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 #define EXT4_MAX_CLUSTER_LOG_SIZE 30
225 #ifdef __KERNEL__
226 # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
227 #else
228 # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
229 #endif
230 #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
231 #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
232 EXT4_SB(s)->s_cluster_bits)
233 #ifdef __KERNEL__
234 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
235 # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
236 #else
237 # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
238 #endif
239 #ifdef __KERNEL__
240 #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
241 #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
242 #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
243 #else
244 #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
245 EXT4_GOOD_OLD_INODE_SIZE : \
246 (s)->s_inode_size)
247 #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
248 EXT4_GOOD_OLD_FIRST_INO : \
249 (s)->s_first_ino)
250 #endif
251 #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
252
253 /* Translate a block number to a cluster number */
254 #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
255 /* Translate a cluster number to a block number */
256 #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
257 /* Translate # of blks to # of clusters */
258 #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
259 (sbi)->s_cluster_bits)
260 /* Mask out the low bits to get the starting block of the cluster */
261 #define EXT4_PBLK_CMASK(s, pblk) ((pblk) & \
262 ~((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
263 #define EXT4_LBLK_CMASK(s, lblk) ((lblk) & \
264 ~((ext4_lblk_t) (s)->s_cluster_ratio - 1))
265 /* Get the cluster offset */
266 #define EXT4_PBLK_COFF(s, pblk) ((pblk) & \
267 ((ext4_fsblk_t) (s)->s_cluster_ratio - 1))
268 #define EXT4_LBLK_COFF(s, lblk) ((lblk) & \
269 ((ext4_lblk_t) (s)->s_cluster_ratio - 1))
270
271 /*
272 * Structure of a blocks group descriptor
273 */
274 struct ext4_group_desc
275 {
276 __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
277 __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
278 __le32 bg_inode_table_lo; /* Inodes table block */
279 __le16 bg_free_blocks_count_lo;/* Free blocks count */
280 __le16 bg_free_inodes_count_lo;/* Free inodes count */
281 __le16 bg_used_dirs_count_lo; /* Directories count */
282 __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
283 __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
284 __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
285 __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
286 __le16 bg_itable_unused_lo; /* Unused inodes count */
287 __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
288 __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
289 __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
290 __le32 bg_inode_table_hi; /* Inodes table block MSB */
291 __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
292 __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
293 __le16 bg_used_dirs_count_hi; /* Directories count MSB */
294 __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
295 __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
296 __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
297 __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
298 __u32 bg_reserved;
299 };
300
301 #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
302 (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
303 sizeof(__le16))
304 #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
305 (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
306 sizeof(__le16))
307
308 /*
309 * Structure of a flex block group info
310 */
311
312 struct flex_groups {
313 atomic64_t free_clusters;
314 atomic_t free_inodes;
315 atomic_t used_dirs;
316 };
317
318 #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
319 #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
320 #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
321
322 /*
323 * Macro-instructions used to manage group descriptors
324 */
325 #define EXT4_MIN_DESC_SIZE 32
326 #define EXT4_MIN_DESC_SIZE_64BIT 64
327 #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
328 #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
329 #ifdef __KERNEL__
330 # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
331 # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
332 # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
333 # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
334 # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
335 #else
336 # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
337 # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
338 # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
339 #endif
340
341 /*
342 * Constants relative to the data blocks
343 */
344 #define EXT4_NDIR_BLOCKS 12
345 #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
346 #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
347 #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
348 #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
349
350 /*
351 * Inode flags
352 */
353 #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
354 #define EXT4_UNRM_FL 0x00000002 /* Undelete */
355 #define EXT4_COMPR_FL 0x00000004 /* Compress file */
356 #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
357 #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
358 #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
359 #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
360 #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
361 /* Reserved for compression usage... */
362 #define EXT4_DIRTY_FL 0x00000100
363 #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
364 #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
365 /* nb: was previously EXT2_ECOMPR_FL */
366 #define EXT4_ENCRYPT_FL 0x00000800 /* encrypted file */
367 /* End compression flags --- maybe not all used */
368 #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
369 #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
370 #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
371 #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
372 #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
373 #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
374 #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
375 #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
376 #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
377 #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
378 #define EXT4_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */
379 #define EXT4_PROJINHERIT_FL 0x20000000 /* Create with parents projid */
380 #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
381
382 #define EXT4_FL_USER_VISIBLE 0x004BDFFF /* User visible flags */
383 #define EXT4_FL_USER_MODIFIABLE 0x004380FF /* User modifiable flags */
384
385 /* Flags that should be inherited by new inodes from their parent. */
386 #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
387 EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
388 EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
389 EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
390
391 /* Flags that are appropriate for regular files (all but dir-specific ones). */
392 #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
393
394 /* Flags that are appropriate for non-directories/regular files. */
395 #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
396
397 /* Mask out flags that are inappropriate for the given type of inode. */
ext4_mask_flags(umode_t mode,__u32 flags)398 static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
399 {
400 if (S_ISDIR(mode))
401 return flags;
402 else if (S_ISREG(mode))
403 return flags & EXT4_REG_FLMASK;
404 else
405 return flags & EXT4_OTHER_FLMASK;
406 }
407
408 /*
409 * Inode flags used for atomic set/get
410 */
411 enum {
412 EXT4_INODE_SECRM = 0, /* Secure deletion */
413 EXT4_INODE_UNRM = 1, /* Undelete */
414 EXT4_INODE_COMPR = 2, /* Compress file */
415 EXT4_INODE_SYNC = 3, /* Synchronous updates */
416 EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
417 EXT4_INODE_APPEND = 5, /* writes to file may only append */
418 EXT4_INODE_NODUMP = 6, /* do not dump file */
419 EXT4_INODE_NOATIME = 7, /* do not update atime */
420 /* Reserved for compression usage... */
421 EXT4_INODE_DIRTY = 8,
422 EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
423 EXT4_INODE_NOCOMPR = 10, /* Don't compress */
424 EXT4_INODE_ENCRYPT = 11, /* Encrypted file */
425 /* End compression flags --- maybe not all used */
426 EXT4_INODE_INDEX = 12, /* hash-indexed directory */
427 EXT4_INODE_IMAGIC = 13, /* AFS directory */
428 EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
429 EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
430 EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
431 EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
432 EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
433 EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
434 EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
435 EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
436 EXT4_INODE_INLINE_DATA = 28, /* Data in inode. */
437 EXT4_INODE_PROJINHERIT = 29, /* Create with parents projid */
438 EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
439 };
440
441 /*
442 * Since it's pretty easy to mix up bit numbers and hex values, we use a
443 * build-time check to make sure that EXT4_XXX_FL is consistent with respect to
444 * EXT4_INODE_XXX. If all is well, the macros will be dropped, so, it won't cost
445 * any extra space in the compiled kernel image, otherwise, the build will fail.
446 * It's important that these values are the same, since we are using
447 * EXT4_INODE_XXX to test for flag values, but EXT4_XXX_FL must be consistent
448 * with the values of FS_XXX_FL defined in include/linux/fs.h and the on-disk
449 * values found in ext2, ext3 and ext4 filesystems, and of course the values
450 * defined in e2fsprogs.
451 *
452 * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
453 */
454 #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
455 #define CHECK_FLAG_VALUE(FLAG) BUILD_BUG_ON(!TEST_FLAG_VALUE(FLAG))
456
ext4_check_flag_values(void)457 static inline void ext4_check_flag_values(void)
458 {
459 CHECK_FLAG_VALUE(SECRM);
460 CHECK_FLAG_VALUE(UNRM);
461 CHECK_FLAG_VALUE(COMPR);
462 CHECK_FLAG_VALUE(SYNC);
463 CHECK_FLAG_VALUE(IMMUTABLE);
464 CHECK_FLAG_VALUE(APPEND);
465 CHECK_FLAG_VALUE(NODUMP);
466 CHECK_FLAG_VALUE(NOATIME);
467 CHECK_FLAG_VALUE(DIRTY);
468 CHECK_FLAG_VALUE(COMPRBLK);
469 CHECK_FLAG_VALUE(NOCOMPR);
470 CHECK_FLAG_VALUE(ENCRYPT);
471 CHECK_FLAG_VALUE(INDEX);
472 CHECK_FLAG_VALUE(IMAGIC);
473 CHECK_FLAG_VALUE(JOURNAL_DATA);
474 CHECK_FLAG_VALUE(NOTAIL);
475 CHECK_FLAG_VALUE(DIRSYNC);
476 CHECK_FLAG_VALUE(TOPDIR);
477 CHECK_FLAG_VALUE(HUGE_FILE);
478 CHECK_FLAG_VALUE(EXTENTS);
479 CHECK_FLAG_VALUE(EA_INODE);
480 CHECK_FLAG_VALUE(EOFBLOCKS);
481 CHECK_FLAG_VALUE(INLINE_DATA);
482 CHECK_FLAG_VALUE(PROJINHERIT);
483 CHECK_FLAG_VALUE(RESERVED);
484 }
485
486 /* Used to pass group descriptor data when online resize is done */
487 struct ext4_new_group_input {
488 __u32 group; /* Group number for this data */
489 __u64 block_bitmap; /* Absolute block number of block bitmap */
490 __u64 inode_bitmap; /* Absolute block number of inode bitmap */
491 __u64 inode_table; /* Absolute block number of inode table start */
492 __u32 blocks_count; /* Total number of blocks in this group */
493 __u16 reserved_blocks; /* Number of reserved blocks in this group */
494 __u16 unused;
495 };
496
497 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
498 struct compat_ext4_new_group_input {
499 u32 group;
500 compat_u64 block_bitmap;
501 compat_u64 inode_bitmap;
502 compat_u64 inode_table;
503 u32 blocks_count;
504 u16 reserved_blocks;
505 u16 unused;
506 };
507 #endif
508
509 /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
510 struct ext4_new_group_data {
511 __u32 group;
512 __u64 block_bitmap;
513 __u64 inode_bitmap;
514 __u64 inode_table;
515 __u32 blocks_count;
516 __u16 reserved_blocks;
517 __u16 unused;
518 __u32 free_blocks_count;
519 };
520
521 /* Indexes used to index group tables in ext4_new_group_data */
522 enum {
523 BLOCK_BITMAP = 0, /* block bitmap */
524 INODE_BITMAP, /* inode bitmap */
525 INODE_TABLE, /* inode tables */
526 GROUP_TABLE_COUNT,
527 };
528
529 /*
530 * Flags used by ext4_map_blocks()
531 */
532 /* Allocate any needed blocks and/or convert an unwritten
533 extent to be an initialized ext4 */
534 #define EXT4_GET_BLOCKS_CREATE 0x0001
535 /* Request the creation of an unwritten extent */
536 #define EXT4_GET_BLOCKS_UNWRIT_EXT 0x0002
537 #define EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT (EXT4_GET_BLOCKS_UNWRIT_EXT|\
538 EXT4_GET_BLOCKS_CREATE)
539 /* Caller is from the delayed allocation writeout path
540 * finally doing the actual allocation of delayed blocks */
541 #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
542 /* caller is from the direct IO path, request to creation of an
543 unwritten extents if not allocated, split the unwritten
544 extent if blocks has been preallocated already*/
545 #define EXT4_GET_BLOCKS_PRE_IO 0x0008
546 #define EXT4_GET_BLOCKS_CONVERT 0x0010
547 #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
548 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
549 /* Convert extent to initialized after IO complete */
550 #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
551 EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT)
552 /* Eventual metadata allocation (due to growing extent tree)
553 * should not fail, so try to use reserved blocks for that.*/
554 #define EXT4_GET_BLOCKS_METADATA_NOFAIL 0x0020
555 /* Don't normalize allocation size (used for fallocate) */
556 #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
557 /* Request will not result in inode size update (user for fallocate) */
558 #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080
559 /* Do not take i_data_sem locking in ext4_map_blocks */
560 #define EXT4_GET_BLOCKS_NO_LOCK 0x0100
561 /* Convert written extents to unwritten */
562 #define EXT4_GET_BLOCKS_CONVERT_UNWRITTEN 0x0200
563
564 /*
565 * The bit position of these flags must not overlap with any of the
566 * EXT4_GET_BLOCKS_*. They are used by ext4_find_extent(),
567 * read_extent_tree_block(), ext4_split_extent_at(),
568 * ext4_ext_insert_extent(), and ext4_ext_create_new_leaf().
569 * EXT4_EX_NOCACHE is used to indicate that the we shouldn't be
570 * caching the extents when reading from the extent tree while a
571 * truncate or punch hole operation is in progress.
