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. */
ext4_mask_flags(umode_t mode,__u32 flags)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
ext4_check_flag_values(void)456 static inline void ext4_check_flag_values(void)
457 {
458 CHECK_FLAG_VALUE(SECRM);
459 CHECK_FLAG_VALUE(UNRM);
460 CHECK_FLAG_VALUE(COMPR);
461 CHECK_FLAG_VALUE(SYNC);
462 CHECK_FLAG_VALUE(IMMUTABLE);
463 CHECK_FLAG_VALUE(APPEND);
464 CHECK_FLAG_VALUE(NODUMP);
465 CHECK_FLAG_VALUE(NOATIME);
466 CHECK_FLAG_VALUE(DIRTY);
467 CHECK_FLAG_VALUE(COMPRBLK);
468 CHECK_FLAG_VALUE(NOCOMPR);
469 CHECK_FLAG_VALUE(ENCRYPT);
470 CHECK_FLAG_VALUE(INDEX);
471 CHECK_FLAG_VALUE(IMAGIC);
472 CHECK_FLAG_VALUE(JOURNAL_DATA);
473 CHECK_FLAG_VALUE(NOTAIL);
474 CHECK_FLAG_VALUE(DIRSYNC);
475 CHECK_FLAG_VALUE(TOPDIR);
476 CHECK_FLAG_VALUE(HUGE_FILE);
477 CHECK_FLAG_VALUE(EXTENTS);
478 CHECK_FLAG_VALUE(EA_INODE);
479 CHECK_FLAG_VALUE(EOFBLOCKS);
480 CHECK_FLAG_VALUE(INLINE_DATA);
481 CHECK_FLAG_VALUE(PROJINHERIT);
482 CHECK_FLAG_VALUE(RESERVED);
483 }
484
485 /* Used to pass group descriptor data when online resize is done */
486 struct ext4_new_group_input {
487 __u32 group; /* Group number for this data */
488 __u64 block_bitmap; /* Absolute block number of block bitmap */
489 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
490 __u64 inode_table; /* Absolute block number of inode table start */
491 __u32 blocks_count; /* Total number of blocks in this group */
492 __u16 reserved_blocks; /* Number of reserved blocks in this group */
493 __u16 unused;
494 };
495
496 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
497 struct compat_ext4_new_group_input {
498 u32 group;
499 compat_u64 block_bitmap;
500 compat_u64 inode_bitmap;
501 compat_u64 inode_table;
502 u32 blocks_count;
503 u16 reserved_blocks;
504 u16 unused;
505 };
506 #endif
507
508 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
509 struct ext4_new_group_data {
510 __u32 group;
511 __u64 block_bitmap;
512 __u64 inode_bitmap;
513 __u64 inode_table;
514 __u32 blocks_count;
515 __u16 reserved_blocks;
516 __u16 unused;
517 __u32 free_blocks_count;
518 };
519
520 /* Indexes used to index group tables in ext4_new_group_data */
521 enum {
522 BLOCK_BITMAP = 0, /* block bitmap */
523 INODE_BITMAP, /* inode bitmap */
524 INODE_TABLE, /* inode tables */
525 GROUP_TABLE_COUNT,
526 };
527
528 /*
529 * Flags used by ext4_map_blocks()
530 */
531 /* Allocate any needed blocks and/or convert an unwritten
532 extent to be an initialized ext4 */
533 #define EXT4_GET_BLOCKS_CREATE 0x0001
534 /* Request the creation of an unwritten extent */
535 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002
536 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\
537 EXT4_GET_BLOCKS_CREATE)
538 /* Caller is from the delayed allocation writeout path
539 * finally doing the actual allocation of delayed blocks */
540 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
541 /* caller is from the direct IO path, request to creation of an
542 unwritten extents if not allocated, split the unwritten
543 extent if blocks has been preallocated already*/
544 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
545 #define EXT4_GET_BLOCKS_CONVERT 0x0010
546 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
547 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
548 /* Convert extent to initialized after IO complete */
549 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
550 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
551 /* Eventual metadata allocation (due to growing extent tree)
552 * should not fail, so try to use reserved blocks for that.*/
553 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
554 /* Don't normalize allocation size (used for fallocate) */
555 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
556 /* Request will not result in inode size update (user for fallocate) */
557 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080
558 /* Do not take i_data_sem locking in ext4_map_blocks */
559 #define EXT4_GET_BLOCKS_NO_LOCK 0x0100
560 /* Convert written extents to unwritten */
561 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0200
562
563 /*
564 * The bit position of these flags must not overlap with any of the
565 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(),
566 * read_extent_tree_block(), ext4_split_extent_at(),
567 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
568 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
569 * caching the extents when reading from the extent tree while a
570 * truncate or punch hole operation is in progress.
571 */
572 #define EXT4_EX_NOCACHE 0x40000000
573 #define EXT4_EX_FORCE_CACHE 0x20000000
574
575 /*
576 * Flags used by ext4_free_blocks
577 */
578 #define EXT4_FREE_BLOCKS_METADATA 0x0001
579 #define EXT4_FREE_BLOCKS_FORGET 0x0002
580 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
581 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
582 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
583 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
584
585 /* Encryption algorithms */
586 #define EXT4_ENCRYPTION_MODE_INVALID 0
587 #define EXT4_ENCRYPTION_MODE_AES_256_XTS 1
588 #define EXT4_ENCRYPTION_MODE_AES_256_GCM 2
589 #define EXT4_ENCRYPTION_MODE_AES_256_CBC 3
590 #define EXT4_ENCRYPTION_MODE_AES_256_CTS 4
591
592 #include "ext4_crypto.h"
593
594 /*
595 * ioctl commands
596 */
597 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
598 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
599 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
600 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
601 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
602 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
603 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
604 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
605 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
606 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
607 #define EXT4_IOC_MIGRATE _IO('f', 9)
608 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
609 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
610 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
611 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
612 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
613 #define EXT4_IOC_SWAP_BOOT _IO('f', 17)
614 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18)
615 #define EXT4_IOC_SET_ENCRYPTION_POLICY _IOR('f', 19, struct ext4_encryption_policy)
616 #define EXT4_IOC_GET_ENCRYPTION_PWSALT _IOW('f', 20, __u8[16])
617 #define EXT4_IOC_GET_ENCRYPTION_POLICY _IOW('f', 21, struct ext4_encryption_policy)
618
619 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
620 /*
621 * ioctl commands in 32 bit emulation
622 */
623 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
624 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
625 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
626 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
627 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
628 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
629 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
630 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
631 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
632 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
633 #endif
634
635 /* Max physical block we can address w/o extents */
636 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
637
638 /*
639 * Structure of an inode on the disk
640 */
641 struct ext4_inode {
642 __le16 i_mode; /* File mode */
643 __le16 i_uid; /* Low 16 bits of Owner Uid */
644 __le32 i_size_lo; /* Size in bytes */
645 __le32 i_atime; /* Access time */
646 __le32 i_ctime; /* Inode Change time */
647 __le32 i_mtime; /* Modification time */
648 __le32 i_dtime; /* Deletion Time */
649 __le16 i_gid; /* Low 16 bits of Group Id */
650 __le16 i_links_count; /* Links count */
651 __le32 i_blocks_lo; /* Blocks count */
652 __le32 i_flags; /* File flags */
653 union {
654 struct {
655 __le32 l_i_version;
656 } linux1;
657 struct {
658 __u32 h_i_translator;
659 } hurd1;
660 struct {
661 __u32 m_i_reserved1;
662 } masix1;
663 } osd1; /* OS dependent 1 */
664 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
665 __le32 i_generation; /* File version (for NFS) */
666 __le32 i_file_acl_lo; /* File ACL */
667 __le32 i_size_high;
668 __le32 i_obso_faddr; /* Obsoleted fragment address */
669 union {
670 struct {
671 __le16 l_i_blocks_high; /* were l_i_reserved1 */
672 __le16 l_i_file_acl_high;
673 __le16 l_i_uid_high; /* these 2 fields */
674 __le16 l_i_gid_high; /* were reserved2[0] */
675 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
676 __le16 l_i_reserved;
677 } linux2;
678 struct {
679 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
680 __u16 h_i_mode_high;
681 __u16 h_i_uid_high;
682 __u16 h_i_gid_high;
683 __u32 h_i_author;
684 } hurd2;
685 struct {
686 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
687 __le16 m_i_file_acl_high;
688 __u32 m_i_reserved2[2];
689 } masix2;
690 } osd2; /* OS dependent 2 */
691 __le16 i_extra_isize;
692 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
693 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
694 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
695 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
696 __le32 i_crtime; /* File Creation time */
697 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
698 __le32 i_version_hi; /* high 32 bits for 64-bit version */
699 __le32 i_projid; /* Project ID */
700 };
701
702 struct move_extent {
703 __u32 reserved; /* should be zero */
704 __u32 donor_fd; /* donor file descriptor */
705 __u64 orig_start; /* logical start offset in block for orig */
706 __u64 donor_start; /* logical start offset in block for donor */
707 __u64 len; /* block length to be moved */
708 __u64 moved_len; /* moved block length */
709 };
710
711 #define EXT4_EPOCH_BITS 2
712 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
713 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
714
715 /*
716 * Extended fields will fit into an inode if the filesystem was formatted
717 * with large inodes (-I 256 or larger) and there are not currently any EAs
718 * consuming all of the available space. For new inodes we always reserve
719 * enough space for the kernel's known extended fields, but for inodes
720 * created with an old kernel this might not have been the case. None of
721 * the extended inode fields is critical for correct filesystem operation.
722 * This macro checks if a certain field fits in the inode. Note that
723 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
724 */
725 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
726 ((offsetof(typeof(*ext4_inode), field) + \
727 sizeof((ext4_inode)->field)) \
728 <= (EXT4_GOOD_OLD_INODE_SIZE + \
729 (einode)->i_extra_isize)) \
730
731 /*
732 * We use an encoding that preserves the times for extra epoch "00":
733 *
734 * extra msb of adjust for signed
735 * epoch 32-bit 32-bit tv_sec to
736 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
737 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
738 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
739 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
740 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
741 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
742 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
743 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
744 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
745 *
746 * Note that previous versions of the kernel on 64-bit systems would
747 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
748 * 1970. e2fsck will correct this, assuming that it is run on the
749 * affected filesystem before 2242.
750 */
751
ext4_encode_extra_time(struct timespec * time)752 static inline __le32 ext4_encode_extra_time(struct timespec *time)
753 {
754 u32 extra = sizeof(time->tv_sec) > 4 ?
755 ((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK : 0;
756 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
757 }
758
ext4_decode_extra_time(struct timespec * time,__le32 extra)759 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
760 {
761 if (unlikely(sizeof(time->tv_sec) > 4 &&
762 (extra & cpu_to_le32(EXT4_EPOCH_MASK)))) {
763 #if LINUX_VERSION_CODE < KERNEL_VERSION(4,20,0)
764 /* Handle legacy encoding of pre-1970 dates with epoch
765 * bits 1,1. We assume that by kernel version 4.20,
766 * everyone will have run fsck over the affected
767 * filesystems to correct the problem. (This
768 * backwards compatibility may be removed before this
769 * time, at the discretion of the ext4 developers.)
