]> git.proxmox.com Git - mirror_ubuntu-focal-kernel.git/blob - fs/ext4/ext4.h
Merge tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi
[mirror_ubuntu-focal-kernel.git] / fs / ext4 / ext4.h
1 // SPDX-License-Identifier: GPL-2.0
2 /*
3 * ext4.h
4 *
5 * Copyright (C) 1992, 1993, 1994, 1995
6 * Remy Card (card@masi.ibp.fr)
7 * Laboratoire MASI - Institut Blaise Pascal
8 * Universite Pierre et Marie Curie (Paris VI)
9 *
10 * from
11 *
12 * linux/include/linux/minix_fs.h
13 *
14 * Copyright (C) 1991, 1992 Linus Torvalds
15 */
16
17 #ifndef _EXT4_H
18 #define _EXT4_H
19
20 #include <linux/types.h>
21 #include <linux/blkdev.h>
22 #include <linux/magic.h>
23 #include <linux/jbd2.h>
24 #include <linux/quota.h>
25 #include <linux/rwsem.h>
26 #include <linux/rbtree.h>
27 #include <linux/seqlock.h>
28 #include <linux/mutex.h>
29 #include <linux/timer.h>
30 #include <linux/version.h>
31 #include <linux/wait.h>
32 #include <linux/sched/signal.h>
33 #include <linux/blockgroup_lock.h>
34 #include <linux/percpu_counter.h>
35 #include <linux/ratelimit.h>
36 #include <crypto/hash.h>
37 #include <linux/falloc.h>
38 #include <linux/percpu-rwsem.h>
39 #ifdef __KERNEL__
40 #include <linux/compat.h>
41 #endif
42
43 #include <linux/fscrypt.h>
44
45 #include <linux/compiler.h>
46
47 /*
48 * The fourth extended filesystem constants/structures
49 */
50
51 /*
52 * with AGGRESSIVE_CHECK allocator runs consistency checks over
53 * structures. these checks slow things down a lot
54 */
55 #define AGGRESSIVE_CHECK__
56
57 /*
58 * with DOUBLE_CHECK defined mballoc creates persistent in-core
59 * bitmaps, maintains and uses them to check for double allocations
60 */
61 #define DOUBLE_CHECK__
62
63 /*
64 * Define EXT4FS_DEBUG to produce debug messages
65 */
66 #undef EXT4FS_DEBUG
67
68 /*
69 * Debug code
70 */
71 #ifdef EXT4FS_DEBUG
72 #define ext4_debug(f, a...) \
73 do { \
74 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
75 __FILE__, __LINE__, __func__); \
76 printk(KERN_DEBUG f, ## a); \
77 } while (0)
78 #else
79 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
80 #endif
81
82 /*
83 * Turn on EXT_DEBUG to get lots of info about extents operations.
84 */
85 #define EXT_DEBUG__
86 #ifdef EXT_DEBUG
87 #define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__)
88 #else
89 #define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
90 #endif
91
92 /* data type for block offset of block group */
93 typedef int ext4_grpblk_t;
94
95 /* data type for filesystem-wide blocks number */
96 typedef unsigned long long ext4_fsblk_t;
97
98 /* data type for file logical block number */
99 typedef __u32 ext4_lblk_t;
100
101 /* data type for block group number */
102 typedef unsigned int ext4_group_t;
103
104 enum SHIFT_DIRECTION {
105 SHIFT_LEFT = 0,
106 SHIFT_RIGHT,
107 };
108
109 /*
110 * Flags used in mballoc's allocation_context flags field.
111 *
112 * Also used to show what's going on for debugging purposes when the
113 * flag field is exported via the traceport interface
114 */
115
116 /* prefer goal again. length */
117 #define EXT4_MB_HINT_MERGE 0x0001
118 /* blocks already reserved */
119 #define EXT4_MB_HINT_RESERVED 0x0002
120 /* metadata is being allocated */
121 #define EXT4_MB_HINT_METADATA 0x0004
122 /* first blocks in the file */
123 #define EXT4_MB_HINT_FIRST 0x0008
124 /* search for the best chunk */
125 #define EXT4_MB_HINT_BEST 0x0010
126 /* data is being allocated */
127 #define EXT4_MB_HINT_DATA 0x0020
128 /* don't preallocate (for tails) */
129 #define EXT4_MB_HINT_NOPREALLOC 0x0040
130 /* allocate for locality group */
131 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080
132 /* allocate goal blocks or none */
133 #define EXT4_MB_HINT_GOAL_ONLY 0x0100
134 /* goal is meaningful */
135 #define EXT4_MB_HINT_TRY_GOAL 0x0200
136 /* blocks already pre-reserved by delayed allocation */
137 #define EXT4_MB_DELALLOC_RESERVED 0x0400
138 /* We are doing stream allocation */
139 #define EXT4_MB_STREAM_ALLOC 0x0800
140 /* Use reserved root blocks if needed */
141 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000
142 /* Use blocks from reserved pool */
143 #define EXT4_MB_USE_RESERVED 0x2000
144
145 struct ext4_allocation_request {
146 /* target inode for block we're allocating */
147 struct inode *inode;
148 /* how many blocks we want to allocate */
149 unsigned int len;
150 /* logical block in target inode */
151 ext4_lblk_t logical;
152 /* the closest logical allocated block to the left */
153 ext4_lblk_t lleft;
154 /* the closest logical allocated block to the right */
155 ext4_lblk_t lright;
156 /* phys. target (a hint) */
157 ext4_fsblk_t goal;
158 /* phys. block for the closest logical allocated block to the left */
159 ext4_fsblk_t pleft;
160 /* phys. block for the closest logical allocated block to the right */
161 ext4_fsblk_t pright;
162 /* flags. see above EXT4_MB_HINT_* */
163 unsigned int flags;
164 };
165
166 /*
167 * Logical to physical block mapping, used by ext4_map_blocks()
168 *
169 * This structure is used to pass requests into ext4_map_blocks() as
170 * well as to store the information returned by ext4_map_blocks(). It
171 * takes less room on the stack than a struct buffer_head.
172 */
173 #define EXT4_MAP_NEW (1 << BH_New)
174 #define EXT4_MAP_MAPPED (1 << BH_Mapped)
175 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten)
176 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary)
177 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
178 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
179
180 struct ext4_map_blocks {
181 ext4_fsblk_t m_pblk;
182 ext4_lblk_t m_lblk;
183 unsigned int m_len;
184 unsigned int m_flags;
185 };
186
187 /*
188 * Flags for ext4_io_end->flags
189 */
190 #define EXT4_IO_END_UNWRITTEN 0x0001
191
192 /*
193 * For converting unwritten extents on a work queue. 'handle' is used for
194 * buffered writeback.
195 */
196 typedef struct ext4_io_end {
197 struct list_head list; /* per-file finished IO list */
198 handle_t *handle; /* handle reserved for extent
199 * conversion */
200 struct inode *inode; /* file being written to */
201 struct bio *bio; /* Linked list of completed
202 * bios covering the extent */
203 unsigned int flag; /* unwritten or not */
204 atomic_t count; /* reference counter */
205 loff_t offset; /* offset in the file */
206 ssize_t size; /* size of the extent */
207 } ext4_io_end_t;
208
209 struct ext4_io_submit {
210 struct writeback_control *io_wbc;
211 struct bio *io_bio;
212 ext4_io_end_t *io_end;
213 sector_t io_next_block;
214 };
215
216 /*
217 * Special inodes numbers
218 */
219 #define EXT4_BAD_INO 1 /* Bad blocks inode */
220 #define EXT4_ROOT_INO 2 /* Root inode */
221 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */
222 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
223 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
224 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
225 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
226 #define EXT4_JOURNAL_INO 8 /* Journal inode */
227
228 /* First non-reserved inode for old ext4 filesystems */
229 #define EXT4_GOOD_OLD_FIRST_INO 11
230
231 /*
232 * Maximal count of links to a file
233 */
234 #define EXT4_LINK_MAX 65000
235
236 /*
237 * Macro-instructions used to manage several block sizes
238 */
239 #define EXT4_MIN_BLOCK_SIZE 1024
240 #define EXT4_MAX_BLOCK_SIZE 65536
241 #define EXT4_MIN_BLOCK_LOG_SIZE 10
242 #define EXT4_MAX_BLOCK_LOG_SIZE 16
243 #define EXT4_MAX_CLUSTER_LOG_SIZE 30
244 #ifdef __KERNEL__
245 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
246 #else
247 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
248 #endif
249 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
250 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
251 EXT4_SB(s)->s_cluster_bits)
252 #ifdef __KERNEL__
253 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
254 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
255 #else
256 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
257 #endif
258 #ifdef __KERNEL__
259 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
260 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
261 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
262 #else
263 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
264 EXT4_GOOD_OLD_INODE_SIZE : \
265 (s)->s_inode_size)
266 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
267 EXT4_GOOD_OLD_FIRST_INO : \
268 (s)->s_first_ino)
269 #endif
270 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
271 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \
272 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \
273 blkbits))
274
275 /* Translate a block number to a cluster number */
276 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
277 /* Translate a cluster number to a block number */
278 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
279 /* Translate # of blks to # of clusters */
280 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
281 (sbi)->s_cluster_bits)
282 /* Mask out the low bits to get the starting block of the cluster */
283 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
284 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
285 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
286 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
287 /* Get the cluster offset */
288 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
289 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
290 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
291 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
292
293 /*
294 * Structure of a blocks group descriptor
295 */
296 struct ext4_group_desc
297 {
298 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
299 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
300 __le32 bg_inode_table_lo; /* Inodes table block */
301 __le16 bg_free_blocks_count_lo;/* Free blocks count */
302 __le16 bg_free_inodes_count_lo;/* Free inodes count */
303 __le16 bg_used_dirs_count_lo; /* Directories count */
304 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
305 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
306 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
307 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
308 __le16 bg_itable_unused_lo; /* Unused inodes count */
309 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
310 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
311 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
312 __le32 bg_inode_table_hi; /* Inodes table block MSB */
313 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
314 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
315 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
316 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
317 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
318 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
319 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
320 __u32 bg_reserved;
321 };
322
323 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
324 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
325 sizeof(__le16))
326 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
327 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
328 sizeof(__le16))
329
330 /*
331 * Structure of a flex block group info
332 */
333
334 struct flex_groups {
335 atomic64_t free_clusters;
336 atomic_t free_inodes;
337 atomic_t used_dirs;
338 };
339
340 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
341 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
342 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
343
344 /*
345 * Macro-instructions used to manage group descriptors
346 */
347 #define EXT4_MIN_DESC_SIZE 32
348 #define EXT4_MIN_DESC_SIZE_64BIT 64
349 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
350 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
351 #ifdef __KERNEL__
352 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
353 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
354 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
355 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
356 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
357 #else
358 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
359 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
360 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
361 #endif
362
363 /*
364 * Constants relative to the data blocks
365 */
366 #define EXT4_NDIR_BLOCKS 12
367 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
368 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
369 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
370 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
371
372 /*
373 * Inode flags
374 */
375 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
376 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
377 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
378 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
379 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
380 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
381 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
382 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
383 /* Reserved for compression usage... */
384 #define EXT4_DIRTY_FL 0x00000100
385 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
386 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
387 /* nb: was previously EXT2_ECOMPR_FL */
388 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */
389 /* End compression flags --- maybe not all used */
390 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
391 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
392 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
393 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
394 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
395 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
396 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
397 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
398 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
399 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
400 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
401 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
402 #define EXT4_CASEFOLD_FL 0x40000000 /* Casefolded file */
403 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
404
405 #define EXT4_FL_USER_VISIBLE 0x704BDFFF /* User visible flags */
406 #define EXT4_FL_USER_MODIFIABLE 0x604BC0FF /* User modifiable flags */
407
408 /* Flags we can manipulate with through EXT4_IOC_FSSETXATTR */
409 #define EXT4_FL_XFLAG_VISIBLE (EXT4_SYNC_FL | \
410 EXT4_IMMUTABLE_FL | \
411 EXT4_APPEND_FL | \
412 EXT4_NODUMP_FL | \
413 EXT4_NOATIME_FL | \
414 EXT4_PROJINHERIT_FL)
415
416 /* Flags that should be inherited by new inodes from their parent. */
417 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
418 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
419 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
420 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
421 EXT4_PROJINHERIT_FL | EXT4_CASEFOLD_FL)
422
423 /* Flags that are appropriate for regular files (all but dir-specific ones). */
424 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL | EXT4_CASEFOLD_FL))
425
426 /* Flags that are appropriate for non-directories/regular files. */
427 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
428
429 /* The only flags that should be swapped */
430 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL)
431
432 /* Mask out flags that are inappropriate for the given type of inode. */
433 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
434 {
435 if (S_ISDIR(mode))
436 return flags;
437 else if (S_ISREG(mode))
438 return flags & EXT4_REG_FLMASK;
439 else
440 return flags & EXT4_OTHER_FLMASK;
441 }
442
443 /*
444 * Inode flags used for atomic set/get
445 */
446 enum {
447 EXT4_INODE_SECRM = 0, /* Secure deletion */
448 EXT4_INODE_UNRM = 1, /* Undelete */
449 EXT4_INODE_COMPR = 2, /* Compress file */
450 EXT4_INODE_SYNC = 3, /* Synchronous updates */
451 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
452 EXT4_INODE_APPEND = 5, /* writes to file may only append */
453 EXT4_INODE_NODUMP = 6, /* do not dump file */
454 EXT4_INODE_NOATIME = 7, /* do not update atime */
455 /* Reserved for compression usage... */
456 EXT4_INODE_DIRTY = 8,
457 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
458 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
459 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */
460 /* End compression flags --- maybe not all used */
461 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
462 EXT4_INODE_IMAGIC = 13, /* AFS directory */
463 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
464 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
465 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
466 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
467 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
468 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
469 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
470 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
471 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
472 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */
473 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
474 };
475
476 /*
477 * Since it's pretty easy to mix up bit numbers and hex values, we use a
478 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
479 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
480 * any extra space in the compiled kernel image, otherwise, the build will fail.