572 */
573 #define EXT4_EX_NOCACHE 0x40000000
574 #define EXT4_EX_FORCE_CACHE 0x20000000
575
576 /*
577 * Flags used by ext4_free_blocks
578 */
579 #define EXT4_FREE_BLOCKS_METADATA 0x0001
580 #define EXT4_FREE_BLOCKS_FORGET 0x0002
581 #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
582 #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
583 #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
584 #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
585
586 /* Encryption algorithms */
587 #define EXT4_ENCRYPTION_MODE_INVALID 0
588 #define EXT4_ENCRYPTION_MODE_AES_256_XTS 1
589 #define EXT4_ENCRYPTION_MODE_AES_256_GCM 2
590 #define EXT4_ENCRYPTION_MODE_AES_256_CBC 3
591 #define EXT4_ENCRYPTION_MODE_AES_256_CTS 4
592 #define EXT4_ENCRYPTION_MODE_SPECK128_256_XTS 7
593 #define EXT4_ENCRYPTION_MODE_SPECK128_256_CTS 8
594 #define EXT4_ENCRYPTION_MODE_AES_256_HEH 126
595
596 #include "ext4_crypto.h"
597
598 /*
599 * ioctl commands
600 */
601 #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
602 #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
603 #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
604 #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
605 #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
606 #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
607 #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
608 #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
609 #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
610 #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
611 #define EXT4_IOC_MIGRATE _IO('f', 9)
612 /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
613 /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
614 #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
615 #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
616 #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
617 #define EXT4_IOC_SWAP_BOOT _IO('f', 17)
618 #define EXT4_IOC_PRECACHE_EXTENTS _IO('f', 18)
619 #define EXT4_IOC_SET_ENCRYPTION_POLICY _IOR('f', 19, struct ext4_encryption_policy)
620 #define EXT4_IOC_GET_ENCRYPTION_PWSALT _IOW('f', 20, __u8[16])
621 #define EXT4_IOC_GET_ENCRYPTION_POLICY _IOW('f', 21, struct ext4_encryption_policy)
622
623 #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
624 /*
625 * ioctl commands in 32 bit emulation
626 */
627 #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
628 #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
629 #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
630 #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
631 #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
632 #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
633 #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
634 #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
635 #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
636 #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
637 #endif
638
639 /* Max physical block we can address w/o extents */
640 #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
641
642 /*
643 * Structure of an inode on the disk
644 */
645 struct ext4_inode {
646 __le16 i_mode; /* File mode */
647 __le16 i_uid; /* Low 16 bits of Owner Uid */
648 __le32 i_size_lo; /* Size in bytes */
649 __le32 i_atime; /* Access time */
650 __le32 i_ctime; /* Inode Change time */
651 __le32 i_mtime; /* Modification time */
652 __le32 i_dtime; /* Deletion Time */
653 __le16 i_gid; /* Low 16 bits of Group Id */
654 __le16 i_links_count; /* Links count */
655 __le32 i_blocks_lo; /* Blocks count */
656 __le32 i_flags; /* File flags */
657 union {
658 struct {
659 __le32 l_i_version;
660 } linux1;
661 struct {
662 __u32 h_i_translator;
663 } hurd1;
664 struct {
665 __u32 m_i_reserved1;
666 } masix1;
667 } osd1; /* OS dependent 1 */
668 __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
669 __le32 i_generation; /* File version (for NFS) */
670 __le32 i_file_acl_lo; /* File ACL */
671 __le32 i_size_high;
672 __le32 i_obso_faddr; /* Obsoleted fragment address */
673 union {
674 struct {
675 __le16 l_i_blocks_high; /* were l_i_reserved1 */
676 __le16 l_i_file_acl_high;
677 __le16 l_i_uid_high; /* these 2 fields */
678 __le16 l_i_gid_high; /* were reserved2[0] */
679 __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
680 __le16 l_i_reserved;
681 } linux2;
682 struct {
683 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
684 __u16 h_i_mode_high;
685 __u16 h_i_uid_high;
686 __u16 h_i_gid_high;
687 __u32 h_i_author;
688 } hurd2;
689 struct {
690 __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
691 __le16 m_i_file_acl_high;
692 __u32 m_i_reserved2[2];
693 } masix2;
694 } osd2; /* OS dependent 2 */
695 __le16 i_extra_isize;
696 __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
697 __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
698 __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
699 __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
700 __le32 i_crtime; /* File Creation time */
701 __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
702 __le32 i_version_hi; /* high 32 bits for 64-bit version */
703 __le32 i_projid; /* Project ID */
704 };
705
706 struct move_extent {
707 __u32 reserved; /* should be zero */
708 __u32 donor_fd; /* donor file descriptor */
709 __u64 orig_start; /* logical start offset in block for orig */
710 __u64 donor_start; /* logical start offset in block for donor */
711 __u64 len; /* block length to be moved */
712 __u64 moved_len; /* moved block length */
713 };
714
715 #define EXT4_EPOCH_BITS 2
716 #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
717 #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
718
719 /*
720 * Extended fields will fit into an inode if the filesystem was formatted
721 * with large inodes (-I 256 or larger) and there are not currently any EAs
722 * consuming all of the available space. For new inodes we always reserve
723 * enough space for the kernel's known extended fields, but for inodes
724 * created with an old kernel this might not have been the case. None of
725 * the extended inode fields is critical for correct filesystem operation.
726 * This macro checks if a certain field fits in the inode. Note that
727 * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
728 */
729 #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
730 ((offsetof(typeof(*ext4_inode), field) + \
731 sizeof((ext4_inode)->field)) \
732 <= (EXT4_GOOD_OLD_INODE_SIZE + \
733 (einode)->i_extra_isize)) \
734
735 /*
736 * We use an encoding that preserves the times for extra epoch "00":
737 *
738 * extra msb of adjust for signed
739 * epoch 32-bit 32-bit tv_sec to
740 * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
741 * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
742 * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
743 * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
744 * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
745 * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
746 * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
747 * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
748 * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
749 *
750 * Note that previous versions of the kernel on 64-bit systems would
751 * incorrectly use extra epoch bits 1,1 for dates between 1901 and
752 * 1970. e2fsck will correct this, assuming that it is run on the
753 * affected filesystem before 2242.
754 */
755
ext4_encode_extra_time(struct timespec * time)756 static inline __le32 ext4_encode_extra_time(struct timespec *time)
757 {
758 u32 extra = sizeof(time->tv_sec) > 4 ?
759 ((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK : 0;
760 return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
761 }
762
ext4_decode_extra_time(struct timespec * time,__le32 extra)763 static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
764 {
765 if (unlikely(sizeof(time->tv_sec) > 4 &&
766 (extra & cpu_to_le32(EXT4_EPOCH_MASK)))) {
767 #if LINUX_VERSION_CODE < KERNEL_VERSION(4,20,0)
768 /* Handle legacy encoding of pre-1970 dates with epoch
769 * bits 1,1. We assume that by kernel version 4.20,
770 * everyone will have run fsck over the affected
771 * filesystems to correct the problem. (This
772 * backwards compatibility may be removed before this
773 * time, at the discretion of the ext4 developers.)
774 */
775 u64 extra_bits = le32_to_cpu(extra) & EXT4_EPOCH_MASK;
776 if (extra_bits == 3 && ((time->tv_sec) & 0x80000000) != 0)
777 extra_bits = 0;
778 time->tv_sec += extra_bits << 32;
779 #else
780 time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
781 #endif
782 }
783 time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
784 }
785
786 #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
787 do { \
788 (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
789 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
790 (raw_inode)->xtime ## _extra = \
791 ext4_encode_extra_time(&(inode)->xtime); \
792 } while (0)
793
794 #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
795 do { \
796 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
797 (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
798 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
799 (raw_inode)->xtime ## _extra = \
800 ext4_encode_extra_time(&(einode)->xtime); \
801 } while (0)
802
803 #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
804 do { \
805 (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
806 if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
807 ext4_decode_extra_time(&(inode)->xtime, \
808 raw_inode->xtime ## _extra); \
809 else \
810 (inode)->xtime.tv_nsec = 0; \
811 } while (0)
812
813 #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
814 do { \
815 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
816 (einode)->xtime.tv_sec = \
817 (signed)le32_to_cpu((raw_inode)->xtime); \
818 else \
819 (einode)->xtime.tv_sec = 0; \
820 if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
821 ext4_decode_extra_time(&(einode)->xtime, \
822 raw_inode->xtime ## _extra); \
823 else \
824 (einode)->xtime.tv_nsec = 0; \
825 } while (0)
826
827 #define i_disk_version osd1.linux1.l_i_version
828
829 #if defined(__KERNEL__) || defined(__linux__)
830 #define i_reserved1 osd1.linux1.l_i_reserved1
831 #define i_file_acl_high osd2.linux2.l_i_file_acl_high
832 #define i_blocks_high osd2.linux2.l_i_blocks_high
833 #define i_uid_low i_uid
834 #define i_gid_low i_gid
835 #define i_uid_high osd2.linux2.l_i_uid_high
836 #define i_gid_high osd2.linux2.l_i_gid_high
837 #define i_checksum_lo osd2.linux2.l_i_checksum_lo
838
839 #elif defined(__GNU__)
840
841 #define i_translator osd1.hurd1.h_i_translator
842 #define i_uid_high osd2.hurd2.h_i_uid_high
843 #define i_gid_high osd2.hurd2.h_i_gid_high
844 #define i_author osd2.hurd2.h_i_author
845
846 #elif defined(__masix__)
847
848 #define i_reserved1 osd1.masix1.m_i_reserved1
849 #define i_file_acl_high osd2.masix2.m_i_file_acl_high
850 #define i_reserved2 osd2.masix2.m_i_reserved2
851
852 #endif /* defined(__KERNEL__) || defined(__linux__) */
853
854 #include "extents_status.h"
855
856 /*
857 * Lock subclasses for i_data_sem in the ext4_inode_info structure.
858 *
859 * These are needed to avoid lockdep false positives when we need to
860 * allocate blocks to the quota inode during ext4_map_blocks(), while
861 * holding i_data_sem for a normal (non-quota) inode. Since we don't
862 * do quota tracking for the quota inode, this avoids deadlock (as
863 * well as infinite recursion, since it isn't turtles all the way
864 * down...)
865 *
866 * I_DATA_SEM_NORMAL - Used for most inodes
867 * I_DATA_SEM_OTHER - Used by move_inode.c for the second normal inode
868 * where the second inode has larger inode number
869 * than the first
870 * I_DATA_SEM_QUOTA - Used for quota inodes only
871 */
872 enum {
873 I_DATA_SEM_NORMAL = 0,
874 I_DATA_SEM_OTHER,
875 I_DATA_SEM_QUOTA,
876 };
877
878
879 /*
880 * fourth extended file system inode data in memory
881 */
882 struct ext4_inode_info {
883 __le32 i_data[15]; /* unconverted */
884 __u32 i_dtime;
885 ext4_fsblk_t i_file_acl;
886
887 /*
888 * i_block_group is the number of the block group which contains
889 * this file's inode. Constant across the lifetime of the inode,
890 * it is ued for making block allocation decisions - we try to
891 * place a file's data blocks near its inode block, and new inodes
892 * near to their parent directory's inode.
893 */
894 ext4_group_t i_block_group;
895 ext4_lblk_t i_dir_start_lookup;
896 #if (BITS_PER_LONG < 64)
897 unsigned long i_state_flags; /* Dynamic state flags */
898 #endif
899 unsigned long i_flags;
900
901 /*
902 * Extended attributes can be read independently of the main file
903 * data. Taking i_mutex even when reading would cause contention
904 * between readers of EAs and writers of regular file data, so
905 * instead we synchronize on xattr_sem when reading or changing
906 * EAs.
907 */
908 struct rw_semaphore xattr_sem;
909
910 struct list_head i_orphan; /* unlinked but open inodes */
911
912 /*
913 * i_disksize keeps track of what the inode size is ON DISK, not
914 * in memory. During truncate, i_size is set to the new size by
915 * the VFS prior to calling ext4_truncate(), but the filesystem won't
916 * set i_disksize to 0 until the truncate is actually under way.
917 *
918 * The intent is that i_disksize always represents the blocks which
919 * are used by this file. This allows recovery to restart truncate
920 * on orphans if we crash during truncate. We actually write i_disksize
921 * into the on-disk inode when writing inodes out, instead of i_size.
922 *
923 * The only time when i_disksize and i_size may be different is when
924 * a truncate is in progress. The only things which change i_disksize
925 * are ext4_get_block (growth) and ext4_truncate (shrinkth).
926 */
927 loff_t i_disksize;
928
929 /*
930 * i_data_sem is for serialising ext4_truncate() against
931 * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
932 * data tree are chopped off during truncate. We can't do that in
933 * ext4 because whenever we perform intermediate commits during
934 * truncate, the inode and all the metadata blocks *must* be in a
935 * consistent state which allows truncation of the orphans to restart
936 * during recovery. Hence we must fix the get_block-vs-truncate race
937 * by other means, so we have i_data_sem.
938 */
939 struct rw_semaphore i_data_sem;
940 /*
941 * i_mmap_sem is for serializing page faults with truncate / punch hole
942 * operations. We have to make sure that new page cannot be faulted in
943 * a section of the inode that is being punched. We cannot easily use
944 * i_data_sem for this since we need protection for the whole punch
945 * operation and i_data_sem ranks below transaction start so we have
946 * to occasionally drop it.
947 */
948 struct rw_semaphore i_mmap_sem;
949 struct inode vfs_inode;
950 struct jbd2_inode *jinode;
951
952 spinlock_t i_raw_lock; /* protects updates to the raw inode */
953
954 /*
955 * File creation time. Its function is same as that of
956 * struct timespec i_{a,c,m}time in the generic inode.
957 */
958 struct timespec i_crtime;
959
960 /* mballoc */
961 struct list_head i_prealloc_list;
962 spinlock_t i_prealloc_lock;
963
964 /* extents status tree */
965 struct ext4_es_tree i_es_tree;
966 rwlock_t i_es_lock;
967 struct list_head i_es_list;
968 unsigned int i_es_all_nr; /* protected by i_es_lock */
969 unsigned int i_es_shk_nr; /* protected by i_es_lock */
970 ext4_lblk_t i_es_shrink_lblk; /* Offset where we start searching for
971 extents to shrink. Protected by
972 i_es_lock */
973
974 /* ialloc */
975 ext4_group_t i_last_alloc_group;
976
977 /* allocation reservation info for delalloc */
978 /* In case of bigalloc, these refer to clusters rather than blocks */
979 unsigned int i_reserved_data_blocks;
980 unsigned int i_reserved_meta_blocks;
981 unsigned int i_allocated_meta_blocks;
982 ext4_lblk_t i_da_metadata_calc_last_lblock;
983 int i_da_metadata_calc_len;
984
985 /* on-disk additional length */
986 __u16 i_extra_isize;
987
988 /* Indicate the inline data space. */
989 u16 i_inline_off;
990 u16 i_inline_size;
991
992 #ifdef CONFIG_QUOTA
993 /* quota space reservation, managed internally by quota code */
994 qsize_t i_reserved_quota;
995 #endif
996
997 /* Lock protecting lists below */
998 spinlock_t i_completed_io_lock;
999 /*
1000 * Completed IOs that need unwritten extents handling and have
1001 * transaction reserved
1002 */
1003 struct list_head i_rsv_conversion_list;
1004 /*
1005 * Completed IOs that need unwritten extents handling and don't have
1006 * transaction reserved
1007 */
1008 atomic_t i_ioend_count; /* Number of outstanding io_end structs */
1009 atomic_t i_unwritten; /* Nr. of inflight conversions pending */
1010 struct work_struct i_rsv_conversion_work;
1011
1012 spinlock_t i_block_reservation_lock;
1013
1014 /*
1015 * Transactions that contain inode's metadata needed to complete
1016 * fsync and fdatasync, respectively.
1017 */
1018 tid_t i_sync_tid;
1019 tid_t i_datasync_tid;
1020
1021 #ifdef CONFIG_QUOTA
1022 struct dquot *i_dquot[MAXQUOTAS];
1023 #endif
1024
1025 /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
1026 __u32 i_csum_seed;
1027
1028 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1029 /* Encryption params */
1030 struct ext4_crypt_info *i_crypt_info;
1031 #endif
1032 };
1033
1034 /*
1035 * File system states
1036 */
1037 #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
1038 #define EXT4_ERROR_FS 0x0002 /* Errors detected */
1039 #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
1040
1041 /*
1042 * Misc. filesystem flags
1043 */
1044 #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
1045 #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
1046 #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
1047
1048 /*
1049 * Mount flags set via mount options or defaults
1050 */
1051 #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
1052 #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
1053 #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
1054 #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
1055 #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
1056 #define EXT4_MOUNT_ERRORS_MASK 0x00070
1057 #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
1058 #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
1059 #ifdef CONFIG_FS_DAX
1060 #define EXT4_MOUNT_DAX 0x00200 /* Direct Access */
1061 #else
1062 #define EXT4_MOUNT_DAX 0
1063 #endif
1064 #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
1065 #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
1066 #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
1067 #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
1068 #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
1069 #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
1070 #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
1071 #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
1072 #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
1073 #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
1074 #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
1075 #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
1076 #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
1077 #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
1078 #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
1079 #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
1080 #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
1081 #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
1082 #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
1083 #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
1084 #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
1085
1086 /*
1087 * Mount flags set either automatically (could not be set by mount option)
1088 * based on per file system feature or property or in special cases such as
1089 * distinguishing between explicit mount option definition and default.
