?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Merge tag 'xtensa-20180225' of git://github.com/jcmvbkbc/linux-xtensa
[cris-mirror.git] / fs / ext4 / ext4.h
blob3241475a1733492e8c88a58dfd9668ad330bbcb9
1 // SPDX-License-Identifier: GPL-2.0
2 /*
3 * ext4.h
4 *
5 * Copyright (C) 1992, 1993, 1994, 1995
6 * Remy Card (card@masi.ibp.fr)
7 * Laboratoire MASI - Institut Blaise Pascal
8 * Universite Pierre et Marie Curie (Paris VI)
9 *
10 * from
11 *
12 * linux/include/linux/minix_fs.h
13 *
14 * Copyright (C) 1991, 1992 Linus Torvalds
15 */
16
17 #ifndef _EXT4_H
18 #define _EXT4_H
19
20 #include <linux/types.h>
21 #include <linux/blkdev.h>
22 #include <linux/magic.h>
23 #include <linux/jbd2.h>
24 #include <linux/quota.h>
25 #include <linux/rwsem.h>
26 #include <linux/rbtree.h>
27 #include <linux/seqlock.h>
28 #include <linux/mutex.h>
29 #include <linux/timer.h>
30 #include <linux/version.h>
31 #include <linux/wait.h>
32 #include <linux/sched/signal.h>
33 #include <linux/blockgroup_lock.h>
34 #include <linux/percpu_counter.h>
35 #include <linux/ratelimit.h>
36 #include <crypto/hash.h>
37 #include <linux/falloc.h>
38 #include <linux/percpu-rwsem.h>
39 #ifdef __KERNEL__
40 #include <linux/compat.h>
41 #endif
42
43 #define __FS_HAS_ENCRYPTION IS_ENABLED(CONFIG_EXT4_FS_ENCRYPTION)
44 #include <linux/fscrypt.h>
45
46 /*
47 * The fourth extended filesystem constants/structures
48 */
49
50 /*
51 * with AGGRESSIVE_CHECK allocator runs consistency checks over
52 * structures. these checks slow things down a lot
53 */
54 #define AGGRESSIVE_CHECK__
55
56 /*
57 * with DOUBLE_CHECK defined mballoc creates persistent in-core
58 * bitmaps, maintains and uses them to check for double allocations
59 */
60 #define DOUBLE_CHECK__
61
62 /*
63 * Define EXT4FS_DEBUG to produce debug messages
64 */
65 #undef EXT4FS_DEBUG
66
67 /*
68 * Debug code
69 */
70 #ifdef EXT4FS_DEBUG
71 #define ext4_debug(f, a...) \
72 do { \
73 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
74 __FILE__, __LINE__, __func__); \
75 printk(KERN_DEBUG f, ## a); \
76 } while (0)
77 #else
78 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
79 #endif
80
81 /*
82 * Turn on EXT_DEBUG to get lots of info about extents operations.
83 */
84 #define EXT_DEBUG__
85 #ifdef EXT_DEBUG
86 #define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__)
87 #else
88 #define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
89 #endif
90
91 /* data type for block offset of block group */
92 typedef int ext4_grpblk_t;
93
94 /* data type for filesystem-wide blocks number */
95 typedef unsigned long long ext4_fsblk_t;
96
97 /* data type for file logical block number */
98 typedef __u32 ext4_lblk_t;
99
100 /* data type for block group number */
101 typedef unsigned int ext4_group_t;
102
103 enum SHIFT_DIRECTION {
104 SHIFT_LEFT = 0,
105 SHIFT_RIGHT,
106 };
107
108 /*
109 * Flags used in mballoc's allocation_context flags field.
110 *
111 * Also used to show what's going on for debugging purposes when the
112 * flag field is exported via the traceport interface
113 */
114
115 /* prefer goal again. length */
116 #define EXT4_MB_HINT_MERGE 0x0001
117 /* blocks already reserved */
118 #define EXT4_MB_HINT_RESERVED 0x0002
119 /* metadata is being allocated */
120 #define EXT4_MB_HINT_METADATA 0x0004
121 /* first blocks in the file */
122 #define EXT4_MB_HINT_FIRST 0x0008
123 /* search for the best chunk */
124 #define EXT4_MB_HINT_BEST 0x0010
125 /* data is being allocated */
126 #define EXT4_MB_HINT_DATA 0x0020
127 /* don't preallocate (for tails) */
128 #define EXT4_MB_HINT_NOPREALLOC 0x0040
129 /* allocate for locality group */
130 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080
131 /* allocate goal blocks or none */
132 #define EXT4_MB_HINT_GOAL_ONLY 0x0100
133 /* goal is meaningful */
134 #define EXT4_MB_HINT_TRY_GOAL 0x0200
135 /* blocks already pre-reserved by delayed allocation */
136 #define EXT4_MB_DELALLOC_RESERVED 0x0400
137 /* We are doing stream allocation */
138 #define EXT4_MB_STREAM_ALLOC 0x0800
139 /* Use reserved root blocks if needed */
140 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000
141 /* Use blocks from reserved pool */
142 #define EXT4_MB_USE_RESERVED 0x2000
143
144 struct ext4_allocation_request {
145 /* target inode for block we're allocating */
146 struct inode *inode;
147 /* how many blocks we want to allocate */
148 unsigned int len;
149 /* logical block in target inode */
150 ext4_lblk_t logical;
151 /* the closest logical allocated block to the left */
152 ext4_lblk_t lleft;
153 /* the closest logical allocated block to the right */
154 ext4_lblk_t lright;
155 /* phys. target (a hint) */
156 ext4_fsblk_t goal;
157 /* phys. block for the closest logical allocated block to the left */
158 ext4_fsblk_t pleft;
159 /* phys. block for the closest logical allocated block to the right */
160 ext4_fsblk_t pright;
161 /* flags. see above EXT4_MB_HINT_* */
162 unsigned int flags;
163 };
164
165 /*
166 * Logical to physical block mapping, used by ext4_map_blocks()
167 *
168 * This structure is used to pass requests into ext4_map_blocks() as
169 * well as to store the information returned by ext4_map_blocks(). It
170 * takes less room on the stack than a struct buffer_head.
171 */
172 #define EXT4_MAP_NEW (1 << BH_New)
173 #define EXT4_MAP_MAPPED (1 << BH_Mapped)
174 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten)
175 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary)
176 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
177 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
178
179 struct ext4_map_blocks {
180 ext4_fsblk_t m_pblk;
181 ext4_lblk_t m_lblk;
182 unsigned int m_len;
183 unsigned int m_flags;
184 };
185
186 /*
187 * Flags for ext4_io_end->flags
188 */
189 #define EXT4_IO_END_UNWRITTEN 0x0001
190
191 /*
192 * For converting unwritten extents on a work queue. 'handle' is used for
193 * buffered writeback.
194 */
195 typedef struct ext4_io_end {
196 struct list_head list; /* per-file finished IO list */
197 handle_t *handle; /* handle reserved for extent
198 * conversion */
199 struct inode *inode; /* file being written to */
200 struct bio *bio; /* Linked list of completed
201 * bios covering the extent */
202 unsigned int flag; /* unwritten or not */
203 atomic_t count; /* reference counter */
204 loff_t offset; /* offset in the file */
205 ssize_t size; /* size of the extent */
206 } ext4_io_end_t;
207
208 struct ext4_io_submit {
209 struct writeback_control *io_wbc;
210 struct bio *io_bio;
211 ext4_io_end_t *io_end;
212 sector_t io_next_block;
213 };
214
215 /*
216 * Special inodes numbers
217 */
218 #define EXT4_BAD_INO 1 /* Bad blocks inode */
219 #define EXT4_ROOT_INO 2 /* Root inode */
220 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */
221 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
222 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
223 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
224 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
225 #define EXT4_JOURNAL_INO 8 /* Journal inode */
226
227 /* First non-reserved inode for old ext4 filesystems */
228 #define EXT4_GOOD_OLD_FIRST_INO 11
229
230 /*
231 * Maximal count of links to a file
232 */
233 #define EXT4_LINK_MAX 65000
234
235 /*
236 * Macro-instructions used to manage several block sizes
237 */
238 #define EXT4_MIN_BLOCK_SIZE 1024
239 #define EXT4_MAX_BLOCK_SIZE 65536
240 #define EXT4_MIN_BLOCK_LOG_SIZE 10
241 #define EXT4_MAX_BLOCK_LOG_SIZE 16
242 #define EXT4_MAX_CLUSTER_LOG_SIZE 30
243 #ifdef __KERNEL__
244 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
245 #else
246 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
247 #endif
248 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
249 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
250 EXT4_SB(s)->s_cluster_bits)
251 #ifdef __KERNEL__
252 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
253 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
254 #else
255 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
256 #endif
257 #ifdef __KERNEL__
258 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
259 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
260 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
261 #else
262 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
263 EXT4_GOOD_OLD_INODE_SIZE : \
264 (s)->s_inode_size)
265 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
266 EXT4_GOOD_OLD_FIRST_INO : \
267 (s)->s_first_ino)
268 #endif
269 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
270 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \
271 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \
272 blkbits))
273
274 /* Translate a block number to a cluster number */
275 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
276 /* Translate a cluster number to a block number */
277 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
278 /* Translate # of blks to # of clusters */
279 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
280 (sbi)->s_cluster_bits)
281 /* Mask out the low bits to get the starting block of the cluster */
282 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
283 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
284 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
285 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
286 /* Get the cluster offset */
287 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
288 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
289 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
290 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
291
292 /*
293 * Structure of a blocks group descriptor
294 */
295 struct ext4_group_desc
296 {
297 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
298 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
299 __le32 bg_inode_table_lo; /* Inodes table block */
300 __le16 bg_free_blocks_count_lo;/* Free blocks count */
301 __le16 bg_free_inodes_count_lo;/* Free inodes count */
302 __le16 bg_used_dirs_count_lo; /* Directories count */
303 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
304 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
305 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
306 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
307 __le16 bg_itable_unused_lo; /* Unused inodes count */
308 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
309 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
310 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
311 __le32 bg_inode_table_hi; /* Inodes table block MSB */
312 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
313 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
314 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
315 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
316 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
317 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
318 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
319 __u32 bg_reserved;
320 };
321
322 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
323 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
324 sizeof(__le16))
325 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
326 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
327 sizeof(__le16))
328
329 /*
330 * Structure of a flex block group info
331 */
332
333 struct flex_groups {
334 atomic64_t free_clusters;
335 atomic_t free_inodes;
336 atomic_t used_dirs;
337 };
338
339 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
340 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
341 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
342
343 /*
344 * Macro-instructions used to manage group descriptors
345 */
346 #define EXT4_MIN_DESC_SIZE 32
347 #define EXT4_MIN_DESC_SIZE_64BIT 64
348 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
349 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
350 #ifdef __KERNEL__
351 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
352 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
353 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
354 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
355 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
356 #else
357 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
358 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
359 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
360 #endif
361
362 /*
363 * Constants relative to the data blocks
364 */
365 #define EXT4_NDIR_BLOCKS 12
366 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
367 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
368 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
369 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
370
371 /*
372 * Inode flags
373 */
374 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
375 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
376 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
377 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
378 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
379 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
380 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
381 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
382 /* Reserved for compression usage... */
383 #define EXT4_DIRTY_FL 0x00000100
384 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
385 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
386 /* nb: was previously EXT2_ECOMPR_FL */
387 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */
388 /* End compression flags --- maybe not all used */
389 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
390 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
391 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
392 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
393 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
394 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
395 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
396 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
397 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
398 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
399 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
400 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
401 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
402
403 #define EXT4_FL_USER_VISIBLE 0x304BDFFF /* User visible flags */
404 #define EXT4_FL_USER_MODIFIABLE 0x204BC0FF /* User modifiable flags */
405
406 /* Flags we can manipulate with through EXT4_IOC_FSSETXATTR */
407 #define EXT4_FL_XFLAG_VISIBLE (EXT4_SYNC_FL | \
408 EXT4_IMMUTABLE_FL | \
409 EXT4_APPEND_FL | \
410 EXT4_NODUMP_FL | \
411 EXT4_NOATIME_FL | \
412 EXT4_PROJINHERIT_FL)
413
414 /* Flags that should be inherited by new inodes from their parent. */
415 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
416 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
417 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
418 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
419 EXT4_PROJINHERIT_FL)
420
421 /* Flags that are appropriate for regular files (all but dir-specific ones). */
422 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
423
424 /* Flags that are appropriate for non-directories/regular files. */
425 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
426
427 /* Mask out flags that are inappropriate for the given type of inode. */
428 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
429 {
430 if (S_ISDIR(mode))
431 return flags;
432 else if (S_ISREG(mode))
433 return flags & EXT4_REG_FLMASK;
434 else
435 return flags & EXT4_OTHER_FLMASK;
436 }
437
438 /*
439 * Inode flags used for atomic set/get
440 */
441 enum {
442 EXT4_INODE_SECRM = 0, /* Secure deletion */
443 EXT4_INODE_UNRM = 1, /* Undelete */
444 EXT4_INODE_COMPR = 2, /* Compress file */
445 EXT4_INODE_SYNC = 3, /* Synchronous updates */
446 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
447 EXT4_INODE_APPEND = 5, /* writes to file may only append */
448 EXT4_INODE_NODUMP = 6, /* do not dump file */
449 EXT4_INODE_NOATIME = 7, /* do not update atime */
450 /* Reserved for compression usage... */
451 EXT4_INODE_DIRTY = 8,
452 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
453 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
454 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */
455 /* End compression flags --- maybe not all used */
456 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
457 EXT4_INODE_IMAGIC = 13, /* AFS directory */
458 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
459 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
460 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
461 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
462 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
463 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
464 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
465 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
466 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
467 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */
468 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
469 };
470
471 /*
472 * Since it's pretty easy to mix up bit numbers and hex values, we use a
473 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
474 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
475 * any extra space in the compiled kernel image, otherwise, the build will fail.
