]> git.proxmox.com Git - mirror_ubuntu-zesty-kernel.git/blob - fs/ext4/ext4.h
ext4 crypto: simplify interfaces to directory entry insert functions
[mirror_ubuntu-zesty-kernel.git] / fs / ext4 / ext4.h
1 /*
2 * ext4.h
3 *
4 * Copyright (C) 1992, 1993, 1994, 1995
5 * Remy Card (card@masi.ibp.fr)
6 * Laboratoire MASI - Institut Blaise Pascal
7 * Universite Pierre et Marie Curie (Paris VI)
8 *
9 * from
10 *
11 * linux/include/linux/minix_fs.h
12 *
13 * Copyright (C) 1991, 1992 Linus Torvalds
14 */
15
16 #ifndef _EXT4_H
17 #define _EXT4_H
18
19 #include <linux/types.h>
20 #include <linux/blkdev.h>
21 #include <linux/magic.h>
22 #include <linux/jbd2.h>
23 #include <linux/quota.h>
24 #include <linux/rwsem.h>
25 #include <linux/rbtree.h>
26 #include <linux/seqlock.h>
27 #include <linux/mutex.h>
28 #include <linux/timer.h>
29 #include <linux/version.h>
30 #include <linux/wait.h>
31 #include <linux/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 /* Convert written extents to unwritten */
559 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0100
560 /* Write zeros to newly created written extents */
561 #define EXT4_GET_BLOCKS_ZERO 0x0200
562 #define EXT4_GET_BLOCKS_CREATE_ZERO (EXT4_GET_BLOCKS_CREATE |\
563 EXT4_GET_BLOCKS_ZERO)
564
565 /*
566 * The bit position of these flags must not overlap with any of the
567 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(),
568 * read_extent_tree_block(), ext4_split_extent_at(),
569 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
570 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
571 * caching the extents when reading from the extent tree while a
572 * truncate or punch hole operation is in progress.
573 */
574 #define EXT4_EX_NOCACHE 0x40000000
575 #define EXT4_EX_FORCE_CACHE 0x20000000
576
577 /*
578 * Flags used by ext4_free_blocks
579 */
580 #define EXT4_FREE_BLOCKS_METADATA 0x0001
581 #define EXT4_FREE_BLOCKS_FORGET 0x0002
582 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
583 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
584 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
585 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
586
587 /* Encryption algorithms */
588 #define EXT4_ENCRYPTION_MODE_INVALID 0
589 #define EXT4_ENCRYPTION_MODE_AES_256_XTS 1
590 #define EXT4_ENCRYPTION_MODE_AES_256_GCM 2
591 #define EXT4_ENCRYPTION_MODE_AES_256_CBC 3
592 #define EXT4_ENCRYPTION_MODE_AES_256_CTS 4
593
594 #include "ext4_crypto.h"
595
596 /*
597 * ioctl commands
598 */
599 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
600 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
601 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
602 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
603 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
604 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
605 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
606 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
607 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
608 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
609 #define EXT4_IOC_MIGRATE _IO('f', 9)
610 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
611 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
612 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
613 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
614 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
615 #define EXT4_IOC_SWAP_BOOT _IO('f', 17)
616 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18)
617 #define EXT4_IOC_SET_ENCRYPTION_POLICY _IOR('f', 19, struct ext4_encryption_policy)
618 #define EXT4_IOC_GET_ENCRYPTION_PWSALT _IOW('f', 20, __u8[16])
619 #define EXT4_IOC_GET_ENCRYPTION_POLICY _IOW('f', 21, struct ext4_encryption_policy)
620
621 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
622 /*
623 * ioctl commands in 32 bit emulation
624 */
625 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
626 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
627 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
628 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
629 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
630 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
631 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
632 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
633 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
634 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
635 #endif
636
637 /* Max physical block we can address w/o extents */
638 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
639
640 /*
641 * Structure of an inode on the disk
642 */
643 struct ext4_inode {
644 __le16 i_mode; /* File mode */
645 __le16 i_uid; /* Low 16 bits of Owner Uid */
646 __le32 i_size_lo; /* Size in bytes */
647 __le32 i_atime; /* Access time */
648 __le32 i_ctime; /* Inode Change time */
649 __le32 i_mtime; /* Modification time */
650 __le32 i_dtime; /* Deletion Time */
651 __le16 i_gid; /* Low 16 bits of Group Id */
652 __le16 i_links_count; /* Links count */
653 __le32 i_blocks_lo; /* Blocks count */
654 __le32 i_flags; /* File flags */
655 union {
656 struct {
657 __le32 l_i_version;
658 } linux1;
659 struct {
660 __u32 h_i_translator;
661 } hurd1;
662 struct {
663 __u32 m_i_reserved1;
664 } masix1;
665 } osd1; /* OS dependent 1 */
666 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
667 __le32 i_generation; /* File version (for NFS) */
668 __le32 i_file_acl_lo; /* File ACL */
669 __le32 i_size_high;
670 __le32 i_obso_faddr; /* Obsoleted fragment address */
671 union {
672 struct {
673 __le16 l_i_blocks_high; /* were l_i_reserved1 */
674 __le16 l_i_file_acl_high;
675 __le16 l_i_uid_high; /* these 2 fields */
676 __le16 l_i_gid_high; /* were reserved2[0] */
677 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
678 __le16 l_i_reserved;
679 } linux2;
680 struct {
681 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
682 __u16 h_i_mode_high;
683 __u16 h_i_uid_high;
684 __u16 h_i_gid_high;
685 __u32 h_i_author;
686 } hurd2;
687 struct {
688 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
689 __le16 m_i_file_acl_high;
690 __u32 m_i_reserved2[2];
691 } masix2;
692 } osd2; /* OS dependent 2 */
693 __le16 i_extra_isize;
694 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
695 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
696 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
697 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
698 __le32 i_crtime; /* File Creation time */
699 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
700 __le32 i_version_hi; /* high 32 bits for 64-bit version */
701 __le32 i_projid; /* Project ID */
702 };
703
704 struct move_extent {
705 __u32 reserved; /* should be zero */
706 __u32 donor_fd; /* donor file descriptor */
707 __u64 orig_start; /* logical start offset in block for orig */
708 __u64 donor_start; /* logical start offset in block for donor */
709 __u64 len; /* block length to be moved */
710 __u64 moved_len; /* moved block length */
711 };
712
713 #define EXT4_EPOCH_BITS 2
714 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
715 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
716
717 /*
718 * Extended fields will fit into an inode if the filesystem was formatted
719 * with large inodes (-I 256 or larger) and there are not currently any EAs
720 * consuming all of the available space. For new inodes we always reserve
721 * enough space for the kernel's known extended fields, but for inodes
722 * created with an old kernel this might not have been the case. None of
723 * the extended inode fields is critical for correct filesystem operation.
724 * This macro checks if a certain field fits in the inode. Note that
725 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
726 */
727 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
728 ((offsetof(typeof(*ext4_inode), field) + \
729 sizeof((ext4_inode)->field)) \
730 <= (EXT4_GOOD_OLD_INODE_SIZE + \
731 (einode)->i_extra_isize)) \
732
733 /*
734 * We use an encoding that preserves the times for extra epoch "00":
735 *
736 * extra msb of adjust for signed
737 * epoch 32-bit 32-bit tv_sec to
738 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
739 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
740 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
741 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
742 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
743 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
744 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
745 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
746 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
747 *
748 * Note that previous versions of the kernel on 64-bit systems would
749 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
750 * 1970. e2fsck will correct this, assuming that it is run on the
751 * affected filesystem before 2242.
752 */
753
754 static inline __le32 ext4_encode_extra_time(struct timespec *time)
755 {
756 u32 extra = sizeof(time->tv_sec) > 4 ?
757 ((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK : 0;
758 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
759 }
760
761 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
762 {
763 if (unlikely(sizeof(time->tv_sec) > 4 &&
764 (extra & cpu_to_le32(EXT4_EPOCH_MASK)))) {
765 #if LINUX_VERSION_CODE < KERNEL_VERSION(4,20,0)
766 /* Handle legacy encoding of pre-1970 dates with epoch
767 * bits 1,1. We assume that by kernel version 4.20,
768 * everyone will have run fsck over the affected
769 * filesystems to correct the problem. (This
770 * backwards compatibility may be removed before this
771 * time, at the discretion of the ext4 developers.)
772 */
773 u64 extra_bits = le32_to_cpu(extra) & EXT4_EPOCH_MASK;
774 if (extra_bits == 3 && ((time->tv_sec) & 0x80000000) != 0)
775 extra_bits = 0;
776 time->tv_sec += extra_bits << 32;
777 #else
778 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
779 #endif
780 }
781 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
782 }
783
784 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
785 do { \
786 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
787 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
788 (raw_inode)->xtime ## _extra = \
789 ext4_encode_extra_time(&(inode)->xtime); \
790 } while (0)
791
792 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
793 do { \
794 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
795 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
796 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
797 (raw_inode)->xtime ## _extra = \
798 ext4_encode_extra_time(&(einode)->xtime); \
799 } while (0)
800
801 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
802 do { \
803 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
804 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
805 ext4_decode_extra_time(&(inode)->xtime, \
806 raw_inode->xtime ## _extra); \
807 else \
808 (inode)->xtime.tv_nsec = 0; \
809 } while (0)
810
811 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
812 do { \
813 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
814 (einode)->xtime.tv_sec = \
815 (signed)le32_to_cpu((raw_inode)->xtime); \
816 else \
817 (einode)->xtime.tv_sec = 0; \
818 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
819 ext4_decode_extra_time(&(einode)->xtime, \
820 raw_inode->xtime ## _extra); \
821 else \
822 (einode)->xtime.tv_nsec = 0; \
823 } while (0)
824
825 #define i_disk_version osd1.linux1.l_i_version
826
827 #if defined(__KERNEL__) || defined(__linux__)
828 #define i_reserved1 osd1.linux1.l_i_reserved1
829 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
830 #define i_blocks_high osd2.linux2.l_i_blocks_high
831 #define i_uid_low i_uid
832 #define i_gid_low i_gid
833 #define i_uid_high osd2.linux2.l_i_uid_high
834 #define i_gid_high osd2.linux2.l_i_gid_high
835 #define i_checksum_lo osd2.linux2.l_i_checksum_lo
836
837 #elif defined(__GNU__)
838
839 #define i_translator osd1.hurd1.h_i_translator
840 #define i_uid_high osd2.hurd2.h_i_uid_high
841 #define i_gid_high osd2.hurd2.h_i_gid_high
842 #define i_author osd2.hurd2.h_i_author
843
844 #elif defined(__masix__)
845
846 #define i_reserved1 osd1.masix1.m_i_reserved1
847 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
848 #define i_reserved2 osd2.masix2.m_i_reserved2
849
850 #endif /* defined(__KERNEL__) || defined(__linux__) */
851
852 #include "extents_status.h"
853
854 /*
855 * fourth extended file system inode data in memory
856 */
857 struct ext4_inode_info {
858 __le32 i_data[15]; /* unconverted */
859 __u32 i_dtime;
860 ext4_fsblk_t i_file_acl;
861
862 /*
863 * i_block_group is the number of the block group which contains
864 * this file's inode. Constant across the lifetime of the inode,
865 * it is ued for making block allocation decisions - we try to
866 * place a file's data blocks near its inode block, and new inodes
867 * near to their parent directory's inode.