1090 */
1091 #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
1092 specified delalloc */
1093 #define EXT4_MOUNT2_STD_GROUP_SIZE 0x00000002 /* We have standard group
1094 size of blocksize * 8
1095 blocks */
1096 #define EXT4_MOUNT2_HURD_COMPAT 0x00000004 /* Support HURD-castrated
1097 file systems */
1098
1099 #define EXT4_MOUNT2_EXPLICIT_JOURNAL_CHECKSUM 0x00000008 /* User explicitly
1100 specified journal checksum */
1101
1102 #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
1103 ~EXT4_MOUNT_##opt
1104 #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
1105 EXT4_MOUNT_##opt
1106 #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
1107 EXT4_MOUNT_##opt)
1108
1109 #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
1110 ~EXT4_MOUNT2_##opt
1111 #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
1112 EXT4_MOUNT2_##opt
1113 #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
1114 EXT4_MOUNT2_##opt)
1115
1116 #define ext4_test_and_set_bit __test_and_set_bit_le
1117 #define ext4_set_bit __set_bit_le
1118 #define ext4_set_bit_atomic ext2_set_bit_atomic
1119 #define ext4_test_and_clear_bit __test_and_clear_bit_le
1120 #define ext4_clear_bit __clear_bit_le
1121 #define ext4_clear_bit_atomic ext2_clear_bit_atomic
1122 #define ext4_test_bit test_bit_le
1123 #define ext4_find_next_zero_bit find_next_zero_bit_le
1124 #define ext4_find_next_bit find_next_bit_le
1125
1126 extern void ext4_set_bits(void *bm, int cur, int len);
1127
1128 /*
1129 * Maximal mount counts between two filesystem checks
1130 */
1131 #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
1132 #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
1133
1134 /*
1135 * Behaviour when detecting errors
1136 */
1137 #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
1138 #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
1139 #define EXT4_ERRORS_PANIC 3 /* Panic */
1140 #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
1141
1142 /* Metadata checksum algorithm codes */
1143 #define EXT4_CRC32C_CHKSUM 1
1144
1145 /*
1146 * Structure of the super block
1147 */
1148 struct ext4_super_block {
1149 /*00*/ __le32 s_inodes_count; /* Inodes count */
1150 __le32 s_blocks_count_lo; /* Blocks count */
1151 __le32 s_r_blocks_count_lo; /* Reserved blocks count */
1152 __le32 s_free_blocks_count_lo; /* Free blocks count */
1153 /*10*/ __le32 s_free_inodes_count; /* Free inodes count */
1154 __le32 s_first_data_block; /* First Data Block */
1155 __le32 s_log_block_size; /* Block size */
1156 __le32 s_log_cluster_size; /* Allocation cluster size */
1157 /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
1158 __le32 s_clusters_per_group; /* # Clusters per group */
1159 __le32 s_inodes_per_group; /* # Inodes per group */
1160 __le32 s_mtime; /* Mount time */
1161 /*30*/ __le32 s_wtime; /* Write time */
1162 __le16 s_mnt_count; /* Mount count */
1163 __le16 s_max_mnt_count; /* Maximal mount count */
1164 __le16 s_magic; /* Magic signature */
1165 __le16 s_state; /* File system state */
1166 __le16 s_errors; /* Behaviour when detecting errors */
1167 __le16 s_minor_rev_level; /* minor revision level */
1168 /*40*/ __le32 s_lastcheck; /* time of last check */
1169 __le32 s_checkinterval; /* max. time between checks */
1170 __le32 s_creator_os; /* OS */
1171 __le32 s_rev_level; /* Revision level */
1172 /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
1173 __le16 s_def_resgid; /* Default gid for reserved blocks */
1174 /*
1175 * These fields are for EXT4_DYNAMIC_REV superblocks only.
1176 *
1177 * Note: the difference between the compatible feature set and
1178 * the incompatible feature set is that if there is a bit set
1179 * in the incompatible feature set that the kernel doesn't
1180 * know about, it should refuse to mount the filesystem.
1181 *
1182 * e2fsck's requirements are more strict; if it doesn't know
1183 * about a feature in either the compatible or incompatible
1184 * feature set, it must abort and not try to meddle with
1185 * things it doesn't understand...
1186 */
1187 __le32 s_first_ino; /* First non-reserved inode */
1188 __le16 s_inode_size; /* size of inode structure */
1189 __le16 s_block_group_nr; /* block group # of this superblock */
1190 __le32 s_feature_compat; /* compatible feature set */
1191 /*60*/ __le32 s_feature_incompat; /* incompatible feature set */
1192 __le32 s_feature_ro_compat; /* readonly-compatible feature set */
1193 /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
1194 /*78*/ char s_volume_name[16]; /* volume name */
1195 /*88*/ char s_last_mounted[64]; /* directory where last mounted */
1196 /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
1197 /*
1198 * Performance hints. Directory preallocation should only
1199 * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
1200 */
1201 __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
1202 __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
1203 __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
1204 /*
1205 * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
1206 */
1207 /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
1208 /*E0*/ __le32 s_journal_inum; /* inode number of journal file */
1209 __le32 s_journal_dev; /* device number of journal file */
1210 __le32 s_last_orphan; /* start of list of inodes to delete */
1211 __le32 s_hash_seed[4]; /* HTREE hash seed */
1212 __u8 s_def_hash_version; /* Default hash version to use */
1213 __u8 s_jnl_backup_type;
1214 __le16 s_desc_size; /* size of group descriptor */
1215 /*100*/ __le32 s_default_mount_opts;
1216 __le32 s_first_meta_bg; /* First metablock block group */
1217 __le32 s_mkfs_time; /* When the filesystem was created */
1218 __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
1219 /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
1220 /*150*/ __le32 s_blocks_count_hi; /* Blocks count */
1221 __le32 s_r_blocks_count_hi; /* Reserved blocks count */
1222 __le32 s_free_blocks_count_hi; /* Free blocks count */
1223 __le16 s_min_extra_isize; /* All inodes have at least # bytes */
1224 __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
1225 __le32 s_flags; /* Miscellaneous flags */
1226 __le16 s_raid_stride; /* RAID stride */
1227 __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
1228 __le64 s_mmp_block; /* Block for multi-mount protection */
1229 __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
1230 __u8 s_log_groups_per_flex; /* FLEX_BG group size */
1231 __u8 s_checksum_type; /* metadata checksum algorithm used */
1232 __u8 s_encryption_level; /* versioning level for encryption */
1233 __u8 s_reserved_pad; /* Padding to next 32bits */
1234 __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
1235 __le32 s_snapshot_inum; /* Inode number of active snapshot */
1236 __le32 s_snapshot_id; /* sequential ID of active snapshot */
1237 __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
1238 snapshot's future use */
1239 __le32 s_snapshot_list; /* inode number of the head of the
1240 on-disk snapshot list */
1241 #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
1242 __le32 s_error_count; /* number of fs errors */
1243 __le32 s_first_error_time; /* first time an error happened */
1244 __le32 s_first_error_ino; /* inode involved in first error */
1245 __le64 s_first_error_block; /* block involved of first error */
1246 __u8 s_first_error_func[32]; /* function where the error happened */
1247 __le32 s_first_error_line; /* line number where error happened */
1248 __le32 s_last_error_time; /* most recent time of an error */
1249 __le32 s_last_error_ino; /* inode involved in last error */
1250 __le32 s_last_error_line; /* line number where error happened */
1251 __le64 s_last_error_block; /* block involved of last error */
1252 __u8 s_last_error_func[32]; /* function where the error happened */
1253 #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
1254 __u8 s_mount_opts[64];
1255 __le32 s_usr_quota_inum; /* inode for tracking user quota */
1256 __le32 s_grp_quota_inum; /* inode for tracking group quota */
1257 __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
1258 __le32 s_backup_bgs[2]; /* groups with sparse_super2 SBs */
1259 __u8 s_encrypt_algos[4]; /* Encryption algorithms in use */
1260 __u8 s_encrypt_pw_salt[16]; /* Salt used for string2key algorithm */
1261 __le32 s_lpf_ino; /* Location of the lost+found inode */
1262 __le32 s_prj_quota_inum; /* inode for tracking project quota */
1263 __le32 s_checksum_seed; /* crc32c(uuid) if csum_seed set */
1264 __le32 s_reserved[98]; /* Padding to the end of the block */
1265 __le32 s_checksum; /* crc32c(superblock) */
1266 };
1267
1268 #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
1269
1270 #ifdef __KERNEL__
1271
1272 /*
1273 * run-time mount flags
1274 */
1275 #define EXT4_MF_MNTDIR_SAMPLED 0x0001
1276 #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
1277 #define EXT4_MF_TEST_DUMMY_ENCRYPTION 0x0004
1278
1279 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1280 #define DUMMY_ENCRYPTION_ENABLED(sbi) (unlikely((sbi)->s_mount_flags & \
1281 EXT4_MF_TEST_DUMMY_ENCRYPTION))
1282 #else
1283 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0)
1284 #endif
1285
1286 /* Number of quota types we support */
1287 #define EXT4_MAXQUOTAS 2
1288
1289 /*
1290 * fourth extended-fs super-block data in memory
1291 */
1292 struct ext4_sb_info {
1293 unsigned long s_desc_size; /* Size of a group descriptor in bytes */
1294 unsigned long s_inodes_per_block;/* Number of inodes per block */
1295 unsigned long s_blocks_per_group;/* Number of blocks in a group */
1296 unsigned long s_clusters_per_group; /* Number of clusters in a group */
1297 unsigned long s_inodes_per_group;/* Number of inodes in a group */
1298 unsigned long s_itb_per_group; /* Number of inode table blocks per group */
1299 unsigned long s_gdb_count; /* Number of group descriptor blocks */
1300 unsigned long s_desc_per_block; /* Number of group descriptors per block */
1301 ext4_group_t s_groups_count; /* Number of groups in the fs */
1302 ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
1303 unsigned long s_overhead; /* # of fs overhead clusters */
1304 unsigned int s_cluster_ratio; /* Number of blocks per cluster */
1305 unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
1306 loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
1307 struct buffer_head * s_sbh; /* Buffer containing the super block */
1308 struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
1309 struct buffer_head * __rcu *s_group_desc;
1310 unsigned int s_mount_opt;
1311 unsigned int s_mount_opt2;
1312 unsigned int s_mount_flags;
1313 unsigned int s_def_mount_opt;
1314 ext4_fsblk_t s_sb_block;
1315 atomic64_t s_resv_clusters;
1316 kuid_t s_resuid;
1317 kgid_t s_resgid;
1318 unsigned short s_mount_state;
1319 unsigned short s_pad;
1320 int s_addr_per_block_bits;
1321 int s_desc_per_block_bits;
1322 int s_inode_size;
1323 int s_first_ino;
1324 unsigned int s_inode_readahead_blks;
1325 unsigned int s_inode_goal;
1326 spinlock_t s_next_gen_lock;
1327 u32 s_next_generation;
1328 u32 s_hash_seed[4];
1329 int s_def_hash_version;
1330 int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
1331 struct percpu_counter s_freeclusters_counter;
1332 struct percpu_counter s_freeinodes_counter;
1333 struct percpu_counter s_dirs_counter;
1334 struct percpu_counter s_dirtyclusters_counter;
1335 struct blockgroup_lock *s_blockgroup_lock;
1336 struct proc_dir_entry *s_proc;
1337 struct kobject s_kobj;
1338 struct completion s_kobj_unregister;
1339 struct super_block *s_sb;
1340
1341 /* Journaling */
1342 struct journal_s *s_journal;
1343 struct list_head s_orphan;
1344 struct mutex s_orphan_lock;
1345 unsigned long s_resize_flags; /* Flags indicating if there
1346 is a resizer */
1347 unsigned long s_commit_interval;
1348 u32 s_max_batch_time;
1349 u32 s_min_batch_time;
1350 struct block_device *journal_bdev;
1351 #ifdef CONFIG_QUOTA
1352 char *s_qf_names[EXT4_MAXQUOTAS]; /* Names of quota files with journalled quota */
1353 int s_jquota_fmt; /* Format of quota to use */
1354 #endif
1355 unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
1356 struct rb_root system_blks;
1357
1358 #ifdef EXTENTS_STATS
1359 /* ext4 extents stats */
1360 unsigned long s_ext_min;
1361 unsigned long s_ext_max;
1362 unsigned long s_depth_max;
1363 spinlock_t s_ext_stats_lock;
1364 unsigned long s_ext_blocks;
1365 unsigned long s_ext_extents;
1366 #endif
1367
1368 /* for buddy allocator */
1369 struct ext4_group_info ** __rcu *s_group_info;
1370 struct inode *s_buddy_cache;
1371 spinlock_t s_md_lock;
1372 unsigned short *s_mb_offsets;
1373 unsigned int *s_mb_maxs;
1374 unsigned int s_group_info_size;
1375
1376 /* tunables */
1377 unsigned long s_stripe;
1378 unsigned int s_mb_stream_request;
1379 unsigned int s_mb_max_to_scan;
1380 unsigned int s_mb_min_to_scan;
1381 unsigned int s_mb_stats;
1382 unsigned int s_mb_order2_reqs;
1383 unsigned int s_mb_group_prealloc;
1384 unsigned int s_max_dir_size_kb;
1385 /* where last allocation was done - for stream allocation */
1386 unsigned long s_mb_last_group;
1387 unsigned long s_mb_last_start;
1388
1389 /* stats for buddy allocator */
1390 atomic_t s_bal_reqs; /* number of reqs with len > 1 */
1391 atomic_t s_bal_success; /* we found long enough chunks */
1392 atomic_t s_bal_allocated; /* in blocks */
1393 atomic_t s_bal_ex_scanned; /* total extents scanned */
1394 atomic_t s_bal_goals; /* goal hits */
1395 atomic_t s_bal_breaks; /* too long searches */
1396 atomic_t s_bal_2orders; /* 2^order hits */
1397 spinlock_t s_bal_lock;
1398 unsigned long s_mb_buddies_generated;
1399 unsigned long long s_mb_generation_time;
1400 atomic_t s_mb_lost_chunks;
1401 atomic_t s_mb_preallocated;
1402 atomic_t s_mb_discarded;
1403 atomic_t s_lock_busy;
1404
1405 /* locality groups */
1406 struct ext4_locality_group __percpu *s_locality_groups;
1407
1408 /* for write statistics */
1409 unsigned long s_sectors_written_start;
1410 u64 s_kbytes_written;
1411
1412 /* the size of zero-out chunk */
1413 unsigned int s_extent_max_zeroout_kb;
1414
1415 unsigned int s_log_groups_per_flex;
1416 struct flex_groups * __rcu *s_flex_groups;
1417 ext4_group_t s_flex_groups_allocated;
1418
1419 /* workqueue for reserved extent conversions (buffered io) */
1420 struct workqueue_struct *rsv_conversion_wq;
1421
1422 /* timer for periodic error stats printing */
1423 struct timer_list s_err_report;
1424
1425 /* Lazy inode table initialization info */
1426 struct ext4_li_request *s_li_request;
1427 /* Wait multiplier for lazy initialization thread */
1428 unsigned int s_li_wait_mult;
1429
1430 /* Kernel thread for multiple mount protection */
1431 struct task_struct *s_mmp_tsk;
1432
1433 /* record the last minlen when FITRIM is called. */
1434 atomic_t s_last_trim_minblks;
1435
1436 /* Reference to checksum algorithm driver via cryptoapi */
1437 struct crypto_shash *s_chksum_driver;
1438
1439 /* Precomputed FS UUID checksum for seeding other checksums */
1440 __u32 s_csum_seed;
1441
1442 /* Reclaim extents from extent status tree */
1443 struct shrinker s_es_shrinker;
1444 struct list_head s_es_list; /* List of inodes with reclaimable extents */
1445 long s_es_nr_inode;
1446 struct ext4_es_stats s_es_stats;
1447 struct mb2_cache *s_mb_cache;
1448 spinlock_t s_es_lock ____cacheline_aligned_in_smp;
1449
1450 /* Ratelimit ext4 messages. */
1451 struct ratelimit_state s_err_ratelimit_state;
1452 struct ratelimit_state s_warning_ratelimit_state;
1453 struct ratelimit_state s_msg_ratelimit_state;
1454 };
1455
EXT4_SB(struct super_block * sb)1456 static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
1457 {
1458 return sb->s_fs_info;
1459 }
EXT4_I(struct inode * inode)1460 static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
1461 {
1462 return container_of(inode, struct ext4_inode_info, vfs_inode);
1463 }
1464
ext4_current_time(struct inode * inode)1465 static inline struct timespec ext4_current_time(struct inode *inode)
1466 {
1467 return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
1468 current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
1469 }
1470
ext4_valid_inum(struct super_block * sb,unsigned long ino)1471 static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
1472 {
1473 return ino == EXT4_ROOT_INO ||
1474 (ino >= EXT4_FIRST_INO(sb) &&
1475 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
1476 }
1477
ext4_set_io_unwritten_flag(struct inode * inode,struct ext4_io_end * io_end)1478 static inline void ext4_set_io_unwritten_flag(struct inode *inode,
1479 struct ext4_io_end *io_end)
1480 {
1481 if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
1482 io_end->flag |= EXT4_IO_END_UNWRITTEN;
1483 atomic_inc(&EXT4_I(inode)->i_unwritten);
1484 }
1485 }
1486
