1 /*
2 * ext4.h
3 *
4 * Copyright (C) 1992, 1993, 1994, 1995
5 * Remy Card (card@masi.ibp.fr)
6 * Laboratoire MASI - Institut Blaise Pascal
7 * Universite Pierre et Marie Curie (Paris VI)
8 *
9 * from
10 *
11 * linux/include/linux/minix_fs.h
12 *
13 * Copyright (C) 1991, 1992 Linus Torvalds
14 */
15
16 #ifndef _EXT4_H
17 #define _EXT4_H
18
19 #include <linux/types.h>
20 #include <linux/blkdev.h>
21 #include <linux/magic.h>
22 #include <linux/jbd2.h>
23 #include <linux/quota.h>
24 #include <linux/rwsem.h>
25 #include <linux/rbtree.h>
26 #include <linux/seqlock.h>
27 #include <linux/mutex.h>
28 #include <linux/timer.h>
29 #include <linux/version.h>
30 #include <linux/wait.h>
31 #include <linux/blockgroup_lock.h>
32 #include <linux/percpu_counter.h>
33 #include <linux/ratelimit.h>
34 #include <crypto/hash.h>
35 #include <linux/falloc.h>
36 #include <linux/percpu-rwsem.h>
37 #ifdef __KERNEL__
38 #include <linux/compat.h>
39 #endif
40
41 #define __FS_HAS_ENCRYPTION IS_ENABLED(CONFIG_EXT4_FS_ENCRYPTION)
42 #include <linux/fscrypt.h>
43
44 /*
45 * The fourth extended filesystem constants/structures
46 */
47
48 /*
49 * with AGGRESSIVE_CHECK allocator runs consistency checks over
50 * structures. these checks slow things down a lot
51 */
52 #define AGGRESSIVE_CHECK__
53
54 /*
55 * with DOUBLE_CHECK defined mballoc creates persistent in-core
56 * bitmaps, maintains and uses them to check for double allocations
57 */
58 #define DOUBLE_CHECK__
59
60 /*
61 * Define EXT4FS_DEBUG to produce debug messages
62 */
63 #undef EXT4FS_DEBUG
64
65 /*
66 * Debug code
67 */
68 #ifdef EXT4FS_DEBUG
69 #define ext4_debug(f, a...) \
70 do { \
71 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
72 __FILE__, __LINE__, __func__); \
73 printk(KERN_DEBUG f, ## a); \
74 } while (0)
75 #else
76 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
77 #endif
78
79 /*
80 * Turn on EXT_DEBUG to get lots of info about extents operations.
81 */
82 #define EXT_DEBUG__
83 #ifdef EXT_DEBUG
84 #define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__)
85 #else
86 #define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
87 #endif
88
89 /* data type for block offset of block group */
90 typedef int ext4_grpblk_t;
91
92 /* data type for filesystem-wide blocks number */
93 typedef unsigned long long ext4_fsblk_t;
94
95 /* data type for file logical block number */
96 typedef __u32 ext4_lblk_t;
97
98 /* data type for block group number */
99 typedef unsigned int ext4_group_t;
100
101 enum SHIFT_DIRECTION {
102 SHIFT_LEFT = 0,
103 SHIFT_RIGHT,
104 };
105
106 /*
107 * Flags used in mballoc's allocation_context flags field.
108 *
109 * Also used to show what's going on for debugging purposes when the
110 * flag field is exported via the traceport interface
111 */
112
113 /* prefer goal again. length */
114 #define EXT4_MB_HINT_MERGE 0x0001
115 /* blocks already reserved */
116 #define EXT4_MB_HINT_RESERVED 0x0002
117 /* metadata is being allocated */
118 #define EXT4_MB_HINT_METADATA 0x0004
119 /* first blocks in the file */
120 #define EXT4_MB_HINT_FIRST 0x0008
121 /* search for the best chunk */
122 #define EXT4_MB_HINT_BEST 0x0010
123 /* data is being allocated */
124 #define EXT4_MB_HINT_DATA 0x0020
125 /* don't preallocate (for tails) */
126 #define EXT4_MB_HINT_NOPREALLOC 0x0040
127 /* allocate for locality group */
128 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080
129 /* allocate goal blocks or none */
130 #define EXT4_MB_HINT_GOAL_ONLY 0x0100
131 /* goal is meaningful */
132 #define EXT4_MB_HINT_TRY_GOAL 0x0200
133 /* blocks already pre-reserved by delayed allocation */
134 #define EXT4_MB_DELALLOC_RESERVED 0x0400
135 /* We are doing stream allocation */
136 #define EXT4_MB_STREAM_ALLOC 0x0800
137 /* Use reserved root blocks if needed */
138 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000
139 /* Use blocks from reserved pool */
140 #define EXT4_MB_USE_RESERVED 0x2000
141
142 struct ext4_allocation_request {
143 /* target inode for block we're allocating */
144 struct inode *inode;
145 /* how many blocks we want to allocate */
146 unsigned int len;
147 /* logical block in target inode */
148 ext4_lblk_t logical;
149 /* the closest logical allocated block to the left */
150 ext4_lblk_t lleft;
151 /* the closest logical allocated block to the right */
152 ext4_lblk_t lright;
153 /* phys. target (a hint) */
154 ext4_fsblk_t goal;
155 /* phys. block for the closest logical allocated block to the left */
156 ext4_fsblk_t pleft;
157 /* phys. block for the closest logical allocated block to the right */
158 ext4_fsblk_t pright;
159 /* flags. see above EXT4_MB_HINT_* */
160 unsigned int flags;
161 };
162
163 /*
164 * Logical to physical block mapping, used by ext4_map_blocks()
165 *
166 * This structure is used to pass requests into ext4_map_blocks() as
167 * well as to store the information returned by ext4_map_blocks(). It
168 * takes less room on the stack than a struct buffer_head.
169 */
170 #define EXT4_MAP_NEW (1 << BH_New)
171 #define EXT4_MAP_MAPPED (1 << BH_Mapped)
172 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten)
173 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary)
174 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
175 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
176
177 struct ext4_map_blocks {
178 ext4_fsblk_t m_pblk;
179 ext4_lblk_t m_lblk;
180 unsigned int m_len;
181 unsigned int m_flags;
182 };
183
184 /*
185 * Flags for ext4_io_end->flags
186 */
187 #define EXT4_IO_END_UNWRITTEN 0x0001
188
189 /*
190 * For converting unwritten extents on a work queue. 'handle' is used for
191 * buffered writeback.
192 */
193 typedef struct ext4_io_end {
194 struct list_head list; /* per-file finished IO list */
195 handle_t *handle; /* handle reserved for extent
196 * conversion */
197 struct inode *inode; /* file being written to */
198 struct bio *bio; /* Linked list of completed
199 * bios covering the extent */
200 unsigned int flag; /* unwritten or not */
201 atomic_t count; /* reference counter */
202 loff_t offset; /* offset in the file */
203 ssize_t size; /* size of the extent */
204 } ext4_io_end_t;
205
206 struct ext4_io_submit {
207 struct writeback_control *io_wbc;
208 struct bio *io_bio;
209 ext4_io_end_t *io_end;
210 sector_t io_next_block;
211 };
212
213 /*
214 * Special inodes numbers
215 */
216 #define EXT4_BAD_INO 1 /* Bad blocks inode */
217 #define EXT4_ROOT_INO 2 /* Root inode */
218 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */
219 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
220 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
221 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
222 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
223 #define EXT4_JOURNAL_INO 8 /* Journal inode */
224
225 /* First non-reserved inode for old ext4 filesystems */
226 #define EXT4_GOOD_OLD_FIRST_INO 11
227
228 /*
229 * Maximal count of links to a file
230 */
231 #define EXT4_LINK_MAX 65000
232
233 /*
234 * Macro-instructions used to manage several block sizes
235 */
236 #define EXT4_MIN_BLOCK_SIZE 1024
237 #define EXT4_MAX_BLOCK_SIZE 65536
238 #define EXT4_MIN_BLOCK_LOG_SIZE 10
239 #define EXT4_MAX_BLOCK_LOG_SIZE 16
240 #define EXT4_MAX_CLUSTER_LOG_SIZE 30
241 #ifdef __KERNEL__
242 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
243 #else
244 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
245 #endif
246 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
247 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
248 EXT4_SB(s)->s_cluster_bits)
249 #ifdef __KERNEL__
250 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
251 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
252 #else
253 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
254 #endif
255 #ifdef __KERNEL__
256 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
257 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
258 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
259 #else
260 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
261 EXT4_GOOD_OLD_INODE_SIZE : \
262 (s)->s_inode_size)
263 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
264 EXT4_GOOD_OLD_FIRST_INO : \
265 (s)->s_first_ino)
266 #endif
267 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
268 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \
269 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \
270 blkbits))
271
272 /* Translate a block number to a cluster number */
273 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
274 /* Translate a cluster number to a block number */
275 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
276 /* Translate # of blks to # of clusters */
277 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
278 (sbi)->s_cluster_bits)
279 /* Mask out the low bits to get the starting block of the cluster */
280 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
281 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
282 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
283 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
284 /* Get the cluster offset */
285 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
286 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
287 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
288 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
289
290 /*
291 * Structure of a blocks group descriptor
292 */
293 struct ext4_group_desc
294 {
295 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
296 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
297 __le32 bg_inode_table_lo; /* Inodes table block */
298 __le16 bg_free_blocks_count_lo;/* Free blocks count */
299 __le16 bg_free_inodes_count_lo;/* Free inodes count */
300 __le16 bg_used_dirs_count_lo; /* Directories count */
301 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
302 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
303 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
304 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
305 __le16 bg_itable_unused_lo; /* Unused inodes count */
306 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
307 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
308 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
309 __le32 bg_inode_table_hi; /* Inodes table block MSB */
310 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
311 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
312 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
313 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
314 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
315 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
316 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
317 __u32 bg_reserved;
318 };
319
320 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
321 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
322 sizeof(__le16))
323 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
324 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
325 sizeof(__le16))
326
327 /*
328 * Structure of a flex block group info
329 */
330
331 struct flex_groups {
332 atomic64_t free_clusters;
333 atomic_t free_inodes;
334 atomic_t used_dirs;
335 };
336
337 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
338 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
339 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
340
341 /*
342 * Macro-instructions used to manage group descriptors
343 */
344 #define EXT4_MIN_DESC_SIZE 32
345 #define EXT4_MIN_DESC_SIZE_64BIT 64
346 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
347 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
348 #ifdef __KERNEL__
349 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
350 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
351 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
352 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
353 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
354 #else
355 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
356 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
357 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
358 #endif
359
360 /*
361 * Constants relative to the data blocks
362 */
363 #define EXT4_NDIR_BLOCKS 12
364 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
365 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
366 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
367 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
368
369 /*
370 * Inode flags
371 */
372 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
373 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
374 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
375 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
376 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
377 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
378 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
379 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
380 /* Reserved for compression usage... */
381 #define EXT4_DIRTY_FL 0x00000100
382 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
383 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
384 /* nb: was previously EXT2_ECOMPR_FL */
385 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */
386 /* End compression flags --- maybe not all used */
387 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
388 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
389 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
390 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
391 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
392 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
393 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
394 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
395 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
396 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
397 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
398 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
399 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
400
401 #define EXT4_FL_USER_VISIBLE 0x304BDFFF /* User visible flags */
402 #define EXT4_FL_USER_MODIFIABLE 0x204380FF /* User modifiable flags */
403
404 #define EXT4_FL_XFLAG_VISIBLE (EXT4_SYNC_FL | \
405 EXT4_IMMUTABLE_FL | \
406 EXT4_APPEND_FL | \
407 EXT4_NODUMP_FL | \
408 EXT4_NOATIME_FL | \
409 EXT4_PROJINHERIT_FL)
410
411 /* Flags that should be inherited by new inodes from their parent. */
412 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
413 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
414 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
415 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
416 EXT4_PROJINHERIT_FL)
417
418 /* Flags that are appropriate for regular files (all but dir-specific ones). */
419 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
420
421 /* Flags that are appropriate for non-directories/regular files. */
422 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
423
424 /* Mask out flags that are inappropriate for the given type of inode. */
ext4_mask_flags(umode_t mode,__u32 flags)425 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
426 {
427 if (S_ISDIR(mode))
428 return flags;
429 else if (S_ISREG(mode))
430 return flags & EXT4_REG_FLMASK;
431 else
432 return flags & EXT4_OTHER_FLMASK;
433 }
434
435 /*
436 * Inode flags used for atomic set/get
437 */
438 enum {
439 EXT4_INODE_SECRM = 0, /* Secure deletion */
440 EXT4_INODE_UNRM = 1, /* Undelete */
441 EXT4_INODE_COMPR = 2, /* Compress file */
442 EXT4_INODE_SYNC = 3, /* Synchronous updates */
443 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
444 EXT4_INODE_APPEND = 5, /* writes to file may only append */
445 EXT4_INODE_NODUMP = 6, /* do not dump file */
446 EXT4_INODE_NOATIME = 7, /* do not update atime */
447 /* Reserved for compression usage... */
448 EXT4_INODE_DIRTY = 8,
449 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
450 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
451 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */
452 /* End compression flags --- maybe not all used */
453 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
454 EXT4_INODE_IMAGIC = 13, /* AFS directory */
455 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
456 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
457 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
458 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
459 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
460 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
461 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
462 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
463 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
464 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */
465 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
466 };
467
468 /*
469 * Since it's pretty easy to mix up bit numbers and hex values, we use a
470 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
471 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
472 * any extra space in the compiled kernel image, otherwise, the build will fail.
