?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
mm: fix exec activate_mm vs TLB shootdown and lazy tlb switching race
[linux/fpc-iii.git] / fs / ext4 / ext4.h
blob6c5b4301ee37ca787ce900dc04f620d603abf73c
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 #ifdef CONFIG_EXT4_FS_ENCRYPTION
38 #include <linux/fscrypt_supp.h>
39 #else
40 #include <linux/fscrypt_notsupp.h>
41 #endif
42 #include <linux/falloc.h>
43 #include <linux/percpu-rwsem.h>
44 #ifdef __KERNEL__
45 #include <linux/compat.h>
46 #endif
47
48 /*
49 * The fourth extended filesystem constants/structures
50 */
51
52 /*
53 * with AGGRESSIVE_CHECK allocator runs consistency checks over
54 * structures. these checks slow things down a lot
55 */
56 #define AGGRESSIVE_CHECK__
57
58 /*
59 * with DOUBLE_CHECK defined mballoc creates persistent in-core
60 * bitmaps, maintains and uses them to check for double allocations
61 */
62 #define DOUBLE_CHECK__
63
64 /*
65 * Define EXT4FS_DEBUG to produce debug messages
66 */
67 #undef EXT4FS_DEBUG
68
69 /*
70 * Debug code
71 */
72 #ifdef EXT4FS_DEBUG
73 #define ext4_debug(f, a...) \
74 do { \
75 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
76 __FILE__, __LINE__, __func__); \
77 printk(KERN_DEBUG f, ## a); \
78 } while (0)
79 #else
80 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
81 #endif
82
83 /*
84 * Turn on EXT_DEBUG to get lots of info about extents operations.
85 */
86 #define EXT_DEBUG__
87 #ifdef EXT_DEBUG
88 #define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__)
89 #else
90 #define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
91 #endif
92
93 /* data type for block offset of block group */
94 typedef int ext4_grpblk_t;
95
96 /* data type for filesystem-wide blocks number */
97 typedef unsigned long long ext4_fsblk_t;
98
99 /* data type for file logical block number */
100 typedef __u32 ext4_lblk_t;
101
102 /* data type for block group number */
103 typedef unsigned int ext4_group_t;
104
105 enum SHIFT_DIRECTION {
106 SHIFT_LEFT = 0,
107 SHIFT_RIGHT,
108 };
109
110 /*
111 * Flags used in mballoc's allocation_context flags field.
112 *
113 * Also used to show what's going on for debugging purposes when the
114 * flag field is exported via the traceport interface
115 */
116
117 /* prefer goal again. length */
118 #define EXT4_MB_HINT_MERGE 0x0001
119 /* blocks already reserved */
120 #define EXT4_MB_HINT_RESERVED 0x0002
121 /* metadata is being allocated */
122 #define EXT4_MB_HINT_METADATA 0x0004
123 /* first blocks in the file */
124 #define EXT4_MB_HINT_FIRST 0x0008
125 /* search for the best chunk */
126 #define EXT4_MB_HINT_BEST 0x0010
127 /* data is being allocated */
128 #define EXT4_MB_HINT_DATA 0x0020
129 /* don't preallocate (for tails) */
130 #define EXT4_MB_HINT_NOPREALLOC 0x0040
131 /* allocate for locality group */
132 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080
133 /* allocate goal blocks or none */
134 #define EXT4_MB_HINT_GOAL_ONLY 0x0100
135 /* goal is meaningful */
136 #define EXT4_MB_HINT_TRY_GOAL 0x0200
137 /* blocks already pre-reserved by delayed allocation */
138 #define EXT4_MB_DELALLOC_RESERVED 0x0400
139 /* We are doing stream allocation */
140 #define EXT4_MB_STREAM_ALLOC 0x0800
141 /* Use reserved root blocks if needed */
142 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000
143 /* Use blocks from reserved pool */
144 #define EXT4_MB_USE_RESERVED 0x2000
145
146 struct ext4_allocation_request {
147 /* target inode for block we're allocating */
148 struct inode *inode;
149 /* how many blocks we want to allocate */
150 unsigned int len;
151 /* logical block in target inode */
152 ext4_lblk_t logical;
153 /* the closest logical allocated block to the left */
154 ext4_lblk_t lleft;
155 /* the closest logical allocated block to the right */
156 ext4_lblk_t lright;
157 /* phys. target (a hint) */
158 ext4_fsblk_t goal;
159 /* phys. block for the closest logical allocated block to the left */
160 ext4_fsblk_t pleft;
161 /* phys. block for the closest logical allocated block to the right */
162 ext4_fsblk_t pright;
163 /* flags. see above EXT4_MB_HINT_* */
164 unsigned int flags;
165 };
166
167 /*
168 * Logical to physical block mapping, used by ext4_map_blocks()
169 *
170 * This structure is used to pass requests into ext4_map_blocks() as
171 * well as to store the information returned by ext4_map_blocks(). It
172 * takes less room on the stack than a struct buffer_head.
173 */
174 #define EXT4_MAP_NEW (1 << BH_New)
175 #define EXT4_MAP_MAPPED (1 << BH_Mapped)
176 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten)
177 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary)
178 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
179 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
180
181 struct ext4_map_blocks {
182 ext4_fsblk_t m_pblk;
183 ext4_lblk_t m_lblk;
184 unsigned int m_len;
185 unsigned int m_flags;
186 };
187
188 /*
189 * Flags for ext4_io_end->flags
190 */
191 #define EXT4_IO_END_UNWRITTEN 0x0001
192
193 /*
194 * For converting unwritten extents on a work queue. 'handle' is used for
195 * buffered writeback.
196 */
197 typedef struct ext4_io_end {
198 struct list_head list; /* per-file finished IO list */
199 handle_t *handle; /* handle reserved for extent
200 * conversion */
201 struct inode *inode; /* file being written to */
202 struct bio *bio; /* Linked list of completed
203 * bios covering the extent */
204 unsigned int flag; /* unwritten or not */
205 atomic_t count; /* reference counter */
206 loff_t offset; /* offset in the file */
207 ssize_t size; /* size of the extent */
208 } ext4_io_end_t;
209
210 struct ext4_io_submit {
211 struct writeback_control *io_wbc;
212 struct bio *io_bio;
213 ext4_io_end_t *io_end;
214 sector_t io_next_block;
215 };
216
217 /*
218 * Special inodes numbers
219 */
220 #define EXT4_BAD_INO 1 /* Bad blocks inode */
221 #define EXT4_ROOT_INO 2 /* Root inode */
222 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */
223 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
224 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
225 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
226 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
227 #define EXT4_JOURNAL_INO 8 /* Journal inode */
228
229 /* First non-reserved inode for old ext4 filesystems */
230 #define EXT4_GOOD_OLD_FIRST_INO 11
231
232 /*
233 * Maximal count of links to a file
234 */
235 #define EXT4_LINK_MAX 65000
236
237 /*
238 * Macro-instructions used to manage several block sizes
239 */
240 #define EXT4_MIN_BLOCK_SIZE 1024
241 #define EXT4_MAX_BLOCK_SIZE 65536
242 #define EXT4_MIN_BLOCK_LOG_SIZE 10
243 #define EXT4_MAX_BLOCK_LOG_SIZE 16
244 #define EXT4_MAX_CLUSTER_LOG_SIZE 30
245 #ifdef __KERNEL__
246 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
247 #else
248 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
249 #endif
250 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
251 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
252 EXT4_SB(s)->s_cluster_bits)
253 #ifdef __KERNEL__
254 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
255 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
256 #else
257 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
258 #endif
259 #ifdef __KERNEL__
260 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
261 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
262 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
263 #else
264 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
265 EXT4_GOOD_OLD_INODE_SIZE : \
266 (s)->s_inode_size)
267 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
268 EXT4_GOOD_OLD_FIRST_INO : \
269 (s)->s_first_ino)
270 #endif
271 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
272 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \
273 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \
274 blkbits))
275
276 /* Translate a block number to a cluster number */
277 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
278 /* Translate a cluster number to a block number */
279 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
280 /* Translate # of blks to # of clusters */
281 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
282 (sbi)->s_cluster_bits)
283 /* Mask out the low bits to get the starting block of the cluster */
284 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
285 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
286 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
287 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
288 /* Get the cluster offset */
289 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
290 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
291 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
292 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
293
294 /*
295 * Structure of a blocks group descriptor
296 */
297 struct ext4_group_desc
298 {
299 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
300 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
301 __le32 bg_inode_table_lo; /* Inodes table block */
302 __le16 bg_free_blocks_count_lo;/* Free blocks count */
303 __le16 bg_free_inodes_count_lo;/* Free inodes count */
304 __le16 bg_used_dirs_count_lo; /* Directories count */
305 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
306 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
307 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
308 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
309 __le16 bg_itable_unused_lo; /* Unused inodes count */
310 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
311 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
312 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
313 __le32 bg_inode_table_hi; /* Inodes table block MSB */
314 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
315 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
316 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
317 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
318 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
319 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
320 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
321 __u32 bg_reserved;
322 };
323
324 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
325 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
326 sizeof(__le16))
327 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
328 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
329 sizeof(__le16))
330
331 /*
332 * Structure of a flex block group info
333 */
334
335 struct flex_groups {
336 atomic64_t free_clusters;
337 atomic_t free_inodes;
338 atomic_t used_dirs;
339 };
340
341 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
342 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
343 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
344
345 /*
346 * Macro-instructions used to manage group descriptors
347 */
348 #define EXT4_MIN_DESC_SIZE 32
349 #define EXT4_MIN_DESC_SIZE_64BIT 64
350 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
351 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
352 #ifdef __KERNEL__
353 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
354 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
355 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
356 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
357 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
358 #else
359 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
360 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
361 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
362 #endif
363
364 /*
365 * Constants relative to the data blocks
366 */
367 #define EXT4_NDIR_BLOCKS 12
368 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
369 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
370 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
371 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
372
373 /*
374 * Inode flags
375 */
376 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
377 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
378 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
379 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
380 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
381 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
382 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
383 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
384 /* Reserved for compression usage... */
385 #define EXT4_DIRTY_FL 0x00000100
386 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
387 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
388 /* nb: was previously EXT2_ECOMPR_FL */
389 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */
390 /* End compression flags --- maybe not all used */
391 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
392 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
393 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
394 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
395 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
396 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
397 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
398 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
399 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
400 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
401 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
402 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
403 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
404
405 #define EXT4_FL_USER_VISIBLE 0x304BDFFF /* User visible flags */
406 #define EXT4_FL_USER_MODIFIABLE 0x204BC0FF /* User modifiable flags */
407
408 /* Flags we can manipulate with through EXT4_IOC_FSSETXATTR */
409 #define EXT4_FL_XFLAG_VISIBLE (EXT4_SYNC_FL | \
410 EXT4_IMMUTABLE_FL | \
411 EXT4_APPEND_FL | \
412 EXT4_NODUMP_FL | \
413 EXT4_NOATIME_FL | \
414 EXT4_PROJINHERIT_FL)
415
416 /* Flags that should be inherited by new inodes from their parent. */
417 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
418 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
419 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
420 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
421 EXT4_PROJINHERIT_FL)
422
423 /* Flags that are appropriate for regular files (all but dir-specific ones). */
424 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
425
426 /* Flags that are appropriate for non-directories/regular files. */
427 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
428
429 /* The only flags that should be swapped */
430 #define EXT4_FL_SHOULD_SWAP (EXT4_HUGE_FILE_FL | EXT4_EXTENTS_FL)
431
432 /* Mask out flags that are inappropriate for the given type of inode. */
433 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
434 {
435 if (S_ISDIR(mode))
436 return flags;
437 else if (S_ISREG(mode))
438 return flags & EXT4_REG_FLMASK;
439 else
440 return flags & EXT4_OTHER_FLMASK;
441 }
442
443 /*
444 * Inode flags used for atomic set/get
445 */
446 enum {
447 EXT4_INODE_SECRM = 0, /* Secure deletion */
448 EXT4_INODE_UNRM = 1, /* Undelete */
449 EXT4_INODE_COMPR = 2, /* Compress file */
450 EXT4_INODE_SYNC = 3, /* Synchronous updates */
451 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
452 EXT4_INODE_APPEND = 5, /* writes to file may only append */
453 EXT4_INODE_NODUMP = 6, /* do not dump file */
454 EXT4_INODE_NOATIME = 7, /* do not update atime */
455 /* Reserved for compression usage... */
456 EXT4_INODE_DIRTY = 8,
457 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
458 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
459 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */
460 /* End compression flags --- maybe not all used */
461 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
462 EXT4_INODE_IMAGIC = 13, /* AFS directory */
463 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
464 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
465 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
466 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
467 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
468 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
469 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
470 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
471 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
472 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */
473 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
474 };
475
476 /*
477 * Since it's pretty easy to mix up bit numbers and hex values, we use a
478 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
479 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
480 * any extra space in the compiled kernel image, otherwise, the build will fail.
481 * It's important that these values are the same, since we are using
482 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
483 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
484 * values found in ext2, ext3 and ext4 filesystems, and of course the values
485 * defined in e2fsprogs.
