]> git.proxmox.com Git - mirror_ubuntu-artful-kernel.git/blob - fs/ext4/ext4.h
Merge tag 'iio-for-4.13b' of git://git.kernel.org/pub/scm/linux/kernel/git/jic23...
[mirror_ubuntu-artful-kernel.git] / fs / ext4 / ext4.h
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/sched/signal.h>
32 #include <linux/blockgroup_lock.h>
33 #include <linux/percpu_counter.h>
34 #include <linux/ratelimit.h>
35 #include <crypto/hash.h>
36 #ifdef CONFIG_EXT4_FS_ENCRYPTION
37 #include <linux/fscrypt_supp.h>
38 #else
39 #include <linux/fscrypt_notsupp.h>
40 #endif
41 #include <linux/falloc.h>
42 #include <linux/percpu-rwsem.h>
43 #ifdef __KERNEL__
44 #include <linux/compat.h>
45 #endif
46
47 /*
48 * The fourth extended filesystem constants/structures
49 */
50
51 /*
52 * with AGGRESSIVE_CHECK allocator runs consistency checks over
53 * structures. these checks slow things down a lot
54 */
55 #define AGGRESSIVE_CHECK__
56
57 /*
58 * with DOUBLE_CHECK defined mballoc creates persistent in-core
59 * bitmaps, maintains and uses them to check for double allocations
60 */
61 #define DOUBLE_CHECK__
62
63 /*
64 * Define EXT4FS_DEBUG to produce debug messages
65 */
66 #undef EXT4FS_DEBUG
67
68 /*
69 * Debug code
70 */
71 #ifdef EXT4FS_DEBUG
72 #define ext4_debug(f, a...) \
73 do { \
74 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
75 __FILE__, __LINE__, __func__); \
76 printk(KERN_DEBUG f, ## a); \
77 } while (0)
78 #else
79 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
80 #endif
81
82 /*
83 * Turn on EXT_DEBUG to get lots of info about extents operations.
84 */
85 #define EXT_DEBUG__
86 #ifdef EXT_DEBUG
87 #define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__)
88 #else
89 #define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
90 #endif
91
92 /* data type for block offset of block group */
93 typedef int ext4_grpblk_t;
94
95 /* data type for filesystem-wide blocks number */
96 typedef unsigned long long ext4_fsblk_t;
97
98 /* data type for file logical block number */
99 typedef __u32 ext4_lblk_t;
100
101 /* data type for block group number */
102 typedef unsigned int ext4_group_t;
103
104 enum SHIFT_DIRECTION {
105 SHIFT_LEFT = 0,
106 SHIFT_RIGHT,
107 };
108
109 /*
110 * Flags used in mballoc's allocation_context flags field.
111 *
112 * Also used to show what's going on for debugging purposes when the
113 * flag field is exported via the traceport interface
114 */
115
116 /* prefer goal again. length */
117 #define EXT4_MB_HINT_MERGE 0x0001
118 /* blocks already reserved */
119 #define EXT4_MB_HINT_RESERVED 0x0002
120 /* metadata is being allocated */
121 #define EXT4_MB_HINT_METADATA 0x0004
122 /* first blocks in the file */
123 #define EXT4_MB_HINT_FIRST 0x0008
124 /* search for the best chunk */
125 #define EXT4_MB_HINT_BEST 0x0010
126 /* data is being allocated */
127 #define EXT4_MB_HINT_DATA 0x0020
128 /* don't preallocate (for tails) */
129 #define EXT4_MB_HINT_NOPREALLOC 0x0040
130 /* allocate for locality group */
131 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080
132 /* allocate goal blocks or none */
133 #define EXT4_MB_HINT_GOAL_ONLY 0x0100
134 /* goal is meaningful */
135 #define EXT4_MB_HINT_TRY_GOAL 0x0200
136 /* blocks already pre-reserved by delayed allocation */
137 #define EXT4_MB_DELALLOC_RESERVED 0x0400
138 /* We are doing stream allocation */
139 #define EXT4_MB_STREAM_ALLOC 0x0800
140 /* Use reserved root blocks if needed */
141 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000
142 /* Use blocks from reserved pool */
143 #define EXT4_MB_USE_RESERVED 0x2000
144
145 struct ext4_allocation_request {
146 /* target inode for block we're allocating */
147 struct inode *inode;
148 /* how many blocks we want to allocate */
149 unsigned int len;
150 /* logical block in target inode */
151 ext4_lblk_t logical;
152 /* the closest logical allocated block to the left */
153 ext4_lblk_t lleft;
154 /* the closest logical allocated block to the right */
155 ext4_lblk_t lright;
156 /* phys. target (a hint) */
157 ext4_fsblk_t goal;
158 /* phys. block for the closest logical allocated block to the left */
159 ext4_fsblk_t pleft;
160 /* phys. block for the closest logical allocated block to the right */
161 ext4_fsblk_t pright;
162 /* flags. see above EXT4_MB_HINT_* */
163 unsigned int flags;
164 };
165
166 /*
167 * Logical to physical block mapping, used by ext4_map_blocks()
168 *
169 * This structure is used to pass requests into ext4_map_blocks() as
170 * well as to store the information returned by ext4_map_blocks(). It
171 * takes less room on the stack than a struct buffer_head.
172 */
173 #define EXT4_MAP_NEW (1 << BH_New)
174 #define EXT4_MAP_MAPPED (1 << BH_Mapped)
175 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten)
176 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary)
177 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
178 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
179
180 struct ext4_map_blocks {
181 ext4_fsblk_t m_pblk;
182 ext4_lblk_t m_lblk;
183 unsigned int m_len;
184 unsigned int m_flags;
185 };
186
187 /*
188 * Flags for ext4_io_end->flags
189 */
190 #define EXT4_IO_END_UNWRITTEN 0x0001
191
192 /*
193 * For converting unwritten extents on a work queue. 'handle' is used for
194 * buffered writeback.
195 */
196 typedef struct ext4_io_end {
197 struct list_head list; /* per-file finished IO list */
198 handle_t *handle; /* handle reserved for extent
199 * conversion */
200 struct inode *inode; /* file being written to */
201 struct bio *bio; /* Linked list of completed
202 * bios covering the extent */
203 unsigned int flag; /* unwritten or not */
204 atomic_t count; /* reference counter */
205 loff_t offset; /* offset in the file */
206 ssize_t size; /* size of the extent */
207 } ext4_io_end_t;
208
209 struct ext4_io_submit {
210 struct writeback_control *io_wbc;
211 struct bio *io_bio;
212 ext4_io_end_t *io_end;
213 sector_t io_next_block;
214 };
215
216 /*
217 * Special inodes numbers
218 */
219 #define EXT4_BAD_INO 1 /* Bad blocks inode */
220 #define EXT4_ROOT_INO 2 /* Root inode */
221 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */
222 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
223 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
224 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
225 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
226 #define EXT4_JOURNAL_INO 8 /* Journal inode */
227
228 /* First non-reserved inode for old ext4 filesystems */
229 #define EXT4_GOOD_OLD_FIRST_INO 11
230
231 /*
232 * Maximal count of links to a file
233 */
234 #define EXT4_LINK_MAX 65000
235
236 /*
237 * Macro-instructions used to manage several block sizes
238 */
239 #define EXT4_MIN_BLOCK_SIZE 1024
240 #define EXT4_MAX_BLOCK_SIZE 65536
241 #define EXT4_MIN_BLOCK_LOG_SIZE 10
242 #define EXT4_MAX_BLOCK_LOG_SIZE 16
243 #define EXT4_MAX_CLUSTER_LOG_SIZE 30
244 #ifdef __KERNEL__
245 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
246 #else
247 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
248 #endif
249 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
250 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
251 EXT4_SB(s)->s_cluster_bits)
252 #ifdef __KERNEL__
253 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
254 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
255 #else
256 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
257 #endif
258 #ifdef __KERNEL__
259 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
260 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
261 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
262 #else
263 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
264 EXT4_GOOD_OLD_INODE_SIZE : \
265 (s)->s_inode_size)
266 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
267 EXT4_GOOD_OLD_FIRST_INO : \
268 (s)->s_first_ino)
269 #endif
270 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
271 #define EXT4_MAX_BLOCKS(size, offset, blkbits) \
272 ((EXT4_BLOCK_ALIGN(size + offset, blkbits) >> blkbits) - (offset >> \
273 blkbits))
274
275 /* Translate a block number to a cluster number */
276 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
277 /* Translate a cluster number to a block number */
278 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
279 /* Translate # of blks to # of clusters */
280 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
281 (sbi)->s_cluster_bits)
282 /* Mask out the low bits to get the starting block of the cluster */
283 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
284 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
285 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
286 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
287 /* Get the cluster offset */
288 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
289 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
290 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
291 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
292
293 /*
294 * Structure of a blocks group descriptor
295 */
296 struct ext4_group_desc
297 {
298 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
299 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
300 __le32 bg_inode_table_lo; /* Inodes table block */
301 __le16 bg_free_blocks_count_lo;/* Free blocks count */
302 __le16 bg_free_inodes_count_lo;/* Free inodes count */
303 __le16 bg_used_dirs_count_lo; /* Directories count */
304 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
305 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
306 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
307 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
308 __le16 bg_itable_unused_lo; /* Unused inodes count */
309 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
310 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
311 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
312 __le32 bg_inode_table_hi; /* Inodes table block MSB */
313 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
314 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
315 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
316 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
317 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
318 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
319 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
320 __u32 bg_reserved;
321 };
322
323 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
324 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
325 sizeof(__le16))
326 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
327 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
328 sizeof(__le16))
329
330 /*
331 * Structure of a flex block group info
332 */
333
334 struct flex_groups {
335 atomic64_t free_clusters;
336 atomic_t free_inodes;
337 atomic_t used_dirs;
338 };
339
340 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
341 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
342 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
343
344 /*
345 * Macro-instructions used to manage group descriptors
346 */
347 #define EXT4_MIN_DESC_SIZE 32
348 #define EXT4_MIN_DESC_SIZE_64BIT 64
349 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
350 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
351 #ifdef __KERNEL__
352 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
353 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
354 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
355 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
356 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
357 #else
358 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
359 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
360 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
361 #endif
362
363 /*
364 * Constants relative to the data blocks
365 */
366 #define EXT4_NDIR_BLOCKS 12
367 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
368 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
369 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
370 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
371
372 /*
373 * Inode flags
374 */
375 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
376 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
377 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
378 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
379 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
380 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
381 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
382 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
383 /* Reserved for compression usage... */
384 #define EXT4_DIRTY_FL 0x00000100
385 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
386 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
387 /* nb: was previously EXT2_ECOMPR_FL */
388 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */
389 /* End compression flags --- maybe not all used */
390 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
391 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
392 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
393 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
394 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
395 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
396 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
397 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
398 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
399 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
400 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
401 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
402 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
403
404 #define EXT4_FL_USER_VISIBLE 0x304BDFFF /* User visible flags */
405 #define EXT4_FL_USER_MODIFIABLE 0x204BC0FF /* User modifiable flags */
406
407 /* Flags we can manipulate with through EXT4_IOC_FSSETXATTR */
408 #define EXT4_FL_XFLAG_VISIBLE (EXT4_SYNC_FL | \
409 EXT4_IMMUTABLE_FL | \
410 EXT4_APPEND_FL | \
411 EXT4_NODUMP_FL | \
412 EXT4_NOATIME_FL | \
413 EXT4_PROJINHERIT_FL)
414
415 /* Flags that should be inherited by new inodes from their parent. */
416 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
417 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
418 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
419 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
420 EXT4_PROJINHERIT_FL)
421
422 /* Flags that are appropriate for regular files (all but dir-specific ones). */
423 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
424
425 /* Flags that are appropriate for non-directories/regular files. */
426 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
427
428 /* Mask out flags that are inappropriate for the given type of inode. */
429 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
430 {
431 if (S_ISDIR(mode))
432 return flags;
433 else if (S_ISREG(mode))
434 return flags & EXT4_REG_FLMASK;
435 else
436 return flags & EXT4_OTHER_FLMASK;
437 }
438
439 /*
440 * Inode flags used for atomic set/get
441 */
442 enum {
443 EXT4_INODE_SECRM = 0, /* Secure deletion */
444 EXT4_INODE_UNRM = 1, /* Undelete */
445 EXT4_INODE_COMPR = 2, /* Compress file */
446 EXT4_INODE_SYNC = 3, /* Synchronous updates */
447 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
448 EXT4_INODE_APPEND = 5, /* writes to file may only append */
449 EXT4_INODE_NODUMP = 6, /* do not dump file */
450 EXT4_INODE_NOATIME = 7, /* do not update atime */
451 /* Reserved for compression usage... */
452 EXT4_INODE_DIRTY = 8,
453 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
454 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
455 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */
456 /* End compression flags --- maybe not all used */
457 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
458 EXT4_INODE_IMAGIC = 13, /* AFS directory */
459 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
460 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
461 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
462 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
463 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
464 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
465 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
466 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
467 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
468 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */
469 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
470 };
471
472 /*
473 * Since it's pretty easy to mix up bit numbers and hex values, we use a
474 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
475 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
476 * any extra space in the compiled kernel image, otherwise, the build will fail.