473 * It's important that these values are the same, since we are using
474 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
475 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
476 * values found in ext2, ext3 and ext4 filesystems, and of course the values
477 * defined in e2fsprogs.
478 *
479 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
480 */
481 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
482 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
483
ext4_check_flag_values(void)484 static inline void ext4_check_flag_values(void)
485 {
486 CHECK_FLAG_VALUE(SECRM);
487 CHECK_FLAG_VALUE(UNRM);
488 CHECK_FLAG_VALUE(COMPR);
489 CHECK_FLAG_VALUE(SYNC);
490 CHECK_FLAG_VALUE(IMMUTABLE);
491 CHECK_FLAG_VALUE(APPEND);
492 CHECK_FLAG_VALUE(NODUMP);
493 CHECK_FLAG_VALUE(NOATIME);
494 CHECK_FLAG_VALUE(DIRTY);
495 CHECK_FLAG_VALUE(COMPRBLK);
496 CHECK_FLAG_VALUE(NOCOMPR);
497 CHECK_FLAG_VALUE(ENCRYPT);
498 CHECK_FLAG_VALUE(INDEX);
499 CHECK_FLAG_VALUE(IMAGIC);
500 CHECK_FLAG_VALUE(JOURNAL_DATA);
501 CHECK_FLAG_VALUE(NOTAIL);
502 CHECK_FLAG_VALUE(DIRSYNC);
503 CHECK_FLAG_VALUE(TOPDIR);
504 CHECK_FLAG_VALUE(HUGE_FILE);
505 CHECK_FLAG_VALUE(EXTENTS);
506 CHECK_FLAG_VALUE(EA_INODE);
507 CHECK_FLAG_VALUE(EOFBLOCKS);
508 CHECK_FLAG_VALUE(INLINE_DATA);
509 CHECK_FLAG_VALUE(PROJINHERIT);
510 CHECK_FLAG_VALUE(RESERVED);
511 }
512
513 /* Used to pass group descriptor data when online resize is done */
514 struct ext4_new_group_input {
515 __u32 group; /* Group number for this data */
516 __u64 block_bitmap; /* Absolute block number of block bitmap */
517 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
518 __u64 inode_table; /* Absolute block number of inode table start */
519 __u32 blocks_count; /* Total number of blocks in this group */
520 __u16 reserved_blocks; /* Number of reserved blocks in this group */
521 __u16 unused;
522 };
523
524 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
525 struct compat_ext4_new_group_input {
526 u32 group;
527 compat_u64 block_bitmap;
528 compat_u64 inode_bitmap;
529 compat_u64 inode_table;
530 u32 blocks_count;
531 u16 reserved_blocks;
532 u16 unused;
533 };
534 #endif
535
536 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
537 struct ext4_new_group_data {
538 __u32 group;
539 __u64 block_bitmap;
540 __u64 inode_bitmap;
541 __u64 inode_table;
542 __u32 blocks_count;
543 __u16 reserved_blocks;
544 __u16 unused;
545 __u32 free_blocks_count;
546 };
547
548 /* Indexes used to index group tables in ext4_new_group_data */
549 enum {
550 BLOCK_BITMAP = 0, /* block bitmap */
551 INODE_BITMAP, /* inode bitmap */
552 INODE_TABLE, /* inode tables */
553 GROUP_TABLE_COUNT,
554 };
555
556 /*
557 * Flags used by ext4_map_blocks()
558 */
559 /* Allocate any needed blocks and/or convert an unwritten
560 extent to be an initialized ext4 */
561 #define EXT4_GET_BLOCKS_CREATE 0x0001
562 /* Request the creation of an unwritten extent */
563 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002
564 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\
565 EXT4_GET_BLOCKS_CREATE)
566 /* Caller is from the delayed allocation writeout path
567 * finally doing the actual allocation of delayed blocks */
568 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
569 /* caller is from the direct IO path, request to creation of an
570 unwritten extents if not allocated, split the unwritten
571 extent if blocks has been preallocated already*/
572 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
573 #define EXT4_GET_BLOCKS_CONVERT 0x0010
574 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
575 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
576 /* Convert extent to initialized after IO complete */
577 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
578 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
579 /* Eventual metadata allocation (due to growing extent tree)
580 * should not fail, so try to use reserved blocks for that.*/
581 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
582 /* Don't normalize allocation size (used for fallocate) */
583 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
584 /* Request will not result in inode size update (user for fallocate) */
585 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080
586 /* Convert written extents to unwritten */
587 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100
588 /* Write zeros to newly created written extents */
589 #define EXT4_GET_BLOCKS_ZERO 0x0200
590 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\
591 EXT4_GET_BLOCKS_ZERO)
592 /* Caller will submit data before dropping transaction handle. This
593 * allows jbd2 to avoid submitting data before commit. */
594 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400
595
596 /*
597 * The bit position of these flags must not overlap with any of the
598 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(),
599 * read_extent_tree_block(), ext4_split_extent_at(),
600 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
601 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
602 * caching the extents when reading from the extent tree while a
603 * truncate or punch hole operation is in progress.
604 */
605 #define EXT4_EX_NOCACHE 0x40000000
606 #define EXT4_EX_FORCE_CACHE 0x20000000
607
608 /*
609 * Flags used by ext4_free_blocks
610 */
611 #define EXT4_FREE_BLOCKS_METADATA 0x0001
612 #define EXT4_FREE_BLOCKS_FORGET 0x0002
613 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
614 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
615 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
616 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
617
618 /*
619 * ioctl commands
620 */
621 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
622 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
623 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
624 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
625 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
626 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
627 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
628 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
629 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
630 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
631 #define EXT4_IOC_MIGRATE _IO('f', 9)
632 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
633 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
634 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
635 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
636 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
637 #define EXT4_IOC_SWAP_BOOT _IO('f', 17)
638 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18)
639 #define EXT4_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY
640 #define EXT4_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT
641 #define EXT4_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY
642
643 #ifndef FS_IOC_FSGETXATTR
644 /* Until the uapi changes get merged for project quota... */
645
646 #define FS_IOC_FSGETXATTR _IOR('X', 31, struct fsxattr)
647 #define FS_IOC_FSSETXATTR _IOW('X', 32, struct fsxattr)
648
649 /*
650 * Structure for FS_IOC_FSGETXATTR and FS_IOC_FSSETXATTR.
651 */
652 struct fsxattr {
653 __u32 fsx_xflags; /* xflags field value (get/set) */
654 __u32 fsx_extsize; /* extsize field value (get/set)*/
655 __u32 fsx_nextents; /* nextents field value (get) */
656 __u32 fsx_projid; /* project identifier (get/set) */
657 unsigned char fsx_pad[12];
658 };
659
660 /*
661 * Flags for the fsx_xflags field
662 */
663 #define FS_XFLAG_REALTIME 0x00000001 /* data in realtime volume */
664 #define FS_XFLAG_PREALLOC 0x00000002 /* preallocated file extents */
665 #define FS_XFLAG_IMMUTABLE 0x00000008 /* file cannot be modified */
666 #define FS_XFLAG_APPEND 0x00000010 /* all writes append */
667 #define FS_XFLAG_SYNC 0x00000020 /* all writes synchronous */
668 #define FS_XFLAG_NOATIME 0x00000040 /* do not update access time */
669 #define FS_XFLAG_NODUMP 0x00000080 /* do not include in backups */
670 #define FS_XFLAG_RTINHERIT 0x00000100 /* create with rt bit set */
671 #define FS_XFLAG_PROJINHERIT 0x00000200 /* create with parents projid */
672 #define FS_XFLAG_NOSYMLINKS 0x00000400 /* disallow symlink creation */
673 #define FS_XFLAG_EXTSIZE 0x00000800 /* extent size allocator hint */
674 #define FS_XFLAG_EXTSZINHERIT 0x00001000 /* inherit inode extent size */
675 #define FS_XFLAG_NODEFRAG 0x00002000 /* do not defragment */
676 #define FS_XFLAG_FILESTREAM 0x00004000 /* use filestream allocator */
677 #define FS_XFLAG_HASATTR 0x80000000 /* no DIFLAG for this */
678 #endif /* !defined(FS_IOC_FSGETXATTR) */
679
680 #define EXT4_IOC_FSGETXATTR FS_IOC_FSGETXATTR
681 #define EXT4_IOC_FSSETXATTR FS_IOC_FSSETXATTR
682
683 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
684 /*
685 * ioctl commands in 32 bit emulation
686 */
687 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
688 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
689 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
690 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
691 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
692 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
693 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
694 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
695 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
696 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
697 #endif
698
699 /* Max physical block we can address w/o extents */
700 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
701
702 /*
703 * Structure of an inode on the disk
704 */
705 struct ext4_inode {
706 __le16 i_mode; /* File mode */
707 __le16 i_uid; /* Low 16 bits of Owner Uid */
708 __le32 i_size_lo; /* Size in bytes */
709 __le32 i_atime; /* Access time */
710 __le32 i_ctime; /* Inode Change time */
711 __le32 i_mtime; /* Modification time */
712 __le32 i_dtime; /* Deletion Time */
713 __le16 i_gid; /* Low 16 bits of Group Id */
714 __le16 i_links_count; /* Links count */
715 __le32 i_blocks_lo; /* Blocks count */
716 __le32 i_flags; /* File flags */
717 union {
718 struct {
719 __le32 l_i_version;
720 } linux1;
721 struct {
722 __u32 h_i_translator;
723 } hurd1;
724 struct {
725 __u32 m_i_reserved1;
726 } masix1;
727 } osd1; /* OS dependent 1 */
728 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
729 __le32 i_generation; /* File version (for NFS) */
730 __le32 i_file_acl_lo; /* File ACL */
731 __le32 i_size_high;
732 __le32 i_obso_faddr; /* Obsoleted fragment address */
733 union {
734 struct {
735 __le16 l_i_blocks_high; /* were l_i_reserved1 */
736 __le16 l_i_file_acl_high;
737 __le16 l_i_uid_high; /* these 2 fields */
738 __le16 l_i_gid_high; /* were reserved2[0] */
739 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
740 __le16 l_i_reserved;
741 } linux2;
742 struct {
743 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
744 __u16 h_i_mode_high;
745 __u16 h_i_uid_high;
746 __u16 h_i_gid_high;
747 __u32 h_i_author;
748 } hurd2;
749 struct {
750 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
751 __le16 m_i_file_acl_high;
752 __u32 m_i_reserved2[2];
753 } masix2;
754 } osd2; /* OS dependent 2 */
755 __le16 i_extra_isize;
756 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
757 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
758 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
759 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
760 __le32 i_crtime; /* File Creation time */
761 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
762 __le32 i_version_hi; /* high 32 bits for 64-bit version */
763 __le32 i_projid; /* Project ID */
764 };
765
766 struct move_extent {
767 __u32 reserved; /* should be zero */
768 __u32 donor_fd; /* donor file descriptor */
769 __u64 orig_start; /* logical start offset in block for orig */
770 __u64 donor_start; /* logical start offset in block for donor */
771 __u64 len; /* block length to be moved */
772 __u64 moved_len; /* moved block length */
773 };
774
775 #define EXT4_EPOCH_BITS 2
776 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
777 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
778
779 /*
780 * Extended fields will fit into an inode if the filesystem was formatted
781 * with large inodes (-I 256 or larger) and there are not currently any EAs
782 * consuming all of the available space. For new inodes we always reserve
783 * enough space for the kernel's known extended fields, but for inodes
784 * created with an old kernel this might not have been the case. None of
785 * the extended inode fields is critical for correct filesystem operation.
786 * This macro checks if a certain field fits in the inode. Note that
787 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
788 */
789 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
790 ((offsetof(typeof(*ext4_inode), field) + \
791 sizeof((ext4_inode)->field)) \
792 <= (EXT4_GOOD_OLD_INODE_SIZE + \
793 (einode)->i_extra_isize)) \
794
795 /*
796 * We use an encoding that preserves the times for extra epoch "00":
797 *
798 * extra msb of adjust for signed
799 * epoch 32-bit 32-bit tv_sec to
800 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
801 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
802 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
803 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
804 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
805 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
806 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
807 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
808 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
809 *
810 * Note that previous versions of the kernel on 64-bit systems would
811 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
812 * 1970. e2fsck will correct this, assuming that it is run on the
813 * affected filesystem before 2242.
814 */
815
ext4_encode_extra_time(struct timespec * time)816 static inline __le32 ext4_encode_extra_time(struct timespec *time)
817 {
818 u32 extra = sizeof(time->tv_sec) > 4 ?
819 ((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK : 0;
820 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
821 }
822
ext4_decode_extra_time(struct timespec * time,__le32 extra)823 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
824 {
825 if (unlikely(sizeof(time->tv_sec) > 4 &&
826 (extra & cpu_to_le32(EXT4_EPOCH_MASK)))) {
827 #if LINUX_VERSION_CODE < KERNEL_VERSION(4,20,0)
828 /* Handle legacy encoding of pre-1970 dates with epoch
829 * bits 1,1. We assume that by kernel version 4.20,
830 * everyone will have run fsck over the affected
831 * filesystems to correct the problem. (This
832 * backwards compatibility may be removed before this
833 * time, at the discretion of the ext4 developers.)
