no_printk(fmt, ##__VA_ARGS__); \
2648 __ext4_warning_inode(inode, "", 0, " "); \
2649 } while (0)
2650 #define ext4_msg(sb, level, fmt, ...) \
2651 do { \
2652 no_printk(fmt, ##__VA_ARGS__); \
2653 __ext4_msg(sb, "", " "); \
2654 } while (0)
2655 #define dump_mmp_msg(sb, mmp, msg) \
2656 __dump_mmp_msg(sb, mmp, "", 0, "")
2657 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2658 do { \
2659 no_printk(fmt, ##__VA_ARGS__); \
2660 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
2661 } while (0)
2662
2663 #endif
2664
2665 extern void ext4_update_dynamic_rev(struct super_block *sb);
2666 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2667 __u32 compat);
2668 extern int ext4_update_rocompat_feature(handle_t *handle,
2669 struct super_block *sb, __u32 rocompat);
2670 extern int ext4_update_incompat_feature(handle_t *handle,
2671 struct super_block *sb, __u32 incompat);
2672 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2673 struct ext4_group_desc *bg);
2674 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2675 struct ext4_group_desc *bg);
2676 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2677 struct ext4_group_desc *bg);
2678 extern __u32 ext4_free_group_clusters(struct super_block *sb,
2679 struct ext4_group_desc *bg);
2680 extern __u32 ext4_free_inodes_count(struct super_block *sb,
2681 struct ext4_group_desc *bg);
2682 extern __u32 ext4_used_dirs_count(struct super_block *sb,
2683 struct ext4_group_desc *bg);
2684 extern __u32 ext4_itable_unused_count(struct super_block *sb,
2685 struct ext4_group_desc *bg);
2686 extern void ext4_block_bitmap_set(struct super_block *sb,
2687 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2688 extern void ext4_inode_bitmap_set(struct super_block *sb,
2689 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2690 extern void ext4_inode_table_set(struct super_block *sb,
2691 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2692 extern void ext4_free_group_clusters_set(struct super_block *sb,
2693 struct ext4_group_desc *bg,
2694 __u32 count);
2695 extern void ext4_free_inodes_set(struct super_block *sb,
2696 struct ext4_group_desc *bg, __u32 count);
2697 extern void ext4_used_dirs_set(struct super_block *sb,
2698 struct ext4_group_desc *bg, __u32 count);
2699 extern void ext4_itable_unused_set(struct super_block *sb,
2700 struct ext4_group_desc *bg, __u32 count);
2701 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2702 struct ext4_group_desc *gdp);
2703 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2704 struct ext4_group_desc *gdp);
2705 extern int ext4_register_li_request(struct super_block *sb,
2706 ext4_group_t first_not_zeroed);
2707
ext4_has_metadata_csum(struct super_block * sb)2708 static inline int ext4_has_metadata_csum(struct super_block *sb)
2709 {
2710 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
2711 !EXT4_SB(sb)->s_chksum_driver);
2712
2713 return ext4_has_feature_metadata_csum(sb) &&
2714 (EXT4_SB(sb)->s_chksum_driver != NULL);
2715 }
2716
ext4_has_group_desc_csum(struct super_block * sb)2717 static inline int ext4_has_group_desc_csum(struct super_block *sb)
2718 {
2719 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb);
2720 }
2721
ext4_blocks_count(struct ext4_super_block * es)2722 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2723 {
2724 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2725 le32_to_cpu(es->s_blocks_count_lo);
2726 }
2727
ext4_r_blocks_count(struct ext4_super_block * es)2728 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2729 {
2730 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2731 le32_to_cpu(es->s_r_blocks_count_lo);
2732 }
2733
ext4_free_blocks_count(struct ext4_super_block * es)2734 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2735 {
2736 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2737 le32_to_cpu(es->s_free_blocks_count_lo);
2738 }
2739
ext4_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)2740 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2741 ext4_fsblk_t blk)
2742 {
2743 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2744 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2745 }
2746
ext4_free_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)2747 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2748 ext4_fsblk_t blk)
2749 {
2750 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2751 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2752 }
