Merge remote-tracking branch 'mmc-uh/next'
[deliverable/linux.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/blockgroup_lock.h>
32 #include <linux/percpu_counter.h>
33 #include <linux/ratelimit.h>
34 #include <crypto/hash.h>
35 #include <linux/fscrypto.h>
36 #include <linux/falloc.h>
37 #include <linux/percpu-rwsem.h>
38 #ifdef __KERNEL__
39 #include <linux/compat.h>
40 #endif
41
42 /*
43 * The fourth extended filesystem constants/structures
44 */
45
46 /*
47 * with AGGRESSIVE_CHECK allocator runs consistency checks over
48 * structures. these checks slow things down a lot
49 */
50 #define AGGRESSIVE_CHECK__
51
52 /*
53 * with DOUBLE_CHECK defined mballoc creates persistent in-core
54 * bitmaps, maintains and uses them to check for double allocations
55 */
56 #define DOUBLE_CHECK__
57
58 /*
59 * Define EXT4FS_DEBUG to produce debug messages
60 */
61 #undef EXT4FS_DEBUG
62
63 /*
64 * Debug code
65 */
66 #ifdef EXT4FS_DEBUG
67 #define ext4_debug(f, a...) \
68 do { \
69 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
70 __FILE__, __LINE__, __func__); \
71 printk(KERN_DEBUG f, ## a); \
72 } while (0)
73 #else
74 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
75 #endif
76
77 /*
78 * Turn on EXT_DEBUG to get lots of info about extents operations.
79 */
80 #define EXT_DEBUG__
81 #ifdef EXT_DEBUG
82 #define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__)
83 #else
84 #define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
85 #endif
86
87 /* data type for block offset of block group */
88 typedef int ext4_grpblk_t;
89
90 /* data type for filesystem-wide blocks number */
91 typedef unsigned long long ext4_fsblk_t;
92
93 /* data type for file logical block number */
94 typedef __u32 ext4_lblk_t;
95
96 /* data type for block group number */
97 typedef unsigned int ext4_group_t;
98
99 enum SHIFT_DIRECTION {
100 SHIFT_LEFT = 0,
101 SHIFT_RIGHT,
102 };
103
104 /*
105 * Flags used in mballoc's allocation_context flags field.
106 *
107 * Also used to show what's going on for debugging purposes when the
108 * flag field is exported via the traceport interface
109 */
110
111 /* prefer goal again. length */
112 #define EXT4_MB_HINT_MERGE 0x0001
113 /* blocks already reserved */
114 #define EXT4_MB_HINT_RESERVED 0x0002
115 /* metadata is being allocated */
116 #define EXT4_MB_HINT_METADATA 0x0004
117 /* first blocks in the file */
118 #define EXT4_MB_HINT_FIRST 0x0008
119 /* search for the best chunk */
120 #define EXT4_MB_HINT_BEST 0x0010
121 /* data is being allocated */
122 #define EXT4_MB_HINT_DATA 0x0020
123 /* don't preallocate (for tails) */
124 #define EXT4_MB_HINT_NOPREALLOC 0x0040
125 /* allocate for locality group */
126 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080
127 /* allocate goal blocks or none */
128 #define EXT4_MB_HINT_GOAL_ONLY 0x0100
129 /* goal is meaningful */
130 #define EXT4_MB_HINT_TRY_GOAL 0x0200
131 /* blocks already pre-reserved by delayed allocation */
132 #define EXT4_MB_DELALLOC_RESERVED 0x0400
133 /* We are doing stream allocation */
134 #define EXT4_MB_STREAM_ALLOC 0x0800
135 /* Use reserved root blocks if needed */
136 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000
137 /* Use blocks from reserved pool */
138 #define EXT4_MB_USE_RESERVED 0x2000
139
140 struct ext4_allocation_request {
141 /* target inode for block we're allocating */
142 struct inode *inode;
143 /* how many blocks we want to allocate */
144 unsigned int len;
145 /* logical block in target inode */
146 ext4_lblk_t logical;
147 /* the closest logical allocated block to the left */
148 ext4_lblk_t lleft;
149 /* the closest logical allocated block to the right */
150 ext4_lblk_t lright;
151 /* phys. target (a hint) */
152 ext4_fsblk_t goal;
153 /* phys. block for the closest logical allocated block to the left */
154 ext4_fsblk_t pleft;
155 /* phys. block for the closest logical allocated block to the right */
156 ext4_fsblk_t pright;
157 /* flags. see above EXT4_MB_HINT_* */
158 unsigned int flags;
159 };
160
161 /*
162 * Logical to physical block mapping, used by ext4_map_blocks()
163 *
164 * This structure is used to pass requests into ext4_map_blocks() as
165 * well as to store the information returned by ext4_map_blocks(). It
166 * takes less room on the stack than a struct buffer_head.
167 */
168 #define EXT4_MAP_NEW (1 << BH_New)
169 #define EXT4_MAP_MAPPED (1 << BH_Mapped)
170 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten)
171 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary)
172 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
173 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
174
175 struct ext4_map_blocks {
176 ext4_fsblk_t m_pblk;
177 ext4_lblk_t m_lblk;
178 unsigned int m_len;
179 unsigned int m_flags;
180 };
181
182 /*
183 * Flags for ext4_io_end->flags
184 */
185 #define EXT4_IO_END_UNWRITTEN 0x0001
186
187 /*
188 * For converting unwritten extents on a work queue. 'handle' is used for
189 * buffered writeback.
190 */
191 typedef struct ext4_io_end {
192 struct list_head list; /* per-file finished IO list */
193 handle_t *handle; /* handle reserved for extent
194 * conversion */
195 struct inode *inode; /* file being written to */
196 struct bio *bio; /* Linked list of completed
197 * bios covering the extent */
198 unsigned int flag; /* unwritten or not */
199 atomic_t count; /* reference counter */
200 loff_t offset; /* offset in the file */
201 ssize_t size; /* size of the extent */
202 } ext4_io_end_t;
203
204 struct ext4_io_submit {
205 struct writeback_control *io_wbc;
206 struct bio *io_bio;
207 ext4_io_end_t *io_end;
208 sector_t io_next_block;
209 };
210
211 /*
212 * Special inodes numbers
213 */
214 #define EXT4_BAD_INO 1 /* Bad blocks inode */
215 #define EXT4_ROOT_INO 2 /* Root inode */
216 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */
217 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
218 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
219 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
220 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
221 #define EXT4_JOURNAL_INO 8 /* Journal inode */
222
223 /* First non-reserved inode for old ext4 filesystems */
224 #define EXT4_GOOD_OLD_FIRST_INO 11
225
226 /*
227 * Maximal count of links to a file
228 */
229 #define EXT4_LINK_MAX 65000
230
231 /*
232 * Macro-instructions used to manage several block sizes
233 */
234 #define EXT4_MIN_BLOCK_SIZE 1024
235 #define EXT4_MAX_BLOCK_SIZE 65536
236 #define EXT4_MIN_BLOCK_LOG_SIZE 10
237 #define EXT4_MAX_BLOCK_LOG_SIZE 16
238 #ifdef __KERNEL__
239 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
240 #else
241 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
242 #endif
243 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
244 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
245 EXT4_SB(s)->s_cluster_bits)
246 #ifdef __KERNEL__
247 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
248 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
249 #else
250 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
251 #endif
252 #ifdef __KERNEL__
253 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
254 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
255 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
256 #else
257 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
258 EXT4_GOOD_OLD_INODE_SIZE : \
259 (s)->s_inode_size)
260 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
261 EXT4_GOOD_OLD_FIRST_INO : \
262 (s)->s_first_ino)
263 #endif
264 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
265
266 /* Translate a block number to a cluster number */
267 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
268 /* Translate a cluster number to a block number */
269 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
270 /* Translate # of blks to # of clusters */
271 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
272 (sbi)->s_cluster_bits)
273 /* Mask out the low bits to get the starting block of the cluster */
274 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
275 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
276 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
277 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
278 /* Get the cluster offset */
279 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
280 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
281 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
282 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
283
284 /*
285 * Structure of a blocks group descriptor
286 */
287 struct ext4_group_desc
288 {
289 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
290 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
291 __le32 bg_inode_table_lo; /* Inodes table block */
292 __le16 bg_free_blocks_count_lo;/* Free blocks count */
293 __le16 bg_free_inodes_count_lo;/* Free inodes count */
294 __le16 bg_used_dirs_count_lo; /* Directories count */
295 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
296 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
297 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
298 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
299 __le16 bg_itable_unused_lo; /* Unused inodes count */
300 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
301 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
302 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
303 __le32 bg_inode_table_hi; /* Inodes table block MSB */
304 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
305 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
306 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
307 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
308 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
309 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
310 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
311 __u32 bg_reserved;
312 };
313
314 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
315 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
316 sizeof(__le16))
317 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
318 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
319 sizeof(__le16))
320
321 /*
322 * Structure of a flex block group info
323 */
324
325 struct flex_groups {
326 atomic64_t free_clusters;
327 atomic_t free_inodes;
328 atomic_t used_dirs;
329 };
330
331 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
332 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
333 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
334
335 /*
336 * Macro-instructions used to manage group descriptors
337 */
338 #define EXT4_MIN_DESC_SIZE 32
339 #define EXT4_MIN_DESC_SIZE_64BIT 64
340 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
341 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
342 #ifdef __KERNEL__
343 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
344 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
345 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
346 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
347 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
348 #else
349 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
350 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
351 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
352 #endif
353
354 /*
355 * Constants relative to the data blocks
356 */
357 #define EXT4_NDIR_BLOCKS 12
358 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
359 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
360 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
361 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
362
363 /*
364 * Inode flags
365 */
366 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
367 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
368 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
369 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
370 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
371 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
372 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
373 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
374 /* Reserved for compression usage... */
375 #define EXT4_DIRTY_FL 0x00000100
376 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
377 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
378 /* nb: was previously EXT2_ECOMPR_FL */
379 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */
380 /* End compression flags --- maybe not all used */
381 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
382 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
383 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
384 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
385 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
386 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
387 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
388 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
389 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
390 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
391 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
392 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
393 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
394
395 #define EXT4_FL_USER_VISIBLE 0x304BDFFF /* User visible flags */
396 #define EXT4_FL_USER_MODIFIABLE 0x204380FF /* User modifiable flags */
397
398 #define EXT4_FL_XFLAG_VISIBLE (EXT4_SYNC_FL | \
399 EXT4_IMMUTABLE_FL | \
400 EXT4_APPEND_FL | \
401 EXT4_NODUMP_FL | \
402 EXT4_NOATIME_FL | \
403 EXT4_PROJINHERIT_FL)
404
405 /* Flags that should be inherited by new inodes from their parent. */
406 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
407 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
408 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
409 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL |\
410 EXT4_PROJINHERIT_FL)
411
412 /* Flags that are appropriate for regular files (all but dir-specific ones). */
413 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
414
415 /* Flags that are appropriate for non-directories/regular files. */
416 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
417
418 /* Mask out flags that are inappropriate for the given type of inode. */
419 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
420 {
421 if (S_ISDIR(mode))
422 return flags;
423 else if (S_ISREG(mode))
424 return flags & EXT4_REG_FLMASK;
425 else
426 return flags & EXT4_OTHER_FLMASK;
427 }
428
429 /*
430 * Inode flags used for atomic set/get
431 */
432 enum {
433 EXT4_INODE_SECRM = 0, /* Secure deletion */
434 EXT4_INODE_UNRM = 1, /* Undelete */
435 EXT4_INODE_COMPR = 2, /* Compress file */
436 EXT4_INODE_SYNC = 3, /* Synchronous updates */
437 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
438 EXT4_INODE_APPEND = 5, /* writes to file may only append */
439 EXT4_INODE_NODUMP = 6, /* do not dump file */
440 EXT4_INODE_NOATIME = 7, /* do not update atime */
441 /* Reserved for compression usage... */
442 EXT4_INODE_DIRTY = 8,
443 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
444 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
445 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */
446 /* End compression flags --- maybe not all used */
447 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
448 EXT4_INODE_IMAGIC = 13, /* AFS directory */
449 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
450 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
451 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
452 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
453 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
454 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
455 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
456 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
457 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
458 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */
459 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
460 };
461
462 /*
463 * Since it's pretty easy to mix up bit numbers and hex values, we use a
464 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
465 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
466 * any extra space in the compiled kernel image, otherwise, the build will fail.
467 * It's important that these values are the same, since we are using
468 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
469 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
470 * values found in ext2, ext3 and ext4 filesystems, and of course the values
471 * defined in e2fsprogs.