770 */
771 u64 extra_bits = le32_to_cpu(extra) & EXT4_EPOCH_MASK;
772 if (extra_bits == 3 && ((time->tv_sec) & 0x80000000) != 0)
773 extra_bits = 0;
774 time->tv_sec += extra_bits << 32;
775 #else
776 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
777 #endif
778 }
779 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
780 }
781
782 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
783 do { \
784 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
785 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
786 (raw_inode)->xtime ## _extra = \
787 ext4_encode_extra_time(&(inode)->xtime); \
788 } while (0)
789
790 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
791 do { \
792 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
793 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
794 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
795 (raw_inode)->xtime ## _extra = \
796 ext4_encode_extra_time(&(einode)->xtime); \
797 } while (0)
798
799 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
800 do { \
801 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
802 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
803 ext4_decode_extra_time(&(inode)->xtime, \
804 raw_inode->xtime ## _extra); \
805 else \
806 (inode)->xtime.tv_nsec = 0; \
807 } while (0)
808
809 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
810 do { \
811 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
812 (einode)->xtime.tv_sec = \
813 (signed)le32_to_cpu((raw_inode)->xtime); \
814 else \
815 (einode)->xtime.tv_sec = 0; \
816 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
817 ext4_decode_extra_time(&(einode)->xtime, \
818 raw_inode->xtime ## _extra); \
819 else \
820 (einode)->xtime.tv_nsec = 0; \
821 } while (0)
822
823 #define i_disk_version osd1.linux1.l_i_version
824
825 #if defined(__KERNEL__) || defined(__linux__)
826 #define i_reserved1 osd1.linux1.l_i_reserved1
827 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
828 #define i_blocks_high osd2.linux2.l_i_blocks_high
829 #define i_uid_low i_uid
830 #define i_gid_low i_gid
831 #define i_uid_high osd2.linux2.l_i_uid_high
832 #define i_gid_high osd2.linux2.l_i_gid_high
833 #define i_checksum_lo osd2.linux2.l_i_checksum_lo
834
835 #elif defined(__GNU__)
836
837 #define i_translator osd1.hurd1.h_i_translator
838 #define i_uid_high osd2.hurd2.h_i_uid_high
839 #define i_gid_high osd2.hurd2.h_i_gid_high
840 #define i_author osd2.hurd2.h_i_author
841
842 #elif defined(__masix__)
843
844 #define i_reserved1 osd1.masix1.m_i_reserved1
845 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
846 #define i_reserved2 osd2.masix2.m_i_reserved2
847
848 #endif /* defined(__KERNEL__) || defined(__linux__) */
849
850 #include "extents_status.h"
851
852 /*
853 * Lock subclasses for i_data_sem in the ext4_inode_info structure.
854 *
855 * These are needed to avoid lockdep false positives when we need to
856 * allocate blocks to the quota inode during ext4_map_blocks(), while
857 * holding i_data_sem for a normal (non-quota) inode. Since we don't
858 * do quota tracking for the quota inode, this avoids deadlock (as
859 * well as infinite recursion, since it isn't turtles all the way
860 * down...)
861 *
862 * I_DATA_SEM_NORMAL - Used for most inodes
863 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode
864 * where the second inode has larger inode number
865 * than the first
866 * I_DATA_SEM_QUOTA - Used for quota inodes only
867 */
868 enum {
869 I_DATA_SEM_NORMAL = 0,
870 I_DATA_SEM_OTHER,
871 I_DATA_SEM_QUOTA,
872 };
873
874
875 /*
876 * fourth extended file system inode data in memory
877 */
878 struct ext4_inode_info {
879 __le32 i_data[15]; /* unconverted */
880 __u32 i_dtime;
881 ext4_fsblk_t i_file_acl;
882
883 /*
884 * i_block_group is the number of the block group which contains
885 * this file's inode. Constant across the lifetime of the inode,
886 * it is ued for making block allocation decisions - we try to
887 * place a file's data blocks near its inode block, and new inodes
888 * near to their parent directory's inode.
889 */
890 ext4_group_t i_block_group;
891 ext4_lblk_t i_dir_start_lookup;
892 #if (BITS_PER_LONG < 64)
893 unsigned long i_state_flags; /* Dynamic state flags */
894 #endif
895 unsigned long i_flags;
896
897 /*
898 * Extended attributes can be read independently of the main file
899 * data. Taking i_mutex even when reading would cause contention
900 * between readers of EAs and writers of regular file data, so
901 * instead we synchronize on xattr_sem when reading or changing
902 * EAs.
903 */
904 struct rw_semaphore xattr_sem;
905
906 struct list_head i_orphan; /* unlinked but open inodes */
907
908 /*
909 * i_disksize keeps track of what the inode size is ON DISK, not
910 * in memory. During truncate, i_size is set to the new size by
911 * the VFS prior to calling ext4_truncate(), but the filesystem won't
912 * set i_disksize to 0 until the truncate is actually under way.
913 *
914 * The intent is that i_disksize always represents the blocks which
915 * are used by this file. This allows recovery to restart truncate
916 * on orphans if we crash during truncate. We actually write i_disksize
917 * into the on-disk inode when writing inodes out, instead of i_size.
918 *
919 * The only time when i_disksize and i_size may be different is when
920 * a truncate is in progress. The only things which change i_disksize
921 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
922 */
923 loff_t i_disksize;
924
925 /*
926 * i_data_sem is for serialising ext4_truncate() against
927 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
928 * data tree are chopped off during truncate. We can't do that in
929 * ext4 because whenever we perform intermediate commits during
930 * truncate, the inode and all the metadata blocks *must* be in a
931 * consistent state which allows truncation of the orphans to restart
932 * during recovery. Hence we must fix the get_block-vs-truncate race
933 * by other means, so we have i_data_sem.
934 */
935 struct rw_semaphore i_data_sem;
936 /*
937 * i_mmap_sem is for serializing page faults with truncate / punch hole
938 * operations. We have to make sure that new page cannot be faulted in
939 * a section of the inode that is being punched. We cannot easily use
940 * i_data_sem for this since we need protection for the whole punch
941 * operation and i_data_sem ranks below transaction start so we have
942 * to occasionally drop it.
943 */
944 struct rw_semaphore i_mmap_sem;
945 struct inode vfs_inode;
946 struct jbd2_inode *jinode;
947
948 spinlock_t i_raw_lock; /* protects updates to the raw inode */
949
950 /*
951 * File creation time. Its function is same as that of
952 * struct timespec i_{a,c,m}time in the generic inode.
953 */
954 struct timespec i_crtime;
955
956 /* mballoc */
957 struct list_head i_prealloc_list;
958 spinlock_t i_prealloc_lock;
959
960 /* extents status tree */
961 struct ext4_es_tree i_es_tree;
962 rwlock_t i_es_lock;
963 struct list_head i_es_list;
964 unsigned int i_es_all_nr; /* protected by i_es_lock */
965 unsigned int i_es_shk_nr; /* protected by i_es_lock */
966 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for
967 extents to shrink. Protected by
968 i_es_lock */
969
970 /* ialloc */
971 ext4_group_t i_last_alloc_group;
972
973 /* allocation reservation info for delalloc */
974 /* In case of bigalloc, these refer to clusters rather than blocks */
975 unsigned int i_reserved_data_blocks;
976 unsigned int i_reserved_meta_blocks;
977 unsigned int i_allocated_meta_blocks;
978 ext4_lblk_t i_da_metadata_calc_last_lblock;
979 int i_da_metadata_calc_len;
980
981 /* on-disk additional length */
982 __u16 i_extra_isize;
983
984 /* Indicate the inline data space. */
985 u16 i_inline_off;
986 u16 i_inline_size;
987
988 #ifdef CONFIG_QUOTA
989 /* quota space reservation, managed internally by quota code */
990 qsize_t i_reserved_quota;
991 #endif
992
993 /* Lock protecting lists below */
994 spinlock_t i_completed_io_lock;
995 /*
996 * Completed IOs that need unwritten extents handling and have
997 * transaction reserved
998 */
999 struct list_head i_rsv_conversion_list;
1000 /*
1001 * Completed IOs that need unwritten extents handling and don't have
1002 * transaction reserved
1003 */
1004 atomic_t i_ioend_count; /* Number of outstanding io_end structs */
1005 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1006 struct work_struct i_rsv_conversion_work;
1007
1008 spinlock_t i_block_reservation_lock;
1009
1010 /*
1011 * Transactions that contain inode's metadata needed to complete
1012 * fsync and fdatasync, respectively.
1013 */
1014 tid_t i_sync_tid;
1015 tid_t i_datasync_tid;
1016
1017 #ifdef CONFIG_QUOTA
1018 struct dquot *i_dquot[MAXQUOTAS];
1019 #endif
1020
1021 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1022 __u32 i_csum_seed;
1023
1024 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1025 /* Encryption params */
1026 struct ext4_crypt_info *i_crypt_info;
1027 #endif
1028 };
1029
1030 /*
1031 * File system states
1032 */
1033 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
1034 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
1035 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
1036
1037 /*
1038 * Misc. filesystem flags
1039 */
1040 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
1041 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
1042 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
1043
1044 /*
1045 * Mount flags set via mount options or defaults
1046 */
1047 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
1048 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
1049 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
1050 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
1051 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
1052 #define EXT4_MOUNT_ERRORS_MASK 0x00070
1053 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
1054 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
1055 #ifdef CONFIG_FS_DAX
1056 #define EXT4_MOUNT_DAX 0x00200 /* Direct Access */
1057 #else
1058 #define EXT4_MOUNT_DAX 0
1059 #endif
1060 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
1061 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
1062 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
1063 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
1064 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
1065 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
1066 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
1067 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
1068 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
1069 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
1070 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
1071 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
1072 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
1073 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
1074 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
1075 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1076 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
1077 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
1078 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
1079 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
1080 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
1081
1082 /*
1083 * Mount flags set either automatically (could not be set by mount option)
1084 * based on per file system feature or property or in special cases such as
1085 * distinguishing between explicit mount option definition and default.
1086 */
1087 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1088 specified delalloc */
1089 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1090 size of blocksize * 8
1091 blocks */
1092 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1093 file systems */
1094
1095 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly
1096 specified journal checksum */
1097
1098 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1099 ~EXT4_MOUNT_##opt
1100 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1101 EXT4_MOUNT_##opt
1102 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1103 EXT4_MOUNT_##opt)
1104
1105 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1106 ~EXT4_MOUNT2_##opt
1107 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1108 EXT4_MOUNT2_##opt
1109 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1110 EXT4_MOUNT2_##opt)
1111
1112 #define ext4_test_and_set_bit __test_and_set_bit_le
1113 #define ext4_set_bit __set_bit_le
1114 #define ext4_set_bit_atomic ext2_set_bit_atomic
1115 #define ext4_test_and_clear_bit __test_and_clear_bit_le
1116 #define ext4_clear_bit __clear_bit_le
1117 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
1118 #define ext4_test_bit test_bit_le
1119 #define ext4_find_next_zero_bit find_next_zero_bit_le
1120 #define ext4_find_next_bit find_next_bit_le
1121
1122 extern void ext4_set_bits(void *bm, int cur, int len);
1123
1124 /*
1125 * Maximal mount counts between two filesystem checks
1126 */
1127 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1128 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1129
1130 /*
1131 * Behaviour when detecting errors
1132 */
1133 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1134 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1135 #define EXT4_ERRORS_PANIC 3 /* Panic */
1136 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1137
1138 /* Metadata checksum algorithm codes */
1139 #define EXT4_CRC32C_CHKSUM 1
1140
1141 /*
1142 * Structure of the super block
1143 */
1144 struct ext4_super_block {
1145 /*00*/ __le32 s_inodes_count; /* Inodes count */
1146 __le32 s_blocks_count_lo; /* Blocks count */
1147 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1148 __le32 s_free_blocks_count_lo; /* Free blocks count */
1149 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1150 __le32 s_first_data_block; /* First Data Block */
1151 __le32 s_log_block_size; /* Block size */
1152 __le32 s_log_cluster_size; /* Allocation cluster size */
1153 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1154 __le32 s_clusters_per_group; /* # Clusters per group */
1155 __le32 s_inodes_per_group; /* # Inodes per group */
1156 __le32 s_mtime; /* Mount time */
1157 /*30*/ __le32 s_wtime; /* Write time */
1158 __le16 s_mnt_count; /* Mount count */
1159 __le16 s_max_mnt_count; /* Maximal mount count */
1160 __le16 s_magic; /* Magic signature */
1161 __le16 s_state; /* File system state */
1162 __le16 s_errors; /* Behaviour when detecting errors */
1163 __le16 s_minor_rev_level; /* minor revision level */
1164 /*40*/ __le32 s_lastcheck; /* time of last check */
1165 __le32 s_checkinterval; /* max. time between checks */
1166 __le32 s_creator_os; /* OS */
1167 __le32 s_rev_level; /* Revision level */
1168 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1169 __le16 s_def_resgid; /* Default gid for reserved blocks */
1170 /*
1171 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1172 *
1173 * Note: the difference between the compatible feature set and
1174 * the incompatible feature set is that if there is a bit set
1175 * in the incompatible feature set that the kernel doesn't
1176 * know about, it should refuse to mount the filesystem.