477 * It's important that these values are the same, since we are using
478 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
479 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
480 * values found in ext2, ext3 and ext4 filesystems, and of course the values
481 * defined in e2fsprogs.
482 *
483 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
484 */
485 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
486 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
487
488 static inline void ext4_check_flag_values(void)
489 {
490 CHECK_FLAG_VALUE(SECRM);
491 CHECK_FLAG_VALUE(UNRM);
492 CHECK_FLAG_VALUE(COMPR);
493 CHECK_FLAG_VALUE(SYNC);
494 CHECK_FLAG_VALUE(IMMUTABLE);
495 CHECK_FLAG_VALUE(APPEND);
496 CHECK_FLAG_VALUE(NODUMP);
497 CHECK_FLAG_VALUE(NOATIME);
498 CHECK_FLAG_VALUE(DIRTY);
499 CHECK_FLAG_VALUE(COMPRBLK);
500 CHECK_FLAG_VALUE(NOCOMPR);
501 CHECK_FLAG_VALUE(ENCRYPT);
502 CHECK_FLAG_VALUE(INDEX);
503 CHECK_FLAG_VALUE(IMAGIC);
504 CHECK_FLAG_VALUE(JOURNAL_DATA);
505 CHECK_FLAG_VALUE(NOTAIL);
506 CHECK_FLAG_VALUE(DIRSYNC);
507 CHECK_FLAG_VALUE(TOPDIR);
508 CHECK_FLAG_VALUE(HUGE_FILE);
509 CHECK_FLAG_VALUE(EXTENTS);
510 CHECK_FLAG_VALUE(EA_INODE);
511 CHECK_FLAG_VALUE(EOFBLOCKS);
512 CHECK_FLAG_VALUE(INLINE_DATA);
513 CHECK_FLAG_VALUE(PROJINHERIT);
514 CHECK_FLAG_VALUE(RESERVED);
515 }
516
517 /* Used to pass group descriptor data when online resize is done */
518 struct ext4_new_group_input {
519 __u32 group; /* Group number for this data */
520 __u64 block_bitmap; /* Absolute block number of block bitmap */
521 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
522 __u64 inode_table; /* Absolute block number of inode table start */
523 __u32 blocks_count; /* Total number of blocks in this group */
524 __u16 reserved_blocks; /* Number of reserved blocks in this group */
525 __u16 unused;
526 };
527
528 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
529 struct compat_ext4_new_group_input {
530 u32 group;
531 compat_u64 block_bitmap;
532 compat_u64 inode_bitmap;
533 compat_u64 inode_table;
534 u32 blocks_count;
535 u16 reserved_blocks;
536 u16 unused;
537 };
538 #endif
539
540 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
541 struct ext4_new_group_data {
542 __u32 group;
543 __u64 block_bitmap;
544 __u64 inode_bitmap;
545 __u64 inode_table;
546 __u32 blocks_count;
547 __u16 reserved_blocks;
548 __u16 unused;
549 __u32 free_blocks_count;
550 };
551
552 /* Indexes used to index group tables in ext4_new_group_data */
553 enum {
554 BLOCK_BITMAP = 0, /* block bitmap */
555 INODE_BITMAP, /* inode bitmap */
556 INODE_TABLE, /* inode tables */
557 GROUP_TABLE_COUNT,
558 };
559
560 /*
561 * Flags used by ext4_map_blocks()
562 */
563 /* Allocate any needed blocks and/or convert an unwritten
564 extent to be an initialized ext4 */
565 #define EXT4_GET_BLOCKS_CREATE 0x0001
566 /* Request the creation of an unwritten extent */
567 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002
568 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\
569 EXT4_GET_BLOCKS_CREATE)
570 /* Caller is from the delayed allocation writeout path
571 * finally doing the actual allocation of delayed blocks */
572 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
573 /* caller is from the direct IO path, request to creation of an
574 unwritten extents if not allocated, split the unwritten
575 extent if blocks has been preallocated already*/
576 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
577 #define EXT4_GET_BLOCKS_CONVERT 0x0010
578 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
579 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
580 /* Convert extent to initialized after IO complete */
581 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
582 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
583 /* Eventual metadata allocation (due to growing extent tree)
584 * should not fail, so try to use reserved blocks for that.*/
585 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
586 /* Don't normalize allocation size (used for fallocate) */
587 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
588 /* Request will not result in inode size update (user for fallocate) */
589 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080
590 /* Convert written extents to unwritten */
591 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100
592 /* Write zeros to newly created written extents */
593 #define EXT4_GET_BLOCKS_ZERO 0x0200
594 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\
595 EXT4_GET_BLOCKS_ZERO)
596 /* Caller will submit data before dropping transaction handle. This
597 * allows jbd2 to avoid submitting data before commit. */
598 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400
599
600 /*
601 * The bit position of these flags must not overlap with any of the
602 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(),
603 * read_extent_tree_block(), ext4_split_extent_at(),
604 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
605 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
606 * caching the extents when reading from the extent tree while a
607 * truncate or punch hole operation is in progress.
608 */
609 #define EXT4_EX_NOCACHE 0x40000000
610 #define EXT4_EX_FORCE_CACHE 0x20000000
611
612 /*
613 * Flags used by ext4_free_blocks
614 */
615 #define EXT4_FREE_BLOCKS_METADATA 0x0001
616 #define EXT4_FREE_BLOCKS_FORGET 0x0002
617 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
618 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
619 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
620 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
621
622 /*
623 * ioctl commands
624 */
625 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
626 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
627 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
628 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
629 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
630 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
631 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
632 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
633 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
634 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
635 #define EXT4_IOC_MIGRATE _IO('f', 9)
636 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
637 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
638 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
639 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
640 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
641 #define EXT4_IOC_SWAP_BOOT _IO('f', 17)
642 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18)
643 #define EXT4_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY
644 #define EXT4_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT
645 #define EXT4_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY
646
647 #ifndef FS_IOC_FSGETXATTR
648 /* Until the uapi changes get merged for project quota... */
649
650 #define FS_IOC_FSGETXATTR _IOR('X', 31, struct fsxattr)
651 #define FS_IOC_FSSETXATTR _IOW('X', 32, struct fsxattr)
652
653 /*
654 * Structure for FS_IOC_FSGETXATTR and FS_IOC_FSSETXATTR.
655 */
656 struct fsxattr {
657 __u32 fsx_xflags; /* xflags field value (get/set) */
658 __u32 fsx_extsize; /* extsize field value (get/set)*/
659 __u32 fsx_nextents; /* nextents field value (get) */
660 __u32 fsx_projid; /* project identifier (get/set) */
661 unsigned char fsx_pad[12];
662 };
663
664 /*
665 * Flags for the fsx_xflags field
666 */
667 #define FS_XFLAG_REALTIME 0x00000001 /* data in realtime volume */
668 #define FS_XFLAG_PREALLOC 0x00000002 /* preallocated file extents */
669 #define FS_XFLAG_IMMUTABLE 0x00000008 /* file cannot be modified */
670 #define FS_XFLAG_APPEND 0x00000010 /* all writes append */
671 #define FS_XFLAG_SYNC 0x00000020 /* all writes synchronous */
672 #define FS_XFLAG_NOATIME 0x00000040 /* do not update access time */
673 #define FS_XFLAG_NODUMP 0x00000080 /* do not include in backups */
674 #define FS_XFLAG_RTINHERIT 0x00000100 /* create with rt bit set */
675 #define FS_XFLAG_PROJINHERIT 0x00000200 /* create with parents projid */
676 #define FS_XFLAG_NOSYMLINKS 0x00000400 /* disallow symlink creation */
677 #define FS_XFLAG_EXTSIZE 0x00000800 /* extent size allocator hint */
678 #define FS_XFLAG_EXTSZINHERIT 0x00001000 /* inherit inode extent size */
679 #define FS_XFLAG_NODEFRAG 0x00002000 /* do not defragment */
680 #define FS_XFLAG_FILESTREAM 0x00004000 /* use filestream allocator */
681 #define FS_XFLAG_HASATTR 0x80000000 /* no DIFLAG for this */
682 #endif /* !defined(FS_IOC_FSGETXATTR) */
683
684 #define EXT4_IOC_FSGETXATTR FS_IOC_FSGETXATTR
685 #define EXT4_IOC_FSSETXATTR FS_IOC_FSSETXATTR
686
687 #define EXT4_IOC_SHUTDOWN _IOR ('X', 125, __u32)
688
689 /*
690 * Flags for going down operation
691 */
692 #define EXT4_GOING_FLAGS_DEFAULT 0x0 /* going down */
693 #define EXT4_GOING_FLAGS_LOGFLUSH 0x1 /* flush log but not data */
694 #define EXT4_GOING_FLAGS_NOLOGFLUSH 0x2 /* don't flush log nor data */
695
696
697 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
698 /*
699 * ioctl commands in 32 bit emulation
700 */
701 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
702 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
703 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
704 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
705 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
706 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
707 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
708 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
709 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
710 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
711 #endif
712
713 /* Max physical block we can address w/o extents */
714 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
715
716 /*
717 * Structure of an inode on the disk
718 */
719 struct ext4_inode {
720 __le16 i_mode; /* File mode */
721 __le16 i_uid; /* Low 16 bits of Owner Uid */
722 __le32 i_size_lo; /* Size in bytes */
723 __le32 i_atime; /* Access time */
724 __le32 i_ctime; /* Inode Change time */
725 __le32 i_mtime; /* Modification time */
726 __le32 i_dtime; /* Deletion Time */
727 __le16 i_gid; /* Low 16 bits of Group Id */
728 __le16 i_links_count; /* Links count */
729 __le32 i_blocks_lo; /* Blocks count */
730 __le32 i_flags; /* File flags */
731 union {
732 struct {
733 __le32 l_i_version;
734 } linux1;
735 struct {
736 __u32 h_i_translator;
737 } hurd1;
738 struct {
739 __u32 m_i_reserved1;
740 } masix1;
741 } osd1; /* OS dependent 1 */
742 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
743 __le32 i_generation; /* File version (for NFS) */
744 __le32 i_file_acl_lo; /* File ACL */
745 __le32 i_size_high;
746 __le32 i_obso_faddr; /* Obsoleted fragment address */
747 union {
748 struct {
749 __le16 l_i_blocks_high; /* were l_i_reserved1 */
750 __le16 l_i_file_acl_high;
751 __le16 l_i_uid_high; /* these 2 fields */
752 __le16 l_i_gid_high; /* were reserved2[0] */
753 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
754 __le16 l_i_reserved;
755 } linux2;
756 struct {
757 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
758 __u16 h_i_mode_high;
759 __u16 h_i_uid_high;
760 __u16 h_i_gid_high;
761 __u32 h_i_author;
762 } hurd2;
763 struct {
764 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
765 __le16 m_i_file_acl_high;
766 __u32 m_i_reserved2[2];
767 } masix2;
768 } osd2; /* OS dependent 2 */
769 __le16 i_extra_isize;
770 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
771 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
772 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
773 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
774 __le32 i_crtime; /* File Creation time */
775 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
776 __le32 i_version_hi; /* high 32 bits for 64-bit version */
777 __le32 i_projid; /* Project ID */
778 };
779
780 struct move_extent {
781 __u32 reserved; /* should be zero */
782 __u32 donor_fd; /* donor file descriptor */
783 __u64 orig_start; /* logical start offset in block for orig */
784 __u64 donor_start; /* logical start offset in block for donor */
785 __u64 len; /* block length to be moved */
786 __u64 moved_len; /* moved block length */
787 };
788
789 #define EXT4_EPOCH_BITS 2
790 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
791 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
792
793 /*
794 * Extended fields will fit into an inode if the filesystem was formatted
795 * with large inodes (-I 256 or larger) and there are not currently any EAs
796 * consuming all of the available space. For new inodes we always reserve
797 * enough space for the kernel's known extended fields, but for inodes
798 * created with an old kernel this might not have been the case. None of
799 * the extended inode fields is critical for correct filesystem operation.
800 * This macro checks if a certain field fits in the inode. Note that
801 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
802 */
803 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
804 ((offsetof(typeof(*ext4_inode), field) + \
805 sizeof((ext4_inode)->field)) \
806 <= (EXT4_GOOD_OLD_INODE_SIZE + \
807 (einode)->i_extra_isize)) \
808
809 /*
810 * We use an encoding that preserves the times for extra epoch "00":
811 *
812 * extra msb of adjust for signed
813 * epoch 32-bit 32-bit tv_sec to
814 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
815 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
816 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
817 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
818 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
819 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
820 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
821 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
822 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
823 *
824 * Note that previous versions of the kernel on 64-bit systems would
825 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
826 * 1970. e2fsck will correct this, assuming that it is run on the
827 * affected filesystem before 2242.