472 *
473 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
474 */
475 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
476 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
477
478 static inline void ext4_check_flag_values(void)
479 {
480 CHECK_FLAG_VALUE(SECRM);
481 CHECK_FLAG_VALUE(UNRM);
482 CHECK_FLAG_VALUE(COMPR);
483 CHECK_FLAG_VALUE(SYNC);
484 CHECK_FLAG_VALUE(IMMUTABLE);
485 CHECK_FLAG_VALUE(APPEND);
486 CHECK_FLAG_VALUE(NODUMP);
487 CHECK_FLAG_VALUE(NOATIME);
488 CHECK_FLAG_VALUE(DIRTY);
489 CHECK_FLAG_VALUE(COMPRBLK);
490 CHECK_FLAG_VALUE(NOCOMPR);
491 CHECK_FLAG_VALUE(ENCRYPT);
492 CHECK_FLAG_VALUE(INDEX);
493 CHECK_FLAG_VALUE(IMAGIC);
494 CHECK_FLAG_VALUE(JOURNAL_DATA);
495 CHECK_FLAG_VALUE(NOTAIL);
496 CHECK_FLAG_VALUE(DIRSYNC);
497 CHECK_FLAG_VALUE(TOPDIR);
498 CHECK_FLAG_VALUE(HUGE_FILE);
499 CHECK_FLAG_VALUE(EXTENTS);
500 CHECK_FLAG_VALUE(EA_INODE);
501 CHECK_FLAG_VALUE(EOFBLOCKS);
502 CHECK_FLAG_VALUE(INLINE_DATA);
503 CHECK_FLAG_VALUE(PROJINHERIT);
504 CHECK_FLAG_VALUE(RESERVED);
505 }
506
507 /* Used to pass group descriptor data when online resize is done */
508 struct ext4_new_group_input {
509 __u32 group; /* Group number for this data */
510 __u64 block_bitmap; /* Absolute block number of block bitmap */
511 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
512 __u64 inode_table; /* Absolute block number of inode table start */
513 __u32 blocks_count; /* Total number of blocks in this group */
514 __u16 reserved_blocks; /* Number of reserved blocks in this group */
515 __u16 unused;
516 };
517
518 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
519 struct compat_ext4_new_group_input {
520 u32 group;
521 compat_u64 block_bitmap;
522 compat_u64 inode_bitmap;
523 compat_u64 inode_table;
524 u32 blocks_count;
525 u16 reserved_blocks;
526 u16 unused;
527 };
528 #endif
529
530 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
531 struct ext4_new_group_data {
532 __u32 group;
533 __u64 block_bitmap;
534 __u64 inode_bitmap;
535 __u64 inode_table;
536 __u32 blocks_count;
537 __u16 reserved_blocks;
538 __u16 unused;
539 __u32 free_blocks_count;
540 };
541
542 /* Indexes used to index group tables in ext4_new_group_data */
543 enum {
544 BLOCK_BITMAP = 0, /* block bitmap */
545 INODE_BITMAP, /* inode bitmap */
546 INODE_TABLE, /* inode tables */
547 GROUP_TABLE_COUNT,
548 };
549
550 /*
551 * Flags used by ext4_map_blocks()
552 */
553 /* Allocate any needed blocks and/or convert an unwritten
554 extent to be an initialized ext4 */
555 #define EXT4_GET_BLOCKS_CREATE 0x0001
556 /* Request the creation of an unwritten extent */
557 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002
558 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\
559 EXT4_GET_BLOCKS_CREATE)
560 /* Caller is from the delayed allocation writeout path
561 * finally doing the actual allocation of delayed blocks */
562 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
563 /* caller is from the direct IO path, request to creation of an
564 unwritten extents if not allocated, split the unwritten
565 extent if blocks has been preallocated already*/
566 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
567 #define EXT4_GET_BLOCKS_CONVERT 0x0010
568 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
569 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
570 /* Convert extent to initialized after IO complete */
571 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
572 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
573 /* Eventual metadata allocation (due to growing extent tree)
574 * should not fail, so try to use reserved blocks for that.*/
575 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
576 /* Don't normalize allocation size (used for fallocate) */
577 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
578 /* Request will not result in inode size update (user for fallocate) */
579 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080
580 /* Convert written extents to unwritten */
581 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100
582 /* Write zeros to newly created written extents */
583 #define EXT4_GET_BLOCKS_ZERO 0x0200
584 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\
585 EXT4_GET_BLOCKS_ZERO)
586 /* Caller will submit data before dropping transaction handle. This
587 * allows jbd2 to avoid submitting data before commit. */
588 #define EXT4_GET_BLOCKS_IO_SUBMIT 0x0400
589
590 /*
591 * The bit position of these flags must not overlap with any of the
592 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(),
593 * read_extent_tree_block(), ext4_split_extent_at(),
594 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
595 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
596 * caching the extents when reading from the extent tree while a
597 * truncate or punch hole operation is in progress.
598 */
599 #define EXT4_EX_NOCACHE 0x40000000
600 #define EXT4_EX_FORCE_CACHE 0x20000000
601
602 /*
603 * Flags used by ext4_free_blocks
604 */
605 #define EXT4_FREE_BLOCKS_METADATA 0x0001
606 #define EXT4_FREE_BLOCKS_FORGET 0x0002
607 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
608 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
609 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
610 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
611
612 /*
613 * ioctl commands
614 */
615 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
616 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
617 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
618 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
619 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
620 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
621 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
622 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
623 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
624 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
625 #define EXT4_IOC_MIGRATE _IO('f', 9)
626 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
627 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
628 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
629 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
630 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
631 #define EXT4_IOC_SWAP_BOOT _IO('f', 17)
632 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18)
633 #define EXT4_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY
634 #define EXT4_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT
635 #define EXT4_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY
636
637 #ifndef FS_IOC_FSGETXATTR
638 /* Until the uapi changes get merged for project quota... */
639
640 #define FS_IOC_FSGETXATTR _IOR('X', 31, struct fsxattr)
641 #define FS_IOC_FSSETXATTR _IOW('X', 32, struct fsxattr)
642
643 /*
644 * Structure for FS_IOC_FSGETXATTR and FS_IOC_FSSETXATTR.
645 */
646 struct fsxattr {
647 __u32 fsx_xflags; /* xflags field value (get/set) */
648 __u32 fsx_extsize; /* extsize field value (get/set)*/
649 __u32 fsx_nextents; /* nextents field value (get) */
650 __u32 fsx_projid; /* project identifier (get/set) */
651 unsigned char fsx_pad[12];
652 };
653
654 /*
655 * Flags for the fsx_xflags field
656 */
657 #define FS_XFLAG_REALTIME 0x00000001 /* data in realtime volume */
658 #define FS_XFLAG_PREALLOC 0x00000002 /* preallocated file extents */
659 #define FS_XFLAG_IMMUTABLE 0x00000008 /* file cannot be modified */
660 #define FS_XFLAG_APPEND 0x00000010 /* all writes append */
661 #define FS_XFLAG_SYNC 0x00000020 /* all writes synchronous */
662 #define FS_XFLAG_NOATIME 0x00000040 /* do not update access time */
663 #define FS_XFLAG_NODUMP 0x00000080 /* do not include in backups */
664 #define FS_XFLAG_RTINHERIT 0x00000100 /* create with rt bit set */
665 #define FS_XFLAG_PROJINHERIT 0x00000200 /* create with parents projid */
666 #define FS_XFLAG_NOSYMLINKS 0x00000400 /* disallow symlink creation */
667 #define FS_XFLAG_EXTSIZE 0x00000800 /* extent size allocator hint */
668 #define FS_XFLAG_EXTSZINHERIT 0x00001000 /* inherit inode extent size */
669 #define FS_XFLAG_NODEFRAG 0x00002000 /* do not defragment */
670 #define FS_XFLAG_FILESTREAM 0x00004000 /* use filestream allocator */
671 #define FS_XFLAG_HASATTR 0x80000000 /* no DIFLAG for this */
672 #endif /* !defined(FS_IOC_FSGETXATTR) */
673
674 #define EXT4_IOC_FSGETXATTR FS_IOC_FSGETXATTR
675 #define EXT4_IOC_FSSETXATTR FS_IOC_FSSETXATTR
676
677 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
678 /*
679 * ioctl commands in 32 bit emulation
680 */
681 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
682 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
683 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
684 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
685 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
686 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
687 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
688 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
689 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
690 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
691 #endif
692
693 /* Max physical block we can address w/o extents */
694 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
695
696 /*
697 * Structure of an inode on the disk
698 */
699 struct ext4_inode {
700 __le16 i_mode; /* File mode */
701 __le16 i_uid; /* Low 16 bits of Owner Uid */
702 __le32 i_size_lo; /* Size in bytes */
703 __le32 i_atime; /* Access time */
704 __le32 i_ctime; /* Inode Change time */
705 __le32 i_mtime; /* Modification time */
706 __le32 i_dtime; /* Deletion Time */
707 __le16 i_gid; /* Low 16 bits of Group Id */
708 __le16 i_links_count; /* Links count */
709 __le32 i_blocks_lo; /* Blocks count */
710 __le32 i_flags; /* File flags */
711 union {
712 struct {
713 __le32 l_i_version;
714 } linux1;
715 struct {
716 __u32 h_i_translator;
717 } hurd1;
718 struct {
719 __u32 m_i_reserved1;
720 } masix1;
721 } osd1; /* OS dependent 1 */
722 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
723 __le32 i_generation; /* File version (for NFS) */
724 __le32 i_file_acl_lo; /* File ACL */
725 __le32 i_size_high;
726 __le32 i_obso_faddr; /* Obsoleted fragment address */
727 union {
728 struct {
729 __le16 l_i_blocks_high; /* were l_i_reserved1 */
730 __le16 l_i_file_acl_high;
731 __le16 l_i_uid_high; /* these 2 fields */
732 __le16 l_i_gid_high; /* were reserved2[0] */
733 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
734 __le16 l_i_reserved;
735 } linux2;
736 struct {
737 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
738 __u16 h_i_mode_high;
739 __u16 h_i_uid_high;
740 __u16 h_i_gid_high;
741 __u32 h_i_author;
742 } hurd2;
743 struct {
744 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
745 __le16 m_i_file_acl_high;
746 __u32 m_i_reserved2[2];
747 } masix2;
748 } osd2; /* OS dependent 2 */
749 __le16 i_extra_isize;
750 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
751 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
752 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
753 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
754 __le32 i_crtime; /* File Creation time */
755 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
756 __le32 i_version_hi; /* high 32 bits for 64-bit version */
757 __le32 i_projid; /* Project ID */
758 };
759
760 struct move_extent {
761 __u32 reserved; /* should be zero */
762 __u32 donor_fd; /* donor file descriptor */
763 __u64 orig_start; /* logical start offset in block for orig */
764 __u64 donor_start; /* logical start offset in block for donor */
765 __u64 len; /* block length to be moved */
766 __u64 moved_len; /* moved block length */
767 };
768
769 #define EXT4_EPOCH_BITS 2
770 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
771 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
772
773 /*
774 * Extended fields will fit into an inode if the filesystem was formatted
775 * with large inodes (-I 256 or larger) and there are not currently any EAs
776 * consuming all of the available space. For new inodes we always reserve
777 * enough space for the kernel's known extended fields, but for inodes
778 * created with an old kernel this might not have been the case. None of
779 * the extended inode fields is critical for correct filesystem operation.
780 * This macro checks if a certain field fits in the inode. Note that
781 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
782 */
783 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
784 ((offsetof(typeof(*ext4_inode), field) + \
785 sizeof((ext4_inode)->field)) \
786 <= (EXT4_GOOD_OLD_INODE_SIZE + \
787 (einode)->i_extra_isize)) \
788
789 /*
790 * We use an encoding that preserves the times for extra epoch "00":
791 *
792 * extra msb of adjust for signed
793 * epoch 32-bit 32-bit tv_sec to
794 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
795 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
796 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
797 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
798 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
799 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
800 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
801 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
802 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
803 *
804 * Note that previous versions of the kernel on 64-bit systems would
805 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
806 * 1970. e2fsck will correct this, assuming that it is run on the
807 * affected filesystem before 2242.
808 */
809
810 static inline __le32 ext4_encode_extra_time(struct timespec *time)
811 {
812 u32 extra = sizeof(time->tv_sec) > 4 ?
813 ((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK : 0;
814 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
815 }
816
817 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
818 {
819 if (unlikely(sizeof(time->tv_sec) > 4 &&
820 (extra & cpu_to_le32(EXT4_EPOCH_MASK)))) {
821 #if LINUX_VERSION_CODE < KERNEL_VERSION(4,20,0)
822 /* Handle legacy encoding of pre-1970 dates with epoch
823 * bits 1,1. We assume that by kernel version 4.20,
824 * everyone will have run fsck over the affected
825 * filesystems to correct the problem. (This
826 * backwards compatibility may be removed before this
827 * time, at the discretion of the ext4 developers.)