1177 *
1178 * e2fsck's requirements are more strict; if it doesn't know
1179 * about a feature in either the compatible or incompatible
1180 * feature set, it must abort and not try to meddle with
1181 * things it doesn't understand...
1182 */
1183 __le32 s_first_ino; /* First non-reserved inode */
1184 __le16 s_inode_size; /* size of inode structure */
1185 __le16 s_block_group_nr; /* block group # of this superblock */
1186 __le32 s_feature_compat; /* compatible feature set */
1187 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1188 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1189 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1190 /*78*/ char s_volume_name[16]; /* volume name */
1191 /*88*/ char s_last_mounted[64]; /* directory where last mounted */
1192 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1193 /*
1194 * Performance hints. Directory preallocation should only
1195 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1196 */
1197 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1198 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1199 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1200 /*
1201 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1202 */
1203 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1204 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1205 __le32 s_journal_dev; /* device number of journal file */
1206 __le32 s_last_orphan; /* start of list of inodes to delete */
1207 __le32 s_hash_seed[4]; /* HTREE hash seed */
1208 __u8 s_def_hash_version; /* Default hash version to use */
1209 __u8 s_jnl_backup_type;
1210 __le16 s_desc_size; /* size of group descriptor */
1211 /*100*/ __le32 s_default_mount_opts;
1212 __le32 s_first_meta_bg; /* First metablock block group */
1213 __le32 s_mkfs_time; /* When the filesystem was created */
1214 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1215 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1216 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1217 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1218 __le32 s_free_blocks_count_hi; /* Free blocks count */
1219 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1220 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1221 __le32 s_flags; /* Miscellaneous flags */
1222 __le16 s_raid_stride; /* RAID stride */
1223 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1224 __le64 s_mmp_block; /* Block for multi-mount protection */
1225 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1226 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1227 __u8 s_checksum_type; /* metadata checksum algorithm used */
1228 __u8 s_encryption_level; /* versioning level for encryption */
1229 __u8 s_reserved_pad; /* Padding to next 32bits */
1230 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1231 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1232 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1233 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1234 snapshot's future use */
1235 __le32 s_snapshot_list; /* inode number of the head of the
1236 on-disk snapshot list */
1237 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1238 __le32 s_error_count; /* number of fs errors */
1239 __le32 s_first_error_time; /* first time an error happened */
1240 __le32 s_first_error_ino; /* inode involved in first error */
1241 __le64 s_first_error_block; /* block involved of first error */
1242 __u8 s_first_error_func[32]; /* function where the error happened */
1243 __le32 s_first_error_line; /* line number where error happened */
1244 __le32 s_last_error_time; /* most recent time of an error */
1245 __le32 s_last_error_ino; /* inode involved in last error */
1246 __le32 s_last_error_line; /* line number where error happened */
1247 __le64 s_last_error_block; /* block involved of last error */
1248 __u8 s_last_error_func[32]; /* function where the error happened */
1249 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1250 __u8 s_mount_opts[64];
1251 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1252 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1253 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1254 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
1255 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */
1256 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */
1257 __le32 s_lpf_ino; /* Location of the lost+found inode */
1258 __le32 s_prj_quota_inum; /* inode for tracking project quota */
1259 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */
1260 __le32 s_reserved[98]; /* Padding to the end of the block */
1261 __le32 s_checksum; /* crc32c(superblock) */
1262 };
1263
1264 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1265
1266 #ifdef __KERNEL__
1267
1268 /*
1269 * run-time mount flags
1270 */
1271 #define EXT4_MF_MNTDIR_SAMPLED 0x0001
1272 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1273 #define EXT4_MF_TEST_DUMMY_ENCRYPTION 0x0004
1274
1275 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1276 #define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \
1277 EXT4_MF_TEST_DUMMY_ENCRYPTION))
1278 #else
1279 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1280 #endif
1281
1282 /* Number of quota types we support */
1283 #define EXT4_MAXQUOTAS 2
1284
1285 /*
1286 * fourth extended-fs super-block data in memory
1287 */
1288 struct ext4_sb_info {
1289 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1290 unsigned long s_inodes_per_block;/* Number of inodes per block */
1291 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1292 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1293 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1294 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1295 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1296 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1297 ext4_group_t s_groups_count; /* Number of groups in the fs */
1298 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1299 unsigned long s_overhead; /* # of fs overhead clusters */
1300 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1301 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1302 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1303 struct buffer_head * s_sbh; /* Buffer containing the super block */
1304 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1305 struct buffer_head **s_group_desc;
1306 unsigned int s_mount_opt;
1307 unsigned int s_mount_opt2;
1308 unsigned int s_mount_flags;
1309 unsigned int s_def_mount_opt;
1310 ext4_fsblk_t s_sb_block;
1311 atomic64_t s_resv_clusters;
1312 kuid_t s_resuid;
1313 kgid_t s_resgid;
1314 unsigned short s_mount_state;
1315 unsigned short s_pad;
1316 int s_addr_per_block_bits;
1317 int s_desc_per_block_bits;
1318 int s_inode_size;
1319 int s_first_ino;
1320 unsigned int s_inode_readahead_blks;
1321 unsigned int s_inode_goal;
1322 spinlock_t s_next_gen_lock;
1323 u32 s_next_generation;
1324 u32 s_hash_seed[4];
1325 int s_def_hash_version;
1326 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1327 struct percpu_counter s_freeclusters_counter;
1328 struct percpu_counter s_freeinodes_counter;
1329 struct percpu_counter s_dirs_counter;
1330 struct percpu_counter s_dirtyclusters_counter;
1331 struct blockgroup_lock *s_blockgroup_lock;
1332 struct proc_dir_entry *s_proc;
1333 struct kobject s_kobj;
1334 struct completion s_kobj_unregister;
1335 struct super_block *s_sb;
1336
1337 /* Journaling */
1338 struct journal_s *s_journal;
1339 struct list_head s_orphan;
1340 struct mutex s_orphan_lock;
1341 unsigned long s_resize_flags; /* Flags indicating if there
1342 is a resizer */
1343 unsigned long s_commit_interval;
1344 u32 s_max_batch_time;
1345 u32 s_min_batch_time;
1346 struct block_device *journal_bdev;
1347 #ifdef CONFIG_QUOTA
1348 char *s_qf_names[EXT4_MAXQUOTAS]; /* Names of quota files with journalled quota */
1349 int s_jquota_fmt; /* Format of quota to use */
1350 #endif
1351 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1352 struct rb_root system_blks;
1353
1354 #ifdef EXTENTS_STATS
1355 /* ext4 extents stats */
1356 unsigned long s_ext_min;
1357 unsigned long s_ext_max;
1358 unsigned long s_depth_max;
1359 spinlock_t s_ext_stats_lock;
1360 unsigned long s_ext_blocks;
1361 unsigned long s_ext_extents;
1362 #endif
1363
1364 /* for buddy allocator */
1365 struct ext4_group_info ***s_group_info;
1366 struct inode *s_buddy_cache;
1367 spinlock_t s_md_lock;
1368 unsigned short *s_mb_offsets;
1369 unsigned int *s_mb_maxs;
1370 unsigned int s_group_info_size;
1371
1372 /* tunables */
1373 unsigned long s_stripe;
1374 unsigned int s_mb_stream_request;
1375 unsigned int s_mb_max_to_scan;
1376 unsigned int s_mb_min_to_scan;
1377 unsigned int s_mb_stats;
1378 unsigned int s_mb_order2_reqs;
1379 unsigned int s_mb_group_prealloc;
1380 unsigned int s_max_dir_size_kb;
1381 /* where last allocation was done - for stream allocation */
1382 unsigned long s_mb_last_group;
1383 unsigned long s_mb_last_start;
1384
1385 /* stats for buddy allocator */
1386 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1387 atomic_t s_bal_success; /* we found long enough chunks */
1388 atomic_t s_bal_allocated; /* in blocks */
1389 atomic_t s_bal_ex_scanned; /* total extents scanned */
1390 atomic_t s_bal_goals; /* goal hits */
1391 atomic_t s_bal_breaks; /* too long searches */
1392 atomic_t s_bal_2orders; /* 2^order hits */
1393 spinlock_t s_bal_lock;
1394 unsigned long s_mb_buddies_generated;
1395 unsigned long long s_mb_generation_time;
1396 atomic_t s_mb_lost_chunks;
1397 atomic_t s_mb_preallocated;
1398 atomic_t s_mb_discarded;
1399 atomic_t s_lock_busy;
1400
1401 /* locality groups */
1402 struct ext4_locality_group __percpu *s_locality_groups;
1403
1404 /* for write statistics */
1405 unsigned long s_sectors_written_start;
1406 u64 s_kbytes_written;
1407
1408 /* the size of zero-out chunk */
1409 unsigned int s_extent_max_zeroout_kb;
1410
1411 unsigned int s_log_groups_per_flex;
1412 struct flex_groups *s_flex_groups;
1413 ext4_group_t s_flex_groups_allocated;
1414
1415 /* workqueue for reserved extent conversions (buffered io) */
1416 struct workqueue_struct *rsv_conversion_wq;
1417
1418 /* timer for periodic error stats printing */
1419 struct timer_list s_err_report;
1420
1421 /* Lazy inode table initialization info */
1422 struct ext4_li_request *s_li_request;
1423 /* Wait multiplier for lazy initialization thread */
1424 unsigned int s_li_wait_mult;
1425
1426 /* Kernel thread for multiple mount protection */
1427 struct task_struct *s_mmp_tsk;
1428
1429 /* record the last minlen when FITRIM is called. */
1430 atomic_t s_last_trim_minblks;
1431
1432 /* Reference to checksum algorithm driver via cryptoapi */
1433 struct crypto_shash *s_chksum_driver;
1434
1435 /* Precomputed FS UUID checksum for seeding other checksums */
1436 __u32 s_csum_seed;
1437
1438 /* Reclaim extents from extent status tree */
1439 struct shrinker s_es_shrinker;
1440 struct list_head s_es_list; /* List of inodes with reclaimable extents */
1441 long s_es_nr_inode;
1442 struct ext4_es_stats s_es_stats;
1443 struct mb_cache *s_mb_cache;
1444 spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1445
1446 /* Ratelimit ext4 messages. */
1447 struct ratelimit_state s_err_ratelimit_state;
1448 struct ratelimit_state s_warning_ratelimit_state;
1449 struct ratelimit_state s_msg_ratelimit_state;
1450 };
1451
EXT4_SB(struct super_block * sb)1452 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1453 {
1454 return sb->s_fs_info;
1455 }
EXT4_I(struct inode * inode)1456 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1457 {
1458 return container_of(inode, struct ext4_inode_info, vfs_inode);
1459 }
1460
ext4_current_time(struct inode * inode)1461 static inline struct timespec ext4_current_time(struct inode *inode)
1462 {
1463 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1464 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1465 }
1466
ext4_valid_inum(struct super_block * sb,unsigned long ino)1467 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1468 {
1469 return ino == EXT4_ROOT_INO ||
1470 ino == EXT4_USR_QUOTA_INO ||
1471 ino == EXT4_GRP_QUOTA_INO ||
1472 ino == EXT4_BOOT_LOADER_INO ||
1473 ino == EXT4_JOURNAL_INO ||
1474 ino == EXT4_RESIZE_INO ||
1475 (ino >= EXT4_FIRST_INO(sb) &&
1476 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1477 }
1478
ext4_set_io_unwritten_flag(struct inode * inode,struct ext4_io_end * io_end)1479 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
1480 struct ext4_io_end *io_end)
1481 {
1482 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
1483 io_end->flag |= EXT4_IO_END_UNWRITTEN;
1484 atomic_inc(&EXT4_I(inode)->i_unwritten);
1485 }
1486 }
1487
ext4_inode_aio(struct inode * inode)1488 static inline ext4_io_end_t *ext4_inode_aio(struct inode *inode)
1489 {
1490 return inode->i_private;
1491 }
1492
ext4_inode_aio_set(struct inode * inode,ext4_io_end_t * io)1493 static inline void ext4_inode_aio_set(struct inode *inode, ext4_io_end_t *io)
1494 {
1495 inode->i_private = io;
1496 }
1497
1498 /*
1499 * Inode dynamic state flags
1500 */
1501 enum {
1502 EXT4_STATE_JDATA, /* journaled data exists */
1503 EXT4_STATE_NEW, /* inode is newly created */
1504 EXT4_STATE_XATTR, /* has in-inode xattrs */
1505 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1506 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1507 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1508 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
1509 EXT4_STATE_NEWENTRY, /* File just added to dir */
1510 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read
1511 nolocking */
1512 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1513 EXT4_STATE_ORDERED_MODE, /* data=ordered mode */
1514 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
1515 };
1516
1517 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1518 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1519 { \
1520 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1521 } \
1522 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1523 { \
1524 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1525 } \
1526 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1527 { \
1528 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1529 }
1530
1531 /* Add these declarations here only so that these functions can be
1532 * found by name. Otherwise, they are very hard to locate. */
1533 static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1534 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1535 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1536 EXT4_INODE_BIT_FNS(flag, flags, 0)
1537
1538 /* Add these declarations here only so that these functions can be
1539 * found by name. Otherwise, they are very hard to locate. */
1540 static inline int ext4_test_inode_state(struct inode *inode, int bit);
1541 static inline void ext4_set_inode_state(struct inode *inode, int bit);
1542 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1543 #if (BITS_PER_LONG < 64)
1544 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1545
ext4_clear_state_flags(struct ext4_inode_info * ei)1546 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1547 {
1548 (ei)->i_state_flags = 0;
1549 }
1550 #else
1551 EXT4_INODE_BIT_FNS(state, flags, 32)
1552
ext4_clear_state_flags(struct ext4_inode_info * ei)1553 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1554 {