486 *
487 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
488 */
489 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
490 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
491
492 static inline void ext4_check_flag_values(void)
493 {
494 CHECK_FLAG_VALUE(SECRM);
495 CHECK_FLAG_VALUE(UNRM);
496 CHECK_FLAG_VALUE(COMPR);
497 CHECK_FLAG_VALUE(SYNC);
498 CHECK_FLAG_VALUE(IMMUTABLE);
499 CHECK_FLAG_VALUE(APPEND);
500 CHECK_FLAG_VALUE(NODUMP);
501 CHECK_FLAG_VALUE(NOATIME);
502 CHECK_FLAG_VALUE(DIRTY);
503 CHECK_FLAG_VALUE(COMPRBLK);
504 CHECK_FLAG_VALUE(NOCOMPR);
505 CHECK_FLAG_VALUE(ENCRYPT);
506 CHECK_FLAG_VALUE(INDEX);
507 CHECK_FLAG_VALUE(IMAGIC);
508 CHECK_FLAG_VALUE(JOURNAL_DATA);
509 CHECK_FLAG_VALUE(NOTAIL);
510 CHECK_FLAG_VALUE(DIRSYNC);
511 CHECK_FLAG_VALUE(TOPDIR);
512 CHECK_FLAG_VALUE(HUGE_FILE);
513 CHECK_FLAG_VALUE(EXTENTS);
514 CHECK_FLAG_VALUE(EA_INODE);
515 CHECK_FLAG_VALUE(EOFBLOCKS);
516 CHECK_FLAG_VALUE(INLINE_DATA);
517 CHECK_FLAG_VALUE(PROJINHERIT);
518 CHECK_FLAG_VALUE(RESERVED);
519 }
520
521 /* Used to pass group descriptor data when online resize is done */
522 struct ext4_new_group_input {
523 __u32 group; /* Group number for this data */
524 __u64 block_bitmap; /* Absolute block number of block bitmap */
525 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
526 __u64 inode_table; /* Absolute block number of inode table start */
527 __u32 blocks_count; /* Total number of blocks in this group */
528 __u16 reserved_blocks; /* Number of reserved blocks in this group */
529 __u16 unused;
530 };
531
532 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
533 struct compat_ext4_new_group_input {
534 u32 group;
535 compat_u64 block_bitmap;
536 compat_u64 inode_bitmap;
537 compat_u64 inode_table;
538 u32 blocks_count;
539 u16 reserved_blocks;
540 u16 unused;
541 };
542 #endif
543
544 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
545 struct ext4_new_group_data {
546 __u32 group;
547 __u64 block_bitmap;
548 __u64 inode_bitmap;
549 __u64 inode_table;
550 __u32 blocks_count;
551 __u16 reserved_blocks;
552 __u16 unused;
553 __u32 free_blocks_count;
554 };
555
556 /* Indexes used to index group tables in ext4_new_group_data */
557 enum {
558 BLOCK_BITMAP = 0, /* block bitmap */
559 INODE_BITMAP, /* inode bitmap */
560 INODE_TABLE, /* inode tables */
561 GROUP_TABLE_COUNT,
562 };
563
564 /*
565 * Flags used by ext4_map_blocks()
566 */
567 /* Allocate any needed blocks and/or convert an unwritten
568 extent to be an initialized ext4 */
569 #define EXT4_GET_BLOCKS_CREATE 0x0001
570 /* Request the creation of an unwritten extent */
571 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002
572 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\
573 EXT4_GET_BLOCKS_CREATE)
574 /* Caller is from the delayed allocation writeout path
575 * finally doing the actual allocation of delayed blocks */
576 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
577 /* caller is from the direct IO path, request to creation of an
578 unwritten extents if not allocated, split the unwritten
579 extent if blocks has been preallocated already*/
580 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
581 #define EXT4_GET_BLOCKS_CONVERT 0x0010
582 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
583 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
584 /* Convert extent to initialized after IO complete */
585 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
586 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
587 /* Eventual metadata allocation (due to growing extent tree)
588 * should not fail, so try to use reserved blocks for that.*/
589 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
590 /* Don't normalize allocation size (used for fallocate) */
591 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
592 /* Request will not result in inode size update (user for fallocate) */
593 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080
594 /* Convert written extents to unwritten */
595 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100
596 /* Write zeros to newly created written extents */
597 #define EXT4_GET_BLOCKS_ZERO 0x0200
598 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\
599 EXT4_GET_BLOCKS_ZERO)
600 /* Caller will submit data before dropping transaction handle. This
601 * allows jbd2 to avoid submitting data before commit. */
602 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400
603
604 /*
605 * The bit position of these flags must not overlap with any of the
606 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(),
607 * read_extent_tree_block(), ext4_split_extent_at(),
608 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
609 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
610 * caching the extents when reading from the extent tree while a
611 * truncate or punch hole operation is in progress.
612 */
613 #define EXT4_EX_NOCACHE 0x40000000
614 #define EXT4_EX_FORCE_CACHE 0x20000000
615
616 /*
617 * Flags used by ext4_free_blocks
618 */
619 #define EXT4_FREE_BLOCKS_METADATA 0x0001
620 #define EXT4_FREE_BLOCKS_FORGET 0x0002
621 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
622 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
623 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
624 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
625
626 /*
627 * ioctl commands
628 */
629 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
630 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
631 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
632 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
633 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
634 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
635 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
636 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
637 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
638 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
639 #define EXT4_IOC_MIGRATE _IO('f', 9)
640 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
641 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
642 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
643 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
644 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
645 #define EXT4_IOC_SWAP_BOOT _IO('f', 17)
646 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18)
647 #define EXT4_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY
648 #define EXT4_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT
649 #define EXT4_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY
650
651 #ifndef FS_IOC_FSGETXATTR
652 /* Until the uapi changes get merged for project quota... */
653
654 #define FS_IOC_FSGETXATTR _IOR('X', 31, struct fsxattr)
655 #define FS_IOC_FSSETXATTR _IOW('X', 32, struct fsxattr)
656
657 /*
658 * Structure for FS_IOC_FSGETXATTR and FS_IOC_FSSETXATTR.
659 */
660 struct fsxattr {
661 __u32 fsx_xflags; /* xflags field value (get/set) */
662 __u32 fsx_extsize; /* extsize field value (get/set)*/
663 __u32 fsx_nextents; /* nextents field value (get) */
664 __u32 fsx_projid; /* project identifier (get/set) */
665 unsigned char fsx_pad[12];
666 };
667
668 /*
669 * Flags for the fsx_xflags field
670 */
671 #define FS_XFLAG_REALTIME 0x00000001 /* data in realtime volume */
672 #define FS_XFLAG_PREALLOC 0x00000002 /* preallocated file extents */
673 #define FS_XFLAG_IMMUTABLE 0x00000008 /* file cannot be modified */
674 #define FS_XFLAG_APPEND 0x00000010 /* all writes append */
675 #define FS_XFLAG_SYNC 0x00000020 /* all writes synchronous */
676 #define FS_XFLAG_NOATIME 0x00000040 /* do not update access time */
677 #define FS_XFLAG_NODUMP 0x00000080 /* do not include in backups */
678 #define FS_XFLAG_RTINHERIT 0x00000100 /* create with rt bit set */
679 #define FS_XFLAG_PROJINHERIT 0x00000200 /* create with parents projid */
680 #define FS_XFLAG_NOSYMLINKS 0x00000400 /* disallow symlink creation */
681 #define FS_XFLAG_EXTSIZE 0x00000800 /* extent size allocator hint */
682 #define FS_XFLAG_EXTSZINHERIT 0x00001000 /* inherit inode extent size */
683 #define FS_XFLAG_NODEFRAG 0x00002000 /* do not defragment */
684 #define FS_XFLAG_FILESTREAM 0x00004000 /* use filestream allocator */
685 #define FS_XFLAG_HASATTR 0x80000000 /* no DIFLAG for this */
686 #endif /* !defined(FS_IOC_FSGETXATTR) */
687
688 #define EXT4_IOC_FSGETXATTR FS_IOC_FSGETXATTR
689 #define EXT4_IOC_FSSETXATTR FS_IOC_FSSETXATTR
690
691 #define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32)
692
693 /*
694 * Flags for going down operation
695 */
696 #define EXT4_GOING_FLAGS_DEFAULT 0x0 /* going down */
697 #define EXT4_GOING_FLAGS_LOGFLUSH 0x1 /* flush log but not data */
698 #define EXT4_GOING_FLAGS_NOLOGFLUSH 0x2 /* don't flush log nor data */
699
700
701 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
702 /*
703 * ioctl commands in 32 bit emulation
704 */
705 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
706 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
707 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
708 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
709 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
710 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
711 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
712 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
713 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
714 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
715 #endif
716
717 /* Max physical block we can address w/o extents */
718 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
719
720 /* Max logical block we can support */
721 #define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFF
722
723 /*
724 * Structure of an inode on the disk
725 */
726 struct ext4_inode {
727 __le16 i_mode; /* File mode */
728 __le16 i_uid; /* Low 16 bits of Owner Uid */
729 __le32 i_size_lo; /* Size in bytes */
730 __le32 i_atime; /* Access time */
731 __le32 i_ctime; /* Inode Change time */
732 __le32 i_mtime; /* Modification time */
733 __le32 i_dtime; /* Deletion Time */
734 __le16 i_gid; /* Low 16 bits of Group Id */
735 __le16 i_links_count; /* Links count */
736 __le32 i_blocks_lo; /* Blocks count */
737 __le32 i_flags; /* File flags */
738 union {
739 struct {
740 __le32 l_i_version;
741 } linux1;
742 struct {
743 __u32 h_i_translator;
744 } hurd1;
745 struct {
746 __u32 m_i_reserved1;
747 } masix1;
748 } osd1; /* OS dependent 1 */
749 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
750 __le32 i_generation; /* File version (for NFS) */
751 __le32 i_file_acl_lo; /* File ACL */
752 __le32 i_size_high;
753 __le32 i_obso_faddr; /* Obsoleted fragment address */
754 union {
755 struct {
756 __le16 l_i_blocks_high; /* were l_i_reserved1 */
757 __le16 l_i_file_acl_high;
758 __le16 l_i_uid_high; /* these 2 fields */
759 __le16 l_i_gid_high; /* were reserved2[0] */
760 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
761 __le16 l_i_reserved;
762 } linux2;
763 struct {
764 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
765 __u16 h_i_mode_high;
766 __u16 h_i_uid_high;
767 __u16 h_i_gid_high;
768 __u32 h_i_author;
769 } hurd2;
770 struct {
771 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
772 __le16 m_i_file_acl_high;
773 __u32 m_i_reserved2[2];
774 } masix2;
775 } osd2; /* OS dependent 2 */
776 __le16 i_extra_isize;
777 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
778 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
779 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
780 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
781 __le32 i_crtime; /* File Creation time */
782 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
783 __le32 i_version_hi; /* high 32 bits for 64-bit version */
784 __le32 i_projid; /* Project ID */
785 };
786
787 struct move_extent {
788 __u32 reserved; /* should be zero */
789 __u32 donor_fd; /* donor file descriptor */
790 __u64 orig_start; /* logical start offset in block for orig */
791 __u64 donor_start; /* logical start offset in block for donor */
792 __u64 len; /* block length to be moved */
793 __u64 moved_len; /* moved block length */
794 };
795
796 #define EXT4_EPOCH_BITS 2
797 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
798 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
799
800 /*
801 * Extended fields will fit into an inode if the filesystem was formatted
802 * with large inodes (-I 256 or larger) and there are not currently any EAs
803 * consuming all of the available space. For new inodes we always reserve
804 * enough space for the kernel's known extended fields, but for inodes
805 * created with an old kernel this might not have been the case. None of
806 * the extended inode fields is critical for correct filesystem operation.
807 * This macro checks if a certain field fits in the inode. Note that
808 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
809 */
810 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
811 ((offsetof(typeof(*ext4_inode), field) + \
812 sizeof((ext4_inode)->field)) \
813 <= (EXT4_GOOD_OLD_INODE_SIZE + \
814 (einode)->i_extra_isize)) \
815
816 /*
817 * We use an encoding that preserves the times for extra epoch "00":
818 *
819 * extra msb of adjust for signed
820 * epoch 32-bit 32-bit tv_sec to
821 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
822 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
823 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
824 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
825 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
826 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
827 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
828 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
829 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
830 *
831 * Note that previous versions of the kernel on 64-bit systems would
832 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
833 * 1970. e2fsck will correct this, assuming that it is run on the
834 * affected filesystem before 2242.
835 */
836
837 static inline __le32 ext4_encode_extra_time(struct timespec *time)
838 {
839 u32 extra = sizeof(time->tv_sec) > 4 ?
840 ((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK : 0;
841 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
842 }
843
844 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
845 {
846 if (unlikely(sizeof(time->tv_sec) > 4 &&
847 (extra & cpu_to_le32(EXT4_EPOCH_MASK)))) {
848
849 #if 1
850 /* Handle legacy encoding of pre-1970 dates with epoch
851 * bits 1,1. (This backwards compatibility may be removed
852 * at the discretion of the ext4 developers.)