868 */
869 ext4_group_t i_block_group;
870 ext4_lblk_t i_dir_start_lookup;
871 #if (BITS_PER_LONG < 64)
872 unsigned long i_state_flags; /* Dynamic state flags */
873 #endif
874 unsigned long i_flags;
875
876 /*
877 * Extended attributes can be read independently of the main file
878 * data. Taking i_mutex even when reading would cause contention
879 * between readers of EAs and writers of regular file data, so
880 * instead we synchronize on xattr_sem when reading or changing
881 * EAs.
882 */
883 struct rw_semaphore xattr_sem;
884
885 struct list_head i_orphan; /* unlinked but open inodes */
886
887 /*
888 * i_disksize keeps track of what the inode size is ON DISK, not
889 * in memory. During truncate, i_size is set to the new size by
890 * the VFS prior to calling ext4_truncate(), but the filesystem won't
891 * set i_disksize to 0 until the truncate is actually under way.
892 *
893 * The intent is that i_disksize always represents the blocks which
894 * are used by this file. This allows recovery to restart truncate
895 * on orphans if we crash during truncate. We actually write i_disksize
896 * into the on-disk inode when writing inodes out, instead of i_size.
897 *
898 * The only time when i_disksize and i_size may be different is when
899 * a truncate is in progress. The only things which change i_disksize
900 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
901 */
902 loff_t i_disksize;
903
904 /*
905 * i_data_sem is for serialising ext4_truncate() against
906 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
907 * data tree are chopped off during truncate. We can't do that in
908 * ext4 because whenever we perform intermediate commits during
909 * truncate, the inode and all the metadata blocks *must* be in a
910 * consistent state which allows truncation of the orphans to restart
911 * during recovery. Hence we must fix the get_block-vs-truncate race
912 * by other means, so we have i_data_sem.
913 */
914 struct rw_semaphore i_data_sem;
915 /*
916 * i_mmap_sem is for serializing page faults with truncate / punch hole
917 * operations. We have to make sure that new page cannot be faulted in
918 * a section of the inode that is being punched. We cannot easily use
919 * i_data_sem for this since we need protection for the whole punch
920 * operation and i_data_sem ranks below transaction start so we have
921 * to occasionally drop it.
922 */
923 struct rw_semaphore i_mmap_sem;
924 struct inode vfs_inode;
925 struct jbd2_inode *jinode;
926
927 spinlock_t i_raw_lock; /* protects updates to the raw inode */
928
929 /*
930 * File creation time. Its function is same as that of
931 * struct timespec i_{a,c,m}time in the generic inode.
932 */
933 struct timespec i_crtime;
934
935 /* mballoc */
936 struct list_head i_prealloc_list;
937 spinlock_t i_prealloc_lock;
938
939 /* extents status tree */
940 struct ext4_es_tree i_es_tree;
941 rwlock_t i_es_lock;
942 struct list_head i_es_list;
943 unsigned int i_es_all_nr; /* protected by i_es_lock */
944 unsigned int i_es_shk_nr; /* protected by i_es_lock */
945 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for
946 extents to shrink. Protected by
947 i_es_lock */
948
949 /* ialloc */
950 ext4_group_t i_last_alloc_group;
951
952 /* allocation reservation info for delalloc */
953 /* In case of bigalloc, these refer to clusters rather than blocks */
954 unsigned int i_reserved_data_blocks;
955 unsigned int i_reserved_meta_blocks;
956 unsigned int i_allocated_meta_blocks;
957 ext4_lblk_t i_da_metadata_calc_last_lblock;
958 int i_da_metadata_calc_len;
959
960 /* on-disk additional length */
961 __u16 i_extra_isize;
962
963 /* Indicate the inline data space. */
964 u16 i_inline_off;
965 u16 i_inline_size;
966
967 #ifdef CONFIG_QUOTA
968 /* quota space reservation, managed internally by quota code */
969 qsize_t i_reserved_quota;
970 #endif
971
972 /* Lock protecting lists below */
973 spinlock_t i_completed_io_lock;
974 /*
975 * Completed IOs that need unwritten extents handling and have
976 * transaction reserved
977 */
978 struct list_head i_rsv_conversion_list;
979 /*
980 * Completed IOs that need unwritten extents handling and don't have
981 * transaction reserved
982 */
983 atomic_t i_ioend_count; /* Number of outstanding io_end structs */
984 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
985 struct work_struct i_rsv_conversion_work;
986
987 spinlock_t i_block_reservation_lock;
988
989 /*
990 * Transactions that contain inode's metadata needed to complete
991 * fsync and fdatasync, respectively.
992 */
993 tid_t i_sync_tid;
994 tid_t i_datasync_tid;
995
996 #ifdef CONFIG_QUOTA
997 struct dquot *i_dquot[MAXQUOTAS];
998 #endif
999
1000 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1001 __u32 i_csum_seed;
1002
1003 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1004 /* Encryption params */
1005 struct ext4_crypt_info *i_crypt_info;
1006 #endif
1007 };
1008
1009 /*
1010 * File system states
1011 */
1012 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
1013 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
1014 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
1015
1016 /*
1017 * Misc. filesystem flags
1018 */
1019 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
1020 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
1021 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
1022
1023 /*
1024 * Mount flags set via mount options or defaults
1025 */
1026 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
1027 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
1028 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
1029 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
1030 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
1031 #define EXT4_MOUNT_ERRORS_MASK 0x00070
1032 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
1033 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
1034 #ifdef CONFIG_FS_DAX
1035 #define EXT4_MOUNT_DAX 0x00200 /* Direct Access */
1036 #else
1037 #define EXT4_MOUNT_DAX 0
1038 #endif
1039 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
1040 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
1041 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
1042 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
1043 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
1044 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
1045 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
1046 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
1047 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
1048 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
1049 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
1050 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
1051 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
1052 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
1053 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
1054 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1055 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
1056 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
1057 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
1058 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
1059 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
1060
1061 /*
1062 * Mount flags set either automatically (could not be set by mount option)
1063 * based on per file system feature or property or in special cases such as
1064 * distinguishing between explicit mount option definition and default.
1065 */
1066 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1067 specified delalloc */
1068 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1069 size of blocksize * 8
1070 blocks */
1071 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1072 file systems */
1073
1074 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly
1075 specified journal checksum */
1076
1077 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1078 ~EXT4_MOUNT_##opt
1079 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1080 EXT4_MOUNT_##opt
1081 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1082 EXT4_MOUNT_##opt)
1083
1084 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1085 ~EXT4_MOUNT2_##opt
1086 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1087 EXT4_MOUNT2_##opt
1088 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1089 EXT4_MOUNT2_##opt)
1090
1091 #define ext4_test_and_set_bit __test_and_set_bit_le
1092 #define ext4_set_bit __set_bit_le
1093 #define ext4_set_bit_atomic ext2_set_bit_atomic
1094 #define ext4_test_and_clear_bit __test_and_clear_bit_le
1095 #define ext4_clear_bit __clear_bit_le
1096 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
1097 #define ext4_test_bit test_bit_le
1098 #define ext4_find_next_zero_bit find_next_zero_bit_le
1099 #define ext4_find_next_bit find_next_bit_le
1100
1101 extern void ext4_set_bits(void *bm, int cur, int len);
1102
1103 /*
1104 * Maximal mount counts between two filesystem checks
1105 */
1106 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1107 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1108
1109 /*
1110 * Behaviour when detecting errors
1111 */
1112 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1113 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1114 #define EXT4_ERRORS_PANIC 3 /* Panic */
1115 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1116
1117 /* Metadata checksum algorithm codes */
1118 #define EXT4_CRC32C_CHKSUM 1
1119
1120 /*
1121 * Structure of the super block
1122 */
1123 struct ext4_super_block {
1124 /*00*/ __le32 s_inodes_count; /* Inodes count */
1125 __le32 s_blocks_count_lo; /* Blocks count */
1126 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1127 __le32 s_free_blocks_count_lo; /* Free blocks count */
1128 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1129 __le32 s_first_data_block; /* First Data Block */
1130 __le32 s_log_block_size; /* Block size */
1131 __le32 s_log_cluster_size; /* Allocation cluster size */
1132 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1133 __le32 s_clusters_per_group; /* # Clusters per group */
1134 __le32 s_inodes_per_group; /* # Inodes per group */
1135 __le32 s_mtime; /* Mount time */
1136 /*30*/ __le32 s_wtime; /* Write time */
1137 __le16 s_mnt_count; /* Mount count */
1138 __le16 s_max_mnt_count; /* Maximal mount count */
1139 __le16 s_magic; /* Magic signature */
1140 __le16 s_state; /* File system state */
1141 __le16 s_errors; /* Behaviour when detecting errors */
1142 __le16 s_minor_rev_level; /* minor revision level */
1143 /*40*/ __le32 s_lastcheck; /* time of last check */
1144 __le32 s_checkinterval; /* max. time between checks */
1145 __le32 s_creator_os; /* OS */
1146 __le32 s_rev_level; /* Revision level */
1147 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1148 __le16 s_def_resgid; /* Default gid for reserved blocks */
1149 /*
1150 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1151 *
1152 * Note: the difference between the compatible feature set and
1153 * the incompatible feature set is that if there is a bit set
1154 * in the incompatible feature set that the kernel doesn't
1155 * know about, it should refuse to mount the filesystem.
1156 *
1157 * e2fsck's requirements are more strict; if it doesn't know
1158 * about a feature in either the compatible or incompatible
1159 * feature set, it must abort and not try to meddle with
1160 * things it doesn't understand...