1555 /* We depend on the fact that callers will set i_flags */
1556 }
1557 #endif
1558 #else
1559 /* Assume that user mode programs are passing in an ext4fs superblock, not
1560 * a kernel struct super_block. This will allow us to call the feature-test
1561 * macros from user land. */
1562 #define EXT4_SB(sb) (sb)
1563 #endif
1564
1565 /*
1566 * Returns true if the inode is inode is encrypted
1567 */
ext4_encrypted_inode(struct inode * inode)1568 static inline int ext4_encrypted_inode(struct inode *inode)
1569 {
1570 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1571 return ext4_test_inode_flag(inode, EXT4_INODE_ENCRYPT);
1572 #else
1573 return 0;
1574 #endif
1575 }
1576
1577 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1578
1579 /*
1580 * Codes for operating systems
1581 */
1582 #define EXT4_OS_LINUX 0
1583 #define EXT4_OS_HURD 1
1584 #define EXT4_OS_MASIX 2
1585 #define EXT4_OS_FREEBSD 3
1586 #define EXT4_OS_LITES 4
1587
1588 /*
1589 * Revision levels
1590 */
1591 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1592 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1593
1594 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1595 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1596
1597 #define EXT4_GOOD_OLD_INODE_SIZE 128
1598
1599 /*
1600 * Feature set definitions
1601 */
1602
1603 /* Use the ext4_{has,set,clear}_feature_* helpers; these will be removed */
1604 #define EXT4_HAS_COMPAT_FEATURE(sb,mask) \
1605 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1606 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \
1607 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1608 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \
1609 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1610 #define EXT4_SET_COMPAT_FEATURE(sb,mask) \
1611 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1612 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \
1613 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1614 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \
1615 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1616 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \
1617 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1618 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
1619 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1620 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \
1621 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1622
1623 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1624 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1625 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1626 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1627 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1628 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1629 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
1630
1631 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1632 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1633 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1634 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1635 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1636 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1637 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1638 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1639 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1640 /*
1641 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1642 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1643 * all other data structures' checksums. However, the METADATA_CSUM and
1644 * GDT_CSUM bits are mutually exclusive.
1645 */
1646 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
1647 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000
1648 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000
1649
1650 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1651 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1652 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1653 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1654 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1655 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1656 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1657 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1658 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1659 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1660 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1661 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000
1662 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
1663 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
1664 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000
1665
1666 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
1667 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1668 { \
1669 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1670 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
1671 } \
1672 static inline void ext4_set_feature_##name(struct super_block *sb) \
1673 { \
1674 EXT4_SB(sb)->s_es->s_feature_compat |= \
1675 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1676 } \
1677 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1678 { \
1679 EXT4_SB(sb)->s_es->s_feature_compat &= \
1680 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1681 }
1682
1683 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
1684 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1685 { \
1686 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1687 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
1688 } \
1689 static inline void ext4_set_feature_##name(struct super_block *sb) \
1690 { \
1691 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
1692 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1693 } \
1694 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1695 { \
1696 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
1697 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1698 }
1699
1700 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
1701 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1702 { \
1703 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1704 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
1705 } \
1706 static inline void ext4_set_feature_##name(struct super_block *sb) \
1707 { \
1708 EXT4_SB(sb)->s_es->s_feature_incompat |= \
1709 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1710 } \
1711 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1712 { \
1713 EXT4_SB(sb)->s_es->s_feature_incompat &= \
1714 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1715 }
1716
EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc,DIR_PREALLOC)1717 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC)
1718 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES)
1719 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL)
1720 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR)
1721 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE)
1722 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX)
1723 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2)
1724
1725 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER)
1726 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE)
1727 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR)
1728 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE)
1729 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM)
1730 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK)
1731 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE)
1732 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA)
1733 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC)
1734 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM)
1735 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY)
1736 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT)
1737
1738 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION)
1739 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE)
1740 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER)
1741 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV)
1742 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG)
1743 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS)
1744 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT)
1745 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP)
1746 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG)
1747 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE)
1748 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA)
1749 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED)
1750 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR)
1751 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA)
1752 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT)
1753
1754 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1755 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1756 EXT4_FEATURE_INCOMPAT_META_BG)
1757 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1758 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1759 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1760
1761 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1762 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1763 EXT4_FEATURE_INCOMPAT_RECOVER| \
1764 EXT4_FEATURE_INCOMPAT_META_BG)
1765 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1766 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1767 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1768
1769 #define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1770 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1771 EXT4_FEATURE_INCOMPAT_RECOVER| \
1772 EXT4_FEATURE_INCOMPAT_META_BG| \
1773 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1774 EXT4_FEATURE_INCOMPAT_64BIT| \
1775 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1776 EXT4_FEATURE_INCOMPAT_MMP | \
1777 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
1778 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
1779 EXT4_FEATURE_INCOMPAT_CSUM_SEED)
1780 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1781 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1782 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1783 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1784 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1785 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1786 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1787 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1788 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1789 EXT4_FEATURE_RO_COMPAT_QUOTA)
1790
1791 #define EXTN_FEATURE_FUNCS(ver) \
1792 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
1793 { \
1794 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1795 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
1796 } \
1797 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
1798 { \
1799 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1800 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
1801 } \
1802 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
1803 { \
1804 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1805 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
1806 }
1807
1808 EXTN_FEATURE_FUNCS(2)
1809 EXTN_FEATURE_FUNCS(3)
1810 EXTN_FEATURE_FUNCS(4)
1811
1812 static inline bool ext4_has_compat_features(struct super_block *sb)
1813 {
1814 return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
1815 }
ext4_has_ro_compat_features(struct super_block * sb)1816 static inline bool ext4_has_ro_compat_features(struct super_block *sb)
1817 {
1818 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
1819 }
ext4_has_incompat_features(struct super_block * sb)1820 static inline bool ext4_has_incompat_features(struct super_block *sb)
1821 {
1822 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
1823 }
1824
1825 /*
1826 * Default values for user and/or group using reserved blocks
1827 */
1828 #define EXT4_DEF_RESUID 0
1829 #define EXT4_DEF_RESGID 0
1830
1831 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
1832
1833 /*
1834 * Default mount options
1835 */
1836 #define EXT4_DEFM_DEBUG 0x0001
1837 #define EXT4_DEFM_BSDGROUPS 0x0002
1838 #define EXT4_DEFM_XATTR_USER 0x0004
1839 #define EXT4_DEFM_ACL 0x0008
1840 #define EXT4_DEFM_UID16 0x0010
1841 #define EXT4_DEFM_JMODE 0x0060
1842 #define EXT4_DEFM_JMODE_DATA 0x0020
1843 #define EXT4_DEFM_JMODE_ORDERED 0x0040
1844 #define EXT4_DEFM_JMODE_WBACK 0x0060
1845 #define EXT4_DEFM_NOBARRIER 0x0100
1846 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1847 #define EXT4_DEFM_DISCARD 0x0400
1848 #define EXT4_DEFM_NODELALLOC 0x0800
1849
1850 /*
1851 * Default journal batch times
1852 */
1853 #define EXT4_DEF_MIN_BATCH_TIME 0
1854 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1855
1856 /*
1857 * Minimum number of groups in a flexgroup before we separate out
1858 * directories into the first block group of a flexgroup
1859 */
1860 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1861
1862 /*
1863 * Structure of a directory entry
1864 */
1865 #define EXT4_NAME_LEN 255
1866
1867 struct ext4_dir_entry {
1868 __le32 inode; /* Inode number */
1869 __le16 rec_len; /* Directory entry length */
1870 __le16 name_len; /* Name length */
1871 char name[EXT4_NAME_LEN]; /* File name */
1872 };
1873
1874 /*
1875 * The new version of the directory entry. Since EXT4 structures are
1876 * stored in intel byte order, and the name_len field could never be
1877 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1878 * file_type field.
1879 */
1880 struct ext4_dir_entry_2 {
1881 __le32 inode; /* Inode number */
1882 __le16 rec_len; /* Directory entry length */
1883 __u8 name_len; /* Name length */
1884 __u8 file_type;
1885 char name[EXT4_NAME_LEN]; /* File name */
1886 };
1887
1888 /*
1889 * This is a bogus directory entry at the end of each leaf block that
1890 * records checksums.
1891 */
1892 struct ext4_dir_entry_tail {
1893 __le32 det_reserved_zero1; /* Pretend to be unused */
1894 __le16 det_rec_len; /* 12 */
1895 __u8 det_reserved_zero2; /* Zero name length */
1896 __u8 det_reserved_ft; /* 0xDE, fake file type */
1897 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
1898 };
1899
1900 #define EXT4_DIRENT_TAIL(block, blocksize) \
1901 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1902 ((blocksize) - \
1903 sizeof(struct ext4_dir_entry_tail))))
1904
1905 /*
1906 * Ext4 directory file types. Only the low 3 bits are used. The
1907 * other bits are reserved for now.
1908 */
1909 #define EXT4_FT_UNKNOWN 0
1910 #define EXT4_FT_REG_FILE 1
1911 #define EXT4_FT_DIR 2
1912 #define EXT4_FT_CHRDEV 3
1913 #define EXT4_FT_BLKDEV 4
1914 #define EXT4_FT_FIFO 5
1915 #define EXT4_FT_SOCK 6
1916 #define EXT4_FT_SYMLINK 7
1917
1918 #define EXT4_FT_MAX 8
1919
1920 #define EXT4_FT_DIR_CSUM 0xDE
1921
1922 /*
1923 * EXT4_DIR_PAD defines the directory entries boundaries
1924 *
1925 * NOTE: It must be a multiple of 4
1926 */
1927 #define EXT4_DIR_PAD 4
1928 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1929 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1930 ~EXT4_DIR_ROUND)
1931 #define EXT4_MAX_REC_LEN ((1<<16)-1)
1932
1933 /*
1934 * If we ever get support for fs block sizes > page_size, we'll need
1935 * to remove the #if statements in the next two functions...