481 * It's important that these values are the same, since we are using
482 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
483 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
484 * values found in ext2, ext3 and ext4 filesystems, and of course the values
485 * defined in e2fsprogs.
486 *
487 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
488 */
489 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
490 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
491
492 static inline void ext4_check_flag_values(void)
493 {
494 CHECK_FLAG_VALUE(SECRM);
495 CHECK_FLAG_VALUE(UNRM);
496 CHECK_FLAG_VALUE(COMPR);
497 CHECK_FLAG_VALUE(SYNC);
498 CHECK_FLAG_VALUE(IMMUTABLE);
499 CHECK_FLAG_VALUE(APPEND);
500 CHECK_FLAG_VALUE(NODUMP);
501 CHECK_FLAG_VALUE(NOATIME);
502 CHECK_FLAG_VALUE(DIRTY);
503 CHECK_FLAG_VALUE(COMPRBLK);
504 CHECK_FLAG_VALUE(NOCOMPR);
505 CHECK_FLAG_VALUE(ENCRYPT);
506 CHECK_FLAG_VALUE(INDEX);
507 CHECK_FLAG_VALUE(IMAGIC);
508 CHECK_FLAG_VALUE(JOURNAL_DATA);
509 CHECK_FLAG_VALUE(NOTAIL);
510 CHECK_FLAG_VALUE(DIRSYNC);
511 CHECK_FLAG_VALUE(TOPDIR);
512 CHECK_FLAG_VALUE(HUGE_FILE);
513 CHECK_FLAG_VALUE(EXTENTS);
514 CHECK_FLAG_VALUE(EA_INODE);
515 CHECK_FLAG_VALUE(EOFBLOCKS);
516 CHECK_FLAG_VALUE(INLINE_DATA);
517 CHECK_FLAG_VALUE(PROJINHERIT);
518 CHECK_FLAG_VALUE(RESERVED);
519 }
520
521 /* Used to pass group descriptor data when online resize is done */
522 struct ext4_new_group_input {
523 __u32 group; /* Group number for this data */
524 __u64 block_bitmap; /* Absolute block number of block bitmap */
525 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
526 __u64 inode_table; /* Absolute block number of inode table start */
527 __u32 blocks_count; /* Total number of blocks in this group */
528 __u16 reserved_blocks; /* Number of reserved blocks in this group */
529 __u16 unused;
530 };
531
532 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
533 struct compat_ext4_new_group_input {
534 u32 group;
535 compat_u64 block_bitmap;
536 compat_u64 inode_bitmap;
537 compat_u64 inode_table;
538 u32 blocks_count;
539 u16 reserved_blocks;
540 u16 unused;
541 };
542 #endif
543
544 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
545 struct ext4_new_group_data {
546 __u32 group;
547 __u64 block_bitmap;
548 __u64 inode_bitmap;
549 __u64 inode_table;
550 __u32 blocks_count;
551 __u16 reserved_blocks;
552 __u16 mdata_blocks;
553 __u32 free_clusters_count;
554 };
555
556 /* Indexes used to index group tables in ext4_new_group_data */
557 enum {
558 BLOCK_BITMAP = 0, /* block bitmap */
559 INODE_BITMAP, /* inode bitmap */
560 INODE_TABLE, /* inode tables */
561 GROUP_TABLE_COUNT,
562 };
563
564 /*
565 * Flags used by ext4_map_blocks()
566 */
567 /* Allocate any needed blocks and/or convert an unwritten
568 extent to be an initialized ext4 */
569 #define EXT4_GET_BLOCKS_CREATE 0x0001
570 /* Request the creation of an unwritten extent */
571 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002
572 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\
573 EXT4_GET_BLOCKS_CREATE)
574 /* Caller is from the delayed allocation writeout path
575 * finally doing the actual allocation of delayed blocks */
576 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
577 /* caller is from the direct IO path, request to creation of an
578 unwritten extents if not allocated, split the unwritten
579 extent if blocks has been preallocated already*/
580 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
581 #define EXT4_GET_BLOCKS_CONVERT 0x0010
582 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
583 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
584 /* Convert extent to initialized after IO complete */
585 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
586 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
587 /* Eventual metadata allocation (due to growing extent tree)
588 * should not fail, so try to use reserved blocks for that.*/
589 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
590 /* Don't normalize allocation size (used for fallocate) */
591 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
592 /* Request will not result in inode size update (user for fallocate) */
593 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080
594 /* Convert written extents to unwritten */
595 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100
596 /* Write zeros to newly created written extents */
597 #define EXT4_GET_BLOCKS_ZERO 0x0200
598 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\
599 EXT4_GET_BLOCKS_ZERO)
600 /* Caller will submit data before dropping transaction handle. This
601 * allows jbd2 to avoid submitting data before commit. */
602 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400
603
604 /*
605 * The bit position of these flags must not overlap with any of the
606 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(),
607 * read_extent_tree_block(), ext4_split_extent_at(),
608 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
609 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
610 * caching the extents when reading from the extent tree while a
611 * truncate or punch hole operation is in progress.
612 */
613 #define EXT4_EX_NOCACHE 0x40000000
614 #define EXT4_EX_FORCE_CACHE 0x20000000
615
616 /*
617 * Flags used by ext4_free_blocks
618 */
619 #define EXT4_FREE_BLOCKS_METADATA 0x0001
620 #define EXT4_FREE_BLOCKS_FORGET 0x0002
621 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
622 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
623 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
624 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
625 #define EXT4_FREE_BLOCKS_RERESERVE_CLUSTER 0x0040
626
627 /*
628 * ioctl commands
629 */
630 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
631 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
632 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
633 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
634 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
635 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
636 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
637 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
638 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
639 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
640 #define EXT4_IOC_MIGRATE _IO('f', 9)
641 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
642 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
643 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
644 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
645 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
646 #define EXT4_IOC_SWAP_BOOT _IO('f', 17)
647 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18)
648 #define EXT4_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY
649 #define EXT4_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT
650 #define EXT4_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY
651
652 #define EXT4_IOC_FSGETXATTR FS_IOC_FSGETXATTR
653 #define EXT4_IOC_FSSETXATTR FS_IOC_FSSETXATTR
654
655 #define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32)
656
657 /*
658 * Flags for going down operation
659 */
660 #define EXT4_GOING_FLAGS_DEFAULT 0x0 /* going down */
661 #define EXT4_GOING_FLAGS_LOGFLUSH 0x1 /* flush log but not data */
662 #define EXT4_GOING_FLAGS_NOLOGFLUSH 0x2 /* don't flush log nor data */
663
664
665 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
666 /*
667 * ioctl commands in 32 bit emulation
668 */
669 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
670 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
671 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
672 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
673 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
674 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
675 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
676 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
677 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
678 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
679 #endif
680
681 /* Max physical block we can address w/o extents */
682 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
683
684 /* Max logical block we can support */
685 #define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFF
686
687 /*
688 * Structure of an inode on the disk
689 */
690 struct ext4_inode {
691 __le16 i_mode; /* File mode */
692 __le16 i_uid; /* Low 16 bits of Owner Uid */
693 __le32 i_size_lo; /* Size in bytes */
694 __le32 i_atime; /* Access time */
695 __le32 i_ctime; /* Inode Change time */
696 __le32 i_mtime; /* Modification time */
697 __le32 i_dtime; /* Deletion Time */
698 __le16 i_gid; /* Low 16 bits of Group Id */
699 __le16 i_links_count; /* Links count */
700 __le32 i_blocks_lo; /* Blocks count */
701 __le32 i_flags; /* File flags */
702 union {
703 struct {
704 __le32 l_i_version;
705 } linux1;
706 struct {
707 __u32 h_i_translator;
708 } hurd1;
709 struct {
710 __u32 m_i_reserved1;
711 } masix1;
712 } osd1; /* OS dependent 1 */
713 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
714 __le32 i_generation; /* File version (for NFS) */
715 __le32 i_file_acl_lo; /* File ACL */
716 __le32 i_size_high;
717 __le32 i_obso_faddr; /* Obsoleted fragment address */
718 union {
719 struct {
720 __le16 l_i_blocks_high; /* were l_i_reserved1 */
721 __le16 l_i_file_acl_high;
722 __le16 l_i_uid_high; /* these 2 fields */
723 __le16 l_i_gid_high; /* were reserved2[0] */
724 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
725 __le16 l_i_reserved;
726 } linux2;
727 struct {
728 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
729 __u16 h_i_mode_high;
730 __u16 h_i_uid_high;
731 __u16 h_i_gid_high;
732 __u32 h_i_author;
733 } hurd2;
734 struct {
735 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
736 __le16 m_i_file_acl_high;
737 __u32 m_i_reserved2[2];
738 } masix2;
739 } osd2; /* OS dependent 2 */
740 __le16 i_extra_isize;
741 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
742 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
743 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
744 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
745 __le32 i_crtime; /* File Creation time */
746 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
747 __le32 i_version_hi; /* high 32 bits for 64-bit version */
748 __le32 i_projid; /* Project ID */
749 };
750
751 struct move_extent {
752 __u32 reserved; /* should be zero */
753 __u32 donor_fd; /* donor file descriptor */
754 __u64 orig_start; /* logical start offset in block for orig */
755 __u64 donor_start; /* logical start offset in block for donor */
756 __u64 len; /* block length to be moved */
757 __u64 moved_len; /* moved block length */
758 };
759
760 #define EXT4_EPOCH_BITS 2
761 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
762 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
763
764 /*
765 * Extended fields will fit into an inode if the filesystem was formatted
766 * with large inodes (-I 256 or larger) and there are not currently any EAs
767 * consuming all of the available space. For new inodes we always reserve
768 * enough space for the kernel's known extended fields, but for inodes
769 * created with an old kernel this might not have been the case. None of
770 * the extended inode fields is critical for correct filesystem operation.
771 * This macro checks if a certain field fits in the inode. Note that
772 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
773 */
774 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
775 ((offsetof(typeof(*ext4_inode), field) + \
776 sizeof((ext4_inode)->field)) \
777 <= (EXT4_GOOD_OLD_INODE_SIZE + \
778 (einode)->i_extra_isize)) \
779
780 /*
781 * We use an encoding that preserves the times for extra epoch "00":
782 *
783 * extra msb of adjust for signed
784 * epoch 32-bit 32-bit tv_sec to
785 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
786 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
787 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
788 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
789 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
790 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
791 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
792 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
793 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
794 *
795 * Note that previous versions of the kernel on 64-bit systems would
796 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
797 * 1970. e2fsck will correct this, assuming that it is run on the
798 * affected filesystem before 2242.
799 */
800
801 static inline __le32 ext4_encode_extra_time(struct timespec64 *time)
802 {
803 u32 extra =((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK;
804 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
805 }
806
807 static inline void ext4_decode_extra_time(struct timespec64 *time,
808 __le32 extra)
809 {
810 if (unlikely(extra & cpu_to_le32(EXT4_EPOCH_MASK))) {
811
812 #if 1
813 /* Handle legacy encoding of pre-1970 dates with epoch
814 * bits 1,1. (This backwards compatibility may be removed
815 * at the discretion of the ext4 developers.)