1161 */
1162 __le32 s_first_ino; /* First non-reserved inode */
1163 __le16 s_inode_size; /* size of inode structure */
1164 __le16 s_block_group_nr; /* block group # of this superblock */
1165 __le32 s_feature_compat; /* compatible feature set */
1166 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1167 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1168 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1169 /*78*/ char s_volume_name[16]; /* volume name */
1170 /*88*/ char s_last_mounted[64]; /* directory where last mounted */
1171 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1172 /*
1173 * Performance hints. Directory preallocation should only
1174 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1175 */
1176 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1177 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1178 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1179 /*
1180 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1181 */
1182 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1183 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1184 __le32 s_journal_dev; /* device number of journal file */
1185 __le32 s_last_orphan; /* start of list of inodes to delete */
1186 __le32 s_hash_seed[4]; /* HTREE hash seed */
1187 __u8 s_def_hash_version; /* Default hash version to use */
1188 __u8 s_jnl_backup_type;
1189 __le16 s_desc_size; /* size of group descriptor */
1190 /*100*/ __le32 s_default_mount_opts;
1191 __le32 s_first_meta_bg; /* First metablock block group */
1192 __le32 s_mkfs_time; /* When the filesystem was created */
1193 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1194 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1195 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1196 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1197 __le32 s_free_blocks_count_hi; /* Free blocks count */
1198 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1199 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1200 __le32 s_flags; /* Miscellaneous flags */
1201 __le16 s_raid_stride; /* RAID stride */
1202 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1203 __le64 s_mmp_block; /* Block for multi-mount protection */
1204 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1205 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1206 __u8 s_checksum_type; /* metadata checksum algorithm used */
1207 __u8 s_encryption_level; /* versioning level for encryption */
1208 __u8 s_reserved_pad; /* Padding to next 32bits */
1209 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1210 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1211 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1212 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1213 snapshot's future use */
1214 __le32 s_snapshot_list; /* inode number of the head of the
1215 on-disk snapshot list */
1216 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1217 __le32 s_error_count; /* number of fs errors */
1218 __le32 s_first_error_time; /* first time an error happened */
1219 __le32 s_first_error_ino; /* inode involved in first error */
1220 __le64 s_first_error_block; /* block involved of first error */
1221 __u8 s_first_error_func[32]; /* function where the error happened */
1222 __le32 s_first_error_line; /* line number where error happened */
1223 __le32 s_last_error_time; /* most recent time of an error */
1224 __le32 s_last_error_ino; /* inode involved in last error */
1225 __le32 s_last_error_line; /* line number where error happened */
1226 __le64 s_last_error_block; /* block involved of last error */
1227 __u8 s_last_error_func[32]; /* function where the error happened */
1228 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1229 __u8 s_mount_opts[64];
1230 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1231 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1232 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1233 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
1234 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */
1235 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */
1236 __le32 s_lpf_ino; /* Location of the lost+found inode */
1237 __le32 s_prj_quota_inum; /* inode for tracking project quota */
1238 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */
1239 __le32 s_reserved[98]; /* Padding to the end of the block */
1240 __le32 s_checksum; /* crc32c(superblock) */
1241 };
1242
1243 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1244
1245 #ifdef __KERNEL__
1246
1247 /*
1248 * run-time mount flags
1249 */
1250 #define EXT4_MF_MNTDIR_SAMPLED 0x0001
1251 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1252 #define EXT4_MF_TEST_DUMMY_ENCRYPTION 0x0004
1253
1254 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1255 #define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \
1256 EXT4_MF_TEST_DUMMY_ENCRYPTION))
1257 #else
1258 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1259 #endif
1260
1261 /* Number of quota types we support */
1262 #define EXT4_MAXQUOTAS 2
1263
1264 /*
1265 * fourth extended-fs super-block data in memory
1266 */
1267 struct ext4_sb_info {
1268 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1269 unsigned long s_inodes_per_block;/* Number of inodes per block */
1270 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1271 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1272 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1273 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1274 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1275 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1276 ext4_group_t s_groups_count; /* Number of groups in the fs */
1277 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1278 unsigned long s_overhead; /* # of fs overhead clusters */
1279 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1280 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1281 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1282 struct buffer_head * s_sbh; /* Buffer containing the super block */
1283 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1284 struct buffer_head **s_group_desc;
1285 unsigned int s_mount_opt;
1286 unsigned int s_mount_opt2;
1287 unsigned int s_mount_flags;
1288 unsigned int s_def_mount_opt;
1289 ext4_fsblk_t s_sb_block;
1290 atomic64_t s_resv_clusters;
1291 kuid_t s_resuid;
1292 kgid_t s_resgid;
1293 unsigned short s_mount_state;
1294 unsigned short s_pad;
1295 int s_addr_per_block_bits;
1296 int s_desc_per_block_bits;
1297 int s_inode_size;
1298 int s_first_ino;
1299 unsigned int s_inode_readahead_blks;
1300 unsigned int s_inode_goal;
1301 spinlock_t s_next_gen_lock;
1302 u32 s_next_generation;
1303 u32 s_hash_seed[4];
1304 int s_def_hash_version;
1305 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1306 struct percpu_counter s_freeclusters_counter;
1307 struct percpu_counter s_freeinodes_counter;
1308 struct percpu_counter s_dirs_counter;
1309 struct percpu_counter s_dirtyclusters_counter;
1310 struct blockgroup_lock *s_blockgroup_lock;
1311 struct proc_dir_entry *s_proc;
1312 struct kobject s_kobj;
1313 struct completion s_kobj_unregister;
1314 struct super_block *s_sb;
1315
1316 /* Journaling */
1317 struct journal_s *s_journal;
1318 struct list_head s_orphan;
1319 struct mutex s_orphan_lock;
1320 unsigned long s_resize_flags; /* Flags indicating if there
1321 is a resizer */
1322 unsigned long s_commit_interval;
1323 u32 s_max_batch_time;
1324 u32 s_min_batch_time;
1325 struct block_device *journal_bdev;
1326 #ifdef CONFIG_QUOTA
1327 char *s_qf_names[EXT4_MAXQUOTAS]; /* Names of quota files with journalled quota */
1328 int s_jquota_fmt; /* Format of quota to use */
1329 #endif
1330 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1331 struct rb_root system_blks;
1332
1333 #ifdef EXTENTS_STATS
1334 /* ext4 extents stats */
1335 unsigned long s_ext_min;
1336 unsigned long s_ext_max;
1337 unsigned long s_depth_max;
1338 spinlock_t s_ext_stats_lock;
1339 unsigned long s_ext_blocks;
1340 unsigned long s_ext_extents;
1341 #endif
1342
1343 /* for buddy allocator */
1344 struct ext4_group_info ***s_group_info;
1345 struct inode *s_buddy_cache;
1346 spinlock_t s_md_lock;
1347 unsigned short *s_mb_offsets;
1348 unsigned int *s_mb_maxs;
1349 unsigned int s_group_info_size;
1350
1351 /* tunables */
1352 unsigned long s_stripe;
1353 unsigned int s_mb_stream_request;
1354 unsigned int s_mb_max_to_scan;
1355 unsigned int s_mb_min_to_scan;
1356 unsigned int s_mb_stats;
1357 unsigned int s_mb_order2_reqs;
1358 unsigned int s_mb_group_prealloc;
1359 unsigned int s_max_dir_size_kb;
1360 /* where last allocation was done - for stream allocation */
1361 unsigned long s_mb_last_group;
1362 unsigned long s_mb_last_start;
1363
1364 /* stats for buddy allocator */
1365 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1366 atomic_t s_bal_success; /* we found long enough chunks */
1367 atomic_t s_bal_allocated; /* in blocks */
1368 atomic_t s_bal_ex_scanned; /* total extents scanned */
1369 atomic_t s_bal_goals; /* goal hits */
1370 atomic_t s_bal_breaks; /* too long searches */
1371 atomic_t s_bal_2orders; /* 2^order hits */
1372 spinlock_t s_bal_lock;
1373 unsigned long s_mb_buddies_generated;
1374 unsigned long long s_mb_generation_time;
1375 atomic_t s_mb_lost_chunks;
1376 atomic_t s_mb_preallocated;
1377 atomic_t s_mb_discarded;
1378 atomic_t s_lock_busy;
1379
1380 /* locality groups */
1381 struct ext4_locality_group __percpu *s_locality_groups;
1382
1383 /* for write statistics */
1384 unsigned long s_sectors_written_start;
1385 u64 s_kbytes_written;
1386
1387 /* the size of zero-out chunk */
1388 unsigned int s_extent_max_zeroout_kb;
1389
1390 unsigned int s_log_groups_per_flex;
1391 struct flex_groups *s_flex_groups;
1392 ext4_group_t s_flex_groups_allocated;
1393
1394 /* workqueue for reserved extent conversions (buffered io) */
1395 struct workqueue_struct *rsv_conversion_wq;
1396
1397 /* timer for periodic error stats printing */
1398 struct timer_list s_err_report;
1399
1400 /* Lazy inode table initialization info */
1401 struct ext4_li_request *s_li_request;
1402 /* Wait multiplier for lazy initialization thread */
1403 unsigned int s_li_wait_mult;
1404
1405 /* Kernel thread for multiple mount protection */
1406 struct task_struct *s_mmp_tsk;
1407
1408 /* record the last minlen when FITRIM is called. */
1409 atomic_t s_last_trim_minblks;
1410
1411 /* Reference to checksum algorithm driver via cryptoapi */
1412 struct crypto_shash *s_chksum_driver;
1413
1414 /* Precomputed FS UUID checksum for seeding other checksums */
1415 __u32 s_csum_seed;
1416
1417 /* Reclaim extents from extent status tree */
1418 struct shrinker s_es_shrinker;
1419 struct list_head s_es_list; /* List of inodes with reclaimable extents */
1420 long s_es_nr_inode;
1421 struct ext4_es_stats s_es_stats;
1422 struct mb_cache *s_mb_cache;
1423 spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1424
1425 /* Ratelimit ext4 messages. */
1426 struct ratelimit_state s_err_ratelimit_state;
1427 struct ratelimit_state s_warning_ratelimit_state;
1428 struct ratelimit_state s_msg_ratelimit_state;
1429 };
1430
1431 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1432 {
1433 return sb->s_fs_info;
1434 }
1435 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1436 {
1437 return container_of(inode, struct ext4_inode_info, vfs_inode);
1438 }
1439
1440 static inline struct timespec ext4_current_time(struct inode *inode)
1441 {
1442 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1443 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1444 }
1445
1446 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1447 {
1448 return ino == EXT4_ROOT_INO ||
1449 ino == EXT4_USR_QUOTA_INO ||
1450 ino == EXT4_GRP_QUOTA_INO ||
1451 ino == EXT4_BOOT_LOADER_INO ||
1452 ino == EXT4_JOURNAL_INO ||
1453 ino == EXT4_RESIZE_INO ||
1454 (ino >= EXT4_FIRST_INO(sb) &&
1455 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1456 }
1457
1458 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
1459 struct ext4_io_end *io_end)
1460 {
1461 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
1462 io_end->flag |= EXT4_IO_END_UNWRITTEN;
1463 atomic_inc(&EXT4_I(inode)->i_unwritten);
1464 }
1465 }
1466
1467 static inline ext4_io_end_t *ext4_inode_aio(struct inode *inode)
1468 {
1469 return inode->i_private;
1470 }
1471
1472 static inline void ext4_inode_aio_set(struct inode *inode, ext4_io_end_t *io)
1473 {
1474 inode->i_private = io;
1475 }
1476
1477 /*
1478 * Inode dynamic state flags
1479 */
1480 enum {
1481 EXT4_STATE_JDATA, /* journaled data exists */
1482 EXT4_STATE_NEW, /* inode is newly created */
1483 EXT4_STATE_XATTR, /* has in-inode xattrs */
1484 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1485 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1486 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1487 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
1488 EXT4_STATE_NEWENTRY, /* File just added to dir */
1489 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read
1490 nolocking */
1491 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1492 EXT4_STATE_ORDERED_MODE, /* data=ordered mode */
1493 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
1494 };
1495
1496 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1497 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1498 { \
1499 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1500 } \
1501 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1502 { \
1503 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1504 } \
1505 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1506 { \
1507 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1508 }
1509
1510 /* Add these declarations here only so that these functions can be
1511 * found by name. Otherwise, they are very hard to locate. */
1512 static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1513 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1514 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1515 EXT4_INODE_BIT_FNS(flag, flags, 0)
1516
1517 /* Add these declarations here only so that these functions can be
1518 * found by name. Otherwise, they are very hard to locate. */
1519 static inline int ext4_test_inode_state(struct inode *inode, int bit);
1520 static inline void ext4_set_inode_state(struct inode *inode, int bit);
1521 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1522 #if (BITS_PER_LONG < 64)
1523 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1524
1525 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1526 {
1527 (ei)->i_state_flags = 0;
1528 }
1529 #else
1530 EXT4_INODE_BIT_FNS(state, flags, 32)
1531
1532 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1533 {
1534 /* We depend on the fact that callers will set i_flags */
1535 }
1536 #endif
1537 #else
1538 /* Assume that user mode programs are passing in an ext4fs superblock, not
1539 * a kernel struct super_block. This will allow us to call the feature-test
1540 * macros from user land. */
1541 #define EXT4_SB(sb) (sb)
1542 #endif
1543
1544 /*
1545 * Returns true if the inode is inode is encrypted
1546 */
1547 static inline int ext4_encrypted_inode(struct inode *inode)
1548 {
1549 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1550 return ext4_test_inode_flag(inode, EXT4_INODE_ENCRYPT);
1551 #else
1552 return 0;
1553 #endif
1554 }
1555
1556 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1557
1558 /*
1559 * Codes for operating systems
1560 */
1561 #define EXT4_OS_LINUX 0
1562 #define EXT4_OS_HURD 1
1563 #define EXT4_OS_MASIX 2
1564 #define EXT4_OS_FREEBSD 3
1565 #define EXT4_OS_LITES 4
1566
1567 /*
1568 * Revision levels
1569 */
1570 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1571 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1572
1573 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1574 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1575
1576 #define EXT4_GOOD_OLD_INODE_SIZE 128
1577
1578 /*
1579 * Feature set definitions
1580 */
1581
1582 /* Use the ext4_{has,set,clear}_feature_* helpers; these will be removed */
1583 #define EXT4_HAS_COMPAT_FEATURE(sb,mask) \
1584 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1585 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \
1586 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1587 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \
1588 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1589 #define EXT4_SET_COMPAT_FEATURE(sb,mask) \
1590 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1591 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \
1592 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1593 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \
1594 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1595 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \
1596 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1597 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
1598 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1599 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \
1600 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1601
1602 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1603 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1604 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1605 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1606 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1607 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1608 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
1609
1610 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1611 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1612 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1613 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1614 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1615 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1616 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1617 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1618 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1619 /*
1620 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1621 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1622 * all other data structures' checksums. However, the METADATA_CSUM and
1623 * GDT_CSUM bits are mutually exclusive.