1936 */
1937 static inline unsigned int
ext4_rec_len_from_disk(__le16 dlen,unsigned blocksize)1938 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1939 {
1940 unsigned len = le16_to_cpu(dlen);
1941
1942 #if (PAGE_CACHE_SIZE >= 65536)
1943 if (len == EXT4_MAX_REC_LEN || len == 0)
1944 return blocksize;
1945 return (len & 65532) | ((len & 3) << 16);
1946 #else
1947 return len;
1948 #endif
1949 }
1950
ext4_rec_len_to_disk(unsigned len,unsigned blocksize)1951 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1952 {
1953 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1954 BUG();
1955 #if (PAGE_CACHE_SIZE >= 65536)
1956 if (len < 65536)
1957 return cpu_to_le16(len);
1958 if (len == blocksize) {
1959 if (blocksize == 65536)
1960 return cpu_to_le16(EXT4_MAX_REC_LEN);
1961 else
1962 return cpu_to_le16(0);
1963 }
1964 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1965 #else
1966 return cpu_to_le16(len);
1967 #endif
1968 }
1969
1970 /*
1971 * Hash Tree Directory indexing
1972 * (c) Daniel Phillips, 2001
1973 */
1974
1975 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
1976 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
1977 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1978 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1979
1980 /* Legal values for the dx_root hash_version field: */
1981
1982 #define DX_HASH_LEGACY 0
1983 #define DX_HASH_HALF_MD4 1
1984 #define DX_HASH_TEA 2
1985 #define DX_HASH_LEGACY_UNSIGNED 3
1986 #define DX_HASH_HALF_MD4_UNSIGNED 4
1987 #define DX_HASH_TEA_UNSIGNED 5
1988
ext4_chksum(struct ext4_sb_info * sbi,u32 crc,const void * address,unsigned int length)1989 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
1990 const void *address, unsigned int length)
1991 {
1992 struct {
1993 struct shash_desc shash;
1994 char ctx[4];
1995 } desc;
1996 int err;
1997
1998 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
1999
2000 desc.shash.tfm = sbi->s_chksum_driver;
2001 desc.shash.flags = 0;
2002 *(u32 *)desc.ctx = crc;
2003
2004 err = crypto_shash_update(&desc.shash, address, length);
2005 BUG_ON(err);
2006
2007 return *(u32 *)desc.ctx;
2008 }
2009
2010 #ifdef __KERNEL__
2011
2012 /* hash info structure used by the directory hash */
2013 struct dx_hash_info
2014 {
2015 u32 hash;
2016 u32 minor_hash;
2017 int hash_version;
2018 u32 *seed;
2019 };
2020
2021
2022 /* 32 and 64 bit signed EOF for dx directories */
2023 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
2024 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
2025
2026
2027 /*
2028 * Control parameters used by ext4_htree_next_block
2029 */
2030 #define HASH_NB_ALWAYS 1
2031
2032 struct ext4_filename {
2033 const struct qstr *usr_fname;
2034 struct ext4_str disk_name;
2035 struct dx_hash_info hinfo;
2036 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2037 struct ext4_str crypto_buf;
2038 #endif
2039 };
2040
2041 #define fname_name(p) ((p)->disk_name.name)
2042 #define fname_len(p) ((p)->disk_name.len)
2043
2044 /*
2045 * Describe an inode's exact location on disk and in memory
2046 */
2047 struct ext4_iloc
2048 {
2049 struct buffer_head *bh;
2050 unsigned long offset;
2051 ext4_group_t block_group;
2052 };
2053
ext4_raw_inode(struct ext4_iloc * iloc)2054 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2055 {
2056 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2057 }
2058
2059 /*
2060 * This structure is stuffed into the struct file's private_data field
2061 * for directories. It is where we put information so that we can do
2062 * readdir operations in hash tree order.
2063 */
2064 struct dir_private_info {
2065 struct rb_root root;
2066 struct rb_node *curr_node;
2067 struct fname *extra_fname;
2068 loff_t last_pos;
2069 __u32 curr_hash;
2070 __u32 curr_minor_hash;
2071 __u32 next_hash;
2072 };
2073
2074 /* calculate the first block number of the group */
2075 static inline ext4_fsblk_t
ext4_group_first_block_no(struct super_block * sb,ext4_group_t group_no)2076 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2077 {
2078 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2079 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2080 }
2081
2082 /*
2083 * Special error return code only used by dx_probe() and its callers.
2084 */
2085 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
2086
2087 /*
2088 * Timeout and state flag for lazy initialization inode thread.
2089 */
2090 #define EXT4_DEF_LI_WAIT_MULT 10
2091 #define EXT4_DEF_LI_MAX_START_DELAY 5
2092 #define EXT4_LAZYINIT_QUIT 0x0001
2093 #define EXT4_LAZYINIT_RUNNING 0x0002
2094
2095 /*
2096 * Lazy inode table initialization info
2097 */
2098 struct ext4_lazy_init {
2099 unsigned long li_state;
2100 struct list_head li_request_list;
2101 struct mutex li_list_mtx;
2102 };
2103
2104 struct ext4_li_request {
2105 struct super_block *lr_super;
2106 struct ext4_sb_info *lr_sbi;
2107 ext4_group_t lr_next_group;
2108 struct list_head lr_request;
2109 unsigned long lr_next_sched;
2110 unsigned long lr_timeout;
2111 };
2112
2113 struct ext4_features {
2114 struct kobject f_kobj;
2115 struct completion f_kobj_unregister;
2116 };
2117
2118 /*
2119 * This structure will be used for multiple mount protection. It will be
2120 * written into the block number saved in the s_mmp_block field in the
2121 * superblock. Programs that check MMP should assume that if
2122 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2123 * to use the filesystem, regardless of how old the timestamp is.
2124 */
2125 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
2126 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2127 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
2128 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
2129
2130 struct mmp_struct {
2131 __le32 mmp_magic; /* Magic number for MMP */
2132 __le32 mmp_seq; /* Sequence no. updated periodically */
2133
2134 /*
2135 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2136 * purposes and do not affect the correctness of the algorithm
2137 */
2138 __le64 mmp_time; /* Time last updated */
2139 char mmp_nodename[64]; /* Node which last updated MMP block */
2140 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
2141
2142 /*
2143 * mmp_check_interval is used to verify if the MMP block has been
2144 * updated on the block device. The value is updated based on the
2145 * maximum time to write the MMP block during an update cycle.
2146 */
2147 __le16 mmp_check_interval;
2148
2149 __le16 mmp_pad1;
2150 __le32 mmp_pad2[226];
2151 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
2152 };
2153
2154 /* arguments passed to the mmp thread */
2155 struct mmpd_data {
2156 struct buffer_head *bh; /* bh from initial read_mmp_block() */
2157 struct super_block *sb; /* super block of the fs */
2158 };
2159
2160 /*
2161 * Check interval multiplier
2162 * The MMP block is written every update interval and initially checked every
2163 * update interval x the multiplier (the value is then adapted based on the
2164 * write latency). The reason is that writes can be delayed under load and we
2165 * don't want readers to incorrectly assume that the filesystem is no longer
2166 * in use.
2167 */
2168 #define EXT4_MMP_CHECK_MULT 2UL
2169
2170 /*
2171 * Minimum interval for MMP checking in seconds.
2172 */
2173 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
2174
2175 /*
2176 * Maximum interval for MMP checking in seconds.
2177 */
2178 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
2179
2180 /*
2181 * Function prototypes
2182 */
2183
2184 /*
2185 * Ok, these declarations are also in <linux/kernel.h> but none of the
2186 * ext4 source programs needs to include it so they are duplicated here.
2187 */
2188 # define NORET_TYPE /**/
2189 # define ATTRIB_NORET __attribute__((noreturn))
2190 # define NORET_AND noreturn,
2191
2192 /* bitmap.c */
2193 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2194 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2195 struct ext4_group_desc *gdp,
2196 struct buffer_head *bh, int sz);
2197 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2198 struct ext4_group_desc *gdp,
2199 struct buffer_head *bh, int sz);
2200 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2201 struct ext4_group_desc *gdp,
2202 struct buffer_head *bh);
2203 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2204 struct ext4_group_desc *gdp,
2205 struct buffer_head *bh);
2206
2207 /* balloc.c */
2208 extern void ext4_get_group_no_and_offset(struct super_block *sb,
2209 ext4_fsblk_t blocknr,
2210 ext4_group_t *blockgrpp,
2211 ext4_grpblk_t *offsetp);
2212 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2213 ext4_fsblk_t block);
2214
2215 extern unsigned int ext4_block_group(struct super_block *sb,
2216 ext4_fsblk_t blocknr);
2217 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2218 ext4_fsblk_t blocknr);
2219 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2220 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2221 ext4_group_t group);
2222 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2223 ext4_fsblk_t goal,
2224 unsigned int flags,
2225 unsigned long *count,
2226 int *errp);
2227 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2228 s64 nclusters, unsigned int flags);
2229 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2230 extern void ext4_check_blocks_bitmap(struct super_block *);
2231 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2232 ext4_group_t block_group,
2233 struct buffer_head ** bh);
2234 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2235
2236 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2237 ext4_group_t block_group);
2238 extern int ext4_wait_block_bitmap(struct super_block *sb,
2239 ext4_group_t block_group,
2240 struct buffer_head *bh);
2241 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2242 ext4_group_t block_group);
2243 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2244 ext4_group_t block_group,
2245 struct ext4_group_desc *gdp);
2246 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2247
2248 /* crypto_policy.c */
2249 int ext4_is_child_context_consistent_with_parent(struct inode *parent,
2250 struct inode *child);
2251 int ext4_inherit_context(struct inode *parent, struct inode *child);
2252 void ext4_to_hex(char *dst, char *src, size_t src_size);
2253 int ext4_process_policy(const struct ext4_encryption_policy *policy,
2254 struct inode *inode);
2255 int ext4_get_policy(struct inode *inode,
2256 struct ext4_encryption_policy *policy);
2257
2258 /* crypto.c */
2259 extern struct kmem_cache *ext4_crypt_info_cachep;
2260 bool ext4_valid_contents_enc_mode(uint32_t mode);
2261 uint32_t ext4_validate_encryption_key_size(uint32_t mode, uint32_t size);
2262 extern struct workqueue_struct *ext4_read_workqueue;
2263 struct ext4_crypto_ctx *ext4_get_crypto_ctx(struct inode *inode);
2264 void ext4_release_crypto_ctx(struct ext4_crypto_ctx *ctx);
2265 void ext4_restore_control_page(struct page *data_page);
2266 struct page *ext4_encrypt(struct inode *inode,
2267 struct page *plaintext_page);
2268 int ext4_decrypt(struct page *page);
2269 int ext4_encrypted_zeroout(struct inode *inode, struct ext4_extent *ex);
2270
2271 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2272 int ext4_init_crypto(void);
2273 void ext4_exit_crypto(void);
ext4_sb_has_crypto(struct super_block * sb)2274 static inline int ext4_sb_has_crypto(struct super_block *sb)
2275 {
2276 return ext4_has_feature_encrypt(sb);
2277 }
2278 #else
ext4_init_crypto(void)2279 static inline int ext4_init_crypto(void) { return 0; }
ext4_exit_crypto(void)2280 static inline void ext4_exit_crypto(void) { }
ext4_sb_has_crypto(struct super_block * sb)2281 static inline int ext4_sb_has_crypto(struct super_block *sb)
2282 {
2283 return 0;
2284 }
2285 #endif
2286
2287 /* crypto_fname.c */
2288 bool ext4_valid_filenames_enc_mode(uint32_t mode);
2289 u32 ext4_fname_crypto_round_up(u32 size, u32 blksize);
2290 unsigned ext4_fname_encrypted_size(struct inode *inode, u32 ilen);
2291 int ext4_fname_crypto_alloc_buffer(struct inode *inode,
2292 u32 ilen, struct ext4_str *crypto_str);
2293 int _ext4_fname_disk_to_usr(struct inode *inode,
2294 struct dx_hash_info *hinfo,
2295 const struct ext4_str *iname,
2296 struct ext4_str *oname);
2297 int ext4_fname_disk_to_usr(struct inode *inode,
2298 struct dx_hash_info *hinfo,
2299 const struct ext4_dir_entry_2 *de,
2300 struct ext4_str *oname);
2301 int ext4_fname_usr_to_disk(struct inode *inode,
2302 const struct qstr *iname,
2303 struct ext4_str *oname);
2304 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2305 void ext4_fname_crypto_free_buffer(struct ext4_str *crypto_str);