816 */
817 u64 extra_bits = le32_to_cpu(extra) & EXT4_EPOCH_MASK;
818 if (extra_bits == 3 && ((time->tv_sec) & 0x80000000) != 0)
819 extra_bits = 0;
820 time->tv_sec += extra_bits << 32;
821 #else
822 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
823 #endif
824 }
825 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
826 }
827
828 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
829 do { \
830 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
831 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) {\
832 (raw_inode)->xtime ## _extra = \
833 ext4_encode_extra_time(&(inode)->xtime); \
834 } \
835 } while (0)
836
837 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
838 do { \
839 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
840 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
841 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
842 (raw_inode)->xtime ## _extra = \
843 ext4_encode_extra_time(&(einode)->xtime); \
844 } while (0)
845
846 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
847 do { \
848 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
849 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) { \
850 ext4_decode_extra_time(&(inode)->xtime, \
851 raw_inode->xtime ## _extra); \
852 } \
853 else \
854 (inode)->xtime.tv_nsec = 0; \
855 } while (0)
856
857
858 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
859 do { \
860 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
861 (einode)->xtime.tv_sec = \
862 (signed)le32_to_cpu((raw_inode)->xtime); \
863 else \
864 (einode)->xtime.tv_sec = 0; \
865 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
866 ext4_decode_extra_time(&(einode)->xtime, \
867 raw_inode->xtime ## _extra); \
868 else \
869 (einode)->xtime.tv_nsec = 0; \
870 } while (0)
871
872 #define i_disk_version osd1.linux1.l_i_version
873
874 #if defined(__KERNEL__) || defined(__linux__)
875 #define i_reserved1 osd1.linux1.l_i_reserved1
876 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
877 #define i_blocks_high osd2.linux2.l_i_blocks_high
878 #define i_uid_low i_uid
879 #define i_gid_low i_gid
880 #define i_uid_high osd2.linux2.l_i_uid_high
881 #define i_gid_high osd2.linux2.l_i_gid_high
882 #define i_checksum_lo osd2.linux2.l_i_checksum_lo
883
884 #elif defined(__GNU__)
885
886 #define i_translator osd1.hurd1.h_i_translator
887 #define i_uid_high osd2.hurd2.h_i_uid_high
888 #define i_gid_high osd2.hurd2.h_i_gid_high
889 #define i_author osd2.hurd2.h_i_author
890
891 #elif defined(__masix__)
892
893 #define i_reserved1 osd1.masix1.m_i_reserved1
894 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
895 #define i_reserved2 osd2.masix2.m_i_reserved2
896
897 #endif /* defined(__KERNEL__) || defined(__linux__) */
898
899 #include "extents_status.h"
900
901 /*
902 * Lock subclasses for i_data_sem in the ext4_inode_info structure.
903 *
904 * These are needed to avoid lockdep false positives when we need to
905 * allocate blocks to the quota inode during ext4_map_blocks(), while
906 * holding i_data_sem for a normal (non-quota) inode. Since we don't
907 * do quota tracking for the quota inode, this avoids deadlock (as
908 * well as infinite recursion, since it isn't turtles all the way
909 * down...)
910 *
911 * I_DATA_SEM_NORMAL - Used for most inodes
912 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode
913 * where the second inode has larger inode number
914 * than the first
915 * I_DATA_SEM_QUOTA - Used for quota inodes only
916 */
917 enum {
918 I_DATA_SEM_NORMAL = 0,
919 I_DATA_SEM_OTHER,
920 I_DATA_SEM_QUOTA,
921 };
922
923
924 /*
925 * fourth extended file system inode data in memory
926 */
927 struct ext4_inode_info {
928 __le32 i_data[15]; /* unconverted */
929 __u32 i_dtime;
930 ext4_fsblk_t i_file_acl;
931
932 /*
933 * i_block_group is the number of the block group which contains
934 * this file's inode. Constant across the lifetime of the inode,
935 * it is used for making block allocation decisions - we try to
936 * place a file's data blocks near its inode block, and new inodes
937 * near to their parent directory's inode.
938 */
939 ext4_group_t i_block_group;
940 ext4_lblk_t i_dir_start_lookup;
941 #if (BITS_PER_LONG < 64)
942 unsigned long i_state_flags; /* Dynamic state flags */
943 #endif
944 unsigned long i_flags;
945
946 /*
947 * Extended attributes can be read independently of the main file
948 * data. Taking i_mutex even when reading would cause contention
949 * between readers of EAs and writers of regular file data, so
950 * instead we synchronize on xattr_sem when reading or changing
951 * EAs.
952 */
953 struct rw_semaphore xattr_sem;
954
955 struct list_head i_orphan; /* unlinked but open inodes */
956
957 /*
958 * i_disksize keeps track of what the inode size is ON DISK, not
959 * in memory. During truncate, i_size is set to the new size by
960 * the VFS prior to calling ext4_truncate(), but the filesystem won't
961 * set i_disksize to 0 until the truncate is actually under way.
962 *
963 * The intent is that i_disksize always represents the blocks which
964 * are used by this file. This allows recovery to restart truncate
965 * on orphans if we crash during truncate. We actually write i_disksize
966 * into the on-disk inode when writing inodes out, instead of i_size.
967 *
968 * The only time when i_disksize and i_size may be different is when
969 * a truncate is in progress. The only things which change i_disksize
970 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
971 */
972 loff_t i_disksize;
973
974 /*
975 * i_data_sem is for serialising ext4_truncate() against
976 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
977 * data tree are chopped off during truncate. We can't do that in
978 * ext4 because whenever we perform intermediate commits during
979 * truncate, the inode and all the metadata blocks *must* be in a
980 * consistent state which allows truncation of the orphans to restart
981 * during recovery. Hence we must fix the get_block-vs-truncate race
982 * by other means, so we have i_data_sem.
983 */
984 struct rw_semaphore i_data_sem;
985 /*
986 * i_mmap_sem is for serializing page faults with truncate / punch hole
987 * operations. We have to make sure that new page cannot be faulted in
988 * a section of the inode that is being punched. We cannot easily use
989 * i_data_sem for this since we need protection for the whole punch
990 * operation and i_data_sem ranks below transaction start so we have
991 * to occasionally drop it.
992 */
993 struct rw_semaphore i_mmap_sem;
994 struct inode vfs_inode;
995 struct jbd2_inode *jinode;
996
997 spinlock_t i_raw_lock; /* protects updates to the raw inode */
998
999 /*
1000 * File creation time. Its function is same as that of
1001 * struct timespec64 i_{a,c,m}time in the generic inode.
1002 */
1003 struct timespec64 i_crtime;
1004
1005 /* mballoc */
1006 struct list_head i_prealloc_list;
1007 spinlock_t i_prealloc_lock;
1008
1009 /* extents status tree */
1010 struct ext4_es_tree i_es_tree;
1011 rwlock_t i_es_lock;
1012 struct list_head i_es_list;
1013 unsigned int i_es_all_nr; /* protected by i_es_lock */
1014 unsigned int i_es_shk_nr; /* protected by i_es_lock */
1015 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for
1016 extents to shrink. Protected by
1017 i_es_lock */
1018
1019 /* ialloc */
1020 ext4_group_t i_last_alloc_group;
1021
1022 /* allocation reservation info for delalloc */
1023 /* In case of bigalloc, this refer to clusters rather than blocks */
1024 unsigned int i_reserved_data_blocks;
1025 ext4_lblk_t i_da_metadata_calc_last_lblock;
1026 int i_da_metadata_calc_len;
1027
1028 /* pending cluster reservations for bigalloc file systems */
1029 struct ext4_pending_tree i_pending_tree;
1030
1031 /* on-disk additional length */
1032 __u16 i_extra_isize;
1033
1034 /* Indicate the inline data space. */
1035 u16 i_inline_off;
1036 u16 i_inline_size;
1037
1038 #ifdef CONFIG_QUOTA
1039 /* quota space reservation, managed internally by quota code */
1040 qsize_t i_reserved_quota;
1041 #endif
1042
1043 /* Lock protecting lists below */
1044 spinlock_t i_completed_io_lock;
1045 /*
1046 * Completed IOs that need unwritten extents handling and have
1047 * transaction reserved
1048 */
1049 struct list_head i_rsv_conversion_list;
1050 struct work_struct i_rsv_conversion_work;
1051 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1052
1053 spinlock_t i_block_reservation_lock;
1054
1055 /*
1056 * Transactions that contain inode's metadata needed to complete
1057 * fsync and fdatasync, respectively.
1058 */
1059 tid_t i_sync_tid;
1060 tid_t i_datasync_tid;
1061
1062 #ifdef CONFIG_QUOTA
1063 struct dquot *i_dquot[MAXQUOTAS];
1064 #endif
1065
1066 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1067 __u32 i_csum_seed;
1068
1069 kprojid_t i_projid;
1070 };
1071
1072 /*
1073 * File system states
1074 */
1075 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
1076 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
1077 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
1078
1079 /*
1080 * Misc. filesystem flags
1081 */
1082 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
1083 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
1084 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
1085
1086 /*
1087 * Mount flags set via mount options or defaults
1088 */
1089 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */
1090 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
1091 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
1092 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
1093 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
1094 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
1095 #define EXT4_MOUNT_ERRORS_MASK 0x00070
1096 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
1097 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
1098 #ifdef CONFIG_FS_DAX
1099 #define EXT4_MOUNT_DAX 0x00200 /* Direct Access */
1100 #else
1101 #define EXT4_MOUNT_DAX 0
1102 #endif
1103 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
1104 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
1105 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
1106 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
1107 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
1108 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
1109 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
1110 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
1111 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
1112 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
1113 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */
1114 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota,
1115 * enable enforcement for hidden
1116 * quota files */
1117 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable
1118 * enforcement for hidden quota
1119 * files */
1120 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota
1121 * enforcement */
1122 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
1123 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
1124 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1125 #define EXT4_MOUNT_WARN_ON_ERROR 0x2000000 /* Trigger WARN_ON on error */
1126 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
1127 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
1128 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
1129 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
1130 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
1131
1132 /*
1133 * Mount flags set either automatically (could not be set by mount option)
1134 * based on per file system feature or property or in special cases such as
1135 * distinguishing between explicit mount option definition and default.
1136 */
1137 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1138 specified delalloc */
1139 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1140 size of blocksize * 8
1141 blocks */
1142 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1143 file systems */
1144
1145 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly
1146 specified journal checksum */
1147
1148 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1149 ~EXT4_MOUNT_##opt
1150 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1151 EXT4_MOUNT_##opt
1152 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1153 EXT4_MOUNT_##opt)
1154
1155 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1156 ~EXT4_MOUNT2_##opt
1157 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1158 EXT4_MOUNT2_##opt
1159 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1160 EXT4_MOUNT2_##opt)
1161
1162 #define ext4_test_and_set_bit __test_and_set_bit_le
1163 #define ext4_set_bit __set_bit_le
1164 #define ext4_set_bit_atomic ext2_set_bit_atomic
1165 #define ext4_test_and_clear_bit __test_and_clear_bit_le
1166 #define ext4_clear_bit __clear_bit_le
1167 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
1168 #define ext4_test_bit test_bit_le
1169 #define ext4_find_next_zero_bit find_next_zero_bit_le
1170 #define ext4_find_next_bit find_next_bit_le
1171
1172 extern void ext4_set_bits(void *bm, int cur, int len);
1173
1174 /*
1175 * Maximal mount counts between two filesystem checks
1176 */
1177 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1178 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1179
1180 /*
1181 * Behaviour when detecting errors
1182 */
1183 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1184 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1185 #define EXT4_ERRORS_PANIC 3 /* Panic */
1186 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1187
1188 /* Metadata checksum algorithm codes */
1189 #define EXT4_CRC32C_CHKSUM 1
1190
1191 /*
1192 * Structure of the super block
1193 */
1194 struct ext4_super_block {
1195 /*00*/ __le32 s_inodes_count; /* Inodes count */
1196 __le32 s_blocks_count_lo; /* Blocks count */
1197 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1198 __le32 s_free_blocks_count_lo; /* Free blocks count */
1199 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1200 __le32 s_first_data_block; /* First Data Block */
1201 __le32 s_log_block_size; /* Block size */
1202 __le32 s_log_cluster_size; /* Allocation cluster size */
1203 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1204 __le32 s_clusters_per_group; /* # Clusters per group */
1205 __le32 s_inodes_per_group; /* # Inodes per group */
1206 __le32 s_mtime; /* Mount time */
1207 /*30*/ __le32 s_wtime; /* Write time */
1208 __le16 s_mnt_count; /* Mount count */
1209 __le16 s_max_mnt_count; /* Maximal mount count */
1210 __le16 s_magic; /* Magic signature */
1211 __le16 s_state; /* File system state */
1212 __le16 s_errors; /* Behaviour when detecting errors */
1213 __le16 s_minor_rev_level; /* minor revision level */
1214 /*40*/ __le32 s_lastcheck; /* time of last check */
1215 __le32 s_checkinterval; /* max. time between checks */
1216 __le32 s_creator_os; /* OS */
1217 __le32 s_rev_level; /* Revision level */
1218 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1219 __le16 s_def_resgid; /* Default gid for reserved blocks */
1220 /*
1221 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1222 *
1223 * Note: the difference between the compatible feature set and
1224 * the incompatible feature set is that if there is a bit set
1225 * in the incompatible feature set that the kernel doesn't
1226 * know about, it should refuse to mount the filesystem.
1227 *
1228 * e2fsck's requirements are more strict; if it doesn't know
1229 * about a feature in either the compatible or incompatible
1230 * feature set, it must abort and not try to meddle with
1231 * things it doesn't understand...
