?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
irqchip: Fix dependencies for archs w/o HAS_IOMEM
[linux/fpc-iii.git] / fs / ext4 / ext4.h
blobcc7ca4e87144a540332213ce03b63e80e601f40e
1 /*
2 * ext4.h
3 *
4 * Copyright (C) 1992, 1993, 1994, 1995
5 * Remy Card (card@masi.ibp.fr)
6 * Laboratoire MASI - Institut Blaise Pascal
7 * Universite Pierre et Marie Curie (Paris VI)
8 *
9 * from
10 *
11 * linux/include/linux/minix_fs.h
12 *
13 * Copyright (C) 1991, 1992 Linus Torvalds
14 */
15
16 #ifndef _EXT4_H
17 #define _EXT4_H
18
19 #include <linux/types.h>
20 #include <linux/blkdev.h>
21 #include <linux/magic.h>
22 #include <linux/jbd2.h>
23 #include <linux/quota.h>
24 #include <linux/rwsem.h>
25 #include <linux/rbtree.h>
26 #include <linux/seqlock.h>
27 #include <linux/mutex.h>
28 #include <linux/timer.h>
29 #include <linux/version.h>
30 #include <linux/wait.h>
31 #include <linux/blockgroup_lock.h>
32 #include <linux/percpu_counter.h>
33 #include <linux/ratelimit.h>
34 #include <crypto/hash.h>
35 #include <linux/falloc.h>
36 #ifdef __KERNEL__
37 #include <linux/compat.h>
38 #endif
39
40 /*
41 * The fourth extended filesystem constants/structures
42 */
43
44 /*
45 * Define EXT4FS_DEBUG to produce debug messages
46 */
47 #undef EXT4FS_DEBUG
48
49 /*
50 * Debug code
51 */
52 #ifdef EXT4FS_DEBUG
53 #define ext4_debug(f, a...) \
54 do { \
55 printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
56 __FILE__, __LINE__, __func__); \
57 printk(KERN_DEBUG f, ## a); \
58 } while (0)
59 #else
60 #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
61 #endif
62
63 /*
64 * Turn on EXT_DEBUG to get lots of info about extents operations.
65 */
66 #define EXT_DEBUG__
67 #ifdef EXT_DEBUG
68 #define ext_debug(fmt, ...) printk(fmt, ##__VA_ARGS__)
69 #else
70 #define ext_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
71 #endif
72
73 /* data type for block offset of block group */
74 typedef int ext4_grpblk_t;
75
76 /* data type for filesystem-wide blocks number */
77 typedef unsigned long long ext4_fsblk_t;
78
79 /* data type for file logical block number */
80 typedef __u32 ext4_lblk_t;
81
82 /* data type for block group number */
83 typedef unsigned int ext4_group_t;
84
85 enum SHIFT_DIRECTION {
86 SHIFT_LEFT = 0,
87 SHIFT_RIGHT,
88 };
89
90 /*
91 * Flags used in mballoc's allocation_context flags field.
92 *
93 * Also used to show what's going on for debugging purposes when the
94 * flag field is exported via the traceport interface
95 */
96
97 /* prefer goal again. length */
98 #define EXT4_MB_HINT_MERGE 0x0001
99 /* blocks already reserved */
100 #define EXT4_MB_HINT_RESERVED 0x0002
101 /* metadata is being allocated */
102 #define EXT4_MB_HINT_METADATA 0x0004
103 /* first blocks in the file */
104 #define EXT4_MB_HINT_FIRST 0x0008
105 /* search for the best chunk */
106 #define EXT4_MB_HINT_BEST 0x0010
107 /* data is being allocated */
108 #define EXT4_MB_HINT_DATA 0x0020
109 /* don't preallocate (for tails) */
110 #define EXT4_MB_HINT_NOPREALLOC 0x0040
111 /* allocate for locality group */
112 #define EXT4_MB_HINT_GROUP_ALLOC 0x0080
113 /* allocate goal blocks or none */
114 #define EXT4_MB_HINT_GOAL_ONLY 0x0100
115 /* goal is meaningful */
116 #define EXT4_MB_HINT_TRY_GOAL 0x0200
117 /* blocks already pre-reserved by delayed allocation */
118 #define EXT4_MB_DELALLOC_RESERVED 0x0400
119 /* We are doing stream allocation */
120 #define EXT4_MB_STREAM_ALLOC 0x0800
121 /* Use reserved root blocks if needed */
122 #define EXT4_MB_USE_ROOT_BLOCKS 0x1000
123 /* Use blocks from reserved pool */
124 #define EXT4_MB_USE_RESERVED 0x2000
125
126 struct ext4_allocation_request {
127 /* target inode for block we're allocating */
128 struct inode *inode;
129 /* how many blocks we want to allocate */
130 unsigned int len;
131 /* logical block in target inode */
132 ext4_lblk_t logical;
133 /* the closest logical allocated block to the left */
134 ext4_lblk_t lleft;
135 /* the closest logical allocated block to the right */
136 ext4_lblk_t lright;
137 /* phys. target (a hint) */
138 ext4_fsblk_t goal;
139 /* phys. block for the closest logical allocated block to the left */
140 ext4_fsblk_t pleft;
141 /* phys. block for the closest logical allocated block to the right */
142 ext4_fsblk_t pright;
143 /* flags. see above EXT4_MB_HINT_* */
144 unsigned int flags;
145 };
146
147 /*
148 * Logical to physical block mapping, used by ext4_map_blocks()
149 *
150 * This structure is used to pass requests into ext4_map_blocks() as
151 * well as to store the information returned by ext4_map_blocks(). It
152 * takes less room on the stack than a struct buffer_head.
153 */
154 #define EXT4_MAP_NEW (1 << BH_New)
155 #define EXT4_MAP_MAPPED (1 << BH_Mapped)
156 #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten)
157 #define EXT4_MAP_BOUNDARY (1 << BH_Boundary)
158 #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
159 EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY)
160
161 struct ext4_map_blocks {
162 ext4_fsblk_t m_pblk;
163 ext4_lblk_t m_lblk;
164 unsigned int m_len;
165 unsigned int m_flags;
166 };
167
168 /*
169 * Flags for ext4_io_end->flags
170 */
171 #define EXT4_IO_END_UNWRITTEN 0x0001
172
173 /*
174 * For converting unwritten extents on a work queue. 'handle' is used for
175 * buffered writeback.
176 */
177 typedef struct ext4_io_end {
178 struct list_head list; /* per-file finished IO list */
179 handle_t *handle; /* handle reserved for extent
180 * conversion */
181 struct inode *inode; /* file being written to */
182 struct bio *bio; /* Linked list of completed
183 * bios covering the extent */
184 unsigned int flag; /* unwritten or not */
185 loff_t offset; /* offset in the file */
186 ssize_t size; /* size of the extent */
187 atomic_t count; /* reference counter */
188 } ext4_io_end_t;
189
190 struct ext4_io_submit {
191 struct writeback_control *io_wbc;
192 struct bio *io_bio;
193 ext4_io_end_t *io_end;
194 sector_t io_next_block;
195 };
196
197 /*
198 * Special inodes numbers
199 */
200 #define EXT4_BAD_INO 1 /* Bad blocks inode */
201 #define EXT4_ROOT_INO 2 /* Root inode */
202 #define EXT4_USR_QUOTA_INO 3 /* User quota inode */
203 #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
204 #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
205 #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
206 #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
207 #define EXT4_JOURNAL_INO 8 /* Journal inode */
208
209 /* First non-reserved inode for old ext4 filesystems */
210 #define EXT4_GOOD_OLD_FIRST_INO 11
211
212 /*
213 * Maximal count of links to a file
214 */
215 #define EXT4_LINK_MAX 65000
216
217 /*
218 * Macro-instructions used to manage several block sizes
219 */
220 #define EXT4_MIN_BLOCK_SIZE 1024
221 #define EXT4_MAX_BLOCK_SIZE 65536
222 #define EXT4_MIN_BLOCK_LOG_SIZE 10
223 #define EXT4_MAX_BLOCK_LOG_SIZE 16
224 #ifdef __KERNEL__
225 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
226 #else
227 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
228 #endif
229 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
230 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
231 EXT4_SB(s)->s_cluster_bits)
232 #ifdef __KERNEL__
233 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
234 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
235 #else
236 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
237 #endif
238 #ifdef __KERNEL__
239 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
240 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
241 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
242 #else
243 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
244 EXT4_GOOD_OLD_INODE_SIZE : \
245 (s)->s_inode_size)
246 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
247 EXT4_GOOD_OLD_FIRST_INO : \
248 (s)->s_first_ino)
249 #endif
250 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
251
252 /* Translate a block number to a cluster number */
253 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
254 /* Translate a cluster number to a block number */
255 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
256 /* Translate # of blks to # of clusters */
257 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
258 (sbi)->s_cluster_bits)
259 /* Mask out the low bits to get the starting block of the cluster */
260 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
261 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
262 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
263 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
264 /* Get the cluster offset */
265 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
266 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
267 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
268 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
269
270 /*
271 * Structure of a blocks group descriptor
272 */
273 struct ext4_group_desc
274 {
275 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
276 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
277 __le32 bg_inode_table_lo; /* Inodes table block */
278 __le16 bg_free_blocks_count_lo;/* Free blocks count */
279 __le16 bg_free_inodes_count_lo;/* Free inodes count */
280 __le16 bg_used_dirs_count_lo; /* Directories count */
281 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
282 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
283 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
284 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
285 __le16 bg_itable_unused_lo; /* Unused inodes count */
286 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
287 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
288 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
289 __le32 bg_inode_table_hi; /* Inodes table block MSB */
290 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
291 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
292 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
293 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
294 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
295 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
296 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
297 __u32 bg_reserved;
298 };
299
300 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
301 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
302 sizeof(__le16))
303 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
304 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
305 sizeof(__le16))
306
307 /*
308 * Structure of a flex block group info
309 */
310
311 struct flex_groups {
312 atomic64_t free_clusters;
313 atomic_t free_inodes;
314 atomic_t used_dirs;
315 };
316
317 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
318 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
319 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
320
321 /*
322 * Macro-instructions used to manage group descriptors
323 */
324 #define EXT4_MIN_DESC_SIZE 32
325 #define EXT4_MIN_DESC_SIZE_64BIT 64
326 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
327 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
328 #ifdef __KERNEL__
329 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
330 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
331 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
332 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
333 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
334 #else
335 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
336 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
337 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
338 #endif
339
340 /*
341 * Constants relative to the data blocks
342 */
343 #define EXT4_NDIR_BLOCKS 12
344 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
345 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
346 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
347 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
348
349 /*
350 * Inode flags
351 */
352 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
353 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
354 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
355 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
356 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
357 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
358 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
359 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
360 /* Reserved for compression usage... */
361 #define EXT4_DIRTY_FL 0x00000100
362 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
363 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
364 /* nb: was previously EXT2_ECOMPR_FL */
365 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */
366 /* End compression flags --- maybe not all used */
367 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
368 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
369 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
370 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
371 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
372 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
373 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
374 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
375 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
376 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
377 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
378 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
379 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
380
381 #define EXT4_FL_USER_VISIBLE 0x004BDFFF /* User visible flags */
382 #define EXT4_FL_USER_MODIFIABLE 0x004380FF /* User modifiable flags */
383
384 /* Flags that should be inherited by new inodes from their parent. */
385 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
386 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
387 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
388 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
389
390 /* Flags that are appropriate for regular files (all but dir-specific ones). */
391 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
392
393 /* Flags that are appropriate for non-directories/regular files. */
394 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
395
396 /* Mask out flags that are inappropriate for the given type of inode. */
397 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
398 {
399 if (S_ISDIR(mode))
400 return flags;
401 else if (S_ISREG(mode))
402 return flags & EXT4_REG_FLMASK;
403 else
404 return flags & EXT4_OTHER_FLMASK;
405 }
406
407 /*
408 * Inode flags used for atomic set/get
409 */
410 enum {
411 EXT4_INODE_SECRM = 0, /* Secure deletion */
412 EXT4_INODE_UNRM = 1, /* Undelete */
413 EXT4_INODE_COMPR = 2, /* Compress file */
414 EXT4_INODE_SYNC = 3, /* Synchronous updates */
415 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
416 EXT4_INODE_APPEND = 5, /* writes to file may only append */
417 EXT4_INODE_NODUMP = 6, /* do not dump file */
418 EXT4_INODE_NOATIME = 7, /* do not update atime */
419 /* Reserved for compression usage... */
420 EXT4_INODE_DIRTY = 8,
421 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
422 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
423 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */
424 /* End compression flags --- maybe not all used */
425 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
426 EXT4_INODE_IMAGIC = 13, /* AFS directory */
427 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
428 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
429 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
430 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
431 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
432 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
433 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
434 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
435 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
436 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */
437 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
438 };
439
440 /*
441 * Since it's pretty easy to mix up bit numbers and hex values, we use a
442 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
443 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
444 * any extra space in the compiled kernel image, otherwise, the build will fail.