1624 */
1625 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
1626 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000
1627 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000
1628
1629 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1630 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1631 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1632 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1633 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1634 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1635 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1636 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1637 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1638 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1639 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1640 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000
1641 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
1642 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
1643 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000
1644
1645 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
1646 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1647 { \
1648 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1649 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
1650 } \
1651 static inline void ext4_set_feature_##name(struct super_block *sb) \
1652 { \
1653 EXT4_SB(sb)->s_es->s_feature_compat |= \
1654 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1655 } \
1656 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1657 { \
1658 EXT4_SB(sb)->s_es->s_feature_compat &= \
1659 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1660 }
1661
1662 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
1663 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1664 { \
1665 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1666 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
1667 } \
1668 static inline void ext4_set_feature_##name(struct super_block *sb) \
1669 { \
1670 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
1671 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1672 } \
1673 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1674 { \
1675 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
1676 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1677 }
1678
1679 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
1680 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1681 { \
1682 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1683 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
1684 } \
1685 static inline void ext4_set_feature_##name(struct super_block *sb) \
1686 { \
1687 EXT4_SB(sb)->s_es->s_feature_incompat |= \
1688 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1689 } \
1690 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1691 { \
1692 EXT4_SB(sb)->s_es->s_feature_incompat &= \
1693 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1694 }
1695
1696 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC)
1697 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES)
1698 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL)
1699 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR)
1700 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE)
1701 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX)
1702 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2)
1703
1704 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER)
1705 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE)
1706 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR)
1707 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE)
1708 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM)
1709 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK)
1710 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE)
1711 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA)
1712 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC)
1713 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM)
1714 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY)
1715 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT)
1716
1717 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION)
1718 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE)
1719 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER)
1720 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV)
1721 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG)
1722 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS)
1723 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT)
1724 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP)
1725 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG)
1726 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE)
1727 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA)
1728 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED)
1729 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR)
1730 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA)
1731 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT)
1732
1733 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1734 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1735 EXT4_FEATURE_INCOMPAT_META_BG)
1736 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1737 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1738 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1739
1740 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1741 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1742 EXT4_FEATURE_INCOMPAT_RECOVER| \
1743 EXT4_FEATURE_INCOMPAT_META_BG)
1744 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1745 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1746 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1747
1748 #define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1749 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1750 EXT4_FEATURE_INCOMPAT_RECOVER| \
1751 EXT4_FEATURE_INCOMPAT_META_BG| \
1752 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1753 EXT4_FEATURE_INCOMPAT_64BIT| \
1754 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1755 EXT4_FEATURE_INCOMPAT_MMP | \
1756 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
1757 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
1758 EXT4_FEATURE_INCOMPAT_CSUM_SEED)
1759 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1760 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1761 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1762 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1763 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1764 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1765 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1766 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1767 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1768 EXT4_FEATURE_RO_COMPAT_QUOTA)
1769
1770 #define EXTN_FEATURE_FUNCS(ver) \
1771 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
1772 { \
1773 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1774 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
1775 } \
1776 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
1777 { \
1778 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1779 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
1780 } \
1781 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
1782 { \
1783 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1784 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
1785 }
1786
1787 EXTN_FEATURE_FUNCS(2)
1788 EXTN_FEATURE_FUNCS(3)
1789 EXTN_FEATURE_FUNCS(4)
1790
1791 static inline bool ext4_has_compat_features(struct super_block *sb)
1792 {
1793 return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
1794 }
1795 static inline bool ext4_has_ro_compat_features(struct super_block *sb)
1796 {
1797 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
1798 }
1799 static inline bool ext4_has_incompat_features(struct super_block *sb)
1800 {
1801 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
1802 }
1803
1804 /*
1805 * Default values for user and/or group using reserved blocks
1806 */
1807 #define EXT4_DEF_RESUID 0
1808 #define EXT4_DEF_RESGID 0
1809
1810 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
1811
1812 /*
1813 * Default mount options
1814 */
1815 #define EXT4_DEFM_DEBUG 0x0001
1816 #define EXT4_DEFM_BSDGROUPS 0x0002
1817 #define EXT4_DEFM_XATTR_USER 0x0004
1818 #define EXT4_DEFM_ACL 0x0008
1819 #define EXT4_DEFM_UID16 0x0010
1820 #define EXT4_DEFM_JMODE 0x0060
1821 #define EXT4_DEFM_JMODE_DATA 0x0020
1822 #define EXT4_DEFM_JMODE_ORDERED 0x0040
1823 #define EXT4_DEFM_JMODE_WBACK 0x0060
1824 #define EXT4_DEFM_NOBARRIER 0x0100
1825 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1826 #define EXT4_DEFM_DISCARD 0x0400
1827 #define EXT4_DEFM_NODELALLOC 0x0800
1828
1829 /*
1830 * Default journal batch times
1831 */
1832 #define EXT4_DEF_MIN_BATCH_TIME 0
1833 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1834
1835 /*
1836 * Minimum number of groups in a flexgroup before we separate out
1837 * directories into the first block group of a flexgroup
1838 */
1839 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1840
1841 /*
1842 * Structure of a directory entry
1843 */
1844 #define EXT4_NAME_LEN 255
1845
1846 struct ext4_dir_entry {
1847 __le32 inode; /* Inode number */
1848 __le16 rec_len; /* Directory entry length */
1849 __le16 name_len; /* Name length */
1850 char name[EXT4_NAME_LEN]; /* File name */
1851 };
1852
1853 /*
1854 * The new version of the directory entry. Since EXT4 structures are
1855 * stored in intel byte order, and the name_len field could never be
1856 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1857 * file_type field.
1858 */
1859 struct ext4_dir_entry_2 {
1860 __le32 inode; /* Inode number */
1861 __le16 rec_len; /* Directory entry length */
1862 __u8 name_len; /* Name length */
1863 __u8 file_type;
1864 char name[EXT4_NAME_LEN]; /* File name */
1865 };
1866
1867 /*
1868 * This is a bogus directory entry at the end of each leaf block that
1869 * records checksums.
1870 */
1871 struct ext4_dir_entry_tail {
1872 __le32 det_reserved_zero1; /* Pretend to be unused */
1873 __le16 det_rec_len; /* 12 */
1874 __u8 det_reserved_zero2; /* Zero name length */
1875 __u8 det_reserved_ft; /* 0xDE, fake file type */
1876 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
1877 };
1878
1879 #define EXT4_DIRENT_TAIL(block, blocksize) \
1880 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1881 ((blocksize) - \
1882 sizeof(struct ext4_dir_entry_tail))))
1883
1884 /*
1885 * Ext4 directory file types. Only the low 3 bits are used. The
1886 * other bits are reserved for now.
1887 */
1888 #define EXT4_FT_UNKNOWN 0
1889 #define EXT4_FT_REG_FILE 1
1890 #define EXT4_FT_DIR 2
1891 #define EXT4_FT_CHRDEV 3
1892 #define EXT4_FT_BLKDEV 4
1893 #define EXT4_FT_FIFO 5
1894 #define EXT4_FT_SOCK 6
1895 #define EXT4_FT_SYMLINK 7
1896
1897 #define EXT4_FT_MAX 8
1898
1899 #define EXT4_FT_DIR_CSUM 0xDE
1900
1901 /*
1902 * EXT4_DIR_PAD defines the directory entries boundaries
1903 *
1904 * NOTE: It must be a multiple of 4
1905 */
1906 #define EXT4_DIR_PAD 4
1907 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1908 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1909 ~EXT4_DIR_ROUND)
1910 #define EXT4_MAX_REC_LEN ((1<<16)-1)
1911
1912 /*
1913 * If we ever get support for fs block sizes > page_size, we'll need
1914 * to remove the #if statements in the next two functions...