476 * It's important that these values are the same, since we are using
477 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
478 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
479 * values found in ext2, ext3 and ext4 filesystems, and of course the values
480 * defined in e2fsprogs.
481 *
482 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
483 */
484 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
485 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
486
487 static inline void ext4_check_flag_values(void)
488 {
489 CHECK_FLAG_VALUE(SECRM);
490 CHECK_FLAG_VALUE(UNRM);
491 CHECK_FLAG_VALUE(COMPR);
492 CHECK_FLAG_VALUE(SYNC);
493 CHECK_FLAG_VALUE(IMMUTABLE);
494 CHECK_FLAG_VALUE(APPEND);
495 CHECK_FLAG_VALUE(NODUMP);
496 CHECK_FLAG_VALUE(NOATIME);
497 CHECK_FLAG_VALUE(DIRTY);
498 CHECK_FLAG_VALUE(COMPRBLK);
499 CHECK_FLAG_VALUE(NOCOMPR);
500 CHECK_FLAG_VALUE(ENCRYPT);
501 CHECK_FLAG_VALUE(INDEX);
502 CHECK_FLAG_VALUE(IMAGIC);
503 CHECK_FLAG_VALUE(JOURNAL_DATA);
504 CHECK_FLAG_VALUE(NOTAIL);
505 CHECK_FLAG_VALUE(DIRSYNC);
506 CHECK_FLAG_VALUE(TOPDIR);
507 CHECK_FLAG_VALUE(HUGE_FILE);
508 CHECK_FLAG_VALUE(EXTENTS);
509 CHECK_FLAG_VALUE(EA_INODE);
510 CHECK_FLAG_VALUE(EOFBLOCKS);
511 CHECK_FLAG_VALUE(INLINE_DATA);
512 CHECK_FLAG_VALUE(PROJINHERIT);
513 CHECK_FLAG_VALUE(RESERVED);
514 }
515
516 /* Used to pass group descriptor data when online resize is done */
517 struct ext4_new_group_input {
518 __u32 group; /* Group number for this data */
519 __u64 block_bitmap; /* Absolute block number of block bitmap */
520 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
521 __u64 inode_table; /* Absolute block number of inode table start */
522 __u32 blocks_count; /* Total number of blocks in this group */
523 __u16 reserved_blocks; /* Number of reserved blocks in this group */
524 __u16 unused;
525 };
526
527 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
528 struct compat_ext4_new_group_input {
529 u32 group;
530 compat_u64 block_bitmap;
531 compat_u64 inode_bitmap;
532 compat_u64 inode_table;
533 u32 blocks_count;
534 u16 reserved_blocks;
535 u16 unused;
536 };
537 #endif
538
539 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
540 struct ext4_new_group_data {
541 __u32 group;
542 __u64 block_bitmap;
543 __u64 inode_bitmap;
544 __u64 inode_table;
545 __u32 blocks_count;
546 __u16 reserved_blocks;
547 __u16 mdata_blocks;
548 __u32 free_clusters_count;
549 };
550
551 /* Indexes used to index group tables in ext4_new_group_data */
552 enum {
553 BLOCK_BITMAP = 0, /* block bitmap */
554 INODE_BITMAP, /* inode bitmap */
555 INODE_TABLE, /* inode tables */
556 GROUP_TABLE_COUNT,
557 };
558
559 /*
560 * Flags used by ext4_map_blocks()
561 */
562 /* Allocate any needed blocks and/or convert an unwritten
563 extent to be an initialized ext4 */
564 #define EXT4_GET_BLOCKS_CREATE 0x0001
565 /* Request the creation of an unwritten extent */
566 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002
567 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\
568 EXT4_GET_BLOCKS_CREATE)
569 /* Caller is from the delayed allocation writeout path
570 * finally doing the actual allocation of delayed blocks */
571 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
572 /* caller is from the direct IO path, request to creation of an
573 unwritten extents if not allocated, split the unwritten
574 extent if blocks has been preallocated already*/
575 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
576 #define EXT4_GET_BLOCKS_CONVERT 0x0010
577 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
578 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
579 /* Convert extent to initialized after IO complete */
580 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
581 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
582 /* Eventual metadata allocation (due to growing extent tree)
583 * should not fail, so try to use reserved blocks for that.*/
584 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
585 /* Don't normalize allocation size (used for fallocate) */
586 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
587 /* Request will not result in inode size update (user for fallocate) */
588 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080
589 /* Convert written extents to unwritten */
590 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100
591 /* Write zeros to newly created written extents */
592 #define EXT4_GET_BLOCKS_ZERO 0x0200
593 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\
594 EXT4_GET_BLOCKS_ZERO)
595 /* Caller will submit data before dropping transaction handle. This
596 * allows jbd2 to avoid submitting data before commit. */
597 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400
598
599 /*
600 * The bit position of these flags must not overlap with any of the
601 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(),
602 * read_extent_tree_block(), ext4_split_extent_at(),
603 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
604 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
605 * caching the extents when reading from the extent tree while a
606 * truncate or punch hole operation is in progress.
607 */
608 #define EXT4_EX_NOCACHE 0x40000000
609 #define EXT4_EX_FORCE_CACHE 0x20000000
610
611 /*
612 * Flags used by ext4_free_blocks
613 */
614 #define EXT4_FREE_BLOCKS_METADATA 0x0001
615 #define EXT4_FREE_BLOCKS_FORGET 0x0002
616 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
617 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
618 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
619 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
620
621 /*
622 * ioctl commands
623 */
624 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
625 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
626 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
627 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
628 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
629 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
630 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
631 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
632 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
633 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
634 #define EXT4_IOC_MIGRATE _IO('f', 9)
635 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
636 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
637 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
638 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
639 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
640 #define EXT4_IOC_SWAP_BOOT _IO('f', 17)
641 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18)
642 #define EXT4_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY
643 #define EXT4_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT
644 #define EXT4_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY
645
646 #define EXT4_IOC_FSGETXATTR FS_IOC_FSGETXATTR
647 #define EXT4_IOC_FSSETXATTR FS_IOC_FSSETXATTR
648
649 #define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32)
650
651 /*
652 * Flags for going down operation
653 */
654 #define EXT4_GOING_FLAGS_DEFAULT 0x0 /* going down */
655 #define EXT4_GOING_FLAGS_LOGFLUSH 0x1 /* flush log but not data */
656 #define EXT4_GOING_FLAGS_NOLOGFLUSH 0x2 /* don't flush log nor data */
657
658
659 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
660 /*
661 * ioctl commands in 32 bit emulation
662 */
663 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
664 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
665 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
666 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
667 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
668 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
669 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
670 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
671 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
672 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
673 #endif
674
675 /* Max physical block we can address w/o extents */
676 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
677
678 /*
679 * Structure of an inode on the disk
680 */
681 struct ext4_inode {
682 __le16 i_mode; /* File mode */
683 __le16 i_uid; /* Low 16 bits of Owner Uid */
684 __le32 i_size_lo; /* Size in bytes */
685 __le32 i_atime; /* Access time */
686 __le32 i_ctime; /* Inode Change time */
687 __le32 i_mtime; /* Modification time */
688 __le32 i_dtime; /* Deletion Time */
689 __le16 i_gid; /* Low 16 bits of Group Id */
690 __le16 i_links_count; /* Links count */
691 __le32 i_blocks_lo; /* Blocks count */
692 __le32 i_flags; /* File flags */
693 union {
694 struct {
695 __le32 l_i_version;
696 } linux1;
697 struct {
698 __u32 h_i_translator;
699 } hurd1;
700 struct {
701 __u32 m_i_reserved1;
702 } masix1;
703 } osd1; /* OS dependent 1 */
704 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
705 __le32 i_generation; /* File version (for NFS) */
706 __le32 i_file_acl_lo; /* File ACL */
707 __le32 i_size_high;
708 __le32 i_obso_faddr; /* Obsoleted fragment address */
709 union {
710 struct {
711 __le16 l_i_blocks_high; /* were l_i_reserved1 */
712 __le16 l_i_file_acl_high;
713 __le16 l_i_uid_high; /* these 2 fields */
714 __le16 l_i_gid_high; /* were reserved2[0] */
715 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
716 __le16 l_i_reserved;
717 } linux2;
718 struct {
719 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
720 __u16 h_i_mode_high;
721 __u16 h_i_uid_high;
722 __u16 h_i_gid_high;
723 __u32 h_i_author;
724 } hurd2;
725 struct {
726 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
727 __le16 m_i_file_acl_high;
728 __u32 m_i_reserved2[2];
729 } masix2;
730 } osd2; /* OS dependent 2 */
731 __le16 i_extra_isize;
732 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
733 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
734 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
735 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
736 __le32 i_crtime; /* File Creation time */
737 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
738 __le32 i_version_hi; /* high 32 bits for 64-bit version */
739 __le32 i_projid; /* Project ID */
740 };
741
742 struct move_extent {
743 __u32 reserved; /* should be zero */
744 __u32 donor_fd; /* donor file descriptor */
745 __u64 orig_start; /* logical start offset in block for orig */
746 __u64 donor_start; /* logical start offset in block for donor */
747 __u64 len; /* block length to be moved */
748 __u64 moved_len; /* moved block length */
749 };
750
751 #define EXT4_EPOCH_BITS 2
752 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
753 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
754
755 /*
756 * Extended fields will fit into an inode if the filesystem was formatted
757 * with large inodes (-I 256 or larger) and there are not currently any EAs
758 * consuming all of the available space. For new inodes we always reserve
759 * enough space for the kernel's known extended fields, but for inodes
760 * created with an old kernel this might not have been the case. None of
761 * the extended inode fields is critical for correct filesystem operation.
762 * This macro checks if a certain field fits in the inode. Note that
763 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
764 */
765 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
766 ((offsetof(typeof(*ext4_inode), field) + \
767 sizeof((ext4_inode)->field)) \
768 <= (EXT4_GOOD_OLD_INODE_SIZE + \
769 (einode)->i_extra_isize)) \
770
771 /*
772 * We use an encoding that preserves the times for extra epoch "00":
773 *
774 * extra msb of adjust for signed
775 * epoch 32-bit 32-bit tv_sec to
776 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
777 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
778 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
779 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
780 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
781 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
782 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
783 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
784 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
785 *
786 * Note that previous versions of the kernel on 64-bit systems would
787 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
788 * 1970. e2fsck will correct this, assuming that it is run on the
789 * affected filesystem before 2242.
790 */
791
792 static inline __le32 ext4_encode_extra_time(struct timespec *time)
793 {
794 u32 extra = sizeof(time->tv_sec) > 4 ?
795 ((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK : 0;
796 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
797 }
798
799 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
800 {
801 if (unlikely(sizeof(time->tv_sec) > 4 &&
802 (extra & cpu_to_le32(EXT4_EPOCH_MASK)))) {
803
804 #if 1
805 /* Handle legacy encoding of pre-1970 dates with epoch
806 * bits 1,1. (This backwards compatibility may be removed
807 * at the discretion of the ext4 developers.)