1232 */
1233 __le32 s_first_ino; /* First non-reserved inode */
1234 __le16 s_inode_size; /* size of inode structure */
1235 __le16 s_block_group_nr; /* block group # of this superblock */
1236 __le32 s_feature_compat; /* compatible feature set */
1237 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1238 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1239 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1240 /*78*/ char s_volume_name[16]; /* volume name */
1241 /*88*/ char s_last_mounted[64] __nonstring; /* directory where last mounted */
1242 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1243 /*
1244 * Performance hints. Directory preallocation should only
1245 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1246 */
1247 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1248 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1249 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1250 /*
1251 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1252 */
1253 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1254 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1255 __le32 s_journal_dev; /* device number of journal file */
1256 __le32 s_last_orphan; /* start of list of inodes to delete */
1257 __le32 s_hash_seed[4]; /* HTREE hash seed */
1258 __u8 s_def_hash_version; /* Default hash version to use */
1259 __u8 s_jnl_backup_type;
1260 __le16 s_desc_size; /* size of group descriptor */
1261 /*100*/ __le32 s_default_mount_opts;
1262 __le32 s_first_meta_bg; /* First metablock block group */
1263 __le32 s_mkfs_time; /* When the filesystem was created */
1264 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1265 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1266 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1267 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1268 __le32 s_free_blocks_count_hi; /* Free blocks count */
1269 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1270 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1271 __le32 s_flags; /* Miscellaneous flags */
1272 __le16 s_raid_stride; /* RAID stride */
1273 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1274 __le64 s_mmp_block; /* Block for multi-mount protection */
1275 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1276 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1277 __u8 s_checksum_type; /* metadata checksum algorithm used */
1278 __u8 s_encryption_level; /* versioning level for encryption */
1279 __u8 s_reserved_pad; /* Padding to next 32bits */
1280 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1281 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1282 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1283 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1284 snapshot's future use */
1285 __le32 s_snapshot_list; /* inode number of the head of the
1286 on-disk snapshot list */
1287 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1288 __le32 s_error_count; /* number of fs errors */
1289 __le32 s_first_error_time; /* first time an error happened */
1290 __le32 s_first_error_ino; /* inode involved in first error */
1291 __le64 s_first_error_block; /* block involved of first error */
1292 __u8 s_first_error_func[32] __nonstring; /* function where the error happened */
1293 __le32 s_first_error_line; /* line number where error happened */
1294 __le32 s_last_error_time; /* most recent time of an error */
1295 __le32 s_last_error_ino; /* inode involved in last error */
1296 __le32 s_last_error_line; /* line number where error happened */
1297 __le64 s_last_error_block; /* block involved of last error */
1298 __u8 s_last_error_func[32] __nonstring; /* function where the error happened */
1299 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1300 __u8 s_mount_opts[64];
1301 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1302 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1303 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1304 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
1305 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */
1306 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */
1307 __le32 s_lpf_ino; /* Location of the lost+found inode */
1308 __le32 s_prj_quota_inum; /* inode for tracking project quota */
1309 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */
1310 __u8 s_wtime_hi;
1311 __u8 s_mtime_hi;
1312 __u8 s_mkfs_time_hi;
1313 __u8 s_lastcheck_hi;
1314 __u8 s_first_error_time_hi;
1315 __u8 s_last_error_time_hi;
1316 __u8 s_pad[2];
1317 __le16 s_encoding; /* Filename charset encoding */
1318 __le16 s_encoding_flags; /* Filename charset encoding flags */
1319 __le32 s_reserved[95]; /* 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_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 #define EXT4_ENC_UTF8_12_1 1
1345
1346 /*
1347 * Flags for ext4_sb_info.s_encoding_flags.
1348 */
1349 #define EXT4_ENC_STRICT_MODE_FL (1 << 0)
1350
1351 #define ext4_has_strict_mode(sbi) \
1352 (sbi->s_encoding_flags & EXT4_ENC_STRICT_MODE_FL)
1353
1354 /*
1355 * fourth extended-fs super-block data in memory
1356 */
1357 struct ext4_sb_info {
1358 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1359 unsigned long s_inodes_per_block;/* Number of inodes per block */
1360 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1361 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1362 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1363 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1364 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1365 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1366 ext4_group_t s_groups_count; /* Number of groups in the fs */
1367 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1368 unsigned long s_overhead; /* # of fs overhead clusters */
1369 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1370 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1371 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1372 struct buffer_head * s_sbh; /* Buffer containing the super block */
1373 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1374 struct buffer_head **s_group_desc;
1375 unsigned int s_mount_opt;
1376 unsigned int s_mount_opt2;
1377 unsigned int s_mount_flags;
1378 unsigned int s_def_mount_opt;
1379 ext4_fsblk_t s_sb_block;
1380 atomic64_t s_resv_clusters;
1381 kuid_t s_resuid;
1382 kgid_t s_resgid;
1383 unsigned short s_mount_state;
1384 unsigned short s_pad;
1385 int s_addr_per_block_bits;
1386 int s_desc_per_block_bits;
1387 int s_inode_size;
1388 int s_first_ino;
1389 unsigned int s_inode_readahead_blks;
1390 unsigned int s_inode_goal;
1391 u32 s_hash_seed[4];
1392 int s_def_hash_version;
1393 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1394 struct percpu_counter s_freeclusters_counter;
1395 struct percpu_counter s_freeinodes_counter;
1396 struct percpu_counter s_dirs_counter;
1397 struct percpu_counter s_dirtyclusters_counter;
1398 struct blockgroup_lock *s_blockgroup_lock;
1399 struct proc_dir_entry *s_proc;
1400 struct kobject s_kobj;
1401 struct completion s_kobj_unregister;
1402 struct super_block *s_sb;
1403 #ifdef CONFIG_UNICODE
1404 struct unicode_map *s_encoding;
1405 __u16 s_encoding_flags;
1406 #endif
1407
1408 /* Journaling */
1409 struct journal_s *s_journal;
1410 struct list_head s_orphan;
1411 struct mutex s_orphan_lock;
1412 unsigned long s_ext4_flags; /* Ext4 superblock flags */
1413 unsigned long s_commit_interval;
1414 u32 s_max_batch_time;
1415 u32 s_min_batch_time;
1416 struct block_device *journal_bdev;
1417 #ifdef CONFIG_QUOTA
1418 /* Names of quota files with journalled quota */
1419 char __rcu *s_qf_names[EXT4_MAXQUOTAS];
1420 int s_jquota_fmt; /* Format of quota to use */
1421 #endif
1422 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1423 struct rb_root system_blks;
1424
1425 #ifdef EXTENTS_STATS
1426 /* ext4 extents stats */
1427 unsigned long s_ext_min;
1428 unsigned long s_ext_max;
1429 unsigned long s_depth_max;
1430 spinlock_t s_ext_stats_lock;
1431 unsigned long s_ext_blocks;
1432 unsigned long s_ext_extents;
1433 #endif
1434
1435 /* for buddy allocator */
1436 struct ext4_group_info ***s_group_info;
1437 struct inode *s_buddy_cache;
1438 spinlock_t s_md_lock;
1439 unsigned short *s_mb_offsets;
1440 unsigned int *s_mb_maxs;
1441 unsigned int s_group_info_size;
1442 unsigned int s_mb_free_pending;
1443 struct list_head s_freed_data_list; /* List of blocks to be freed
1444 after commit completed */
1445
1446 /* tunables */
1447 unsigned long s_stripe;
1448 unsigned int s_mb_stream_request;
1449 unsigned int s_mb_max_to_scan;
1450 unsigned int s_mb_min_to_scan;
1451 unsigned int s_mb_stats;
1452 unsigned int s_mb_order2_reqs;
1453 unsigned int s_mb_group_prealloc;
1454 unsigned int s_max_dir_size_kb;
1455 /* where last allocation was done - for stream allocation */
1456 unsigned long s_mb_last_group;
1457 unsigned long s_mb_last_start;
1458
1459 /* stats for buddy allocator */
1460 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1461 atomic_t s_bal_success; /* we found long enough chunks */
1462 atomic_t s_bal_allocated; /* in blocks */
1463 atomic_t s_bal_ex_scanned; /* total extents scanned */
1464 atomic_t s_bal_goals; /* goal hits */
1465 atomic_t s_bal_breaks; /* too long searches */
1466 atomic_t s_bal_2orders; /* 2^order hits */
1467 spinlock_t s_bal_lock;
1468 unsigned long s_mb_buddies_generated;
1469 unsigned long long s_mb_generation_time;
1470 atomic_t s_mb_lost_chunks;
1471 atomic_t s_mb_preallocated;
1472 atomic_t s_mb_discarded;
1473 atomic_t s_lock_busy;
1474
1475 /* locality groups */
1476 struct ext4_locality_group __percpu *s_locality_groups;
1477
1478 /* for write statistics */
1479 unsigned long s_sectors_written_start;
1480 u64 s_kbytes_written;
1481
1482 /* the size of zero-out chunk */
1483 unsigned int s_extent_max_zeroout_kb;
1484
1485 unsigned int s_log_groups_per_flex;
1486 struct flex_groups *s_flex_groups;
1487 ext4_group_t s_flex_groups_allocated;
1488
1489 /* workqueue for reserved extent conversions (buffered io) */
1490 struct workqueue_struct *rsv_conversion_wq;
1491
1492 /* timer for periodic error stats printing */
1493 struct timer_list s_err_report;
1494
1495 /* Lazy inode table initialization info */
1496 struct ext4_li_request *s_li_request;
1497 /* Wait multiplier for lazy initialization thread */
1498 unsigned int s_li_wait_mult;
1499
1500 /* Kernel thread for multiple mount protection */
1501 struct task_struct *s_mmp_tsk;
1502
1503 /* record the last minlen when FITRIM is called. */
1504 atomic_t s_last_trim_minblks;
1505
1506 /* Reference to checksum algorithm driver via cryptoapi */
1507 struct crypto_shash *s_chksum_driver;
1508
1509 /* Precomputed FS UUID checksum for seeding other checksums */
1510 __u32 s_csum_seed;
1511
1512 /* Reclaim extents from extent status tree */
1513 struct shrinker s_es_shrinker;
1514 struct list_head s_es_list; /* List of inodes with reclaimable extents */
1515 long s_es_nr_inode;
1516 struct ext4_es_stats s_es_stats;
1517 struct mb_cache *s_ea_block_cache;
1518 struct mb_cache *s_ea_inode_cache;
1519 spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1520
1521 /* Ratelimit ext4 messages. */
1522 struct ratelimit_state s_err_ratelimit_state;
1523 struct ratelimit_state s_warning_ratelimit_state;
1524 struct ratelimit_state s_msg_ratelimit_state;
1525
1526 /* Barrier between changing inodes' journal flags and writepages ops. */
1527 struct percpu_rw_semaphore s_journal_flag_rwsem;
1528 struct dax_device *s_daxdev;
1529 };
1530
1531 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1532 {
1533 return sb->s_fs_info;
1534 }
1535 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1536 {
1537 return container_of(inode, struct ext4_inode_info, vfs_inode);
1538 }
1539
1540 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1541 {
1542 return ino == EXT4_ROOT_INO ||
1543 (ino >= EXT4_FIRST_INO(sb) &&
1544 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1545 }
1546
1547 /*
1548 * Inode dynamic state flags
1549 */
1550 enum {
1551 EXT4_STATE_JDATA, /* journaled data exists */
1552 EXT4_STATE_NEW, /* inode is newly created */
1553 EXT4_STATE_XATTR, /* has in-inode xattrs */
1554 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1555 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1556 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1557 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
1558 EXT4_STATE_NEWENTRY, /* File just added to dir */
1559 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1560 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
1561 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */
1562 };
1563
1564 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1565 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1566 { \
1567 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1568 } \
1569 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1570 { \
1571 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1572 } \
1573 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1574 { \
1575 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1576 }
1577
1578 /* Add these declarations here only so that these functions can be
1579 * found by name. Otherwise, they are very hard to locate. */
1580 static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1581 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1582 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1583 EXT4_INODE_BIT_FNS(flag, flags, 0)
1584
1585 /* Add these declarations here only so that these functions can be
1586 * found by name. Otherwise, they are very hard to locate. */
1587 static inline int ext4_test_inode_state(struct inode *inode, int bit);
1588 static inline void ext4_set_inode_state(struct inode *inode, int bit);
1589 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1590 #if (BITS_PER_LONG < 64)
1591 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1592
1593 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1594 {
1595 (ei)->i_state_flags = 0;
1596 }
1597 #else
1598 EXT4_INODE_BIT_FNS(state, flags, 32)
1599
1600 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1601 {
1602 /* We depend on the fact that callers will set i_flags */
1603 }
1604 #endif
1605 #else
1606 /* Assume that user mode programs are passing in an ext4fs superblock, not
1607 * a kernel struct super_block. This will allow us to call the feature-test
1608 * macros from user land. */
1609 #define EXT4_SB(sb) (sb)
1610 #endif
1611
1612 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1613
1614 /*
1615 * Codes for operating systems
1616 */
1617 #define EXT4_OS_LINUX 0
1618 #define EXT4_OS_HURD 1
1619 #define EXT4_OS_MASIX 2
1620 #define EXT4_OS_FREEBSD 3
1621 #define EXT4_OS_LITES 4
1622
1623 /*
1624 * Revision levels
1625 */
1626 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1627 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1628
1629 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1630 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1631
1632 #define EXT4_GOOD_OLD_INODE_SIZE 128
1633
1634 /*
1635 * Feature set definitions
1636 */
1637
1638 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1639 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1640 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1641 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1642 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1643 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1644 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
1645
1646 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1647 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1648 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1649 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1650 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1651 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1652 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1653 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1654 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1655 /*
1656 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1657 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1658 * all other data structures' checksums. However, the METADATA_CSUM and
1659 * GDT_CSUM bits are mutually exclusive.