828 */
829 u64 extra_bits = le32_to_cpu(extra) & EXT4_EPOCH_MASK;
830 if (extra_bits == 3 && ((time->tv_sec) & 0x80000000) != 0)
831 extra_bits = 0;
832 time->tv_sec += extra_bits << 32;
833 #else
834 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
835 #endif
836 }
837 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
838 }
839
840 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
841 do { \
842 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
843 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
844 (raw_inode)->xtime ## _extra = \
845 ext4_encode_extra_time(&(inode)->xtime); \
846 } while (0)
847
848 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
849 do { \
850 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
851 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
852 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
853 (raw_inode)->xtime ## _extra = \
854 ext4_encode_extra_time(&(einode)->xtime); \
855 } while (0)
856
857 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
858 do { \
859 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
860 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
861 ext4_decode_extra_time(&(inode)->xtime, \
862 raw_inode->xtime ## _extra); \
863 else \
864 (inode)->xtime.tv_nsec = 0; \
865 } while (0)
866
867 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
868 do { \
869 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
870 (einode)->xtime.tv_sec = \
871 (signed)le32_to_cpu((raw_inode)->xtime); \
872 else \
873 (einode)->xtime.tv_sec = 0; \
874 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
875 ext4_decode_extra_time(&(einode)->xtime, \
876 raw_inode->xtime ## _extra); \
877 else \
878 (einode)->xtime.tv_nsec = 0; \
879 } while (0)
880
881 #define i_disk_version osd1.linux1.l_i_version
882
883 #if defined(__KERNEL__) || defined(__linux__)
884 #define i_reserved1 osd1.linux1.l_i_reserved1
885 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
886 #define i_blocks_high osd2.linux2.l_i_blocks_high
887 #define i_uid_low i_uid
888 #define i_gid_low i_gid
889 #define i_uid_high osd2.linux2.l_i_uid_high
890 #define i_gid_high osd2.linux2.l_i_gid_high
891 #define i_checksum_lo osd2.linux2.l_i_checksum_lo
892
893 #elif defined(__GNU__)
894
895 #define i_translator osd1.hurd1.h_i_translator
896 #define i_uid_high osd2.hurd2.h_i_uid_high
897 #define i_gid_high osd2.hurd2.h_i_gid_high
898 #define i_author osd2.hurd2.h_i_author
899
900 #elif defined(__masix__)
901
902 #define i_reserved1 osd1.masix1.m_i_reserved1
903 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
904 #define i_reserved2 osd2.masix2.m_i_reserved2
905
906 #endif /* defined(__KERNEL__) || defined(__linux__) */
907
908 #include "extents_status.h"
909
910 /*
911 * Lock subclasses for i_data_sem in the ext4_inode_info structure.
912 *
913 * These are needed to avoid lockdep false positives when we need to
914 * allocate blocks to the quota inode during ext4_map_blocks(), while
915 * holding i_data_sem for a normal (non-quota) inode. Since we don't
916 * do quota tracking for the quota inode, this avoids deadlock (as
917 * well as infinite recursion, since it isn't turtles all the way
918 * down...)
919 *
920 * I_DATA_SEM_NORMAL - Used for most inodes
921 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode
922 * where the second inode has larger inode number
923 * than the first
924 * I_DATA_SEM_QUOTA - Used for quota inodes only
925 */
926 enum {
927 I_DATA_SEM_NORMAL = 0,
928 I_DATA_SEM_OTHER,
929 I_DATA_SEM_QUOTA,
930 };
931
932
933 /*
934 * fourth extended file system inode data in memory
935 */
936 struct ext4_inode_info {
937 __le32 i_data[15]; /* unconverted */
938 __u32 i_dtime;
939 ext4_fsblk_t i_file_acl;
940
941 /*
942 * i_block_group is the number of the block group which contains
943 * this file's inode. Constant across the lifetime of the inode,
944 * it is ued for making block allocation decisions - we try to
945 * place a file's data blocks near its inode block, and new inodes
946 * near to their parent directory's inode.
947 */
948 ext4_group_t i_block_group;
949 ext4_lblk_t i_dir_start_lookup;
950 #if (BITS_PER_LONG < 64)
951 unsigned long i_state_flags; /* Dynamic state flags */
952 #endif
953 unsigned long i_flags;
954
955 /*
956 * Extended attributes can be read independently of the main file
957 * data. Taking i_mutex even when reading would cause contention
958 * between readers of EAs and writers of regular file data, so
959 * instead we synchronize on xattr_sem when reading or changing
960 * EAs.
961 */
962 struct rw_semaphore xattr_sem;
963
964 struct list_head i_orphan; /* unlinked but open inodes */
965
966 /*
967 * i_disksize keeps track of what the inode size is ON DISK, not
968 * in memory. During truncate, i_size is set to the new size by
969 * the VFS prior to calling ext4_truncate(), but the filesystem won't
970 * set i_disksize to 0 until the truncate is actually under way.
971 *
972 * The intent is that i_disksize always represents the blocks which
973 * are used by this file. This allows recovery to restart truncate
974 * on orphans if we crash during truncate. We actually write i_disksize
975 * into the on-disk inode when writing inodes out, instead of i_size.
976 *
977 * The only time when i_disksize and i_size may be different is when
978 * a truncate is in progress. The only things which change i_disksize
979 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
980 */
981 loff_t i_disksize;
982
983 /*
984 * i_data_sem is for serialising ext4_truncate() against
985 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
986 * data tree are chopped off during truncate. We can't do that in
987 * ext4 because whenever we perform intermediate commits during
988 * truncate, the inode and all the metadata blocks *must* be in a
989 * consistent state which allows truncation of the orphans to restart
990 * during recovery. Hence we must fix the get_block-vs-truncate race
991 * by other means, so we have i_data_sem.
992 */
993 struct rw_semaphore i_data_sem;
994 /*
995 * i_mmap_sem is for serializing page faults with truncate / punch hole
996 * operations. We have to make sure that new page cannot be faulted in
997 * a section of the inode that is being punched. We cannot easily use
998 * i_data_sem for this since we need protection for the whole punch
999 * operation and i_data_sem ranks below transaction start so we have
1000 * to occasionally drop it.
1001 */
1002 struct rw_semaphore i_mmap_sem;
1003 struct inode vfs_inode;
1004 struct jbd2_inode *jinode;
1005
1006 spinlock_t i_raw_lock; /* protects updates to the raw inode */
1007
1008 /*
1009 * File creation time. Its function is same as that of
1010 * struct timespec i_{a,c,m}time in the generic inode.
1011 */
1012 struct timespec i_crtime;
1013
1014 /* mballoc */
1015 struct list_head i_prealloc_list;
1016 spinlock_t i_prealloc_lock;
1017
1018 /* extents status tree */
1019 struct ext4_es_tree i_es_tree;
1020 rwlock_t i_es_lock;
1021 struct list_head i_es_list;
1022 unsigned int i_es_all_nr; /* protected by i_es_lock */
1023 unsigned int i_es_shk_nr; /* protected by i_es_lock */
1024 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for
1025 extents to shrink. Protected by
1026 i_es_lock */
1027
1028 /* ialloc */
1029 ext4_group_t i_last_alloc_group;
1030
1031 /* allocation reservation info for delalloc */
1032 /* In case of bigalloc, these refer to clusters rather than blocks */
1033 unsigned int i_reserved_data_blocks;
1034 unsigned int i_reserved_meta_blocks;
1035 unsigned int i_allocated_meta_blocks;
1036 ext4_lblk_t i_da_metadata_calc_last_lblock;
1037 int i_da_metadata_calc_len;
1038
1039 /* on-disk additional length */
1040 __u16 i_extra_isize;
1041
1042 /* Indicate the inline data space. */
1043 u16 i_inline_off;
1044 u16 i_inline_size;
1045
1046 #ifdef CONFIG_QUOTA
1047 /* quota space reservation, managed internally by quota code */
1048 qsize_t i_reserved_quota;
1049 #endif
1050
1051 /* Lock protecting lists below */
1052 spinlock_t i_completed_io_lock;
1053 /*
1054 * Completed IOs that need unwritten extents handling and have
1055 * transaction reserved
1056 */
1057 struct list_head i_rsv_conversion_list;
1058 struct work_struct i_rsv_conversion_work;
1059 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1060
1061 spinlock_t i_block_reservation_lock;
1062
1063 /*
1064 * Transactions that contain inode's metadata needed to complete
1065 * fsync and fdatasync, respectively.
1066 */
1067 tid_t i_sync_tid;
1068 tid_t i_datasync_tid;
1069
1070 #ifdef CONFIG_QUOTA
1071 struct dquot *i_dquot[MAXQUOTAS];
1072 #endif
1073
1074 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1075 __u32 i_csum_seed;
1076
1077 kprojid_t i_projid;
1078 };
1079
1080 /*
1081 * File system states
1082 */
1083 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
1084 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
1085 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
1086
1087 /*
1088 * Misc. filesystem flags
1089 */
1090 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
1091 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
1092 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
1093
1094 /*
1095 * Mount flags set via mount options or defaults
1096 */
1097 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
1098 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
1099 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
1100 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
1101 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
1102 #define EXT4_MOUNT_ERRORS_MASK 0x00070
1103 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
1104 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
1105 #ifdef CONFIG_FS_DAX
1106 #define EXT4_MOUNT_DAX 0x00200 /* Direct Access */
1107 #else
1108 #define EXT4_MOUNT_DAX 0
1109 #endif
1110 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
1111 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
1112 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
1113 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
1114 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
1115 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
1116 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
1117 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
1118 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
1119 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
1120 #define EXT4_MOUNT_QUOTA 0x40000 /* Some quota option set */
1121 #define EXT4_MOUNT_USRQUOTA 0x80000 /* "old" user quota,
1122 * enable enforcement for hidden
1123 * quota files */
1124 #define EXT4_MOUNT_GRPQUOTA 0x100000 /* "old" group quota, enable
1125 * enforcement for hidden quota
1126 * files */
1127 #define EXT4_MOUNT_PRJQUOTA 0x200000 /* Enable project quota
1128 * enforcement */
1129 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
1130 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
1131 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1132 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
1133 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
1134 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
1135 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
1136 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
1137
1138 /*
1139 * Mount flags set either automatically (could not be set by mount option)
1140 * based on per file system feature or property or in special cases such as
1141 * distinguishing between explicit mount option definition and default.
1142 */
1143 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1144 specified delalloc */
1145 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1146 size of blocksize * 8
1147 blocks */
1148 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1149 file systems */
1150
1151 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly
1152 specified journal checksum */
1153
1154 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1155 ~EXT4_MOUNT_##opt
1156 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1157 EXT4_MOUNT_##opt
1158 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1159 EXT4_MOUNT_##opt)
1160
1161 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1162 ~EXT4_MOUNT2_##opt
1163 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1164 EXT4_MOUNT2_##opt
1165 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1166 EXT4_MOUNT2_##opt)
1167
1168 #define ext4_test_and_set_bit __test_and_set_bit_le
1169 #define ext4_set_bit __set_bit_le
1170 #define ext4_set_bit_atomic ext2_set_bit_atomic
1171 #define ext4_test_and_clear_bit __test_and_clear_bit_le
1172 #define ext4_clear_bit __clear_bit_le
1173 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
1174 #define ext4_test_bit test_bit_le
1175 #define ext4_find_next_zero_bit find_next_zero_bit_le
1176 #define ext4_find_next_bit find_next_bit_le
1177
1178 extern void ext4_set_bits(void *bm, int cur, int len);
1179
1180 /*
1181 * Maximal mount counts between two filesystem checks
1182 */
1183 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1184 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1185
1186 /*
1187 * Behaviour when detecting errors
1188 */
1189 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1190 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1191 #define EXT4_ERRORS_PANIC 3 /* Panic */
1192 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1193
1194 /* Metadata checksum algorithm codes */
1195 #define EXT4_CRC32C_CHKSUM 1
1196
1197 /*
1198 * Structure of the super block
1199 */
1200 struct ext4_super_block {
1201 /*00*/ __le32 s_inodes_count; /* Inodes count */
1202 __le32 s_blocks_count_lo; /* Blocks count */
1203 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1204 __le32 s_free_blocks_count_lo; /* Free blocks count */
1205 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1206 __le32 s_first_data_block; /* First Data Block */
1207 __le32 s_log_block_size; /* Block size */
1208 __le32 s_log_cluster_size; /* Allocation cluster size */
1209 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1210 __le32 s_clusters_per_group; /* # Clusters per group */
1211 __le32 s_inodes_per_group; /* # Inodes per group */
1212 __le32 s_mtime; /* Mount time */
1213 /*30*/ __le32 s_wtime; /* Write time */
1214 __le16 s_mnt_count; /* Mount count */
1215 __le16 s_max_mnt_count; /* Maximal mount count */
1216 __le16 s_magic; /* Magic signature */
1217 __le16 s_state; /* File system state */
1218 __le16 s_errors; /* Behaviour when detecting errors */
1219 __le16 s_minor_rev_level; /* minor revision level */
1220 /*40*/ __le32 s_lastcheck; /* time of last check */
1221 __le32 s_checkinterval; /* max. time between checks */
1222 __le32 s_creator_os; /* OS */
1223 __le32 s_rev_level; /* Revision level */
1224 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1225 __le16 s_def_resgid; /* Default gid for reserved blocks */
1226 /*
1227 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1228 *
1229 * Note: the difference between the compatible feature set and
1230 * the incompatible feature set is that if there is a bit set
1231 * in the incompatible feature set that the kernel doesn't
1232 * know about, it should refuse to mount the filesystem.
1233 *
1234 * e2fsck's requirements are more strict; if it doesn't know
1235 * about a feature in either the compatible or incompatible
1236 * feature set, it must abort and not try to meddle with
1237 * things it doesn't understand...