834 */
835 u64 extra_bits = le32_to_cpu(extra) & EXT4_EPOCH_MASK;
836 if (extra_bits == 3 && ((time->tv_sec) & 0x80000000) != 0)
837 extra_bits = 0;
838 time->tv_sec += extra_bits << 32;
839 #else
840 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
841 #endif
842 }
843 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
844 }
845
846 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
847 do { \
848 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
849 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
850 (raw_inode)->xtime ## _extra = \
851 ext4_encode_extra_time(&(inode)->xtime); \
852 } while (0)
853
854 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
855 do { \
856 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
857 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
858 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
859 (raw_inode)->xtime ## _extra = \
860 ext4_encode_extra_time(&(einode)->xtime); \
861 } while (0)
862
863 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
864 do { \
865 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
866 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
867 ext4_decode_extra_time(&(inode)->xtime, \
868 raw_inode->xtime ## _extra); \
869 else \
870 (inode)->xtime.tv_nsec = 0; \
871 } while (0)
872
873 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
874 do { \
875 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
876 (einode)->xtime.tv_sec = \
877 (signed)le32_to_cpu((raw_inode)->xtime); \
878 else \
879 (einode)->xtime.tv_sec = 0; \
880 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
881 ext4_decode_extra_time(&(einode)->xtime, \
882 raw_inode->xtime ## _extra); \
883 else \
884 (einode)->xtime.tv_nsec = 0; \
885 } while (0)
886
887 #define i_disk_version osd1.linux1.l_i_version
888
889 #if defined(__KERNEL__) || defined(__linux__)
890 #define i_reserved1 osd1.linux1.l_i_reserved1
891 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
892 #define i_blocks_high osd2.linux2.l_i_blocks_high
893 #define i_uid_low i_uid
894 #define i_gid_low i_gid
895 #define i_uid_high osd2.linux2.l_i_uid_high
896 #define i_gid_high osd2.linux2.l_i_gid_high
897 #define i_checksum_lo osd2.linux2.l_i_checksum_lo
898
899 #elif defined(__GNU__)
900
901 #define i_translator osd1.hurd1.h_i_translator
902 #define i_uid_high osd2.hurd2.h_i_uid_high
903 #define i_gid_high osd2.hurd2.h_i_gid_high
904 #define i_author osd2.hurd2.h_i_author
905
906 #elif defined(__masix__)
907
908 #define i_reserved1 osd1.masix1.m_i_reserved1
909 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
910 #define i_reserved2 osd2.masix2.m_i_reserved2
911
912 #endif /* defined(__KERNEL__) || defined(__linux__) */
913
914 #include "extents_status.h"
915
916 /*
917 * Lock subclasses for i_data_sem in the ext4_inode_info structure.
918 *
919 * These are needed to avoid lockdep false positives when we need to
920 * allocate blocks to the quota inode during ext4_map_blocks(), while
921 * holding i_data_sem for a normal (non-quota) inode. Since we don't
922 * do quota tracking for the quota inode, this avoids deadlock (as
923 * well as infinite recursion, since it isn't turtles all the way
924 * down...)
925 *
926 * I_DATA_SEM_NORMAL - Used for most inodes
927 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode
928 * where the second inode has larger inode number
929 * than the first
930 * I_DATA_SEM_QUOTA - Used for quota inodes only
931 */
932 enum {
933 I_DATA_SEM_NORMAL = 0,
934 I_DATA_SEM_OTHER,
935 I_DATA_SEM_QUOTA,
936 };
937
938
939 /*
940 * fourth extended file system inode data in memory
941 */
942 struct ext4_inode_info {
943 __le32 i_data[15]; /* unconverted */
944 __u32 i_dtime;
945 ext4_fsblk_t i_file_acl;
946
947 /*
948 * i_block_group is the number of the block group which contains
949 * this file's inode. Constant across the lifetime of the inode,
950 * it is ued for making block allocation decisions - we try to
951 * place a file's data blocks near its inode block, and new inodes
952 * near to their parent directory's inode.
953 */
954 ext4_group_t i_block_group;
955 ext4_lblk_t i_dir_start_lookup;
956 #if (BITS_PER_LONG < 64)
957 unsigned long i_state_flags; /* Dynamic state flags */
958 #endif
959 unsigned long i_flags;
960
961 /*
962 * Extended attributes can be read independently of the main file
963 * data. Taking i_mutex even when reading would cause contention
964 * between readers of EAs and writers of regular file data, so
965 * instead we synchronize on xattr_sem when reading or changing
966 * EAs.
967 */
968 struct rw_semaphore xattr_sem;
969
970 struct list_head i_orphan; /* unlinked but open inodes */
971
972 /*
973 * i_disksize keeps track of what the inode size is ON DISK, not
974 * in memory. During truncate, i_size is set to the new size by
975 * the VFS prior to calling ext4_truncate(), but the filesystem won't
976 * set i_disksize to 0 until the truncate is actually under way.
977 *
978 * The intent is that i_disksize always represents the blocks which
979 * are used by this file. This allows recovery to restart truncate
980 * on orphans if we crash during truncate. We actually write i_disksize
981 * into the on-disk inode when writing inodes out, instead of i_size.
982 *
983 * The only time when i_disksize and i_size may be different is when
984 * a truncate is in progress. The only things which change i_disksize
985 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
986 */
987 loff_t i_disksize;
988
989 /*
990 * i_data_sem is for serialising ext4_truncate() against
991 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
992 * data tree are chopped off during truncate. We can't do that in
993 * ext4 because whenever we perform intermediate commits during
994 * truncate, the inode and all the metadata blocks *must* be in a
995 * consistent state which allows truncation of the orphans to restart
996 * during recovery. Hence we must fix the get_block-vs-truncate race
997 * by other means, so we have i_data_sem.
998 */
999 struct rw_semaphore i_data_sem;
1000 /*
1001 * i_mmap_sem is for serializing page faults with truncate / punch hole
1002 * operations. We have to make sure that new page cannot be faulted in
1003 * a section of the inode that is being punched. We cannot easily use
1004 * i_data_sem for this since we need protection for the whole punch
1005 * operation and i_data_sem ranks below transaction start so we have
1006 * to occasionally drop it.
1007 */
1008 struct rw_semaphore i_mmap_sem;
1009 struct inode vfs_inode;
1010 struct jbd2_inode *jinode;
1011
1012 spinlock_t i_raw_lock; /* protects updates to the raw inode */
1013
1014 /*
1015 * File creation time. Its function is same as that of
1016 * struct timespec i_{a,c,m}time in the generic inode.
1017 */
1018 struct timespec i_crtime;
1019
1020 /* mballoc */
1021 struct list_head i_prealloc_list;
1022 spinlock_t i_prealloc_lock;
1023
1024 /* extents status tree */
1025 struct ext4_es_tree i_es_tree;
1026 rwlock_t i_es_lock;
1027 struct list_head i_es_list;
1028 unsigned int i_es_all_nr; /* protected by i_es_lock */
1029 unsigned int i_es_shk_nr; /* protected by i_es_lock */
1030 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for
1031 extents to shrink. Protected by
1032 i_es_lock */
1033
1034 /* ialloc */
1035 ext4_group_t i_last_alloc_group;
1036
1037 /* allocation reservation info for delalloc */
1038 /* In case of bigalloc, these refer to clusters rather than blocks */
1039 unsigned int i_reserved_data_blocks;
1040 unsigned int i_reserved_meta_blocks;
1041 unsigned int i_allocated_meta_blocks;
1042 ext4_lblk_t i_da_metadata_calc_last_lblock;
1043 int i_da_metadata_calc_len;
1044
1045 /* on-disk additional length */
1046 __u16 i_extra_isize;
1047
1048 /* Indicate the inline data space. */
1049 u16 i_inline_off;
1050 u16 i_inline_size;
1051
1052 #ifdef CONFIG_QUOTA
1053 /* quota space reservation, managed internally by quota code */
1054 qsize_t i_reserved_quota;
1055 #endif
1056
1057 /* Lock protecting lists below */
1058 spinlock_t i_completed_io_lock;
1059 /*
1060 * Completed IOs that need unwritten extents handling and have
1061 * transaction reserved
1062 */
1063 struct list_head i_rsv_conversion_list;
1064 struct work_struct i_rsv_conversion_work;
1065 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1066
1067 spinlock_t i_block_reservation_lock;
1068
1069 /*
1070 * Transactions that contain inode's metadata needed to complete
1071 * fsync and fdatasync, respectively.
1072 */
1073 tid_t i_sync_tid;
1074 tid_t i_datasync_tid;
1075
1076 #ifdef CONFIG_QUOTA
1077 struct dquot *i_dquot[MAXQUOTAS];
1078 #endif
1079
1080 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1081 __u32 i_csum_seed;
1082
1083 kprojid_t i_projid;
1084 };
1085
1086 /*
1087 * File system states
1088 */
1089 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
1090 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
1091 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
1092
1093 /*
1094 * Misc. filesystem flags
1095 */
1096 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
1097 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
1098 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
1099
1100 /*
1101 * Mount flags set via mount options or defaults
1102 */
1103 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
1104 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
1105 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
1106 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
1107 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
1108 #define EXT4_MOUNT_ERRORS_MASK 0x00070
1109 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
1110 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
1111 #ifdef CONFIG_FS_DAX
1112 #define EXT4_MOUNT_DAX 0x00200 /* Direct Access */
1113 #else
1114 #define EXT4_MOUNT_DAX 0
1115 #endif
1116 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
1117 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
1118 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
1119 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
1120 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
1121 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
1122 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
1123 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
1124 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
1125 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
1126 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */
1127 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota,
1128 * enable enforcement for hidden
1129 * quota files */
1130 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable
1131 * enforcement for hidden quota
1132 * files */
1133 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota
1134 * enforcement */
1135 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
1136 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
1137 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1138 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
1139 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
1140 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
1141 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
1142 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
1143
1144 /*
1145 * Mount flags set either automatically (could not be set by mount option)
1146 * based on per file system feature or property or in special cases such as
1147 * distinguishing between explicit mount option definition and default.
1148 */
1149 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1150 specified delalloc */
1151 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1152 size of blocksize * 8
1153 blocks */
1154 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1155 file systems */
1156
1157 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly
1158 specified journal checksum */
1159
1160 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1161 ~EXT4_MOUNT_##opt
1162 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1163 EXT4_MOUNT_##opt
1164 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1165 EXT4_MOUNT_##opt)
1166
1167 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1168 ~EXT4_MOUNT2_##opt
1169 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1170 EXT4_MOUNT2_##opt
1171 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1172 EXT4_MOUNT2_##opt)
1173
1174 #define ext4_test_and_set_bit __test_and_set_bit_le
1175 #define ext4_set_bit __set_bit_le
1176 #define ext4_set_bit_atomic ext2_set_bit_atomic
1177 #define ext4_test_and_clear_bit __test_and_clear_bit_le
1178 #define ext4_clear_bit __clear_bit_le
1179 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
1180 #define ext4_test_bit test_bit_le
1181 #define ext4_find_next_zero_bit find_next_zero_bit_le
1182 #define ext4_find_next_bit find_next_bit_le
1183
1184 extern void ext4_set_bits(void *bm, int cur, int len);
1185
1186 /*
1187 * Maximal mount counts between two filesystem checks
1188 */
1189 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1190 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1191
1192 /*
1193 * Behaviour when detecting errors
1194 */
1195 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1196 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1197 #define EXT4_ERRORS_PANIC 3 /* Panic */
1198 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1199
1200 /* Metadata checksum algorithm codes */
1201 #define EXT4_CRC32C_CHKSUM 1
1202
1203 /*
1204 * Structure of the super block
1205 */
1206 struct ext4_super_block {
1207 /*00*/ __le32 s_inodes_count; /* Inodes count */
1208 __le32 s_blocks_count_lo; /* Blocks count */
1209 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1210 __le32 s_free_blocks_count_lo; /* Free blocks count */
1211 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1212 __le32 s_first_data_block; /* First Data Block */
1213 __le32 s_log_block_size; /* Block size */
1214 __le32 s_log_cluster_size; /* Allocation cluster size */
1215 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1216 __le32 s_clusters_per_group; /* # Clusters per group */
1217 __le32 s_inodes_per_group; /* # Inodes per group */
1218 __le32 s_mtime; /* Mount time */
1219 /*30*/ __le32 s_wtime; /* Write time */
1220 __le16 s_mnt_count; /* Mount count */
1221 __le16 s_max_mnt_count; /* Maximal mount count */
1222 __le16 s_magic; /* Magic signature */
1223 __le16 s_state; /* File system state */
1224 __le16 s_errors; /* Behaviour when detecting errors */
1225 __le16 s_minor_rev_level; /* minor revision level */
1226 /*40*/ __le32 s_lastcheck; /* time of last check */
1227 __le32 s_checkinterval; /* max. time between checks */
1228 __le32 s_creator_os; /* OS */
1229 __le32 s_rev_level; /* Revision level */
1230 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1231 __le16 s_def_resgid; /* Default gid for reserved blocks */
1232 /*
1233 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1234 *
1235 * Note: the difference between the compatible feature set and
1236 * the incompatible feature set is that if there is a bit set
1237 * in the incompatible feature set that the kernel doesn't
1238 * know about, it should refuse to mount the filesystem.