1915 */
1916 static inline unsigned int
1917 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1918 {
1919 unsigned len = le16_to_cpu(dlen);
1920
1921 #if (PAGE_CACHE_SIZE >= 65536)
1922 if (len == EXT4_MAX_REC_LEN || len == 0)
1923 return blocksize;
1924 return (len & 65532) | ((len & 3) << 16);
1925 #else
1926 return len;
1927 #endif
1928 }
1929
1930 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1931 {
1932 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1933 BUG();
1934 #if (PAGE_CACHE_SIZE >= 65536)
1935 if (len < 65536)
1936 return cpu_to_le16(len);
1937 if (len == blocksize) {
1938 if (blocksize == 65536)
1939 return cpu_to_le16(EXT4_MAX_REC_LEN);
1940 else
1941 return cpu_to_le16(0);
1942 }
1943 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1944 #else
1945 return cpu_to_le16(len);
1946 #endif
1947 }
1948
1949 /*
1950 * Hash Tree Directory indexing
1951 * (c) Daniel Phillips, 2001
1952 */
1953
1954 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
1955 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
1956 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1957 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1958
1959 /* Legal values for the dx_root hash_version field: */
1960
1961 #define DX_HASH_LEGACY 0
1962 #define DX_HASH_HALF_MD4 1
1963 #define DX_HASH_TEA 2
1964 #define DX_HASH_LEGACY_UNSIGNED 3
1965 #define DX_HASH_HALF_MD4_UNSIGNED 4
1966 #define DX_HASH_TEA_UNSIGNED 5
1967
1968 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
1969 const void *address, unsigned int length)
1970 {
1971 struct {
1972 struct shash_desc shash;
1973 char ctx[4];
1974 } desc;
1975 int err;
1976
1977 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
1978
1979 desc.shash.tfm = sbi->s_chksum_driver;
1980 desc.shash.flags = 0;
1981 *(u32 *)desc.ctx = crc;
1982
1983 err = crypto_shash_update(&desc.shash, address, length);
1984 BUG_ON(err);
1985
1986 return *(u32 *)desc.ctx;
1987 }
1988
1989 #ifdef __KERNEL__
1990
1991 /* hash info structure used by the directory hash */
1992 struct dx_hash_info
1993 {
1994 u32 hash;
1995 u32 minor_hash;
1996 int hash_version;
1997 u32 *seed;
1998 };
1999
2000
2001 /* 32 and 64 bit signed EOF for dx directories */
2002 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
2003 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
2004
2005
2006 /*
2007 * Control parameters used by ext4_htree_next_block
2008 */
2009 #define HASH_NB_ALWAYS 1
2010
2011 struct ext4_filename {
2012 const struct qstr *usr_fname;
2013 struct ext4_str disk_name;
2014 struct dx_hash_info hinfo;
2015 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2016 struct ext4_str crypto_buf;
2017 #endif
2018 };
2019
2020 #define fname_name(p) ((p)->disk_name.name)
2021 #define fname_len(p) ((p)->disk_name.len)
2022
2023 /*
2024 * Describe an inode's exact location on disk and in memory
2025 */
2026 struct ext4_iloc
2027 {
2028 struct buffer_head *bh;
2029 unsigned long offset;
2030 ext4_group_t block_group;
2031 };
2032
2033 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2034 {
2035 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2036 }
2037
2038 /*
2039 * This structure is stuffed into the struct file's private_data field
2040 * for directories. It is where we put information so that we can do
2041 * readdir operations in hash tree order.
2042 */
2043 struct dir_private_info {
2044 struct rb_root root;
2045 struct rb_node *curr_node;
2046 struct fname *extra_fname;
2047 loff_t last_pos;
2048 __u32 curr_hash;
2049 __u32 curr_minor_hash;
2050 __u32 next_hash;
2051 };
2052
2053 /* calculate the first block number of the group */
2054 static inline ext4_fsblk_t
2055 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2056 {
2057 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2058 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2059 }
2060
2061 /*
2062 * Special error return code only used by dx_probe() and its callers.
2063 */
2064 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
2065
2066 /*
2067 * Timeout and state flag for lazy initialization inode thread.
2068 */
2069 #define EXT4_DEF_LI_WAIT_MULT 10
2070 #define EXT4_DEF_LI_MAX_START_DELAY 5
2071 #define EXT4_LAZYINIT_QUIT 0x0001
2072 #define EXT4_LAZYINIT_RUNNING 0x0002
2073
2074 /*
2075 * Lazy inode table initialization info
2076 */
2077 struct ext4_lazy_init {
2078 unsigned long li_state;
2079 struct list_head li_request_list;
2080 struct mutex li_list_mtx;
2081 };
2082
2083 struct ext4_li_request {
2084 struct super_block *lr_super;
2085 struct ext4_sb_info *lr_sbi;
2086 ext4_group_t lr_next_group;
2087 struct list_head lr_request;
2088 unsigned long lr_next_sched;
2089 unsigned long lr_timeout;
2090 };
2091
2092 struct ext4_features {
2093 struct kobject f_kobj;
2094 struct completion f_kobj_unregister;
2095 };
2096
2097 /*
2098 * This structure will be used for multiple mount protection. It will be
2099 * written into the block number saved in the s_mmp_block field in the
2100 * superblock. Programs that check MMP should assume that if
2101 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2102 * to use the filesystem, regardless of how old the timestamp is.
2103 */
2104 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
2105 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2106 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
2107 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
2108
2109 struct mmp_struct {
2110 __le32 mmp_magic; /* Magic number for MMP */
2111 __le32 mmp_seq; /* Sequence no. updated periodically */
2112
2113 /*
2114 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2115 * purposes and do not affect the correctness of the algorithm
2116 */
2117 __le64 mmp_time; /* Time last updated */
2118 char mmp_nodename[64]; /* Node which last updated MMP block */
2119 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
2120
2121 /*
2122 * mmp_check_interval is used to verify if the MMP block has been
2123 * updated on the block device. The value is updated based on the
2124 * maximum time to write the MMP block during an update cycle.
2125 */
2126 __le16 mmp_check_interval;
2127
2128 __le16 mmp_pad1;
2129 __le32 mmp_pad2[226];
2130 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
2131 };
2132
2133 /* arguments passed to the mmp thread */
2134 struct mmpd_data {
2135 struct buffer_head *bh; /* bh from initial read_mmp_block() */
2136 struct super_block *sb; /* super block of the fs */
2137 };
2138
2139 /*
2140 * Check interval multiplier
2141 * The MMP block is written every update interval and initially checked every
2142 * update interval x the multiplier (the value is then adapted based on the
2143 * write latency). The reason is that writes can be delayed under load and we
2144 * don't want readers to incorrectly assume that the filesystem is no longer
2145 * in use.
2146 */
2147 #define EXT4_MMP_CHECK_MULT 2UL
2148
2149 /*
2150 * Minimum interval for MMP checking in seconds.
2151 */
2152 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
2153
2154 /*
2155 * Maximum interval for MMP checking in seconds.
2156 */
2157 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
2158
2159 /*
2160 * Function prototypes
2161 */
2162
2163 /*
2164 * Ok, these declarations are also in <linux/kernel.h> but none of the
2165 * ext4 source programs needs to include it so they are duplicated here.
2166 */
2167 # define NORET_TYPE /**/
2168 # define ATTRIB_NORET __attribute__((noreturn))
2169 # define NORET_AND noreturn,
2170
2171 /* bitmap.c */
2172 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2173 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2174 struct ext4_group_desc *gdp,
2175 struct buffer_head *bh, int sz);
2176 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2177 struct ext4_group_desc *gdp,
2178 struct buffer_head *bh, int sz);
2179 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2180 struct ext4_group_desc *gdp,
2181 struct buffer_head *bh);
2182 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2183 struct ext4_group_desc *gdp,
2184 struct buffer_head *bh);
2185
2186 /* balloc.c */
2187 extern void ext4_get_group_no_and_offset(struct super_block *sb,
2188 ext4_fsblk_t blocknr,
2189 ext4_group_t *blockgrpp,
2190 ext4_grpblk_t *offsetp);
2191 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2192 ext4_fsblk_t block);
2193
2194 extern unsigned int ext4_block_group(struct super_block *sb,
2195 ext4_fsblk_t blocknr);
2196 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2197 ext4_fsblk_t blocknr);
2198 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2199 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2200 ext4_group_t group);
2201 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2202 ext4_fsblk_t goal,
2203 unsigned int flags,
2204 unsigned long *count,
2205 int *errp);
2206 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2207 s64 nclusters, unsigned int flags);
2208 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2209 extern void ext4_check_blocks_bitmap(struct super_block *);
2210 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2211 ext4_group_t block_group,
2212 struct buffer_head ** bh);
2213 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2214
2215 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2216 ext4_group_t block_group);
2217 extern int ext4_wait_block_bitmap(struct super_block *sb,
2218 ext4_group_t block_group,
2219 struct buffer_head *bh);
2220 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2221 ext4_group_t block_group);
2222 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2223 ext4_group_t block_group,
2224 struct ext4_group_desc *gdp);
2225 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2226
2227 /* crypto_policy.c */
2228 int ext4_is_child_context_consistent_with_parent(struct inode *parent,
2229 struct inode *child);
2230 int ext4_inherit_context(struct inode *parent, struct inode *child);
2231 void ext4_to_hex(char *dst, char *src, size_t src_size);
2232 int ext4_process_policy(const struct ext4_encryption_policy *policy,
2233 struct inode *inode);
2234 int ext4_get_policy(struct inode *inode,
2235 struct ext4_encryption_policy *policy);
2236
2237 /* crypto.c */
2238 extern struct kmem_cache *ext4_crypt_info_cachep;
2239 bool ext4_valid_contents_enc_mode(uint32_t mode);
2240 uint32_t ext4_validate_encryption_key_size(uint32_t mode, uint32_t size);
2241 extern struct workqueue_struct *ext4_read_workqueue;
2242 struct ext4_crypto_ctx *ext4_get_crypto_ctx(struct inode *inode);
2243 void ext4_release_crypto_ctx(struct ext4_crypto_ctx *ctx);
2244 void ext4_restore_control_page(struct page *data_page);
2245 struct page *ext4_encrypt(struct inode *inode,
2246 struct page *plaintext_page);
2247 int ext4_decrypt(struct page *page);
2248 int ext4_encrypted_zeroout(struct inode *inode, ext4_lblk_t lblk,
2249 ext4_fsblk_t pblk, ext4_lblk_t len);
2250
2251 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2252 int ext4_init_crypto(void);
2253 void ext4_exit_crypto(void);
2254 static inline int ext4_sb_has_crypto(struct super_block *sb)
2255 {
2256 return ext4_has_feature_encrypt(sb);
2257 }
2258 #else
2259 static inline int ext4_init_crypto(void) { return 0; }
2260 static inline void ext4_exit_crypto(void) { }
2261 static inline int ext4_sb_has_crypto(struct super_block *sb)
2262 {
2263 return 0;
2264 }
2265 #endif
2266
2267 /* crypto_fname.c */
2268 bool ext4_valid_filenames_enc_mode(uint32_t mode);
2269 u32 ext4_fname_crypto_round_up(u32 size, u32 blksize);
2270 unsigned ext4_fname_encrypted_size(struct inode *inode, u32 ilen);
2271 int ext4_fname_crypto_alloc_buffer(struct inode *inode,
2272 u32 ilen, struct ext4_str *crypto_str);
2273 int _ext4_fname_disk_to_usr(struct inode *inode,
2274 struct dx_hash_info *hinfo,
2275 const struct ext4_str *iname,
2276 struct ext4_str *oname);
2277 int ext4_fname_disk_to_usr(struct inode *inode,
2278 struct dx_hash_info *hinfo,
2279 const struct ext4_dir_entry_2 *de,
2280 struct ext4_str *oname);
2281 int ext4_fname_usr_to_disk(struct inode *inode,
2282 const struct qstr *iname,
2283 struct ext4_str *oname);
2284 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2285 void ext4_fname_crypto_free_buffer(struct ext4_str *crypto_str);