853 */
854 u64 extra_bits = le32_to_cpu(extra) & EXT4_EPOCH_MASK;
855 if (extra_bits == 3 && ((time->tv_sec) & 0x80000000) != 0)
856 extra_bits = 0;
857 time->tv_sec += extra_bits << 32;
858 #else
859 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
860 #endif
861 }
862 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
863 }
864
865 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
866 do { \
867 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
868 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
869 (raw_inode)->xtime ## _extra = \
870 ext4_encode_extra_time(&(inode)->xtime); \
871 } while (0)
872
873 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
874 do { \
875 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
876 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
877 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
878 (raw_inode)->xtime ## _extra = \
879 ext4_encode_extra_time(&(einode)->xtime); \
880 } while (0)
881
882 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
883 do { \
884 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
885 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
886 ext4_decode_extra_time(&(inode)->xtime, \
887 raw_inode->xtime ## _extra); \
888 else \
889 (inode)->xtime.tv_nsec = 0; \
890 } while (0)
891
892 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
893 do { \
894 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
895 (einode)->xtime.tv_sec = \
896 (signed)le32_to_cpu((raw_inode)->xtime); \
897 else \
898 (einode)->xtime.tv_sec = 0; \
899 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
900 ext4_decode_extra_time(&(einode)->xtime, \
901 raw_inode->xtime ## _extra); \
902 else \
903 (einode)->xtime.tv_nsec = 0; \
904 } while (0)
905
906 #define i_disk_version osd1.linux1.l_i_version
907
908 #if defined(__KERNEL__) || defined(__linux__)
909 #define i_reserved1 osd1.linux1.l_i_reserved1
910 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
911 #define i_blocks_high osd2.linux2.l_i_blocks_high
912 #define i_uid_low i_uid
913 #define i_gid_low i_gid
914 #define i_uid_high osd2.linux2.l_i_uid_high
915 #define i_gid_high osd2.linux2.l_i_gid_high
916 #define i_checksum_lo osd2.linux2.l_i_checksum_lo
917
918 #elif defined(__GNU__)
919
920 #define i_translator osd1.hurd1.h_i_translator
921 #define i_uid_high osd2.hurd2.h_i_uid_high
922 #define i_gid_high osd2.hurd2.h_i_gid_high
923 #define i_author osd2.hurd2.h_i_author
924
925 #elif defined(__masix__)
926
927 #define i_reserved1 osd1.masix1.m_i_reserved1
928 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
929 #define i_reserved2 osd2.masix2.m_i_reserved2
930
931 #endif /* defined(__KERNEL__) || defined(__linux__) */
932
933 #include "extents_status.h"
934
935 /*
936 * Lock subclasses for i_data_sem in the ext4_inode_info structure.
937 *
938 * These are needed to avoid lockdep false positives when we need to
939 * allocate blocks to the quota inode during ext4_map_blocks(), while
940 * holding i_data_sem for a normal (non-quota) inode. Since we don't
941 * do quota tracking for the quota inode, this avoids deadlock (as
942 * well as infinite recursion, since it isn't turtles all the way
943 * down...)
944 *
945 * I_DATA_SEM_NORMAL - Used for most inodes
946 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode
947 * where the second inode has larger inode number
948 * than the first
949 * I_DATA_SEM_QUOTA - Used for quota inodes only
950 */
951 enum {
952 I_DATA_SEM_NORMAL = 0,
953 I_DATA_SEM_OTHER,
954 I_DATA_SEM_QUOTA,
955 };
956
957
958 /*
959 * fourth extended file system inode data in memory
960 */
961 struct ext4_inode_info {
962 __le32 i_data[15]; /* unconverted */
963 __u32 i_dtime;
964 ext4_fsblk_t i_file_acl;
965
966 /*
967 * i_block_group is the number of the block group which contains
968 * this file's inode. Constant across the lifetime of the inode,
969 * it is used for making block allocation decisions - we try to
970 * place a file's data blocks near its inode block, and new inodes
971 * near to their parent directory's inode.
972 */
973 ext4_group_t i_block_group;
974 ext4_lblk_t i_dir_start_lookup;
975 #if (BITS_PER_LONG < 64)
976 unsigned long i_state_flags; /* Dynamic state flags */
977 #endif
978 unsigned long i_flags;
979
980 /*
981 * Extended attributes can be read independently of the main file
982 * data. Taking i_mutex even when reading would cause contention
983 * between readers of EAs and writers of regular file data, so
984 * instead we synchronize on xattr_sem when reading or changing
985 * EAs.
986 */
987 struct rw_semaphore xattr_sem;
988
989 struct list_head i_orphan; /* unlinked but open inodes */
990
991 /*
992 * i_disksize keeps track of what the inode size is ON DISK, not
993 * in memory. During truncate, i_size is set to the new size by
994 * the VFS prior to calling ext4_truncate(), but the filesystem won't
995 * set i_disksize to 0 until the truncate is actually under way.
996 *
997 * The intent is that i_disksize always represents the blocks which
998 * are used by this file. This allows recovery to restart truncate
999 * on orphans if we crash during truncate. We actually write i_disksize
1000 * into the on-disk inode when writing inodes out, instead of i_size.
1001 *
1002 * The only time when i_disksize and i_size may be different is when
1003 * a truncate is in progress. The only things which change i_disksize
1004 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
1005 */
1006 loff_t i_disksize;
1007
1008 /*
1009 * i_data_sem is for serialising ext4_truncate() against
1010 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
1011 * data tree are chopped off during truncate. We can't do that in
1012 * ext4 because whenever we perform intermediate commits during
1013 * truncate, the inode and all the metadata blocks *must* be in a
1014 * consistent state which allows truncation of the orphans to restart
1015 * during recovery. Hence we must fix the get_block-vs-truncate race
1016 * by other means, so we have i_data_sem.
1017 */
1018 struct rw_semaphore i_data_sem;
1019 /*
1020 * i_mmap_sem is for serializing page faults with truncate / punch hole
1021 * operations. We have to make sure that new page cannot be faulted in
1022 * a section of the inode that is being punched. We cannot easily use
1023 * i_data_sem for this since we need protection for the whole punch
1024 * operation and i_data_sem ranks below transaction start so we have
1025 * to occasionally drop it.
1026 */
1027 struct rw_semaphore i_mmap_sem;
1028 struct inode vfs_inode;
1029 struct jbd2_inode *jinode;
1030
1031 spinlock_t i_raw_lock; /* protects updates to the raw inode */
1032
1033 /*
1034 * File creation time. Its function is same as that of
1035 * struct timespec i_{a,c,m}time in the generic inode.
1036 */
1037 struct timespec i_crtime;
1038
1039 /* mballoc */
1040 struct list_head i_prealloc_list;
1041 spinlock_t i_prealloc_lock;
1042
1043 /* extents status tree */
1044 struct ext4_es_tree i_es_tree;
1045 rwlock_t i_es_lock;
1046 struct list_head i_es_list;
1047 unsigned int i_es_all_nr; /* protected by i_es_lock */
1048 unsigned int i_es_shk_nr; /* protected by i_es_lock */
1049 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for
1050 extents to shrink. Protected by
1051 i_es_lock */
1052
1053 /* ialloc */
1054 ext4_group_t i_last_alloc_group;
1055
1056 /* allocation reservation info for delalloc */
1057 /* In case of bigalloc, this refer to clusters rather than blocks */
1058 unsigned int i_reserved_data_blocks;
1059 ext4_lblk_t i_da_metadata_calc_last_lblock;
1060 int i_da_metadata_calc_len;
1061
1062 /* on-disk additional length */
1063 __u16 i_extra_isize;
1064
1065 /* Indicate the inline data space. */
1066 u16 i_inline_off;
1067 u16 i_inline_size;
1068
1069 #ifdef CONFIG_QUOTA
1070 /* quota space reservation, managed internally by quota code */
1071 qsize_t i_reserved_quota;
1072 #endif
1073
1074 /* Lock protecting lists below */
1075 spinlock_t i_completed_io_lock;
1076 /*
1077 * Completed IOs that need unwritten extents handling and have
1078 * transaction reserved
1079 */
1080 struct list_head i_rsv_conversion_list;
1081 struct work_struct i_rsv_conversion_work;
1082 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1083
1084 spinlock_t i_block_reservation_lock;
1085
1086 /*
1087 * Transactions that contain inode's metadata needed to complete
1088 * fsync and fdatasync, respectively.
1089 */
1090 tid_t i_sync_tid;
1091 tid_t i_datasync_tid;
1092
1093 #ifdef CONFIG_QUOTA
1094 struct dquot *i_dquot[MAXQUOTAS];
1095 #endif
1096
1097 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1098 __u32 i_csum_seed;
1099
1100 kprojid_t i_projid;
1101 };
1102
1103 /*
1104 * File system states
1105 */
1106 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
1107 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
1108 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
1109
1110 /*
1111 * Misc. filesystem flags
1112 */
1113 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
1114 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
1115 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
1116
1117 /*
1118 * Mount flags set via mount options or defaults
1119 */
1120 #define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */
1121 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
1122 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
1123 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
1124 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
1125 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
1126 #define EXT4_MOUNT_ERRORS_MASK 0x00070
1127 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
1128 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
1129 #ifdef CONFIG_FS_DAX
1130 #define EXT4_MOUNT_DAX 0x00200 /* Direct Access */
1131 #else
1132 #define EXT4_MOUNT_DAX 0
1133 #endif
1134 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
1135 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
1136 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
1137 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
1138 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
1139 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
1140 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
1141 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
1142 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
1143 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
1144 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */
1145 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota,
1146 * enable enforcement for hidden
1147 * quota files */
1148 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable
1149 * enforcement for hidden quota
1150 * files */
1151 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota
1152 * enforcement */
1153 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
1154 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
1155 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1156 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
1157 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
1158 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
1159 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
1160 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
1161
1162 /*
1163 * Mount flags set either automatically (could not be set by mount option)
1164 * based on per file system feature or property or in special cases such as
1165 * distinguishing between explicit mount option definition and default.
1166 */
1167 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1168 specified delalloc */
1169 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1170 size of blocksize * 8
1171 blocks */
1172 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1173 file systems */
1174
1175 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly
1176 specified journal checksum */
1177
1178 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1179 ~EXT4_MOUNT_##opt
1180 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1181 EXT4_MOUNT_##opt
1182 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1183 EXT4_MOUNT_##opt)
1184
1185 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1186 ~EXT4_MOUNT2_##opt
1187 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1188 EXT4_MOUNT2_##opt
1189 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1190 EXT4_MOUNT2_##opt)
1191
1192 #define ext4_test_and_set_bit __test_and_set_bit_le
1193 #define ext4_set_bit __set_bit_le
1194 #define ext4_set_bit_atomic ext2_set_bit_atomic
1195 #define ext4_test_and_clear_bit __test_and_clear_bit_le
1196 #define ext4_clear_bit __clear_bit_le
1197 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
1198 #define ext4_test_bit test_bit_le
1199 #define ext4_find_next_zero_bit find_next_zero_bit_le
1200 #define ext4_find_next_bit find_next_bit_le
1201
1202 extern void ext4_set_bits(void *bm, int cur, int len);
1203
1204 /*
1205 * Maximal mount counts between two filesystem checks
1206 */
1207 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1208 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1209
1210 /*
1211 * Behaviour when detecting errors
1212 */
1213 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1214 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1215 #define EXT4_ERRORS_PANIC 3 /* Panic */
1216 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1217
1218 /* Metadata checksum algorithm codes */
1219 #define EXT4_CRC32C_CHKSUM 1
1220
1221 /*
1222 * Structure of the super block
1223 */
1224 struct ext4_super_block {
1225 /*00*/ __le32 s_inodes_count; /* Inodes count */
1226 __le32 s_blocks_count_lo; /* Blocks count */
1227 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1228 __le32 s_free_blocks_count_lo; /* Free blocks count */
1229 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1230 __le32 s_first_data_block; /* First Data Block */
1231 __le32 s_log_block_size; /* Block size */
1232 __le32 s_log_cluster_size; /* Allocation cluster size */
1233 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1234 __le32 s_clusters_per_group; /* # Clusters per group */
1235 __le32 s_inodes_per_group; /* # Inodes per group */
1236 __le32 s_mtime; /* Mount time */
1237 /*30*/ __le32 s_wtime; /* Write time */
1238 __le16 s_mnt_count; /* Mount count */
1239 __le16 s_max_mnt_count; /* Maximal mount count */
1240 __le16 s_magic; /* Magic signature */
1241 __le16 s_state; /* File system state */
1242 __le16 s_errors; /* Behaviour when detecting errors */
1243 __le16 s_minor_rev_level; /* minor revision level */
1244 /*40*/ __le32 s_lastcheck; /* time of last check */
1245 __le32 s_checkinterval; /* max. time between checks */
1246 __le32 s_creator_os; /* OS */
1247 __le32 s_rev_level; /* Revision level */
1248 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1249 __le16 s_def_resgid; /* Default gid for reserved blocks */
1250 /*
1251 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1252 *
1253 * Note: the difference between the compatible feature set and
1254 * the incompatible feature set is that if there is a bit set
1255 * in the incompatible feature set that the kernel doesn't
1256 * know about, it should refuse to mount the filesystem.