445 * It's important that these values are the same, since we are using
446 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
447 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
448 * values found in ext2, ext3 and ext4 filesystems, and of course the values
449 * defined in e2fsprogs.
450 *
451 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
452 */
453 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
454 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
455
456 static inline void ext4_check_flag_values(void)
457 {
458 CHECK_FLAG_VALUE(SECRM);
459 CHECK_FLAG_VALUE(UNRM);
460 CHECK_FLAG_VALUE(COMPR);
461 CHECK_FLAG_VALUE(SYNC);
462 CHECK_FLAG_VALUE(IMMUTABLE);
463 CHECK_FLAG_VALUE(APPEND);
464 CHECK_FLAG_VALUE(NODUMP);
465 CHECK_FLAG_VALUE(NOATIME);
466 CHECK_FLAG_VALUE(DIRTY);
467 CHECK_FLAG_VALUE(COMPRBLK);
468 CHECK_FLAG_VALUE(NOCOMPR);
469 CHECK_FLAG_VALUE(ENCRYPT);
470 CHECK_FLAG_VALUE(INDEX);
471 CHECK_FLAG_VALUE(IMAGIC);
472 CHECK_FLAG_VALUE(JOURNAL_DATA);
473 CHECK_FLAG_VALUE(NOTAIL);
474 CHECK_FLAG_VALUE(DIRSYNC);
475 CHECK_FLAG_VALUE(TOPDIR);
476 CHECK_FLAG_VALUE(HUGE_FILE);
477 CHECK_FLAG_VALUE(EXTENTS);
478 CHECK_FLAG_VALUE(EA_INODE);
479 CHECK_FLAG_VALUE(EOFBLOCKS);
480 CHECK_FLAG_VALUE(INLINE_DATA);
481 CHECK_FLAG_VALUE(PROJINHERIT);
482 CHECK_FLAG_VALUE(RESERVED);
483 }
484
485 /* Used to pass group descriptor data when online resize is done */
486 struct ext4_new_group_input {
487 __u32 group; /* Group number for this data */
488 __u64 block_bitmap; /* Absolute block number of block bitmap */
489 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
490 __u64 inode_table; /* Absolute block number of inode table start */
491 __u32 blocks_count; /* Total number of blocks in this group */
492 __u16 reserved_blocks; /* Number of reserved blocks in this group */
493 __u16 unused;
494 };
495
496 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
497 struct compat_ext4_new_group_input {
498 u32 group;
499 compat_u64 block_bitmap;
500 compat_u64 inode_bitmap;
501 compat_u64 inode_table;
502 u32 blocks_count;
503 u16 reserved_blocks;
504 u16 unused;
505 };
506 #endif
507
508 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
509 struct ext4_new_group_data {
510 __u32 group;
511 __u64 block_bitmap;
512 __u64 inode_bitmap;
513 __u64 inode_table;
514 __u32 blocks_count;
515 __u16 reserved_blocks;
516 __u16 unused;
517 __u32 free_blocks_count;
518 };
519
520 /* Indexes used to index group tables in ext4_new_group_data */
521 enum {
522 BLOCK_BITMAP = 0, /* block bitmap */
523 INODE_BITMAP, /* inode bitmap */
524 INODE_TABLE, /* inode tables */
525 GROUP_TABLE_COUNT,
526 };
527
528 /*
529 * Flags used by ext4_map_blocks()
530 */
531 /* Allocate any needed blocks and/or convert an unwritten
532 extent to be an initialized ext4 */
533 #define EXT4_GET_BLOCKS_CREATE 0x0001
534 /* Request the creation of an unwritten extent */
535 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002
536 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\
537 EXT4_GET_BLOCKS_CREATE)
538 /* Caller is from the delayed allocation writeout path
539 * finally doing the actual allocation of delayed blocks */
540 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
541 /* caller is from the direct IO path, request to creation of an
542 unwritten extents if not allocated, split the unwritten
543 extent if blocks has been preallocated already*/
544 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
545 #define EXT4_GET_BLOCKS_CONVERT 0x0010
546 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
547 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
548 /* Convert extent to initialized after IO complete */
549 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
550 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
551 /* Eventual metadata allocation (due to growing extent tree)
552 * should not fail, so try to use reserved blocks for that.*/
553 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
554 /* Don't normalize allocation size (used for fallocate) */
555 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
556 /* Request will not result in inode size update (user for fallocate) */
557 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080
558 /* Do not take i_data_sem locking in ext4_map_blocks */
559 #define EXT4_GET_BLOCKS_NO_LOCK 0x0100
560 /* Convert written extents to unwritten */
561 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0200
562
563 /*
564 * The bit position of these flags must not overlap with any of the
565 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(),
566 * read_extent_tree_block(), ext4_split_extent_at(),
567 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
568 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
569 * caching the extents when reading from the extent tree while a
570 * truncate or punch hole operation is in progress.
571 */
572 #define EXT4_EX_NOCACHE 0x40000000
573 #define EXT4_EX_FORCE_CACHE 0x20000000
574
575 /*
576 * Flags used by ext4_free_blocks
577 */
578 #define EXT4_FREE_BLOCKS_METADATA 0x0001
579 #define EXT4_FREE_BLOCKS_FORGET 0x0002
580 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
581 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
582 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
583 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
584
585 /* Encryption algorithms */
586 #define EXT4_ENCRYPTION_MODE_INVALID 0
587 #define EXT4_ENCRYPTION_MODE_AES_256_XTS 1
588 #define EXT4_ENCRYPTION_MODE_AES_256_GCM 2
589 #define EXT4_ENCRYPTION_MODE_AES_256_CBC 3
590 #define EXT4_ENCRYPTION_MODE_AES_256_CTS 4
591
592 #include "ext4_crypto.h"
593
594 /*
595 * ioctl commands
596 */
597 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
598 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
599 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
600 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
601 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
602 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
603 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
604 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
605 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
606 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
607 #define EXT4_IOC_MIGRATE _IO('f', 9)
608 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
609 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
610 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
611 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
612 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
613 #define EXT4_IOC_SWAP_BOOT _IO('f', 17)
614 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18)
615 #define EXT4_IOC_SET_ENCRYPTION_POLICY _IOR('f', 19, struct ext4_encryption_policy)
616 #define EXT4_IOC_GET_ENCRYPTION_PWSALT _IOW('f', 20, __u8[16])
617 #define EXT4_IOC_GET_ENCRYPTION_POLICY _IOW('f', 21, struct ext4_encryption_policy)
618
619 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
620 /*
621 * ioctl commands in 32 bit emulation
622 */
623 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
624 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
625 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
626 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
627 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
628 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
629 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
630 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
631 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
632 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
633 #endif
634
635 /* Max physical block we can address w/o extents */
636 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
637
638 /*
639 * Structure of an inode on the disk
640 */
641 struct ext4_inode {
642 __le16 i_mode; /* File mode */
643 __le16 i_uid; /* Low 16 bits of Owner Uid */
644 __le32 i_size_lo; /* Size in bytes */
645 __le32 i_atime; /* Access time */
646 __le32 i_ctime; /* Inode Change time */
647 __le32 i_mtime; /* Modification time */
648 __le32 i_dtime; /* Deletion Time */
649 __le16 i_gid; /* Low 16 bits of Group Id */
650 __le16 i_links_count; /* Links count */
651 __le32 i_blocks_lo; /* Blocks count */
652 __le32 i_flags; /* File flags */
653 union {
654 struct {
655 __le32 l_i_version;
656 } linux1;
657 struct {
658 __u32 h_i_translator;
659 } hurd1;
660 struct {
661 __u32 m_i_reserved1;
662 } masix1;
663 } osd1; /* OS dependent 1 */
664 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
665 __le32 i_generation; /* File version (for NFS) */
666 __le32 i_file_acl_lo; /* File ACL */
667 __le32 i_size_high;
668 __le32 i_obso_faddr; /* Obsoleted fragment address */
669 union {
670 struct {
671 __le16 l_i_blocks_high; /* were l_i_reserved1 */
672 __le16 l_i_file_acl_high;
673 __le16 l_i_uid_high; /* these 2 fields */
674 __le16 l_i_gid_high; /* were reserved2[0] */
675 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
676 __le16 l_i_reserved;
677 } linux2;
678 struct {
679 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
680 __u16 h_i_mode_high;
681 __u16 h_i_uid_high;
682 __u16 h_i_gid_high;
683 __u32 h_i_author;
684 } hurd2;
685 struct {
686 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
687 __le16 m_i_file_acl_high;
688 __u32 m_i_reserved2[2];
689 } masix2;
690 } osd2; /* OS dependent 2 */
691 __le16 i_extra_isize;
692 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
693 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
694 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
695 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
696 __le32 i_crtime; /* File Creation time */
697 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
698 __le32 i_version_hi; /* high 32 bits for 64-bit version */
699 __le32 i_projid; /* Project ID */
700 };
701
702 struct move_extent {
703 __u32 reserved; /* should be zero */
704 __u32 donor_fd; /* donor file descriptor */
705 __u64 orig_start; /* logical start offset in block for orig */
706 __u64 donor_start; /* logical start offset in block for donor */
707 __u64 len; /* block length to be moved */
708 __u64 moved_len; /* moved block length */
709 };
710
711 #define EXT4_EPOCH_BITS 2
712 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
713 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
714
715 /*
716 * Extended fields will fit into an inode if the filesystem was formatted
717 * with large inodes (-I 256 or larger) and there are not currently any EAs
718 * consuming all of the available space. For new inodes we always reserve
719 * enough space for the kernel's known extended fields, but for inodes
720 * created with an old kernel this might not have been the case. None of
721 * the extended inode fields is critical for correct filesystem operation.
722 * This macro checks if a certain field fits in the inode. Note that
723 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
724 */
725 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
726 ((offsetof(typeof(*ext4_inode), field) + \
727 sizeof((ext4_inode)->field)) \
728 <= (EXT4_GOOD_OLD_INODE_SIZE + \
729 (einode)->i_extra_isize)) \
730
731 /*
732 * We use an encoding that preserves the times for extra epoch "00":
733 *
734 * extra msb of adjust for signed
735 * epoch 32-bit 32-bit tv_sec to
736 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
737 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
738 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
739 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
740 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
741 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
742 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
743 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
744 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
745 *
746 * Note that previous versions of the kernel on 64-bit systems would
747 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
748 * 1970. e2fsck will correct this, assuming that it is run on the
749 * affected filesystem before 2242.
750 */
751
752 static inline __le32 ext4_encode_extra_time(struct timespec *time)
753 {
754 u32 extra = sizeof(time->tv_sec) > 4 ?
755 ((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK : 0;
756 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
757 }
758
759 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
760 {
761 if (unlikely(sizeof(time->tv_sec) > 4 &&
762 (extra & cpu_to_le32(EXT4_EPOCH_MASK)))) {
763 #if LINUX_VERSION_CODE < KERNEL_VERSION(4,20,0)
764 /* Handle legacy encoding of pre-1970 dates with epoch
765 * bits 1,1. We assume that by kernel version 4.20,
766 * everyone will have run fsck over the affected
767 * filesystems to correct the problem. (This
768 * backwards compatibility may be removed before this
769 * time, at the discretion of the ext4 developers.)
770 */
771 u64 extra_bits = le32_to_cpu(extra) & EXT4_EPOCH_MASK;
772 if (extra_bits == 3 && ((time->tv_sec) & 0x80000000) != 0)
773 extra_bits = 0;
774 time->tv_sec += extra_bits << 32;
775 #else
776 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
777 #endif
778 }
779 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
780 }
781
782 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
783 do { \
784 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
785 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
786 (raw_inode)->xtime ## _extra = \
787 ext4_encode_extra_time(&(inode)->xtime); \
788 } while (0)
789
790 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
791 do { \
792 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
793 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
794 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
795 (raw_inode)->xtime ## _extra = \
796 ext4_encode_extra_time(&(einode)->xtime); \
797 } while (0)
798
799 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
800 do { \
801 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
802 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
803 ext4_decode_extra_time(&(inode)->xtime, \
804 raw_inode->xtime ## _extra); \
805 else \
806 (inode)->xtime.tv_nsec = 0; \
807 } while (0)
808
809 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
810 do { \
811 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
812 (einode)->xtime.tv_sec = \
813 (signed)le32_to_cpu((raw_inode)->xtime); \
814 else \
815 (einode)->xtime.tv_sec = 0; \
816 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
817 ext4_decode_extra_time(&(einode)->xtime, \
818 raw_inode->xtime ## _extra); \
819 else \
820 (einode)->xtime.tv_nsec = 0; \
821 } while (0)
822
823 #define i_disk_version osd1.linux1.l_i_version
824
825 #if defined(__KERNEL__) || defined(__linux__)
826 #define i_reserved1 osd1.linux1.l_i_reserved1
827 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
828 #define i_blocks_high osd2.linux2.l_i_blocks_high
829 #define i_uid_low i_uid
830 #define i_gid_low i_gid
831 #define i_uid_high osd2.linux2.l_i_uid_high
832 #define i_gid_high osd2.linux2.l_i_gid_high
833 #define i_checksum_lo osd2.linux2.l_i_checksum_lo
834
835 #elif defined(__GNU__)
836
837 #define i_translator osd1.hurd1.h_i_translator
838 #define i_uid_high osd2.hurd2.h_i_uid_high
839 #define i_gid_high osd2.hurd2.h_i_gid_high
840 #define i_author osd2.hurd2.h_i_author
841
842 #elif defined(__masix__)
843
844 #define i_reserved1 osd1.masix1.m_i_reserved1
845 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
846 #define i_reserved2 osd2.masix2.m_i_reserved2
847
848 #endif /* defined(__KERNEL__) || defined(__linux__) */
849
850 #include "extents_status.h"
851
852 /*
853 * fourth extended file system inode data in memory
854 */
855 struct ext4_inode_info {
856 __le32 i_data[15]; /* unconverted */
857 __u32 i_dtime;
858 ext4_fsblk_t i_file_acl;
859
860 /*
861 * i_block_group is the number of the block group which contains
862 * this file's inode. Constant across the lifetime of the inode,
863 * it is ued for making block allocation decisions - we try to
864 * place a file's data blocks near its inode block, and new inodes
865 * near to their parent directory's inode.