808 */
809 u64 extra_bits = le32_to_cpu(extra) & EXT4_EPOCH_MASK;
810 if (extra_bits == 3 && ((time->tv_sec) & 0x80000000) != 0)
811 extra_bits = 0;
812 time->tv_sec += extra_bits << 32;
813 #else
814 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
815 #endif
816 }
817 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
818 }
819
820 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
821 do { \
822 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
823 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
824 (raw_inode)->xtime ## _extra = \
825 ext4_encode_extra_time(&(inode)->xtime); \
826 } while (0)
827
828 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
829 do { \
830 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
831 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
832 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
833 (raw_inode)->xtime ## _extra = \
834 ext4_encode_extra_time(&(einode)->xtime); \
835 } while (0)
836
837 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
838 do { \
839 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
840 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
841 ext4_decode_extra_time(&(inode)->xtime, \
842 raw_inode->xtime ## _extra); \
843 else \
844 (inode)->xtime.tv_nsec = 0; \
845 } while (0)
846
847 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
848 do { \
849 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
850 (einode)->xtime.tv_sec = \
851 (signed)le32_to_cpu((raw_inode)->xtime); \
852 else \
853 (einode)->xtime.tv_sec = 0; \
854 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
855 ext4_decode_extra_time(&(einode)->xtime, \
856 raw_inode->xtime ## _extra); \
857 else \
858 (einode)->xtime.tv_nsec = 0; \
859 } while (0)
860
861 #define i_disk_version osd1.linux1.l_i_version
862
863 #if defined(__KERNEL__) || defined(__linux__)
864 #define i_reserved1 osd1.linux1.l_i_reserved1
865 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
866 #define i_blocks_high osd2.linux2.l_i_blocks_high
867 #define i_uid_low i_uid
868 #define i_gid_low i_gid
869 #define i_uid_high osd2.linux2.l_i_uid_high
870 #define i_gid_high osd2.linux2.l_i_gid_high
871 #define i_checksum_lo osd2.linux2.l_i_checksum_lo
872
873 #elif defined(__GNU__)
874
875 #define i_translator osd1.hurd1.h_i_translator
876 #define i_uid_high osd2.hurd2.h_i_uid_high
877 #define i_gid_high osd2.hurd2.h_i_gid_high
878 #define i_author osd2.hurd2.h_i_author
879
880 #elif defined(__masix__)
881
882 #define i_reserved1 osd1.masix1.m_i_reserved1
883 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
884 #define i_reserved2 osd2.masix2.m_i_reserved2
885
886 #endif /* defined(__KERNEL__) || defined(__linux__) */
887
888 #include "extents_status.h"
889
890 /*
891 * Lock subclasses for i_data_sem in the ext4_inode_info structure.
892 *
893 * These are needed to avoid lockdep false positives when we need to
894 * allocate blocks to the quota inode during ext4_map_blocks(), while
895 * holding i_data_sem for a normal (non-quota) inode. Since we don't
896 * do quota tracking for the quota inode, this avoids deadlock (as
897 * well as infinite recursion, since it isn't turtles all the way
898 * down...)
899 *
900 * I_DATA_SEM_NORMAL - Used for most inodes
901 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode
902 * where the second inode has larger inode number
903 * than the first
904 * I_DATA_SEM_QUOTA - Used for quota inodes only
905 */
906 enum {
907 I_DATA_SEM_NORMAL = 0,
908 I_DATA_SEM_OTHER,
909 I_DATA_SEM_QUOTA,
910 };
911
912
913 /*
914 * fourth extended file system inode data in memory
915 */
916 struct ext4_inode_info {
917 __le32 i_data[15]; /* unconverted */
918 __u32 i_dtime;
919 ext4_fsblk_t i_file_acl;
920
921 /*
922 * i_block_group is the number of the block group which contains
923 * this file's inode. Constant across the lifetime of the inode,
924 * it is used for making block allocation decisions - we try to
925 * place a file's data blocks near its inode block, and new inodes
926 * near to their parent directory's inode.
927 */
928 ext4_group_t i_block_group;
929 ext4_lblk_t i_dir_start_lookup;
930 #if (BITS_PER_LONG < 64)
931 unsigned long i_state_flags; /* Dynamic state flags */
932 #endif
933 unsigned long i_flags;
934
935 /*
936 * Extended attributes can be read independently of the main file
937 * data. Taking i_mutex even when reading would cause contention
938 * between readers of EAs and writers of regular file data, so
939 * instead we synchronize on xattr_sem when reading or changing
940 * EAs.
941 */
942 struct rw_semaphore xattr_sem;
943
944 struct list_head i_orphan; /* unlinked but open inodes */
945
946 /*
947 * i_disksize keeps track of what the inode size is ON DISK, not
948 * in memory. During truncate, i_size is set to the new size by
949 * the VFS prior to calling ext4_truncate(), but the filesystem won't
950 * set i_disksize to 0 until the truncate is actually under way.
951 *
952 * The intent is that i_disksize always represents the blocks which
953 * are used by this file. This allows recovery to restart truncate
954 * on orphans if we crash during truncate. We actually write i_disksize
955 * into the on-disk inode when writing inodes out, instead of i_size.
956 *
957 * The only time when i_disksize and i_size may be different is when
958 * a truncate is in progress. The only things which change i_disksize
959 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
960 */
961 loff_t i_disksize;
962
963 /*
964 * i_data_sem is for serialising ext4_truncate() against
965 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
966 * data tree are chopped off during truncate. We can't do that in
967 * ext4 because whenever we perform intermediate commits during
968 * truncate, the inode and all the metadata blocks *must* be in a
969 * consistent state which allows truncation of the orphans to restart
970 * during recovery. Hence we must fix the get_block-vs-truncate race
971 * by other means, so we have i_data_sem.
972 */
973 struct rw_semaphore i_data_sem;
974 /*
975 * i_mmap_sem is for serializing page faults with truncate / punch hole
976 * operations. We have to make sure that new page cannot be faulted in
977 * a section of the inode that is being punched. We cannot easily use
978 * i_data_sem for this since we need protection for the whole punch
979 * operation and i_data_sem ranks below transaction start so we have
980 * to occasionally drop it.
981 */
982 struct rw_semaphore i_mmap_sem;
983 struct inode vfs_inode;
984 struct jbd2_inode *jinode;
985
986 spinlock_t i_raw_lock; /* protects updates to the raw inode */
987
988 /*
989 * File creation time. Its function is same as that of
990 * struct timespec i_{a,c,m}time in the generic inode.
991 */
992 struct timespec i_crtime;
993
994 /* mballoc */
995 struct list_head i_prealloc_list;
996 spinlock_t i_prealloc_lock;
997
998 /* extents status tree */
999 struct ext4_es_tree i_es_tree;
1000 rwlock_t i_es_lock;
1001 struct list_head i_es_list;
1002 unsigned int i_es_all_nr; /* protected by i_es_lock */
1003 unsigned int i_es_shk_nr; /* protected by i_es_lock */
1004 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for
1005 extents to shrink. Protected by
1006 i_es_lock */
1007
1008 /* ialloc */
1009 ext4_group_t i_last_alloc_group;
1010
1011 /* allocation reservation info for delalloc */
1012 /* In case of bigalloc, this refer to clusters rather than blocks */
1013 unsigned int i_reserved_data_blocks;
1014 ext4_lblk_t i_da_metadata_calc_last_lblock;
1015 int i_da_metadata_calc_len;
1016
1017 /* on-disk additional length */
1018 __u16 i_extra_isize;
1019
1020 /* Indicate the inline data space. */
1021 u16 i_inline_off;
1022 u16 i_inline_size;
1023
1024 #ifdef CONFIG_QUOTA
1025 /* quota space reservation, managed internally by quota code */
1026 qsize_t i_reserved_quota;
1027 #endif
1028
1029 /* Lock protecting lists below */
1030 spinlock_t i_completed_io_lock;
1031 /*
1032 * Completed IOs that need unwritten extents handling and have
1033 * transaction reserved
1034 */
1035 struct list_head i_rsv_conversion_list;
1036 struct work_struct i_rsv_conversion_work;
1037 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1038
1039 spinlock_t i_block_reservation_lock;
1040
1041 /*
1042 * Transactions that contain inode's metadata needed to complete
1043 * fsync and fdatasync, respectively.
1044 */
1045 tid_t i_sync_tid;
1046 tid_t i_datasync_tid;
1047
1048 #ifdef CONFIG_QUOTA
1049 struct dquot *i_dquot[MAXQUOTAS];
1050 #endif
1051
1052 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1053 __u32 i_csum_seed;
1054
1055 kprojid_t i_projid;
1056 };
1057
1058 /*
1059 * File system states
1060 */
1061 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
1062 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
1063 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
1064
1065 /*
1066 * Misc. filesystem flags
1067 */
1068 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
1069 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
1070 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
1071
1072 /*
1073 * Mount flags set via mount options or defaults
1074 */
1075 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */
1076 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
1077 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
1078 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
1079 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
1080 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
1081 #define EXT4_MOUNT_ERRORS_MASK 0x00070
1082 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
1083 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
1084 #ifdef CONFIG_FS_DAX
1085 #define EXT4_MOUNT_DAX 0x00200 /* Direct Access */
1086 #else
1087 #define EXT4_MOUNT_DAX 0
1088 #endif
1089 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
1090 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
1091 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
1092 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
1093 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
1094 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
1095 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
1096 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
1097 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
1098 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
1099 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */
1100 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota,
1101 * enable enforcement for hidden
1102 * quota files */
1103 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable
1104 * enforcement for hidden quota
1105 * files */
1106 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota
1107 * enforcement */
1108 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
1109 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
1110 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1111 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
1112 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
1113 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
1114 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
1115 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
1116
1117 /*
1118 * Mount flags set either automatically (could not be set by mount option)
1119 * based on per file system feature or property or in special cases such as
1120 * distinguishing between explicit mount option definition and default.
1121 */
1122 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1123 specified delalloc */
1124 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1125 size of blocksize * 8
1126 blocks */
1127 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1128 file systems */
1129
1130 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly
1131 specified journal checksum */
1132
1133 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1134 ~EXT4_MOUNT_##opt
1135 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1136 EXT4_MOUNT_##opt
1137 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1138 EXT4_MOUNT_##opt)
1139
1140 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1141 ~EXT4_MOUNT2_##opt
1142 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1143 EXT4_MOUNT2_##opt
1144 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1145 EXT4_MOUNT2_##opt)
1146
1147 #define ext4_test_and_set_bit __test_and_set_bit_le
1148 #define ext4_set_bit __set_bit_le
1149 #define ext4_set_bit_atomic ext2_set_bit_atomic
1150 #define ext4_test_and_clear_bit __test_and_clear_bit_le
1151 #define ext4_clear_bit __clear_bit_le
1152 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
1153 #define ext4_test_bit test_bit_le
1154 #define ext4_find_next_zero_bit find_next_zero_bit_le
1155 #define ext4_find_next_bit find_next_bit_le
1156
1157 extern void ext4_set_bits(void *bm, int cur, int len);
1158
1159 /*
1160 * Maximal mount counts between two filesystem checks
1161 */
1162 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1163 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1164
1165 /*
1166 * Behaviour when detecting errors
1167 */
1168 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1169 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1170 #define EXT4_ERRORS_PANIC 3 /* Panic */
1171 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1172
1173 /* Metadata checksum algorithm codes */
1174 #define EXT4_CRC32C_CHKSUM 1
1175
1176 /*
1177 * Structure of the super block
1178 */
1179 struct ext4_super_block {
1180 /*00*/ __le32 s_inodes_count; /* Inodes count */
1181 __le32 s_blocks_count_lo; /* Blocks count */
1182 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1183 __le32 s_free_blocks_count_lo; /* Free blocks count */
1184 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1185 __le32 s_first_data_block; /* First Data Block */
1186 __le32 s_log_block_size; /* Block size */
1187 __le32 s_log_cluster_size; /* Allocation cluster size */
1188 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1189 __le32 s_clusters_per_group; /* # Clusters per group */
1190 __le32 s_inodes_per_group; /* # Inodes per group */
1191 __le32 s_mtime; /* Mount time */
1192 /*30*/ __le32 s_wtime; /* Write time */
1193 __le16 s_mnt_count; /* Mount count */
1194 __le16 s_max_mnt_count; /* Maximal mount count */
1195 __le16 s_magic; /* Magic signature */
1196 __le16 s_state; /* File system state */
1197 __le16 s_errors; /* Behaviour when detecting errors */
1198 __le16 s_minor_rev_level; /* minor revision level */
1199 /*40*/ __le32 s_lastcheck; /* time of last check */
1200 __le32 s_checkinterval; /* max. time between checks */
1201 __le32 s_creator_os; /* OS */
1202 __le32 s_rev_level; /* Revision level */
1203 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1204 __le16 s_def_resgid; /* Default gid for reserved blocks */
1205 /*
1206 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1207 *
1208 * Note: the difference between the compatible feature set and
1209 * the incompatible feature set is that if there is a bit set
1210 * in the incompatible feature set that the kernel doesn't
1211 * know about, it should refuse to mount the filesystem.
1212 *
1213 * e2fsck's requirements are more strict; if it doesn't know
1214 * about a feature in either the compatible or incompatible
1215 * feature set, it must abort and not try to meddle with
1216 * things it doesn't understand...