1257 *
1258 * e2fsck's requirements are more strict; if it doesn't know
1259 * about a feature in either the compatible or incompatible
1260 * feature set, it must abort and not try to meddle with
1261 * things it doesn't understand...
1262 */
1263 __le32 s_first_ino; /* First non-reserved inode */
1264 __le16 s_inode_size; /* size of inode structure */
1265 __le16 s_block_group_nr; /* block group # of this superblock */
1266 __le32 s_feature_compat; /* compatible feature set */
1267 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1268 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1269 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1270 /*78*/ char s_volume_name[16]; /* volume name */
1271 /*88*/ char s_last_mounted[64]; /* directory where last mounted */
1272 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1273 /*
1274 * Performance hints. Directory preallocation should only
1275 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1276 */
1277 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1278 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1279 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1280 /*
1281 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1282 */
1283 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1284 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1285 __le32 s_journal_dev; /* device number of journal file */
1286 __le32 s_last_orphan; /* start of list of inodes to delete */
1287 __le32 s_hash_seed[4]; /* HTREE hash seed */
1288 __u8 s_def_hash_version; /* Default hash version to use */
1289 __u8 s_jnl_backup_type;
1290 __le16 s_desc_size; /* size of group descriptor */
1291 /*100*/ __le32 s_default_mount_opts;
1292 __le32 s_first_meta_bg; /* First metablock block group */
1293 __le32 s_mkfs_time; /* When the filesystem was created */
1294 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1295 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1296 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1297 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1298 __le32 s_free_blocks_count_hi; /* Free blocks count */
1299 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1300 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1301 __le32 s_flags; /* Miscellaneous flags */
1302 __le16 s_raid_stride; /* RAID stride */
1303 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1304 __le64 s_mmp_block; /* Block for multi-mount protection */
1305 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1306 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1307 __u8 s_checksum_type; /* metadata checksum algorithm used */
1308 __u8 s_encryption_level; /* versioning level for encryption */
1309 __u8 s_reserved_pad; /* Padding to next 32bits */
1310 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1311 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1312 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1313 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1314 snapshot's future use */
1315 __le32 s_snapshot_list; /* inode number of the head of the
1316 on-disk snapshot list */
1317 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1318 __le32 s_error_count; /* number of fs errors */
1319 __le32 s_first_error_time; /* first time an error happened */
1320 __le32 s_first_error_ino; /* inode involved in first error */
1321 __le64 s_first_error_block; /* block involved of first error */
1322 __u8 s_first_error_func[32]; /* function where the error happened */
1323 __le32 s_first_error_line; /* line number where error happened */
1324 __le32 s_last_error_time; /* most recent time of an error */
1325 __le32 s_last_error_ino; /* inode involved in last error */
1326 __le32 s_last_error_line; /* line number where error happened */
1327 __le64 s_last_error_block; /* block involved of last error */
1328 __u8 s_last_error_func[32]; /* function where the error happened */
1329 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1330 __u8 s_mount_opts[64];
1331 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1332 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1333 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1334 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
1335 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */
1336 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */
1337 __le32 s_lpf_ino; /* Location of the lost+found inode */
1338 __le32 s_prj_quota_inum; /* inode for tracking project quota */
1339 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */
1340 __le32 s_reserved[98]; /* Padding to the end of the block */
1341 __le32 s_checksum; /* crc32c(superblock) */
1342 };
1343
1344 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1345
1346 #ifdef __KERNEL__
1347
1348 /*
1349 * run-time mount flags
1350 */
1351 #define EXT4_MF_MNTDIR_SAMPLED 0x0001
1352 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1353 #define EXT4_MF_TEST_DUMMY_ENCRYPTION 0x0004
1354
1355 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1356 #define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \
1357 EXT4_MF_TEST_DUMMY_ENCRYPTION))
1358 #else
1359 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1360 #endif
1361
1362 /* Number of quota types we support */
1363 #define EXT4_MAXQUOTAS 3
1364
1365 /*
1366 * fourth extended-fs super-block data in memory
1367 */
1368 struct ext4_sb_info {
1369 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1370 unsigned long s_inodes_per_block;/* Number of inodes per block */
1371 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1372 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1373 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1374 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1375 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1376 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1377 ext4_group_t s_groups_count; /* Number of groups in the fs */
1378 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1379 unsigned long s_overhead; /* # of fs overhead clusters */
1380 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1381 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1382 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1383 struct buffer_head * s_sbh; /* Buffer containing the super block */
1384 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1385 struct buffer_head * __rcu *s_group_desc;
1386 unsigned int s_mount_opt;
1387 unsigned int s_mount_opt2;
1388 unsigned int s_mount_flags;
1389 unsigned int s_def_mount_opt;
1390 ext4_fsblk_t s_sb_block;
1391 atomic64_t s_resv_clusters;
1392 kuid_t s_resuid;
1393 kgid_t s_resgid;
1394 unsigned short s_mount_state;
1395 unsigned short s_pad;
1396 int s_addr_per_block_bits;
1397 int s_desc_per_block_bits;
1398 int s_inode_size;
1399 int s_first_ino;
1400 unsigned int s_inode_readahead_blks;
1401 unsigned int s_inode_goal;
1402 spinlock_t s_next_gen_lock;
1403 u32 s_next_generation;
1404 u32 s_hash_seed[4];
1405 int s_def_hash_version;
1406 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1407 struct percpu_counter s_freeclusters_counter;
1408 struct percpu_counter s_freeinodes_counter;
1409 struct percpu_counter s_dirs_counter;
1410 struct percpu_counter s_dirtyclusters_counter;
1411 struct blockgroup_lock *s_blockgroup_lock;
1412 struct proc_dir_entry *s_proc;
1413 struct kobject s_kobj;
1414 struct completion s_kobj_unregister;
1415 struct super_block *s_sb;
1416
1417 /* Journaling */
1418 struct journal_s *s_journal;
1419 struct list_head s_orphan;
1420 struct mutex s_orphan_lock;
1421 unsigned long s_ext4_flags; /* Ext4 superblock flags */
1422 unsigned long s_commit_interval;
1423 u32 s_max_batch_time;
1424 u32 s_min_batch_time;
1425 struct block_device *journal_bdev;
1426 #ifdef CONFIG_QUOTA
1427 /* Names of quota files with journalled quota */
1428 char __rcu *s_qf_names[EXT4_MAXQUOTAS];
1429 int s_jquota_fmt; /* Format of quota to use */
1430 #endif
1431 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1432 struct rb_root system_blks;
1433
1434 #ifdef EXTENTS_STATS
1435 /* ext4 extents stats */
1436 unsigned long s_ext_min;
1437 unsigned long s_ext_max;
1438 unsigned long s_depth_max;
1439 spinlock_t s_ext_stats_lock;
1440 unsigned long s_ext_blocks;
1441 unsigned long s_ext_extents;
1442 #endif
1443
1444 /* for buddy allocator */
1445 struct ext4_group_info ** __rcu *s_group_info;
1446 struct inode *s_buddy_cache;
1447 spinlock_t s_md_lock;
1448 unsigned short *s_mb_offsets;
1449 unsigned int *s_mb_maxs;
1450 unsigned int s_group_info_size;
1451 unsigned int s_mb_free_pending;
1452 struct list_head s_freed_data_list; /* List of blocks to be freed
1453 after commit completed */
1454
1455 /* tunables */
1456 unsigned long s_stripe;
1457 unsigned int s_mb_stream_request;
1458 unsigned int s_mb_max_to_scan;
1459 unsigned int s_mb_min_to_scan;
1460 unsigned int s_mb_stats;
1461 unsigned int s_mb_order2_reqs;
1462 unsigned int s_mb_group_prealloc;
1463 unsigned int s_max_dir_size_kb;
1464 /* where last allocation was done - for stream allocation */
1465 unsigned long s_mb_last_group;
1466 unsigned long s_mb_last_start;
1467
1468 /* stats for buddy allocator */
1469 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1470 atomic_t s_bal_success; /* we found long enough chunks */
1471 atomic_t s_bal_allocated; /* in blocks */
1472 atomic_t s_bal_ex_scanned; /* total extents scanned */
1473 atomic_t s_bal_goals; /* goal hits */
1474 atomic_t s_bal_breaks; /* too long searches */
1475 atomic_t s_bal_2orders; /* 2^order hits */
1476 spinlock_t s_bal_lock;
1477 unsigned long s_mb_buddies_generated;
1478 unsigned long long s_mb_generation_time;
1479 atomic_t s_mb_lost_chunks;
1480 atomic_t s_mb_preallocated;
1481 atomic_t s_mb_discarded;
1482 atomic_t s_lock_busy;
1483
1484 /* locality groups */
1485 struct ext4_locality_group __percpu *s_locality_groups;
1486
1487 /* for write statistics */
1488 unsigned long s_sectors_written_start;
1489 u64 s_kbytes_written;
1490
1491 /* the size of zero-out chunk */
1492 unsigned int s_extent_max_zeroout_kb;
1493
1494 unsigned int s_log_groups_per_flex;
1495 struct flex_groups * __rcu *s_flex_groups;
1496 ext4_group_t s_flex_groups_allocated;
1497
1498 /* workqueue for reserved extent conversions (buffered io) */
1499 struct workqueue_struct *rsv_conversion_wq;
1500
1501 /* timer for periodic error stats printing */
1502 struct timer_list s_err_report;
1503
1504 /* Lazy inode table initialization info */
1505 struct ext4_li_request *s_li_request;
1506 /* Wait multiplier for lazy initialization thread */
1507 unsigned int s_li_wait_mult;
1508
1509 /* Kernel thread for multiple mount protection */
1510 struct task_struct *s_mmp_tsk;
1511
1512 /* record the last minlen when FITRIM is called. */
1513 atomic_t s_last_trim_minblks;
1514
1515 /* Reference to checksum algorithm driver via cryptoapi */
1516 struct crypto_shash *s_chksum_driver;
1517
1518 /* Precomputed FS UUID checksum for seeding other checksums */
1519 __u32 s_csum_seed;
1520
1521 /* Reclaim extents from extent status tree */
1522 struct shrinker s_es_shrinker;
1523 struct list_head s_es_list; /* List of inodes with reclaimable extents */
1524 long s_es_nr_inode;
1525 struct ext4_es_stats s_es_stats;
1526 struct mb_cache *s_ea_block_cache;
1527 struct mb_cache *s_ea_inode_cache;
1528 spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1529
1530 /* Ratelimit ext4 messages. */
1531 struct ratelimit_state s_err_ratelimit_state;
1532 struct ratelimit_state s_warning_ratelimit_state;
1533 struct ratelimit_state s_msg_ratelimit_state;
1534
1535 /*
1536 * Barrier between writepages ops and changing any inode's JOURNAL_DATA
1537 * or EXTENTS flag.
1538 */
1539 struct percpu_rw_semaphore s_writepages_rwsem;
1540 struct dax_device *s_daxdev;
1541 };
1542
1543 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1544 {
1545 return sb->s_fs_info;
1546 }
1547 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1548 {
1549 return container_of(inode, struct ext4_inode_info, vfs_inode);
1550 }
1551
1552 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1553 {
1554 return ino == EXT4_ROOT_INO ||
1555 (ino >= EXT4_FIRST_INO(sb) &&
1556 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1557 }
1558
1559 /*
1560 * Returns: sbi->field[index]
1561 * Used to access an array element from the following sbi fields which require
1562 * rcu protection to avoid dereferencing an invalid pointer due to reassignment
1563 * - s_group_desc
1564 * - s_group_info
1565 * - s_flex_group
1566 */
1567 #define sbi_array_rcu_deref(sbi, field, index) \
1568 ({ \
1569 typeof(*((sbi)->field)) _v; \
1570 rcu_read_lock(); \
1571 _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index]; \
1572 rcu_read_unlock(); \
1573 _v; \
1574 })
1575
1576 /*
1577 * Inode dynamic state flags
1578 */
1579 enum {
1580 EXT4_STATE_JDATA, /* journaled data exists */
1581 EXT4_STATE_NEW, /* inode is newly created */
1582 EXT4_STATE_XATTR, /* has in-inode xattrs */
1583 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1584 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1585 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1586 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
1587 EXT4_STATE_NEWENTRY, /* File just added to dir */
1588 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read
1589 nolocking */
1590 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1591 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
1592 EXT4_STATE_LUSTRE_EA_INODE, /* Lustre-style ea_inode */
1593 };
1594
1595 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1596 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1597 { \
1598 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1599 } \
1600 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1601 { \
1602 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1603 } \
1604 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1605 { \
1606 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1607 }
1608
1609 /* Add these declarations here only so that these functions can be
1610 * found by name. Otherwise, they are very hard to locate. */
1611 static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1612 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1613 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1614 EXT4_INODE_BIT_FNS(flag, flags, 0)
1615
1616 /* Add these declarations here only so that these functions can be
1617 * found by name. Otherwise, they are very hard to locate. */
1618 static inline int ext4_test_inode_state(struct inode *inode, int bit);
1619 static inline void ext4_set_inode_state(struct inode *inode, int bit);
1620 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1621 #if (BITS_PER_LONG < 64)
1622 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1623
1624 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1625 {
1626 (ei)->i_state_flags = 0;
1627 }
1628 #else
1629 EXT4_INODE_BIT_FNS(state, flags, 32)
1630
1631 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1632 {
1633 /* We depend on the fact that callers will set i_flags */
1634 }
1635 #endif
1636 #else
1637 /* Assume that user mode programs are passing in an ext4fs superblock, not
1638 * a kernel struct super_block. This will allow us to call the feature-test
1639 * macros from user land. */
1640 #define EXT4_SB(sb) (sb)
1641 #endif
1642
1643 /*
1644 * Returns true if the inode is inode is encrypted
1645 */
1646 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1647
1648 /*
1649 * Codes for operating systems
1650 */
1651 #define EXT4_OS_LINUX 0
1652 #define EXT4_OS_HURD 1
1653 #define EXT4_OS_MASIX 2
1654 #define EXT4_OS_FREEBSD 3
1655 #define EXT4_OS_LITES 4
1656
1657 /*
1658 * Revision levels
1659 */
1660 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1661 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1662
1663 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1664 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1665
1666 #define EXT4_GOOD_OLD_INODE_SIZE 128
1667
1668 /*
1669 * Feature set definitions
1670 */
1671
1672 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1673 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1674 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1675 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1676 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1677 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1678 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
1679
1680 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1681 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1682 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1683 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1684 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1685 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1686 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1687 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1688 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1689 /*
1690 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1691 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1692 * all other data structures' checksums. However, the METADATA_CSUM and
1693 * GDT_CSUM bits are mutually exclusive.