2753
ext4_r_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)2754 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2755 ext4_fsblk_t blk)
2756 {
2757 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2758 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2759 }
2760
ext4_isize(struct super_block * sb,struct ext4_inode * raw_inode)2761 static inline loff_t ext4_isize(struct super_block *sb,
2762 struct ext4_inode *raw_inode)
2763 {
2764 if (ext4_has_feature_largedir(sb) ||
2765 S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2766 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2767 le32_to_cpu(raw_inode->i_size_lo);
2768
2769 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2770 }
2771
ext4_isize_set(struct ext4_inode * raw_inode,loff_t i_size)2772 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2773 {
2774 raw_inode->i_size_lo = cpu_to_le32(i_size);
2775 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2776 }
2777
2778 static inline
ext4_get_group_info(struct super_block * sb,ext4_group_t group)2779 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2780 ext4_group_t group)
2781 {
2782 struct ext4_group_info ***grp_info;
2783 long indexv, indexh;
2784 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
2785 grp_info = EXT4_SB(sb)->s_group_info;
2786 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2787 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2788 return grp_info[indexv][indexh];
2789 }
2790
2791 /*
2792 * Reading s_groups_count requires using smp_rmb() afterwards. See
2793 * the locking protocol documented in the comments of ext4_group_add()
2794 * in resize.c
2795 */
ext4_get_groups_count(struct super_block * sb)2796 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2797 {
2798 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
2799
2800 smp_rmb();
2801 return ngroups;
2802 }
2803
ext4_flex_group(struct ext4_sb_info * sbi,ext4_group_t block_group)2804 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2805 ext4_group_t block_group)
2806 {
2807 return block_group >> sbi->s_log_groups_per_flex;
2808 }
2809
ext4_flex_bg_size(struct ext4_sb_info * sbi)2810 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2811 {
2812 return 1 << sbi->s_log_groups_per_flex;
2813 }
2814
2815 #define ext4_std_error(sb, errno) \
2816 do { \
2817 if ((errno)) \
2818 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
2819 } while (0)
2820
2821 #ifdef CONFIG_SMP
2822 /* Each CPU can accumulate percpu_counter_batch clusters in their local
2823 * counters. So we need to make sure we have free clusters more
2824 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
2825 */
2826 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2827 #else
2828 #define EXT4_FREECLUSTERS_WATERMARK 0
2829 #endif
2830
2831 /* Update i_disksize. Requires i_mutex to avoid races with truncate */
ext4_update_i_disksize(struct inode * inode,loff_t newsize)2832 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2833 {
2834 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
2835 !inode_is_locked(inode));
2836 down_write(&EXT4_I(inode)->i_data_sem);
2837 if (newsize > EXT4_I(inode)->i_disksize)
2838 EXT4_I(inode)->i_disksize = newsize;
2839 up_write(&EXT4_I(inode)->i_data_sem);
2840 }
2841
2842 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
ext4_update_inode_size(struct inode * inode,loff_t newsize)2843 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
2844 {
2845 int changed = 0;
2846
2847 if (newsize > inode->i_size) {
2848 i_size_write(inode, newsize);
2849 changed = 1;
2850 }
2851 if (newsize > EXT4_I(inode)->i_disksize) {
2852 ext4_update_i_disksize(inode, newsize);
2853 changed |= 2;
2854 }
2855 return changed;
2856 }
2857
2858 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
2859 loff_t len);
2860
2861 struct ext4_group_info {
2862 unsigned long bb_state;
2863 struct rb_root bb_free_root;
2864 ext4_grpblk_t bb_first_free; /* first free block */
2865 ext4_grpblk_t bb_free; /* total free blocks */
2866 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
2867 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
2868 struct list_head bb_prealloc_list;
2869 #ifdef DOUBLE_CHECK
2870 void *bb_bitmap;
2871 #endif
2872 struct rw_semaphore alloc_sem;
2873 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
2874 * regions, index is order.