2306 int ext4_fname_setup_filename(struct inode *dir, const struct qstr *iname,
2307 int lookup, struct ext4_filename *fname);
2308 void ext4_fname_free_filename(struct ext4_filename *fname);
2309 #else
2310 static inline
ext4_setup_fname_crypto(struct inode * inode)2311 int ext4_setup_fname_crypto(struct inode *inode)
2312 {
2313 return 0;
2314 }
ext4_fname_crypto_free_buffer(struct ext4_str * p)2315 static inline void ext4_fname_crypto_free_buffer(struct ext4_str *p) { }
ext4_fname_setup_filename(struct inode * dir,const struct qstr * iname,int lookup,struct ext4_filename * fname)2316 static inline int ext4_fname_setup_filename(struct inode *dir,
2317 const struct qstr *iname,
2318 int lookup, struct ext4_filename *fname)
2319 {
2320 fname->usr_fname = iname;
2321 fname->disk_name.name = (unsigned char *) iname->name;
2322 fname->disk_name.len = iname->len;
2323 return 0;
2324 }
ext4_fname_free_filename(struct ext4_filename * fname)2325 static inline void ext4_fname_free_filename(struct ext4_filename *fname) { }
2326 #endif
2327
2328
2329 /* crypto_key.c */
2330 void ext4_free_crypt_info(struct ext4_crypt_info *ci);
2331 void ext4_free_encryption_info(struct inode *inode, struct ext4_crypt_info *ci);
2332 int _ext4_get_encryption_info(struct inode *inode);
2333
2334 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2335 int ext4_has_encryption_key(struct inode *inode);
2336
ext4_get_encryption_info(struct inode * inode)2337 static inline int ext4_get_encryption_info(struct inode *inode)
2338 {
2339 struct ext4_crypt_info *ci = EXT4_I(inode)->i_crypt_info;
2340
2341 if (!ci ||
2342 (ci->ci_keyring_key &&
2343 (ci->ci_keyring_key->flags & ((1 << KEY_FLAG_INVALIDATED) |
2344 (1 << KEY_FLAG_REVOKED) |
2345 (1 << KEY_FLAG_DEAD)))))
2346 return _ext4_get_encryption_info(inode);
2347 return 0;
2348 }
2349
ext4_encryption_info(struct inode * inode)2350 static inline struct ext4_crypt_info *ext4_encryption_info(struct inode *inode)
2351 {
2352 return EXT4_I(inode)->i_crypt_info;
2353 }
2354
2355 #else
ext4_has_encryption_key(struct inode * inode)2356 static inline int ext4_has_encryption_key(struct inode *inode)
2357 {
2358 return 0;
2359 }
ext4_get_encryption_info(struct inode * inode)2360 static inline int ext4_get_encryption_info(struct inode *inode)
2361 {
2362 return 0;
2363 }
ext4_encryption_info(struct inode * inode)2364 static inline struct ext4_crypt_info *ext4_encryption_info(struct inode *inode)
2365 {
2366 return NULL;
2367 }
2368 #endif
2369
2370
2371 /* dir.c */
2372 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2373 struct file *,
2374 struct ext4_dir_entry_2 *,
2375 struct buffer_head *, char *, int,
2376 unsigned int);
2377 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2378 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2379 (de), (bh), (buf), (size), (offset)))
2380 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2381 __u32 minor_hash,
2382 struct ext4_dir_entry_2 *dirent,
2383 struct ext4_str *ent_name);
2384 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2385 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2386 struct buffer_head *bh,
2387 void *buf, int buf_size,
2388 struct ext4_filename *fname,
2389 struct ext4_dir_entry_2 **dest_de);
2390 int ext4_insert_dentry(struct inode *dir,
2391 struct inode *inode,
2392 struct ext4_dir_entry_2 *de,
2393 int buf_size,
2394 struct ext4_filename *fname);
ext4_update_dx_flag(struct inode * inode)2395 static inline void ext4_update_dx_flag(struct inode *inode)
2396 {
2397 if (!ext4_has_feature_dir_index(inode->i_sb))
2398 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2399 }
2400 static unsigned char ext4_filetype_table[] = {
2401 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2402 };
2403
get_dtype(struct super_block * sb,int filetype)2404 static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2405 {
2406 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2407 return DT_UNKNOWN;
2408
2409 return ext4_filetype_table[filetype];
2410 }
2411 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2412 void *buf, int buf_size);
2413
2414 /* fsync.c */
2415 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2416
2417 /* hash.c */
2418 extern int ext4fs_dirhash(const char *name, int len, struct
2419 dx_hash_info *hinfo);
2420
2421 /* ialloc.c */
2422 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2423 const struct qstr *qstr, __u32 goal,
2424 uid_t *owner, int handle_type,
2425 unsigned int line_no, int nblocks);
2426
2427 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner) \
2428 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2429 0, 0, 0)
2430 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2431 type, nblocks) \
2432 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2433 (type), __LINE__, (nblocks))
2434
2435
2436 extern void ext4_free_inode(handle_t *, struct inode *);
2437 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2438 extern unsigned long ext4_count_free_inodes(struct super_block *);
2439 extern unsigned long ext4_count_dirs(struct super_block *);
2440 extern void ext4_check_inodes_bitmap(struct super_block *);
2441 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2442 extern int ext4_init_inode_table(struct super_block *sb,
2443 ext4_group_t group, int barrier);
2444 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2445
2446 /* mballoc.c */
2447 extern const struct file_operations ext4_seq_mb_groups_fops;
2448 extern long ext4_mb_stats;
2449 extern long ext4_mb_max_to_scan;
2450 extern int ext4_mb_init(struct super_block *);
2451 extern int ext4_mb_release(struct super_block *);
2452 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2453 struct ext4_allocation_request *, int *);
2454 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2455 extern void ext4_discard_preallocations(struct inode *);
2456 extern int __init ext4_init_mballoc(void);
2457 extern void ext4_exit_mballoc(void);
2458 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2459 struct buffer_head *bh, ext4_fsblk_t block,
2460 unsigned long count, int flags);
2461 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2462 ext4_group_t ngroups);
2463 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2464 ext4_group_t i, struct ext4_group_desc *desc);
2465 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2466 ext4_fsblk_t block, unsigned long count);
2467 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *);
2468
2469 /* inode.c */
2470 int ext4_inode_is_fast_symlink(struct inode *inode);
2471 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2472 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2473 int ext4_get_block_write(struct inode *inode, sector_t iblock,
2474 struct buffer_head *bh_result, int create);
2475 int ext4_get_block_dax(struct inode *inode, sector_t iblock,
2476 struct buffer_head *bh_result, int create);
2477 int ext4_get_block(struct inode *inode, sector_t iblock,
2478 struct buffer_head *bh_result, int create);
2479 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2480 struct buffer_head *bh, int create);
2481 int ext4_walk_page_buffers(handle_t *handle,
2482 struct buffer_head *head,
2483 unsigned from,
2484 unsigned to,
2485 int *partial,
2486 int (*fn)(handle_t *handle,
2487 struct buffer_head *bh));
2488 int do_journal_get_write_access(handle_t *handle,
2489 struct buffer_head *bh);
2490 #define FALL_BACK_TO_NONDELALLOC 1
2491 #define CONVERT_INLINE_DATA 2
2492
2493 extern struct inode *ext4_iget(struct super_block *, unsigned long);
2494 extern struct inode *ext4_iget_normal(struct super_block *, unsigned long);
2495 extern int ext4_write_inode(struct inode *, struct writeback_control *);
2496 extern int ext4_setattr(struct dentry *, struct iattr *);
2497 extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
2498 struct kstat *stat);
2499 extern void ext4_evict_inode(struct inode *);
2500 extern void ext4_clear_inode(struct inode *);
2501 extern int ext4_sync_inode(handle_t *, struct inode *);
2502 extern void ext4_dirty_inode(struct inode *, int);
2503 extern int ext4_change_inode_journal_flag(struct inode *, int);
2504 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2505 extern int ext4_inode_attach_jinode(struct inode *inode);
2506 extern int ext4_can_truncate(struct inode *inode);
2507 extern void ext4_truncate(struct inode *);
2508 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2509 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
2510 extern void ext4_set_inode_flags(struct inode *);
2511 extern void ext4_get_inode_flags(struct ext4_inode_info *);
2512 extern int ext4_alloc_da_blocks(struct inode *inode);
2513 extern void ext4_set_aops(struct inode *inode);
2514 extern int ext4_writepage_trans_blocks(struct inode *);
2515 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2516 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2517 loff_t lstart, loff_t lend);
2518 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
2519 extern int ext4_filemap_fault(struct vm_area_struct *vma, struct vm_fault *vmf);
2520 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2521 extern void ext4_da_update_reserve_space(struct inode *inode,
2522 int used, int quota_claim);
2523
2524 /* indirect.c */
2525 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2526 struct ext4_map_blocks *map, int flags);
2527 extern ssize_t ext4_ind_direct_IO(struct kiocb *iocb, struct iov_iter *iter,
2528 loff_t offset);
2529 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2530 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2531 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2532 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2533 ext4_lblk_t start, ext4_lblk_t end);
2534
2535 /* ioctl.c */
2536 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2537 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2538
2539 /* migrate.c */
2540 extern int ext4_ext_migrate(struct inode *);
2541 extern int ext4_ind_migrate(struct inode *inode);
2542
2543 /* namei.c */
2544 extern int ext4_dirent_csum_verify(struct inode *inode,
2545 struct ext4_dir_entry *dirent);
2546 extern int ext4_orphan_add(handle_t *, struct inode *);
2547 extern int ext4_orphan_del(handle_t *, struct inode *);
2548 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2549 __u32 start_minor_hash, __u32 *next_hash);
2550 extern int ext4_search_dir(struct buffer_head *bh,
2551 char *search_buf,
2552 int buf_size,
2553 struct inode *dir,
2554 struct ext4_filename *fname,
2555 const struct qstr *d_name,
2556 unsigned int offset,
2557 struct ext4_dir_entry_2 **res_dir);
2558 extern int ext4_generic_delete_entry(handle_t *handle,
2559 struct inode *dir,
2560 struct ext4_dir_entry_2 *de_del,
2561 struct buffer_head *bh,
2562 void *entry_buf,
2563 int buf_size,
2564 int csum_size);
2565 extern int ext4_empty_dir(struct inode *inode);
2566
2567 /* resize.c */
2568 extern int ext4_group_add(struct super_block *sb,
2569 struct ext4_new_group_data *input);
2570 extern int ext4_group_extend(struct super_block *sb,
2571 struct ext4_super_block *es,
2572 ext4_fsblk_t n_blocks_count);
2573 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2574
2575 /* super.c */
2576 extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
2577 extern int ext4_calculate_overhead(struct super_block *sb);
2578 extern void ext4_superblock_csum_set(struct super_block *sb);
2579 extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2580 extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2581 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2582 ext4_group_t ngroup);
2583 extern const char *ext4_decode_error(struct super_block *sb, int errno,
2584 char nbuf[16]);
2585
2586 extern __printf(4, 5)
2587 void __ext4_error(struct super_block *, const char *, unsigned int,
2588 const char *, ...);
2589 extern __printf(5, 6)
2590 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2591 const char *, ...);
2592 extern __printf(5, 6)
2593 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2594 const char *, ...);
2595 extern void __ext4_std_error(struct super_block *, const char *,
2596 unsigned int, int);
2597 extern __printf(4, 5)
2598 void __ext4_abort(struct super_block *, const char *, unsigned int,
2599 const char *, ...);
2600 extern __printf(4, 5)
2601 void __ext4_warning(struct super_block *, const char *, unsigned int,
2602 const char *, ...);
2603 extern __printf(4, 5)
2604 void __ext4_warning_inode(const struct inode *inode, const char *function,
2605 unsigned int line, const char *fmt, ...);
2606 extern __printf(3, 4)
2607 void __ext4_msg(struct super_block *, const char *, const char *, ...);