2286 int ext4_fname_setup_filename(struct inode *dir, const struct qstr *iname,
2287 int lookup, struct ext4_filename *fname);
2288 void ext4_fname_free_filename(struct ext4_filename *fname);
2289 #else
2290 static inline
2291 int ext4_setup_fname_crypto(struct inode *inode)
2292 {
2293 return 0;
2294 }
2295 static inline void ext4_fname_crypto_free_buffer(struct ext4_str *p) { }
2296 static inline int ext4_fname_setup_filename(struct inode *dir,
2297 const struct qstr *iname,
2298 int lookup, struct ext4_filename *fname)
2299 {
2300 fname->usr_fname = iname;
2301 fname->disk_name.name = (unsigned char *) iname->name;
2302 fname->disk_name.len = iname->len;
2303 return 0;
2304 }
2305 static inline void ext4_fname_free_filename(struct ext4_filename *fname) { }
2306 #endif
2307
2308
2309 /* crypto_key.c */
2310 void ext4_free_crypt_info(struct ext4_crypt_info *ci);
2311 void ext4_free_encryption_info(struct inode *inode, struct ext4_crypt_info *ci);
2312 int _ext4_get_encryption_info(struct inode *inode);
2313
2314 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2315 int ext4_has_encryption_key(struct inode *inode);
2316
2317 static inline int ext4_get_encryption_info(struct inode *inode)
2318 {
2319 struct ext4_crypt_info *ci = EXT4_I(inode)->i_crypt_info;
2320
2321 if (!ci ||
2322 (ci->ci_keyring_key &&
2323 (ci->ci_keyring_key->flags & ((1 << KEY_FLAG_INVALIDATED) |
2324 (1 << KEY_FLAG_REVOKED) |
2325 (1 << KEY_FLAG_DEAD)))))
2326 return _ext4_get_encryption_info(inode);
2327 return 0;
2328 }
2329
2330 static inline struct ext4_crypt_info *ext4_encryption_info(struct inode *inode)
2331 {
2332 return EXT4_I(inode)->i_crypt_info;
2333 }
2334
2335 #else
2336 static inline int ext4_has_encryption_key(struct inode *inode)
2337 {
2338 return 0;
2339 }
2340 static inline int ext4_get_encryption_info(struct inode *inode)
2341 {
2342 return 0;
2343 }
2344 static inline struct ext4_crypt_info *ext4_encryption_info(struct inode *inode)
2345 {
2346 return NULL;
2347 }
2348 #endif
2349
2350
2351 /* dir.c */
2352 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2353 struct file *,
2354 struct ext4_dir_entry_2 *,
2355 struct buffer_head *, char *, int,
2356 unsigned int);
2357 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2358 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2359 (de), (bh), (buf), (size), (offset)))
2360 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2361 __u32 minor_hash,
2362 struct ext4_dir_entry_2 *dirent,
2363 struct ext4_str *ent_name);
2364 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2365 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2366 struct buffer_head *bh,
2367 void *buf, int buf_size,
2368 struct ext4_filename *fname,
2369 struct ext4_dir_entry_2 **dest_de);
2370 int ext4_insert_dentry(struct inode *dir,
2371 struct inode *inode,
2372 struct ext4_dir_entry_2 *de,
2373 int buf_size,
2374 struct ext4_filename *fname);
2375 static inline void ext4_update_dx_flag(struct inode *inode)
2376 {
2377 if (!ext4_has_feature_dir_index(inode->i_sb))
2378 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2379 }
2380 static unsigned char ext4_filetype_table[] = {
2381 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2382 };
2383
2384 static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2385 {
2386 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2387 return DT_UNKNOWN;
2388
2389 return ext4_filetype_table[filetype];
2390 }
2391 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2392 void *buf, int buf_size);
2393
2394 /* fsync.c */
2395 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2396
2397 /* hash.c */
2398 extern int ext4fs_dirhash(const char *name, int len, struct
2399 dx_hash_info *hinfo);
2400
2401 /* ialloc.c */
2402 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2403 const struct qstr *qstr, __u32 goal,
2404 uid_t *owner, int handle_type,
2405 unsigned int line_no, int nblocks);
2406
2407 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner) \
2408 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2409 0, 0, 0)
2410 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2411 type, nblocks) \
2412 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2413 (type), __LINE__, (nblocks))
2414
2415
2416 extern void ext4_free_inode(handle_t *, struct inode *);
2417 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2418 extern unsigned long ext4_count_free_inodes(struct super_block *);
2419 extern unsigned long ext4_count_dirs(struct super_block *);
2420 extern void ext4_check_inodes_bitmap(struct super_block *);
2421 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2422 extern int ext4_init_inode_table(struct super_block *sb,
2423 ext4_group_t group, int barrier);
2424 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2425
2426 /* mballoc.c */
2427 extern const struct file_operations ext4_seq_mb_groups_fops;
2428 extern long ext4_mb_stats;
2429 extern long ext4_mb_max_to_scan;
2430 extern int ext4_mb_init(struct super_block *);
2431 extern int ext4_mb_release(struct super_block *);
2432 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2433 struct ext4_allocation_request *, int *);
2434 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2435 extern void ext4_discard_preallocations(struct inode *);
2436 extern int __init ext4_init_mballoc(void);
2437 extern void ext4_exit_mballoc(void);
2438 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2439 struct buffer_head *bh, ext4_fsblk_t block,
2440 unsigned long count, int flags);
2441 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2442 ext4_group_t ngroups);
2443 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2444 ext4_group_t i, struct ext4_group_desc *desc);
2445 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2446 ext4_fsblk_t block, unsigned long count);
2447 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2448
2449 /* inode.c */
2450 int ext4_inode_is_fast_symlink(struct inode *inode);
2451 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2452 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2453 int ext4_get_block_write(struct inode *inode, sector_t iblock,
2454 struct buffer_head *bh_result, int create);
2455 int ext4_dax_mmap_get_block(struct inode *inode, sector_t iblock,
2456 struct buffer_head *bh_result, int create);
2457 int ext4_get_block(struct inode *inode, sector_t iblock,
2458 struct buffer_head *bh_result, int create);
2459 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2460 struct buffer_head *bh, int create);
2461 int ext4_walk_page_buffers(handle_t *handle,
2462 struct buffer_head *head,
2463 unsigned from,
2464 unsigned to,
2465 int *partial,
2466 int (*fn)(handle_t *handle,
2467 struct buffer_head *bh));
2468 int do_journal_get_write_access(handle_t *handle,
2469 struct buffer_head *bh);
2470 #define FALL_BACK_TO_NONDELALLOC 1
2471 #define CONVERT_INLINE_DATA 2
2472
2473 extern struct inode *ext4_iget(struct super_block *, unsigned long);
2474 extern struct inode *ext4_iget_normal(struct super_block *, unsigned long);
2475 extern int ext4_write_inode(struct inode *, struct writeback_control *);
2476 extern int ext4_setattr(struct dentry *, struct iattr *);
2477 extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
2478 struct kstat *stat);
2479 extern void ext4_evict_inode(struct inode *);
2480 extern void ext4_clear_inode(struct inode *);
2481 extern int ext4_sync_inode(handle_t *, struct inode *);
2482 extern void ext4_dirty_inode(struct inode *, int);
2483 extern int ext4_change_inode_journal_flag(struct inode *, int);
2484 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2485 extern int ext4_inode_attach_jinode(struct inode *inode);
2486 extern int ext4_can_truncate(struct inode *inode);
2487 extern void ext4_truncate(struct inode *);
2488 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2489 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
2490 extern void ext4_set_inode_flags(struct inode *);
2491 extern void ext4_get_inode_flags(struct ext4_inode_info *);
2492 extern int ext4_alloc_da_blocks(struct inode *inode);
2493 extern void ext4_set_aops(struct inode *inode);
2494 extern int ext4_writepage_trans_blocks(struct inode *);
2495 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2496 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2497 loff_t lstart, loff_t lend);
2498 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
2499 extern int ext4_filemap_fault(struct vm_area_struct *vma, struct vm_fault *vmf);
2500 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2501 extern void ext4_da_update_reserve_space(struct inode *inode,
2502 int used, int quota_claim);
2503 extern int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk,
2504 ext4_fsblk_t pblk, ext4_lblk_t len);
2505
2506 /* indirect.c */
2507 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2508 struct ext4_map_blocks *map, int flags);
2509 extern ssize_t ext4_ind_direct_IO(struct kiocb *iocb, struct iov_iter *iter,
2510 loff_t offset);
2511 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2512 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2513 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2514 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2515 ext4_lblk_t start, ext4_lblk_t end);
2516
2517 /* ioctl.c */
2518 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2519 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2520
2521 /* migrate.c */
2522 extern int ext4_ext_migrate(struct inode *);
2523 extern int ext4_ind_migrate(struct inode *inode);
2524
2525 /* namei.c */
2526 extern int ext4_dirent_csum_verify(struct inode *inode,
2527 struct ext4_dir_entry *dirent);
2528 extern int ext4_orphan_add(handle_t *, struct inode *);
2529 extern int ext4_orphan_del(handle_t *, struct inode *);
2530 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2531 __u32 start_minor_hash, __u32 *next_hash);
2532 extern int ext4_search_dir(struct buffer_head *bh,
2533 char *search_buf,
2534 int buf_size,
2535 struct inode *dir,
2536 struct ext4_filename *fname,
2537 const struct qstr *d_name,
2538 unsigned int offset,
2539 struct ext4_dir_entry_2 **res_dir);
2540 extern int ext4_generic_delete_entry(handle_t *handle,
2541 struct inode *dir,
2542 struct ext4_dir_entry_2 *de_del,
2543 struct buffer_head *bh,
2544 void *entry_buf,
2545 int buf_size,
2546 int csum_size);
2547 extern int ext4_empty_dir(struct inode *inode);
2548
2549 /* resize.c */
2550 extern int ext4_group_add(struct super_block *sb,
2551 struct ext4_new_group_data *input);
2552 extern int ext4_group_extend(struct super_block *sb,
2553 struct ext4_super_block *es,
2554 ext4_fsblk_t n_blocks_count);
2555 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2556
2557 /* super.c */
2558 extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
2559 extern int ext4_calculate_overhead(struct super_block *sb);
2560 extern void ext4_superblock_csum_set(struct super_block *sb);
2561 extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2562 extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2563 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2564 ext4_group_t ngroup);
2565 extern const char *ext4_decode_error(struct super_block *sb, int errno,
2566 char nbuf[16]);
2567
2568 extern __printf(4, 5)
2569 void __ext4_error(struct super_block *, const char *, unsigned int,
2570 const char *, ...);
2571 extern __printf(5, 6)
2572 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2573 const char *, ...);
2574 extern __printf(5, 6)
2575 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2576 const char *, ...);
2577 extern void __ext4_std_error(struct super_block *, const char *,
2578 unsigned int, int);
2579 extern __printf(4, 5)
2580 void __ext4_abort(struct super_block *, const char *, unsigned int,
2581 const char *, ...);
2582 extern __printf(4, 5)
2583 void __ext4_warning(struct super_block *, const char *, unsigned int,
2584 const char *, ...);
2585 extern __printf(4, 5)
2586 void __ext4_warning_inode(const struct inode *inode, const char *function,
2587 unsigned int line, const char *fmt, ...);
2588 extern __printf(3, 4)
2589 void __ext4_msg(struct super_block *, const char *, const char *, ...);