1660 */
1661 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
1662 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000
1663 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000
1664
1665 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1666 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1667 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1668 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1669 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1670 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1671 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1672 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1673 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1674 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1675 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1676 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000
1677 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
1678 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
1679 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000
1680 #define EXT4_FEATURE_INCOMPAT_CASEFOLD 0x20000
1681
1682 extern void ext4_update_dynamic_rev(struct super_block *sb);
1683
1684 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
1685 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1686 { \
1687 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1688 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
1689 } \
1690 static inline void ext4_set_feature_##name(struct super_block *sb) \
1691 { \
1692 ext4_update_dynamic_rev(sb); \
1693 EXT4_SB(sb)->s_es->s_feature_compat |= \
1694 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1695 } \
1696 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1697 { \
1698 EXT4_SB(sb)->s_es->s_feature_compat &= \
1699 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1700 }
1701
1702 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
1703 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1704 { \
1705 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1706 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
1707 } \
1708 static inline void ext4_set_feature_##name(struct super_block *sb) \
1709 { \
1710 ext4_update_dynamic_rev(sb); \
1711 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
1712 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1713 } \
1714 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1715 { \
1716 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
1717 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1718 }
1719
1720 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
1721 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1722 { \
1723 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1724 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
1725 } \
1726 static inline void ext4_set_feature_##name(struct super_block *sb) \
1727 { \
1728 ext4_update_dynamic_rev(sb); \
1729 EXT4_SB(sb)->s_es->s_feature_incompat |= \
1730 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1731 } \
1732 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1733 { \
1734 EXT4_SB(sb)->s_es->s_feature_incompat &= \
1735 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1736 }
1737
1738 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC)
1739 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES)
1740 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL)
1741 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR)
1742 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE)
1743 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX)
1744 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2)
1745
1746 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER)
1747 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE)
1748 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR)
1749 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE)
1750 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM)
1751 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK)
1752 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE)
1753 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA)
1754 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC)
1755 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM)
1756 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY)
1757 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT)
1758
1759 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION)
1760 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE)
1761 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER)
1762 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV)
1763 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG)
1764 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS)
1765 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT)
1766 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP)
1767 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG)
1768 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE)
1769 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA)
1770 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED)
1771 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR)
1772 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA)
1773 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT)
1774 EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD)
1775
1776 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1777 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1778 EXT4_FEATURE_INCOMPAT_META_BG)
1779 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1780 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1781 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1782
1783 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1784 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1785 EXT4_FEATURE_INCOMPAT_RECOVER| \
1786 EXT4_FEATURE_INCOMPAT_META_BG)
1787 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1788 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1789 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1790
1791 #define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1792 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1793 EXT4_FEATURE_INCOMPAT_RECOVER| \
1794 EXT4_FEATURE_INCOMPAT_META_BG| \
1795 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1796 EXT4_FEATURE_INCOMPAT_64BIT| \
1797 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1798 EXT4_FEATURE_INCOMPAT_EA_INODE| \
1799 EXT4_FEATURE_INCOMPAT_MMP | \
1800 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
1801 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
1802 EXT4_FEATURE_INCOMPAT_CASEFOLD | \
1803 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \
1804 EXT4_FEATURE_INCOMPAT_LARGEDIR)
1805 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1806 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1807 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1808 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1809 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1810 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1811 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1812 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1813 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1814 EXT4_FEATURE_RO_COMPAT_QUOTA |\
1815 EXT4_FEATURE_RO_COMPAT_PROJECT)
1816
1817 #define EXTN_FEATURE_FUNCS(ver) \
1818 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
1819 { \
1820 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1821 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
1822 } \
1823 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
1824 { \
1825 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1826 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
1827 } \
1828 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
1829 { \
1830 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1831 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
1832 }
1833
1834 EXTN_FEATURE_FUNCS(2)
1835 EXTN_FEATURE_FUNCS(3)
1836 EXTN_FEATURE_FUNCS(4)
1837
1838 static inline bool ext4_has_compat_features(struct super_block *sb)
1839 {
1840 return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
1841 }
1842 static inline bool ext4_has_ro_compat_features(struct super_block *sb)
1843 {
1844 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
1845 }
1846 static inline bool ext4_has_incompat_features(struct super_block *sb)
1847 {
1848 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
1849 }
1850
1851 /*
1852 * Superblock flags
1853 */
1854 #define EXT4_FLAGS_RESIZING 0
1855 #define EXT4_FLAGS_SHUTDOWN 1
1856
1857 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi)
1858 {
1859 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags);
1860 }
1861
1862
1863 /*
1864 * Default values for user and/or group using reserved blocks
1865 */
1866 #define EXT4_DEF_RESUID 0
1867 #define EXT4_DEF_RESGID 0
1868
1869 /*
1870 * Default project ID
1871 */
1872 #define EXT4_DEF_PROJID 0
1873
1874 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
1875
1876 /*
1877 * Default mount options
1878 */
1879 #define EXT4_DEFM_DEBUG 0x0001
1880 #define EXT4_DEFM_BSDGROUPS 0x0002
1881 #define EXT4_DEFM_XATTR_USER 0x0004
1882 #define EXT4_DEFM_ACL 0x0008
1883 #define EXT4_DEFM_UID16 0x0010
1884 #define EXT4_DEFM_JMODE 0x0060
1885 #define EXT4_DEFM_JMODE_DATA 0x0020
1886 #define EXT4_DEFM_JMODE_ORDERED 0x0040
1887 #define EXT4_DEFM_JMODE_WBACK 0x0060
1888 #define EXT4_DEFM_NOBARRIER 0x0100
1889 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1890 #define EXT4_DEFM_DISCARD 0x0400
1891 #define EXT4_DEFM_NODELALLOC 0x0800
1892
1893 /*
1894 * Default journal batch times
1895 */
1896 #define EXT4_DEF_MIN_BATCH_TIME 0
1897 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1898
1899 /*
1900 * Minimum number of groups in a flexgroup before we separate out
1901 * directories into the first block group of a flexgroup
1902 */
1903 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1904
1905 /*
1906 * Structure of a directory entry
1907 */
1908 #define EXT4_NAME_LEN 255
1909
1910 struct ext4_dir_entry {
1911 __le32 inode; /* Inode number */
1912 __le16 rec_len; /* Directory entry length */
1913 __le16 name_len; /* Name length */
1914 char name[EXT4_NAME_LEN]; /* File name */
1915 };
1916
1917 /*
1918 * The new version of the directory entry. Since EXT4 structures are
1919 * stored in intel byte order, and the name_len field could never be
1920 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1921 * file_type field.
1922 */
1923 struct ext4_dir_entry_2 {
1924 __le32 inode; /* Inode number */
1925 __le16 rec_len; /* Directory entry length */
1926 __u8 name_len; /* Name length */
1927 __u8 file_type;
1928 char name[EXT4_NAME_LEN]; /* File name */
1929 };
1930
1931 /*
1932 * This is a bogus directory entry at the end of each leaf block that
1933 * records checksums.
1934 */
1935 struct ext4_dir_entry_tail {
1936 __le32 det_reserved_zero1; /* Pretend to be unused */
1937 __le16 det_rec_len; /* 12 */
1938 __u8 det_reserved_zero2; /* Zero name length */
1939 __u8 det_reserved_ft; /* 0xDE, fake file type */
1940 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
1941 };
1942
1943 #define EXT4_DIRENT_TAIL(block, blocksize) \
1944 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1945 ((blocksize) - \
1946 sizeof(struct ext4_dir_entry_tail))))
1947
1948 /*
1949 * Ext4 directory file types. Only the low 3 bits are used. The
1950 * other bits are reserved for now.
1951 */
1952 #define EXT4_FT_UNKNOWN 0
1953 #define EXT4_FT_REG_FILE 1
1954 #define EXT4_FT_DIR 2
1955 #define EXT4_FT_CHRDEV 3
1956 #define EXT4_FT_BLKDEV 4
1957 #define EXT4_FT_FIFO 5
1958 #define EXT4_FT_SOCK 6
1959 #define EXT4_FT_SYMLINK 7
1960
1961 #define EXT4_FT_MAX 8
1962
1963 #define EXT4_FT_DIR_CSUM 0xDE
1964
1965 /*
1966 * EXT4_DIR_PAD defines the directory entries boundaries
1967 *
1968 * NOTE: It must be a multiple of 4
1969 */
1970 #define EXT4_DIR_PAD 4
1971 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1972 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1973 ~EXT4_DIR_ROUND)
1974 #define EXT4_MAX_REC_LEN ((1<<16)-1)
1975
1976 /*
1977 * If we ever get support for fs block sizes > page_size, we'll need
1978 * to remove the #if statements in the next two functions...
1979 */
1980 static inline unsigned int
1981 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1982 {
1983 unsigned len = le16_to_cpu(dlen);
1984
1985 #if (PAGE_SIZE >= 65536)
1986 if (len == EXT4_MAX_REC_LEN || len == 0)
1987 return blocksize;
1988 return (len & 65532) | ((len & 3) << 16);
1989 #else
1990 return len;
1991 #endif
1992 }
1993
1994 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1995 {
1996 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1997 BUG();
1998 #if (PAGE_SIZE >= 65536)
1999 if (len < 65536)
2000 return cpu_to_le16(len);
2001 if (len == blocksize) {
2002 if (blocksize == 65536)
2003 return cpu_to_le16(EXT4_MAX_REC_LEN);
2004 else
2005 return cpu_to_le16(0);
2006 }
2007 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
2008 #else
2009 return cpu_to_le16(len);
2010 #endif
2011 }
2012
2013 /*
2014 * Hash Tree Directory indexing
2015 * (c) Daniel Phillips, 2001
2016 */
2017
2018 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
2019 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2020 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \
2021 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir)))
2022 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2023
2024 /* Legal values for the dx_root hash_version field: */
2025
2026 #define DX_HASH_LEGACY 0
2027 #define DX_HASH_HALF_MD4 1
2028 #define DX_HASH_TEA 2
2029 #define DX_HASH_LEGACY_UNSIGNED 3
2030 #define DX_HASH_HALF_MD4_UNSIGNED 4
2031 #define DX_HASH_TEA_UNSIGNED 5
2032
2033 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2034 const void *address, unsigned int length)
2035 {
2036 struct {
2037 struct shash_desc shash;
2038 char ctx[4];
2039 } desc;
2040
2041 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2042
2043 desc.shash.tfm = sbi->s_chksum_driver;
2044 *(u32 *)desc.ctx = crc;
2045
2046 BUG_ON(crypto_shash_update(&desc.shash, address, length));
2047
2048 return *(u32 *)desc.ctx;
2049 }
2050
2051 #ifdef __KERNEL__
2052
2053 /* hash info structure used by the directory hash */
2054 struct dx_hash_info
2055 {
2056 u32 hash;
2057 u32 minor_hash;
2058 int hash_version;
2059 u32 *seed;
2060 };
2061
2062
2063 /* 32 and 64 bit signed EOF for dx directories */
2064 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
2065 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
2066
2067
2068 /*
2069 * Control parameters used by ext4_htree_next_block
2070 */
2071 #define HASH_NB_ALWAYS 1
2072
2073 struct ext4_filename {
2074 const struct qstr *usr_fname;
2075 struct fscrypt_str disk_name;
2076 struct dx_hash_info hinfo;
2077 #ifdef CONFIG_FS_ENCRYPTION
2078 struct fscrypt_str crypto_buf;
2079 #endif
2080 };
2081
2082 #define fname_name(p) ((p)->disk_name.name)
2083 #define fname_len(p) ((p)->disk_name.len)
2084
2085 /*
2086 * Describe an inode's exact location on disk and in memory
2087 */
2088 struct ext4_iloc
2089 {
2090 struct buffer_head *bh;
2091 unsigned long offset;
2092 ext4_group_t block_group;
2093 };
2094
2095 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2096 {
2097 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2098 }
2099
2100 static inline bool ext4_is_quota_file(struct inode *inode)
2101 {
2102 return IS_NOQUOTA(inode) &&
2103 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL);
2104 }
2105
2106 /*
2107 * This structure is stuffed into the struct file's private_data field
2108 * for directories. It is where we put information so that we can do
2109 * readdir operations in hash tree order.