866 */
867 ext4_group_t i_block_group;
868 ext4_lblk_t i_dir_start_lookup;
869 #if (BITS_PER_LONG < 64)
870 unsigned long i_state_flags; /* Dynamic state flags */
871 #endif
872 unsigned long i_flags;
873
874 /*
875 * Extended attributes can be read independently of the main file
876 * data. Taking i_mutex even when reading would cause contention
877 * between readers of EAs and writers of regular file data, so
878 * instead we synchronize on xattr_sem when reading or changing
879 * EAs.
880 */
881 struct rw_semaphore xattr_sem;
882
883 struct list_head i_orphan; /* unlinked but open inodes */
884
885 /*
886 * i_disksize keeps track of what the inode size is ON DISK, not
887 * in memory. During truncate, i_size is set to the new size by
888 * the VFS prior to calling ext4_truncate(), but the filesystem won't
889 * set i_disksize to 0 until the truncate is actually under way.
890 *
891 * The intent is that i_disksize always represents the blocks which
892 * are used by this file. This allows recovery to restart truncate
893 * on orphans if we crash during truncate. We actually write i_disksize
894 * into the on-disk inode when writing inodes out, instead of i_size.
895 *
896 * The only time when i_disksize and i_size may be different is when
897 * a truncate is in progress. The only things which change i_disksize
898 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
899 */
900 loff_t i_disksize;
901
902 /*
903 * i_data_sem is for serialising ext4_truncate() against
904 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
905 * data tree are chopped off during truncate. We can't do that in
906 * ext4 because whenever we perform intermediate commits during
907 * truncate, the inode and all the metadata blocks *must* be in a
908 * consistent state which allows truncation of the orphans to restart
909 * during recovery. Hence we must fix the get_block-vs-truncate race
910 * by other means, so we have i_data_sem.
911 */
912 struct rw_semaphore i_data_sem;
913 struct inode vfs_inode;
914 struct jbd2_inode *jinode;
915
916 spinlock_t i_raw_lock; /* protects updates to the raw inode */
917
918 /*
919 * File creation time. Its function is same as that of
920 * struct timespec i_{a,c,m}time in the generic inode.
921 */
922 struct timespec i_crtime;
923
924 /* mballoc */
925 struct list_head i_prealloc_list;
926 spinlock_t i_prealloc_lock;
927
928 /* extents status tree */
929 struct ext4_es_tree i_es_tree;
930 rwlock_t i_es_lock;
931 struct list_head i_es_list;
932 unsigned int i_es_all_nr; /* protected by i_es_lock */
933 unsigned int i_es_shk_nr; /* protected by i_es_lock */
934 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for
935 extents to shrink. Protected by
936 i_es_lock */
937
938 /* ialloc */
939 ext4_group_t i_last_alloc_group;
940
941 /* allocation reservation info for delalloc */
942 /* In case of bigalloc, these refer to clusters rather than blocks */
943 unsigned int i_reserved_data_blocks;
944 unsigned int i_reserved_meta_blocks;
945 unsigned int i_allocated_meta_blocks;
946 ext4_lblk_t i_da_metadata_calc_last_lblock;
947 int i_da_metadata_calc_len;
948
949 /* on-disk additional length */
950 __u16 i_extra_isize;
951
952 /* Indicate the inline data space. */
953 u16 i_inline_off;
954 u16 i_inline_size;
955
956 #ifdef CONFIG_QUOTA
957 /* quota space reservation, managed internally by quota code */
958 qsize_t i_reserved_quota;
959 #endif
960
961 /* Lock protecting lists below */
962 spinlock_t i_completed_io_lock;
963 /*
964 * Completed IOs that need unwritten extents handling and have
965 * transaction reserved
966 */
967 struct list_head i_rsv_conversion_list;
968 /*
969 * Completed IOs that need unwritten extents handling and don't have
970 * transaction reserved
971 */
972 atomic_t i_ioend_count; /* Number of outstanding io_end structs */
973 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
974 struct work_struct i_rsv_conversion_work;
975
976 spinlock_t i_block_reservation_lock;
977
978 /*
979 * Transactions that contain inode's metadata needed to complete
980 * fsync and fdatasync, respectively.
981 */
982 tid_t i_sync_tid;
983 tid_t i_datasync_tid;
984
985 #ifdef CONFIG_QUOTA
986 struct dquot *i_dquot[MAXQUOTAS];
987 #endif
988
989 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
990 __u32 i_csum_seed;
991
992 #ifdef CONFIG_EXT4_FS_ENCRYPTION
993 /* Encryption params */
994 struct ext4_crypt_info *i_crypt_info;
995 #endif
996 };
997
998 /*
999 * File system states
1000 */
1001 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
1002 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
1003 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
1004
1005 /*
1006 * Misc. filesystem flags
1007 */
1008 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
1009 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
1010 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
1011
1012 /*
1013 * Mount flags set via mount options or defaults
1014 */
1015 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
1016 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
1017 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
1018 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
1019 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
1020 #define EXT4_MOUNT_ERRORS_MASK 0x00070
1021 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
1022 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
1023 #ifdef CONFIG_FS_DAX
1024 #define EXT4_MOUNT_DAX 0x00200 /* Direct Access */
1025 #else
1026 #define EXT4_MOUNT_DAX 0
1027 #endif
1028 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
1029 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
1030 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
1031 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
1032 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
1033 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
1034 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
1035 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
1036 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
1037 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
1038 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
1039 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
1040 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
1041 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
1042 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
1043 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1044 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
1045 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
1046 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
1047 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
1048 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
1049
1050 /*
1051 * Mount flags set either automatically (could not be set by mount option)
1052 * based on per file system feature or property or in special cases such as
1053 * distinguishing between explicit mount option definition and default.
1054 */
1055 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1056 specified delalloc */
1057 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1058 size of blocksize * 8
1059 blocks */
1060 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1061 file systems */
1062
1063 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly
1064 specified journal checksum */
1065
1066 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1067 ~EXT4_MOUNT_##opt
1068 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1069 EXT4_MOUNT_##opt
1070 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1071 EXT4_MOUNT_##opt)
1072
1073 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1074 ~EXT4_MOUNT2_##opt
1075 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1076 EXT4_MOUNT2_##opt
1077 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1078 EXT4_MOUNT2_##opt)
1079
1080 #define ext4_test_and_set_bit __test_and_set_bit_le
1081 #define ext4_set_bit __set_bit_le
1082 #define ext4_set_bit_atomic ext2_set_bit_atomic
1083 #define ext4_test_and_clear_bit __test_and_clear_bit_le
1084 #define ext4_clear_bit __clear_bit_le
1085 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
1086 #define ext4_test_bit test_bit_le
1087 #define ext4_find_next_zero_bit find_next_zero_bit_le
1088 #define ext4_find_next_bit find_next_bit_le
1089
1090 extern void ext4_set_bits(void *bm, int cur, int len);
1091
1092 /*
1093 * Maximal mount counts between two filesystem checks
1094 */
1095 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1096 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1097
1098 /*
1099 * Behaviour when detecting errors
1100 */
1101 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1102 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1103 #define EXT4_ERRORS_PANIC 3 /* Panic */
1104 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1105
1106 /* Metadata checksum algorithm codes */
1107 #define EXT4_CRC32C_CHKSUM 1
1108
1109 /*
1110 * Structure of the super block
1111 */
1112 struct ext4_super_block {
1113 /*00*/ __le32 s_inodes_count; /* Inodes count */
1114 __le32 s_blocks_count_lo; /* Blocks count */
1115 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1116 __le32 s_free_blocks_count_lo; /* Free blocks count */
1117 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1118 __le32 s_first_data_block; /* First Data Block */
1119 __le32 s_log_block_size; /* Block size */
1120 __le32 s_log_cluster_size; /* Allocation cluster size */
1121 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1122 __le32 s_clusters_per_group; /* # Clusters per group */
1123 __le32 s_inodes_per_group; /* # Inodes per group */
1124 __le32 s_mtime; /* Mount time */
1125 /*30*/ __le32 s_wtime; /* Write time */
1126 __le16 s_mnt_count; /* Mount count */
1127 __le16 s_max_mnt_count; /* Maximal mount count */
1128 __le16 s_magic; /* Magic signature */
1129 __le16 s_state; /* File system state */
1130 __le16 s_errors; /* Behaviour when detecting errors */
1131 __le16 s_minor_rev_level; /* minor revision level */
1132 /*40*/ __le32 s_lastcheck; /* time of last check */
1133 __le32 s_checkinterval; /* max. time between checks */
1134 __le32 s_creator_os; /* OS */
1135 __le32 s_rev_level; /* Revision level */
1136 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1137 __le16 s_def_resgid; /* Default gid for reserved blocks */
1138 /*
1139 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1140 *
1141 * Note: the difference between the compatible feature set and
1142 * the incompatible feature set is that if there is a bit set
1143 * in the incompatible feature set that the kernel doesn't
1144 * know about, it should refuse to mount the filesystem.
1145 *
1146 * e2fsck's requirements are more strict; if it doesn't know
1147 * about a feature in either the compatible or incompatible
1148 * feature set, it must abort and not try to meddle with
1149 * things it doesn't understand...