1239 *
1240 * e2fsck's requirements are more strict; if it doesn't know
1241 * about a feature in either the compatible or incompatible
1242 * feature set, it must abort and not try to meddle with
1243 * things it doesn't understand...
1244 */
1245 __le32 s_first_ino; /* First non-reserved inode */
1246 __le16 s_inode_size; /* size of inode structure */
1247 __le16 s_block_group_nr; /* block group # of this superblock */
1248 __le32 s_feature_compat; /* compatible feature set */
1249 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1250 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1251 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1252 /*78*/ char s_volume_name[16]; /* volume name */
1253 /*88*/ char s_last_mounted[64]; /* directory where last mounted */
1254 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1255 /*
1256 * Performance hints. Directory preallocation should only
1257 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1258 */
1259 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1260 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1261 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1262 /*
1263 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1264 */
1265 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1266 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1267 __le32 s_journal_dev; /* device number of journal file */
1268 __le32 s_last_orphan; /* start of list of inodes to delete */
1269 __le32 s_hash_seed[4]; /* HTREE hash seed */
1270 __u8 s_def_hash_version; /* Default hash version to use */
1271 __u8 s_jnl_backup_type;
1272 __le16 s_desc_size; /* size of group descriptor */
1273 /*100*/ __le32 s_default_mount_opts;
1274 __le32 s_first_meta_bg; /* First metablock block group */
1275 __le32 s_mkfs_time; /* When the filesystem was created */
1276 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1277 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1278 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1279 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1280 __le32 s_free_blocks_count_hi; /* Free blocks count */
1281 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1282 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1283 __le32 s_flags; /* Miscellaneous flags */
1284 __le16 s_raid_stride; /* RAID stride */
1285 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1286 __le64 s_mmp_block; /* Block for multi-mount protection */
1287 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1288 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1289 __u8 s_checksum_type; /* metadata checksum algorithm used */
1290 __u8 s_encryption_level; /* versioning level for encryption */
1291 __u8 s_reserved_pad; /* Padding to next 32bits */
1292 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1293 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1294 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1295 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1296 snapshot's future use */
1297 __le32 s_snapshot_list; /* inode number of the head of the
1298 on-disk snapshot list */
1299 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1300 __le32 s_error_count; /* number of fs errors */
1301 __le32 s_first_error_time; /* first time an error happened */
1302 __le32 s_first_error_ino; /* inode involved in first error */
1303 __le64 s_first_error_block; /* block involved of first error */
1304 __u8 s_first_error_func[32]; /* function where the error happened */
1305 __le32 s_first_error_line; /* line number where error happened */
1306 __le32 s_last_error_time; /* most recent time of an error */
1307 __le32 s_last_error_ino; /* inode involved in last error */
1308 __le32 s_last_error_line; /* line number where error happened */
1309 __le64 s_last_error_block; /* block involved of last error */
1310 __u8 s_last_error_func[32]; /* function where the error happened */
1311 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1312 __u8 s_mount_opts[64];
1313 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1314 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1315 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1316 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
1317 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */
1318 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */
1319 __le32 s_lpf_ino; /* Location of the lost+found inode */
1320 __le32 s_prj_quota_inum; /* inode for tracking project quota */
1321 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */
1322 __le32 s_reserved[98]; /* Padding to the end of the block */
1323 __le32 s_checksum; /* crc32c(superblock) */
1324 };
1325
1326 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1327
1328 #ifdef __KERNEL__
1329
1330 /*
1331 * run-time mount flags
1332 */
1333 #define EXT4_MF_MNTDIR_SAMPLED 0x0001
1334 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1335 #define EXT4_MF_TEST_DUMMY_ENCRYPTION 0x0004
1336
1337 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1338 #define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \
1339 EXT4_MF_TEST_DUMMY_ENCRYPTION))
1340 #else
1341 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1342 #endif
1343
1344 /* Number of quota types we support */
1345 #define EXT4_MAXQUOTAS 3
1346
1347 /*
1348 * fourth extended-fs super-block data in memory
1349 */
1350 struct ext4_sb_info {
1351 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1352 unsigned long s_inodes_per_block;/* Number of inodes per block */
1353 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1354 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1355 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1356 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1357 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1358 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1359 ext4_group_t s_groups_count; /* Number of groups in the fs */
1360 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1361 unsigned long s_overhead; /* # of fs overhead clusters */
1362 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1363 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1364 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1365 struct buffer_head * s_sbh; /* Buffer containing the super block */
1366 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1367 struct buffer_head **s_group_desc;
1368 unsigned int s_mount_opt;
1369 unsigned int s_mount_opt2;
1370 unsigned int s_mount_flags;
1371 unsigned int s_def_mount_opt;
1372 ext4_fsblk_t s_sb_block;
1373 atomic64_t s_resv_clusters;
1374 kuid_t s_resuid;
1375 kgid_t s_resgid;
1376 unsigned short s_mount_state;
1377 unsigned short s_pad;
1378 int s_addr_per_block_bits;
1379 int s_desc_per_block_bits;
1380 int s_inode_size;
1381 int s_first_ino;
1382 unsigned int s_inode_readahead_blks;
1383 unsigned int s_inode_goal;
1384 spinlock_t s_next_gen_lock;
1385 u32 s_next_generation;
1386 u32 s_hash_seed[4];
1387 int s_def_hash_version;
1388 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1389 struct percpu_counter s_freeclusters_counter;
1390 struct percpu_counter s_freeinodes_counter;
1391 struct percpu_counter s_dirs_counter;
1392 struct percpu_counter s_dirtyclusters_counter;
1393 struct blockgroup_lock *s_blockgroup_lock;
1394 struct proc_dir_entry *s_proc;
1395 struct kobject s_kobj;
1396 struct completion s_kobj_unregister;
1397 struct super_block *s_sb;
1398
1399 /* Journaling */
1400 struct journal_s *s_journal;
1401 struct list_head s_orphan;
1402 struct mutex s_orphan_lock;
1403 unsigned long s_resize_flags; /* Flags indicating if there
1404 is a resizer */
1405 unsigned long s_commit_interval;
1406 u32 s_max_batch_time;
1407 u32 s_min_batch_time;
1408 struct block_device *journal_bdev;
1409 #ifdef CONFIG_QUOTA
1410 char *s_qf_names[EXT4_MAXQUOTAS]; /* Names of quota files with journalled quota */
1411 int s_jquota_fmt; /* Format of quota to use */
1412 #endif
1413 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1414 struct rb_root system_blks;
1415
1416 #ifdef EXTENTS_STATS
1417 /* ext4 extents stats */
1418 unsigned long s_ext_min;
1419 unsigned long s_ext_max;
1420 unsigned long s_depth_max;
1421 spinlock_t s_ext_stats_lock;
1422 unsigned long s_ext_blocks;
1423 unsigned long s_ext_extents;
1424 #endif
1425
1426 /* for buddy allocator */
1427 struct ext4_group_info ***s_group_info;
1428 struct inode *s_buddy_cache;
1429 spinlock_t s_md_lock;
1430 unsigned short *s_mb_offsets;
1431 unsigned int *s_mb_maxs;
1432 unsigned int s_group_info_size;
1433 unsigned int s_mb_free_pending;
1434
1435 /* tunables */
1436 unsigned long s_stripe;
1437 unsigned int s_mb_stream_request;
1438 unsigned int s_mb_max_to_scan;
1439 unsigned int s_mb_min_to_scan;
1440 unsigned int s_mb_stats;
1441 unsigned int s_mb_order2_reqs;
1442 unsigned int s_mb_group_prealloc;
1443 unsigned int s_max_dir_size_kb;
1444 /* where last allocation was done - for stream allocation */
1445 unsigned long s_mb_last_group;
1446 unsigned long s_mb_last_start;
1447
1448 /* stats for buddy allocator */
1449 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1450 atomic_t s_bal_success; /* we found long enough chunks */
1451 atomic_t s_bal_allocated; /* in blocks */
1452 atomic_t s_bal_ex_scanned; /* total extents scanned */
1453 atomic_t s_bal_goals; /* goal hits */
1454 atomic_t s_bal_breaks; /* too long searches */
1455 atomic_t s_bal_2orders; /* 2^order hits */
1456 spinlock_t s_bal_lock;
1457 unsigned long s_mb_buddies_generated;
1458 unsigned long long s_mb_generation_time;
1459 atomic_t s_mb_lost_chunks;
1460 atomic_t s_mb_preallocated;
1461 atomic_t s_mb_discarded;
1462 atomic_t s_lock_busy;
1463
1464 /* locality groups */
1465 struct ext4_locality_group __percpu *s_locality_groups;
1466
1467 /* for write statistics */
1468 unsigned long s_sectors_written_start;
1469 u64 s_kbytes_written;
1470
1471 /* the size of zero-out chunk */
1472 unsigned int s_extent_max_zeroout_kb;
1473
1474 unsigned int s_log_groups_per_flex;
1475 struct flex_groups *s_flex_groups;
1476 ext4_group_t s_flex_groups_allocated;
1477
1478 /* workqueue for reserved extent conversions (buffered io) */
1479 struct workqueue_struct *rsv_conversion_wq;
1480
1481 /* timer for periodic error stats printing */
1482 struct timer_list s_err_report;
1483
1484 /* Lazy inode table initialization info */
1485 struct ext4_li_request *s_li_request;
1486 /* Wait multiplier for lazy initialization thread */
1487 unsigned int s_li_wait_mult;
1488
1489 /* Kernel thread for multiple mount protection */
1490 struct task_struct *s_mmp_tsk;
1491
1492 /* record the last minlen when FITRIM is called. */
1493 atomic_t s_last_trim_minblks;
1494
1495 /* Reference to checksum algorithm driver via cryptoapi */
1496 struct crypto_shash *s_chksum_driver;
1497
1498 /* Precomputed FS UUID checksum for seeding other checksums */
1499 __u32 s_csum_seed;
1500
1501 /* Reclaim extents from extent status tree */
1502 struct shrinker s_es_shrinker;
1503 struct list_head s_es_list; /* List of inodes with reclaimable extents */
1504 long s_es_nr_inode;
1505 struct ext4_es_stats s_es_stats;
1506 struct mb_cache *s_mb_cache;
1507 spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1508
1509 /* Ratelimit ext4 messages. */
1510 struct ratelimit_state s_err_ratelimit_state;
1511 struct ratelimit_state s_warning_ratelimit_state;
1512 struct ratelimit_state s_msg_ratelimit_state;
1513
1514 /* Barrier between changing inodes' journal flags and writepages ops. */
1515 struct percpu_rw_semaphore s_journal_flag_rwsem;
1516 };
1517
EXT4_SB(struct super_block * sb)1518 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1519 {
1520 return sb->s_fs_info;
1521 }
EXT4_I(struct inode * inode)1522 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1523 {
1524 return container_of(inode, struct ext4_inode_info, vfs_inode);
1525 }
1526
ext4_current_time(struct inode * inode)1527 static inline struct timespec ext4_current_time(struct inode *inode)
1528 {
1529 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1530 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1531 }
1532
ext4_valid_inum(struct super_block * sb,unsigned long ino)1533 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1534 {
1535 return ino == EXT4_ROOT_INO ||
1536 ino == EXT4_USR_QUOTA_INO ||
1537 ino == EXT4_GRP_QUOTA_INO ||
1538 ino == EXT4_BOOT_LOADER_INO ||
1539 ino == EXT4_JOURNAL_INO ||
1540 ino == EXT4_RESIZE_INO ||
1541 (ino >= EXT4_FIRST_INO(sb) &&
1542 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1543 }
1544
1545 /*
1546 * Inode dynamic state flags
1547 */
1548 enum {
1549 EXT4_STATE_JDATA, /* journaled data exists */
1550 EXT4_STATE_NEW, /* inode is newly created */
1551 EXT4_STATE_XATTR, /* has in-inode xattrs */
1552 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1553 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1554 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1555 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
1556 EXT4_STATE_NEWENTRY, /* File just added to dir */
1557 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read
1558 nolocking */
1559 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1560 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
1561 };
1562
1563 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1564 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1565 { \
1566 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1567 } \
1568 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1569 { \
1570 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1571 } \
1572 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1573 { \
1574 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1575 }
1576
1577 /* Add these declarations here only so that these functions can be
1578 * found by name. Otherwise, they are very hard to locate. */
1579 static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1580 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1581 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1582 EXT4_INODE_BIT_FNS(flag, flags, 0)
1583
1584 /* Add these declarations here only so that these functions can be
1585 * found by name. Otherwise, they are very hard to locate. */
1586 static inline int ext4_test_inode_state(struct inode *inode, int bit);
1587 static inline void ext4_set_inode_state(struct inode *inode, int bit);
1588 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1589 #if (BITS_PER_LONG < 64)
1590 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1591
ext4_clear_state_flags(struct ext4_inode_info * ei)1592 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1593 {
1594 (ei)->i_state_flags = 0;
1595 }
1596 #else
1597 EXT4_INODE_BIT_FNS(state, flags, 32)
1598
ext4_clear_state_flags(struct ext4_inode_info * ei)1599 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1600 {
1601 /* We depend on the fact that callers will set i_flags */
1602 }
1603 #endif
1604 #else
1605 /* Assume that user mode programs are passing in an ext4fs superblock, not
1606 * a kernel struct super_block. This will allow us to call the feature-test
1607 * macros from user land. */
1608 #define EXT4_SB(sb) (sb)
1609 #endif
1610
1611 /*
1612 * Returns true if the inode is inode is encrypted
1613 */
1614 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1615
1616 /*
1617 * Codes for operating systems
1618 */
1619 #define EXT4_OS_LINUX 0
1620 #define EXT4_OS_HURD 1
1621 #define EXT4_OS_MASIX 2
1622 #define EXT4_OS_FREEBSD 3
1623 #define EXT4_OS_LITES 4
1624
1625 /*
1626 * Revision levels
1627 */
1628 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1629 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1630
1631 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1632 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1633
1634 #define EXT4_GOOD_OLD_INODE_SIZE 128
1635
1636 /*
1637 * Feature set definitions
1638 */
1639
1640 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1641 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1642 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1643 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1644 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1645 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1646 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
1647
1648 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1649 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1650 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1651 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1652 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1653 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1654 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1655 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1656 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1657 /*
1658 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1659 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1660 * all other data structures' checksums. However, the METADATA_CSUM and
1661 * GDT_CSUM bits are mutually exclusive.