828 */
829
830 static inline __le32 ext4_encode_extra_time(struct timespec *time)
831 {
832 u32 extra = sizeof(time->tv_sec) > 4 ?
833 ((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK : 0;
834 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
835 }
836
837 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
838 {
839 if (unlikely(sizeof(time->tv_sec) > 4 &&
840 (extra & cpu_to_le32(EXT4_EPOCH_MASK)))) {
841 #if LINUX_VERSION_CODE < KERNEL_VERSION(4,20,0)
842 /* Handle legacy encoding of pre-1970 dates with epoch
843 * bits 1,1. We assume that by kernel version 4.20,
844 * everyone will have run fsck over the affected
845 * filesystems to correct the problem. (This
846 * backwards compatibility may be removed before this
847 * time, at the discretion of the ext4 developers.)
848 */
849 u64 extra_bits = le32_to_cpu(extra) & EXT4_EPOCH_MASK;
850 if (extra_bits == 3 && ((time->tv_sec) & 0x80000000) != 0)
851 extra_bits = 0;
852 time->tv_sec += extra_bits << 32;
853 #else
854 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
855 #endif
856 }
857 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
858 }
859
860 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
861 do { \
862 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
863 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
864 (raw_inode)->xtime ## _extra = \
865 ext4_encode_extra_time(&(inode)->xtime); \
866 } while (0)
867
868 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
869 do { \
870 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
871 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
872 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
873 (raw_inode)->xtime ## _extra = \
874 ext4_encode_extra_time(&(einode)->xtime); \
875 } while (0)
876
877 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
878 do { \
879 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
880 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
881 ext4_decode_extra_time(&(inode)->xtime, \
882 raw_inode->xtime ## _extra); \
883 else \
884 (inode)->xtime.tv_nsec = 0; \
885 } while (0)
886
887 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
888 do { \
889 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
890 (einode)->xtime.tv_sec = \
891 (signed)le32_to_cpu((raw_inode)->xtime); \
892 else \
893 (einode)->xtime.tv_sec = 0; \
894 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
895 ext4_decode_extra_time(&(einode)->xtime, \
896 raw_inode->xtime ## _extra); \
897 else \
898 (einode)->xtime.tv_nsec = 0; \
899 } while (0)
900
901 #define i_disk_version osd1.linux1.l_i_version
902
903 #if defined(__KERNEL__) || defined(__linux__)
904 #define i_reserved1 osd1.linux1.l_i_reserved1
905 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
906 #define i_blocks_high osd2.linux2.l_i_blocks_high
907 #define i_uid_low i_uid
908 #define i_gid_low i_gid
909 #define i_uid_high osd2.linux2.l_i_uid_high
910 #define i_gid_high osd2.linux2.l_i_gid_high
911 #define i_checksum_lo osd2.linux2.l_i_checksum_lo
912
913 #elif defined(__GNU__)
914
915 #define i_translator osd1.hurd1.h_i_translator
916 #define i_uid_high osd2.hurd2.h_i_uid_high
917 #define i_gid_high osd2.hurd2.h_i_gid_high
918 #define i_author osd2.hurd2.h_i_author
919
920 #elif defined(__masix__)
921
922 #define i_reserved1 osd1.masix1.m_i_reserved1
923 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
924 #define i_reserved2 osd2.masix2.m_i_reserved2
925
926 #endif /* defined(__KERNEL__) || defined(__linux__) */
927
928 #include "extents_status.h"
929
930 /*
931 * Lock subclasses for i_data_sem in the ext4_inode_info structure.
932 *
933 * These are needed to avoid lockdep false positives when we need to
934 * allocate blocks to the quota inode during ext4_map_blocks(), while
935 * holding i_data_sem for a normal (non-quota) inode. Since we don't
936 * do quota tracking for the quota inode, this avoids deadlock (as
937 * well as infinite recursion, since it isn't turtles all the way
938 * down...)
939 *
940 * I_DATA_SEM_NORMAL - Used for most inodes
941 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode
942 * where the second inode has larger inode number
943 * than the first
944 * I_DATA_SEM_QUOTA - Used for quota inodes only
945 */
946 enum {
947 I_DATA_SEM_NORMAL = 0,
948 I_DATA_SEM_OTHER,
949 I_DATA_SEM_QUOTA,
950 };
951
952
953 /*
954 * fourth extended file system inode data in memory
955 */
956 struct ext4_inode_info {
957 __le32 i_data[15]; /* unconverted */
958 __u32 i_dtime;
959 ext4_fsblk_t i_file_acl;
960
961 /*
962 * i_block_group is the number of the block group which contains
963 * this file's inode. Constant across the lifetime of the inode,
964 * it is ued for making block allocation decisions - we try to
965 * place a file's data blocks near its inode block, and new inodes
966 * near to their parent directory's inode.
967 */
968 ext4_group_t i_block_group;
969 ext4_lblk_t i_dir_start_lookup;
970 #if (BITS_PER_LONG < 64)
971 unsigned long i_state_flags; /* Dynamic state flags */
972 #endif
973 unsigned long i_flags;
974
975 /*
976 * Extended attributes can be read independently of the main file
977 * data. Taking i_mutex even when reading would cause contention
978 * between readers of EAs and writers of regular file data, so
979 * instead we synchronize on xattr_sem when reading or changing
980 * EAs.
981 */
982 struct rw_semaphore xattr_sem;
983
984 struct list_head i_orphan; /* unlinked but open inodes */
985
986 /*
987 * i_disksize keeps track of what the inode size is ON DISK, not
988 * in memory. During truncate, i_size is set to the new size by
989 * the VFS prior to calling ext4_truncate(), but the filesystem won't
990 * set i_disksize to 0 until the truncate is actually under way.
991 *
992 * The intent is that i_disksize always represents the blocks which
993 * are used by this file. This allows recovery to restart truncate
994 * on orphans if we crash during truncate. We actually write i_disksize
995 * into the on-disk inode when writing inodes out, instead of i_size.
996 *
997 * The only time when i_disksize and i_size may be different is when
998 * a truncate is in progress. The only things which change i_disksize
999 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
1000 */
1001 loff_t i_disksize;
1002
1003 /*
1004 * i_data_sem is for serialising ext4_truncate() against
1005 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
1006 * data tree are chopped off during truncate. We can't do that in
1007 * ext4 because whenever we perform intermediate commits during
1008 * truncate, the inode and all the metadata blocks *must* be in a
1009 * consistent state which allows truncation of the orphans to restart
1010 * during recovery. Hence we must fix the get_block-vs-truncate race
1011 * by other means, so we have i_data_sem.
1012 */
1013 struct rw_semaphore i_data_sem;
1014 /*
1015 * i_mmap_sem is for serializing page faults with truncate / punch hole
1016 * operations. We have to make sure that new page cannot be faulted in
1017 * a section of the inode that is being punched. We cannot easily use
1018 * i_data_sem for this since we need protection for the whole punch
1019 * operation and i_data_sem ranks below transaction start so we have
1020 * to occasionally drop it.
1021 */
1022 struct rw_semaphore i_mmap_sem;
1023 struct inode vfs_inode;
1024 struct jbd2_inode *jinode;
1025
1026 spinlock_t i_raw_lock; /* protects updates to the raw inode */
1027
1028 /*
1029 * File creation time. Its function is same as that of
1030 * struct timespec i_{a,c,m}time in the generic inode.
1031 */
1032 struct timespec i_crtime;
1033
1034 /* mballoc */
1035 struct list_head i_prealloc_list;
1036 spinlock_t i_prealloc_lock;
1037
1038 /* extents status tree */
1039 struct ext4_es_tree i_es_tree;
1040 rwlock_t i_es_lock;
1041 struct list_head i_es_list;
1042 unsigned int i_es_all_nr; /* protected by i_es_lock */
1043 unsigned int i_es_shk_nr; /* protected by i_es_lock */
1044 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for
1045 extents to shrink. Protected by
1046 i_es_lock */
1047
1048 /* ialloc */
1049 ext4_group_t i_last_alloc_group;
1050
1051 /* allocation reservation info for delalloc */
1052 /* In case of bigalloc, these refer to clusters rather than blocks */
1053 unsigned int i_reserved_data_blocks;
1054 unsigned int i_reserved_meta_blocks;
1055 unsigned int i_allocated_meta_blocks;
1056 ext4_lblk_t i_da_metadata_calc_last_lblock;
1057 int i_da_metadata_calc_len;
1058
1059 /* on-disk additional length */
1060 __u16 i_extra_isize;
1061
1062 /* Indicate the inline data space. */
1063 u16 i_inline_off;
1064 u16 i_inline_size;
1065
1066 #ifdef CONFIG_QUOTA
1067 /* quota space reservation, managed internally by quota code */
1068 qsize_t i_reserved_quota;
1069 #endif
1070
1071 /* Lock protecting lists below */
1072 spinlock_t i_completed_io_lock;
1073 /*
1074 * Completed IOs that need unwritten extents handling and have
1075 * transaction reserved
1076 */
1077 struct list_head i_rsv_conversion_list;
1078 struct work_struct i_rsv_conversion_work;
1079 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1080
1081 spinlock_t i_block_reservation_lock;
1082
1083 /*
1084 * Transactions that contain inode's metadata needed to complete
1085 * fsync and fdatasync, respectively.
1086 */
1087 tid_t i_sync_tid;
1088 tid_t i_datasync_tid;
1089
1090 #ifdef CONFIG_QUOTA
1091 struct dquot *i_dquot[MAXQUOTAS];
1092 #endif
1093
1094 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1095 __u32 i_csum_seed;
1096
1097 kprojid_t i_projid;
1098 };
1099
1100 /*
1101 * File system states
1102 */
1103 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
1104 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
1105 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
1106
1107 /*
1108 * Misc. filesystem flags
1109 */
1110 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
1111 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
1112 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
1113
1114 /*
1115 * Mount flags set via mount options or defaults
1116 */
1117 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
1118 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
1119 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
1120 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
1121 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
1122 #define EXT4_MOUNT_ERRORS_MASK 0x00070
1123 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
1124 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
1125 #ifdef CONFIG_FS_DAX
1126 #define EXT4_MOUNT_DAX 0x00200 /* Direct Access */
1127 #else
1128 #define EXT4_MOUNT_DAX 0
1129 #endif
1130 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
1131 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
1132 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
1133 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
1134 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
1135 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
1136 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
1137 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
1138 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
1139 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
1140 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */
1141 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota,
1142 * enable enforcement for hidden
1143 * quota files */
1144 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable
1145 * enforcement for hidden quota
1146 * files */
1147 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota
1148 * enforcement */
1149 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
1150 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
1151 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1152 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
1153 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
1154 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
1155 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
1156 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
1157
1158 /*
1159 * Mount flags set either automatically (could not be set by mount option)
1160 * based on per file system feature or property or in special cases such as
1161 * distinguishing between explicit mount option definition and default.
1162 */
1163 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1164 specified delalloc */
1165 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1166 size of blocksize * 8
1167 blocks */
1168 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1169 file systems */
1170
1171 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly
1172 specified journal checksum */
1173
1174 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1175 ~EXT4_MOUNT_##opt
1176 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1177 EXT4_MOUNT_##opt
1178 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1179 EXT4_MOUNT_##opt)
1180
1181 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1182 ~EXT4_MOUNT2_##opt
1183 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1184 EXT4_MOUNT2_##opt
1185 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1186 EXT4_MOUNT2_##opt)
1187
1188 #define ext4_test_and_set_bit __test_and_set_bit_le
1189 #define ext4_set_bit __set_bit_le
1190 #define ext4_set_bit_atomic ext2_set_bit_atomic
1191 #define ext4_test_and_clear_bit __test_and_clear_bit_le
1192 #define ext4_clear_bit __clear_bit_le
1193 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
1194 #define ext4_test_bit test_bit_le
1195 #define ext4_find_next_zero_bit find_next_zero_bit_le
1196 #define ext4_find_next_bit find_next_bit_le
1197
1198 extern void ext4_set_bits(void *bm, int cur, int len);
1199
1200 /*
1201 * Maximal mount counts between two filesystem checks
1202 */
1203 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1204 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1205
1206 /*
1207 * Behaviour when detecting errors
1208 */
1209 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1210 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1211 #define EXT4_ERRORS_PANIC 3 /* Panic */
1212 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1213
1214 /* Metadata checksum algorithm codes */
1215 #define EXT4_CRC32C_CHKSUM 1
1216
1217 /*
1218 * Structure of the super block
1219 */
1220 struct ext4_super_block {
1221 /*00*/ __le32 s_inodes_count; /* Inodes count */
1222 __le32 s_blocks_count_lo; /* Blocks count */
1223 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1224 __le32 s_free_blocks_count_lo; /* Free blocks count */
1225 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1226 __le32 s_first_data_block; /* First Data Block */
1227 __le32 s_log_block_size; /* Block size */
1228 __le32 s_log_cluster_size; /* Allocation cluster size */
1229 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1230 __le32 s_clusters_per_group; /* # Clusters per group */
1231 __le32 s_inodes_per_group; /* # Inodes per group */
1232 __le32 s_mtime; /* Mount time */
1233 /*30*/ __le32 s_wtime; /* Write time */
1234 __le16 s_mnt_count; /* Mount count */
1235 __le16 s_max_mnt_count; /* Maximal mount count */
1236 __le16 s_magic; /* Magic signature */
1237 __le16 s_state; /* File system state */
1238 __le16 s_errors; /* Behaviour when detecting errors */
1239 __le16 s_minor_rev_level; /* minor revision level */
1240 /*40*/ __le32 s_lastcheck; /* time of last check */
1241 __le32 s_checkinterval; /* max. time between checks */
1242 __le32 s_creator_os; /* OS */
1243 __le32 s_rev_level; /* Revision level */
1244 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1245 __le16 s_def_resgid; /* Default gid for reserved blocks */
1246 /*
1247 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1248 *
1249 * Note: the difference between the compatible feature set and
1250 * the incompatible feature set is that if there is a bit set
1251 * in the incompatible feature set that the kernel doesn't
1252 * know about, it should refuse to mount the filesystem.