2875 * bb_counters[3] = 5 means
2876 * 5 free 8-block regions. */
2877 };
2878
2879 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
2880 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
2881 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
2882 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
2883 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \
2884 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT)
2885 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \
2886 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT)
2887
2888 #define EXT4_MB_GRP_NEED_INIT(grp) \
2889 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2890 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
2891 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2892 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
2893 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2894
2895 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
2896 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2897 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
2898 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2899 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
2900 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2901
2902 #define EXT4_MAX_CONTENTION 8
2903 #define EXT4_CONTENTION_THRESHOLD 2
2904
ext4_group_lock_ptr(struct super_block * sb,ext4_group_t group)2905 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2906 ext4_group_t group)
2907 {
2908 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2909 }
2910
2911 /*
2912 * Returns true if the filesystem is busy enough that attempts to
2913 * access the block group locks has run into contention.
2914 */
ext4_fs_is_busy(struct ext4_sb_info * sbi)2915 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2916 {
2917 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2918 }
2919
ext4_lock_group(struct super_block * sb,ext4_group_t group)2920 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2921 {
2922 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2923 if (spin_trylock(lock))
2924 /*
2925 * We're able to grab the lock right away, so drop the
2926 * lock contention counter.
2927 */
2928 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2929 else {
2930 /*
2931 * The lock is busy, so bump the contention counter,
2932 * and then wait on the spin lock.
2933 */
2934 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2935 EXT4_MAX_CONTENTION);
2936 spin_lock(lock);
2937 }
2938 }
2939
ext4_unlock_group(struct super_block * sb,ext4_group_t group)2940 static inline void ext4_unlock_group(struct super_block *sb,
2941 ext4_group_t group)
2942 {
2943 spin_unlock(ext4_group_lock_ptr(sb, group));
2944 }
2945
2946 /*
2947 * Block validity checking
2948 */
2949 #define ext4_check_indirect_blockref(inode, bh) \
2950 ext4_check_blockref(__func__, __LINE__, inode, \
2951 (__le32 *)(bh)->b_data, \
2952 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2953
2954 #define ext4_ind_check_inode(inode) \
2955 ext4_check_blockref(__func__, __LINE__, inode, \
2956 EXT4_I(inode)->i_data, \
2957 EXT4_NDIR_BLOCKS)
2958
2959 /*
2960 * Inodes and files operations
2961 */
2962
2963 /* dir.c */
2964 extern const struct file_operations ext4_dir_operations;
2965
2966 /* file.c */
2967 extern const struct inode_operations ext4_file_inode_operations;
2968 extern const struct file_operations ext4_file_operations;
2969 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
2970
2971 /* inline.c */
2972 extern int ext4_get_max_inline_size(struct inode *inode);
2973 extern int ext4_find_inline_data_nolock(struct inode *inode);
2974 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
2975 unsigned int len);
2976 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
2977
2978 extern int ext4_readpage_inline(struct inode *inode, struct page *page);
2979 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
2980 struct inode *inode,
2981 loff_t pos, unsigned len,
2982 unsigned flags,
2983 struct page **pagep);
2984 extern int ext4_write_inline_data_end(struct inode *inode,
2985 loff_t pos, unsigned len,
2986 unsigned copied,
2987 struct page *page);
2988 extern struct buffer_head *
2989 ext4_journalled_write_inline_data(struct inode *inode,
2990 unsigned len,
2991 struct page *page);
2992 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
2993 struct inode *inode,
2994 loff_t pos, unsigned len,
2995 unsigned flags,
2996 struct page **pagep,
2997 void **fsdata);
2998 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