1662 */
1663 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
1664 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000
1665 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000
1666
1667 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1668 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1669 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1670 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1671 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1672 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1673 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1674 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1675 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1676 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1677 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1678 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000
1679 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
1680 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
1681 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000
1682
1683 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
1684 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1685 { \
1686 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1687 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
1688 } \
1689 static inline void ext4_set_feature_##name(struct super_block *sb) \
1690 { \
1691 EXT4_SB(sb)->s_es->s_feature_compat |= \
1692 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1693 } \
1694 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1695 { \
1696 EXT4_SB(sb)->s_es->s_feature_compat &= \
1697 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1698 }
1699
1700 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
1701 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1702 { \
1703 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1704 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
1705 } \
1706 static inline void ext4_set_feature_##name(struct super_block *sb) \
1707 { \
1708 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
1709 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1710 } \
1711 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1712 { \
1713 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
1714 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1715 }
1716
1717 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
1718 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1719 { \
1720 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1721 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
1722 } \
1723 static inline void ext4_set_feature_##name(struct super_block *sb) \
1724 { \
1725 EXT4_SB(sb)->s_es->s_feature_incompat |= \
1726 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1727 } \
1728 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1729 { \
1730 EXT4_SB(sb)->s_es->s_feature_incompat &= \
1731 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1732 }
1733
EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc,DIR_PREALLOC)1734 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC)
1735 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES)
1736 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL)
1737 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR)
1738 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE)
1739 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX)
1740 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2)
1741
1742 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER)
1743 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE)
1744 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR)
1745 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE)
1746 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM)
1747 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK)
1748 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE)
1749 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA)
1750 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC)
1751 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM)
1752 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY)
1753 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT)
1754
1755 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION)
1756 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE)
1757 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER)
1758 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV)
1759 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG)
1760 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS)
1761 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT)
1762 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP)
1763 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG)
1764 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE)
1765 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA)
1766 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED)
1767 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR)
1768 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA)
1769 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT)
1770
1771 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1772 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1773 EXT4_FEATURE_INCOMPAT_META_BG)
1774 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1775 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1776 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1777
1778 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1779 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1780 EXT4_FEATURE_INCOMPAT_RECOVER| \
1781 EXT4_FEATURE_INCOMPAT_META_BG)
1782 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1783 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1784 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1785
1786 #define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1787 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1788 EXT4_FEATURE_INCOMPAT_RECOVER| \
1789 EXT4_FEATURE_INCOMPAT_META_BG| \
1790 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1791 EXT4_FEATURE_INCOMPAT_64BIT| \
1792 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1793 EXT4_FEATURE_INCOMPAT_MMP | \
1794 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
1795 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
1796 EXT4_FEATURE_INCOMPAT_CSUM_SEED)
1797 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1798 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1799 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1800 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1801 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1802 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1803 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1804 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1805 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1806 EXT4_FEATURE_RO_COMPAT_QUOTA |\
1807 EXT4_FEATURE_RO_COMPAT_PROJECT)
1808
1809 #define EXTN_FEATURE_FUNCS(ver) \
1810 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
1811 { \
1812 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1813 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
1814 } \
1815 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
1816 { \
1817 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1818 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
1819 } \
1820 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
1821 { \
1822 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1823 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
1824 }
1825
1826 EXTN_FEATURE_FUNCS(2)
1827 EXTN_FEATURE_FUNCS(3)
1828 EXTN_FEATURE_FUNCS(4)
1829
1830 static inline bool ext4_has_compat_features(struct super_block *sb)
1831 {
1832 return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
1833 }
ext4_has_ro_compat_features(struct super_block * sb)1834 static inline bool ext4_has_ro_compat_features(struct super_block *sb)
1835 {
1836 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
1837 }
ext4_has_incompat_features(struct super_block * sb)1838 static inline bool ext4_has_incompat_features(struct super_block *sb)
1839 {
1840 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
1841 }
1842
1843 /*
1844 * Default values for user and/or group using reserved blocks
1845 */
1846 #define EXT4_DEF_RESUID 0
1847 #define EXT4_DEF_RESGID 0
1848
1849 /*
1850 * Default project ID
1851 */
1852 #define EXT4_DEF_PROJID 0
1853
1854 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
1855
1856 /*
1857 * Default mount options
1858 */
1859 #define EXT4_DEFM_DEBUG 0x0001
1860 #define EXT4_DEFM_BSDGROUPS 0x0002
1861 #define EXT4_DEFM_XATTR_USER 0x0004
1862 #define EXT4_DEFM_ACL 0x0008
1863 #define EXT4_DEFM_UID16 0x0010
1864 #define EXT4_DEFM_JMODE 0x0060
1865 #define EXT4_DEFM_JMODE_DATA 0x0020
1866 #define EXT4_DEFM_JMODE_ORDERED 0x0040
1867 #define EXT4_DEFM_JMODE_WBACK 0x0060
1868 #define EXT4_DEFM_NOBARRIER 0x0100
1869 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1870 #define EXT4_DEFM_DISCARD 0x0400
1871 #define EXT4_DEFM_NODELALLOC 0x0800
1872
1873 /*
1874 * Default journal batch times
1875 */
1876 #define EXT4_DEF_MIN_BATCH_TIME 0
1877 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1878
1879 /*
1880 * Minimum number of groups in a flexgroup before we separate out
1881 * directories into the first block group of a flexgroup
1882 */
1883 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1884
1885 /*
1886 * Structure of a directory entry
1887 */
1888 #define EXT4_NAME_LEN 255
1889
1890 struct ext4_dir_entry {
1891 __le32 inode; /* Inode number */
1892 __le16 rec_len; /* Directory entry length */
1893 __le16 name_len; /* Name length */
1894 char name[EXT4_NAME_LEN]; /* File name */
1895 };
1896
1897 /*
1898 * The new version of the directory entry. Since EXT4 structures are
1899 * stored in intel byte order, and the name_len field could never be
1900 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1901 * file_type field.
1902 */
1903 struct ext4_dir_entry_2 {
1904 __le32 inode; /* Inode number */
1905 __le16 rec_len; /* Directory entry length */
1906 __u8 name_len; /* Name length */
1907 __u8 file_type;
1908 char name[EXT4_NAME_LEN]; /* File name */
1909 };
1910
1911 /*
1912 * This is a bogus directory entry at the end of each leaf block that
1913 * records checksums.
1914 */
1915 struct ext4_dir_entry_tail {
1916 __le32 det_reserved_zero1; /* Pretend to be unused */
1917 __le16 det_rec_len; /* 12 */
1918 __u8 det_reserved_zero2; /* Zero name length */
1919 __u8 det_reserved_ft; /* 0xDE, fake file type */
1920 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
1921 };
1922
1923 #define EXT4_DIRENT_TAIL(block, blocksize) \
1924 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1925 ((blocksize) - \
1926 sizeof(struct ext4_dir_entry_tail))))
1927
1928 /*
1929 * Ext4 directory file types. Only the low 3 bits are used. The
1930 * other bits are reserved for now.
1931 */
1932 #define EXT4_FT_UNKNOWN 0
1933 #define EXT4_FT_REG_FILE 1
1934 #define EXT4_FT_DIR 2
1935 #define EXT4_FT_CHRDEV 3
1936 #define EXT4_FT_BLKDEV 4
1937 #define EXT4_FT_FIFO 5
1938 #define EXT4_FT_SOCK 6
1939 #define EXT4_FT_SYMLINK 7
1940
1941 #define EXT4_FT_MAX 8
1942
1943 #define EXT4_FT_DIR_CSUM 0xDE
1944
1945 /*
1946 * EXT4_DIR_PAD defines the directory entries boundaries
1947 *
1948 * NOTE: It must be a multiple of 4
1949 */
1950 #define EXT4_DIR_PAD 4
1951 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1952 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1953 ~EXT4_DIR_ROUND)
1954 #define EXT4_MAX_REC_LEN ((1<<16)-1)
1955
1956 /*
1957 * If we ever get support for fs block sizes > page_size, we'll need
1958 * to remove the #if statements in the next two functions...
1959 */
1960 static inline unsigned int
ext4_rec_len_from_disk(__le16 dlen,unsigned blocksize)1961 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1962 {
1963 unsigned len = le16_to_cpu(dlen);
1964
1965 #if (PAGE_SIZE >= 65536)
1966 if (len == EXT4_MAX_REC_LEN || len == 0)
1967 return blocksize;
1968 return (len & 65532) | ((len & 3) << 16);
1969 #else
1970 return len;
1971 #endif
1972 }
1973
ext4_rec_len_to_disk(unsigned len,unsigned blocksize)1974 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1975 {
1976 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1977 BUG();
1978 #if (PAGE_SIZE >= 65536)
1979 if (len < 65536)
1980 return cpu_to_le16(len);
1981 if (len == blocksize) {
1982 if (blocksize == 65536)
1983 return cpu_to_le16(EXT4_MAX_REC_LEN);
1984 else
1985 return cpu_to_le16(0);
1986 }
1987 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1988 #else
1989 return cpu_to_le16(len);
1990 #endif
1991 }
1992
1993 /*
1994 * Hash Tree Directory indexing
1995 * (c) Daniel Phillips, 2001
1996 */
1997
1998 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
1999 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2000 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
2001 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2002
2003 /* Legal values for the dx_root hash_version field: */
2004
2005 #define DX_HASH_LEGACY 0
2006 #define DX_HASH_HALF_MD4 1
2007 #define DX_HASH_TEA 2
2008 #define DX_HASH_LEGACY_UNSIGNED 3
2009 #define DX_HASH_HALF_MD4_UNSIGNED 4
2010 #define DX_HASH_TEA_UNSIGNED 5
2011
ext4_chksum(struct ext4_sb_info * sbi,u32 crc,const void * address,unsigned int length)2012 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2013 const void *address, unsigned int length)
2014 {
2015 struct {
2016 struct shash_desc shash;
2017 char ctx[4];
2018 } desc;
2019 int err;
2020
2021 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2022
2023 desc.shash.tfm = sbi->s_chksum_driver;
2024 desc.shash.flags = 0;
2025 *(u32 *)desc.ctx = crc;
2026
2027 err = crypto_shash_update(&desc.shash, address, length);
2028 BUG_ON(err);
2029
2030 return *(u32 *)desc.ctx;
2031 }
2032
2033 #ifdef __KERNEL__
2034
2035 /* hash info structure used by the directory hash */
2036 struct dx_hash_info
2037 {
2038 u32 hash;
2039 u32 minor_hash;
2040 int hash_version;
2041 u32 *seed;
2042 };
2043
2044
2045 /* 32 and 64 bit signed EOF for dx directories */
2046 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
2047 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
2048
2049
2050 /*
2051 * Control parameters used by ext4_htree_next_block
2052 */
2053 #define HASH_NB_ALWAYS 1
2054
2055 struct ext4_filename {
2056 const struct qstr *usr_fname;
2057 struct fscrypt_str disk_name;
2058 struct dx_hash_info hinfo;
2059 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2060 struct fscrypt_str crypto_buf;
2061 #endif
2062 };
2063
2064 #define fname_name(p) ((p)->disk_name.name)
2065 #define fname_len(p) ((p)->disk_name.len)
2066
2067 /*
2068 * Describe an inode's exact location on disk and in memory
2069 */
2070 struct ext4_iloc
2071 {
2072 struct buffer_head *bh;
2073 unsigned long offset;
2074 ext4_group_t block_group;
2075 };
2076
ext4_raw_inode(struct ext4_iloc * iloc)2077 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2078 {
2079 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2080 }
2081
2082 /*
2083 * This structure is stuffed into the struct file's private_data field
2084 * for directories. It is where we put information so that we can do
2085 * readdir operations in hash tree order.
2086 */
2087 struct dir_private_info {
2088 struct rb_root root;
2089 struct rb_node *curr_node;
2090 struct fname *extra_fname;
2091 loff_t last_pos;
2092 __u32 curr_hash;
2093 __u32 curr_minor_hash;
2094 __u32 next_hash;
2095 };
2096
2097 /* calculate the first block number of the group */
2098 static inline ext4_fsblk_t
ext4_group_first_block_no(struct super_block * sb,ext4_group_t group_no)2099 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2100 {
2101 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2102 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2103 }
2104
2105 /*
2106 * Special error return code only used by dx_probe() and its callers.