1150 */
1151 __le32 s_first_ino; /* First non-reserved inode */
1152 __le16 s_inode_size; /* size of inode structure */
1153 __le16 s_block_group_nr; /* block group # of this superblock */
1154 __le32 s_feature_compat; /* compatible feature set */
1155 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1156 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1157 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1158 /*78*/ char s_volume_name[16]; /* volume name */
1159 /*88*/ char s_last_mounted[64]; /* directory where last mounted */
1160 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1161 /*
1162 * Performance hints. Directory preallocation should only
1163 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1164 */
1165 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1166 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1167 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1168 /*
1169 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1170 */
1171 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1172 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1173 __le32 s_journal_dev; /* device number of journal file */
1174 __le32 s_last_orphan; /* start of list of inodes to delete */
1175 __le32 s_hash_seed[4]; /* HTREE hash seed */
1176 __u8 s_def_hash_version; /* Default hash version to use */
1177 __u8 s_jnl_backup_type;
1178 __le16 s_desc_size; /* size of group descriptor */
1179 /*100*/ __le32 s_default_mount_opts;
1180 __le32 s_first_meta_bg; /* First metablock block group */
1181 __le32 s_mkfs_time; /* When the filesystem was created */
1182 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1183 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1184 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1185 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1186 __le32 s_free_blocks_count_hi; /* Free blocks count */
1187 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1188 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1189 __le32 s_flags; /* Miscellaneous flags */
1190 __le16 s_raid_stride; /* RAID stride */
1191 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1192 __le64 s_mmp_block; /* Block for multi-mount protection */
1193 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1194 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1195 __u8 s_checksum_type; /* metadata checksum algorithm used */
1196 __u8 s_encryption_level; /* versioning level for encryption */
1197 __u8 s_reserved_pad; /* Padding to next 32bits */
1198 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1199 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1200 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1201 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1202 snapshot's future use */
1203 __le32 s_snapshot_list; /* inode number of the head of the
1204 on-disk snapshot list */
1205 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1206 __le32 s_error_count; /* number of fs errors */
1207 __le32 s_first_error_time; /* first time an error happened */
1208 __le32 s_first_error_ino; /* inode involved in first error */
1209 __le64 s_first_error_block; /* block involved of first error */
1210 __u8 s_first_error_func[32]; /* function where the error happened */
1211 __le32 s_first_error_line; /* line number where error happened */
1212 __le32 s_last_error_time; /* most recent time of an error */
1213 __le32 s_last_error_ino; /* inode involved in last error */
1214 __le32 s_last_error_line; /* line number where error happened */
1215 __le64 s_last_error_block; /* block involved of last error */
1216 __u8 s_last_error_func[32]; /* function where the error happened */
1217 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1218 __u8 s_mount_opts[64];
1219 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1220 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1221 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1222 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
1223 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */
1224 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */
1225 __le32 s_lpf_ino; /* Location of the lost+found inode */
1226 __le32 s_prj_quota_inum; /* inode for tracking project quota */
1227 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */
1228 __le32 s_reserved[98]; /* Padding to the end of the block */
1229 __le32 s_checksum; /* crc32c(superblock) */
1230 };
1231
1232 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1233
1234 #ifdef __KERNEL__
1235
1236 /*
1237 * run-time mount flags
1238 */
1239 #define EXT4_MF_MNTDIR_SAMPLED 0x0001
1240 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1241 #define EXT4_MF_TEST_DUMMY_ENCRYPTION 0x0004
1242
1243 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1244 #define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \
1245 EXT4_MF_TEST_DUMMY_ENCRYPTION))
1246 #else
1247 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1248 #endif
1249
1250 /* Number of quota types we support */
1251 #define EXT4_MAXQUOTAS 2
1252
1253 /*
1254 * fourth extended-fs super-block data in memory
1255 */
1256 struct ext4_sb_info {
1257 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1258 unsigned long s_inodes_per_block;/* Number of inodes per block */
1259 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1260 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1261 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1262 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1263 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1264 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1265 ext4_group_t s_groups_count; /* Number of groups in the fs */
1266 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1267 unsigned long s_overhead; /* # of fs overhead clusters */
1268 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1269 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1270 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1271 struct buffer_head * s_sbh; /* Buffer containing the super block */
1272 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1273 struct buffer_head **s_group_desc;
1274 unsigned int s_mount_opt;
1275 unsigned int s_mount_opt2;
1276 unsigned int s_mount_flags;
1277 unsigned int s_def_mount_opt;
1278 ext4_fsblk_t s_sb_block;
1279 atomic64_t s_resv_clusters;
1280 kuid_t s_resuid;
1281 kgid_t s_resgid;
1282 unsigned short s_mount_state;
1283 unsigned short s_pad;
1284 int s_addr_per_block_bits;
1285 int s_desc_per_block_bits;
1286 int s_inode_size;
1287 int s_first_ino;
1288 unsigned int s_inode_readahead_blks;
1289 unsigned int s_inode_goal;
1290 spinlock_t s_next_gen_lock;
1291 u32 s_next_generation;
1292 u32 s_hash_seed[4];
1293 int s_def_hash_version;
1294 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1295 struct percpu_counter s_freeclusters_counter;
1296 struct percpu_counter s_freeinodes_counter;
1297 struct percpu_counter s_dirs_counter;
1298 struct percpu_counter s_dirtyclusters_counter;
1299 struct blockgroup_lock *s_blockgroup_lock;
1300 struct proc_dir_entry *s_proc;
1301 struct kobject s_kobj;
1302 struct completion s_kobj_unregister;
1303 struct super_block *s_sb;
1304
1305 /* Journaling */
1306 struct journal_s *s_journal;
1307 struct list_head s_orphan;
1308 struct mutex s_orphan_lock;
1309 unsigned long s_resize_flags; /* Flags indicating if there
1310 is a resizer */
1311 unsigned long s_commit_interval;
1312 u32 s_max_batch_time;
1313 u32 s_min_batch_time;
1314 struct block_device *journal_bdev;
1315 #ifdef CONFIG_QUOTA
1316 char *s_qf_names[EXT4_MAXQUOTAS]; /* Names of quota files with journalled quota */
1317 int s_jquota_fmt; /* Format of quota to use */
1318 #endif
1319 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1320 struct rb_root system_blks;
1321
1322 #ifdef EXTENTS_STATS
1323 /* ext4 extents stats */
1324 unsigned long s_ext_min;
1325 unsigned long s_ext_max;
1326 unsigned long s_depth_max;
1327 spinlock_t s_ext_stats_lock;
1328 unsigned long s_ext_blocks;
1329 unsigned long s_ext_extents;
1330 #endif
1331
1332 /* for buddy allocator */
1333 struct ext4_group_info ***s_group_info;
1334 struct inode *s_buddy_cache;
1335 spinlock_t s_md_lock;
1336 unsigned short *s_mb_offsets;
1337 unsigned int *s_mb_maxs;
1338 unsigned int s_group_info_size;
1339
1340 /* tunables */
1341 unsigned long s_stripe;
1342 unsigned int s_mb_stream_request;
1343 unsigned int s_mb_max_to_scan;
1344 unsigned int s_mb_min_to_scan;
1345 unsigned int s_mb_stats;
1346 unsigned int s_mb_order2_reqs;
1347 unsigned int s_mb_group_prealloc;
1348 unsigned int s_max_dir_size_kb;
1349 /* where last allocation was done - for stream allocation */
1350 unsigned long s_mb_last_group;
1351 unsigned long s_mb_last_start;
1352
1353 /* stats for buddy allocator */
1354 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1355 atomic_t s_bal_success; /* we found long enough chunks */
1356 atomic_t s_bal_allocated; /* in blocks */
1357 atomic_t s_bal_ex_scanned; /* total extents scanned */
1358 atomic_t s_bal_goals; /* goal hits */
1359 atomic_t s_bal_breaks; /* too long searches */
1360 atomic_t s_bal_2orders; /* 2^order hits */
1361 spinlock_t s_bal_lock;
1362 unsigned long s_mb_buddies_generated;
1363 unsigned long long s_mb_generation_time;
1364 atomic_t s_mb_lost_chunks;
1365 atomic_t s_mb_preallocated;
1366 atomic_t s_mb_discarded;
1367 atomic_t s_lock_busy;
1368
1369 /* locality groups */
1370 struct ext4_locality_group __percpu *s_locality_groups;
1371
1372 /* for write statistics */
1373 unsigned long s_sectors_written_start;
1374 u64 s_kbytes_written;
1375
1376 /* the size of zero-out chunk */
1377 unsigned int s_extent_max_zeroout_kb;
1378
1379 unsigned int s_log_groups_per_flex;
1380 struct flex_groups *s_flex_groups;
1381 ext4_group_t s_flex_groups_allocated;
1382
1383 /* workqueue for reserved extent conversions (buffered io) */
1384 struct workqueue_struct *rsv_conversion_wq;
1385
1386 /* timer for periodic error stats printing */
1387 struct timer_list s_err_report;
1388
1389 /* Lazy inode table initialization info */
1390 struct ext4_li_request *s_li_request;
1391 /* Wait multiplier for lazy initialization thread */
1392 unsigned int s_li_wait_mult;
1393
1394 /* Kernel thread for multiple mount protection */
1395 struct task_struct *s_mmp_tsk;
1396
1397 /* record the last minlen when FITRIM is called. */
1398 atomic_t s_last_trim_minblks;
1399
1400 /* Reference to checksum algorithm driver via cryptoapi */
1401 struct crypto_shash *s_chksum_driver;
1402
1403 /* Precomputed FS UUID checksum for seeding other checksums */
1404 __u32 s_csum_seed;
1405
1406 /* Reclaim extents from extent status tree */
1407 struct shrinker s_es_shrinker;
1408 struct list_head s_es_list; /* List of inodes with reclaimable extents */
1409 long s_es_nr_inode;
1410 struct ext4_es_stats s_es_stats;
1411 struct mb_cache *s_mb_cache;
1412 spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1413
1414 /* Ratelimit ext4 messages. */
1415 struct ratelimit_state s_err_ratelimit_state;
1416 struct ratelimit_state s_warning_ratelimit_state;
1417 struct ratelimit_state s_msg_ratelimit_state;
1418 };
1419
1420 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1421 {
1422 return sb->s_fs_info;
1423 }
1424 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1425 {
1426 return container_of(inode, struct ext4_inode_info, vfs_inode);
1427 }
1428
1429 static inline struct timespec ext4_current_time(struct inode *inode)
1430 {
1431 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1432 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1433 }
1434
1435 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1436 {
1437 return ino == EXT4_ROOT_INO ||
1438 ino == EXT4_USR_QUOTA_INO ||
1439 ino == EXT4_GRP_QUOTA_INO ||
1440 ino == EXT4_BOOT_LOADER_INO ||
1441 ino == EXT4_JOURNAL_INO ||
1442 ino == EXT4_RESIZE_INO ||
1443 (ino >= EXT4_FIRST_INO(sb) &&
1444 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1445 }
1446
1447 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
1448 struct ext4_io_end *io_end)
1449 {
1450 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
1451 io_end->flag |= EXT4_IO_END_UNWRITTEN;
1452 atomic_inc(&EXT4_I(inode)->i_unwritten);
1453 }
1454 }
1455
1456 static inline ext4_io_end_t *ext4_inode_aio(struct inode *inode)
1457 {
1458 return inode->i_private;
1459 }
1460
1461 static inline void ext4_inode_aio_set(struct inode *inode, ext4_io_end_t *io)
1462 {
1463 inode->i_private = io;
1464 }
1465
1466 /*
1467 * Inode dynamic state flags
1468 */
1469 enum {
1470 EXT4_STATE_JDATA, /* journaled data exists */
1471 EXT4_STATE_NEW, /* inode is newly created */
1472 EXT4_STATE_XATTR, /* has in-inode xattrs */
1473 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1474 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1475 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1476 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
1477 EXT4_STATE_NEWENTRY, /* File just added to dir */
1478 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read
1479 nolocking */
1480 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1481 EXT4_STATE_ORDERED_MODE, /* data=ordered mode */
1482 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
1483 };
1484
1485 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1486 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1487 { \
1488 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1489 } \
1490 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1491 { \
1492 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1493 } \
1494 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1495 { \
1496 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1497 }
1498
1499 /* Add these declarations here only so that these functions can be
1500 * found by name. Otherwise, they are very hard to locate. */
1501 static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1502 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1503 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1504 EXT4_INODE_BIT_FNS(flag, flags, 0)
1505
1506 /* Add these declarations here only so that these functions can be
1507 * found by name. Otherwise, they are very hard to locate. */
1508 static inline int ext4_test_inode_state(struct inode *inode, int bit);
1509 static inline void ext4_set_inode_state(struct inode *inode, int bit);
1510 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1511 #if (BITS_PER_LONG < 64)
1512 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1513
1514 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1515 {
1516 (ei)->i_state_flags = 0;
1517 }
1518 #else
1519 EXT4_INODE_BIT_FNS(state, flags, 32)
1520
1521 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1522 {
1523 /* We depend on the fact that callers will set i_flags */
1524 }
1525 #endif
1526 #else
1527 /* Assume that user mode programs are passing in an ext4fs superblock, not
1528 * a kernel struct super_block. This will allow us to call the feature-test
1529 * macros from user land. */
1530 #define EXT4_SB(sb) (sb)
1531 #endif
1532
1533 /*
1534 * Returns true if the inode is inode is encrypted
1535 */
1536 static inline int ext4_encrypted_inode(struct inode *inode)
1537 {
1538 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1539 return ext4_test_inode_flag(inode, EXT4_INODE_ENCRYPT);
1540 #else
1541 return 0;
1542 #endif
1543 }
1544
1545 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1546
1547 /*
1548 * Codes for operating systems
1549 */
1550 #define EXT4_OS_LINUX 0
1551 #define EXT4_OS_HURD 1
1552 #define EXT4_OS_MASIX 2
1553 #define EXT4_OS_FREEBSD 3
1554 #define EXT4_OS_LITES 4
1555
1556 /*
1557 * Revision levels
1558 */
1559 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1560 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1561
1562 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1563 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1564
1565 #define EXT4_GOOD_OLD_INODE_SIZE 128
1566
1567 /*
1568 * Feature set definitions
1569 */
1570
1571 /* Use the ext4_{has,set,clear}_feature_* helpers; these will be removed */
1572 #define EXT4_HAS_COMPAT_FEATURE(sb,mask) \
1573 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1574 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \
1575 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1576 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \
1577 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1578 #define EXT4_SET_COMPAT_FEATURE(sb,mask) \
1579 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1580 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \
1581 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1582 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \
1583 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1584 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \
1585 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1586 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
1587 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1588 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \
1589 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1590
1591 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1592 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1593 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1594 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1595 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1596 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1597 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
1598
1599 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1600 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1601 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1602 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1603 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1604 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1605 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1606 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1607 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1608 /*
1609 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1610 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1611 * all other data structures' checksums. However, the METADATA_CSUM and
1612 * GDT_CSUM bits are mutually exclusive.