1238 */
1239 __le32 s_first_ino; /* First non-reserved inode */
1240 __le16 s_inode_size; /* size of inode structure */
1241 __le16 s_block_group_nr; /* block group # of this superblock */
1242 __le32 s_feature_compat; /* compatible feature set */
1243 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1244 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1245 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1246 /*78*/ char s_volume_name[16]; /* volume name */
1247 /*88*/ char s_last_mounted[64]; /* directory where last mounted */
1248 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1249 /*
1250 * Performance hints. Directory preallocation should only
1251 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1252 */
1253 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1254 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1255 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1256 /*
1257 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1258 */
1259 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1260 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1261 __le32 s_journal_dev; /* device number of journal file */
1262 __le32 s_last_orphan; /* start of list of inodes to delete */
1263 __le32 s_hash_seed[4]; /* HTREE hash seed */
1264 __u8 s_def_hash_version; /* Default hash version to use */
1265 __u8 s_jnl_backup_type;
1266 __le16 s_desc_size; /* size of group descriptor */
1267 /*100*/ __le32 s_default_mount_opts;
1268 __le32 s_first_meta_bg; /* First metablock block group */
1269 __le32 s_mkfs_time; /* When the filesystem was created */
1270 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1271 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1272 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1273 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1274 __le32 s_free_blocks_count_hi; /* Free blocks count */
1275 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1276 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1277 __le32 s_flags; /* Miscellaneous flags */
1278 __le16 s_raid_stride; /* RAID stride */
1279 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1280 __le64 s_mmp_block; /* Block for multi-mount protection */
1281 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1282 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1283 __u8 s_checksum_type; /* metadata checksum algorithm used */
1284 __u8 s_encryption_level; /* versioning level for encryption */
1285 __u8 s_reserved_pad; /* Padding to next 32bits */
1286 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1287 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1288 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1289 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1290 snapshot's future use */
1291 __le32 s_snapshot_list; /* inode number of the head of the
1292 on-disk snapshot list */
1293 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1294 __le32 s_error_count; /* number of fs errors */
1295 __le32 s_first_error_time; /* first time an error happened */
1296 __le32 s_first_error_ino; /* inode involved in first error */
1297 __le64 s_first_error_block; /* block involved of first error */
1298 __u8 s_first_error_func[32]; /* function where the error happened */
1299 __le32 s_first_error_line; /* line number where error happened */
1300 __le32 s_last_error_time; /* most recent time of an error */
1301 __le32 s_last_error_ino; /* inode involved in last error */
1302 __le32 s_last_error_line; /* line number where error happened */
1303 __le64 s_last_error_block; /* block involved of last error */
1304 __u8 s_last_error_func[32]; /* function where the error happened */
1305 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1306 __u8 s_mount_opts[64];
1307 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1308 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1309 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1310 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
1311 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */
1312 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */
1313 __le32 s_lpf_ino; /* Location of the lost+found inode */
1314 __le32 s_prj_quota_inum; /* inode for tracking project quota */
1315 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */
1316 __le32 s_reserved[98]; /* Padding to the end of the block */
1317 __le32 s_checksum; /* crc32c(superblock) */
1318 };
1319
1320 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1321
1322 #ifdef __KERNEL__
1323
1324 /*
1325 * run-time mount flags
1326 */
1327 #define EXT4_MF_MNTDIR_SAMPLED 0x0001
1328 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1329 #define EXT4_MF_TEST_DUMMY_ENCRYPTION 0x0004
1330
1331 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1332 #define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \
1333 EXT4_MF_TEST_DUMMY_ENCRYPTION))
1334 #else
1335 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1336 #endif
1337
1338 /* Number of quota types we support */
1339 #define EXT4_MAXQUOTAS 3
1340
1341 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1342 #define EXT4_KEY_DESC_PREFIX "ext4:"
1343 #define EXT4_KEY_DESC_PREFIX_SIZE 5
1344 #endif
1345
1346 /*
1347 * fourth extended-fs super-block data in memory
1348 */
1349 struct ext4_sb_info {
1350 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1351 unsigned long s_inodes_per_block;/* Number of inodes per block */
1352 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1353 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1354 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1355 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1356 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1357 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1358 ext4_group_t s_groups_count; /* Number of groups in the fs */
1359 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1360 unsigned long s_overhead; /* # of fs overhead clusters */
1361 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1362 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1363 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1364 struct buffer_head * s_sbh; /* Buffer containing the super block */
1365 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1366 struct buffer_head **s_group_desc;
1367 unsigned int s_mount_opt;
1368 unsigned int s_mount_opt2;
1369 unsigned int s_mount_flags;
1370 unsigned int s_def_mount_opt;
1371 ext4_fsblk_t s_sb_block;
1372 atomic64_t s_resv_clusters;
1373 kuid_t s_resuid;
1374 kgid_t s_resgid;
1375 unsigned short s_mount_state;
1376 unsigned short s_pad;
1377 int s_addr_per_block_bits;
1378 int s_desc_per_block_bits;
1379 int s_inode_size;
1380 int s_first_ino;
1381 unsigned int s_inode_readahead_blks;
1382 unsigned int s_inode_goal;
1383 spinlock_t s_next_gen_lock;
1384 u32 s_next_generation;
1385 u32 s_hash_seed[4];
1386 int s_def_hash_version;
1387 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1388 struct percpu_counter s_freeclusters_counter;
1389 struct percpu_counter s_freeinodes_counter;
1390 struct percpu_counter s_dirs_counter;
1391 struct percpu_counter s_dirtyclusters_counter;
1392 struct blockgroup_lock *s_blockgroup_lock;
1393 struct proc_dir_entry *s_proc;
1394 struct kobject s_kobj;
1395 struct completion s_kobj_unregister;
1396 struct super_block *s_sb;
1397
1398 /* Journaling */
1399 struct journal_s *s_journal;
1400 struct list_head s_orphan;
1401 struct mutex s_orphan_lock;
1402 unsigned long s_resize_flags; /* Flags indicating if there
1403 is a resizer */
1404 unsigned long s_commit_interval;
1405 u32 s_max_batch_time;
1406 u32 s_min_batch_time;
1407 struct block_device *journal_bdev;
1408 #ifdef CONFIG_QUOTA
1409 char *s_qf_names[EXT4_MAXQUOTAS]; /* Names of quota files with journalled quota */
1410 int s_jquota_fmt; /* Format of quota to use */
1411 #endif
1412 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1413 struct rb_root system_blks;
1414
1415 #ifdef EXTENTS_STATS
1416 /* ext4 extents stats */
1417 unsigned long s_ext_min;
1418 unsigned long s_ext_max;
1419 unsigned long s_depth_max;
1420 spinlock_t s_ext_stats_lock;
1421 unsigned long s_ext_blocks;
1422 unsigned long s_ext_extents;
1423 #endif
1424
1425 /* for buddy allocator */
1426 struct ext4_group_info ***s_group_info;
1427 struct inode *s_buddy_cache;
1428 spinlock_t s_md_lock;
1429 unsigned short *s_mb_offsets;
1430 unsigned int *s_mb_maxs;
1431 unsigned int s_group_info_size;
1432 unsigned int s_mb_free_pending;
1433
1434 /* tunables */
1435 unsigned long s_stripe;
1436 unsigned int s_mb_stream_request;
1437 unsigned int s_mb_max_to_scan;
1438 unsigned int s_mb_min_to_scan;
1439 unsigned int s_mb_stats;
1440 unsigned int s_mb_order2_reqs;
1441 unsigned int s_mb_group_prealloc;
1442 unsigned int s_max_dir_size_kb;
1443 /* where last allocation was done - for stream allocation */
1444 unsigned long s_mb_last_group;
1445 unsigned long s_mb_last_start;
1446
1447 /* stats for buddy allocator */
1448 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1449 atomic_t s_bal_success; /* we found long enough chunks */
1450 atomic_t s_bal_allocated; /* in blocks */
1451 atomic_t s_bal_ex_scanned; /* total extents scanned */
1452 atomic_t s_bal_goals; /* goal hits */
1453 atomic_t s_bal_breaks; /* too long searches */
1454 atomic_t s_bal_2orders; /* 2^order hits */
1455 spinlock_t s_bal_lock;
1456 unsigned long s_mb_buddies_generated;
1457 unsigned long long s_mb_generation_time;
1458 atomic_t s_mb_lost_chunks;
1459 atomic_t s_mb_preallocated;
1460 atomic_t s_mb_discarded;
1461 atomic_t s_lock_busy;
1462
1463 /* locality groups */
1464 struct ext4_locality_group __percpu *s_locality_groups;
1465
1466 /* for write statistics */
1467 unsigned long s_sectors_written_start;
1468 u64 s_kbytes_written;
1469
1470 /* the size of zero-out chunk */
1471 unsigned int s_extent_max_zeroout_kb;
1472
1473 unsigned int s_log_groups_per_flex;
1474 struct flex_groups *s_flex_groups;
1475 ext4_group_t s_flex_groups_allocated;
1476
1477 /* workqueue for reserved extent conversions (buffered io) */
1478 struct workqueue_struct *rsv_conversion_wq;
1479
1480 /* timer for periodic error stats printing */
1481 struct timer_list s_err_report;
1482
1483 /* Lazy inode table initialization info */
1484 struct ext4_li_request *s_li_request;
1485 /* Wait multiplier for lazy initialization thread */
1486 unsigned int s_li_wait_mult;
1487
1488 /* Kernel thread for multiple mount protection */
1489 struct task_struct *s_mmp_tsk;
1490
1491 /* record the last minlen when FITRIM is called. */
1492 atomic_t s_last_trim_minblks;
1493
1494 /* Reference to checksum algorithm driver via cryptoapi */
1495 struct crypto_shash *s_chksum_driver;
1496
1497 /* Precomputed FS UUID checksum for seeding other checksums */
1498 __u32 s_csum_seed;
1499
1500 /* Reclaim extents from extent status tree */
1501 struct shrinker s_es_shrinker;
1502 struct list_head s_es_list; /* List of inodes with reclaimable extents */
1503 long s_es_nr_inode;
1504 struct ext4_es_stats s_es_stats;
1505 struct mb_cache *s_mb_cache;
1506 spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1507
1508 /* Ratelimit ext4 messages. */
1509 struct ratelimit_state s_err_ratelimit_state;
1510 struct ratelimit_state s_warning_ratelimit_state;
1511 struct ratelimit_state s_msg_ratelimit_state;
1512
1513 /* Barrier between changing inodes' journal flags and writepages ops. */
1514 struct percpu_rw_semaphore s_journal_flag_rwsem;
1515
1516 /* Encryption support */
1517 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1518 u8 key_prefix[EXT4_KEY_DESC_PREFIX_SIZE];
1519 u8 key_prefix_size;
1520 #endif
1521 };
1522
1523 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1524 {
1525 return sb->s_fs_info;
1526 }
1527 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1528 {
1529 return container_of(inode, struct ext4_inode_info, vfs_inode);
1530 }
1531
1532 static inline struct timespec ext4_current_time(struct inode *inode)
1533 {
1534 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1535 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1536 }
1537
1538 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1539 {
1540 return ino == EXT4_ROOT_INO ||
1541 ino == EXT4_USR_QUOTA_INO ||
1542 ino == EXT4_GRP_QUOTA_INO ||
1543 ino == EXT4_BOOT_LOADER_INO ||
1544 ino == EXT4_JOURNAL_INO ||
1545 ino == EXT4_RESIZE_INO ||
1546 (ino >= EXT4_FIRST_INO(sb) &&
1547 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1548 }
1549
1550 /*
1551 * Inode dynamic state flags
1552 */
1553 enum {
1554 EXT4_STATE_JDATA, /* journaled data exists */
1555 EXT4_STATE_NEW, /* inode is newly created */
1556 EXT4_STATE_XATTR, /* has in-inode xattrs */
1557 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1558 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1559 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1560 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
1561 EXT4_STATE_NEWENTRY, /* File just added to dir */
1562 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read
1563 nolocking */
1564 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1565 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
1566 };
1567
1568 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1569 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1570 { \
1571 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1572 } \
1573 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1574 { \
1575 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1576 } \
1577 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1578 { \
1579 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1580 }
1581
1582 /* Add these declarations here only so that these functions can be
1583 * found by name. Otherwise, they are very hard to locate. */
1584 static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1585 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1586 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1587 EXT4_INODE_BIT_FNS(flag, flags, 0)
1588
1589 /* Add these declarations here only so that these functions can be
1590 * found by name. Otherwise, they are very hard to locate. */
1591 static inline int ext4_test_inode_state(struct inode *inode, int bit);
1592 static inline void ext4_set_inode_state(struct inode *inode, int bit);
1593 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1594 #if (BITS_PER_LONG < 64)
1595 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1596
1597 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1598 {
1599 (ei)->i_state_flags = 0;
1600 }
1601 #else
1602 EXT4_INODE_BIT_FNS(state, flags, 32)
1603
1604 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1605 {
1606 /* We depend on the fact that callers will set i_flags */
1607 }
1608 #endif
1609 #else
1610 /* Assume that user mode programs are passing in an ext4fs superblock, not
1611 * a kernel struct super_block. This will allow us to call the feature-test
1612 * macros from user land. */
1613 #define EXT4_SB(sb) (sb)
1614 #endif
1615
1616 /*
1617 * Returns true if the inode is inode is encrypted
1618 */
1619 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1620
1621 /*
1622 * Codes for operating systems
1623 */
1624 #define EXT4_OS_LINUX 0
1625 #define EXT4_OS_HURD 1
1626 #define EXT4_OS_MASIX 2
1627 #define EXT4_OS_FREEBSD 3
1628 #define EXT4_OS_LITES 4
1629
1630 /*
1631 * Revision levels
1632 */
1633 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1634 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1635
1636 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1637 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1638
1639 #define EXT4_GOOD_OLD_INODE_SIZE 128
1640
1641 /*
1642 * Feature set definitions
1643 */
1644
1645 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1646 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1647 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1648 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1649 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1650 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1651 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
1652
1653 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1654 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1655 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1656 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1657 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1658 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1659 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1660 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1661 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1662 /*
1663 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1664 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1665 * all other data structures' checksums. However, the METADATA_CSUM and
1666 * GDT_CSUM bits are mutually exclusive.