2107 */
2108 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
2109
2110 /*
2111 * Timeout and state flag for lazy initialization inode thread.
2112 */
2113 #define EXT4_DEF_LI_WAIT_MULT 10
2114 #define EXT4_DEF_LI_MAX_START_DELAY 5
2115 #define EXT4_LAZYINIT_QUIT 0x0001
2116 #define EXT4_LAZYINIT_RUNNING 0x0002
2117
2118 /*
2119 * Lazy inode table initialization info
2120 */
2121 struct ext4_lazy_init {
2122 unsigned long li_state;
2123 struct list_head li_request_list;
2124 struct mutex li_list_mtx;
2125 };
2126
2127 struct ext4_li_request {
2128 struct super_block *lr_super;
2129 struct ext4_sb_info *lr_sbi;
2130 ext4_group_t lr_next_group;
2131 struct list_head lr_request;
2132 unsigned long lr_next_sched;
2133 unsigned long lr_timeout;
2134 };
2135
2136 struct ext4_features {
2137 struct kobject f_kobj;
2138 struct completion f_kobj_unregister;
2139 };
2140
2141 /*
2142 * This structure will be used for multiple mount protection. It will be
2143 * written into the block number saved in the s_mmp_block field in the
2144 * superblock. Programs that check MMP should assume that if
2145 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2146 * to use the filesystem, regardless of how old the timestamp is.
2147 */
2148 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
2149 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2150 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
2151 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
2152
2153 struct mmp_struct {
2154 __le32 mmp_magic; /* Magic number for MMP */
2155 __le32 mmp_seq; /* Sequence no. updated periodically */
2156
2157 /*
2158 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2159 * purposes and do not affect the correctness of the algorithm
2160 */
2161 __le64 mmp_time; /* Time last updated */
2162 char mmp_nodename[64]; /* Node which last updated MMP block */
2163 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
2164
2165 /*
2166 * mmp_check_interval is used to verify if the MMP block has been
2167 * updated on the block device. The value is updated based on the
2168 * maximum time to write the MMP block during an update cycle.
2169 */
2170 __le16 mmp_check_interval;
2171
2172 __le16 mmp_pad1;
2173 __le32 mmp_pad2[226];
2174 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
2175 };
2176
2177 /* arguments passed to the mmp thread */
2178 struct mmpd_data {
2179 struct buffer_head *bh; /* bh from initial read_mmp_block() */
2180 struct super_block *sb; /* super block of the fs */
2181 };
2182
2183 /*
2184 * Check interval multiplier
2185 * The MMP block is written every update interval and initially checked every
2186 * update interval x the multiplier (the value is then adapted based on the
2187 * write latency). The reason is that writes can be delayed under load and we
2188 * don't want readers to incorrectly assume that the filesystem is no longer
2189 * in use.
2190 */
2191 #define EXT4_MMP_CHECK_MULT 2UL
2192
2193 /*
2194 * Minimum interval for MMP checking in seconds.
2195 */
2196 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
2197
2198 /*
2199 * Maximum interval for MMP checking in seconds.
2200 */
2201 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
2202
2203 /*
2204 * Function prototypes
2205 */
2206
2207 /*
2208 * Ok, these declarations are also in <linux/kernel.h> but none of the
2209 * ext4 source programs needs to include it so they are duplicated here.
2210 */
2211 # define NORET_TYPE /**/
2212 # define ATTRIB_NORET __attribute__((noreturn))
2213 # define NORET_AND noreturn,
2214
2215 /* bitmap.c */
2216 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2217 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2218 struct ext4_group_desc *gdp,
2219 struct buffer_head *bh, int sz);
2220 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2221 struct ext4_group_desc *gdp,
2222 struct buffer_head *bh, int sz);
2223 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2224 struct ext4_group_desc *gdp,
2225 struct buffer_head *bh);
2226 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2227 struct ext4_group_desc *gdp,
2228 struct buffer_head *bh);
2229
2230 /* balloc.c */
2231 extern void ext4_get_group_no_and_offset(struct super_block *sb,
2232 ext4_fsblk_t blocknr,
2233 ext4_group_t *blockgrpp,
2234 ext4_grpblk_t *offsetp);
2235 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2236 ext4_fsblk_t block);
2237
2238 extern unsigned int ext4_block_group(struct super_block *sb,
2239 ext4_fsblk_t blocknr);
2240 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2241 ext4_fsblk_t blocknr);
2242 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2243 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2244 ext4_group_t group);
2245 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2246 ext4_fsblk_t goal,
2247 unsigned int flags,
2248 unsigned long *count,
2249 int *errp);
2250 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2251 s64 nclusters, unsigned int flags);
2252 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2253 extern void ext4_check_blocks_bitmap(struct super_block *);
2254 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2255 ext4_group_t block_group,
2256 struct buffer_head ** bh);
2257 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2258
2259 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2260 ext4_group_t block_group);
2261 extern int ext4_wait_block_bitmap(struct super_block *sb,
2262 ext4_group_t block_group,
2263 struct buffer_head *bh);
2264 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2265 ext4_group_t block_group);
2266 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2267 ext4_group_t block_group,
2268 struct ext4_group_desc *gdp);
2269 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2270
ext4_encrypted_inode(struct inode * inode)2271 static inline bool ext4_encrypted_inode(struct inode *inode)
2272 {
2273 return ext4_test_inode_flag(inode, EXT4_INODE_ENCRYPT);
2274 }
2275
2276 #ifdef CONFIG_EXT4_FS_ENCRYPTION
ext4_fname_setup_filename(struct inode * dir,const struct qstr * iname,int lookup,struct ext4_filename * fname)2277 static inline int ext4_fname_setup_filename(struct inode *dir,
2278 const struct qstr *iname,
2279 int lookup, struct ext4_filename *fname)
2280 {
2281 struct fscrypt_name name;
2282 int err;
2283
2284 memset(fname, 0, sizeof(struct ext4_filename));
2285
2286 err = fscrypt_setup_filename(dir, iname, lookup, &name);
2287
2288 fname->usr_fname = name.usr_fname;
2289 fname->disk_name = name.disk_name;
2290 fname->hinfo.hash = name.hash;
2291 fname->hinfo.minor_hash = name.minor_hash;
2292 fname->crypto_buf = name.crypto_buf;
2293 return err;
2294 }
2295
ext4_fname_free_filename(struct ext4_filename * fname)2296 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2297 {
2298 struct fscrypt_name name;
2299
2300 name.crypto_buf = fname->crypto_buf;
2301 fscrypt_free_filename(&name);
2302
2303 fname->crypto_buf.name = NULL;
2304 fname->usr_fname = NULL;
2305 fname->disk_name.name = NULL;
2306 }
2307 #else
ext4_fname_setup_filename(struct inode * dir,const struct qstr * iname,int lookup,struct ext4_filename * fname)2308 static inline int ext4_fname_setup_filename(struct inode *dir,
2309 const struct qstr *iname,
2310 int lookup, struct ext4_filename *fname)
2311 {
2312 fname->usr_fname = iname;
2313 fname->disk_name.name = (unsigned char *) iname->name;
2314 fname->disk_name.len = iname->len;
2315 return 0;
2316 }
ext4_fname_free_filename(struct ext4_filename * fname)2317 static inline void ext4_fname_free_filename(struct ext4_filename *fname) { }
2318
2319 #endif
2320
2321 /* dir.c */
2322 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2323 struct file *,
2324 struct ext4_dir_entry_2 *,
2325 struct buffer_head *, char *, int,
2326 unsigned int);
2327 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2328 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2329 (de), (bh), (buf), (size), (offset)))
2330 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2331 __u32 minor_hash,
2332 struct ext4_dir_entry_2 *dirent,
2333 struct fscrypt_str *ent_name);
2334 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2335 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2336 struct buffer_head *bh,
2337 void *buf, int buf_size,
2338 struct ext4_filename *fname,
2339 struct ext4_dir_entry_2 **dest_de);
2340 void ext4_insert_dentry(struct inode *inode,
2341 struct ext4_dir_entry_2 *de,
2342 int buf_size,
2343 struct ext4_filename *fname);
ext4_update_dx_flag(struct inode * inode)2344 static inline void ext4_update_dx_flag(struct inode *inode)
2345 {
2346 if (!ext4_has_feature_dir_index(inode->i_sb))
2347 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2348 }
2349 static const unsigned char ext4_filetype_table[] = {
2350 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2351 };
2352
get_dtype(struct super_block * sb,int filetype)2353 static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2354 {
2355 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2356 return DT_UNKNOWN;
2357
2358 return ext4_filetype_table[filetype];
2359 }
2360 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2361 void *buf, int buf_size);
2362
2363 /* fsync.c */
2364 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2365
2366 /* hash.c */
2367 extern int ext4fs_dirhash(const char *name, int len, struct
2368 dx_hash_info *hinfo);
2369
2370 /* ialloc.c */
2371 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2372 const struct qstr *qstr, __u32 goal,
2373 uid_t *owner, int handle_type,
2374 unsigned int line_no, int nblocks);
2375
2376 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner) \
2377 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2378 0, 0, 0)
2379 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2380 type, nblocks) \
2381 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2382 (type), __LINE__, (nblocks))
2383
2384
2385 extern void ext4_free_inode(handle_t *, struct inode *);
2386 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2387 extern unsigned long ext4_count_free_inodes(struct super_block *);
2388 extern unsigned long ext4_count_dirs(struct super_block *);
2389 extern void ext4_check_inodes_bitmap(struct super_block *);
2390 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2391 extern int ext4_init_inode_table(struct super_block *sb,
2392 ext4_group_t group, int barrier);
2393 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2394
2395 /* mballoc.c */
2396 extern const struct file_operations ext4_seq_mb_groups_fops;
2397 extern long ext4_mb_stats;
2398 extern long ext4_mb_max_to_scan;
2399 extern int ext4_mb_init(struct super_block *);
2400 extern int ext4_mb_release(struct super_block *);
2401 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2402 struct ext4_allocation_request *, int *);
2403 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2404 extern void ext4_discard_preallocations(struct inode *);
2405 extern int __init ext4_init_mballoc(void);
2406 extern void ext4_exit_mballoc(void);
2407 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2408 struct buffer_head *bh, ext4_fsblk_t block,
2409 unsigned long count, int flags);
2410 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2411 ext4_group_t ngroups);
2412 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2413 ext4_group_t i, struct ext4_group_desc *desc);
2414 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2415 ext4_fsblk_t block, unsigned long count);
2416 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *,
2417 unsigned long blkdev_flags);
2418
2419 /* inode.c */
2420 int ext4_inode_is_fast_symlink(struct inode *inode);
2421 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2422 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2423 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2424 struct buffer_head *bh_result, int create);
2425 int ext4_dax_get_block(struct inode *inode, sector_t iblock,
2426 struct buffer_head *bh_result, int create);
2427 int ext4_get_block(struct inode *inode, sector_t iblock,
2428 struct buffer_head *bh_result, int create);
2429 int ext4_dio_get_block(struct inode *inode, sector_t iblock,
2430 struct buffer_head *bh_result, int create);
2431 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2432 struct buffer_head *bh, int create);
2433 int ext4_walk_page_buffers(handle_t *handle,
2434 struct buffer_head *head,
2435 unsigned from,
2436 unsigned to,
2437 int *partial,
2438 int (*fn)(handle_t *handle,
2439 struct buffer_head *bh));
2440 int do_journal_get_write_access(handle_t *handle,
2441 struct buffer_head *bh);
2442 #define FALL_BACK_TO_NONDELALLOC 1
2443 #define CONVERT_INLINE_DATA 2
2444
2445 extern struct inode *ext4_iget(struct super_block *, unsigned long);
2446 extern struct inode *ext4_iget_normal(struct super_block *, unsigned long);
2447 extern int ext4_write_inode(struct inode *, struct writeback_control *);
2448 extern int ext4_setattr(struct dentry *, struct iattr *);
2449 extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
2450 struct kstat *stat);
2451 extern void ext4_evict_inode(struct inode *);
2452 extern void ext4_clear_inode(struct inode *);
2453 extern int ext4_sync_inode(handle_t *, struct inode *);
2454 extern void ext4_dirty_inode(struct inode *, int);
2455 extern int ext4_change_inode_journal_flag(struct inode *, int);
2456 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2457 extern int ext4_inode_attach_jinode(struct inode *inode);
2458 extern int ext4_can_truncate(struct inode *inode);
2459 extern void ext4_truncate(struct inode *);
2460 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2461 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
2462 extern void ext4_set_inode_flags(struct inode *);
2463 extern void ext4_get_inode_flags(struct ext4_inode_info *);
2464 extern int ext4_alloc_da_blocks(struct inode *inode);
2465 extern void ext4_set_aops(struct inode *inode);
2466 extern int ext4_writepage_trans_blocks(struct inode *);
2467 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2468 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2469 loff_t lstart, loff_t lend);
2470 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
2471 extern int ext4_filemap_fault(struct vm_area_struct *vma, struct vm_fault *vmf);
2472 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2473 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
2474 extern void ext4_da_update_reserve_space(struct inode *inode,
2475 int used, int quota_claim);
2476 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
2477 ext4_fsblk_t pblk, ext4_lblk_t len);
2478 extern int ext4_get_next_extent(struct inode *inode, ext4_lblk_t lblk,