1613 */
1614 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
1615 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000
1616 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000
1617
1618 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1619 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1620 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1621 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1622 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1623 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1624 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1625 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1626 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1627 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1628 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1629 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000
1630 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
1631 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
1632 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000
1633
1634 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
1635 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1636 { \
1637 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1638 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
1639 } \
1640 static inline void ext4_set_feature_##name(struct super_block *sb) \
1641 { \
1642 EXT4_SB(sb)->s_es->s_feature_compat |= \
1643 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1644 } \
1645 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1646 { \
1647 EXT4_SB(sb)->s_es->s_feature_compat &= \
1648 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1649 }
1650
1651 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
1652 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1653 { \
1654 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1655 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
1656 } \
1657 static inline void ext4_set_feature_##name(struct super_block *sb) \
1658 { \
1659 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
1660 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1661 } \
1662 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1663 { \
1664 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
1665 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1666 }
1667
1668 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
1669 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1670 { \
1671 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1672 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
1673 } \
1674 static inline void ext4_set_feature_##name(struct super_block *sb) \
1675 { \
1676 EXT4_SB(sb)->s_es->s_feature_incompat |= \
1677 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1678 } \
1679 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1680 { \
1681 EXT4_SB(sb)->s_es->s_feature_incompat &= \
1682 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1683 }
1684
1685 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC)
1686 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES)
1687 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL)
1688 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR)
1689 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE)
1690 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX)
1691 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2)
1692
1693 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER)
1694 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE)
1695 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR)
1696 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE)
1697 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM)
1698 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK)
1699 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE)
1700 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA)
1701 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC)
1702 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM)
1703 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY)
1704 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT)
1705
1706 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION)
1707 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE)
1708 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER)
1709 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV)
1710 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG)
1711 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS)
1712 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT)
1713 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP)
1714 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG)
1715 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE)
1716 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA)
1717 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED)
1718 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR)
1719 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA)
1720 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT)
1721
1722 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1723 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1724 EXT4_FEATURE_INCOMPAT_META_BG)
1725 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1726 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1727 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1728
1729 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1730 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1731 EXT4_FEATURE_INCOMPAT_RECOVER| \
1732 EXT4_FEATURE_INCOMPAT_META_BG)
1733 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1734 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1735 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1736
1737 #define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1738 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1739 EXT4_FEATURE_INCOMPAT_RECOVER| \
1740 EXT4_FEATURE_INCOMPAT_META_BG| \
1741 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1742 EXT4_FEATURE_INCOMPAT_64BIT| \
1743 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1744 EXT4_FEATURE_INCOMPAT_MMP | \
1745 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
1746 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
1747 EXT4_FEATURE_INCOMPAT_CSUM_SEED)
1748 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1749 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1750 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1751 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1752 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1753 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1754 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1755 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1756 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1757 EXT4_FEATURE_RO_COMPAT_QUOTA)
1758
1759 #define EXTN_FEATURE_FUNCS(ver) \
1760 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
1761 { \
1762 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1763 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
1764 } \
1765 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
1766 { \
1767 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1768 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
1769 } \
1770 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
1771 { \
1772 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1773 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
1774 }
1775
1776 EXTN_FEATURE_FUNCS(2)
1777 EXTN_FEATURE_FUNCS(3)
1778 EXTN_FEATURE_FUNCS(4)
1779
1780 static inline bool ext4_has_compat_features(struct super_block *sb)
1781 {
1782 return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
1783 }
1784 static inline bool ext4_has_ro_compat_features(struct super_block *sb)
1785 {
1786 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
1787 }
1788 static inline bool ext4_has_incompat_features(struct super_block *sb)
1789 {
1790 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
1791 }
1792
1793 /*
1794 * Default values for user and/or group using reserved blocks
1795 */
1796 #define EXT4_DEF_RESUID 0
1797 #define EXT4_DEF_RESGID 0
1798
1799 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
1800
1801 /*
1802 * Default mount options
1803 */
1804 #define EXT4_DEFM_DEBUG 0x0001
1805 #define EXT4_DEFM_BSDGROUPS 0x0002
1806 #define EXT4_DEFM_XATTR_USER 0x0004
1807 #define EXT4_DEFM_ACL 0x0008
1808 #define EXT4_DEFM_UID16 0x0010
1809 #define EXT4_DEFM_JMODE 0x0060
1810 #define EXT4_DEFM_JMODE_DATA 0x0020
1811 #define EXT4_DEFM_JMODE_ORDERED 0x0040
1812 #define EXT4_DEFM_JMODE_WBACK 0x0060
1813 #define EXT4_DEFM_NOBARRIER 0x0100
1814 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1815 #define EXT4_DEFM_DISCARD 0x0400
1816 #define EXT4_DEFM_NODELALLOC 0x0800
1817
1818 /*
1819 * Default journal batch times
1820 */
1821 #define EXT4_DEF_MIN_BATCH_TIME 0
1822 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1823
1824 /*
1825 * Minimum number of groups in a flexgroup before we separate out
1826 * directories into the first block group of a flexgroup
1827 */
1828 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1829
1830 /*
1831 * Structure of a directory entry
1832 */
1833 #define EXT4_NAME_LEN 255
1834
1835 struct ext4_dir_entry {
1836 __le32 inode; /* Inode number */
1837 __le16 rec_len; /* Directory entry length */
1838 __le16 name_len; /* Name length */
1839 char name[EXT4_NAME_LEN]; /* File name */
1840 };
1841
1842 /*
1843 * The new version of the directory entry. Since EXT4 structures are
1844 * stored in intel byte order, and the name_len field could never be
1845 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1846 * file_type field.
1847 */
1848 struct ext4_dir_entry_2 {
1849 __le32 inode; /* Inode number */
1850 __le16 rec_len; /* Directory entry length */
1851 __u8 name_len; /* Name length */
1852 __u8 file_type;
1853 char name[EXT4_NAME_LEN]; /* File name */
1854 };
1855
1856 /*
1857 * This is a bogus directory entry at the end of each leaf block that
1858 * records checksums.
1859 */
1860 struct ext4_dir_entry_tail {
1861 __le32 det_reserved_zero1; /* Pretend to be unused */
1862 __le16 det_rec_len; /* 12 */
1863 __u8 det_reserved_zero2; /* Zero name length */
1864 __u8 det_reserved_ft; /* 0xDE, fake file type */
1865 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
1866 };
1867
1868 #define EXT4_DIRENT_TAIL(block, blocksize) \
1869 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1870 ((blocksize) - \
1871 sizeof(struct ext4_dir_entry_tail))))
1872
1873 /*
1874 * Ext4 directory file types. Only the low 3 bits are used. The
1875 * other bits are reserved for now.
1876 */
1877 #define EXT4_FT_UNKNOWN 0
1878 #define EXT4_FT_REG_FILE 1
1879 #define EXT4_FT_DIR 2
1880 #define EXT4_FT_CHRDEV 3
1881 #define EXT4_FT_BLKDEV 4
1882 #define EXT4_FT_FIFO 5
1883 #define EXT4_FT_SOCK 6
1884 #define EXT4_FT_SYMLINK 7
1885
1886 #define EXT4_FT_MAX 8
1887
1888 #define EXT4_FT_DIR_CSUM 0xDE
1889
1890 /*
1891 * EXT4_DIR_PAD defines the directory entries boundaries
1892 *
1893 * NOTE: It must be a multiple of 4
1894 */
1895 #define EXT4_DIR_PAD 4
1896 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1897 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1898 ~EXT4_DIR_ROUND)
1899 #define EXT4_MAX_REC_LEN ((1<<16)-1)
1900
1901 /*
1902 * If we ever get support for fs block sizes > page_size, we'll need
1903 * to remove the #if statements in the next two functions...
1904 */
1905 static inline unsigned int
1906 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1907 {
1908 unsigned len = le16_to_cpu(dlen);
1909
1910 #if (PAGE_CACHE_SIZE >= 65536)
1911 if (len == EXT4_MAX_REC_LEN || len == 0)
1912 return blocksize;
1913 return (len & 65532) | ((len & 3) << 16);
1914 #else
1915 return len;
1916 #endif
1917 }
1918
1919 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1920 {
1921 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1922 BUG();
1923 #if (PAGE_CACHE_SIZE >= 65536)
1924 if (len < 65536)
1925 return cpu_to_le16(len);
1926 if (len == blocksize) {
1927 if (blocksize == 65536)
1928 return cpu_to_le16(EXT4_MAX_REC_LEN);
1929 else
1930 return cpu_to_le16(0);
1931 }
1932 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1933 #else
1934 return cpu_to_le16(len);
1935 #endif
1936 }
1937
1938 /*
1939 * Hash Tree Directory indexing
1940 * (c) Daniel Phillips, 2001
1941 */
1942
1943 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
1944 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
1945 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1946 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1947
1948 /* Legal values for the dx_root hash_version field: */
1949
1950 #define DX_HASH_LEGACY 0
1951 #define DX_HASH_HALF_MD4 1
1952 #define DX_HASH_TEA 2
1953 #define DX_HASH_LEGACY_UNSIGNED 3
1954 #define DX_HASH_HALF_MD4_UNSIGNED 4
1955 #define DX_HASH_TEA_UNSIGNED 5
1956
1957 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
1958 const void *address, unsigned int length)
1959 {
1960 struct {
1961 struct shash_desc shash;
1962 char ctx[4];
1963 } desc;
1964 int err;
1965
1966 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
1967
1968 desc.shash.tfm = sbi->s_chksum_driver;
1969 desc.shash.flags = 0;
1970 *(u32 *)desc.ctx = crc;
1971
1972 err = crypto_shash_update(&desc.shash, address, length);
1973 BUG_ON(err);
1974
1975 return *(u32 *)desc.ctx;
1976 }
1977
1978 #ifdef __KERNEL__
1979
1980 /* hash info structure used by the directory hash */
1981 struct dx_hash_info
1982 {
1983 u32 hash;
1984 u32 minor_hash;
1985 int hash_version;
1986 u32 *seed;
1987 };
1988
1989
1990 /* 32 and 64 bit signed EOF for dx directories */
1991 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
1992 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
1993
1994
1995 /*
1996 * Control parameters used by ext4_htree_next_block
1997 */
1998 #define HASH_NB_ALWAYS 1
1999
2000 struct ext4_filename {
2001 const struct qstr *usr_fname;
2002 struct ext4_str disk_name;
2003 struct dx_hash_info hinfo;
2004 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2005 struct ext4_str crypto_buf;
2006 #endif
2007 };
2008
2009 #define fname_name(p) ((p)->disk_name.name)
2010 #define fname_len(p) ((p)->disk_name.len)
2011
2012 /*
2013 * Describe an inode's exact location on disk and in memory
2014 */
2015 struct ext4_iloc
2016 {
2017 struct buffer_head *bh;
2018 unsigned long offset;
2019 ext4_group_t block_group;
2020 };
2021
2022 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2023 {
2024 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2025 }
2026
2027 /*
2028 * This structure is stuffed into the struct file's private_data field
2029 * for directories. It is where we put information so that we can do
2030 * readdir operations in hash tree order.
2031 */
2032 struct dir_private_info {
2033 struct rb_root root;
2034 struct rb_node *curr_node;
2035 struct fname *extra_fname;
2036 loff_t last_pos;
2037 __u32 curr_hash;
2038 __u32 curr_minor_hash;
2039 __u32 next_hash;
2040 };
2041
2042 /* calculate the first block number of the group */
2043 static inline ext4_fsblk_t
2044 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2045 {
2046 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2047 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2048 }
2049
2050 /*
2051 * Special error return code only used by dx_probe() and its callers.
2052 */
2053 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
2054
2055 /*
2056 * Timeout and state flag for lazy initialization inode thread.
2057 */
2058 #define EXT4_DEF_LI_WAIT_MULT 10
2059 #define EXT4_DEF_LI_MAX_START_DELAY 5
2060 #define EXT4_LAZYINIT_QUIT 0x0001
2061 #define EXT4_LAZYINIT_RUNNING 0x0002
2062
2063 /*
2064 * Lazy inode table initialization info
2065 */
2066 struct ext4_lazy_init {
2067 unsigned long li_state;
2068 struct list_head li_request_list;
2069 struct mutex li_list_mtx;
2070 };
2071
2072 struct ext4_li_request {
2073 struct super_block *lr_super;
2074 struct ext4_sb_info *lr_sbi;
2075 ext4_group_t lr_next_group;
2076 struct list_head lr_request;
2077 unsigned long lr_next_sched;
2078 unsigned long lr_timeout;
2079 };
2080
2081 struct ext4_features {
2082 struct kobject f_kobj;
2083 struct completion f_kobj_unregister;
2084 };
2085
2086 /*
2087 * This structure will be used for multiple mount protection. It will be
2088 * written into the block number saved in the s_mmp_block field in the
2089 * superblock. Programs that check MMP should assume that if
2090 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2091 * to use the filesystem, regardless of how old the timestamp is.