ext4_inode_aio(struct inode * inode)1487 static inline ext4_io_end_t *ext4_inode_aio(struct inode *inode)
1488 {
1489 return inode->i_private;
1490 }
1491
ext4_inode_aio_set(struct inode * inode,ext4_io_end_t * io)1492 static inline void ext4_inode_aio_set(struct inode *inode, ext4_io_end_t *io)
1493 {
1494 inode->i_private = io;
1495 }
1496
1497 /*
1498 * Returns: sbi->field[index]
1499 * Used to access an array element from the following sbi fields which require
1500 * rcu protection to avoid dereferencing an invalid pointer due to reassignment
1501 * - s_group_desc
1502 * - s_group_info
1503 * - s_flex_group
1504 */
1505 #define sbi_array_rcu_deref(sbi, field, index) \
1506 ({ \
1507 typeof(*((sbi)->field)) _v; \
1508 rcu_read_lock(); \
1509 _v = ((typeof(_v)*)rcu_dereference((sbi)->field))[index]; \
1510 rcu_read_unlock(); \
1511 _v; \
1512 })
1513
1514 /*
1515 * Inode dynamic state flags
1516 */
1517 enum {
1518 EXT4_STATE_JDATA, /* journaled data exists */
1519 EXT4_STATE_NEW, /* inode is newly created */
1520 EXT4_STATE_XATTR, /* has in-inode xattrs */
1521 EXT4_STATE_NO_EXPAND, /* No space for expansion */
1522 EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
1523 EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
1524 EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
1525 EXT4_STATE_NEWENTRY, /* File just added to dir */
1526 EXT4_STATE_DIOREAD_LOCK, /* Disable support for dio read
1527 nolocking */
1528 EXT4_STATE_MAY_INLINE_DATA, /* may have in-inode data */
1529 EXT4_STATE_ORDERED_MODE, /* data=ordered mode */
1530 EXT4_STATE_EXT_PRECACHED, /* extents have been precached */
1531 };
1532
1533 #define EXT4_INODE_BIT_FNS(name, field, offset) \
1534 static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
1535 { \
1536 return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1537 } \
1538 static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
1539 { \
1540 set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1541 } \
1542 static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
1543 { \
1544 clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
1545 }
1546
1547 /* Add these declarations here only so that these functions can be
1548 * found by name. Otherwise, they are very hard to locate. */
1549 static inline int ext4_test_inode_flag(struct inode *inode, int bit);
1550 static inline void ext4_set_inode_flag(struct inode *inode, int bit);
1551 static inline void ext4_clear_inode_flag(struct inode *inode, int bit);
1552 EXT4_INODE_BIT_FNS(flag, flags, 0)
1553
1554 /* Add these declarations here only so that these functions can be
1555 * found by name. Otherwise, they are very hard to locate. */
1556 static inline int ext4_test_inode_state(struct inode *inode, int bit);
1557 static inline void ext4_set_inode_state(struct inode *inode, int bit);
1558 static inline void ext4_clear_inode_state(struct inode *inode, int bit);
1559 #if (BITS_PER_LONG < 64)
1560 EXT4_INODE_BIT_FNS(state, state_flags, 0)
1561
ext4_clear_state_flags(struct ext4_inode_info * ei)1562 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1563 {
1564 (ei)->i_state_flags = 0;
1565 }
1566 #else
1567 EXT4_INODE_BIT_FNS(state, flags, 32)
1568
ext4_clear_state_flags(struct ext4_inode_info * ei)1569 static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
1570 {
1571 /* We depend on the fact that callers will set i_flags */
1572 }
1573 #endif
1574 #else
1575 /* Assume that user mode programs are passing in an ext4fs superblock, not
1576 * a kernel struct super_block. This will allow us to call the feature-test
1577 * macros from user land. */
1578 #define EXT4_SB(sb) (sb)
1579 #endif
1580
1581 /*
1582 * Returns true if the inode is inode is encrypted
1583 */
ext4_encrypted_inode(struct inode * inode)1584 static inline int ext4_encrypted_inode(struct inode *inode)
1585 {
1586 #ifdef CONFIG_EXT4_FS_ENCRYPTION
1587 return ext4_test_inode_flag(inode, EXT4_INODE_ENCRYPT);
1588 #else
1589 return 0;
1590 #endif
1591 }
1592
1593 #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
1594
1595 /*
1596 * Codes for operating systems
1597 */
1598 #define EXT4_OS_LINUX 0
1599 #define EXT4_OS_HURD 1
1600 #define EXT4_OS_MASIX 2
1601 #define EXT4_OS_FREEBSD 3
1602 #define EXT4_OS_LITES 4
1603
1604 /*
1605 * Revision levels
1606 */
1607 #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
1608 #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
1609
1610 #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
1611 #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
1612
1613 #define EXT4_GOOD_OLD_INODE_SIZE 128
1614
1615 /*
1616 * Feature set definitions
1617 */
1618
1619 /* Use the ext4_{has,set,clear}_feature_* helpers; these will be removed */
1620 #define EXT4_HAS_COMPAT_FEATURE(sb,mask) \
1621 ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
1622 #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \
1623 ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
1624 #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \
1625 ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
1626 #define EXT4_SET_COMPAT_FEATURE(sb,mask) \
1627 EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
1628 #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \
1629 EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
1630 #define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \
1631 EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
1632 #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \
1633 EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
1634 #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
1635 EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
1636 #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \
1637 EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
1638
1639 #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
1640 #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
1641 #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
1642 #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
1643 #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
1644 #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
1645 #define EXT4_FEATURE_COMPAT_SPARSE_SUPER2 0x0200
1646
1647 #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
1648 #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
1649 #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
1650 #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
1651 #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
1652 #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
1653 #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
1654 #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
1655 #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
1656 /*
1657 * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
1658 * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
1659 * all other data structures' checksums. However, the METADATA_CSUM and
1660 * GDT_CSUM bits are mutually exclusive.
1661 */
1662 #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
1663 #define EXT4_FEATURE_RO_COMPAT_READONLY 0x1000
1664 #define EXT4_FEATURE_RO_COMPAT_PROJECT 0x2000
1665
1666 #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
1667 #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
1668 #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
1669 #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
1670 #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
1671 #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
1672 #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
1673 #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
1674 #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
1675 #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
1676 #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
1677 #define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000
1678 #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
1679 #define EXT4_FEATURE_INCOMPAT_INLINE_DATA 0x8000 /* data in inode */
1680 #define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000
1681
1682 #define EXT4_FEATURE_COMPAT_FUNCS(name, flagname) \
1683 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1684 { \
1685 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1686 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname)) != 0); \
1687 } \
1688 static inline void ext4_set_feature_##name(struct super_block *sb) \
1689 { \
1690 EXT4_SB(sb)->s_es->s_feature_compat |= \
1691 cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1692 } \
1693 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1694 { \
1695 EXT4_SB(sb)->s_es->s_feature_compat &= \
1696 ~cpu_to_le32(EXT4_FEATURE_COMPAT_##flagname); \
1697 }
1698
1699 #define EXT4_FEATURE_RO_COMPAT_FUNCS(name, flagname) \
1700 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1701 { \
1702 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1703 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname)) != 0); \
1704 } \
1705 static inline void ext4_set_feature_##name(struct super_block *sb) \
1706 { \
1707 EXT4_SB(sb)->s_es->s_feature_ro_compat |= \
1708 cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1709 } \
1710 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1711 { \
1712 EXT4_SB(sb)->s_es->s_feature_ro_compat &= \
1713 ~cpu_to_le32(EXT4_FEATURE_RO_COMPAT_##flagname); \
1714 }
1715
1716 #define EXT4_FEATURE_INCOMPAT_FUNCS(name, flagname) \
1717 static inline bool ext4_has_feature_##name(struct super_block *sb) \
1718 { \
1719 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1720 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname)) != 0); \
1721 } \
1722 static inline void ext4_set_feature_##name(struct super_block *sb) \
1723 { \
1724 EXT4_SB(sb)->s_es->s_feature_incompat |= \
1725 cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1726 } \
1727 static inline void ext4_clear_feature_##name(struct super_block *sb) \
1728 { \
1729 EXT4_SB(sb)->s_es->s_feature_incompat &= \
1730 ~cpu_to_le32(EXT4_FEATURE_INCOMPAT_##flagname); \
1731 }
1732
EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc,DIR_PREALLOC)1733 EXT4_FEATURE_COMPAT_FUNCS(dir_prealloc, DIR_PREALLOC)
1734 EXT4_FEATURE_COMPAT_FUNCS(imagic_inodes, IMAGIC_INODES)
1735 EXT4_FEATURE_COMPAT_FUNCS(journal, HAS_JOURNAL)
1736 EXT4_FEATURE_COMPAT_FUNCS(xattr, EXT_ATTR)
1737 EXT4_FEATURE_COMPAT_FUNCS(resize_inode, RESIZE_INODE)
1738 EXT4_FEATURE_COMPAT_FUNCS(dir_index, DIR_INDEX)
1739 EXT4_FEATURE_COMPAT_FUNCS(sparse_super2, SPARSE_SUPER2)
1740
1741 EXT4_FEATURE_RO_COMPAT_FUNCS(sparse_super, SPARSE_SUPER)
1742 EXT4_FEATURE_RO_COMPAT_FUNCS(large_file, LARGE_FILE)
1743 EXT4_FEATURE_RO_COMPAT_FUNCS(btree_dir, BTREE_DIR)
1744 EXT4_FEATURE_RO_COMPAT_FUNCS(huge_file, HUGE_FILE)
1745 EXT4_FEATURE_RO_COMPAT_FUNCS(gdt_csum, GDT_CSUM)
1746 EXT4_FEATURE_RO_COMPAT_FUNCS(dir_nlink, DIR_NLINK)
1747 EXT4_FEATURE_RO_COMPAT_FUNCS(extra_isize, EXTRA_ISIZE)
1748 EXT4_FEATURE_RO_COMPAT_FUNCS(quota, QUOTA)
1749 EXT4_FEATURE_RO_COMPAT_FUNCS(bigalloc, BIGALLOC)
1750 EXT4_FEATURE_RO_COMPAT_FUNCS(metadata_csum, METADATA_CSUM)
1751 EXT4_FEATURE_RO_COMPAT_FUNCS(readonly, READONLY)
1752 EXT4_FEATURE_RO_COMPAT_FUNCS(project, PROJECT)
1753
1754 EXT4_FEATURE_INCOMPAT_FUNCS(compression, COMPRESSION)
1755 EXT4_FEATURE_INCOMPAT_FUNCS(filetype, FILETYPE)
1756 EXT4_FEATURE_INCOMPAT_FUNCS(journal_needs_recovery, RECOVER)
1757 EXT4_FEATURE_INCOMPAT_FUNCS(journal_dev, JOURNAL_DEV)
1758 EXT4_FEATURE_INCOMPAT_FUNCS(meta_bg, META_BG)
1759 EXT4_FEATURE_INCOMPAT_FUNCS(extents, EXTENTS)
1760 EXT4_FEATURE_INCOMPAT_FUNCS(64bit, 64BIT)
1761 EXT4_FEATURE_INCOMPAT_FUNCS(mmp, MMP)
1762 EXT4_FEATURE_INCOMPAT_FUNCS(flex_bg, FLEX_BG)
1763 EXT4_FEATURE_INCOMPAT_FUNCS(ea_inode, EA_INODE)
1764 EXT4_FEATURE_INCOMPAT_FUNCS(dirdata, DIRDATA)
1765 EXT4_FEATURE_INCOMPAT_FUNCS(csum_seed, CSUM_SEED)
1766 EXT4_FEATURE_INCOMPAT_FUNCS(largedir, LARGEDIR)
1767 EXT4_FEATURE_INCOMPAT_FUNCS(inline_data, INLINE_DATA)
1768 EXT4_FEATURE_INCOMPAT_FUNCS(encrypt, ENCRYPT)
1769
1770 #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1771 #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1772 EXT4_FEATURE_INCOMPAT_META_BG)
1773 #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1774 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1775 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1776
1777 #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1778 #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1779 EXT4_FEATURE_INCOMPAT_RECOVER| \
1780 EXT4_FEATURE_INCOMPAT_META_BG)
1781 #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1782 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1783 EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
1784
1785 #define EXT4_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
1786 #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
1787 EXT4_FEATURE_INCOMPAT_RECOVER| \
1788 EXT4_FEATURE_INCOMPAT_META_BG| \
1789 EXT4_FEATURE_INCOMPAT_EXTENTS| \
1790 EXT4_FEATURE_INCOMPAT_64BIT| \
1791 EXT4_FEATURE_INCOMPAT_FLEX_BG| \
1792 EXT4_FEATURE_INCOMPAT_MMP | \
1793 EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
1794 EXT4_FEATURE_INCOMPAT_ENCRYPT | \
1795 EXT4_FEATURE_INCOMPAT_CSUM_SEED)
1796 #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
1797 EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
1798 EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
1799 EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
1800 EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
1801 EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
1802 EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
1803 EXT4_FEATURE_RO_COMPAT_BIGALLOC |\
1804 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM|\
1805 EXT4_FEATURE_RO_COMPAT_QUOTA)
1806
1807 #define EXTN_FEATURE_FUNCS(ver) \
1808 static inline bool ext4_has_unknown_ext##ver##_compat_features(struct super_block *sb) \
1809 { \
1810 return ((EXT4_SB(sb)->s_es->s_feature_compat & \
1811 cpu_to_le32(~EXT##ver##_FEATURE_COMPAT_SUPP)) != 0); \
1812 } \
1813 static inline bool ext4_has_unknown_ext##ver##_ro_compat_features(struct super_block *sb) \
1814 { \
1815 return ((EXT4_SB(sb)->s_es->s_feature_ro_compat & \
1816 cpu_to_le32(~EXT##ver##_FEATURE_RO_COMPAT_SUPP)) != 0); \
1817 } \
1818 static inline bool ext4_has_unknown_ext##ver##_incompat_features(struct super_block *sb) \
1819 { \
1820 return ((EXT4_SB(sb)->s_es->s_feature_incompat & \
1821 cpu_to_le32(~EXT##ver##_FEATURE_INCOMPAT_SUPP)) != 0); \
1822 }
1823
1824 EXTN_FEATURE_FUNCS(2)
1825 EXTN_FEATURE_FUNCS(3)
1826 EXTN_FEATURE_FUNCS(4)
1827
1828 static inline bool ext4_has_compat_features(struct super_block *sb)
1829 {
1830 return (EXT4_SB(sb)->s_es->s_feature_compat != 0);
1831 }
ext4_has_ro_compat_features(struct super_block * sb)1832 static inline bool ext4_has_ro_compat_features(struct super_block *sb)
1833 {
1834 return (EXT4_SB(sb)->s_es->s_feature_ro_compat != 0);
1835 }
ext4_has_incompat_features(struct super_block * sb)1836 static inline bool ext4_has_incompat_features(struct super_block *sb)
1837 {
1838 return (EXT4_SB(sb)->s_es->s_feature_incompat != 0);
1839 }
1840
1841 /*
1842 * Default values for user and/or group using reserved blocks
1843 */
1844 #define EXT4_DEF_RESUID 0
1845 #define EXT4_DEF_RESGID 0
1846
1847 #define EXT4_DEF_INODE_READAHEAD_BLKS 32
1848
1849 /*
1850 * Default mount options
1851 */
1852 #define EXT4_DEFM_DEBUG 0x0001
1853 #define EXT4_DEFM_BSDGROUPS 0x0002
1854 #define EXT4_DEFM_XATTR_USER 0x0004
1855 #define EXT4_DEFM_ACL 0x0008
1856 #define EXT4_DEFM_UID16 0x0010
1857 #define EXT4_DEFM_JMODE 0x0060
1858 #define EXT4_DEFM_JMODE_DATA 0x0020
1859 #define EXT4_DEFM_JMODE_ORDERED 0x0040
1860 #define EXT4_DEFM_JMODE_WBACK 0x0060
1861 #define EXT4_DEFM_NOBARRIER 0x0100
1862 #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
1863 #define EXT4_DEFM_DISCARD 0x0400
1864 #define EXT4_DEFM_NODELALLOC 0x0800
1865
1866 /*
1867 * Default journal batch times
1868 */
1869 #define EXT4_DEF_MIN_BATCH_TIME 0
1870 #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
1871
1872 /*
1873 * Minimum number of groups in a flexgroup before we separate out
1874 * directories into the first block group of a flexgroup
1875 */
1876 #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
1877
1878 /*
1879 * Structure of a directory entry
1880 */
1881 #define EXT4_NAME_LEN 255
1882
1883 struct ext4_dir_entry {
1884 __le32 inode; /* Inode number */
1885 __le16 rec_len; /* Directory entry length */
1886 __le16 name_len; /* Name length */
1887 char name[EXT4_NAME_LEN]; /* File name */
1888 };
1889
1890 /*
1891 * The new version of the directory entry. Since EXT4 structures are
1892 * stored in intel byte order, and the name_len field could never be
1893 * bigger than 255 chars, it's safe to reclaim the extra byte for the
1894 * file_type field.