1253 *
1254 * e2fsck's requirements are more strict; if it doesn't know
1255 * about a feature in either the compatible or incompatible
1256 * feature set, it must abort and not try to meddle with
1257 * things it doesn't understand...
1258 */
1259 __le32 s_first_ino; /* First non-reserved inode */
1260 __le16 s_inode_size; /* size of inode structure */
1261 __le16 s_block_group_nr; /* block group # of this superblock */
1262 __le32 s_feature_compat; /* compatible feature set */
1263 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1264 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1265 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1266 /*78*/ char s_volume_name[16]; /* volume name */
1267 /*88*/ char s_last_mounted[64]; /* directory where last mounted */
1268 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1269 /*
1270 * Performance hints. Directory preallocation should only
1271 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1272 */
1273 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1274 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1275 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1276 /*
1277 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1278 */
1279 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1280 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1281 __le32 s_journal_dev; /* device number of journal file */
1282 __le32 s_last_orphan; /* start of list of inodes to delete */
1283 __le32 s_hash_seed[4]; /* HTREE hash seed */
1284 __u8 s_def_hash_version; /* Default hash version to use */
1285 __u8 s_jnl_backup_type;
1286 __le16 s_desc_size; /* size of group descriptor */
1287 /*100*/ __le32 s_default_mount_opts;
1288 __le32 s_first_meta_bg; /* First metablock block group */
1289 __le32 s_mkfs_time; /* When the filesystem was created */
1290 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1291 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1292 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1293 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1294 __le32 s_free_blocks_count_hi; /* Free blocks count */
1295 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1296 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1297 __le32 s_flags; /* Miscellaneous flags */
1298 __le16 s_raid_stride; /* RAID stride */
1299 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1300 __le64 s_mmp_block; /* Block for multi-mount protection */
1301 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1302 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1303 __u8 s_checksum_type; /* metadata checksum algorithm used */
1304 __u8 s_encryption_level; /* versioning level for encryption */
1305 __u8 s_reserved_pad; /* Padding to next 32bits */
1306 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1307 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1308 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1309 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1310 snapshot's future use */
1311 __le32 s_snapshot_list; /* inode number of the head of the
1312 on-disk snapshot list */
1313 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1314 __le32 s_error_count; /* number of fs errors */
1315 __le32 s_first_error_time; /* first time an error happened */
1316 __le32 s_first_error_ino; /* inode involved in first error */
1317 __le64 s_first_error_block; /* block involved of first error */
1318 __u8 s_first_error_func[32]; /* function where the error happened */
1319 __le32 s_first_error_line; /* line number where error happened */
1320 __le32 s_last_error_time; /* most recent time of an error */
1321 __le32 s_last_error_ino; /* inode involved in last error */
1322 __le32 s_last_error_line; /* line number where error happened */
1323 __le64 s_last_error_block; /* block involved of last error */
1324 __u8 s_last_error_func[32]; /* function where the error happened */
1325 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1326 __u8 s_mount_opts[64];
1327 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1328 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1329 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1330 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
1331 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */
1332 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */
1333 __le32 s_lpf_ino; /* Location of the lost+found inode */
1334 __le32 s_prj_quota_inum; /* inode for tracking project quota */
1335 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */
1336 __le32 s_reserved[98]; /* Padding to the end of the block */
1337 __le32 s_checksum; /* crc32c(superblock) */
1338 };
1339
1340 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1341
1342 #ifdef __KERNEL__
1343
1344 /*
1345 * run-time mount flags
1346 */
1347 #define EXT4_MF_MNTDIR_SAMPLED 0x0001
1348 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1349 #define EXT4_MF_TEST_DUMMY_ENCRYPTION 0x0004
1350
1351 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1352 #define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \
1353 EXT4_MF_TEST_DUMMY_ENCRYPTION))
1354 #else
1355 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1356 #endif
1357
1358 /* Number of quota types we support */
1359 #define EXT4_MAXQUOTAS 3
1360
1361 /*
1362 * fourth extended-fs super-block data in memory
1363 */
1364 struct ext4_sb_info {
1365 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1366 unsigned long s_inodes_per_block;/* Number of inodes per block */
1367 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1368 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1369 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1370 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1371 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1372 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1373 ext4_group_t s_groups_count; /* Number of groups in the fs */
1374 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1375 unsigned long s_overhead; /* # of fs overhead clusters */
1376 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1377 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1378 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1379 struct buffer_head * s_sbh; /* Buffer containing the super block */
1380 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1381 struct buffer_head **s_group_desc;
1382 unsigned int s_mount_opt;
1383 unsigned int s_mount_opt2;
1384 unsigned int s_mount_flags;
1385 unsigned int s_def_mount_opt;
1386 ext4_fsblk_t s_sb_block;
1387 atomic64_t s_resv_clusters;
1388 kuid_t s_resuid;
1389 kgid_t s_resgid;
1390 unsigned short s_mount_state;
1391 unsigned short s_pad;
1392 int s_addr_per_block_bits;
1393 int s_desc_per_block_bits;
1394 int s_inode_size;
1395 int s_first_ino;
1396 unsigned int s_inode_readahead_blks;
1397 unsigned int s_inode_goal;
1398 spinlock_t s_next_gen_lock;
1399 u32 s_next_generation;
1400 u32 s_hash_seed[4];
1401 int s_def_hash_version;
1402 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1403 struct percpu_counter s_freeclusters_counter;
1404 struct percpu_counter s_freeinodes_counter;
1405 struct percpu_counter s_dirs_counter;
1406 struct percpu_counter s_dirtyclusters_counter;
1407 struct blockgroup_lock *s_blockgroup_lock;
1408 struct proc_dir_entry *s_proc;
1409 struct kobject s_kobj;
1410 struct completion s_kobj_unregister;
1411 struct super_block *s_sb;
1412
1413 /* Journaling */
1414 struct journal_s *s_journal;
1415 struct list_head s_orphan;
1416 struct mutex s_orphan_lock;
1417 unsigned long s_ext4_flags; /* Ext4 superblock flags */
1418 unsigned long s_commit_interval;
1419 u32 s_max_batch_time;
1420 u32 s_min_batch_time;
1421 struct block_device *journal_bdev;
1422 #ifdef CONFIG_QUOTA
1423 char *s_qf_names[EXT4_MAXQUOTAS]; /* Names of quota files with journalled quota */
1424 int s_jquota_fmt; /* Format of quota to use */
1425 #endif
1426 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1427 struct rb_root system_blks;
1428
1429 #ifdef EXTENTS_STATS
1430 /* ext4 extents stats */
1431 unsigned long s_ext_min;
1432 unsigned long s_ext_max;
1433 unsigned long s_depth_max;
1434 spinlock_t s_ext_stats_lock;
1435 unsigned long s_ext_blocks;
1436 unsigned long s_ext_extents;
1437 #endif
1438
1439 /* for buddy allocator */
1440 struct ext4_group_info ***s_group_info;
1441 struct inode *s_buddy_cache;
1442 spinlock_t s_md_lock;
1443 unsigned short *s_mb_offsets;
1444 unsigned int *s_mb_maxs;
1445 unsigned int s_group_info_size;
1446 unsigned int s_mb_free_pending;
1447
1448 /* tunables */
1449 unsigned long s_stripe;
1450 unsigned int s_mb_stream_request;
1451 unsigned int s_mb_max_to_scan;
1452 unsigned int s_mb_min_to_scan;
1453 unsigned int s_mb_stats;
1454 unsigned int s_mb_order2_reqs;
1455 unsigned int s_mb_group_prealloc;
1456 unsigned int s_max_dir_size_kb;
1457 /* where last allocation was done - for stream allocation */
1458 unsigned long s_mb_last_group;
1459 unsigned long s_mb_last_start;
1460
1461 /* stats for buddy allocator */
1462 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1463 atomic_t s_bal_success; /* we found long enough chunks */
1464 atomic_t s_bal_allocated; /* in blocks */
1465 atomic_t s_bal_ex_scanned; /* total extents scanned */
1466 atomic_t s_bal_goals; /* goal hits */
1467 atomic_t s_bal_breaks; /* too long searches */
1468 atomic_t s_bal_2orders; /* 2^order hits */
1469 spinlock_t s_bal_lock;
1470 unsigned long s_mb_buddies_generated;
1471 unsigned long long s_mb_generation_time;
1472 atomic_t s_mb_lost_chunks;
1473 atomic_t s_mb_preallocated;
1474 atomic_t s_mb_discarded;
1475 atomic_t s_lock_busy;
1476
1477 /* locality groups */
1478 struct ext4_locality_group __percpu *s_locality_groups;
1479
1480 /* for write statistics */
1481 unsigned long s_sectors_written_start;
1482 u64 s_kbytes_written;
1483
1484 /* the size of zero-out chunk */
1485 unsigned int s_extent_max_zeroout_kb;
1486
1487 unsigned int s_log_groups_per_flex;
1488 struct flex_groups *s_flex_groups;
1489 ext4_group_t s_flex_groups_allocated;
1490
1491 /* workqueue for reserved extent conversions (buffered io) */
1492 struct workqueue_struct *rsv_conversion_wq;
1493
1494 /* timer for periodic error stats printing */
1495 struct timer_list s_err_report;
1496
1497 /* Lazy inode table initialization info */
1498 struct ext4_li_request *s_li_request;
1499 /* Wait multiplier for lazy initialization thread */
1500 unsigned int s_li_wait_mult;
1501
1502 /* Kernel thread for multiple mount protection */
1503 struct task_struct *s_mmp_tsk;
1504
1505 /* record the last minlen when FITRIM is called. */
1506 atomic_t s_last_trim_minblks;
1507
1508 /* Reference to checksum algorithm driver via cryptoapi */
1509 struct crypto_shash *s_chksum_driver;
1510
1511 /* Precomputed FS UUID checksum for seeding other checksums */
1512 __u32 s_csum_seed;
1513
1514 /* Reclaim extents from extent status tree */
1515 struct shrinker s_es_shrinker;
1516 struct list_head s_es_list; /* List of inodes with reclaimable extents */
1517 long s_es_nr_inode;
1518 struct ext4_es_stats s_es_stats;
1519 struct mb_cache *s_mb_cache;
1520 spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1521
1522 /* Ratelimit ext4 messages. */
1523 struct ratelimit_state s_err_ratelimit_state;
1524 struct ratelimit_state s_warning_ratelimit_state;
1525 struct ratelimit_state s_msg_ratelimit_state;
1526
1527 /* Barrier between changing inodes' journal flags and writepages ops. */
1528 struct percpu_rw_semaphore s_journal_flag_rwsem;
1529 };
1530
1531 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1532 {
1533 return sb->s_fs_info;
1534 }
1535 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1536 {
1537 return container_of(inode, struct ext4_inode_info, vfs_inode);
1538 }
1539
1540 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1541 {
1542 return ino == EXT4_ROOT_INO ||
1543 ino == EXT4_USR_QUOTA_INO ||
1544 ino == EXT4_GRP_QUOTA_INO ||
1545 ino == EXT4_BOOT_LOADER_INO ||
1546 ino == EXT4_JOURNAL_INO ||
1547 ino == EXT4_RESIZE_INO ||
1548 (ino >= EXT4_FIRST_INO(sb) &&
1549 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1550 }
1551
1552 /*
1553 * Inode dynamic state flags
1554 */
1555 enum {
1556 EXT4_STATE_JDATA, /* journaled data exists */
1557 EXT4_STATE_NEW, /* inode is newly created */
1558 EXT4_STATE_XATTR, /* has in-inode xattrs */
1559 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1560 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1561 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1562 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
1563 EXT4_STATE_NEWENTRY, /* File just added to dir */
1564 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read
1565 nolocking */
1566 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1567 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
1568 };
1569
1570 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1571 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1572 { \
1573 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1574 } \
1575 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1576 { \
1577 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1578 } \
1579 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1580 { \
1581 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1582 }
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_flag(struct inode *inode, int bit);
1587 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1588 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1589 EXT4_INODE_BIT_FNS(flag, flags, 0)
1590
1591 /* Add these declarations here only so that these functions can be
1592 * found by name. Otherwise, they are very hard to locate. */
1593 static inline int ext4_test_inode_state(struct inode *inode, int bit);
1594 static inline void ext4_set_inode_state(struct inode *inode, int bit);
1595 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1596 #if (BITS_PER_LONG < 64)
1597 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1598
1599 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1600 {
1601 (ei)->i_state_flags = 0;
1602 }
1603 #else
1604 EXT4_INODE_BIT_FNS(state, flags, 32)
1605
1606 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1607 {
1608 /* We depend on the fact that callers will set i_flags */
1609 }
1610 #endif
1611 #else
1612 /* Assume that user mode programs are passing in an ext4fs superblock, not
1613 * a kernel struct super_block. This will allow us to call the feature-test
1614 * macros from user land. */
1615 #define EXT4_SB(sb) (sb)
1616 #endif
1617
1618 /*
1619 * Returns true if the inode is inode is encrypted
1620 */
1621 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1622
1623 /*
1624 * Codes for operating systems
1625 */
1626 #define EXT4_OS_LINUX 0
1627 #define EXT4_OS_HURD 1
1628 #define EXT4_OS_MASIX 2
1629 #define EXT4_OS_FREEBSD 3
1630 #define EXT4_OS_LITES 4
1631
1632 /*
1633 * Revision levels
1634 */
1635 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1636 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1637
1638 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1639 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1640
1641 #define EXT4_GOOD_OLD_INODE_SIZE 128
1642
1643 /*
1644 * Feature set definitions
1645 */
1646
1647 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1648 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1649 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1650 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1651 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1652 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1653 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
1654
1655 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1656 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1657 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1658 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1659 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1660 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1661 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1662 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1663 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1664 /*
1665 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1666 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1667 * all other data structures' checksums. However, the METADATA_CSUM and
1668 * GDT_CSUM bits are mutually exclusive.