2092 */
2093 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
2094 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2095 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
2096 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
2097
2098 struct mmp_struct {
2099 __le32 mmp_magic; /* Magic number for MMP */
2100 __le32 mmp_seq; /* Sequence no. updated periodically */
2101
2102 /*
2103 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2104 * purposes and do not affect the correctness of the algorithm
2105 */
2106 __le64 mmp_time; /* Time last updated */
2107 char mmp_nodename[64]; /* Node which last updated MMP block */
2108 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
2109
2110 /*
2111 * mmp_check_interval is used to verify if the MMP block has been
2112 * updated on the block device. The value is updated based on the
2113 * maximum time to write the MMP block during an update cycle.
2114 */
2115 __le16 mmp_check_interval;
2116
2117 __le16 mmp_pad1;
2118 __le32 mmp_pad2[226];
2119 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
2120 };
2121
2122 /* arguments passed to the mmp thread */
2123 struct mmpd_data {
2124 struct buffer_head *bh; /* bh from initial read_mmp_block() */
2125 struct super_block *sb; /* super block of the fs */
2126 };
2127
2128 /*
2129 * Check interval multiplier
2130 * The MMP block is written every update interval and initially checked every
2131 * update interval x the multiplier (the value is then adapted based on the
2132 * write latency). The reason is that writes can be delayed under load and we
2133 * don't want readers to incorrectly assume that the filesystem is no longer
2134 * in use.
2135 */
2136 #define EXT4_MMP_CHECK_MULT 2UL
2137
2138 /*
2139 * Minimum interval for MMP checking in seconds.
2140 */
2141 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
2142
2143 /*
2144 * Maximum interval for MMP checking in seconds.
2145 */
2146 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
2147
2148 /*
2149 * Function prototypes
2150 */
2151
2152 /*
2153 * Ok, these declarations are also in <linux/kernel.h> but none of the
2154 * ext4 source programs needs to include it so they are duplicated here.
2155 */
2156 # define NORET_TYPE /**/
2157 # define ATTRIB_NORET __attribute__((noreturn))
2158 # define NORET_AND noreturn,
2159
2160 /* bitmap.c */
2161 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2162 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2163 struct ext4_group_desc *gdp,
2164 struct buffer_head *bh, int sz);
2165 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2166 struct ext4_group_desc *gdp,
2167 struct buffer_head *bh, int sz);
2168 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2169 struct ext4_group_desc *gdp,
2170 struct buffer_head *bh);
2171 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2172 struct ext4_group_desc *gdp,
2173 struct buffer_head *bh);
2174
2175 /* balloc.c */
2176 extern void ext4_get_group_no_and_offset(struct super_block *sb,
2177 ext4_fsblk_t blocknr,
2178 ext4_group_t *blockgrpp,
2179 ext4_grpblk_t *offsetp);
2180 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2181 ext4_fsblk_t block);
2182
2183 extern unsigned int ext4_block_group(struct super_block *sb,
2184 ext4_fsblk_t blocknr);
2185 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2186 ext4_fsblk_t blocknr);
2187 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2188 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2189 ext4_group_t group);
2190 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2191 ext4_fsblk_t goal,
2192 unsigned int flags,
2193 unsigned long *count,
2194 int *errp);
2195 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2196 s64 nclusters, unsigned int flags);
2197 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2198 extern void ext4_check_blocks_bitmap(struct super_block *);
2199 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2200 ext4_group_t block_group,
2201 struct buffer_head ** bh);
2202 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2203
2204 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2205 ext4_group_t block_group);
2206 extern int ext4_wait_block_bitmap(struct super_block *sb,
2207 ext4_group_t block_group,
2208 struct buffer_head *bh);
2209 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2210 ext4_group_t block_group);
2211 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2212 ext4_group_t block_group,
2213 struct ext4_group_desc *gdp);
2214 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2215
2216 /* crypto_policy.c */
2217 int ext4_is_child_context_consistent_with_parent(struct inode *parent,
2218 struct inode *child);
2219 int ext4_inherit_context(struct inode *parent, struct inode *child);
2220 void ext4_to_hex(char *dst, char *src, size_t src_size);
2221 int ext4_process_policy(const struct ext4_encryption_policy *policy,
2222 struct inode *inode);
2223 int ext4_get_policy(struct inode *inode,
2224 struct ext4_encryption_policy *policy);
2225
2226 /* crypto.c */
2227 extern struct kmem_cache *ext4_crypt_info_cachep;
2228 bool ext4_valid_contents_enc_mode(uint32_t mode);
2229 uint32_t ext4_validate_encryption_key_size(uint32_t mode, uint32_t size);
2230 extern struct workqueue_struct *ext4_read_workqueue;
2231 struct ext4_crypto_ctx *ext4_get_crypto_ctx(struct inode *inode);
2232 void ext4_release_crypto_ctx(struct ext4_crypto_ctx *ctx);
2233 void ext4_restore_control_page(struct page *data_page);
2234 struct page *ext4_encrypt(struct inode *inode,
2235 struct page *plaintext_page);
2236 int ext4_decrypt(struct page *page);
2237 int ext4_encrypted_zeroout(struct inode *inode, struct ext4_extent *ex);
2238
2239 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2240 int ext4_init_crypto(void);
2241 void ext4_exit_crypto(void);
2242 static inline int ext4_sb_has_crypto(struct super_block *sb)
2243 {
2244 return ext4_has_feature_encrypt(sb);
2245 }
2246 #else
2247 static inline int ext4_init_crypto(void) { return 0; }
2248 static inline void ext4_exit_crypto(void) { }
2249 static inline int ext4_sb_has_crypto(struct super_block *sb)
2250 {
2251 return 0;
2252 }
2253 #endif
2254
2255 /* crypto_fname.c */
2256 bool ext4_valid_filenames_enc_mode(uint32_t mode);
2257 u32 ext4_fname_crypto_round_up(u32 size, u32 blksize);
2258 unsigned ext4_fname_encrypted_size(struct inode *inode, u32 ilen);
2259 int ext4_fname_crypto_alloc_buffer(struct inode *inode,
2260 u32 ilen, struct ext4_str *crypto_str);
2261 int _ext4_fname_disk_to_usr(struct inode *inode,
2262 struct dx_hash_info *hinfo,
2263 const struct ext4_str *iname,
2264 struct ext4_str *oname);
2265 int ext4_fname_disk_to_usr(struct inode *inode,
2266 struct dx_hash_info *hinfo,
2267 const struct ext4_dir_entry_2 *de,
2268 struct ext4_str *oname);
2269 int ext4_fname_usr_to_disk(struct inode *inode,
2270 const struct qstr *iname,
2271 struct ext4_str *oname);
2272 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2273 void ext4_fname_crypto_free_buffer(struct ext4_str *crypto_str);
2274 int ext4_fname_setup_filename(struct inode *dir, const struct qstr *iname,
2275 int lookup, struct ext4_filename *fname);
2276 void ext4_fname_free_filename(struct ext4_filename *fname);
2277 #else
2278 static inline
2279 int ext4_setup_fname_crypto(struct inode *inode)
2280 {
2281 return 0;
2282 }
2283 static inline void ext4_fname_crypto_free_buffer(struct ext4_str *p) { }
2284 static inline int ext4_fname_setup_filename(struct inode *dir,
2285 const struct qstr *iname,
2286 int lookup, struct ext4_filename *fname)
2287 {
2288 fname->usr_fname = iname;
2289 fname->disk_name.name = (unsigned char *) iname->name;
2290 fname->disk_name.len = iname->len;
2291 return 0;
2292 }
2293 static inline void ext4_fname_free_filename(struct ext4_filename *fname) { }
2294 #endif
2295
2296
2297 /* crypto_key.c */
2298 void ext4_free_crypt_info(struct ext4_crypt_info *ci);
2299 void ext4_free_encryption_info(struct inode *inode, struct ext4_crypt_info *ci);
2300 int _ext4_get_encryption_info(struct inode *inode);
2301
2302 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2303 int ext4_has_encryption_key(struct inode *inode);
2304
2305 static inline int ext4_get_encryption_info(struct inode *inode)
2306 {
2307 struct ext4_crypt_info *ci = EXT4_I(inode)->i_crypt_info;
2308
2309 if (!ci ||
2310 (ci->ci_keyring_key &&
2311 (ci->ci_keyring_key->flags & ((1 << KEY_FLAG_INVALIDATED) |
2312 (1 << KEY_FLAG_REVOKED) |
2313 (1 << KEY_FLAG_DEAD)))))
2314 return _ext4_get_encryption_info(inode);
2315 return 0;
2316 }
2317
2318 static inline struct ext4_crypt_info *ext4_encryption_info(struct inode *inode)
2319 {
2320 return EXT4_I(inode)->i_crypt_info;
2321 }
2322
2323 #else
2324 static inline int ext4_has_encryption_key(struct inode *inode)
2325 {
2326 return 0;
2327 }
2328 static inline int ext4_get_encryption_info(struct inode *inode)
2329 {
2330 return 0;
2331 }
2332 static inline struct ext4_crypt_info *ext4_encryption_info(struct inode *inode)
2333 {
2334 return NULL;
2335 }
2336 #endif
2337
2338
2339 /* dir.c */
2340 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2341 struct file *,
2342 struct ext4_dir_entry_2 *,
2343 struct buffer_head *, char *, int,
2344 unsigned int);
2345 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2346 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2347 (de), (bh), (buf), (size), (offset)))
2348 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2349 __u32 minor_hash,
2350 struct ext4_dir_entry_2 *dirent,
2351 struct ext4_str *ent_name);
2352 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2353 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2354 struct buffer_head *bh,
2355 void *buf, int buf_size,
2356 struct ext4_filename *fname,
2357 struct ext4_dir_entry_2 **dest_de);
2358 int ext4_insert_dentry(struct inode *dir,
2359 struct inode *inode,
2360 struct ext4_dir_entry_2 *de,
2361 int buf_size,
2362 struct ext4_filename *fname);
2363 static inline void ext4_update_dx_flag(struct inode *inode)
2364 {
2365 if (!ext4_has_feature_dir_index(inode->i_sb))
2366 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2367 }
2368 static unsigned char ext4_filetype_table[] = {
2369 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2370 };
2371
2372 static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2373 {
2374 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2375 return DT_UNKNOWN;
2376
2377 return ext4_filetype_table[filetype];
2378 }
2379 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2380 void *buf, int buf_size);
2381
2382 /* fsync.c */
2383 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2384
2385 /* hash.c */
2386 extern int ext4fs_dirhash(const char *name, int len, struct
2387 dx_hash_info *hinfo);
2388
2389 /* ialloc.c */
2390 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2391 const struct qstr *qstr, __u32 goal,
2392 uid_t *owner, int handle_type,
2393 unsigned int line_no, int nblocks);
2394
2395 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner) \
2396 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2397 0, 0, 0)
2398 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2399 type, nblocks) \
2400 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2401 (type), __LINE__, (nblocks))
2402
2403
2404 extern void ext4_free_inode(handle_t *, struct inode *);
2405 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2406 extern unsigned long ext4_count_free_inodes(struct super_block *);
2407 extern unsigned long ext4_count_dirs(struct super_block *);
2408 extern void ext4_check_inodes_bitmap(struct super_block *);
2409 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2410 extern int ext4_init_inode_table(struct super_block *sb,
2411 ext4_group_t group, int barrier);
2412 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2413
2414 /* mballoc.c */
2415 extern const struct file_operations ext4_seq_mb_groups_fops;
2416 extern long ext4_mb_stats;
2417 extern long ext4_mb_max_to_scan;
2418 extern int ext4_mb_init(struct super_block *);
2419 extern int ext4_mb_release(struct super_block *);
2420 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2421 struct ext4_allocation_request *, int *);
2422 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2423 extern void ext4_discard_preallocations(struct inode *);
2424 extern int __init ext4_init_mballoc(void);
2425 extern void ext4_exit_mballoc(void);
2426 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2427 struct buffer_head *bh, ext4_fsblk_t block,
2428 unsigned long count, int flags);
2429 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2430 ext4_group_t ngroups);
2431 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2432 ext4_group_t i, struct ext4_group_desc *desc);
2433 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2434 ext4_fsblk_t block, unsigned long count);
2435 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2436
2437 /* inode.c */
2438 int ext4_inode_is_fast_symlink(struct inode *inode);
2439 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2440 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2441 int ext4_get_block_write(struct inode *inode, sector_t iblock,
2442 struct buffer_head *bh_result, int create);
2443 int ext4_get_block_dax(struct inode *inode, sector_t iblock,
2444 struct buffer_head *bh_result, int create);
2445 int ext4_get_block(struct inode *inode, sector_t iblock,
2446 struct buffer_head *bh_result, int create);
2447 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2448 struct buffer_head *bh, int create);
2449 int ext4_walk_page_buffers(handle_t *handle,
2450 struct buffer_head *head,
2451 unsigned from,
2452 unsigned to,
2453 int *partial,
2454 int (*fn)(handle_t *handle,
2455 struct buffer_head *bh));
2456 int do_journal_get_write_access(handle_t *handle,
2457 struct buffer_head *bh);
2458 #define FALL_BACK_TO_NONDELALLOC 1
2459 #define CONVERT_INLINE_DATA 2
2460
2461 extern struct inode *ext4_iget(struct super_block *, unsigned long);
2462 extern struct inode *ext4_iget_normal(struct super_block *, unsigned long);
2463 extern int ext4_write_inode(struct inode *, struct writeback_control *);
2464 extern int ext4_setattr(struct dentry *, struct iattr *);
2465 extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
2466 struct kstat *stat);
2467 extern void ext4_evict_inode(struct inode *);
2468 extern void ext4_clear_inode(struct inode *);
2469 extern int ext4_sync_inode(handle_t *, struct inode *);
2470 extern void ext4_dirty_inode(struct inode *, int);
2471 extern int ext4_change_inode_journal_flag(struct inode *, int);
2472 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2473 extern int ext4_inode_attach_jinode(struct inode *inode);