1895 */
1896 struct ext4_dir_entry_2 {
1897 __le32 inode; /* Inode number */
1898 __le16 rec_len; /* Directory entry length */
1899 __u8 name_len; /* Name length */
1900 __u8 file_type;
1901 char name[EXT4_NAME_LEN]; /* File name */
1902 };
1903
1904 /*
1905 * This is a bogus directory entry at the end of each leaf block that
1906 * records checksums.
1907 */
1908 struct ext4_dir_entry_tail {
1909 __le32 det_reserved_zero1; /* Pretend to be unused */
1910 __le16 det_rec_len; /* 12 */
1911 __u8 det_reserved_zero2; /* Zero name length */
1912 __u8 det_reserved_ft; /* 0xDE, fake file type */
1913 __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
1914 };
1915
1916 #define EXT4_DIRENT_TAIL(block, blocksize) \
1917 ((struct ext4_dir_entry_tail *)(((void *)(block)) + \
1918 ((blocksize) - \
1919 sizeof(struct ext4_dir_entry_tail))))
1920
1921 /*
1922 * Ext4 directory file types. Only the low 3 bits are used. The
1923 * other bits are reserved for now.
1924 */
1925 #define EXT4_FT_UNKNOWN 0
1926 #define EXT4_FT_REG_FILE 1
1927 #define EXT4_FT_DIR 2
1928 #define EXT4_FT_CHRDEV 3
1929 #define EXT4_FT_BLKDEV 4
1930 #define EXT4_FT_FIFO 5
1931 #define EXT4_FT_SOCK 6
1932 #define EXT4_FT_SYMLINK 7
1933
1934 #define EXT4_FT_MAX 8
1935
1936 #define EXT4_FT_DIR_CSUM 0xDE
1937
1938 /*
1939 * EXT4_DIR_PAD defines the directory entries boundaries
1940 *
1941 * NOTE: It must be a multiple of 4
1942 */
1943 #define EXT4_DIR_PAD 4
1944 #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
1945 #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
1946 ~EXT4_DIR_ROUND)
1947 #define EXT4_MAX_REC_LEN ((1<<16)-1)
1948
1949 /*
1950 * If we ever get support for fs block sizes > page_size, we'll need
1951 * to remove the #if statements in the next two functions...
1952 */
1953 static inline unsigned int
ext4_rec_len_from_disk(__le16 dlen,unsigned blocksize)1954 ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
1955 {
1956 unsigned len = le16_to_cpu(dlen);
1957
1958 #if (PAGE_CACHE_SIZE >= 65536)
1959 if (len == EXT4_MAX_REC_LEN || len == 0)
1960 return blocksize;
1961 return (len & 65532) | ((len & 3) << 16);
1962 #else
1963 return len;
1964 #endif
1965 }
1966
ext4_rec_len_to_disk(unsigned len,unsigned blocksize)1967 static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
1968 {
1969 if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
1970 BUG();
1971 #if (PAGE_CACHE_SIZE >= 65536)
1972 if (len < 65536)
1973 return cpu_to_le16(len);
1974 if (len == blocksize) {
1975 if (blocksize == 65536)
1976 return cpu_to_le16(EXT4_MAX_REC_LEN);
1977 else
1978 return cpu_to_le16(0);
1979 }
1980 return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
1981 #else
1982 return cpu_to_le16(len);
1983 #endif
1984 }
1985
1986 /*
1987 * Hash Tree Directory indexing
1988 * (c) Daniel Phillips, 2001
1989 */
1990
1991 #define is_dx(dir) (ext4_has_feature_dir_index((dir)->i_sb) && \
1992 ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
1993 #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
1994 #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
1995
1996 /* Legal values for the dx_root hash_version field: */
1997
1998 #define DX_HASH_LEGACY 0
1999 #define DX_HASH_HALF_MD4 1
2000 #define DX_HASH_TEA 2
2001 #define DX_HASH_LEGACY_UNSIGNED 3
2002 #define DX_HASH_HALF_MD4_UNSIGNED 4
2003 #define DX_HASH_TEA_UNSIGNED 5
2004
ext4_chksum(struct ext4_sb_info * sbi,u32 crc,const void * address,unsigned int length)2005 static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
2006 const void *address, unsigned int length)
2007 {
2008 struct {
2009 struct shash_desc shash;
2010 char ctx[4];
2011 } desc;
2012 int err;
2013
2014 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver)!=sizeof(desc.ctx));
2015
2016 desc.shash.tfm = sbi->s_chksum_driver;
2017 desc.shash.flags = 0;
2018 *(u32 *)desc.ctx = crc;
2019
2020 err = crypto_shash_update(&desc.shash, address, length);
2021 BUG_ON(err);
2022
2023 return *(u32 *)desc.ctx;
2024 }
2025
2026 #ifdef __KERNEL__
2027
2028 /* hash info structure used by the directory hash */
2029 struct dx_hash_info
2030 {
2031 u32 hash;
2032 u32 minor_hash;
2033 int hash_version;
2034 u32 *seed;
2035 };
2036
2037
2038 /* 32 and 64 bit signed EOF for dx directories */
2039 #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
2040 #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
2041
2042
2043 /*
2044 * Control parameters used by ext4_htree_next_block
2045 */
2046 #define HASH_NB_ALWAYS 1
2047
2048 struct ext4_filename {
2049 const struct qstr *usr_fname;
2050 struct ext4_str disk_name;
2051 struct dx_hash_info hinfo;
2052 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2053 struct ext4_str crypto_buf;
2054 #endif
2055 };
2056
2057 #define fname_name(p) ((p)->disk_name.name)
2058 #define fname_len(p) ((p)->disk_name.len)
2059
2060 /*
2061 * Describe an inode's exact location on disk and in memory
2062 */
2063 struct ext4_iloc
2064 {
2065 struct buffer_head *bh;
2066 unsigned long offset;
2067 ext4_group_t block_group;
2068 };
2069
ext4_raw_inode(struct ext4_iloc * iloc)2070 static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
2071 {
2072 return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
2073 }
2074
2075 /*
2076 * This structure is stuffed into the struct file's private_data field
2077 * for directories. It is where we put information so that we can do
2078 * readdir operations in hash tree order.
2079 */
2080 struct dir_private_info {
2081 struct rb_root root;
2082 struct rb_node *curr_node;
2083 struct fname *extra_fname;
2084 loff_t last_pos;
2085 __u32 curr_hash;
2086 __u32 curr_minor_hash;
2087 __u32 next_hash;
2088 };
2089
2090 /* calculate the first block number of the group */
2091 static inline ext4_fsblk_t
ext4_group_first_block_no(struct super_block * sb,ext4_group_t group_no)2092 ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
2093 {
2094 return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
2095 le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
2096 }
2097
2098 /*
2099 * Special error return code only used by dx_probe() and its callers.
2100 */
2101 #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
2102
2103 /*
2104 * Timeout and state flag for lazy initialization inode thread.
2105 */
2106 #define EXT4_DEF_LI_WAIT_MULT 10
2107 #define EXT4_DEF_LI_MAX_START_DELAY 5
2108 #define EXT4_LAZYINIT_QUIT 0x0001
2109 #define EXT4_LAZYINIT_RUNNING 0x0002
2110
2111 /*
2112 * Lazy inode table initialization info
2113 */
2114 struct ext4_lazy_init {
2115 unsigned long li_state;
2116 struct list_head li_request_list;
2117 struct mutex li_list_mtx;
2118 };
2119
2120 struct ext4_li_request {
2121 struct super_block *lr_super;
2122 struct ext4_sb_info *lr_sbi;
2123 ext4_group_t lr_next_group;
2124 struct list_head lr_request;
2125 unsigned long lr_next_sched;
2126 unsigned long lr_timeout;
2127 };
2128
2129 struct ext4_features {
2130 struct kobject f_kobj;
2131 struct completion f_kobj_unregister;
2132 };
2133
2134 /*
2135 * This structure will be used for multiple mount protection. It will be
2136 * written into the block number saved in the s_mmp_block field in the
2137 * superblock. Programs that check MMP should assume that if
2138 * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
2139 * to use the filesystem, regardless of how old the timestamp is.
2140 */
2141 #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
2142 #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
2143 #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
2144 #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
2145
2146 struct mmp_struct {
2147 __le32 mmp_magic; /* Magic number for MMP */
2148 __le32 mmp_seq; /* Sequence no. updated periodically */
2149
2150 /*
2151 * mmp_time, mmp_nodename & mmp_bdevname are only used for information
2152 * purposes and do not affect the correctness of the algorithm
2153 */
2154 __le64 mmp_time; /* Time last updated */
2155 char mmp_nodename[64]; /* Node which last updated MMP block */
2156 char mmp_bdevname[32]; /* Bdev which last updated MMP block */
2157
2158 /*
2159 * mmp_check_interval is used to verify if the MMP block has been
2160 * updated on the block device. The value is updated based on the
2161 * maximum time to write the MMP block during an update cycle.
2162 */
2163 __le16 mmp_check_interval;
2164
2165 __le16 mmp_pad1;
2166 __le32 mmp_pad2[226];
2167 __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
2168 };
2169
2170 /* arguments passed to the mmp thread */
2171 struct mmpd_data {
2172 struct buffer_head *bh; /* bh from initial read_mmp_block() */
2173 struct super_block *sb; /* super block of the fs */
2174 };
2175
2176 /*
2177 * Check interval multiplier
2178 * The MMP block is written every update interval and initially checked every
2179 * update interval x the multiplier (the value is then adapted based on the
2180 * write latency). The reason is that writes can be delayed under load and we
2181 * don't want readers to incorrectly assume that the filesystem is no longer
2182 * in use.
2183 */
2184 #define EXT4_MMP_CHECK_MULT 2UL
2185
2186 /*
2187 * Minimum interval for MMP checking in seconds.
2188 */
2189 #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
2190
2191 /*
2192 * Maximum interval for MMP checking in seconds.
2193 */
2194 #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
2195
2196 /*
2197 * Function prototypes
2198 */
2199
2200 /*
2201 * Ok, these declarations are also in <linux/kernel.h> but none of the
2202 * ext4 source programs needs to include it so they are duplicated here.