1667 */
1668 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
1669 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000
1670 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000
1671
1672 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1673 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1674 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1675 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1676 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1677 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1678 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1679 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1680 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1681 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1682 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1683 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000
1684 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
1685 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
1686 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000
1687
1688 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
1689 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1690 { \
1691 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1692 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
1693 } \
1694 static inline void ext4_set_feature_##name(struct super_block *sb) \
1695 { \
1696 EXT4_SB(sb)->s_es->s_feature_compat |= \
1697 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1698 } \
1699 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1700 { \
1701 EXT4_SB(sb)->s_es->s_feature_compat &= \
1702 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1703 }
1704
1705 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
1706 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1707 { \
1708 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1709 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
1710 } \
1711 static inline void ext4_set_feature_##name(struct super_block *sb) \
1712 { \
1713 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
1714 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1715 } \
1716 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1717 { \
1718 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
1719 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1720 }
1721
1722 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
1723 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1724 { \
1725 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1726 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
1727 } \
1728 static inline void ext4_set_feature_##name(struct super_block *sb) \
1729 { \
1730 EXT4_SB(sb)->s_es->s_feature_incompat |= \
1731 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1732 } \
1733 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1734 { \
1735 EXT4_SB(sb)->s_es->s_feature_incompat &= \
1736 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1737 }
1738
1739 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC)
1740 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES)
1741 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL)
1742 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR)
1743 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE)
1744 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX)
1745 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2)
1746
1747 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER)
1748 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE)
1749 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR)
1750 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE)
1751 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM)
1752 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK)
1753 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE)
1754 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA)
1755 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC)
1756 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM)
1757 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY)
1758 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT)
1759
1760 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION)
1761 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE)
1762 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER)
1763 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV)
1764 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG)
1765 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS)
1766 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT)
1767 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP)
1768 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG)
1769 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE)
1770 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA)
1771 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED)
1772 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR)
1773 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA)
1774 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT)
1775
1776 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1777 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1778 EXT4_FEATURE_INCOMPAT_META_BG)
1779 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1780 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1781 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1782
1783 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1784 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1785 EXT4_FEATURE_INCOMPAT_RECOVER| \
1786 EXT4_FEATURE_INCOMPAT_META_BG)
1787 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1788 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1789 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1790
1791 #define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1792 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1793 EXT4_FEATURE_INCOMPAT_RECOVER| \
1794 EXT4_FEATURE_INCOMPAT_META_BG| \
1795 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1796 EXT4_FEATURE_INCOMPAT_64BIT| \
1797 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1798 EXT4_FEATURE_INCOMPAT_MMP | \
1799 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
1800 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
1801 EXT4_FEATURE_INCOMPAT_CSUM_SEED)
1802 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1803 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1804 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1805 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1806 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1807 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1808 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1809 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1810 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1811 EXT4_FEATURE_RO_COMPAT_QUOTA |\
1812 EXT4_FEATURE_RO_COMPAT_PROJECT)
1813
1814 #define EXTN_FEATURE_FUNCS(ver) \
1815 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
1816 { \
1817 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1818 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
1819 } \
1820 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
1821 { \
1822 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1823 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
1824 } \
1825 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
1826 { \
1827 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1828 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
1829 }
1830
1831 EXTN_FEATURE_FUNCS(2)
1832 EXTN_FEATURE_FUNCS(3)
1833 EXTN_FEATURE_FUNCS(4)
1834
1835 static inline bool ext4_has_compat_features(struct super_block *sb)
1836 {
1837 return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
1838 }
1839 static inline bool ext4_has_ro_compat_features(struct super_block *sb)
1840 {
1841 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
1842 }
1843 static inline bool ext4_has_incompat_features(struct super_block *sb)
1844 {
1845 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
1846 }
1847
1848 /*
1849 * Default values for user and/or group using reserved blocks
1850 */
1851 #define EXT4_DEF_RESUID 0
1852 #define EXT4_DEF_RESGID 0
1853
1854 /*
1855 * Default project ID
1856 */
1857 #define EXT4_DEF_PROJID 0
1858
1859 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
1860
1861 /*
1862 * Default mount options
1863 */
1864 #define EXT4_DEFM_DEBUG 0x0001
1865 #define EXT4_DEFM_BSDGROUPS 0x0002
1866 #define EXT4_DEFM_XATTR_USER 0x0004
1867 #define EXT4_DEFM_ACL 0x0008
1868 #define EXT4_DEFM_UID16 0x0010
1869 #define EXT4_DEFM_JMODE 0x0060
1870 #define EXT4_DEFM_JMODE_DATA 0x0020
1871 #define EXT4_DEFM_JMODE_ORDERED 0x0040
1872 #define EXT4_DEFM_JMODE_WBACK 0x0060
1873 #define EXT4_DEFM_NOBARRIER 0x0100
1874 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1875 #define EXT4_DEFM_DISCARD 0x0400
1876 #define EXT4_DEFM_NODELALLOC 0x0800
1877
1878 /*
1879 * Default journal batch times
1880 */
1881 #define EXT4_DEF_MIN_BATCH_TIME 0
1882 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1883
1884 /*
1885 * Minimum number of groups in a flexgroup before we separate out
1886 * directories into the first block group of a flexgroup
1887 */
1888 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1889
1890 /*
1891 * Structure of a directory entry
1892 */
1893 #define EXT4_NAME_LEN 255
1894
1895 struct ext4_dir_entry {
1896 __le32 inode; /* Inode number */
1897 __le16 rec_len; /* Directory entry length */
1898 __le16 name_len; /* Name length */
1899 char name[EXT4_NAME_LEN]; /* File name */
1900 };
1901
1902 /*
1903 * The new version of the directory entry. Since EXT4 structures are
1904 * stored in intel byte order, and the name_len field could never be
1905 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1906 * file_type field.
1907 */
1908 struct ext4_dir_entry_2 {
1909 __le32 inode; /* Inode number */
1910 __le16 rec_len; /* Directory entry length */
1911 __u8 name_len; /* Name length */
1912 __u8 file_type;
1913 char name[EXT4_NAME_LEN]; /* File name */
1914 };
1915
1916 /*
1917 * This is a bogus directory entry at the end of each leaf block that
1918 * records checksums.
1919 */
1920 struct ext4_dir_entry_tail {
1921 __le32 det_reserved_zero1; /* Pretend to be unused */
1922 __le16 det_rec_len; /* 12 */
1923 __u8 det_reserved_zero2; /* Zero name length */
1924 __u8 det_reserved_ft; /* 0xDE, fake file type */
1925 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
1926 };
1927
1928 #define EXT4_DIRENT_TAIL(block, blocksize) \
1929 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1930 ((blocksize) - \
1931 sizeof(struct ext4_dir_entry_tail))))
1932
1933 /*
1934 * Ext4 directory file types. Only the low 3 bits are used. The
1935 * other bits are reserved for now.
1936 */
1937 #define EXT4_FT_UNKNOWN 0
1938 #define EXT4_FT_REG_FILE 1
1939 #define EXT4_FT_DIR 2
1940 #define EXT4_FT_CHRDEV 3
1941 #define EXT4_FT_BLKDEV 4
1942 #define EXT4_FT_FIFO 5
1943 #define EXT4_FT_SOCK 6
1944 #define EXT4_FT_SYMLINK 7
1945
1946 #define EXT4_FT_MAX 8
1947
1948 #define EXT4_FT_DIR_CSUM 0xDE
1949
1950 /*
1951 * EXT4_DIR_PAD defines the directory entries boundaries
1952 *
1953 * NOTE: It must be a multiple of 4
1954 */
1955 #define EXT4_DIR_PAD 4
1956 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1957 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1958 ~EXT4_DIR_ROUND)
1959 #define EXT4_MAX_REC_LEN ((1<<16)-1)
1960
1961 /*
1962 * If we ever get support for fs block sizes > page_size, we'll need
1963 * to remove the #if statements in the next two functions...
1964 */
1965 static inline unsigned int
1966 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1967 {
1968 unsigned len = le16_to_cpu(dlen);
1969
1970 #if (PAGE_SIZE >= 65536)
1971 if (len == EXT4_MAX_REC_LEN || len == 0)
1972 return blocksize;
1973 return (len & 65532) | ((len & 3) << 16);
1974 #else
1975 return len;
1976 #endif
1977 }
1978
1979 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1980 {
1981 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1982 BUG();
1983 #if (PAGE_SIZE >= 65536)
1984 if (len < 65536)
1985 return cpu_to_le16(len);
1986 if (len == blocksize) {
1987 if (blocksize == 65536)
1988 return cpu_to_le16(EXT4_MAX_REC_LEN);
1989 else
1990 return cpu_to_le16(0);
1991 }
1992 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1993 #else
1994 return cpu_to_le16(len);
1995 #endif
1996 }
1997
1998 /*
1999 * Hash Tree Directory indexing
2000 * (c) Daniel Phillips, 2001
2001 */
2002
2003 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
2004 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
2005 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
2006 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
2007
2008 /* Legal values for the dx_root hash_version field: */
2009
2010 #define DX_HASH_LEGACY 0
2011 #define DX_HASH_HALF_MD4 1
2012 #define DX_HASH_TEA 2
2013 #define DX_HASH_LEGACY_UNSIGNED 3
2014 #define DX_HASH_HALF_MD4_UNSIGNED 4
2015 #define DX_HASH_TEA_UNSIGNED 5
2016
2017 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2018 const void *address, unsigned int length)
2019 {
2020 struct {
2021 struct shash_desc shash;
2022 char ctx[4];
2023 } desc;
2024 int err;
2025
2026 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2027
2028 desc.shash.tfm = sbi->s_chksum_driver;
2029 desc.shash.flags = 0;
2030 *(u32 *)desc.ctx = crc;
2031
2032 err = crypto_shash_update(&desc.shash, address, length);
2033 BUG_ON(err);
2034
2035 return *(u32 *)desc.ctx;
2036 }
2037
2038 #ifdef __KERNEL__
2039
2040 /* hash info structure used by the directory hash */
2041 struct dx_hash_info
2042 {
2043 u32 hash;
2044 u32 minor_hash;
2045 int hash_version;
2046 u32 *seed;
2047 };
2048
2049
2050 /* 32 and 64 bit signed EOF for dx directories */
2051 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
2052 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
2053
2054
2055 /*
2056 * Control parameters used by ext4_htree_next_block
2057 */
2058 #define HASH_NB_ALWAYS 1
2059
2060 struct ext4_filename {
2061 const struct qstr *usr_fname;
2062 struct fscrypt_str disk_name;
2063 struct dx_hash_info hinfo;
2064 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2065 struct fscrypt_str crypto_buf;
2066 #endif
2067 };
2068
2069 #define fname_name(p) ((p)->disk_name.name)
2070 #define fname_len(p) ((p)->disk_name.len)
2071
2072 /*
2073 * Describe an inode's exact location on disk and in memory
2074 */
2075 struct ext4_iloc
2076 {
2077 struct buffer_head *bh;
2078 unsigned long offset;
2079 ext4_group_t block_group;
2080 };
2081
2082 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2083 {
2084 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2085 }
2086
2087 /*
2088 * This structure is stuffed into the struct file's private_data field
2089 * for directories. It is where we put information so that we can do
2090 * readdir operations in hash tree order.
2091 */
2092 struct dir_private_info {
2093 struct rb_root root;
2094 struct rb_node *curr_node;
2095 struct fname *extra_fname;
2096 loff_t last_pos;
2097 __u32 curr_hash;
2098 __u32 curr_minor_hash;
2099 __u32 next_hash;
2100 };
2101
2102 /* calculate the first block number of the group */
2103 static inline ext4_fsblk_t
2104 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2105 {
2106 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2107 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2108 }
2109
2110 /*
2111 * Special error return code only used by dx_probe() and its callers.
2112 */
2113 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
2114
2115 /*
2116 * Timeout and state flag for lazy initialization inode thread.