2110 */
2111 struct dir_private_info {
2112 struct rb_root root;
2113 struct rb_node *curr_node;
2114 struct fname *extra_fname;
2115 loff_t last_pos;
2116 __u32 curr_hash;
2117 __u32 curr_minor_hash;
2118 __u32 next_hash;
2119 };
2120
2121 /* calculate the first block number of the group */
2122 static inline ext4_fsblk_t
2123 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2124 {
2125 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2126 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2127 }
2128
2129 /*
2130 * Special error return code only used by dx_probe() and its callers.
2131 */
2132 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
2133
2134 /* htree levels for ext4 */
2135 #define EXT4_HTREE_LEVEL_COMPAT 2
2136 #define EXT4_HTREE_LEVEL 3
2137
2138 static inline int ext4_dir_htree_level(struct super_block *sb)
2139 {
2140 return ext4_has_feature_largedir(sb) ?
2141 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT;
2142 }
2143
2144 /*
2145 * Timeout and state flag for lazy initialization inode thread.
2146 */
2147 #define EXT4_DEF_LI_WAIT_MULT 10
2148 #define EXT4_DEF_LI_MAX_START_DELAY 5
2149 #define EXT4_LAZYINIT_QUIT 0x0001
2150 #define EXT4_LAZYINIT_RUNNING 0x0002
2151
2152 /*
2153 * Lazy inode table initialization info
2154 */
2155 struct ext4_lazy_init {
2156 unsigned long li_state;
2157 struct list_head li_request_list;
2158 struct mutex li_list_mtx;
2159 };
2160
2161 struct ext4_li_request {
2162 struct super_block *lr_super;
2163 struct ext4_sb_info *lr_sbi;
2164 ext4_group_t lr_next_group;
2165 struct list_head lr_request;
2166 unsigned long lr_next_sched;
2167 unsigned long lr_timeout;
2168 };
2169
2170 struct ext4_features {
2171 struct kobject f_kobj;
2172 struct completion f_kobj_unregister;
2173 };
2174
2175 /*
2176 * This structure will be used for multiple mount protection. It will be
2177 * written into the block number saved in the s_mmp_block field in the
2178 * superblock. Programs that check MMP should assume that if
2179 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2180 * to use the filesystem, regardless of how old the timestamp is.
2181 */
2182 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
2183 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2184 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
2185 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
2186
2187 struct mmp_struct {
2188 __le32 mmp_magic; /* Magic number for MMP */
2189 __le32 mmp_seq; /* Sequence no. updated periodically */
2190
2191 /*
2192 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2193 * purposes and do not affect the correctness of the algorithm
2194 */
2195 __le64 mmp_time; /* Time last updated */
2196 char mmp_nodename[64]; /* Node which last updated MMP block */
2197 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
2198
2199 /*
2200 * mmp_check_interval is used to verify if the MMP block has been
2201 * updated on the block device. The value is updated based on the
2202 * maximum time to write the MMP block during an update cycle.
2203 */
2204 __le16 mmp_check_interval;
2205
2206 __le16 mmp_pad1;
2207 __le32 mmp_pad2[226];
2208 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
2209 };
2210
2211 /* arguments passed to the mmp thread */
2212 struct mmpd_data {
2213 struct buffer_head *bh; /* bh from initial read_mmp_block() */
2214 struct super_block *sb; /* super block of the fs */
2215 };
2216
2217 /*
2218 * Check interval multiplier
2219 * The MMP block is written every update interval and initially checked every
2220 * update interval x the multiplier (the value is then adapted based on the
2221 * write latency). The reason is that writes can be delayed under load and we
2222 * don't want readers to incorrectly assume that the filesystem is no longer
2223 * in use.
2224 */
2225 #define EXT4_MMP_CHECK_MULT 2UL
2226
2227 /*
2228 * Minimum interval for MMP checking in seconds.
2229 */
2230 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
2231
2232 /*
2233 * Maximum interval for MMP checking in seconds.
2234 */
2235 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
2236
2237 /*
2238 * Function prototypes
2239 */
2240
2241 /*
2242 * Ok, these declarations are also in <linux/kernel.h> but none of the
2243 * ext4 source programs needs to include it so they are duplicated here.
2244 */
2245 # define NORET_TYPE /**/
2246 # define ATTRIB_NORET __attribute__((noreturn))
2247 # define NORET_AND noreturn,
2248
2249 /* bitmap.c */
2250 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2251 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2252 struct ext4_group_desc *gdp,
2253 struct buffer_head *bh, int sz);
2254 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2255 struct ext4_group_desc *gdp,
2256 struct buffer_head *bh, int sz);
2257 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2258 struct ext4_group_desc *gdp,
2259 struct buffer_head *bh);
2260 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2261 struct ext4_group_desc *gdp,
2262 struct buffer_head *bh);
2263
2264 /* balloc.c */
2265 extern void ext4_get_group_no_and_offset(struct super_block *sb,
2266 ext4_fsblk_t blocknr,
2267 ext4_group_t *blockgrpp,
2268 ext4_grpblk_t *offsetp);
2269 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2270 ext4_fsblk_t block);
2271
2272 extern unsigned int ext4_block_group(struct super_block *sb,
2273 ext4_fsblk_t blocknr);
2274 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2275 ext4_fsblk_t blocknr);
2276 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2277 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2278 ext4_group_t group);
2279 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2280 ext4_fsblk_t goal,
2281 unsigned int flags,
2282 unsigned long *count,
2283 int *errp);
2284 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2285 s64 nclusters, unsigned int flags);
2286 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2287 extern void ext4_check_blocks_bitmap(struct super_block *);
2288 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2289 ext4_group_t block_group,
2290 struct buffer_head ** bh);
2291 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2292
2293 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2294 ext4_group_t block_group);
2295 extern int ext4_wait_block_bitmap(struct super_block *sb,
2296 ext4_group_t block_group,
2297 struct buffer_head *bh);
2298 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2299 ext4_group_t block_group);
2300 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2301 ext4_group_t block_group,
2302 struct ext4_group_desc *gdp);
2303 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2304
2305 #ifdef CONFIG_FS_ENCRYPTION
2306 static inline void ext4_fname_from_fscrypt_name(struct ext4_filename *dst,
2307 const struct fscrypt_name *src)
2308 {
2309 memset(dst, 0, sizeof(*dst));
2310
2311 dst->usr_fname = src->usr_fname;
2312 dst->disk_name = src->disk_name;
2313 dst->hinfo.hash = src->hash;
2314 dst->hinfo.minor_hash = src->minor_hash;
2315 dst->crypto_buf = src->crypto_buf;
2316 }
2317
2318 static inline int ext4_fname_setup_filename(struct inode *dir,
2319 const struct qstr *iname,
2320 int lookup,
2321 struct ext4_filename *fname)
2322 {
2323 struct fscrypt_name name;
2324 int err;
2325
2326 err = fscrypt_setup_filename(dir, iname, lookup, &name);
2327 if (err)
2328 return err;
2329
2330 ext4_fname_from_fscrypt_name(fname, &name);
2331 return 0;
2332 }
2333
2334 static inline int ext4_fname_prepare_lookup(struct inode *dir,
2335 struct dentry *dentry,
2336 struct ext4_filename *fname)
2337 {
2338 struct fscrypt_name name;
2339 int err;
2340
2341 err = fscrypt_prepare_lookup(dir, dentry, &name);
2342 if (err)
2343 return err;
2344
2345 ext4_fname_from_fscrypt_name(fname, &name);
2346 return 0;
2347 }
2348
2349 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2350 {
2351 struct fscrypt_name name;
2352
2353 name.crypto_buf = fname->crypto_buf;
2354 fscrypt_free_filename(&name);
2355
2356 fname->crypto_buf.name = NULL;
2357 fname->usr_fname = NULL;
2358 fname->disk_name.name = NULL;
2359 }
2360 #else /* !CONFIG_FS_ENCRYPTION */
2361 static inline int ext4_fname_setup_filename(struct inode *dir,
2362 const struct qstr *iname,
2363 int lookup,
2364 struct ext4_filename *fname)
2365 {
2366 fname->usr_fname = iname;
2367 fname->disk_name.name = (unsigned char *) iname->name;
2368 fname->disk_name.len = iname->len;
2369 return 0;
2370 }
2371
2372 static inline int ext4_fname_prepare_lookup(struct inode *dir,
2373 struct dentry *dentry,
2374 struct ext4_filename *fname)
2375 {
2376 return ext4_fname_setup_filename(dir, &dentry->d_name, 1, fname);
2377 }
2378
2379 static inline void ext4_fname_free_filename(struct ext4_filename *fname) { }
2380 #endif /* !CONFIG_FS_ENCRYPTION */
2381
2382 /* dir.c */
2383 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2384 struct file *,
2385 struct ext4_dir_entry_2 *,
2386 struct buffer_head *, char *, int,
2387 unsigned int);
2388 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2389 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2390 (de), (bh), (buf), (size), (offset)))
2391 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2392 __u32 minor_hash,
2393 struct ext4_dir_entry_2 *dirent,
2394 struct fscrypt_str *ent_name);
2395 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2396 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2397 struct buffer_head *bh,
2398 void *buf, int buf_size,
2399 struct ext4_filename *fname,
2400 struct ext4_dir_entry_2 **dest_de);
2401 void ext4_insert_dentry(struct inode *inode,
2402 struct ext4_dir_entry_2 *de,
2403 int buf_size,
2404 struct ext4_filename *fname);
2405 static inline void ext4_update_dx_flag(struct inode *inode)
2406 {
2407 if (!ext4_has_feature_dir_index(inode->i_sb))
2408 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2409 }
2410 static const unsigned char ext4_filetype_table[] = {
2411 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2412 };
2413
2414 static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2415 {
2416 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2417 return DT_UNKNOWN;
2418
2419 return ext4_filetype_table[filetype];
2420 }
2421 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2422 void *buf, int buf_size);
2423
2424 /* fsync.c */
2425 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2426
2427 /* hash.c */
2428 extern int ext4fs_dirhash(const struct inode *dir, const char *name, int len,
2429 struct dx_hash_info *hinfo);
2430
2431 /* ialloc.c */
2432 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2433 const struct qstr *qstr, __u32 goal,
2434 uid_t *owner, __u32 i_flags,
2435 int handle_type, unsigned int line_no,
2436 int nblocks);
2437
2438 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \
2439 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2440 i_flags, 0, 0, 0)
2441 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2442 type, nblocks) \
2443 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2444 0, (type), __LINE__, (nblocks))
2445
2446
2447 extern void ext4_free_inode(handle_t *, struct inode *);
2448 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2449 extern unsigned long ext4_count_free_inodes(struct super_block *);
2450 extern unsigned long ext4_count_dirs(struct super_block *);
2451 extern void ext4_check_inodes_bitmap(struct super_block *);
2452 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2453 extern int ext4_init_inode_table(struct super_block *sb,
2454 ext4_group_t group, int barrier);
2455 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2456
2457 /* mballoc.c */
2458 extern const struct seq_operations ext4_mb_seq_groups_ops;
2459 extern long ext4_mb_stats;
2460 extern long ext4_mb_max_to_scan;
2461 extern int ext4_mb_init(struct super_block *);
2462 extern int ext4_mb_release(struct super_block *);
2463 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2464 struct ext4_allocation_request *, int *);
2465 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2466 extern void ext4_discard_preallocations(struct inode *);
2467 extern int __init ext4_init_mballoc(void);
2468 extern void ext4_exit_mballoc(void);
2469 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2470 struct buffer_head *bh, ext4_fsblk_t block,
2471 unsigned long count, int flags);
2472 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2473 ext4_group_t ngroups);
2474 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2475 ext4_group_t i, struct ext4_group_desc *desc);
2476 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2477 ext4_fsblk_t block, unsigned long count);
2478 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2479 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid);
2480
2481 /* inode.c */
2482 int ext4_inode_is_fast_symlink(struct inode *inode);
2483 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2484 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2485 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count,
2486 bool wait, struct buffer_head **bhs);
2487 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2488 struct buffer_head *bh_result, int create);
2489 int ext4_get_block(struct inode *inode, sector_t iblock,
2490 struct buffer_head *bh_result, int create);
2491 int ext4_dio_get_block(struct inode *inode, sector_t iblock,
2492 struct buffer_head *bh_result, int create);
2493 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2494 struct buffer_head *bh, int create);
2495 int ext4_walk_page_buffers(handle_t *handle,
2496 struct buffer_head *head,
2497 unsigned from,
2498 unsigned to,
2499 int *partial,
2500 int (*fn)(handle_t *handle,
2501 struct buffer_head *bh));
2502 int do_journal_get_write_access(handle_t *handle,
2503 struct buffer_head *bh);
2504 #define FALL_BACK_TO_NONDELALLOC 1
2505 #define CONVERT_INLINE_DATA 2
2506
2507 typedef enum {
2508 EXT4_IGET_NORMAL = 0,
2509 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */
2510 EXT4_IGET_HANDLE = 0x0002 /* Inode # is from a handle */
2511 } ext4_iget_flags;
2512
2513 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
2514 ext4_iget_flags flags, const char *function,
2515 unsigned int line);
2516
2517 #define ext4_iget(sb, ino, flags) \
2518 __ext4_iget((sb), (ino), (flags), __func__, __LINE__)
2519
2520 extern int ext4_write_inode(struct inode *, struct writeback_control *);
2521 extern int ext4_setattr(struct dentry *, struct iattr *);
2522 extern int ext4_getattr(const struct path *, struct kstat *, u32, unsigned int);
2523 extern void ext4_evict_inode(struct inode *);
2524 extern void ext4_clear_inode(struct inode *);
2525 extern int ext4_file_getattr(const struct path *, struct kstat *, u32, unsigned int);
2526 extern int ext4_sync_inode(handle_t *, struct inode *);
2527 extern void ext4_dirty_inode(struct inode *, int);
2528 extern int ext4_change_inode_journal_flag(struct inode *, int);
2529 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2530 extern int ext4_inode_attach_jinode(struct inode *inode);