2608 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2609 const char *, unsigned int, const char *);
2610 extern __printf(7, 8)
2611 void __ext4_grp_locked_error(const char *, unsigned int,
2612 struct super_block *, ext4_group_t,
2613 unsigned long, ext4_fsblk_t,
2614 const char *, ...);
2615
2616 #define EXT4_ERROR_INODE(inode, fmt, a...) \
2617 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
2618
2619 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
2620 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
2621
2622 #define EXT4_ERROR_FILE(file, block, fmt, a...) \
2623 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
2624
2625 #ifdef CONFIG_PRINTK
2626
2627 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2628 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__)
2629 #define ext4_error_file(file, func, line, block, fmt, ...) \
2630 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2631 #define ext4_error(sb, fmt, ...) \
2632 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2633 #define ext4_abort(sb, fmt, ...) \
2634 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2635 #define ext4_warning(sb, fmt, ...) \
2636 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2637 #define ext4_warning_inode(inode, fmt, ...) \
2638 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
2639 #define ext4_msg(sb, level, fmt, ...) \
2640 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2641 #define dump_mmp_msg(sb, mmp, msg) \
2642 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2643 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2644 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2645 fmt, ##__VA_ARGS__)
2646
2647 #else
2648
2649 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2650 do { \
2651 no_printk(fmt, ##__VA_ARGS__); \
2652 __ext4_error_inode(inode, "", 0, block, " "); \
2653 } while (0)
2654 #define ext4_error_file(file, func, line, block, fmt, ...) \
2655 do { \
2656 no_printk(fmt, ##__VA_ARGS__); \
2657 __ext4_error_file(file, "", 0, block, " "); \
2658 } while (0)
2659 #define ext4_error(sb, fmt, ...) \
2660 do { \
2661 no_printk(fmt, ##__VA_ARGS__); \
2662 __ext4_error(sb, "", 0, " "); \
2663 } while (0)
2664 #define ext4_abort(sb, fmt, ...) \
2665 do { \
2666 no_printk(fmt, ##__VA_ARGS__); \
2667 __ext4_abort(sb, "", 0, " "); \
2668 } while (0)
2669 #define ext4_warning(sb, fmt, ...) \
2670 do { \
2671 no_printk(fmt, ##__VA_ARGS__); \
2672 __ext4_warning(sb, "", 0, " "); \
2673 } while (0)
2674 #define ext4_warning_inode(inode, fmt, ...) \
2675 do { \
2676 no_printk(fmt, ##__VA_ARGS__); \
2677 __ext4_warning_inode(inode, "", 0, " "); \
2678 } while (0)
2679 #define ext4_msg(sb, level, fmt, ...) \
2680 do { \
2681 no_printk(fmt, ##__VA_ARGS__); \
2682 __ext4_msg(sb, "", " "); \
2683 } while (0)
2684 #define dump_mmp_msg(sb, mmp, msg) \
2685 __dump_mmp_msg(sb, mmp, "", 0, "")
2686 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2687 do { \
2688 no_printk(fmt, ##__VA_ARGS__); \
2689 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
2690 } while (0)
2691
2692 #endif
2693
2694 extern void ext4_update_dynamic_rev(struct super_block *sb);
2695 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2696 __u32 compat);
2697 extern int ext4_update_rocompat_feature(handle_t *handle,
2698 struct super_block *sb, __u32 rocompat);
2699 extern int ext4_update_incompat_feature(handle_t *handle,
2700 struct super_block *sb, __u32 incompat);
2701 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2702 struct ext4_group_desc *bg);
2703 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2704 struct ext4_group_desc *bg);
2705 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2706 struct ext4_group_desc *bg);
2707 extern __u32 ext4_free_group_clusters(struct super_block *sb,
2708 struct ext4_group_desc *bg);
2709 extern __u32 ext4_free_inodes_count(struct super_block *sb,
2710 struct ext4_group_desc *bg);
2711 extern __u32 ext4_used_dirs_count(struct super_block *sb,
2712 struct ext4_group_desc *bg);
2713 extern __u32 ext4_itable_unused_count(struct super_block *sb,
2714 struct ext4_group_desc *bg);
2715 extern void ext4_block_bitmap_set(struct super_block *sb,
2716 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2717 extern void ext4_inode_bitmap_set(struct super_block *sb,
2718 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2719 extern void ext4_inode_table_set(struct super_block *sb,
2720 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2721 extern void ext4_free_group_clusters_set(struct super_block *sb,
2722 struct ext4_group_desc *bg,
2723 __u32 count);
2724 extern void ext4_free_inodes_set(struct super_block *sb,
2725 struct ext4_group_desc *bg, __u32 count);
2726 extern void ext4_used_dirs_set(struct super_block *sb,
2727 struct ext4_group_desc *bg, __u32 count);
2728 extern void ext4_itable_unused_set(struct super_block *sb,
2729 struct ext4_group_desc *bg, __u32 count);
2730 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2731 struct ext4_group_desc *gdp);
2732 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2733 struct ext4_group_desc *gdp);
2734 extern int ext4_register_li_request(struct super_block *sb,
2735 ext4_group_t first_not_zeroed);
2736
ext4_has_group_desc_csum(struct super_block * sb)2737 static inline int ext4_has_group_desc_csum(struct super_block *sb)
2738 {
2739 return ext4_has_feature_gdt_csum(sb) ||
2740 EXT4_SB(sb)->s_chksum_driver != NULL;
2741 }
2742
ext4_has_metadata_csum(struct super_block * sb)2743 static inline int ext4_has_metadata_csum(struct super_block *sb)
2744 {
2745 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
2746 !EXT4_SB(sb)->s_chksum_driver);
2747
2748 return (EXT4_SB(sb)->s_chksum_driver != NULL);
2749 }
ext4_blocks_count(struct ext4_super_block * es)2750 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2751 {
2752 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2753 le32_to_cpu(es->s_blocks_count_lo);
2754 }
2755
ext4_r_blocks_count(struct ext4_super_block * es)2756 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2757 {
2758 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2759 le32_to_cpu(es->s_r_blocks_count_lo);
2760 }
2761
ext4_free_blocks_count(struct ext4_super_block * es)2762 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2763 {
2764 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2765 le32_to_cpu(es->s_free_blocks_count_lo);
2766 }
2767
ext4_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)2768 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2769 ext4_fsblk_t blk)
2770 {
2771 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2772 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2773 }
2774
ext4_free_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)2775 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2776 ext4_fsblk_t blk)
2777 {
2778 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2779 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2780 }
2781
ext4_r_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)2782 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2783 ext4_fsblk_t blk)
2784 {
2785 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2786 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2787 }
2788
ext4_isize(struct ext4_inode * raw_inode)2789 static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
2790 {
2791 if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2792 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2793 le32_to_cpu(raw_inode->i_size_lo);
2794 else
2795 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2796 }
2797
ext4_isize_set(struct ext4_inode * raw_inode,loff_t i_size)2798 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2799 {
2800 raw_inode->i_size_lo = cpu_to_le32(i_size);
2801 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2802 }
2803
2804 static inline
ext4_get_group_info(struct super_block * sb,ext4_group_t group)2805 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2806 ext4_group_t group)
2807 {
2808 struct ext4_group_info ***grp_info;
2809 long indexv, indexh;
2810 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
2811 grp_info = EXT4_SB(sb)->s_group_info;
2812 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2813 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2814 return grp_info[indexv][indexh];
2815 }
2816
2817 /*
2818 * Reading s_groups_count requires using smp_rmb() afterwards. See
2819 * the locking protocol documented in the comments of ext4_group_add()
2820 * in resize.c
2821 */
ext4_get_groups_count(struct super_block * sb)2822 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2823 {
2824 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
2825
2826 smp_rmb();
2827 return ngroups;
2828 }
2829
ext4_flex_group(struct ext4_sb_info * sbi,ext4_group_t block_group)2830 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2831 ext4_group_t block_group)
2832 {
2833 return block_group >> sbi->s_log_groups_per_flex;
2834 }
2835
ext4_flex_bg_size(struct ext4_sb_info * sbi)2836 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2837 {
2838 return 1 << sbi->s_log_groups_per_flex;
2839 }
2840
2841 #define ext4_std_error(sb, errno) \
2842 do { \
2843 if ((errno)) \
2844 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
2845 } while (0)
2846
2847 #ifdef CONFIG_SMP
2848 /* Each CPU can accumulate percpu_counter_batch clusters in their local
2849 * counters. So we need to make sure we have free clusters more
2850 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
2851 */
2852 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2853 #else
2854 #define EXT4_FREECLUSTERS_WATERMARK 0
2855 #endif
2856
2857 /* Update i_disksize. Requires i_mutex to avoid races with truncate */
ext4_update_i_disksize(struct inode * inode,loff_t newsize)2858 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2859 {
2860 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
2861 !mutex_is_locked(&inode->i_mutex));
2862 down_write(&EXT4_I(inode)->i_data_sem);
2863 if (newsize > EXT4_I(inode)->i_disksize)
2864 EXT4_I(inode)->i_disksize = newsize;
2865 up_write(&EXT4_I(inode)->i_data_sem);
2866 }
2867
2868 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
ext4_update_inode_size(struct inode * inode,loff_t newsize)2869 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
2870 {
2871 int changed = 0;
2872
2873 if (newsize > inode->i_size) {
2874 i_size_write(inode, newsize);
2875 changed = 1;
2876 }
2877 if (newsize > EXT4_I(inode)->i_disksize) {
2878 ext4_update_i_disksize(inode, newsize);
2879 changed |= 2;
2880 }
2881 return changed;
2882 }
2883
2884 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
2885 loff_t len);
2886
2887 struct ext4_group_info {
2888 unsigned long bb_state;
2889 struct rb_root bb_free_root;
2890 ext4_grpblk_t bb_first_free; /* first free block */
2891 ext4_grpblk_t bb_free; /* total free blocks */
2892 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
2893 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
2894 struct list_head bb_prealloc_list;
2895 #ifdef DOUBLE_CHECK
2896 void *bb_bitmap;
2897 #endif
2898 struct rw_semaphore alloc_sem;
2899 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
2900 * regions, index is order.
2901 * bb_counters[3] = 5 means
2902 * 5 free 8-block regions. */
2903 };
2904
2905 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
2906 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
2907 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
2908 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
2909
2910 #define EXT4_MB_GRP_NEED_INIT(grp) \
2911 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2912 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
2913 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2914 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
2915 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2916
2917 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
2918 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2919 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
2920 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2921 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
2922 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2923
2924 #define EXT4_MAX_CONTENTION 8
2925 #define EXT4_CONTENTION_THRESHOLD 2
2926
ext4_group_lock_ptr(struct super_block * sb,ext4_group_t group)2927 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2928 ext4_group_t group)
2929 {
2930 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2931 }
2932
2933 /*
2934 * Returns true if the filesystem is busy enough that attempts to
2935 * access the block group locks has run into contention.