2479 unsigned int map_len,
2480 struct extent_status *result);
2481
2482 /* indirect.c */
2483 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2484 struct ext4_map_blocks *map, int flags);
2485 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2486 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2487 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2488 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2489 ext4_lblk_t start, ext4_lblk_t end);
2490
2491 /* ioctl.c */
2492 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2493 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2494
2495 /* migrate.c */
2496 extern int ext4_ext_migrate(struct inode *);
2497 extern int ext4_ind_migrate(struct inode *inode);
2498
2499 /* namei.c */
2500 extern int ext4_dirent_csum_verify(struct inode *inode,
2501 struct ext4_dir_entry *dirent);
2502 extern int ext4_orphan_add(handle_t *, struct inode *);
2503 extern int ext4_orphan_del(handle_t *, struct inode *);
2504 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2505 __u32 start_minor_hash, __u32 *next_hash);
2506 extern int ext4_search_dir(struct buffer_head *bh,
2507 char *search_buf,
2508 int buf_size,
2509 struct inode *dir,
2510 struct ext4_filename *fname,
2511 const struct qstr *d_name,
2512 unsigned int offset,
2513 struct ext4_dir_entry_2 **res_dir);
2514 extern int ext4_generic_delete_entry(handle_t *handle,
2515 struct inode *dir,
2516 struct ext4_dir_entry_2 *de_del,
2517 struct buffer_head *bh,
2518 void *entry_buf,
2519 int buf_size,
2520 int csum_size);
2521 extern bool ext4_empty_dir(struct inode *inode);
2522
2523 /* resize.c */
2524 extern int ext4_group_add(struct super_block *sb,
2525 struct ext4_new_group_data *input);
2526 extern int ext4_group_extend(struct super_block *sb,
2527 struct ext4_super_block *es,
2528 ext4_fsblk_t n_blocks_count);
2529 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2530
2531 /* super.c */
2532 extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
2533 extern int ext4_calculate_overhead(struct super_block *sb);
2534 extern void ext4_superblock_csum_set(struct super_block *sb);
2535 extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2536 extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2537 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2538 ext4_group_t ngroup);
2539 extern const char *ext4_decode_error(struct super_block *sb, int errno,
2540 char nbuf[16]);
2541
2542 extern __printf(4, 5)
2543 void __ext4_error(struct super_block *, const char *, unsigned int,
2544 const char *, ...);
2545 extern __printf(5, 6)
2546 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2547 const char *, ...);
2548 extern __printf(5, 6)
2549 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2550 const char *, ...);
2551 extern void __ext4_std_error(struct super_block *, const char *,
2552 unsigned int, int);
2553 extern __printf(4, 5)
2554 void __ext4_abort(struct super_block *, const char *, unsigned int,
2555 const char *, ...);
2556 extern __printf(4, 5)
2557 void __ext4_warning(struct super_block *, const char *, unsigned int,
2558 const char *, ...);
2559 extern __printf(4, 5)
2560 void __ext4_warning_inode(const struct inode *inode, const char *function,
2561 unsigned int line, const char *fmt, ...);
2562 extern __printf(3, 4)
2563 void __ext4_msg(struct super_block *, const char *, const char *, ...);
2564 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2565 const char *, unsigned int, const char *);
2566 extern __printf(7, 8)
2567 void __ext4_grp_locked_error(const char *, unsigned int,
2568 struct super_block *, ext4_group_t,
2569 unsigned long, ext4_fsblk_t,
2570 const char *, ...);
2571
2572 #define EXT4_ERROR_INODE(inode, fmt, a...) \
2573 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
2574
2575 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
2576 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
2577
2578 #define EXT4_ERROR_FILE(file, block, fmt, a...) \
2579 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
2580
2581 #ifdef CONFIG_PRINTK
2582
2583 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2584 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__)
2585 #define ext4_error_file(file, func, line, block, fmt, ...) \
2586 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2587 #define ext4_error(sb, fmt, ...) \
2588 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2589 #define ext4_abort(sb, fmt, ...) \
2590 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2591 #define ext4_warning(sb, fmt, ...) \
2592 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2593 #define ext4_warning_inode(inode, fmt, ...) \
2594 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
2595 #define ext4_msg(sb, level, fmt, ...) \
2596 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2597 #define dump_mmp_msg(sb, mmp, msg) \
2598 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2599 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2600 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2601 fmt, ##__VA_ARGS__)
2602
2603 #else
2604
2605 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2606 do { \
2607 no_printk(fmt, ##__VA_ARGS__); \
2608 __ext4_error_inode(inode, "", 0, block, " "); \
2609 } while (0)
2610 #define ext4_error_file(file, func, line, block, fmt, ...) \
2611 do { \
2612 no_printk(fmt, ##__VA_ARGS__); \
2613 __ext4_error_file(file, "", 0, block, " "); \
2614 } while (0)
2615 #define ext4_error(sb, fmt, ...) \
2616 do { \
2617 no_printk(fmt, ##__VA_ARGS__); \
2618 __ext4_error(sb, "", 0, " "); \
2619 } while (0)
2620 #define ext4_abort(sb, fmt, ...) \
2621 do { \
2622 no_printk(fmt, ##__VA_ARGS__); \
2623 __ext4_abort(sb, "", 0, " "); \
2624 } while (0)
2625 #define ext4_warning(sb, fmt, ...) \
2626 do { \
2627 no_printk(fmt, ##__VA_ARGS__); \
2628 __ext4_warning(sb, "", 0, " "); \
2629 } while (0)
2630 #define ext4_warning_inode(inode, fmt, ...) \
2631 do { \
2632 no_printk(fmt, ##__VA_ARGS__); \
2633 __ext4_warning_inode(inode, "", 0, " "); \
2634 } while (0)
2635 #define ext4_msg(sb, level, fmt, ...) \
2636 do { \
2637 no_printk(fmt, ##__VA_ARGS__); \
2638 __ext4_msg(sb, "", " "); \
2639 } while (0)
2640 #define dump_mmp_msg(sb, mmp, msg) \
2641 __dump_mmp_msg(sb, mmp, "", 0, "")
2642 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2643 do { \
2644 no_printk(fmt, ##__VA_ARGS__); \
2645 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
2646 } while (0)
2647
2648 #endif
2649
2650 extern void ext4_update_dynamic_rev(struct super_block *sb);
2651 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2652 __u32 compat);
2653 extern int ext4_update_rocompat_feature(handle_t *handle,
2654 struct super_block *sb, __u32 rocompat);
2655 extern int ext4_update_incompat_feature(handle_t *handle,
2656 struct super_block *sb, __u32 incompat);
2657 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2658 struct ext4_group_desc *bg);
2659 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2660 struct ext4_group_desc *bg);
2661 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2662 struct ext4_group_desc *bg);
2663 extern __u32 ext4_free_group_clusters(struct super_block *sb,
2664 struct ext4_group_desc *bg);
2665 extern __u32 ext4_free_inodes_count(struct super_block *sb,
2666 struct ext4_group_desc *bg);
2667 extern __u32 ext4_used_dirs_count(struct super_block *sb,
2668 struct ext4_group_desc *bg);
2669 extern __u32 ext4_itable_unused_count(struct super_block *sb,
2670 struct ext4_group_desc *bg);
2671 extern void ext4_block_bitmap_set(struct super_block *sb,
2672 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2673 extern void ext4_inode_bitmap_set(struct super_block *sb,
2674 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2675 extern void ext4_inode_table_set(struct super_block *sb,
2676 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2677 extern void ext4_free_group_clusters_set(struct super_block *sb,
2678 struct ext4_group_desc *bg,
2679 __u32 count);
2680 extern void ext4_free_inodes_set(struct super_block *sb,
2681 struct ext4_group_desc *bg, __u32 count);
2682 extern void ext4_used_dirs_set(struct super_block *sb,
2683 struct ext4_group_desc *bg, __u32 count);
2684 extern void ext4_itable_unused_set(struct super_block *sb,
2685 struct ext4_group_desc *bg, __u32 count);
2686 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2687 struct ext4_group_desc *gdp);
2688 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2689 struct ext4_group_desc *gdp);
2690 extern int ext4_register_li_request(struct super_block *sb,
2691 ext4_group_t first_not_zeroed);
2692
ext4_has_group_desc_csum(struct super_block * sb)2693 static inline int ext4_has_group_desc_csum(struct super_block *sb)
2694 {
2695 return ext4_has_feature_gdt_csum(sb) ||
2696 EXT4_SB(sb)->s_chksum_driver != NULL;
2697 }
2698
ext4_has_metadata_csum(struct super_block * sb)2699 static inline int ext4_has_metadata_csum(struct super_block *sb)
2700 {
2701 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
2702 !EXT4_SB(sb)->s_chksum_driver);
2703
2704 return (EXT4_SB(sb)->s_chksum_driver != NULL);
2705 }
ext4_blocks_count(struct ext4_super_block * es)2706 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2707 {
2708 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2709 le32_to_cpu(es->s_blocks_count_lo);
2710 }
2711
ext4_r_blocks_count(struct ext4_super_block * es)2712 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2713 {
2714 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2715 le32_to_cpu(es->s_r_blocks_count_lo);
2716 }
2717
ext4_free_blocks_count(struct ext4_super_block * es)2718 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2719 {
2720 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2721 le32_to_cpu(es->s_free_blocks_count_lo);
2722 }
2723
ext4_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)2724 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2725 ext4_fsblk_t blk)
2726 {
2727 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2728 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2729 }
2730
ext4_free_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)2731 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2732 ext4_fsblk_t blk)
2733 {
2734 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2735 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2736 }
2737
ext4_r_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)2738 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2739 ext4_fsblk_t blk)
2740 {
2741 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2742 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2743 }
2744
ext4_isize(struct ext4_inode * raw_inode)2745 static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
2746 {
2747 if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2748 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2749 le32_to_cpu(raw_inode->i_size_lo);
2750 else
2751 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2752 }
2753
ext4_isize_set(struct ext4_inode * raw_inode,loff_t i_size)2754 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2755 {
2756 raw_inode->i_size_lo = cpu_to_le32(i_size);
2757 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2758 }
2759
2760 static inline
ext4_get_group_info(struct super_block * sb,ext4_group_t group)2761 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2762 ext4_group_t group)
2763 {
2764 struct ext4_group_info ***grp_info;
2765 long indexv, indexh;
2766 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
2767 grp_info = EXT4_SB(sb)->s_group_info;
2768 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2769 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2770 return grp_info[indexv][indexh];
2771 }
2772
2773 /*
2774 * Reading s_groups_count requires using smp_rmb() afterwards. See
2775 * the locking protocol documented in the comments of ext4_group_add()
2776 * in resize.c
2777 */
ext4_get_groups_count(struct super_block * sb)2778 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2779 {
2780 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
2781
2782 smp_rmb();
2783 return ngroups;
2784 }
2785
ext4_flex_group(struct ext4_sb_info * sbi,ext4_group_t block_group)2786 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2787 ext4_group_t block_group)
2788 {
2789 return block_group >> sbi->s_log_groups_per_flex;
2790 }
2791
ext4_flex_bg_size(struct ext4_sb_info * sbi)2792 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2793 {
2794 return 1 << sbi->s_log_groups_per_flex;
2795 }
2796
2797 #define ext4_std_error(sb, errno) \
2798 do { \
2799 if ((errno)) \
2800 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
2801 } while (0)
2802
2803 #ifdef CONFIG_SMP
2804 /* Each CPU can accumulate percpu_counter_batch clusters in their local
2805 * counters. So we need to make sure we have free clusters more
2806 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
2807 */
2808 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2809 #else
2810 #define EXT4_FREECLUSTERS_WATERMARK 0
2811 #endif
2812
2813 /* Update i_disksize. Requires i_mutex to avoid races with truncate */
ext4_update_i_disksize(struct inode * inode,loff_t newsize)2814 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2815 {
2816 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
2817 !inode_is_locked(inode));
2818 down_write(&EXT4_I(inode)->i_data_sem);
2819 if (newsize > EXT4_I(inode)->i_disksize)
2820 EXT4_I(inode)->i_disksize = newsize;
2821 up_write(&EXT4_I(inode)->i_data_sem);
2822 }
2823
2824 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
ext4_update_inode_size(struct inode * inode,loff_t newsize)2825 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
2826 {
2827 int changed = 0;
2828
2829 if (newsize > inode->i_size) {
2830 i_size_write(inode, newsize);
2831 changed = 1;
2832 }
2833 if (newsize > EXT4_I(inode)->i_disksize) {
2834 ext4_update_i_disksize(inode, newsize);
2835 changed |= 2;
2836 }
2837 return changed;
2838 }
2839
2840 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
2841 loff_t len);
2842
2843 struct ext4_group_info {
2844 unsigned long bb_state;
2845 struct rb_root bb_free_root;
2846 ext4_grpblk_t bb_first_free; /* first free block */
2847 ext4_grpblk_t bb_free; /* total free blocks */
2848 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
2849 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
2850 struct list_head bb_prealloc_list;
2851 #ifdef DOUBLE_CHECK
2852 void *bb_bitmap;
2853 #endif
2854 struct rw_semaphore alloc_sem;
2855 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
2856 * regions, index is order.