2531 extern int ext4_can_truncate(struct inode *inode);
2532 extern int ext4_truncate(struct inode *);
2533 extern int ext4_break_layouts(struct inode *);
2534 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2535 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
2536 extern void ext4_set_inode_flags(struct inode *);
2537 extern int ext4_alloc_da_blocks(struct inode *inode);
2538 extern void ext4_set_aops(struct inode *inode);
2539 extern int ext4_writepage_trans_blocks(struct inode *);
2540 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2541 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2542 loff_t lstart, loff_t lend);
2543 extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf);
2544 extern vm_fault_t ext4_filemap_fault(struct vm_fault *vmf);
2545 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2546 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
2547 extern void ext4_da_release_space(struct inode *inode, int to_free);
2548 extern void ext4_da_update_reserve_space(struct inode *inode,
2549 int used, int quota_claim);
2550 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
2551 ext4_fsblk_t pblk, ext4_lblk_t len);
2552
2553 /* indirect.c */
2554 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2555 struct ext4_map_blocks *map, int flags);
2556 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2557 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2558 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2559 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2560 ext4_lblk_t start, ext4_lblk_t end);
2561
2562 /* ioctl.c */
2563 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2564 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2565
2566 /* migrate.c */
2567 extern int ext4_ext_migrate(struct inode *);
2568 extern int ext4_ind_migrate(struct inode *inode);
2569
2570 /* namei.c */
2571 extern int ext4_dirent_csum_verify(struct inode *inode,
2572 struct ext4_dir_entry *dirent);
2573 extern int ext4_orphan_add(handle_t *, struct inode *);
2574 extern int ext4_orphan_del(handle_t *, struct inode *);
2575 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2576 __u32 start_minor_hash, __u32 *next_hash);
2577 extern int ext4_search_dir(struct buffer_head *bh,
2578 char *search_buf,
2579 int buf_size,
2580 struct inode *dir,
2581 struct ext4_filename *fname,
2582 unsigned int offset,
2583 struct ext4_dir_entry_2 **res_dir);
2584 extern int ext4_generic_delete_entry(handle_t *handle,
2585 struct inode *dir,
2586 struct ext4_dir_entry_2 *de_del,
2587 struct buffer_head *bh,
2588 void *entry_buf,
2589 int buf_size,
2590 int csum_size);
2591 extern bool ext4_empty_dir(struct inode *inode);
2592
2593 /* resize.c */
2594 extern int ext4_group_add(struct super_block *sb,
2595 struct ext4_new_group_data *input);
2596 extern int ext4_group_extend(struct super_block *sb,
2597 struct ext4_super_block *es,
2598 ext4_fsblk_t n_blocks_count);
2599 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2600
2601 /* super.c */
2602 extern struct buffer_head *ext4_sb_bread(struct super_block *sb,
2603 sector_t block, int op_flags);
2604 extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
2605 extern int ext4_calculate_overhead(struct super_block *sb);
2606 extern void ext4_superblock_csum_set(struct super_block *sb);
2607 extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2608 extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2609 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2610 ext4_group_t ngroup);
2611 extern const char *ext4_decode_error(struct super_block *sb, int errno,
2612 char nbuf[16]);
2613 extern void ext4_mark_group_bitmap_corrupted(struct super_block *sb,
2614 ext4_group_t block_group,
2615 unsigned int flags);
2616
2617 extern __printf(4, 5)
2618 void __ext4_error(struct super_block *, const char *, unsigned int,
2619 const char *, ...);
2620 extern __printf(5, 6)
2621 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2622 const char *, ...);
2623 extern __printf(5, 6)
2624 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2625 const char *, ...);
2626 extern void __ext4_std_error(struct super_block *, const char *,
2627 unsigned int, int);
2628 extern __printf(4, 5)
2629 void __ext4_abort(struct super_block *, const char *, unsigned int,
2630 const char *, ...);
2631 extern __printf(4, 5)
2632 void __ext4_warning(struct super_block *, const char *, unsigned int,
2633 const char *, ...);
2634 extern __printf(4, 5)
2635 void __ext4_warning_inode(const struct inode *inode, const char *function,
2636 unsigned int line, const char *fmt, ...);
2637 extern __printf(3, 4)
2638 void __ext4_msg(struct super_block *, const char *, const char *, ...);
2639 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2640 const char *, unsigned int, const char *);
2641 extern __printf(7, 8)
2642 void __ext4_grp_locked_error(const char *, unsigned int,
2643 struct super_block *, ext4_group_t,
2644 unsigned long, ext4_fsblk_t,
2645 const char *, ...);
2646
2647 #define EXT4_ERROR_INODE(inode, fmt, a...) \
2648 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
2649
2650 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
2651 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
2652
2653 #define EXT4_ERROR_FILE(file, block, fmt, a...) \
2654 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
2655
2656 #ifdef CONFIG_PRINTK
2657
2658 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2659 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__)
2660 #define ext4_error_file(file, func, line, block, fmt, ...) \
2661 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2662 #define ext4_error(sb, fmt, ...) \
2663 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2664 #define ext4_abort(sb, fmt, ...) \
2665 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2666 #define ext4_warning(sb, fmt, ...) \
2667 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2668 #define ext4_warning_inode(inode, fmt, ...) \
2669 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
2670 #define ext4_msg(sb, level, fmt, ...) \
2671 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2672 #define dump_mmp_msg(sb, mmp, msg) \
2673 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2674 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2675 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2676 fmt, ##__VA_ARGS__)
2677
2678 #else
2679
2680 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2681 do { \
2682 no_printk(fmt, ##__VA_ARGS__); \
2683 __ext4_error_inode(inode, "", 0, block, " "); \
2684 } while (0)
2685 #define ext4_error_file(file, func, line, block, fmt, ...) \
2686 do { \
2687 no_printk(fmt, ##__VA_ARGS__); \
2688 __ext4_error_file(file, "", 0, block, " "); \
2689 } while (0)
2690 #define ext4_error(sb, fmt, ...) \
2691 do { \
2692 no_printk(fmt, ##__VA_ARGS__); \
2693 __ext4_error(sb, "", 0, " "); \
2694 } while (0)
2695 #define ext4_abort(sb, fmt, ...) \
2696 do { \
2697 no_printk(fmt, ##__VA_ARGS__); \
2698 __ext4_abort(sb, "", 0, " "); \
2699 } while (0)
2700 #define ext4_warning(sb, fmt, ...) \
2701 do { \
2702 no_printk(fmt, ##__VA_ARGS__); \
2703 __ext4_warning(sb, "", 0, " "); \
2704 } while (0)
2705 #define ext4_warning_inode(inode, fmt, ...) \
2706 do { \
2707 no_printk(fmt, ##__VA_ARGS__); \
2708 __ext4_warning_inode(inode, "", 0, " "); \
2709 } while (0)
2710 #define ext4_msg(sb, level, fmt, ...) \
2711 do { \
2712 no_printk(fmt, ##__VA_ARGS__); \
2713 __ext4_msg(sb, "", " "); \
2714 } while (0)
2715 #define dump_mmp_msg(sb, mmp, msg) \
2716 __dump_mmp_msg(sb, mmp, "", 0, "")
2717 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2718 do { \
2719 no_printk(fmt, ##__VA_ARGS__); \
2720 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
2721 } while (0)
2722
2723 #endif
2724
2725 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2726 __u32 compat);
2727 extern int ext4_update_rocompat_feature(handle_t *handle,
2728 struct super_block *sb, __u32 rocompat);
2729 extern int ext4_update_incompat_feature(handle_t *handle,
2730 struct super_block *sb, __u32 incompat);
2731 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2732 struct ext4_group_desc *bg);
2733 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2734 struct ext4_group_desc *bg);
2735 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2736 struct ext4_group_desc *bg);
2737 extern __u32 ext4_free_group_clusters(struct super_block *sb,
2738 struct ext4_group_desc *bg);
2739 extern __u32 ext4_free_inodes_count(struct super_block *sb,
2740 struct ext4_group_desc *bg);
2741 extern __u32 ext4_used_dirs_count(struct super_block *sb,
2742 struct ext4_group_desc *bg);
2743 extern __u32 ext4_itable_unused_count(struct super_block *sb,
2744 struct ext4_group_desc *bg);
2745 extern void ext4_block_bitmap_set(struct super_block *sb,
2746 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2747 extern void ext4_inode_bitmap_set(struct super_block *sb,
2748 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2749 extern void ext4_inode_table_set(struct super_block *sb,
2750 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2751 extern void ext4_free_group_clusters_set(struct super_block *sb,
2752 struct ext4_group_desc *bg,
2753 __u32 count);
2754 extern void ext4_free_inodes_set(struct super_block *sb,
2755 struct ext4_group_desc *bg, __u32 count);
2756 extern void ext4_used_dirs_set(struct super_block *sb,
2757 struct ext4_group_desc *bg, __u32 count);
2758 extern void ext4_itable_unused_set(struct super_block *sb,
2759 struct ext4_group_desc *bg, __u32 count);
2760 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2761 struct ext4_group_desc *gdp);
2762 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2763 struct ext4_group_desc *gdp);
2764 extern int ext4_register_li_request(struct super_block *sb,
2765 ext4_group_t first_not_zeroed);
2766
2767 static inline int ext4_has_metadata_csum(struct super_block *sb)
2768 {
2769 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
2770 !EXT4_SB(sb)->s_chksum_driver);
2771
2772 return ext4_has_feature_metadata_csum(sb) &&
2773 (EXT4_SB(sb)->s_chksum_driver != NULL);
2774 }
2775
2776 static inline int ext4_has_group_desc_csum(struct super_block *sb)
2777 {
2778 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb);
2779 }
2780
2781 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2782 {
2783 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2784 le32_to_cpu(es->s_blocks_count_lo);
2785 }
2786
2787 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2788 {
2789 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2790 le32_to_cpu(es->s_r_blocks_count_lo);
2791 }
2792
2793 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2794 {
2795 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2796 le32_to_cpu(es->s_free_blocks_count_lo);
2797 }
2798
2799 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2800 ext4_fsblk_t blk)
2801 {
2802 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2803 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2804 }
2805
2806 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2807 ext4_fsblk_t blk)
2808 {
2809 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2810 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2811 }
2812
2813 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2814 ext4_fsblk_t blk)
2815 {
2816 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2817 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2818 }
2819
2820 static inline loff_t ext4_isize(struct super_block *sb,
2821 struct ext4_inode *raw_inode)
2822 {
2823 if (ext4_has_feature_largedir(sb) ||
2824 S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2825 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2826 le32_to_cpu(raw_inode->i_size_lo);
2827
2828 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2829 }
2830
2831 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2832 {
2833 raw_inode->i_size_lo = cpu_to_le32(i_size);
2834 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2835 }
2836
2837 static inline
2838 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2839 ext4_group_t group)
2840 {
2841 struct ext4_group_info ***grp_info;
2842 long indexv, indexh;
2843 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
2844 grp_info = EXT4_SB(sb)->s_group_info;
2845 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2846 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2847 return grp_info[indexv][indexh];
2848 }
2849
2850 /*
2851 * Reading s_groups_count requires using smp_rmb() afterwards. See
2852 * the locking protocol documented in the comments of ext4_group_add()
2853 * in resize.c
2854 */
2855 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2856 {
2857 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
2858
2859 smp_rmb();
2860 return ngroups;
2861 }
2862
2863 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2864 ext4_group_t block_group)
2865 {
2866 return block_group >> sbi->s_log_groups_per_flex;
2867 }
2868
2869 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2870 {
2871 return 1 << sbi->s_log_groups_per_flex;
2872 }
2873
2874 #define ext4_std_error(sb, errno) \
2875 do { \
2876 if ((errno)) \
2877 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
2878 } while (0)
2879
2880 #ifdef CONFIG_SMP
2881 /* Each CPU can accumulate percpu_counter_batch clusters in their local
2882 * counters. So we need to make sure we have free clusters more
2883 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
2884 */
2885 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2886 #else
2887 #define EXT4_FREECLUSTERS_WATERMARK 0
2888 #endif
2889
2890 /* Update i_disksize. Requires i_mutex to avoid races with truncate */
2891 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2892 {
2893 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
2894 !inode_is_locked(inode));
2895 down_write(&EXT4_I(inode)->i_data_sem);
2896 if (newsize > EXT4_I(inode)->i_disksize)
2897 EXT4_I(inode)->i_disksize = newsize;
2898 up_write(&EXT4_I(inode)->i_data_sem);
2899 }
2900
2901 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
2902 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
2903 {
2904 int changed = 0;
2905
2906 if (newsize > inode->i_size) {
2907 i_size_write(inode, newsize);
2908 changed = 1;
2909 }
2910 if (newsize > EXT4_I(inode)->i_disksize) {
2911 ext4_update_i_disksize(inode, newsize);
2912 changed |= 2;
2913 }
2914 return changed;
2915 }
2916
2917 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
2918 loff_t len);
2919
2920 struct ext4_group_info {
2921 unsigned long bb_state;
2922 struct rb_root bb_free_root;
2923 ext4_grpblk_t bb_first_free; /* first free block */
2924 ext4_grpblk_t bb_free; /* total free blocks */
2925 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
2926 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
2927 struct list_head bb_prealloc_list;
2928 #ifdef DOUBLE_CHECK
2929 void *bb_bitmap;
2930 #endif
2931 struct rw_semaphore alloc_sem;
2932 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
2933 * regions, index is order.