1669 */
1670 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
1671 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000
1672 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000
1673
1674 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1675 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1676 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1677 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1678 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1679 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1680 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1681 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1682 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1683 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1684 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1685 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000
1686 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
1687 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
1688 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000
1689
1690 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
1691 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1692 { \
1693 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1694 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
1695 } \
1696 static inline void ext4_set_feature_##name(struct super_block *sb) \
1697 { \
1698 EXT4_SB(sb)->s_es->s_feature_compat |= \
1699 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1700 } \
1701 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1702 { \
1703 EXT4_SB(sb)->s_es->s_feature_compat &= \
1704 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1705 }
1706
1707 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
1708 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1709 { \
1710 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1711 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
1712 } \
1713 static inline void ext4_set_feature_##name(struct super_block *sb) \
1714 { \
1715 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
1716 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1717 } \
1718 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1719 { \
1720 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
1721 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1722 }
1723
1724 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
1725 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1726 { \
1727 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1728 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
1729 } \
1730 static inline void ext4_set_feature_##name(struct super_block *sb) \
1731 { \
1732 EXT4_SB(sb)->s_es->s_feature_incompat |= \
1733 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1734 } \
1735 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1736 { \
1737 EXT4_SB(sb)->s_es->s_feature_incompat &= \
1738 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1739 }
1740
1741 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC)
1742 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES)
1743 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL)
1744 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR)
1745 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE)
1746 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX)
1747 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2)
1748
1749 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER)
1750 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE)
1751 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR)
1752 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE)
1753 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM)
1754 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK)
1755 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE)
1756 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA)
1757 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC)
1758 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM)
1759 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY)
1760 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT)
1761
1762 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION)
1763 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE)
1764 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER)
1765 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV)
1766 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG)
1767 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS)
1768 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT)
1769 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP)
1770 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG)
1771 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE)
1772 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA)
1773 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED)
1774 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR)
1775 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA)
1776 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT)
1777
1778 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1779 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1780 EXT4_FEATURE_INCOMPAT_META_BG)
1781 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1782 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1783 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1784
1785 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1786 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1787 EXT4_FEATURE_INCOMPAT_RECOVER| \
1788 EXT4_FEATURE_INCOMPAT_META_BG)
1789 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1790 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1791 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1792
1793 #define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1794 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1795 EXT4_FEATURE_INCOMPAT_RECOVER| \
1796 EXT4_FEATURE_INCOMPAT_META_BG| \
1797 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1798 EXT4_FEATURE_INCOMPAT_64BIT| \
1799 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1800 EXT4_FEATURE_INCOMPAT_MMP | \
1801 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
1802 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
1803 EXT4_FEATURE_INCOMPAT_CSUM_SEED)
1804 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1805 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1806 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1807 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1808 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1809 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1810 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1811 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1812 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1813 EXT4_FEATURE_RO_COMPAT_QUOTA |\
1814 EXT4_FEATURE_RO_COMPAT_PROJECT)
1815
1816 #define EXTN_FEATURE_FUNCS(ver) \
1817 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
1818 { \
1819 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1820 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
1821 } \
1822 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
1823 { \
1824 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1825 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
1826 } \
1827 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
1828 { \
1829 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1830 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
1831 }
1832
1833 EXTN_FEATURE_FUNCS(2)
1834 EXTN_FEATURE_FUNCS(3)
1835 EXTN_FEATURE_FUNCS(4)
1836
1837 static inline bool ext4_has_compat_features(struct super_block *sb)
1838 {
1839 return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
1840 }
1841 static inline bool ext4_has_ro_compat_features(struct super_block *sb)
1842 {
1843 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
1844 }
1845 static inline bool ext4_has_incompat_features(struct super_block *sb)
1846 {
1847 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
1848 }
1849
1850 /*
1851 * Superblock flags
1852 */
1853 #define EXT4_FLAGS_RESIZING 0
1854 #define EXT4_FLAGS_SHUTDOWN 1
1855
1856 static inline int ext4_forced_shutdown(struct ext4_sb_info *sbi)
1857 {
1858 return test_bit(EXT4_FLAGS_SHUTDOWN, &sbi->s_ext4_flags);
1859 }
1860
1861
1862 /*
1863 * Default values for user and/or group using reserved blocks
1864 */
1865 #define EXT4_DEF_RESUID 0
1866 #define EXT4_DEF_RESGID 0
1867
1868 /*
1869 * Default project ID
1870 */
1871 #define EXT4_DEF_PROJID 0
1872
1873 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
1874
1875 /*
1876 * Default mount options
1877 */
1878 #define EXT4_DEFM_DEBUG 0x0001
1879 #define EXT4_DEFM_BSDGROUPS 0x0002
1880 #define EXT4_DEFM_XATTR_USER 0x0004
1881 #define EXT4_DEFM_ACL 0x0008
1882 #define EXT4_DEFM_UID16 0x0010
1883 #define EXT4_DEFM_JMODE 0x0060
1884 #define EXT4_DEFM_JMODE_DATA 0x0020
1885 #define EXT4_DEFM_JMODE_ORDERED 0x0040
1886 #define EXT4_DEFM_JMODE_WBACK 0x0060
1887 #define EXT4_DEFM_NOBARRIER 0x0100
1888 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1889 #define EXT4_DEFM_DISCARD 0x0400
1890 #define EXT4_DEFM_NODELALLOC 0x0800
1891
1892 /*
1893 * Default journal batch times
1894 */
1895 #define EXT4_DEF_MIN_BATCH_TIME 0
1896 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1897
1898 /*
1899 * Minimum number of groups in a flexgroup before we separate out
1900 * directories into the first block group of a flexgroup
1901 */
1902 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1903
1904 /*
1905 * Structure of a directory entry
1906 */
1907 #define EXT4_NAME_LEN 255
1908
1909 struct ext4_dir_entry {
1910 __le32 inode; /* Inode number */
1911 __le16 rec_len; /* Directory entry length */
1912 __le16 name_len; /* Name length */
1913 char name[EXT4_NAME_LEN]; /* File name */
1914 };
1915
1916 /*
1917 * The new version of the directory entry. Since EXT4 structures are
1918 * stored in intel byte order, and the name_len field could never be
1919 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1920 * file_type field.
1921 */
1922 struct ext4_dir_entry_2 {
1923 __le32 inode; /* Inode number */
1924 __le16 rec_len; /* Directory entry length */
1925 __u8 name_len; /* Name length */
1926 __u8 file_type;
1927 char name[EXT4_NAME_LEN]; /* File name */
1928 };
1929
1930 /*
1931 * This is a bogus directory entry at the end of each leaf block that
1932 * records checksums.
1933 */
1934 struct ext4_dir_entry_tail {
1935 __le32 det_reserved_zero1; /* Pretend to be unused */
1936 __le16 det_rec_len; /* 12 */
1937 __u8 det_reserved_zero2; /* Zero name length */
1938 __u8 det_reserved_ft; /* 0xDE, fake file type */
1939 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
1940 };
1941
1942 #define EXT4_DIRENT_TAIL(block, blocksize) \
1943 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1944 ((blocksize) - \
1945 sizeof(struct ext4_dir_entry_tail))))
1946
1947 /*
1948 * Ext4 directory file types. Only the low 3 bits are used. The
1949 * other bits are reserved for now.
1950 */
1951 #define EXT4_FT_UNKNOWN 0
1952 #define EXT4_FT_REG_FILE 1
1953 #define EXT4_FT_DIR 2
1954 #define EXT4_FT_CHRDEV 3
1955 #define EXT4_FT_BLKDEV 4
1956 #define EXT4_FT_FIFO 5
1957 #define EXT4_FT_SOCK 6
1958 #define EXT4_FT_SYMLINK 7
1959
1960 #define EXT4_FT_MAX 8
1961
1962 #define EXT4_FT_DIR_CSUM 0xDE
1963
1964 /*
1965 * EXT4_DIR_PAD defines the directory entries boundaries
1966 *
1967 * NOTE: It must be a multiple of 4
1968 */
1969 #define EXT4_DIR_PAD 4
1970 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1971 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1972 ~EXT4_DIR_ROUND)
1973 #define EXT4_MAX_REC_LEN ((1<<16)-1)
1974
1975 /*
1976 * If we ever get support for fs block sizes > page_size, we'll need
1977 * to remove the #if statements in the next two functions...
1978 */
1979 static inline unsigned int
1980 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1981 {
1982 unsigned len = le16_to_cpu(dlen);
1983
1984 #if (PAGE_SIZE >= 65536)
1985 if (len == EXT4_MAX_REC_LEN || len == 0)
1986 return blocksize;
1987 return (len & 65532) | ((len & 3) << 16);
1988 #else
1989 return len;
1990 #endif
1991 }
1992
1993 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1994 {
1995 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1996 BUG();
1997 #if (PAGE_SIZE >= 65536)
1998 if (len < 65536)
1999 return cpu_to_le16(len);
2000 if (len == blocksize) {
2001 if (blocksize == 65536)
2002 return cpu_to_le16(EXT4_MAX_REC_LEN);
2003 else
2004 return cpu_to_le16(0);
2005 }
2006 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
2007 #else
2008 return cpu_to_le16(len);
2009 #endif
2010 }
2011
2012 /*
2013 * Hash Tree Directory indexing
2014 * (c) Daniel Phillips, 2001
2015 */
2016
2017 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
2018 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2019 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
2020 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2021
2022 /* Legal values for the dx_root hash_version field: */
2023
2024 #define DX_HASH_LEGACY 0
2025 #define DX_HASH_HALF_MD4 1
2026 #define DX_HASH_TEA 2
2027 #define DX_HASH_LEGACY_UNSIGNED 3
2028 #define DX_HASH_HALF_MD4_UNSIGNED 4
2029 #define DX_HASH_TEA_UNSIGNED 5
2030
2031 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2032 const void *address, unsigned int length)
2033 {
2034 struct {
2035 struct shash_desc shash;
2036 char ctx[4];
2037 } desc;
2038 int err;
2039
2040 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2041
2042 desc.shash.tfm = sbi->s_chksum_driver;
2043 desc.shash.flags = 0;
2044 *(u32 *)desc.ctx = crc;
2045
2046 err = crypto_shash_update(&desc.shash, address, length);
2047 BUG_ON(err);
2048
2049 return *(u32 *)desc.ctx;
2050 }
2051
2052 #ifdef __KERNEL__
2053
2054 /* hash info structure used by the directory hash */
2055 struct dx_hash_info
2056 {
2057 u32 hash;
2058 u32 minor_hash;
2059 int hash_version;
2060 u32 *seed;
2061 };
2062
2063
2064 /* 32 and 64 bit signed EOF for dx directories */
2065 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
2066 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
2067
2068
2069 /*
2070 * Control parameters used by ext4_htree_next_block
2071 */
2072 #define HASH_NB_ALWAYS 1
2073
2074 struct ext4_filename {
2075 const struct qstr *usr_fname;
2076 struct fscrypt_str disk_name;
2077 struct dx_hash_info hinfo;
2078 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2079 struct fscrypt_str crypto_buf;
2080 #endif
2081 };
2082
2083 #define fname_name(p) ((p)->disk_name.name)
2084 #define fname_len(p) ((p)->disk_name.len)
2085
2086 /*
2087 * Describe an inode's exact location on disk and in memory
2088 */
2089 struct ext4_iloc
2090 {
2091 struct buffer_head *bh;
2092 unsigned long offset;
2093 ext4_group_t block_group;
2094 };
2095
2096 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2097 {
2098 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2099 }
2100
2101 /*
2102 * This structure is stuffed into the struct file's private_data field
2103 * for directories. It is where we put information so that we can do
2104 * readdir operations in hash tree order.