1217 */
1218 __le32 s_first_ino; /* First non-reserved inode */
1219 __le16 s_inode_size; /* size of inode structure */
1220 __le16 s_block_group_nr; /* block group # of this superblock */
1221 __le32 s_feature_compat; /* compatible feature set */
1222 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1223 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1224 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1225 /*78*/ char s_volume_name[16]; /* volume name */
1226 /*88*/ char s_last_mounted[64]; /* directory where last mounted */
1227 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1228 /*
1229 * Performance hints. Directory preallocation should only
1230 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1231 */
1232 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1233 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1234 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1235 /*
1236 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1237 */
1238 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1239 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1240 __le32 s_journal_dev; /* device number of journal file */
1241 __le32 s_last_orphan; /* start of list of inodes to delete */
1242 __le32 s_hash_seed[4]; /* HTREE hash seed */
1243 __u8 s_def_hash_version; /* Default hash version to use */
1244 __u8 s_jnl_backup_type;
1245 __le16 s_desc_size; /* size of group descriptor */
1246 /*100*/ __le32 s_default_mount_opts;
1247 __le32 s_first_meta_bg; /* First metablock block group */
1248 __le32 s_mkfs_time; /* When the filesystem was created */
1249 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1250 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1251 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1252 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1253 __le32 s_free_blocks_count_hi; /* Free blocks count */
1254 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1255 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1256 __le32 s_flags; /* Miscellaneous flags */
1257 __le16 s_raid_stride; /* RAID stride */
1258 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1259 __le64 s_mmp_block; /* Block for multi-mount protection */
1260 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1261 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1262 __u8 s_checksum_type; /* metadata checksum algorithm used */
1263 __u8 s_encryption_level; /* versioning level for encryption */
1264 __u8 s_reserved_pad; /* Padding to next 32bits */
1265 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1266 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1267 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1268 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1269 snapshot's future use */
1270 __le32 s_snapshot_list; /* inode number of the head of the
1271 on-disk snapshot list */
1272 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1273 __le32 s_error_count; /* number of fs errors */
1274 __le32 s_first_error_time; /* first time an error happened */
1275 __le32 s_first_error_ino; /* inode involved in first error */
1276 __le64 s_first_error_block; /* block involved of first error */
1277 __u8 s_first_error_func[32]; /* function where the error happened */
1278 __le32 s_first_error_line; /* line number where error happened */
1279 __le32 s_last_error_time; /* most recent time of an error */
1280 __le32 s_last_error_ino; /* inode involved in last error */
1281 __le32 s_last_error_line; /* line number where error happened */
1282 __le64 s_last_error_block; /* block involved of last error */
1283 __u8 s_last_error_func[32]; /* function where the error happened */
1284 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1285 __u8 s_mount_opts[64];
1286 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1287 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1288 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1289 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
1290 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */
1291 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */
1292 __le32 s_lpf_ino; /* Location of the lost+found inode */
1293 __le32 s_prj_quota_inum; /* inode for tracking project quota */
1294 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */
1295 __le32 s_reserved[98]; /* Padding to the end of the block */
1296 __le32 s_checksum; /* crc32c(superblock) */
1297 };
1298
1299 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1300
1301 #ifdef __KERNEL__
1302
1303 /*
1304 * run-time mount flags
1305 */
1306 #define EXT4_MF_MNTDIR_SAMPLED 0x0001
1307 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1308 #define EXT4_MF_TEST_DUMMY_ENCRYPTION 0x0004
1309
1310 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1311 #define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \
1312 EXT4_MF_TEST_DUMMY_ENCRYPTION))
1313 #else
1314 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1315 #endif
1316
1317 /* Number of quota types we support */
1318 #define EXT4_MAXQUOTAS 3
1319
1320 /*
1321 * fourth extended-fs super-block data in memory
1322 */
1323 struct ext4_sb_info {
1324 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1325 unsigned long s_inodes_per_block;/* Number of inodes per block */
1326 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1327 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1328 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1329 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1330 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1331 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1332 ext4_group_t s_groups_count; /* Number of groups in the fs */
1333 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1334 unsigned long s_overhead; /* # of fs overhead clusters */
1335 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1336 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1337 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1338 struct buffer_head * s_sbh; /* Buffer containing the super block */
1339 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1340 struct buffer_head **s_group_desc;
1341 unsigned int s_mount_opt;
1342 unsigned int s_mount_opt2;
1343 unsigned int s_mount_flags;
1344 unsigned int s_def_mount_opt;
1345 ext4_fsblk_t s_sb_block;
1346 atomic64_t s_resv_clusters;
1347 kuid_t s_resuid;
1348 kgid_t s_resgid;
1349 unsigned short s_mount_state;
1350 unsigned short s_pad;
1351 int s_addr_per_block_bits;
1352 int s_desc_per_block_bits;
1353 int s_inode_size;
1354 int s_first_ino;
1355 unsigned int s_inode_readahead_blks;
1356 unsigned int s_inode_goal;
1357 u32 s_hash_seed[4];
1358 int s_def_hash_version;
1359 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1360 struct percpu_counter s_freeclusters_counter;
1361 struct percpu_counter s_freeinodes_counter;
1362 struct percpu_counter s_dirs_counter;
1363 struct percpu_counter s_dirtyclusters_counter;
1364 struct blockgroup_lock *s_blockgroup_lock;
1365 struct proc_dir_entry *s_proc;
1366 struct kobject s_kobj;
1367 struct completion s_kobj_unregister;
1368 struct super_block *s_sb;
1369
1370 /* Journaling */
1371 struct journal_s *s_journal;
1372 struct list_head s_orphan;
1373 struct mutex s_orphan_lock;
1374 unsigned long s_ext4_flags; /* Ext4 superblock flags */
1375 unsigned long s_commit_interval;
1376 u32 s_max_batch_time;
1377 u32 s_min_batch_time;
1378 struct block_device *journal_bdev;
1379 #ifdef CONFIG_QUOTA
1380 char *s_qf_names[EXT4_MAXQUOTAS]; /* Names of quota files with journalled quota */
1381 int s_jquota_fmt; /* Format of quota to use */
1382 #endif
1383 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1384 struct rb_root system_blks;
1385
1386 #ifdef EXTENTS_STATS
1387 /* ext4 extents stats */
1388 unsigned long s_ext_min;
1389 unsigned long s_ext_max;
1390 unsigned long s_depth_max;
1391 spinlock_t s_ext_stats_lock;
1392 unsigned long s_ext_blocks;
1393 unsigned long s_ext_extents;
1394 #endif
1395
1396 /* for buddy allocator */
1397 struct ext4_group_info ***s_group_info;
1398 struct inode *s_buddy_cache;
1399 spinlock_t s_md_lock;
1400 unsigned short *s_mb_offsets;
1401 unsigned int *s_mb_maxs;
1402 unsigned int s_group_info_size;
1403 unsigned int s_mb_free_pending;
1404 struct list_head s_freed_data_list; /* List of blocks to be freed
1405 after commit completed */
1406
1407 /* tunables */
1408 unsigned long s_stripe;
1409 unsigned int s_mb_stream_request;
1410 unsigned int s_mb_max_to_scan;
1411 unsigned int s_mb_min_to_scan;
1412 unsigned int s_mb_stats;
1413 unsigned int s_mb_order2_reqs;
1414 unsigned int s_mb_group_prealloc;
1415 unsigned int s_max_dir_size_kb;
1416 /* where last allocation was done - for stream allocation */
1417 unsigned long s_mb_last_group;
1418 unsigned long s_mb_last_start;
1419
1420 /* stats for buddy allocator */
1421 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1422 atomic_t s_bal_success; /* we found long enough chunks */
1423 atomic_t s_bal_allocated; /* in blocks */
1424 atomic_t s_bal_ex_scanned; /* total extents scanned */
1425 atomic_t s_bal_goals; /* goal hits */
1426 atomic_t s_bal_breaks; /* too long searches */
1427 atomic_t s_bal_2orders; /* 2^order hits */
1428 spinlock_t s_bal_lock;
1429 unsigned long s_mb_buddies_generated;
1430 unsigned long long s_mb_generation_time;
1431 atomic_t s_mb_lost_chunks;
1432 atomic_t s_mb_preallocated;
1433 atomic_t s_mb_discarded;
1434 atomic_t s_lock_busy;
1435
1436 /* locality groups */
1437 struct ext4_locality_group __percpu *s_locality_groups;
1438
1439 /* for write statistics */
1440 unsigned long s_sectors_written_start;
1441 u64 s_kbytes_written;
1442
1443 /* the size of zero-out chunk */
1444 unsigned int s_extent_max_zeroout_kb;
1445
1446 unsigned int s_log_groups_per_flex;
1447 struct flex_groups *s_flex_groups;
1448 ext4_group_t s_flex_groups_allocated;
1449
1450 /* workqueue for reserved extent conversions (buffered io) */
1451 struct workqueue_struct *rsv_conversion_wq;
1452
1453 /* timer for periodic error stats printing */
1454 struct timer_list s_err_report;
1455
1456 /* Lazy inode table initialization info */
1457 struct ext4_li_request *s_li_request;
1458 /* Wait multiplier for lazy initialization thread */
1459 unsigned int s_li_wait_mult;
1460
1461 /* Kernel thread for multiple mount protection */
1462 struct task_struct *s_mmp_tsk;
1463
1464 /* record the last minlen when FITRIM is called. */
1465 atomic_t s_last_trim_minblks;
1466
1467 /* Reference to checksum algorithm driver via cryptoapi */
1468 struct crypto_shash *s_chksum_driver;
1469
1470 /* Precomputed FS UUID checksum for seeding other checksums */
1471 __u32 s_csum_seed;
1472
1473 /* Reclaim extents from extent status tree */
1474 struct shrinker s_es_shrinker;
1475 struct list_head s_es_list; /* List of inodes with reclaimable extents */
1476 long s_es_nr_inode;
1477 struct ext4_es_stats s_es_stats;
1478 struct mb_cache *s_ea_block_cache;
1479 struct mb_cache *s_ea_inode_cache;
1480 spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1481
1482 /* Ratelimit ext4 messages. */
1483 struct ratelimit_state s_err_ratelimit_state;
1484 struct ratelimit_state s_warning_ratelimit_state;
1485 struct ratelimit_state s_msg_ratelimit_state;
1486
1487 /* Barrier between changing inodes' journal flags and writepages ops. */
1488 struct percpu_rw_semaphore s_journal_flag_rwsem;
1489 struct dax_device *s_daxdev;
1490 };
1491
1492 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1493 {
1494 return sb->s_fs_info;
1495 }
1496 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1497 {
1498 return container_of(inode, struct ext4_inode_info, vfs_inode);
1499 }
1500
1501 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1502 {
1503 return ino == EXT4_ROOT_INO ||
1504 ino == EXT4_USR_QUOTA_INO ||
1505 ino == EXT4_GRP_QUOTA_INO ||
1506 ino == EXT4_BOOT_LOADER_INO ||
1507 ino == EXT4_JOURNAL_INO ||
1508 ino == EXT4_RESIZE_INO ||
1509 (ino >= EXT4_FIRST_INO(sb) &&
1510 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1511 }
1512
1513 /*
1514 * Inode dynamic state flags
1515 */
1516 enum {
1517 EXT4_STATE_JDATA, /* journaled data exists */
1518 EXT4_STATE_NEW, /* inode is newly created */
1519 EXT4_STATE_XATTR, /* has in-inode xattrs */
1520 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1521 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1522 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1523 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
1524 EXT4_STATE_NEWENTRY, /* File just added to dir */
1525 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read
1526 nolocking */
1527 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1528 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
1529 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */
1530 };
1531
1532 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1533 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1534 { \
1535 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1536 } \
1537 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1538 { \
1539 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1540 } \
1541 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1542 { \
1543 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1544 }
1545
1546 /* Add these declarations here only so that these functions can be
1547 * found by name. Otherwise, they are very hard to locate. */
1548 static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1549 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1550 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1551 EXT4_INODE_BIT_FNS(flag, flags, 0)
1552
1553 /* Add these declarations here only so that these functions can be
1554 * found by name. Otherwise, they are very hard to locate. */
1555 static inline int ext4_test_inode_state(struct inode *inode, int bit);
1556 static inline void ext4_set_inode_state(struct inode *inode, int bit);
1557 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1558 #if (BITS_PER_LONG < 64)
1559 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1560
1561 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1562 {
1563 (ei)->i_state_flags = 0;
1564 }
1565 #else
1566 EXT4_INODE_BIT_FNS(state, flags, 32)
1567
1568 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1569 {
1570 /* We depend on the fact that callers will set i_flags */
1571 }
1572 #endif
1573 #else
1574 /* Assume that user mode programs are passing in an ext4fs superblock, not
1575 * a kernel struct super_block. This will allow us to call the feature-test
1576 * macros from user land. */
1577 #define EXT4_SB(sb) (sb)
1578 #endif
1579
1580 /*
1581 * Returns true if the inode is inode is encrypted
1582 */
1583 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1584
1585 /*
1586 * Codes for operating systems
1587 */
1588 #define EXT4_OS_LINUX 0
1589 #define EXT4_OS_HURD 1
1590 #define EXT4_OS_MASIX 2
1591 #define EXT4_OS_FREEBSD 3
1592 #define EXT4_OS_LITES 4
1593
1594 /*
1595 * Revision levels
1596 */
1597 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1598 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1599
1600 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1601 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1602
1603 #define EXT4_GOOD_OLD_INODE_SIZE 128
1604
1605 /*
1606 * Feature set definitions
1607 */
1608
1609 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1610 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1611 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1612 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1613 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1614 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1615 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
1616
1617 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1618 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1619 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1620 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1621 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1622 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1623 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1624 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1625 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1626 /*
1627 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1628 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1629 * all other data structures' checksums. However, the METADATA_CSUM and
1630 * GDT_CSUM bits are mutually exclusive.