2474 extern int ext4_can_truncate(struct inode *inode);
2475 extern void ext4_truncate(struct inode *);
2476 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2477 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
2478 extern void ext4_set_inode_flags(struct inode *);
2479 extern void ext4_get_inode_flags(struct ext4_inode_info *);
2480 extern int ext4_alloc_da_blocks(struct inode *inode);
2481 extern void ext4_set_aops(struct inode *inode);
2482 extern int ext4_writepage_trans_blocks(struct inode *);
2483 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2484 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2485 loff_t lstart, loff_t lend);
2486 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
2487 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2488 extern void ext4_da_update_reserve_space(struct inode *inode,
2489 int used, int quota_claim);
2490
2491 /* indirect.c */
2492 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2493 struct ext4_map_blocks *map, int flags);
2494 extern ssize_t ext4_ind_direct_IO(struct kiocb *iocb, struct iov_iter *iter,
2495 loff_t offset);
2496 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2497 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2498 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2499 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2500 ext4_lblk_t start, ext4_lblk_t end);
2501
2502 /* ioctl.c */
2503 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2504 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2505
2506 /* migrate.c */
2507 extern int ext4_ext_migrate(struct inode *);
2508 extern int ext4_ind_migrate(struct inode *inode);
2509
2510 /* namei.c */
2511 extern int ext4_dirent_csum_verify(struct inode *inode,
2512 struct ext4_dir_entry *dirent);
2513 extern int ext4_orphan_add(handle_t *, struct inode *);
2514 extern int ext4_orphan_del(handle_t *, struct inode *);
2515 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2516 __u32 start_minor_hash, __u32 *next_hash);
2517 extern int ext4_search_dir(struct buffer_head *bh,
2518 char *search_buf,
2519 int buf_size,
2520 struct inode *dir,
2521 struct ext4_filename *fname,
2522 const struct qstr *d_name,
2523 unsigned int offset,
2524 struct ext4_dir_entry_2 **res_dir);
2525 extern int ext4_generic_delete_entry(handle_t *handle,
2526 struct inode *dir,
2527 struct ext4_dir_entry_2 *de_del,
2528 struct buffer_head *bh,
2529 void *entry_buf,
2530 int buf_size,
2531 int csum_size);
2532 extern int ext4_empty_dir(struct inode *inode);
2533
2534 /* resize.c */
2535 extern int ext4_group_add(struct super_block *sb,
2536 struct ext4_new_group_data *input);
2537 extern int ext4_group_extend(struct super_block *sb,
2538 struct ext4_super_block *es,
2539 ext4_fsblk_t n_blocks_count);
2540 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2541
2542 /* super.c */
2543 extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
2544 extern int ext4_calculate_overhead(struct super_block *sb);
2545 extern void ext4_superblock_csum_set(struct super_block *sb);
2546 extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2547 extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2548 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2549 ext4_group_t ngroup);
2550 extern const char *ext4_decode_error(struct super_block *sb, int errno,
2551 char nbuf[16]);
2552
2553 extern __printf(4, 5)
2554 void __ext4_error(struct super_block *, const char *, unsigned int,
2555 const char *, ...);
2556 extern __printf(5, 6)
2557 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2558 const char *, ...);
2559 extern __printf(5, 6)
2560 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2561 const char *, ...);
2562 extern void __ext4_std_error(struct super_block *, const char *,
2563 unsigned int, int);
2564 extern __printf(4, 5)
2565 void __ext4_abort(struct super_block *, const char *, unsigned int,
2566 const char *, ...);
2567 extern __printf(4, 5)
2568 void __ext4_warning(struct super_block *, const char *, unsigned int,
2569 const char *, ...);
2570 extern __printf(4, 5)
2571 void __ext4_warning_inode(const struct inode *inode, const char *function,
2572 unsigned int line, const char *fmt, ...);
2573 extern __printf(3, 4)
2574 void __ext4_msg(struct super_block *, const char *, const char *, ...);
2575 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2576 const char *, unsigned int, const char *);
2577 extern __printf(7, 8)
2578 void __ext4_grp_locked_error(const char *, unsigned int,
2579 struct super_block *, ext4_group_t,
2580 unsigned long, ext4_fsblk_t,
2581 const char *, ...);
2582
2583 #define EXT4_ERROR_INODE(inode, fmt, a...) \
2584 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
2585
2586 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
2587 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
2588
2589 #define EXT4_ERROR_FILE(file, block, fmt, a...) \
2590 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
2591
2592 #ifdef CONFIG_PRINTK
2593
2594 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2595 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__)
2596 #define ext4_error_file(file, func, line, block, fmt, ...) \
2597 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2598 #define ext4_error(sb, fmt, ...) \
2599 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2600 #define ext4_abort(sb, fmt, ...) \
2601 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2602 #define ext4_warning(sb, fmt, ...) \
2603 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2604 #define ext4_warning_inode(inode, fmt, ...) \
2605 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
2606 #define ext4_msg(sb, level, fmt, ...) \
2607 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2608 #define dump_mmp_msg(sb, mmp, msg) \
2609 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2610 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2611 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2612 fmt, ##__VA_ARGS__)
2613
2614 #else
2615
2616 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2617 do { \
2618 no_printk(fmt, ##__VA_ARGS__); \
2619 __ext4_error_inode(inode, "", 0, block, " "); \
2620 } while (0)
2621 #define ext4_error_file(file, func, line, block, fmt, ...) \
2622 do { \
2623 no_printk(fmt, ##__VA_ARGS__); \
2624 __ext4_error_file(file, "", 0, block, " "); \
2625 } while (0)
2626 #define ext4_error(sb, fmt, ...) \
2627 do { \
2628 no_printk(fmt, ##__VA_ARGS__); \
2629 __ext4_error(sb, "", 0, " "); \
2630 } while (0)
2631 #define ext4_abort(sb, fmt, ...) \
2632 do { \
2633 no_printk(fmt, ##__VA_ARGS__); \
2634 __ext4_abort(sb, "", 0, " "); \
2635 } while (0)
2636 #define ext4_warning(sb, fmt, ...) \
2637 do { \
2638 no_printk(fmt, ##__VA_ARGS__); \
2639 __ext4_warning(sb, "", 0, " "); \
2640 } while (0)
2641 #define ext4_warning_inode(inode, fmt, ...) \
2642 do { \
2643 no_printk(fmt, ##__VA_ARGS__); \
2644 __ext4_warning_inode(inode, "", 0, " "); \
2645 } while (0)
2646 #define ext4_msg(sb, level, fmt, ...) \
2647 do { \
2648 no_printk(fmt, ##__VA_ARGS__); \
2649 __ext4_msg(sb, "", " "); \
2650 } while (0)
2651 #define dump_mmp_msg(sb, mmp, msg) \
2652 __dump_mmp_msg(sb, mmp, "", 0, "")
2653 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2654 do { \
2655 no_printk(fmt, ##__VA_ARGS__); \
2656 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
2657 } while (0)
2658
2659 #endif
2660
2661 extern void ext4_update_dynamic_rev(struct super_block *sb);
2662 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2663 __u32 compat);
2664 extern int ext4_update_rocompat_feature(handle_t *handle,
2665 struct super_block *sb, __u32 rocompat);
2666 extern int ext4_update_incompat_feature(handle_t *handle,
2667 struct super_block *sb, __u32 incompat);
2668 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2669 struct ext4_group_desc *bg);
2670 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2671 struct ext4_group_desc *bg);
2672 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2673 struct ext4_group_desc *bg);
2674 extern __u32 ext4_free_group_clusters(struct super_block *sb,
2675 struct ext4_group_desc *bg);
2676 extern __u32 ext4_free_inodes_count(struct super_block *sb,
2677 struct ext4_group_desc *bg);
2678 extern __u32 ext4_used_dirs_count(struct super_block *sb,
2679 struct ext4_group_desc *bg);
2680 extern __u32 ext4_itable_unused_count(struct super_block *sb,
2681 struct ext4_group_desc *bg);
2682 extern void ext4_block_bitmap_set(struct super_block *sb,
2683 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2684 extern void ext4_inode_bitmap_set(struct super_block *sb,
2685 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2686 extern void ext4_inode_table_set(struct super_block *sb,
2687 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2688 extern void ext4_free_group_clusters_set(struct super_block *sb,
2689 struct ext4_group_desc *bg,
2690 __u32 count);
2691 extern void ext4_free_inodes_set(struct super_block *sb,
2692 struct ext4_group_desc *bg, __u32 count);
2693 extern void ext4_used_dirs_set(struct super_block *sb,
2694 struct ext4_group_desc *bg, __u32 count);
2695 extern void ext4_itable_unused_set(struct super_block *sb,
2696 struct ext4_group_desc *bg, __u32 count);
2697 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2698 struct ext4_group_desc *gdp);
2699 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2700 struct ext4_group_desc *gdp);
2701 extern int ext4_register_li_request(struct super_block *sb,
2702 ext4_group_t first_not_zeroed);
2703
2704 static inline int ext4_has_group_desc_csum(struct super_block *sb)
2705 {
2706 return ext4_has_feature_gdt_csum(sb) ||
2707 EXT4_SB(sb)->s_chksum_driver != NULL;
2708 }
2709
2710 static inline int ext4_has_metadata_csum(struct super_block *sb)
2711 {
2712 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
2713 !EXT4_SB(sb)->s_chksum_driver);
2714
2715 return (EXT4_SB(sb)->s_chksum_driver != NULL);
2716 }
2717 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2718 {
2719 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2720 le32_to_cpu(es->s_blocks_count_lo);
2721 }
2722
2723 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2724 {
2725 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2726 le32_to_cpu(es->s_r_blocks_count_lo);
2727 }
2728
2729 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2730 {
2731 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2732 le32_to_cpu(es->s_free_blocks_count_lo);
2733 }
2734
2735 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2736 ext4_fsblk_t blk)
2737 {
2738 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2739 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2740 }
2741
2742 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2743 ext4_fsblk_t blk)
2744 {
2745 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2746 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2747 }
2748
2749 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2750 ext4_fsblk_t blk)
2751 {
2752 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2753 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2754 }
2755
2756 static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
2757 {
2758 if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2759 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2760 le32_to_cpu(raw_inode->i_size_lo);
2761 else
2762 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2763 }
2764
2765 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2766 {
2767 raw_inode->i_size_lo = cpu_to_le32(i_size);
2768 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2769 }
2770
2771 static inline
2772 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2773 ext4_group_t group)
2774 {
2775 struct ext4_group_info ***grp_info;
2776 long indexv, indexh;
2777 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
2778 grp_info = EXT4_SB(sb)->s_group_info;
2779 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2780 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2781 return grp_info[indexv][indexh];
2782 }
2783
2784 /*
2785 * Reading s_groups_count requires using smp_rmb() afterwards. See
2786 * the locking protocol documented in the comments of ext4_group_add()
2787 * in resize.c
2788 */
2789 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2790 {
2791 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
2792
2793 smp_rmb();
2794 return ngroups;
2795 }
2796
2797 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2798 ext4_group_t block_group)
2799 {
2800 return block_group >> sbi->s_log_groups_per_flex;
2801 }
2802
2803 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2804 {
2805 return 1 << sbi->s_log_groups_per_flex;
2806 }
2807
2808 #define ext4_std_error(sb, errno) \
2809 do { \
2810 if ((errno)) \
2811 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
2812 } while (0)
2813
2814 #ifdef CONFIG_SMP
2815 /* Each CPU can accumulate percpu_counter_batch clusters in their local
2816 * counters. So we need to make sure we have free clusters more
2817 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
2818 */
2819 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2820 #else
2821 #define EXT4_FREECLUSTERS_WATERMARK 0
2822 #endif
2823
2824 /* Update i_disksize. Requires i_mutex to avoid races with truncate */
2825 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2826 {
2827 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
2828 !mutex_is_locked(&inode->i_mutex));
2829 down_write(&EXT4_I(inode)->i_data_sem);
2830 if (newsize > EXT4_I(inode)->i_disksize)
2831 EXT4_I(inode)->i_disksize = newsize;
2832 up_write(&EXT4_I(inode)->i_data_sem);
2833 }
2834
2835 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
2836 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
2837 {
2838 int changed = 0;
2839
2840 if (newsize > inode->i_size) {
2841 i_size_write(inode, newsize);
2842 changed = 1;
2843 }
2844 if (newsize > EXT4_I(inode)->i_disksize) {
2845 ext4_update_i_disksize(inode, newsize);
2846 changed |= 2;
2847 }
2848 return changed;
2849 }
2850
2851 struct ext4_group_info {
2852 unsigned long bb_state;
2853 struct rb_root bb_free_root;
2854 ext4_grpblk_t bb_first_free; /* first free block */
2855 ext4_grpblk_t bb_free; /* total free blocks */
2856 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
2857 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
2858 struct list_head bb_prealloc_list;
2859 #ifdef DOUBLE_CHECK
2860 void *bb_bitmap;
2861 #endif
2862 struct rw_semaphore alloc_sem;
2863 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
2864 * regions, index is order.