2105 */
2106 struct dir_private_info {
2107 struct rb_root root;
2108 struct rb_node *curr_node;
2109 struct fname *extra_fname;
2110 loff_t last_pos;
2111 __u32 curr_hash;
2112 __u32 curr_minor_hash;
2113 __u32 next_hash;
2114 };
2115
2116 /* calculate the first block number of the group */
2117 static inline ext4_fsblk_t
2118 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2119 {
2120 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2121 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2122 }
2123
2124 /*
2125 * Special error return code only used by dx_probe() and its callers.
2126 */
2127 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
2128
2129 /*
2130 * Timeout and state flag for lazy initialization inode thread.
2131 */
2132 #define EXT4_DEF_LI_WAIT_MULT 10
2133 #define EXT4_DEF_LI_MAX_START_DELAY 5
2134 #define EXT4_LAZYINIT_QUIT 0x0001
2135 #define EXT4_LAZYINIT_RUNNING 0x0002
2136
2137 /*
2138 * Lazy inode table initialization info
2139 */
2140 struct ext4_lazy_init {
2141 unsigned long li_state;
2142 struct list_head li_request_list;
2143 struct mutex li_list_mtx;
2144 };
2145
2146 struct ext4_li_request {
2147 struct super_block *lr_super;
2148 struct ext4_sb_info *lr_sbi;
2149 ext4_group_t lr_next_group;
2150 struct list_head lr_request;
2151 unsigned long lr_next_sched;
2152 unsigned long lr_timeout;
2153 };
2154
2155 struct ext4_features {
2156 struct kobject f_kobj;
2157 struct completion f_kobj_unregister;
2158 };
2159
2160 /*
2161 * This structure will be used for multiple mount protection. It will be
2162 * written into the block number saved in the s_mmp_block field in the
2163 * superblock. Programs that check MMP should assume that if
2164 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2165 * to use the filesystem, regardless of how old the timestamp is.
2166 */
2167 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
2168 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2169 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
2170 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
2171
2172 struct mmp_struct {
2173 __le32 mmp_magic; /* Magic number for MMP */
2174 __le32 mmp_seq; /* Sequence no. updated periodically */
2175
2176 /*
2177 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2178 * purposes and do not affect the correctness of the algorithm
2179 */
2180 __le64 mmp_time; /* Time last updated */
2181 char mmp_nodename[64]; /* Node which last updated MMP block */
2182 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
2183
2184 /*
2185 * mmp_check_interval is used to verify if the MMP block has been
2186 * updated on the block device. The value is updated based on the
2187 * maximum time to write the MMP block during an update cycle.
2188 */
2189 __le16 mmp_check_interval;
2190
2191 __le16 mmp_pad1;
2192 __le32 mmp_pad2[226];
2193 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
2194 };
2195
2196 /* arguments passed to the mmp thread */
2197 struct mmpd_data {
2198 struct buffer_head *bh; /* bh from initial read_mmp_block() */
2199 struct super_block *sb; /* super block of the fs */
2200 };
2201
2202 /*
2203 * Check interval multiplier
2204 * The MMP block is written every update interval and initially checked every
2205 * update interval x the multiplier (the value is then adapted based on the
2206 * write latency). The reason is that writes can be delayed under load and we
2207 * don't want readers to incorrectly assume that the filesystem is no longer
2208 * in use.
2209 */
2210 #define EXT4_MMP_CHECK_MULT 2UL
2211
2212 /*
2213 * Minimum interval for MMP checking in seconds.
2214 */
2215 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
2216
2217 /*
2218 * Maximum interval for MMP checking in seconds.
2219 */
2220 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
2221
2222 /*
2223 * Function prototypes
2224 */
2225
2226 /*
2227 * Ok, these declarations are also in <linux/kernel.h> but none of the
2228 * ext4 source programs needs to include it so they are duplicated here.
2229 */
2230 # define NORET_TYPE /**/
2231 # define ATTRIB_NORET __attribute__((noreturn))
2232 # define NORET_AND noreturn,
2233
2234 /* bitmap.c */
2235 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2236 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2237 struct ext4_group_desc *gdp,
2238 struct buffer_head *bh, int sz);
2239 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2240 struct ext4_group_desc *gdp,
2241 struct buffer_head *bh, int sz);
2242 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2243 struct ext4_group_desc *gdp,
2244 struct buffer_head *bh);
2245 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2246 struct ext4_group_desc *gdp,
2247 struct buffer_head *bh);
2248
2249 /* balloc.c */
2250 extern void ext4_get_group_no_and_offset(struct super_block *sb,
2251 ext4_fsblk_t blocknr,
2252 ext4_group_t *blockgrpp,
2253 ext4_grpblk_t *offsetp);
2254 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2255 ext4_fsblk_t block);
2256
2257 extern unsigned int ext4_block_group(struct super_block *sb,
2258 ext4_fsblk_t blocknr);
2259 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2260 ext4_fsblk_t blocknr);
2261 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2262 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2263 ext4_group_t group);
2264 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2265 ext4_fsblk_t goal,
2266 unsigned int flags,
2267 unsigned long *count,
2268 int *errp);
2269 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2270 s64 nclusters, unsigned int flags);
2271 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2272 extern void ext4_check_blocks_bitmap(struct super_block *);
2273 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2274 ext4_group_t block_group,
2275 struct buffer_head ** bh);
2276 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2277
2278 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2279 ext4_group_t block_group);
2280 extern int ext4_wait_block_bitmap(struct super_block *sb,
2281 ext4_group_t block_group,
2282 struct buffer_head *bh);
2283 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2284 ext4_group_t block_group);
2285 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2286 ext4_group_t block_group,
2287 struct ext4_group_desc *gdp);
2288 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2289
2290 static inline bool ext4_encrypted_inode(struct inode *inode)
2291 {
2292 return ext4_test_inode_flag(inode, EXT4_INODE_ENCRYPT);
2293 }
2294
2295 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2296 static inline int ext4_fname_setup_filename(struct inode *dir,
2297 const struct qstr *iname,
2298 int lookup, struct ext4_filename *fname)
2299 {
2300 struct fscrypt_name name;
2301 int err;
2302
2303 memset(fname, 0, sizeof(struct ext4_filename));
2304
2305 err = fscrypt_setup_filename(dir, iname, lookup, &name);
2306
2307 fname->usr_fname = name.usr_fname;
2308 fname->disk_name = name.disk_name;
2309 fname->hinfo.hash = name.hash;
2310 fname->hinfo.minor_hash = name.minor_hash;
2311 fname->crypto_buf = name.crypto_buf;
2312 return err;
2313 }
2314
2315 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2316 {
2317 struct fscrypt_name name;
2318
2319 name.crypto_buf = fname->crypto_buf;
2320 fscrypt_free_filename(&name);
2321
2322 fname->crypto_buf.name = NULL;
2323 fname->usr_fname = NULL;
2324 fname->disk_name.name = NULL;
2325 }
2326 #else
2327 static inline int ext4_fname_setup_filename(struct inode *dir,
2328 const struct qstr *iname,
2329 int lookup, struct ext4_filename *fname)
2330 {
2331 fname->usr_fname = iname;
2332 fname->disk_name.name = (unsigned char *) iname->name;
2333 fname->disk_name.len = iname->len;
2334 return 0;
2335 }
2336 static inline void ext4_fname_free_filename(struct ext4_filename *fname) { }
2337
2338 #endif
2339
2340 /* dir.c */
2341 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2342 struct file *,
2343 struct ext4_dir_entry_2 *,
2344 struct buffer_head *, char *, int,
2345 unsigned int);
2346 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2347 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2348 (de), (bh), (buf), (size), (offset)))
2349 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2350 __u32 minor_hash,
2351 struct ext4_dir_entry_2 *dirent,
2352 struct fscrypt_str *ent_name);
2353 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2354 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2355 struct buffer_head *bh,
2356 void *buf, int buf_size,
2357 struct ext4_filename *fname,
2358 struct ext4_dir_entry_2 **dest_de);
2359 void ext4_insert_dentry(struct inode *inode,
2360 struct ext4_dir_entry_2 *de,
2361 int buf_size,
2362 struct ext4_filename *fname);
2363 static inline void ext4_update_dx_flag(struct inode *inode)
2364 {
2365 if (!ext4_has_feature_dir_index(inode->i_sb))
2366 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2367 }
2368 static const unsigned char ext4_filetype_table[] = {
2369 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2370 };
2371
2372 static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2373 {
2374 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2375 return DT_UNKNOWN;
2376
2377 return ext4_filetype_table[filetype];
2378 }
2379 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2380 void *buf, int buf_size);
2381
2382 /* fsync.c */
2383 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2384
2385 /* hash.c */
2386 extern int ext4fs_dirhash(const char *name, int len, struct
2387 dx_hash_info *hinfo);
2388
2389 /* ialloc.c */
2390 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2391 const struct qstr *qstr, __u32 goal,
2392 uid_t *owner, int handle_type,
2393 unsigned int line_no, int nblocks);
2394
2395 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner) \
2396 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2397 0, 0, 0)
2398 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2399 type, nblocks) \
2400 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2401 (type), __LINE__, (nblocks))
2402
2403
2404 extern void ext4_free_inode(handle_t *, struct inode *);
2405 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2406 extern unsigned long ext4_count_free_inodes(struct super_block *);
2407 extern unsigned long ext4_count_dirs(struct super_block *);
2408 extern void ext4_check_inodes_bitmap(struct super_block *);
2409 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2410 extern int ext4_init_inode_table(struct super_block *sb,
2411 ext4_group_t group, int barrier);
2412 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2413
2414 /* mballoc.c */
2415 extern const struct file_operations ext4_seq_mb_groups_fops;
2416 extern long ext4_mb_stats;
2417 extern long ext4_mb_max_to_scan;
2418 extern int ext4_mb_init(struct super_block *);
2419 extern int ext4_mb_release(struct super_block *);
2420 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2421 struct ext4_allocation_request *, int *);
2422 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2423 extern void ext4_discard_preallocations(struct inode *);
2424 extern int __init ext4_init_mballoc(void);
2425 extern void ext4_exit_mballoc(void);
2426 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2427 struct buffer_head *bh, ext4_fsblk_t block,
2428 unsigned long count, int flags);
2429 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2430 ext4_group_t ngroups);
2431 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2432 ext4_group_t i, struct ext4_group_desc *desc);
2433 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2434 ext4_fsblk_t block, unsigned long count);
2435 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2436
2437 /* inode.c */
2438 int ext4_inode_is_fast_symlink(struct inode *inode);
2439 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2440 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2441 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2442 struct buffer_head *bh_result, int create);
2443 int ext4_get_block(struct inode *inode, sector_t iblock,
2444 struct buffer_head *bh_result, int create);
2445 int ext4_dio_get_block(struct inode *inode, sector_t iblock,
2446 struct buffer_head *bh_result, int create);
2447 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2448 struct buffer_head *bh, int create);
2449 int ext4_walk_page_buffers(handle_t *handle,
2450 struct buffer_head *head,
2451 unsigned from,
2452 unsigned to,
2453 int *partial,
2454 int (*fn)(handle_t *handle,
2455 struct buffer_head *bh));
2456 int do_journal_get_write_access(handle_t *handle,
2457 struct buffer_head *bh);
2458 #define FALL_BACK_TO_NONDELALLOC 1
2459 #define CONVERT_INLINE_DATA 2
2460
2461 extern struct inode *ext4_iget(struct super_block *, unsigned long);
2462 extern struct inode *ext4_iget_normal(struct super_block *, unsigned long);
2463 extern int ext4_write_inode(struct inode *, struct writeback_control *);
2464 extern int ext4_setattr(struct dentry *, struct iattr *);
2465 extern int ext4_getattr(const struct path *, struct kstat *, u32, unsigned int);
2466 extern void ext4_evict_inode(struct inode *);
2467 extern void ext4_clear_inode(struct inode *);
2468 extern int ext4_file_getattr(const struct path *, struct kstat *, u32, unsigned int);
2469 extern int ext4_sync_inode(handle_t *, struct inode *);
2470 extern void ext4_dirty_inode(struct inode *, int);
2471 extern int ext4_change_inode_journal_flag(struct inode *, int);
2472 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2473 extern int ext4_inode_attach_jinode(struct inode *inode);
2474 extern int ext4_can_truncate(struct inode *inode);
2475 extern int ext4_truncate(struct inode *);
2476 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2477 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
2478 extern void ext4_set_inode_flags(struct inode *);
2479 extern int ext4_alloc_da_blocks(struct inode *inode);
2480 extern void ext4_set_aops(struct inode *inode);
2481 extern int ext4_writepage_trans_blocks(struct inode *);
2482 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2483 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2484 loff_t lstart, loff_t lend);
2485 extern int ext4_page_mkwrite(struct vm_fault *vmf);