2999 unsigned len, unsigned copied,
3000 struct page *page);
3001 extern int ext4_try_add_inline_entry(handle_t *handle,
3002 struct ext4_filename *fname,
3003 struct inode *dir, struct inode *inode);
3004 extern int ext4_try_create_inline_dir(handle_t *handle,
3005 struct inode *parent,
3006 struct inode *inode);
3007 extern int ext4_read_inline_dir(struct file *filp,
3008 struct dir_context *ctx,
3009 int *has_inline_data);
3010 extern int htree_inlinedir_to_tree(struct file *dir_file,
3011 struct inode *dir, ext4_lblk_t block,
3012 struct dx_hash_info *hinfo,
3013 __u32 start_hash, __u32 start_minor_hash,
3014 int *has_inline_data);
3015 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3016 struct ext4_filename *fname,
3017 struct ext4_dir_entry_2 **res_dir,
3018 int *has_inline_data);
3019 extern int ext4_delete_inline_entry(handle_t *handle,
3020 struct inode *dir,
3021 struct ext4_dir_entry_2 *de_del,
3022 struct buffer_head *bh,
3023 int *has_inline_data);
3024 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3025 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3026 struct ext4_dir_entry_2 **parent_de,
3027 int *retval);
3028 extern int ext4_inline_data_fiemap(struct inode *inode,
3029 struct fiemap_extent_info *fieinfo,
3030 int *has_inline, __u64 start, __u64 len);
3031
3032 struct iomap;
3033 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap);
3034
3035 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3036
3037 extern int ext4_convert_inline_data(struct inode *inode);
3038
ext4_has_inline_data(struct inode * inode)3039 static inline int ext4_has_inline_data(struct inode *inode)
3040 {
3041 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3042 EXT4_I(inode)->i_inline_off;
3043 }
3044
3045 /* namei.c */
3046 extern const struct inode_operations ext4_dir_inode_operations;
3047 extern const struct inode_operations ext4_special_inode_operations;
3048 extern struct dentry *ext4_get_parent(struct dentry *child);
3049 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3050 struct ext4_dir_entry_2 *de,
3051 int blocksize, int csum_size,
3052 unsigned int parent_ino, int dotdot_real_len);
3053 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t,
3054 unsigned int blocksize);
3055 extern int ext4_handle_dirty_dirent_node(handle_t *handle,
3056 struct inode *inode,
3057 struct buffer_head *bh);
3058 #define S_SHIFT 12
3059 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = {
3060 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
3061 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
3062 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
3063 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
3064 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
3065 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
3066 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
3067 };
3068
ext4_set_de_type(struct super_block * sb,struct ext4_dir_entry_2 * de,umode_t mode)3069 static inline void ext4_set_de_type(struct super_block *sb,
3070 struct ext4_dir_entry_2 *de,
3071 umode_t mode) {
3072 if (ext4_has_feature_filetype(sb))
3073 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3074 }
3075
3076 /* readpages.c */
3077 extern int ext4_mpage_readpages(struct address_space *mapping,
3078 struct list_head *pages, struct page *page,
3079 unsigned nr_pages, bool is_readahead);
3080
3081 /* symlink.c */
3082 extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3083 extern const struct inode_operations ext4_symlink_inode_operations;
3084 extern const struct inode_operations ext4_fast_symlink_inode_operations;
3085
3086 /* sysfs.c */
3087 extern int ext4_register_sysfs(struct super_block *sb);
3088 extern void ext4_unregister_sysfs(struct super_block *sb);
3089 extern int __init ext4_init_sysfs(void);
3090 extern void ext4_exit_sysfs(void);
3091
3092 /* block_validity */
3093 extern void ext4_release_system_zone(struct super_block *sb);
3094 extern int ext4_setup_system_zone(struct super_block *sb);
3095 extern int __init ext4_init_system_zone(void);
3096 extern void ext4_exit_system_zone(void);
3097 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
3098 ext4_fsblk_t start_blk,
3099 unsigned int count);
3100 extern int ext4_check_blockref(const char *, unsigned int,
3101 struct inode *, __le32 *, unsigned int);
3102
3103 /* extents.c */
3104 struct ext4_ext_path;
3105 struct ext4_extent;
3106
3107 /*
3108 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3109 * __le32.