2117 */
2118 #define EXT4_DEF_LI_WAIT_MULT 10
2119 #define EXT4_DEF_LI_MAX_START_DELAY 5
2120 #define EXT4_LAZYINIT_QUIT 0x0001
2121 #define EXT4_LAZYINIT_RUNNING 0x0002
2122
2123 /*
2124 * Lazy inode table initialization info
2125 */
2126 struct ext4_lazy_init {
2127 unsigned long li_state;
2128 struct list_head li_request_list;
2129 struct mutex li_list_mtx;
2130 };
2131
2132 struct ext4_li_request {
2133 struct super_block *lr_super;
2134 struct ext4_sb_info *lr_sbi;
2135 ext4_group_t lr_next_group;
2136 struct list_head lr_request;
2137 unsigned long lr_next_sched;
2138 unsigned long lr_timeout;
2139 };
2140
2141 struct ext4_features {
2142 struct kobject f_kobj;
2143 struct completion f_kobj_unregister;
2144 };
2145
2146 /*
2147 * This structure will be used for multiple mount protection. It will be
2148 * written into the block number saved in the s_mmp_block field in the
2149 * superblock. Programs that check MMP should assume that if
2150 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2151 * to use the filesystem, regardless of how old the timestamp is.
2152 */
2153 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
2154 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2155 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
2156 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
2157
2158 struct mmp_struct {
2159 __le32 mmp_magic; /* Magic number for MMP */
2160 __le32 mmp_seq; /* Sequence no. updated periodically */
2161
2162 /*
2163 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2164 * purposes and do not affect the correctness of the algorithm
2165 */
2166 __le64 mmp_time; /* Time last updated */
2167 char mmp_nodename[64]; /* Node which last updated MMP block */
2168 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
2169
2170 /*
2171 * mmp_check_interval is used to verify if the MMP block has been
2172 * updated on the block device. The value is updated based on the
2173 * maximum time to write the MMP block during an update cycle.
2174 */
2175 __le16 mmp_check_interval;
2176
2177 __le16 mmp_pad1;
2178 __le32 mmp_pad2[226];
2179 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
2180 };
2181
2182 /* arguments passed to the mmp thread */
2183 struct mmpd_data {
2184 struct buffer_head *bh; /* bh from initial read_mmp_block() */
2185 struct super_block *sb; /* super block of the fs */
2186 };
2187
2188 /*
2189 * Check interval multiplier
2190 * The MMP block is written every update interval and initially checked every
2191 * update interval x the multiplier (the value is then adapted based on the
2192 * write latency). The reason is that writes can be delayed under load and we
2193 * don't want readers to incorrectly assume that the filesystem is no longer
2194 * in use.
2195 */
2196 #define EXT4_MMP_CHECK_MULT 2UL
2197
2198 /*
2199 * Minimum interval for MMP checking in seconds.
2200 */
2201 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
2202
2203 /*
2204 * Maximum interval for MMP checking in seconds.
2205 */
2206 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
2207
2208 /*
2209 * Function prototypes
2210 */
2211
2212 /*
2213 * Ok, these declarations are also in <linux/kernel.h> but none of the
2214 * ext4 source programs needs to include it so they are duplicated here.
2215 */
2216 # define NORET_TYPE /**/
2217 # define ATTRIB_NORET __attribute__((noreturn))
2218 # define NORET_AND noreturn,
2219
2220 /* bitmap.c */
2221 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2222 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2223 struct ext4_group_desc *gdp,
2224 struct buffer_head *bh, int sz);
2225 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2226 struct ext4_group_desc *gdp,
2227 struct buffer_head *bh, int sz);
2228 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2229 struct ext4_group_desc *gdp,
2230 struct buffer_head *bh);
2231 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2232 struct ext4_group_desc *gdp,
2233 struct buffer_head *bh);
2234
2235 /* balloc.c */
2236 extern void ext4_get_group_no_and_offset(struct super_block *sb,
2237 ext4_fsblk_t blocknr,
2238 ext4_group_t *blockgrpp,
2239 ext4_grpblk_t *offsetp);
2240 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2241 ext4_fsblk_t block);
2242
2243 extern unsigned int ext4_block_group(struct super_block *sb,
2244 ext4_fsblk_t blocknr);
2245 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2246 ext4_fsblk_t blocknr);
2247 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2248 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2249 ext4_group_t group);
2250 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2251 ext4_fsblk_t goal,
2252 unsigned int flags,
2253 unsigned long *count,
2254 int *errp);
2255 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2256 s64 nclusters, unsigned int flags);
2257 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2258 extern void ext4_check_blocks_bitmap(struct super_block *);
2259 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2260 ext4_group_t block_group,
2261 struct buffer_head ** bh);
2262 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2263
2264 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2265 ext4_group_t block_group);
2266 extern int ext4_wait_block_bitmap(struct super_block *sb,
2267 ext4_group_t block_group,
2268 struct buffer_head *bh);
2269 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2270 ext4_group_t block_group);
2271 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2272 ext4_group_t block_group,
2273 struct ext4_group_desc *gdp);
2274 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2275
2276 static inline int ext4_sb_has_crypto(struct super_block *sb)
2277 {
2278 return ext4_has_feature_encrypt(sb);
2279 }
2280
2281 static inline bool ext4_encrypted_inode(struct inode *inode)
2282 {
2283 return ext4_test_inode_flag(inode, EXT4_INODE_ENCRYPT);
2284 }
2285
2286 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2287 static inline int ext4_fname_setup_filename(struct inode *dir,
2288 const struct qstr *iname,
2289 int lookup, struct ext4_filename *fname)
2290 {
2291 struct fscrypt_name name;
2292 int err;
2293
2294 memset(fname, 0, sizeof(struct ext4_filename));
2295
2296 err = fscrypt_setup_filename(dir, iname, lookup, &name);
2297
2298 fname->usr_fname = name.usr_fname;
2299 fname->disk_name = name.disk_name;
2300 fname->hinfo.hash = name.hash;
2301 fname->hinfo.minor_hash = name.minor_hash;
2302 fname->crypto_buf = name.crypto_buf;
2303 return err;
2304 }
2305
2306 static inline void ext4_fname_free_filename(struct ext4_filename *fname)
2307 {
2308 struct fscrypt_name name;
2309
2310 name.crypto_buf = fname->crypto_buf;
2311 fscrypt_free_filename(&name);
2312
2313 fname->crypto_buf.name = NULL;
2314 fname->usr_fname = NULL;
2315 fname->disk_name.name = NULL;
2316 }
2317 #else
2318 static inline int ext4_fname_setup_filename(struct inode *dir,
2319 const struct qstr *iname,
2320 int lookup, struct ext4_filename *fname)
2321 {
2322 fname->usr_fname = iname;
2323 fname->disk_name.name = (unsigned char *) iname->name;
2324 fname->disk_name.len = iname->len;
2325 return 0;
2326 }
2327 static inline void ext4_fname_free_filename(struct ext4_filename *fname) { }
2328
2329 #define fscrypt_set_d_op(i)
2330 #define fscrypt_get_ctx fscrypt_notsupp_get_ctx
2331 #define fscrypt_release_ctx fscrypt_notsupp_release_ctx
2332 #define fscrypt_encrypt_page fscrypt_notsupp_encrypt_page
2333 #define fscrypt_decrypt_page fscrypt_notsupp_decrypt_page
2334 #define fscrypt_decrypt_bio_pages fscrypt_notsupp_decrypt_bio_pages
2335 #define fscrypt_pullback_bio_page fscrypt_notsupp_pullback_bio_page
2336 #define fscrypt_restore_control_page fscrypt_notsupp_restore_control_page
2337 #define fscrypt_zeroout_range fscrypt_notsupp_zeroout_range
2338 #define fscrypt_process_policy fscrypt_notsupp_process_policy
2339 #define fscrypt_get_policy fscrypt_notsupp_get_policy
2340 #define fscrypt_has_permitted_context fscrypt_notsupp_has_permitted_context
2341 #define fscrypt_inherit_context fscrypt_notsupp_inherit_context
2342 #define fscrypt_get_encryption_info fscrypt_notsupp_get_encryption_info
2343 #define fscrypt_put_encryption_info fscrypt_notsupp_put_encryption_info
2344 #define fscrypt_setup_filename fscrypt_notsupp_setup_filename
2345 #define fscrypt_free_filename fscrypt_notsupp_free_filename
2346 #define fscrypt_fname_encrypted_size fscrypt_notsupp_fname_encrypted_size
2347 #define fscrypt_fname_alloc_buffer fscrypt_notsupp_fname_alloc_buffer
2348 #define fscrypt_fname_free_buffer fscrypt_notsupp_fname_free_buffer
2349 #define fscrypt_fname_disk_to_usr fscrypt_notsupp_fname_disk_to_usr
2350 #define fscrypt_fname_usr_to_disk fscrypt_notsupp_fname_usr_to_disk
2351 #endif
2352
2353 /* dir.c */
2354 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2355 struct file *,
2356 struct ext4_dir_entry_2 *,
2357 struct buffer_head *, char *, int,
2358 unsigned int);
2359 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2360 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2361 (de), (bh), (buf), (size), (offset)))
2362 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2363 __u32 minor_hash,
2364 struct ext4_dir_entry_2 *dirent,
2365 struct fscrypt_str *ent_name);
2366 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2367 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2368 struct buffer_head *bh,
2369 void *buf, int buf_size,
2370 struct ext4_filename *fname,
2371 struct ext4_dir_entry_2 **dest_de);
2372 int ext4_insert_dentry(struct inode *dir,
2373 struct inode *inode,
2374 struct ext4_dir_entry_2 *de,
2375 int buf_size,
2376 struct ext4_filename *fname);
2377 static inline void ext4_update_dx_flag(struct inode *inode)
2378 {
2379 if (!ext4_has_feature_dir_index(inode->i_sb))
2380 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2381 }
2382 static unsigned char ext4_filetype_table[] = {
2383 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2384 };
2385
2386 static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2387 {
2388 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2389 return DT_UNKNOWN;
2390
2391 return ext4_filetype_table[filetype];
2392 }
2393 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2394 void *buf, int buf_size);
2395
2396 /* fsync.c */
2397 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2398
2399 /* hash.c */
2400 extern int ext4fs_dirhash(const char *name, int len, struct
2401 dx_hash_info *hinfo);
2402
2403 /* ialloc.c */
2404 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2405 const struct qstr *qstr, __u32 goal,
2406 uid_t *owner, int handle_type,
2407 unsigned int line_no, int nblocks);
2408
2409 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner) \
2410 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2411 0, 0, 0)
2412 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2413 type, nblocks) \
2414 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2415 (type), __LINE__, (nblocks))
2416
2417
2418 extern void ext4_free_inode(handle_t *, struct inode *);
2419 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2420 extern unsigned long ext4_count_free_inodes(struct super_block *);
2421 extern unsigned long ext4_count_dirs(struct super_block *);
2422 extern void ext4_check_inodes_bitmap(struct super_block *);
2423 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2424 extern int ext4_init_inode_table(struct super_block *sb,
2425 ext4_group_t group, int barrier);
2426 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2427
2428 /* mballoc.c */
2429 extern const struct file_operations ext4_seq_mb_groups_fops;
2430 extern long ext4_mb_stats;
2431 extern long ext4_mb_max_to_scan;
2432 extern int ext4_mb_init(struct super_block *);
2433 extern int ext4_mb_release(struct super_block *);
2434 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2435 struct ext4_allocation_request *, int *);
2436 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2437 extern void ext4_discard_preallocations(struct inode *);
2438 extern int __init ext4_init_mballoc(void);
2439 extern void ext4_exit_mballoc(void);
2440 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2441 struct buffer_head *bh, ext4_fsblk_t block,
2442 unsigned long count, int flags);
2443 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2444 ext4_group_t ngroups);
2445 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2446 ext4_group_t i, struct ext4_group_desc *desc);
2447 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2448 ext4_fsblk_t block, unsigned long count);
2449 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2450
2451 /* inode.c */
2452 int ext4_inode_is_fast_symlink(struct inode *inode);
2453 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2454 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2455 int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
2456 struct buffer_head *bh_result, int create);
2457 int ext4_dax_get_block(struct inode *inode, sector_t iblock,
2458 struct buffer_head *bh_result, int create);
2459 int ext4_get_block(struct inode *inode, sector_t iblock,
2460 struct buffer_head *bh_result, int create);
2461 int ext4_dio_get_block(struct inode *inode, sector_t iblock,
2462 struct buffer_head *bh_result, int create);
2463 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2464 struct buffer_head *bh, int create);
2465 int ext4_walk_page_buffers(handle_t *handle,
2466 struct buffer_head *head,
2467 unsigned from,
2468 unsigned to,
2469 int *partial,
2470 int (*fn)(handle_t *handle,
2471 struct buffer_head *bh));
2472 int do_journal_get_write_access(handle_t *handle,
2473 struct buffer_head *bh);
2474 #define FALL_BACK_TO_NONDELALLOC 1
2475 #define CONVERT_INLINE_DATA 2
2476
2477 extern struct inode *ext4_iget(struct super_block *, unsigned long);
2478 extern struct inode *ext4_iget_normal(struct super_block *, unsigned long);
2479 extern int ext4_write_inode(struct inode *, struct writeback_control *);
2480 extern int ext4_setattr(struct dentry *, struct iattr *);
2481 extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
2482 struct kstat *stat);
2483 extern void ext4_evict_inode(struct inode *);
2484 extern void ext4_clear_inode(struct inode *);
2485 extern int ext4_sync_inode(handle_t *, struct inode *);
2486 extern void ext4_dirty_inode(struct inode *, int);
2487 extern int ext4_change_inode_journal_flag(struct inode *, int);
2488 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2489 extern int ext4_inode_attach_jinode(struct inode *inode);
2490 extern int ext4_can_truncate(struct inode *inode);
2491 extern void ext4_truncate(struct inode *);
2492 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2493 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