2486 extern int ext4_filemap_fault(struct vm_fault *vmf);
2487 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2488 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
2489 extern void ext4_da_update_reserve_space(struct inode *inode,
2490 int used, int quota_claim);
2491 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
2492 ext4_fsblk_t pblk, ext4_lblk_t len);
2493 extern int ext4_get_next_extent(struct inode *inode, ext4_lblk_t lblk,
2494 unsigned int map_len,
2495 struct extent_status *result);
2496
2497 /* indirect.c */
2498 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2499 struct ext4_map_blocks *map, int flags);
2500 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2501 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2502 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2503 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2504 ext4_lblk_t start, ext4_lblk_t end);
2505
2506 /* ioctl.c */
2507 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2508 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2509
2510 /* migrate.c */
2511 extern int ext4_ext_migrate(struct inode *);
2512 extern int ext4_ind_migrate(struct inode *inode);
2513
2514 /* namei.c */
2515 extern int ext4_dirent_csum_verify(struct inode *inode,
2516 struct ext4_dir_entry *dirent);
2517 extern int ext4_orphan_add(handle_t *, struct inode *);
2518 extern int ext4_orphan_del(handle_t *, struct inode *);
2519 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2520 __u32 start_minor_hash, __u32 *next_hash);
2521 extern int ext4_search_dir(struct buffer_head *bh,
2522 char *search_buf,
2523 int buf_size,
2524 struct inode *dir,
2525 struct ext4_filename *fname,
2526 unsigned int offset,
2527 struct ext4_dir_entry_2 **res_dir);
2528 extern int ext4_generic_delete_entry(handle_t *handle,
2529 struct inode *dir,
2530 struct ext4_dir_entry_2 *de_del,
2531 struct buffer_head *bh,
2532 void *entry_buf,
2533 int buf_size,
2534 int csum_size);
2535 extern bool ext4_empty_dir(struct inode *inode);
2536
2537 /* resize.c */
2538 extern int ext4_group_add(struct super_block *sb,
2539 struct ext4_new_group_data *input);
2540 extern int ext4_group_extend(struct super_block *sb,
2541 struct ext4_super_block *es,
2542 ext4_fsblk_t n_blocks_count);
2543 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2544
2545 /* super.c */
2546 extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
2547 extern int ext4_calculate_overhead(struct super_block *sb);
2548 extern void ext4_superblock_csum_set(struct super_block *sb);
2549 extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2550 extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2551 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2552 ext4_group_t ngroup);
2553 extern const char *ext4_decode_error(struct super_block *sb, int errno,
2554 char nbuf[16]);
2555
2556 extern __printf(4, 5)
2557 void __ext4_error(struct super_block *, const char *, unsigned int,
2558 const char *, ...);
2559 extern __printf(5, 6)
2560 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2561 const char *, ...);
2562 extern __printf(5, 6)
2563 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2564 const char *, ...);
2565 extern void __ext4_std_error(struct super_block *, const char *,
2566 unsigned int, int);
2567 extern __printf(4, 5)
2568 void __ext4_abort(struct super_block *, const char *, unsigned int,
2569 const char *, ...);
2570 extern __printf(4, 5)
2571 void __ext4_warning(struct super_block *, const char *, unsigned int,
2572 const char *, ...);
2573 extern __printf(4, 5)
2574 void __ext4_warning_inode(const struct inode *inode, const char *function,
2575 unsigned int line, const char *fmt, ...);
2576 extern __printf(3, 4)
2577 void __ext4_msg(struct super_block *, const char *, const char *, ...);
2578 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2579 const char *, unsigned int, const char *);
2580 extern __printf(7, 8)
2581 void __ext4_grp_locked_error(const char *, unsigned int,
2582 struct super_block *, ext4_group_t,
2583 unsigned long, ext4_fsblk_t,
2584 const char *, ...);
2585
2586 #define EXT4_ERROR_INODE(inode, fmt, a...) \
2587 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
2588
2589 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
2590 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
2591
2592 #define EXT4_ERROR_FILE(file, block, fmt, a...) \
2593 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
2594
2595 #ifdef CONFIG_PRINTK
2596
2597 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2598 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__)
2599 #define ext4_error_file(file, func, line, block, fmt, ...) \
2600 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2601 #define ext4_error(sb, fmt, ...) \
2602 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2603 #define ext4_abort(sb, fmt, ...) \
2604 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2605 #define ext4_warning(sb, fmt, ...) \
2606 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2607 #define ext4_warning_inode(inode, fmt, ...) \
2608 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
2609 #define ext4_msg(sb, level, fmt, ...) \
2610 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2611 #define dump_mmp_msg(sb, mmp, msg) \
2612 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2613 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2614 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2615 fmt, ##__VA_ARGS__)
2616
2617 #else
2618
2619 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2620 do { \
2621 no_printk(fmt, ##__VA_ARGS__); \
2622 __ext4_error_inode(inode, "", 0, block, " "); \
2623 } while (0)
2624 #define ext4_error_file(file, func, line, block, fmt, ...) \
2625 do { \
2626 no_printk(fmt, ##__VA_ARGS__); \
2627 __ext4_error_file(file, "", 0, block, " "); \
2628 } while (0)
2629 #define ext4_error(sb, fmt, ...) \
2630 do { \
2631 no_printk(fmt, ##__VA_ARGS__); \
2632 __ext4_error(sb, "", 0, " "); \
2633 } while (0)
2634 #define ext4_abort(sb, fmt, ...) \
2635 do { \
2636 no_printk(fmt, ##__VA_ARGS__); \
2637 __ext4_abort(sb, "", 0, " "); \
2638 } while (0)
2639 #define ext4_warning(sb, fmt, ...) \
2640 do { \
2641 no_printk(fmt, ##__VA_ARGS__); \
2642 __ext4_warning(sb, "", 0, " "); \
2643 } while (0)
2644 #define ext4_warning_inode(inode, fmt, ...) \
2645 do { \
2646 no_printk(fmt, ##__VA_ARGS__); \
2647 __ext4_warning_inode(inode, "", 0, " "); \
2648 } while (0)
2649 #define ext4_msg(sb, level, fmt, ...) \
2650 do { \
2651 no_printk(fmt, ##__VA_ARGS__); \
2652 __ext4_msg(sb, "", " "); \
2653 } while (0)
2654 #define dump_mmp_msg(sb, mmp, msg) \
2655 __dump_mmp_msg(sb, mmp, "", 0, "")
2656 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2657 do { \
2658 no_printk(fmt, ##__VA_ARGS__); \
2659 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
2660 } while (0)
2661
2662 #endif
2663
2664 extern void ext4_update_dynamic_rev(struct super_block *sb);
2665 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2666 __u32 compat);
2667 extern int ext4_update_rocompat_feature(handle_t *handle,
2668 struct super_block *sb, __u32 rocompat);
2669 extern int ext4_update_incompat_feature(handle_t *handle,
2670 struct super_block *sb, __u32 incompat);
2671 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2672 struct ext4_group_desc *bg);
2673 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2674 struct ext4_group_desc *bg);
2675 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2676 struct ext4_group_desc *bg);
2677 extern __u32 ext4_free_group_clusters(struct super_block *sb,
2678 struct ext4_group_desc *bg);
2679 extern __u32 ext4_free_inodes_count(struct super_block *sb,
2680 struct ext4_group_desc *bg);
2681 extern __u32 ext4_used_dirs_count(struct super_block *sb,
2682 struct ext4_group_desc *bg);
2683 extern __u32 ext4_itable_unused_count(struct super_block *sb,
2684 struct ext4_group_desc *bg);
2685 extern void ext4_block_bitmap_set(struct super_block *sb,
2686 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2687 extern void ext4_inode_bitmap_set(struct super_block *sb,
2688 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2689 extern void ext4_inode_table_set(struct super_block *sb,
2690 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2691 extern void ext4_free_group_clusters_set(struct super_block *sb,
2692 struct ext4_group_desc *bg,
2693 __u32 count);
2694 extern void ext4_free_inodes_set(struct super_block *sb,
2695 struct ext4_group_desc *bg, __u32 count);
2696 extern void ext4_used_dirs_set(struct super_block *sb,
2697 struct ext4_group_desc *bg, __u32 count);
2698 extern void ext4_itable_unused_set(struct super_block *sb,
2699 struct ext4_group_desc *bg, __u32 count);
2700 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2701 struct ext4_group_desc *gdp);
2702 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2703 struct ext4_group_desc *gdp);
2704 extern int ext4_register_li_request(struct super_block *sb,
2705 ext4_group_t first_not_zeroed);
2706
2707 static inline int ext4_has_group_desc_csum(struct super_block *sb)
2708 {
2709 return ext4_has_feature_gdt_csum(sb) ||
2710 EXT4_SB(sb)->s_chksum_driver != NULL;
2711 }
2712
2713 static inline int ext4_has_metadata_csum(struct super_block *sb)
2714 {
2715 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
2716 !EXT4_SB(sb)->s_chksum_driver);
2717
2718 return (EXT4_SB(sb)->s_chksum_driver != NULL);
2719 }
2720 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2721 {
2722 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2723 le32_to_cpu(es->s_blocks_count_lo);
2724 }
2725
2726 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2727 {
2728 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2729 le32_to_cpu(es->s_r_blocks_count_lo);
2730 }
2731
2732 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2733 {
2734 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2735 le32_to_cpu(es->s_free_blocks_count_lo);
2736 }
2737
2738 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2739 ext4_fsblk_t blk)
2740 {
2741 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2742 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2743 }
2744
2745 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2746 ext4_fsblk_t blk)
2747 {
2748 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2749 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2750 }
2751
2752 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2753 ext4_fsblk_t blk)
2754 {
2755 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2756 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2757 }
2758
2759 static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
2760 {
2761 if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2762 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2763 le32_to_cpu(raw_inode->i_size_lo);
2764 else
2765 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2766 }
2767
2768 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2769 {
2770 raw_inode->i_size_lo = cpu_to_le32(i_size);
2771 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2772 }
2773
2774 static inline
2775 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2776 ext4_group_t group)
2777 {
2778 struct ext4_group_info ***grp_info;
2779 long indexv, indexh;
2780 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
2781 grp_info = EXT4_SB(sb)->s_group_info;
2782 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2783 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2784 return grp_info[indexv][indexh];
2785 }
2786
2787 /*
2788 * Reading s_groups_count requires using smp_rmb() afterwards. See
2789 * the locking protocol documented in the comments of ext4_group_add()
2790 * in resize.c
2791 */
2792 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2793 {
2794 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
2795
2796 smp_rmb();
2797 return ngroups;
2798 }
2799
2800 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2801 ext4_group_t block_group)
2802 {
2803 return block_group >> sbi->s_log_groups_per_flex;
2804 }
2805
2806 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2807 {
2808 return 1 << sbi->s_log_groups_per_flex;
2809 }
2810
2811 #define ext4_std_error(sb, errno) \
2812 do { \
2813 if ((errno)) \
2814 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
2815 } while (0)
2816
2817 #ifdef CONFIG_SMP
2818 /* Each CPU can accumulate percpu_counter_batch clusters in their local
2819 * counters. So we need to make sure we have free clusters more
2820 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
2821 */
2822 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2823 #else
2824 #define EXT4_FREECLUSTERS_WATERMARK 0
2825 #endif
2826
2827 /* Update i_disksize. Requires i_mutex to avoid races with truncate */
2828 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2829 {
2830 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
2831 !inode_is_locked(inode));
2832 down_write(&EXT4_I(inode)->i_data_sem);
2833 if (newsize > EXT4_I(inode)->i_disksize)
2834 EXT4_I(inode)->i_disksize = newsize;
2835 up_write(&EXT4_I(inode)->i_data_sem);
2836 }
2837
2838 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
2839 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
2840 {
2841 int changed = 0;
2842
2843 if (newsize > inode->i_size) {
2844 i_size_write(inode, newsize);
2845 changed = 1;
2846 }
2847 if (newsize > EXT4_I(inode)->i_disksize) {
2848 ext4_update_i_disksize(inode, newsize);
2849 changed |= 2;
2850 }
2851 return changed;
2852 }
2853
2854 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
2855 loff_t len);
2856
2857 struct ext4_group_info {
2858 unsigned long bb_state;
2859 struct rb_root bb_free_root;
2860 ext4_grpblk_t bb_first_free; /* first free block */
2861 ext4_grpblk_t bb_free; /* total free blocks */
2862 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
2863 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
2864 struct list_head bb_prealloc_list;
2865 #ifdef DOUBLE_CHECK
2866 void *bb_bitmap;
2867 #endif
2868 struct rw_semaphore alloc_sem;
2869 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
2870 * regions, index is order.