1631 */
1632 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
1633 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000
1634 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000
1635
1636 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1637 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1638 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1639 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1640 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1641 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1642 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1643 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1644 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1645 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1646 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1647 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000
1648 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
1649 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
1650 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000
1651
1652 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
1653 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1654 { \
1655 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1656 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
1657 } \
1658 static inline void ext4_set_feature_##name(struct super_block *sb) \
1659 { \
1660 EXT4_SB(sb)->s_es->s_feature_compat |= \
1661 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1662 } \
1663 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1664 { \
1665 EXT4_SB(sb)->s_es->s_feature_compat &= \
1666 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1667 }
1668
1669 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
1670 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1671 { \
1672 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1673 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
1674 } \
1675 static inline void ext4_set_feature_##name(struct super_block *sb) \
1676 { \
1677 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
1678 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1679 } \
1680 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1681 { \
1682 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
1683 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1684 }
1685
1686 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
1687 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1688 { \
1689 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1690 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
1691 } \
1692 static inline void ext4_set_feature_##name(struct super_block *sb) \
1693 { \
1694 EXT4_SB(sb)->s_es->s_feature_incompat |= \
1695 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1696 } \
1697 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1698 { \
1699 EXT4_SB(sb)->s_es->s_feature_incompat &= \
1700 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1701 }
1702
1703 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC)
1704 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES)
1705 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL)
1706 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR)
1707 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE)
1708 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX)
1709 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2)
1710
1711 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER)
1712 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE)
1713 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR)
1714 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE)
1715 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM)
1716 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK)
1717 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE)
1718 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA)
1719 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC)
1720 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM)
1721 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY)
1722 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT)
1723
1724 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION)
1725 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE)
1726 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER)
1727 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV)
1728 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG)
1729 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS)
1730 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT)
1731 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP)
1732 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG)
1733 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE)
1734 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA)
1735 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED)
1736 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR)
1737 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA)
1738 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT)
1739
1740 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1741 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1742 EXT4_FEATURE_INCOMPAT_META_BG)
1743 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1744 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1745 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1746
1747 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1748 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1749 EXT4_FEATURE_INCOMPAT_RECOVER| \
1750 EXT4_FEATURE_INCOMPAT_META_BG)
1751 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1752 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1753 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1754
1755 #define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1756 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1757 EXT4_FEATURE_INCOMPAT_RECOVER| \
1758 EXT4_FEATURE_INCOMPAT_META_BG| \
1759 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1760 EXT4_FEATURE_INCOMPAT_64BIT| \
1761 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1762 EXT4_FEATURE_INCOMPAT_EA_INODE| \
1763 EXT4_FEATURE_INCOMPAT_MMP | \
1764 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
1765 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
1766 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \
1767 EXT4_FEATURE_INCOMPAT_LARGEDIR)
1768 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1769 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1770 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1771 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1772 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1773 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1774 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1775 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1776 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1777 EXT4_FEATURE_RO_COMPAT_QUOTA |\
1778 EXT4_FEATURE_RO_COMPAT_PROJECT)
1779
1780 #define EXTN_FEATURE_FUNCS(ver) \
1781 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
1782 { \
1783 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1784 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
1785 } \
1786 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
1787 { \
1788 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1789 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
1790 } \
1791 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
1792 { \
1793 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1794 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
1795 }
1796
1797 EXTN_FEATURE_FUNCS(2)
1798 EXTN_FEATURE_FUNCS(3)
1799 EXTN_FEATURE_FUNCS(4)
1800
1801 static inline bool ext4_has_compat_features(struct super_block *sb)
1802 {
1803 return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
1804 }
1805 static inline bool ext4_has_ro_compat_features(struct super_block *sb)
1806 {
1807 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
1808 }
1809 static inline bool ext4_has_incompat_features(struct super_block *sb)
1810 {
1811 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
1812 }
1813
1814 /*
1815 * Superblock flags
1816 */
1817 #define EXT4_FLAGS_RESIZING 0
1818 #define EXT4_FLAGS_SHUTDOWN 1
1819
1820 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi)
1821 {
1822 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags);
1823 }
1824
1825
1826 /*
1827 * Default values for user and/or group using reserved blocks
1828 */
1829 #define EXT4_DEF_RESUID 0
1830 #define EXT4_DEF_RESGID 0
1831
1832 /*
1833 * Default project ID
1834 */
1835 #define EXT4_DEF_PROJID 0
1836
1837 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
1838
1839 /*
1840 * Default mount options
1841 */
1842 #define EXT4_DEFM_DEBUG 0x0001
1843 #define EXT4_DEFM_BSDGROUPS 0x0002
1844 #define EXT4_DEFM_XATTR_USER 0x0004
1845 #define EXT4_DEFM_ACL 0x0008
1846 #define EXT4_DEFM_UID16 0x0010
1847 #define EXT4_DEFM_JMODE 0x0060
1848 #define EXT4_DEFM_JMODE_DATA 0x0020
1849 #define EXT4_DEFM_JMODE_ORDERED 0x0040
1850 #define EXT4_DEFM_JMODE_WBACK 0x0060
1851 #define EXT4_DEFM_NOBARRIER 0x0100
1852 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1853 #define EXT4_DEFM_DISCARD 0x0400
1854 #define EXT4_DEFM_NODELALLOC 0x0800
1855
1856 /*
1857 * Default journal batch times
1858 */
1859 #define EXT4_DEF_MIN_BATCH_TIME 0
1860 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1861
1862 /*
1863 * Minimum number of groups in a flexgroup before we separate out
1864 * directories into the first block group of a flexgroup
1865 */
1866 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1867
1868 /*
1869 * Structure of a directory entry
1870 */
1871 #define EXT4_NAME_LEN 255
1872
1873 struct ext4_dir_entry {
1874 __le32 inode; /* Inode number */
1875 __le16 rec_len; /* Directory entry length */
1876 __le16 name_len; /* Name length */
1877 char name[EXT4_NAME_LEN]; /* File name */
1878 };
1879
1880 /*
1881 * The new version of the directory entry. Since EXT4 structures are
1882 * stored in intel byte order, and the name_len field could never be
1883 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1884 * file_type field.
1885 */
1886 struct ext4_dir_entry_2 {
1887 __le32 inode; /* Inode number */
1888 __le16 rec_len; /* Directory entry length */
1889 __u8 name_len; /* Name length */
1890 __u8 file_type;
1891 char name[EXT4_NAME_LEN]; /* File name */
1892 };
1893
1894 /*
1895 * This is a bogus directory entry at the end of each leaf block that
1896 * records checksums.
1897 */
1898 struct ext4_dir_entry_tail {
1899 __le32 det_reserved_zero1; /* Pretend to be unused */
1900 __le16 det_rec_len; /* 12 */
1901 __u8 det_reserved_zero2; /* Zero name length */
1902 __u8 det_reserved_ft; /* 0xDE, fake file type */
1903 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
1904 };
1905
1906 #define EXT4_DIRENT_TAIL(block, blocksize) \
1907 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1908 ((blocksize) - \
1909 sizeof(struct ext4_dir_entry_tail))))
1910
1911 /*
1912 * Ext4 directory file types. Only the low 3 bits are used. The
1913 * other bits are reserved for now.
1914 */
1915 #define EXT4_FT_UNKNOWN 0
1916 #define EXT4_FT_REG_FILE 1
1917 #define EXT4_FT_DIR 2
1918 #define EXT4_FT_CHRDEV 3
1919 #define EXT4_FT_BLKDEV 4
1920 #define EXT4_FT_FIFO 5
1921 #define EXT4_FT_SOCK 6
1922 #define EXT4_FT_SYMLINK 7
1923
1924 #define EXT4_FT_MAX 8
1925
1926 #define EXT4_FT_DIR_CSUM 0xDE
1927
1928 /*
1929 * EXT4_DIR_PAD defines the directory entries boundaries
1930 *
1931 * NOTE: It must be a multiple of 4
1932 */
1933 #define EXT4_DIR_PAD 4
1934 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1935 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1936 ~EXT4_DIR_ROUND)
1937 #define EXT4_MAX_REC_LEN ((1<<16)-1)
1938
1939 /*
1940 * If we ever get support for fs block sizes > page_size, we'll need
1941 * to remove the #if statements in the next two functions...
1942 */
1943 static inline unsigned int
1944 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1945 {
1946 unsigned len = le16_to_cpu(dlen);
1947
1948 #if (PAGE_SIZE >= 65536)
1949 if (len == EXT4_MAX_REC_LEN || len == 0)
1950 return blocksize;
1951 return (len & 65532) | ((len & 3) << 16);
1952 #else
1953 return len;
1954 #endif
1955 }
1956
1957 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1958 {
1959 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1960 BUG();
1961 #if (PAGE_SIZE >= 65536)
1962 if (len < 65536)
1963 return cpu_to_le16(len);
1964 if (len == blocksize) {
1965 if (blocksize == 65536)
1966 return cpu_to_le16(EXT4_MAX_REC_LEN);
1967 else
1968 return cpu_to_le16(0);
1969 }
1970 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1971 #else
1972 return cpu_to_le16(len);
1973 #endif
1974 }
1975
1976 /*
1977 * Hash Tree Directory indexing
1978 * (c) Daniel Phillips, 2001
1979 */
1980
1981 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
1982 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
1983 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \
1984 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir)))
1985 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1986
1987 /* Legal values for the dx_root hash_version field: */
1988
1989 #define DX_HASH_LEGACY 0
1990 #define DX_HASH_HALF_MD4 1
1991 #define DX_HASH_TEA 2
1992 #define DX_HASH_LEGACY_UNSIGNED 3
1993 #define DX_HASH_HALF_MD4_UNSIGNED 4
1994 #define DX_HASH_TEA_UNSIGNED 5
1995
1996 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
1997 const void *address, unsigned int length)
1998 {
1999 struct {
2000 struct shash_desc shash;
2001 char ctx[4];
2002 } desc;
2003
2004 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2005
2006 desc.shash.tfm = sbi->s_chksum_driver;
2007 desc.shash.flags = 0;
2008 *(u32 *)desc.ctx = crc;
2009
2010 BUG_ON(crypto_shash_update(&desc.shash, address, length));
2011
2012 return *(u32 *)desc.ctx;
2013 }
2014
2015 #ifdef __KERNEL__
2016
2017 /* hash info structure used by the directory hash */
2018 struct dx_hash_info
2019 {
2020 u32 hash;
2021 u32 minor_hash;
2022 int hash_version;
2023 u32 *seed;
2024 };
2025
2026
2027 /* 32 and 64 bit signed EOF for dx directories */
2028 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
2029 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
2030
2031
2032 /*
2033 * Control parameters used by ext4_htree_next_block
2034 */
2035 #define HASH_NB_ALWAYS 1
2036
2037 struct ext4_filename {
2038 const struct qstr *usr_fname;
2039 struct fscrypt_str disk_name;
2040 struct dx_hash_info hinfo;
2041 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2042 struct fscrypt_str crypto_buf;
2043 #endif
2044 };
2045
2046 #define fname_name(p) ((p)->disk_name.name)
2047 #define fname_len(p) ((p)->disk_name.len)
2048
2049 /*
2050 * Describe an inode's exact location on disk and in memory
2051 */
2052 struct ext4_iloc
2053 {
2054 struct buffer_head *bh;
2055 unsigned long offset;
2056 ext4_group_t block_group;
2057 };
2058
2059 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2060 {
2061 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2062 }
2063
2064 static inline bool ext4_is_quota_file(struct inode *inode)
2065 {
2066 return IS_NOQUOTA(inode) &&
2067 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL);
2068 }
2069
2070 /*
2071 * This structure is stuffed into the struct file's private_data field
2072 * for directories. It is where we put information so that we can do
2073 * readdir operations in hash tree order.