2494 extern void ext4_set_inode_flags(struct inode *);
2495 extern void ext4_get_inode_flags(struct ext4_inode_info *);
2496 extern int ext4_alloc_da_blocks(struct inode *inode);
2497 extern void ext4_set_aops(struct inode *inode);
2498 extern int ext4_writepage_trans_blocks(struct inode *);
2499 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2500 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2501 loff_t lstart, loff_t lend);
2502 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
2503 extern int ext4_filemap_fault(struct vm_area_struct *vma, struct vm_fault *vmf);
2504 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2505 extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
2506 extern void ext4_da_update_reserve_space(struct inode *inode,
2507 int used, int quota_claim);
2508 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
2509 ext4_fsblk_t pblk, ext4_lblk_t len);
2510 extern int ext4_get_next_extent(struct inode *inode, ext4_lblk_t lblk,
2511 unsigned int map_len,
2512 struct extent_status *result);
2513
2514 /* indirect.c */
2515 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2516 struct ext4_map_blocks *map, int flags);
2517 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2518 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2519 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2520 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2521 ext4_lblk_t start, ext4_lblk_t end);
2522
2523 /* ioctl.c */
2524 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2525 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2526
2527 /* migrate.c */
2528 extern int ext4_ext_migrate(struct inode *);
2529 extern int ext4_ind_migrate(struct inode *inode);
2530
2531 /* namei.c */
2532 extern int ext4_dirent_csum_verify(struct inode *inode,
2533 struct ext4_dir_entry *dirent);
2534 extern int ext4_orphan_add(handle_t *, struct inode *);
2535 extern int ext4_orphan_del(handle_t *, struct inode *);
2536 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2537 __u32 start_minor_hash, __u32 *next_hash);
2538 extern int ext4_search_dir(struct buffer_head *bh,
2539 char *search_buf,
2540 int buf_size,
2541 struct inode *dir,
2542 struct ext4_filename *fname,
2543 const struct qstr *d_name,
2544 unsigned int offset,
2545 struct ext4_dir_entry_2 **res_dir);
2546 extern int ext4_generic_delete_entry(handle_t *handle,
2547 struct inode *dir,
2548 struct ext4_dir_entry_2 *de_del,
2549 struct buffer_head *bh,
2550 void *entry_buf,
2551 int buf_size,
2552 int csum_size);
2553 extern bool ext4_empty_dir(struct inode *inode);
2554
2555 /* resize.c */
2556 extern int ext4_group_add(struct super_block *sb,
2557 struct ext4_new_group_data *input);
2558 extern int ext4_group_extend(struct super_block *sb,
2559 struct ext4_super_block *es,
2560 ext4_fsblk_t n_blocks_count);
2561 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2562
2563 /* super.c */
2564 extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
2565 extern int ext4_calculate_overhead(struct super_block *sb);
2566 extern void ext4_superblock_csum_set(struct super_block *sb);
2567 extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2568 extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2569 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2570 ext4_group_t ngroup);
2571 extern const char *ext4_decode_error(struct super_block *sb, int errno,
2572 char nbuf[16]);
2573
2574 extern __printf(4, 5)
2575 void __ext4_error(struct super_block *, const char *, unsigned int,
2576 const char *, ...);
2577 extern __printf(5, 6)
2578 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2579 const char *, ...);
2580 extern __printf(5, 6)
2581 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2582 const char *, ...);
2583 extern void __ext4_std_error(struct super_block *, const char *,
2584 unsigned int, int);
2585 extern __printf(4, 5)
2586 void __ext4_abort(struct super_block *, const char *, unsigned int,
2587 const char *, ...);
2588 extern __printf(4, 5)
2589 void __ext4_warning(struct super_block *, const char *, unsigned int,
2590 const char *, ...);
2591 extern __printf(4, 5)
2592 void __ext4_warning_inode(const struct inode *inode, const char *function,
2593 unsigned int line, const char *fmt, ...);
2594 extern __printf(3, 4)
2595 void __ext4_msg(struct super_block *, const char *, const char *, ...);
2596 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2597 const char *, unsigned int, const char *);
2598 extern __printf(7, 8)
2599 void __ext4_grp_locked_error(const char *, unsigned int,
2600 struct super_block *, ext4_group_t,
2601 unsigned long, ext4_fsblk_t,
2602 const char *, ...);
2603
2604 #define EXT4_ERROR_INODE(inode, fmt, a...) \
2605 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
2606
2607 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
2608 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
2609
2610 #define EXT4_ERROR_FILE(file, block, fmt, a...) \
2611 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
2612
2613 #ifdef CONFIG_PRINTK
2614
2615 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2616 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__)
2617 #define ext4_error_file(file, func, line, block, fmt, ...) \
2618 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2619 #define ext4_error(sb, fmt, ...) \
2620 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2621 #define ext4_abort(sb, fmt, ...) \
2622 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2623 #define ext4_warning(sb, fmt, ...) \
2624 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2625 #define ext4_warning_inode(inode, fmt, ...) \
2626 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
2627 #define ext4_msg(sb, level, fmt, ...) \
2628 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2629 #define dump_mmp_msg(sb, mmp, msg) \
2630 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2631 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2632 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2633 fmt, ##__VA_ARGS__)
2634
2635 #else
2636
2637 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2638 do { \
2639 no_printk(fmt, ##__VA_ARGS__); \
2640 __ext4_error_inode(inode, "", 0, block, " "); \
2641 } while (0)
2642 #define ext4_error_file(file, func, line, block, fmt, ...) \
2643 do { \
2644 no_printk(fmt, ##__VA_ARGS__); \
2645 __ext4_error_file(file, "", 0, block, " "); \
2646 } while (0)
2647 #define ext4_error(sb, fmt, ...) \
2648 do { \
2649 no_printk(fmt, ##__VA_ARGS__); \
2650 __ext4_error(sb, "", 0, " "); \
2651 } while (0)
2652 #define ext4_abort(sb, fmt, ...) \
2653 do { \
2654 no_printk(fmt, ##__VA_ARGS__); \
2655 __ext4_abort(sb, "", 0, " "); \
2656 } while (0)
2657 #define ext4_warning(sb, fmt, ...) \
2658 do { \
2659 no_printk(fmt, ##__VA_ARGS__); \
2660 __ext4_warning(sb, "", 0, " "); \
2661 } while (0)
2662 #define ext4_warning_inode(inode, fmt, ...) \
2663 do { \
2664 no_printk(fmt, ##__VA_ARGS__); \
2665 __ext4_warning_inode(inode, "", 0, " "); \
2666 } while (0)
2667 #define ext4_msg(sb, level, fmt, ...) \
2668 do { \
2669 no_printk(fmt, ##__VA_ARGS__); \
2670 __ext4_msg(sb, "", " "); \
2671 } while (0)
2672 #define dump_mmp_msg(sb, mmp, msg) \
2673 __dump_mmp_msg(sb, mmp, "", 0, "")
2674 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2675 do { \
2676 no_printk(fmt, ##__VA_ARGS__); \
2677 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
2678 } while (0)
2679
2680 #endif
2681
2682 extern void ext4_update_dynamic_rev(struct super_block *sb);
2683 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2684 __u32 compat);
2685 extern int ext4_update_rocompat_feature(handle_t *handle,
2686 struct super_block *sb, __u32 rocompat);
2687 extern int ext4_update_incompat_feature(handle_t *handle,
2688 struct super_block *sb, __u32 incompat);
2689 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2690 struct ext4_group_desc *bg);
2691 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2692 struct ext4_group_desc *bg);
2693 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2694 struct ext4_group_desc *bg);
2695 extern __u32 ext4_free_group_clusters(struct super_block *sb,
2696 struct ext4_group_desc *bg);
2697 extern __u32 ext4_free_inodes_count(struct super_block *sb,
2698 struct ext4_group_desc *bg);
2699 extern __u32 ext4_used_dirs_count(struct super_block *sb,
2700 struct ext4_group_desc *bg);
2701 extern __u32 ext4_itable_unused_count(struct super_block *sb,
2702 struct ext4_group_desc *bg);
2703 extern void ext4_block_bitmap_set(struct super_block *sb,
2704 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2705 extern void ext4_inode_bitmap_set(struct super_block *sb,
2706 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2707 extern void ext4_inode_table_set(struct super_block *sb,
2708 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2709 extern void ext4_free_group_clusters_set(struct super_block *sb,
2710 struct ext4_group_desc *bg,
2711 __u32 count);
2712 extern void ext4_free_inodes_set(struct super_block *sb,
2713 struct ext4_group_desc *bg, __u32 count);
2714 extern void ext4_used_dirs_set(struct super_block *sb,
2715 struct ext4_group_desc *bg, __u32 count);
2716 extern void ext4_itable_unused_set(struct super_block *sb,
2717 struct ext4_group_desc *bg, __u32 count);
2718 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2719 struct ext4_group_desc *gdp);
2720 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2721 struct ext4_group_desc *gdp);
2722 extern int ext4_register_li_request(struct super_block *sb,
2723 ext4_group_t first_not_zeroed);
2724
2725 static inline int ext4_has_group_desc_csum(struct super_block *sb)
2726 {
2727 return ext4_has_feature_gdt_csum(sb) ||
2728 EXT4_SB(sb)->s_chksum_driver != NULL;
2729 }
2730
2731 static inline int ext4_has_metadata_csum(struct super_block *sb)
2732 {
2733 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
2734 !EXT4_SB(sb)->s_chksum_driver);
2735
2736 return (EXT4_SB(sb)->s_chksum_driver != NULL);
2737 }
2738 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2739 {
2740 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2741 le32_to_cpu(es->s_blocks_count_lo);
2742 }
2743
2744 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2745 {
2746 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2747 le32_to_cpu(es->s_r_blocks_count_lo);
2748 }
2749
2750 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2751 {
2752 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2753 le32_to_cpu(es->s_free_blocks_count_lo);
2754 }
2755
2756 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2757 ext4_fsblk_t blk)
2758 {
2759 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2760 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2761 }
2762
2763 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2764 ext4_fsblk_t blk)
2765 {
2766 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2767 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2768 }
2769
2770 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2771 ext4_fsblk_t blk)
2772 {
2773 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2774 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2775 }
2776
2777 static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
2778 {
2779 if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2780 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2781 le32_to_cpu(raw_inode->i_size_lo);
2782 else
2783 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2784 }
2785
2786 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2787 {
2788 raw_inode->i_size_lo = cpu_to_le32(i_size);
2789 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2790 }
2791
2792 static inline
2793 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2794 ext4_group_t group)
2795 {
2796 struct ext4_group_info ***grp_info;
2797 long indexv, indexh;
2798 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
2799 grp_info = EXT4_SB(sb)->s_group_info;
2800 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2801 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2802 return grp_info[indexv][indexh];
2803 }
2804
2805 /*
2806 * Reading s_groups_count requires using smp_rmb() afterwards. See
2807 * the locking protocol documented in the comments of ext4_group_add()
2808 * in resize.c
2809 */
2810 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2811 {
2812 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
2813
2814 smp_rmb();
2815 return ngroups;
2816 }
2817
2818 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2819 ext4_group_t block_group)
2820 {
2821 return block_group >> sbi->s_log_groups_per_flex;
2822 }
2823
2824 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2825 {
2826 return 1 << sbi->s_log_groups_per_flex;
2827 }
2828
2829 #define ext4_std_error(sb, errno) \
2830 do { \
2831 if ((errno)) \
2832 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
2833 } while (0)
2834
2835 #ifdef CONFIG_SMP
2836 /* Each CPU can accumulate percpu_counter_batch clusters in their local
2837 * counters. So we need to make sure we have free clusters more
2838 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
2839 */
2840 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2841 #else
2842 #define EXT4_FREECLUSTERS_WATERMARK 0
2843 #endif
2844
2845 /* Update i_disksize. Requires i_mutex to avoid races with truncate */
2846 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2847 {
2848 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
2849 !inode_is_locked(inode));
2850 down_write(&EXT4_I(inode)->i_data_sem);
2851 if (newsize > EXT4_I(inode)->i_disksize)
2852 EXT4_I(inode)->i_disksize = newsize;
2853 up_write(&EXT4_I(inode)->i_data_sem);
2854 }
2855
2856 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
2857 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
2858 {
2859 int changed = 0;
2860
2861 if (newsize > inode->i_size) {
2862 i_size_write(inode, newsize);
2863 changed = 1;
2864 }
2865 if (newsize > EXT4_I(inode)->i_disksize) {
2866 ext4_update_i_disksize(inode, newsize);
2867 changed |= 2;
2868 }
2869 return changed;
2870 }
2871
2872 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
2873 loff_t len);
2874
2875 struct ext4_group_info {
2876 unsigned long bb_state;
2877 struct rb_root bb_free_root;
2878 ext4_grpblk_t bb_first_free; /* first free block */
2879 ext4_grpblk_t bb_free; /* total free blocks */
2880 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
2881 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
2882 struct list_head bb_prealloc_list;
2883 #ifdef DOUBLE_CHECK
2884 void *bb_bitmap;
2885 #endif
2886 struct rw_semaphore alloc_sem;
2887 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
2888 * regions, index is order.