2936 */
ext4_fs_is_busy(struct ext4_sb_info * sbi)2937 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2938 {
2939 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2940 }
2941
ext4_lock_group(struct super_block * sb,ext4_group_t group)2942 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2943 {
2944 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2945 if (spin_trylock(lock))
2946 /*
2947 * We're able to grab the lock right away, so drop the
2948 * lock contention counter.
2949 */
2950 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2951 else {
2952 /*
2953 * The lock is busy, so bump the contention counter,
2954 * and then wait on the spin lock.
2955 */
2956 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2957 EXT4_MAX_CONTENTION);
2958 spin_lock(lock);
2959 }
2960 }
2961
ext4_unlock_group(struct super_block * sb,ext4_group_t group)2962 static inline void ext4_unlock_group(struct super_block *sb,
2963 ext4_group_t group)
2964 {
2965 spin_unlock(ext4_group_lock_ptr(sb, group));
2966 }
2967
2968 /*
2969 * Block validity checking
2970 */
2971 #define ext4_check_indirect_blockref(inode, bh) \
2972 ext4_check_blockref(__func__, __LINE__, inode, \
2973 (__le32 *)(bh)->b_data, \
2974 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2975
2976 #define ext4_ind_check_inode(inode) \
2977 ext4_check_blockref(__func__, __LINE__, inode, \
2978 EXT4_I(inode)->i_data, \
2979 EXT4_NDIR_BLOCKS)
2980
2981 /*
2982 * Inodes and files operations
2983 */
2984
2985 /* dir.c */
2986 extern const struct file_operations ext4_dir_operations;
2987
2988 /* file.c */
2989 extern const struct inode_operations ext4_file_inode_operations;
2990 extern const struct file_operations ext4_file_operations;
2991 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
2992
2993 /* inline.c */
2994 extern int ext4_get_max_inline_size(struct inode *inode);
2995 extern int ext4_find_inline_data_nolock(struct inode *inode);
2996 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
2997 unsigned int len);
2998 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
2999
3000 extern int ext4_readpage_inline(struct inode *inode, struct page *page);
3001 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3002 struct inode *inode,
3003 loff_t pos, unsigned len,
3004 unsigned flags,
3005 struct page **pagep);
3006 extern int ext4_write_inline_data_end(struct inode *inode,
3007 loff_t pos, unsigned len,
3008 unsigned copied,
3009 struct page *page);
3010 extern struct buffer_head *
3011 ext4_journalled_write_inline_data(struct inode *inode,
3012 unsigned len,
3013 struct page *page);
3014 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
3015 struct inode *inode,
3016 loff_t pos, unsigned len,
3017 unsigned flags,
3018 struct page **pagep,
3019 void **fsdata);
3020 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
3021 unsigned len, unsigned copied,
3022 struct page *page);
3023 extern int ext4_try_add_inline_entry(handle_t *handle,
3024 struct ext4_filename *fname,
3025 struct dentry *dentry,
3026 struct inode *inode);
3027 extern int ext4_try_create_inline_dir(handle_t *handle,
3028 struct inode *parent,
3029 struct inode *inode);
3030 extern int ext4_read_inline_dir(struct file *filp,
3031 struct dir_context *ctx,
3032 int *has_inline_data);
3033 extern int htree_inlinedir_to_tree(struct file *dir_file,
3034 struct inode *dir, ext4_lblk_t block,
3035 struct dx_hash_info *hinfo,
3036 __u32 start_hash, __u32 start_minor_hash,
3037 int *has_inline_data);
3038 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3039 struct ext4_filename *fname,
3040 const struct qstr *d_name,
3041 struct ext4_dir_entry_2 **res_dir,
3042 int *has_inline_data);
3043 extern int ext4_delete_inline_entry(handle_t *handle,
3044 struct inode *dir,
3045 struct ext4_dir_entry_2 *de_del,
3046 struct buffer_head *bh,
3047 int *has_inline_data);
3048 extern int empty_inline_dir(struct inode *dir, int *has_inline_data);
3049 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3050 struct ext4_dir_entry_2 **parent_de,
3051 int *retval);
3052 extern int ext4_inline_data_fiemap(struct inode *inode,
3053 struct fiemap_extent_info *fieinfo,
3054 int *has_inline, __u64 start, __u64 len);
3055 extern int ext4_try_to_evict_inline_data(handle_t *handle,
3056 struct inode *inode,
3057 int needed);
3058 extern void ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3059
3060 extern int ext4_convert_inline_data(struct inode *inode);
3061
ext4_has_inline_data(struct inode * inode)3062 static inline int ext4_has_inline_data(struct inode *inode)
3063 {
3064 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3065 EXT4_I(inode)->i_inline_off;
3066 }
3067
3068 /* namei.c */
3069 extern const struct inode_operations ext4_dir_inode_operations;
3070 extern const struct inode_operations ext4_special_inode_operations;
3071 extern struct dentry *ext4_get_parent(struct dentry *child);
3072 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3073 struct ext4_dir_entry_2 *de,
3074 int blocksize, int csum_size,
3075 unsigned int parent_ino, int dotdot_real_len);
3076 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t,
3077 unsigned int blocksize);
3078 extern int ext4_handle_dirty_dirent_node(handle_t *handle,
3079 struct inode *inode,
3080 struct buffer_head *bh);
3081 #define S_SHIFT 12
3082 static unsigned char ext4_type_by_mode[S_IFMT >> S_SHIFT] = {
3083 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
3084 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
3085 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
3086 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
3087 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
3088 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
3089 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
3090 };
3091
ext4_set_de_type(struct super_block * sb,struct ext4_dir_entry_2 * de,umode_t mode)3092 static inline void ext4_set_de_type(struct super_block *sb,
3093 struct ext4_dir_entry_2 *de,
3094 umode_t mode) {
3095 if (ext4_has_feature_filetype(sb))
3096 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3097 }
3098
3099 /* readpages.c */
3100 extern int ext4_mpage_readpages(struct address_space *mapping,
3101 struct list_head *pages, struct page *page,
3102 unsigned nr_pages);
3103
3104 /* symlink.c */
3105 extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3106 extern const struct inode_operations ext4_symlink_inode_operations;
3107 extern const struct inode_operations ext4_fast_symlink_inode_operations;
3108
3109 /* sysfs.c */
3110 extern int ext4_register_sysfs(struct super_block *sb);
3111 extern void ext4_unregister_sysfs(struct super_block *sb);
3112 extern int __init ext4_init_sysfs(void);
3113 extern void ext4_exit_sysfs(void);
3114
3115 /* block_validity */
3116 extern void ext4_release_system_zone(struct super_block *sb);
3117 extern int ext4_setup_system_zone(struct super_block *sb);
3118 extern int __init ext4_init_system_zone(void);
3119 extern void ext4_exit_system_zone(void);
3120 extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
3121 ext4_fsblk_t start_blk,
3122 unsigned int count);
3123 extern int ext4_check_blockref(const char *, unsigned int,
3124 struct inode *, __le32 *, unsigned int);
3125
3126 /* extents.c */
3127 struct ext4_ext_path;
3128 struct ext4_extent;
3129
3130 /*
3131 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3132 * __le32.
3133 */
3134 #define EXT_MAX_BLOCKS 0xffffffff
3135
3136 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
3137 extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
3138 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3139 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3140 struct ext4_map_blocks *map, int flags);
3141 extern void ext4_ext_truncate(handle_t *, struct inode *);
3142 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3143 ext4_lblk_t end);
3144 extern void ext4_ext_init(struct super_block *);
3145 extern void ext4_ext_release(struct super_block *);
3146 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3147 loff_t len);
3148 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3149 loff_t offset, ssize_t len);
3150 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3151 struct ext4_map_blocks *map, int flags);
3152 extern int ext4_ext_calc_metadata_amount(struct inode *inode,
3153 ext4_lblk_t lblocks);
3154 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3155 int num,
3156 struct ext4_ext_path *path);
3157 extern int ext4_can_extents_be_merged(struct inode *inode,
3158 struct ext4_extent *ex1,
3159 struct ext4_extent *ex2);
3160 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3161 struct ext4_ext_path **,
3162 struct ext4_extent *, int);
3163 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3164 struct ext4_ext_path **,
3165 int flags);
3166 extern void ext4_ext_drop_refs(struct ext4_ext_path *);
3167 extern int ext4_ext_check_inode(struct inode *inode);
3168 extern int ext4_find_delalloc_range(struct inode *inode,
3169 ext4_lblk_t lblk_start,
3170 ext4_lblk_t lblk_end);
3171 extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk);
3172 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3173 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3174 __u64 start, __u64 len);
3175 extern int ext4_ext_precache(struct inode *inode);
3176 extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len);
3177 extern int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len);
3178 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3179 struct inode *inode2, ext4_lblk_t lblk1,
3180 ext4_lblk_t lblk2, ext4_lblk_t count,
3181 int mark_unwritten,int *err);
3182
3183 /* move_extent.c */
3184 extern void ext4_double_down_write_data_sem(struct inode *first,
3185 struct inode *second);
3186 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3187 struct inode *donor_inode);
3188 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3189 __u64 start_orig, __u64 start_donor,
3190 __u64 len, __u64 *moved_len);
3191
3192 /* page-io.c */
3193 extern int __init ext4_init_pageio(void);
3194 extern void ext4_exit_pageio(void);
3195 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3196 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3197 extern int ext4_put_io_end(ext4_io_end_t *io_end);
3198 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3199 extern void ext4_io_submit_init(struct ext4_io_submit *io,
3200 struct writeback_control *wbc);
3201 extern void ext4_end_io_rsv_work(struct work_struct *work);
3202 extern void ext4_io_submit(struct ext4_io_submit *io);
3203 extern int ext4_bio_write_page(struct ext4_io_submit *io,
3204 struct page *page,
3205 int len,
3206 struct writeback_control *wbc,
3207 bool keep_towrite);
3208
3209 /* mmp.c */
3210 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3211
3212 /*
3213 * Add new method to test whether block and inode bitmaps are properly
3214 * initialized. With uninit_bg reading the block from disk is not enough
3215 * to mark the bitmap uptodate. We need to also zero-out the bitmap
3216 */
3217 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3218
bitmap_uptodate(struct buffer_head * bh)3219 static inline int bitmap_uptodate(struct buffer_head *bh)
3220 {
3221 return (buffer_uptodate(bh) &&
3222 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3223 }
set_bitmap_uptodate(struct buffer_head * bh)3224 static inline void set_bitmap_uptodate(struct buffer_head *bh)
3225 {
3226 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3227 }
3228
3229 /*
3230 * Disable DIO read nolock optimization, so new dioreaders will be forced
3231 * to grab i_mutex
3232 */
ext4_inode_block_unlocked_dio(struct inode * inode)3233 static inline void ext4_inode_block_unlocked_dio(struct inode *inode)
3234 {
3235 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
3236 smp_mb();
3237 }
ext4_inode_resume_unlocked_dio(struct inode * inode)3238 static inline void ext4_inode_resume_unlocked_dio(struct inode *inode)
3239 {
3240 smp_mb();
3241 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
3242 }
3243
3244 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
3245
3246 /* For ioend & aio unwritten conversion wait queues */
3247 #define EXT4_WQ_HASH_SZ 37
3248 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
3249 EXT4_WQ_HASH_SZ])
3250 #define ext4_aio_mutex(v) (&ext4__aio_mutex[((unsigned long)(v)) %\
3251 EXT4_WQ_HASH_SZ])
3252 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3253 extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ];
3254
3255 #define EXT4_RESIZING 0
3256 extern int ext4_resize_begin(struct super_block *sb);
3257 extern void ext4_resize_end(struct super_block *sb);
3258
3259 #endif /* __KERNEL__ */
3260
3261 #define EFSBADCRC EBADMSG /* Bad CRC detected */
3262 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
3263
3264 #endif /* _EXT4_H */
3265