2203 */
2204 # define NORET_TYPE /**/
2205 # define ATTRIB_NORET __attribute__((noreturn))
2206 # define NORET_AND noreturn,
2207
2208 /* bitmap.c */
2209 extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
2210 void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2211 struct ext4_group_desc *gdp,
2212 struct buffer_head *bh, int sz);
2213 int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2214 struct ext4_group_desc *gdp,
2215 struct buffer_head *bh, int sz);
2216 void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
2217 struct ext4_group_desc *gdp,
2218 struct buffer_head *bh);
2219 int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
2220 struct ext4_group_desc *gdp,
2221 struct buffer_head *bh);
2222
2223 /* balloc.c */
2224 extern void ext4_get_group_no_and_offset(struct super_block *sb,
2225 ext4_fsblk_t blocknr,
2226 ext4_group_t *blockgrpp,
2227 ext4_grpblk_t *offsetp);
2228 extern ext4_group_t ext4_get_group_number(struct super_block *sb,
2229 ext4_fsblk_t block);
2230
2231 extern unsigned int ext4_block_group(struct super_block *sb,
2232 ext4_fsblk_t blocknr);
2233 extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
2234 ext4_fsblk_t blocknr);
2235 extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
2236 extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
2237 ext4_group_t group);
2238 extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
2239 ext4_fsblk_t goal,
2240 unsigned int flags,
2241 unsigned long *count,
2242 int *errp);
2243 extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
2244 s64 nclusters, unsigned int flags);
2245 extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
2246 extern void ext4_check_blocks_bitmap(struct super_block *);
2247 extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
2248 ext4_group_t block_group,
2249 struct buffer_head ** bh);
2250 extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
2251
2252 extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
2253 ext4_group_t block_group);
2254 extern int ext4_wait_block_bitmap(struct super_block *sb,
2255 ext4_group_t block_group,
2256 struct buffer_head *bh);
2257 extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
2258 ext4_group_t block_group);
2259 extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
2260 ext4_group_t block_group,
2261 struct ext4_group_desc *gdp);
2262 ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
2263
2264 /* crypto_policy.c */
2265 int ext4_is_child_context_consistent_with_parent(struct inode *parent,
2266 struct inode *child);
2267 int ext4_inherit_context(struct inode *parent, struct inode *child);
2268 void ext4_to_hex(char *dst, char *src, size_t src_size);
2269 int ext4_process_policy(const struct ext4_encryption_policy *policy,
2270 struct inode *inode);
2271 int ext4_get_policy(struct inode *inode,
2272 struct ext4_encryption_policy *policy);
2273
2274 /* crypto.c */
2275 extern struct kmem_cache *ext4_crypt_info_cachep;
2276 bool ext4_valid_enc_modes(uint32_t contents_mode, uint32_t filenames_mode);
2277 uint32_t ext4_validate_encryption_key_size(uint32_t mode, uint32_t size);
2278 extern struct workqueue_struct *ext4_read_workqueue;
2279 struct ext4_crypto_ctx *ext4_get_crypto_ctx(struct inode *inode,
2280 gfp_t gfp_flags);
2281 void ext4_release_crypto_ctx(struct ext4_crypto_ctx *ctx);
2282 void ext4_restore_control_page(struct page *data_page);
2283 struct page *ext4_encrypt(struct inode *inode,
2284 struct page *plaintext_page,
2285 gfp_t gfp_flags);
2286 int ext4_decrypt(struct page *page);
2287 int ext4_encrypted_zeroout(struct inode *inode, struct ext4_extent *ex);
2288 extern const struct dentry_operations ext4_encrypted_d_ops;
2289
2290 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2291 int ext4_init_crypto(void);
2292 void ext4_exit_crypto(void);
ext4_sb_has_crypto(struct super_block * sb)2293 static inline int ext4_sb_has_crypto(struct super_block *sb)
2294 {
2295 return ext4_has_feature_encrypt(sb);
2296 }
2297 #else
ext4_init_crypto(void)2298 static inline int ext4_init_crypto(void) { return 0; }
ext4_exit_crypto(void)2299 static inline void ext4_exit_crypto(void) { }
ext4_sb_has_crypto(struct super_block * sb)2300 static inline int ext4_sb_has_crypto(struct super_block *sb)
2301 {
2302 return 0;
2303 }
2304 #endif
2305
2306 /* crypto_fname.c */
2307 u32 ext4_fname_crypto_round_up(u32 size, u32 blksize);
2308 unsigned ext4_fname_encrypted_size(struct inode *inode, u32 ilen);
2309 int ext4_fname_crypto_alloc_buffer(struct inode *inode,
2310 u32 ilen, struct ext4_str *crypto_str);
2311 int _ext4_fname_disk_to_usr(struct inode *inode,
2312 struct dx_hash_info *hinfo,
2313 const struct ext4_str *iname,
2314 struct ext4_str *oname);
2315 int ext4_fname_disk_to_usr(struct inode *inode,
2316 struct dx_hash_info *hinfo,
2317 const struct ext4_dir_entry_2 *de,
2318 struct ext4_str *oname);
2319 int ext4_fname_usr_to_disk(struct inode *inode,
2320 const struct qstr *iname,
2321 struct ext4_str *oname);
2322 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2323 void ext4_fname_crypto_free_buffer(struct ext4_str *crypto_str);
2324 int ext4_fname_setup_filename(struct inode *dir, const struct qstr *iname,
2325 int lookup, struct ext4_filename *fname);
2326 void ext4_fname_free_filename(struct ext4_filename *fname);
2327 #else
2328 static inline
ext4_setup_fname_crypto(struct inode * inode)2329 int ext4_setup_fname_crypto(struct inode *inode)
2330 {
2331 return 0;
2332 }
ext4_fname_crypto_free_buffer(struct ext4_str * p)2333 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)2334 static inline int ext4_fname_setup_filename(struct inode *dir,
2335 const struct qstr *iname,
2336 int lookup, struct ext4_filename *fname)
2337 {
2338 fname->usr_fname = iname;
2339 fname->disk_name.name = (unsigned char *) iname->name;
2340 fname->disk_name.len = iname->len;
2341 return 0;
2342 }
ext4_fname_free_filename(struct ext4_filename * fname)2343 static inline void ext4_fname_free_filename(struct ext4_filename *fname) { }
2344 #endif
2345
2346
2347 /* crypto_key.c */
2348 void ext4_free_crypt_info(struct ext4_crypt_info *ci);
2349 void ext4_free_encryption_info(struct inode *inode, struct ext4_crypt_info *ci);
2350
2351 #ifdef CONFIG_EXT4_FS_ENCRYPTION
2352 int ext4_has_encryption_key(struct inode *inode);
2353
2354 int ext4_get_encryption_info(struct inode *inode);
2355
ext4_encryption_info(struct inode * inode)2356 static inline struct ext4_crypt_info *ext4_encryption_info(struct inode *inode)
2357 {
2358 return EXT4_I(inode)->i_crypt_info;
2359 }
2360
2361 #else
ext4_has_encryption_key(struct inode * inode)2362 static inline int ext4_has_encryption_key(struct inode *inode)
2363 {
2364 return 0;
2365 }
ext4_get_encryption_info(struct inode * inode)2366 static inline int ext4_get_encryption_info(struct inode *inode)
2367 {
2368 return 0;
2369 }
ext4_encryption_info(struct inode * inode)2370 static inline struct ext4_crypt_info *ext4_encryption_info(struct inode *inode)
2371 {
2372 return NULL;
2373 }
2374 #endif
2375
2376
2377 /* dir.c */
2378 extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
2379 struct file *,
2380 struct ext4_dir_entry_2 *,
2381 struct buffer_head *, char *, int,
2382 unsigned int);
2383 #define ext4_check_dir_entry(dir, filp, de, bh, buf, size, offset) \
2384 unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
2385 (de), (bh), (buf), (size), (offset)))
2386 extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
2387 __u32 minor_hash,
2388 struct ext4_dir_entry_2 *dirent,
2389 struct ext4_str *ent_name);
2390 extern void ext4_htree_free_dir_info(struct dir_private_info *p);
2391 extern int ext4_find_dest_de(struct inode *dir, struct inode *inode,
2392 struct buffer_head *bh,
2393 void *buf, int buf_size,
2394 struct ext4_filename *fname,
2395 struct ext4_dir_entry_2 **dest_de);
2396 int ext4_insert_dentry(struct inode *dir,
2397 struct inode *inode,
2398 struct ext4_dir_entry_2 *de,
2399 int buf_size,
2400 struct ext4_filename *fname);
ext4_update_dx_flag(struct inode * inode)2401 static inline void ext4_update_dx_flag(struct inode *inode)
2402 {
2403 if (!ext4_has_feature_dir_index(inode->i_sb) &&
2404 ext4_test_inode_flag(inode, EXT4_INODE_INDEX)) {
2405 /* ext4_iget() should have caught this... */
2406 WARN_ON_ONCE(ext4_has_feature_metadata_csum(inode->i_sb));
2407 ext4_clear_inode_flag(inode, EXT4_INODE_INDEX);
2408 }
2409 }
2410 static unsigned char ext4_filetype_table[] = {
2411 DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
2412 };
2413
get_dtype(struct super_block * sb,int filetype)2414 static inline unsigned char get_dtype(struct super_block *sb, int filetype)
2415 {
2416 if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
2417 return DT_UNKNOWN;
2418
2419 return ext4_filetype_table[filetype];
2420 }
2421 extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
2422 void *buf, int buf_size);
2423
2424 /* fsync.c */
2425 extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
2426
2427 /* hash.c */
2428 extern int ext4fs_dirhash(const char *name, int len, struct
2429 dx_hash_info *hinfo);
2430
2431 /* ialloc.c */
2432 extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
2433 const struct qstr *qstr, __u32 goal,
2434 uid_t *owner, int handle_type,
2435 unsigned int line_no, int nblocks);
2436
2437 #define ext4_new_inode(handle, dir, mode, qstr, goal, owner) \
2438 __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), \
2439 0, 0, 0)
2440 #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, \
2441 type, nblocks) \
2442 __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), \
2443 (type), __LINE__, (nblocks))
2444
2445
2446 extern void ext4_free_inode(handle_t *, struct inode *);
2447 extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
2448 extern unsigned long ext4_count_free_inodes(struct super_block *);
2449 extern unsigned long ext4_count_dirs(struct super_block *);
2450 extern void ext4_check_inodes_bitmap(struct super_block *);
2451 extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
2452 extern int ext4_init_inode_table(struct super_block *sb,
2453 ext4_group_t group, int barrier);
2454 extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
2455
2456 /* mballoc.c */
2457 extern const struct file_operations ext4_seq_mb_groups_fops;
2458 extern long ext4_mb_stats;
2459 extern long ext4_mb_max_to_scan;
2460 extern int ext4_mb_init(struct super_block *);
2461 extern int ext4_mb_release(struct super_block *);
2462 extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
2463 struct ext4_allocation_request *, int *);
2464 extern int ext4_mb_reserve_blocks(struct super_block *, int);
2465 extern void ext4_discard_preallocations(struct inode *);
2466 extern int __init ext4_init_mballoc(void);
2467 extern void ext4_exit_mballoc(void);
2468 extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
2469 struct buffer_head *bh, ext4_fsblk_t block,
2470 unsigned long count, int flags);
2471 extern int ext4_mb_alloc_groupinfo(struct super_block *sb,
2472 ext4_group_t ngroups);
2473 extern int ext4_mb_add_groupinfo(struct super_block *sb,
2474 ext4_group_t i, struct ext4_group_desc *desc);
2475 extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
2476 ext4_fsblk_t block, unsigned long count);
2477 extern int ext4_trim_fs(struct super_block *, struct fstrim_range *,
2478 unsigned long blkdev_flags);
2479
2480 /* inode.c */
2481 int ext4_inode_is_fast_symlink(struct inode *inode);
2482 struct buffer_head *ext4_getblk(handle_t *, struct inode *, ext4_lblk_t, int);
2483 struct buffer_head *ext4_bread(handle_t *, struct inode *, ext4_lblk_t, int);
2484 int ext4_get_block_write(struct inode *inode, sector_t iblock,
2485 struct buffer_head *bh_result, int create);
2486 int ext4_get_block_dax(struct inode *inode, sector_t iblock,
2487 struct buffer_head *bh_result, int create);
2488 int ext4_get_block(struct inode *inode, sector_t iblock,
2489 struct buffer_head *bh_result, int create);
2490 int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
2491 struct buffer_head *bh, int create);
2492 int ext4_walk_page_buffers(handle_t *handle,
2493 struct buffer_head *head,
2494 unsigned from,
2495 unsigned to,
2496 int *partial,
2497 int (*fn)(handle_t *handle,
2498 struct buffer_head *bh));
2499 int do_journal_get_write_access(handle_t *handle,
2500 struct buffer_head *bh);
2501 #define FALL_BACK_TO_NONDELALLOC 1
2502 #define CONVERT_INLINE_DATA 2
2503
2504 typedef enum {
2505 EXT4_IGET_NORMAL = 0,
2506 EXT4_IGET_SPECIAL = 0x0001, /* OK to iget a system inode */
2507 EXT4_IGET_HANDLE = 0x0002 /* Inode # is from a handle */
2508 } ext4_iget_flags;
2509
2510 extern struct inode *__ext4_iget(struct super_block *sb, unsigned long ino,
2511 ext4_iget_flags flags, const char *function,
2512 unsigned int line);
2513
2514 #define ext4_iget(sb, ino, flags) \
2515 __ext4_iget((sb), (ino), (flags), __func__, __LINE__)
2516
2517 extern int ext4_write_inode(struct inode *, struct writeback_control *);
2518 extern int ext4_setattr(struct dentry *, struct iattr *);
2519 extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
2520 struct kstat *stat);
2521 extern void ext4_evict_inode(struct inode *);
2522 extern void ext4_clear_inode(struct inode *);
2523 extern int ext4_sync_inode(handle_t *, struct inode *);
2524 extern void ext4_dirty_inode(struct inode *, int);
2525 extern int ext4_change_inode_journal_flag(struct inode *, int);
2526 extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
2527 extern int ext4_inode_attach_jinode(struct inode *inode);
2528 extern int ext4_can_truncate(struct inode *inode);
2529 extern void ext4_truncate(struct inode *);
2530 extern int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length);
2531 extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
2532 extern void ext4_set_inode_flags(struct inode *);
2533 extern void ext4_get_inode_flags(struct ext4_inode_info *);
2534 extern int ext4_alloc_da_blocks(struct inode *inode);
2535 extern void ext4_set_aops(struct inode *inode);
2536 extern int ext4_writepage_trans_blocks(struct inode *);
2537 extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
2538 extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
2539 loff_t lstart, loff_t lend);
2540 extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
2541 extern int ext4_filemap_fault(struct vm_area_struct *vma, struct vm_fault *vmf);
2542 extern qsize_t *ext4_get_reserved_space(struct inode *inode);
2543 extern void ext4_da_update_reserve_space(struct inode *inode,
2544 int used, int quota_claim);
2545
2546 /* indirect.c */
2547 extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
2548 struct ext4_map_blocks *map, int flags);
2549 extern ssize_t ext4_ind_direct_IO(struct kiocb *iocb, struct iov_iter *iter,
2550 loff_t offset);
2551 extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