2074 */
2075 struct dir_private_info {
2076 struct rb_root root;
2077 struct rb_node *curr_node;
2078 struct fname *extra_fname;
2079 loff_t last_pos;
2080 __u32 curr_hash;
2081 __u32 curr_minor_hash;
2082 __u32 next_hash;
2083 };
2084
2085 /* calculate the first block number of the group */
2086 static inline ext4_fsblk_t
2087 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2088 {
2089 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2090 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2091 }
2092
2093 /*
2094 * Special error return code only used by dx_probe() and its callers.
2095 */
2096 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
2097
2098 /* htree levels for ext4 */
2099 #define EXT4_HTREE_LEVEL_COMPAT 2
2100 #define EXT4_HTREE_LEVEL 3
2101
2102 static inline int ext4_dir_htree_level(struct super_block *sb)
2103 {
2104 return ext4_has_feature_largedir(sb) ?
2105 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT;
2106 }
2107
2108 /*
2109 * Timeout and state flag for lazy initialization inode thread.
2110 */
2111 #define EXT4_DEF_LI_WAIT_MULT 10
2112 #define EXT4_DEF_LI_MAX_START_DELAY 5
2113 #define EXT4_LAZYINIT_QUIT 0x0001
2114 #define EXT4_LAZYINIT_RUNNING 0x0002
2115
2116 /*
2117 * Lazy inode table initialization info
2118 */
2119 struct ext4_lazy_init {
2120 unsigned long li_state;
2121 struct list_head li_request_list;
2122 struct mutex li_list_mtx;
2123 };
2124
2125 struct ext4_li_request {
2126 struct super_block *lr_super;
2127 struct ext4_sb_info *lr_sbi;
2128 ext4_group_t lr_next_group;
2129 struct list_head lr_request;
2130 unsigned long lr_next_sched;
2131 unsigned long lr_timeout;
2132 };
2133
2134 struct ext4_features {
2135 struct kobject f_kobj;
2136 struct completion f_kobj_unregister;
2137 };
2138
2139 /*
2140 * This structure will be used for multiple mount protection. It will be
2141 * written into the block number saved in the s_mmp_block field in the
2142 * superblock. Programs that check MMP should assume that if
2143 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2144 * to use the filesystem, regardless of how old the timestamp is.
2145 */
2146 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
2147 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2148 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
2149 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
2150
2151 struct mmp_struct {
2152 __le32 mmp_magic; /* Magic number for MMP */
2153 __le32 mmp_seq; /* Sequence no. updated periodically */
2154
2155 /*
2156 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2157 * purposes and do not affect the correctness of the algorithm
2158 */
2159 __le64 mmp_time; /* Time last updated */
2160 char mmp_nodename[64]; /* Node which last updated MMP block */
2161 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
2162
2163 /*
2164 * mmp_check_interval is used to verify if the MMP block has been
2165 * updated on the block device. The value is updated based on the
2166 * maximum time to write the MMP block during an update cycle.
2167 */
2168 __le16 mmp_check_interval;
2169
2170 __le16 mmp_pad1;
2171 __le32 mmp_pad2[226];
2172 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
2173 };
2174
2175 /* arguments passed to the mmp thread */
2176 struct mmpd_data {
2177 struct buffer_head *bh; /* bh from initial read_mmp_block() */
2178 struct super_block *sb; /* super block of the fs */
2179 };
2180
2181 /*
2182 * Check interval multiplier
2183 * The MMP block is written every update interval and initially checked every
2184 * update interval x the multiplier (the value is then adapted based on the
2185 * write latency). The reason is that writes can be delayed under load and we
2186 * don't want readers to incorrectly assume that the filesystem is no longer
2187 * in use.
2188 */
2189 #define EXT4_MMP_CHECK_MULT 2UL
2190
2191 /*
2192 * Minimum interval for MMP checking in seconds.
2193 */
2194 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
2195
2196 /*
2197 * Maximum interval for MMP checking in seconds.
2198 */
2199 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
2200
2201 /*
2202 * Function prototypes
2203 */
2204
2205 /*
2206 * Ok, these declarations are also in <linux/kernel.h> but none of the
2207 * ext4 source programs needs to include it so they are duplicated here.
2208 */
2209 # define NORET_TYPE /**/
2210 # define ATTRIB_NORET __attribute__((noreturn))
2211 # define NORET_AND noreturn,
2212
2213 /* bitmap.c */
2214 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2215 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2216 struct ext4_group_desc *gdp,
2217 struct buffer_head *bh, int sz);
2218 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2219 struct ext4_group_desc *gdp,
2220 struct buffer_head *bh, int sz);
2221 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2222 struct ext4_group_desc *gdp,
2223 struct buffer_head *bh);
2224 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2225 struct ext4_group_desc *gdp,
2226 struct buffer_head *bh);
2227
2228 /* balloc.c */
2229 extern void ext4_get_group_no_and_offset(struct super_block *sb,
2230 ext4_fsblk_t blocknr,
2231 ext4_group_t *blockgrpp,
2232 ext4_grpblk_t *offsetp);
2233 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2234 ext4_fsblk_t block);
2235
2236 extern unsigned int ext4_block_group(struct super_block *sb,
2237 ext4_fsblk_t blocknr);
2238 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2239 ext4_fsblk_t blocknr);
2240 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2241 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2242 ext4_group_t group);
2243 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2244 ext4_fsblk_t goal,
2245 unsigned int flags,
2246 unsigned long *count,
2247 int *errp);
2248 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2249 s64 nclusters, unsigned int flags);
2250 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2251 extern void ext4_check_blocks_bitmap(struct super_block *);
2252 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2253 ext4_group_t block_group,
2254 struct buffer_head ** bh);
2255 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2256
2257 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2258 ext4_group_t block_group);
2259 extern int ext4_wait_block_bitmap(struct super_block *sb,
2260 ext4_group_t block_group,
2261 struct buffer_head *bh);
2262 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2263 ext4_group_t block_group);
2264 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2265 ext4_group_t block_group,
2266 struct ext4_group_desc *gdp);
2267 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2268
2269 static inline bool ext4_encrypted_inode(struct inode *inode)
2270 {
2271 return ext4_test_inode_flag(inode, EXT4_INODE_ENCRYPT);
2272 }
2273
2274 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2275 static inline int ext4_fname_setup_filename(struct inode *dir,
2276 const struct qstr *iname,
2277 int lookup, struct ext4_filename *fname)
2278 {
2279 struct fscrypt_name name;
2280 int err;
2281
2282 memset(fname, 0, sizeof(struct ext4_filename));
2283
2284 err = fscrypt_setup_filename(dir, iname, lookup, &name);
2285
2286 fname->usr_fname = name.usr_fname;
2287 fname->disk_name = name.disk_name;
2288 fname->hinfo.hash = name.hash;
2289 fname->hinfo.minor_hash = name.minor_hash;
2290 fname->crypto_buf = name.crypto_buf;
2291 return err;
2292 }
2293
2294 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2295 {
2296 struct fscrypt_name name;
2297
2298 name.crypto_buf = fname->crypto_buf;
2299 fscrypt_free_filename(&name);
2300
2301 fname->crypto_buf.name = NULL;
2302 fname->usr_fname = NULL;
2303 fname->disk_name.name = NULL;
2304 }
2305 #else
2306 static inline int ext4_fname_setup_filename(struct inode *dir,
2307 const struct qstr *iname,
2308 int lookup, struct ext4_filename *fname)
2309 {
2310 fname->usr_fname = iname;
2311 fname->disk_name.name = (unsigned char *) iname->name;
2312 fname->disk_name.len = iname->len;
2313 return 0;
2314 }
2315 static inline void ext4_fname_free_filename(struct ext4_filename *fname) { }
2316
2317 #endif
2318
2319 /* dir.c */
2320 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2321 struct file *,
2322 struct ext4_dir_entry_2 *,
2323 struct buffer_head *, char *, int,
2324 unsigned int);
2325 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2326 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2327 (de), (bh), (buf), (size), (offset)))
2328 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2329 __u32 minor_hash,
2330 struct ext4_dir_entry_2 *dirent,
2331 struct fscrypt_str *ent_name);
2332 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2333 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2334 struct buffer_head *bh,
2335 void *buf, int buf_size,
2336 struct ext4_filename *fname,
2337 struct ext4_dir_entry_2 **dest_de);
2338 void ext4_insert_dentry(struct inode *inode,
2339 struct ext4_dir_entry_2 *de,
2340 int buf_size,
2341 struct ext4_filename *fname);
2342 static inline void ext4_update_dx_flag(struct inode *inode)
2343 {
2344 if (!ext4_has_feature_dir_index(inode->i_sb))
2345 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2346 }
2347 static const unsigned char ext4_filetype_table[] = {
2348 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2349 };
2350
2351 static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2352 {
2353 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2354 return DT_UNKNOWN;
2355
2356 return ext4_filetype_table[filetype];
2357 }
2358 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2359 void *buf, int buf_size);
2360
2361 /* fsync.c */
2362 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2363
2364 /* hash.c */
2365 extern int ext4fs_dirhash(const char *name, int len, struct
2366 dx_hash_info *hinfo);
2367
2368 /* ialloc.c */
2369 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2370 const struct qstr *qstr, __u32 goal,
2371 uid_t *owner, __u32 i_flags,
2372 int handle_type, unsigned int line_no,
2373 int nblocks);
2374
2375 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \
2376 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2377 i_flags, 0, 0, 0)
2378 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2379 type, nblocks) \
2380 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2381 0, (type), __LINE__, (nblocks))
2382
2383
2384 extern void ext4_free_inode(handle_t *, struct inode *);
2385 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2386 extern unsigned long ext4_count_free_inodes(struct super_block *);
2387 extern unsigned long ext4_count_dirs(struct super_block *);
2388 extern void ext4_check_inodes_bitmap(struct super_block *);
2389 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2390 extern int ext4_init_inode_table(struct super_block *sb,
2391 ext4_group_t group, int barrier);
2392 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2393
2394 /* mballoc.c */
2395 extern const struct file_operations ext4_seq_mb_groups_fops;
2396 extern long ext4_mb_stats;
2397 extern long ext4_mb_max_to_scan;
2398 extern int ext4_mb_init(struct super_block *);
2399 extern int ext4_mb_release(struct super_block *);
2400 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2401 struct ext4_allocation_request *, int *);
2402 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2403 extern void ext4_discard_preallocations(struct inode *);
2404 extern int __init ext4_init_mballoc(void);
2405 extern void ext4_exit_mballoc(void);
2406 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2407 struct buffer_head *bh, ext4_fsblk_t block,
2408 unsigned long count, int flags);
2409 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2410 ext4_group_t ngroups);
2411 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2412 ext4_group_t i, struct ext4_group_desc *desc);
2413 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2414 ext4_fsblk_t block, unsigned long count);
2415 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2416 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid);
2417
2418 /* inode.c */
2419 int ext4_inode_is_fast_symlink(struct inode *inode);
2420 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2421 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2422 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count,
2423 bool wait, struct buffer_head **bhs);
2424 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2425 struct buffer_head *bh_result, int create);
2426 int ext4_get_block(struct inode *inode, sector_t iblock,
2427 struct buffer_head *bh_result, int create);
2428 int ext4_dio_get_block(struct inode *inode, sector_t iblock,
2429 struct buffer_head *bh_result, int create);
2430 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2431 struct buffer_head *bh, int create);
2432 int ext4_walk_page_buffers(handle_t *handle,
2433 struct buffer_head *head,
2434 unsigned from,
2435 unsigned to,
2436 int *partial,
2437 int (*fn)(handle_t *handle,
2438 struct buffer_head *bh));
2439 int do_journal_get_write_access(handle_t *handle,
2440 struct buffer_head *bh);
2441 #define FALL_BACK_TO_NONDELALLOC 1
2442 #define CONVERT_INLINE_DATA 2
2443
2444 extern struct inode *ext4_iget(struct super_block *, unsigned long);
2445 extern struct inode *ext4_iget_normal(struct super_block *, unsigned long);
2446 extern int ext4_write_inode(struct inode *, struct writeback_control *);
2447 extern int ext4_setattr(struct dentry *, struct iattr *);
2448 extern int ext4_getattr(const struct path *, struct kstat *, u32, unsigned int);
2449 extern void ext4_evict_inode(struct inode *);
2450 extern void ext4_clear_inode(struct inode *);
2451 extern int ext4_file_getattr(const struct path *, struct kstat *, u32, unsigned int);
2452 extern int ext4_sync_inode(handle_t *, struct inode *);
2453 extern void ext4_dirty_inode(struct inode *, int);
2454 extern int ext4_change_inode_journal_flag(struct inode *, int);
2455 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2456 extern int ext4_inode_attach_jinode(struct inode *inode);
2457 extern int ext4_can_truncate(struct inode *inode);
2458 extern int ext4_truncate(struct inode *);
2459 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2460 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
2461 extern void ext4_set_inode_flags(struct inode *);
2462 extern int ext4_alloc_da_blocks(struct inode *inode);