2871 * bb_counters[3] = 5 means
2872 * 5 free 8-block regions. */
2873 };
2874
2875 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
2876 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
2877 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
2878 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
2879
2880 #define EXT4_MB_GRP_NEED_INIT(grp) \
2881 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2882 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
2883 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2884 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
2885 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2886
2887 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
2888 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2889 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
2890 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2891 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
2892 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2893
2894 #define EXT4_MAX_CONTENTION 8
2895 #define EXT4_CONTENTION_THRESHOLD 2
2896
2897 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2898 ext4_group_t group)
2899 {
2900 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2901 }
2902
2903 /*
2904 * Returns true if the filesystem is busy enough that attempts to
2905 * access the block group locks has run into contention.
2906 */
2907 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2908 {
2909 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2910 }
2911
2912 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2913 {
2914 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2915 if (spin_trylock(lock))
2916 /*
2917 * We're able to grab the lock right away, so drop the
2918 * lock contention counter.
2919 */
2920 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2921 else {
2922 /*
2923 * The lock is busy, so bump the contention counter,
2924 * and then wait on the spin lock.
2925 */
2926 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2927 EXT4_MAX_CONTENTION);
2928 spin_lock(lock);
2929 }
2930 }
2931
2932 static inline void ext4_unlock_group(struct super_block *sb,
2933 ext4_group_t group)
2934 {
2935 spin_unlock(ext4_group_lock_ptr(sb, group));
2936 }
2937
2938 /*
2939 * Block validity checking
2940 */
2941 #define ext4_check_indirect_blockref(inode, bh) \
2942 ext4_check_blockref(__func__, __LINE__, inode, \
2943 (__le32 *)(bh)->b_data, \
2944 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2945
2946 #define ext4_ind_check_inode(inode) \
2947 ext4_check_blockref(__func__, __LINE__, inode, \
2948 EXT4_I(inode)->i_data, \
2949 EXT4_NDIR_BLOCKS)
2950
2951 /*
2952 * Inodes and files operations
2953 */
2954
2955 /* dir.c */
2956 extern const struct file_operations ext4_dir_operations;
2957
2958 /* file.c */
2959 extern const struct inode_operations ext4_file_inode_operations;
2960 extern const struct file_operations ext4_file_operations;
2961 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
2962
2963 /* inline.c */
2964 extern int ext4_get_max_inline_size(struct inode *inode);
2965 extern int ext4_find_inline_data_nolock(struct inode *inode);
2966 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
2967 unsigned int len);
2968 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
2969
2970 extern int ext4_readpage_inline(struct inode *inode, struct page *page);
2971 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
2972 struct inode *inode,
2973 loff_t pos, unsigned len,
2974 unsigned flags,
2975 struct page **pagep);
2976 extern int ext4_write_inline_data_end(struct inode *inode,
2977 loff_t pos, unsigned len,
2978 unsigned copied,
2979 struct page *page);
2980 extern struct buffer_head *
2981 ext4_journalled_write_inline_data(struct inode *inode,
2982 unsigned len,
2983 struct page *page);
2984 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
2985 struct inode *inode,
2986 loff_t pos, unsigned len,
2987 unsigned flags,
2988 struct page **pagep,
2989 void **fsdata);
2990 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
2991 unsigned len, unsigned copied,
2992 struct page *page);
2993 extern int ext4_try_add_inline_entry(handle_t *handle,
2994 struct ext4_filename *fname,
2995 struct inode *dir, struct inode *inode);
2996 extern int ext4_try_create_inline_dir(handle_t *handle,
2997 struct inode *parent,
2998 struct inode *inode);
2999 extern int ext4_read_inline_dir(struct file *filp,
3000 struct dir_context *ctx,
3001 int *has_inline_data);
3002 extern int htree_inlinedir_to_tree(struct file *dir_file,
3003 struct inode *dir, ext4_lblk_t block,
3004 struct dx_hash_info *hinfo,
3005 __u32 start_hash, __u32 start_minor_hash,
3006 int *has_inline_data);
3007 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3008 struct ext4_filename *fname,
3009 struct ext4_dir_entry_2 **res_dir,
3010 int *has_inline_data);
3011 extern int ext4_delete_inline_entry(handle_t *handle,
3012 struct inode *dir,
3013 struct ext4_dir_entry_2 *de_del,
3014 struct buffer_head *bh,
3015 int *has_inline_data);
3016 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3017 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3018 struct ext4_dir_entry_2 **parent_de,
3019 int *retval);
3020 extern int ext4_inline_data_fiemap(struct inode *inode,
3021 struct fiemap_extent_info *fieinfo,
3022 int *has_inline, __u64 start, __u64 len);
3023 extern int ext4_try_to_evict_inline_data(handle_t *handle,
3024 struct inode *inode,
3025 int needed);
3026 extern int ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3027
3028 extern int ext4_convert_inline_data(struct inode *inode);
3029
3030 static inline int ext4_has_inline_data(struct inode *inode)
3031 {
3032 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3033 EXT4_I(inode)->i_inline_off;
3034 }
3035
3036 /* namei.c */
3037 extern const struct inode_operations ext4_dir_inode_operations;
3038 extern const struct inode_operations ext4_special_inode_operations;
3039 extern struct dentry *ext4_get_parent(struct dentry *child);
3040 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3041 struct ext4_dir_entry_2 *de,
3042 int blocksize, int csum_size,
3043 unsigned int parent_ino, int dotdot_real_len);
3044 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t,
3045 unsigned int blocksize);
3046 extern int ext4_handle_dirty_dirent_node(handle_t *handle,
3047 struct inode *inode,
3048 struct buffer_head *bh);
3049 #define S_SHIFT 12
3050 static const unsigned char ext4_type_by_mode[S_IFMT >> S_SHIFT] = {
3051 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
3052 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
3053 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
3054 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
3055 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
3056 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
3057 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
3058 };
3059
3060 static inline void ext4_set_de_type(struct super_block *sb,
3061 struct ext4_dir_entry_2 *de,
3062 umode_t mode) {
3063 if (ext4_has_feature_filetype(sb))
3064 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3065 }
3066
3067 /* readpages.c */
3068 extern int ext4_mpage_readpages(struct address_space *mapping,
3069 struct list_head *pages, struct page *page,
3070 unsigned nr_pages);
3071
3072 /* symlink.c */
3073 extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3074 extern const struct inode_operations ext4_symlink_inode_operations;
3075 extern const struct inode_operations ext4_fast_symlink_inode_operations;
3076
3077 /* sysfs.c */
3078 extern int ext4_register_sysfs(struct super_block *sb);
3079 extern void ext4_unregister_sysfs(struct super_block *sb);
3080 extern int __init ext4_init_sysfs(void);
3081 extern void ext4_exit_sysfs(void);
3082
3083 /* block_validity */
3084 extern void ext4_release_system_zone(struct super_block *sb);
3085 extern int ext4_setup_system_zone(struct super_block *sb);
3086 extern int __init ext4_init_system_zone(void);
3087 extern void ext4_exit_system_zone(void);
3088 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
3089 ext4_fsblk_t start_blk,
3090 unsigned int count);
3091 extern int ext4_check_blockref(const char *, unsigned int,
3092 struct inode *, __le32 *, unsigned int);
3093
3094 /* extents.c */
3095 struct ext4_ext_path;
3096 struct ext4_extent;
3097
3098 /*
3099 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3100 * __le32.
3101 */
3102 #define EXT_MAX_BLOCKS 0xffffffff
3103
3104 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
3105 extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
3106 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3107 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3108 struct ext4_map_blocks *map, int flags);
3109 extern int ext4_ext_truncate(handle_t *, struct inode *);
3110 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3111 ext4_lblk_t end);
3112 extern void ext4_ext_init(struct super_block *);
3113 extern void ext4_ext_release(struct super_block *);
3114 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3115 loff_t len);
3116 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3117 loff_t offset, ssize_t len);
3118 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3119 struct ext4_map_blocks *map, int flags);
3120 extern int ext4_ext_calc_metadata_amount(struct inode *inode,
3121 ext4_lblk_t lblocks);
3122 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3123 int num,
3124 struct ext4_ext_path *path);
3125 extern int ext4_can_extents_be_merged(struct inode *inode,
3126 struct ext4_extent *ex1,
3127 struct ext4_extent *ex2);
3128 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3129 struct ext4_ext_path **,
3130 struct ext4_extent *, int);
3131 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3132 struct ext4_ext_path **,
3133 int flags);
3134 extern void ext4_ext_drop_refs(struct ext4_ext_path *);
3135 extern int ext4_ext_check_inode(struct inode *inode);
3136 extern int ext4_find_delalloc_range(struct inode *inode,
3137 ext4_lblk_t lblk_start,
3138 ext4_lblk_t lblk_end);
3139 extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk);
3140 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3141 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3142 __u64 start, __u64 len);
3143 extern int ext4_ext_precache(struct inode *inode);
3144 extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len);
3145 extern int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len);
3146 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3147 struct inode *inode2, ext4_lblk_t lblk1,
3148 ext4_lblk_t lblk2, ext4_lblk_t count,
3149 int mark_unwritten,int *err);
3150
3151 /* move_extent.c */
3152 extern void ext4_double_down_write_data_sem(struct inode *first,
3153 struct inode *second);
3154 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3155 struct inode *donor_inode);
3156 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3157 __u64 start_orig, __u64 start_donor,
3158 __u64 len, __u64 *moved_len);
3159
3160 /* page-io.c */
3161 extern int __init ext4_init_pageio(void);
3162 extern void ext4_exit_pageio(void);
3163 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3164 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3165 extern int ext4_put_io_end(ext4_io_end_t *io_end);
3166 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3167 extern void ext4_io_submit_init(struct ext4_io_submit *io,
3168 struct writeback_control *wbc);
3169 extern void ext4_end_io_rsv_work(struct work_struct *work);
3170 extern void ext4_io_submit(struct ext4_io_submit *io);
3171 extern int ext4_bio_write_page(struct ext4_io_submit *io,
3172 struct page *page,
3173 int len,
3174 struct writeback_control *wbc,
3175 bool keep_towrite);
3176
3177 /* mmp.c */
3178 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3179
3180 /*
3181 * Add new method to test whether block and inode bitmaps are properly
3182 * initialized. With uninit_bg reading the block from disk is not enough
3183 * to mark the bitmap uptodate. We need to also zero-out the bitmap
3184 */
3185 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3186
3187 static inline int bitmap_uptodate(struct buffer_head *bh)
3188 {
3189 return (buffer_uptodate(bh) &&
3190 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3191 }
3192 static inline void set_bitmap_uptodate(struct buffer_head *bh)
3193 {
3194 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3195 }
3196
3197 /*
3198 * Disable DIO read nolock optimization, so new dioreaders will be forced
3199 * to grab i_mutex
3200 */
3201 static inline void ext4_inode_block_unlocked_dio(struct inode *inode)
3202 {
3203 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
3204 smp_mb();
3205 }
3206 static inline void ext4_inode_resume_unlocked_dio(struct inode *inode)
3207 {
3208 smp_mb();
3209 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
3210 }
3211
3212 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
3213
3214 /* For ioend & aio unwritten conversion wait queues */
3215 #define EXT4_WQ_HASH_SZ 37
3216 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
3217 EXT4_WQ_HASH_SZ])
3218 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3219
3220 extern int ext4_resize_begin(struct super_block *sb);
3221 extern void ext4_resize_end(struct super_block *sb);
3222
3223 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
3224 struct ext4_io_end *io_end)
3225 {
3226 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
3227 io_end->flag |= EXT4_IO_END_UNWRITTEN;
3228 atomic_inc(&EXT4_I(inode)->i_unwritten);
3229 }
3230 }
3231
3232 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3233 {
3234 struct inode *inode = io_end->inode;
3235
3236 if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
3237 io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3238 /* Wake up anyone waiting on unwritten extent conversion */
3239 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
3240 wake_up_all(ext4_ioend_wq(inode));
3241 }
3242 }
3243
3244 extern const struct iomap_ops ext4_iomap_ops;
3245
3246 #endif /* __KERNEL__ */
3247
3248 #define EFSBADCRC EBADMSG /* Bad CRC detected */
3249 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
3250
3251 #endif /* _EXT4_H */