2934 * bb_counters[3] = 5 means
2935 * 5 free 8-block regions. */
2936 };
2937
2938 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
2939 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
2940 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
2941 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
2942 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT \
2943 (1 << EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT)
2944 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT \
2945 (1 << EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT)
2946
2947 #define EXT4_MB_GRP_NEED_INIT(grp) \
2948 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2949 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
2950 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2951 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
2952 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2953
2954 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
2955 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2956 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
2957 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2958 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
2959 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2960
2961 #define EXT4_MAX_CONTENTION 8
2962 #define EXT4_CONTENTION_THRESHOLD 2
2963
2964 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2965 ext4_group_t group)
2966 {
2967 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2968 }
2969
2970 /*
2971 * Returns true if the filesystem is busy enough that attempts to
2972 * access the block group locks has run into contention.
2973 */
2974 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2975 {
2976 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2977 }
2978
2979 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2980 {
2981 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2982 if (spin_trylock(lock))
2983 /*
2984 * We're able to grab the lock right away, so drop the
2985 * lock contention counter.
2986 */
2987 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2988 else {
2989 /*
2990 * The lock is busy, so bump the contention counter,
2991 * and then wait on the spin lock.
2992 */
2993 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2994 EXT4_MAX_CONTENTION);
2995 spin_lock(lock);
2996 }
2997 }
2998
2999 static inline void ext4_unlock_group(struct super_block *sb,
3000 ext4_group_t group)
3001 {
3002 spin_unlock(ext4_group_lock_ptr(sb, group));
3003 }
3004
3005 /*
3006 * Block validity checking
3007 */
3008 #define ext4_check_indirect_blockref(inode, bh) \
3009 ext4_check_blockref(__func__, __LINE__, inode, \
3010 (__le32 *)(bh)->b_data, \
3011 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
3012
3013 #define ext4_ind_check_inode(inode) \
3014 ext4_check_blockref(__func__, __LINE__, inode, \
3015 EXT4_I(inode)->i_data, \
3016 EXT4_NDIR_BLOCKS)
3017
3018 /*
3019 * Inodes and files operations
3020 */
3021
3022 /* dir.c */
3023 extern const struct file_operations ext4_dir_operations;
3024
3025 #ifdef CONFIG_UNICODE
3026 extern const struct dentry_operations ext4_dentry_ops;
3027 #endif
3028
3029 /* file.c */
3030 extern const struct inode_operations ext4_file_inode_operations;
3031 extern const struct file_operations ext4_file_operations;
3032 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
3033
3034 /* inline.c */
3035 extern int ext4_get_max_inline_size(struct inode *inode);
3036 extern int ext4_find_inline_data_nolock(struct inode *inode);
3037 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
3038 unsigned int len);
3039 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
3040
3041 extern int ext4_readpage_inline(struct inode *inode, struct page *page);
3042 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3043 struct inode *inode,
3044 loff_t pos, unsigned len,
3045 unsigned flags,
3046 struct page **pagep);
3047 extern int ext4_write_inline_data_end(struct inode *inode,
3048 loff_t pos, unsigned len,
3049 unsigned copied,
3050 struct page *page);
3051 extern struct buffer_head *
3052 ext4_journalled_write_inline_data(struct inode *inode,
3053 unsigned len,
3054 struct page *page);
3055 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
3056 struct inode *inode,
3057 loff_t pos, unsigned len,
3058 unsigned flags,
3059 struct page **pagep,
3060 void **fsdata);
3061 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
3062 unsigned len, unsigned copied,
3063 struct page *page);
3064 extern int ext4_try_add_inline_entry(handle_t *handle,
3065 struct ext4_filename *fname,
3066 struct inode *dir, struct inode *inode);
3067 extern int ext4_try_create_inline_dir(handle_t *handle,
3068 struct inode *parent,
3069 struct inode *inode);
3070 extern int ext4_read_inline_dir(struct file *filp,
3071 struct dir_context *ctx,
3072 int *has_inline_data);
3073 extern int htree_inlinedir_to_tree(struct file *dir_file,
3074 struct inode *dir, ext4_lblk_t block,
3075 struct dx_hash_info *hinfo,
3076 __u32 start_hash, __u32 start_minor_hash,
3077 int *has_inline_data);
3078 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3079 struct ext4_filename *fname,
3080 struct ext4_dir_entry_2 **res_dir,
3081 int *has_inline_data);
3082 extern int ext4_delete_inline_entry(handle_t *handle,
3083 struct inode *dir,
3084 struct ext4_dir_entry_2 *de_del,
3085 struct buffer_head *bh,
3086 int *has_inline_data);
3087 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3088 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3089 struct ext4_dir_entry_2 **parent_de,
3090 int *retval);
3091 extern int ext4_inline_data_fiemap(struct inode *inode,
3092 struct fiemap_extent_info *fieinfo,
3093 int *has_inline, __u64 start, __u64 len);
3094
3095 struct iomap;
3096 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap);
3097
3098 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3099
3100 extern int ext4_convert_inline_data(struct inode *inode);
3101
3102 static inline int ext4_has_inline_data(struct inode *inode)
3103 {
3104 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3105 EXT4_I(inode)->i_inline_off;
3106 }
3107
3108 /* namei.c */
3109 extern const struct inode_operations ext4_dir_inode_operations;
3110 extern const struct inode_operations ext4_special_inode_operations;
3111 extern struct dentry *ext4_get_parent(struct dentry *child);
3112 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3113 struct ext4_dir_entry_2 *de,
3114 int blocksize, int csum_size,
3115 unsigned int parent_ino, int dotdot_real_len);
3116 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t,
3117 unsigned int blocksize);
3118 extern int ext4_handle_dirty_dirent_node(handle_t *handle,
3119 struct inode *inode,
3120 struct buffer_head *bh);
3121 extern int ext4_ci_compare(const struct inode *parent,
3122 const struct qstr *name,
3123 const struct qstr *entry);
3124
3125 #define S_SHIFT 12
3126 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = {
3127 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
3128 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
3129 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
3130 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
3131 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
3132 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
3133 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
3134 };
3135
3136 static inline void ext4_set_de_type(struct super_block *sb,
3137 struct ext4_dir_entry_2 *de,
3138 umode_t mode) {
3139 if (ext4_has_feature_filetype(sb))
3140 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3141 }
3142
3143 /* readpages.c */
3144 extern int ext4_mpage_readpages(struct address_space *mapping,
3145 struct list_head *pages, struct page *page,
3146 unsigned nr_pages, bool is_readahead);
3147
3148 /* symlink.c */
3149 extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3150 extern const struct inode_operations ext4_symlink_inode_operations;
3151 extern const struct inode_operations ext4_fast_symlink_inode_operations;
3152
3153 /* sysfs.c */
3154 extern int ext4_register_sysfs(struct super_block *sb);
3155 extern void ext4_unregister_sysfs(struct super_block *sb);
3156 extern int __init ext4_init_sysfs(void);
3157 extern void ext4_exit_sysfs(void);
3158
3159 /* block_validity */
3160 extern void ext4_release_system_zone(struct super_block *sb);
3161 extern int ext4_setup_system_zone(struct super_block *sb);
3162 extern int __init ext4_init_system_zone(void);
3163 extern void ext4_exit_system_zone(void);
3164 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
3165 ext4_fsblk_t start_blk,
3166 unsigned int count);
3167 extern int ext4_check_blockref(const char *, unsigned int,
3168 struct inode *, __le32 *, unsigned int);
3169
3170 /* extents.c */
3171 struct ext4_ext_path;
3172 struct ext4_extent;
3173
3174 /*
3175 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3176 * __le32.
3177 */
3178 #define EXT_MAX_BLOCKS 0xffffffff
3179
3180 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
3181 extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
3182 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3183 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3184 struct ext4_map_blocks *map, int flags);
3185 extern int ext4_ext_truncate(handle_t *, struct inode *);
3186 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3187 ext4_lblk_t end);
3188 extern void ext4_ext_init(struct super_block *);
3189 extern void ext4_ext_release(struct super_block *);
3190 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3191 loff_t len);
3192 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3193 loff_t offset, ssize_t len);
3194 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3195 struct ext4_map_blocks *map, int flags);
3196 extern int ext4_ext_calc_metadata_amount(struct inode *inode,
3197 ext4_lblk_t lblocks);
3198 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3199 int num,
3200 struct ext4_ext_path *path);
3201 extern int ext4_can_extents_be_merged(struct inode *inode,
3202 struct ext4_extent *ex1,
3203 struct ext4_extent *ex2);
3204 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3205 struct ext4_ext_path **,
3206 struct ext4_extent *, int);
3207 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3208 struct ext4_ext_path **,
3209 int flags);
3210 extern void ext4_ext_drop_refs(struct ext4_ext_path *);
3211 extern int ext4_ext_check_inode(struct inode *inode);
3212 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3213 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3214 __u64 start, __u64 len);
3215 extern int ext4_ext_precache(struct inode *inode);
3216 extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len);
3217 extern int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len);
3218 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3219 struct inode *inode2, ext4_lblk_t lblk1,
3220 ext4_lblk_t lblk2, ext4_lblk_t count,
3221 int mark_unwritten,int *err);
3222 extern int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu);
3223
3224 /* move_extent.c */
3225 extern void ext4_double_down_write_data_sem(struct inode *first,
3226 struct inode *second);
3227 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3228 struct inode *donor_inode);
3229 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3230 __u64 start_orig, __u64 start_donor,
3231 __u64 len, __u64 *moved_len);
3232
3233 /* page-io.c */
3234 extern int __init ext4_init_pageio(void);
3235 extern void ext4_exit_pageio(void);
3236 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3237 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3238 extern int ext4_put_io_end(ext4_io_end_t *io_end);
3239 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3240 extern void ext4_io_submit_init(struct ext4_io_submit *io,
3241 struct writeback_control *wbc);
3242 extern void ext4_end_io_rsv_work(struct work_struct *work);
3243 extern void ext4_io_submit(struct ext4_io_submit *io);
3244 extern int ext4_bio_write_page(struct ext4_io_submit *io,
3245 struct page *page,
3246 int len,
3247 struct writeback_control *wbc,
3248 bool keep_towrite);
3249
3250 /* mmp.c */
3251 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3252
3253 /*
3254 * Add new method to test whether block and inode bitmaps are properly
3255 * initialized. With uninit_bg reading the block from disk is not enough
3256 * to mark the bitmap uptodate. We need to also zero-out the bitmap
3257 */
3258 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3259
3260 static inline int bitmap_uptodate(struct buffer_head *bh)
3261 {
3262 return (buffer_uptodate(bh) &&
3263 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3264 }
3265 static inline void set_bitmap_uptodate(struct buffer_head *bh)
3266 {
3267 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3268 }
3269
3270 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
3271
3272 /* For ioend & aio unwritten conversion wait queues */
3273 #define EXT4_WQ_HASH_SZ 37
3274 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
3275 EXT4_WQ_HASH_SZ])
3276 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3277
3278 extern int ext4_resize_begin(struct super_block *sb);
3279 extern void ext4_resize_end(struct super_block *sb);
3280
3281 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
3282 struct ext4_io_end *io_end)
3283 {
3284 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
3285 io_end->flag |= EXT4_IO_END_UNWRITTEN;
3286 atomic_inc(&EXT4_I(inode)->i_unwritten);
3287 }
3288 }
3289
3290 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3291 {
3292 struct inode *inode = io_end->inode;
3293
3294 if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
3295 io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3296 /* Wake up anyone waiting on unwritten extent conversion */
3297 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
3298 wake_up_all(ext4_ioend_wq(inode));
3299 }
3300 }
3301
3302 extern const struct iomap_ops ext4_iomap_ops;
3303
3304 #endif /* __KERNEL__ */
3305
3306 #define EFSBADCRC EBADMSG /* Bad CRC detected */
3307 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
3308
3309 #endif /* _EXT4_H */