2865 * bb_counters[3] = 5 means
2866 * 5 free 8-block regions. */
2867 };
2868
2869 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
2870 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
2871 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
2872 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
2873
2874 #define EXT4_MB_GRP_NEED_INIT(grp) \
2875 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2876 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
2877 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2878 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
2879 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2880
2881 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
2882 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2883 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
2884 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2885 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
2886 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2887
2888 #define EXT4_MAX_CONTENTION 8
2889 #define EXT4_CONTENTION_THRESHOLD 2
2890
2891 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2892 ext4_group_t group)
2893 {
2894 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2895 }
2896
2897 /*
2898 * Returns true if the filesystem is busy enough that attempts to
2899 * access the block group locks has run into contention.
2900 */
2901 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2902 {
2903 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2904 }
2905
2906 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2907 {
2908 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2909 if (spin_trylock(lock))
2910 /*
2911 * We're able to grab the lock right away, so drop the
2912 * lock contention counter.
2913 */
2914 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2915 else {
2916 /*
2917 * The lock is busy, so bump the contention counter,
2918 * and then wait on the spin lock.
2919 */
2920 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2921 EXT4_MAX_CONTENTION);
2922 spin_lock(lock);
2923 }
2924 }
2925
2926 static inline void ext4_unlock_group(struct super_block *sb,
2927 ext4_group_t group)
2928 {
2929 spin_unlock(ext4_group_lock_ptr(sb, group));
2930 }
2931
2932 /*
2933 * Block validity checking
2934 */
2935 #define ext4_check_indirect_blockref(inode, bh) \
2936 ext4_check_blockref(__func__, __LINE__, inode, \
2937 (__le32 *)(bh)->b_data, \
2938 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2939
2940 #define ext4_ind_check_inode(inode) \
2941 ext4_check_blockref(__func__, __LINE__, inode, \
2942 EXT4_I(inode)->i_data, \
2943 EXT4_NDIR_BLOCKS)
2944
2945 /*
2946 * Inodes and files operations
2947 */
2948
2949 /* dir.c */
2950 extern const struct file_operations ext4_dir_operations;
2951
2952 /* file.c */
2953 extern const struct inode_operations ext4_file_inode_operations;
2954 extern const struct file_operations ext4_file_operations;
2955 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
2956
2957 /* inline.c */
2958 extern int ext4_get_max_inline_size(struct inode *inode);
2959 extern int ext4_find_inline_data_nolock(struct inode *inode);
2960 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
2961 unsigned int len);
2962 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
2963
2964 extern int ext4_readpage_inline(struct inode *inode, struct page *page);
2965 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
2966 struct inode *inode,
2967 loff_t pos, unsigned len,
2968 unsigned flags,
2969 struct page **pagep);
2970 extern int ext4_write_inline_data_end(struct inode *inode,
2971 loff_t pos, unsigned len,
2972 unsigned copied,
2973 struct page *page);
2974 extern struct buffer_head *
2975 ext4_journalled_write_inline_data(struct inode *inode,
2976 unsigned len,
2977 struct page *page);
2978 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
2979 struct inode *inode,
2980 loff_t pos, unsigned len,
2981 unsigned flags,
2982 struct page **pagep,
2983 void **fsdata);
2984 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
2985 unsigned len, unsigned copied,
2986 struct page *page);
2987 extern int ext4_try_add_inline_entry(handle_t *handle,
2988 struct ext4_filename *fname,
2989 struct dentry *dentry,
2990 struct inode *inode);
2991 extern int ext4_try_create_inline_dir(handle_t *handle,
2992 struct inode *parent,
2993 struct inode *inode);
2994 extern int ext4_read_inline_dir(struct file *filp,
2995 struct dir_context *ctx,
2996 int *has_inline_data);
2997 extern int htree_inlinedir_to_tree(struct file *dir_file,
2998 struct inode *dir, ext4_lblk_t block,
2999 struct dx_hash_info *hinfo,
3000 __u32 start_hash, __u32 start_minor_hash,
3001 int *has_inline_data);
3002 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3003 struct ext4_filename *fname,
3004 const struct qstr *d_name,
3005 struct ext4_dir_entry_2 **res_dir,
3006 int *has_inline_data);
3007 extern int ext4_delete_inline_entry(handle_t *handle,
3008 struct inode *dir,
3009 struct ext4_dir_entry_2 *de_del,
3010 struct buffer_head *bh,
3011 int *has_inline_data);
3012 extern int empty_inline_dir(struct inode *dir, int *has_inline_data);
3013 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3014 struct ext4_dir_entry_2 **parent_de,
3015 int *retval);
3016 extern int ext4_inline_data_fiemap(struct inode *inode,
3017 struct fiemap_extent_info *fieinfo,
3018 int *has_inline, __u64 start, __u64 len);
3019 extern int ext4_try_to_evict_inline_data(handle_t *handle,
3020 struct inode *inode,
3021 int needed);
3022 extern void ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3023
3024 extern int ext4_convert_inline_data(struct inode *inode);
3025
3026 static inline int ext4_has_inline_data(struct inode *inode)
3027 {
3028 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3029 EXT4_I(inode)->i_inline_off;
3030 }
3031
3032 /* namei.c */
3033 extern const struct inode_operations ext4_dir_inode_operations;
3034 extern const struct inode_operations ext4_special_inode_operations;
3035 extern struct dentry *ext4_get_parent(struct dentry *child);
3036 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3037 struct ext4_dir_entry_2 *de,
3038 int blocksize, int csum_size,
3039 unsigned int parent_ino, int dotdot_real_len);
3040 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t,
3041 unsigned int blocksize);
3042 extern int ext4_handle_dirty_dirent_node(handle_t *handle,
3043 struct inode *inode,
3044 struct buffer_head *bh);
3045 #define S_SHIFT 12
3046 static unsigned char ext4_type_by_mode[S_IFMT >> S_SHIFT] = {
3047 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
3048 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
3049 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
3050 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
3051 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
3052 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
3053 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
3054 };
3055
3056 static inline void ext4_set_de_type(struct super_block *sb,
3057 struct ext4_dir_entry_2 *de,
3058 umode_t mode) {
3059 if (ext4_has_feature_filetype(sb))
3060 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3061 }
3062
3063 /* readpages.c */
3064 extern int ext4_mpage_readpages(struct address_space *mapping,
3065 struct list_head *pages, struct page *page,
3066 unsigned nr_pages);
3067
3068 /* symlink.c */
3069 extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3070 extern const struct inode_operations ext4_symlink_inode_operations;
3071 extern const struct inode_operations ext4_fast_symlink_inode_operations;
3072
3073 /* sysfs.c */
3074 extern int ext4_register_sysfs(struct super_block *sb);
3075 extern void ext4_unregister_sysfs(struct super_block *sb);
3076 extern int __init ext4_init_sysfs(void);
3077 extern void ext4_exit_sysfs(void);
3078
3079 /* block_validity */
3080 extern void ext4_release_system_zone(struct super_block *sb);
3081 extern int ext4_setup_system_zone(struct super_block *sb);
3082 extern int __init ext4_init_system_zone(void);
3083 extern void ext4_exit_system_zone(void);
3084 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
3085 ext4_fsblk_t start_blk,
3086 unsigned int count);
3087 extern int ext4_check_blockref(const char *, unsigned int,
3088 struct inode *, __le32 *, unsigned int);
3089
3090 /* extents.c */
3091 struct ext4_ext_path;
3092 struct ext4_extent;
3093
3094 /*
3095 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3096 * __le32.
3097 */
3098 #define EXT_MAX_BLOCKS 0xffffffff
3099
3100 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
3101 extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
3102 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3103 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3104 struct ext4_map_blocks *map, int flags);
3105 extern void ext4_ext_truncate(handle_t *, struct inode *);
3106 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3107 ext4_lblk_t end);
3108 extern void ext4_ext_init(struct super_block *);
3109 extern void ext4_ext_release(struct super_block *);
3110 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3111 loff_t len);
3112 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3113 loff_t offset, ssize_t len);
3114 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3115 struct ext4_map_blocks *map, int flags);
3116 extern int ext4_ext_calc_metadata_amount(struct inode *inode,
3117 ext4_lblk_t lblocks);
3118 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3119 int num,
3120 struct ext4_ext_path *path);
3121 extern int ext4_can_extents_be_merged(struct inode *inode,
3122 struct ext4_extent *ex1,
3123 struct ext4_extent *ex2);
3124 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3125 struct ext4_ext_path **,
3126 struct ext4_extent *, int);
3127 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3128 struct ext4_ext_path **,
3129 int flags);
3130 extern void ext4_ext_drop_refs(struct ext4_ext_path *);
3131 extern int ext4_ext_check_inode(struct inode *inode);
3132 extern int ext4_find_delalloc_range(struct inode *inode,
3133 ext4_lblk_t lblk_start,
3134 ext4_lblk_t lblk_end);
3135 extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk);
3136 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3137 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3138 __u64 start, __u64 len);
3139 extern int ext4_ext_precache(struct inode *inode);
3140 extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len);
3141 extern int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len);
3142 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3143 struct inode *inode2, ext4_lblk_t lblk1,
3144 ext4_lblk_t lblk2, ext4_lblk_t count,
3145 int mark_unwritten,int *err);
3146
3147 /* move_extent.c */
3148 extern void ext4_double_down_write_data_sem(struct inode *first,
3149 struct inode *second);
3150 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3151 struct inode *donor_inode);
3152 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3153 __u64 start_orig, __u64 start_donor,
3154 __u64 len, __u64 *moved_len);
3155
3156 /* page-io.c */
3157 extern int __init ext4_init_pageio(void);
3158 extern void ext4_exit_pageio(void);
3159 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3160 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3161 extern int ext4_put_io_end(ext4_io_end_t *io_end);
3162 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3163 extern void ext4_io_submit_init(struct ext4_io_submit *io,
3164 struct writeback_control *wbc);
3165 extern void ext4_end_io_rsv_work(struct work_struct *work);
3166 extern void ext4_io_submit(struct ext4_io_submit *io);
3167 extern int ext4_bio_write_page(struct ext4_io_submit *io,
3168 struct page *page,
3169 int len,
3170 struct writeback_control *wbc,
3171 bool keep_towrite);
3172
3173 /* mmp.c */
3174 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3175
3176 /*
3177 * Add new method to test whether block and inode bitmaps are properly
3178 * initialized. With uninit_bg reading the block from disk is not enough
3179 * to mark the bitmap uptodate. We need to also zero-out the bitmap
3180 */
3181 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3182
3183 static inline int bitmap_uptodate(struct buffer_head *bh)
3184 {
3185 return (buffer_uptodate(bh) &&
3186 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3187 }
3188 static inline void set_bitmap_uptodate(struct buffer_head *bh)
3189 {
3190 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3191 }
3192
3193 /*
3194 * Disable DIO read nolock optimization, so new dioreaders will be forced
3195 * to grab i_mutex
3196 */
3197 static inline void ext4_inode_block_unlocked_dio(struct inode *inode)
3198 {
3199 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
3200 smp_mb();
3201 }
3202 static inline void ext4_inode_resume_unlocked_dio(struct inode *inode)
3203 {
3204 smp_mb();
3205 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
3206 }
3207
3208 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
3209
3210 /* For ioend & aio unwritten conversion wait queues */
3211 #define EXT4_WQ_HASH_SZ 37
3212 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
3213 EXT4_WQ_HASH_SZ])
3214 #define ext4_aio_mutex(v) (&ext4__aio_mutex[((unsigned long)(v)) %\
3215 EXT4_WQ_HASH_SZ])
3216 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3217 extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ];
3218
3219 #define EXT4_RESIZING 0
3220 extern int ext4_resize_begin(struct super_block *sb);
3221 extern void ext4_resize_end(struct super_block *sb);
3222
3223 #endif /* __KERNEL__ */
3224
3225 #define EFSBADCRC EBADMSG /* Bad CRC detected */
3226 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
3227
3228 #endif /* _EXT4_H */
Linux with added FPC-III support