1694 */
1695 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
1696 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000
1697 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000
1698
1699 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1700 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1701 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1702 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1703 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1704 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1705 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1706 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1707 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1708 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1709 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1710 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000
1711 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
1712 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
1713 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000
1714
1715 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
1716 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1717 { \
1718 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1719 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
1720 } \
1721 static inline void ext4_set_feature_##name(struct super_block *sb) \
1722 { \
1723 EXT4_SB(sb)->s_es->s_feature_compat |= \
1724 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1725 } \
1726 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1727 { \
1728 EXT4_SB(sb)->s_es->s_feature_compat &= \
1729 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1730 }
1731
1732 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
1733 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1734 { \
1735 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1736 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
1737 } \
1738 static inline void ext4_set_feature_##name(struct super_block *sb) \
1739 { \
1740 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
1741 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1742 } \
1743 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1744 { \
1745 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
1746 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1747 }
1748
1749 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
1750 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1751 { \
1752 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1753 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
1754 } \
1755 static inline void ext4_set_feature_##name(struct super_block *sb) \
1756 { \
1757 EXT4_SB(sb)->s_es->s_feature_incompat |= \
1758 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1759 } \
1760 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1761 { \
1762 EXT4_SB(sb)->s_es->s_feature_incompat &= \
1763 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1764 }
1765
1766 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC)
1767 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES)
1768 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL)
1769 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR)
1770 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE)
1771 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX)
1772 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2)
1773
1774 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER)
1775 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE)
1776 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR)
1777 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE)
1778 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM)
1779 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK)
1780 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE)
1781 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA)
1782 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC)
1783 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM)
1784 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY)
1785 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT)
1786
1787 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION)
1788 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE)
1789 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER)
1790 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV)
1791 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG)
1792 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS)
1793 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT)
1794 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP)
1795 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG)
1796 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE)
1797 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA)
1798 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED)
1799 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR)
1800 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA)
1801 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT)
1802
1803 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1804 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1805 EXT4_FEATURE_INCOMPAT_META_BG)
1806 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1807 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1808 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1809
1810 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1811 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1812 EXT4_FEATURE_INCOMPAT_RECOVER| \
1813 EXT4_FEATURE_INCOMPAT_META_BG)
1814 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1815 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1816 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1817
1818 #define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1819 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1820 EXT4_FEATURE_INCOMPAT_RECOVER| \
1821 EXT4_FEATURE_INCOMPAT_META_BG| \
1822 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1823 EXT4_FEATURE_INCOMPAT_64BIT| \
1824 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1825 EXT4_FEATURE_INCOMPAT_EA_INODE| \
1826 EXT4_FEATURE_INCOMPAT_MMP | \
1827 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
1828 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
1829 EXT4_FEATURE_INCOMPAT_CSUM_SEED | \
1830 EXT4_FEATURE_INCOMPAT_LARGEDIR)
1831 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1832 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1833 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1834 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1835 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1836 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1837 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1838 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1839 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1840 EXT4_FEATURE_RO_COMPAT_QUOTA |\
1841 EXT4_FEATURE_RO_COMPAT_PROJECT)
1842
1843 #define EXTN_FEATURE_FUNCS(ver) \
1844 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
1845 { \
1846 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1847 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
1848 } \
1849 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
1850 { \
1851 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1852 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
1853 } \
1854 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
1855 { \
1856 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1857 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
1858 }
1859
1860 EXTN_FEATURE_FUNCS(2)
1861 EXTN_FEATURE_FUNCS(3)
1862 EXTN_FEATURE_FUNCS(4)
1863
1864 static inline bool ext4_has_compat_features(struct super_block *sb)
1865 {
1866 return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
1867 }
1868 static inline bool ext4_has_ro_compat_features(struct super_block *sb)
1869 {
1870 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
1871 }
1872 static inline bool ext4_has_incompat_features(struct super_block *sb)
1873 {
1874 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
1875 }
1876
1877 /*
1878 * Superblock flags
1879 */
1880 #define EXT4_FLAGS_RESIZING 0
1881 #define EXT4_FLAGS_SHUTDOWN 1
1882
1883 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi)
1884 {
1885 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags);
1886 }
1887
1888
1889 /*
1890 * Default values for user and/or group using reserved blocks
1891 */
1892 #define EXT4_DEF_RESUID 0
1893 #define EXT4_DEF_RESGID 0
1894
1895 /*
1896 * Default project ID
1897 */
1898 #define EXT4_DEF_PROJID 0
1899
1900 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
1901
1902 /*
1903 * Default mount options
1904 */
1905 #define EXT4_DEFM_DEBUG 0x0001
1906 #define EXT4_DEFM_BSDGROUPS 0x0002
1907 #define EXT4_DEFM_XATTR_USER 0x0004
1908 #define EXT4_DEFM_ACL 0x0008
1909 #define EXT4_DEFM_UID16 0x0010
1910 #define EXT4_DEFM_JMODE 0x0060
1911 #define EXT4_DEFM_JMODE_DATA 0x0020
1912 #define EXT4_DEFM_JMODE_ORDERED 0x0040
1913 #define EXT4_DEFM_JMODE_WBACK 0x0060
1914 #define EXT4_DEFM_NOBARRIER 0x0100
1915 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1916 #define EXT4_DEFM_DISCARD 0x0400
1917 #define EXT4_DEFM_NODELALLOC 0x0800
1918
1919 /*
1920 * Default journal batch times
1921 */
1922 #define EXT4_DEF_MIN_BATCH_TIME 0
1923 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1924
1925 /*
1926 * Minimum number of groups in a flexgroup before we separate out
1927 * directories into the first block group of a flexgroup
1928 */
1929 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1930
1931 /*
1932 * Structure of a directory entry
1933 */
1934 #define EXT4_NAME_LEN 255
1935
1936 struct ext4_dir_entry {
1937 __le32 inode; /* Inode number */
1938 __le16 rec_len; /* Directory entry length */
1939 __le16 name_len; /* Name length */
1940 char name[EXT4_NAME_LEN]; /* File name */
1941 };
1942
1943 /*
1944 * The new version of the directory entry. Since EXT4 structures are
1945 * stored in intel byte order, and the name_len field could never be
1946 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1947 * file_type field.
1948 */
1949 struct ext4_dir_entry_2 {
1950 __le32 inode; /* Inode number */
1951 __le16 rec_len; /* Directory entry length */
1952 __u8 name_len; /* Name length */
1953 __u8 file_type;
1954 char name[EXT4_NAME_LEN]; /* File name */
1955 };
1956
1957 /*
1958 * This is a bogus directory entry at the end of each leaf block that
1959 * records checksums.
1960 */
1961 struct ext4_dir_entry_tail {
1962 __le32 det_reserved_zero1; /* Pretend to be unused */
1963 __le16 det_rec_len; /* 12 */
1964 __u8 det_reserved_zero2; /* Zero name length */
1965 __u8 det_reserved_ft; /* 0xDE, fake file type */
1966 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
1967 };
1968
1969 #define EXT4_DIRENT_TAIL(block, blocksize) \
1970 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1971 ((blocksize) - \
1972 sizeof(struct ext4_dir_entry_tail))))
1973
1974 /*
1975 * Ext4 directory file types. Only the low 3 bits are used. The
1976 * other bits are reserved for now.
1977 */
1978 #define EXT4_FT_UNKNOWN 0
1979 #define EXT4_FT_REG_FILE 1
1980 #define EXT4_FT_DIR 2
1981 #define EXT4_FT_CHRDEV 3
1982 #define EXT4_FT_BLKDEV 4
1983 #define EXT4_FT_FIFO 5
1984 #define EXT4_FT_SOCK 6
1985 #define EXT4_FT_SYMLINK 7
1986
1987 #define EXT4_FT_MAX 8
1988
1989 #define EXT4_FT_DIR_CSUM 0xDE
1990
1991 /*
1992 * EXT4_DIR_PAD defines the directory entries boundaries
1993 *
1994 * NOTE: It must be a multiple of 4
1995 */
1996 #define EXT4_DIR_PAD 4
1997 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1998 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1999 ~EXT4_DIR_ROUND)
2000 #define EXT4_MAX_REC_LEN ((1<<16)-1)
2001
2002 /*
2003 * If we ever get support for fs block sizes > page_size, we'll need
2004 * to remove the #if statements in the next two functions...
2005 */
2006 static inline unsigned int
2007 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
2008 {
2009 unsigned len = le16_to_cpu(dlen);
2010
2011 #if (PAGE_SIZE >= 65536)
2012 if (len == EXT4_MAX_REC_LEN || len == 0)
2013 return blocksize;
2014 return (len & 65532) | ((len & 3) << 16);
2015 #else
2016 return len;
2017 #endif
2018 }
2019
2020 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
2021 {
2022 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
2023 BUG();
2024 #if (PAGE_SIZE >= 65536)
2025 if (len < 65536)
2026 return cpu_to_le16(len);
2027 if (len == blocksize) {
2028 if (blocksize == 65536)
2029 return cpu_to_le16(EXT4_MAX_REC_LEN);
2030 else
2031 return cpu_to_le16(0);
2032 }
2033 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
2034 #else
2035 return cpu_to_le16(len);
2036 #endif
2037 }
2038
2039 /*
2040 * Hash Tree Directory indexing
2041 * (c) Daniel Phillips, 2001
2042 */
2043
2044 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
2045 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2046 #define EXT4_DIR_LINK_MAX(dir) unlikely((dir)->i_nlink >= EXT4_LINK_MAX && \
2047 !(ext4_has_feature_dir_nlink((dir)->i_sb) && is_dx(dir)))
2048 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2049
2050 /* Legal values for the dx_root hash_version field: */
2051
2052 #define DX_HASH_LEGACY 0
2053 #define DX_HASH_HALF_MD4 1
2054 #define DX_HASH_TEA 2
2055 #define DX_HASH_LEGACY_UNSIGNED 3
2056 #define DX_HASH_HALF_MD4_UNSIGNED 4
2057 #define DX_HASH_TEA_UNSIGNED 5
2058
2059 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2060 const void *address, unsigned int length)
2061 {
2062 struct {
2063 struct shash_desc shash;
2064 char ctx[4];
2065 } desc;
2066 int err;
2067
2068 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2069
2070 desc.shash.tfm = sbi->s_chksum_driver;
2071 desc.shash.flags = 0;
2072 *(u32 *)desc.ctx = crc;
2073
2074 err = crypto_shash_update(&desc.shash, address, length);
2075 BUG_ON(err);
2076
2077 return *(u32 *)desc.ctx;
2078 }
2079
2080 #ifdef __KERNEL__
2081
2082 /* hash info structure used by the directory hash */
2083 struct dx_hash_info
2084 {
2085 u32 hash;
2086 u32 minor_hash;
2087 int hash_version;
2088 u32 *seed;
2089 };
2090
2091
2092 /* 32 and 64 bit signed EOF for dx directories */
2093 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
2094 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
2095
2096
2097 /*
2098 * Control parameters used by ext4_htree_next_block
2099 */
2100 #define HASH_NB_ALWAYS 1
2101
2102 struct ext4_filename {
2103 const struct qstr *usr_fname;
2104 struct fscrypt_str disk_name;
2105 struct dx_hash_info hinfo;
2106 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2107 struct fscrypt_str crypto_buf;
2108 #endif
2109 };
2110
2111 #define fname_name(p) ((p)->disk_name.name)
2112 #define fname_len(p) ((p)->disk_name.len)
2113
2114 /*
2115 * Describe an inode's exact location on disk and in memory
2116 */
2117 struct ext4_iloc
2118 {
2119 struct buffer_head *bh;
2120 unsigned long offset;
2121 ext4_group_t block_group;
2122 };
2123
2124 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2125 {
2126 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2127 }
2128
2129 static inline bool ext4_is_quota_file(struct inode *inode)
2130 {
2131 return IS_NOQUOTA(inode) &&
2132 !(EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL);
2133 }
2134
2135 /*
2136 * This structure is stuffed into the struct file's private_data field
2137 * for directories. It is where we put information so that we can do
2138 * readdir operations in hash tree order.