2857 * bb_counters[3] = 5 means
2858 * 5 free 8-block regions. */
2859 };
2860
2861 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
2862 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
2863 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
2864 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
2865
2866 #define EXT4_MB_GRP_NEED_INIT(grp) \
2867 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2868 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
2869 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2870 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
2871 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2872
2873 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
2874 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2875 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
2876 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2877 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
2878 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2879
2880 #define EXT4_MAX_CONTENTION 8
2881 #define EXT4_CONTENTION_THRESHOLD 2
2882
ext4_group_lock_ptr(struct super_block * sb,ext4_group_t group)2883 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2884 ext4_group_t group)
2885 {
2886 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2887 }
2888
2889 /*
2890 * Returns true if the filesystem is busy enough that attempts to
2891 * access the block group locks has run into contention.
2892 */
ext4_fs_is_busy(struct ext4_sb_info * sbi)2893 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2894 {
2895 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2896 }
2897
ext4_lock_group(struct super_block * sb,ext4_group_t group)2898 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2899 {
2900 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2901 if (spin_trylock(lock))
2902 /*
2903 * We're able to grab the lock right away, so drop the
2904 * lock contention counter.
2905 */
2906 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2907 else {
2908 /*
2909 * The lock is busy, so bump the contention counter,
2910 * and then wait on the spin lock.
2911 */
2912 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2913 EXT4_MAX_CONTENTION);
2914 spin_lock(lock);
2915 }
2916 }
2917
ext4_unlock_group(struct super_block * sb,ext4_group_t group)2918 static inline void ext4_unlock_group(struct super_block *sb,
2919 ext4_group_t group)
2920 {
2921 spin_unlock(ext4_group_lock_ptr(sb, group));
2922 }
2923
2924 /*
2925 * Block validity checking
2926 */
2927 #define ext4_check_indirect_blockref(inode, bh) \
2928 ext4_check_blockref(__func__, __LINE__, inode, \
2929 (__le32 *)(bh)->b_data, \
2930 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2931
2932 #define ext4_ind_check_inode(inode) \
2933 ext4_check_blockref(__func__, __LINE__, inode, \
2934 EXT4_I(inode)->i_data, \
2935 EXT4_NDIR_BLOCKS)
2936
2937 /*
2938 * Inodes and files operations
2939 */
2940
2941 /* dir.c */
2942 extern const struct file_operations ext4_dir_operations;
2943
2944 /* file.c */
2945 extern const struct inode_operations ext4_file_inode_operations;
2946 extern const struct file_operations ext4_file_operations;
2947 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
2948
2949 /* inline.c */
2950 extern int ext4_get_max_inline_size(struct inode *inode);
2951 extern int ext4_find_inline_data_nolock(struct inode *inode);
2952 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
2953 unsigned int len);
2954 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
2955
2956 extern int ext4_readpage_inline(struct inode *inode, struct page *page);
2957 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
2958 struct inode *inode,
2959 loff_t pos, unsigned len,
2960 unsigned flags,
2961 struct page **pagep);
2962 extern int ext4_write_inline_data_end(struct inode *inode,
2963 loff_t pos, unsigned len,
2964 unsigned copied,
2965 struct page *page);
2966 extern struct buffer_head *
2967 ext4_journalled_write_inline_data(struct inode *inode,
2968 unsigned len,
2969 struct page *page);
2970 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
2971 struct inode *inode,
2972 loff_t pos, unsigned len,
2973 unsigned flags,
2974 struct page **pagep,
2975 void **fsdata);
2976 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
2977 unsigned len, unsigned copied,
2978 struct page *page);
2979 extern int ext4_try_add_inline_entry(handle_t *handle,
2980 struct ext4_filename *fname,
2981 struct inode *dir, struct inode *inode);
2982 extern int ext4_try_create_inline_dir(handle_t *handle,
2983 struct inode *parent,
2984 struct inode *inode);
2985 extern int ext4_read_inline_dir(struct file *filp,
2986 struct dir_context *ctx,
2987 int *has_inline_data);
2988 extern int htree_inlinedir_to_tree(struct file *dir_file,
2989 struct inode *dir, ext4_lblk_t block,
2990 struct dx_hash_info *hinfo,
2991 __u32 start_hash, __u32 start_minor_hash,
2992 int *has_inline_data);
2993 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
2994 struct ext4_filename *fname,
2995 const struct qstr *d_name,
2996 struct ext4_dir_entry_2 **res_dir,
2997 int *has_inline_data);
2998 extern int ext4_delete_inline_entry(handle_t *handle,
2999 struct inode *dir,
3000 struct ext4_dir_entry_2 *de_del,
3001 struct buffer_head *bh,
3002 int *has_inline_data);
3003 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3004 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3005 struct ext4_dir_entry_2 **parent_de,
3006 int *retval);
3007 extern int ext4_inline_data_fiemap(struct inode *inode,
3008 struct fiemap_extent_info *fieinfo,
3009 int *has_inline, __u64 start, __u64 len);
3010 extern int ext4_try_to_evict_inline_data(handle_t *handle,
3011 struct inode *inode,
3012 int needed);
3013 extern void ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3014
3015 extern int ext4_convert_inline_data(struct inode *inode);
3016
ext4_has_inline_data(struct inode * inode)3017 static inline int ext4_has_inline_data(struct inode *inode)
3018 {
3019 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3020 EXT4_I(inode)->i_inline_off;
3021 }
3022
3023 /* namei.c */
3024 extern const struct inode_operations ext4_dir_inode_operations;
3025 extern const struct inode_operations ext4_special_inode_operations;
3026 extern struct dentry *ext4_get_parent(struct dentry *child);
3027 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3028 struct ext4_dir_entry_2 *de,
3029 int blocksize, int csum_size,
3030 unsigned int parent_ino, int dotdot_real_len);
3031 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t,
3032 unsigned int blocksize);
3033 extern int ext4_handle_dirty_dirent_node(handle_t *handle,
3034 struct inode *inode,
3035 struct buffer_head *bh);
3036 #define S_SHIFT 12
3037 static const unsigned char ext4_type_by_mode[S_IFMT >> S_SHIFT] = {
3038 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
3039 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
3040 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
3041 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
3042 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
3043 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
3044 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
3045 };
3046
ext4_set_de_type(struct super_block * sb,struct ext4_dir_entry_2 * de,umode_t mode)3047 static inline void ext4_set_de_type(struct super_block *sb,
3048 struct ext4_dir_entry_2 *de,
3049 umode_t mode) {
3050 if (ext4_has_feature_filetype(sb))
3051 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3052 }
3053
3054 /* readpages.c */
3055 extern int ext4_mpage_readpages(struct address_space *mapping,
3056 struct list_head *pages, struct page *page,
3057 unsigned nr_pages);
3058
3059 /* symlink.c */
3060 extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3061 extern const struct inode_operations ext4_symlink_inode_operations;
3062 extern const struct inode_operations ext4_fast_symlink_inode_operations;
3063
3064 /* sysfs.c */
3065 extern int ext4_register_sysfs(struct super_block *sb);
3066 extern void ext4_unregister_sysfs(struct super_block *sb);
3067 extern int __init ext4_init_sysfs(void);
3068 extern void ext4_exit_sysfs(void);
3069
3070 /* block_validity */
3071 extern void ext4_release_system_zone(struct super_block *sb);
3072 extern int ext4_setup_system_zone(struct super_block *sb);
3073 extern int __init ext4_init_system_zone(void);
3074 extern void ext4_exit_system_zone(void);
3075 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
3076 ext4_fsblk_t start_blk,
3077 unsigned int count);
3078 extern int ext4_check_blockref(const char *, unsigned int,
3079 struct inode *, __le32 *, unsigned int);
3080
3081 /* extents.c */
3082 struct ext4_ext_path;
3083 struct ext4_extent;
3084
3085 /*
3086 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3087 * __le32.
3088 */
3089 #define EXT_MAX_BLOCKS 0xffffffff
3090
3091 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
3092 extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
3093 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3094 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3095 struct ext4_map_blocks *map, int flags);
3096 extern void ext4_ext_truncate(handle_t *, struct inode *);
3097 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3098 ext4_lblk_t end);
3099 extern void ext4_ext_init(struct super_block *);
3100 extern void ext4_ext_release(struct super_block *);
3101 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3102 loff_t len);
3103 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3104 loff_t offset, ssize_t len);
3105 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3106 struct ext4_map_blocks *map, int flags);
3107 extern int ext4_ext_calc_metadata_amount(struct inode *inode,
3108 ext4_lblk_t lblocks);
3109 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3110 int num,
3111 struct ext4_ext_path *path);
3112 extern int ext4_can_extents_be_merged(struct inode *inode,
3113 struct ext4_extent *ex1,
3114 struct ext4_extent *ex2);
3115 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3116 struct ext4_ext_path **,
3117 struct ext4_extent *, int);
3118 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3119 struct ext4_ext_path **,
3120 int flags);
3121 extern void ext4_ext_drop_refs(struct ext4_ext_path *);
3122 extern int ext4_ext_check_inode(struct inode *inode);
3123 extern int ext4_find_delalloc_range(struct inode *inode,
3124 ext4_lblk_t lblk_start,
3125 ext4_lblk_t lblk_end);
3126 extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk);
3127 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3128 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3129 __u64 start, __u64 len);
3130 extern int ext4_ext_precache(struct inode *inode);
3131 extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len);
3132 extern int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len);
3133 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3134 struct inode *inode2, ext4_lblk_t lblk1,
3135 ext4_lblk_t lblk2, ext4_lblk_t count,
3136 int mark_unwritten,int *err);
3137
3138 /* move_extent.c */
3139 extern void ext4_double_down_write_data_sem(struct inode *first,
3140 struct inode *second);
3141 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3142 struct inode *donor_inode);
3143 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3144 __u64 start_orig, __u64 start_donor,
3145 __u64 len, __u64 *moved_len);
3146
3147 /* page-io.c */
3148 extern int __init ext4_init_pageio(void);
3149 extern void ext4_exit_pageio(void);
3150 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3151 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3152 extern int ext4_put_io_end(ext4_io_end_t *io_end);
3153 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3154 extern void ext4_io_submit_init(struct ext4_io_submit *io,
3155 struct writeback_control *wbc);
3156 extern void ext4_end_io_rsv_work(struct work_struct *work);
3157 extern void ext4_io_submit(struct ext4_io_submit *io);
3158 extern int ext4_bio_write_page(struct ext4_io_submit *io,
3159 struct page *page,
3160 int len,
3161 struct writeback_control *wbc,
3162 bool keep_towrite);
3163
3164 /* mmp.c */
3165 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3166
3167 /*
3168 * Add new method to test whether block and inode bitmaps are properly
3169 * initialized. With uninit_bg reading the block from disk is not enough
3170 * to mark the bitmap uptodate. We need to also zero-out the bitmap
3171 */
3172 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3173
bitmap_uptodate(struct buffer_head * bh)3174 static inline int bitmap_uptodate(struct buffer_head *bh)
3175 {
3176 return (buffer_uptodate(bh) &&
3177 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3178 }
set_bitmap_uptodate(struct buffer_head * bh)3179 static inline void set_bitmap_uptodate(struct buffer_head *bh)
3180 {
3181 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3182 }
3183
3184 /*
3185 * Disable DIO read nolock optimization, so new dioreaders will be forced
3186 * to grab i_mutex
3187 */
ext4_inode_block_unlocked_dio(struct inode * inode)3188 static inline void ext4_inode_block_unlocked_dio(struct inode *inode)
3189 {
3190 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
3191 smp_mb();
3192 }
ext4_inode_resume_unlocked_dio(struct inode * inode)3193 static inline void ext4_inode_resume_unlocked_dio(struct inode *inode)
3194 {
3195 smp_mb();
3196 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
3197 }
3198
3199 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
3200
3201 /* For ioend & aio unwritten conversion wait queues */
3202 #define EXT4_WQ_HASH_SZ 37
3203 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
3204 EXT4_WQ_HASH_SZ])
3205 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3206
3207 #define EXT4_RESIZING 0
3208 extern int ext4_resize_begin(struct super_block *sb);
3209 extern void ext4_resize_end(struct super_block *sb);
3210
ext4_set_io_unwritten_flag(struct inode * inode,struct ext4_io_end * io_end)3211 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
3212 struct ext4_io_end *io_end)
3213 {
3214 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
3215 io_end->flag |= EXT4_IO_END_UNWRITTEN;
3216 atomic_inc(&EXT4_I(inode)->i_unwritten);
3217 }
3218 }
3219
ext4_clear_io_unwritten_flag(ext4_io_end_t * io_end)3220 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3221 {
3222 struct inode *inode = io_end->inode;
3223
3224 if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
3225 io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3226 /* Wake up anyone waiting on unwritten extent conversion */
3227 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
3228 wake_up_all(ext4_ioend_wq(inode));
3229 }
3230 }
3231
ext4_aligned_io(struct inode * inode,loff_t off,loff_t len)3232 static inline bool ext4_aligned_io(struct inode *inode, loff_t off, loff_t len)
3233 {
3234 int blksize = 1 << inode->i_blkbits;
3235
3236 return IS_ALIGNED(off, blksize) && IS_ALIGNED(len, blksize);
3237 }
3238
3239 #endif /* __KERNEL__ */
3240
3241 #define EFSBADCRC EBADMSG /* Bad CRC detected */
3242 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
3243
3244 #endif /* _EXT4_H */
3245