2552 extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks);
2553 extern void ext4_ind_truncate(handle_t *, struct inode *inode);
2554 extern int ext4_ind_remove_space(handle_t *handle, struct inode *inode,
2555 ext4_lblk_t start, ext4_lblk_t end);
2556
2557 /* ioctl.c */
2558 extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
2559 extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
2560
2561 /* migrate.c */
2562 extern int ext4_ext_migrate(struct inode *);
2563 extern int ext4_ind_migrate(struct inode *inode);
2564
2565 /* namei.c */
2566 extern int ext4_dirent_csum_verify(struct inode *inode,
2567 struct ext4_dir_entry *dirent);
2568 extern int ext4_orphan_add(handle_t *, struct inode *);
2569 extern int ext4_orphan_del(handle_t *, struct inode *);
2570 extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
2571 __u32 start_minor_hash, __u32 *next_hash);
2572 extern int ext4_search_dir(struct buffer_head *bh,
2573 char *search_buf,
2574 int buf_size,
2575 struct inode *dir,
2576 struct ext4_filename *fname,
2577 const struct qstr *d_name,
2578 unsigned int offset,
2579 struct ext4_dir_entry_2 **res_dir);
2580 extern int ext4_generic_delete_entry(handle_t *handle,
2581 struct inode *dir,
2582 struct ext4_dir_entry_2 *de_del,
2583 struct buffer_head *bh,
2584 void *entry_buf,
2585 int buf_size,
2586 int csum_size);
2587 extern int ext4_empty_dir(struct inode *inode);
2588
2589 /* resize.c */
2590 extern void ext4_kvfree_array_rcu(void *to_free);
2591 extern int ext4_group_add(struct super_block *sb,
2592 struct ext4_new_group_data *input);
2593 extern int ext4_group_extend(struct super_block *sb,
2594 struct ext4_super_block *es,
2595 ext4_fsblk_t n_blocks_count);
2596 extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
2597
2598 /* super.c */
2599 extern int ext4_seq_options_show(struct seq_file *seq, void *offset);
2600 extern int ext4_calculate_overhead(struct super_block *sb);
2601 extern void ext4_superblock_csum_set(struct super_block *sb);
2602 extern void *ext4_kvmalloc(size_t size, gfp_t flags);
2603 extern void *ext4_kvzalloc(size_t size, gfp_t flags);
2604 extern int ext4_alloc_flex_bg_array(struct super_block *sb,
2605 ext4_group_t ngroup);
2606 extern const char *ext4_decode_error(struct super_block *sb, int errno,
2607 char nbuf[16]);
2608
2609 extern __printf(4, 5)
2610 void __ext4_error(struct super_block *, const char *, unsigned int,
2611 const char *, ...);
2612 extern __printf(5, 6)
2613 void __ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
2614 const char *, ...);
2615 extern __printf(5, 6)
2616 void __ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
2617 const char *, ...);
2618 extern void __ext4_std_error(struct super_block *, const char *,
2619 unsigned int, int);
2620 extern __printf(4, 5)
2621 void __ext4_abort(struct super_block *, const char *, unsigned int,
2622 const char *, ...);
2623 extern __printf(4, 5)
2624 void __ext4_warning(struct super_block *, const char *, unsigned int,
2625 const char *, ...);
2626 extern __printf(4, 5)
2627 void __ext4_warning_inode(const struct inode *inode, const char *function,
2628 unsigned int line, const char *fmt, ...);
2629 extern __printf(3, 4)
2630 void __ext4_msg(struct super_block *, const char *, const char *, ...);
2631 extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
2632 const char *, unsigned int, const char *);
2633 extern __printf(7, 8)
2634 void __ext4_grp_locked_error(const char *, unsigned int,
2635 struct super_block *, ext4_group_t,
2636 unsigned long, ext4_fsblk_t,
2637 const char *, ...);
2638
2639 #define EXT4_ERROR_INODE(inode, fmt, a...) \
2640 ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
2641
2642 #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
2643 ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
2644
2645 #define EXT4_ERROR_FILE(file, block, fmt, a...) \
2646 ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
2647
2648 #ifdef CONFIG_PRINTK
2649
2650 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2651 __ext4_error_inode(inode, func, line, block, fmt, ##__VA_ARGS__)
2652 #define ext4_error_file(file, func, line, block, fmt, ...) \
2653 __ext4_error_file(file, func, line, block, fmt, ##__VA_ARGS__)
2654 #define ext4_error(sb, fmt, ...) \
2655 __ext4_error(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2656 #define ext4_abort(sb, fmt, ...) \
2657 __ext4_abort(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2658 #define ext4_warning(sb, fmt, ...) \
2659 __ext4_warning(sb, __func__, __LINE__, fmt, ##__VA_ARGS__)
2660 #define ext4_warning_inode(inode, fmt, ...) \
2661 __ext4_warning_inode(inode, __func__, __LINE__, fmt, ##__VA_ARGS__)
2662 #define ext4_msg(sb, level, fmt, ...) \
2663 __ext4_msg(sb, level, fmt, ##__VA_ARGS__)
2664 #define dump_mmp_msg(sb, mmp, msg) \
2665 __dump_mmp_msg(sb, mmp, __func__, __LINE__, msg)
2666 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2667 __ext4_grp_locked_error(__func__, __LINE__, sb, grp, ino, block, \
2668 fmt, ##__VA_ARGS__)
2669
2670 #else
2671
2672 #define ext4_error_inode(inode, func, line, block, fmt, ...) \
2673 do { \
2674 no_printk(fmt, ##__VA_ARGS__); \
2675 __ext4_error_inode(inode, "", 0, block, " "); \
2676 } while (0)
2677 #define ext4_error_file(file, func, line, block, fmt, ...) \
2678 do { \
2679 no_printk(fmt, ##__VA_ARGS__); \
2680 __ext4_error_file(file, "", 0, block, " "); \
2681 } while (0)
2682 #define ext4_error(sb, fmt, ...) \
2683 do { \
2684 no_printk(fmt, ##__VA_ARGS__); \
2685 __ext4_error(sb, "", 0, " "); \
2686 } while (0)
2687 #define ext4_abort(sb, fmt, ...) \
2688 do { \
2689 no_printk(fmt, ##__VA_ARGS__); \
2690 __ext4_abort(sb, "", 0, " "); \
2691 } while (0)
2692 #define ext4_warning(sb, fmt, ...) \
2693 do { \
2694 no_printk(fmt, ##__VA_ARGS__); \
2695 __ext4_warning(sb, "", 0, " "); \
2696 } while (0)
2697 #define ext4_warning_inode(inode, fmt, ...) \
2698 do { \
2699 no_printk(fmt, ##__VA_ARGS__); \
2700 __ext4_warning_inode(inode, "", 0, " "); \
2701 } while (0)
2702 #define ext4_msg(sb, level, fmt, ...) \
2703 do { \
2704 no_printk(fmt, ##__VA_ARGS__); \
2705 __ext4_msg(sb, "", " "); \
2706 } while (0)
2707 #define dump_mmp_msg(sb, mmp, msg) \
2708 __dump_mmp_msg(sb, mmp, "", 0, "")
2709 #define ext4_grp_locked_error(sb, grp, ino, block, fmt, ...) \
2710 do { \
2711 no_printk(fmt, ##__VA_ARGS__); \
2712 __ext4_grp_locked_error("", 0, sb, grp, ino, block, " "); \
2713 } while (0)
2714
2715 #endif
2716
2717 extern void ext4_update_dynamic_rev(struct super_block *sb);
2718 extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
2719 __u32 compat);
2720 extern int ext4_update_rocompat_feature(handle_t *handle,
2721 struct super_block *sb, __u32 rocompat);
2722 extern int ext4_update_incompat_feature(handle_t *handle,
2723 struct super_block *sb, __u32 incompat);
2724 extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
2725 struct ext4_group_desc *bg);
2726 extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
2727 struct ext4_group_desc *bg);
2728 extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
2729 struct ext4_group_desc *bg);
2730 extern __u32 ext4_free_group_clusters(struct super_block *sb,
2731 struct ext4_group_desc *bg);
2732 extern __u32 ext4_free_inodes_count(struct super_block *sb,
2733 struct ext4_group_desc *bg);
2734 extern __u32 ext4_used_dirs_count(struct super_block *sb,
2735 struct ext4_group_desc *bg);
2736 extern __u32 ext4_itable_unused_count(struct super_block *sb,
2737 struct ext4_group_desc *bg);
2738 extern void ext4_block_bitmap_set(struct super_block *sb,
2739 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2740 extern void ext4_inode_bitmap_set(struct super_block *sb,
2741 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2742 extern void ext4_inode_table_set(struct super_block *sb,
2743 struct ext4_group_desc *bg, ext4_fsblk_t blk);
2744 extern void ext4_free_group_clusters_set(struct super_block *sb,
2745 struct ext4_group_desc *bg,
2746 __u32 count);
2747 extern void ext4_free_inodes_set(struct super_block *sb,
2748 struct ext4_group_desc *bg, __u32 count);
2749 extern void ext4_used_dirs_set(struct super_block *sb,
2750 struct ext4_group_desc *bg, __u32 count);
2751 extern void ext4_itable_unused_set(struct super_block *sb,
2752 struct ext4_group_desc *bg, __u32 count);
2753 extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
2754 struct ext4_group_desc *gdp);
2755 extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
2756 struct ext4_group_desc *gdp);
2757 extern int ext4_register_li_request(struct super_block *sb,
2758 ext4_group_t first_not_zeroed);
2759
ext4_has_group_desc_csum(struct super_block * sb)2760 static inline int ext4_has_group_desc_csum(struct super_block *sb)
2761 {
2762 return ext4_has_feature_gdt_csum(sb) ||
2763 EXT4_SB(sb)->s_chksum_driver != NULL;
2764 }
2765
ext4_has_metadata_csum(struct super_block * sb)2766 static inline int ext4_has_metadata_csum(struct super_block *sb)
2767 {
2768 WARN_ON_ONCE(ext4_has_feature_metadata_csum(sb) &&
2769 !EXT4_SB(sb)->s_chksum_driver);
2770
2771 return (EXT4_SB(sb)->s_chksum_driver != NULL);
2772 }
ext4_blocks_count(struct ext4_super_block * es)2773 static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
2774 {
2775 return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
2776 le32_to_cpu(es->s_blocks_count_lo);
2777 }
2778
ext4_r_blocks_count(struct ext4_super_block * es)2779 static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
2780 {
2781 return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
2782 le32_to_cpu(es->s_r_blocks_count_lo);
2783 }
2784
ext4_free_blocks_count(struct ext4_super_block * es)2785 static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
2786 {
2787 return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
2788 le32_to_cpu(es->s_free_blocks_count_lo);
2789 }
2790
ext4_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)2791 static inline void ext4_blocks_count_set(struct ext4_super_block *es,
2792 ext4_fsblk_t blk)
2793 {
2794 es->s_blocks_count_lo = cpu_to_le32((u32)blk);
2795 es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
2796 }
2797
ext4_free_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)2798 static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
2799 ext4_fsblk_t blk)
2800 {
2801 es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
2802 es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
2803 }
2804
ext4_r_blocks_count_set(struct ext4_super_block * es,ext4_fsblk_t blk)2805 static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
2806 ext4_fsblk_t blk)
2807 {
2808 es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
2809 es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
2810 }
2811
ext4_isize(struct ext4_inode * raw_inode)2812 static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
2813 {
2814 if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
2815 return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
2816 le32_to_cpu(raw_inode->i_size_lo);
2817 else
2818 return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
2819 }
2820
ext4_isize_set(struct ext4_inode * raw_inode,loff_t i_size)2821 static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
2822 {
2823 raw_inode->i_size_lo = cpu_to_le32(i_size);
2824 raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
2825 }
2826
2827 static inline
ext4_get_group_info(struct super_block * sb,ext4_group_t group)2828 struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
2829 ext4_group_t group)
2830 {
2831 struct ext4_group_info **grp_info;
2832 long indexv, indexh;
2833 BUG_ON(group >= EXT4_SB(sb)->s_groups_count);
2834 indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
2835 indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
2836 grp_info = sbi_array_rcu_deref(EXT4_SB(sb), s_group_info, indexv);
2837 return grp_info[indexh];
2838 }
2839
2840 /*
2841 * Reading s_groups_count requires using smp_rmb() afterwards. See
2842 * the locking protocol documented in the comments of ext4_group_add()
2843 * in resize.c
2844 */
ext4_get_groups_count(struct super_block * sb)2845 static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
2846 {
2847 ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
2848
2849 smp_rmb();
2850 return ngroups;
2851 }
2852
ext4_flex_group(struct ext4_sb_info * sbi,ext4_group_t block_group)2853 static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
2854 ext4_group_t block_group)
2855 {
2856 return block_group >> sbi->s_log_groups_per_flex;
2857 }
2858
ext4_flex_bg_size(struct ext4_sb_info * sbi)2859 static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
2860 {
2861 return 1 << sbi->s_log_groups_per_flex;
2862 }
2863
2864 #define ext4_std_error(sb, errno) \
2865 do { \
2866 if ((errno)) \
2867 __ext4_std_error((sb), __func__, __LINE__, (errno)); \
2868 } while (0)
2869
2870 #ifdef CONFIG_SMP
2871 /* Each CPU can accumulate percpu_counter_batch clusters in their local
2872 * counters. So we need to make sure we have free clusters more
2873 * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
2874 */
2875 #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
2876 #else
2877 #define EXT4_FREECLUSTERS_WATERMARK 0
2878 #endif
2879
2880 /* Update i_disksize. Requires i_mutex to avoid races with truncate */
ext4_update_i_disksize(struct inode * inode,loff_t newsize)2881 static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
2882 {
2883 WARN_ON_ONCE(S_ISREG(inode->i_mode) &&
2884 !mutex_is_locked(&inode->i_mutex));
2885 down_write(&EXT4_I(inode)->i_data_sem);
2886 if (newsize > EXT4_I(inode)->i_disksize)
2887 WRITE_ONCE(EXT4_I(inode)->i_disksize, newsize);
2888 up_write(&EXT4_I(inode)->i_data_sem);
2889 }
2890
2891 /* Update i_size, i_disksize. Requires i_mutex to avoid races with truncate */
ext4_update_inode_size(struct inode * inode,loff_t newsize)2892 static inline int ext4_update_inode_size(struct inode *inode, loff_t newsize)
2893 {
2894 int changed = 0;
2895
2896 if (newsize > inode->i_size) {
2897 i_size_write(inode, newsize);
2898 changed = 1;
2899 }
2900 if (newsize > EXT4_I(inode)->i_disksize) {
2901 ext4_update_i_disksize(inode, newsize);
2902 changed |= 2;
2903 }
2904 return changed;
2905 }
2906
2907 int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
2908 loff_t len);
2909
2910 struct ext4_group_info {
2911 unsigned long bb_state;
2912 struct rb_root bb_free_root;
2913 ext4_grpblk_t bb_first_free; /* first free block */
2914 ext4_grpblk_t bb_free; /* total free blocks */
2915 ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
2916 ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
2917 struct list_head bb_prealloc_list;
2918 #ifdef DOUBLE_CHECK
2919 void *bb_bitmap;
2920 #endif
2921 struct rw_semaphore alloc_sem;
2922 ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
2923 * regions, index is order.