2139 */
2140 struct dir_private_info {
2141 struct rb_root root;
2142 struct rb_node *curr_node;
2143 struct fname *extra_fname;
2144 loff_t last_pos;
2145 __u32 curr_hash;
2146 __u32 curr_minor_hash;
2147 __u32 next_hash;
2148 };
2149
2150 /* calculate the first block number of the group */
2151 static inline ext4_fsblk_t
2152 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2153 {
2154 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2155 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2156 }
2157
2158 /*
2159 * Special error return code only used by dx_probe() and its callers.
2160 */
2161 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
2162
2163 /* htree levels for ext4 */
2164 #define EXT4_HTREE_LEVEL_COMPAT 2
2165 #define EXT4_HTREE_LEVEL 3
2166
2167 static inline int ext4_dir_htree_level(struct super_block *sb)
2168 {
2169 return ext4_has_feature_largedir(sb) ?
2170 EXT4_HTREE_LEVEL : EXT4_HTREE_LEVEL_COMPAT;
2171 }
2172
2173 /*
2174 * Timeout and state flag for lazy initialization inode thread.
2175 */
2176 #define EXT4_DEF_LI_WAIT_MULT 10
2177 #define EXT4_DEF_LI_MAX_START_DELAY 5
2178 #define EXT4_LAZYINIT_QUIT 0x0001
2179 #define EXT4_LAZYINIT_RUNNING 0x0002
2180
2181 /*
2182 * Lazy inode table initialization info
2183 */
2184 struct ext4_lazy_init {
2185 unsigned long li_state;
2186 struct list_head li_request_list;
2187 struct mutex li_list_mtx;
2188 };
2189
2190 struct ext4_li_request {
2191 struct super_block *lr_super;
2192 struct ext4_sb_info *lr_sbi;
2193 ext4_group_t lr_next_group;
2194 struct list_head lr_request;
2195 unsigned long lr_next_sched;
2196 unsigned long lr_timeout;
2197 };
2198
2199 struct ext4_features {
2200 struct kobject f_kobj;
2201 struct completion f_kobj_unregister;
2202 };
2203
2204 /*
2205 * This structure will be used for multiple mount protection. It will be
2206 * written into the block number saved in the s_mmp_block field in the
2207 * superblock. Programs that check MMP should assume that if
2208 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2209 * to use the filesystem, regardless of how old the timestamp is.
2210 */
2211 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
2212 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2213 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
2214 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
2215
2216 struct mmp_struct {
2217 __le32 mmp_magic; /* Magic number for MMP */
2218 __le32 mmp_seq; /* Sequence no. updated periodically */
2219
2220 /*
2221 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2222 * purposes and do not affect the correctness of the algorithm
2223 */
2224 __le64 mmp_time; /* Time last updated */
2225 char mmp_nodename[64]; /* Node which last updated MMP block */
2226 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
2227
2228 /*
2229 * mmp_check_interval is used to verify if the MMP block has been
2230 * updated on the block device. The value is updated based on the
2231 * maximum time to write the MMP block during an update cycle.
2232 */
2233 __le16 mmp_check_interval;
2234
2235 __le16 mmp_pad1;
2236 __le32 mmp_pad2[226];
2237 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
2238 };
2239
2240 /* arguments passed to the mmp thread */
2241 struct mmpd_data {
2242 struct buffer_head *bh; /* bh from initial read_mmp_block() */
2243 struct super_block *sb; /* super block of the fs */
2244 };
2245
2246 /*
2247 * Check interval multiplier
2248 * The MMP block is written every update interval and initially checked every
2249 * update interval x the multiplier (the value is then adapted based on the
2250 * write latency). The reason is that writes can be delayed under load and we
2251 * don't want readers to incorrectly assume that the filesystem is no longer
2252 * in use.
2253 */
2254 #define EXT4_MMP_CHECK_MULT 2UL
2255
2256 /*
2257 * Minimum interval for MMP checking in seconds.
2258 */
2259 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
2260
2261 /*
2262 * Maximum interval for MMP checking in seconds.
2263 */
2264 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
2265
2266 /*
2267 * Function prototypes
2268 */
2269
2270 /*
2271 * Ok, these declarations are also in <linux/kernel.h> but none of the
2272 * ext4 source programs needs to include it so they are duplicated here.
2273 */
2274 # define NORET_TYPE /**/
2275 # define ATTRIB_NORET __attribute__((noreturn))
2276 # define NORET_AND noreturn,
2277
2278 /* bitmap.c */
2279 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2280 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2281 struct ext4_group_desc *gdp,
2282 struct buffer_head *bh, int sz);
2283 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2284 struct ext4_group_desc *gdp,
2285 struct buffer_head *bh, int sz);
2286 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2287 struct ext4_group_desc *gdp,
2288 struct buffer_head *bh);
2289 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2290 struct ext4_group_desc *gdp,
2291 struct buffer_head *bh);
2292
2293 /* balloc.c */
2294 extern void ext4_get_group_no_and_offset(struct super_block *sb,
2295 ext4_fsblk_t blocknr,
2296 ext4_group_t *blockgrpp,
2297 ext4_grpblk_t *offsetp);
2298 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2299 ext4_fsblk_t block);
2300
2301 extern unsigned int ext4_block_group(struct super_block *sb,
2302 ext4_fsblk_t blocknr);
2303 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2304 ext4_fsblk_t blocknr);
2305 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2306 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2307 ext4_group_t group);
2308 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2309 ext4_fsblk_t goal,
2310 unsigned int flags,
2311 unsigned long *count,
2312 int *errp);
2313 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2314 s64 nclusters, unsigned int flags);
2315 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2316 extern void ext4_check_blocks_bitmap(struct super_block *);
2317 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2318 ext4_group_t block_group,
2319 struct buffer_head ** bh);
2320 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2321
2322 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2323 ext4_group_t block_group);
2324 extern int ext4_wait_block_bitmap(struct super_block *sb,
2325 ext4_group_t block_group,
2326 struct buffer_head *bh);
2327 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2328 ext4_group_t block_group);
2329 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2330 ext4_group_t block_group,
2331 struct ext4_group_desc *gdp);
2332 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2333
2334 static inline bool ext4_encrypted_inode(struct inode *inode)
2335 {
2336 return ext4_test_inode_flag(inode, EXT4_INODE_ENCRYPT);
2337 }
2338
2339 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2340 static inline int ext4_fname_setup_filename(struct inode *dir,
2341 const struct qstr *iname,
2342 int lookup, struct ext4_filename *fname)
2343 {
2344 struct fscrypt_name name;
2345 int err;
2346
2347 memset(fname, 0, sizeof(struct ext4_filename));
2348
2349 err = fscrypt_setup_filename(dir, iname, lookup, &name);
2350
2351 fname->usr_fname = name.usr_fname;
2352 fname->disk_name = name.disk_name;
2353 fname->hinfo.hash = name.hash;
2354 fname->hinfo.minor_hash = name.minor_hash;
2355 fname->crypto_buf = name.crypto_buf;
2356 return err;
2357 }
2358
2359 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2360 {
2361 struct fscrypt_name name;
2362
2363 name.crypto_buf = fname->crypto_buf;
2364 fscrypt_free_filename(&name);
2365
2366 fname->crypto_buf.name = NULL;
2367 fname->usr_fname = NULL;
2368 fname->disk_name.name = NULL;
2369 }
2370 #else
2371 static inline int ext4_fname_setup_filename(struct inode *dir,
2372 const struct qstr *iname,
2373 int lookup, struct ext4_filename *fname)
2374 {
2375 fname->usr_fname = iname;
2376 fname->disk_name.name = (unsigned char *) iname->name;
2377 fname->disk_name.len = iname->len;
2378 return 0;
2379 }
2380 static inline void ext4_fname_free_filename(struct ext4_filename *fname) { }
2381
2382 #endif
2383
2384 /* dir.c */
2385 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2386 struct file *,
2387 struct ext4_dir_entry_2 *,
2388 struct buffer_head *, char *, int,
2389 unsigned int);
2390 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2391 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2392 (de), (bh), (buf), (size), (offset)))
2393 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2394 __u32 minor_hash,
2395 struct ext4_dir_entry_2 *dirent,
2396 struct fscrypt_str *ent_name);
2397 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2398 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2399 struct buffer_head *bh,
2400 void *buf, int buf_size,
2401 struct ext4_filename *fname,
2402 struct ext4_dir_entry_2 **dest_de);
2403 void ext4_insert_dentry(struct inode *inode,
2404 struct ext4_dir_entry_2 *de,
2405 int buf_size,
2406 struct ext4_filename *fname);
2407 static inline void ext4_update_dx_flag(struct inode *inode)
2408 {
2409 if (!ext4_has_feature_dir_index(inode->i_sb)) {
2410 /* ext4_iget() should have caught this... */
2411 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb));
2412 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2413 }
2414 }
2415 static const unsigned char ext4_filetype_table[] = {
2416 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2417 };
2418
2419 static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2420 {
2421 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2422 return DT_UNKNOWN;
2423
2424 return ext4_filetype_table[filetype];
2425 }
2426 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2427 void *buf, int buf_size);
2428
2429 /* fsync.c */
2430 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2431
2432 /* hash.c */
2433 extern int ext4fs_dirhash(const char *name, int len, struct
2434 dx_hash_info *hinfo);
2435
2436 /* ialloc.c */
2437 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2438 const struct qstr *qstr, __u32 goal,
2439 uid_t *owner, __u32 i_flags,
2440 int handle_type, unsigned int line_no,
2441 int nblocks);
2442
2443 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner, i_flags) \
2444 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2445 i_flags, 0, 0, 0)
2446 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2447 type, nblocks) \
2448 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2449 0, (type), __LINE__, (nblocks))
2450
2451
2452 extern void ext4_free_inode(handle_t *, struct inode *);
2453 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2454 extern unsigned long ext4_count_free_inodes(struct super_block *);
2455 extern unsigned long ext4_count_dirs(struct super_block *);
2456 extern void ext4_check_inodes_bitmap(struct super_block *);
2457 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2458 extern int ext4_init_inode_table(struct super_block *sb,
2459 ext4_group_t group, int barrier);
2460 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2461
2462 /* mballoc.c */
2463 extern const struct file_operations ext4_seq_mb_groups_fops;
2464 extern long ext4_mb_stats;
2465 extern long ext4_mb_max_to_scan;
2466 extern int ext4_mb_init(struct super_block *);
2467 extern int ext4_mb_release(struct super_block *);
2468 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2469 struct ext4_allocation_request *, int *);
2470 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2471 extern void ext4_discard_preallocations(struct inode *);
2472 extern int __init ext4_init_mballoc(void);
2473 extern void ext4_exit_mballoc(void);
2474 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2475 struct buffer_head *bh, ext4_fsblk_t block,
2476 unsigned long count, int flags);
2477 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2478 ext4_group_t ngroups);
2479 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2480 ext4_group_t i, struct ext4_group_desc *desc);
2481 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2482 ext4_fsblk_t block, unsigned long count);
2483 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2484 extern void ext4_process_freed_data(struct super_block *sb, tid_t commit_tid);
2485
2486 /* inode.c */
2487 int ext4_inode_is_fast_symlink(struct inode *inode);
2488 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2489 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2490 int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count,
2491 bool wait, struct buffer_head **bhs);
2492 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2493 struct buffer_head *bh_result, int create);
2494 int ext4_get_block(struct inode *inode, sector_t iblock,
2495 struct buffer_head *bh_result, int create);
2496 int ext4_dio_get_block(struct inode *inode, sector_t iblock,
2497 struct buffer_head *bh_result, int create);
2498 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2499 struct buffer_head *bh, int create);
2500 int ext4_walk_page_buffers(handle_t *handle,
2501 struct buffer_head *head,
2502 unsigned from,
2503 unsigned to,
2504 int *partial,
2505 int (*fn)(handle_t *handle,
2506 struct buffer_head *bh));
2507 int do_journal_get_write_access(handle_t *handle,
2508 struct buffer_head *bh);
2509 #define FALL_BACK_TO_NONDELALLOC 1
2510 #define CONVERT_INLINE_DATA 2
2511
2512 typedef enum {
2513 EXT4_IGET_NORMAL = 0,
2514 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */
2515 EXT4_IGET_HANDLE = 0x0002 /* Inode # is from a handle */
2516 } ext4_iget_flags;
2517
2518 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
2519 ext4_iget_flags flags, const char *function,
2520 unsigned int line);
2521
2522 #define ext4_iget(sb, ino, flags) \
2523 __ext4_iget((sb), (ino), (flags), __func__, __LINE__)
2524
2525 extern int ext4_write_inode(struct inode *, struct writeback_control *);
2526 extern int ext4_setattr(struct dentry *, struct iattr *);
2527 extern int ext4_getattr(const struct path *, struct kstat *, u32, unsigned int);
2528 extern void ext4_evict_inode(struct inode *);
2529 extern void ext4_clear_inode(struct inode *);
2530 extern int ext4_file_getattr(const struct path *, struct kstat *, u32, unsigned int);
2531 extern int ext4_sync_inode(handle_t *, struct inode *);
2532 extern void ext4_dirty_inode(struct inode *, int);
2533 extern int ext4_change_inode_journal_flag(struct inode *, int);
2534 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2535 extern int ext4_inode_attach_jinode(struct inode *inode);
2536 extern int ext4_can_truncate(struct inode *inode);
2537 extern int ext4_truncate(struct inode *);
2538 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2539 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
2540 extern void ext4_set_inode_flags(struct inode *);
2541 extern int ext4_alloc_da_blocks(struct inode *inode);