2590 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2591 const char *, unsigned int, const char *);
2592 extern __printf(7, 8)
2593 void __ext4_grp_locked_error(const char *, unsigned int,
2594 struct super_block *, ext4_group_t,
2595 unsigned long, ext4_fsblk_t,
2596 const char *, ...);
2597
2598 #define EXT4_ERROR_INODE(inode, fmt, a...) \
2599 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
2600
2601 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
2602 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
2603
2604 #define EXT4_ERROR_FILE(file, block, fmt, a...) \
2605 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
2606
2607 #ifdef CONFIG_PRINTK
2608
2609 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2610 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__)
2611 #define ext4_error_file(file, func, line, block, fmt, ...) \
2612 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2613 #define ext4_error(sb, fmt, ...) \
2614 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2615 #define ext4_abort(sb, fmt, ...) \
2616 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2617 #define ext4_warning(sb, fmt, ...) \
2618 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2619 #define ext4_warning_inode(inode, fmt, ...) \
2620 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
2621 #define ext4_msg(sb, level, fmt, ...) \
2622 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2623 #define dump_mmp_msg(sb, mmp, msg) \
2624 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2625 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2626 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2627 fmt, ##__VA_ARGS__)
2628
2629 #else
2630
2631 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2632 do { \
2633 no_printk(fmt, ##__VA_ARGS__); \
2634 __ext4_error_inode(inode, "", 0, block, " "); \
2635 } while (0)
2636 #define ext4_error_file(file, func, line, block, fmt, ...) \
2637 do { \
2638 no_printk(fmt, ##__VA_ARGS__); \
2639 __ext4_error_file(file, "", 0, block, " "); \
2640 } while (0)
2641 #define ext4_error(sb, fmt, ...) \
2642 do { \
2643 no_printk(fmt, ##__VA_ARGS__); \
2644 __ext4_error(sb, "", 0, " "); \
2645 } while (0)
2646 #define ext4_abort(sb, fmt, ...) \
2647 do { \
2648 no_printk(fmt, ##__VA_ARGS__); \
2649 __ext4_abort(sb, "", 0, " "); \
2650 } while (0)
2651 #define ext4_warning(sb, fmt, ...) \
2652 do { \
2653 no_printk(fmt, ##__VA_ARGS__); \
2654 __ext4_warning(sb, "", 0, " "); \
2655 } while (0)
2656 #define ext4_warning_inode(inode, fmt, ...) \
2657 do { \
2658 no_printk(fmt, ##__VA_ARGS__); \
2659 __ext4_warning_inode(inode, "", 0, " "); \
2660 } while (0)
2661 #define ext4_msg(sb, level, fmt, ...) \
2662 do { \
2663 no_printk(fmt, ##__VA_ARGS__); \
2664 __ext4_msg(sb, "", " "); \
2665 } while (0)
2666 #define dump_mmp_msg(sb, mmp, msg) \
2667 __dump_mmp_msg(sb, mmp, "", 0, "")
2668 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2669 do { \
2670 no_printk(fmt, ##__VA_ARGS__); \
2671 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
2672 } while (0)
2673
2674 #endif
2675
2676 extern void ext4_update_dynamic_rev(struct super_block *sb);
2677 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2678 __u32 compat);
2679 extern int ext4_update_rocompat_feature(handle_t *handle,
2680 struct super_block *sb, __u32 rocompat);
2681 extern int ext4_update_incompat_feature(handle_t *handle,
2682 struct super_block *sb, __u32 incompat);
2683 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2684 struct ext4_group_desc *bg);
2685 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2686 struct ext4_group_desc *bg);
2687 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2688 struct ext4_group_desc *bg);
2689 extern __u32 ext4_free_group_clusters(struct super_block *sb,
2690 struct ext4_group_desc *bg);
2691 extern __u32 ext4_free_inodes_count(struct super_block *sb,
2692 struct ext4_group_desc *bg);
2693 extern __u32 ext4_used_dirs_count(struct super_block *sb,
2694 struct ext4_group_desc *bg);
2695 extern __u32 ext4_itable_unused_count(struct super_block *sb,
2696 struct ext4_group_desc *bg);
2697 extern void ext4_block_bitmap_set(struct super_block *sb,
2698 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2699 extern void ext4_inode_bitmap_set(struct super_block *sb,
2700 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2701 extern void ext4_inode_table_set(struct super_block *sb,
2702 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2703 extern void ext4_free_group_clusters_set(struct super_block *sb,
2704 struct ext4_group_desc *bg,
2705 __u32 count);
2706 extern void ext4_free_inodes_set(struct super_block *sb,
2707 struct ext4_group_desc *bg, __u32 count);
2708 extern void ext4_used_dirs_set(struct super_block *sb,
2709 struct ext4_group_desc *bg, __u32 count);
2710 extern void ext4_itable_unused_set(struct super_block *sb,
2711 struct ext4_group_desc *bg, __u32 count);
2712 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2713 struct ext4_group_desc *gdp);
2714 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2715 struct ext4_group_desc *gdp);
2716 extern int ext4_register_li_request(struct super_block *sb,
2717 ext4_group_t first_not_zeroed);
2718
2719 static inline int ext4_has_group_desc_csum(struct super_block *sb)
2720 {
2721 return ext4_has_feature_gdt_csum(sb) ||
2722 EXT4_SB(sb)->s_chksum_driver != NULL;
2723 }
2724
2725 static inline int ext4_has_metadata_csum(struct super_block *sb)
2726 {
2727 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
2728 !EXT4_SB(sb)->s_chksum_driver);
2729
2730 return (EXT4_SB(sb)->s_chksum_driver != NULL);
2731 }
2732 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2733 {
2734 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2735 le32_to_cpu(es->s_blocks_count_lo);
2736 }
2737
2738 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2739 {
2740 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2741 le32_to_cpu(es->s_r_blocks_count_lo);
2742 }
2743
2744 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2745 {
2746 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2747 le32_to_cpu(es->s_free_blocks_count_lo);
2748 }
2749
2750 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2751 ext4_fsblk_t blk)
2752 {
2753 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2754 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2755 }
2756
2757 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2758 ext4_fsblk_t blk)
2759 {
2760 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2761 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2762 }
2763
2764 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2765 ext4_fsblk_t blk)
2766 {
2767 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2768 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2769 }
2770
2771 static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
2772 {
2773 if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2774 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2775 le32_to_cpu(raw_inode->i_size_lo);
2776 else
2777 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2778 }
2779
2780 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2781 {
2782 raw_inode->i_size_lo = cpu_to_le32(i_size);
2783 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2784 }
2785
2786 static inline
2787 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2788 ext4_group_t group)
2789 {
2790 struct ext4_group_info ***grp_info;
2791 long indexv, indexh;
2792 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
2793 grp_info = EXT4_SB(sb)->s_group_info;
2794 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2795 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2796 return grp_info[indexv][indexh];
2797 }
2798
2799 /*
2800 * Reading s_groups_count requires using smp_rmb() afterwards. See
2801 * the locking protocol documented in the comments of ext4_group_add()
2802 * in resize.c
2803 */
2804 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2805 {
2806 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
2807
2808 smp_rmb();
2809 return ngroups;
2810 }
2811
2812 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2813 ext4_group_t block_group)
2814 {
2815 return block_group >> sbi->s_log_groups_per_flex;
2816 }
2817
2818 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2819 {
2820 return 1 << sbi->s_log_groups_per_flex;
2821 }
2822
2823 #define ext4_std_error(sb, errno) \
2824 do { \
2825 if ((errno)) \
2826 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
2827 } while (0)
2828
2829 #ifdef CONFIG_SMP
2830 /* Each CPU can accumulate percpu_counter_batch clusters in their local
2831 * counters. So we need to make sure we have free clusters more
2832 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
2833 */
2834 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2835 #else
2836 #define EXT4_FREECLUSTERS_WATERMARK 0
2837 #endif
2838
2839 /* Update i_disksize. Requires i_mutex to avoid races with truncate */
2840 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2841 {
2842 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
2843 !mutex_is_locked(&inode->i_mutex));
2844 down_write(&EXT4_I(inode)->i_data_sem);
2845 if (newsize > EXT4_I(inode)->i_disksize)
2846 EXT4_I(inode)->i_disksize = newsize;
2847 up_write(&EXT4_I(inode)->i_data_sem);
2848 }
2849
2850 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
2851 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
2852 {
2853 int changed = 0;
2854
2855 if (newsize > inode->i_size) {
2856 i_size_write(inode, newsize);
2857 changed = 1;
2858 }
2859 if (newsize > EXT4_I(inode)->i_disksize) {
2860 ext4_update_i_disksize(inode, newsize);
2861 changed |= 2;
2862 }
2863 return changed;
2864 }
2865
2866 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
2867 loff_t len);
2868
2869 struct ext4_group_info {
2870 unsigned long bb_state;
2871 struct rb_root bb_free_root;
2872 ext4_grpblk_t bb_first_free; /* first free block */
2873 ext4_grpblk_t bb_free; /* total free blocks */
2874 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
2875 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
2876 struct list_head bb_prealloc_list;
2877 #ifdef DOUBLE_CHECK
2878 void *bb_bitmap;
2879 #endif
2880 struct rw_semaphore alloc_sem;
2881 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
2882 * regions, index is order.
2883 * bb_counters[3] = 5 means
2884 * 5 free 8-block regions. */
2885 };
2886
2887 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
2888 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
2889 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
2890 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
2891
2892 #define EXT4_MB_GRP_NEED_INIT(grp) \
2893 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2894 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
2895 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2896 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
2897 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2898
2899 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
2900 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2901 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
2902 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2903 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
2904 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2905
2906 #define EXT4_MAX_CONTENTION 8
2907 #define EXT4_CONTENTION_THRESHOLD 2
2908
2909 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2910 ext4_group_t group)
2911 {
2912 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2913 }
2914
2915 /*
2916 * Returns true if the filesystem is busy enough that attempts to
2917 * access the block group locks has run into contention.
2918 */
2919 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2920 {
2921 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2922 }
2923
2924 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2925 {
2926 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2927 if (spin_trylock(lock))
2928 /*
2929 * We're able to grab the lock right away, so drop the
2930 * lock contention counter.