2924 * bb_counters[3] = 5 means
2925 * 5 free 8-block regions. */
2926 };
2927
2928 #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
2929 #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
2930 #define EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT 2
2931 #define EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT 3
2932
2933 #define EXT4_MB_GRP_NEED_INIT(grp) \
2934 (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
2935 #define EXT4_MB_GRP_BBITMAP_CORRUPT(grp) \
2936 (test_bit(EXT4_GROUP_INFO_BBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2937 #define EXT4_MB_GRP_IBITMAP_CORRUPT(grp) \
2938 (test_bit(EXT4_GROUP_INFO_IBITMAP_CORRUPT_BIT, &((grp)->bb_state)))
2939
2940 #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
2941 (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2942 #define EXT4_MB_GRP_SET_TRIMMED(grp) \
2943 (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2944 #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
2945 (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
2946
2947 #define EXT4_MAX_CONTENTION 8
2948 #define EXT4_CONTENTION_THRESHOLD 2
2949
ext4_group_lock_ptr(struct super_block * sb,ext4_group_t group)2950 static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
2951 ext4_group_t group)
2952 {
2953 return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
2954 }
2955
2956 /*
2957 * Returns true if the filesystem is busy enough that attempts to
2958 * access the block group locks has run into contention.
2959 */
ext4_fs_is_busy(struct ext4_sb_info * sbi)2960 static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
2961 {
2962 return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
2963 }
2964
ext4_lock_group(struct super_block * sb,ext4_group_t group)2965 static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
2966 {
2967 spinlock_t *lock = ext4_group_lock_ptr(sb, group);
2968 if (spin_trylock(lock))
2969 /*
2970 * We're able to grab the lock right away, so drop the
2971 * lock contention counter.
2972 */
2973 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
2974 else {
2975 /*
2976 * The lock is busy, so bump the contention counter,
2977 * and then wait on the spin lock.
2978 */
2979 atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
2980 EXT4_MAX_CONTENTION);
2981 spin_lock(lock);
2982 }
2983 }
2984
ext4_unlock_group(struct super_block * sb,ext4_group_t group)2985 static inline void ext4_unlock_group(struct super_block *sb,
2986 ext4_group_t group)
2987 {
2988 spin_unlock(ext4_group_lock_ptr(sb, group));
2989 }
2990
2991 /*
2992 * Block validity checking
2993 */
2994 #define ext4_check_indirect_blockref(inode, bh) \
2995 ext4_check_blockref(__func__, __LINE__, inode, \
2996 (__le32 *)(bh)->b_data, \
2997 EXT4_ADDR_PER_BLOCK((inode)->i_sb))
2998
2999 #define ext4_ind_check_inode(inode) \
3000 ext4_check_blockref(__func__, __LINE__, inode, \
3001 EXT4_I(inode)->i_data, \
3002 EXT4_NDIR_BLOCKS)
3003
3004 /*
3005 * Inodes and files operations
3006 */
3007
3008 /* dir.c */
3009 extern const struct file_operations ext4_dir_operations;
3010
3011 /* file.c */
3012 extern const struct inode_operations ext4_file_inode_operations;
3013 extern const struct file_operations ext4_file_operations;
3014 extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
3015
3016 /* inline.c */
3017 extern int ext4_get_max_inline_size(struct inode *inode);
3018 extern int ext4_find_inline_data_nolock(struct inode *inode);
3019 extern int ext4_init_inline_data(handle_t *handle, struct inode *inode,
3020 unsigned int len);
3021 extern int ext4_destroy_inline_data(handle_t *handle, struct inode *inode);
3022
3023 extern int ext4_readpage_inline(struct inode *inode, struct page *page);
3024 extern int ext4_try_to_write_inline_data(struct address_space *mapping,
3025 struct inode *inode,
3026 loff_t pos, unsigned len,
3027 unsigned flags,
3028 struct page **pagep);
3029 extern int ext4_write_inline_data_end(struct inode *inode,
3030 loff_t pos, unsigned len,
3031 unsigned copied,
3032 struct page *page);
3033 extern struct buffer_head *
3034 ext4_journalled_write_inline_data(struct inode *inode,
3035 unsigned len,
3036 struct page *page);
3037 extern int ext4_da_write_inline_data_begin(struct address_space *mapping,
3038 struct inode *inode,
3039 loff_t pos, unsigned len,
3040 unsigned flags,
3041 struct page **pagep,
3042 void **fsdata);
3043 extern int ext4_da_write_inline_data_end(struct inode *inode, loff_t pos,
3044 unsigned len, unsigned copied,
3045 struct page *page);
3046 extern int ext4_try_add_inline_entry(handle_t *handle,
3047 struct ext4_filename *fname,
3048 struct dentry *dentry,
3049 struct inode *inode);
3050 extern int ext4_try_create_inline_dir(handle_t *handle,
3051 struct inode *parent,
3052 struct inode *inode);
3053 extern int ext4_read_inline_dir(struct file *filp,
3054 struct dir_context *ctx,
3055 int *has_inline_data);
3056 extern int htree_inlinedir_to_tree(struct file *dir_file,
3057 struct inode *dir, ext4_lblk_t block,
3058 struct dx_hash_info *hinfo,
3059 __u32 start_hash, __u32 start_minor_hash,
3060 int *has_inline_data);
3061 extern struct buffer_head *ext4_find_inline_entry(struct inode *dir,
3062 struct ext4_filename *fname,
3063 const struct qstr *d_name,
3064 struct ext4_dir_entry_2 **res_dir,
3065 int *has_inline_data);
3066 extern int ext4_delete_inline_entry(handle_t *handle,
3067 struct inode *dir,
3068 struct ext4_dir_entry_2 *de_del,
3069 struct buffer_head *bh,
3070 int *has_inline_data);
3071 extern int empty_inline_dir(struct inode *dir, int *has_inline_data);
3072 extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode,
3073 struct ext4_dir_entry_2 **parent_de,
3074 int *retval);
3075 extern int ext4_inline_data_fiemap(struct inode *inode,
3076 struct fiemap_extent_info *fieinfo,
3077 int *has_inline, __u64 start, __u64 len);
3078 extern void ext4_inline_data_truncate(struct inode *inode, int *has_inline);
3079
3080 extern int ext4_convert_inline_data(struct inode *inode);
3081
ext4_has_inline_data(struct inode * inode)3082 static inline int ext4_has_inline_data(struct inode *inode)
3083 {
3084 return ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA) &&
3085 EXT4_I(inode)->i_inline_off;
3086 }
3087
3088 /* namei.c */
3089 extern const struct inode_operations ext4_dir_inode_operations;
3090 extern const struct inode_operations ext4_special_inode_operations;
3091 extern struct dentry *ext4_get_parent(struct dentry *child);
3092 extern struct ext4_dir_entry_2 *ext4_init_dot_dotdot(struct inode *inode,
3093 struct ext4_dir_entry_2 *de,
3094 int blocksize, int csum_size,
3095 unsigned int parent_ino, int dotdot_real_len);
3096 extern void initialize_dirent_tail(struct ext4_dir_entry_tail *t,
3097 unsigned int blocksize);
3098 extern int ext4_handle_dirty_dirent_node(handle_t *handle,
3099 struct inode *inode,
3100 struct buffer_head *bh);
3101 #define S_SHIFT 12
3102 static unsigned char ext4_type_by_mode[S_IFMT >> S_SHIFT] = {
3103 [S_IFREG >> S_SHIFT] = EXT4_FT_REG_FILE,
3104 [S_IFDIR >> S_SHIFT] = EXT4_FT_DIR,
3105 [S_IFCHR >> S_SHIFT] = EXT4_FT_CHRDEV,
3106 [S_IFBLK >> S_SHIFT] = EXT4_FT_BLKDEV,
3107 [S_IFIFO >> S_SHIFT] = EXT4_FT_FIFO,
3108 [S_IFSOCK >> S_SHIFT] = EXT4_FT_SOCK,
3109 [S_IFLNK >> S_SHIFT] = EXT4_FT_SYMLINK,
3110 };
3111
ext4_set_de_type(struct super_block * sb,struct ext4_dir_entry_2 * de,umode_t mode)3112 static inline void ext4_set_de_type(struct super_block *sb,
3113 struct ext4_dir_entry_2 *de,
3114 umode_t mode) {
3115 if (ext4_has_feature_filetype(sb))
3116 de->file_type = ext4_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
3117 }
3118
3119 /* readpages.c */
3120 extern int ext4_mpage_readpages(struct address_space *mapping,
3121 struct list_head *pages, struct page *page,
3122 unsigned nr_pages);
3123
3124 /* symlink.c */
3125 extern const struct inode_operations ext4_encrypted_symlink_inode_operations;
3126 extern const struct inode_operations ext4_symlink_inode_operations;
3127 extern const struct inode_operations ext4_fast_symlink_inode_operations;
3128
3129 /* sysfs.c */
3130 extern int ext4_register_sysfs(struct super_block *sb);
3131 extern void ext4_unregister_sysfs(struct super_block *sb);
3132 extern int __init ext4_init_sysfs(void);
3133 extern void ext4_exit_sysfs(void);
3134
3135 /* block_validity */
3136 extern void ext4_release_system_zone(struct super_block *sb);
3137 extern int ext4_setup_system_zone(struct super_block *sb);
3138 extern int __init ext4_init_system_zone(void);
3139 extern void ext4_exit_system_zone(void);
3140 extern int ext4_inode_block_valid(struct inode *inode,
3141 ext4_fsblk_t start_blk,
3142 unsigned int count);
3143 extern int ext4_check_blockref(const char *, unsigned int,
3144 struct inode *, __le32 *, unsigned int);
3145
3146 /* extents.c */
3147 struct ext4_ext_path;
3148 struct ext4_extent;
3149
3150 /*
3151 * Maximum number of logical blocks in a file; ext4_extent's ee_block is
3152 * __le32.
3153 */
3154 #define EXT_MAX_BLOCKS 0xffffffff
3155
3156 extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
3157 extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
3158 extern int ext4_ext_index_trans_blocks(struct inode *inode, int extents);
3159 extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
3160 struct ext4_map_blocks *map, int flags);
3161 extern void ext4_ext_truncate(handle_t *, struct inode *);
3162 extern int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
3163 ext4_lblk_t end);
3164 extern void ext4_ext_init(struct super_block *);
3165 extern void ext4_ext_release(struct super_block *);
3166 extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
3167 loff_t len);
3168 extern int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
3169 loff_t offset, ssize_t len);
3170 extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
3171 struct ext4_map_blocks *map, int flags);
3172 extern int ext4_ext_calc_metadata_amount(struct inode *inode,
3173 ext4_lblk_t lblocks);
3174 extern int ext4_ext_calc_credits_for_single_extent(struct inode *inode,
3175 int num,
3176 struct ext4_ext_path *path);
3177 extern int ext4_can_extents_be_merged(struct inode *inode,
3178 struct ext4_extent *ex1,
3179 struct ext4_extent *ex2);
3180 extern int ext4_ext_insert_extent(handle_t *, struct inode *,
3181 struct ext4_ext_path **,
3182 struct ext4_extent *, int);
3183 extern struct ext4_ext_path *ext4_find_extent(struct inode *, ext4_lblk_t,
3184 struct ext4_ext_path **,
3185 int flags);
3186 extern void ext4_ext_drop_refs(struct ext4_ext_path *);
3187 extern int ext4_ext_check_inode(struct inode *inode);
3188 extern int ext4_find_delalloc_range(struct inode *inode,
3189 ext4_lblk_t lblk_start,
3190 ext4_lblk_t lblk_end);
3191 extern int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk);
3192 extern ext4_lblk_t ext4_ext_next_allocated_block(struct ext4_ext_path *path);
3193 extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
3194 __u64 start, __u64 len);
3195 extern int ext4_ext_precache(struct inode *inode);
3196 extern int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len);
3197 extern int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len);
3198 extern int ext4_swap_extents(handle_t *handle, struct inode *inode1,
3199 struct inode *inode2, ext4_lblk_t lblk1,
3200 ext4_lblk_t lblk2, ext4_lblk_t count,
3201 int mark_unwritten,int *err);
3202
3203 /* move_extent.c */
3204 extern void ext4_double_down_write_data_sem(struct inode *first,
3205 struct inode *second);
3206 extern void ext4_double_up_write_data_sem(struct inode *orig_inode,
3207 struct inode *donor_inode);
3208 extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
3209 __u64 start_orig, __u64 start_donor,
3210 __u64 len, __u64 *moved_len);
3211
3212 /* page-io.c */
3213 extern int __init ext4_init_pageio(void);
3214 extern void ext4_exit_pageio(void);
3215 extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
3216 extern ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end);
3217 extern int ext4_put_io_end(ext4_io_end_t *io_end);
3218 extern void ext4_put_io_end_defer(ext4_io_end_t *io_end);
3219 extern void ext4_io_submit_init(struct ext4_io_submit *io,
3220 struct writeback_control *wbc);
3221 extern void ext4_end_io_rsv_work(struct work_struct *work);
3222 extern void ext4_io_submit(struct ext4_io_submit *io);
3223 extern int ext4_bio_write_page(struct ext4_io_submit *io,
3224 struct page *page,
3225 int len,
3226 struct writeback_control *wbc,
3227 bool keep_towrite);
3228
3229 /* mmp.c */
3230 extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
3231
3232 /*
3233 * Add new method to test whether block and inode bitmaps are properly
3234 * initialized. With uninit_bg reading the block from disk is not enough
3235 * to mark the bitmap uptodate. We need to also zero-out the bitmap
3236 */
3237 #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
3238
bitmap_uptodate(struct buffer_head * bh)3239 static inline int bitmap_uptodate(struct buffer_head *bh)
3240 {
3241 return (buffer_uptodate(bh) &&
3242 test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
3243 }
set_bitmap_uptodate(struct buffer_head * bh)3244 static inline void set_bitmap_uptodate(struct buffer_head *bh)
3245 {
3246 set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
3247 }
3248
3249 /*
3250 * Disable DIO read nolock optimization, so new dioreaders will be forced
3251 * to grab i_mutex
3252 */
ext4_inode_block_unlocked_dio(struct inode * inode)3253 static inline void ext4_inode_block_unlocked_dio(struct inode *inode)
3254 {
3255 ext4_set_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
3256 smp_mb();
3257 }
ext4_inode_resume_unlocked_dio(struct inode * inode)3258 static inline void ext4_inode_resume_unlocked_dio(struct inode *inode)
3259 {
3260 smp_mb();
3261 ext4_clear_inode_state(inode, EXT4_STATE_DIOREAD_LOCK);
3262 }
3263
3264 #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
3265
3266 /* For ioend & aio unwritten conversion wait queues */
3267 #define EXT4_WQ_HASH_SZ 37
3268 #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
3269 EXT4_WQ_HASH_SZ])
3270 #define ext4_aio_mutex(v) (&ext4__aio_mutex[((unsigned long)(v)) %\
3271 EXT4_WQ_HASH_SZ])
3272 extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
3273 extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ];
3274
3275 #define EXT4_RESIZING 0
3276 extern int ext4_resize_begin(struct super_block *sb);
3277 extern void ext4_resize_end(struct super_block *sb);
3278
3279 #endif /* __KERNEL__ */
3280
3281 #define EFSBADCRC EBADMSG /* Bad CRC detected */
3282 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */
3283
3284 #endif /* _EXT4_H */
3285