2542 extern void ext4_set_aops(struct inode *inode);
2543 extern int ext4_writepage_trans_blocks(struct inode *);
2544 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2545 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2546 loff_t lstart, loff_t lend);
2547 extern int ext4_page_mkwrite(struct vm_fault *vmf);
2548 extern int ext4_filemap_fault(struct vm_fault *vmf);
2549 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2550 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
2551 extern void ext4_da_update_reserve_space(struct inode *inode,
2552 int used, int quota_claim);
2553 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
2554 ext4_fsblk_t pblk, ext4_lblk_t len);
2555 extern int ext4_get_next_extent(struct inode *inode, ext4_lblk_t lblk,
2556 unsigned int map_len,
2557 struct extent_status *result);
2558
2559 /* indirect.c */
2560 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2561 struct ext4_map_blocks *map, int flags);
2562 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2563 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2564 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2565 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2566 ext4_lblk_t start, ext4_lblk_t end);
2567
2568 /* ioctl.c */
2569 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2570 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2571
2572 /* migrate.c */
2573 extern int ext4_ext_migrate(struct inode *);
2574 extern int ext4_ind_migrate(struct inode *inode);
2575
2576 /* namei.c */
2577 extern int ext4_dirent_csum_verify(struct inode *inode,
2578 struct ext4_dir_entry *dirent);
2579 extern int ext4_orphan_add(handle_t *, struct inode *);
2580 extern int ext4_orphan_del(handle_t *, struct inode *);
2581 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2582 __u32 start_minor_hash, __u32 *next_hash);
2583 extern int ext4_search_dir(struct buffer_head *bh,
2584 char *search_buf,
2585 int buf_size,
2586 struct inode *dir,
2587 struct ext4_filename *fname,
2588 unsigned int offset,
2589 struct ext4_dir_entry_2 **res_dir);
2590 extern int ext4_generic_delete_entry(handle_t *handle,
2591 struct inode *dir,
2592 struct ext4_dir_entry_2 *de_del,
2593 struct buffer_head *bh,
2594 void *entry_buf,
2595 int buf_size,
2596 int csum_size);
2597 extern bool ext4_empty_dir(struct inode *inode);
2598
2599 /* resize.c */
2600 extern void ext4_kvfree_array_rcu(void *to_free);
2601 extern int ext4_group_add(struct super_block *sb,
2602 struct ext4_new_group_data *input);
2603 extern int ext4_group_extend(struct super_block *sb,
2604 struct ext4_super_block *es,
2605 ext4_fsblk_t n_blocks_count);
2606 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2607
2608 /* super.c */
2609 extern struct buffer_head *ext4_sb_bread(struct super_block *sb,
2610 sector_t block, int op_flags);
2611 extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
2612 extern int ext4_calculate_overhead(struct super_block *sb);
2613 extern void ext4_superblock_csum_set(struct super_block *sb);
2614 extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2615 extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2616 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2617 ext4_group_t ngroup);
2618 extern const char *ext4_decode_error(struct super_block *sb, int errno,
2619 char nbuf[16]);
2620
2621 extern __printf(4, 5)
2622 void __ext4_error(struct super_block *, const char *, unsigned int,
2623 const char *, ...);
2624 extern __printf(5, 6)
2625 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2626 const char *, ...);
2627 extern __printf(5, 6)
2628 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2629 const char *, ...);
2630 extern void __ext4_std_error(struct super_block *, const char *,
2631 unsigned int, int);
2632 extern __printf(4, 5)
2633 void __ext4_abort(struct super_block *, const char *, unsigned int,
2634 const char *, ...);
2635 extern __printf(4, 5)
2636 void __ext4_warning(struct super_block *, const char *, unsigned int,
2637 const char *, ...);
2638 extern __printf(4, 5)
2639 void __ext4_warning_inode(const struct inode *inode, const char *function,
2640 unsigned int line, const char *fmt, ...);
2641 extern __printf(3, 4)
2642 void __ext4_msg(struct super_block *, const char *, const char *, ...);
2643 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2644 const char *, unsigned int, const char *);
2645 extern __printf(7, 8)
2646 void __ext4_grp_locked_error(const char *, unsigned int,
2647 struct super_block *, ext4_group_t,
2648 unsigned long, ext4_fsblk_t,
2649 const char *, ...);
2650
2651 #define EXT4_ERROR_INODE(inode, fmt, a...) \
2652 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
2653
2654 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
2655 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
2656
2657 #define EXT4_ERROR_FILE(file, block, fmt, a...) \
2658 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
2659
2660 #ifdef CONFIG_PRINTK
2661
2662 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2663 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__)
2664 #define ext4_error_file(file, func, line, block, fmt, ...) \
2665 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2666 #define ext4_error(sb, fmt, ...) \
2667 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2668 #define ext4_abort(sb, fmt, ...) \
2669 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2670 #define ext4_warning(sb, fmt, ...) \
2671 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2672 #define ext4_warning_inode(inode, fmt, ...) \
2673 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
2674 #define ext4_msg(sb, level, fmt, ...) \
2675 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2676 #define dump_mmp_msg(sb, mmp, msg) \
2677 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2678 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2679 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2680 fmt, ##__VA_ARGS__)
2681
2682 #else
2683
2684 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2685 do { \
2686 no_printk(fmt, ##__VA_ARGS__); \
2687 __ext4_error_inode(inode, "", 0, block, " "); \
2688 } while (0)
2689 #define ext4_error_file(file, func, line, block, fmt, ...) \
2690 do { \
2691 no_printk(fmt, ##__VA_ARGS__); \
2692 __ext4_error_file(file, "", 0, block, " "); \
2693 } while (0)
2694 #define ext4_error(sb, fmt, ...) \
2695 do { \
2696 no_printk(fmt, ##__VA_ARGS__); \
2697 __ext4_error(sb, "", 0, " "); \
2698 } while (0)
2699 #define ext4_abort(sb, fmt, ...) \
2700 do { \
2701 no_printk(fmt, ##__VA_ARGS__); \
2702 __ext4_abort(sb, "", 0, " "); \
2703 } while (0)
2704 #define ext4_warning(sb, fmt, ...) \
2705 do { \
2706 no_printk(fmt, ##__VA_ARGS__); \
2707 __ext4_warning(sb, "", 0, " "); \
2708 } while (0)
2709 #define ext4_warning_inode(inode, fmt, ...) \
2710 do { \
2711 no_printk(fmt, ##__VA_ARGS__); \
2712 __ext4_warning_inode(inode, "", 0, " "); \
2713 } while (0)
2714 #define ext4_msg(sb, level, fmt, ...) \
2715 do { \
2716 no_printk(fmt, ##__VA_ARGS__); \
2717 __ext4_msg(sb, "", " "); \
2718 } while (0)
2719 #define dump_mmp_msg(sb, mmp, msg) \
2720 __dump_mmp_msg(sb, mmp, "", 0, "")
2721 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2722 do { \
2723 no_printk(fmt, ##__VA_ARGS__); \
2724 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
2725 } while (0)
2726
2727 #endif
2728
2729 extern void ext4_update_dynamic_rev(struct super_block *sb);
2730 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2731 __u32 compat);
2732 extern int ext4_update_rocompat_feature(handle_t *handle,
2733 struct super_block *sb, __u32 rocompat);
2734 extern int ext4_update_incompat_feature(handle_t *handle,
2735 struct super_block *sb, __u32 incompat);
2736 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2737 struct ext4_group_desc *bg);
2738 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2739 struct ext4_group_desc *bg);
2740 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2741 struct ext4_group_desc *bg);
2742 extern __u32 ext4_free_group_clusters(struct super_block *sb,
2743 struct ext4_group_desc *bg);
2744 extern __u32 ext4_free_inodes_count(struct super_block *sb,
2745 struct ext4_group_desc *bg);
2746 extern __u32 ext4_used_dirs_count(struct super_block *sb,
2747 struct ext4_group_desc *bg);
2748 extern __u32 ext4_itable_unused_count(struct super_block *sb,
2749 struct ext4_group_desc *bg);
2750 extern void ext4_block_bitmap_set(struct super_block *sb,
2751 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2752 extern void ext4_inode_bitmap_set(struct super_block *sb,
2753 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2754 extern void ext4_inode_table_set(struct super_block *sb,
2755 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2756 extern void ext4_free_group_clusters_set(struct super_block *sb,
2757 struct ext4_group_desc *bg,
2758 __u32 count);
2759 extern void ext4_free_inodes_set(struct super_block *sb,
2760 struct ext4_group_desc *bg, __u32 count);
2761 extern void ext4_used_dirs_set(struct super_block *sb,
2762 struct ext4_group_desc *bg, __u32 count);
2763 extern void ext4_itable_unused_set(struct super_block *sb,
2764 struct ext4_group_desc *bg, __u32 count);
2765 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2766 struct ext4_group_desc *gdp);
2767 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2768 struct ext4_group_desc *gdp);
2769 extern int ext4_register_li_request(struct super_block *sb,
2770 ext4_group_t first_not_zeroed);
2771
2772 static inline int ext4_has_metadata_csum(struct super_block *sb)
2773 {
2774 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
2775 !EXT4_SB(sb)->s_chksum_driver);
2776
2777 return ext4_has_feature_metadata_csum(sb) &&
2778 (EXT4_SB(sb)->s_chksum_driver != NULL);
2779 }
2780
2781 static inline int ext4_has_group_desc_csum(struct super_block *sb)
2782 {
2783 return ext4_has_feature_gdt_csum(sb) || ext4_has_metadata_csum(sb);
2784 }
2785
2786 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2787 {
2788 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2789 le32_to_cpu(es->s_blocks_count_lo);
2790 }
2791
2792 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2793 {
2794 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2795 le32_to_cpu(es->s_r_blocks_count_lo);
2796 }
2797
2798 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2799 {
2800 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2801 le32_to_cpu(es->s_free_blocks_count_lo);
2802 }
2803
2804 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2805 ext4_fsblk_t blk)
2806 {
2807 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2808 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2809 }
2810
2811 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2812 ext4_fsblk_t blk)
2813 {
2814 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2815 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2816 }
2817
2818 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2819 ext4_fsblk_t blk)
2820 {
2821 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2822 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2823 }
2824
2825 static inline loff_t ext4_isize(struct super_block *sb,
2826 struct ext4_inode *raw_inode)
2827 {
2828 if (ext4_has_feature_largedir(sb) ||
2829 S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2830 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2831 le32_to_cpu(raw_inode->i_size_lo);
2832
2833 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2834 }
2835
2836 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2837 {
2838 raw_inode->i_size_lo = cpu_to_le32(i_size);
2839 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2840 }
2841
2842 static inline
2843 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2844 ext4_group_t group)
2845 {
2846 struct ext4_group_info **grp_info;
2847 long indexv, indexh;
2848 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
2849 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2850 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2851 grp_info = sbi_array_rcu_deref(EXT4_SB(sb), s_group_info, indexv);
2852 return grp_info[indexh];
2853 }
2854
2855 /*
2856 * Reading s_groups_count requires using smp_rmb() afterwards. See
2857 * the locking protocol documented in the comments of ext4_group_add()
2858 * in resize.c
2859 */
2860 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2861 {
2862 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
2863
2864 smp_rmb();
2865 return ngroups;
2866 }
2867
2868 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2869 ext4_group_t block_group)
2870 {
2871 return block_group >> sbi->s_log_groups_per_flex;
2872 }
2873
2874 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2875 {
2876 return 1 << sbi->s_log_groups_per_flex;
2877 }
2878
2879 #define ext4_std_error(sb, errno) \
2880 do { \
2881 if ((errno)) \
2882 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
2883 } while (0)
2884
2885 #ifdef CONFIG_SMP
2886 /* Each CPU can accumulate percpu_counter_batch clusters in their local
2887 * counters. So we need to make sure we have free clusters more
2888 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
2889 */
2890 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2891 #else
2892 #define EXT4_FREECLUSTERS_WATERMARK 0
2893 #endif
2894
2895 /* Update i_disksize. Requires i_mutex to avoid races with truncate */
2896 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2897 {
2898 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
2899 !inode_is_locked(inode));
2900 down_write(&EXT4_I(inode)->i_data_sem);
2901 if (newsize > EXT4_I(inode)->i_disksize)
2902 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize);
2903 up_write(&EXT4_I(inode)->i_data_sem);
2904 }
2905
2906 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
2907 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
2908 {
2909 int changed = 0;
2910
2911 if (newsize > inode->i_size) {
2912 i_size_write(inode, newsize);
2913 changed = 1;
2914 }
2915 if (newsize > EXT4_I(inode)->i_disksize) {
2916 ext4_update_i_disksize(inode, newsize);
2917 changed |= 2;
2918 }
2919 return changed;
2920 }
2921
2922 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
2923 loff_t len);
2924
2925 struct ext4_group_info {
2926 unsigned long bb_state;
2927 struct rb_root bb_free_root;
2928 ext4_grpblk_t bb_first_free; /* first free block */
2929 ext4_grpblk_t bb_free; /* total free blocks */
2930 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
2931 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
2932 struct list_head bb_prealloc_list;
2933 #ifdef DOUBLE_CHECK
2934 void *bb_bitmap;
2935 #endif
2936 struct rw_semaphore alloc_sem;
2937 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
2938 * regions, index is order.