2931 */
2932 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2933 else {
2934 /*
2935 * The lock is busy, so bump the contention counter,
2936 * and then wait on the spin lock.
2937 */
2938 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2939 EXT4_MAX_CONTENTION);
2940 spin_lock(lock);
2941 }
2942 }
2943
2944 static inline void ext4_unlock_group(struct super_block *sb,
2945 ext4_group_t group)
2946 {
2947 spin_unlock(ext4_group_lock_ptr(sb, group));
2948 }
2949
2950 /*
2951 * Block validity checking
2952 */
2953 #define ext4_check_indirect_blockref(inode, bh) \
2954 ext4_check_blockref(__func__, __LINE__, inode, \
2955 (__le32 *)(bh)->b_data, \
2956 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2957
2958 #define ext4_ind_check_inode(inode) \
2959 ext4_check_blockref(__func__, __LINE__, inode, \
2960 EXT4_I(inode)->i_data, \
2961 EXT4_NDIR_BLOCKS)
2962
2963 /*
2964 * Inodes and files operations
2965 */
2966
2967 /* dir.c */
2968 extern const struct file_operations ext4_dir_operations;
2969
2970 /* file.c */
2971 extern const struct inode_operations ext4_file_inode_operations;
2972 extern const struct file_operations ext4_file_operations;
2973 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
2974
2975 /* inline.c */
2976 extern int ext4_get_max_inline_size(struct inode *inode);
2977 extern int ext4_find_inline_data_nolock(struct inode *inode);
2978 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
2979 unsigned int len);
2980 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
2981
2982 extern int ext4_readpage_inline(struct inode *inode, struct page *page);
2983 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
2984 struct inode *inode,
2985 loff_t pos, unsigned len,
2986 unsigned flags,
2987 struct page **pagep);
2988 extern int ext4_write_inline_data_end(struct inode *inode,
2989 loff_t pos, unsigned len,
2990 unsigned copied,
2991 struct page *page);
2992 extern struct buffer_head *
2993 ext4_journalled_write_inline_data(struct inode *inode,
2994 unsigned len,
2995 struct page *page);
2996 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
2997 struct inode *inode,
2998 loff_t pos, unsigned len,
2999 unsigned flags,
3000 struct page **pagep,
3001 void **fsdata);
3002 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
3003 unsigned len, unsigned copied,
3004 struct page *page);
3005 extern int ext4_try_add_inline_entry(handle_t *handle,
3006 struct ext4_filename *fname,
3007 struct inode *dir, struct inode *inode);
3008 extern int ext4_try_create_inline_dir(handle_t *handle,
3009 struct inode *parent,
3010 struct inode *inode);
3011 extern int ext4_read_inline_dir(struct file *filp,
3012 struct dir_context *ctx,
3013 int *has_inline_data);
3014 extern int htree_inlinedir_to_tree(struct file *dir_file,
3015 struct inode *dir, ext4_lblk_t block,
3016 struct dx_hash_info *hinfo,
3017 __u32 start_hash, __u32 start_minor_hash,
3018 int *has_inline_data);
3019 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3020 struct ext4_filename *fname,
3021 const struct qstr *d_name,
3022 struct ext4_dir_entry_2 **res_dir,
3023 int *has_inline_data);
3024 extern int ext4_delete_inline_entry(handle_t *handle,
3025 struct inode *dir,
3026 struct ext4_dir_entry_2 *de_del,
3027 struct buffer_head *bh,
3028 int *has_inline_data);
3029 extern int empty_inline_dir(struct inode *dir, int *has_inline_data);
3030 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3031 struct ext4_dir_entry_2 **parent_de,
3032 int *retval);
3033 extern int ext4_inline_data_fiemap(struct inode *inode,
3034 struct fiemap_extent_info *fieinfo,
3035 int *has_inline, __u64 start, __u64 len);
3036 extern int ext4_try_to_evict_inline_data(handle_t *handle,
3037 struct inode *inode,
3038 int needed);
3039 extern void ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3040
3041 extern int ext4_convert_inline_data(struct inode *inode);
3042
3043 static inline int ext4_has_inline_data(struct inode *inode)
3044 {
3045 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3046 EXT4_I(inode)->i_inline_off;
3047 }
3048
3049 /* namei.c */
3050 extern const struct inode_operations ext4_dir_inode_operations;
3051 extern const struct inode_operations ext4_special_inode_operations;
3052 extern struct dentry *ext4_get_parent(struct dentry *child);
3053 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3054 struct ext4_dir_entry_2 *de,
3055 int blocksize, int csum_size,
3056 unsigned int parent_ino, int dotdot_real_len);
3057 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t,
3058 unsigned int blocksize);
3059 extern int ext4_handle_dirty_dirent_node(handle_t *handle,
3060 struct inode *inode,
3061 struct buffer_head *bh);
3062 #define S_SHIFT 12
3063 static unsigned char ext4_type_by_mode[S_IFMT >> S_SHIFT] = {
3064 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
3065 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
3066 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
3067 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
3068 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
3069 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
3070 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
3071 };
3072
3073 static inline void ext4_set_de_type(struct super_block *sb,
3074 struct ext4_dir_entry_2 *de,
3075 umode_t mode) {
3076 if (ext4_has_feature_filetype(sb))
3077 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3078 }
3079
3080 /* readpages.c */
3081 extern int ext4_mpage_readpages(struct address_space *mapping,
3082 struct list_head *pages, struct page *page,
3083 unsigned nr_pages);
3084
3085 /* symlink.c */
3086 extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3087 extern const struct inode_operations ext4_symlink_inode_operations;
3088 extern const struct inode_operations ext4_fast_symlink_inode_operations;
3089
3090 /* sysfs.c */
3091 extern int ext4_register_sysfs(struct super_block *sb);
3092 extern void ext4_unregister_sysfs(struct super_block *sb);
3093 extern int __init ext4_init_sysfs(void);
3094 extern void ext4_exit_sysfs(void);
3095
3096 /* block_validity */
3097 extern void ext4_release_system_zone(struct super_block *sb);
3098 extern int ext4_setup_system_zone(struct super_block *sb);
3099 extern int __init ext4_init_system_zone(void);
3100 extern void ext4_exit_system_zone(void);
3101 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
3102 ext4_fsblk_t start_blk,
3103 unsigned int count);
3104 extern int ext4_check_blockref(const char *, unsigned int,
3105 struct inode *, __le32 *, unsigned int);
3106
3107 /* extents.c */
3108 struct ext4_ext_path;
3109 struct ext4_extent;
3110
3111 /*
3112 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3113 * __le32.
3114 */
3115 #define EXT_MAX_BLOCKS 0xffffffff
3116
3117 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
3118 extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
3119 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3120 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3121 struct ext4_map_blocks *map, int flags);
3122 extern void ext4_ext_truncate(handle_t *, struct inode *);
3123 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3124 ext4_lblk_t end);
3125 extern void ext4_ext_init(struct super_block *);
3126 extern void ext4_ext_release(struct super_block *);
3127 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3128 loff_t len);
3129 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3130 loff_t offset, ssize_t len);
3131 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3132 struct ext4_map_blocks *map, int flags);
3133 extern int ext4_ext_calc_metadata_amount(struct inode *inode,
3134 ext4_lblk_t lblocks);
3135 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3136 int num,
3137 struct ext4_ext_path *path);
3138 extern int ext4_can_extents_be_merged(struct inode *inode,
3139 struct ext4_extent *ex1,
3140 struct ext4_extent *ex2);
3141 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3142 struct ext4_ext_path **,
3143 struct ext4_extent *, int);
3144 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3145 struct ext4_ext_path **,
3146 int flags);
3147 extern void ext4_ext_drop_refs(struct ext4_ext_path *);
3148 extern int ext4_ext_check_inode(struct inode *inode);
3149 extern int ext4_find_delalloc_range(struct inode *inode,
3150 ext4_lblk_t lblk_start,
3151 ext4_lblk_t lblk_end);
3152 extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk);
3153 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3154 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3155 __u64 start, __u64 len);
3156 extern int ext4_ext_precache(struct inode *inode);
3157 extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len);
3158 extern int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len);
3159 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3160 struct inode *inode2, ext4_lblk_t lblk1,
3161 ext4_lblk_t lblk2, ext4_lblk_t count,
3162 int mark_unwritten,int *err);
3163
3164 /* move_extent.c */
3165 extern void ext4_double_down_write_data_sem(struct inode *first,
3166 struct inode *second);
3167 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3168 struct inode *donor_inode);
3169 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3170 __u64 start_orig, __u64 start_donor,
3171 __u64 len, __u64 *moved_len);
3172
3173 /* page-io.c */
3174 extern int __init ext4_init_pageio(void);
3175 extern void ext4_exit_pageio(void);
3176 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3177 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3178 extern int ext4_put_io_end(ext4_io_end_t *io_end);
3179 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3180 extern void ext4_io_submit_init(struct ext4_io_submit *io,
3181 struct writeback_control *wbc);
3182 extern void ext4_end_io_rsv_work(struct work_struct *work);
3183 extern void ext4_io_submit(struct ext4_io_submit *io);
3184 extern int ext4_bio_write_page(struct ext4_io_submit *io,
3185 struct page *page,
3186 int len,
3187 struct writeback_control *wbc,
3188 bool keep_towrite);
3189
3190 /* mmp.c */
3191 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3192
3193 /*
3194 * Add new method to test whether block and inode bitmaps are properly
3195 * initialized. With uninit_bg reading the block from disk is not enough
3196 * to mark the bitmap uptodate. We need to also zero-out the bitmap
3197 */
3198 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3199
3200 static inline int bitmap_uptodate(struct buffer_head *bh)
3201 {
3202 return (buffer_uptodate(bh) &&
3203 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3204 }
3205 static inline void set_bitmap_uptodate(struct buffer_head *bh)
3206 {
3207 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3208 }
3209
3210 /*
3211 * Disable DIO read nolock optimization, so new dioreaders will be forced
3212 * to grab i_mutex
3213 */
3214 static inline void ext4_inode_block_unlocked_dio(struct inode *inode)
3215 {
3216 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
3217 smp_mb();
3218 }
3219 static inline void ext4_inode_resume_unlocked_dio(struct inode *inode)
3220 {
3221 smp_mb();
3222 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
3223 }
3224
3225 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
3226
3227 /* For ioend & aio unwritten conversion wait queues */
3228 #define EXT4_WQ_HASH_SZ 37
3229 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
3230 EXT4_WQ_HASH_SZ])
3231 #define ext4_aio_mutex(v) (&ext4__aio_mutex[((unsigned long)(v)) %\
3232 EXT4_WQ_HASH_SZ])
3233 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3234 extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ];
3235
3236 #define EXT4_RESIZING 0
3237 extern int ext4_resize_begin(struct super_block *sb);
3238 extern void ext4_resize_end(struct super_block *sb);
3239
3240 #endif /* __KERNEL__ */
3241
3242 #define EFSBADCRC EBADMSG /* Bad CRC detected */
3243 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
3244
3245 #endif /* _EXT4_H */