2463 extern void ext4_set_aops(struct inode *inode);
2464 extern int ext4_writepage_trans_blocks(struct inode *);
2465 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2466 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2467 loff_t lstart, loff_t lend);
2468 extern int ext4_page_mkwrite(struct vm_fault *vmf);
2469 extern int ext4_filemap_fault(struct vm_fault *vmf);
2470 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2471 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
2472 extern void ext4_da_update_reserve_space(struct inode *inode,
2473 int used, int quota_claim);
2474 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
2475 ext4_fsblk_t pblk, ext4_lblk_t len);
2476
2477 /* indirect.c */
2478 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2479 struct ext4_map_blocks *map, int flags);
2480 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2481 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2482 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2483 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2484 ext4_lblk_t start, ext4_lblk_t end);
2485
2486 /* ioctl.c */
2487 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2488 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2489
2490 /* migrate.c */
2491 extern int ext4_ext_migrate(struct inode *);
2492 extern int ext4_ind_migrate(struct inode *inode);
2493
2494 /* namei.c */
2495 extern int ext4_dirent_csum_verify(struct inode *inode,
2496 struct ext4_dir_entry *dirent);
2497 extern int ext4_orphan_add(handle_t *, struct inode *);
2498 extern int ext4_orphan_del(handle_t *, struct inode *);
2499 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2500 __u32 start_minor_hash, __u32 *next_hash);
2501 extern int ext4_search_dir(struct buffer_head *bh,
2502 char *search_buf,
2503 int buf_size,
2504 struct inode *dir,
2505 struct ext4_filename *fname,
2506 unsigned int offset,
2507 struct ext4_dir_entry_2 **res_dir);
2508 extern int ext4_generic_delete_entry(handle_t *handle,
2509 struct inode *dir,
2510 struct ext4_dir_entry_2 *de_del,
2511 struct buffer_head *bh,
2512 void *entry_buf,
2513 int buf_size,
2514 int csum_size);
2515 extern bool ext4_empty_dir(struct inode *inode);
2516
2517 /* resize.c */
2518 extern int ext4_group_add(struct super_block *sb,
2519 struct ext4_new_group_data *input);
2520 extern int ext4_group_extend(struct super_block *sb,
2521 struct ext4_super_block *es,
2522 ext4_fsblk_t n_blocks_count);
2523 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2524
2525 /* super.c */
2526 extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
2527 extern int ext4_calculate_overhead(struct super_block *sb);
2528 extern void ext4_superblock_csum_set(struct super_block *sb);
2529 extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2530 extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2531 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2532 ext4_group_t ngroup);
2533 extern const char *ext4_decode_error(struct super_block *sb, int errno,
2534 char nbuf[16]);
2535
2536 extern __printf(4, 5)
2537 void __ext4_error(struct super_block *, const char *, unsigned int,
2538 const char *, ...);
2539 extern __printf(5, 6)
2540 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2541 const char *, ...);
2542 extern __printf(5, 6)
2543 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2544 const char *, ...);
2545 extern void __ext4_std_error(struct super_block *, const char *,
2546 unsigned int, int);
2547 extern __printf(4, 5)
2548 void __ext4_abort(struct super_block *, const char *, unsigned int,
2549 const char *, ...);
2550 extern __printf(4, 5)
2551 void __ext4_warning(struct super_block *, const char *, unsigned int,
2552 const char *, ...);
2553 extern __printf(4, 5)
2554 void __ext4_warning_inode(const struct inode *inode, const char *function,
2555 unsigned int line, const char *fmt, ...);
2556 extern __printf(3, 4)
2557 void __ext4_msg(struct super_block *, const char *, const char *, ...);
2558 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2559 const char *, unsigned int, const char *);
2560 extern __printf(7, 8)
2561 void __ext4_grp_locked_error(const char *, unsigned int,
2562 struct super_block *, ext4_group_t,
2563 unsigned long, ext4_fsblk_t,
2564 const char *, ...);
2565
2566 #define EXT4_ERROR_INODE(inode, fmt, a...) \
2567 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
2568
2569 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
2570 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
2571
2572 #define EXT4_ERROR_FILE(file, block, fmt, a...) \
2573 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
2574
2575 #ifdef CONFIG_PRINTK
2576
2577 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2578 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__)
2579 #define ext4_error_file(file, func, line, block, fmt, ...) \
2580 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2581 #define ext4_error(sb, fmt, ...) \
2582 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2583 #define ext4_abort(sb, fmt, ...) \
2584 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2585 #define ext4_warning(sb, fmt, ...) \
2586 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2587 #define ext4_warning_inode(inode, fmt, ...) \
2588 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
2589 #define ext4_msg(sb, level, fmt, ...) \
2590 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2591 #define dump_mmp_msg(sb, mmp, msg) \
2592 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2593 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2594 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2595 fmt, ##__VA_ARGS__)
2596
2597 #else
2598
2599 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2600 do { \
2601 no_printk(fmt, ##__VA_ARGS__); \
2602 __ext4_error_inode(inode, "", 0, block, " "); \
2603 } while (0)
2604 #define ext4_error_file(file, func, line, block, fmt, ...) \
2605 do { \
2606 no_printk(fmt, ##__VA_ARGS__); \
2607 __ext4_error_file(file, "", 0, block, " "); \
2608 } while (0)
2609 #define ext4_error(sb, fmt, ...) \
2610 do { \
2611 no_printk(fmt, ##__VA_ARGS__); \
2612 __ext4_error(sb, "", 0, " "); \
2613 } while (0)
2614 #define ext4_abort(sb, fmt, ...) \
2615 do { \
2616 no_printk(fmt, ##__VA_ARGS__); \
2617 __ext4_abort(sb, "", 0, " "); \
2618 } while (0)
2619 #define ext4_warning(sb, fmt, ...) \
2620 do { \
2621 no_printk(fmt, ##__VA_ARGS__); \
2622 __ext4_warning(sb, "", 0, " "); \
2623 } while (0)
2624 #define ext4_warning_inode(inode, fmt, ...) \
2625 do { \
2626 no_printk(fmt, ##__VA_ARGS__); \
2627 __ext4_warning_inode(inode, "", 0, " "); \
2628 } while (0)
2629 #define ext4_msg(sb, level, fmt, ...) \
2630 do { \
2631 no_printk(fmt, ##__VA_ARGS__); \
2632 __ext4_msg(sb, "", " "); \
2633 } while (0)
2634 #define dump_mmp_msg(sb, mmp, msg) \
2635 __dump_mmp_msg(sb, mmp, "", 0, "")
2636 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2637 do { \
2638 no_printk(fmt, ##__VA_ARGS__); \
2639 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
2640 } while (0)
2641
2642 #endif
2643
2644 extern void ext4_update_dynamic_rev(struct super_block *sb);
2645 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2646 __u32 compat);
2647 extern int ext4_update_rocompat_feature(handle_t *handle,
2648 struct super_block *sb, __u32 rocompat);
2649 extern int ext4_update_incompat_feature(handle_t *handle,
2650 struct super_block *sb, __u32 incompat);
2651 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2652 struct ext4_group_desc *bg);
2653 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2654 struct ext4_group_desc *bg);
2655 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2656 struct ext4_group_desc *bg);
2657 extern __u32 ext4_free_group_clusters(struct super_block *sb,
2658 struct ext4_group_desc *bg);
2659 extern __u32 ext4_free_inodes_count(struct super_block *sb,
2660 struct ext4_group_desc *bg);
2661 extern __u32 ext4_used_dirs_count(struct super_block *sb,
2662 struct ext4_group_desc *bg);
2663 extern __u32 ext4_itable_unused_count(struct super_block *sb,
2664 struct ext4_group_desc *bg);
2665 extern void ext4_block_bitmap_set(struct super_block *sb,
2666 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2667 extern void ext4_inode_bitmap_set(struct super_block *sb,
2668 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2669 extern void ext4_inode_table_set(struct super_block *sb,
2670 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2671 extern void ext4_free_group_clusters_set(struct super_block *sb,
2672 struct ext4_group_desc *bg,
2673 __u32 count);
2674 extern void ext4_free_inodes_set(struct super_block *sb,
2675 struct ext4_group_desc *bg, __u32 count);
2676 extern void ext4_used_dirs_set(struct super_block *sb,
2677 struct ext4_group_desc *bg, __u32 count);
2678 extern void ext4_itable_unused_set(struct super_block *sb,
2679 struct ext4_group_desc *bg, __u32 count);
2680 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2681 struct ext4_group_desc *gdp);
2682 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2683 struct ext4_group_desc *gdp);
2684 extern int ext4_register_li_request(struct super_block *sb,
2685 ext4_group_t first_not_zeroed);
2686
2687 static inline int ext4_has_metadata_csum(struct super_block *sb)
2688 {
2689 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
2690 !EXT4_SB(sb)->s_chksum_driver);
2691
2692 return ext4_has_feature_metadata_csum(sb) &&
2693 (EXT4_SB(sb)->s_chksum_driver != NULL);
2694 }
2695
2696 static inline int ext4_has_group_desc_csum(struct super_block *sb)
2697 {
2698 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb);
2699 }
2700
2701 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2702 {
2703 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2704 le32_to_cpu(es->s_blocks_count_lo);
2705 }
2706
2707 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2708 {
2709 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2710 le32_to_cpu(es->s_r_blocks_count_lo);
2711 }
2712
2713 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2714 {
2715 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2716 le32_to_cpu(es->s_free_blocks_count_lo);
2717 }
2718
2719 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2720 ext4_fsblk_t blk)
2721 {
2722 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2723 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2724 }
2725
2726 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2727 ext4_fsblk_t blk)
2728 {
2729 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2730 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2731 }
2732
2733 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2734 ext4_fsblk_t blk)
2735 {
2736 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2737 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2738 }
2739
2740 static inline loff_t ext4_isize(struct super_block *sb,
2741 struct ext4_inode *raw_inode)
2742 {
2743 if (ext4_has_feature_largedir(sb) ||
2744 S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2745 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2746 le32_to_cpu(raw_inode->i_size_lo);
2747
2748 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2749 }
2750
2751 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2752 {
2753 raw_inode->i_size_lo = cpu_to_le32(i_size);
2754 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2755 }
2756
2757 static inline
2758 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2759 ext4_group_t group)
2760 {
2761 struct ext4_group_info ***grp_info;
2762 long indexv, indexh;
2763 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
2764 grp_info = EXT4_SB(sb)->s_group_info;
2765 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2766 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2767 return grp_info[indexv][indexh];
2768 }
2769
2770 /*
2771 * Reading s_groups_count requires using smp_rmb() afterwards. See
2772 * the locking protocol documented in the comments of ext4_group_add()
2773 * in resize.c
2774 */
2775 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2776 {
2777 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
2778
2779 smp_rmb();
2780 return ngroups;
2781 }
2782
2783 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2784 ext4_group_t block_group)
2785 {
2786 return block_group >> sbi->s_log_groups_per_flex;
2787 }
2788
2789 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2790 {
2791 return 1 << sbi->s_log_groups_per_flex;
2792 }
2793
2794 #define ext4_std_error(sb, errno) \
2795 do { \
2796 if ((errno)) \
2797 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
2798 } while (0)
2799
2800 #ifdef CONFIG_SMP
2801 /* Each CPU can accumulate percpu_counter_batch clusters in their local
2802 * counters. So we need to make sure we have free clusters more
2803 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
2804 */
2805 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2806 #else
2807 #define EXT4_FREECLUSTERS_WATERMARK 0
2808 #endif
2809
2810 /* Update i_disksize. Requires i_mutex to avoid races with truncate */
2811 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2812 {
2813 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
2814 !inode_is_locked(inode));
2815 down_write(&EXT4_I(inode)->i_data_sem);
2816 if (newsize > EXT4_I(inode)->i_disksize)
2817 EXT4_I(inode)->i_disksize = newsize;
2818 up_write(&EXT4_I(inode)->i_data_sem);
2819 }
2820
2821 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
2822 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
2823 {
2824 int changed = 0;
2825
2826 if (newsize > inode->i_size) {
2827 i_size_write(inode, newsize);
2828 changed = 1;
2829 }
2830 if (newsize > EXT4_I(inode)->i_disksize) {
2831 ext4_update_i_disksize(inode, newsize);
2832 changed |= 2;
2833 }
2834 return changed;
2835 }
2836
2837 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
2838 loff_t len);
2839
2840 struct ext4_group_info {
2841 unsigned long bb_state;
2842 struct rb_root bb_free_root;
2843 ext4_grpblk_t bb_first_free; /* first free block */
2844 ext4_grpblk_t bb_free; /* total free blocks */
2845 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
2846 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
2847 struct list_head bb_prealloc_list;
2848 #ifdef DOUBLE_CHECK
2849 void *bb_bitmap;
2850 #endif
2851 struct rw_semaphore alloc_sem;
2852 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
2853 * regions, index is order.