3110 */
3111 #define EXT_MAX_BLOCKS 0xffffffff
3112
3113 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
3114 extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
3115 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3116 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3117 struct ext4_map_blocks *map, int flags);
3118 extern int ext4_ext_truncate(handle_t *, struct inode *);
3119 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3120 ext4_lblk_t end);
3121 extern void ext4_ext_init(struct super_block *);
3122 extern void ext4_ext_release(struct super_block *);
3123 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3124 loff_t len);
3125 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3126 loff_t offset, ssize_t len);
3127 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3128 struct ext4_map_blocks *map, int flags);
3129 extern int ext4_ext_calc_metadata_amount(struct inode *inode,
3130 ext4_lblk_t lblocks);
3131 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3132 int num,
3133 struct ext4_ext_path *path);
3134 extern int ext4_can_extents_be_merged(struct inode *inode,
3135 struct ext4_extent *ex1,
3136 struct ext4_extent *ex2);
3137 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3138 struct ext4_ext_path **,
3139 struct ext4_extent *, int);
3140 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3141 struct ext4_ext_path **,
3142 int flags);
3143 extern void ext4_ext_drop_refs(struct ext4_ext_path *);
3144 extern int ext4_ext_check_inode(struct inode *inode);
3145 extern int ext4_find_delalloc_range(struct inode *inode,
3146 ext4_lblk_t lblk_start,
3147 ext4_lblk_t lblk_end);
3148 extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk);
3149 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3150 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3151 __u64 start, __u64 len);
3152 extern int ext4_ext_precache(struct inode *inode);
3153 extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len);
3154 extern int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len);
3155 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3156 struct inode *inode2, ext4_lblk_t lblk1,
3157 ext4_lblk_t lblk2, ext4_lblk_t count,
3158 int mark_unwritten,int *err);
3159
3160 /* move_extent.c */
3161 extern void ext4_double_down_write_data_sem(struct inode *first,
3162 struct inode *second);
3163 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3164 struct inode *donor_inode);
3165 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3166 __u64 start_orig, __u64 start_donor,
3167 __u64 len, __u64 *moved_len);
3168
3169 /* page-io.c */
3170 extern int __init ext4_init_pageio(void);
3171 extern void ext4_exit_pageio(void);
3172 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3173 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3174 extern int ext4_put_io_end(ext4_io_end_t *io_end);
3175 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3176 extern void ext4_io_submit_init(struct ext4_io_submit *io,
3177 struct writeback_control *wbc);
3178 extern void ext4_end_io_rsv_work(struct work_struct *work);
3179 extern void ext4_io_submit(struct ext4_io_submit *io);
3180 extern int ext4_bio_write_page(struct ext4_io_submit *io,
3181 struct page *page,
3182 int len,
3183 struct writeback_control *wbc,
3184 bool keep_towrite);
3185
3186 /* mmp.c */
3187 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3188
3189 /*
3190 * Add new method to test whether block and inode bitmaps are properly
3191 * initialized. With uninit_bg reading the block from disk is not enough
3192 * to mark the bitmap uptodate. We need to also zero-out the bitmap
3193 */
3194 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3195
bitmap_uptodate(struct buffer_head * bh)3196 static inline int bitmap_uptodate(struct buffer_head *bh)
3197 {
3198 return (buffer_uptodate(bh) &&
3199 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3200 }
set_bitmap_uptodate(struct buffer_head * bh)3201 static inline void set_bitmap_uptodate(struct buffer_head *bh)
3202 {
3203 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3204 }
3205
3206 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
3207
3208 /* For ioend & aio unwritten conversion wait queues */
3209 #define EXT4_WQ_HASH_SZ 37
3210 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
3211 EXT4_WQ_HASH_SZ])
3212 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3213
3214 extern int ext4_resize_begin(struct super_block *sb);
3215 extern void ext4_resize_end(struct super_block *sb);
3216
ext4_set_io_unwritten_flag(struct inode * inode,struct ext4_io_end * io_end)3217 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
3218 struct ext4_io_end *io_end)
3219 {
3220 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
3221 io_end->flag |= EXT4_IO_END_UNWRITTEN;
3222 atomic_inc(&EXT4_I(inode)->i_unwritten);
3223 }
3224 }
3225
ext4_clear_io_unwritten_flag(ext4_io_end_t * io_end)3226 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3227 {
3228 struct inode *inode = io_end->inode;
3229
3230 if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
3231 io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3232 /* Wake up anyone waiting on unwritten extent conversion */
3233 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
3234 wake_up_all(ext4_ioend_wq(inode));
3235 }
3236 }
3237
3238 extern const struct iomap_ops ext4_iomap_ops;
3239
3240 #endif /* __KERNEL__ */
3241
3242 #define EFSBADCRC EBADMSG /* Bad CRC detected */
3243 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
3244
3245 #endif /* _EXT4_H */
3246