2889 * bb_counters[3] = 5 means
2890 * 5 free 8-block regions. */
2891 };
2892
2893 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
2894 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
2895 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
2896 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
2897
2898 #define EXT4_MB_GRP_NEED_INIT(grp) \
2899 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2900 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
2901 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2902 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
2903 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2904
2905 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
2906 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2907 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
2908 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2909 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
2910 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2911
2912 #define EXT4_MAX_CONTENTION 8
2913 #define EXT4_CONTENTION_THRESHOLD 2
2914
2915 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2916 ext4_group_t group)
2917 {
2918 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2919 }
2920
2921 /*
2922 * Returns true if the filesystem is busy enough that attempts to
2923 * access the block group locks has run into contention.
2924 */
2925 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2926 {
2927 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2928 }
2929
2930 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2931 {
2932 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2933 if (spin_trylock(lock))
2934 /*
2935 * We're able to grab the lock right away, so drop the
2936 * lock contention counter.
2937 */
2938 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2939 else {
2940 /*
2941 * The lock is busy, so bump the contention counter,
2942 * and then wait on the spin lock.
2943 */
2944 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2945 EXT4_MAX_CONTENTION);
2946 spin_lock(lock);
2947 }
2948 }
2949
2950 static inline void ext4_unlock_group(struct super_block *sb,
2951 ext4_group_t group)
2952 {
2953 spin_unlock(ext4_group_lock_ptr(sb, group));
2954 }
2955
2956 /*
2957 * Block validity checking
2958 */
2959 #define ext4_check_indirect_blockref(inode, bh) \
2960 ext4_check_blockref(__func__, __LINE__, inode, \
2961 (__le32 *)(bh)->b_data, \
2962 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2963
2964 #define ext4_ind_check_inode(inode) \
2965 ext4_check_blockref(__func__, __LINE__, inode, \
2966 EXT4_I(inode)->i_data, \
2967 EXT4_NDIR_BLOCKS)
2968
2969 /*
2970 * Inodes and files operations
2971 */
2972
2973 /* dir.c */
2974 extern const struct file_operations ext4_dir_operations;
2975
2976 /* file.c */
2977 extern const struct inode_operations ext4_file_inode_operations;
2978 extern const struct file_operations ext4_file_operations;
2979 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
2980
2981 /* inline.c */
2982 extern int ext4_get_max_inline_size(struct inode *inode);
2983 extern int ext4_find_inline_data_nolock(struct inode *inode);
2984 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
2985 unsigned int len);
2986 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
2987
2988 extern int ext4_readpage_inline(struct inode *inode, struct page *page);
2989 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
2990 struct inode *inode,
2991 loff_t pos, unsigned len,
2992 unsigned flags,
2993 struct page **pagep);
2994 extern int ext4_write_inline_data_end(struct inode *inode,
2995 loff_t pos, unsigned len,
2996 unsigned copied,
2997 struct page *page);
2998 extern struct buffer_head *
2999 ext4_journalled_write_inline_data(struct inode *inode,
3000 unsigned len,
3001 struct page *page);
3002 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
3003 struct inode *inode,
3004 loff_t pos, unsigned len,
3005 unsigned flags,
3006 struct page **pagep,
3007 void **fsdata);
3008 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
3009 unsigned len, unsigned copied,
3010 struct page *page);
3011 extern int ext4_try_add_inline_entry(handle_t *handle,
3012 struct ext4_filename *fname,
3013 struct inode *dir, struct inode *inode);
3014 extern int ext4_try_create_inline_dir(handle_t *handle,
3015 struct inode *parent,
3016 struct inode *inode);
3017 extern int ext4_read_inline_dir(struct file *filp,
3018 struct dir_context *ctx,
3019 int *has_inline_data);
3020 extern int htree_inlinedir_to_tree(struct file *dir_file,
3021 struct inode *dir, ext4_lblk_t block,
3022 struct dx_hash_info *hinfo,
3023 __u32 start_hash, __u32 start_minor_hash,
3024 int *has_inline_data);
3025 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3026 struct ext4_filename *fname,
3027 const struct qstr *d_name,
3028 struct ext4_dir_entry_2 **res_dir,
3029 int *has_inline_data);
3030 extern int ext4_delete_inline_entry(handle_t *handle,
3031 struct inode *dir,
3032 struct ext4_dir_entry_2 *de_del,
3033 struct buffer_head *bh,
3034 int *has_inline_data);
3035 extern bool empty_inline_dir(struct inode *dir, int *has_inline_data);
3036 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3037 struct ext4_dir_entry_2 **parent_de,
3038 int *retval);
3039 extern int ext4_inline_data_fiemap(struct inode *inode,
3040 struct fiemap_extent_info *fieinfo,
3041 int *has_inline, __u64 start, __u64 len);
3042 extern int ext4_try_to_evict_inline_data(handle_t *handle,
3043 struct inode *inode,
3044 int needed);
3045 extern void ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3046
3047 extern int ext4_convert_inline_data(struct inode *inode);
3048
3049 static inline int ext4_has_inline_data(struct inode *inode)
3050 {
3051 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3052 EXT4_I(inode)->i_inline_off;
3053 }
3054
3055 /* namei.c */
3056 extern const struct inode_operations ext4_dir_inode_operations;
3057 extern const struct inode_operations ext4_special_inode_operations;
3058 extern struct dentry *ext4_get_parent(struct dentry *child);
3059 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3060 struct ext4_dir_entry_2 *de,
3061 int blocksize, int csum_size,
3062 unsigned int parent_ino, int dotdot_real_len);
3063 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t,
3064 unsigned int blocksize);
3065 extern int ext4_handle_dirty_dirent_node(handle_t *handle,
3066 struct inode *inode,
3067 struct buffer_head *bh);
3068 #define S_SHIFT 12
3069 static unsigned char ext4_type_by_mode[S_IFMT >> S_SHIFT] = {
3070 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
3071 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
3072 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
3073 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
3074 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
3075 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
3076 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
3077 };
3078
3079 static inline void ext4_set_de_type(struct super_block *sb,
3080 struct ext4_dir_entry_2 *de,
3081 umode_t mode) {
3082 if (ext4_has_feature_filetype(sb))
3083 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3084 }
3085
3086 /* readpages.c */
3087 extern int ext4_mpage_readpages(struct address_space *mapping,
3088 struct list_head *pages, struct page *page,
3089 unsigned nr_pages);
3090
3091 /* symlink.c */
3092 extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3093 extern const struct inode_operations ext4_symlink_inode_operations;
3094 extern const struct inode_operations ext4_fast_symlink_inode_operations;
3095
3096 /* sysfs.c */
3097 extern int ext4_register_sysfs(struct super_block *sb);
3098 extern void ext4_unregister_sysfs(struct super_block *sb);
3099 extern int __init ext4_init_sysfs(void);
3100 extern void ext4_exit_sysfs(void);
3101
3102 /* block_validity */
3103 extern void ext4_release_system_zone(struct super_block *sb);
3104 extern int ext4_setup_system_zone(struct super_block *sb);
3105 extern int __init ext4_init_system_zone(void);
3106 extern void ext4_exit_system_zone(void);
3107 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
3108 ext4_fsblk_t start_blk,
3109 unsigned int count);
3110 extern int ext4_check_blockref(const char *, unsigned int,
3111 struct inode *, __le32 *, unsigned int);
3112
3113 /* extents.c */
3114 struct ext4_ext_path;
3115 struct ext4_extent;
3116
3117 /*
3118 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3119 * __le32.
3120 */
3121 #define EXT_MAX_BLOCKS 0xffffffff
3122
3123 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
3124 extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
3125 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3126 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3127 struct ext4_map_blocks *map, int flags);
3128 extern void ext4_ext_truncate(handle_t *, struct inode *);
3129 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3130 ext4_lblk_t end);
3131 extern void ext4_ext_init(struct super_block *);
3132 extern void ext4_ext_release(struct super_block *);
3133 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3134 loff_t len);
3135 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3136 loff_t offset, ssize_t len);
3137 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3138 struct ext4_map_blocks *map, int flags);
3139 extern int ext4_ext_calc_metadata_amount(struct inode *inode,
3140 ext4_lblk_t lblocks);
3141 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3142 int num,
3143 struct ext4_ext_path *path);
3144 extern int ext4_can_extents_be_merged(struct inode *inode,
3145 struct ext4_extent *ex1,
3146 struct ext4_extent *ex2);
3147 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3148 struct ext4_ext_path **,
3149 struct ext4_extent *, int);
3150 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3151 struct ext4_ext_path **,
3152 int flags);
3153 extern void ext4_ext_drop_refs(struct ext4_ext_path *);
3154 extern int ext4_ext_check_inode(struct inode *inode);
3155 extern int ext4_find_delalloc_range(struct inode *inode,
3156 ext4_lblk_t lblk_start,
3157 ext4_lblk_t lblk_end);
3158 extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk);
3159 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3160 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3161 __u64 start, __u64 len);
3162 extern int ext4_ext_precache(struct inode *inode);
3163 extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len);
3164 extern int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len);
3165 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3166 struct inode *inode2, ext4_lblk_t lblk1,
3167 ext4_lblk_t lblk2, ext4_lblk_t count,
3168 int mark_unwritten,int *err);
3169
3170 /* move_extent.c */
3171 extern void ext4_double_down_write_data_sem(struct inode *first,
3172 struct inode *second);
3173 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3174 struct inode *donor_inode);
3175 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3176 __u64 start_orig, __u64 start_donor,
3177 __u64 len, __u64 *moved_len);
3178
3179 /* page-io.c */
3180 extern int __init ext4_init_pageio(void);
3181 extern void ext4_exit_pageio(void);
3182 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3183 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3184 extern int ext4_put_io_end(ext4_io_end_t *io_end);
3185 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3186 extern void ext4_io_submit_init(struct ext4_io_submit *io,
3187 struct writeback_control *wbc);
3188 extern void ext4_end_io_rsv_work(struct work_struct *work);
3189 extern void ext4_io_submit(struct ext4_io_submit *io);
3190 extern int ext4_bio_write_page(struct ext4_io_submit *io,
3191 struct page *page,
3192 int len,
3193 struct writeback_control *wbc,
3194 bool keep_towrite);
3195
3196 /* mmp.c */
3197 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3198
3199 /*
3200 * Add new method to test whether block and inode bitmaps are properly
3201 * initialized. With uninit_bg reading the block from disk is not enough
3202 * to mark the bitmap uptodate. We need to also zero-out the bitmap
3203 */
3204 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3205
3206 static inline int bitmap_uptodate(struct buffer_head *bh)
3207 {
3208 return (buffer_uptodate(bh) &&
3209 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3210 }
3211 static inline void set_bitmap_uptodate(struct buffer_head *bh)
3212 {
3213 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3214 }
3215
3216 /*
3217 * Disable DIO read nolock optimization, so new dioreaders will be forced
3218 * to grab i_mutex
3219 */
3220 static inline void ext4_inode_block_unlocked_dio(struct inode *inode)
3221 {
3222 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
3223 smp_mb();
3224 }
3225 static inline void ext4_inode_resume_unlocked_dio(struct inode *inode)
3226 {
3227 smp_mb();
3228 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
3229 }
3230
3231 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
3232
3233 /* For ioend & aio unwritten conversion wait queues */
3234 #define EXT4_WQ_HASH_SZ 37
3235 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
3236 EXT4_WQ_HASH_SZ])
3237 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3238
3239 #define EXT4_RESIZING 0
3240 extern int ext4_resize_begin(struct super_block *sb);
3241 extern void ext4_resize_end(struct super_block *sb);
3242
3243 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
3244 struct ext4_io_end *io_end)
3245 {
3246 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
3247 io_end->flag |= EXT4_IO_END_UNWRITTEN;
3248 atomic_inc(&EXT4_I(inode)->i_unwritten);
3249 }
3250 }
3251
3252 static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
3253 {
3254 struct inode *inode = io_end->inode;
3255
3256 if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
3257 io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
3258 /* Wake up anyone waiting on unwritten extent conversion */
3259 if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
3260 wake_up_all(ext4_ioend_wq(inode));
3261 }
3262 }
3263
3264 static inline bool ext4_aligned_io(struct inode *inode, loff_t off, loff_t len)
3265 {
3266 int blksize = 1 << inode->i_blkbits;
3267
3268 return IS_ALIGNED(off, blksize) && IS_ALIGNED(len, blksize);
3269 }
3270
3271 #endif /* __KERNEL__ */
3272
3273 #define EFSBADCRC EBADMSG /* Bad CRC detected */
3274 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
3275
3276 #endif /* _EXT4_H */
This page took 0.131638 seconds and 5 git commands to generate.