2939 * bb_counters[3] = 5 means
2940 * 5 free 8-block regions. */
2941 };
2942
2943 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
2944 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
2945 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
2946 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
2947
2948 #define EXT4_MB_GRP_NEED_INIT(grp) \
2949 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2950 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
2951 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2952 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
2953 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2954
2955 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
2956 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2957 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
2958 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2959 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
2960 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2961
2962 #define EXT4_MAX_CONTENTION 8
2963 #define EXT4_CONTENTION_THRESHOLD 2
2964
2965 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2966 ext4_group_t group)
2967 {
2968 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2969 }
2970
2971 /*
2972 * Returns true if the filesystem is busy enough that attempts to
2973 * access the block group locks has run into contention.
2974 */
2975 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2976 {
2977 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2978 }
2979
2980 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2981 {
2982 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2983 if (spin_trylock(lock))
2984 /*
2985 * We're able to grab the lock right away, so drop the
2986 * lock contention counter.
2987 */
2988 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2989 else {
2990 /*
2991 * The lock is busy, so bump the contention counter,
2992 * and then wait on the spin lock.
2993 */
2994 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2995 EXT4_MAX_CONTENTION);
2996 spin_lock(lock);
2997 }
2998 }
2999
3000 static inline void ext4_unlock_group(struct super_block *sb,
3001 ext4_group_t group)
3002 {
3003 spin_unlock(ext4_group_lock_ptr(sb, group));
3004 }
3005
3006 /*
3007 * Block validity checking
3008 */
3009 #define ext4_check_indirect_blockref(inode, bh) \
3010 ext4_check_blockref(__func__, __LINE__, inode, \
3011 (__le32 *)(bh)->b_data, \
3012 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
3013
3014 #define ext4_ind_check_inode(inode) \
3015 ext4_check_blockref(__func__, __LINE__, inode, \
3016 EXT4_I(inode)->i_data, \
3017 EXT4_NDIR_BLOCKS)
3018
3019 /*
3020 * Inodes and files operations
3021 */
3022
3023 /* dir.c */
3024 extern const struct file_operations ext4_dir_operations;
3025
3026 /* file.c */
3027 extern const struct inode_operations ext4_file_inode_operations;
3028 extern const struct file_operations ext4_file_operations;
3029 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
3030
3031 /* inline.c */
3032 extern int ext4_get_max_inline_size(struct inode *inode);
3033 extern int ext4_find_inline_data_nolock(struct inode *inode);
3034 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
3035 unsigned int len);
3036 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
3037
3038 extern int ext4_readpage_inline(struct inode *inode, struct page *page);
3039 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3040 struct inode *inode,
3041 loff_t pos, unsigned len,
3042 unsigned flags,
3043 struct page **pagep);
3044 extern int ext4_write_inline_data_end(struct inode *inode,
3045 loff_t pos, unsigned len,
3046 unsigned copied,
3047 struct page *page);
3048 extern struct buffer_head *
3049 ext4_journalled_write_inline_data(struct inode *inode,
3050 unsigned len,
3051 struct page *page);
3052 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
3053 struct inode *inode,
3054 loff_t pos, unsigned len,
3055 unsigned flags,
3056 struct page **pagep,
3057 void **fsdata);
3058 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
3059 unsigned len, unsigned copied,
3060 struct page *page);
3061 extern int ext4_try_add_inline_entry(handle_t *handle,
3062 struct ext4_filename *fname,
3063 struct inode *dir, struct inode *inode);
3064 extern int ext4_try_create_inline_dir(handle_t *handle,
3065 struct inode *parent,
3066 struct inode *inode);
3067 extern int ext4_read_inline_dir(struct file *filp,
3068 struct dir_context *ctx,
3069 int *has_inline_data);
3070 extern int htree_inlinedir_to_tree(struct file *dir_file,
3071 struct inode *dir, ext4_lblk_t block,
3072 struct dx_hash_info *hinfo,
3073 __u32 start_hash, __u32 start_minor_hash,
3074 int *has_inline_data);
3075 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3076 struct ext4_filename *fname,
3077 struct ext4_dir_entry_2 **res_dir,
3078 int *has_inline_data);
3079 extern int ext4_delete_inline_entry(handle_t *handle,
3080 struct inode *dir,
3081 struct ext4_dir_entry_2 *de_del,
3082 struct buffer_head *bh,
3083 int *has_inline_data);
3084 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3085 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3086 struct ext4_dir_entry_2 **parent_de,
3087 int *retval);
3088 extern int ext4_inline_data_fiemap(struct inode *inode,
3089 struct fiemap_extent_info *fieinfo,
3090 int *has_inline, __u64 start, __u64 len);
3091 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3092
3093 extern int ext4_convert_inline_data(struct inode *inode);
3094
3095 static inline int ext4_has_inline_data(struct inode *inode)
3096 {
3097 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3098 EXT4_I(inode)->i_inline_off;
3099 }
3100
3101 /* namei.c */
3102 extern const struct inode_operations ext4_dir_inode_operations;
3103 extern const struct inode_operations ext4_special_inode_operations;
3104 extern struct dentry *ext4_get_parent(struct dentry *child);
3105 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3106 struct ext4_dir_entry_2 *de,
3107 int blocksize, int csum_size,
3108 unsigned int parent_ino, int dotdot_real_len);
3109 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t,
3110 unsigned int blocksize);
3111 extern int ext4_handle_dirty_dirent_node(handle_t *handle,
3112 struct inode *inode,
3113 struct buffer_head *bh);
3114 #define S_SHIFT 12
3115 static const unsigned char ext4_type_by_mode[(S_IFMT >> S_SHIFT) + 1] = {
3116 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
3117 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
3118 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
3119 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
3120 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
3121 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
3122 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
3123 };
3124
3125 static inline void ext4_set_de_type(struct super_block *sb,
3126 struct ext4_dir_entry_2 *de,
3127 umode_t mode) {
3128 if (ext4_has_feature_filetype(sb))
3129 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3130 }
3131
3132 /* readpages.c */
3133 extern int ext4_mpage_readpages(struct address_space *mapping,
3134 struct list_head *pages, struct page *page,
3135 unsigned nr_pages);
3136
3137 /* symlink.c */
3138 extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3139 extern const struct inode_operations ext4_symlink_inode_operations;
3140 extern const struct inode_operations ext4_fast_symlink_inode_operations;
3141
3142 /* sysfs.c */
3143 extern int ext4_register_sysfs(struct super_block *sb);
3144 extern void ext4_unregister_sysfs(struct super_block *sb);
3145 extern int __init ext4_init_sysfs(void);
3146 extern void ext4_exit_sysfs(void);
3147
3148 /* block_validity */
3149 extern void ext4_release_system_zone(struct super_block *sb);
3150 extern int ext4_setup_system_zone(struct super_block *sb);
3151 extern int __init ext4_init_system_zone(void);
3152 extern void ext4_exit_system_zone(void);
3153 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
3154 ext4_fsblk_t start_blk,
3155 unsigned int count);
3156 extern int ext4_check_blockref(const char *, unsigned int,
3157 struct inode *, __le32 *, unsigned int);
3158
3159 /* extents.c */
3160 struct ext4_ext_path;
3161 struct ext4_extent;
3162
3163 /*
3164 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3165 * __le32.
3166 */
3167 #define EXT_MAX_BLOCKS 0xffffffff
3168
3169 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
3170 extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
3171 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3172 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3173 struct ext4_map_blocks *map, int flags);
3174 extern int ext4_ext_truncate(handle_t *, struct inode *);
3175 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3176 ext4_lblk_t end);
3177 extern void ext4_ext_init(struct super_block *);
3178 extern void ext4_ext_release(struct super_block *);
3179 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3180 loff_t len);
3181 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3182 loff_t offset, ssize_t len);
3183 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3184 struct ext4_map_blocks *map, int flags);
3185 extern int ext4_ext_calc_metadata_amount(struct inode *inode,
3186 ext4_lblk_t lblocks);
3187 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3188 int num,
3189 struct ext4_ext_path *path);
3190 extern int ext4_can_extents_be_merged(struct inode *inode,
3191 struct ext4_extent *ex1,
3192 struct ext4_extent *ex2);
3193 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3194 struct ext4_ext_path **,
3195 struct ext4_extent *, int);
3196 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3197 struct ext4_ext_path **,
3198 int flags);
3199 extern void ext4_ext_drop_refs(struct ext4_ext_path *);
3200 extern int ext4_ext_check_inode(struct inode *inode);
3201 extern int ext4_find_delalloc_range(struct inode *inode,
3202 ext4_lblk_t lblk_start,
3203 ext4_lblk_t lblk_end);
3204 extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk);
3205 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3206 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3207 __u64 start, __u64 len);
3208 extern int ext4_ext_precache(struct inode *inode);
3209 extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len);
3210 extern int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len);
3211 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3212 struct inode *inode2, ext4_lblk_t lblk1,
3213 ext4_lblk_t lblk2, ext4_lblk_t count,
3214 int mark_unwritten,int *err);
3215
3216 /* move_extent.c */
3217 extern void ext4_double_down_write_data_sem(struct inode *first,
3218 struct inode *second);
3219 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3220 struct inode *donor_inode);
3221 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3222 __u64 start_orig, __u64 start_donor,
3223 __u64 len, __u64 *moved_len);
3224
3225 /* page-io.c */
3226 extern int __init ext4_init_pageio(void);
3227 extern void ext4_exit_pageio(void);
3228 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3229 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3230 extern int ext4_put_io_end(ext4_io_end_t *io_end);
3231 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3232 extern void ext4_io_submit_init(struct ext4_io_submit *io,
3233 struct writeback_control *wbc);
3234 extern void ext4_end_io_rsv_work(struct work_struct *work);
3235 extern void ext4_io_submit(struct ext4_io_submit *io);
3236 extern int ext4_bio_write_page(struct ext4_io_submit *io,
3237 struct page *page,
3238 int len,
3239 struct writeback_control *wbc,
3240 bool keep_towrite);
3241
3242 /* mmp.c */
3243 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3244
3245 /*
3246 * Add new method to test whether block and inode bitmaps are properly
3247 * initialized. With uninit_bg reading the block from disk is not enough
3248 * to mark the bitmap uptodate. We need to also zero-out the bitmap
3249 */
3250 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3251
3252 static inline int bitmap_uptodate(struct buffer_head *bh)
3253 {
3254 return (buffer_uptodate(bh) &&
3255 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3256 }
3257 static inline void set_bitmap_uptodate(struct buffer_head *bh)
3258 {
3259 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3260 }
3261
3262 /*
3263 * Disable DIO read nolock optimization, so new dioreaders will be forced
3264 * to grab i_mutex
3265 */
3266 static inline void ext4_inode_block_unlocked_dio(struct inode *inode)
3267 {
3268 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
3269 smp_mb();
3270 }
3271 static inline void ext4_inode_resume_unlocked_dio(struct inode *inode)
3272 {
3273 smp_mb();
3274 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
3275 }
3276
3277 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
3278
3279 /* For ioend & aio unwritten conversion wait queues */
3280 #define EXT4_WQ_HASH_SZ 37
3281 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
3282 EXT4_WQ_HASH_SZ])
3283 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3284
3285 extern int ext4_resize_begin(struct super_block *sb);
3286 extern void ext4_resize_end(struct super_block *sb);
3287
3288 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
3289 struct ext4_io_end *io_end)
3290 {
3291 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
3292 io_end->flag |= EXT4_IO_END_UNWRITTEN;
3293 atomic_inc(&EXT4_I(inode)->i_unwritten);
3294 }
3295 }
3296
3297 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3298 {
3299 struct inode *inode = io_end->inode;
3300
3301 if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
3302 io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3303 /* Wake up anyone waiting on unwritten extent conversion */
3304 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
3305 wake_up_all(ext4_ioend_wq(inode));
3306 }
3307 }
3308
3309 extern const struct iomap_ops ext4_iomap_ops;
3310
3311 #endif /* __KERNEL__ */
3312
3313 #define EFSBADCRC EBADMSG /* Bad CRC detected */
3314 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
3315
3316 #endif /* _EXT4_H */
Linux with added FPC-III support