2854 * bb_counters[3] = 5 means
2855 * 5 free 8-block regions. */
2856 };
2857
2858 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
2859 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
2860 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
2861 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
2862
2863 #define EXT4_MB_GRP_NEED_INIT(grp) \
2864 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2865 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
2866 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2867 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
2868 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2869
2870 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
2871 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2872 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
2873 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2874 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
2875 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2876
2877 #define EXT4_MAX_CONTENTION 8
2878 #define EXT4_CONTENTION_THRESHOLD 2
2879
2880 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2881 ext4_group_t group)
2882 {
2883 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2884 }
2885
2886 /*
2887 * Returns true if the filesystem is busy enough that attempts to
2888 * access the block group locks has run into contention.
2889 */
2890 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2891 {
2892 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2893 }
2894
2895 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2896 {
2897 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2898 if (spin_trylock(lock))
2899 /*
2900 * We're able to grab the lock right away, so drop the
2901 * lock contention counter.
2902 */
2903 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2904 else {
2905 /*
2906 * The lock is busy, so bump the contention counter,
2907 * and then wait on the spin lock.
2908 */
2909 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2910 EXT4_MAX_CONTENTION);
2911 spin_lock(lock);
2912 }
2913 }
2914
2915 static inline void ext4_unlock_group(struct super_block *sb,
2916 ext4_group_t group)
2917 {
2918 spin_unlock(ext4_group_lock_ptr(sb, group));
2919 }
2920
2921 /*
2922 * Block validity checking
2923 */
2924 #define ext4_check_indirect_blockref(inode, bh) \
2925 ext4_check_blockref(__func__, __LINE__, inode, \
2926 (__le32 *)(bh)->b_data, \
2927 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2928
2929 #define ext4_ind_check_inode(inode) \
2930 ext4_check_blockref(__func__, __LINE__, inode, \
2931 EXT4_I(inode)->i_data, \
2932 EXT4_NDIR_BLOCKS)
2933
2934 /*
2935 * Inodes and files operations
2936 */
2937
2938 /* dir.c */
2939 extern const struct file_operations ext4_dir_operations;
2940
2941 /* file.c */
2942 extern const struct inode_operations ext4_file_inode_operations;
2943 extern const struct file_operations ext4_file_operations;
2944 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
2945
2946 /* inline.c */
2947 extern int ext4_get_max_inline_size(struct inode *inode);
2948 extern int ext4_find_inline_data_nolock(struct inode *inode);
2949 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
2950 unsigned int len);
2951 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
2952
2953 extern int ext4_readpage_inline(struct inode *inode, struct page *page);
2954 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
2955 struct inode *inode,
2956 loff_t pos, unsigned len,
2957 unsigned flags,
2958 struct page **pagep);
2959 extern int ext4_write_inline_data_end(struct inode *inode,
2960 loff_t pos, unsigned len,
2961 unsigned copied,
2962 struct page *page);
2963 extern struct buffer_head *
2964 ext4_journalled_write_inline_data(struct inode *inode,
2965 unsigned len,
2966 struct page *page);
2967 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
2968 struct inode *inode,
2969 loff_t pos, unsigned len,
2970 unsigned flags,
2971 struct page **pagep,
2972 void **fsdata);
2973 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
2974 unsigned len, unsigned copied,
2975 struct page *page);
2976 extern int ext4_try_add_inline_entry(handle_t *handle,
2977 struct ext4_filename *fname,
2978 struct inode *dir, struct inode *inode);
2979 extern int ext4_try_create_inline_dir(handle_t *handle,
2980 struct inode *parent,
2981 struct inode *inode);
2982 extern int ext4_read_inline_dir(struct file *filp,
2983 struct dir_context *ctx,
2984 int *has_inline_data);
2985 extern int htree_inlinedir_to_tree(struct file *dir_file,
2986 struct inode *dir, ext4_lblk_t block,
2987 struct dx_hash_info *hinfo,
2988 __u32 start_hash, __u32 start_minor_hash,
2989 int *has_inline_data);
2990 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
2991 struct ext4_filename *fname,
2992 struct ext4_dir_entry_2 **res_dir,
2993 int *has_inline_data);
2994 extern int ext4_delete_inline_entry(handle_t *handle,
2995 struct inode *dir,
2996 struct ext4_dir_entry_2 *de_del,
2997 struct buffer_head *bh,
2998 int *has_inline_data);
2999 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3000 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3001 struct ext4_dir_entry_2 **parent_de,
3002 int *retval);
3003 extern int ext4_inline_data_fiemap(struct inode *inode,
3004 struct fiemap_extent_info *fieinfo,
3005 int *has_inline, __u64 start, __u64 len);
3006
3007 struct iomap;
3008 extern int ext4_inline_data_iomap(struct inode *inode, struct iomap *iomap);
3009
3010 extern int ext4_try_to_evict_inline_data(handle_t *handle,
3011 struct inode *inode,
3012 int needed);
3013 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3014
3015 extern int ext4_convert_inline_data(struct inode *inode);
3016
3017 static inline int ext4_has_inline_data(struct inode *inode)
3018 {
3019 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3020 EXT4_I(inode)->i_inline_off;
3021 }
3022
3023 /* namei.c */
3024 extern const struct inode_operations ext4_dir_inode_operations;
3025 extern const struct inode_operations ext4_special_inode_operations;
3026 extern struct dentry *ext4_get_parent(struct dentry *child);
3027 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3028 struct ext4_dir_entry_2 *de,
3029 int blocksize, int csum_size,
3030 unsigned int parent_ino, int dotdot_real_len);
3031 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t,
3032 unsigned int blocksize);
3033 extern int ext4_handle_dirty_dirent_node(handle_t *handle,
3034 struct inode *inode,
3035 struct buffer_head *bh);
3036 #define S_SHIFT 12
3037 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = {
3038 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
3039 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
3040 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
3041 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
3042 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
3043 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
3044 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
3045 };
3046
3047 static inline void ext4_set_de_type(struct super_block *sb,
3048 struct ext4_dir_entry_2 *de,
3049 umode_t mode) {
3050 if (ext4_has_feature_filetype(sb))
3051 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3052 }
3053
3054 /* readpages.c */
3055 extern int ext4_mpage_readpages(struct address_space *mapping,
3056 struct list_head *pages, struct page *page,
3057 unsigned nr_pages);
3058
3059 /* symlink.c */
3060 extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3061 extern const struct inode_operations ext4_symlink_inode_operations;
3062 extern const struct inode_operations ext4_fast_symlink_inode_operations;
3063
3064 /* sysfs.c */
3065 extern int ext4_register_sysfs(struct super_block *sb);
3066 extern void ext4_unregister_sysfs(struct super_block *sb);
3067 extern int __init ext4_init_sysfs(void);
3068 extern void ext4_exit_sysfs(void);
3069
3070 /* block_validity */
3071 extern void ext4_release_system_zone(struct super_block *sb);
3072 extern int ext4_setup_system_zone(struct super_block *sb);
3073 extern int __init ext4_init_system_zone(void);
3074 extern void ext4_exit_system_zone(void);
3075 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
3076 ext4_fsblk_t start_blk,
3077 unsigned int count);
3078 extern int ext4_check_blockref(const char *, unsigned int,
3079 struct inode *, __le32 *, unsigned int);
3080
3081 /* extents.c */
3082 struct ext4_ext_path;
3083 struct ext4_extent;
3084
3085 /*
3086 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3087 * __le32.
3088 */
3089 #define EXT_MAX_BLOCKS 0xffffffff
3090
3091 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
3092 extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
3093 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3094 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3095 struct ext4_map_blocks *map, int flags);
3096 extern int ext4_ext_truncate(handle_t *, struct inode *);
3097 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3098 ext4_lblk_t end);
3099 extern void ext4_ext_init(struct super_block *);
3100 extern void ext4_ext_release(struct super_block *);
3101 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3102 loff_t len);
3103 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3104 loff_t offset, ssize_t len);
3105 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3106 struct ext4_map_blocks *map, int flags);
3107 extern int ext4_ext_calc_metadata_amount(struct inode *inode,
3108 ext4_lblk_t lblocks);
3109 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3110 int num,
3111 struct ext4_ext_path *path);
3112 extern int ext4_can_extents_be_merged(struct inode *inode,
3113 struct ext4_extent *ex1,
3114 struct ext4_extent *ex2);
3115 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3116 struct ext4_ext_path **,
3117 struct ext4_extent *, int);
3118 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3119 struct ext4_ext_path **,
3120 int flags);
3121 extern void ext4_ext_drop_refs(struct ext4_ext_path *);
3122 extern int ext4_ext_check_inode(struct inode *inode);
3123 extern int ext4_find_delalloc_range(struct inode *inode,
3124 ext4_lblk_t lblk_start,
3125 ext4_lblk_t lblk_end);
3126 extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk);
3127 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3128 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3129 __u64 start, __u64 len);
3130 extern int ext4_ext_precache(struct inode *inode);
3131 extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len);
3132 extern int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len);
3133 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3134 struct inode *inode2, ext4_lblk_t lblk1,
3135 ext4_lblk_t lblk2, ext4_lblk_t count,
3136 int mark_unwritten,int *err);
3137
3138 /* move_extent.c */
3139 extern void ext4_double_down_write_data_sem(struct inode *first,
3140 struct inode *second);
3141 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3142 struct inode *donor_inode);
3143 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3144 __u64 start_orig, __u64 start_donor,
3145 __u64 len, __u64 *moved_len);
3146
3147 /* page-io.c */
3148 extern int __init ext4_init_pageio(void);
3149 extern void ext4_exit_pageio(void);
3150 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3151 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3152 extern int ext4_put_io_end(ext4_io_end_t *io_end);
3153 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3154 extern void ext4_io_submit_init(struct ext4_io_submit *io,
3155 struct writeback_control *wbc);
3156 extern void ext4_end_io_rsv_work(struct work_struct *work);
3157 extern void ext4_io_submit(struct ext4_io_submit *io);
3158 extern int ext4_bio_write_page(struct ext4_io_submit *io,
3159 struct page *page,
3160 int len,
3161 struct writeback_control *wbc,
3162 bool keep_towrite);
3163
3164 /* mmp.c */
3165 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3166
3167 /*
3168 * Add new method to test whether block and inode bitmaps are properly
3169 * initialized. With uninit_bg reading the block from disk is not enough
3170 * to mark the bitmap uptodate. We need to also zero-out the bitmap
3171 */
3172 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3173
3174 static inline int bitmap_uptodate(struct buffer_head *bh)
3175 {
3176 return (buffer_uptodate(bh) &&
3177 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3178 }
3179 static inline void set_bitmap_uptodate(struct buffer_head *bh)
3180 {
3181 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3182 }
3183
3184 /*
3185 * Disable DIO read nolock optimization, so new dioreaders will be forced
3186 * to grab i_mutex
3187 */
3188 static inline void ext4_inode_block_unlocked_dio(struct inode *inode)
3189 {
3190 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
3191 smp_mb();
3192 }
3193 static inline void ext4_inode_resume_unlocked_dio(struct inode *inode)
3194 {
3195 smp_mb();
3196 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
3197 }
3198
3199 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
3200
3201 /* For ioend & aio unwritten conversion wait queues */
3202 #define EXT4_WQ_HASH_SZ 37
3203 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
3204 EXT4_WQ_HASH_SZ])
3205 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3206
3207 extern int ext4_resize_begin(struct super_block *sb);
3208 extern void ext4_resize_end(struct super_block *sb);
3209
3210 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
3211 struct ext4_io_end *io_end)
3212 {
3213 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
3214 io_end->flag |= EXT4_IO_END_UNWRITTEN;
3215 atomic_inc(&EXT4_I(inode)->i_unwritten);
3216 }
3217 }
3218
3219 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3220 {
3221 struct inode *inode = io_end->inode;
3222
3223 if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
3224 io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3225 /* Wake up anyone waiting on unwritten extent conversion */
3226 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
3227 wake_up_all(ext4_ioend_wq(inode));
3228 }
3229 }
3230
3231 extern const struct iomap_ops ext4_iomap_ops;
3232
3233 #endif /* __KERNEL__ */
3234
3235 #define EFSBADCRC EBADMSG /* Bad CRC detected */
3236 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
3237
3238 #endif /* _EXT4_H */
CRIS linux kernel tree